Biomass Gasification and Solid Oxide Fuel Cell

DISS. ETH NO.
17856
Electricity from wood through the combination

of gasification and solid oxide fuel cells
Systems analysis and Proof-of-concept
A dissertation submitted to
ETH ZURICH
for the degree of

Doctor of Sciences
presented by
Florian-Patrice Nagel
Dipl.-Ing. Universität Stuttgart
Born 05.12.1977
citizen of
the Federal Republic of Germany
accepted on the recommendation of

Prof. Dr. A. Wokaun, examiner
Dr. F. Maréchal, co-examiner
Dr. S. Biollaz, co-examiner
2008
Für meine Eltern
in Liebe und Dankbarkeit
Danksagung
If we knew what it was we were doing, it would not be called research, would it?
Albert Einstein
Prof. Dr. Alexander Wokaun gilt mein besonderer Dank für die Möglichkeit diese Disser-
tation anzufertigen und die stets vorzügliche Betreuung dieser Arbeit. Ihr Vertrauen, Herr
Prof. Wokaun, in meine Lernfähigkeit und Beharrlichkeit war für mich Verpflichtung und
Ansporn zugleich. Die Quartalsbesprechungen haben meine Arbeit wesentlich beein-
flusst und bereichert.
Die herzliche Aufnahme in die „Thermal Process Engineering“ Gruppe, die Formulierung
des Dissertationsthemas und die Übernahme des Koreferats kann ich Herrn Dr. Serge
Biollaz nicht hoch genug anrechnen. Ich danke Dir, Serge, für die vielen anregenden
Diskussionen die aufgrund Deiner Erfahrung fast immer zur Verbesserung des Ver-
suchsaufbaus geführt haben. Vielen herzlichen Dank für die freundschaftliche Zusam-
menarbeit und für die Unterstützung in Zeiten persönlicher Not und Ungewissheit ob ich
diese Arbeit gesundheitsbedingt abschließen kann.
Für die Übernahme des Koreferats und interessante Diskussionen zum Thema der Mo-
dellierung bedanke ich mich herzlich bei Dr. François Maréchal.
Dr. Tilman Schildhauer könnte ich eine eigene Danksagung widmen und hätte immer
noch nicht genug Platz um meinen Dank ausreichend auszudrücken. Die unzähligen Dis-
kussionen, Fragen und Anregungen in allen Bereichen der Dissertation haben mich nicht
nur akademisch sondern auch menschlich bereichert. Danke für die viele Zeit die Du,
Tilman, in mich und diese Dissertation investiert hast, ohne die diese Arbeit nicht in die-
ser Form möglich gewesen wäre.
Dr. Samuel Stucki danke ich herzlich für die freundliche Aufnahme in das Labor für
Energie- und Stoffkreisläufe und die Einführung in die Welt der Elektrochemie.
Dr. Alexander Schuler von der Hexis AG danke ich für die fachliche Unterstützung und
die Bereitstellung mehrerer Brennstoffzellenmodule ohne die der experimentelle Teil die-
ser Arbeit nicht denkbar gewesen wäre.
Für die entspannenden Gespräche in den Kaffeepausen die mir stets neuen Schwung für
meine Arbeit gaben danke ich Gisela Herlein. Ohne Dich, Gisela, wäre meine Zeit am
LEM nur halb so schön gewesen.
Ein besonderer Dank gilt dem Technikteam des LEM. Für die Lösung unlösbarer Steue-
rungsprobleme beim Aufbau des weltweit ersten Vergaser- Brennstoffzellen Systems
danke ich Peter Hottinger. Peter, der erste HoVer-BZM Versuch, bei dem wir uns vorka-
men wie Spaceshuttlepiloten beim Landeanflug auf den alten Hong Kong Airport werde
ich nie vergessen.
Für sein Talent auch die verrücktesten Ideen aufs Papier und zum schlussendlich sogar
zum funktionieren zu bringen danke ich Thomas Marti. Thomas, ich werde immer mit ei-
nem Lächeln im Gesicht an die Zeit mit dem Schraubenschlüssel in der Hand zurück
denken.
Peter Binkert danke ich post mortem für seine Geduld mit meinen Wünschen und die
Beharrlichkeit an der Dreh- und Fräsmaschine. Spaß hat es gemacht, Peter.
Die perfekten Zyklone und viele andere Bauteile im HoVex-Veruschsaufbau verdanke ich
Marcel Hottiger.
Für die unzähligen Stunden mit dem Schweißapparat in der Hand durch die der HoVex-
Aufbau erst Realität wurde danke ich Markus Schriber.
Die Messungen der abstrusesten Gase, die so manche Überraschung an den Tag brach-
ten, hat Herr Jörg Schneebeli durchgeführt. Danke Jörg, für deine stete Bereitschaft mir
in Diagnostikfragen zu helfen.
Weiterer Dank gebührt Dr. Frédéric Vogel, Christian Pitta, Dr. Sudip Ghosh, Manuel
Damsohn, Markus Jenne, Martin Künstle, Stefan Decker, Mohand Nait-Atmane, Thorsten
Schulz, Jan Kopyscinski und Johannes Judex für die Freundschaft sowie allen anderen
nicht namentlich genannten LEM-Mitarbeiter mit denen ich arbeiten durfte und die immer
noch Zeit für das ein oder andere Späßchen gefunden haben.
Last but not least möchte ich für die außerordentlich entspannte und positive Arbeitsat-
mosphäre am PSI und dem AXPO Naturstrom Fond für die finanzielle Unterstützung
meiner Arbeit danken.
Meine Eltern, Dr. Karl-Heinz Nagel und Chantal Nagel, und meine beiden Schwestern,
Nathalie McCaughey und Sandra Nagel Bolliger, haben mich stets unterstützt und mir
durch eine unfallbedingte Zeit voller dunkler Zweifel hinweggeholfen. Ihre Liebe und ihr
Vertrauen in mich und nicht zuletzt die familiäre Geborgenheit haben mir die notwendige
Kraft gegeben meine Reise in die Welt der Wissenschaft fortzusetzen und letztendlich
abzuschließen.
Villigen, im Juli 2008

Florian-Patrice Nagel
Abstract
The availability of energy in general and electricity in particular has become a major con-
cern given the growing global population, numerous fast developing nations and deplet-
ing fossil fuel reserves. Further, the excessive use of fossil fuels with their attributed car-
bon dioxide emissions has lead to changes in the global ecosystem with negative conse-
quences for life on earth. Against this background it comes as no surprise that renewable,
carbon dioxide neutral energy sources have gained increased interest recently. Biomass
has a large potential as renewable and carbon dioxide neutral feedstock for electricity
generation but is comparably expensive. An approach to overcome this economical
drawback is the combination of high-temperature fuel cells with biomass gasification
processes, which is commonly referred to as "Biomass Integrated Gasification Fuel Cell"
systems (B-IGFC). Solid oxide fuel cells (SOFC) are considered as promising candidates
for the application in B-IGFC system due to their less stringent requirements to the fuel
gas quality compared to other fuel cell types and their capability of directly oxidizing car-
bonaceous fuel gases with high efficiencies at small scales. However, the composition of
the fuel gas plays an important role for its conversion through SOFCs which leads to
strong system interactions that must be well understood to allow for the exploitation of
the full potential of the B-IGFC approach. This thesis presents the demonstration of the
B-IGFC technology on kW-scale and a thermo-economic system analysis aiming at the
identification of promising B-IGFC systems with power outputs around 1 MWel.
The PSI B-IGFC system concept comprises an updraft biomass gasification reactor, a
hot gas cyclone, a catalytic partial oxidation (CPO) unit and a 1 kW Hexis SOFC system.
The experiments conducted in this thesis initially concentrated on the characterization of
all processes steps in the PSI B-IGFC system. The first milestone was reached with the
stable operation of the lab-scale updraft gasifier over 165 h non-stop. Experiments with
short stacks showed that tars are to some extent a fuel for SOFCs. A commercial CPO
catalyst was investigated, revealing satisfying conversion performance for not only oxy-
genated tars and aromatics but also organic sulfur compounds. The PSI B-IGFC concept
was subsequently operated for 28 h non-stop employing the above mentioned 1 kW
Hexis SOFC system. Compared to operation with partially oxidized methane, the SOFC
delivered 40 % less current when operated with water and nitrogen diluted producer gas.
Overall, the demonstration unit was operated without problems and valuable experience
for future improvements was gathered. The application of effective means for the re-
moval of micro particles has been identified as very important in this respect.
The system analysis is based on a finite volume SOFC model developed in this thesis
that includes all relevant charge, mass and heat transport processes. Cell internal steam
I
reforming is considered through a Langmuir-Hinshelwood type applied kinetic model. For
the system analysis, the model was applied to anode- and electrolyte-supported planar
cell designs with co- and counter-current flow pattern as well as the standard tubular cell
design promoted by the Siemens AG. The model was validated against experimental and
literature data and proved to behave physically correct with smaller mass and energy
balance errors than comparable models. The robustness of the general trends predicted
for various fuel gas compositions was attested by means of a sensitivity analysis.
Based on measured producer gas compositions originating from downdraft and updraft
gasification processes as well as the fluidized bed steam gasification process, seven
B-IGFC systems with different gas processing strategies were defined. The correspond-
ing gas compositions at the fuel inlet of the SOFCs were computed with ASPEN PLUS.
Compared to pre-reformed methane, the power output of SOFCs decreases by 25 to
70 % when operated with producer gases. The magnitude of the power output decrease
mainly depends on the degree of dilution of the producer gas and on the anode catalyst
activity which both determine the importance of activation losses. The tubular cell design
appears to feature a highly active anode catalyst according to the employed model pa-
rameters. Diffusion limitations in contrast do not gain importance for any of the investi-
gated producer gases and cell designs. The required air-to-fuel ratio to maintain a given
mean cell temperature increases with decreasing internal reforming potential of the fuel
gas. This mechanism is considerably less pronounced for counter-current cell designs
compared to their co-current correspondents. The thermal stress resulting from the op-
eration of SOFCs with producer gases is generally lower than for operation with pre-
reformed methane. This may slow down the corresponding cell degradation processes.
The power outputs predicted by the SOFC model were used as input for overall system
simulations performed with ASPEN PLUS aiming at the investigation of the interactions
between the gasification processes, gas processing technologies and SOFC designs.
The heat integration was conducted by means of a generalized heat exchanger network.
The comparably low operational temperature of the zinc oxide trap beds, generally em-
ployed for the removal of hydrogen sulfide, calls for additional humidification of the pro-
ducer gases to prevent thermodynamic carbon deposition except in the system where a
catalytic partial oxidation unit is employed for tar conversion. The net system efficiency is
preeminently determined by the cold gas efficiency of the biomass gasification process
and the auxiliary power requirements, which directly correlate with the air-to-fuel ratios.
The adiabatic methanation and catalytic partial oxidation are effective means to lower the
required air-to-fuel ratios by increasing the internal reforming potential of the correspond-
ing producer gases, while simultaneously converting undesired organic species.
Consequently with 32.1 %, the highest net AC system efficiency in the analysis is
reached by the combination of the updraft gasification with an adiabatic methanation and
II
the tubular cell design. The updraft gasification yields a producer gas with high cold gas
efficiency and considerable internal reforming potential, which is further increased
through the adiabatic methanation. The high internal reforming potential leads to a very
low air-to-fuel ratio and thus low auxiliary power needs. The steam reforming of the pro-
duced methane can be interpreted as the final gasification step using SOFC waste heat.
With the mass and energy flows determined for of the different B-IGFC systems through
the overall system simulations, all major equipment pieces were sized and pricewise
rated using pertinent cost functions. The fuel cell balance of plant equipment and the cor-
responding heat exchangers in particular are the most important cost drivers amounting
for up to 50 %, while the fuel cell itself accounts for up to 25 % of the total system costs.
This emphasizes the importance of the air-to-fuel ratios which together with the opera-
tional cell temperatures determine the size of the heat exchangers. The total plant costs
were related to the corresponding net system power outputs to yield the specific plant
costs. The downdraft gasification based systems have the highest specific plant costs
due to costly wood drying, large heat exchangers and low power outputs resulting from
high activation losses as a consequence of gas dilution. The fluidized bed steam gasifi-
cation based systems yield lower specific plant costs, mainly because of low activation
losses and thus high power outputs. The updraft gasification based systems feature the
lowest specific plant costs as a result of low total plant costs and auxiliary power needs.
The power production costs (PPC) relate the total plant costs to the corresponding sys-
tem efficiencies. The basis for the calculations was a detailed cost analysis for a given
planning horizon including costs not only for operation and maintenance, feedstock and
utilities but also capital costs resulting from plant costs depreciation and interests. The
annuity method was employed to evenly distribute the net value of all costs throughout
the planning horizon. On average, 40 % of the PPC arise from fuel expenses, the re-
mainder being attributed to capital costs. Hence, reduction of the specific plant costs is
considerably more effective than system efficiency increases with respect to PPC reduc-
tion. With 0.1154 €/kWhel, the updraft gasification based system discussed above yields
the lowest PPCs in the analysis. Increasing the operational voltage of the SOFC from
0.6 V to 0.7 V results in a net system efficiency increase from 32.1 % to 37.1 %. With
0.1912 €/kWhel, the resulting reduced power output yields however considerably higher
PPCs. In contrast, reducing the auxiliary power needs by adjusting the air-to-fuel ratio to
the lowest possible value leads to higher net system efficiencies and lower specific plant
costs, thus considerably lowering the PPCs.
Finally, the revenues from heat sales are crucial for the economical viability of the inves-
tigated B-IGFC systems without bottoming cycles. Future systems analysis should focus
on B-IGFC systems with bottoming cycles. Therefore, the developed SOFC model has to
be integrated in Flowsheeting- Software packages such as e.g. ASPEN PLUS.
III
IV
Zusammenfassung
Angesichts der wachsenden Weltbevölkerung, zahlreicher wirtschaftlich aufsteigernder

Länder und zur Neige gehender fossiler Energiequellen ist die Verfügbarkeit von Energie
und besonders von Elektrizität zu einem Besorgnis geworden. Zudem hat der übermäßi-
ge Verbrauch fossiler Energiequellen durch die damit verbundenen Kohlendioxidemissi-
onen zu bedenklichen Veränderungen des Ökosystems der Erde geführt. Vor diesem
Hintergrund ist es nicht verwunderlich, dass erneuerbare, CO2-neutrale Energiequellen
zunehmend an Interesse gewonnen haben. Unter den erneuerbaren Energien hat Bio-
masse ein vielversprechendes Potential für die Elektrizitätserzeugung, ist allerdings ver-
gleichsweise teuer. Ein Ansatz um diesen Nachteil auszugleichen ist die Kopplung der
Hochtemperaturbrennstoffzellen-Technologie mit der Biomassevergasungstechnologie
der gemeinhin als "Biomass Integrated Gasification Fuel Cell Systems" (B-IGFC) be-
zeichnet wird. Aufgrund Ihrer geringeren Anforderungen an die Brenngasreinheit im Ver-
gleich zu anderen Brennstoffzellen und der Möglichkeit kohlenwasserstoffhaltige Brenn-
gase im kleinen Leistungsbereich hocheffizient direkt zu Strom zu wandeln werden Fest-
keramikbrennstoffzellen (SOFC) als sehr interessant für den Einsatz in B-IGFC Syste-
men eingestuft. Die Zusammensetzung des Brenngases spielt eine besonders wichtige
Rolle für dessen elektrochemische Umwandlung in SOFCs, was zu starken Wechselwir-
kungen zwischen den einzelnen Prozessen in B-IGFC Systemen führt. Diese müssen
verstanden werden um das volle Potential der B-IGFC Technologie auszunutzen.
In der vorliegenden Arbeit wird über die Demonstration der B-IGFC Technologie im kW-
Maßstab berichtet. Zudem wird eine thermo-ökonomische Systemanalyse vorgestellt, die
zur Identifikation viel versprechender B-IGFC Systemkonzepte im 1 MWel Leistungsbe-
reich durchgeführt wurde.
Das PSI B-IGFC Systemkonzept umfasst einen Gegenstromholzvergaser, einen Heiß-
gaszyklon, eine katalytisch partielle Oxidationsstufe sowie ein 1 kW SOFC- System der
Hexis AG. Zunächst wurden Experimente zur Charakterisierung aller oben genannten
Prozessschritte durchgeführt. Ein erster Meilenstein wurde mit dem stabilen und unun-
terbrochenen Betrieb des Gegenstromholzvergasers über 165 Stunden erreicht. Versu-
che mit Experimentalzellstapeln haben ergeben, dass Teere bis zu einem gewissen Grad
als Brennstoff für SOFCs angesehen werden können. Untersuchungen eines kommer-
ziellen Katalysators für die katalytisch partielle Oxidation haben befriedigende Umsatzra-
ten sauerstoffhaltiger und aromatischer Teere sowie organischer Schwefelverbindungen
bestätigt. Das PSI B-IGFC System wurde im Anschluss an die Charakterisierung aller
wichtigen Prozessschritte ununterbrochen für eine Dauer von 28 Stunden betrieben. Im
Vergleich zum Betrieb des SOFC- Systems mit partiell oxidiertem Methan wurde mit
V
Holzgas aufgrund dessen Verdünnung mit Stickstoff und Wasser 40 % weniger Strom
erzeugt. Insgesamt konnte der Versuchsaufbau ohne nennenswerte Probleme betrieben
und wertvolle Erfahrungen für künftige Verbesserungen gesammelt werden. Insbesonde-
re hat sich gezeigt, dass eine effiziente Kleinstpartikelabscheidung für künftige Langzeit-
experimente von hoher Wichtigkeit ist.
Die in der vorliegenden Arbeit beschriebenen Ergebnisse der Systemanalyse beruht auf
einem eigens dafür entwickelten SOFC Modell. Das Modell wurde nach dem Finite-
Volumen-Ansatz aufgebaut und berücksichtigt alle wichtigen Ladungs-, Masse- und
Wärmetransportprozesse die in SOFCs stattfinden. Mögliche zellinterne Dampfreformie-
rungsreaktionen werden über eine Langmuir-Hinshelwood Kinetik beschrieben. Für die
Systemanalyse wurden planare Anoden- und Elektrolyt-gestützte Zelltypen in Gleich-
und Gegenstromausführung, sowie der röhrenförmige Zelltyp der Siemens AG mit dem
Modell abgebildet und gegen Messdaten und Literaturwerte validiert. Es konnte gezeigt
werden, dass das aufgebaute Model physikalisch sinnvolle Ergebnisse liefert. Die er-
reichte Schließung der Massen- und Energiebilanzen ist deutlich besser als die ver-
gleichbarer Modelle in der einschlägigen Literatur. Um sicherzustellen, dass die vorher-
gesagten Trends für verschiedene Brenngaszusammensetzungen nicht die Folge der
festgelegten Modellparameter ist, wurde eine entsprechende Sensitivitätsanalyse durch-
geführt. Die vom Modell vorhergesagten Trends haben sich als sehr robust erwiesen.
Aufbauend auf gemessenen Zusammensetzungen von Produktgasen aus Gleich- und
Gegenstromholzvergasungsprozessen sowie einem Wirbelschichtdampfvergasungspro-
zess wurden sieben verschiedene B-IGFC Systemkonzepte definiert. Die aus den ver-
schiedenen B-IGFC Systemkonzepte resultierenden Gaszusammensetzungen am Eintritt
der eingesetzten SOFCs wurden mittels ASPEN PLUS berechnet.
Im Vergleich zu vorreformiertem Methan ist die Leistung der holzgasbetriebenen SOFCs
um 25 bis 70 % reduziert. Die Höhe der Leistungsminderung hängt maßgeblich vom
"Verdünnungsgrad" des Holzgases und der Aktivität des Anodenmaterials ab, die beide
zusammen die Aktivierungsspannungsverluste bestimmen. Gemäß den verwendeten
Modellparametern scheint im untersuchten röhrenförmigen Zelltyp ein hochaktives Kata-
lysatormaterial zum Einsatz zu kommen. Nennenswerte Diffusionslimitierung wurde für
keines der untersuchten Holzgase und keinen der untersuchten Zelltypen vorhergesagt.
Die zur Einhaltung der vorgegebenen mittleren Zelltemperatur benötigte Kühlungsluftzahl
nimmt mit sinkendem Potential des Brenngases für zellinterne Reformierung zu. Dieser
Effekt ist für Zelltypen in Gegenstromausführung deutlich schwächer ausgeprägt als für
solche in Gleichstromausführung. Die aus dem Betrieb der verschiedenen Zelltypen mit
Holzgas resultierenden thermischen Belastungen sind grundsätzlich geringer als die die
sich mit vorreformiertem Methan als Brenngas einstellen. Dies könnte sich positiv auf die
entsprechenden Zelldegradationsmechanismen auswirken.
VI
Die vom SOFC Modell vorhergesagten Leistungswerte wurden als Eingabewerte für Ge-
samtsystemrechnungen genutzt. Die Rechnungen wurden mit ASPEN PLUS durchge-
führt um die Wechselwirkungen zwischen den verschiedenen Vergasungsprozessen,
Gasreinigungsvarianten und Zelltypen zu untersuchen. Die Wärmeintegration der be-
trachteten Gesamtsysteme wurde über ein verallgemeinertes Wärmetauschernetzwerk
abgebildet.
Die vergleichsweise niedrige Betriebstemperatur des standardmäßig zur Schwefelwas-
serstoffabscheidung eingesetzten Zinkoxids macht eine zusätzliche Befeuchtung der
verschiedenen Holzgase zwingend um eine thermodynamisch begünstigte Kohlenstoff-
bildung zu vermeiden. Einzige Ausnahme ist dabei des B-IGFC Systemkonzept in dem
eine katalytisch partielle Oxidationsstufe zur teilweisen Teerumwandlung zum Einsatz
kommt. Der Netto-Systemwirkungsgrad wird hauptsächlich vom Kaltgaswirkungsgrad
des Vergasungsprozesses und vom Eigenstrombedarf bestimmt, der direkt von der zur
Zellkühlung benötigten Luftzahl abhängt. Die adiabate Methanierung und die katalytisch
partielle Oxidation sind Prozesschritte, die sich gut für die Senkung der benötigten Luft-
zahl eignen indem sie das Potential der entsprechenden Brenngase für zellinterne Re-
formierung steigern und gleichzeitig den Abbau unerwünschter organischer Verbindun-
gen erlauben.
Folglich wurde der höchste Netto-Systemwirkungsgrad von 32.1 % in dieser Systemana-
lyse von der Kombination der Gegenstromholzvergasung mit einer adiabaten Methanie-
rungsstufe und des röhrenförmigen Zelltyps erreicht. Die Gegenstromvergasung liefert
bereits ein Holzgas mit vergleichsweise hohem Potential für zellinterne Reformierung.
Dieses wird in der adiabaten Methanierungsstufe zusätzlich gesteigert, was zu einer ent-
sprechend niedrigen Kühlungsluftzahl und geringem Eigenstrombedarf dieses B-IGFC
Systemkonzepts führt. Die zellinterne Dampfreformierung des hohen Methananteils im
so aufbereiteten Holzgas entspricht im weitesten Sinne dem letzten Schritt der Biomas-
severgasung unter direkter Nutzung der Hochtemperaturabwärme der SOFC.
Die über Gesamtsystemrechnungen für jedes der untersuchten B-IGFC Systemkonzepte
ermittelten Massen- und Energieströme wurden für die Auslegung der wichtigsten Appa-
rate genutzt. Die Gesamtsystemkosten wurden mit Hilfe entsprechende Kostenfunktio-
nen und Wichtungsfaktoren bestimmt. Die Anlagenperipherie der Brennstoffzellen und
insbesondere die benötigten Wärmetauscher machen bis zu 50 % der Gesamtsystem-
kosten aus, während die Brennstoffzellenstapel selbst nur bis zu 25 % der Gesamtsys-
temkosten verursachen. Dies verdeutlicht nochmals den Stellenwert der Kühlungsluftzahl,
aus der, zusammen mit dem Temperaturniveau des entsprechenden Brennstoffzellen-
stapels, die benötigte Wärmetauscherfläche folgt.
Die Gesamtsystemkosten wurden mit den entsprechenden Gesamtsystemleistungen zu
den spezifischen Anlagenkosten verrechnet. Gleichstromholzvergaser basierte Systeme
VII
weisen die höchsten spezifischen Anlagenkosten in dieser Systemanalyse auf. Die
Gründe dafür sind die benötigte kostenintensive Holztrocknung, große Wärmetauscher-
flächen und relative geringe Gesamtsystemleistungen, die auf hohe Aktivierungsverluste
durch die stark verdünnten Holzgase zurückzuführen sind. Im Vergleich dazu sind die
spezifischen Anlagenkosten der B-IGFC Systemkonzepte mit Wirbelschichtdampfholz-
vergasern etwas geringer. Dies ergibt sich trotz der insgesamt höheren Gesamtsystem-
kosten aus den deutlich kleineren Aktivierungsverlusten. Die auf Gegenstromholzverga-
sern beruhenden B-IGFC Systemkonzepte haben die geringsten spezifischen Anlagen-
kosten dieser Systemanalyse aufgrund der insgesamt niedriger Gesamtsystemkosten
und zusätzlich befriedigenden Gesamtsystemleistungen die sich hauptsächlich aus nied-
rigen Eigenstrombedarfswerten ergeben.
Die Stromgestehungskosten bilden die Synthese aus den Gesamtsystemkosten und den
entsprechenden Netto-Systemwirkungsgraden. Grundlage für deren Berechnung war ei-
ne detaillierte Kostenanalyse für einen festgelegten Planungszeitraum. Die Kostenanaly-
se umfasst Betriebs- und Wartungskosten, Brennstoffkosten und Betriebsmittelkosten
sowie alle Kosten die aus der Anlagenabschreibung und Darlehnszinsen entstehen. Zur
gleichmäßigen Verteilung des Nettobarwerts aller anfallenden Kosten über den Pla-
nungszeitraum wurde die Annuitätenmethode angewandt.
Etwa 40 % der Stromgestehungskosten können dem Brennstoffbedarf zugeordnet wer-
den während der Rest im weitesten Sinne von den spezifischen Anlagenkosten abhängt.
Folglich ist die Senkung der spezifischen Anlagenkosten wirksamer um geringere Strom-
gestehungskosten zu erreichen als die Steigerung des Netto-Systemwirkungsgrades.
Das bereits weiter oben beschriebene gegenstromholzvergasungsbasierte System erzielt
mit 0.1154 €/kWhel die geringsten Stromgestehungskosten in dieser Systemanalyse. Die
Erhöhung der Betriebsspannung des eingesetzten Brennstoffzellenstapels von 0.6 V auf
0.7 V ergibt eine Zunahme des Netto-Systemwirkungsgrads von 32.1 auf 37.1 %. Im Ge-
genzug steigen die Stromgestehungskosten auf 0.1912 €/kWhel, was auf die deutlich re-
duzierte Gesamtsystemleistung zurückzuführen ist. Dagegen birgt die Senkung des Ei-
genstrombedarfs durch optimierte Brennstoffzellenstapelbetriebsparameter zur Senkung
der Kühlungsluftzahl erhebliches Potential zur Senkung der Stromgestehungskosten.
Dies folgt aus der gesteigerten Gesamtsystemleistung und den gleichzeitig reduzierten
Kosten für die Anlagenperipherie der Brennstoffzellen.
Schlussendlich sind die Einkünfte aus dem Verkauf von Nutzwärme entscheidend für die
Wirtschaftlichkeit von B-IGFC Systemen, falls die Wärme nicht anlagenintern zur zusätz-
lichen Stromerzeugung verwendet wird. Der Einfluss entsprechender "Bottoming cycles"
auf die Wirtschaftlichkeit von B-IGFC Systemen ist in weitererführenden Arbeiten zu un-
tersuchen. Dazu muss das aufgebaute SOFC Modell in Flowsheeting- Programme wie
z.B. ASPEN PLUS integriert werden.
VIII
IX
X
Table of Contents
1 Introduction................................................................................................................ 1
1.1 Motivation.............................................................................................................. 2
1.2 Objectives ............................................................................................................. 4
1.3 Methodology ......................................................................................................... 4
1.4 Organization of the thesis ..................................................................................... 5
2 Technology overview ................................................................................................ 9
2.1 Biomass .............................................................................................................. 10
2.1.1 Definition .................................................................................................... 10
2.1.2 Classification.............................................................................................. 10
2.1.3 Construction............................................................................................... 10
2.1.4 Properties................................................................................................... 12
2.2 Gasification ......................................................................................................... 16
2.2.1 Biomass conversion overview.................................................................... 16
2.2.2 Fundamentals of thermochemical conversion ........................................... 17
2.2.3 Wood gasification technology overview ..................................................... 21
2.3 Gas processing................................................................................................... 31
2.3.1 Gas cleaning .............................................................................................. 31
2.3.2 Gas conditioning ........................................................................................ 39
2.4 Solid oxide fuel cell ............................................................................................. 43
2.4.1 Fuel cells in general ................................................................................... 43
2.4.2 Technology features .................................................................................. 46
2.4.3 Working principle........................................................................................ 47
2.4.4 Cell designs................................................................................................ 50
2.4.5 Support designs ......................................................................................... 52
2.4.6 Materials..................................................................................................... 54
2.4.7 Cell failure and performance degradation.................................................. 61
3 Experiments ............................................................................................................ 71
3.1 The PSI B-IGFC system ..................................................................................... 71
3.1.1 System description..................................................................................... 71
3.1.2 Discussion of the chosen approach ........................................................... 72
XI
3.2 Gas analysis ....................................................................................................... 76
3.3 Proof-of-feasibility tests ...................................................................................... 79
3.3.1 Lab scale tests under load conditions........................................................ 79
3.3.2 Lab scale tests under open circuit conditions ............................................ 81
3.4 Characterization of important unit operations ..................................................... 84
3.4.1 Updraft gasification reactor ........................................................................ 84
3.4.2 Hot gas particle removal ............................................................................ 88
3.4.3 Catalytic partial oxidation ........................................................................... 92
3.4.4 Sulfur absorption by zinc oxide .................................................................. 96
3.4.5 Solid oxide fuel cell .................................................................................... 97
3.5 Proof-of-concept of the PSI B-IGFC system..................................................... 103
3.5.1 First demonstration unit tests and derived modifications......................... 104
3.5.2 Present demonstration unit configuration ................................................ 106
3.5.3 Long-term test results .............................................................................. 108
4 Modeling work literature review............................................................................. 117
4.1 Modeling approaches for steady-state investigations....................................... 117
4.1.1 Molecular level models ............................................................................ 118
4.1.2 Electrode level models............................................................................. 118
4.1.3 Cell level models...................................................................................... 120
4.1.4 Stack level models ................................................................................... 123
4.1.5 Plant level models.................................................................................... 124
4.2 Recent studies of biomass fed SOFC cycles.................................................... 124
5 Modeling................................................................................................................ 129
5.1 Overall modeling approach............................................................................... 129
5.2 Flowsheeting models ........................................................................................ 131
5.2.1 Biomass gasifier....................................................................................... 131
5.2.2 Gas humidifier.......................................................................................... 132
5.2.3 Heat integration network .......................................................................... 132
5.2.4 Steam reforming and methanation reactors............................................. 133
5.2.5 Catalytic partial oxidation reactors ........................................................... 134
5.2.6 Sulfur adsorbent beds.............................................................................. 135
5.2.7 SOFC ....................................................................................................... 135
XII
5.3 ATHENA SOFC model ..................................................................................... 136
5.3.1 Model description..................................................................................... 136
5.3.2 Planar model verification.......................................................................... 181
5.3.3 Tubular model validation.......................................................................... 186
5.3.4 Triangular model validation...................................................................... 191
5.3.5 Sensitivity analysis................................................................................... 193
5.4 Economic model ............................................................................................... 219
5.4.1 Equipment purchase cost estimate functions .......................................... 220
5.4.2 Equipment sizing and purchase cost estimation...................................... 222
5.4.3 Total investment, direct and indirect plant costs...................................... 227
5.4.4 Net AC power efficiency........................................................................... 228
5.4.5 Power production costs............................................................................ 229
6 System analysis .................................................................................................... 235
6.1 Analysis constraints .......................................................................................... 235
6.1.1 SOFC simulation assumptions................................................................. 235
6.1.2 Overall system simulation assumptions................................................... 239
6.2 System definition............................................................................................... 240
6.2.1 Natural gas reference system .................................................................. 240
6.2.2 B-IGFC systems....................................................................................... 241
6.3 Technical evaluation ......................................................................................... 249
6.3.1 Simulation results of gas processing sections ......................................... 249
6.3.2 Simulation results of SOFC operation with producer gases .................... 253
6.3.3 System simulation results ........................................................................ 270
6.4 Economic evaluation......................................................................................... 279
6.4.1 Direct plant cost estimates....................................................................... 279
6.4.2 Power production costs............................................................................ 283
6.4.3 Sensitivity analysis................................................................................... 286
7 Concluding remarks .............................................................................................. 293
7.1 Conclusions ...................................................................................................... 293
7.1.1 Experiments ............................................................................................. 293
7.1.2 System analysis ....................................................................................... 294
7.2 Recommendations for future work.................................................................... 297
XIII
7.2.1 Experiments ............................................................................................. 297
7.2.2 Modeling and system analysis ................................................................. 299
Notation.......................................................................................................................... 303
References..................................................................................................................... 311
XIV
XV
XVI
1 Introduction
1 Introduction
The on-call availability of energy was a prerequisite for the development

of mankind, its culture and technology. After the discovery of fossil fuels
and the start of their large-scale use approx. 200 years ago, the devel-
opment has strongly accelerated. Today, the global energy consumption
is primarily covered by fossil fuels, Figure 1-1, and amounts to a yearly
average of 17500 kWh per capita translating into a continuous power
need of around 2000 W per capita, [1]. However, the majority of people
are denied access to their share. For instance, the average person living
on the African continent has to content himself with approx. 650 W,
while the average Western European resident requires almost ten times
as much, 6000 W, and the average North American tops the consump-
tion at 12000 W, [1]. Aggravating the picture many of the fossil fuels are
extracted in regions of the world, which are politically unstable und less-
developed. These issues combined with the fact that fossil fuel reserves
are limited, [2], generate serious conflict potential between the well- and
the less-developed nations.
Figure 1-1:
Development of
global energy
consumption dur-
ing the last 150
years, [1]
Note:
1 Etajoule = 1018 J
A fairer distribution of the fossils fuels and the foreseeable end of re-
serves is however only one part of the problem arising from the pres-
ently excessive use of fossil fuels, which will only continue to increase
due to the rapidly growing global population and numerous fast develop-
ing nations. In the year 2000 only, mankind consumed an amount of fos-
1
1 Introduction
sil fuels that took approx. one million years to be formed. The emissions
of CO2 attributed to the combustion of fossil fuels have thus reached
levels impacting negatively on the ecosystem, endangering not only the
security of food and water supplies but also irreversibly altering the cli-
mate conditions on earth. Thus, a continuity of the developed nations
currently taken as granted and a fair chance for every human being to
relish a comparable quality of life crucially hinges on finding ways to re-
duce the pace at which global energy consumption increases and to
sustainably satisfy the global energy needs. It comes as no surprise
then that renewable energy sources such as solar (direct), wind, tide,
wave, geothermal and biomass have gained increased interest recently
given their time unlimited abundance and global availability. In the ongo-
ing and often controversial debates between citizens, governments, in-
dustry and scientists regarding the energy problem, the electricity gen-
eration has always played a key role and will continue to gain signifi-
cance in the near future.
1.1 Motivation
Among the renewable energy sources biomass has a large potential for
electricity generation. If the harvest is performed in a sustainable way,
wood is a CO2-neutral energy source. Within the next 20 years, the
worldwide installed capacity of biomass power plants is expected to in-
crease by approx. 65 GW, [3]. This forecasted increase however still
leaves most of the potential untapped. A prerequisite to achieve this in-
crease and to exploit its full potential is to overcome the economical
drawbacks related to biomass as feed stock.
Currently most power plants using wood as feedstock are based on
grate firing and steam cycles, where high electrical efficiencies are only
possible on multi-MW scale. Wood is an expensive feedstock due to the
elaborate harvesting processes. This makes the sale of the produced
heat mandatory for the economical feasibility of power generation using
wood. Consequently the scarcity of major heat consumers and connec-
tion points to large-scale district heating grids limit the number of eco-
nomically feasible power plant locations. Small-scale wood power plants
may overcome the heat sales problem due to comparably small
amounts of heat that can be sold to residential heating. However, the
low efficiencies of small-scale steam cycles waste this advantage to a
major extent.
2
1 Introduction
The gasification of wood instead of its combustion allows the application

of more efficient energy conversion cycles than steam cycles such as in-
ternal combustion engines (ICE). These ICE based combined heat and
power plants (ICE CHP) reach electrical efficiencies up to three times
larger than small-scale steam-cycle based wood power plants in the
range up to a few MW, [4]. In a first step, the wood is converted to a
combustible gas, commonly referred to as producer gas (PG), through a
gasification reactor. The composition of the PG depends on the em-
ployed gasification reactor type, gasification agent and operating condi-
tions of the gasification process. The PG consists to a major extent of
hydrogen and carbon monoxide, the rest being carbon dioxide, methane,
water, impurities (e.g. tars, alkali salts, sulfur, particles etc.) and nitrogen.
Especially the hydrocarboneous species that are liquid at room tempera-
ture, summarized under the term "tars", cause problems in ICE CHP
plants. This is because the PG has to be cooled prior to its conversion in
the ICE for thermodynamic efficiency reasons. During the cool down,
tars condensate and can cause problems for the operation of ICEs.
Thus, tars have to be largely removed from the PG. The removal of tars
often involves complex and expensive processes, which add to the spe-
cific system costs. Nevertheless, the efficiency advantages of ICE CHP
plants are high enough to justify this additional effort compared to clas-
sical steam-cycle based biomass power plants.
A promising approach to achieve higher electrical efficiencies than those
of ICE CHP plants is the use fuel cells (FC) for the conversion of the PG
to electricity. This combination is commonly referred to as "Biomass In-
tegrated Gasification Fuel Cell System" (B-IGFC). FCs directly convert
the chemical energy of a fuel gas into electricity and heat through elec-
trochemical reactions. Regarding its application in B-IGFC systems, the
solid oxide fuel cell (SOFC) has drawn a lot of interest in recent years.
SOFCs impose less stringent requirements to the fuel gas quality com-
pared to all other FCs and are capable of directly oxidizing carbona-
ceous gases with high efficiencies at small scales. Further, SOFCs re-
quire fuel gas inlet temperatures well above the dew point of tars and
operate at temperatures that are high enough to allow for the conversion
of tars and their subsequent use for the electricity production. Ideally,
only a particle removal would be required between the wood gasification
reactor and the SOFC. Especially with respect to its hydrocarbon, tar,
alkali salts and sulfur content, the composition of the PG plays a more
important role for its conversion in SOFCs than in ICEs. The interactions
3
1 Introduction
between the gasification process and the SOFC need to be well under-
stood to allow for the exploitation of the full potential of the B-IGFC ap-
proach.
1.2 Objectives
The present thesis aims at demonstrating the B-IGFC technology on
kW-scale and at identifying promising B-IGFC system concepts with
power outputs around 1 MWel. With respect to the latter the most impor-
tant questions are:
How strongly does the producer gas composition resulting from dif-
ferent wood gasification processes influence the temperature man-
agement and the power output of available SOFC designs and what
is the impact on the corresponding cooling requirements?
Are there producer gases that lead to critical operating conditions in
SOFCs such as critical temperature gradients and cell temperatures?
Do different producer gases yield the same effects in SOFCs of dif-
ferent design?
What role does the eventual treatment of tars and removal of sulfur
play for the overall system efficiency and the specific plant costs?
What are the major cost drivers in B-IGFC systems?
What power production costs are feasible and what are the cost dif-
ferences between the different cell designs?
1.3 Methodology
The chosen approach to fulfill the above aims is based on dedicated ex-
periments and numerical systems analysis:
Investigations of different operational modes of an in-house devel-
oped lab-scale updraft gasification reactor are performed aiming at a
stable long-term operation and producer gas composition.
The removal of impurities such as particles and sulfur species as well
as the partial degradation of tars are experimentally characterized to
explore the performance of state-of-the-art and currently developed
gas processing technologies.
An experimental setup covering the complete chain from wood to
electricity is erected and operated to identify the most important tech-
nical problems, which may arise in B-IGFC systems.
4
1 Introduction
The knowledge gained from the above mentioned experimental investi-

gations is used for the definition of promising and realistic B-IGFC sys-
tem concepts including all major equipment pieces. The interactions be-
tween different gasification processes, gas processing technologies and
SOFC designs are investigated through numerical systems analysis:
The core of the system analysis is a detailed one-dimensional nu-
merical SOFC model. The model is developed to predict the impact of
different producer gas compositions on the power output delivered by
state-of-the-art cell designs for power plant applications. Besides, the
model shall allow for the identification of critical operating conditions
in SOFCs provoked by different producer gases.
The defined B-IGFC systems are simulated by means of the com-
mercial flowsheeting software package ASPEN PLUS. The system
simulations aim at the determination of all relevant mass and energy
streams in the investigated B-IGFC systems as well as an approxi-
mate sizing of the required heat exchanger network.
The predicted SOFC power output and the results of the system
simulations are the basis for the economic simulation of the defined
B-IGFC systems. The economic model is developed to yield plant
cost estimates and power production costs.
1.4 Organization of the thesis

Chapter 2 gives an overview of important aspects of the B-IGFC tech-
nology. At first, the properties of biomass as feedstock for power gen-
eration are explained in section 2.1. Section 2.2 provides the fundamen-
tals of thermochemical biomass conversion processes focusing on the
differences between producer gases originating from different biomass
gasification processes. The most important technical processes for the
removal of impurities and gas conditioning are discussed in section 2.3.
Section 2.4 gives a brief introduction to the solid oxide fuel cell technol-
ogy concentrating on the state-of-the-art, development trends as well as
cell failure and degradation mechanisms.
Chapter 3 provides details about the experimental work conducted in
the framework of this thesis. The B-IGFC system envisaged for the dem-
onstration of the B-IGFC technology on kW-scale is explained in section
3.1. Further, the experimental setups used for the characterization of dif-
ferent gas cleaning and conditioning technologies are presented and the
corresponding results are discussed in sections 3.2 to 3.4. The history of
5
1 Introduction
experiments leading to the first-time long-term demonstration experi-

ment of the B-IGFC technology on kW-scale is outlined in section 3.5.
Chapter 4 summarizes the different modeling approaches for steady-
state systems analysis of SOFC-based power generation systems in
section 4.1. The modeling standard of recently published system analy-
sis dealing with B-IGFC systems is presented in section 4.2.
Chapter 5 explains the numerical model framework developed for the
envisaged system analysis. The overall modeling approach is charted in
section 5.1 and centers on the data exchange between the detailed
SOFC model, the flowsheeting system model and the economic model.
Section 5.2 provides a detailed account of the lumped models devel-
oped to simulate the gas cleaning and conditioning technologies consid-
ered in the system analysis. The detailed one-dimensional SOFC model
is explained in detail in section 5.3. Further, the model validation and an
extensive sensitivity analysis is presented, the latter being focused on
the determination of the impact of important model parameters on the
temperature and power output trends predicted for different fuel gases.
Section 5.4 explains the economic model built on cost functions, equip-
ment sizing procedures and power production cost calculations.
Chapter 6 addresses the conducted system analysis. The major as-
sumptions made and model settings used for the analysis are summa-
rized in section 6.1. The investigated B-IGFC systems are described in
detail in section 6.2. Section 6.3 treats the simulation of the gas proc-
essing performed in the various B-IGFC systems. The predicted PG
compositions are then investigated with respect to the resulting operat-
ing conditions in different SOFC designs. The discussion centers on the
impact of the different PGs on the interactions between the heat, mass
and charge transport processes taking place in various SOFC designs.
The predicted power output of the SOFC designs provides the back-
ground for overall system simulations which in turn are the basis for the
discussion of the expected system efficiencies and efficiency losses.
The results of the economic model are presented in section 6.4. At first,
the plant cost estimates are analyzed to appoint the cost drivers of B-
IGFC systems. The power production cost structure is explained subse-
quently. Finally, a sensitivity study of economic model parameters and
analysis assumptions is presented.
Chapter 7 summarizes the main findings, gives the resulting conclu-
sions and provides recommendations for future research.
6
7
8
2 Technology overview
The combination of biomass gasification with SOFCs is a highly promis-

ing approach to exploit the potential of biomass in combined heat and
power generation. The according systems are commonly referred to as
"Biomass Integrated Gasification Fuel Cell Systems" (B-IGFC). Figure
2-1 gives an outline of the core components of B-IGFC systems.
Figure 2-1: Biomass Gasification

Technical outline of agent
a B-IGFC system
Gasification
Gasification
PG FC
FCSystem
System
Raw suitable
PG for FC
- -Reactor
Reactortype
type - -FC
FCType
Type
- -Gasification
Gasificationagent
agent
Gas
Gasprocessing
processing - -Catalysts
Catalysts
- -Feedstock
Feedstock - -FC
FCdesign
design
- -Operating
Operating - -Operating
Operating
Conditions
Conditions Conditions
Conditions
Heat Electricity
In a first step the solid biomass is converted to a combustible gas mix-

ture. The composition of the gas mixture depends on the employed re-
actor type, gasification agent, feedstock and operating conditions of the
gasification process. It consists to a major extent of hydrogen and car-
bon monoxide, the rest being carbon dioxide, methane, other hydrocar-
bon species, water, diverse impurities (e.g. tars, alkali salts, sulfur, soot
particles etc.) and nitrogen in case of air as gasification agent. The im-
purities are potentially performance degrading and have to be removed
to some extent in order to meet the requirements of the employed fuel
cell. The requirements depend on the specific fuel cell (FC) type and its
design, catalyst materials and the operating conditions.
The strong interactions between the composition of the gas mixture ob-
tained from the gasification process and the fuel cell entail that optimal
system integration is crucial for overall energy efficient and cost effective
system. The subsequent sections provide an introduction to the most
important aspects and components of B-IGFC systems.
9
2.1 Biomass
2.1.1 Definition
Biomass is a diverse term. In biological and ecological disciplines, the
accumulation of all living matter (plants, bacteria, animals and men) or
specific species in a given area or of a biological community is referred
to as biomass. In contrast, the energy and chemical industry defines
biomass as all living or recently deceased biological matter suitable as
raw material for industrial processing. In this work, the term biomass re-
fers to plant tissue or more specifically to wood.
2.1.2 Classification
The variety of plant matter can be classified based on the origin and the
type of growth of the plants, [5]:
Terrestrial plants
Woody plants, i.e. perennial lignocellulosic crops
Long term growth (15-50 year); Short term single stem

forestry (6-15 year) and multiple stem plants (3-5 year)
Herbaceous plants and grasses, i.e. switchgrass
High sugar-/ starch-producing crops, i.e. sugarcane
Oil-producing plants, i.e. rapeseed
Aquatic plants, i.e. algae

Another, more process-oriented, classification is based on the intrinsic
moisture content which is that of the biomass material without the influ-
ence of weather effects. Plants with an intrinsic moisture content requir-
ing more energy for the drying than their chemical energy content are
classified as wet, [6], while wood is classified as dry biomass.
2.1.3 Construction
Plants combine water and carbon dioxide to sugar building blocks, Eq. 1.
n ⋅ H 2O + n ⋅ CO2 + Light ⎯Chlorophyl
⎯ ⎯⎯l → (CH 2O) n + n ⋅ O2 Eq. 1
The required energy is produced from light via photosynthesis based on

chlorophyll. On average, between 0.1 and 1 % of the available light is
stored as chemical energy in plants, [7].
10
The sugar building blocks are the starting point for the major fractions
found in all terrestrial plants, [8], lignin, hemicellulose and cellulose.
Lignin
Lignin makes up about 25-35 mass-% (dry) of wood and 10-25 mass-
% (dry) of other plants. It is a complex irregular aromatic polymer, of-
ten linked to cellulose and hemicellulose compounds in the plant cell
walls. The high lignin fraction of wood accounts for its rigidity and
slow growth compared to herbaceous plants, Figure 2-2, [6].
Figure 2-2:
Molecular structure
of lignin, [9]
Hemicellulose
In contrast to cellulose, hemicellulose comprises five sugars, namely
xylose, arabinose, galactose, glucose and mannose. Hemicellulose
consists of 50 to 200 sugar units and is amorphous, see Figure 2-3.
Plants consist to approx. 15-30 mass-% (dry) of hemicellulose.
Figure 2-3:
Molecular structure
of hemicellulose,
[9]
11
Cellulose
Plants consist to approx. 40-80 mass-% (dry) of cellulose, being a
linear polysaccaride occasionally made up of over 10000 glucose
units, see Figure 2-4. The elemental formula of polymeric cellulose is
(C6H10O5)n.
Figure 2-4:
Molecular structure
of cellulose, [9]
Cellulose, hemicellulose and lignin represent approx. 95-98 mass-%

(dry) of woody biomass. The rest can be allocated to extractives and
ash, [10]. Extractives are defined as compounds of the biomass that are
not an integral part of the plant structure and that are soluble in e.g. wa-
ter or methanol. Ash is defined as the non-combustible, inorganic frac-
tion of the biomass. It includes alkali salts, heavy metals, sulphur, chlo-
rine and silicates. The ash fraction of wood is mostly below 1 mass-%
(dry). Table 2-1 illustrates the variation spectrum.
Table 2-1: Dry fraction Eucalyptus

Switchgrass Corn stover Corn grain
Average chemical [mass-%] tree
composition of Cellulose 48 43 36 3
various terrestrial
biomass species, Hemicellulose 14 33 23 6
[7] Lignin 29 9 17 2
Extractives 6 8 6 82
Ash 1 6 10 0
Residues 2 1 8 7
2.1.4 Properties
This work focuses on woody biomass for power generation, the most
important material properties of which are discussed in the following.
2.1.4.1 Moisture
The intrinsic moisture content comprises the water in the pores and hy-
droscopic bound water in the solid structure of the woody material, [10].
The extrinsic moisture content accounts for complementary moisture
originating from the weather conditions during harvesting and storage.
Regarding wood as feedstock for power generation applications, the ex-
12
trinsic moisture content, also referred to as water load, is of greater in-

terest than the intrinsic:
ms − md mH 2O
M = =
md md
with Eq. 2
M =water load; ms = moist mass; mH 2O = extrinsic water mass
md = mass of specimen after 24 h drying at 102 °C at 1 atm
In contrast to the moisture content, the humidity is based on the wet

mass of the specimen. It is defines as:
ms − md mH 2O
w= =
ms ms
Eq. 3
with w = humidity; ms = moist mass
md = mass of specimen after 24 h drying at 102 °C at 1 atm
The moisture content can be converted to a humidity value via Eq. 4.

w
M = Eq. 4
1− w
2.1.4.2 Proximate and ultimate analysis

The "proximate analysis" of solid fuels is standardized in the ASTM
E870-82. The analysis yields information about moisture, volatile matter,
ash and residue as well as fixed carbon content of solid fuels. The vola-
tile matter content is determined by heating a fuel sample up to 950 °C
without adding oxygen or steam and maintaining this temperature for 7
minutes. The mass difference of the sample before and after the proce-
dure represents the volatile matter content. The volatile matter fraction is
calculated as follows:
mstart − mend
X VM =
mstart
with Eq. 5
mstart = original sample mass
mend =sample mass at end of volatile matter test procedure
The ash content is determined by heating a fuel sample up to 600 °C in

air and maintaining this temperature for 4 to 6 hours. The remaining ma-
13
terial is considered as ash. The ash fraction is the quotient of ash mass
and sample mass before the test procedure.
mend
X Ash =
mstart
with Eq. 6
mstart = original sample mass
mend =sample mass at end of ash test procedure
The fixed carbon fraction is used to estimate the amount of coke that
can be obtained from a fuel sample. It is calculated according to:
X FC = 100 − X VM − X Ash Eq. 7
The proximate analysis gives a first measure for the suitability of the
analyzed material for thermochemical conversion technologies.
The "ultimate analysis" provides information about the elemental com-
position of solid fuels. Plants mainly consist of carbon, hydrogen, oxy-
gen and nitrogen. Besides, a variety of other chemical elements can be
found in biomass with more or less important concentrations. The impor-
tance of these trace elements depends on the subsequent processes.
For thermochemical conversion, sulfur, chlorine, potassium and sodium
are most important. The rest of the biomass material (e.g. heavy metals)
is usually summed up under the term ash. Table 2-2 gives the proximate
and ultimate analysis of biomass materials in comparison to coal.
Table 2-2: M XVM XFC C H O N S Cl K Na Ash

Proximate (italic) [%] [mass-% (dry)]
and ultimate
Average
(bold) analysis for 20 82 17 51.9 6.0 41.8 0.1 0.01 0.02 0.04 0.02 1
Wood
various biomass
and fossil fuels, Corn
16 59 21 50.2 6.7 42.1 0.44 0.44 0.11 - - 4
[10] Straw
Lignite 34 44 47 70.0 5.2 22.8 1.99 - - - - 9
Bitumi-
11 39 51 80.9 6.1 9.6 1.55 1.88 - - - 10
nous coal
The main difference between biomass and solid fossil fuels can be
found in the oxygen to carbon and the hydrogen to carbon ratios, [11].
14
Figure 2-5: 1.8
/ kg
J /k g
40 MJ/k
Van Krevelen dia- Cellulose
35 MJ
30 M
gram for various 1.6
dry solid fuels, [12],
1.4 Wood
with calculated dry-
Lignin
basis iso-LHV lines
Atomic H/C Ratio

1.2 Biomass
via Eq. 8 exclu-
sively considering 1.0 Peat
the carbon, hydro- Lignite
0.8
gen and oxygen
fractions 0.6 Coal
0.4
J/kg
g
0.2 Anthracite
/k
/k
MJ
MJ
25 M
20
15
0.0
0.0 0.2 0.4 0.6 0.8
Atomic O/C Ratio
Figure 2-5 illustrates the significance of these ratios on the lower heating
value of young biomass and solid fossil fuels in a so called Van Kreve-
len diagram. The comparably higher heating value of solid fossil fuels
can be explained with the low number of carbon-oxygen bonds, [6],
which again is a result of the natural diffusion process of biomass bound
oxygen under high pressure and temperature. However, hydrogen also
diffuses away under these conditions. Hence, at a certain stage of the
process, the heating value begins to decrease asymptotically against
the heating value of pure carbon which is 34.8 MJ/kg (daf).
The alkali metal content (Na, K, Ca, etc.) of biomass is of special inter-
est regarding thermochemical conversion processes. Alkali metals can
react with silica present in the biomass or originating from soil material
introduced into the process during harvesting. The resulting sticky, liquid
phase can provoke slagging in furnaces and boilers. Another known
problem which can arise from alkali metals is high temperature corrosion
of e.g. gas turbine blades. The catalysts employed in fuel cells could be
poisoned by alkali metals, [13]. However, alkali metals were also found
to have a positive effect as they reduce the tar formation, while increas-
ing the char formation, in pyrolysis and gasification processes, [14, 15].
2.1.4.3 Caloric value

The caloric value (CV) of a material is defined as the heat per mass,
volumetric or molar unit which is released during combustion. There are
two different expressions for the CV:
15
The quantity of heat which is obtained by bringing all the products of

combustion back to the pre-combustion temperature, usually stan-
dard conditions, is defined as the higher heating value (HHV).
The lower heating value (LHV) is mostly employed for applications
that can not use the heat of condensation of steam. The LHV equals
the HHV less the evaporation heat of water at standard conditions.
The caloric value is commonly determined using bomb calorimeters.
The LHV of dry wood is typically in the range between 17 and 20 MJ/kg
(daf), [16], see also Figure 2-5. The CV is often calculated via empirical
correlations based on proximate and ultimate analysis results. In this
work, the LHV was calculated using the Boie correlation, Eq. 8.
⎡ MJ ⎤
LHV ⎢ ⎥ = 34.8 ⋅ c + 93.9 ⋅ h − 10.8 ⋅ o + 10.5 ⋅ s + 6.3 ⋅ n − 2.44 ⋅ w
⎣ kg ⎦
Eq. 8
with c, h, o, s, n, w : carbon, hydrogen, oxygen, sulfur, nitrogen &
water mass fraction on wet, ash free basis
Besides, the Dulong, Grummel & Davis and Mott & Spooner correlations
are used, [17]. The results may significantly diverge. Hence, the correla-
tion choice is usually a question of minimizing energy balance errors.
2.2 Gasification
2.2.1 Biomass conversion overview
The type of biomass has a strong impact on the choice of conversion
technology due to varying moisture contents and proportions of the
three main components; cellulose, hemicellulose and lignin. In addition,
the choice of conversion technology depends on the subsequent energy
converting device and its specific requirements.
Figure 2-6: Thermochemical Physical / Chemical Biochemical

Biomass conver-
Combustion Gasification
Gasification Pyrolysis
Pyrolysis Extraction Digestion
Digestion Fermentation
Fermentation
sion technologies, Combustion Extraction
intermediate prod-
ucts and end-use
Heat Combustible
Combustiblegas Combustible
Combustibleliquid
applications for the Heat gas liquid
combined genera-
tion of power and
Steam
Steamcycle
heat cycle Fuel
Fuelcell
cell Combustion
Combustionengine
engine Gas
Gasturbine
turbine
Heat
Heatand
andElectricity
Electricity
16
Figure 2-6 gives an overview of conversion technologies for heat and

power generation from biomass with the respective intermediate prod-
ucts and possible end-use energy converting devices. The following
three categories of conversion technologies are differentiated:
Biochemical conversion
Physical/chemical conversion
Thermochemical conversion
Digestion and fermentation of biomass are biochemical conversion tech-
nologies with an energy efficiency of approx. 20 to 40 %, [18]. Biomass
fermentation usually yields liquid ethanol, while digestion yields gaseous
methane. The micro-organisms used in both processes can convert the
sugar-based cellulose and hemicellulose as well as starch but not stable
organic materials such as lignin. Hence, biochemical conversion of
woody biomass results in high amounts of residues and is not indicated.
In contrast, algae are well suited to biochemical conversion, [18].
Mechanical extraction, as physical/chemical conversion technology,
yields a combustible liquid referred to as bio-oil. This oil is usually further
processed to bio-diesel via transersterification and used for mobile ap-
plications. Plants such as rapeseed, with high moisture and extractible
fraction are predestined for mechanical extraction technologies. Physi-
cal/chemical conversion of woody biomass is unfavorable due to its low
extractible fraction. Thermochemical conversion technologies are dis-
cussed in the following sections.
2.2.2 Fundamentals of thermochemical conversion

Thermochemical conversion allows complete conversion of lignin and is
therefore mainly indicated for dry biomass such as wood, [6]. The most
important thermochemical conversion technologies are combustion, py-
rolysis and gasification. The three technologies mainly differ in the prod-
uct composition and amount of used oxidant if any. The latter is de-
scribed by either the air-to-fuel ratio, which is the quotient of the effec-
tive and the stoichiometric amount of air or oxygen for complete com-
bustion, Eq. 9, or the steam-to-fuel ratio for steam gasification proc-
esses.
mact ,oxi
λ= Eq. 9
mstoic ,oxi
All thermochemical conversion technologies involve all or some of the

sub-processes "oxidation", "reduction", "devolatilization" and "drying".
17
Oxidation
Heat is produced via exothermal partial or total oxidation of carbon,
Eq. 10 and Eq. 11, at around 1000 to 1600 °C, [19]. The carbon origi-
nates from the devolatilization and reduction sub-processes.
⎛ kJ ⎞
C + 0.5 O2 → CO with ⎜ − ΔH R0 = 110.5 ⎟ Eq. 10
⎝ mol ⎠
⎛ kJ ⎞
C + O2 → CO2 with ⎜ − ΔH R0 = 398.8 ⎟ Eq. 11
⎝ mol ⎠
Reduction
Most of the hydrogen, carbon monoxide, carbon dioxide and methane
found in producer gases from thermochemical biomass conversion is
produced in the reduction sub-process at temperatures between 600
and 1000 °C, [19]. The required heat, steam and carbon dioxide is
usually produced in the oxidation sub-process. The carbon steam re-
action, Eq. 12, is the principal reaction for the gasification of carbon.
Besides, carbon can be gasified via the Boudouard reaction, Eq. 13.
⎛ kJ ⎞
C + H 2O ↔ CO + H 2 with ⎜ − ΔH R0 = −131.3 ⎟ Eq. 12
⎝ mol ⎠
⎛ kJ ⎞
C + CO2 ↔ 2CO with ⎜ − ΔH R0 = −172.5 ⎟ Eq. 13
⎝ mol ⎠
Another important reaction in the reduction sub-process is the exo-

thermic water-gas shift reaction where carbon monoxide and steam
react to hydrogen and carbon dioxide, Eq. 14. The water-gas shift re-
action is heterogeneously catalyzed on carbon at temperatures below
1100 °C but may occur homogenously at higher temperatures, [20].
The exothermic hydrogasification reaction yields methane as product
from carbon and hydrogen, Eq. 15. Methane can also be produced
via the exothermic methanation reaction, Eq. 16. However in sum, the
reactions taking place in the reduction sub-process are endothermic.
⎛ kJ ⎞
CO + H 2O ↔ CO2 + H 2 with ⎜ − ΔH R0 = 41.0 ⎟ Eq. 14
⎝ mol ⎠
⎛ kJ ⎞
C + 2 H 2 ↔ CH 4 with ⎜ − ΔH R0 = 74.8 ⎟ Eq. 15
⎝ mol ⎠
⎛ kJ ⎞
CO + 3H 2 ↔ CH 4 + H 2O with ⎜ − ΔH R0 = 206.1 ⎟ Eq. 16
⎝ mol ⎠
18
Devolatilization
Devolatilization denotes the thermal decomposition of biomass with-
out the participation of an oxidant at temperatures between 200 and
600 °C, [19]. The three main constituents of biomass; cellulose,
hemi-cellulose and lignin, see section 2.1.3, exhibit different thermal
decomposition behaviors which originate from their molecular struc-
ture, [5]. Devolatilization yields gaseous, liquid and solid products.
Gaseous products are mainly hydrogen, carbon monoxide and diox-
ide as well as methane and higher hydrocarbons. Liquid products are
hydrocarboneous species that condense at room temperature, gen-
erally summarized under the term "tars". Non reacted carbon is the
main solid product. Due to transport phenomena in pores and on the
surface, the kinetics of devolatilization depends on the shrinking be-
havior and structure of the feed particles, [21]. Further, the heating
rate and the heat transfer play important roles. The thermal decom-
position reactions are endothermic. A more detailed discussion of the
devolatilization of biomass is given in [20].
Drying
The moisture of biomass is vaporized in the drying sub-process at
temperatures below 200 °C. The amount and nature of moisture de-
termine the required heat quantity. Especially the vaporization of the
water bound in the cell walls involves strengthening of the wood fi-
bers and is accompanied by shrinkage of the biomass, [5]. More de-
tails on recent developments in biomass drying are given in [22].
Table 2-3 summarizes thermochemical conversion technologies with
typical air/steam-to-fuel ratios, process temperatures and products.
Table 2-3: Temperature Air-to-fuel ratio Product

Technology
Thermochemical [°C] [-] [-]
conversion tech-
Pyrolysis 400-700 0 Combustible liquid, gas and solid
nologies differen-
tiated by Oxidant: Air Combustible gas with low LHV
air/steam-to-fuel 0.2-0.5 Non combustible solid
Gasification 700-900
ratio and desired Oxidant: Steam Combustible gas with high LHV
products, [4] 0.4-3 Non combustible solid
Combustion 800-1300 >1 Non combustible gas and solid
The technical pyrolysis process comprises the sub-processes "drying",

"devolatilization" and to a minor extent "reduction" which usually occur
simultaneously. During the thermal decomposition of biomass at tem-
19
peratures between 400 and 700 °C without addition of oxygen or steam,

a mixture of combustible gaseous, liquid (at room temperature) and solid
products is produced. The product distribution as well as the composi-
tion of the produced gaseous and liquid species depend on the applied
process conditions such as temperature, pressure, heating rate, fuel
particle size and composition of the biomass, [20]. The required heat for
the pyrolysis is produced via the oxidation of undesired products, mostly
the solid fraction. The liquid product, named pyrolysis oil, can contain up
to 80 % of the energy originally stored in the converted biomass, [23].
Combustion involves the simultaneously occurring sub-processes "dry-
ing", "devolatilization", "reduction" and "oxidation". During the oxidation
with air-to-fuel ratios higher than one, the chemical energy of the bio-
mass is entirely converted to sensible heat of exhaust gases consisting
of steam, carbon dioxide and nitrogen for air as oxidant. The volatile
compounds of the biomass and the products of char reduction are oxi-
dized in the gas phase. However, char may also be directly oxidized in
surface reactions which are not well understood yet, [24]. Depending on
the biomass humidity, the exhaust gases have a temperature of 800 to
1300 °C. The produced heat can be converted to mechanical work via
e.g. steam cycles or Stirling engines, [4].
Gasification comprises the sub-processes "drying", "devolatilization",
"reduction" and "oxidation". Depending on the reactor design, the sub-
processes may occur simultaneously or sequentially. The desired prod-
uct of biomass gasification is a combustible gas with high heating value.
Its composition is mainly determined by the gasification agent and the
process conditions, [25]. Air as gasification agent yields a nitrogen di-
luted gas mixture which mostly contains hydrocarbons. This gas mixture
is commonly referred to as producer gas in contrast to syngas, which
stands for nitrogen-free gases from oxygen, steam or hydrogen gasifica-
tion with negligible hydrocarbon content, [7]. The gasification is auto-
thermal if the required heat is produced via partial combustion of the
feed within the same reactor. The gasification is allothermal if the heat is
produced in a spatially separated second reactor and introduced to the
gasification reactor by means of a bed material, heat pipes etc. The effi-
ciency of a gasification process is given by the cold gas efficiency which
relates the chemical energy content of the produced gas to that of the
biomass before conversion. Depending on the process conditions, the
cold gas efficiencies of today's biomass gasification processes range
from 50 to over 90 %, [25].
20
2.2.3 Wood gasification technology overview

The primary aim of every wood gasification technology is the efficient
conversion of the energy content of the wood into the chemical energy
content of the produced gas. The gas composition and the impurity sig-
nature depend on the gasification agent and the process conditions
which are a consequence of the gasification reactor design and vice
versa. The impurity signature of a producer gas is principally character-
ized by six different types of impurities and their phase of appearance,
Figure 2-7. The total amount of impurities is independent of the reactor
design but depends strongly on the concentration of the according
educts in the feedstock.
Figure 2-7:
Impurities
Types of impurities
and their typical
phase of appear-
ance in biomass- Solid Liquid Gaseous
derived producer
gases
Alkali and Nitrogenous Sulfurous
Particles Tars Halides
heavy metals species species
As the gasification is the origin for all impurities found in producer gases,
a short summary is given at this point as a basis for the discussion of
different gasification technologies. The removal of impurities and their
impact on the operation of SOFCs is discussed in the sections 2.3.1 and
2.4.7. As can be seen in Figure 2-7, impurities can be present in the
produced gas in the solid, liquid and gaseous phase. Following types of
impurities can be distinguished:
Particles are solid agglomerations of unreacted carbon and ash.
Alkali metals can be present in the solid state condensed on particles
at temperatures below 600 °C, [26], and in the liquid phase as aero-
sols or even gaseous phase at temperatures above 800 °C, [25].
Tars are a mixture of diverse organic species of oxygenated and
aromatic character formed during the thermal decomposition of the
biomass in the devolatilization sub-process and subsequent reactions
with reactive char. The complex mechanisms of primary and secon-
dary tar formation are discussed in depth in [20, 27]. Depending on
their molecular structure, tars may be in the liquid state as aerosols
21
up to temperatures around 400 °C, e.g. the boiling point of benzan-

thracene is approx. 435 °C.
Halides, of which hydrochloric acid (HCl) is the most important repre-
sentative, originate from the chlorine content of the biomass. Hydro-
chloric acid is in the gaseous phase due to its boiling point at 57 °C.
Nitrogenous species such as ammonia (NH3) and hydrogen cyanide
(HCN) are typically in the gaseous phase due to their boiling point be-
low 30 °C and originate from the nitrogen content of the biomass. Be-
tween 50 and 80 % of the nitrogen originally bound in the biomass
are released as gaseous species during gasification, [25].
Sulfurous species in producer gases are generally in the gaseous
phase with hydrogen sulfide (H2S) as most prevalent species. How-
ever, carbonyl sulfide (COS) and organic sulfur species, of which
thiophene (C4H4S) is an important representative, can also be pre-
sent in producer gases.
More generally, fixed bed, moving bed and entrained flow biomass gasi-
fication reactors can be distinguished. Figure 2-8 outlines the most im-
portant gasification reactor designs. The main difference is the gas/solid
flow pattern which in return determines the sequence of the above de-
scribed sup-processes of gasification.
Figure 2-8: Fuel Fuel in fixed bed

Gasification reactors
Gasification agent Fuel and bed material in fluidized bed
with gas/solid flow
patterns, adapted Producer gas Entrained flow oxidation zone
from [20]
Drying Drying
Devolatilization Devolatilization
Reduction Oxidation
Oxidation
Reduction
Updraft Downdraft Stationary Circulating Entrained

fixed bed fixed bed fluidized bed fluidized bed flow
Updraft wood gasifiers are fixed bed reactors. The wood is introduced
at the top of the gasifier while the gasification agent, typically air, is in-
troduced at the bottom through a grate. The producer gas is extracted at
the top of the gasifier after passing through the wood material present in
the gasifier. In contrast to the gasification agent and the producer gas,
22
the wood particles move downwards in the gasifier, running through the
sub-processes of gasification in the order "drying", "devolatilization",
"reduction" and finally "oxidation". Char is partially combusted in the oxi-
dation zone at temperatures between 1600 and 1000 °C. After the oxi-
dant is spent, the hot exhaust gas reacts with char in the reduction zone
at temperatures between 1000 and 600 °C. Further up in the gasifier,
the temperature of the producer gas becomes to low for reducing reac-
tions. The thermal decomposition of the wood takes place in the subse-
quent devolatilization zone at temperatures between 600 and 200 °C.
Before exiting the gasifier, the producer gas dries the fresh wood at tem-
peratures below 200 °C in the drying zone.
Updraft gasification producer gases are relatively cold with outlet tem-
peratures typically between 200 and 300 °C. In [28], producer gas tem-
peratures of 75 °C are reported. The cold gas efficiency including all
combustible species in the producer gas can reach values in excess of
90 % as a consequence of the low sensible heat content of the producer
gas, [4]. Due to the use of air as gasification agent, the heating value of
updraft gasification producer gases is low in the range of 5 MJ/mn3 on a
dry and tar free basis (dtf), [20]. Since the tars formed in the devolatiliza-
tion zone only pass the colder drying zone before exiting the updraft
gasifier, cracking is not likely to occur resulting in very high tar loads
with values up to 150 g per mn3 (dtf), [4]. The tar mix mainly consists of
oxygenated species such as phenols and acids as well as light aromat-
ics, [29]. Heavier aromatic species might be present in the condensed
phase. The particle load is usually low due to the filtration effect of the
wood material bed which the producer gas passes on its way to the
gasifier exit at its top, [4]. Alkali metals can be assumed to stay almost
completely in the solid phase. For the same reason as for the high tar
load, the presence of a notable amount of organic sulfur species is likely.
The updraft gasification process is very robust towards the size of the
wood particles and their humidity. The large internal drying zone allows
the conversion of wood with up to 50 % humidity. Updraft gasification
reactors are well scalable to a size of 20 MW thermal input, [28], and the
costs are relatively low due to the simple reactor concept.
Downdraft wood gasifiers are the second common type of fixed bed
reactors. The wood is introduced at the top of the gasifier, while the
gasification air is either introduced there as well or right above the
hearth zone in the lower part of the reactor. The producer gas exits the
gasifier at the bottom. Hence, wood particles and gasification air move
23
both downwards in the gasifier running through the sub-processes of

gasification in the order "drying", "devolatilization", "oxidation" and finally
"reduction". In the upper part of downdraft gasification reactors, the
wood is dried and decomposed at temperatures up to 600 °C. In con-
trast to the updraft gasification, the required heat does not directly come
from the producer gas but is introduced to these zones via gasification
air pre-heating and/or heating of the reactor walls, [19]. The char and
the tars produced in the devolatilization zone are partially combusted in
the oxidation zone at temperatures up to 1600 °C. The oxidation zone is
situated directly above the so called throat which is characteristic for
downdraft gasification reactors. This reduced cross-sectional area as-
sures a homogenous layer of hot char and is the location of the reduc-
tion zone. Tars originating from the devolatilization zone that were not
cracked in the oxidation zone are further degraded at temperatures
around 1000 to 600 °C.
Downdraft gasification producer gases typically have temperatures be-
tween 500 and 900 °C, [25], and according cold gas efficiency values of
70 to 80 %, [4]. Similar to the updraft gasification, the heating value of
downdraft producer gases is low in the range of 5 MJ/mn3 (dtf), [20]. Due
to the reactor internal tar cracking, tar loads are low with maximum val-
ues of 6 g per mn3 (dtf), [4]. However, the tar species in the tar mix are
mainly aromatic and hence more stable than those found in updraft gasi-
fication producer gases, [30]. The particle load is higher than that of up-
draft gasification producer gases. Due to the high producer gas tem-
peratures, alkali metals may be present in the liquid phase. Similar to
most of the tar species, organic sulfur species should be degraded in
the oxidation and reduction zones. Sulfur is therefore most likely to be
present in form of hydrogen sulfide and carbonyl sulfide.
The downdraft gasification process is sensible towards the size and ge-
ometry of the wood particles because it strongly influences the homoge-
neous distribution of the air in the oxidation zone and the controlled
bridge building in the throat area, [19]. The poor heat exchange of the
oxidation zone with the drying and devolatilization zones limits the wood
humidity to values below 20 %, [30]. The maximum size of downdraft
gasification reactors is around 2 MW thermal as the homogeneous gasi-
fication agent distribution in the hearth zone becomes increasingly diffi-
cult with larger diameters of the throat, [19].
Figure 2-9 shows the technical outlines of a commercial updraft and a
commercial downdraft gasification reactor. The updraft reactor (left) has
24
a rated thermal capacity of 4.8 MW and was built during 1993 in Har-
boore, Denmark, by the Babcock & Wilcox Volund A/S. The producer
gas has a temperature of 75 °C and an average tar load of 70 g/mn3 (dtf).
The gas is cleaned by means of a gas scrubber and electrostatic pre-
cipitator. With two gas engines producing 1536 kW electrical power, the
overall power efficiency departing from wood chips with approx. 50 %
humidity is in excess of 30 %, [29]. 3.2 MW of heat at 90 °C are deliv-
ered to a district heating grid. Since commissioning, the updraft gasifica-
tion reactor has been operated several ten thousand hours with 8000
hours per year in average. The gas engines were commissioned in 2003
and operated until 2005 for over ten thousand hours, [25].
The downdraft reactor (right) has a rated thermal input of 0.8 MW. It was
built during 2001 in Spiez, Switzerland, by Pyroforce AG. The almost tar
free producer gas exits the gasifier at approx. 650 °C. Gas cleaning in-
volves the addition of an adsorbent to the producer gas and its subse-
quent removal together with all other particles in a particle filter. A single
gas engine is installed producing a maximum of 200 kW of power and
270 kW of heat. Since commissioning, the gasification reactor has been
operated for 25000 hours and the gas engine for 15000 hours, [31].
Figure 2-9: Fuel

Technical outlines
Producer
of the updraft gasi- gas
fication reactor in
Harboore/DK (left), Biomass
Drying
[28], and the Pyro-
Air
force downdraft Drying
gasification reactor
Devolatilization
in Spiez/CH (right), Devolatil.
[32] Reduction Air
Oxidation Oxidation Producer
gas
Ash
Steam
and air Reduction
Ash
Fluidized bed gasifiers are categorized via their fluid dynamics and
heat transfer into stationary and circulating fluidized bed gasifier types.
Generally, the wood is introduced near the bottom of the gasifier where
25
it is immediately mixed with a bed material such as sand. The portion of

bed material in the reactor volume is typically over 90 %, [25]. The gasi-
fication agent is usually introduced through a nozzle floor or a frit. Due to
the mixing of the wood with the bed material, the sub-processes of gasi-
fication do not take place in a defined zone of the reactor but throughout
the whole reactor volume which yields almost isothermal conditions. The
temperature in the gasification reactors is usually kept below 950 °C to
prevent slagging of the bed material, [25]. The producer gas is extracted
via a cyclone at the top of the gasifier.
In stationary fluidized bed gasifiers, the flow velocity of the gasifica-
tion agent is adjusted so that only few bed particles are transported out
of the bed region. Thus, mainly ash particles are blown out of the fluid-
ized bed. The heat required for the gasification reactions is usually pro-
duced via partial combustion of the feed. Thus, air is commonly used as
gasification agent. However, there are also reactors employing mixtures
of air and steam or oxygen and steam. A recently developed approach
is the use of high temperature heat pipes as means of heat input to the
gasification reactor, [33].
The producer gas has temperatures around 800 to 950 °C, [25], which
allows cold gas efficiency values around 70 to 85 %, [4]. Where air is
used as gasification agent, the heating value of the produced gas is in
the range of 3.6 to 5.9 MJ/mn3 (dtf), [4]. Similar to downdraft gasification,
the high temperature prevalent in fluidized bed reactors yields a tar mix
mainly consisting of aromatic species. The tar load is however consid-
erably higher than that of downdraft gasifiers with 2 to 50 g/mn3 (dtf), [25].
This is because the formed aromatic tar species do not experience tem-
peratures as high as prevalent in the oxidation zone of downdraft gasifi-
ers and are hence degraded to a much lesser extent inside the gasifica-
tion reactor. As a matter of principle, the particle load is high. Alkali met-
als are mainly in the liquid phase due to the high producer gas tempera-
ture, [34], which also accounts for the predominant presence of hydro-
gen sulfide and carbonyl sulfide as sulfur species. However, a notable
amount of organic sulfur species may also be present.
The stationary fluidized bed gasification process is sensible towards the
size and geometry of the wood particles as these properties determine
the fluid dynamic behavior of the feed. To maintain the comparably high
temperatures in the gasification bed, the feed humidity usually has to be
lower than 15 %. This is particularly important for fluidized bed reactor
concepts where the required heat is not produced via partial combustion
26
of the feed but introduced by other means such as heat pipes, [33]. Sta-
tionary fluidized bed gasification is very well scalable for plant sizes be-
tween 1 to 50 MW thermal input, [4].
Circulating fluidized bed gasifiers feature almost the same properties
as their stationary counterparts, except that the flow velocities are con-
siderably higher. Consequently the mixing of bed material and wood
particles is better which yields fully isothermal conditions. Moreover, bed
material and unreacted char are continuously discharged from the gasi-
fication zone. In standard circulating fluidized bed reactors, the dis-
charged bed material and char are reintroduced to the gasification zone
where the required heat for the gasification reactions is produced via di-
rect partial combustion of the feed. Two-zone circulating fluidized bed
reactors however use the discharge mechanism for the spatial separa-
tion of gasification and combustion reactions. This way, air can be used
as oxidant in the combustion zone and steam as gasification agent in
the gasification zone yielding a non-diluted product gas. The heat for the
gasification reactions is transported from the combustion zone to the
gasification zone via the circulating bed material.
The producer gas from circulating fluidized bed reactors has basically
the same properties as that of stationary fluidized bed reactors. Excep-
tions are the higher heating values of two-zone concepts around 14.2 to
18.1 MJ/mn3 (dtf), [4], and the typically lower tar load in the order of 1 to
20 g/mn3 (dtf), [25]. This can be assigned to the frequently used reactive
bed materials such as olivine, [35]. Reasonable reactor sizes are be-
tween 20 to 200 MW thermal input, [4].
Figure 2-10 shows two fluidized bed gasification reactors employing
steam as gasification agent and air for the combustion of unreacted char
to produce the required heat for the gasification reactions. The station-
ary fluidized bed reactor (left) uses sodium heat pipes for the heat ex-
change between the combustion and the gasification chamber. The
"Biomass heat pipe reformer" (BHPR) was developed and patented by
the Technische Universität München. The first prototype reactor with a
thermal capacity of 150 kW was commissioned in 2002 and operated for
72 hours non-stop. The producer gas had a temperature around 800 °C
and was directly flared. In future, the product gas shall be produced with
a cold gas efficiency around 70 % , [33], and converted to electricity via
micro gas turbines and high-temperature fuel cells.
27
Figure 2-10:
Technical outlines Biomass Steam Producer gas Producer gas Exhausts
of the Biomass
heat pipe reformer
Combustion chamber
from TU Munich/D Exhausts
(left), [36], and the
Fast internally cir- Gasification
chamber
culating fluidized
bed reactor in Gasification
Güssing/AT (right), Filtration Biomass
chamber
[37] bed
Bed
Combustion Heatpipes material
chamber
Steam
Combustion
Combustion air air
The "Fast internally circulating fluidized bed" (FICFB) reactor depicted

on the right of Figure 2-10 employs the bed material for the heat ex-
change between the combustion and the gasification chamber. The re-
actor was built in Güssing/Austria by the Repotec Umwelttechnik GmbH.
Commissioning was completed in 2002 with a rated thermal capacity of
8 MW. The product gas exits the gasifier at approx. 850 °C with several
g/mn3 (dtf) of tar load. The gas cleaning comprises a hot gas filtration
and a gas scrubber. One single gas engines produces 2 MW of electri-
cal power, which translates in an overall power efficiency of 25 %, and
4.5 MW of heat at 120 °C. With over 25000 operating hours of the gasi-
fication reactor and the gas engine, the FICFB plant in Güssing is one of
the most successful biomass gasification power plants in existence.
Entrained flow gasifiers were initially developed for the gasification of
coal which is fed to the reactors as either dry powder or water slurry.
Oxygen is commonly employed as gasification agent yielding very high
reactor temperatures between 1250 to 2000 °C. The conversion of
wood to powder or slurry requires more effort than the simple grinding of
coal. Entrained flow wood gasifiers are therefore usually combined with
a pyrolysis reactor in which the sub-processes "drying" and "devolatiliza-
tion" take place at temperatures around 600 °C. The heat required for
the pyrolysis process may either come from exhaust gases or be pro-
duced via partial combustion of the feed. The pyrolysis reactor is then in
fact a pre-gasification reactor. The sub-processes "oxidation" and "re-
28
duction" occur in the actual entrained flow gasification reactor. As the

pyrolysis yields a mixture of gaseous, liquid and solid products the com-
position of which can be controlled via the pyrolysis process conditions,
the form of the feed depends on the entrained flow gasifier design and
can be either a gas/solid or a liquid/solid mixture.
Entrained flow gasification producer gases have temperatures between
800 and 1200 °C, [38, 39]. The cold gas efficiency of this process is in
the range of 50 to 70 % depending on the reactor temperature. Despite
the use of oxygen as gasification agent which inhibits nitrogen dilution,
the heating value of entrained flow gasification syngases is only in the
range of 8.8 to 9.3 MJ/mn3 (dtf). This is because the gases are almost
free of hydrocarbon species with high heating values. The syngases are
generally tar and particle free. Alkali salts may be present in the liquid
and even gaseous phase. Sulfur is present in form of hydrogen sulfide
and carbonyl sulfide only.
Entrained flow gasifiers are at the most sensible towards the ash melting
point and content of the feed. The humidity of the gasified feed usually
plays a less important role for the gasification itself but may be limited by
the upstream pyrolysis. Due to the complexity of the involved reactors,
entrained flow gasification is a large scale technology aiming at thermal
inputs up to several hundred MW.
Figure 2-11:
Technical outlines
of the Carbo-V
gasifier developed
by CHOREN/D
(left), [39], and the
GSP gasifier de-
veloped by the
Deutsches Brenn-
stoffinstitut/DDR,
[38]
Figure 2-11 shows the technical outlines of two entrained flow biomass
gasifiers. CHOREN Industries GmbH has developed the Carbo-V® gasi-
fier (left) and is currently building a Biomass-to-liquids (BtL) plant in
Freiberg, Germany, based on this gasification technology. The biomass
29
is first gasified with very little air, producing a highly tar laden gas and
char. The gas is partially oxidized with oxygen in the entrained flow gasi-
fier. The hot gases are cooled down to approx. 800 °C in the reduction
section of the gasifier by means of quenching with grinded char from the
low temperature gasification. The char is simultaneously gasified, [39].
The GSP gasifier (Gaskombinat Schwarze Pumpe, right) was developed
for the gasification of salt-containing brown coal and therefore suits the
gasification of biomass with high ash content well. In fact, the ash acts
as protective layer for the reactor walls and is therefore required. As the
gasifier requires a liquid feed, the Forschungszentrum Karlsruhe has
developed a pyrolysis process that is suited for the production of slurry
from various types of biomass. With 1200 °C, [38], the producer gas is
hotter than that of the CHOREN gasifier as no reactor internal quench-
ing is conducted. Table 2-4 summarizes the above paragraphs.
Table 2-4: Down- Fluid. Circul. Entrain.

Unit Updraft
Comparison of draft Bed FB flow
biomass gasifica- Air/O2/
tion technologies Gasification agent [-] Air Air
H2 O
H2 O O2
based on [4, 19, H2 10-14 15-21 15-22 17-36 29-40
25, 30, 40] CO 15-20 10-22 13-15 36-51 39-45
CO2 [mol-%]
8-10 11-13 13-15 7-15 18-20
CH4 2-3 1-5 2-4 0.1-0.6 0.05-0.1
C2 - 0.5-2 - 1.4-7.5 -
N2 53-65 37-63 44-57 0-39 0.1-9
LHV [MJ/mn3 (dtf)] 3.7-5.1 4.0-5.6 3.6-5.9 14.2-18.1 8.8-9.3
Gas temperature [°C] 75-300 500-900 800-950 800-950 800-1000
Cold gas efficiency [%] >90incl.tar 65-75 70-85 60-70 50-70
Particle load 0.1-3 0.02-8 20-100 8-100 -
[g/mn3 (dtf)]
Tar load 10-150 0.01-6 2-50 1-20 0
mostly oxygen- oxygen-
Tar signature oxygen- aromatic ated and ated and none
ated aromatic aromatic
liquid/
Alkali phase solid liquid liquid liquid
gaseous
partially mainly partially partially
Sulfur signature organic inorganic organic organic
inorganic
[-]
Feed size & non- homoge- homoge- homoge- Suspen-
geometry sensitive neous neous neous sion
Feed humidity <50 % <20 % <15 % <15 % -
sensitive,
sensitive,
Process robustness stable bridge stable stable
slagging
building
Reactor size [MWth] 0.1-20 0.1-2 1-50 20-200 30-600
30
2.3 Gas processing

The properties of the raw producer gas and the requirements of the em-
ployed fuel cell determine the design and the components of the gas
processing in B-IGFC systems, Figure 2-1. The term "Gas processing"
stands for the combination of gas cleaning and conditioning. Gas clean-
ing summarizes all required steps for the removal of impurities, while
gas conditioning subsumes all processes involved in the conversion of
the producer gas to a fuel gas suitable for the fuel cell in question.
2.3.1 Gas cleaning

Gas cleaning systems for producer gases usually comprise several unit
operations in series of which each serves to the removal of at least one
impurity type or specie. Currently a variety of technologies and appara-
tuses can be employed for the removal of impurities from gas streams.
The state-of-the-art technologies and recent research results are out-
lined below focusing on the removal of the impurities discussed in sec-
tion 2.2.3. The given information is partially based on [25, 34].
2.3.1.1 Particles
Particles in producer gases can cause erosion and blocking in the
equipment downstream of the gasifier. Further, particles are subject to
emission limits. For these reasons, the removal of particles is an integral
part of every gas cleaning system for producer gases from biomass
gasification. Ordered by their maximum operational temperature, the fol-
lowing particle removal technologies can be distinguished:
Wet scrubbers
Electrostatic precipitators
Cyclones
Barrier filters
Wet scrubbers employ a liquid, often water, for the removal of particles
from gas streams. Their maximum operational temperature is usually
below 100 °C which requires a cool down of the producer gas. The par-
ticles are captured in wash fluid droplets which are then removed from
the gas stream in a demister. The most common wet scrubber type is
the venturi scrubber. It allows removing particles below 1 μm diameter
and achieving particle loads around 10 to 20 mg/mn3 (dtf). However,
venturi scrubbers cause a pressure loss between 30 and 200 mbar.
31
Electrostatic precipitators consist of two zones. In the first zone, the

product gas with the particles flows past high voltage electrodes that
transfer a charge to the particles but do not interfere with the product
gas. In the second zone, the product gas with the charged particles
passes collector plates with the opposite polarity of the high voltage
electrodes in the first zone. The charged particles consequently precipi-
tate towards the collector plates where they deposit. Depending on the
method of particle removal from the collector plates, electrostatic pre-
cipitators can be classified in wet and dry. Dry precipitators use me-
chanical means, while wet employ a liquid, usually water, to flush the
deposited particles from the collector plates. Therefore, the maximum
operational temperature of wet precipitators is limited to 65 °C, in con-
trast to their dry counterparts which may be operated at temperatures up
to 500 °C. Wet and dry electrostatic precipitators achieve particle loads
below 25 mg/mn3 (dtf), [29], and cause negligible pressure losses.
Cyclones are predestined for the removal of large quantities of large
particles. For this reason they are an integral part of most fluidized bed
gasification reactors. The particles are separated from the producer gas
using the centrifugal force resulting from constraining the gas flow to a
circular path, Figure 2-12 (left).
Figure 2-12:
Clean gas
Working principle
of cyclones and Pulse gas
rigid barrier filters, Raw gas Valve Injector
[41] Venturi
Clean gas
nozzle
Dip pipe Pulse
control
Cyclone Dust
body cake
Filter
element
Particle Raw gas

collecting
basin
Particles Particles
As a matter of principle, cyclones are relatively simple, robust and cheap

devices. Their maximum operational temperature is solely limited by the
32
material of construction. Especially for the application of high tempera-

ture fuel cells and gas turbines, this is a considerable advantage as no
cool down of the producer gas is required and consequently no sensible
heat of the gas is lost. Most commonly, several cyclones are connected
in series thereby improving the collection efficiency. These multiclones
reliably remove up to 90 % of the particles with a maximum diameter of
5 μm causing less than 10 mbar pressure loss. The collection efficiency
of smaller particles is unsatisfactory even for high performance cyclones.
This is particularly problematic as the char particles in producer gases
from biomass gasification are predominantly of this smaller size fraction.
Consequently particle loads of producer gases de-dusted by cyclones
may still contain particle loads up to 30 g/mn3 (dtf).
Barrier filters use porous filter elements through which gases can
penetrate but inhibit the passage of particles, Figure 2-12 (right). The
collection efficiency increases with the thickness of the deposited dust
cake. The filter elements may consist of rigid metallic or ceramic mate-
rial or flexible woven material. Besides, barrier filters may consist of
packed beds of ceramic spheres, sand etc. The removal of the depos-
ited particles is usually carried out by periodically passing a clean gas
flow through the filter elements in reverse direction of the normal gas
flow. For reasons of costs reduction, barrier filters are usually employed
in combination with cyclones to minimize the amount of cleaning pulses.
Bag filters mainly use the filtration effect of the deposited dust cake. Be-
sides the reverse flushing of the filter elements, the dust cake can be
removed by shaking. The maximum operational temperature is below
250 °C, which is acceptable in systems where the producer gas has to
be cooled down for the end use device such as gas engines. The tars
typically present in producer gases start condensing at the temperatures
below 400 °C and may cause serious jamming problems. For this rea-
son, bag filters are uncommon for the filtration of producer gases. How-
ever, precoat materials were recently demonstrated to prevent jamming,
thus enabling the application of bag filters for producer gas filtration.
Packed bed filters can be operated in a wide range of temperatures. The
greatest advantage of packed bed filters is that the bed materials can be
easily exchanged once blocked by. e.g. tars. Therefore packed bed fil-
ters are often used for combined dust and tar removal.
Rigid barrier filters feature the possibility to retain the sensible heat of
producer gases due to maximum operation temperatures up to 900 °C.
33
Hence they are predestined to the combination of biomass gasification

with gas turbines and high-temperature fuel cells. Ceramic filter ele-
ments are however fragile and may break due to thermal stress induced
by thermal cycling. Further, the ceramic materials are prone to react with
alkali vapors leading to decomposition and plugging. Metallic filter ele-
ments have been reported to be prone to jamming especially with the
high tar loads present in producer gases, see section 3.4.2. Further, par-
tial producer gas cooling might be necessary to prevent metal sintering.
Depending on the pore diameter of the used porous material, barrier fil-
ters can almost completely remove particulates down to 0.5 μm diame-
ter. However, the according pressure loss is considerable.
2.3.1.2 Alkali metals

Alkali salts evaporate at temperatures above 800 °C. This volatilization
is further facilitated by the chlorine present in biomass, [42, 43], see
Table 2-2. Vaporized alkali salts cause damage to ceramic filter materi-
als and corrosion of e.g. gas turbine blades.
State-of-the-art alkali removal from producer gases comprises the cool-
ing of the gas below 600 °C, where alkali salts condense forming parti-
cles with diameters less than 5 μm. The particles are subsequently re-
moved by means of particle filtration, [26]. After filtration with rigid filter
elements at temperatures around 400 to 500 °C the alkali concentration
in the filtered gas is typically in the order of 0.1 ppmw which translates in
a removal efficiency above 99 %, [44]. The main drawback of this re-
moval method is the inherent loss of sensible heat.
Recent research has shown that activated bauxite as alkali adsorbent
achieves removal efficiencies for sodium and potassium between 95
and 99 % at temperatures between 650 and 750 °C, [45]. Reliable alkali
removal systems based on these findings are however not available yet.
2.3.1.3 Tars
Producer gases can contain considerable amounts of tars. At tempera-
tures below their dew point and/or at elevated pressures, tars condense
causing operational problems due to the formation of droplets which ac-
cumulate to sticky films on cold surfaces of e.g. pipes and other equip-
ment. Besides the condensation related issues, tars may cause carbon
deposition problems at very elevated temperatures. Systems employing
gas engines and gas turbines as end use devices require almost com-
34
plete tar removal to avoid condensation problems, while systems using

high-temperature fuel cells may cope with moderate tar loads. Besides
the primary tar reduction via optimized gasification processes, see sec-
tion 2.2.3, tars may be removed by physical means or decomposed by
thermal and catalytic means.
Physical tar removal is the most frequently employed technology, in-
volving the controlled condensation of tars and the subsequent removal
of the tar droplets in wet scrubbers or electrostatic precipitators following
the same principles as for the removal of particles. The maximum opera-
tional temperature is limited to values below 65 °C for wet scrubbers.
Electrostatic precipitators may be operated at temperatures up to
150 °C if the collected tars exhibit low viscosity and thus do not need to
be actively flushed from the collector plates with water, but drip down on
their own. In the cleaned producer gas, tar loads between 20 and 40
mg/mn3 (dtf) can be achieved with water based wet scrubbers. Oil based
wet scrubbers may achieve tar loads below 10 mg/mn3 (dtf) due to the
additional solution of tars in the oil. The heat loss resulting from the re-
quired gas cool down is the main disadvantage of this technology. The
main advantage is that physical tar removal and particle removal may be
carried out in one single apparatus reducing the equipment cost.
Thermal tar decomposition requires high temperatures, ranging from
900 °C for oxygenated tar species originating from updraft gasification
to 1200 °C and more for the refractory tar species formed during e.g.
downdraft gasification, [34]. Electrically heated surfaces are one possi-
bility to generate the required temperatures for thermal tar decomposi-
tion. The major drawbacks of this technology are incomplete tar decom-
position and poor energy efficiency. Another possibility is the partial oxi-
dation of the producer gas. Using oxygen, temperatures over 1200 °C
can be achieved at low air-to-fuel ratios where only little of the producer
gas energy content is converted to heat. However, oxygen is expensive.
In contrast, air is abundant but the heating of the nitrogen introduced to
the producer gas consumes a considerable amount of energy. For both
oxidants the energy content of the tars is preserved in the producer gas
either in form of sensible heat or heating value of the tar decomposition
products. This is the main advantage of thermal tar decomposition tech-
nologies compared to physical tar removal technologies.
Catalytic tar decomposition aims at the conversion of tars at tempera-
tures between 800 and 900 °C. The catalyst materials may be placed in
the gasification reactor (in situ tar decomposition) or in an external reac-
35
tor. The potential advantage of in situ tar decomposition is that the tars
may be decomposed at the same time that they are formed. However,
experiments showed that in fixed bed gasification reactors the contact
time between catalyst and tars is insufficient for complete tar decompo-
sition and that in fluidized bed reactors the attrition of the catalyst parti-
cles causes a fast deactivation. Further, the means of temperature con-
trol for in situ tar decomposition are limited. The latter is the main advan-
tage of external tar decomposition reactors, which however involve addi-
tional costs. State-of-the-art non-metallic catalyst materials for tar de-
composition are dolomites, zeolites and calcites. These materials are
cheap, insensitive towards most impurities present in producer gases
and allow tar decomposition rates of up to 99 %. Metallic tar decomposi-
tion catalysts employ a wide range of materials such as Ni, Mo, Co, Pt
and Ru in pure form or as mixtures supported on silica alumina, zeolites
etc. The main advantages of metallic catalysts are that they allow almost
complete tar decomposition and simultaneous decomposition of ammo-
nia. However, the long-term application in biomass gasification systems
has not been demonstrated yet and the costs are higher than for non-
metallic catalysts. Sections 2.3.2.1 and 2.3.2.2 give additional informa-
tion for the catalytic decomposition of hydrocarbons and tars.
2.3.1.4 Halides
The average chlorine content of untreated wood is low, which yields low
concentrations of hydrochloric acid in producer gases. If required, hy-
drochloric acid can be removed from producer gases using wet scrub-
bers which are well tested and commercially available. To avoid heat
losses, high temperature capable adsorbent materials can be used such
as e.g. lime, Eq. 17.
CaO + 2 HCl → CaCl2 + H 2O Eq. 17
Lime can be used up to temperatures slightly below the melting tem-

perature of the formed calcium chloride being 770 °C. Hydrochloric acid
adsorption is a well tested technology.
2.3.1.5 Nitrogenous species

Ammonia (NH3) is the most important nitrogen-containing impurity in
producer gases. During combustion at high temperatures in gas engines
and gas turbines, ammonia is likely to be converted to nitrogen oxides
(NOx), which are subject to strict legal limits and difficult to remove from
36
the exhaust gases. Therefore, ammonia is commonly removed from the

producer gas prior to its combustion in gas engines and gas turbines. In
contrast, recent research has shown that ammonia is very well con-
verted in high-temperature fuel cells with very little formation of NOx, [46,
47]. This strongly indicates that ammonia removal might not be neces-
sary when using high-temperature fuel cells for the conversion of pro-
ducer gases. Nevertheless, a short summary of the available ammonia
removal technologies is given below.
Ammonia can be removed from producer gases either by catalytic de-
composition or by wet scrubbing. The latter technology, using the good
solubility of ammonia in water, is well known and reliable. The catalytic
decomposition of ammonia is usually carried out with catalysts similar to
those used for the tar decomposition. Dolomites, nickel- and iron-based
catalysts have been shown to achieve ammonia conversion rates in ex-
cess of 99 % at temperatures higher than 900 °C. However, thus far this
technology has not been applied in commercial systems. Further, the
employed catalysts are prone to deactivation by sulfur and halide com-
pounds.
2.3.1.6 Sulfurous species

The average sulfur content of untreated wood is 600 mg/kg dry and ash
free (daf), [48]. During combustion of the producer gas, the sulfurous
species are converted to sulfur oxides (SOx) which can be reliably re-
moved from the exhaust gases to meet legal emission limits. The re-
moval of sulfur species from the producer gas prior to the combustion in
gas engines and gas turbines is therefore uncommon. In contrast, the
nickel-based catalysts frequently used in high-temperature fuel cells are
prone to sulfur poisoning, which reduces the power output noticeably
and may also cause permanent damage. To prevent this and to mini-
mize sulfur-induced power losses, the sulfur concentration in the pro-
ducer gas has to be well below 10 ppmV, see section 3.4.5. The typi-
cally targeted sulfur concentration of below 1 ppmV can be reached by
either absorptive or adsorptive methods.
Alkaline wet scrubbers have long been used for the removal of sulfur
species from natural gas and syngases from coal gasification. The most
important commercially available processes are the Rectisol process
developed by the Linde AG and the Purisol process from the Lurgi
GmbH. These processes employ different solvents which can be regen-
erated under separation of hydrogen sulfide containing gases that are
37
usually converted to elemental sulfur via the Claus process. These

processes are however complex and, from the economical point of view,
hardly applicable to biomass gasification systems. Besides, the wet
scrubbing processes require a gas cool down which is particularly unfa-
vorable when employing high-temperature fuel cells as these require
high producer gas temperatures.
The most common adsorptive sulfur removal process is based on the
adsorption of hydrogen sulfide onto zinc oxide, Eq. 18. Despite that zinc
oxide and zinc sulfide are non-volatile up to 1700 °C, in reducing at-
mospheres zinc oxide is likely to be reduced to pure zinc which has a
considerably lower melting temperature. The maximum operation tem-
perature of zinc oxide adsorbents is therefore limited to 350 to 450 °C.
ZnO + H 2 S → ZnS + H 2O Eq. 18
Depending on the operational temperature and the steam content of the

producer gas, sulfur concentrations in the ppb range can be reached
without the need of a heavy cool down of the producer gas. The sensi-
ble heat of the producer gas can thus largely be preserved. However,
the formed zinc sulfide has to be recycled, which involves costs. Further,
hydrochloric acid removal is mandatory before the producer gas gets in
contact with the zinc oxide adsorbent. This is because hydrochloric acid
may react with the zinc oxide forming zinc chloride which is volatile at
temperatures above its melting point at 275 °C, Eq. 19. The removal of
hydrochloric acid was discussed in section 2.3.1.4.
ZnO + 2 HCl → ZnCl2 + H 2O Eq. 19
Recently, the adsorption of sulfur on metal-oxides such as titanium, iron,

magnesium and aluminum oxides at temperatures around 500 °C has
engendered some interest. In contrast to zinc sulfide, the formed metal
sulfides can be regenerated with steam producing hydrogen sulfide con-
taining exhaust gases. However, these materials are still under investi-
gation and not yet commercially available. The same applies to the ad-
sorption of sulfur onto dolomite and lime particles which are introduced
in the product gas stream and removed by means of particle removal.
Adsorption capacities in excess of 85 % were reported under lab condi-
tions. Activated carbon may also be used for the sulfur removal from
producer gases at temperatures up to 550 °C, [49]. However, the typi-
cally high steam content of producer gases might cause difficulties and
has not yet been investigated.
38
The most important drawback of all discussed adsorption methods is

that organic sulfur compounds are not directly adsorbed. Hence a con-
version of these species to hydrogen sulfide is required. The state-of-
the-art in the petrochemical industry is the conversion via hydrogenation
over Ni/Mo or Cr/Mo catalysts with excess hydrogen. As the hydrogen
content of producer gases is usually moderate, hydrogenation is most
probably not feasible in biomass gasification systems. However, due to
the tar-like nature of organic sulfur species, tar removal or decomposi-
tion technologies can be used for the conversion to hydrogen sulfide or
the direct removal from producer gases.
2.3.2 Gas conditioning

Hydrogen is the most easily convertible fuel for fuel cells. However, it is
difficult to store and prone to explosion, [50]. Further, it is not readily
available as a fuel but needs to be produced from e.g. natural gas or
biomass or by electrolysis. In addition, there is no area-wide hydrogen
distribution infrastructure available currently. Therefore, research activi-
ties focus on the system-integrated generation of hydrogen-rich gases
from hydrocarbons. Besides hydrocarbons, producer gases can contain
oxygenated organic compounds. The amount of oxygenates depends on
the biomass conversion process and can reach important levels.
State-of-the-art gas conditioning processes are steam reforming (STR),
autothermal reforming (ATR) and catalytic (CPO) or non-catalytic partial
oxidation (PO). These processes may take place in external reactors or
directly in the fuel cell.
2.3.2.1 Steam reforming

The STR reaction was discovered in the early 19th century. Since then,
the STR process using methane as educt has been developed by sev-
eral groups and companies and is now one of the cheapest, [51], and
most energy efficient, [15], processes for syngas production. Syngas is
a mixture of hydrogen and carbon monoxide. The endothermic STR re-
action of hydrocarbons and methane in particular are stated below.
⎛ m⎞
Cn H m + n ⋅ H 2O → n ⋅ CO + ⎜ n + ⎟ ⋅ H 2 with (− ΔH < 0) Eq. 20
⎝ 2⎠
⎛ kJ ⎞
CH 4 + H 2O → CO + 3 ⋅ H 2 with ⎜ − ΔH R0 = −206 ⎟ Eq. 21
⎝ mol ⎠
39
Eq. 21 shows the high heat requirement of the STR and explains why
reactor designs are typically limited by heat transfer rather than the re-
action kinetics. STR of most hydrocarbons takes place only over suit-
able catalysts and at high temperatures typically over 800 °C. Group VIII
metals are employed as STR catalysts, [52], of which Ni is the mostly
used material due to cost considerations. SOFCs offer all factors for an
integrated STR process, see also section 2.4.3. The conversion can be
performed cell internally or externally using the SOFC waste heat.
The main problem of the STR process is carbon deposition. High steam
partial pressures help to prevent carbon deposition on the catalyst mate-
rial. Typically, steam-to-carbon ratios (SC) of two and higher, [53], are
required to prevent carbon deposition. The tendency towards carbon
deposition increases with rising molecular weight of the hydrocarbon
species in the educt gas, especially for polycyclic molecules, [54].
Poisoning of the STR catalyst is another issue. Hydrocarbon fuels are
usually contaminated with small amounts of sulfur, which is likely to
chemisorb on any metallic surface. Hence, a desulphurization step has
to be carried out prior to the reformer, see section 2.3.1.6.
Oxygenated organic compounds can be steam reformed using Ni-based
catalysts, [55, 56]. The corresponding reaction can be stated as:
⎛ m ⎞
Cn H m Ok + (n − k ) ⋅ H 2O → n ⋅ CO + ⎜ n + − k ⎟ ⋅ H 2 with (− ΔH < 0) Eq. 22
⎝ 2 ⎠
Although Eq. 22 appears similar to Eq. 20, there are significant differ-
ences between the reaction mechanisms of pure hydrocarbons and
oxygenates. According to [55], the oxygenated organic compounds
found in biomass-derived gases are thermally unstable and decompose
without forming significant amounts of coke at the operating conditions
of STR processes. The formed intermediates are instantly converted to
hydrogen and carbon monoxide. Experiments showed that oxygenate
STR allows lower reforming temperatures and higher space velocities
than the comparable hydrocarbon STR process, [57]. The following sec-
tions provide an overview of state-of-the-art hydrocarbon STR reactors
and integration possibilities of the STR process into SOFC designs.
2.3.2.1.1 External steam reforming

The heat integration of external STR reactors in SOFC systems is un-
dertaken with heat exchangers. Heat on a high temperature level is re-
covered from the flue gas of the SOFC and transferred to the external
40
STR reactor. The advantage of this system design is that relatively

cheap standard STR reactors can be used, [51], see Figure 2-13. How-
ever, the inherent heat transfer limitations usually lower the system effi-
ciency. In addition, the STR process requires steam at an appropriate
temperature. This steam is usually generated using de-ionized water
and heat from the flue gas of the SOFC, while the steam content the
SOFC flue gas remains unused.
Figure 2-13:
External reforming
reactor with inte-
grated heat ex-
changer designed
by Haldor Topsoe,
[58]
2.3.2.1.2 Indirect internal steam reforming

Indirect internal reforming uses the operation temperature of the SOFC
and the steam content of the flue gas via a recycle, [51], see Figure 2-14.
Figure 2-14:
Indirect internal
reforming concept
by the Siemens
AG, [59]
41
This integration strategy allows simplified system designs and generates

additional cooling means for the SOFC stack via direct radiative heat
exchange between the stack and the reforming reactor. The latter is also
employed to reduce the required cooling air mass flow. Nevertheless,
the integration of a separate STR reactor imposes important constraints
for the stack design and is not necessarily possible for all cell designs.
2.3.2.1.3 Direct internal steam reforming

State-of-the-art SOFCs have nickel cermet anodes which provide suffi-
cient STR activity at the typical operation temperatures, see section also
2.4.6.1. Therefore, the STR reactions may directly occur on the anode.
The endothermic character of the STR can be used to chemically cool
the cells, reducing the electricity need of the cooling air blower and
hence increasing the net electrical efficiency of SOFC systems. The re-
quired steam and heat are continuously produced in exothermic electro-
chemical hydrogen oxidation and consumed in STR reactions. Well bal-
anced, a more uniform temperature distribution over the cells can result,
see also section 2.4.3. The overall system design is further simplified as
no separate STR reactor is needed, [51].
2.3.2.2 Partial oxidation and autothermal reforming

The partial oxidation (PO) of hydrocarbons is an alternative to the STR
process. Hydrocarbons are degraded to a mixture of hydrogen and car-
bon monoxide in a strongly exothermic reaction which can be written as:
n m
Cn H m + ⋅ O2 → n ⋅ CO + ⋅ H 2 with (− ΔH > 0) Eq. 23
2 2
The hydrocarbons react with oxygen at temperatures between 850 and

1700 °C whereas the provided amount of oxygen is insufficient for the
complete combustion, [60]. PO reactors are largely used to process liq-
uid fuels as evaporation and conversion can take place in one step. PO
is also used to process gaseous fuels for mobile or small stationary ap-
plications due to the compact reactor design. No flue gas recycles or de-
ionized water is needed to run a PO reactor, making the process simple.
The main drawback of the PO process is that fuel is burned to produce
the required high temperatures resulting in moderate efficiency, [61].
Recently, catalysts have been developed to lower the process tempera-
ture and increase the energy efficiency. This process, which has been
termed catalytic partial oxidation (CPO), takes place at temperatures be-
42
low 800 °C, [62]. In-situ temperature profile measurements of CPO

monoliths have shown that two different reaction zones establish during
operation. In the hotspot zone, oxygen is consumed in oxidation reac-
tions producing temperatures around 1000 °C. The remaining hydrocar-
bons then take part in reforming reactions which are sustained by the
heat and the steam generated in the hotspot zone. For more information,
see [63].
The combination of STR and CPO is referred to as autothermal reform-
ing (ATR). In contrast to the PO and the CPO, where only molecular
oxygen is used as oxygen source, the ATR uses molecular oxygen and
steam as oxygen sources. The reaction can be written as follows:
n n ⎛n m⎞
Cn H m + ⋅ O2 + ⋅ H 2O → n ⋅ CO + ⎜ + ⎟ ⋅ H 2 with (−ΔH ≥ 0) Eq. 24
4 2 ⎝2 2 ⎠
The ATR process requires less steam than the STR process. The re-
quired heat is generated via partial oxidation of the fuel, avoiding the
need for a complex heat management, [58]. Both processes are well
suitable for the conversion of biomass-derived fuel gases. For oxygen-
ated organic compounds, Eq. 25 and Eq. 26 are reformulated as follows:
n−k m
C n H m Ok + ⋅ O2 → n ⋅ CO + ⋅ H 2 with (− ΔH > 0) Eq. 25
2 2
n−k n−k ⎛n−k m⎞
Cn H m Ok + ⋅ O2 + ⋅ H 2O → n ⋅ CO + ⎜ + ⎟ ⋅ H2
4 2 ⎝ 2 2⎠
Eq. 26
with (− ΔH ≥ 0)
CPO and ATR of oxygenated organic compounds have been shown us-
ing pyrolysis gas from cedar wood powder, [64]. It was found, that the
tar content of the pyrolysis gas was almost completely converted to hy-
drogen and carbon monoxide.
2.4 Solid oxide fuel cell

2.4.1 Fuel cells in general
Fuel cells are energy converters which directly convert the chemical en-
ergy of a fuel into electricity and heat via electrochemical reactions. Fuel
cells produce electricity as long as fuel is supplied. The principle was
discovered in 1838 by the two scientists Christian Friedrich Schönbein
and Sir William Grove, [65]. However, the fuel cell technology did not
gain significance until the 1960s, where alkaline fuel cells were first em-
43
ployed in the NASA space program to produce electricity and drinking

water in space crafts, [58]. Since then, fuel cells have been developed to
cover applications such as combined heat and power generation, vehi-
cle propulsion and replacement for batteries.
Fuel cells could play an important role in future energy conversion due
to their high efficiencies, modular design, low noise and environmental
impact. Although significant progress has been made in recent years,
there are still many technical and economical problems, e.g. insufficient
durability and the need for expensive catalyst materials. These problems
have to be resolved to allow for a strong market penetration of fuel cell
technologies.
Figure 2-15 gives a comparison of different fuel cell types. Currently the
most important fuel cell (FC) types are the polymer electrolyte mem-
brane FC (PEMFC), the molten carbonate FC (MCFC) and the solid ox-
ide FC (SOFC). The direct methanol FC (DMFC) is derived from the
PEMFC. Besides, there are the alkaline FC (AFC) and the phosphoric
acid FC (PAFC), which are however no longer in fashion.
The different FC types are distinguished by the electrolyte used and
their operating temperature. The MCFC and the SOFC are high-
temperature, the PAFC mid-temperature and the rest low-temperature
FCs. The last-mentioned (PEMFC, AFC) are aiming at mobile and port-
able applications, while SOFCs, MCFCs and PAFCs aim at stationary
heat and power generation.
Fuel cells can operate on a variety of fuels ranging from pure hydrogen
to gasified solid fuels, such as biomass or coal. Presently, hydrogen is
the only fuel which can be oxidized electrochemically in PEMFC, AFC
and PAFC. In addition, SOFCs and MCFCs can also electrochemically
convert carbon monoxide and hydrocarbons. Fuel gases from renew-
able sources such as producer gas from biomass gasification are of
special importance for the future. In this respect, MCFCs and SOFCs
have the greatest potential as their actual development is focused on
pre-reformed natural gas as fuel which exhibits similar properties as pro-
ducer gases.
44
Figure 2-15:
Comparison of fuel
cell technologies
Pictures: [66]
45
2.4.2 Technology features

Compared to other fuel cell types, SOFCs feature:
The highest fuel to electricity efficiency, [67]. Fuel cells in general are
not limited by the Carnot efficiency.
A wide operation temperature range (500 up to 1000 °C), [68].
Great potential for hybridization due to the typically high operation
temperature and incomplete fuel utilization, [67]. The generated sur-
plus of high quality heat enables new power plant designs with Car-
not heat engines and/or steam turbines as bottoming cycles. Ther-
modynamic systems analyses revealed that most hybrid systems
proposed in the literature have the potential for highly efficient heat
and power generation, especially when the systems are operated un-
der elevated pressure. The NOx emissions of such hybrid systems
are expected to be low due to mostly electrochemical oxidation of the
fuel, which takes place at comparably low temperatures.
The application of low-cost catalyst materials due to the high opera-
tion temperatures. Typically, nickel is employed as anode catalyst in
SOFCs while platinum is employed in PEMFCs, [68].
High fuel flexibility. SOFCs can directly convert hydrogen, carbon
monoxide and hydrocarbons, [68]. Furthermore, the employed cata-
lysts and the high operation temperature allow for internal reforming
of hydrocarbons and shifting of carbon monoxide. Hence, no complex
gas processing aiming at the production of pure hydrogen is needed
for the use of natural gas, syngas or producer gases.
Good scalability from watts to multi- megawatts, [67].
Almost noiseless operation. SOFCs for the application as replace-
ment for batteries in cars, so called auxiliary power units (APUs), are
developed by several groups, [67].
However, the SOFC technology is not mature yet. The actual status of
development ranges from fundamental research mainly in the field of
material science to pre-commercial system operations and development
of market penetration strategies. The most important challenges which
have to be addressed prior to a successful market penetration are:
The manufacturing processes of the ceramic raw materials and cell
components as well as the cell- and stack assembly processes are
not yet optimized. Consequently, the stray area of cell performance
and lifetime is usually wide and the number of deficient cells is high.
46
Simplifying the production processes should allow for significant cost

reductions. To achieve this, every step in the stack production proc-
ess has to be well understood, [68].
The elevated operation temperatures induce high requirements for
the employed sealing and interconnect materials. Currently these re-
quirements can only be met, if at all, by expensive materials, [68].
Start-up and shut-down usually take long and the tolerance of SOFCs
towards temperature cycles is low.
In the emerging distributed power and heat generation market, SOFCs
will face competition not only from established less efficient, but, less
expensive, technologies such as gas engines, but also from micro tur-
bines and not least MCFCs assuming comparable reliability and costs.
2.4.3 Working principle

Walter Hermann Nernst’s discovery in 1899 that zirconia (ZrO2) is con-
ductive for oxygen ions at elevated temperatures was the starting point
for the SOFC technology, [58]. SOFCs, as all other fuel cells too, consist
of two porous electrodes which are separated by an ion-conducting
electrolyte. Figure 2-16 depicts the working principle of an internally re-
forming, anode-supported planar SOFC.
Figure 2-16:
Working principle
of an internally
reforming anode-
supported SOFC,
[69]
Hydrocarbon containing fuels are most challenging for SOFCs as a vari-

ety of reactions take place in the cell simultaneously. The fuel gas dif-
47
fuses through the anode and participates in STR, see section 2.3.2.1,
and shifting reactions (SR), Eq. 14.
The required steam for both reactions is generated in the electrochemi-
cal oxidation of hydrogen at the anode triple phase boundary (TPB)
where the fuel gas meets electrode and electrolyte material simultane-
ously. The hydrogen produced in the reforming reactions further diffuses
to the anode TPB, while the major part of the formed carbon monoxide
is converted to hydrogen and carbon dioxide via exothermal shifting re-
actions. Some of the carbon monoxide may be directly converted to car-
bon dioxide at the anode TPB. The heat of the shifting and of electro-
chemical reactions is partially consumed by the reforming reactions. At
the TPB of the anode, hydrogen and carbon monoxide are oxidized by
oxygen ions, at the same time releasing electrons:
H 2 + O 2− → H 2O + 2e − Eq. 27
CO + O 2− → CO2 + 2e − Eq. 28
The free electrons flow through an external circuit to the cathode where
oxygen is converted to oxygen ions at the TPB of the cathode consum-
ing the free electrons:
1
O2 + 2e − → O 2− Eq. 29
2
The oxygen ions pass the electrolyte via vacancies in the crystal struc-
ture, hence closing the electrical circuit. In sum, the electrochemical re-
actions can be written as:
1
H2 + O2 → H 2O Eq. 30
2
1
CO + O2 → CO2 Eq. 31
2
The amount of generated electrical power depends on the current of

free electrons and the voltage of the cell. Figure 2-17 shows a sche-
matic voltage-current curve and the voltage loss shares. The Nernst po-
tential (reversible potential) is the maximum possible voltage at non-
standard conditions. The obtained voltage is lower than the Nernst volt-
age when a current flows. The Nernst voltage is therefore commonly re-
ferred to as “open circuit voltage” (OCV). The relation between voltage
losses, operation temperature, current density and species concentra-
tions constitutes the fuel cell characteristics.
48
Figure 2-17: 1.4

Qualitative de- Standard potential
pendency of volt- 1.2
age losses and
current Nernst potential
1.0 Activation or reaction rate losses
Voltage [V/cell]
0.8 Ohmic or resistance losses
0.6 Operation voltage
0.4
Concentration or gas transport losses

0.2
0.0
Current denisity [mA/cm2]
The voltage losses can be assigned to three mechanisms, Figure 2-17:

Ohmic losses
Electronic (electrodes and interconnect) and ionic (electrolyte) cur-
rents are accompanied by voltage losses. Ohmic losses are propor-
tional to, [70]:
Temperature
Geometry of the current paths
Current density
The magnitude of the ohmic losses depends on the resistivity of the

employed materials and the interfacial resistances between electro-
lyte, electrodes and interconnects. In SOFCs, the ionic resistivity of
the electrolyte is up to 500000 times higher than the electronic resis-
tivity of the electrodes, [70]. This is why electrode supported cells are
favored by most SOFC developers, see also section 2.4.5.
Activation losses
The electrochemical reactions (Eq. 27, Eq. 28, Eq. 29) are always
taking place at the electrodes. At zero current, the reactions occur at
equal rates in both directions. The exchange current represents the
current flowing in one of the two directions. When a current flows, the
voltage losses are usually higher than the ohmic losses alone. The
additional losses are referred to as activation polarization. The activa-
tion polarization depends on, [70]:
49
Temperature
Current density and hence the electron concentration in the

electrodes as well as the oxygen ion mobility in the electrolyte
Product (H2O, CO2) and educt (H2, CO) partial pressures
Size of the TPB
Figure 2-17 shows that the activation polarization increases fast with
rising small current density values and then stabilizes on a certain
level.
Concentration losses
The electrochemical reactions (Eq. 27, Eq. 28, Eq. 29) take place at
the TPB. Due to slow diffusion processes, the product partial pres-
sures are higher and the reactant partial pressures are lower at the
TPB compared to the gas flow channels. Consequently, using the
TPB instead of the gas flow channel partial pressures yields a lower
Nernst voltage. The difference between the TPB and the gas flow
channel Nernst voltage is referred to as concentration loss. Concen-
tration losses become important at high current densities, where they
increase against an asymptotic maximum current, see Figure 2-17. A
further increase of the current is impossible because at this point the
concentrations of the reactants at the TPB are equal to zero due to
instant consumption.
2.4.4 Cell designs

Different cell designs have been developed over the years. Currently
two base designs and combinations thereof can be differentiated, the
first being the tubular and the second being the planar cell design. Each
has different characteristics.
The tubular cell design, developed by the Siemens AG, features the
largest presently achievable active cell area. This is due to the applied
extrusion production process, which allows the production of long cell
tubes. The cell components are deposited as thin layers on a ceramic
cathode tube which is closed at one end, see Figure 2-18. Therefore, no
sealing is needed to separate the fuel and the oxidant flow paths in con-
trast to planar cell designs, [51]. The oxidant (air or oxygen) is intro-
duced via a coaxial ceramic air delivery tube (ADT) near the closed
cathode tube end. It then flows through the space between the ADT and
50
the cathode tube. The fuel flows at the outside of the cell. The depleted
air exits the cell at the open end, where it is combusted with the partially
depleted fuel, 50 to 90 % respectively.
Figure 2-18: Interconnect

Technical outline of
the SOFC tubular Fuel
*)
and flattened tubu-
lar design
Oxidant
*) [71]
Fuel
Interconnect
Cathode
Oxidant
Electrolyte
Anode
The tubular cell design is at present the most reliable and robust design,
[72]. However, the tubular cell design features only low power density,
[73]. This is due to high ohmic in-plane resistance resulting from long
current paths, [51], and high concentration polarization losses resulting
from long diffusion paths for the oxygen from the bulk gas phase to the
cathode TPB. The employed coating techniques, e.g. electrochemical
vapor deposition (EVD), carry a high cost, [51], leading to high overall
cell costs. Finally, thermal stress can be a concern, [73].
The flattened tubular cell design, commonly referred to as HPD design
(high power density), was also developed by the Siemens AG, see
Figure 2-18. It allows for about 30 to 40 % higher volumetric power den-
sities compared to standard tubular cells, [74], due to better stacking
possibilities, [59]. The seal-less design, similar to the standard tubular
design, avoids gas cross-over problems, [74]. Internal ribs were intro-
duced to reduce the high ohmic resistance of the standard tubular de-
sign by shortening the current paths, [75].
The main difference between planar and tubular cells is that their flow
channels are usually defined by the geometry of the interconnector
plates. Three flow configurations can be distinguished, namely co-flow,
cross-flow and counter-flow, see Figure 2-19.
51
Figure 2-19:
Flow configurations
of planar SOFCs Interconnect
Co-Flow
Cathode
Electrolyte
Anode
Ox
id
an
t
Fu
el
Interconnect
Cross-Flow
Cathode
Electrolyte
Anode
t
an
id
Ox
Fu
el
Interconnect
Counter-Flow
Cathode
Electrolyte
Anode
Ox
ida
nt
The planar cell design features higher area and especially volume spe-
cific power densities than standard tubular designs due to more efficient
current collection, [73], and simple in-series connection of the single
cells to stacks, [68]. The interconnectors may consist of ceramic or me-
tallic materials. There exists a high cost reduction potential of the planar
design as cost efficient production processes for high volume manufac-
turing of the flat cell components are available, [51]. The drawback of
these cost efficient manufacturing techniques is that the maximum cell
size is limited, [51]. The most important issue of the planar design is that
sealings are necessary in order to avoid gas cross-over of oxidant and
fuel gases.
Thermal stress is an important problem for planar cells, which are usu-
ally less robust than tubular cells. Thermal stress can result from multi-
ple sources such as material property mismatches, poorly balanced en-
dothermic and exothermic reactions and transients of operational pa-
rameters, [51].
2.4.5 Support designs

The support is the principal structural component of the anode-
electrolyte-cathode assembly. It has to cover conflicting requirements
52
like mechanical strength, in order to withstand thermal and mechanical

stresses, and e.g. gaseous permeability, [76], or low ohmic resistance.
Figure 2-20 depicts an overview of the four most common support de-
signs, which are drawn to scale to give a better impression of the differ-
ences.
Figure 2-20: Cathode supported Electrolyte supported

Support designs, ca. 50 μm
drawn to scale
ca. 2200 μm ca. 200 μm
ca. 50 μm
Anode supported
ca. 50 μm
ca. 10 μm
ca. 600- 2000 μm
Passive supported
ca. 50 μm
ca. 10 μm
ca. 10 μm
ca. 600- 2000 μm
Cathode
Electrolyte
ca. 40 μm
Anode
ca. 100 μm Passive Support
The cathode-support design is widely used for tubular cells. It features

high mechanical strength and has cost advantages as the cathode tubes
are manufactured via relatively cheap extrusion processes. However,
the gaseous permeability is an issue regarding the thick cathode leading
to high cathode polarization losses.
The electrolyte-support design is state-of-the-art for planar cells. It is ro-
bust, [73], but suffers from high ohmic resistance, [75]. To counter-
balance this disadvantage, electrolyte-supported cells are typically op-
erated at high temperatures around 1000 °C.
53
The anode-support design has gained a lot of interest recently. The ma-
jor advantage is that the thickness of the electrolyte can be heavily re-
duced as the mechanical strength of the cell is provided by the anode.
Consequently, the ohmic losses caused by the electrolyte are heavily
reduced, [77]. The good electrical conductivity of the supporting anode
material allows for low operation temperatures around 700 to 800 °C,
[75]. This, and the matching thermal expansion rate of the usually em-
ployed Ni-based anode materials makes the application of ferritic chro-
mium interconnects possible, again lowering the internal resistance of
the cells, [75]. Finally, the thick porous anode features a large specific
surface and void volume which improves the conditions for Ni-catalyzed
reactions, such as steam reforming of hydrocarbons and water gas shift.
Although the anode-support design has significant advantages, it bears
a relatively low mechanical strength compared to other support designs.
Consequently, thermal stress and vibrations are the most important is-
sues, [76].
The passive-support design allows very thin anode, electrolyte and cath-
ode layers. These functional layers can be deposited via coating tech-
niques onto either metallic or ceramic support structures.
Metallic support structures offer high mechanical strength, Redox stabil-
ity, [75], and the potential for operation temperatures below 600 °C. The
start-up times can therefore be reduced to values suitable for auxiliary
power units (APUs), [76]. Other advantages of metallic support struc-
tures are the reduced need for expensive ceramic materials, [78], and
the applicability of conventional joining techniques, [76]. The high ther-
mal conductivity of metallic materials reduces spatial thermal gradients.
Ceramic support structures provide reasonable mechanical strength at
relatively low costs but struggle with low gas permeability and high con-
centration polarization losses.
2.4.6 Materials
Material and fabrication process improvements play an important role in
reducing the stack costs. For instance, sulfur tolerant anode materials
are presently an important research field because these materials would
largely simplify the required gas cleaning and hence lower the costs of
SOFC systems. Another important research area is in lowering the op-
eration temperatures, enabling the use of cheaper materials for all cell
elements. The working principle of the SOFC leads to the conductivity
requirements shown in Table 2-5, [68].
54
Table 2-5: Type of Ionic

Electronic
Conductivity re- Conductivty (via atom vacancies in crys-
(via electron-gas-cloud)
quirements of the Cell component tal structure)
different cell
components Electrode High Desirable
Electrolyte Very low or none High
Interconnector High Very low or none
Other important requirements for SOFC materials are, [68]:

Mechanical and chemical stability at high temperatures in reducing
and oxidizing atmospheres
Low formation risk of highly resistive phases
Similar thermo-mechanical properties
Low costs and ease of manufacture
Compared to the state-of-the-art materials mix, every new material
represents an entirely new matrix of possible material interactions. Re-
search activities focus on the thermal expansion compatibility and pos-
sible reactions at the material interfaces, such as phase decomposition
via diffusion. This makes the search for and investigation of new materi-
als more complex than the pure optimization of specific material proper-
ties such as e.g. the ionic conductivity of electrolyte materials. The sub-
sequent sections provide an overview of state-of-the-art materials and a
brief insight into the research work aiming at the identification and char-
acterization of new promising materials for cell components.
2.4.6.1 Anode
Apart from the properties mentioned above, SOFC anode materials
should feature, [68]:
High surface to volume ratios. The porosity, tortuosity and average
pore radius play an important role for the size of the TPB and the gas
diffusion. These parameters are to a certain extent determined by the
production processes.
High catalytic activity towards electrochemical fuel oxidation, hydro-
carbon reforming and water gas shift reactions.
Chemical compatibility with electrolyte and interconnect materials.
Mechanical strength in case that the cell is anode supported.
Tolerance towards fuel contaminants.
55
State-of-the-art anodes are made of a ceramic metallic composite (cer-

met), typically containing 50 mass-% nickel oxide (NiO) and 50 mass-%
yttria stabilized zirconia (YSZ). The NiO particles dispersed in the po-
rous YSZ are reduced in-situ to metallic nickel. The porosity is typically
25 to 40 vol.-%, [79]. The TBP area, which determines the catalytic ac-
tivity towards electrochemical reactions, is controlled by the YSZ powder
size, the Ni to YSZ-ratio and the fabrication technique of the cell. Figure
2-21 shows the contact area between a state-of-the-art porous anode
and the gastight electrolyte.
Figure 2-21:
SEM picture of the
contact area be-
tween anode and
electrolyte, [80]
The advantages of Ni-YSZ anodes are:

Nickel acts as electronic conductor and as STR catalyst, [73].
The YSZ matrix assures a more or less equal thermal expansion co-
efficient of the anode and the usually employed YSZ electrolyte, [81].
Nickel is a very good inexpensive electronic conductor showing an
excellent oxidation activity for pre-reformed fuels, [81].
The mixed cermet guarantees a large triple phased boundary (TPB).
However, the disadvantages of Ni-YSZ anodes are:
Nickel is easily poisoned by sulfur, commonly used as odorant in
natural gas, [81].
The very high activity of nickel towards STR can lead to cold spots on
the cell resulting in high spatial thermal gradients eventually leading
to cell failures due to cracking of the AEC assembly, [68].
Although Ni-YSZ anodes show a satisfactory performance, there is still
room for improvement. The research community investigates a variety of
new materials. Nickel gadolinium doped cerium oxide (Ni-CGO) anodes
feature:
56
Low sintering temperatures around 1000 °C, allowing for cheap pro-
duction processes, [76].
Improved resistance against Ni-NiO cycling, [73].
Strongly improved electrochemical oxidation activity as CGO is a
mixed ionic and electronic conductor (MIEC). MIECs enable the oxi-
dation of the fuel gas in the entire catalyst volume, [68].
However, Ni-CGO anodes have the following disadvantages:
The low sintering temperature limits the operating temperature and
hence the applicability of Ni-CGO to the anode-support design, [82].
If other dopants than nickel are used, e.g. Cu, Au, etc., the STR activ-
ity is comparably low, [68].
Lanthanum strontium chromium perovskite (LSCr) anodes offer good
Redox and sulfur tolerance. However, the electrochemical performance
is lower than that of nickel-based anodes, [73].
Titanate anodes have drawn a lot of interest recently but their properties
are hardly known yet. The material is stable in the reducing atmosphere
and resistant against coking and sulfur poisoning. However, titanate an-
odes show only poor electronic conductivity, thus requiring highly con-
ductive current collectors to keep the ohmic losses small, [82].
2.4.6.2 Electrolyte
Electrolyte materials should satisfy the following requirements besides
those mentioned in section 2.4.6:
Chemical compatibility with the electrode materials.
Mechanical strength in case that the cell is electrolyte-supported.
Gas tightness to prevent gas cross-over phenomena causing a strong
decrease of the cell voltage and power output, [68].
State-of-the-art SOFCs employ yttria stabilized zirconia (YSZ) as elec-
trolyte material. YSZ typically contains 8 mol-% of yttria. The advan-
tages of YSZ electrolytes are, [82]:
Acceptable ionic conductivity between 800 and 1000 °C.
Excellent chemical stability in reducing and oxidizing atmospheres.
Almost no electronic conductivity, [68].
YSZ shows only low ionic conductivity below 800 °C and is therefore not
well suitable for low temperature SOFCs. Being a very refractory mate-
rial, YSZ requires higher sintering temperatures than the other cell com-
57
ponent materials. Co-sintering is hardly possible, increasing the produc-

tion costs, [82].
Figure 2-22:
Ionic conductivities
of selected high
temperature con-
ductors, [79]
The most important property of electrolyte materials is the ionic conduc-

tivity. A broad range of materials are summarized in Figure 2-22.
Of all zirconium solid solutions, scandia stabilized zirconia (ScSZ) has
the highest ionic conductivity at temperatures above 750 °C, [68]. It is
mechanically strong and allows operation temperatures below 800 °C,
[73]. However, scandia is a very expensive material (4000$/kg) and
ScSZ tends to be unstable at temperatures below 600 °C, [68].
LSGM (Lanthanum Strontium Magnesium Gallate) exhibits high ionic
conductivity and is compatible with perovskite cathodes, [68]. LSGM is
however difficult to process to a single phase and features only fair
chemical stability in the operating conditions of SOFCs, [68], as the
gallate is likely to evaporate, causing problems in other parts of the
SOFC system, [73]. The material is less mechanically stable and may
react with nickel, hence a barrier layer of e.g. doped ceria may be re-
quired, [82]. Further, gallium oxide is an expensive material (approx.
1000 $/kg).
58
2.4.6.3 Cathode
Apart from the properties mentioned in section 2.4.6, SOFC cathode
materials should feature, [68]:
High porosity to allow unrestricted gas diffusion to the cathode TPB.
Catalytic activity towards electrochemical oxygen ionization.
Chemical stability in oxidizing atmospheres.
State-of-the-art SOFCs use La1-xSrxMnO3-δ, (lanthanum strontium man-
ganite or LSM) as cathode material, where x is typically 0.15 to 0.25.
Porosities are in the range of 25 to 40 %. Composite cathodes with YSZ
to LSM ratios of 1:1 recently appeared and may be widely used in the
near future, [79].
The advantages of LSM cathodes are, [82]:
Good phase stability.
Good thermal conductivity match with YSZ electrolytes.
Well characterized in SOFC systems.
However, LSM is an almost pure electronic conductor with low conduc-
tivity. The electrochemically actives sites are therefore limited to the
TPB, leading to a moderate catalytic activity.
Recently investigated mixed ionic-electronic conductors (MIEC) allow
the ionization of oxygen in a larger active volume compared to standard
LSM cathodes, where they are limited to the TPB area. This strongly re-
duces the cathode polarization losses, [82].
Most of the recently investigated cathode materials are mixtures of lan-
thanum and strontium with iron and/or cobalt (LSCF, LSF and LSC). The
main concerns with these materials are low phase stability, yet unknown
interface reactions and poor mechanical properties due to a strong mis-
match of their thermal expansion coefficient with most electrolyte mate-
rials, [73]. For example, LSC cathodes were found to form resistive lay-
ers when combined with YSZ electrolytes. Hence, protective layers or
other electrolyte materials have to be employed with LSC cathodes, [68].
Samarium-based perovskite cobaltites (e.g. Sm5Sr5CoO3-x) are exten-
sively investigated in Japan as they are amongst the best MIE conduc-
tors and feature smaller thermal expansion coefficients than LSCF, thus
fixing the compatibility issue with standard electrolyte materials, [82].
59
2.4.6.4 Interconnect
Interconnects, also named bipolar plates in planar stack systems, estab-
lish the mechanical and electrical connection between the anode of one
cell and the cathode of the next cell in SOFC stacks. There are different
interconnect concepts under development which are suitable for differ-
ent stack designs. Metallic, conducting ceramic and non-conducting ce-
ramic interconnects with conducting vias can be distinguished.
Metallic interconnects usually consist of ferritic chromium. The chro-
mium (up to 17 %, [76]) is used to improve the high-temperature oxida-
tion resistance and to realize a sufficient electronic conductivity, [73].
Metallic interconnects feature:
High electronic conductivity at low temperatures, [75].
Production wise almost unlimited size, [75].
Good thermal conductivity, thus reducing temperature gradients, [75].
High mechanical integrity, [83].
Lower production cost than ceramic interconnects, [83].
However, metallic interconnect exhibit a strong mismatch of the thermal
expansion coefficient with all other cell components, [75]. The chromium
can cause evaporation and deposition induced issues leading to severe
cell degradation, [73]. Protective coatings are therefore mandatory, [84].
Conductive ceramic interconnects are usually made of doped lanthanum
chromite perovskites. The specific dopant (typically, Sr, Ca or Mg) and
concentration allow matching the thermal expansion coefficients of the
cathode material and the other materials used for the remaining fuel cell
components. Conducting ceramic interconnects feature negligible corro-
sion and low degradation. However, there are negative aspects to ce-
ramic interconnects, [75]:
The size is limited by available manufacturing processes.
High sintering temperatures are needed.
The thermal expansion coefficient is different in oxidizing and reduc-
ing atmospheres, which may cause thermal stress issues.
The electrical and thermal conductivity is poor inducing thermal stress.
The ceramic materials are difficult to process because chromium
evaporation at high temperatures can lead to poor densification, [79].
A less common approach is the use of non-conducting ceramic inter-
connects with e.g. silver vias, [85]. This approach is well established in
60
the microelectronics packing industry and combines the advantages of

ceramic and metallic interconnects. However, the co-sintering of the vias
and e.g. YSZ is an elaborate manufacturing process and a good match
of the thermal expansion coefficients of the ceramic and the vias materi-
als is difficult to achieve.
2.4.6.5 Sealant
In addition to anode, electrolyte, cathode and interconnects, SOFCs re-
quire sealant materials to isolate anode and cathode chambers from
each other in planar configurations. Sealants are typically made of
glasses or glass-ceramics, [79].
2.4.7 Cell failure and performance degradation

Long life time with low cell degradation and reliable operation are the
most important prerequisites for a successful market introduction of the
SOFC technology. Cell degradation manifests itself as a constant or
sometimes sudden increase of the cell resistance resulting in a lower
operational voltage or current depending on the operational mode of the
cell. The cell degradation rate is expressed as the percentage of voltage
lost during a certain period of time, where a constant current was drawn
from the cell.
Most cell degradation is caused by the high operational temperature of
SOFCs. It may also be induced by the composition of the fuel or oxidant
gases. The detailed understanding of degradation mechanisms and the
origin of their driving forces is the starting point for technology improve-
ments. The comprehension of cell degradation is therefore currently the
main scientific challenge in the SOFC materials and production proc-
esses research communities. Operational strategies to compensate for
losses due to cell degradation on a short term basis have been identified
already, [73]. For instance, an increase of the operational temperature
by 15 K is sufficient to balance a performance degradation of 0.25 %.
Further, cell degradation induced performance losses may be compen-
sated by reducing the fuel utilization. The following paragraphs give an
overview of the most important cell degradation mechanisms.
2.4.7.1 State-of-the-art
Cell degradation may vary by cell and stack size, fuel composition, fuel
utilization and current load. A standardized degradation test procedure
is yet to be defined by the research community. Consequently, a com-
61
parison of the degradation behavior of different SOFC types is difficult.

However, general trends concerning the impact of the test conditions
and test sample types can be observed. Button cell degradation tests do
not reflect the impact of aging phenomena of sealants and spatial tem-
perature gradients. The resulting degradation rates are usually lower
than for the according SOFC cells or stacks. Tests conducted with
gases containing hydrocarbons usually yield higher degradation rates
due to higher thermal gradients stressing the cells.
Presently, the average cell or stack degradation determined with mostly
pre-reformed natural gas is around 1 % per 1000 hours (%/1000h) with
minimum values of 0.1 %/1000h and maximum values of 3.2 %/1000h,
[59, 75, 86-95]. At the moment, the most reliable SOFC type is the tubu-
lar seal-less cell produced by the Siemens AG.
2.4.7.2 Changes in the structure of cell components

The performance of SOFCs strongly depends on the micro- and nanos-
tructure of the cell components. Changes of properties on these scales
may occur over long periods of time or be induced by specific events.
The most important structural change mechanisms leading to cell deg-
radation are:
Nickel particle sintering at the anode
State-of-the-art Ni-YSZ anodes, see section 2.4.6.1, predominantly
degrade due to the agglomeration and coarsening of the nickel parti-
cles which are dissolved in an YSZ matrix, Figure 2-23.
Figure 2-23:
SEM pictures of a
Ni-YSZ anode after
1500 h of operation
(a, b) and after 8 h
of operation (c, d);
Width of pictures
equals 100 μm,
[96]
62
The thermodynamic driving force for nickel particle sintering is the

poor wettability between metallic Ni and YSZ, [97]. In the reducing
atmosphere of SOFC anode channels where steam is also present,
nickel sintering takes place particularly fast, [98]. The formation of
isolated large nickel islands in the YSZ matrix has two effects. First,
the reduction of the specific nickel surface and of the attributed size
of the triple phase boundary leads to an increase of the activation po-
larization, [99]. And second, the smaller amount of nickel particles
and therewith of direct Ni-Ni contacts results in a lower electrical con-
ductivity of the anode and increased ohmic polarization, [100].
Nickel depletion of the anode
The high operational temperatures call for nickel depletion of Ni-YSZ
anodes which leads to increasing activation and ohmic polarization,
see above. The nickel is discharged in form of either elemental nickel
in the gas phase or gaseous nickel hydroxide (Ni(OH)2) depending on
the steam content of the anode gas, [96]. Ni(OH)2 is the predominant
volatile species at steam contents higher than 50 vol.-%.
Delamination of the anode from the electrolyte
During regular operation, the nickel particles are dissolved in the YSZ
matrix in reduced form. If oxygen is introduced into the anode chan-
nel by e.g. diffusion processes through sealings or if steam partial
pressures are very high due to insufficient fuel supply during load
changes, the reduced nickel particles may be oxidized to nickel oxide.
The associated large volume change may cause the disintegration of
the cermet structure, [97]. The resulting delamination on the anode
and the electrolyte reduces the size of the triple phase boundary and
causes severe permanent cell degradation, [76].
Demixing of electrolyte phases
State-of-the-art electrolytes consist of yttria-stabilized zirconia. Zirco-
nia (ZrO2) is a polymorphic material, Figure 2-24. During cool down of
zirconia oxide from the molten state, cubic crystals are formed at
temperatures lower than 2680 °C. Below 2370 °C, the cubic crystals
change to the tetragonal form. Further cool down below 1170 °C
yields a third change of the lattice structure to the monoclinic form.
The latter structural change involves a volume expansion of 3 to 5 %
which entails that components made of pure zirconia are prone to
dismember during cool down, [101].
63
Figure 2-24:
Lattice structures
of zirconia, [101]
The addition of yttrium oxide (Y2O3) avoids the transformation of the

crystals to the monoclinic form and hence stabilizes the zirconia. With
yttria concentrations higher than 8 mol-%, the cubic form of zirconia is
stable at room temperature. The exchange of tetravalent zirconium
with trivalent yttrium yields oxygen vacancies in the lattice structure.
These vacancies permit the passage of oxygen ions and constitute
the ionic conductivity of YSZ, [101]. The number of vacancies is half
that of the yttrium ions present in the electrolyte.
Typical SOFC electrolytes contain between 3 to 8 mol-% of yttria and
therefore consist of a cubic zirconia crystal matrix with tetragonal em-
beddings which can stress-induced transform into the monoclinic
form. The monoclinic form offers good mechanical properties and is
therefore desired in electrolyte-supported cells. Degradation of YSZ
electrolytes results from the demixing of the YSZ matrix into an yttria-
rich cubic phase and an yttria-poor tetragonal phase, [102]. The
tetragonal phase hinders the passage of oxygen ions and hence de-
creases the ionic conductivity of the electrolyte leading to increased
ohmic losses.
Ordering of vacancies in the electrolyte structure
As discussed above, a uniform distribution of oxygen vacancies
throughout the whole electrolyte is a prerequisite for high ionic con-
ductivity. Long-term ordering processes of vacancies result in irre-
versible degradation of the electrolyte, [102, 103].
64
Formation of low conductive layers between the cathode and the

electrolyte
The most commonly employed cathode material is lanthanum stron-
tium manganite (LSM), see section 2.4.6.3. LSM cathodes tend to
form resistive products with zirconia such as e.g. La2Zr2O7 which en-
tail severe irreversible cell degradation, [68]. Further, gaseous chro-
mium in the oxidant gas originating from metallic interconnects or up-
stream equipment may lead to the formation of a spinel phase near
the cathode triple phase boundary which deteriorates the electro-
chemical properties of LSM, [104].
Pores between the cathode and the electrolyte
Under load conditions, an oxygen potential difference may be created
across the cathode which leads to a redistribution of the cathode ma-
terials. Pore formation and movement may result, leading to a de-
crease of the active triple phase boundary length at the electrolyte-
cathode interface and consequently to higher cathode activation po-
larization. For a more detailed discussion of this complex and not yet
fully understood degradation mechanism please see [105].
2.4.7.3 Carbon deposition

Carbon deposition is a concern in all catalytic conversion processes of
hydrocarbons, including internally reforming SOFCs. Catalyst deactiva-
tion due to carbon deposition may result from encapsulation of nickel
particles by polymeric carbon, blocking of catalyst pores and deteriora-
tion of the catalyst surface by carbon whiskers, [106], Figure 2-25.
Figure 2-25:
Forms of perma- Blocked Carbon whisker
Encapsulated
catalyst pore with nickel particle
nent deactivation nickel particle
of nickel catalysts
by deposited car- Nickel particle
bon, [40]
Carbon
Depending on their origin, carbon deposits are differentiated in carbon

and coke. Carbon is the product of CO decomposition via the reverse
carbon steam reaction, Eq. 12, and the reverse Boudouard reaction, Eq.
13. Carbon originating from the decomposition of hydrocarbons is sum-
65
marized under the term coke. Methane may contribute to coke deposi-
tion via the reverse hydrogasification reaction, Eq. 15. The decomposi-
tion pathway of higher hydrocarbons is however more complex than that
of carbon monoxide and methane. Hydrocarbons adsorb on the catalytic
surface and split off hydrogen. The remaining carbon structures polym-
erize to form coke, [107]. The coking potential of higher hydrocarbons
increases from paraffins via olefins to acetylenes, [108]. Aromats are
also known for their potential towards coke formation, which is several
times higher than that of carbon monoxide and increases with increasing
molecular weight of the aromatic specie, [54]. More detailed information
about carbon and coke formation can be found in [106, 109, 110].
Once deposited, carbon is difficult to remove at temperatures below
800 °C without the presence of a catalyst due to the slow carbon gasifi-
cation reactions. Depending on the SOFC type, the temperatures may
be significantly higher, which results in reduced carbon deposition risk.
However, buildup of carbon deposits was observed in internally reform-
ing SOFCs, [82], causing reduced STR activity and increased anode ac-
tivation losses. The latter is due to shortened active length of the TPB. A
material-based solution for the coking issues is copper doping of the an-
ode material, [81, 111].
2.4.7.4 Structural stress

The anode-electrolyte-cathode assembly (AEC) of SOFCs consists of
three thin ceramic material layers which are jointed via sintering proc-
esses. Additionally, planar SOFCs have glassy sealants which separate
the anode and cathode gas channels at the gas in- and outlets. Cell fail-
ure and irreversible performance degradation due to gas cross-over is
the result, if these fragile ceramic and glassy structures break or disas-
semble. Structural stress is the most important reason for such malfunc-
tions, whereas its order of magnitude depends on the properties of the
employed material, the operating conditions as well as the design and
stacking concept, [112]. Structural stress may arise from:
Different thermal expansion coefficients of the used materials
Anodes, electrolytes, cathodes and sealings are made of different
materials with different thermal expansion coefficients. Mismatches
result in residual stress during operation as the operational tempera-
tures of SOFCs are lower than the temperatures during manufactur-
ing. These stresses can lead to micro crack formation and delamina-
tion of the ceramic and glassy structures, [113]. Residual stress is
66
therefore one of the most important limiting factors for the scaling up
of planar cells over 100 cm2 active area, [114].
Spatial and temporal thermal gradients
Residual stresses are aggravated by spatial and temporal thermal
gradients. Important spatial thermal gradients result from poorly bal-
anced exothermic water gas shift and electrochemical reactions with
endothermic steam reforming reactions of methane, higher hydrocar-
bons and tars, [51]. Cracking of the AEC and attributed cell failure
was frequently reported as consequence of high spatial thermal gra-
dients, [53, 113, 115]. Spatial thermal gradients may be reduced by
metallic interconnects with high thermal conductivity, appropriate flow
configurations, increased fuel utilization and operational voltages,
[113], as well as higher degree of pre-reforming of the fuel gas in
question. Temporal thermal gradients are most important during heat-
up and start-up phases. The effects of temporal thermal gradients are
similar to those of spatial thermal gradients.
External mechanical loading
SOFC stacks are usually charged with weights to assure correct ori-
entation and good contact between the single cells. The induced me-
chanical stresses may cause stability issues within the AEC, [113].
2.4.7.5 Catalyst poisoning

The activity of the employed catalysts at the anode and cathode may be
reduced or even eliminated by contaminants in the fuel or oxidant gas.
Alkali metals may cause deactivation of the STR activity of nickel and
thus reduce the ability of SOFCs to convert hydrocarbon containing fu-
els. This effect is most pronounced for potassium and less for lithium
and sodium, [13]. The impact of alkali metals towards the electrochemi-
cal activity of nickel has however not yet been studied.
Halides may also reduce the activity of nickel. In [95] however, no im-
pact on cell performance was found for 1 ppmV of HCl in the fuel gas.
Sulfur is a well known poison for STR nickel catalysts, [107]. The type
of sulfur species is usually not important for the poisoning, [116]. Sulfur
chemisorbs on nickel, thus blocking active sites. In SOFCs this leads to
reduced STR activity and shortening of the active length of the TPB.
Hence, activation polarization losses increase very fast when sulfur is in-
troduced to SOFCs. Figure 2-26 shows exemplarily that 1 ppmV of H2S
may result into a very fast 15 % performance drop, [117]. Sulfur poison-
67
ing is often reported to be reversible and less pronounced at higher

temperatures, [10, 118-120]. However, irreversible degradation may
also be caused by sulfur as it easily reacts with nickel, forming e.g.
nickel sulfide (Ni2S3), [96]. Ni2S3 melts at 806 °C compared to 1455 °C
of pure nickel. This may lead to enhanced nickel particle sintering and
therefore irreversible degradation of Ni-YSZ anodes, see section 2.4.7.2,
[96].
Chromium is known to be a poison for the state-of-the-art LSM cath-
odes. It may evaporate from metallic interconnects or up-stream equip-
ment and subsequently condensate at the cathode TPB, [104, 121]. The
condensed chromium inhibits the oxygen reduction and leads to in-
creased cathode polarization losses.
Figure 2-26:
Synergetic degra-
dation effects of
phosphorous, ar-
senic and sulfur,
[117]
Arsenic reacts with standard Ni-YSZ anodes and yields Ni5As2, [117].
However, Figure 2-26 shows that the exposure of Ni-YSZ anodes to 1
ppmV AsH3 for a period of 700 h at 800 °C resulted in minor cell degra-
dation. According to [117], this can be explained with strong nickel arse-
nic interaction at the surface of the Ni-YSZ anode. Therefore the arsenic
did not penetrate into the anode and did not provoke a change of the tri-
ple phase boundary structure.
Phosphorous in form of PH3 was shown in [117] to react with nickel,
forming Ni3P which extensively agglomerates inside the anode layer.
Figure 2-26 shows that the operation over 700 hours at 800 °C with 2
ppm of PH3 resulted in considerable degradation. At room temperature,
PH3 can be removed with copper doped zeolites, [122].
68
69
70
3 Experiments
3 Experiments
The experimental research agenda with respect to the development of a

robust B-IGFC system design at Paul Scherrer Institut (PSI) began in
2000. Initial tests showed that aromatic tars such as benzene, toluene
etc. rather than being mere impurities in the fuel gas can be considered
as possible fuel for solid oxide fuel cells (SOFCs). Encouraged by these
findings, system analysis calculations were carried out in 2002 and 2003,
revealing that the link-up of an updraft gasifier with a SOFC via hot gas
processing would be a promising B-IGFC concept, [123, 124]. In 2004,
the proof-of-feasibility of wood gas operated SOFCs was achieved on
lab scale. Further testing showed that tars are not inert but degraded in
SOFCs to hydrogen and carbon monoxide, thus contributing to the elec-
trochemical reactions and power generation. The characterization of all
important unit operations was completed by mid 2007. For the final
proof-of-concept, an experimental setup based on the PSI B-IGFC con-
cept was put in place, [125]. In late 2007, the setup covering the com-
plete chain from wood to electricity was operated several times up to
28 h non-stop and 65 h in total. The following section outlines the most
important experiments conducted, supervised or analyzed by the author.
3.1 The PSI B-IGFC system

3.1.1 System description
The PSI B-IGFC concept is shown in Figure 3-1. In a first step, the wood
is converted to a highly tar-laden producer gas in an updraft wood gasi-
fier. The cold gas efficiency of this gasification process is above 90 %
due to the relatively low temperature of the raw producer gas of around
75 to 250 °C. At these temperatures, inorganic impurities such as e.g.
alkali metals are in the solid phase condensed on particles. Together
with the particles, the condensed impurities are removed from the pro-
ducer gas via a hot gas cyclone and a particle filter employing filter can-
dles. The particle free producer gas is then fed to a catalytic partial oxi-
dation where the tar and hydrocarbon content is decreased to levels
suitable for SOFCs. After the catalytic partial oxidation, the producer gas
is cooled down from approx. 750 to 400 °C in order to allow for the re-
moval of chlorine and sulfur compounds in sorbent trap beds. This tem-
71
3 Experiments
perature is the maximum possible temperature to avoid volatilization of

the employed sorbent materials, namely sodium carbonate and zinc ox-
ide.
Figure 3-1: Wood Air Air 400°C

PSI B-IGFC sys-
tem outline Updraft 150°C 150°C Particle 150°C 750°C 400°C HCl, H2S
Cyclone CPO HX1
gasifier filter removal
Char/Ash Char/Ash
700°C
To stack
HX5 HX4 HX3 HX2
Air
900°C 900°C
HEAT
200°C
POWER C SOFC A
1000°C
1000°C 1000°C
1200°C
Combustor
The producer gas is then heated up again to approx. 700 °C in the gas-
to-gas heat exchanger HX1 and superheated to 900 °C in HX2 before
entering the anode channels of the SOFC. The depleted producer gas is
mixed with the depleted cooling air of the SOFC and completely burned
in a catalytic combustor. The hot flue gas is used to superheat the pro-
ducer gas in HX2 and the cooling air of the SOFC in HX3. Heat at a
temperature level of 200 °C is recuperated in HX4. The remainder of the
flue gas heat is used to pre-heat the SOFC cooling air in HX5 before it is
sent to the stack. The SOFC produces direct current at a given cell volt-
age which is converted to AC power in an inverter.
3.1.2 Discussion of the chosen approach

The starting point for the development of the PSI B-IGFC system was an
analysis of potential threats for SOFCs arising from the composition and
the impurity load of biomass gasification derived fuel gases. The con-
sidered threats were carbon deposition, temperature gradients and sul-
fur poisoning.
Evaluation of biomass gasification processes
Nickel is usually employed as catalyst material in SOFC anodes. At the
typical SOFC operating temperatures of around 600 to 1000 °C, nickel
catalysts are prone to coking and therewith induced deactivation. The
mechanisms of carbon deposition on and removal from nickel catalysts
72
3 Experiments
is the subject of several other investigations, [106, 109, 110, 126]. Car-
bon deposition can either result from CO dissociation following the exo-
thermic Boudouard reaction or from the dehydrogenation of hydrocar-
bons and of aromatics in particular, [127].
The probability of carbon deposition can be roughly estimated with equi-
librium calculations assuming carbon depositions as graphite. This ap-
proach was chosen by several other research teams for similar investi-
gations, [128-131]. Figure 3-2 shows the ternary diagram of the carbon-
hydrogen-oxygen system. The inscribed isotherms divide the diagram
into a region where graphite formation can be expected and one in
which it is unlikely to take place. It can be seen that the probability of
graphite formation decreases with increasing temperatures. Air or oxy-
gen gasification of dry wood shifts the point representing the corre-
sponding producer gas towards the oxygen corner. Increasing wood
humidity and steam gasification shifts it towards the H2O point in the
diagram.
Figure 3-2:
C-H-O ternary dia-
gram for the ther-
modynamic graphite C
0.0
formation at equilib- 1.0
rium conditions with T=1000°C
inscribed fuel gases T= 300°C
as received from 0.2
0.8
different biomass
gasifier types
0.4 Graphite 0.6
formation
CO
region
0.6
0.4
Dry wood
CO2
CH4 Downdraft
0.8
Updraft 0.2
Fluidized bed
1.0
0.0
0.0 0.2 0.4 0.6 0.8 1.0
H H2O O
For fixed bed downdraft gasification, the humidity of the wood is limited
to less than 12 %. The cold gas efficiency of this process is typically
around 80 to 85 %. The corresponding producer gas is close to the con-
73
3 Experiments
nection line between the dry wood point and the oxygen corner of the
diagram as air is employed as gasification agent.
The updraft gasification process also employs air as gasification agent,
but allows for the conversion of wood with humidity up to 50 %. Conse-
quently, the corresponding point in the diagram is close to the connec-
tion line between the dry wood point and the water point. The cold gas
efficiency of this process can be up to 95 %.
For the fluidized bed steam gasification process, the feed humidity
should be kept below 10 % in order to reach satisfying conversion effi-
ciencies in the range of 70 %, [33]. Using steam as gasification agent,
the corresponding producer gas point lies directly on the connection line
between the dry wood point and the water point of the diagram.
Figure 3-2 shows that all raw fuel gases are in the graphite formation
free region at 1000 °C. However at 300 °C, which is considered the low-
est temperature at which carbon formation can occur at a considerable
rate, only the updraft and the fluidized bed producer gases are close to
the graphite formation free zone. The required water addition to prevent
carbon deposition, the low content of hydrocarbons and the conversion
efficiency of 80 % were clear factors against the application of the down-
draft gasification in the PSI B-IGFC system.
The tar species found in producer gases from updraft wood gasifiers are
mainly oxygenates, [29], which decompose in the gas phase at tempera-
tures of 500 °C or higher without forming significant amounts of coke,
[55]. In contrast, the tar load of producer gases originating from fluidized
bed steam gasification typically consists to a major extent of aromatic
species which are prone to coking as discussed above. For this reason
and due to the higher conversion efficiency, the updraft gasification
process was identified as best choice for the application in the PSI B-
IGFC system.
Evaluation of gas processing options
In a next step, the required gas processing was defined based on the
fuel gas requirements imposed by SOFCs and the producer gas proper-
ties of the chosen updraft fixed bed gasification process. Fuel gases for
SOFCs have to be particle free in order to avoid physical blocking of the
gas channels. Another issue are impurities such as e.g. alkali salts. Due
to migration processes of these impurities into the electrode materials,
they can cause serious damage to fuel cells via micro structural
changes of the electrode structure, [13]. The particle load of producer
74
3 Experiments
gases from updraft gasification is usually low and at the typical producer
gas temperatures of updraft gasifiers around 75 to 200 °C, alkali salts
can be assumed to be condensed on particles to a major extent. There-
fore it was concluded, that the particle removal via a cyclone and a par-
ticle filter should resolve both issues.
State-of-the-art gas processing is focused on the application of internal
combustion engines which require fuel gas temperatures below the dew
point of tars. In order to prevent blocking of valves etc. tars are almost
completely removed. SOFCs, in contrast to internal combustion engines,
require high fuel gas temperatures and can, to some extent, use tars as
fuel. The reforming of oxygenated tars, which are mainly present in the
producer gas from the updraft gasification process, was demonstrated in
e.g. [55]. Applying cold gas processing for the complete removal of tars
in a B-IGFC system therefore entails a loss of both fuel and sensible
heat, which ultimately results in lower system efficiencies, [132]. Conse-
quently, hot gas processing at temperatures above the dew point of tars
was considered for the PSI B-IGFC system. This consideration took
place despite the handling of tars being more delicate than their removal,
as especially the heat-up of tar-laden gases can cause serious coking
problems, [133].
Cell internal reforming of hydrocarbons and tars is a possibility for the
chemical cooling of SOFCs, which reduces the cooling air mass flow re-
quired to maintain the operational temperature. Further, internal reform-
ing increases the electrical efficiency of SOFCs as the heat of the elec-
trochemical reactions is used cell internally to increase the heating value
of the fuel gas. However, hydrocarbon and tar reforming reactions are
relatively fast in the presence of nickel at the typical operating tempera-
tures of SOFCs. High hydrocarbon and tar contents can therefore lead
to thermal stress due to cooling of the cell inlet region as a consequence
of the endothermic character of reforming reactions. It was therefore
concluded that the gas processing must allow for a precise partial con-
version of hydrocarbons and tars to yield the best compromise between
thermal stress problems and efficiency benefits.
A further issue which has to be addressed by the gas processing is sul-
fur. Sulfur is known for its tendency to form strong bonds with nickel,
[134-136], and thereby blocking active sites making these unavailable
for reactant molecules. For the poisoning, the type of the sulfur species
is largely irrelevant, [116, 118]. However for the removal of sulfur, the
75
3 Experiments
type of species is of major importance and must be considered in the

definition of the gas processing.
Catalytic partial oxidation (CPO) was identified to possibly cover all the
above requirements. CPO catalysts are usually ceramic monoliths
coated with noble metals, [63]. The average temperature of the process
is below 800 °C, [62]. A characteristic feature of the CPO is that two dif-
ferent reaction zones can be distinguished within the monolith during
operation. Near the monolith inlet, a hotspot with temperatures of approx.
1000 °C indicates that the supplied oxygen is consumed by the partial
combustion of the fuel. In the downstream parts of the monolith, the heat
and the steam originating for the combustion reactions are consumed by
steam reforming reactions. CPO catalysts may be to a large extent sul-
fur tolerant. Furthermore, the high hotspot temperatures are beneficial
for an almost complete conversion of organic sulfur compounds to hy-
drogen sulfide. The hydrogen sulfide can then be removed in high tem-
perature zinc oxide trap beds. The rate of conversion of hydrocarbons
and tars may be controlled through the geometry and the operating con-
ditions of the CPO monoliths.
3.2 Gas analysis

In all experiments, Varian micro gas chromatographs (μ-GCs) were em-
ployed to monitor the dry and clean composition of gas mixtures online.
Figure 3-3:
Outline of used
gas sampling sys-
tem
The columns employed for the separation of different gaseous species

allow the measurement of the following permanent gas species: hydro-
gen, helium, carbon monoxide, carbon dioxide, methane, ethene, pro-
76
3 Experiments
pane, nitrogen and oxygen. All other species have to be removed from
the analyzed gas prior to the injection to the μ-GC. The employed gas
sampling system is based on the design proposed in [137] and has been
further developed at PSI, Figure 3-3.
In a first step, the raw gas is quenched with a solvent. Condensable
species, such as tars, water and other impurities, are thus dissolved.
The solution is cooled to -25 °C to assure that only traces of condens-
able species remain gaseous. After a gas-liquid separation, the dry and
clean gas is pumped to the μ-GC while the used solvent is either reused
for another quench cycle, as depicted in Figure 3-3, or bottled for later
analysis. 1-methoxy-2-propanol is used as solvent due to its good solu-
bility for both organic carboneous species and water.
At times the employed solvent was analyzed in order to determine the
amount and nature of impurities in the raw gas and its water load. The
employed methods were:
Gas chromatography coupled with mass spectrometry (GC/MS) for
the identification of aromatic species such as benzene, toluene etc.
and quantification of their concentration in the raw gas.
Gravimetric method for the determination of the concentration of hy-
drocarbons with dew point above approx. 140 °C in the raw gas.
Following the gravimetric method, a given amount of used solvent is
placed in a vacuum oven at 40 °C and 30 mbara until the weight loss
of the specimen due to evaporation of volatile compounds subsides.
Based on the remaining mass of the sample, which is attributed to the
mass of higher hydrocarbons, the corresponding concentration in the
raw gas is computed.
Karl-Fischer titration method for the analysis of the water concentra-
tion in the raw gas.
Given that offline analysis is time-consuming, mass balance calculations
were performed to estimate the tar and water concentrations of the raw
gas based on the measured dry and clean gas compositions. The gen-
eral assumptions were:
Carbon (C), hydrogen (H), oxygen (O) and nitrogen (N) are the only
considered elements.
There are no uncontrolled leakages in the balanced system.
Nitrogen appears as molecular nitrogen only.
77
3 Experiments
Most of the experiments presented in the subsequent sections involved

a lab scale wood gasifier as gas source. Following assumptions were
made to account for the heterogeneous character of wood and the com-
plexity of the gasification process:
Dry wood contains 50 mass-% C, 44 mass-% O and 6 mass-% H.
An elemental analysis of the used wood pellets was commissioned to
two independent laboratories. The analyzed samples consisted of
49.85 to 50.09 mass-% carbon, 43.41 to 43.65 mass-% oxygen, 5.93
to 6.03 mass-% hydrogen and 0.15 to 0.2 mass-% nitrogen. The sul-
fur mass fraction was found to be about 0.01 to 0.03 mass-%.
The average humidity of the gasified wood is 6 %.
Air consists of 79 vol.-% nitrogen and 21 vol.-% oxygen.
All carbon introduced to the gasifier is converted to gaseous or liquid
species. The formation of char or soot is neglected.
All nitrogen in the producer gas comes from the gasification agent.
Consequently, the total molar flow of the dry and clean producer gas
equals the gasification air molar flow times the quotient of the nitro-
gen fraction of air and raw gas. The molar flows of atomic carbon,
oxygen and hydrogen bound in permanent gases are computed using
the measured molar fractions and according stoiciometric coefficients.
All carbon, which is not bound in permanent gases, is bound in "tars".
"Tar" is approximated by the empirical formula CH1.34O0.81.
The investigated lab scale gasifier is based on the countercurrent
fixed bed process. In this process, the "tars" mainly originate from py-
rolysis sub-processes, which take place close to the gas outlet and
occur in reducing atmospheres. According to [57], the composition of
"tars" originating from fast pyrolysis processes can be approximated
as mixture of 67 % acetic acid, 16.5 % m-cresol, 16.5 % syringol
which translates in the above empirical formula.
All hydrogen and oxygen, which is neither bound in the measured dry
and clean gas species nor bound in "tars", is assumed to be bound in
water molecules.
Where the water load is measured via the Karl-Fischer coulombmetry
method, the empirical formula for "tars" was determined to satisfy the
mass balance. The resulting empirical formula can be used as indica-
tor for the tar signature of the investigated producer gas.
78
3 Experiments
All incoming gaseous mass streams where directly monitored using

mass flow controllers, rotameters or gas meters. Liquid and solid mass
flows where indirectly determined by time averaging the measured ab-
solute mass difference for a given time interval.
3.3 Proof-of-feasibility tests

3.3.1 Lab scale tests under load conditions
The feasibility of the PSI B-IGFC concept was shown by linking-up a
laboratory scale SOFC with an updraft gasifier in 2004, Figure 3-4.
Figure 3-4:
Technical outline
and photograph of
the experimental
setup used for
proof-of-feasibility
tests in 2004
The gasifier had a maximum feed of 2.7 kg/h of wood pellets. For the
experiments, the mass flow was reduced to 1.3 kg/h or 6 kW of thermal
input, respectively. The non-stop operation period was limited to 16
hours. Typical gasifier air-to-fuel ratios were between 0.2 and 0.4. A pla-
nar lab scale stack provided by the Hexis AG (formerly Sulzer Hexis)
was used for the experiments. The SOFC stack consisted of five single
cells with an active area per cell of 60 cm2. The stack was designed for
a nominal electric power output of 60 W when operated with reformate
gas originating from the catalytic partial oxidation of natural gas from the
Swiss grid. The stack was placed in an oven which kept the operational
temperature of the stack at approx. 950 °C.
Due to the stack and system design aiming at natural gas applications,
means of heating had to be introduced to the normally non-heated feed
gas inlet area. This area as well as the pipe system connecting the gasi-
fier and the SOFC system were heated to 400 °C in order to prevent
condensation of tars and thereby blocking. Further, a particle filter was
79
3 Experiments
employed to remove particles which could also cause blocking. No other

gas processing steps were undertaken.
The particle free, humid and tar laden producer gas was fed to the fuel
cell through a jet pump, the reason being that the gasifier had a pro-
ducer gas output of approx. 5 kW (based on the lower heating value -
LHV) and only a few percent of the producer gas flow rate were required
for the operation of the 60W-SOFC. The specifically developed jet pump
delivered approx. four standard liters of fuel gas per minute with the re-
quired overpressure of 20 mbar to the stack. The propellant gas compo-
sition was adjusted to match the main components of the tar-free pro-
ducer gas. The mixing ratio was on average one part of producer gas
per part of propellant gas. Due to the particle filter, the feeding system
could be operated for several hours at a time. The fuel gas volume flow
was monitored with a heated orifice. For more information see [125].
Figure 3-5: 1000 20

55 h operation of the Hy SPG MPG
900 18
5-cell SOFC stack
with wood gas under 800 16
isothermal conditions
700 14
in an oven
Cell voltages [mV]
Stack current [A]

600 12
"Hy": Hydrogen 500 Voltage 10
"SPG": Synthetic gas
"MPG": Mixed syn- 400 8
thetic and real pro- 300 Current 6

ducer gas
200 4
100 2
0 0
0 6 12 18 24 30 36 42 48 54 60 66 72 78
Runtime [h]
The voltage progression of the longest experimental run, shown in

Figure 3-5, was obtained from the cell in the middle of the stack. In total,
the SOFC stack was operated for over 100 h with diluted producer gas
corresponding to an equivalent pure producer gas volume of 15 stan-
dard cubic meters on dry and tar free basis (dtf). The average tar load
was 85 g/mn3 (dtf). Nonstop experiments of up to 16 h were performed.
The producer gas operation of the fuel cell caused a continuous in-
crease of the pressure loss of the fuel cell. Variations of the electro-
chemical performance parameters of the fuel cell showed no effect on
the pressure loss. However by switching to hydrogen as fuel gas for
80
3 Experiments
short periods of time, the pressure loss could be reset to normal levels.
The voltage-current-curve remained almost constant during operation
with real producer gas as well as with synthetic producer gas. The ob-
served minor cell deactivation after the experiments was on normal lev-
els compared to that of natural gas operation of the same duration. The
tar species typically present in updraft gasification producer gases, such
as e.g. phenol and acetic acid, were not found to have a negative impact
on the performance of SOFCs. Unlike most relevant literature, this result
once more supported the hypothesis that SOFCs are, to some degree,
tolerant towards tars.
3.3.2 Lab scale tests under open circuit conditions

The fate of tar species in SOFCs was investigated with the experimental
setup shown in Figure 3-6.
Figure 3-6: Water

Heated Transfer Pipe
Technical outline T,V Vaporizer
p
Furnace
of the experimen- H2 MFC T
tal setup used for SOFC

stack tests in Tar stack
2004 Vaporizer Tar Pump
T,V
N2 MFC Water Pump
H2O Balances Tar
Humidified hydrogen with a tar load of approx. 240 g/mn3 (dtf) was fed
via a heated transfer pipe to a SOFC lab scale stack. The stack was not
electrically connected to a power unit. This was done in order to reduce
the costs of the experiment in case the stack might get destroyed due to
the high tar load. The stack was mounted in a tinder resistant INCONEL
housing placed in an oven heated to 900 °C. The hydrogen mass flow
was controlled via a mass flow controller (MFC). Further, a nitrogen line
was installed for inertization purposes in emergency cases and for the
conservation of deposits in the SOFC for post-test analysis. The hydro-
gen volume flow was monitored manually using a gas meter prior to en-
tering the water vaporizer. The added steam mass flow was set via the
flow rate of the water pump and monitored manually by means of a bal-
ance. A similar approach was used to add a mixture of tar species rep-
resenting the typical composition of tars found in producer gases from
fluidized bed gasifiers, [138]. This mixture was chosen as it was consid-
81
3 Experiments
ered more difficult to degrade than the mixture of oxygenated tars typi-
cally found in updraft producer gases. The mixture consisted of benzene
(63.0 mass-%), toluene (9.5 mass-%) and naphthalene (27.5 mass-%).
The composition of the gas after passing the SOFC was measured
through gas chromatography using the above described sampling sys-
tem. The gas volume was monitored with a gas meter.
After the experiments, mass balance calculations were conducted yield-
ing tar conversions between 92 and almost 99 % for different steam-to-
carbon ratios, see Table 3-1. In comparison, the tar conversion found for
the empty INCONEL housing was in the order of 10 %. Therefore, the
observed tar conversions were attributed to a major extent to the cata-
lytic activity of the investigated nickel anode catalyst.
Table 3-1: C H O C H O H2 CO CO2 CH4 Tar con-

Tar conversion Exp. version
observed on Input [mol/h] Output [mol/h] Output [mol-%] [% input]
Nickel-cermet
1 0.94 12.59 3.33 0.89 8.07 1.14 81.87 12.75 5.27 0.11 94.5
2 0.88 12.41 3.23 0.81 7.56 1.04 82.22 12.67 4.97 0.14 93.1
3 0.97 9.06 1.59 0.96 7.78 1.06 80.13 17.44 2.29 0.14 98.9
4 0.95 7.64 0.86 0.88 7.25 0.86 80.14 18.91 0.19 0.76 92.5
Figure 3-7 compares the measured gas composition at the stack outlet
and the equilibrium composition calculated with ASPEN PLUS.
Figure 3-7: Hydrogen

100 volume flow: 55.6 nl/h 40
Measured anode Tar mixture mass flow: 13.0 g/h Gas phase only
1
off-gas composi- Gas and solid phase
2
tion vs. equilib-
85 4 30
rium calculation
CO,CO2, CH4 & C(S) [mol-%]
H2 3
H2 [mol-%]
H2O (balanced)
70 20
CO
55 10
CH4 CO2
40 0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
Steam-to-carbon ratio [-]
The equilibrium calculations were conducted for the set oven tempera-
ture and the input gas composition considering H2, CO, CO2, CH4, C2H2,
82
3 Experiments
C2H4, C2H6, C3H6, C6H6, C7H8, C10H8, O2, water and graphite as possible
products. The full lines in Figure 3-7 represent the equilibrium molar
fractions of the most important species in the gas phase for varying
steam-to-carbon ratios. The dotted lines stand for the equilibrium molar
fractions considering solid graphite, which is formed for steam-to-carbon
ratios below one at 900 °C. It can be seen that the tars in the input gas
were degraded to hydrogen and carbon monoxide and that the refor-
mate gas almost reached equilibrium. The assumption for SOFC models
that the water gas shift, [139-141], and the steam reforming reaction,
[142-144], are at equilibrium was therewith confirmed.
In order to provoke carbon depositions the steam-to-carbon ratio of the
last experimental point was chosen below one. In fact, the deposits
caused an increasing pressure loss. For a closer analysis, the deposits
were conserved by flushing the stack with nitrogen during cool down.
This allowed not only the verification of the formation of carbon deposi-
tions but also their location was determined. Figure 3-8 shows the top
view (A) of an interconnector plate with carbon deposition at the gas
inlet. View (B) shows that the carbon depositions grew from the inter-
connector plate in the central gas channel, see view (C). This indicates
that carbon depositions are self-catalyzing. An explanation for the loca-
tion of the carbon depositions at the cell inlet is the local temperature,
which must have been considerably lower than the average cell tem-
perature due to endothermal steam reforming reactions.
Figure 3-8:
Carbon depositions
in anode channel
inlet region and
central fuel gas
channel
Overall 0.5 g of deposits were found. Assuming the deposits to be

100 % carbon, this amount corresponds to a deposition rate of 0.07
mol/h, Table 3-1. The formation rate of graphite at equilibrium for the
given feed gas composition was calculated to 0.16 mol/h. It was there-
fore concluded that the carbon deposit formation is kinetically controlled.
83
3 Experiments
In summary, the experiments have shown that the anode catalyst of the
investigated SOFC catalyses the degradation of aromatic tars to hydro-
gen and carbon monoxide. Oxygenated tars should be degraded even
more easily. Water gas shift and steam reforming reactions were found
to be very close to equilibrium while the formation of carbon deposits
was found to be locally limited to the cell inlet and kinetically controlled.
3.4 Characterization of important unit operations

The proof-of-feasibility tests confirmed the main assumptions forming
the basis for the PSI B-IGFC concept. It was shown, that
producer gas originating from the updraft gasification of wood can be
electrochemically converted in a SOFC.
tars are degraded to hydrogen and carbon monoxide and contribute
to the electrochemical reactions and power generation in SOFCs.
The construction of a lab-scale demonstration unit and the achievement
of the proof of concept were defined as next steps. Prior to the assembly
of the PSI B-IGFC system demonstration unit, each of the important unit
operations was characterized experimentally in stand-alone operations.
This way, the optimal operating conditions of the single unit operations
and possible interactions in the overall system were identified.
3.4.1 Updraft gasification reactor

During the proof-of-feasibility tests over 100 hours of SOFC wood gas
operation were accumulated, see section 3.3.1. The non-stop duration
of the experiments was limited to around 16 h due to gasifier problems
such as strong increase of pressure or over-temperatures. The gasifier
had to be regenerated via burnout cycles, thus removing deposits in the
reactor and the piping downstream. During these regeneration cycles,
the SOFC was fed with humidified hydrogen. As it could not be excluded
that the SOFC regenerated as well while being fed with humidified hy-
drogen, drawing conclusions for long-term deactivation processes due
to producer gas operation was not possible. Therefore, the following re-
quirements were defined for the gasifier of the demonstration unit:
Stable non-stop operation for 200 h or more.
Adjustable tar load of the raw gas between 150 g/mn3 (dtf) and approx.
80 g/mn3 (dtf) via gasifier-internal degradation of tars.
84
3 Experiments
Thermal input up to 25 kW in order to allow for the operation of 10 kW

SOFC stacks.
Based on the design of the gasifier used in the first proof-of-feasibility
tests, a new gasifier capable of pressurized operation was built. The
gasifier is fed from a fully automated wood pellet hopper, which can be
operated at pressures up to 3 bara. The hopper has a capacity of 60 kg
of commercial wood pellets. The maximum feed to the gasification reac-
tor is approx. 5 kg of pellets per hour which satisfies the requirement of
25 kW thermal input. The pellets are introduced at the top of the gasifier
while air as gasification agent is introduced at the bottom through a
grate. The grate area is 100 cm2 which corresponds to a maximum of
250 W per cm2 thermal area loading. The firebed is separated from the
grate by alumina balls. Besides the protective function, the alumina balls
promote a homogeneous distribution of the gasification air throughout
the grate area, [145]. The producer gas is taken out near the top of the
gasifier.
An initial commissioning run of the gasifier is shown in Figure 3-9. It can
be seen that the temperature in the middle of the gasification reactor,
the producer gas temperature and the gasifier pressure fluctuated, ulti-
mately leading to a shutdown of the gasifier due to over-pressure and
over-temperature alarm after roughly 24 h of operation. The mass flow
of wood pellets was approx. 2.8 kg/h, which corresponds to 45 % of the
rated capacity.
Figure 3-9: 800 200

24 h operation of
Producer gas
the new updraft 700
temperature
175
gasifier with fluc-
600 150
tuating tempera-
tures and pres-
Temperature [°C]
Pressure [mbar]
500 125
sures Temperature in the
400 middle of the 100
gasification reactor
300 75
Gasifier
200 50
pressure
100 25
0 0
0.00 0.25 0.50 0.75 1.00
Runtime [d]
The gasifier produced about 2.2 mn3 of producer gas on a dry and tar
free basis per hour at an air-to-fuel ratio in the range of 0.25 to 0.27. A
85
3 Experiments
mass balance was applied to the gasifier, taking into consideration the
dry gas composition and measured water content. It was found that the
sum of the tar species maintained a C-H-O ratio corresponding to the
stoiciometric formula CH1.2O0.5. This implied abundance of oxygenated
tar species in the producer gas as also reported in [28, 29] for a com-
mercial updraft gasifier. The tar load was calculated to be as high as
160 g/mn3 (dtf). This is an elevated value even for an updraft gasifier,
where most of the volatile tars escape during the pyrolysis occurring at
the upper part of the reactor. Apart from being detrimental to a SOFC
stack at such levels, the high tar load also rendered stable long-term
operations of the gasifier. This called for further improvements.
Thermodynamic calculations indicated that with the injection of steam
into the gasifier, keeping the reaction zone temperature above 650 °C, a
considerable fraction of the tars could be reformed or formation could be
inhibited. Consequently, a steam generator along with a super heater
coil was integrated with the gasifier, capable of injecting up to 800 g/h of
steam at 600 °C into the alumina ball bed.
Figure 3-10 shows 167 h of operation of the gasifier with constant steam
injection at 600 g/h, which corresponds to 460 g per kg of dry wood pel-
lets or a theoretical humidity of the wood pellets of approx. 30 %.
Figure 3-10: 800 200

167 h operation Producer gas
temperature
progression with 700 175
stable tempera-
tures and stable 600 150
pressure build-up
Temperature [°C]
Pressure [mbar]
500 125
and decline cycles Temperature in the
middle of the
400 gasification reactor 100
300 75
Gasifier
200 pressure 50
100 25
0 0
0 1 2 3 4 5 6 7
Runtime [d]
The temperature in the middle of the gasification reactor and the pro-
ducer gas temperature did not fluctuate and the gasifier pressure
showed no trend for blocking of the gasifier or the downstream piping.
However, the gasifier pressure did not remain at a constant level. It in-
creased up to a maximum of 60 mbar over-pressure, before the deposits
86
3 Experiments
causing this over pressure were removed by the producer gas flow. Be-
sides allowing for a stable operation of the gasifier, the steam injection
had the expected positive impact on the producer gas composition and
tar load. Figure 3-11 shows the improvement of gas composition, start-
ing from no steam injection mode, as the steam rate was gradually in-
creased till 640 g/h. This corresponds to a theoretical wood pellet hu-
midity of approx. 33 %. The air-to-fuel ratio was fixed at 0.26.
Figure 3-11: 20 72
Effect of steam
injection on the
raw producer gas
H2, CO, CO2, CH4, C2H4 [mol-%]
15 54
composition of the
PSI updraft gasi-
fier
N2 [mol-%]
10 36
H2 CH4
CO CO2
C2H4 N2
5 18
0 g/h 150 g/h 400 g/h Water 640 g/h
0 0
0:00 1:00 2:00 3:00 4:00 5:00 6:00
Time [h]
It can be seen that the molar fraction of hydrogen almost doubled with
increasing steam injection from approx. 5 to 9 %. The carbon dioxide
molar fraction increased from approx. 8 to 12 %, while the carbon mon-
oxide molar fraction decreased from 15 to 13 %. This indicates that wa-
ter gas shift reactions were enhanced by the injected excess steam. The
nitrogen fraction decreased from approx. 63 to 56 % with increasing
steam injection. This corresponds to a gain of dry and tar free gas vol-
ume of roughly 13 %.
Mass balance calculations were performed to estimate the tar load of
the producer gas with steam injection, Figure 3-12. It was found that the
steam injection allowed an effective conversion of the tars into prefer-
able permanent gas species, decreasing the tar load by about 50 %.
This also explained the increased producer gas volume. The remaining
tars are not expected to pose problems to a SOFC anode at an operat-
ing temperature around 900 °C due to their oxygenated, thus unstable
nature.
In summary it can be said, that the gasifier built for the demonstration
unit fulfills all the defined requirements, allowing for long-term tests.
87
3 Experiments
Figure 3-12: 300

Comparison of tar Tars
286
and water load of Water
250
the producer gas
produced by the
Tar and water load [g/mn3 dtf]

PSI updraft gasi- 200
fier operated with
and without steam 150
injection 152
100 112
82
50
0
No steam injection 460 g/h steam injection
3.4.2 Hot gas particle removal

For the removal of particles from hot producer gases, a filter unit em-
ploying candle filter elements was built. Despite the current high stan-
dards of candle filtration systems, the design is still based on empirical
findings and experiments. Compared to the typical flue gas filtration, the
unsteady physical and chemical properties of the particles and the pro-
ducer gas originating from the gasification process may cause problems,
[41]. Especially the tar load of producer gases makes predictions about
the technical behavior of hot producer gas filtration systems (HGF) diffi-
cult. Consequently, only little information about the operation of HGF
systems with real producer gas is available. Further, the effect of HGF
systems on the producer gas composition has hardly been investigated.
Eventual positive effects regarding the conversion of tars may be possi-
ble. Therefore, the scope of the HGF characterization experiments was:
To demonstrate a stable operation with real producer gas over a pe-
riod of at least 24 h.
To determine the impact of the HGF system employing different filter
candle elements on the producer gas composition and the tar load.
Figure 3-13 shows the experimental setup employed for this investiga-
tion. The updraft wood gasifier used for the first proof-of-feasibility tests
was coupled with a heated filter housing. The housing was designed to
accommodate a single filter candle of either ceramic or metallic material.
Further, the gas sampling points SP1 and SP2 were installed for the
monitoring of the producer gas composition up- and downstream of the
HGF unit. The raw gas was sucked through the HGF system by different
gas pumps depending on the desired volume flow. The non-filtered gas
was burned in a flare. The volume flow of the filtered producer gas was
88
3 Experiments
monitored with a heated orifice and subsequently sucked through a wet

scrubber where condensable tar species were removed in order to pre-
vent blocking of the employed pumps. After passing the pumps, the pro-
ducer gas was forced through a gas meter and finally into the flare for
combustion.
Figure 3-13:
Updraft wood
gasifier com-
bined with a hot
gas filtration
system and an
oil scrubber Flare Silo Wet scrubber
SP 2
Orifice
Gas pumps
SP 1
Filter housing
Wood gasifier
The time progression of the pressure loss over the investigated HGF
system is a good indicator for its operational condition.
Figure 3-14:
Δpmax
Typical pressure
history of filtra-
Pressure drop Δp
tion systems
with cleaning
cycles, [41, 146] Deep bed filtration
ΔpR
Surface filtration
Filtration
interval
Δp0
Runtime t
Figure 3-14 shows a typical pressure loss time progression of candle fil-
ter element based HGF systems. New filter elements pass through a
commissioning phase prior to reaching stable operating conditions. At
first, the pressure loss over the filter housing Δp0 originates only from
the filter element. Particles deposit in the filter element and induce an
increase of the pressure loss over the filter element. The commissioning
89
3 Experiments
phase is completed after the saturation level of particles in the filter ele-
ment is reached. The particles then deposit only on the surface of the fil-
ter element accumulating to a dense dust cake with lower porosity com-
pared to the fresh filter element. The dust cake is allowed to build up un-
til the pressure loss reaches the design level Δpmax. This initiates a re-
generative cycle, where the dust cake and parts of the particles depos-
ited in the filter element are removed by a reversed gas flow through the
filter element. Due to the flammability of producer gases, pressurized
argon was employed for the regeneration cycles. The filtration interval is
defined as the time period between two consecutive regeneration cycles.
The pressure loss after a regenerative cycle, ΔpR, draws near a given
value for stable operation. If it steadily increases, particles penetrate
deep into the filter element ultimately leading to operational failure.
Two types of filter elements were characterized, namely a non-ceramic
fiber filter candle with a pore diameter less than 1 μm and a metallic fil-
ter candle with a pore diameter less than 5 μm. The gasifier was oper-
ated with a wood pellet mass flow of 1.3 kg/h and an air-to-fuel ratio of
0.36 which translates into a gasification air volume flow of 33 ln/min. The
filtration velocity is defined as the quotient of filtered producer gas vol-
ume flow and filtration area of the investigated filter element. For the
demonstration of a stable operation, the filtration velocity was main-
tained at 1.6 cm/sec. The temperature of the HGF unit was held con-
stant at 400 °C.
Figure 3-15: 200 2
27 h operation of
hot and highly tar
laden gas filtration
Pressure drop over HGF unti [mbar]
160 1.6
employing ce-
Filtration velocity [cm/sec]
ramic filter can- 120 1.2

dles
80 0.8
40 0.4
Gasifier failure
0 0
8:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 0:00 2:00 4:00 6:00 8:00 10:00
Time [hh:mm]
Pressure drop over HGF system Filtration velocity
Figure 3-15 depicts 27 h of non-stop operation of the HGF unit employ-

ing the non-ceramic fiber candle. In total approx. 75 hours of stable op-
90
3 Experiments
eration were achieved with this filter element type. This corresponds to a
total filtered gas volume of approx. 57 mN3, thereof approx 27 mN3 non-
stop. The grey area indicates a gasifier failure which made a burnout
cycle necessary. The HGF unit was operated throughout this burnout
cycle, showing no signs of clogging due to the strongly increased par-
ticulate matter load of the gasifier flue gas. The pressure after cleaning
pulses, ΔpR, leveled off at 80 mbar with Δpmax set to 140 mbar. The re-
generation cycles occurred at regular 60 minutes intervals. No particles
were found in the wet scrubber. It was concluded, that stable operation
was reached with the investigated non-ceramic fiber filter element.
In contrast, the metallic filter candles were clogged almost instantly
when operated at 400 °C. The temperature of the HGF unit was there-
fore increased to 500 °C, which allowed a continuous operation. How-
ever, large amounts of soot were found in the wet scrubber fluid. The
soot was either produced in the metallic filter candle or consisted of char
particles originating from the gasifier which passed the filter element.
For both cases it was concluded, that metallic filter candles are highly
problematic for the filtration of tar laden producer gases.
The impact of the HGF system was assessed via the measurement of
the producer gas composition up- and downstream of the filtration unit.
The tar and water load was estimated with mass balance calculations,
see section 3.2. It was found that the HGF system employing the non-
ceramic filter elements influenced the producer gas composition. Due to
the inert character of the filter element material, it is most probable that
the observed activity is due to the soot cake, which is continuously de-
posited on the candles. The molar fractions of H2, CO, H2O and CH4 in-
creased notably passing the HGF system, while the molar fractions of
C2H2, CO2 and tars decreased. This indicates that tar and ethylene de-
composition processes and reverse water gas shift reactions took place
in the HGF system, see Figure 3-16. It has to be pointed out that these
values were calculated based on the measured species concentration
and the balanced tar and water load. As it cannot be excluded, that the
principal tar signature changed, the values should be understood as
rough trend indicators. More detailed studies are required in order to de-
termine the exact reaction scheme. The variation of the filtration velocity
had no measurable impact on the discussed trends.
The metallic filter elements were two to three times more active than the
non-ceramic fiber filter elements towards the decomposition of tars and
91
3 Experiments
ethylene. With ethylene being a well-known soot pre-cursor, this rein-

forced the conclusion that the soot found in the wet scrubber fluid was
formed in the filter element.
Figure 3-16: 1.25
Conversion of 1.00
measured gas
CO
species and 0.75
balanced tar and 0.50
water load in the

Conversion [mol/h]
0.25
HGF system CH4 C2H4
H2O
employing a 0.00
C2H2 H2
non-ceramic -0.25
Tars
fibre filter ele-

-0.50 CO2
ment
-0.75
-1.00
-1.25
In summary, stable operation with tar-laden producer gas was shown

with non-ceramic fiber filter elements. These inert filter elements in
combination with the soot cake typical for this filtration technology had a
positive impact on the producer gas via tar and ethylene decomposition.
For filter elements made of metallic materials this impact was found to
be noticeably stronger. The high activity of the metallic material leads to
soot formation in the filter element, negating the filtration function.
Therefore metallic filter elements should not be used for the filtration of
tar-laden producer gases.
3.4.3 Catalytic partial oxidation

The catalytic partial oxidation (CPO) was identified as a possible con-
verter of aromatic and oxygenated aromatic tars to hydrogen and carbon
monoxide as well as organic sulfur species to hydrogen sulfide. Initial
experiments with a CPO operated with real wood gas were conducted
and proved the robustness of the catalyst. The results were reported in
[147]. However, to prove that the employed commercial CPO catalyst
actually does convert the typical tar species found in biomass producer
gases, dedicated experiments were conducted. Three representative tar
species were investigated, namely toluene, representing single ring
aromatics, anisole, representing oxygenated aromatics and thiophene,
representing aromatic organic sulfur compounds, see Figure 3-17. The
92
3 Experiments
experiments discussed in the following were part of a diploma thesis co-

supervised by the author, [148].
Figure 3-17:
Representative tar O
species for CPO
characterization
experiments
S
Toluene Anisole Thiophene
With these representative tar species most of the structures of biomass

gasification tars were covered, except primary tars such as e.g. acetic
acid which should not be problematic for conversion due to their low
stability. The tests with toluene provided an indication of the influence of
the catalyst on stable aromats, while the tests with thiophene aimed at
the conversion of organic sulfur species to H2S and the proof of sulfur
tolerance of the investigated CPO catalyst. Despite phenol being a more
frequently encountered oxygenated tar specie in biomass producer
gases, anisole was used for the experiments because it is less difficult
to handle at room temperature and features almost the same properties
with respect to its interaction with catalytic surfaces. The experimental
setup used is outlined in Figure 3-18.
Figure 3-18: Ar Ar
Experimental MFC-Station
setup for catalyst Ar-
H2S
characterization
experiments P-11
CO2
CO
N2
Tar H2O
Saturator Saturator
H2
CH4
Preheater
MS Superheater
Note: Valve points to

the active line
Fixed bed reactor in furnace Condenser

To fume
hood
Catalyst
93
3 Experiments
The setup was designed to allow for catalyst tests with tar-laden and
humidified synthetic gas mixtures. Six mass flow controllers enabled the
generation of a wide range of dry gas mixtures consisting of H2, CO,
CO2, CH4 and N2. Hydrogen sulfide diluted in argon could be added as
well. The dry gas mixture was then preheated to 180 °C and humidified
with deionized water via a water saturator with argon as carrier gas. The
water load was controlled via the saturator temperature which was set to
70 °C for the experiments. The addition of liquid tars followed the same
scheme. The saturator temperatures were 30 °C for thiophene, 60 °C
for toluene and 90 °C for anisole. The tar-laden and humidified gas mix-
ture was then superheated and piped into a packed-bed reactor housing,
which was placed in an electric furnace. The gas samples for the
Thermo VG ProLab quadrupole mass spectrometer (MS) were sucked
through fused silica capillaries which were placed before and after the
catalyst. A condenser located at the reactor outlet was used for the re-
covery of non converted water or organic compounds.
As mentioned in section 3.1.2, CPO catalysts feature two reaction zones,
the first being the oxidation and the second being the steam reforming
zone. It was assumed, that while oxidation occurs fast, producing water,
carbon dioxide and heat, the steam reforming is a slow process. There-
fore, a diluted "post-oxidation" gas composition was defined as matrix
for the tests, consisting of 9 mol-% H2, 8 mol-% CO, 6 mol-% CO2, 15
mol-% H2O and 62 mol-% N2. The temperature range chosen for the ex-
periments was 700 to 800 °C. The weight hourly space velocity (WHSV)
was varied between 2.75 and 7 gtar/(gcatalyst h).
For WHSVs between 2.75 and 5.5 gtoluene/(gcath), toluene conversions
between 38 and 59 % were observed using 119 mg of catalyst at a tem-
perature of 700 °C, see Figure 3-19. At 800 °C the conversion was con-
siderably higher with values between 72 and 90 %. While at 800 °C
most of the toluene was reformed to hydrogen and carbon monoxide, Eq.
1, at 700 °C an important part of the toluene was degraded to benzene,
Eq. 33.
C7 H 8 + 7 H 2O → 7CO + 11H 2 Eq. 32
2C7 H 8 + 2 H 2O → 2C6 H 6 + 2CO + 4 H 2 Eq. 33
The produced benzene however should not be problematic in an SOFC

due to the high reforming activity of the typically used anode catalysts,
see section 3.3.2.
94
3 Experiments
Figure 3-19: 100

Toluene conver-
sion over a CPO 80
catalyst for differ-
ent temperatures
Conversion [%]
60
and WHSVs,
[148]
40
20
0
2 3 4 5 6 7 8
WHSV [gtoluene/(gcat h)]
700°C 800°C
Figure 3-20 depicts the conversion of anisole over the investigated CPO
catalyst. The observed conversions of anisole were between 61 and
83 % at 700 °C and between 78 and 87 % at 800 °C. In contrast to the
conversion of toluene, no formation of benzene was observed during the
reforming of anisole. The high conversions and the lack of benzene as
side product support the assumption that oxygenated tars are easier to
degrade than pure aromatic species.
Figure 3-20: 100

Anisole conver-
sion over a CPO 80
catalyst for differ-
ent temperatures
Conversion [%]
60
and WHSVs,
[148]
40
20
0
2 3 4 5 6 7 8
WHSV [ganisole/ (gcath)]
700°C 800°C
The observed conversions for thiophene were between 34 and 45 % at

700 °C and between 68 and 72 % at 800 °C, Figure 3-21. Keeping in
mind that the oxidation zone of the CPO catalyst exhibits temperatures
up to 1000 °C, a complete conversion of organic sulfur compounds is
probable. This was also enforced by the finding that at higher tempera-
tures, the produced amount of hydrogen sulfide increases, which sug-
gests a cracking mechanism of the thiophene ring.
95
3 Experiments
Figure 3-21: 100

Thiophene con-
version over a 80
CPO catalyst for
different tempera-
Conversion [%]
tures and WHSV, 60
[148]
40
20
0
2 3 4 5 6 7 8
WHSV [gthiophene/(gcath)]
700°C 800°C
In summary the order of reactivity of the three model compounds of

biomass gasification tars towards steam reforming was found to be ani-
sole, thiophene and toluene. It was approved that oxygenated tars are
easier to degrade than pure aromatics as also reported in [55]. Further,
the investigated catalyst was found suitable for the decomposition of or-
ganic sulfur compounds.
3.4.4 Sulfur absorption by zinc oxide

The desulphurization of producer gases from biomass gasification was
investigated in the framework of a diploma thesis co-supervised by the
author, [149]. The experimental setup used for the investigation was al-
ready discussed, see Figure 3-18. Besides the mass spectrometer, a
Sievers/Ionics SCD 355 sulfur chemiluminescence detector with a de-
tection limit below 0.1 ppmV was used to determine the total sulfur con-
centration in the producer gas downstream of the adsorption bed. The
sulfur species of interest were hydrogen sulfide and organic sulfur com-
pounds represented by thiophene. Carbonyl sulfide was not considered
assuming that it is converted to hydrogen sulfide via Eq. 34 at the inves-
tigated conditions.
COS + H 2 O ↔ CO2 + H 2 S Eq. 34
The chemisorption of hydrogen sulfide by zinc oxide follows Eq. 35.

ZnO + H 2 S → ZnS + H 2 O Eq. 35
The experiments with zinc oxide as state-of-the-art sulfur adsorbent

were focused on the separation efficiency, the thermal stability in reduc-
ing atmospheres and the impact on the producer gas composition.
96
3 Experiments
For hydrogen sulfide, the sulfur separation efficiency was close to 100 %,
yielding sulfur concentrations at the adsorbent bed outlet below 1 ppmV.
These values were also reached despite an increase of the water frac-
tion in the producer gas matrix from 5 to 13 mol-%. Tests with thiophene
confirmed, that organic sulfur species are not adsorbed by zinc oxide.
With organic sulfur concentrations up to several ppmVs, this confirms
that a conversion unit for organic sulfur species such as a CPO is most
required in B-IGFC systems. Otherwise sulfur concentrations around 1
ppmV cannot be reached.
The thermal stability of the investigated zinc oxide adsorbent was as-
sessed with thermo-gravimetric measurements. The maximum operation
temperature was found around 450 °C, see Figure 3-22. Starting from
550 °C, zinc oxide is reduced to pure zinc and quickly begins evaporat-
ing. The measured curve compares well with the vapor pressure curve
of pure zinc calculated with ASPEN PLUS.
Figure 3-22: 125 1.25

Thermo gravimet- Mass signal
Zinc oxide reduced
to pure zinc
ric evaporation Zinc vapor pressure from ASPEN
Mass of zinc oxide sample [% of initial value]
curve of zinc ox- 100 1.00

Zinc evaporation
ide sulfur adsorb-
Zinc oxide vapor pressure [bar]

ent, [149]
75 0.75
Thermally stable
operation region
50 0.50
25 0.25
0 0.00
300 400 500 600 700 800 900
Temperature [°C]
The composition of the producer gas matrix was not altered in the pres-
ence of zinc oxide. In summary, the experiments confirmed that the find-
ings reported in [107] with respect to zinc oxide for the desulphurization
of syngases originating from natural gas or coal gasification can be car-
ried forward to biomass gasification applications.
3.4.5 Solid oxide fuel cell

The poisoning effect of sulfur on the electrochemical performance of
SOFCs is the subject of numerous studies, [10, 118, 119]. The corre-
sponding results are however strongly linked to the investigated SOFC
97
3 Experiments
and catalyst materials. The fuel requirements must therefore be deter-

mined for the SOFC in question. However, the primary targeted fuel
provides an initial indication about the sulfur tolerance of a specific
SOFC concept.
The primary targeted fuel of the Hexis SOFC is odorized natural gas.
These cells exhibit therefore some tolerance towards sulfur. For this
reason, and for its proven reliability in numerous long-term field tests
with over 120 running systems, [75], the Hexis SOFC was found a prom-
ising candidate for the proof-of-concept tests. For a test duration of sev-
eral days with a resulting permanent degradation on moderate level, the
minimum required desulphurization level was determined experimentally
using a full-scale Hexis SOFC system.
Figure 3-23:
The Hexis SOFC Air Fuel gas Exhaust gas
system, [150]
Catalytic partial Post-combustion Exhaust channels
oxidation monolith
Insulation
SOFC stack
Air preheater
Figure 3-23 shows the Hexis SOFC system. The Hexis cell design is
planar circular, electrolyte-supported and optimized for co-flow operation.
The stack consists of 63 metallic current collectors and electrode-
98
3 Experiments
electrolyte assemblies. The active area per cell is roughly 100 cm2. The
maximum electrical output of the stack is 1 kW.
Air is introduced to the SOFC stack via an air pre-heater. The pre-
heated air enters the SOFC stack on the cathode side through four air
channels which are formed by the metallic interconnector plates and the
anode-electrolyte-cathode assemblies. After participating in electro-
chemical reactions, the air exits the stack radial and takes part in the
post-combustion of the partially depleted fuel gas. The fuel gas entering
the stack is reformed and heated to a temperature suitable for the SOFC
stack using a catalytic partial oxidation monolith. The processed fuel gas
is introduced to the SOFC stack on the anode side via a central fuel gas
channel. After being partially electrochemically converted to water and
carbon dioxide, the depleted fuel is mixed with the depleted air and
combusted. The exhausts are used for the pre-heating of the fresh air.
Pure methane was used as model fuel for the experiments. A thiophene
saturator, see sections 3.4.3 and 3.4.4, was installed in an ice bed. Ar-
gon was used as carrier gas. The thiophene concentration in the fuel
gas was adjusted via the carrier gas flow. The employed mass flow con-
troller allowed producing thiophene concentrations in the fuel gas up to
400 ppmV. This translates into approx. 80 ppmV at the cell inlet, due to
the volume expansion of the fuel gas resulting from the CPO. To ac-
count for the adsorption of sulfur on the metallic surfaces, the voltage-
current curves were only recorded after several hours of steady-state
operation. Assuming that in steady-state the ad- and desorption of sulfur
on the metallic surfaces of the feed line occur at equal rates, the sulfur
concentration at the cell inlet should correspond to the set value.
Figure 3-24: 65
Voltage-current
curves obtained
from the investi- 60
gated Hexis-stack
for different sulfur
Stack voltage [V]
55
concentrations
ppmV of thiophene in feed gas to
catalytic partial oxidation catalyst
50
45
400 200 90 50 25 12 6 0
40
0 5 10 15 20 25
Stack current [A]
99
3 Experiments
Figure 3-24 shows the measured voltage-current curves for varying

thiophene concentrations. The Hexis SOFC proved to be very sulfur tol-
erant, allowing sound operation even at the highest tested sulfur con-
centrations. The power output declined considerably with increasing sul-
fur concentration due to decreasing open circuit voltages (OCV) and
steeper slopes of the voltage-current curves.
The steeper slopes could be explained by limitations of the electro-
chemical reactions due to the blocking of active sites at the fuel elec-
trode- electrolyte interface as reported in [118]. The decline of the OCV
follows an s-shaped curve, Figure 3-25. This indicates that more than
one mechanism must be active. The present work proposes a combina-
tion of three mechanisms.
Figure 3-25: 1000

Open circuit volt- 1 2
age for different
sulfur concentra- 980
tions 1: Deactivation of anode catalyst
Open circuit voltage [mV]
2: Deactivation of CPO catalyst

960
940
920
900
0 50 100 150 200 250 300 350 400
Sulfur concentration in feed gas [ppmV]
The OCV decrease between 0 and 30 ppmV could be a consequence of

the deactivation of the nickel-based anode catalyst towards steam re-
forming reactions (STR). Slower STR cause flattened Nernst voltage
distributions resulting in lower effective OCV values. Assuming the inter-
connector plates as equipotential, the effective OCV is the voltage at
which the sum of in-plane currents is zero.
The impact of the reforming activity on the OCV was investigated with
the SOFC model discussed in depth in chapter 5. The deactivation of
the catalyst was modeled by reducing the STR reaction rate. Similar to
the experiments, methane, after a catalytic partial oxidation with an air-
to-fuel ratio of 0.27, was assumed as fuel gas for the model calculations
and the overall air-to-fuel ratio was set to three. As in OCV operation the
entire fuel is converted in the post-combustion zones, which directly
100
3 Experiments
heat the stack, isothermal conditions were assumed with a temperature

of 950 °C. The precise geometry of the Hexis cell was not considered
but a generalized geometry was assumed. Despite these simplifications,
the modeling results provide an indication about the impact of the pro-
posed OCV drop mechanism. Figure 3-26, shows the Nernst voltage
distributions for a fully active anode catalyst and a catalyst with 60 % re-
duced activity. The effective OCV drops considerably with decreasing
reforming activity.
Figure 3-26: 1.075

Calculated Nernst
voltage distribu-
tions and effective Inplane currents
open circuit volt-
1.050
ages for 100 %
Nernst voltage [V]
and 40 % reform-
ing activity
1.025
Nernst @ 100% kinetics
Nernst @ 40% kinetics
OCV @ 100% kinetics
OCV @ 40% kinetics
1.000
0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10
Cell length [m]
Figure 3-27 correlates the OCV drop with the reforming activity.
Figure 3-27: 30
Calculated OCV
drop for varying
25
reforming activi-
ties
20
OCV voltage loss [mV]
15
10
0
0 10 20 30 40 50 60 70 80 90 100
Anode reforming activity [% of activity without sulfur poisoning]
Assuming that the first 30 mV of OCV drop in Figure 3-25 result from re-
duced STR activity leads to approx. 90 % deactivation according to
Figure 3-27. This magnitude of deactivation is reasonable for high-
101
3 Experiments
temperature steam reforming catalysts operated at 950 °C with sulfur

concentrations of 10 ppmV, [107]. However, in the experiment the sulfur
concentration at the cell inlet causing 30 mV OCV drop was about 5
ppmV. Considering that in the regions where STR takes place the tem-
peratures can be up to 200 K lower than the average cell temperature
leads to critical sulfur concentrations of 1 ppmV, [107]. Therefore, a
90 % deactivation with 5 ppmV of sulfur in the fuel gas might be possible.
The second mechanism which could play a role for the observed OCV
drop is the deactivation of the CPO catalyst. Lower conversion rates of
the CPO yield less hydrogen and more water due to reduced methane
reforming. The lower hydrogen-to-water ratios yield directly lower OCV
values. To account for a 50 mV of OCV drop, the deactivation of the
CPO was calculated to approx. 30 %.
The double potential formation between hydrogen/water and hydrogen
sulfide/sulfur dioxide could be another explanation for the observed
OCV drop, [119]. However, this mechanism has thus far not been inves-
tigated in depth. Future experiments should focus on hydrogen as a
model fuel, the Nernst voltage of which is independent of the STR activ-
ity, to prove or disprove the proposed hypotheses.
Figure 3-28:
65
Comparison of
voltage- current
curves obtained 60
before and after
sulfur experiments
Stack voltage [V]
55
Before thiophen experiments
After thiophen experiments

50
45
6% degradation
40
0 5 10 15 20 25
Stack current [A]
Figure 3-28 compares the voltage-current curves obtained before and

after the sulfur experiments. It is well known, that nickel-based catalysts
operated at high temperatures can be regenerated after sulfur poisoning
by switching to a sulfur-free fuel, [70, 107]. However, excessive sulfur
concentrations can cause permanent degradation via bulk sulfidation,
[151]. This is probably the reason for the 6 % permanent degradation af-
102
3 Experiments
ter the sulfur experiments. Given that such degradation values are nor-
mal after Redox and thermo cycle test programs, [84], the role of sulfur
as a limiting factor to the proof-of-concept test can be put in perspective.
In summary, the desulphurization is the key to substantially minimize
sulfur poisoning induced power losses. For the proof-of-concept tests, a
desulphurization was found desirable but not necessary. Indeed sulfur
reduces the power output of the Hexis SOFC but does not inhibit its sta-
ble operation. In addition, omitting the desulphurization neglects the
need for a high temperature heat exchanger, simplifying the experimen-
tal setup for the proof-of-concept test considerably.
3.5 Proof-of-concept of the PSI B-IGFC system

The demonstration of a SOFC running with pure producer gas is an im-
portant milestone in the development process of the B-IGFC technology.
A corresponding experiment was recently conducted in the framework of
the BioCellus project and reported in [152]. In this experiment, the inves-
tigated single SOFC membrane was placed in a temperature-controlled
oven and operated with producer gas from a two-stage downdraft gasi-
fier. A slip stream of the particle-free producer gas with approx.
1 mg/mn3 (dtf) tar load was further cleaned in two activated carbon ad-
sorber beds and then humidified in a water saturator before being intro-
duced to the anode compartment of the SOFC test rig. The experiment
lasted 150 h without operational problems or signs of cell degradation.
However, due to the extensive gas cleaning, it is difficult to draw conclu-
sions concerning the tolerance of the tested SOFC towards typical pro-
ducer gas impurities such as tars. Further, the differences regarding the
operational characteristics between an externally heated SOFC mem-
brane, [152], and a thermally self-sustaining SOFC system are large.
For instance, planar SOFC stacks consist of numerous cells where the
uniform fuel distribution is a major concern. Fluctuations of gas composi-
tions may also become critical for the achievement of a suitable and sta-
ble operational temperature for a SOFC stack.
These questions can only be addressed experimentally under realistic
conditions. Hence, once all important unit operations had been charac-
terized, the next steps were to assemble, operate and gradually opti-
mize a demonstration unit, which would realistically represent the com-
plete process chain from wood to electricity.
103
3 Experiments
3.5.1 First demonstration unit tests and derived modifications

In a first step, a Hexis SOFC system that had already completed several
thousand operating hours on natural gas (BZM66) was linked to the PSI
wood gasifier through a hot gas cyclone designed to remove particles
with a minimum diameter of 15 μm. The BZM66 with internal CPO unit,
see Figure 3-23, was operated under open circuit conditions to investi-
gate the impact of the producer gas induced off-design gas flow condi-
tions on the thermal management and the post-combustion. By operat-
ing BZM66 for 24 h, the proof was achieved that producer gas from air-
blown updraft gasification is suitable for thermally self sustaining opera-
tion of the Hexis SOFC system. Nevertheless, the temperatures of the
SOFC stack did fluctuate in the range of 100 K around the design value.
The variations were mainly induced by producer gas flow discontinuities.
The over pressure at the SOFC system inlet was stable at approx. 100
mbar. For more information see [147]. Based on these results, BZM66
was replaced with a new Hexis SOFC system (BZM79) for tests under
load conditions. To reduce the system inlet pressure, the ceramic heat
shield, which was placed above the internal CPO unit, was removed.
For commissioning, BZM79 was operated soundly for 350 h with meth-
ane processed via the internal CPO unit.
During the first producer gas experiment with BZM79, the entire pro-
ducer gas was fed to an external CPO unit for pre-conditioning. Gas flow
discontinuities were reduced via adaptation of the operational parame-
ters of the gasifier. A slipstream of the producer gas, approx. one third,
was supplied to the SOFC stack through the internal CPO unit where
tars should be further degraded. Within the 10 h duration of the experi-
ment, the produced current decreased from initially 13 A to zero. Fur-
thermore, a constant decline of the stack top temperature was observed
while the pressure at the SOFC system inlet was stable at around 15
mbar. The strong reduction of the system inlet pressure confirmed the
positive impact of removing the heat shield. The declining stack top
temperature and current were allocated to a blockage of the valve used
to control the producer gas slip stream to the SOFC system. It was con-
cluded that, with increasing blockage, the producer gas flow to the
SOFC was reduced, yielding a higher than wanted air-to-fuel ratio in the
internal CPO unit. This in turn reduced the open circuit voltage of the
producer gas below the operational voltage, thus leading to zero current
production. The reduced fuel gas heating value and volume flow most
104
3 Experiments
probably resulted in disturbance of the post-combustion zone, which

would account for the changes in the temperature distribution of the
stack. Currently no apparent reason for why these phenomena did fail to
occur with BZM66 has been found.
To avoid blockage of the slipstream valve, it was decided to manually
open and close it regularly during the next test, in an effort to thereby
loosen and discharge possible particle agglomerations in the valve.
The second producer gas experiment with BZM79 lasted 19 h and
was terminated due to a gasifier failure. On average, the stack produced
6 A of current, corresponding to 280W, with maximum values of 11 A
and minimum values around 5 A. More information can be found in [153].
The regular open-close cycles of the slipstream valve allowed maintain-
ing a relatively constant producer gas volume flow throughout the ex-
periment duration. The pressure at the cell inlet was observed to be
lower after each open-close cycle. A possible explanation for this being
that during the short periods of time where the valve was closed, approx.
one minute, the air initially fed to the internal CPO unit reached the cen-
tral fuel gas channel of the stack and removed particle agglomerations
in this area. Keeping in mind the limited efficiency of the used cyclone
with respect to the removal of particles with diameters below 15 μm,
such agglomerations may have formed. After the experiment, the stack
rendered approx. 6 % less power in methane operation. This indicated
that the Redox cycles triggered by the open-close cycles of the slip-
stream valve were detrimental to the stack.
In subsequent tests, the strategy of operating the SOFC with a producer
gas slip stream was abandoned. Instead, the SOFC would be operated
with the complete producer gas volume. To achieve this, the wood mass
flow to the gasifier had to be reduced by two thirds. First tests showed
that the firebed with the reduced wood mass flow was too loose and led
to an unstable gasifier operation. Stable operation was reached again by
reducing the grate area of the gasifier with a liner. Additionally the inter-
nal CPO unit was recognized to be an uncertainty factor in the demon-
stration unit as it inhibited the measurement of the exact producer gas
composition at the stack inlet. Consequently, the internal CPO unit was
taken out of BZM79 and replaced by a second external CPO unit for
methane operation. A new cyclone was also designed and built for the
removal of particles with a minimum diameter of approx. 5 μm and suit-
able for the reduced producer gas volume flow.
105
3 Experiments
3.5.2 Present demonstration unit configuration

Figure 3-29 outlines the present demonstration unit configuration.
Figure 3-29:
Technical outline
and photograph of
the present dem-
onstration unit
configuration
106
3 Experiments
In section (A), a scheme of the PSI wood gasifier is shown. The techni-
cal details were discussed in section 3.4.1. The gasification air is con-
trolled by means of a mass flow controller (MFC). The temperature and
composition of the raw producer gas as well as the pressure loss over
the gasifier are measured approx. 10 cm downstream the gasifier outlet.
Section (B) shows the current configuration of the gas processing setup.
Particles larger than approx. 5 μm are removed in a cyclone and dis-
charged with a screw conveyor. To prevent tar condensation during the
start-up phase of the gasifier, the cyclone and all producer gas flowed
through pipes are electrically heated to 400 °C. The tar load of the pro-
ducer gas is reduced in the first external catalytic partial oxidation unit
(CPO1). The minimum ignition temperature of the CPO catalyst is
approx. 450 °C. CPO1 can hence be electrically heated to over 600 °C.
The air volume fed to CPO1 is set using a MFC. The pressure upstream
and the temperature in the middle of CPO1 are monitored. The compo-
sition of the catalytic partially oxidized producer gas is measured directly
at the reactor outlet. The second catalytic partial oxidation unit (CPO2)
is placed downstream valve V2. The methane and air volumes are both
controlled by means of MFCs. In section (C), the SOFC system is out-
lined. The temperature and pressure of either the catalytic partially oxi-
dized methane or the producer gas are measured at the SOFC system
inlet. The cooling air flow for the system is adjusted using a MFC. The
actual condition of the SOFC stack is monitored via thermocouples in
the top, middle and bottom cells. The electrochemical performance is
determined by measuring the voltages of 5 cell clusters, Cl1 to 5. The
net stack voltage is maintained constant using an electrical load genera-
tor which automatically adjusts the stack current.
Table 3-2: Operational Valve

Description
Valve status for mode 1 2 3 4 5
the four opera-
tional modes of Burning of the unprocessed
1 closed closed closed open open
the demonstra- producer gas
tion unit Burning of the catalytic par-
2 open closed open open closed
tially oxidized gas
SOFC system operation with
3 open open open closed closed
catalytic partially oxidized gas
SOFC system operation with
4 closed open open closed open
unprocessed producer gas
Table 3-2 summarizes the four operational modes of the demonstration

unit and the corresponding valve status, see Figure 3-29. During the
107
3 Experiments
start-up of the gasifier, the producer gas is entirely flared by default

(mode 1). A propane pilot flame ensures the reliable combustion of the
producer gas in the flare. Once the gasifier has reached stable operation,
normally approx. 2 h after ignition, CPO1 can be put on stream (mode 2)
or the producer gas may be directly fed to the SOFC system (mode 4).
In mode 4, valve 4 and the emergency valve are normally closed. How-
ever, in case the control unit registers any malfunction of the SOFC sys-
tem, the emergency valve and the SOFC system nitrogen purge valve
open automatically. Thus, the producer gas is flared and back diffusion
into the SOFC system is prevented. The same procedure applies in
mode 3 when the producer gas is processed via CPO1 before being fed
to the SOFC system. During the switching phase from mode 1 to mode
4 or mode 2 to 3, respectively, the methane and air volume guided
through CPO2 and into the SOFC system are gradually decreased with
increasing producer gas quantities.
3.5.3 Long-term test results

The demonstration unit in the above discussed configuration was oper-
ated non-stop for over 30 h with the SOFC system on-stream for 28 h.
Figure 3-30 shows that the gasifier outlet temperature hardly fluctuated
and was on average 690 °C with a slight increasing trend. The gasifier
was operated with 25 ln/min of air and 1.12 kg/h of wood pellets, yielding
an effective air-to-fuel ratio of approx. 0.32. The average humidity of the
wood pellets was 6 %. 600 g/h of steam at 550 °C were additionally in-
jected into the gasifier, simulating a wood humidity of approx. 39 %. The
CPO1 temperature was stable around 790 °C. The air volume to CPO1
was 865 ln/h, which translates into an air-to-fuel ratio of 0.26. Both, gasi-
fier and CPO1 air volume flows were not changed during the experiment,
Figure 3-31. The temperature at the SOFC inlet was on average 550 °C.
The 140 K temperature difference between the SOFC inlet and the
CPO1 temperature was a consequence of a secondary air injection
through CPO2 to avoid coke depositions in the non-heated area where
the internal CPO monolith was initially placed, see Figure 3-32: Point 5,
300 °C isotherm. With 56 ln/h, the injected air volume yielded an oxygen
concentration in the producer gas at the SOFC system inlet of approx.
0.25 mol-%. An anode reoxidation should not have occurred as the
small oxygen amount should have either reacted with coke depositions,
if any, or with hydrogen or carbon monoxide in the central fuel gas
channel before reaching the anode compartments.
108
3 Experiments
Figure 3-30:
Time plots of gasi- 1000 1000
Temperatures
fier outlet, CPO1
900 900
mid and SOFC sys- Gasifier out CPO mid SOFC in
tem inlet tempera- 800 800
tures as well as
[°C]
[°C]
measured raw and 700 700
CPO gas composi-
tions plus derived 600 600
volume flows, heat-
ing values, steam- 500 500
to-carbon ratios, tar
and water loads 25 70
Raw gas composition
20 63
H2
[N2 mol.-%]
[mol.-%]
15 CO 56
CO2
10 CH4 49
5
C2 42
N2
0 35
25 70
CPO gas composition
20 63
H2
[N2 mol.-%]
[mol.-%]
15 CO 56
CO2
10 CH4 49
5 C2 42
N2
0 35
10 400
Raw gas properties
[mn3/h] & [-] & [kW]
8 320
Vol. [g/mn3 dtf]
6 LHV 240
4 SC 160
Tar
2 Water 80
0 0
10 400
CPO gas properties
[mn3/h] & [-] & [kW]
8 320
Vol.
[g/mn3 dtf]
6 LHV 240
4 SC 160
Tar
2 Water 80
0 0
0 6 12 18 24 30
Run time [h]
109
3 Experiments
Figure 3-30 shows the raw and CPO gas composition measured during
the first 9 h and the last 7 h of the experiment. Note that the discontinu-
ity in the CPO gas progressions in the 5th hour of the experiment is a re-
sult of a change of the sampling point from the middle to the outlet of the
CPO1 reactor. The characteristic properties of the gases, the tar and
water load, volume flow, LHV power and steam-to-carbon ratio (SC),
were calculated based on the measured dry gas compositions. The SC
was defined as the ratio of the total carbon content of the gases to their
steam content. It can be seen, that the gasifier and the CPO1 operated
steady state. Especially the smooth gas volume and steam-to-carbon ra-
tio progressions are of interest. Table 3-3 summarizes the average and
the standard deviation values of the gas compositions and their charac-
teristic properties.
Table 3-3: Raw gas CPO gas

Raw and CPO Parameter Unit Standard Standard
gas composition Average Average
deviation deviation
with characteris-
tic properties H2 8.01 0.56 11.60 1.05
CO 16.02 1.18 7.20 0.98
CO2 11.00 0.62 13.49 0.81
[mol-% (dtf)]
CH4 2.75 0.20 1.36 0.17
C2 1.08 0.08 0.03 0.01
N2 61.15 2.03 66.31 2.14
Tar load 90.27 28.07 61.98 20.57
[g/mn3 (dtf)]
Water load 361.69 11.95 221.04 12.88
Volume flow [mn3/h] 3.50 0.02 4.40 0.03
LHV power [kW] 4.11 0.10 3.69 0.18
SC [-] 1.07 0.01 0.95 0.03
From Table 3-3 it can be seen that the CPO1 predominantly promoted
the water gas shift reaction. In terms of hydrocarbons, it converted al-
most all C2 species and on average about 40 % of the methane in the
raw gas. The tar conversion was on average around 15 %. The low tar
conversion might be a consequence of diffusion limitations in the wide
channels of the employed CPO monolith with 100 channels per square
inch. A finer meshed monolith could not be used due to blocking issues
caused by the moderate particle removal efficiency of the cyclone. The
LHV loss resulting from the partial combustion of the raw producer gas
was approx. 10 %.
110
3 Experiments
Figure 3-31:
Time plots of stack 5 75
current and voltage, Stack current and voltage Current Voltage
4 60
cluster specific av-
erage single cell 3 45
voltages, stack tem-
[V]
[A]
peratures, pressure 2 30
differences and air
volume flows of the 1 15
main components of
the demonstration 0 0
unit
1.0 1.0
Cluster specific cell voltages
0.8 0.8
0.6 0.6
[V]
[V]
0.4 0.4
0.2 Cl1 Cl2 Cl3 Cl4 Cl5 0.2
0.0 0.0
1000 1000
Stack temperatures
950 950
900 900
[°C]
[°C]
850 850
800 800
Top Mid Bottom
750 750
25 75
Pressure differences Gasifier CPO SOFC
20 60
between in - and outlet
15 45
[mbar]
[mbar]
10 30
5 15
0 0
2000 500
Air volume flows Gasifier CPO SOFC
1600 400
1200 300
[ln/min]
[ln/h]
800 200
400 100
0 0
0 6 12 18 24 30
Run time [h]
111
3 Experiments
The SOFC stack produced approx. 3.3 A in methane operation prior to

switching to producer gas. This more than halved current output com-
pared to after the commissioning phase was a consequence of the
harsh test conditions during the experiments reported in section 3.5.1.
The reason for the low current output was that the top cluster did not
reach the average cell voltage of the four other cell clusters. Due to the
in-series connection of the five clusters, the maximum current is limited
by the lowest single cell voltage, which should not under-run 0.6 V.
Hence, the low performance of the top cluster considerably limited the
stack current. The present discussion focuses therefore on the opera-
tional behavior of the SOFC rather than the absolute produced current.
Figure 3-31 depicts the stack current and voltage progression obtained
from the highly tar laden producer gas. The current produced for a stack
voltage of 48 V was around 2 A during the first 15 h, then declined to
around 1.5 A for 7 h and finally swung around 1.75 A for the remaining 6
h of the experiment. This correlates well with the cooling air flow which
averaged 250 ln/min during the first 15 h of the experiment, then
dropped to an average value of 100 ln/min for the following 7 h and than
increased again to 250 ln/min for the remaining 6 h of the experiment.
Numerous possible reasons for this correlation are conceivable. One
reason could be that the higher oxygen partial pressure at the cathode
resulting from the increased cooling air flow yields reduced cathode po-
larization losses. Another reason could be, that the increased air flow
reduced fuel gas diffusion into the cathode compartment, which other-
wise strongly reduced the open circuit voltage and hence the current for
a given operational voltage.
The stack middle and bottom temperatures were quite stable. This is
due to the cooling air flow being controlled to maintain a temperature of
950 °C in the middle of the stack. The fluctuations of the stack top tem-
perature indicate that in this part of the stack, either the fuel gas distribu-
tion or the post-combustion of the depleted fuel gas was disturbed. De-
spite that thermodynamic carbon deposition should not occur under the
given operating conditions, Figure 3-32, it was assumed during the ex-
periment that coke agglomerations could be present in the central fuel
gas channel. To remove these possible agglomerations, the secondary
air injection volume flow was increased from 56 to 600 ln/h for short pe-
riods of time. This corresponds to an air-to-fuel ratio of 0.23 and an oxy-
gen fraction in the gas of 2.2 mol-%. The response of the system to the
increase of the secondary air injection (SAI) was however contradictory.
112
3 Experiments
Figure 3-32:
C-H-O ternary dia-
gram for the ther-
modynamic graphite C
0.0
Lines
formation with in- 1.0 950°C: SOFC design temperature
scribed C-H-O mo- 800°C: SOFC minimum temperature
lar fractions of dry 650°C: Gasifier outlet temperature
550°C: SOFC inlet temperature
and wet wood and 0.2
0.8 300°C: Min. temperature for
producer gas after noticeable graphite formation
the different process Graphite
steps 0.4 formation
0.6
region Points
CO
0.6 1: Dry wood
0.4 2: Wood humidity
3: Steam addition
1 CO2
CH4 2
4: Gasification air
5: CPO air
0.8
3 0.2
4 5
1.0
0.0
0.0 0.2 0.4 0.6 0.8 1.0
H H2O O
The first SAI at around 9.5 h experimental run time resulted in an instant
stack top temperature increase and system pressure decrease indicat-
ing a successful discharge of deposits from the SOFC system. The fol-
lowing SAI at 12.5 h and at 27 h experimental run time had no impact on
the stack top temperature. The system pressure was lowered during the
SAIs due to an automatic decrease of the cooling air flow resulting from
the cooling effect of the increased gas flow at the anode. However after
the SAIs, the system pressure went back to the initial value as the cool-
ing air flow returned to normal. Post-test examination of the central fuel
gas channel showed that it was clear throughout the experiment. The
assumption of coke deposition was hence shown wrong and the opera-
tional conditions were confirmed as safe in this respect. In contrast, evi-
dence was found for ash deposits in cold regions of the system.
Figure 3-33:
Ash deposits on the
mounting of the
thermocouple at the
SOFC system inlet
113
3 Experiments
Figure 3-33 shows ash deposits on the mounting of the thermocouple at

the SOFC system inlet. At this point the temperature was around 550 °C,
which is below the melting point of biomass ash. Besides the SOFC sys-
tem inlet, the spiral cooling air pre-heater is the only part in the whole
demonstration unit where the temperatures are generally below the
melting point of biomass ash. Fouling in heat exchangers for the recov-
ery of heat from flue gases is a common issue. The average ash content
of the gasified wood pellets is approx. 1 mass-%. With a wood pellet
mass flow on dry basis of 1.05 kg/h and an on-stream time of the SOFC
system of approx. 28 h, 294 g of ash could have condensed in the cool-
ing air pre-heater. It is assumed that this amount is sufficient to noticea-
bly disturb the post-combustion of the depleted fuel gas resulting in the
observed fluctuations of the stack top temperature. A continuous ash
deposition also explains the relatively steady system pressure increase
at reasonably constant cooling air and fuel gas flows.
In summary, the gasifier and the catalytic partial oxidation were oper-
ated without problems, delivering a fuel gas to the SOFC system with
relatively constant composition and properties. The tar conversion in the
CPO1 was found to be limited, most probably by the mass transport
through the wide channels of the coarse CPO monolith. Compared to
methane operation, the SOFC system delivered approx. 40 % less cur-
rent. With the chosen operating conditions, carbon deposition was effec-
tively prevented. Ash deposits were however identified as a major ob-
stacle to a smooth SOFC system operation.
114
115
116
4 Modeling work literature review
Modeling power cycles based on the SOFC technology has gained

much interest in recent years. Especially the natural gas fed SOFC-gas
turbine (GT) cycle has been investigated by several groups with different
approaches regarding the size and design of the cycles, the fuel cell
concept and the model type. Consequently, the estimated cycle efficien-
cies vary in the range of 55 % for small scale systems to 77 % for power
plant configurations, [154]. In contrast, system analysis studies of SOFC
based cycles fed with biomass derived gases are scarce. The following
sections give a brief review of commonly used model types with typical
assumptions and field of application, as well as recent works in the field
of B-IGFC systems analysis. The discussion regarding the different
model types focuses on their application for B-IGFC systems analysis.
4.1 Modeling approaches for steady-state investigations

Modeling fuel cells is a challenging task. A number of coupled phenom-
ena have to be considered e.g. heat transfer via convection, solid and
gaseous conduction and radiation, chemical reaction kinetics for e.g. the
steam reforming reaction, voltage losses due to the irreversibility of the
electrochemical reactions, electric and ionic conduction. Therefore, fuel
cell models are generally tailored to the investigation in question. SOFC
model types in the literature vary from lumped models, which are often
used to study power plant cycles, to detailed computational fluid dynam-
ics models (CFD) for cell design optimization where e.g. flow phenom-
ena in the SOFC gas channels are optimized. The permissibility of every
model assumption has to be scrutinized with respect to the impact of the
generated information. The assumptions and the spatial resolution are
the main criteria for the classification of different model types. The fol-
lowing levels of modeling can be distinguished with respect to the latter
criterion, [112]:
Molecular level (nm-scale)
Electrode level (μm-scale)
Cell level (mm-scale)
Stack level (cm-scale)
Plant level (m-scale)
117
4.1.1 Molecular level models

Molecular level models typically deal with reaction kinetics. The aim of
such models is e.g. to find rate limiting reaction steps.
An extrapolation of the results of such models to the plant level for sys-
tems analysis is not feasible as no information about the performance of
the SOFC in question is generated.
4.1.2 Electrode level models

Electrode level models aim at the investigation of charge and mass
transfer phenomena within the anode-electrolyte-cathode (AEC) struc-
ture and interconnect. A typical result of such a model is the share of dif-
ferent voltage loss mechanisms and the total voltage loss for a given
current density, temperature and set of geometrical and structural de-
tails of the modeled SOFC.
The extrapolation of electrode level model results to the plant level for
systems analysis is the most commonly used approach within the B-
IGFC research community. This is mainly due to rapid calculation times
in the order of seconds and the ease of model development. All extrapo-
lated electrode level models (lumped models) in the literature dealing
with B-IGFC systems analysis have the following assumptions in com-
mon:
The SOFC is considered as a continuous stirred tank reactor (CSTR).
All chemical and electrochemical reactions occurring in the SOFC
reach equilibrium.
The mean current density and fuel utilization are model input values.
Faraday's law directly relates the fuel input mol flow to the current
generated by the fuel cell and the fuel utilization at the specific opera-
tional point. The required oxygen ion flow from the cathode to the an-
ode is given by the stoichiometry of the combustion reaction. Hence,
the overall mass balance of the fuel cell is defined by the model input.
Lumped models can be easily calibrated to fit measured data for a spe-
cific cell type, feed gas and operating conditions set. However, investi-
gations of conditions strongly differing from the calibration case have to
be seen as approximation. As a consequence of the CSTR assumption,
lumped models do not allow for the investigation of the local conditions
in the SOFC. Hence, the generated results have to be examined with
respect to the operational feasibility of the investigated SOFC, [155].
118
The main differentiating factor of the plethora of lumped models in the

literature is how the effective operational voltage, the fuel and air input
mol flows are calculated. Three common approaches can be identified:
I. Models with fixed operational voltage
The power output of the SOFC is a function of the current and the op-
erational voltage. By specifying an operational voltage, usually be-
tween 0.6 and 0.8 V, the output power and therefore the efficiency of
the investigated SOFC is defined. Furthermore, the temperatures of
the output fuel and air streams are functions of the power in- and out-
put of the SOFC. Thus, these model types are not predictive as the
model output is defined via the model input. This type of model has
been used in [132, 143, 156, 157]. These investigations aimed at the
impact of different gas cleaning strategies and bottoming cycles down-
stream of the SOFC.
II. Models with measured voltage-current curves
The operational voltage and the input fuel flow are calculated based on
measured voltage-current curves (UI) to reach a targeted power output.
The oxygen ion mol flow is not fixed, but defined by the measured UI
curve. The impact of fuel gas compositions differing from the reference
composition is usually considered through empirical correction correla-
tions. Within the scope of these correlations, results can be regarded
as realistic and quantitative statements with respect to different overall
system designs are valid. [142, 144, 158-160] are examples of such
investigations, mainly focusing on natural gas fed SOFCs. However,
the scope of the empirical correction correlations is narrow. Hence,
one should not act based on the assumption of a real fuel flexibility of
these models, which is however mandatory for B-IGFC systems analy-
sis due to the varying gas compositions resulting from different gasifi-
cation processes.
III. Models with detailed electrochemistry
The calculation of the operational voltage is based on correlations de-
scribing the ideal reversible voltage (Nernst voltage) and the voltage
loss mechanisms. These mechanisms are the activation, the ohmic
and the diffusion voltage losses which depend on the operational tem-
perature, the species partial pressures, the current density and the
geometrical as well as structural details of the modeled SOFC. Fuel
and air flows are calculated analogously to the above outlined model
type with measured voltage-current curves. This type of lumped model
119
possesses some flexibility towards cell design details and fuel gas
compositions. Calculation times are still in the range of seconds.
The main problem with this type of lumped model is the gas composi-
tion, which is used in the calculations of the operational voltage for a
given current density. In reality, the gas composition changes along the
fuel channel and therewith the Nernst voltage as well as the voltage
losses and the local current density. If the inlet gas composition is used
for the lumped model calculations, the voltage and power output is sys-
tematically overestimated, while it is systematically underestimated if
the outlet gas composition is used, [161]. To yield realistic results,
most lumped models use a theoretic gas composition, which is calcu-
lated as arithmetic average of the inlet and the outlet gas composition.
The oxygen ion flow, which is transferred to the anode from the cath-
ode side, is determined from the predefined fuel utilization. The outlet
gas composition is calculated based on the inlet gas composition and
the oxygen ion mol flow assuming chemical equilibrium at the pre-
defined operational temperature.
This lumped model type is the most accurate of the three lumped
model types. However, it lacks information about the local conditions in
the SOFC. Critical conditions such as excessive temperature gradients
cannot be identified. [161-168] are examples for investigations con-
ducted with this model type.
4.1.3 Cell level models

Cell level models are also referred to as repeat element models. They
include voltage, mass and energy balances in one or two dimensions.
These models are generally used for cell design optimization as they al-
low for the prediction of local cell operating conditions, e.g. temperature,
current density and species partial pressure distribution. Typically de-
sign parameter such as cell length, channel width as well as the flow
design, are investigated with these models. The voltage balance is cal-
culated via an electrode level sub-model. The impact of structural details
of the AEC is therefore also considered. Cell level models differ in the
considered dimensions and the numerical solution approach:
One dimensional models (1D)
Tubular cells are usually modeled 1D assuming heat and mass
transfer only in axial direction. Planar co-flow and counter-flow cells
can also be modeled 1D by neglecting boundary effects perpendicu-
120
lar to the gas flow direction, Figure 2-19. According to [169], the er-
ror due to this assumption is negligible when the feed properties are
equal for all gas channels. In consequence, all other properties
along the gas channels should be equal and so should the local
temperature be. Hence, no perpendicular heat transfer should occur.
1D cell level models feature short calculation times in the order of
minutes and are therefore suitable for system analysis calculations.
Two dimensional models (2D)
Planar cross-flow cells, Figure 2-19, have to be modeled 2D due to
the perpendicular gas flow channels. The sizeable additional effort
is why this type of flow design remains largely neglected in system
analysis calculations. However, such a model was used in [154] for
the investigation of an SOFC/gas turbine (GT) cycle.
For cell level models the following discretization methods can be distin-
guished:
Finite element method (FEM)
The idea of this method is to discretize the investigated object, e.g.
flow channel, but also solid structure, and to allocate transfer func-
tions to all considered physical properties, which can change within
a discrete element. The solution is found, when the difference of the
boundary values of all directly adjacent elements is within the toler-
ance. FEM based fuel cell models are usually employed for detailed
investigations of flow phenomena in the gas ducts and channels.
The calculation times are rather long. As a result this approach is
seldom employed for systems analysis calculations.
Numerical volume averaging method (NVAM)
The NVA method is extensively discussed in [170]. The major dif-
ference to the FE method is, that transfer functions are not only de-
fined for physical properties, but also for properties such as pres-
sure, flow velocity etc. which are changing due to the internal struc-
ture of the discretized element. The corresponding transfer functions
are computed via detailed FEM models of the discretized elements.
The overall solution is found analogously to the FE method. NVAM
models feature shorter calculation times than FEM models.
Finite volume method (FVM)
The FV method is often used for one and two dimensional SOFC
models. The discrete elements are assumed as continuously stirred
121
tanks reactors (CSTR). The change of any considered property in a

specific CSTR element is calculated with the output value of the an-
tecedent element. The calculated value is assumed to be valid for
the entire element. Due to this simplification, FVM models feature
shorter calculation times than FEM and NVAM models and are
therefore preeminently suitable for systems analysis calculations.
This is most probably the reason for their popularity.
The information regarding local conditions in the SOFC allows conclu-
sions about potential failure sources, such as carbon deposition and
thermal stress caused by steep temperature gradients. The spectrum of
published studies conducted with cell level models is discussed below.
The model of a tubular SOFC in [171] was developed by assuming iso-
thermal conditions throughout the cell. Although this assumption is a
strong simplification, the authors show that the accuracy of the predicted
power output values is greatly improved by solving the mass balance
and calculating the power output with the local species partial pressures
as compared to lumped models. This approach was also considered in
[141] for a planar cross-flow cell with the same conclusion for the pre-
dicted power output. However, the temperature of the output gas
streams, which needs to be known for investigations of combined cycles,
cannot be predicted. Further, the temperature distribution cannot be
forecasted and thus no information is generated about critical conditions
in the cell. Models including the full energy balance are reported by the
following authors:
The risk of carbon deposition is usually approximated via the computa-
tion of the probability of graphite formation. Therefore, the chemical
equilibrium is calculated for the given local species partial pressures and
temperatures, [129-131, 172]. Fuel starvation effects in a stack of cells,
which are connected in series, are discussed in [173, 174].
The impact of different fuel gas compositions on the fuel cell efficiency
and the risk of carbon deposition is explored in [175-177] with FVM
models. All three studies are in agreement, that the fuel gas composition
has an important impact on the temperature distribution. Hence, it can
be stated that the implementation of the full energy balance generates
valuable information for the prediction of critical operational conditions in
the cells.
More detailed CFD based models are reported in the following studies:
The study in [178] aimed at the investigation of radial temperature distri-
122
bution induced thermal stress. The implemented 3D CFD cell level

model requires approximately 40 minutes of calculation time and is
therefore not suitable for systems analysis calculations. The model em-
ployed in the studies [179-181] is also based on the FE method. The
model was used to investigate different cell designs aiming at the reduc-
tion of e.g. the anode re-oxidation risk by optimizing the flow field ge-
ometry. Analogously to [178], calculation times are rather long.
The extrapolation of cell level model results to the plant level is based
on the assumption, that all cells in the stack or fuel cell system are op-
erated under the same conditions, e.g. fuel gas input, gas temperatures,
heat losses. The error resulting from this assumption is rather small and
the predicted power output and temperature values are close to reality,
[169]. However, boundary effects, such as increased heat losses of the
cells at the outer ends of the stack, or effects resulting from different
stacking or ducting designs can not be studied.
4.1.4 Stack level models

Stack level models are used to investigate the impact of realistic bound-
ary conditions of the repeat cell elements in a stack, e.g. the impact of
different gas duct designs on the fuel gas distribution and therewith on
the temperature and current density distribution in the stack. In conse-
quence, stack level models are mostly three dimensional (3D).
The model types can be differentiated from the point of view of a nu-
merical discretization approach just as the cell level models. The stack
is discretized and the mass and energy balances are solved for all the
discrete elements. Contrary to cell level models, heat conduction has to
be considered in all three dimensions in order to allow for the prediction
of the impact of realistic boundary conditions such as increased heat
losses at the outer stack surfaces to the surroundings.
The study in [141] addresses the cooling effect of the outer stack sur-
face of a seven cell cross-flow stack. The authors show, that the radia-
tive heat losses of the end cell results in a cold spot temperature of
approx. 80K less than the cell in the middle of the stack, which can be
considered almost adiabatic. Further studies conducted with stack level
models are reported in [139, 182].
The surplus of possibilities offered by stack level models compared to
cell level models is associated with a strongly increased modeling effort
and long calculation times of up to several hours. System analysis cal-
123
culations conducted with stack level models have hence not been re-
ported as yet.
4.1.5 Plant level models

Plant level models consider the interaction between the fuel cell stack
and the balance of plant components such as pumps, turbines, com-
pressors and heat exchangers. Investigations focus on the optimization
of the overall plant efficiency through a high degree of heat integration.
The specific system components are generally modeled as lumped
models with measured characteristic performance curves, e.g. voltage-
current curves of the fuel cell stack in question. The scope of the spe-
cific models is typically narrow. Hence, results for strongly differing input
values to that of the performance curves have to be cross-referenced
with corresponding measurements or detailed models. In the long run, it
would be desirable to incorporate all model levels into one model to
asses all possible interactions between the phenomena occurring on the
different length scales.
4.2 Recent studies of biomass fed SOFC cycles

A plethora of system studies dealing with natural gas fed SOFC cycles
can be found in the literature. Most of the studies consider the tubular
Siemens AG design. Model types vary from lumped models to detailed
FEM models validated with voltage-current curves. A detailed discussion
of these studies is not given in this work but can be found in e.g. [140,
141, 154, 155].
In contrast to natural gas fed SOFC cycles, only few studies about bio-
mass producer gas fed SOFC cycles were thus far published. The most
important contributions are briefly summarized in the following. The fo-
cus is on the approach chosen for the investigations as benchmark for
the approach taken in the present work.
The study of Singh, [129], from the University of North Dakota/USA ad-
dresses the carbon deposition risk under typical SOFC operating condi-
tions resulting from the operation with a tar-laden producer gas. The
employed lumped model is based on Faraday's law. Carbon is assumed
to be represented by graphite. The carbon deposition risk is calculated
under the assumption of chemical equilibrium at the characteristic SOFC
operation temperatures. The oxygen partial pressures required for the
equilibrium calculations are determined based on averaged current den-
124
sity values. The main finding is that by increasing the current, the carbon
deposition risk decreases. This phenomenon is associated with the in-
creasing oxygen partial pressure at the anode with increasing current.
Furthermore, the threshold current density value, at which no graphite
formation occurs, is found to increase with higher tar loads.
Kivisaari, [156], from the Royal Institute of Sweden explores the rela-
tions between a MCFC and the balance of plant (BoP) components in B-
IGFC systems. The employed model is of the lumped type based on
Faraday's law with pre-defined voltage. The main finding is that increas-
ing fuel utilization results in reduced cooling requirements of the MCFC.
Omosum, [132], from the Imperial College/United Kingdom has per-
formed a cost and efficiency analysis of two B-IGFC systems with hot
and cold gas processing system. The employed model is of the same
type as Kivisaari’s. Omosum concludes that the hot gas processing has
advantages on the overall efficiency side due to improved heat integra-
tion possibilities. Furthermore, process waste heat can be sold leading
to increased revenues. These two aspects are assumed sufficient to
cover the higher investment costs of the hot gas processing. Omosum
also states that an increased rate of internal reforming leads to lower
cooling requirements of the SOFC and hence reduced BoP costs.
Van Herle, [160], from the Swiss Federal Institute of Technology in
Lausanne/Switzerland uses a lumped model to study the impact of sew-
age gas on the operational behavior of a SOFC operated at full and par-
tial load. The model includes empirical correlations for activation, ohmic
and diffusion losses. The carbon deposition risk is also considered
through equilibrium calculations for graphite formation. The system
analysis reveals an electrical system efficiency of 48.5 % by employing
a pre-reformer as compared to 43 % by employing a catalytic partial oxi-
dation, which also causes higher cooling requirements of the stack.
Proell, [163], from the Technical University of Vienna/Austria analyses
different options of heat integration, bottoming cycles and system pres-
sures focused on a B-IGFC system based on the FICFB gasifier in
Güssing/ Austria. The applied lumped model includes detailed electro-
chemical calculations. The operational voltage is calculated based on
the average gas composition between the cell in- and output gas com-
position. The output gas composition is calculated assuming chemical
equilibrium. The overall electrical system efficiency is calculated to be
40 %. The largest losses are found in the gasification section.
125
Panopoulos, [168], from the National Technical University of Ath-

ens/Greece uses the same lumped model type as Proell. The investiga-
tions, in the framework of the European project "BioCellus", focus on the
coupling between a novel biomass gasification reactor and SOFCs. The
idea is to transfer heat from the SOFC to an allothermal gasification re-
actor via high-temperature heat pipes. The overall electrical efficiency is
calculated to be approx. 36 %. The variation of the fuel gas water con-
tent showed that the efficiency of sulfur removal decreases with increas-
ing water load. Sulfur was assumed as H2S and its adsorption on ZnO
was assumed at equilibrium.
Cordiner, [183], from the University of Rome/Italy employs a cell level
model to investigate the coupling of a biomass gasifier with an SOFC.
The SOFC model is based on the FE method and is implemented in the
CFD software "Fluent". The producer gas composition delivered by the
biomass gasifier is calculated by assuming chemical equilibrium. At set
conditions the investigated dry producer gas consists of 56 vol.-% hy-
drogen, 39 vol.-% carbon monoxide and the remainder being carbon di-
oxide. Cordiner calculates an electrical overall system efficiency of 46 %
with a fuel utilization of 57 %. The unburned fuel is required to cover the
heat needs of the gasification section.
Hernández-Pacheco, [175], from the University of North Dakota/USA
explores the impact of biomass derived producer gases with low CH4
content and different H2/H2O- and CO/CO2-ratios on the electrical effi-
ciency of anode and electrolyte supported SOFCs. The used 1D cell
level SOFC model is based on the FV method and allows for the predic-
tion of the local conditions in the SOFC. This study concludes that high
H2 concentrations in the fuel gas yield the highest efficiencies. However,
the study does not account for the gasification efficiency and the impact
of methane, which is often present in producer gases.
Sucipta, [184], from the Shibaura Institute of Technology/Japan uses a
1D cell level model for his investigation of a SOFC–GT hybrid system
fuelled with producer gases from air-, oxygen- and steam-blown bio-
mass gasification. The origin of the investigated producer gas composi-
tions is not further detailed. The main finding of the study is that the
SOFC-GT system efficiency is lower when operated with producer
gases instead of methane. Amongst the producer gases, the one origi-
nating from steam-blown gasification is found to yield the highest electri-
cal efficiency, while that from air-blown gasification yields the lowest
electrical efficiency. However, this investigation does not account for the
126
efficiency of the gasification processes, which may considerably alter

the reported coherences. Besides, the targeted mean current density is
comparably high leading to operational voltages below 0.5 V. Conse-
quently the proposed operational conditions are neither sustainable for
SOFCs nor are the reported efficiency values from gas to electricity of
around 38 % competitive with modern gas engines.
Regarding published B-IGFC systems analysis, the latter three studies
use the most detailed SOFC model types. However, the authors do not
focus on the impact of real producer gas compositions, but consider in-
stead idealized fuel gases with very low hydrocarbon contents even
though the hydrocarbon content of real producer gases can be as high
as 20 vol.-%. Furthermore the gasification efficiency is not considered in
the latter two studies. This is a serious drawback as the gasification effi-
ciency may vary between 50 to over 95 % depending on the gasification
process, [25]. Finally, the impact of impurities on the system design and
performance is not considered in any of these studies. For instance, the
removal of organic sulfur species, which may be present in producer
gases in considerable concentrations, has a strong impact on the overall
system design which in return affects the overall system efficiency.
127
128
5 Modeling
5 Modeling
5.1 Overall modeling approach

The modeling approach developed in this work aiming at thermo-
economic evaluations of B-IGFC systems is outlined in Figure 5-1.
Figure 5-1:
- Raw gas properties
Overall modeling - Temperatures of
approach gas processing units
- Required water addition
ASPEN PLUS gasifier and ASPEN PLUS flowsheet

gas processing flowsheet with heat integration
- Composition ot the - Use heat at 200 °C

processed gas at - System DC efficiency
SOFC inlet - Heat exchanger areas
EXCEL sheet with

ATHENA SOFC model
cost estimates
- Gas outlet temperature - Net system AC efficiency

- Air and fuel flows - Size of all units
- SOFC DC Power - Overall system cost
output - Power production cost
In a first step, the producer gas composition at the SOFC inlet is deter-
mined using the commercial flowsheeting software ASPEN PLUS (ver-
sion 12.1). ASPEN PLUS was developed for the simulation of chemical
process applications and solves heat and mass balances in complex
systems. The input data for this first calculation involves the raw gas
composition, temperature and yield, the gas processing unit operations
and their operational temperatures as well as the eventually required
additional amount of water or steam to prevent graphite formation.
The predicted gas composition at the SOFC inlet is transferred to a
SOFC model code implemented in the ATHENA equation solver lan-
guage, [185]. The SOFC model yields the gas outlet temperatures, the
required air volume flow to maintain the design temperature of the
SOFC, the fuel volume flow required for a certain operational voltage at
a given fuel utilization and the direct current (DC) power output. The
SOFC model is discussed in section 5.3.
The results of the SOFC model are used as input for a second flow-
sheeting simulation. Figure 5-2 exemplarily shows the ASPEN PLUS
flowsheet of the B-IGFC system 1, see section 6.2.2.
129
5 Modeling
Figure 5-2:
Flowsheet of
"System 1"
Section (A) represents the gasification process model which composes

the raw gas based on measured input data, see section 5.2.1.
Section (B) is the raw gas humidification unit, see section 5.2.2.
Section (C) comprises gas cleaning and conditioning steps depending
on the investigated system design, see sections 5.2.4 to 5.2.6.
Section (D) reproduces the energy and mass balance of the SOFC us-
ing the results of the detailed model calculations, see section 5.2.7.
Section (E) is the generalized heat integration network. The heat of the
post-combusted SOFC flue gas is used to heat up the anode and cath-
ode gas prior to the SOFC inlet. After this, use heat at 200 °C is ex-
130
5 Modeling
tracted from the flue gas. The remaining heat is used for steam genera-
tion, air pre-heating and biomass drying, see section 5.2.3.
The second flowsheeting simulation yields the overall process DC power
efficiency, the amount of use heat at 200 °C for which all other heat re-
quirements are still satisfied, the areas of all heat exchangers in the heat
integration network and other data relevant to component sizing such as
sulfur adsorbent mass, steam reforming reactor heat duties etc.
The results of the second flowsheeting simulation are passed to an EX-
CEL spread sheet where the net AC power efficiency, the equipment
sizes, the overall system costs and the power production costs are de-
termined, see section 5.4.
5.2 Flowsheeting models

ASPEN PLUS features a graphical interface, an extensive library of
physical property data and equations of state as well as a number of
pre-defined unit operation models. Flowsheeting models consist of a
number of combined unit operation models. The most important flow-
sheeting models implemented in this work are discussed below. For
more detailed information about ASPEN PLUS models see [17].
5.2.1 Biomass gasifier

A flowsheet model was implemented to reproduce measured producer
gas compositions. The corresponding data was either gathered through
own measurements or taken from the literature.
Tar is defined as organic matter that can condense under the operating
conditions of process units subsequent to the thermochemical conver-
sion process, [186]. For modeling purpose, the variety of potentially pre-
sent tar species in producer gases was reduced to three representative
species:
Acetic acid represents tar species originating from the thermal de-
composition of cellulose and hemicellulose.
Anisole represents tar species originating from the thermal decom-
position of lignin such as e.g. m-cresol and syringol, [57].
Toluene represents tar species formed in secondary tar reactions
such as e.g. xylene, naphthalene.
Data concerning the sulfur load and signature of producer gases is
scarce. The sulfur load and the signature were therefore estimated, as-
131
5 Modeling
suming that all sulfur present in the gasified wood are bound in either
hydrogen sulfide or thiophene. The latter was chosen as representa-
tive of organic sulfur species. The ratio between organic and inorganic
sulfur species was assumed similar to that of carbon bound in tars to
carbon bound in permanent gas species. These are carbon monoxide,
carbon dioxide, methane and ethene, respectively.
Figure 5-2 (A) shows that the dry raw gas composition "GASDTF" is first
humidified in the unit operation block MIXER "H2O-LOAD". The sulfur
species are added to the raw gas in the MIXER "S-LOAD". Acetic acid
and anisole as primary tar species are introduced to the raw gas via the
MIXER "PT-LOAD". And finally the raw gas is loaded with toluene as
tertiary tar species in the MIXER "TT-LOAD". The HEATER "HEATGAS"
is used to adjust the raw gas temperature to the measured value.
5.2.2 Gas humidifier

A humidification of the producer gas may be required to prevent carbon
deposition problems. In this work, the additional water amount was de-
termined in separate graphite formation equilibrium calculations. As the
probability of graphite formation increases with decreasing temperature,
the calculations were performed at the lowest system temperature. This
way carbon deposition should not occur in the hotter sections of the
considered system, especially the SOFC. For e.g. updraft gasifier based
systems, where the minimum temperature is almost on ambient level,
300 °C was used for the equilibrium calculations assuming that carbon
formation does not occur below this temperature due to kinetic limita-
tions. Figure 5-2 (B) shows that the additional water is added to the raw
gas in the MIXER "HUMIDIFI". Depending on the gas temperature re-
quirements of the downstream equipment, the water is added in form of
either vapor or wet steam or liquid water. Vapor addition results in a
slight gas cool down while liquid water addition may be used for heavy
gas cool down. The steam is produced in the evaporator HEATX "VA-
PORIZE" using flue gas heat, while liquid water is injected via a spray
chamber modeled with the block FSPLIT "SPRAYER".
5.2.3 Heat integration network

To recover heat from the post-combusted SOFC flue gas, a generalized
however not optimized heat integration network was defined, Figure 5-2
(E). In a first step, the readily processed fuel gas is heated or cooled to
the fuel inlet temperature of the SOFC in the heat exchanger block
132
5 Modeling
HEATX "FSUPHEAT" or HEATX "FCOOL", respectively. The flue gas is

then guided through the superheater HEATX "ASUPHEAT", where the
temperature of the pre-heated cooling air is adjusted to the required
SOFC air inlet temperature. After these two high temperature heat ex-
changers, the flue gas is cooled down to 200 °C in the block HEATER
"USEHEAT" to determine the available use heat amount for sales or for
the operation of a bottoming cycle. The 200 °C hot flue gas is then in-
troduced to the evaporator block HEATX "VAPORIZE" of the gas hu-
midifier unit to deliver the required heat of evaporation in case steam
may be added to the raw producer gas. The remaining heat of the flue
gas is used to pre-heat the cooling air in the block HEATX "ARECUP"
and to dry the wet biomass in the block MIXER "DRYER" if external dry-
ing is required by the gasification process. Successful drying is assumed
if the liquid fraction in the flue gas to stack is zero. This indicates that the
temperature of the flue gas is high enough to hold the additionally intro-
duced humidity of the biomass.
For all the heat exchangers with two in- and output streams, the ASPEN
PLUS unit operation model HEATX was employed. HEATX can be used
for the simulation of various shell and tube heat exchangers in co- and
counter-current configuration. However, in the present work the shortcut
mode was used. The flow configuration was counter-current by default.
Based on a user-specified heat duty or cold/hot side outlet temperature
("FSUPHEAT", "FCOOL", "ASUPHEAT") or temperature approach
("ARECUP") or vapor fraction in one of the two outlet streams ("VAPOR-
IZE"), the short cut mode yields a first estimate for the heat exchange
area. A differentiation concerning the phase of the heat exchanging
streams is done through phase specific heat exchange coefficients. 40
W/m2 K were assumed for hot vapor-cold liquid and 31.25 W/m2 K for
vapor-boiling heat exchange ("VAPORIZE"), [187]. For vapor-vapor heat
exchange, 22.5 W/m2 K were assumed, [187].
5.2.4 Steam reforming and methanation reactors

Steam reforming (STR) of methane is currently the most readily ap-
plied process for syngas production, see section 2.3.2. In this work, a
highly active STR catalyst and a relatively homogenous temperature dis-
tribution in STR reactors were assumed, [142, 144, 155, 160, 188, 189].
Further, the STR temperature was assumed 100 K below the SOFC de-
sign temperature and the minimum steam-to-carbon ratio was set to two
by default, [53]. Based on these assumptions, STR reactors were mod-
133
5 Modeling
eled as isothermal equilibrium using the ASPEN PLUS unit operation

model RGIBBS by specifying the reactor temperature and pressure. The
heat to maintain the STR temperature was subtracted from the SOFC
flue gas. RGIBBS computes chemical equilibria as well as simultaneous
phase and chemical equilibria by minimizing the Gibbs free energy un-
der atom balance constraints. For this reason no reaction stoichiometry
is required for the simulation.
The Methanation of producer gases is currently being investigated by
the PSI TPE research group, [40]. The adiabatic and thus partial
methanation of producer gases has been identified as promising gas
processing step in B-IGFC systems. Similar to the STR, highly active
catalysts and homogeneous reactor temperatures were assumed. In-
stead of a reactor temperature, zero heat duty was specified for the
RGIBBS simulation of the heat constrained equilibria.
5.2.5 Catalytic partial oxidation reactors

The catalytic partial oxidation (CPO) is an alternative process to the
STR. Instead of steam, air or oxygen is used as oxidant gas, see section
2.3.2. The scheme of the CPO process flowsheeting model is shown in
Figure 5-3.
Figure 5-3:
ASPEN PLUS AIR
model of catalytic
partial oxidation
CONVER EQUIL CONVER2
PRODGAS PG2CONV CONVPG CPOGAS
FSPLIT RGIBBS MIXER
PGBYPASS
Based on the experimental results reported in [148], see section 3.4.3,

the CPO process was modeled as adiabatic equilibrium with incomplete
conversion. The block FSPLIT "CONVER" diverts 10 % of the incoming
producer gas to a bypass stream. The remainder of the producer gas
together with a set amount of air is introduced to the equilibrium block
RGIBBS "EQUIL" where the adiabatic equilibrium is calculated. The
amount of air depends on the desired air-to-fuel ratio. In the MIXER
"CONVER2", the producer gas at equilibrium and the diverted raw pro-
ducer gas are mixed again.
134
5 Modeling
5.2.6 Sulfur adsorbent beds

Based on the experimental results reported in [149], the hydrogen sul-
fide adsorption on zinc oxide was modeled as isothermal chemical equi-
librium using the ASPEN PLUS model RGIBBS. The block temperature
was set to 400 °C by default to ensure safe operation, see section 3.4.4.
Except for H2S, all gaseous species in the producer gas were marked as
inert. The allowed products for the equilibrium calculations were H2S,
H2O, ZnO, ZnS and metallic Zn. To reproduce the conditions in adsorp-
tion reactors, where excess adsorbent is present compared to hydrogen
sulfide, the zinc oxide adsorbent mol flow was set to 1 mol/sec. A similar
approach was chosen in [168]. Control calculations revealed that the
predicted H2S outlet concentration is independent on the adsorbent mol
flow as long as it is higher than the stoiciometric required.
5.2.7 SOFC
The flowsheeting model shown in Figure 5-2 (D) was implemented to
enable the application of the results of the ATHENA SOFC model, see
section 5.3, to the overall system simulations. The flowsheeting model
reproduces the mass and energy balance results of the more intricate
ATHENA model. The fuel gas mol flow required to achieve a predefined
fuel utilization at a given operational voltage is calculated by the
ATHENA model. In the overall system model, this mol flow is used to
compute the required dry raw gas mol flow "GASDTF" and thus the
mass flow of wood chips. The oxygen ion transfer through the electrolyte
is modeled with the separator block SEP "CATHODE". SEP blocks are
used to divert single species streams to a certain output stream while
the rest of the input stream is guided to given output stream. The oxy-
gen ion mol flow is computed based on the fuel gas mole flow and the
fuel utilization predefined for the ATHENA model calculations. The an-
ode is modeled as equilibrium block RGIBBS "ANODE" with a given
heat duty. The heat duty depends on the heat required to reproduce the
cathode outlet temperature and the power output determined by the
ATHENA model. The depleted anode and cathode off gases are post-
combusted in a catalytic burner for heat recovery reasons. This burner is
modeled with the block RSTOIC "AFTERBUR" as adiabatic combustion.
RSTOIC is used when the reaction stoichiometry and degree of conver-
sion is known and the reaction kinetics are either unknown or irrelevant.
Combustion reactions are predefined in RSTOIC.
135
5 Modeling
5.3 ATHENA SOFC model

5.3.1 Model description
The envisaged degree of detail in the development of any numerical
model largely depends on the desired level of information to be con-
veyed. In the present work, the development of the numerical SOFC
model was focused on investigating the impact of different fuel composi-
tions on state-of-the-art cell designs for power plant applications. Of par-
ticular interest were the expected power output for a given combination
of fuel composition and cell design and the identification of critical oper-
ating conditions such as
local excess of maximum allowed material temperatures
and unacceptable temperature gradients.
Against this background, all phenomena and mechanisms which are in-
fluenced by the fuel composition and the cell design had to be modeled
as detailed as possible or valid assumptions had to be formulated.
5.3.1.1 Assumptions, definitions and model structure

Developing a detailed SOFC model to investigate the impact of different
fuel compositions on alternative cell designs involves the formulation of
assumptions in order to keep calculation time within reasonable limits al-
lowing for overall system calculations. However, assumptions should not
lower the degree of detail of the model such that the desired information
is a result of the assumptions.
The main assumptions of the developed SOFC model are:
Steady-state conditions are assumed.
Gases are assumed as ideal with temperature-dependent heat ca-
pacity, viscosity and thermal conductivity.
The considered SOFC designs can be divided into a number of re-
peating structures, termed control volumes (CV).
Each CV is assumed as continuously stirred tank reactor (CSTR).
Pressure losses in the gas channels are not considered.
Due to their high electrical conductivity, the interconnector or bipolar
plates of planar cell designs are considered equipotential, [141].
The flow regime in the gas channels is assumed as plug flow, [190].
Turbulent flow is not considered, despite its possible occurrence in
136
5 Modeling
the oxidant gas channels due to excessive cooling air flows, [191].
Mass transport in normal direction to the flow direction arises via dif-
fusion only.
A constant Nusselt number with the value 4 is assumed for the calcu-
lation of the convective heat transfer coefficients, [190].
Radiative heat transport between solid structure and gas streams is
ignored as the patch length of the radiation is too short for a signifi-
cant amount of absorbed heat by the gas streams, [191].
Heat is transferred within the solid structure via heat conduction.
All heterogeneous reforming reactions are considered so fast, that no
diffusion into the porous anode takes place.
The heats of reaction of all heterogeneous reactions can be attributed
to the solid structure of the fuel cell.
The geometry- and fuel composition-dependent diffusion of the reac-
tants of the reforming reactions through the gas channel to the cata-
lytic surface is considered.
The water-gas-shift reaction is regarded as non-diffusion limited ho-
mogenous reaction which is always at equilibrium.
The electrochemical oxidation of hydrogen and carbon monoxide are
considered to occur in parallel. Direct electrochemical conversion of
hydrocarbons is disregarded. Instead, hydrocarbons are oxidized
through reforming reactions and subsequent electrochemical conver-
sion of the produced hydrogen and carbon monoxide.
Besides local values such as solid temperature and current density, the
developed model calculates integral values which characterize the op-
erational conditions and the performance of the investigated cell design
and fuel composition. The definitions of the integral values are given in
the following:
The average solid temperature represents the operational tempera-
ture of the cell and is calculate as the arithmetic average of all local
temperatures. It is assumed, that the operation of SOFCs is opti-
mized according to the average solid temperature.
The maximum temperature gradient is the maximum value of the
quotient of the temperature difference between adjacent CVs and
their length. The calculation is performed for all CVs in order to find
the maximum value. In reality this value can hardly be measured.
137
5 Modeling
The fuel utilization is defined as the ratio of the hydrogen equivalent

mol flows at the cell inlet and outlet.
EqH 2 , out
UF = 1 −
EqH 2 ,in
with EqH 2 = yH 2 + yCO + 4 ⋅ yCH 4 + 6 ⋅ yC 2 H 4 Eq. 36
+ 18 ⋅ yC7 H 8 + 4 ⋅ yC 2 H 4O2 + 17 ⋅ yC7 H 8O
The hydrogen equivalent mol flow, EqH2, is calculated for the inlet
and outlet fuel composition assuming that all hydrocarbons are com-
pletely reformed to hydrogen and carbon monoxide and that all car-
bon monoxide is shifted to hydrogen.
The air utilization is defined as the ratio of the oxygen mol flows at the
cell inlet and outlet. It is equal to the quotient of the fuel utilization and
the air-to-fuel ratio.
The air-to-fuel ratio is defined as the ratio between the effective and
the stoichiometry oxygen mol flow for total combustion.
The average current density is calculated as the quotient of the cur-
rent produced by the cell or the considered part of the cell and the
overall area of the cell or the considered part of the cell. This defini-
tion accounts for the fact, that in some cases not the entire cell area
participates in the electrochemical reactions e.g. areas which are
covered by the interconnector plates, see section 5.3.1.3. The aver-
age current density is computed for the total current and for the cur-
rents originating from hydrogen or carbon monoxide conversion sepa-
rately.
The direct current (DC) power produced is calculated as the product
of total produced current and the according operational voltage.
The electrochemical efficiency is defined as the ratio between pro-
duced DC power and the change of enthalpy of the fuel gas.
The fuel cell efficiency is defined as the ratio between DC power and
LHV input to the cell or considered part of the cell, Eq. 182.
The developed model consists of three coupled sub-models, namely the
electrochemical performance model, the mass balance model and the
energy balance model. The chosen modular structure of the developed
model is highly flexible with respect to the integration of further reactions
and species.
Figure 5-4 depicts the model structure and the corresponding iterative
solution algorithm. The three sub-models are coupled via the fuel com-
138
5 Modeling
position and the temperatures of the gases and the solid structure yield-
ing a highly non-linear equation system. The solution can therefore only
be found iteratively.
Figure 5-4: Solution of Initial guess for

antecedent iteration Initial guess first iteration
Model struc- equal initial guess Gas and solid temperature as well as current defined by user
ture and itera- of actual iteration
density distribution for all control volumes (CVs)
tive solution
Values for actual CV: Values for first CV:
algorithm Gas temperatures at antecedent CV outlet Gas temperatures equal inlet temperatures
Gas composition at antecedent CV outlet Gas composition equals inlet fuel composition
Electrochemical performance model Gas composition

solid and gas temperatures
Input: Gas composition, solid and gas temperatures
of antecedent CV and operational voltage
Diffusion coefficients
Ohmic resistance
Output: Current density of actual CV
Mass balance model

Gas composition
solid and gas temperatures
Input: Gas composition, solid temperature,
gas temperatures and current density of antecedent CV
Output: Gas composition, heat source terms

Gas and
Diffusion coefficients
and convective heat streams due to Heats of reaction material
coupled heat and mass transport of actual CV Heat capacities property
models
Energy balance model

Gas composition
Input: solid and gas temperatures
Solid and gas temperatures, gas composition,
heat source terms and convective heat stream due to
coupled mass and heat transport of adjecedent CVs
Output:
Thermal conductivities
Pure convective heat streams between solid structure Heat capacities
and gases, solid temperature and gas temperatures
of the actual CV
Last control volume ?

No
Yes
Gas and solid temperatures as well as current

density and gas composition distribution
equal to the precedent iteration result ?
No
Yes
Solution found
139
5 Modeling
In this work, the modeling package "ATHENA Visual Studio 11.0" was
used to implement the numerical code in the FORTRAN language.
ATHENA includes strong numerical solver algorithms specifically devel-
oped for this kind of numerical problem. Furthermore, ATHENA models
can be easily linked with state-of-the-art flow sheeting software pack-
ages such as e.g. ASPEN PLUS. A more detailed discussion of the
used modeling package can be found elsewhere, [185].
The electrochemical performance model calculates the current density
for a given fuel composition and operational voltage in a control volume.
The core of the model is the voltage balance according to which the
Nernst voltage minus the current density-dependent voltage losses has
to equal the operational voltage.
The mass balance model requires the current density as input for the
calculation of the conversion rates of the electrochemically active spe-
cies. According to the afore mentioned assumptions, these are hydro-
gen, water, carbon monoxide and carbon dioxide. Furthermore, the
mass balance model includes the calculation of reaction rates for all
considered homogenous and heterogeneous reactions based on applied
kinetic models. Besides the calculation of the conversion rates, the
mass balance model also calculates the related heat source terms and
the convective heat transport terms, which are coupled to the mass
transport to and from the catalytic surface.
The energy balance model serves for the calculation of the effective
temperatures in the solid structure and the gas channels based on the
heat source terms and the convective heat transport terms stemming
from the mass balance model. Furthermore, the energy balance model
includes the calculation of the purely convective heat transfer between
the gases and the solid structure as well as the conductive heat trans-
port within the solid structure.
5.3.1.2 Gas and solid material properties

5.3.1.2.1 Thermodynamic properties of gases
The employed thermodynamic property values are the basis of any nu-
merical model of a technical process or apparatus. Much reliable re-
search has been conducted in this area leading to consistent correla-
tions. The final choice of correlation depends on the process tempera-
ture and the temperature variation, which occurs during the process. For
processes that are operated in a narrow temperature band it is valid to
140
5 Modeling
assume temperature independent material properties, hence applying

fixed values. However this approach is not applicable to the present
model due to the operational temperatures, which vary from 650 to
1000 °C depending on the cell designs which had to be covered by the
model.
For the estimation of the thermodynamic properties of the pure gas spe-
cies, correlations according to the DIPPR 801 standard (Design Institute
for Physical Property Data) were implemented as subroutines, [192].
The DIPPR standard includes the Aly-Lee heat capacity correlation,
[193], Eq. 37, where T denotes the corresponding gas temperature.
2 2
⎛ Dx ⎞ ⎛ Fx ⎞
⎜ ⎟ ⎜ ⎟
c p, x = Bx + C x ⎜ T ⎟ +E ⎜ T ⎟ Eq. 37
⎜ ⎛ x⎞
D ⎟ x
⎜ ⎛ x ⎞⎟
F
⎜ sinh ⎜ ⎟ ⎟ ⎜ cosh⎜ ⎟ ⎟
⎝ ⎝ T ⎠⎠ ⎝ ⎝ T ⎠⎠
The coefficients (B,C,D,E and F) for the considered pure specie (sub-
script x) in the model were taken from the flow sheeting package AS-
PEN PLUS, Table 5-1. Gas mixture heat capacities are calculated
through molar fraction weighting.
Table 5-1: Specie B C D E F

Coefficients used
Hydrogen 27.617 9.56 2466.0 3.76 567.6
in the DIPPR
correlation for Carbon monoxide 29.108 8.773 3085.1 8.4553 1538.2
the estimation of Carbon dioxide 29.37 34.54 1428.0 26.4 588.0
gas heat capaci-
ties Water 33.363 26.79 2610.5 8.896 1169.0
Methane 33.298 79.933 2086.9 41.602 991.96
Ethene 33.38 94.79 1596.0 55.1 740.8
Nitrogen 29.1050 8.6149 1701.6 0.1035 909.79
Toluene 58.14 286.3 1440.6 189.8 -650.43
Acetic acid 40.2 136.75 1262.0 70.03 569.7
Anisole 76.37 293.77 1605.1 217.0 751.2
Oxygen 29.103 10.04 2526.5 9.3560 1153.8
The viscosity and thermal conductivity of the pure gas species consid-
ered in the model are also computed according to DIPPR correlations,
Eq. 38 and Eq. 39, respectively.
K x ⋅ T Lx
η xv =
M N Eq. 38
1 + x + 2x
T T
141
5 Modeling
Gx ⋅ T H x
λvx =
I J Eq. 39
1 + x + x2
T T
The coefficients were taken from ASPEN PLUS and are summarized in
Table 5-2 and Table 5-3.
Table 5-2: Specie G H I J

Coefficients used
Hydrogen 0.002653 0.7452 12.0 0.0
in the DIPPR
correlation for Carbon monoxide 0.00059882 0.6863 57.13 501.92
the estimation of Carbon dioxide 3.69 -0.3838 964.0 1860000.0
gas thermal con-
ductivity Water 0.0021606 0.76839 3940.5 -445340.0
Methane 6325.2 0.43041 770400000.0 -3.84e+10
Ethene 0.00001744 1.368 439.3 -38700.0
Nitrogen 0.00033143 0.7722 16.323 373.72
Toluene 0.00002392 1.2694 537.0 0.0
Acetic acid 1.691e-6 1.6692 658.0 -95400.0
Anisole 0.00059858 0.7527 354.04 241830.0
Oxygen 0.00044994 0.7456 56.699 0.0
Table 5-3: Specie K L M N

Coefficients used
Hydrogen 1.7970e-7 0.685 -0.59 140.0
in the DIPPR
correlation for Carbon monoxide 1.1127e-6 0.5338 94.7 0.0
the estimation of Carbon dioxide 2.148e-6 0.46 290.0 0.0
gas viscosity
Water 6.1839e-7 0.67779 847.23 -73930.0
Methane 5.2546e-7 0.59006 105.67 0.0
Ethene 2.0789e-6 0.4163 352.7 0.0
Nitrogen 6.5592e-7 0.6081 54.714 0.0
Toluene 8.7268e-7 0.49397 323.79 0.0
Acetic acid 2.7449e-8 1.0123 7.4948 0.0
Anisole 1.7531e-7 0.72 176.17 0.0
Oxygen 1.1010e-6 0.5634 96.3 0.0
The calculation of the viscosity of mixtures is undertaken according to

the Wilke approximation of the Chapman-Enskog equation, Eq. 40, [194].
2
⎛ ⎛ η v ⎞0.5 ⎛ M ⎞ 0.25 ⎞
⎜1 + ⎜ i ⎟ ⎜ j ⎟ ⎟
⎜ ⎜η v ⎟ ⎜ M ⎟ ⎟
yi ⋅ηiv ⎝ j⎠ ⎝ i⎠ ⎠
η mix
v
=∑ with φij = ⎝ Eq. 40
i ∑ y j ⋅ φij
j
⎛ ⎛ M ⎞⎞
⎜ 8 ⋅ ⎜1 + i ⎟ ⎟
0.5
⎜ ⎜ M j ⎟⎟
⎝ ⎝ ⎠⎠
142
5 Modeling
Eq. 40 involves the molar fractions of the mixture, y, the pure specie vis-
cosity, ηv, and the interaction parameter φij which further involves the
molar weight of the pure species, M.
The calculation of the thermal conductivity of mixtures is done using the
Wassiljewa-Mason-Saxena mixing rule, Eq. 41, [194].
0 .5
⎛ ⎛ η v ⎞ 0.5 ⎛ M ⎞
0.25
⎞
⎜1 + ⎜ i ⎟ ⎜ j ⎟ ⎟
⎜ ⎜η v ⎟ ⎜ M ⎟ ⎟
λvmix = ∑
yi ⋅ λvi
with Aij = ⎝ ⎝ j⎠ ⎝ i ⎠ ⎠ Eq. 41
i ∑ y j ⋅ Aij
j
⎛ ⎛ M ⎞⎞
⎜ 8 ⋅ ⎜1 + i ⎟ ⎟
0.5
⎜ ⎜ M j ⎟⎟
⎝ ⎝ ⎠⎠
Besides the already mentioned variables, Eq. 41 involves the interaction

parameter Aij. The implemented subroutines were checked against AS-
PEN PLUS results and proved correct.
In order to enable the prediction of diffusion limited phenomena, the de-
veloped model includes a subroutine for the calculation of effective dif-
fusion coefficients. Two diffusion mechanisms were considered to de-
scribe the diffusion of gases through porous media such as the elec-
trodes of fuel cells. The first mechanism is the Knudsen diffusion, which
is of importance when the mean free path length of the gas molecules is
large, compared to the pore diameter. In this case, the molecules collide
more often with the pore walls then with each other. The corresponding
Knudsen diffusion coefficient is calculated according to Eq. 42, where
rpore denotes the average pore radius of the porous media and vav,x de-
notes the average molecular speed of the species x. The calculation of
vav,x involves the ideal gas constant R, temperature T and molecular
weight, Mi.
0 .5
2 ⎛ 8 ⋅ R ⋅T ⎞
DK , x = ⋅ rpore ⋅ν av , x with ν av , x = ⎜⎜ ⎟⎟ Eq. 42
3 ⎝ π ⋅ Mi ⎠
The second diffusion mechanism is the molecular diffusion, where the

mean free path length of the molecules is small compared to the pore
diameter. In this case, the gas molecules collide more frequently with
each other than with the pore walls.
For the calculation of the molecular diffusion coefficients two methods
were implemented in the subroutine. The first method according to the
kinetic gas theory features an average variance of 8 % and can be ap-
plied to a wide range of pure species including e.g. acetic acid. Detailed
information about the method based on the kinetic gas theory can be
143
5 Modeling
found in [194]. The second method is the empirical Fuller method de-
rived from an extensive experimental data base and featuring an aver-
age variance of 4 %. The Fuller equation, Eq. 43, comprises the tem-
perature, T, the pressure, p, the molecular weight relation between the
diffusing specie (subscript x) and the matrix specie (subscript z), Mzx,
and the atomic diffusion volumes, Vz and Vx. The atomic diffusion vol-
umes have to be determined experimentally, which is the major draw-
back of the Fuller method.
−1
0.00143 ⋅ T 1.75 ⎛ 1 1 ⎞
Dm , zx = with M zx = 2 ⋅ ⎜⎜ + ⎟⎟ Eq. 43
(
p ⋅ M zx0.5 ⋅ Vz0.33 + Vx0.33 )
2
⎝ Mz Mx ⎠
Based on the atomic diffusion volumes, which are only available for very
common species such as e.g. hydrogen, carbon monoxide, water etc.,
the Fuller method is less flexible than the kinetic gas theory method.
Approximation correlations are available for the atomic diffusion volume
of more exotic species, [194], but the impact on the accuracy of the re-
sults is unknown. Nevertheless, the Fuller method was chosen as stan-
dard method, the reason being that the partial pressures of the exotic
species considered in the developed model are expected to be small
compared to the common species. The employed atomic diffusion vol-
umes are given in Table 5-4.
Table 5-4: Fuller atomic diffusion volume Molecular weight

Specie
Fuller atomic [m3/mol] [g/mol]
diffusion volumes
Hydrogen 6.12 2.01588
and molecular
weight of consid- Carbon monoxide 18.0 28.0104
ered species Carbon dioxide 26.9 44.0098
*Value approxi- Water 13.1 18.01528
mated Methane 25.14 * 16.04276
Ethene 41.04 * 28.05376
Nitrogen 18.5 28.01348
Toluene 111.48 * 92.14052
Acetic acid 53.26 * 60.05256
Anisole 117.59 * 108.13992
Oxygen 16.3 31.9988
Eq. 43 only allows for the calculation of binary diffusion coefficients. In

order to estimate the molecular diffusion coefficient of a specific specie
in gas mixtures, the following mixing rule was implemented, Eq. 44,
[194].
144
5 Modeling
1 − yx
Dm,mix , x =
y
∑i D i Eq. 44
m ,i , x
Eq. 44 comprises the molar fraction of the diffusing specie, yx, the molar
fractions of the all species in the gas mixture, yi, and the binary molecu-
lar diffusion coefficients of the diffusing specie with all other species in
the gas mixture, Dm,i,x.
5.3.1.2.2 Temperature-dependent solid material properties

The materials used in solid oxide fuel cells have to fulfill different tasks
hence influencing the behavior of the cell. The most important properties
in this respect are the electrical and the ionic conductivity, which directly
impact the power output of the cell. As a general rule, the ceramic mate-
rials employed in solid oxide fuel cells feature only poor electrical and
ionic conductivities at room temperatures. At elevated temperatures
however the conductivity improves. The relation between specific con-
ductivity and temperature was mathematically formulated in [191] for
state-of-the-art materials. Equations Eq. 45 to Eq. 48 give the specific
conductivities as functions of the temperature for the anode (ρan), elec-
trolyte (ρel), cathode (ρca) and interconnect (ρic) materials, respectively.
95.0 ⋅ 106 ⎛ − 1150.0 ⎞
ρ an = ⋅ exp⎜ ⎟ Eq. 45
TsK ⎝ TsK ⎠
⎛ 10300.0 ⎞
ρ el = 3.34 ⋅104 ⋅ exp⎜ ⎟ Eq. 46
⎝ TsK ⎠
42.0 ⋅ 106 ⎛ − 1200.0 ⎞

ρca = ⋅ exp⎜ ⎟ Eq. 47
TsK ⎝ TsK ⎠
9.3 ⋅106 ⎛ − 1100.0 ⎞

ρic = ⋅ exp⎜ ⎟ Eq. 48
TsK ⎝ TsK ⎠
The variable TsK denotes the solid material temperature. The corre-
sponding curves are depicted in Figure 5-5.
It can be seen, that even at very elevated temperatures the electric con-
ductivity of the employed materials for the anode and cathode elec-
trodes as well as for the current collectors (also called interconnectors,
IC) are three orders of magnitude higher than the ionic conductivity of
the electrolyte. Further it can be observed, that small temperature varia-
tions from e.g. 1100 K to 1300 K result in an increase of the ionic con-
ductivity by a factor of five.
145
5 Modeling
Figure 5-5: 40000 20

Specific electric Anode conductivity Cathode conductivity
conductivity of Ceramic IC conductivity Electrolyte ionic conductivity
anode, cathode
Specific ionic conductivity [1/Ohm m]

and ceramic inter- 30000 15
Specific electric conductivity

connect material
(left scale) and
[1/Ohm m]
ionic conductivity 20000 10
of electrolyte
(right scale)
10000 5
0 0
300 400 500 600 700 800 900 1000 1100 1200 1300 1400
Temperature [K]
5.3.1.3 Geometries and discretization

The range of fuel cell designs under development mainly differs in two
aspects, which are the geometry and the materials employed in the fuel
cell components. Concerning the term geometry, it is useful to distin-
guish between the macro and the micro geometry of fuel cells. The mi-
cro geometry describes the construction of the anode-electrolyte-
cathode (AEC) assembly, while the macro geometry describes the con-
struction of the cell itself.
The micro geometry has a strong influence on the electrochemical per-
formance of the fuel cell through ohmic losses and transport limitations
due to the diffusion of the reacting species through the electrodes to the
electrolyte where the electrochemical reactions take place.
The macro geometry of a given fuel cell design has an impact on the
electrochemical performance through ohmic losses occurring in the cur-
rent conducting components of the fuel cell, the heat balance through
convective and conductive heat exchange processes and the mass bal-
ance through the catalytically active surfaces areas.
Hence, micro and macro geometry have a strong bearing on the overall
behavior of the fuel cell. Accounting for this was one of the most impor-
tant requirements regarding the developed model. The model had to be
capable of representing fuel cells with different macro geometries, such
as tubular and planar fuel cells, and different micro geometries, such as
electrolyte-supported and anode-supported planar fuel cells. The cho-
sen approach for the geometry description is based on characteristic
lengths and areas. The according definitions of characteristic lengths
and areas for the considered geometries are presented in the following.
146
5 Modeling
5.3.1.3.1 Planar geometry

Figure 5-6 depicts the considered planar geometry. The planar cell stack
consists of bipolar plates, which also form the rectangular gas channels.
The bipolar plates are separated by AEC assemblies. The surface of the
AEC assemblies is consequently partially covered by the ribs of the bi-
polar plates. As assumed above, the cell is divided into repeating struc-
tures according to the number of gas channels.
Figure 5-6:
Generalized pla- Cell width
nar cell geometry

and repeating Cell width
structure descrip-
tion Interconnect
Anode
Electrolyte
Cathode
Fu
el
ga
s
Ox Cell length
ida
nt
gas Macro and micro
geometry of the
planar cell
repeating structure
Interconnect
Anode channel
Interconnect
Cathode channel height
height
on anode side
Channel width
Rib width
Anode channel
Thickness of anode
AEC assembly Thickness of electrolyte
Thickness of cathode
Cathode channel
Cathode
channel
Interconnect
height
Anode channel height on
cathode side
Area coverd by rib
The micro geometry of the planar cell design is entirely expressed by

the thicknesses of the anode, δp,an, electrolyte, δp,el, and cathode, δp,ca.
The complete description of the macro geometry includes the number of
gas channels, nch, the cell length, lc, and width, wce, the anode and
cathode gas channel heights, hp,an, hp,ca, and width, wc, and the inter-
connect or bipolar plate height on the anode and cathode side, hic,an,
hic,ca. Based on these geometrical values, the following characteristic
lengths and areas were defined for the repeating structure of the planar
geometry:
Total active area, Ap,act
Ap , act = (wce nch ) ⋅ lc Eq. 49
147
5 Modeling
Electrochemically active length, lp,elact, where xp,elact denotes the frac-

tion of the area covered by a rib participating in electrochemical reac-
tions
l p , elact = wc + ((wce nch ) − wc ) ⋅ x p , elact Eq. 50
Chemically active length, lp,chact, where xp,chact denotes the fraction of

the area covered by a rib contributing to chemical reactions
l p , chact = wc + ((wce nch ) − wc ) ⋅ x p , chact Eq. 51
Hydraulic diameter of anode gas channel, dp,hyd,an

d p ,hyd ,an = ( 2 ⋅ wc ⋅ h p ,an ) /( wc + h p ,an ) Eq. 52
Hydraulic diameter of cathode gas channel, dp,hyd,ca

d p ,hyd ,ca = ( 2 ⋅ wc ⋅ h p ,ca ) /( wc + h p ,ca ) Eq. 53
Circumferential length of anode gas channel perpendicular to gas

flow direction, lp,circ,an
l p ,circ ,an = 2 ⋅ ( wc + h p ,an ) Eq. 54
Circumferential length of cathode gas channel perpendicular to gas

flow direction, lp,circ,ca
l p ,circ ,ca = 2 ⋅ ( wc + h p ,ca ) Eq. 55
Cross-sectional area of solid structure perpendicular to gas flow di-

rection, Ap,cross
Ap , cross = (wce nch ) ⋅ (hic , an + δ p , an + δ p , el + δ p , ca + hic , ca ) − wc ⋅ (hp , an + hp , ca ) Eq. 56
5.3.1.3.2 Tubular geometry

Figure 5-7 depicts the tubular geometry based on the cell design pro-
moted by the Siemens AG. The standard tubular cell consists of a fuel
cell tube with one closed end and a concentrical air delivery tube (ADT).
The annular gap between the outer surface of the ADT and inner sur-
face of the fuel cell tube forms the cathode gas channel. The anode gas
flows around the outer surface of the fuel cell tube. The geometry of the
anode gas channel is not defined by the cell design itself but rather by
the arrangement of the single fuel cell tube in the whole stack. For mod-
eling purpose however, an annular gap shaped anode gas channel was
148
5 Modeling
assumed. In the case of the tubular geometry, the fuel cell tube itself
was considered as a repeating structure in the complete fuel cell stack.
Similar to the planar cell geometry, the micro geometry of the tubular
cell design is entirely described by the thicknesses of the anode, δt,an,
the electrolyte, δt,el, and the cathode, δt,ca. The complete description of
the tubular cell design macro geometry encompasses the active tube
length, lt, the inner, ri,ADT, and outer radius, ro,ADT, of the air delivery
tube, the inner radius of the fuel cell tube, ri,FC, and the height of the vir-
tual anode channel, ht,an.
Figure 5-7: Tube length

Standard tubular
cell geometry
Fuel gas Oxidant

gas
Part of cathode
Macro and micro tube covered
geometry of the
tubular cell
by interconnect
repeating structure
Inner radius
of air delivery tube Thickness Virtual anode
Outer radius of cathode channel height
of air delivery tube
Thickness Thickness
Inner radius of cathode layer of electrolyte of anode
Similar to the planar geometry, the following characteristic lengths and

areas were defined for the repeating structure of the tubular geometry:
Middle radius of the AEC assembly, rm,AEC
rm , AEC = ri ,FC + (δ t ,ca + δ t ,el + δ t ,an ) / 2 Eq. 57
Outer radius of the cell tube, ro,FC

ro ,FC = ri ,FC + δ t ,ca + δ t ,el + δ t ,an Eq. 58
Total active area, At,act

At ,act = 2π ⋅ rm , AEC ⋅ lt Eq. 59
Electrochemically active circumferential length, lt,elact, where xt,ic de-

notes the fraction of the cell tube surface covered by the interconnect
lt ,elact = 2π ⋅ rm , AEC ⋅ (1 − xt ,ic ) Eq. 60
149
5 Modeling
Chemically active circumferential length, lt,chact, where xt,ic denotes

the fraction of the cell tube surface covered by the interconnect
lt , chact = 2π ⋅ ro , FC ⋅ (1 − xt ,ic ) Eq. 61
Hydraulic diameter of air delivery tube, dt,hyd,ADT

d t ,hyd , ADT = 2 ⋅ ri , ADT Eq. 62
Hydraulic diameter of virtual anode gas channel, dt,hyd,an

d t ,hyd ,an = 2 ⋅ ht ,an Eq. 63
Hydraulic diameter of cathode gas channel, dt,hyd,ca

d t ,hyd ,ca = 2 ⋅ (ri ,FC − ro , ADT ) Eq. 64
Circumferential length of air delivery tube perpendicular to gas flow

direction, lt,circ,ADT
lt ,circ , ADT = π ⋅ ( ro , ADT + ri , ADT ) Eq. 65
Circumferential length of virtual anode gas channel perpendicular to

gas flow direction, lt,circ,an
lt ,circ ,an = 2π ⋅ ro ,FC ⋅ (1 − xt ,ic ) Eq. 66
Circumferential length of cathode gas channel perpendicular to gas

flow direction, lt,circ,ca
lt ,circ ,ca = 2π ⋅ ri ,FC Eq. 67
Cross-sectional area of solid structure perpendicular to gas flow di-

rection, At,cross
At ,cross = (π ⋅ ro2,FC ) − (π ⋅ ri 2,FC ) Eq. 68
5.3.1.3.3 Triangular geometry

Figure 5-8 depicts the triangular geometry based on the latest Siemens
AG cell design referred to as Delta8 design (D8). Similar to the standard
tubular cell design, the D8 employs centered air delivery tubes (ADT) for
the injection of the cathode air. The D8 cell is composed of eight con-
nected triangular-shaped fuel cell tubes with closed ends, Figure 5-8a.
The spaces between the ADTs and inner surface of a triangular fuel cell
tubes form the cathode gas channels, Figure 5-8d.
150
5 Modeling
Figure 5-8:
Triangular cell
geometry
The outer upper surface of the D8 cell and the outer lower surface of the
next D8 cell form the anode gas channels, Figure 5-8d. Note that the D8
cells are stacked like planar cells, Figure 5-8a. For better contacting and
current distribution, a porous contact layer is placed between the indi-
vidual cells. Due to its high electrical conductivity and porosity, the con-
tact layer was not considered in the developed model. The D8 repeating
structure was defined as one triangular-shaped fuel cell tube.
Similar to the above discussed geometries, the micro geometry of the
triangular cell design is entirely described by the thicknesses of the an-
ode, δD8,an, the electrolyte, δD8,el, and the cathode, δD8,ca, Figure 5-8c.
The description of the triangular cell design macro geometry is more
complex than that of the standard tubular and planar cell design. How-
ever, the complexity can be reduced by introducing geometrical de-
pendencies for the curvature radii. For the model, it was assumed, that
the upper outer curvature radius ro equals 1.5 times the D8 cell tube
wall thickness δ. The lower outer curvature radius rm was assumed
equal to the cell tube wall thickness and the inner curvature radius ri
equal to half the cell tube wall thickness, Figure 5-8b.
Based on the above assumptions and the overall cell width, wD8, the cell
length, lD8, the upper triangle half-angle, α, and the number of triangu-
151
5 Modeling
lar-shaped fuel cell tubes forming the D8 cell, ntri, the following charac-
teristic lengths and areas were derived for the triangular geometry:
Electrochemically and chemically active width, lD8,elact
2δ 90 − α
lD8,elact = lD8,chact = + ⋅π ⋅δ + 2 ⋅ X
ntri ⋅ cos α ntri ⋅ 90
Eq. 69
w 5 δ 90 − α
with X = D8 − ⋅ + ⋅π ⋅δ
ntri 2 tan α 72
The active width of one control volume is the arithmetic middle of the
overall active width of the D8 cell and the number of triangular cell tubes.
Note that the additional electrochemically active areas of the lower cur-
vature of the two outer triangular cells were also considered, Figure 5-8d.
Total active area, AD8,act
AD 8,act = lD 8,elact ⋅ lD 8 Eq. 70
Circumferential length of ADT perpendicular to gas flow, lD8,circ,ADT

lD 8, circ, ADT = π ⋅ (rD 8, o , ADT + rD 8, i , ADT ) Eq. 71
Circumferential length of anode channel perpendicular to gas flow,

lD8,circ,an
wD8
lD 8, circ, an = lD8, elact + Eq. 72
ntri
Circumferential length of cathode channel perpendicular to gas flow,

lD8,circ,ca
sin α + 1 ⎛ wD 8 δ + 2 ⋅ sin α ⎞ 90 + α 3⋅δ
lD 8, circ, ca = ⋅⎜ − ⎟⎟ + ⋅π ⋅δ − Eq. 73
sin α ⎜⎝ ntri cos α ⎠ 120 tan (45 − α 2)
Cross-sectional area of cathode channel, AD8,cross,ca

X2 3 ⎛ 90 + α ⎛ α ⎞⎞
AD8,cross,ca = + ⋅ δ 2 ⋅ ⎜⎜ ⋅ π − cot⎜ 45 − ⎟ ⎟⎟ − π ⋅ rD28,o , ADT
4 ⋅ tan α 4 ⎝ 360 ⎝ 2 ⎠⎠
Eq. 74
⎛w δ + 2 ⋅ sin α ⎞
with X = ⎜⎜ D 8 − ⎟
⎝ ntri cos α ⎟⎠
152
5 Modeling
Cross-sectional area of anode channel, AD8,cross,an

9 5 90 − α δ2
AD8,cross ,an = X 2 ⋅ tan α + Y ⋅ ⋅ δ 2 − ⋅ π ⋅ δ 2 ⋅ −
4 4 180 tan α
wD 8 ⋅ cos α + ntri ⋅ δ ⋅ (sin α − 2 )
with X = Eq. 75
2 ⋅ ntri ⋅ sin α
2 ⋅ sin α + cos α − 1 − sin 2 α
and Y =
sin α ⋅ cos α
Cross-sectional area of solid structure, AD8,cross

wD8
AD 8,cross = ⋅ X − AD 8,cross ,ca − AD8,cross ,an − π ⋅ rD28,o , ADT
ntri
Eq. 76
⎛ w ⋅ cos α δ 3 ⎞
with X = ⎜⎜ D 8 − + ⋅ δ ⎟⎟
⎝ 2 ⋅ ntri ⋅ sin α sin α 2 ⎠
Hydraulic diameter of ADT, dD8,hyd,ADT

d D 8, hyd , ADT = 2 ⋅ rD 8, i , ADT Eq. 77
Hydraulic diameter of anode gas channel, dD8,hyd,an

4 ⋅ AD 8,cross ,an
d D 8,hyd ,an = Eq. 78
lD 8,circ,an
Hydraulic diameter of cathode gas channel, dD8,hyd,ca

4 ⋅ AD8,cross ,ca
d D8,hyd ,ca = Eq. 79
lD8,circ,an + 2 ⋅ π ⋅ rD 8,i , ADT
5.3.1.4 Electrochemical performance model

The electrochemical performance model calculates the current density
for a given fuel composition and operational voltage.
Figure 5-9:
Equivalent circuit Hydrogen electrochemistry
Operational
voltage
of the electro-
Current originating from
chemical per- Nernst
hydrogen conversion
voltage
formance model
Activation Diffusion Oxygen electrochemistry
polarization polarization
Current originating from
hydrogen and CO conversion
Carbon monoxide electrochemistry Activation Diffusion Ohmic

polarization polarization polarization
Nernst Current originating from
carbon monoxide conversion
voltage
Activation Diffusion
polarization polarization
153
5 Modeling
To meet the requirement of high fuel flexibility, the electrochemical con-

version of hydrogen and carbon monoxide was considered to occur in
parallel. Furthermore, activation, ohmic and diffusion losses were con-
sidered.
Figure 5-9 shows the equivalent circuit of the electrochemical perform-
ance model, assuming that the voltage losses can be treated as electri-
cal resistors. The corresponding equation system is highly non-linear
and a solution for the current density can only be found numerically.
The starting point for the calculation of the current density is the reversi-
ble potential of the given fuel gas composition, commonly referred to as
the Nernst voltage. The Nernst voltage of the hydrogen oxidation,
ENernst,H2, is calculated according to Eq. 80 and for the carbon monoxide
oxidation, ENernst,CO, following Eq. 81.
⎛ R ⋅ TsK ⎞ ⎛⎜ ⎛ pH ⋅ pO 0.5 ⎞ ⎞
⎜ 2 ⎟⎟
E Nernst ,H 2 = ⎜ ⎟ ⋅ ln Kp + ln 2
Eq. 80
⎝ n⋅F ⎠ ⎝ ⎜ H 2 oxi
⎜ p ⎟⎟
⎝ H 2O ⎠⎠
⎛ R ⋅ TsK ⎞ ⎛⎜ ⎛ pCO ⋅ pO 0.5 ⎞ ⎞

⎟⎟
E Nernst ,CO =⎜ ⎟ ⋅ ln KpCO oxi + ln⎜ 2
Eq. 81
⎝ n ⋅ F ⎠ ⎜⎝ ⎜ p ⎟⎟
⎝ CO2 ⎠⎠
In Eq. 80 and Eq. 81, R denotes the ideal gas constant, TsK stands for
the solid structure temperature, n represents the number of transferred
electrons (in both cases 2), F is the Faraday constant, Kp is the equilib-
rium constant and the factors pi represent the partial pressures of the
products and educts of the oxidation reactions in the bulk gas phase.
In this work, the equilibrium constant values for the hydrogen and car-
bon monoxide oxidation reaction were taken from [195]. The tabulated
values were compiled to yield two fit correlations, which both take the
form of Eq. 82.
⎛ − TsK ⎞ ⎛ − TsK ⎞ ⎛ − TsK ⎞
X = y0 + A1 exp⎜⎜ ⎟⎟ + A2 exp⎜⎜ ⎟⎟ + A3 exp⎜⎜ ⎟⎟ Eq. 82
⎝ t1 ⎠ ⎝ t2 ⎠ ⎝ t3 ⎠
X denotes either the equilibrium constant of the hydrogen oxidation or of

the carbon monoxide oxidation, depending on the employed set of fit
correlation coefficients, Table 5-5.
Figure 5-9 shows that the Nernst voltages of the hydrogen and carbon
monoxide oxidation minus the according current density-dependent
voltage losses have to equal the operational voltage of the fuel cell (Eop),
Eq. 83 and Eq. 84.
154
5 Modeling
Table 5-5: Coefficient Hydrogen oxidation Carbon monoxide oxidation

Coefficients of
y0 -1.093 -3.020
the equilibrium
constant fit corre- A1 42.296 376.372
lation for hydro- t1 1297.858 94.704
gen and carbon
monoxide oxida- A2 106.613 143.468
tion t2 319.283 284.902
A3 284.347 51.766
t3 107.556 1179.242
Eop = E Nernst ,H 2 − η act ,H 2 − η diff ,H 2 − η act ,O2 − η diff ,O2 − η ohm Eq. 83
Eop = E Nernst ,CO − η act ,CO − η diff ,CO − η act ,O2 − η diff ,O2 − η ohm Eq. 84
Consequently the developed model must distinguish between currents

originating from either the oxidation of hydrogen, IH2, or of carbon mon-
oxide, ICO, which sum up to the total produced current, Itot, Eq. 85.
I tot = I H 2 + I CO Eq. 85
From Eq. 83 and Eq. 84 or Figure 5-9 it can be seen, that a current
originating from carbon monoxide oxidation can only be produced, when
the corresponding Nernst voltage minus the voltage losses of the oxy-
gen electrochemistry and the ohmic losses, due to an already flowing
current originating from hydrogen oxidation, still yields a positive value.
The same applies vice versa.
5.3.1.4.1 Activation polarization

Even though no current is drawn from the fuel cell, the electrochemical
reactions, Eq. 86, Eq. 87 (anode) and Eq. 88 (cathode), are taking place.
H 2 + O 2 − ↔ H 2 O + 2e − Eq. 86
CO + O 2− ↔ CO2 + 2e − Eq. 87
O2 + 4e − ↔ 2O 2 − Eq. 88
Without electrical load however, the reactions occur at equal rates in

both directions. This exchange current represents the current flowing in
one of the two directions at equilibrium conditions. In order to generate a
current into one direction a certain potential is required. This potential is
commonly referred to as activation polarization, nact, and mathematically
best described by the Butler-Volmer equation, Eq. 89, where I denotes
the prevalent current density, [196]:
155
5 Modeling
⎡ ⎛ n ⋅ F ⋅ηact ⎞ ⎛ n ⋅ F ⋅ηact ⎞⎤
I = I 0 ⎢exp⎜ β ⋅ ⎟ − exp⎜ − (1 − β ) ⋅ ⎟ Eq. 89
⎣ ⎝ R ⋅ TsK ⎠ ⎝ R ⋅ TsK ⎠⎥⎦
Other correlations such as the Tafel equation, Eq. 90, the linear current-
potential equation, Eq. 91, and more empirical relations can also be
found in the literature and are mostly derivates of the Butler-Volmer
equation. A detailed discussion of the application of these simplified
equations is given in [141].
R ⋅ TsK ⎛I ⎞
η act = ⋅ ln⎜⎜ ⎟⎟ Eq. 90
n ⋅ F ⋅ β ⎝ I0 ⎠
R ⋅ TsK I
η act = ⋅ Eq. 91
n ⋅ F I0
The Butler-Volmer equation, Eq. 89, which was implemented in the de-
veloped model, comprises two important model parameters. The first
parameter is the transfer coefficient β, which represents the part of the
change in polarization leading to a change in the reaction rate constant.
For fuel cell applications and in the developed model, the value is usu-
ally assumed to be 0.5, [182, 196].
The second parameter is the exchange current density I0, which repre-
sents the forward and reverse electrode reaction rate at the equilibrium
potential. High exchange current density values denote a high electro-
chemical reaction rate and hence low activation losses. More detailed
information can be found in [197].
The literature discusses a vast number of models used to calculate the
exchange current density at the anode and cathode. A detailed discus-
sion of these models is given in [182]. In the present work, the anode
exchange current density is calculated according to Eq. 92 for the hy-
drogen electrochemistry and Eq. 93 for the carbon monoxide electro-
chemistry. The exponent near to the stoichiometry of the anodic electro-
chemical reactions results in a direct proportional dependence of the
anode exchange current density to the educt and an inversely propor-
tional dependence to the product partial pressures. With the inverse de-
pendency of the activation losses to the exchange current density, the
exponents yield lower activation losses with increased educt or de-
creased product partial pressures and vice versa, [182]. The cathode
exchange current density is given by Eq. 94. The positive exponent
yields lower cathode activation losses with rising oxygen partial pres-
sures.
156
5 Modeling
−0.5
⎛ pH ⎞⎛ pH 2 O ⎞ ⎛ − Eact , an , H 2 ⎞
I 0, an , H 2 = γ an , H 2 ⎜⎜ 2 ⎟⎟⎜⎜ ⎟⎟ ⋅ exp⎜⎜ ⎟⎟ Eq. 92
⎝ p ⎠⎝ p ⎠ ⎝ R ⋅ TsK ⎠
−0.5
⎛ p ⎞⎛ pCO2 ⎞ ⎛ − Eact , an ,CO ⎞
I 0, an ,CO = γ an ,CO ⎜⎜ CO ⎟⎟⎜⎜ ⎟⎟ ⋅ exp⎜⎜ ⎟⎟ Eq. 93
⎝ p ⎠⎝ p ⎠ ⎝ R ⋅ TsK ⎠
0.25
⎛ pO ⎞ ⎛ − Eact ,ca ⎞
I 0, ca = γ ca ⎜⎜ 2 ⎟⎟ ⋅ exp⎜⎜ ⎟⎟ Eq. 94
⎝ p ⎠ ⎝ R ⋅ TsK ⎠
The material-dependent parameters, activation energies, Eact,an,H2,

Eact,an,CO, Eact,ca, and pre-exponential factors of anode, γan,H2, γan,CO,
and cathode, γca, were taken from [182] for the planar geometry and
[155] for the tubular geometry. In [155, 182], the parameters were esti-
mated based on experimental data.
As only the conversion of hydrogen was considered in [182], no parame-
ters for the carbon monoxide conversion were specified. However, in
[190] it is stated that the activation potential loss of the electrochemical
conversion of carbon monoxide is approx. 1.4 times higher than that for
the electrochemical hydrogen conversion. Based on this, the pre-
exponential factor for the calculation of the carbon monoxide conversion
exchange current density was determined.
It has to be emphasized, that due to the implicit character of the Butler-
Volmer equation its solution can only be determined numerically. This
effort is however justified by higher accuracy, [164], as the straightfor-
ward calculation of the activation polarization is only possible using sim-
plified equations such as Eq. 90 and Eq. 91.
5.3.1.4.2 Ohmic polarization

Electronic currents through the electrodes and interconnects as well as
ionic currents through the electrolyte induce voltage losses in all fuel
cells. For a given resistance and current density, the calculation of the
ohmic voltage loss (ηohm) simply follows Ohm's law, Eq. 95.
η ohm = Requiv ⋅ I tot Eq. 95
The area specific equivalent resistance (Requiv) of SOFCs depends on

the geometry and the conductivity of the current conducting components.
The temperature-dependent conductivity of the ceramic materials em-
ployed in SOFCs is calculated according to Eq. 45 to Eq. 48.
157
5 Modeling
Based on the conductivity values, the area specific equivalent resistance

of the considered cell designs is approximated in the present model us-
ing the transmission line model instead of calculating it through the nu-
merical solution of the Laplace equation. The methodology was devel-
oped by Nisancioglu, [198], and others, e.g. [199], and features a high
flexibility towards different fuel cell designs.
The general idea of the transmission line model is that complex SOFC
geometries can be described by a set of five standard cell sub-units.
Each of these cell sub-units has a characteristic current path pattern and
corresponding analytical formula for the equivalent resistance calcula-
tion, see Table 5-6. In these formulas, L denotes the length of the cell
sub-unit, δ stands for the thickness of the cell sub-unit or parts of it and
ρ represents the conductivity of the corresponding material. The sub-
scripts an, el and ca stand for anode, electrolyte and cathode.
Table 5-6: Geometry factor J Resistance R

Cell sub-unit
Equivalent resis- [-] [Ω m]
tance formulas
for five cell sub-
units with the ρL
δ - δ
arrowed lines
indicating the L 1
current flow path
from cathode to ⎛ ρ ca ⎞
ρ ca ρ elδ el ⎜⎜ ⎟⎟
anode, [198] δ ca
L ⎝ δ ca ⎠
ρ elδ el tanh J
2
⎛ ⎛ ρ ⎞2 ⎛ ρ ⎞2 ⎞
⎜ ⎜ an ⎟ + ⎜ ca ⎟ ⎟ ⋅ cosh J + ⎛⎜ ρ an ρ ca ⎞⎟ ⋅ (2 + J ⋅ sinh J )
1 ⎛ ρ an ρ ca ⎞ ⎜ ⎜⎝ δ an ⎟⎠ ⎜⎝ δ ca ⎟⎠ ⎟ ⎜δ δ ⎟
⎝ an ca ⎠
L ⎜ + ⎟ ⎝ ⎠
ρ elδ el ⎜⎝ δ an δ ca ⎟⎠ 1 ρ an ρ ca
⋅3 + ⋅ sinh J
3 ρ elδ el δ an δ ca
ρ anδ an + ρ elδ el + ρ caδ ca

- L
⎛ ρ an ρ ca ⎞
⎛ ρ an ρ ca ⎞ ρ elδ el ⎜⎜ + ⎟
δ ca ⎟⎠
1
L ⎜ + ⎟ ⎝ δ an
ρ elδ el ⎜⎝ δ an δ ca ⎟⎠
tanh J
5
Application to the tubular geometry

Figure 5-10 shows the cross section with current paths of the tubular
geometry considered in the developed model (a). The transmission line
model approximation (b) consists of two type 3 cell sub-units connected
158
5 Modeling
in parallel, representing the two halves of the cell tube, and one type 2
cell sub-unit, representing the interconnect area, connected in series.
Figure 5-10:
Cross section of
the standard tubu- Interconnect
lar Siemens AG
design (a) and a Cathode
b
transmission line
model (b)
Anode
Electrolyte
Due to the perfect symmetry of the tubular geometry, the equivalent re-
sistance is only calculated for one half-cell and then divided by 2. The
length of the type 3 cell sub-unit (Le) for one half-cell is calculated ac-
cording to Eq. 96 and for cell sub-unit type 2 (Lic) according to Eq. 97.
Le = π ⋅ rm , AEC ⋅ (1 − xt ,ic ) Eq. 96
Lic = π ⋅ rm , AEC ⋅ xt ,ic Eq. 97
The resistance of one half of the cell tube (Re) is given by Eq. 98.
⎛ ⎛ ρ ⎞2 ⎛ ρ ⎞2 ⎞
⎜ ⎜ an ⎟ + ⎜ ca ⎟ ⎟ ⋅ cosh J + ⎛⎜ ρ an ρ ca ⎞⎟ ⋅ (2 + J ⋅ sinh J )
⎜ ⎜⎝ δ an ⎟⎠ ⎜⎝ δ ca ⎟⎠ ⎟ e ⎜δ δ ⎟
⎝ an ca ⎠
e e
Re = ⎝ ⎠
1 ρ an ρca
⋅3 + ⋅ sinh J e
ρelδ el δ an δ ca Eq. 98
1 ⎛ ρ an ρ ca ⎞
with J e = Le ⎜ + ⎟
ρ elδ el ⎜⎝ δ an δ ca ⎟⎠
The resistance of the interconnect (Ric) is calculated according to Eq. 99.

⎛ ρ ca ⎞
ρicδ ic ⎜⎜ ⎟⎟
⎝ δ ca ⎠ 1 ρ Eq. 99
Ric = with J ic = Lic ⋅ ca
tanh J ic ρ elδ el δ ca
159
5 Modeling
The equivalent resistance is calculated following Kirchhoff's law for in-

series connection. In order to obtain the area specific resistance value
for the calculation of the ohmic voltage loss using a current density in-
stead of an absolute current, the equivalent resistance is multiplied with
the electrochemically active circumferential length of the cell tube.
Rt , equiv = lt , elact ⋅ ( Re + Ric ) Eq. 100
Application to the triangular geometry

The triangular geometry exhibits current paths, which can be described
using the same approach as for the standard tubular cell, Figure 5-11.
Figure 5-11:
Cross section of
the triangular
Siemens AG de-
sign (a) and
transmission line
model (b)
The cell sub-unit lengths, Le and Lic, are given by Eq. 101 and Eq. 102.
LD 8, e = 0.5 ⋅ lD 8, elact Eq. 101
wD 8
LD 8,ic = Eq. 102
2 ⋅ ntri
As only one half of the current path is considered, the area specific re-
sistance is obtained by multiplying the equivalent resistance with half of
the electrochemically active width of the repeat element, Eq. 103.
RD 8, equiv = 0.5 ⋅ lD 8,elact ⋅ ( Re + Ric ) Eq. 103
Application to the tubular geometry

The calculation of the area specific equivalent resistance for the consid-
ered planar geometry is done in the same manner as for the tubular ge-
ometry. Figure 5-12 shows a cutout of the considered planar stack and
160
5 Modeling
the cross-section of a repeating structure with the corresponding current

paths (a) and transmission line model approximation (b).
Figure 5-12:
Cross section of a b
the considered Interconnect
planar cell design Cathode channel 6: Type 1
6
(a) and transmis- Interconnect
sion line model (b)
Anode channel 4: Type 1 5: Type 5*
4 5
AEC assembly
3: Type 1
3
Cathode channel
Anode
Electrolyte 1: Type 4 2: Type 5
Anode channel 1 2 Cathode
The equivalent resistance of the repeat element includes 6 cell sub-units.

In the developed model, the current flow through the AEC assembly is
divided into a sub-unit type 4 (R1) and type 5 (R2), Eq. 104 and Eq. 105.
2 ⋅ ( ρ caδ ca + ρ elδ el + ρ anδ an )
R1 = Eq. 104
wc nch − wch
⎛ ρ an ρca ⎞
ρ elδ el ⎜⎜ + ⎟
⎝ δ an δ ca ⎟⎠ wch 1 ⎛ ρ an ρ ca ⎞ Eq. 105
R2 = with J 2 = ⎜ + ⎟
tanh J 2 2 ρelδ el ⎜⎝ δ an δ ca ⎟⎠
The two cell sub-units are connected in parallel. The interconnect con-
tacts anode and cathode of two superposed AEC assemblies. The resis-
tances of the interconnect ribs on the anode (R3) and cathode (R6) side
are calculated using type 1 cell sub-units, Eq. 106 and Eq. 107.
2 ⋅ ρic hp , an
R3 = Eq. 106
wc nch − wch
2 ⋅ ρic hp , ca
R6 = Eq. 107
wc nch − wch
The resistance of the bar between adjoining ribs is calculate by assum-

ing a parallel connection of a type 1 cell sub-unit (R4) and a modified
type 5* (R5), [200].
2 ⋅ ρic ⋅ ((hic , an + hic , ca ) − (hp , an + hp , ca ))
R4 = Eq. 108
wc nch − wch
161
5 Modeling
ρic
R5 =
⎛ ⎛ ⎛ ⎞ ⎞ ⎞⎟
0.41 ⋅ ⎜1 − exp⎜ − 0.6 ⋅ ⎜
wch ⎟⎟ Eq. 109
⎜ ⎜ ⎜ (h + h ) − (h + h ) ⎟ ⎟ ⎟
⎝ ⎝ ⎝ ic , an ic , ca p , an p , ca ⎠⎠⎠
Finally, the equivalent circuit depicted in Figure 5-12 yields Eq. 110.
Similar to the tubular geometry, the area specific resistance is multiplied
with the electrochemically active length of the repeat structure.
⎛ w ⎞ ⎛⎛ 1 ⎞
−1 −1
1 ⎞ ⎛ 1 1 ⎞
R p , equiv = ⎜⎜ ce ⎟⎟ ⋅ ⎜ ⎜⎜ + ⎟⎟ + R3 + ⎜⎜ + ⎟⎟ + R6 ⎟ Eq. 110
⎝ nch ⎠ ⎜⎝ ⎝ R1 R2 ⎠ ⎝ R4 R5 ⎠ ⎟
⎠
5.3.1.4.3 Diffusion polarization

The diffusion polarization losses account for the difference of the Nernst
voltage considering the reactant partial pressures in the bulk gas phase
and at the reaction sites located at the triple phase boundary (TPB)
where electrolyte and electrode material as well as reactant gases meet.
The partial pressures of the educts are generally lower, while those of
the products are generally higher at the TPB then in the bulk gas phase.
The exact partial pressure values depend on the prevalent current den-
sity and the material parameters of the porous electrodes through which
the educts and products have to diffuse in order to reach the TPB. The
prevalent current density determines the concentration gradient between
the TPB and the bulk gas phase. The material parameters of the porous
electrodes include tortuosity, porosity and pore radius. These three ma-
terial parameters together with the fuel composition determine the effec-
tive diffusion coefficient required in order to calculate the partial pres-
sures at the TPB through the application of e.g. Fick's law of diffusion.
Concentration losses become important for highly diluted fuel gases and
at high current densities, where they increase against an asymptotic
maximum. At this specific point a further increase of the current is im-
possible since the reactant partial pressures at the TPB equal zero due
to instant conversion.
The developed model considers diffusion limitation in the anode and the
cathode electrode. At the anode, the educts and the products of the
electrochemical reactions diffuse with equal rates to the TPB and back
into the gas flow channel. This equimolar counter-flow diffusion yields a
zero net diffusion flux. Applying Fick's law of diffusion by assuming, that
the electrochemical reaction rate equals the mass flux via diffusion, the
partial pressures of educts and products at the TPB can be calculated
162
5 Modeling
based on the partial pressures in the bulk gas phase according to

Eq. 111.
⎛ R ⋅ TaK ⋅ I H 2 ⋅ δ an ⋅τ an ⎞
piTPB = pi m ⎜ ⎟
⎜ n ⋅ F ⋅ Deff , H ⋅ ε an ⎟
⎝ 2 ⎠ Eq. 111
with − for i = H 2 , CO
and + for i = H 2O, CO2
In these equations, TaK denotes the temperature of the anode gas flow,
IH2 and ICO represent the currents originating from the hydrogen and
carbon monoxide conversion, δan stands for the thickness of the anode
electrode, τan gives the tortuosity and εan the porosity of the anode elec-
trode material and Deff,H2 and Deff,CO denote the effective diffusion coef-
ficients of hydrogen and carbon monoxide of the anode gas mixture.
The effective diffusion coefficients are calculated based on binary mo-
lecular diffusion coefficients of hydrogen and carbon monoxide with all
other considered species in the gas mixture and the Knudsen diffusion
coefficient. The calculation of binary molecular and Knudsen diffusion
coefficients is discussed in section 5.3.1.2.1. The binary molecular diffu-
sion coefficients are summed up to yield the effective molecular diffusion
coefficient according to Eq. 44. For the calculation of the effective diffu-
sion coefficient it is assumed, that Knudsen and molecular diffusion oc-
cur simultaneously. Therefore, the Bosanquet formula was implemented
in this model, Eq. 112, where x denotes either hydrogen or carbon mon-
oxide.
Dm, x − mix ⋅ DK , x
Deff , x = Eq. 112
Dm , x − mix + DK , x
For known partial pressures of the educts and products in the bulk gas
phase and at the TPB, the diffusion voltage losses are computed ac-
cording to Eq. 113. The temperature used in the calculation is the tem-
perature of the solid structure of the fuel cell and not that of the gas mix-
ture.
⎛ R ⋅ TsK ⎞ ⎛⎜ pi ⋅ p j ⎞⎟
TPB
ηdiff ,i = ⎜ ⎟ ⋅ ln⎜ with i = H 2 , CO and j = H 2O, CO2 Eq. 113

⎝ n ⋅ F ⎠ ⎝ p j ⋅ pi ⎟⎠
TPB
At the cathode, only oxygen takes part in the electrochemical reactions,

where it is consumed without forming products that return to the bulk
gas phase.
163
5 Modeling
pOTPB −p
A⋅ = exp(I tot ⋅ TcK ⋅ B )
( )
2
DK ,O2 ⋅ α ⋅ pOTPB − p − Dm ,O2 − N 2

( )
2
DK ,O2 ⋅ α ⋅ pO2 − p − Dm ,O2 − N 2

with A=
pO2 − p
τ ca ⋅ R ⋅ δ ca ⋅ (Dm ,O − N − DK ,O ⋅ (α − 1)) Eq. 114
and B = 2 2 2
ε ca ⋅ 2 ⋅ n ⋅ F ⋅ Dm,O − N ⋅ DK ,O ⋅ p
2 2 2
⎛ M O2 ⎞
and α = 1 − ⎜ ⎟
⎜ MN ⎟
⎝ 2 ⎠
Eq. 114 was implemented in this model for the implicit calculation of the
partial pressure of oxygen at the cathode TPB, [155]. In contrast to the
diffusion at the anode, the effective diffusion coefficient is calculated for
the non-dilute two-component gas mixture at the cathode according to
[201], which includes the relation between the molecular masses, α, of
the components nitrogen and oxygen. Further, Eq. 114 comprises the
porosity and tortuosity of the cathode electrode, τca and εca, the cathode
electrode thickness, δca, the Knudsen diffusion coefficient of oxygen,
DK,O2, the binary molecular diffusion coefficient of the oxygen-nitrogen
system, Dm,O2-N2, the prevalent total current density, Itot, and the tem-
perature of the cathode gas, TcK. The diffusion voltage loss at the
cathode is calculated analogously to the anode, Eq. 115.
⎛ R ⋅ TsK ⎞ ⎛⎜ pO2 ⎞
⎟
ηdiff ,O = ⎜ ⎟ ⋅ ln TPB Eq. 115
2
⎝ 2 ⋅ n ⋅ F ⎠ ⎜⎝ pO2 ⎟
⎠
5.3.1.5 Mass balance model

The current density values computed by the electrochemical perform-
ance model are related to the reactions Eq. 86 to Eq. 88 via the Faraday
law yielding the area specific reaction rates of the electrochemical reac-
tions and the number of consumed moles of hydrogen, carbon monox-
ide and oxygen per area, respectively. In the present model, the axial
length of the gas flow channels was defined as the integration variable.
In order to account for the 2-dimensional distribution of the electro-
chemical reactions, the area specific reaction rate has to be multiplied
with the electrochemically active length of the considered fuel cell de-
sign, lt,elact or lp,elact, in order to be converted into the integration length
specific reaction rate. Accordingly, Eq. 116 to Eq. 118 were applied in
this model.
164
5 Modeling
⎛ IH ⎞
rH 2 oxi = lt ⋅ ⎜⎜ 2 ⎟⎟ Eq. 116
⎝ 2⋅ F ⎠
p , elact
⎛ I ⎞
rCO oxi = lt ⋅ ⎜ CO ⎟ Eq. 117
⎝ 2⋅ F ⎠
p , elact
⎛ I ⎞
rO2 ion = lt ⋅ ⎜ tot ⎟ Eq. 118
⎝ 4⋅ F ⎠
p , elact
Besides the mentioned electrochemical reactions, this model considers

water-gas-shift, Eq. 119 and Eq. 120, and steam reforming reactions, Eq.
121 to Eq. 125, of methane, ethene, toluene and anisole as well as the
thermal decomposition of acetic acid through applied kinetic models.
CO + H 2O ↔ CO2 + H 2 Eq. 119
CO2 + H 2 ↔ CO + H 2O Eq. 120
CH 4 + H 2O ↔ CO + 3H 2 Eq. 121
C2 H 4 + 2 H 2O → 2CO + 4 H 2 Eq. 122
C7 H 8 + 7 H 2O → 7CO + 11H 2 Eq. 123
C2 H 4O2 → 2CO + 2 H 2 Eq. 124
C7 H 8O + 6 H 2O → 7CO + 10 H 2 Eq. 125
To facilitate the extension of the present model with respect to the con-
sidered chemical reactions and applied kinetic models, a generalized ki-
netic approach was developed, Eq. 126.
A⋅ B ⋅C ⋅ D
rreac = f act ⋅
E⋅F
⎛ − Ea ⎞
with A = k0 ⋅ exp⎜ ⎟
⎝ R ⋅T ⎠
and B = pero1 e1 ⋅ pero2e 2
C = p p1p1 ⋅ p p 2p 2
ro ro
and
ν ν
p p1p1 ⋅ p p p22
D =1−
and
(
K eq ⋅ pνe1e1 ⋅ pνe 2e 2 )
Eq. 126
and E = K eq
and F = (1 + G + H + I ) ro ads
⎛ ΔH ads1 ⎞ ro ads11 ro ads12
with G = K ads
0
1 ⋅ exp⎜ ⎟ ⋅ pads11 ⋅ pads12
⎝ R ⋅T ⎠
⎛ ΔH ads 2 ⎞ roads 21 roads 22
with H = K ads
0
2 ⋅ exp⎜ ⎟ ⋅ pads 21 ⋅ pads 22
⎝ R ⋅T ⎠
⎛ ΔH ads 3 ⎞ roads 31 ro ads 32
with I = K ads
0
3 ⋅ exp⎜ ⎟ ⋅ pads 31 ⋅ pads 32
⎝ R ⋅T ⎠
165
5 Modeling
This approach enables the reproduction of equilibrium, power law and

Langmiur-Hinshelwood kinetic models with up to three adsorption terms
for reactions with a maximum of two educt and product species. Eq. 126
involves the following terms:
A: Reaction constant term
The reaction constant term comprises the pre-exponential factor, k0,
and the activation energy of the reaction, Ea. In this model, the tem-
perature T is assumed to equal the temperature of the solid structure
of the fuel cell.
B: Educt partial pressure term
The educt partial pressure term is the product of the partial pressures
of the two educts, pe1 and pe2, to the power of their corresponding re-
action orders, roe1 and roe2.
C: Product partial pressure term
The product partial pressure term is the product of the partial pres-
sures of the two products, pp1 and pp2, to the power of their corre-
sponding reaction orders, rop1 and rop2.
D: Equilibrium limitation term in the numerator
In the equilibrium term, Keq denotes the equilibrium constant of the
considered reaction and νp1, νp2, νe1, and νe2 the stoiciometric coef-
ficients of the reaction products and educts.
E: Equilibrium limitation term in denominator
F: Overall adsorption term
The overall adsorption term includes a maximum of three adsorption
terms to the power of the according reaction order of the adsorption
reactions, roads.
G, H, I: Adsorption terms
Each of the three adsorption terms includes an adsorption constant,
K0ads1, K0ads2 and K0ads3, a corresponding heat of adsorption ΔHads1,
ΔHads2 and ΔHads3 and a maximum of two adsorption specie partial
pressures to the power of their corresponding reaction order.
Each of the discussed terms can be enabled or disabled in order to re-
produce the envisaged applied kinetic model. Note that Eq. 126 includes
the factor fact. For reactions where an applied kinetic model was specifi-
cally developed, fact is set to unity. However, where no specific kinetic
model is available fact can be set to values other than unity in order to
166
5 Modeling
adapt an available applied kinetic model for another reaction to the reac-
tion in question. The value of fact has to be determined by comparison of
the reaction in question with the reaction for which the applied kinetic
model has been developed. Although this approach lacks a solid data
base, it is nevertheless valuable as it hints towards possible impacts of
hitherto uninvestigated reactions in fuel cells such as e.g. the steam re-
forming of tars.
The water-gas-shift reaction (WGS), Eq. 119, and the reverse water-
gas-shift reaction (REV-WGS), Eq. 120, were assumed to be at equilib-
rium. This assumption is found frequently in the literature, e.g. [141, 202,
203], and was confirmed in experiments conducted by the author, see
section 3.3.2. Besides, the WGS and REV-WGS were considered to be
non-diffusion limited homogenous reactions. The reaction rates were
calculated according to Eq. 127 and Eq. 128.
mol Eq.
rWGS = 10000.0 ⋅ pCO ⋅ pH 2 O
m ⋅ sec⋅ bara 2 127
mol ⎛ pCO ⋅ pH 2O ⎞
rREV −WGS = 10000.0 ⋅ pCO2 ⋅ pH 2 ⋅ ⎜1 − ⎟ Eq.
m ⋅ sec⋅ bara 2 ⎜ K eq ,WGS ⋅ pCO ⋅ pH ⎟ 128
⎝ 2 2 ⎠
Table 5-7 lists the coefficients for the equilibrium constant fit correlation,
Eq. 82, of the WGS and the methane steam reforming reaction (CH4
STR). The equilibrium constant values were taken from [195].
Table 5-7: Coefficient Water-gas-shift reaction Methane steam reforming reaction

Coefficients of
the equilibrium
y0 -2.08873 20.45485
constant fit corre- A1 9.5544 -294.50512
lation for water- t1 354.50643 69.62891
gas-shift and
methane steam A2 6.81354 -144.02845
reforming reac- t2 758.72139 182.27607
tions
A3 45.01004 -70.17964
t3 134.0971 693.27912
In contrast to the WGS, the STR is considered a heterogeneous reac-

tion. A literature review revealed contradictory results with respect to
applied kinetic models for the STR over nickel-cermet materials. The
highest discrepancies were found in the reaction order of water. Achen-
bach found a reaction order of zero, [204], Ahmed and Foger as well as
Lee found negative reaction orders, [205, 206], and Leinfelder found a
167
5 Modeling
positive reaction order, [53]. Drescher has shown, that all these findings
are correct and a result of the chosen operating conditions of the ex-
periments, more precisely of the chosen steam-to-carbon ratio (SC),
[207]. Small SC yield positive reaction orders, SC in the order of 2 yield
reaction orders of zero while high SC lead to negative reaction orders of
water. Drescher developed a Langmuir-Hinshelwood kinetics model for
temperatures around 650 °C.
To allow for an analysis of the impact of the kinetic models on the model
results, Eq. 129 to Eq. 133 and the equilibrium approach were imple-
mented. The equilibrium constant for the CH4 STR is calculated using
the fit correlation coefficients given in Table 5-7. Note that Dreschers ki-
netic model, Eq. 133, was initially related to the mass of catalyst in the
fuel cell. Consequently the pre-exponential factor had to be recalculated
to an area specific value to allow for the application in the presented
model. The author assumed that only the upper 25 microns of the
nickel-cermet participate in the steam reforming reactions. All other data
for the recalculation were taken from [207] in order to keep the kinetic
model concise.
mol ⎛ pCO ⋅ p H3 2 ⎞
rCH 4 STR ,eq = 10000.0 ⋅ p ⋅ p ⋅
H 2O ⎜
⎜ 1 − ⎟ Eq. 129
m ⋅ sec⋅ bara 2 ⎟
⎝ K eq , STR ⋅ pCH 4 ⋅ p H 2O
CH 4
⎠
⎛ J ⎞
mol ⎜ − 82000 ⎟
rCH 4 STR , Ach = 4274.0 2 ⋅ pCH 4 ⋅ exp⎜ mol ⎟ Eq. 130
m ⋅ sec⋅ bara ⎜ R ⋅ TsK ⎟ [204]
⎜ ⎟
⎝ ⎠
⎛ J ⎞
mol ⎜ − 95000 ⎟
rCH 4 STR , Ahm = 8542.0 2 ⋅ pCH
0.85
⋅ pH− 02.O35 ⋅ exp⎜ mol ⎟ Eq. 131
m ⋅ sec⋅ bara 2 4
⎜ R ⋅ TsK ⎟ [205]
⎜ ⎟
⎝ ⎠
⎛ J ⎞
mol ⎜ − 205000 ⎟
rCH 4 STR , Lei = 30.8e + 10 2 ⋅ pCH 4 ⋅ pH 2 O ⋅ exp⎜ mol ⎟ Eq. 132
m ⋅ sec⋅ bara 2 ⎜ R ⋅ TsK ⎟ [53]
⎜ ⎟
⎝ ⎠
⎛ J ⎞
⎜ − 11000 ⎟
mol
288.52 2 ⋅ p ⋅ p ⋅ exp ⎜ mol ⎟
m ⋅ sec⋅ bara 2 ⎜ R ⋅ TsK ⎟
CH 4 H 2O
⎜ ⎟
rCH 4 STR , Dre = ⎝ ⎠ Eq. 133
⎛ ⎛ J ⎞ ⎞ [207]
⎜ ⎜ 39000 ⎟⎟
⎜1 + 16.0 1 1 mol ⎟ ⎟
⋅ pCH 4 + 0.143 ⋅ pH 2O ⋅ exp⎜
⎜ bara bara ⎜ R ⋅ TsK ⎟⎟
⎜ ⎜ ⎟⎟
⎝ ⎝ ⎠⎠
168
5 Modeling
For the reactions Eq. 122 to Eq. 125, no valid kinetic models regarding
the application in a solid oxide fuel cell model were found in the litera-
ture. Therefore, it was assumed, that the STR of higher hydrocarbons
follows the kinetic model of Achenbach, [204]. Further it was assumed
that the reactions occur faster than the CH4 STR due to the higher reac-
tivity of the considered hydrocarboneous species. Following this as-
sumption, the activity factor, fact, was set to 2.5, 1.5, 2.0 and 1.75 for the
reactions Eq. 122, Eq. 123, Eq. 124 and Eq. 125, respectively. The sub-
sequent expressions for the reaction rates of the STR of ethene, toluene
and anisole and the thermal decomposition of acetic acid were formu-
lated as:
⎛ J ⎞
mol ⎜ − 82000 ⎟
rC 2 H 4 STR , Ach = 10685.0 2 ⋅ pC 2 H 4 ⋅ exp⎜ mol ⎟ Eq. 134
m ⋅ sec⋅ bara ⎜ R ⋅ TsK ⎟
⎜ ⎟
⎝ ⎠
⎛ J ⎞
mol ⎜ − 82000 ⎟
rC7 H 8 STR , Ach = 6411.0 2 ⋅ pC7 H 8 ⋅ exp⎜ mol ⎟ Eq. 135
⎜ ⎟
⎝ ⎠
⎛ J ⎞
mol ⎜ − 82000 ⎟
rC7 H 8O STR , Ach = 7479.5 2 ⋅ pC7 H 8 O ⋅ exp⎜ mol ⎟ Eq. 136
⎜ ⎟
⎝ ⎠
⎛ J ⎞
⎜ − 82000 ⎟
mol
rC2 H 4 O2 Dec , Ach = 8548.0 2 ⋅ pC 2 H 4O2 ⋅ exp⎜ mol ⎟ Eq. 137
⎜ ⎟
⎝ ⎠
Depending on the geometry of the anode gas flow channel, heterogene-

ous reactions can be diffusion limited. That is, the diffusion of the react-
ing species from the bulk gas phase to the catalyst surface is slower
than the actual chemical reaction. Therefore, the partial pressures of the
reactants at the catalyst surface can differ significantly from the bulk gas
phase. The calculation of the precise reactant partial pressures at the
catalyst surface requires the solution of the complete concentration field
perpendicular to the gas flow direction. In order to avoid solving the cor-
responding partial differential equations, a mass transfer analogy was
used in this model to compute the reactant partial pressure at the cata-
lyst surface, Eq. 138, [190]. In this equation, βx,diff denotes the mass
169
5 Modeling
transfer coefficient of the diffusion limited considered reactant specie x,

TaK stands for the temperature of the anode gas flow and px and pxcs
represent the reactant specie partial pressure in the bulk gas phase and
at the catalyst surface, respectively.
β x , diff
rx , diff =
R ⋅ TaK
(
⋅ px − pxcs ) Eq. 138
The mass transfer coefficient βx,diff is calculated by considering the ana-

logy of heat and mass transfer according to Eq. 139, including the Nus-
selt number Nu, the molecular diffusion coefficient of the diffusion lim-
ited considered reactant specie x in the gas mixture, Dm,mix,x, and the
hydraulic diameter of the anode gas channel of the according fuel cell
design, dt/p,hyd,an.
Dm , mix , x ⎛ p ⎞
β x , diff = ⋅ Nu ⋅ ⎜⎜ ⎟⎟ Eq. 139
dt p , hyd , an ⎝ p − px ⎠
The calculation of the reactant partial pressure at the catalyst surface

can be avoided, by considering the diffusion process and the chemical
reaction as two mechanisms connected in parallel. Mathematically this
requires the introduction of a reaction conversion coefficient for the con-
sidered chemical reaction, βx,reac, which is based on the reaction rate,
rreac, calculated with the bulk gas phase partial pressures, Eq. 140. Note
that the reaction rate is divided by the partial pressure of the diffusion-
limited specie, px, to the power of its reaction order, rox.
R ⋅ TsK
β x , reac = rreac ⋅ Eq. 140
pxro x
The diffusion limited and integration length specific reaction rate, rdl-reac,
is obtained from Eq. 141.
p xro x β x , diff ⋅ β x , reac
rdl − reac = lt p , chact ⋅ β x , diff − reac ⋅ with β x , diff − reac = Eq. 141
R ⋅ TsK β x , diff + β x , reac
In this equation, lt/p,chact denotes the chemically active length of the con-
sidered fuel cell design and βx,diff-reac represents the diffusion limited re-
action conversion coefficient. The partial pressure of the diffusion-limited
specie, px, to the power of its reaction order, rox, is thus reintroduced.
Knowing the reaction rate of all reactions and neglecting axial diffusion
mass transport, the spatial distribution of the species along the anode
channel can be computed according to following differential equations:
170
5 Modeling
dn& H 2
= RH 2
dx
with RH 2 = − rH 2 oxi + 3 ⋅ rdl − CH 4 STR + rWGS − rREV −WGS + 4 ⋅ rdl − C 2 H 4 STR Eq. 142
+ 11 ⋅ rdl − C7 H 8 STR + 2 ⋅ rdl − C 2 H 4O2 Dec + 10 ⋅ rdl − C7 H 8O STR
dn&CO
= RCO
dx
with RCO = − rCO oxi + rdl − CH 4 STR − rWGS + rREV −WGS + 2 ⋅ rdl − C 2 H 4 STR Eq. 143
+ 7 ⋅ rdl −C 7 H 8 STR + 2 ⋅ rdl −C 2 H 4 O2 Dec + 7 ⋅ rdl −C7 H 8O STR
dn&CO2
= RCO2 with RCO2 = rCO oxi + rWGS − rREV −WGS Eq. 144
dx
dn& H 2O
= RH 2O
dx
with RH 2O = rH 2 oxi − + rdl − CH 4 STR − rWGS + rREV −WGS − 2 ⋅ rdl − C 2 H 4 STR Eq. 145
− 7 ⋅ rdl − C7 H 8 STR − 6 ⋅ rdl − C7 H 8O STR
dn&CH 4
= RCH 4 with RCH 4 = −rdl −CH 4 STR Eq. 146
dx
dn&C 2 H 4
= RC 2 H 4 with RC 2 H 4 = − rdl − C 2 H 4 STR Eq. 147
dx
dn& N 2 , an
= 0 .0 Eq. 148
dx
dn&C7 H 8
= RC 7 H 8 with RC 7 H 8 = − rdl − C7 H 8 STR Eq. 149
dx
dn&C 2 H 4 O2
= RC 2 H 4 O2 with RC 2 H 4 O2 = − rdl − C 2 H 4 O2 Dec Eq. 150
dx
dn&C7 H 8O
= RC 7 H 8O with RC7 H 8O = −rdl − C 7 H 8O STR Eq. 151
dx
The spatial distribution of oxygen and nitrogen in the cathode channel is

computed analogously. Note that the subscripts ca and an in the equa-
tions Eq. 148 and Eq. 152, denote the nitrogen flow in the cathode or
the anode gas channel, respectively. Furthermore the mass balance
equation for oxygen is different for the two modeled flow patterns in the
planar cell design, namely co- and counter flow. In the co-flow case the
oxygen content decreases with the axial coordinate when a current is
produced. For the counter-flow case, the opposite applies. The different
boundary conditions for the considered flow patterns will be addressed
in subsequent sections.
171
5 Modeling
dn& N 2 , ca
= 0.0 Eq. 152
dx
dn&O2
= RO2 with RO2 = −rO2 ion (co − flow) or RO2 = rO2 ion (counter − flow) Eq. 153
dx
The total molar flow in the anode and the cathode channel are calcu-
lated according to Eq. 154 and Eq. 155.
n&an = n&H 2 + n&CO + n&CO2 + n& H 2O + n&CH 4 + n&C2 H 4
Eq. 154
+ n& N 2 , an + n&C7 H 8 + n&C2 H 4O2 + n&C7 H 8O
n&ca = n& N 2 , ca + n&O2 Eq. 155
Based on the calculated reaction rates, the mass balance model also
computes the related heat source term according to Eq. 156.
ΔH r = rH 2 oxi ⋅ ΔH R , H 2 oxi + rCO oxi ⋅ ΔH R ,CO oxi + rdl − CH 4 STR ⋅ ΔH R ,CH 4 STR
+ rWGS ⋅ ΔH R ,WGS + rREV −WGS ⋅ ΔH R , REV −WGS + rdl − C 2 H 4 STR ⋅ ΔH R ,C 2 H 4 STR
Eq. 156
+ rdl − C 7 H 8 STR ⋅ ΔH R ,C 7 H 8 STR + rdl − C 2 H 4O2 Dec ⋅ ΔH R ,C 2 H 4O2 Dec
+ rdl − C 7 H 8O STR ⋅ ΔH R ,C7 H 8O STR
The heats of reaction, ΔHR,reaction, are calculated via Eq. 157, where i
denotes each considered specie, νi gives the stoiciometric coefficient of
the specie in the considered reaction, ΔH0f,i stands for the ideal gas en-
thalpy of formation and T0 is the temperature at standard conditions.
TsK
ΔH R , reaction = ΔH R0 , reaction + ∑ν i ⋅ ∫ c p ,i dT
i T0
with ΔH 0
R , reaction = ∑ν i ⋅ ΔH o
f ,i Eq. 157
i
⎛ D ⎞ ⎛ F ⎞
TsK
and ∫c
T0
p ,i dT = B ⋅ TsK + C ⋅ D ⋅ coth⎜ ⎟ − E ⋅ F ⋅ tanh⎜
⎝ TsK ⎠
⎟−O
⎝ TsK ⎠
The coefficient O and the applied ideal gas enthalpy of formation value
for the considered species are given in Table 5-8, whereas the coeffi-
cients B, C, D, E and F can be found in Table 5-1. All values were taken
from ASPEN PLUS.
By definition, the heats of reaction of all heterogeneous reactions are at-
tributed to the solid structure of the fuel cell. The differing heat capaci-
ties of educt and product species as well as the different temperature of
the gas phase and the solid structure result in an enthalpy flux coupled
to the mass flow from the solid structure to the gas phase and vice versa.
172
5 Modeling
Table 5-8: Ideal gas enthalpy of formation O

Specie
Ideal gas en- [J/mol] [J/mol]
thalpy of forma-
Hydrogen 0.0 29767.51
tion and DIPPR
correlation coef- Carbon monoxide -110530.0 22739.05
ficient O for the Carbon dioxide -393510.0 43153.18
calculation of the
heats of reaction Water -241814.0 69491.22
Methane -74520.0 135578.96
Ethene 52510.0 120989.08
Nitrogen 0.0 23243.35
Toluene 50170.0 309484.72
Acetic acid -432800.0 146450.25
Anisole -67900.0 333408.36
Oxygen 0.0 23257.56
This mass transport coupled enthalpy flux was considered in this model.
Eq. 158 to Eq. 160 give the mass transfer coupled enthalpy flux of the
educts, Q& SH ,ed , an , and products, Q& SH , prod , an , at the anode and the enthalpy
flux at the cathode, Q& SH ,ed ,ca .
Q& SH , ed , an = ∑ rj ⋅ TaK ⋅ ∑ν i , ed ⋅ c p ,i ,TaK

Eq. 158
j i
Q& SH , prod , an = ∑ rj ⋅ TsK ⋅ ∑ν i , prod ⋅ c p ,i ,TsK

Eq. 159
j i
Q& SH , ed , ca = rO2 ion ⋅ TcK ⋅ c p ,O2 ,TcK Eq. 160
In these equations, rj stands for the reaction rate of the considered het-
erogeneous reactions, including the electrochemical reactions, occurring
at the anode, Eq. 121 to Eq. 125 as well as Eq. 86 and Eq. 87, and rO2
ion represents the reaction rate of reaction Eq. 88. TaK, TcK and TsK
denote the anode and cathode gas and the solid structure temperature,
respectively. Further, νi represents the stoiciometric coefficient of educt
(subscript ed) and product (subscript prod) species of the according re-
action. Finally, the heat capacity of the educts and products is calculated
at the gas or the solid structure temperature (subscripts TaK, TcK and
TsK).
5.3.1.6 Energy balance model

Considering the strong heat exchange mechanisms between the solid
cell components, namely heat conduction and radiation, and taking into
account the compactness of the considered fuel cell designs, the as-
sumption of only one effective solid structure temperature for all solid
173
5 Modeling
cell components is reasonable, [190]. However, this assumption does

not necessarily include the gas phase temperatures due to the weak
heat exchange via convection and the negligible radiative heat ex-
change between solid structure and gas phase.
In order to account for these conclusions, the energy balance model de-
termines the axial temperature profile of the solid structure, of the anode
gas and cathode gas as well as the air in the air delivery tube in case of
a tubular design. The calculations are based on the results of the elec-
trochemical performance model and of the mass balance model.
Figure 5-13 depicts the outline of the energy balance model for an infini-
tesimal control volume with the three balance cases of the planar design,
namely the anode and cathode gas channel as well as the solid struc-
ture, and the additional fourth balance case of the tubular cell design,
namely the air delivery tube.
Figure 5-13:
Outline of the
energy balance
model
The energy balance of the anode gas channel includes the sensible
heat stream, n&an ⋅ c p ,an ⋅ TaK , entering the control volume at the coordi-
nate x and leaving it at x+dx. Furthermore, the enthalpy fluxes due to
the different heat capacity of the educt, Q& SH ,ed , an , and product species,
Q& SH , prod , an , reacting at the anode are accounted for. The value of the en-
thalpy fluxes is calculated according to Eq. 158 and Eq. 159, respec-
tively. Finally the conductive heat stream between solid structure and
anode gas channel is considered, Eq. 161. αan denotes the heat trans-
fer coefficient, TaK stands for the anode gas temperature and TsK
174
5 Modeling
represents the solid structure temperature. The heat transfer coefficient

αan is calculated using the Nusselt number Nu, the thermal conductivity
of the anode gas, λan, which is considered temperature and composition
dependent, see Eq. 39 and Eq. 41, and the hydraulic diameter of the
anode gas channel, dt/p,hyd,an.
Nu ⋅ λan
Q& conv , an = α an ⋅ lt p , circ , an ⋅ (TaK − TsK ) with α an = Eq. 161
dt p , hyd , an
The differential equation describing the axial temperature distribution

along the anode gas channel can thus be formulated as follows:
d (n&an ⋅ c p , an ⋅ TaK )
= Q& SH , prod , an − Q& SH , ed , an − Q& conv , an Eq. 162
dx
It should be noted, that this model considers the change of the total mo-
lar flow, nan, and the change of the heat capacity, cp,an, due to the
change of the gas composition over the control volume. In some models
found in the literature, only the temperature change is accounted for by
assuming the heat capacity and sometimes also the molar flows are
fixed at the inlet conditions of the corresponding control volume, [140,
182, 208]. For coarse discretization aiming at short calculation times,
this can cause problems with the gas channel energy balance closure.
The energy balance of the solid structure involves the overall heat of re-
action, Eq. 156, and the mass transfer coupled enthalpy fluxes at the
anode and cathode electrode, Eq. 158 to Eq. 160. Besides this, the con-
vective heat streams between the solid structure and anode, Q& conv , an , as
well as cathode gas channel, Q& conv , ca , have to be taken into account. In
the present model, the produced electrical power, Pel, is regarded as a
source term for the solid structure and is computed according to Eq. 163.
Note that the total current density Itot results from the electrochemical
performance model and Eop is the user-defined operational voltage of
the fuel cell. In order to transform this area specific power density into
an integration length specific value, it has to be multiplied with the elec-
trochemically active length of the fuel cell design in question, lt/p,elact.
Pel = −lt p , elact ⋅ I tot ⋅ Eop Eq. 163
The convective heat stream between the cathode gas channel and the
solid structure is determined according to Eq. 164 considering the tem-
perature difference between cathode gas, TcK, and solid structure, TsK,
the effective heat transfer coefficient, αca, and the circumferential length
175
5 Modeling
of the cathode gas channel, lt/p,circ,ca. Analogously to Eq. 161, the heat
transfer coefficient of the cathode gas channel is calculated with the
temperature and gas composition dependent thermal conductivity of the
cathode gas, λca, and the hydraulic diameter of the cathode gas channel,
dt/p,hyd,ca, of the investigated fuel cell design. The Nusselt number, Nu, is
assumed to equal that of the anode gas channel, as in both cases lami-
nar flow is assumed. It is important to emphasize that the convective
heat stream of the cathode gas channel is considered positive, when
heat is transferred from the solid structure to the cathode gas. This is
due to the cooling task of the air flow in the cathode channel.
Nu ⋅ λca
Q& conv , ca = α ca ⋅ lt p , circ , ca ⋅ (TsK − TcK ) with α ca = Eq. 164
dt p , hyd , ca
Finally in this model, the energy balance of the solid structure includes
solid heat conduction. This yields the second order derivate differential
equation, Eq. 165, for the solid structure temperature, TsK, where λs
symbolizes the average solid heat conduction coefficient of the solid
structure and At/p,cross stands for the cross-sectional area which is per-
pendicular to the gas flow channels.
d 2TsK &
λs ⋅ At p , cross ⋅ = Qconv , ca − Q& conv , an − Q& SH ,ed , an − Q& SH , ed ,ca + Q& SH , prod , an + ΔH r + Pel Eq. 165
dx 2
All terms required to compute the cathode gas temperature of the planar
cell design have been previously mentioned and yield Eq. 166.
d (n&ca ⋅ c p , ca ⋅ TcK )
= Q& conv ,ca − Q& SH , ed , an + Q& conv, ADT (co − flow)
dx Eq. 166
= Q& SH , ed , an − Q& conv, ca (counter − flow)
Analogously to the differential equation of the anode gas channel, the

change of the molar flow and of the heat capacity is considered.
In the presented model, the two considered flow patterns of the planar
cell designs where modeled through the cathode gas balances, see also
the mass balance, e.g. Eq. 153. In principle, the tubular cell design has
a co-flow pattern. However, the design typical air delivery tube, adds
one more term to the cathode gas energy balance, namely the convec-
tive heat stream from the air delivery tube to the cathode gas, Q& conv , ADT .
Q& conv , ADT aggregates the convective heat transport from the cathode air,
with the temperature TcK, to the air delivery tube and from the air deliv-
ery tube (ADT) to the cold air flowing inside the ADT, with the tempera-
176
5 Modeling
ture TadtK, Eq. 167. In this model, the two processes are accounted for
through an effective heat exchange coefficient, αADT,eff, which was pro-
posed in [209]. The calculation of αADT,eff includes the inner, ri,ADT, and
outer radius, ro,ADT, of the ADT, the heat exchange coefficients of the
cathode gas, αca, and inside the ADT, αADT. Also included is the solid
heat conduction coefficient of the ADT material, λs,ADT, which leads to
either enhanced or reduced convective heat exchange in cold or hot re-
gions of the ADT, respectively. αADT is computed analogously to the
other heat exchange coefficients.
Q& conv, ADT = α ADT , eff ⋅ lt , circ, ADT ⋅ (TadtK − TcK )
1 ⎛ 1 1 ⎛r ⎞ 1 ⎞
with α ADT ,eff = ⋅⎜ + ⋅ ln⎜ o , ADT ⎟+ ⎟
ri , ADT ⎜ ri , ADT ⋅ α ADT 4 ⋅ λs , ADT ⎜r ⎟ r ⋅ α ⎟ Eq. 167
⎝ ⎝ i , ADT ⎠ o , ADT ca ⎠
Nu ⋅ λ ADT
and α ADT =
d t , hyd , ADT
The energy balance of the air flow inside the ADT can then be formu-
lated as Eq. 168. Since the air flow inside the ADT is not partaking in
any chemical or electrochemical reactions, the molar flow is constant.
Accordingly the differential equation for the ADT energy balance was
simplified in the present model.
d (c p , ADT ⋅ TadtK )
n& ADT ⋅ = Q& conv , ADT Eq. 168
dx
5.3.1.7 Boundary conditions

The energy balance model alongside the mass balance and the electro-
chemical performance model constitute a highly coupled and strongly
non-linear differential equation system. In this model, the cells are di-
vided into n control volumes (CV). An inclusive solution is found once
the governing equations are solved for each CV. Boundary conditions
need to be defined for the CVs at the outer ends of the cells. The mass
balance boundary condition values are calculated based on the user-
defined operating conditions. The fuel inlet composition is therefore
specified in molar fractions of the considered species, yi. The total molar
flow of the fuel gas at the cell inlet, nan,0, is either calculated based on a
targeted overall current density, Itot,av, and related fuel utilization, UF,
Eq. 169, or by specifying a lower heating value (LHV) based input power,
Pin, Eq. 170.
177
5 Modeling
I tot , av ⋅ Ap t , act
n&an , 0 =
2 ⋅ F ⋅ UF ⋅ EqH 2
Eq. 169
with EqH 2 = y H 2 + yCO + 4 ⋅ yCH 4 + 6 ⋅ yC 2 H 4
+ 18 ⋅ yC7 H 8 + 4 ⋅ yC 2 H 4 O2 + 17 ⋅ yC 7 H 8 O
In Eq. 169, Ap/t,act denotes the electrochemically active area of the con-
sidered fuel cell design and EqH2 represents the hydrogen equivalent of
the prescribed input gas composition.
Pin
n&an , 0 = with nRE = nch ⋅ ncells or nRE = ntubes Eq. 170
LHVin ⋅ nRE
nRE in Eq. 170 stands for the number of repeating elements in the stack.
For the tubular design, the repeating element is a whole cell tube,
whereas for the planar design only one fuel and air channel is consid-
ered. Hence, for the planar design the number of repeating elements is
the product of the number of cells in the stack, ncells, and the number of
channels per cell, nch. Furthermore, LHVin stands for the lower heating
value of the prescribed fuel gas composition.
In this model, the standard cathode gas input composition is equal to
that of air or more precisely 79 mol-% nitrogen and 21 mol-% oxygen.
However, other compositions such as e.g. pure oxygen can also be de-
fined. The inlet molar flow of the cathode gas, nca,0, is either assigned
directly or by specifying an air-to-fuel ratio, λ.
n&O2 , stoic ⋅ λ
n&ca , 0 =
yO2
Eq. 171
with n&O2 , stoic = n&an , 0 ⋅ (0.5 ⋅ yH 2 + 0.5 ⋅ yCO + 2 ⋅ yCH 4
+ 3 ⋅ yC 2 H 4 + 9 ⋅ yC7 H 8 + 2 ⋅ yC 2 H 4O2 + 8.5 ⋅ yC7 H 8O )
In the latter case Eq. 171 applies, where nO2,stoic gives the oxygen molar
flow required for the stoiciometric combustion of the fuel gas. Depending
on the investigated fuel cell design and the considered flow pattern, the
mass balance boundary values were assigned to different CVs, see
Table 5-9 where (0) denotes the first and (n) the last CV.
Table 5-9: Planar Planar Tubular

Mass balance
Mass balance Co-flow Counter-flow
boundary condi-
Anode gas n&an ( 0 ) = n&an , 0 n&an ( 0 ) = n&an , 0 n&an ( 0 ) = n&an , 0
tions
Cathode gas n&ca ( 0 ) = n&ca , 0 n&ca ( n ) = n&ca , 0 n&ca ( 0 ) = n&ca , 0
Air delivery tube - - n& ADT ( n ) = n&ca , 0
178
5 Modeling
The total molar flow values nan, nca and nADT are related to the molar
flows of the single species via the user-defined molar fractions.
Besides the inlet molar flow, the present model requires the definition of
the gas temperatures at the inlet in order to solve the energy balance
equation of the gas channels. The inlet gas temperature of the anode
gas channel, Tan,in, the cathode gas channel, Tca,in and eventually of the
air deliver tube, TADT,in, are user-defined values and assigned as shown
in Table 5-10.
Table 5-10: Planar Planar Tubular

Energy balance
Gas phase en- Co-flow Counter-flow
ergy balance
Anode gas TaK ( 0 ) = Tan ,in TaK ( 0 ) = Tan ,in TaK ( 0 ) = Tan ,in
boundary condi-
tions Cathode gas TcK ( 0 ) = Tca ,in TcK ( n ) = Tca ,in TcK ( 0 ) = TadtK ( 0 )
Air delivery tube - - TadtK ( n ) = TADT ,in
In case of the tubular cell design, the boundary conditions of the cath-
ode gas channel ensure the continuity between the temperature and the
molar flow of the air flow exiting the air delivery tube and entering the
cathode gas channel.
Owing to the considered solid heat conduction, the energy balance of
the solid structure includes a second derivate term. Hence, the solution
of each CV depends on the solutions of the neighboring CVs. Conse-
quently, boundary conditions have to be defined at both ends of the in-
tegration region.
The differential equation applying at the left boundary (first control vol-
ume of the integration region) is given by Eq. 172.
dTsK
λs ⋅ At p,cross ⋅ = Q& hloss Eq. 172
dx
Q& hloss represents the heat loss stream to the surroundings. At the right
boundary, the source terms of the last CV of the integration region have
to be taken into account, Eq. 173.
dTsK
λs ⋅ At p , cross ⋅ =
dx Eq. 173
(Q& conv,ca )
− Q& conv, an − Q& SH , ed , an − Q& SH , ed , ca + Q& SH , prod , an + ΔH r + Pel ⋅ dx − Q& hloss
For the tubular design the cell ends are generally assumed adiabatic. In
contrast, the developed model features the possibility to assume adia-
batic cell ends, radiative or conductive heat losses at the cell ends of
179
5 Modeling
planar cells. Where adiabatic cell ends are assumed for the simulation
of a perfect insulation, Q& hloss is set to zero. The value of the radiative
heat loss streams is calculated according to Eq. 174, [190], where αrad
denotes the radiative heat exchange coefficient, σ stands for the Stefan-
Boltzmann constant and Tsur represents the Temperature of the sur-
roundings.
⎛ Tsur + TsK ⎞
3
Q& hloss =α rad ⋅ At p,cross ⋅ (TsK − Tsur ) with α rad = 4 ⋅ σ ⋅ ⎜ ⎟ Eq. 174
⎝ 2 ⎠
Assuming conductive heat losses through an insulation, Eq. 175, [190],

applies, where αinsul stands for the heat transfer coefficient through the
insulation and Tamb represents the ambient temperature.
Q& hloss = α insul ⋅ At p,cross ⋅ (TsK − Tamb) Eq. 175
5.3.1.8 Solution strategies

The strongly coupled and highly non-linear nature of the model equation
system inhibits an algebraic solution. Instead, it requires a numerical so-
lution starting with an initial conjecture. Convergence is fastly reached,
when the initial conjecture approximates the true solution. If the initial
conjecture is far off the mark, convergence might not be reached.
In the present model, the standard initial conjecture is that the inlet val-
ues apply to all CVs. With this linear initial guess, the developed model
reliably reaches convergence with respect to fuel gases with a hydro-
carbon content of up to 5 vol.-%. Fuel gases with higher hydrocarbon
contents lead to strongly non-linear spatial species distributions, which
are difficult to solve numerically. One way to overcome these conver-
gence problems is to compute interim solutions which are used as initial
conjectures for the next iteration process until the final solution is found.
ATHENA allows the computation of a new solution beginning with a
"saved" solution instead of using the linear initial conjecture. The boxes
"save the final solution" and "start with the saved solution" in the solver
panel under the tab "solution history", need to be activated.
Experience has shown that it is advantageous to begin the solution it-
eration from a gas composition containing hydrogen and carbon monox-
ide at equal rates. In this case the solver needs not to overcome an acti-
vation barrier to compute carbon monoxide originating currents for a hy-
drocarbon containing fuel gas. Usually, three to five interim solutions are
sufficient to find the desired final solution. Following this strategy allows
180
5 Modeling
this model to handle fuel gases with almost unlimited hydrocarbon con-
tent. The investigation of different operating conditions implies fewer
convergence difficulties. Guidelines that might promote a faster outcome
can also be formulated. Changes of the model input that strongly influ-
ence the anode gas mass balance, e.g. increase of the fuel mass flow or
change of the fuel composition, have to be undertaken in small incre-
mental steps. On the other hand changes of e.g. the voltage can be per-
formed in comparably large increments. This is due to the absolute
mass flow values remaining virtually constant. Changes of input gas
temperatures have a strong impact on the energy balance and therefore
also require an iterative approach in small steps.
If the employed numerical solver is unable to find a solution, the values
of the state variables of the last solution need to be examined. If the val-
ues of any of the state variables have reached numbers very close to
zero or are disproportionately high the envisaged operational point could
be physically unfeasible. This physical unfeasibility might be the reason
for the perceived non-convergence. If on the other hand physical infea-
sibility was ruled out the solver could be blocked in a semi-stable solu-
tion point. In this case the only operational strategy is to return to a sta-
ble and physically adequate solution and change the input for the next
interim solution so that the semi-stable point is avoided.
5.3.2 Planar model verification

A quantitative validation of the planar model is almost infeasible as only
few complete data sets of real planar SOFC stacks have been published
thus far. Furthermore, the ohmic losses of a cell depend largely on the
geometry of the bipolar plates, also called interconnector plates (IC).
This model considers a generalized geometry of the bipolar plates for
which data is not available at all.
Nevertheless, a verification of the developed model is carried out by
comparing the results to a benchmark test (BMT) to ensure that the de-
veloped numerical code has reached the BMT quality standard. The
BMT was defined in an International Energy Agency (IEA) program for
the numerical simulation of SOFCs with planar geometry, [112]. In total,
nine academic and industrial institutions participated in the BMT. The
independently developed numerical SOFC models showed good agree-
ment regarding the predicted physical behavior for two fuel gas compo-
sitions, which were humidified hydrogen and 30 % pre-reformed meth-
ane, Table 5-11.
181
5 Modeling
Table 5-11: Specie Unit Fuel gas IEA 1 Fuel gas IEA 2 Cathode gas Source
Fuel and cathode
Hydrogen 90.00 26.26 -
gas compositions
of the BMT Carbon monoxide - 2.94 -
Carbon dioxide - 4.36 -
Water vol.-% 10.00 49.34 - [112]
Methane - 17.10 -
Nitrogen - - 79.00
Oxygen - - 21.00
Table 5-12 gives the BMT geometry data of the planar SOFC stack.
Table 5-12: Model input data Unit Value Source

Geometrical Macro geometry of planar cell
input data of
Flow design - Co- and counter-flow
BMT
Number of cells in stack - 1
Number of gas channels per cell - 18
Cell width m 0.1
Cell length m 0.1
Gas channel width m 0.003 [112]
Anode and cathode channel height m 0.001
Height of IC on anode/ cathode side m 0.00125
Area covered by IC participating in
% 100.0
electrochemistry
Chemically act. area covered by IC % 0.0
Micro geometry of planar cell
Support design - Electrolyte
Anode thickness μm 50.0
[112]
Electrolyte thickness μm 150.0
Cathode thickness μm 50.0
Table 5-13 shows the operational conditions defined for the BMT.

Operational con-
Operational conditions
ditions of BMT
Targeted mean current density A/m2 3000.0
Targeted fuel utilization % 85.0
Air-to-fuel ratio - 7.0
Fuel gas inlet temperature K 1173.15 [112]
Cathode gas inlet temperature K 1173.15
Ambient temperature K 293.15
System pressure bara 1.01325
182
5 Modeling
The BMT participants agreed to consider only ohmic losses for a better
comparability of the model results. The specific ohmic resistances of the
ceramic components were calculated according to the equations Eq. 45
to Eq. 48. For the planar SOFC, ceramic bipolar plates were assumed.
Further, the applied kinetics data for the steam reforming reaction (STR)
were prescribed, Eq. 130. Table 5-14 gives the BMT parameter used for
the energy balance calculations.

Energy balance
Solid heat conduction
input data of
BMT Heat conductivity of anode 2.0
Heat conductivity of electrolyte 2.0
W/ m K [112]
Heat conductivity of cathode 2.0
Heat conductivity of interconnect 2.0
Convective heat transfer
Nusselt number - 4.0 [112]
Besides the described model input data, several assumptions were de-
fined for the BMT, Table 5-15.
Table 5-15: Model settings Description

Model settings of
Electrochemical loss model settings
BMT
Considered electrochemical active species Hydrogen
Activation polarization equation Equal to ohmic loss of electrolyte
Energy balance settings
Solid heat transfer mechanism Non-isothermal with solid heat conduction
Coupled heat and mass transport Considered
Heat loss mechanism at outer surface Conduction through insulating plates
Heat capacity and heat of reaction correlation published in [190]
dT
Definition of sensible heat gradient n& ⋅ c p ⋅
x dx
The most important assumptions were that the activation polarization

losses are equal to the ohmic loss of the electrolyte and that the diffu-
sion losses are negligible. For this reason, a validation of activation and
diffusion loss model parameters is not indicated using the BMT results.
Furthermore, the change of the molar flow and the heat capacity of the
gas mixture was neglected in the calculation of the gas phase energy
balance. Instead the calculations were performed with a simplified defi-
183
5 Modeling
nition of the sensible heat gradient by holding the molar flow and heat
capacity constant at the value of the antecedent control volume for the
respective control volume. Further information can be found in [112].
Figure 5-14 shows the comparison of results of the developed model for
the co-flow pattern considering both IEA fuel gases, Table 5-11.
Figure 5-14:
Comparison of co- Planar co-flow model comparison for IEA benchmark 1 input
flow model results 4500

with BMT results 4000
for fuel composi-
Corresponding units
3500
tions IEA 1 and 3000
IEA 2 2500
2000
1500
1000
500
0
[mV] [10E-2 W] [A/m2] [A/m2] [K] [K] [K/m] [K] [K]
Voltage Power Min. Max. Min. Solid Max. Max. Air outlet Fuel outlet
Current Current temp. Solid Temp. temp. temp.
density density temp. gradient
Minimum value Own Maximun value
Planar co-flow model comparison for IEA benchmark 2 input
4000
3500
Corresponding units
3000
2500
2000
1500
1000
500
0
[mV] [10E-2 W] [A/m2] [A/m2] [K] [K] [K/m] [K] [K]
The terms "Minimum value" and "Maximum value" refer to the minimum
and maximum values among all the BMT participants, whereas "Own"
refers to the results obtained from the author's model.
Figure 5-15 shows the comparison of results for the counter-flow pattern.
It can be seen, that the results of the developed model lie mostly be-
tween the minimum and maximum values of the BMT. Exceptions are
found for the maximum current density (Figure 5-14, fuel gas IEA 1) and
the maximum solid temperature gradient (Figure 5-14 and Figure 5-15,
fuel gas IEA 2).
184
5 Modeling
Figure 5-15:
Comparison of Planar counter-flow model comparison for IEA benchmark 1 input
counter-flow 10000
model results with 9000
BMT results for 8000
Corresponding units
fuel compositions 7000
IEA 1 and IEA 2 6000

5000
4000
3000
2000
1000
0
[mV] [10E-2 W] [A/m2] [A/m2] [K] [K] [K/m] [K] [K]
Planar counter-flow model comparison for IEA benchmark 2 input

14000
12000
Corresponding units
10000
8000
6000
4000
2000
0
[mV] [10E-2 W] [A/m2] [A/m2] [K] [K] [K/m] [K] [K]
The local temperature curves obtained from the author's model were
compared with those of the BMT activity leader, Achenbach, [112], to
find explanations for the discrepancies, Figure 5-16.
Figure 5-16: 1350

Comparison of Own: 20 points Own: 100 points Achenbach
temperature pro-
files for the IEA 1
fuel composition 1300
and counter-flow
Solid temperature [K]
pattern computed
by the Achenbach 1250
model, [112], and
the developed
model with 20 and
100 discretization 1200
points
1150
0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10
Cell length [m]
185
5 Modeling
It was found that different numbers of control volumes were used for the
discretization of the considered planar cell. Figure 5-16 shows the im-
pact of the number of discretization points on the computed temperature
profile. The temperature gradients at the channel ends are determined
by the considered heat losses. Coarse discretization leads to an overes-
timated impact of the cell end heat losses on the temperature profile.
However, the general trends were not affected by the coarse discretiza-
tion which has the advantage of short calculation times around 30 sec-
onds run on a conventional personal computer. In summary it can be
stated that the author's model behaves physically correct and that there
is no systematic discrepancy with the results of other participants in the
BMT. In all the simulations, the error in the mass balance was below
0.5 % and in the energy balance below 2 %. Compared to the state-of-
the-art with 1 % error in the mass balance and up to 6 % error in the en-
ergy balance, [114], the author's model is considered a reliable tool for
the simulation of planar SOFCs.
5.3.3 Tubular model validation

In contrast to the planar geometry, suitable experimental data can be
found in the literature for a quantitative validation of the tubular geome-
try. Moreover, the modeled tubular geometry is not generalized, but
based on the cell design promoted by the Siemens AG. The geometrical
data for the description of the standard tubular Siemens AG cell is given
in Table 5-16.

Geometrical Macro geometry of tubular cell
input data for
Flow design - Co-flow with air delivery tube
tubular Siemens
AG cell model Cell tube length m 1.5
Inner radius of air delivery tube m 0.0025
Outer radius of air delivery tube m 0.004 [140,
Inner radius of cell tube m 0.00866 155]
Thickness of virtual fuel channel
m 0.0023
(depends on stacking of cell tubes)
Percentage of circumferential length
% 10.0
of cell tube covered by IC
Support design - Cathode
Anode thickness μm 100.0 [140,
Electrolyte thickness μm 40.0 155]
186
5 Modeling
The aim of validating the developed model considering the tubular cell
geometry is to ensure that the generalized approach, which is based on
characteristic lengths and areas, has been properly adapted to the tubu-
lar cell design. Although the most important parts of the model code
were already confirmed using the BMT for the planar geometry, there is
still a need to also verify the equations involved in the calculation of the
activation and diffusion losses given these were excluded from the BMT.
The experimental voltage-current curves used for the validation are pub-
lished in [70]. The characterized standard tubular Siemens AG cell was
operated with a fuel gas mixture consisting of 89 vol.-% H2 and 11 vol.-
% H2O at three different cell temperatures, 900 °C, 940 °C and 1000 °C.
The latter feature predestines this data set for model validation, as it al-
lows the model to be checked regarding the correct prediction of tem-
perature effects. Model results obtained for other temperatures can be
assumed valid, in case the model reliably predicts the temperature in-
duced effects of the validation case. When model validation is carried
out with data sets measured at a single temperature, the results are only
valid for this temperature. Hence extrapolated results have to be con-
sidered less reliable.
The voltage-current curves were experimentally determined with con-
stant fuel utilization of 85 % and an air-to-fuel ratio of 4. For the simula-
tion, the model further requires as operational input data the gas inlet
temperatures, system pressure and temperature of the surroundings.
The gas inlet temperatures were assumed 100 K lower than the respec-
tive mean cell temperature, [155]. As the author of [70] stated that the
measurements were carried out using a prototypic cell, it was assumed
that the cell was operated in an oven exhibiting a constant wall tempera-
ture. The surroundings temperature required for the radiational heat
transfer calculation was assumed equal to the mean cell temperature,
[155]. The heat transfer calculation input data is given in Table 5-17.

Energy balance
Solid heat conduction
input data for
tubular Siemens Heat conductivity of anode 6.23
AG cell model Heat conductivity of electrolyte 2.7
validation W/ m K
Heat conductivity of cathode 9.6 [140]
Heat conductivity of air delivery tube 11.8
Nusselt number - 4.0
187
5 Modeling
The electrochemical loss model parameters are given in Table 5-18.

Electrochemical Activation polarization
loss model input
H2 oxidation activation energy J/mol 110000.0
data for tubular [202]
Siemens AG cell H2 oxidation pre-exponential factor A/m2 7000000000.0
model 110000.0
CO oxidation activation energy J/mol [190,
CO oxidation pre-exponential factor A/m2 5000000000.0 202]
O2 reduction activation energy J/mol 149500.0
[155]
O2 reduction pre-exponential factor A/m2 10260000000.0
Ohmic polarization
Electric conductivity of anode 1/Ohm m T-dependent, Eq. 45
Ionic conductivity of electrolyte 1/Ohm m T-dependent, Eq. 46 [191]
Electric conductivity of cathode 1/Ohm m T-dependent, Eq. 47
Electric conductivity of interconnect Ohm m 2.0E-7 [155]
Diffusion polarization
Porosity of anode - 0.4
Tortuosity of anode - 3.0
[140]
Pore diameter of anode and cathode m 1.0E-6
Porosity of cathode - 0.5
Tortuosity of cathode - 1.8 -
The conductivity of the anode, electrolyte and cathode were calculated

using the same correlations as in the BMT. The diffusion polarization lo-
ss parameters were taken from Stiller, [140]. However, the cathode elec-
trode tortuosity was slightly increased from 1.5 to 1.8 in order to closer
reproduce the diffusion limitation at high current densities. The activation
polarization parameters for the hydrogen oxidation and for the oxygen
ionization were taken from Campanari, [202], and Thorud, [155], respec-
tively. Both authors determined the values by means of parameter esti-
mation. A drawback of this approach is, that all uncertainties of the re-
maining model input parameters and of the measurement to which the
model is fitted are summed up in the activation loss parameters. Never-
theless, the lack of complete experimental data sets from which these
parameters could be derived justifies this approach. The activation loss
parameters for the carbon monoxide oxidation were calculated based on
those of the hydrogen oxidation and the relation proposed in [190].
188
5 Modeling
Figure 5-17: 0.80 6%

Comparison of simu-
lated and measured
voltage-current curve
of a standard tubular 0.70 3%
Siemens AG cell
measured at a mean
Voltage [V]
Variance
cell temperature of 0.60 0%
900 °C
0.50 -3%
0.40 -6%
750 1250 1750 2250 2750 3250 3750 4250 4750 5250 5750
2
Current density [A/m ]
Measurement Developed model Variance
Figure 5-17 shows the comparison of simulated (dashed line) and meas-
ured (full line) voltage-current density (UI) characteristics of the tubular
cell operated at 900 °C. The slightly higher slope of the simulated UI
curve indicates an overestimation of the ohmic resistance and an un-
derestimation of the activation polarization at low current density values.
The mean variance of the predicted voltage for 900 °C operational
mean cell temperature compared to the measured values is 1.6 % and
the maximum variance is below +/-3 % (right axis of graph).
Figure 5-18 shows, the simulated and measured U-I curve for 940 °C
mean cell temperature.
Figure 5-18: 0.80 6%

Comparison of simu-
lated and measured
Siemens AG cell
measured at a mean
Voltage [V]
Variance
cell temperature of
0.60 0%
940 °C
0.50 -3%
0.40 -6%
750 1250 1750 2250 2750 3250 3750 4250 4750 5250 5750
2
189
5 Modeling
It can be seen, that the predicted voltage values diverge by less than
1 % for small and medium current density values. At higher current den-
sity values the variance increases up to 5 %, indicating an underestima-
tion of the diffusion polarization losses. However, the activation and oh-
mic losses are predicted well resulting in a good match of the UI curve
slope of the simulated and measured curve. The mean variance be-
tween predicted and measured voltage values for 940 °C mean cell
temperature is 1.3 % and the maximum variance is below +5 %.Figure
5-19 depicts the comparison of simulated and measured voltage-current
curve measured at a mean cell temperature of 1000 °C.
Figure 5-19: 0.80 6%

Comparison of simu-
lated and measured
Siemens AG cell
measured at a mean
Voltage [V]
Variance
cell temperature of 0.60 0%
1000 °C
0.50 -3%
0.40 -6%
750 1250 1750 2250 2750 3250 3750 4250 4750 5250 5750
Current density [A/m2]
Similar to the predicted UI curve for 900 °C, the model underestimates
the activation losses at low current densities for 1000 °C mean cell tem-
perature. Further, the slope of the predicted curve is again steeper than
of the measured UI curve indicating a slight overestimation of the ohmic
losses. At high current density values, the model predicts a slight bend
of the UI curve due to increasing diffusion losses. This phenomenon can
be observed more pronounced for the measurement. Nevertheless, the
prediction of this phenomenon clearly shows that the developed tubular
cell model behaves physically correct. The mean variance between pre-
dicted and measured voltage values for 1000 °C mean cell temperature
is 2.3 % and the maximum variance is below +/-4 %.
In summary, the mean variance of the model results from the measure-
ment results was found to be below 2.5 % for all the three considered
mean cell temperatures. The developed tubular cell model behaves
190
5 Modeling
physically correct and there is no systematic discrepancy to the meas-

urement results. In conclusion it can be stated that the developed gen-
eralized model was properly adapted to the tubular cell geometry pro-
moted by the Siemens AG. Furthermore, the activation and diffusion
loss model equations and related model parameters proved to yield ac-
curate results. Keeping in mind that the model parameters were col-
lected from various sources, the validation was considered successful.
5.3.4 Triangular model validation

The data regarding DELTA8 cells (D8) is scarce as this type of cells has
only recently been developed by the Siemens AG. Nevertheless, Table
5-19 gives the currently available D8 cell geometrical data.

Geometrical Macro geometry of tubular cell
input data for
Flow design - Co-flow with air delivery tube -
triangular D8 cell
model Cell length m 1.0
Cell width m 0.15
[210]
Triangle half angle ° 30.0
Number of triangular tubes per cell - 8
Inner radius of air delivery tube m 0.002 -
Outer radius of air delivery tube m 0.003 -
Support design - Cathode
Electrolyte thickness μm 60.0 [211]
Interconnect thickness μm 100.0
It is important to point out, that the geometry of the air delivery tube was
not taken from a publication but was assumed to yield an adequate
cathode gas channel cross-sectional area. This assumption directly im-
pacts the convective heat transfer between the cathode gas and the air
delivery tube. The convective heat transfer between the cathode gas
and the solid structure however dominates in any case, such that the
overall impact of this assumption is considered less significant.
The experimental voltage-current curve used for the validation of the tri-
angular cell model was measured at the Siemens AG using a D8 cell
segment, [210]. The 1 cm long D8 cell segment was operated under iso-
thermal conditions in an oven at temperatures between 950 and 965 °C
191
5 Modeling
with fuel and air excess. The fuel mixture consisted of 50 vol.-% H2 and
50 vol.-% H2O, being the expected average gas composition of a full-
scale cell operated at 85 % fuel utilization with 89 vol.-% H2 and 11 vol.-
% H2O. Due to the isothermal conditions and the virtually constant reac-
tant partial pressures along the D8 cell segment, the voltage-current
curve was simulated by using the electrochemical performance model
without consideration to the energy and heat balance. With respect to
the model parameters, it was assumed that the anode and cathode ma-
terials used for the D8 cells resemble those used for the standard tubu-
lar cells. Accordingly, the model parameters given in Table 5-18 were
used for the simulation. Exceptions were the electric conductivity of the
interconnect material, which was calculated according to Eq. 48, and the
ionic conductivity of the electrolyte, as D8 cells employ scandia-
stabilized zirconia (ScSZ) instead of the state-of-the-art yttrium-
stabilized zirconia (YSZ) as electrolyte material, [210]. At temperatures
above 900 °C, ScSZ features ionic conductivities about three times
higher than YSZ, [79]. Hence, the conductivity values computed via Eq.
46 were multiplied by three for the D8 cell segment simulation.
Figure 5-20: 0.90

Comparison of simu- Measurement Our model
lated and measured
voltage-current curve 0.80
of a D8 cell segment
measured at a tem- 0.70
perature between
Voltage [V]
950 and 965 °C

0.60
0.50
0.40
0.30
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000
2
Figure 5-20 depicts the comparison of the measured and the simulated
voltage-current curve of the D8 cell segment.
The simulation result for 957 °C fits the measured data well. It is con-
cluded, that the author's model was successfully applied to the triangu-
lar geometry thereby validating the discussed assumptions.
192
5 Modeling
5.3.5 Sensitivity analysis

The complex mechanisms of SOFC electrochemistry are still not fully
understood and are therefore mostly modeled using semi-empirical cor-
relations. This introduces several constants to the model equation sys-
tem, which have to be determined experimentally in order to allow for a
reliable performance prediction of the SOFC in question. The relative
importance of model parameters was assessed with the following sensi-
tivity analysis aiming at the impact on the current density distribution and
local temperatures resulting in the predicted power output.
The applied methodology is that a single model parameter or a set of in-
terdependent parameters is varied at once while all other parameters
are held constant at the reference value. Unless otherwise stated, the
varied model parameter is increased or decreased by 25 and 50 %
based on the corresponding reference value.
The operational voltage is the only actual control variable of SOFCs as
the fuel input flow is a design value defined by the targeted fuel utiliza-
tion and the active cell area. Further, the input air flow follows from the
targeted mean cell temperature and the maximum allowed solid tem-
peratures. In order to conform to this, the fuel utilization is being held
constant by adjusting the operational voltage throughout the whole sen-
sitivity analysis. This allows to better isolate the impact of the varied
model parameter from side effects that are an indirect consequence of
the parameter variation. For instance, increasing the ohmic resistance
with constant operational voltage yields lower fuel utilization. As a con-
sequence of the reduced amount of electrochemical reactions, the mean
cell temperature drops, causing higher ohmic losses and so on. This
self-energizing effect leads to an overestimation of the impact of the in-
creased ohmic loss. Unfortunately, it is not always obvious what side ef-
fects might distort the impact of a model parameter.
5.3.5.1 Reference case

The values of the varied model parameters are given in Table 5-20 for
the reference case. It should be noted that the exchange current density
for the activation polarization loss calculation were computed using the
equations Eq. 92 to Eq. 94, [182].
The sensitivity analysis was carried out for the two BMT fuel gases,
Table 5-11. The IEA 1 gas allows for the explicit investigation of the
model constant variation impact on the electrochemistry. The IEA 2 gas
193
5 Modeling
was investigated to determine the coherences when also internal re-

forming reactions take place. The STR reaction rates were computed via
Eq. 130.

loss model input
data of reference [182]
case H2 oxidation pre-exponential factor A/m2 290000000.0
CO oxidation activation energy J/mol 120000.0
[190]
CO oxidation pre-exponential factor A/m2 207000000.0
[182]
Ohmic polarization
Ionic conductivity of electrolyte 1/Ohm m T-dependent, Eq. 46
[191]
Electric conductivity of ceramic IC 1/Ohm m T-dependent, Eq. 48
Porosity of anode - 0.5
Tortuosity of anode - 3.0
Average pore diameter of anode m 1.0E-6
[141]
Porosity of cathode - 0.5
Tortuosity of cathode - 3.0
Average pore diameter of cathode m 1.0E-6
The cell geometry is planar and equal to that of the IEA BMT, Table
5-12. The considered flow configuration is co-flow. Operational condi-
tions for the sensitivity analysis are given in Table 5-13. The material
property parameters used in the energy balance calculations are given
in Table 5-14. Table 5-21 summarizes the integral model response val-
ues for the reference case and both considered fuel gases.
Table 5-21: Response value Unit Fuel gas IEA 1 Fuel gas IEA 2
Model response K/mm
Maximum solid temperature gradient 2.2 2.0
values for the
reference case Maximum solid temperature 1327.9 1286.9
for IEA 1 and IEA Minimum solid temperature K 1212.0 1146.3
2 gas
Mean solid temperature 1282.7 1221.9
Operational voltage for 85 % fuel utilization V 0.683 0.624
194
5 Modeling
Figure 5-21 depicts the predicted temperature profiles of the solid struc-
ture for the reference case and both considered fuel gases.
Figure 5-21: 1350 100

Solid temperature
distributions and
difference for the
Difference between temperature profiles

reference case for 1300 80
IEA 1 and IEA 2
gas
1250 60
[K]
1200 40
1150 20
1100 0
0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10
Cell lenth [m]
Temperature IEA 1 Temperature IEA 2 Temperature difference
It can be seen, that at the cell inlet (x-axis: cell length equals zero) the
temperature predicted for the fuel gas IEA 1 is approx. 65 K higher than
for the fuel gas IEA 2. That can be explained with the endothermal
steam reforming reactions taking place due to the methane containing
fuel gas IEA 2. At the cell outlet (x-axis: cell length equals 0.1 m), the
difference between the two temperature distributions is decreased to
approx. 40 K. The difference between the temperature profiles of the
two fuel gases is represented by the curve named "Temperature differ-
ence". This curve is considered an indicator for the thermal prediction
behavior of the model towards changes of the fuel composition.
5.3.5.2 Variation of activation polarization parameters

In the author's model, the electrochemical conversion of hydrogen and
carbon monoxide were assumed to occur in parallel. Each of the con-
version processes is described with an activation energy value and a
pre-exponential factor. These model parameters are interdependent as
e.g. increasing only the activation energy of the hydrogen conversion
leads to a compensating carbon monoxide current and vice versa. To
avoid this, the activation energy values of both conversion processes
are varied at once. The same applies for the pre-exponential factors.
195
5 Modeling
5.3.5.2.1 Activation energy

Figure 5-22 depicts the sensitivity of the current density distribution to-
wards the anode activation energy.
Figure 5-22: 9000

Sensitivity of cur-
rent density distri- 8000
bution towards 7000

anode activation
6000
energy for IEA 1
and IEA 2 gas 5000
4000
3000
2000
1000
0
0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10
Cell lenth [m]
IEA 1 -50% IEA1 ref. IEA 1 +50% IEA 2 -50% IEA 2 ref. IEA 2 +50%
For the IEA 1 gas, the maximum current is produced at the cell inlet.
This is due to the high hydrogen partial pressure of the IEA 1 gas. From
thereon the current density constantly decreases along the cell length
because of the ongoing fuel conversion and accordingly decreased
educt partial pressures. In contrast, the IEA 2 gas yields the maximum
educt partial pressures in the region where the methane steam reform-
ing reactions are complete. Consequently, the current density is not
constantly decreasing along the cell length, but is somewhat evenly dis-
tributed with a wide and flat peak.
Decreasing the anode activation energy means to increase the activity
towards electrochemical reactions yielding higher current density values
in the cell inlet region for both gases. The increased fuel consumption in
the cell inlet region causes a considerable current density decrease at
the cell outlet for the IEA 1 gas. However, for the IEA 2 gas the current
density at the cell outlet remains constant, as the fuel consumption in
the cell inlet region is dominated by the hydrogen production through
steam reforming reactions. This leaves more hydrogen over for the latter
part of the cell. In short, the less hydrogen is produced cell internally
through steam reforming, the more does a decrease of the activation
energy yield an increased current density at the cell inlet which accom-
panies a current density decrease at the outlet.
196
5 Modeling
Table 5-22 shows that for both fuel gases, the increase of the anode ac-
tivation energy yields an increase of the temperature extremes, the
maximum temperature gradient and the mean cell temperature.
Table 5-22: Cell outlet Cell inlet Mean cell Operational

Parameter Max. dT
Model response temperature temperature temperature Voltage
value
values for differ- [K/mm] [K] [K] [K] [V]
ent anode acti-
vation energy IEA 1 gas
values for IEA 1 -50 % 2.1 1311.6 1208.1 1274.2 0.750
and IEA 2 gas
-25 % 2.1 1312.6 1208.4 1274.5 0.746
Reference 2.2 1327.9 1212.0 1282.7 0.683
+25 % 3.0 1384.7 1223.9 1320.3 0.446
+50 % 3.9 1449.5 1237.3 1364.5 0.175
IEA 2 gas
-50 % 1.7 1264.2 1142.9 1207.1 0.717
-25 % 1.7 1265.8 1143.2 1208.4 0.710
Reference 2.0 1286.9 1146.3 1221.9 0.624
+25 % 2.8 1345.4 1150.0 1255.2 0.376
+50 % 3.7 1408.4 1153.9 1290.6 0.107
This is the result of the constant fuel utilization, which requires higher
temperatures due to the lower activity towards electrochemical reactions
when the activation energy value is increased.
The operational voltage to meet the defined fuel utilization is strongly
decreased for higher activation energies while it is only little increased
for lower values, Figure 5-23.
Figure 5-23: 20%

Difference be-
tween the pre-
0%
dicted power out- 50% 75% 100% 125% 150%
Relative change to ref. case
put for varying

anode activation -20%
energy values
obtained for IEA 1 -40%
and IEA 2 gas
-60%
-80%
-100%
Percentage of reference value
IEA 1 gas IEA 2 gas
197
5 Modeling
It is important to point out that SOFCs are usually not operated at volt-
ages below 0.5 V. The predicted voltage values for increased activation
energies indicate that in reality a fuel cell employing such an anode
catalyst can not reach 85 % fuel utilization. For the lower activation en-
ergy values, the power output increases by a maximum of 15 % and de-
creases by 80 % for the higher values.
Figure 5-24: 100

Temperature dif-
ference between
the predicted
profile of IEA BMT gas 1 and 2 [K]
80
Difference between temperature
temperature dis-
tributions for vary-
ing anode activa- 60
tion energy values
obtained for IEA 1
and IEA 2 gas 40
20
0
0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10
Cell lenth [m]
50% 75% 100% 125% 150%
Figure 5-24 depicts the sensitivity of the difference between the tem-
perature profiles of the two fuel gases towards the anode activation en-
ergy. The local temperatures are strongly influenced by the locally pro-
duced power, which depends on the prevalent current density and the
operational voltage, see also Table 5-22. For a fixed operational voltage,
high current densities usually lead to high temperatures. However, en-
dothermal reactions such as steam reforming counteract this, provided
that the reaction rate is considerable.
It can be seen, that higher values of the activation energy yield an in-
crease of the difference between the predicted temperatures near the
cell inlet and a higher reforming peak. This can be traced back to the
fact that the temperature at the cell inlet for the fuel gas IEA 2 is domi-
nated by the steam reforming reaction and does not change significantly
by varying the activation energy. In contrast, the temperature at the cell
inlet increases significantly with increasing activation energy for the IEA
1 gas. The reason for this is that the operational voltage was strongly
reduced in order to meet the required overall fuel utilization. As a con-
198
5 Modeling
sequence, the amount of produced power is low even though the current
density is only slightly decreased with increasing activation energy. As
less electrical power is produced from a still considerable amount of
converted fuel, more sensible heat is released from the electrochemical
reactions resulting in higher local temperatures.
Lowering the anode activation energy causes an increase of the tem-
perature difference towards the cell end. Similar to the above discussed
increase of the activation energy, the cell inlet temperature of the IEA 2
gas remains almost equal to the reference case. However with low acti-
vation energy values, the high amount of produced power plus the heat
consumed by the endothermal steam reforming leads to a lower tem-
perature for the IEA 2 gas at the cell outlet compared to the reference
case. For the IEA 1 gas the missing steam reforming cooling is compen-
sated to a smaller extent by low current densities at the cell outlet due to
fuel depletion. Hence, the temperature difference between the two
gases is increased at the cell end.
Figure 5-25 depicts the sensitivity of the current density distribution to-
wards the cathode activation energy. Note that the cathode activation
energy was not decreased by more than 20 % as the associated high
electrochemical activity at the cell inlet for both fuel gases caused sub-
stantial fuel depletion effects at the cell outlet, which resulted in numeri-
cal instability of the employed solver.
Figure 5-25: 9000

Sensitivity of cur-
bution towards 7000

cathode activation
2
energy for IEA 1 6000
and IEA 2 gas 5000
4000
3000
2000
1000
0
0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10
Cell lenth [m]
IEA 1 -20% IEA1 ref. IEA 1 +50% IEA 2 -20% IEA 2 ref. IEA 2 +50%
For the reference case the anode activation losses are slightly higher
than the cathode activation losses. By decreasing the cathode activation
199
5 Modeling
energy the anode activation polarization becomes dominant. As a con-

sequence of the dependency of the anode activation polarization on the
educt partial pressures, the current density increases near the cell inlet
or the end of the reforming region due to the high educt partial pres-
sures and decreases at the cell outlet due to the upstream fuel con-
sumption. This can be observed for both investigated fuel gases.
In contrast, an increase of the cathode activation energy yields a shift
from anode to cathode activation polarization as dominant activation
loss mechanism. The current density then depends on the oxygen par-
tial pressure rather than on the educt partial pressures at the anode.
The oxygen partial pressure does not change significantly along the cell
length due to the considerably smaller air utilization as compared to the
fuel utilization. For the IEA gas 1, the high current density values at the
cell inlet are inhibited by the high cathode polarization and less hydro-
gen is consumed. Consequently, more hydrogen is left over for conver-
sion in the latter parts of the cell. This results in a very even current
density distribution for high cathode activation energy values.
For the IEA 2 gas an increase of the cathode activation energy strongly
reduces the already small current density at the cell inlet. Similar to the
IEA 1 gas, more educts are left for conversion in the latter part of the cell.
However, in contrast to the IEA 1 gas, a current density peak can be ob-
served at the cell outlet as additional hydrogen is produced via steam
reforming reactions and a considerable oxygen partial pressure remains
present in this region. In other words, increasing the cathode activation
energy yields a considerable shift of the typical steam reforming induced
current density peak for hydrocarbon containing fuel gases towards the
cell end. The less hydrocarbons are present in the fuel gas, the more
even the current density distribution becomes.
Table 5-23 shows that for both fuel gases, the increase of the cathode
activation energy yields an increase of the cell outlet temperature
whereas the cell inlet temperature remains almost constant. This results
in higher temperature gradients. The cell inlet temperature stays con-
stant because the smaller amount of fuel conversion at the cell inlet is
compensated by more released sensible heat. This is because less
electrical power is produced due to the lower operational voltage.
200
5 Modeling

Parameter Max. dT
value
ent cathode
activation en- IEA 1 gas
ergy values for -20 % 2.3 1322.4 1218.8 1284.7 0.705
IEA 1 and IEA 2
-10 % 2.3 1323.9 1216.8 1284.2 0.699
gas
Reference 2.2 1327.9 1212.0 1282.7 0.683
+25 % 2.1 1368.1 1205.1 1290.3 0.520
+50 % 2.6 1431.6 1209.1 1319.5 0.258
IEA 2 gas
-20 % 2.2 1280.2 1149.6 1225.6 0.656
-10 % 2.2 1281.7 1148.3 1224.3 0.648
Reference 2.0 1286.9 1146.3 1221.9 0.624
+25 % 2.4 1334.9 1143.8 1234.0 0.424
+50 % 3.3 1397.3 1145.4 1258.6 0.162
The impact of the cathode activation energy on the required operational

voltage to maintain the defined fuel utilization is depicted in Figure 5-26.
Figure 5-26: 20%

Difference be-
tween the pre- 0%
dicted power out- 50% 75% 100% 125% 150%
put for varying

cathode activation -20%
energy values
obtained for IEA 1 -40%
and IEA 2 gas
-60%
-80%
-100%
IEA 1 gas IEA 2 gas
Similar to the analysis of the anode activation energy, it is emphasized

that operational voltages below 0.5 V are usually not applied.
With a maximum of approx. 5 % power output increase, the decrease of
the cathode activation energy has a smaller impact as compared to the
decrease of the anode activation energy. This can be traced back to the
dominating role of the anode activation polarization which was calcu-
lated based on the reference parameters and hence inhibits a strong
power output increase.
201
5 Modeling
Increasing the cathode activation energy yields a maximum power out-

put decrease of approx. 75 %. In this case, the reason for the smaller
impact is that the cathode activation polarization dominates the process
thereby weakening the impact of fuel depletion towards the cell end.
Thus, the power loss, due to higher cathode activation losses, is coun-
teracted by lower anode activation losses towards the cell end.
Figure 5-27: 100

Temperature dif-
ference between
the predicted 80
temperature dis-
ing cathode acti-
Temperature [K]
60
vation energy
values for IEA 1
and IEA 2 gas 40
20
0
0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10
Cell lenth [m]
80% 90% 100% 125% 150%
Figure 5-27 shows the sensitivity of the difference between the tempera-
ture profiles of the two fuel gases towards the cathode activation energy.
Comparing Figure 5-27 and Figure 5-24 one could pre-maturely con-
clude that the cathode activation energy has a weaker impact than the
anode activation energy values on thermal behavior of the fuel cell for
the two investigated fuel gases. However, the current density distribu-
tions for different cathode activation energy values reveal a different pic-
ture.
The strong reforming peak, which was discussed for higher anode acti-
vation energies, cannot be observed for the increased cathode activa-
tion energy values. Similar to the anode activation energy variation, the
temperature at the cell inlet considering the IEA 2 gas is dominated by
the occurring steam reforming reactions and does barely change with
varying cathode activation values.
Considering the IEA 1 gas with higher cathode activation energy, the
temperature at the cell inlet is slightly lower then in the reference case.
The reason for that is that the produced current in this region is strongly
202
5 Modeling
reduced compared to the reference case due to the low oxygen partial
pressure at the cathode compared to the initial hydrogen partial pres-
sure of the IEA 1 gas. However, a strong temperature decline is inhib-
ited by the low operational voltage, which results in a tangible amount of
released sensible heat sufficient to maintain the inlet temperature on a
level close to the reference case. Hence, the temperature difference
found at the cell inlet for increased cathode activation energy values is
slightly lower than for the reference case, which in turn explains the
missing steam reforming peak.
Towards the cell end, the increased cathode activation causes higher
current density values due to less fuel depletion in the front parts of the
cell. This can be observed for both fuel gases. However, the tempera-
ture increase towards the cell end for the IEA 2 gas is stronger due to
the previously discussed current density peak near the cell end. This
explains the slightly lower temperature difference between the two in-
vestigated fuel gases at the cell end for higher cathode activation energy
values as compared to the reference case.
As discussed for the current density distribution, decreasing the cathode
activation energy puts the anode polarization into a dominant position.
As the anode polarization losses are computed using the reference val-
ues, the temperature difference progression remains almost unchanged
by the lowered cathode activation energy.
In conclusion it can be said, that the cathode and anode activation po-
larization energy values are model parameters which have to be han-
dled carefully. The thermal prediction behavior, the predicted current
density distribution and power output are all strongly influenced by these
model parameters. In particular, the impact of the anode and cathode
activation energy values has in each case a different nature. Thus, it is
important to know if anode or cathode activation polarization is dominant.
Dominating anode activation emphasizes the impact of the educt partial
pressure distribution leading to high fuel conversion where educt partial
pressures are high and vice versa. In contrast, high cathode activation
losses yield a relatively even fuel consumption along the cell due to the
almost constant oxygen partial pressures at the cathode resulting from
the usually low air utilization. Regarding the local temperatures, large
activation losses yield a higher amount of released sensible heat par-
tially compensating the lower amount of converted fuel and vice versa.
Unfortunately, these general rules are distorted in case internal reform-
203
5 Modeling
ing reactions take place. These reactions dominate the local tempera-
ture in the region where they occur.
5.3.5.2.2 Pre-exponential factor

Within the scope of variation, the impact of the anode activation pre-
exponential factor on the predicted current density distribution is small
compared to the above discussed activation energy. Increased fuel con-
sumption at the cell inlet takes place for higher pre-exponential factor
values. Analogously to decreased activation energy values, the higher
fuel conversion at the cell inlet yields fuel depletion induced lower cur-
rent densities at the cell outlet. However, the magnitude of changes is
almost negligible. The sensitivity of the predicted power for constant fuel
utilization towards the anode pre-exponential factor is shown in Figure
5-28. In contrast to the activation energy, a decrease of the pre-
exponential factor leads to a decrease of power. This is because, in Eq.
92 to Eq. 94, the exchange current density depends linearly on the pre-
exponential factor, whereas it depends on the multiplicative inverse of
the activation energy.
Figure 5-28: 10%

Difference be-
tween the pre-
dicted power out-
5%
put for varying
anode pre-
exponential factor
values for IEA 1 0%
and IEA 2 gas 50% 75% 100% 125% 150%
-5%
-10%
IEA 1 gas IEA 2 gas
For lower anode pre-exponential factor values, the power output de-
creases by a maximum of 8 %, while it increases by a maximum of 4 %
for higher anode pre-exponential factor values. Similar to the variation of
the activation energy values, the IEA 2 gas shows a more pronounced
response. The reasons for which were discussed above.
The minor changes of the current density distribution appear for both
gases to almost equal rates. Further, as shown in Figure 5-28, the an-
204
5 Modeling
ode pre-exponential factor has a comparably small impact on the opera-

tional voltage or predicted power, respectively, for both gases. Together
this yields almost negligible changes of the temperature difference. The
reason for that is that the exchange current density linearly depends on
the pre-exponential factor in contrast to the exponential dependency on
the activation energy value. Similar results were found in [202]. Figure
5-29 depicts the sensitivity of the temperature distribution trend towards
the anode pre-exponential factor.
Figure 5-29: 100

Temperature dif-
ference between
the predicted
80
temperature dis-
ing anode activa- 60
tion pre-
exponential factor
values for IEA 1 40
and IEA 2 gas

20
0
0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10
Cell lenth [m]
50% 75% 100% 125% 150%
The impact of the cathode activation pre-exponential factor on the pre-

dicted current density distribution is slightly more pronounced than for
the anode activation pre-exponential factor.
Figure 5-30: 10%

Difference be-
tween the pre-
dicted power out-
5%
put for varying
cathode pre-
exponential factor
values for IEA 1 0%
and IEA 2 gas 50% 75% 100% 125% 150%
-5%
-10%
IEA 1 gas IEA 2 gas
205
5 Modeling
In contrast, the opposite is the case for the impact on the predicted
power, Figure 5-30. For lower cathode pre-exponential factor values, the
power output decreases by a maximum of 5 %, while it increases by a
maximum of 2 % for higher cathode pre-exponential factor values. The
reasons for that are the same as already discussed for the cathode acti-
vation energy variation. Higher cathode activation pre-exponential fac-
tors put the anode activation into the dominant role, while lower values
lead to a more homogeneous current density distribution, which equal-
izes fuel depletion losses at the cell ends.
Figure 5-31 depicts the low sensitivity of the temperature distribution
trend towards the cathode pre-exponential factor.
Figure 5-31: 100

Temperature dif-
ference between
the predicted
80
temperature dis-
ing cathode acti- 60
vation pre-
exponential factor
values for IEA 1
40
and IEA 2 gas
20
0
0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10
Cell lenth [m]
50% 75% 100% 125% 150%
In summary it can be said, that the cathode and anode pre-exponential

factor values are model parameters featuring only a minor impact on the
thermal prediction behavior, the predicted current density distribution
and the predicted power output.
5.3.5.3 Variation of diffusion polarization parameters

In order to calculate the diffusion losses both the composition of the gas
mixture and the properties of the porous media through which the specie
of interest diffuses are required. The porous media properties include
the pore diameter, which is important for Knudsen diffusion, and the po-
rosity as well as the tortuosity, see section 5.3.1.4.3. Besides the gas
composition and the properties of the porous media, diffusion losses fur-
206
5 Modeling
ther depend on the prevalent temperature, the diffusion path length and
prevalent current density. The latter is causing the concentration gradi-
ent between the triple phase boundary (TPB) where the electrochemical
reactions take place and the bulk gas phase. The diffusion path length
has a direct impact on the concentration at the end of the diffusion path.
In particular, the educt partial pressure at the TPB decreases with in-
creasing diffusion path length. Hence, with respect to diffusion losses,
the investigated type of support design is of major importance. Cathode
supported cells, such as are the tubular cells of the Siemens AG, feature
a stronger cathodic diffusion limitation then e.g. electrolyte supported
cells, because of their approx. 20 times longer diffusion path through the
cathode. At low and intermediate current densities, diffusion losses
however are usually small or even negligible. Nevertheless, this sensitiv-
ity study was carried out considering an electrolyte-supported planar cell
in order to investigate whether the according material properties could
lead to a change of these coherences.
The variation of the pore diameter was omitted as it merely re-enforces
the impact of Knudsen for smaller pore diameter values or weakens it
for larger pore diameters. Instead, the material parameters influencing
both diffusion mechanisms, namely the porosity and tortuosity, were var-
ied. For the reference case the quotient of the interdependent porosity
and tortuosity has the value 0.166 assuming a tortuosity of 3 and a po-
rosity of 0.5. It was decreased by 70 % to 0.05, which corresponds to a
porosity of 0.25 and a tortuosity of 5 (typical values of "dense mem-
branes"), and increased by 110 % to 0.35, which corresponds to a po-
rosity of 0.7 and a tortuosity of 2, in order to cover the entire range of
technically possible values.
The current density distribution for varied anode material parameters
remained unchanged. Therefore, no discussion on this aspect is under-
taken. In contrast, Figure 5-32 depicts the sensitivity of the current den-
sity distribution towards the cathode material parameters. Increasing the
parameter quotient does not produce any effect. This can be explained
by the magnitude of the diffusion losses for the investigated operational
conditions, which is about two orders of magnitude lower than ohmic
and activation losses. However, the current density progression com-
puted with the dense membrane values shows a considerable decrease
of the current density at the cell inlet for the IEA 1 gas. This indicates
that in this region, the reached current density is close to the diffusion
limited current density of the cathode electrode. In the latter cell parts
207
5 Modeling
higher current density is a result of the higher educt partial pressures

discussed above. For the IEA 2 gas, the same phenomenon can be ob-
served, however to a lesser impact.
Figure 5-32: 9000

Sensitivity of cur-
bution towards 7000

cathode diffusion
material parame- 6000
ters for IEA 1 and 5000

IEA 2 gas
4000
3000
2000
1000
0
0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10
Cell lenth [m]
IEA 1 -70% IEA 1 ref. IEA 1 +110% IEA 2 -70% IEA 2 ref. IEA 2 +110%
For the increased porosity and decreased tortuosity, the power output
remains constant. In contrast, assuming dense membrane properties for
the anode yields a decreased power output of 0.5 %. Decreasing the
cathode material properties quotient results in a 2.5 % decreased power
output. A cause for this somewhat stronger response can be found in
the smaller effective diffusion coefficient of oxygen as compared to hy-
drogen.
Figure 5-33: 100

Temperature dif-
ference between
the predicted
80
temperature dis-
ing cathode diffu- 60
sion material pa-
rameter for IEA 1
and IEA 2 gas 40
20
0
0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10
Cell lenth [m]
30% 60% 100% 155% 210%
208
5 Modeling
Figure 5-33 shows the sensitivity of the temperature difference towards

the cathode diffusion material parameter. The dense membrane like pa-
rameter set leads to a slightly decreased cell inlet temperature for the
IEA 1 gas as less fuel is electrochemically converted than for the refer-
ence case. For the IEA 2 gas the temperature remains on the reference
level due to only minor changes of the prevalent current density and the
dominating steam reforming reactions. Overall the temperature differ-
ence decreases at the cell inlet for the dense membrane parameter set.
In summary it can be stated that for the investigated electrolyte sup-
ported cell the anode diffusion material parameters have almost no im-
pact on the model results. In contrast, the cathode diffusion material pa-
rameters can be set such that diffusion limitations become relevant to
some extent. It is important to point out, that this is especially the case
when other support designs with long current path, such as cathode and
anode, are investigated. Nevertheless, for the investigated support de-
sign and operating conditions the impact on the predicted power output
and the thermal behavior can be considered as negligible.
5.3.5.4 Variation of heat exchange parameters

The author's model considers convective heat transport between the
gases in the gas channels and the solid structure as well as conductive
heat transport within the solid structure. As can be seen in Eq. 161 and
Eq. 164, the convective heat streams depend on the corresponding heat
transfer coefficient, the heat exchange surface area and the temperature
difference between the gases and the solid structure. The heat transfer
coefficient is a function of the Nusselt number (Nu), the thermal conduc-
tivity of the considered gas mixture and the hydraulic diameter of the
corresponding gas channel. In the presented model, Nu is assumed to
be constant at a value of 4. This value lies between the two special
cases of heat transfer with constant heat flux, Nu equal to 4.36, and
with constant surface temperature, Nu equal to 3.66, and is therefore
deemed a valid assumption, [190]. The sensitivity analysis, considering
both mentioned values, was carried out in order to asses the impact of
the presented assumption. An investigation of Nu values higher than
4.36 and lower than 3.66 was not carried out. This is because lower Nu
values are unlikely in typical operating conditions of SOFCs. Higher Nu
values can be found for turbulent flow conditions. However, this case
was excluded from the current model.
209
5 Modeling
The current density distributions of both investigated gases are not

changed by varying the Nu values. Consequently, the change of pre-
dicted power output is in the magnitude of 0.1 %, hence insignificant.
Figure 5-34 depicts the sensitivity of the temperature difference towards
Nu. The lower Nu value does not have a noticeable impact on the pre-
dicted temperature distribution difference for the two investigated fuel
compositions. However, the increased Nu value yields slightly lower
temperatures at the cell inlet for the IEA 1 gas and higher temperatures
for the IEA 2 gas overall resulting in a smaller temperature difference at
the cell inlet.
Figure 5-34: 100

Temperature dif-
ference between
the predicted
80
temperature dis-
tributions for dif-
ferent Nusselt 60
numbers for IEA 1
and IEA 2 gas
40
20
0
0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10
Cell lenth [m]
3.66 4.00 4.36
The observed opposite trends of both gases can be explained through

the driving temperature difference between the solid structure and the
inlet gas temperature. For the IEA 1 gas, the solid temperature is approx.
27 K higher than the inlet gas temperature. The inlet gases are hence
being heated by the solid structure, which in turn is being cooled down
in the process. Increased heat transfer therefore results into increased
cooling of the cell inlet region. In contrast, the strongly endothermal
steam reforming reactions occurring for the IEA 2 gas at the cell inlet
generate solid structure temperatures approx. 30 K below the inlet gas
temperature. Therefore, the solid structure is heated up by the inflowing
gases. Consequently, increased heat transfer leads to increasing solid
structure temperatures explaining the observed trends.
The energy balance of the solid structure is given by Eq. 165. From this
equation it can be derived, that the solid structure temperature depends
210
5 Modeling
on the source terms originating from chemical and electrochemical reac-

tions, the convective heat transport between gases and solid structure
and the conductive heat transport. The conductive heat transport is a
function of the cross-sectional area of the solid structure through which
heat is transported, the temperature difference within the solid structure
and finally the effective solid heat conductivity represented by a material
parameter. For the reference case, the solid heat conductivity coefficient
(λs) was set to 2 W/m K, which is a typical value for the ceramic compo-
nents employed in SOFCs. The effect of the solid heat conductivity was
investigated by increasing λs to 20 W/ m K, representing the value found
in high-temperature stainless steel materials which are used in some
state-of-the-art SOFCs, and decreasing λs to 0.2 W/m K, which covers
the case of an almost non heat conducting material.
Figure 5-35 presents the predicted current density distribution for the in-
vestigated solid heat conductivity coefficients for IEA 1 and IEA 2 gas.
Figure 5-35: 10000

Sensitivity of cur-
9000
rent density distri-
bution towards the 8000
Total current density [A/m2]
solid heat conduc- 7000

tivity coefficient 6000
for IEA 1 and IEA
5000
2 gas
4000
3000
2000
1000
0
0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10
Cell lenth [m]
Higher λs values yield a strong increase of the current density at the cell
inlet and corresponding decrease at the cell outlet. In contrast, lower λs
values show a minor impact on the predicted current density distribution,
the power output and the thermal prediction behavior of the model con-
sidering different fuel gases. These results are sensible, given that the
reference case value of λs is already small. Lowering this value further
reduces the already humble conductive heat transfer to values close to
zero. The remainder of this discussion hence focuses on the impact of
increased λs values, which induce stronger changes to the energy bal-
ance of the cell.
211
5 Modeling

Parameter Max. dT
value
ent solid heat
conductivity IEA 1 gas
coefficients for -90 % 2.4 1324.9 1198.7 1274.6 0.679
IEA 1 and IEA 2
-45 % 2.3 1326.5 1205.8 1278.9 0.681
gas
Reference 2.2 1327.9 1212.0 1282.7 0.683
+450 % 1.8 1338.5 1257.3 1309.2 0.696
+900 % 1.4 1345.2 1284.2 1324.1 0.702
IEA 2 gas
-90 % 2.0 1286.5 1141.5 1216.6 0.618
-45 % 2.0 1286.5 1143.6 1219.1 0.621
Reference 2.0 1286.9 1146.3 1221.9 0.624
+450 % 2.0 1296.3 1186.3 1252.7 0.647
+900 % 1.7 1307.4 1223.5 1276.7 0.659
Table 5-24 gives the cell temperatures and operational voltage values
for varying λs values. Comparing the cell inlet temperature of both fuel
gases for the reference case and the increased solid heat conductivity it
was found that the IEA 2 gas features a stronger increase by approx.
5 K. In detail, the cell inlet temperature for the IEA 1 gas increased by
72 K versus 77 K for the IEA 2 gas, Figure 5-36.
Figure 5-36: 1400

Sensitivity of tem-
perature profile 1350
towards the solid
heat conductivity 1300

coefficient for IEA
1 and IEA 2 gas 1250
1200
1150
1100
0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10
Cell lenth [m]
Similar effects were observable for the cell outlet temperatures which in-
creased by 18 K for the IEA 1 gas and 20 K for the IEA 2 gas. Summed
up these temperature differences result in an increased mean cell tem-
peratures by 13 K for the IEA 2 gas (54 K increase) as compared to the
IEA 1 gas (41 K increase). The differently strong pronounced tempera-
212
5 Modeling
ture increase of the two investigated gases is attributed to the endo-

thermal steam reforming reactions occurring at the cell inlet for the IEA 2
gas, which represent a considerable heat sink and additional driving
force for solid heat conduction. Increased λs values allow more heat to
be transferred from the latter parts of the cell to the cell inlet, resulting
into considerably higher temperature as compared to the reference case
for the IEA 2 gas.
In contrast, for the IEA 1 gas no endothermal reactions take place at the
cell inlet. The heat transfer from the cell outlet to the cell inlet region is
thus only driven by temperature equalization. The more pronounced
mean temperature increase due to the less distinct sub cooling region
for the IEA 2 gas in turn reveals the stronger impact of the solid heat
conductivity coefficient on the predicted power output which is depicted
in Figure 5-37.
Figure 5-37: 7.5%

Difference be-
tween the pre-
dicted power out-
put for varying 5.0%
solid heat conduc-

tivity coefficients
obtained for IEA 1 2.5%
and IEA 2 gas
0.0%
10% 100% 1000%
-2.5%
IEA 1 gas IEA 2 gas
The lower λs values lead to approx. 1 % power output decrease while it

increases by a maximum of 5.5 % for higher λs values.
In this respect, it is important to emphasize that all the model parame-
ters used for the calculation of voltage losses were held constant. Keep-
ing this in mind, the revealed power output dependency on the solid
heat conduction coefficient is considerable. This also provides an addi-
tional justification for the efforts undertaken by the SOFC community to
use metallic materials for the interconnector plates in planar cells.
Despite the inhomogeneous current density distribution shown in Figure
5-35, higher λs values lead to a more homogeneous local temperature
distribution as shown in Figure 5-36. The dominating role of the steam
213
5 Modeling
reforming reactions considering the cell inlet temperatures is weakened

as the corresponding heat requirements are covered by the enhanced
solid heat conduction. The consequences of that are that the predicted
temperature profiles for different fuel gases converge, resulting in a flat-
tened temperature distribution trend progression without considerable
steam reforming peak as for the reference case, Figure 5-38.
Figure 5-38: 100

Temperature dif-
ference between
the predicted
80
temperature dis-

ing solid heat
60
conductivity val-
ues obtained for
IEA 1 and IEA 2
gas 40
20
0
0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10
Cell lenth [m]
10% 55% 100% 550% 1000%
Therefore it can be said that the Nusselt number has almost no impact
on the model results which justifies the according assumption. The heat
conductivity coefficient strongly affects the predicted current density dis-
tribution and local temperature distribution. The current density tends to
be less uniform for higher heat conductivity values whereas the opposite
is the case for the temperature distribution. Further for high solid heat
conduction coefficients, the dominating role of the steam reforming reac-
tions occurring at the cell inlet is weakened. Hence, the differences for
the local temperatures between hydrocarbon containing gases and pure
syngases are reduced. A considerable impact on the predicted power
output was found despite the fact that the model parameters used for
the electrochemical performance prediction were held constant at their
reference value. However, the solid heat conduction is a material pa-
rameter which can be reliably determined. Therefore is it not likely to in-
troduce an uncontrolled uncertainty to the model results.
214
5 Modeling
5.3.5.5 Investigation of different applied kinetics models

In SOFCs, hydrocarbons can be converted to hydrogen and carbon
monoxide through steam reforming reactions (STR) due to the usually
employed nickel based anode material. The STR reaction rate affects
the hydrogen and carbon monoxide as well as the related Nernst volt-
age distribution and the temperatures where the STR reactions take
place. The STR thereby indirectly affects the electrochemical reactions
occurring in SOFCs. To asses this impact, five different STR kinetics
models were investigated, see section 5.3.1.5: Eq. 129 to Eq. 133.
The reaction order of steam is controversially discussed in the literature
related to STR kinetic models. The reference case of the sensitivity
analysis employs the STR applied kinetics power law model (PL) pro-
posed by Achenbach, [204]. Achenbach found a reaction order equal to
zero for steam. Ahmed also formulated a PL kinetics model for the STR,
[205]. In contrast to Achenbach, Ahmed found a negative reaction order
for steam. Another PL model was developed by Leinfelder, who derived
a first-order dependency of the STR on the steam partial pressure from
his experimental data, [53]. Drescher chose a Langmuir-Hinshelwood
approach (LH) to describe the STR of methane over nickel cermet elec-
trodes of SOFCs, [207]. LH approaches feature variable effective reac-
tion orders depending on the prevalent partial pressures and tempera-
tures of the corresponding species. The last investigated approach is
based on the equilibrium constant of the methane STR.
Figure 5-39 shows the axial distribution of methane, hydrogen and car-
bon monoxide for the different STR kinetic models. The Achenbach
model predicts the end of the STR region at around a quarter of the total
cell length. The model proposed by Ahmed predicts slightly higher reac-
tion rates, hence completing the STR reactions after a fifth of the cell
length. Dreschers LH model yields reaction rates, which are about half
of those obtained from the Achenbach model. The Leinfelder model re-
sults are almost congruent with the equilibrium approach, indicating that
the nickel-cermet investigated by Leinfelder was highly active. Note that
the considered diffusion limitation inhibits the completion of the STR re-
actions within the first control volume even though equilibrium is as-
sumed. The faster the STR occurs, the higher is the corresponding hy-
drogen and carbon monoxide peak as the electrochemical consumption
is slower than the production through the STR reactions.
215
5 Modeling
Figure 5-39: 60%

Sensitivity of
methane, hydro-
50%
gen and carbon
monoxide distribu-
Molar fraction [mol.-% wet]

tion towards STR 40%
kinetics
Hydrogen
30%
20%
Carbon monoxide
10%
Methane
0%
0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10
Cell lenth [m]
Figure 5-40 shows the predicted current density and Nernst voltage dis-
tribution for the different STR kinetics models. For faster STR kinetics,
the current density at the cell inlet increases yielding slightly decreased
current density values at the cell outlet due to fuel depletion effects.
Figure 5-40: 4000 0.975

Sensitivity of cur-
rent density and
Nernst voltage
distribution to-
wards STR kinet-
3000 0.950
Nernst voltage [V]

ics
2000 0.925
1000 0.900
0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10
Cell lenth [m]
The reason for the higher current density values at the cell inlet ob-
served for faster STR kinetics can be found in the predicted Nernst volt-
age within the STR region. From Eq. 80 and Eq. 81 it can be derived
that higher hydrogen and carbon monoxide partial pressures yield
higher Nernst voltages. Further, the Nernst voltage increases with de-
creasing temperatures due to the endothermal STR reactions.
216
5 Modeling
Figure 5-41 and Table 5-25 show that the cooling effect in the STR re-
gion is more pronounced for faster STR kinetics as the locally higher
heat requirements cannot be fully covered via heat conduction and con-
vection. However, towards the cell outlet, fast STR kinetics yield higher
temperatures than the slow kinetics. Overall, faster STR kinetics yield
higher average cell temperatures.
Figure 5-41: 1300

Sensitivity of the
temperature dis-
tribution for the
IEA 2 gas towards 1250
STR kinetics
Temperature [K]
1200
1150
1100
0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10
Cell lenth [m]
Drescher LH Achenbach PL Ahmed PL Leinfelder PL Equilibrium
Table 5-25 gives the predicted maximum temperature gradients and op-
erational voltages to maintain 85 % fuel utilization.

Parameter Max. dT
value
ent methane
steam reforming IEA 2 gas
applied kinetics Reference 2.0 1286.9 1146.3 1221.9 0.624
for IEA 2 gas
Drescher LH 2.2 1288.4 1140.7 1223.2 0.626
Ahmed PL 2.1 1286.8 1143.2 1221.4 0.624
Leinfelder PL 3.1 1293.5 1141.0 1229.6 0.630
Equilibrium 3.2 1293.7 1141.2 1230.0 0.631
As a consequence of the more pronounced cooling effect and the result-

ing higher cell temperatures towards the cell outlet, the maximum tem-
perature gradients increase with faster STR kinetics. The increased
mean cell temperatures allow for higher operational voltage values due
217
5 Modeling
to less activation and ohmic voltage losses. However, the gained power
output increase is in the order of 1 % for the equilibrium model as com-
pared to the reference case considering the Achenbach model.
Figure 5-42 depicts the temperature difference obtained for the different
STR kinetics models. The faster the STR reactions occur, the flatter and
the nearer to the cell inlet is the STR peak.
Figure 5-42: 100

Temperature dif-
ference between
the predicted
80
temperature dis-
tributions for dif-
ferent STR kinetic 60
models obtained
for IEA 1 and IEA
2 gas 40
20
0
0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10
Cell lenth [m]
Drescher LH Reference - Achenbach PL Ahmed PL Leinfelder PL Equilibrium
In conclusion it can be stated that the predicted temperature distribution

is considerably affected by the employed STR kinetics model. Faster
STR kinetics yield a stronger pronounced cooling effect of the STR reac-
tions which is counteracted by higher temperatures in the latter parts of
the cell resulting from higher current density values at the cell inlet. Re-
garding the predicted temperature differences for different gases, the
STR kinetics model also has a considerable impact. Especially the re-
forming peak observed in the reference case almost disappears for
faster STR kinetics. In contrast, the power output is only slightly
changed. The equilibrium approach, which yields almost instant meth-
ane conversion, shows power output values increased by approx. 1 %.
5.3.5.6 Conclusions from the sensitivity analysis

It is concluded that the predicted trends regarding the difference be-
tween the temperature distributions predicted for different fuel gases are
not changed but are at the most emphasized or understated by the in-
vestigated model parameters. Especially the activation polarization en-
218
5 Modeling
ergy values, the solid heat conduction coefficient and the applied kinet-
ics of the steam reforming showed a considerable impact. However, the
impact of all variations in model parameters on the predicted tempera-
ture distribution trends remained always smaller than the impact of the
different fuel gases themselves. In this respect, the author's model has
shown a fair robustness. The predicted trends for the power output
based on different fuel gases were also hardly affected by the variations
in model parameters. With respect to the absolute value of the power
output, only the activation polarization activation energy values were
found to have a strong impact. All other investigated model parameters,
including the applied kinetics of the steam reforming reactions, had less
then 10 % variance to the reference case within the variation bandwidth.
5.4 Economic model

For the design of power generation systems, low capital cost and high
electrical efficiencies are generally conflicting requirements. The final
decision to implement a chosen system design is thus based on its prof-
itability. To determine the profitability for different B-IGFC system de-
signs, an economic model was developed. The results of the thermody-
namical system simulations conducted with the models discussed in
sections 5.1 to 5.3 provided the input data for the economic model. The
economic model calculates the size and cost estimates for the most im-
portant equipment pieces, the direct and indirect plant costs, the opera-
tional and capital cost as well as the net AC system efficiency. These
values are then translated into the average power production cost,
which allows a terminal comparison of different system designs.
Figure 5-43 shows that the accuracy of cost estimations strongly de-
pends on the invested effort, which in turn determines its cost. Similar to
other important works in the field of thermo-economic fuel cell systems
analysis, [188, 189, 212], the author's economic model belongs to the
study estimate category. This type of estimate is based on a list of the
major equipment pieces required in the investigated system, including
pumps, blowers, reactors, heat exchangers, furnaces etc. Applying
rough sizing procedures to each piece of equipment, a first cost esti-
mate can be determined. The total equipment cost is further factored to
yield an estimate for the total direct and indirect plant cost. The accuracy
of study estimates is expected to be in the range of +/- 20 to 30 %.
219
5 Modeling
Figure 5-43: High Maximum

Accuracy and cost 40 estimate cost
of different cost
Order of
estimation classes, magnitude
derived from [213] estimate
Expected accuracy of estimate [%]

30 Minimum Low
⎛ P ⎞
0.6
cost estimate
C
=⎜ ⎟
C ref ⎜⎝ Pref ⎟
⎠
Study
estimate
Preli-
20 minary
estimate
Definitive
10 estimate
Detailed estimate
0
1 2 5 10 20 50 100 200 500 1000
Cost of estimate [1000 US$]
5.4.1 Equipment purchase cost estimate functions

Many cost functions have been published over the years, all with the
aim to enable engineers to perform quick and inexpensive equipment
cost estimations. The major drawback of these functions is their rela-
tively low accuracy which mainly depends on how current the underlying
data is and on the methodology of data collection. Even with recent data,
the estimated cost of a single equipment piece can vary between -50 %
and +100 % based on current price quotations of vendors due to widely
differing quality, designs and specifications, [214]. This illustrates that
determining a cost function yielding a good estimate of equipment costs
over a wide range of equipment sizes is a challenging task.
Figure 5-44 shows a comparison of the cost functions for spiral, double
pipe, air cooled as well as shell and tube heat exchangers proposed by
Garrett, [214], and Turton, [213].
Figure 5-44: 1000000

Comparison of
equipment cost
estimates for differ- Air cooled
ent heat exchanger
types using cost
Equipment cost [2001-$]
100000
functions of Garrett
(lines without sym- Shell & tube
bols) and Turton Spiral
(lines with symbols)
10000
Double pipe
1000
1 10 100 1000 10000
Heat transfer area [m2]
220
5 Modeling
The Turton functions yield cost estimates with values on average twice
as high as the Garrett functions. Employing Eq. 176, the renown
"Chemical Engineering magazine Plant Cost Index", CEPCI, was
adopted to adjust the Garrett cost estimates based on data from 1987 to
2001, the year in which the data for the Turton functions was collected.
CEPCI present
C present = ⋅ Creference Eq. 176
CEPCI reference
The CEPCI is a lumped inflation cost indicator, which represents the av-
erage price inflation of all major equipment pieces used in the chemical
industry. Figure 5-45 shows a plot of the CEPCI from 1950 through 2007.
Figure 5-45: 600

Chemical Engi- CEPCI 2007: 530
neering Plant 500
Chemical Engineering Plant Cost Index [-]
Cost Index from

1950 to 2007
400
CEPCI 1987: 320
300
200
100
0
1950 1960 1970 1980 1990 2000 2010
Year [-]
It is likely that technology changes and unequal price inflations for dif-
ferent equipment pieces lead to considerable differences between the
estimated costs obtained from functions developed at various times.
However, keeping the general stray area of cost estimates compared to
vendor quotations in mind, the functions of both authors appear to apply
well to the comparisons of different system designs with respect to their
economical performance. The choice of function is thus a matter of taste
rather than of accuracy. The Turton functions were used in e.g. [212],
while Garretts functions were employed in e.g. [215].
As Garrett offers more specified cost functions than Turton, the cost
functions formulated by Garrett were applied in this work. Being formu-
lated in 1987, the reference CEPCI is 320. The cost estimates were up-
dated to the year 2007 with a CEPCI of 530. Further, the estimated
equipment prizes were converted from US$ to €, assuming a currency
rate of 1.4 US$ per € of the year 2007.
221
5 Modeling
5.4.2 Equipment sizing and purchase cost estimation

The investigated system designs are broken down into six operational
modules comprising typical equipment pieces. The size of the different
equipment pieces was either determined based on a general plant lay-
out assuming a 50 m x 100 m plant footprint, which is the case for
equipment required for wood transportation, exhaust stacks and control
buildings or alternatively based on the temperatures, mass and energy
flows computed with the flowsheeting model and the ATHENA SOFC
model. Following modules were differentiated:
Wood handling module
Gasification module
Gas processing module
Fuel cell balance of plant module
Fuel cell module
Auxiliary equipment module
The cost of the equipment pieces discussed below was estimated using
cost functions named similar in [214], if not stated otherwise. The cost of
high temperature equipment was adjusted through material factors for
stainless steel or other appropriate materials given in [214].
The wood handling module comprises all required steps to deliver
wood chips of a specific moisture and geometry to the gasification reac-
tor in question. This includes conveyor belts for transportation of the
wood, shredders to conform the wood as delivered to wood chips of cer-
tain geometry, storage facilities and wood drying installations.
For the transportation of the wood pieces from the truck dump to the
shredder, 25 m long 16" belt conveyors were assumed adequate. The
capacity of the shredder was assumed three times the daily wood con-
sumption of the investigated B-IGFC system, thus allowing complete
processing of the raw wood within a regular working shift.
9" screw conveyors were chosen for the transportation of the wood
chips. Wood chips are first conveyed from the shredder to a chip pile,
stocking wood for approx. 2 weeks of autonomous operation. This
amount was considered sufficient to bridge considerable wood supply
gaps. For fire protection reasons, a clearance distance of the chip pile
from the other plant equipment of 50 m was assumed. The return con-
veyor is also 50 m long and delivers the wood chips from the chip pile to
either a drying installation or a hopper.
222
5 Modeling
As a matter of simplification a convective wood chips dryer was chosen

in this analysis. This dryer type consists of a belt conveyor on which wet
wood chips are transported through a dryer housing streamed-through
by flue gas. The belt conveyor length and the dryer housing size are
roughly sized, assuming constant temperature and flow of the flue gas.
As mentioned in section 5.2.3, the drying is performed using the sensi-
ble heat of the flue gas after all heat requirements have been covered.
In a first step, the residence time in the dryer required to reach a prede-
fined humidity of the wet wood chips is calculated according to [216].
The calculation involves the wood chips mass flow, geometry, density,
temperature and humidity at the dryer inlet and outlet. Further, the tem-
perature of the gas flow used for the drying at the dryer inlet is required.
Based on the roughly estimated residence time, and assuming a con-
veyor belt width of 7.5 m, the dryer length is determined. The footprint of
the dryer housing equals the belt conveyor length plus 2 m clearance
times its width and an additional meter of clearance. The cost of the
dryer housing is estimated using the "Warehouse" cost function, [214].
Another 9" screw conveyor of 15 m length is used to transport the dried
wood chips to a hopper. The hopper volume is sized such that it stocks
enough wood for five consecutive days of operation. The hopper cost is
estimated employing the "Warehouse" cost function, [214]. From the
hopper to the gasifier, two more 15 m 9" screw conveyors are used.
The gasification module includes the gasification reactor and all addi-
tional equipment required for its operation. The up- and downdraft gasi-
fication processes considered in this work, see section 2.2.3, require
one single reactor whose cost is estimated using the "Cylindrical type
furnace" cost function, [214]. To account for the lower complexity of up-
draft compared to downdraft gasification reactors, the standard material
factor for furnaces was used for the updraft gasifier. The LHV input of
wood to those reactors is known from the flowsheet simulations and suf-
ficient for a rough sizing. Both processes use air as gasification agent.
The air flow rate is determined through a molar nitrogen balance, Eq.
177. The cost for the corresponding air blower is assessed via the "Ro-
tary, centrifugal, turbo 10 psi" cost function, [214].
y N , PG
V&air , gasi = 2 ⋅ V&PG Eq. 177
y N 2 , Air
The investigated fluidized bed gasifier, see section 2.2.3, consists of a

gasification chamber, a combustion chamber, sodium heat pipes for the
223
5 Modeling
heat exchange between the gasification and the combustion chamber, a

combustion air pre-heater, a steam generator and the associated water
pump. Similar to the fixed bed gasification reactors, the cost of the gasi-
fication and the combustion chamber are estimated using the "Cylindri-
cal type furnace" cost function, [214]. For the gasification chamber, the
entire LHV input of wood is considered, while for the combustion cham-
ber the LHV input results from the assumption that 17 mass-% of the dry
wood input in form of char is being combusted, [33]. The same assump-
tion applies to the quantification of the air flow required for the char
combustion with an air-to-fuel ratio of 1.3, [33]. For efficiency reasons,
the combustion air is pre-heated. The size of the required heat ex-
changer is computed via the mean logarithmic temperature difference,
which results from the heating of air from ambient temperature to
400 °C while the combustion flue gas is cooled from 900 to 584 °C.
Similar to all other heat exchanger sizing calculations, 22.5 W/m2 K was
assumed as heat exchange coefficient. For this high-temperature heat
exchanger the "Plate and frame" cost function is used, [214]. The water
mass flow required for the steam gasification reactions taking place in
the gasification chamber at an equivalence ratio of 1.8 is used to size
the corresponding pump. The size and cost of the required evaporator is
determined similar to the combustion air pre-heater using the 584 °C hot
combustion flue gas after the air pre-heating. The heat transfer coeffi-
cients are considered phase dependent with the same value as men-
tioned in section 5.2.3. The sodium heat pipes are important cost factors
of the investigated fluidized bed reactor. According to [168], a single
heat pipe can transport approx. 2.6 kW of heat into the gasification
chamber. For highest efficiencies, approx. 20 % of the LHV input of
wood to the gasification chamber has to be provided by the heat pipes,
[33]. Thus, the number and cost of heat pipes can be straightforwardly
estimated, assuming a unit price of 1000 € in 2007 (2007-€).
The gas processing module covers the equipment required for the gas
cleaning and conditioning. Cyclones and ceramic particle filters are used
for particle removal. Their cost is estimated based on the wet and tar-
laden producer gas flow rate at the gasifier outlet temperature. The
"Electrostatic" cost function was chosen as conservative estimate for ce-
ramic particle filters, [214].
The humidification of the producer gas includes a water pump, a spray
chamber in case liquid water shall be injected aiming at a cooling of the
producer gas and a steam generator operated with flue gas heat. The
224
5 Modeling
water pump is pricewise rated as conventional pump based on its power

computed according to Eq. 178. V&air ,sys stands for the total displacement
volume, Δpsys is the counter pressure which the pump has to handle and
ηblow represents the pump efficiency.
V&air , sys ⋅ Δpsys
PAux = Eq. 178
ηblow
The cost of the spray chamber is determined by its water flow rate, see
section 5.2.2. As the steam generator is an integral part of the heat inte-
gration network discussed in section 5.2.3, its area is known from the
flowsheeting simulation.
State-of-the-art producer gas cleaning systems operate at low tempera-
tures and usually consist of a particle removal unit and a wet scrubber.
The wash liquid, mostly water, has to be treated after its use. The
"Spray chamber" function was used for the prize estimation of wet
scrubbers. The considerably higher cost estimates account for the wash
liquid treatment. The wet scrubbers are sized according to [28], where it
is reported that approx. 2 l of water are spent to clean 1 m3n of wet and
tar-laden producer gas.
For the removal of hydrogen sulfide from producer gases, solely zinc ox-
ide adsorbers were considered using the "Activated carbon adsorber"
cost function, [214]. It was assumed that the adsorber should allow for
one month non-stop operation. The molar amount of hydrogen sulfide
which is adsorbed during this period is known from the flowsheeting
simulations, see section 5.2.6. Assuming a maximum sulfur capacity of
zinc oxide of 10 %, [149], the adsorbent mass and the cost is computed.
The adiabatic methanation of producer gas, see section 5.2.4, is carried
out in a fluidized bed reactor. The cost function "Fluid bed" is based on
the reactor volume, [214], which is sized to yield similar operating condi-
tions as applied in [40].
The catalytic partial oxidation of producer gas is conducted using coated
monoliths. Based on the experiments reported in section 3.5, a tolerable
catalyst load of 2.5 kW LHV producer gas power per monolith with 1" di-
ameter, 3" length and 400 cpsi was assumed to estimate the required
number of monoliths and the CPO reactor volume. The CPO reactor is
pricewise rated employing the "Jacketed reactor" cost function, [214]. Eq.
178 is used for the power requirement calculation of the CPO air blower.
The air flow rate is determined for an air-to-fuel ratio of 0.12.
225
5 Modeling
The fuel cell balance of plant module contains all the heat exchangers
required to adjust the producer gas and the cooling air temperature to
the predefined SOFC inlet temperatures, the cooling air blower, the
DC/AC power converter and the SOFC off-gas combustor. The heat ex-
change areas of the air pre- and super heater, the producer gas cooler
(if required) and super heater are known from the flowsheeting simula-
tion, see section 5.2.3. The costs of all heat exchangers are determined
using the "Plate and frame" cost function, except for the air pre-heater
the cost of which is estimated based on the "Shell and tube" function,
[214], see Figure 5-44.
Similar to all other air blowers, Eq. 178 is used for the power require-
ment calculation of the fuel cell cooling air blower. The air flow rate is
determined by the ATHENA SOFC model.
In 2003, the American Department of Energy together with the Depart-
ment of Defense has launched a design competition for DC/AC power
converters tailored to the application with SOFCs. The cost target of less
than 40 $ per kWel, [217], was adopted in this work.
The oxidation of the depleted SOFC fuel gas is performed in a catalytic
combustor to ensure complete conversion. Its heat duty is known from
the flowsheeting calculations, see section 5.2.7, and used for the cost
estimation based on the "Catalytic incinerator" cost function, [214].
The fuel cell module comprises the SOFC stack and all catalyst, which
have to be exchanged in regular intervals. For 2012, the American Solid
State Energy Conversion Alliance (SECA) has defined a cost target for
SOFC systems of 400 2002-$ per kWel output using natural gas as fuel,
[59]. Despite this objective was formulated for SOFC systems, it was
assumed for the stack cost to be more conservative in this work. In sys-
tems using an adiabatic methanation, the catalyst cost are estimated
based on a price of 50000 2007-€ per ton corresponding to the average
nickel price in 2007. The catalyst mass is determined to yield operating
conditions as applied in [40]. For systems using a CPO, 100 € was as-
sumed as piece prize for a 1" CPO catalyst.
The auxiliary equipment module summarizes all equipment units
which are not directly related with the electricity generation. These are
the stack, the emergency flare and the heat exchanger employed to ex-
tract useful heat from the SOFC flue gas.
The stack is assumed by default 20 m height and pricewise rated using
the "Short chimney" cost function, [214].
226
5 Modeling
In case of an emergency, the producer gas may have to be flared to

prevent serious damage of the equipment downstream of the gasifier.
An elevated emergency flare was therefore considered necessary. The
flare cost is evaluated based on the full producer gas mass flow rate.
The size of the heat exchanger employed to extract useful heat from the
SOFC flue gas is known from the flowsheeting simulation and the cost is
determined similar to all other high-temperature heat exchangers.
5.4.3 Total investment, direct and indirect plant costs

The total investment, TI, for technical installations is the sum of direct
and indirect plant cost plus contingencies and contractor fees. The direct
plant cost, DPC, includes not only the purchase cost of the equipment
but also the cost for shipment, installation and commissioning of each
equipment piece. These additional costs are determined by multiplying
the single equipment cost estimates with a module factor. For the
equipment pieces, which are pricewise rated according to [214], the de-
noted average module factors have been used. For non-standard
equipment such as the SOFC stack, a module factor of 1.5 was gener-
ally assumed. Although this approach is rather general and does not
ameliorate the accuracy of the cost estimation, the estimated direct plant
costs were expected to be in the right general cost range for preliminary
estimates, [214].
The indirect plant costs, IPC, comprise the project development and fi-
nancial cost as well as the interest, engineering and administration cost
during the construction period. All costs attributed to the IPC arising
from borrowing of funds depend on the length of the construction period,
CP, the prevailing interest rate, IR, and the credit given by the debt por-
tion, DP, and the TI. The factors and the equation used to determine the
IPC, the contingencies and the contractor fees are summarized in Table
5-26. These values are in good agreement with e.g. [215].
Table 5-26: Factor assumed [%] or equa-

Cost component Basis
Factors and equation used
tion used for the
Project development cost 8.75 DPC
calculation of the
indirect plant cost Financial cost 2.0 DP*TI
Interest during construction (1 + IR )
CP
-1 DP*TI
Engineering and administra-
5.0 DPC
tion during construction
Contingencies 10.0 DPC+IPC
Contractor fees 5.0 DPC+IPC
227
5 Modeling
For B-IGFC systems, a CP of 1 year, an IR of 8 % and a DP of 75 %

were assumed.
5.4.4 Net AC power efficiency

The performance of power generation systems is rated based on their
ability to convert the energy of a feedstock into electrical AC grid power.
The ATHENA SOFC model computes the DC power efficiency of the
SOFC. Based on this, the ASPEN PLUS flowsheet model yields the DC
power efficiency of the investigated B-IGFC system. The net AC power
efficiency, ηnet,AC, is defined according to Eq. 179.
⎛ PAux ⎞
ηnet , AC = ηCG ⋅ηGP ⋅η SOFC , DC ⋅ xS ,loss ⋅ ⎜⎜1 − ⎟ ⋅η Inv
⎟
Eq. 179
⎝ xS ,loss ⋅ PSOFC , DC ⎠
The cold gas efficiency, ηCG, is given by the ratio between the chemi-
cal energy content of the producer gas (PG) exiting the gasifier to that of
the converted wood, Eq. 180.
V&PG ⋅ LHVPG
ηCG = Eq. 180
m& feed ⋅ LHV feed
The gas processing efficiency, ηGP, relates the chemical energy con-
tent of the processed gas at the SOFC inlet (CG) to that of the raw pro-
ducer gas (PG), Eq. 181. It thus describes chemical energy losses oc-
curring in the gas processing system.
V& ⋅ LHV
ηGP = &CG CG
Eq. 181
VPG ⋅ LHVPG
The SOFC DC efficiency, ηSOFC,DC, includes the fuel utilization and the
electrochemical conversion efficiency of the SOFC. It is defined as the
ratio between the produced DC power and the chemical energy content
of the processed gas at the SOFC inlet (CG), Eq. 182
P
η SOFC, DC = & SOFC , DC Eq. 182
VCG ⋅ LHVCG
Sulfur induced power losses were considered in this economic model

through a correlation derived from sulfur experiments with the Hexis
SOFC, see section 3.4.5. The fair sulfur tolerance of the Hexis SOFC
was assumed for all SOFCs despite the employed anode catalysts
strongly varying for different SOFCs. Eq. 183 yields the expected per-
centage of power output, xS,loss, of the Hexis SOFC operated with sulfur
containing fuel gas.
228
5 Modeling
⎛ yS ⎞ ⎛ yS ⎞
xS ,loss = 0.52695 + 0.20716 ⋅ exp⎜ − ⎟ + 0.26521 ⋅ exp⎜ − ⎟ Eq. 183
⎝ 3.7608 ⎠ ⎝ 27.0748 ⎠
yS represents the molar fraction of the total sulfur content in the fuel gas
in parts per millions. Note that the loss of electrical power due to sulfur is
added to the heat output of the SOFC.
The auxiliary power requirements, PAux, originating from air blowers
for e.g. the SOFC cooling and pumps for e.g. producer gas humidifica-
tion, were calculated according to Eq. 178. As no momentum balance
was performed in the thermodynamical simulations, a counter pressure,
Δpsys, of 375 mbar and a blower or pump efficiency, ηblow, of 75 % was
generally assumed for the auxiliary power requirement estimation.
The inverter efficiency, ηInv, for the conversion of the DC produced by
the SOFC to AC deliverable to the grid was assumed to be 92.5 %.
5.4.5 Power production costs

The calculation of the power production cost is based on the average
yearly power output and cost during a given planning horizon, PH, dis-
counted to the year of start of production, SoP, Figure 5-46. A PH of 20
years was considered and SoP in 2008 was generally assumed. All in-
put data required for the power production cost is summarized in Table
5-27 at the end of this paragraph.
Figure 5-46:
Total investment Wood price szenario Fixed operation cost
Structure of the
Dept portion Plant factor Insurance
power production
Intrest rate Wood consumption Operation
cost calculation, maintenance
Construction period
derived from [218] Utilities
Depreciation period
Variable cost
Captial cost Fuel cost Operational cost
Yearly cost analysis Heat sales revenues
Planning horizon Present value of all yearly

costs throughout the
Imputed interest rate planning horizon
Planning horizon Leveled yearly cost

throughout the
Imputed interest rate planning horizon
Average yearly Average power

electricity production
production cost
229
5 Modeling
The starting point for the calculation is a detailed cost analysis on either
a yearly or monthly basis. Most of the regular and customary recurring
costs are assumed subject to 2 % inflation, IF. This includes costs for
operation and maintenance, feedstock, utilities and replacement of
worn-out equipment. Following cost factors were considered:
Fuel cost
The fuel cost for the operation year i, Cfuel,i, is computed via Eq. 184.
C fuel ,i = m& wood ⋅ 8760 ⋅ PF ⋅ Cwood ⋅ (1 + IF + WI )
(i + SoP − 2007 )
Eq. 184
The wood mass flow per hour, m& wood , is known from the flowsheeting
simulation. The plant factor, PF, describes the availability of the plant
and is assumed to be 85 %. The wood price, Cwood, of 84 2007-€ per
dry ton in the year 2007 is in accordance with the German Coordinat-
ing Office for Renewable Raw Materials, [219]. To account for the in-
creasing future demand of wood, a market-driven wood price in-
crease, WI, of 1 % per annum in addition to regular inflation, IF, was
added.
Insurance cost
For the insurance costs, a fixed rate per annum of 1 % of the total in-
vestment was assumed over the entire planning horizon, Eq. 185.
CIns ,i = 0.01 ⋅ TI Eq. 185
Operation and maintenance cost

The operation and maintenance costs (O+M) are subject to inflation
and were assumed to be 4 % of the DPC per annum, which in-
cludes all expenses for labor and minor repairs, Eq. 186.
CO + M ,i = (0.04 ⋅ DPC ) ⋅ (1 + IF ) Eq. 186
i
Utility cost
Water for the producer gas humidification was considered the only
utility cost factor with a price of 0.94 2007-€ per m3. The required wa-
ter flow rate is known from the flow sheeting simulation.
Cutility ,i = V&water ⋅ 8760 ⋅ PF ⋅ Cwater ⋅ (1 + IF ) Eq. 187
i
Revenues from heat sales

All the useful heat, an amount known from the flowsheeting simula-
tion, is assumed to be sold for a price of 0.04 2008-€ per kWhth. The
corresponding yearly revenues, Rheat,i, are computed via Eq. 188.
230
5 Modeling
Rheat ,i = Pth ⋅ 8760 ⋅ PF ⋅ pheat ⋅ (1 + IF ) Eq. 188

i
Pth stands for the output of useful heat from the B-IGFC system in
question. The heat output increases with proceeding degradation of
the SOFC stack. However, it was assumed that only the heat amount
produced by a new stack is sold.
Depreciation
Linear depreciation was generally assumed.
Standard equipment
The yearly depreciation of the standard equipment, CDep,SE,i, is
calculated according to Eq. 189. Standard equipment comprises
all equipment pieces except the fuel cell stack and any em-
ployed catalysts, the initial cost of which, C FC and Ccat , are0 0
subtracted from the DPC, see section 5.4.2. The depreciation

period, DeP, was assumed to be 10 years.
DPC − CFC 0 − Ccat 0
For i ≤ DeP : CDep , SE ,i =
DeP Eq. 189
For i > DeP : CDep , SE ,i = 0
Fuel cell stack
Fuel cell stacks degrades with growing age. Within the planning
horizon of 20 years, one or several stack exchanges are neces-
sary. Each stack is depreciated during its operation period,
DePFC, which depends on the threshold degradation value, TRD,
the degradation rate, DRFC, and the plant factor, PF. TRD rep-
resents the upper-limiting power degradation value at which a
stack is exchanged. A TRD of 16.8 % and a DRFC of 0.2 % per
1000 h were assumed in this work. The latter is the target for
2012 defined by the SECA for natural gas operated SOFCs,
[220]. The yearly depreciation of the stacks, CDep,FC,i, is deter-
mined via Eq. 190. DePFC was considered on a monthly basis to
allow stack exchange close to the point of time when the TRD is
reached. CFC gives the cost of the stack in operation. Stack
costs were considered subject to inflation. The last stack within
the planning horizon depreciates during the remaining time be-
tween its installation and the end of the planning horizon.
12
CFC
CDep, FC ,i = ∑
m =1 DePFC
Eq. 190
1000 TRD
with DePFC = ⋅
730 DRFC ⋅ PF
231
5 Modeling
Catalysts
Catalysts are subject to deactivation processes and have to be
exchanged in due course. It was assumed that both the
methanation and the CPO catalyst materials are operable for a
maximum period of 24 months in which they are depreciated,
DePCat. The yearly depreciation of catalysts, CDep,cat,i, is com-
puted according to Eq. 191. Similar to the fuel cell stack, cata-
lyst costs were considered subject to inflation. Hence, Ccat de-
notes the costs of the catalyst in operation.
12
Ccat
CDep , cat ,i = ∑ Eq. 191
m =1 DePcat
Amortization of indirect plant cost, contingencies and contractor fees

The amortization of indirect plant costs, contingencies and contractor
fees was considered to take place within a period of 5 years, AmP.
The corresponding yearly costs, CAm,i, are determined via Eq. 192.
TI − DPC
For i ≤ AmP : C Am ,i =
AmP Eq. 192
For i > AmP : C Am ,i =0
Interest expenses
The yearly interest expenses, CIE,i, are due to the debt owing on total
investment, TI, plus the additional cost resulting from fuel cell stack,
CFC, and catalyst exchanges, Ccat. As interest payments in arrear
were assumed, the first depreciation and amortization payments are
effective only from the second year of operation onwards.
⎛ i
⎞
CIE ,i = ⎜ TI + CFC + CCat − ∑ DAi −1 ⎟ ⋅ NCIR
⎝ 1 ⎠ Eq. 193
with DAi = CDep , SE ,i + CDep , FC , i + CDep, cat ,i + C Am, i
To account for the mixed structure of the invested capital, a nominal

capital interest rate, NCIR, is calculated according to WACC method
(Weighted Average Cost of Capital), Eq. 194.
NCIR = RoE ⋅ (1 − DP ) + IR ⋅ DP Eq. 194
RoE denotes the return on equity demanded by the investor. In this

work a moderate RoE of 12 % was assumed.
CCap,i = C Dep, SE , i + C Dep, FC ,i + C Dep ,Cat ,i + C Am,i + C IE , i Eq. 195
The total yearly capital costs are given by Eq. 195. The net value at SoP
of the operational and capital costs accrued over the planning horizon,
232
5 Modeling
COp,net, is determined by summing up the yearly cost less heat revenues

discounted by the nominal capital interest rate, NCIR, Eq. 196.
PH C fuel , i + CIns ,i + CO + M , i + Cutility ,i + CCap ,i − Rheat ,i
COp, net = ∑ Eq. 196
i =1 (1 + NCIR )i
To evenly distribute the net value at SoP of all costs throughout the
planning horizon, the annuity method is used. The annuity factor, AN,
follows from Eq. 197.
NCIR ⋅ (1 + NCIR )
PH
AN = Eq. 197
(1 + NCIR )PH − 1
Finally, the power production cost, PPC, is determined with the annuity
factor, AN, the net value of all costs at SoP, COp,net, the plant factor, PF,
and the average power output, Pel , Eq. 198.
AN ⋅ COp, net
PPC = Eq. 198
Pel ⋅ 8760 ⋅ PF
Pel depends on the TRD and the DRFC and is determined through a
yearly averaging of the produced electricity during the entire planning
horizon.
Table 5-27: Variable Description Unit Value

Input data for
PH Planning horizon years 20
power production
cost calculation SoP Start of production year - 2008
IF Inflation rate % 2.0
PF Plant factor % 85.0
Cwood Wood price 2007-€/dry ton 84.0
WI Yearly wood price increase % 1.0
Cwater Water price SoP-€/m3 0.94
pheat Heat price 2008-€/kWhth 0.04
Depreciation period for standard
DeP years 10
equipment
TRD Threshold degradation % 16.8
DRFC Degradation rate %/1000 h 0.2
DePcat Depreciation period for catalysts months 24
AmP Amortization period years 5
RoE Return of equity % 12.0
IR Interest rate % 8.0
DP Dept portion % 75.0
233
234
6 System analysis
6 System analysis
The main objective of this system analysis was to assess the technical
and economical feasibility of different B-IGFC systems with an electrical
power output of around 1 MWel based on already or soon available bio-
mass gasification processes, gas processing technologies and SOFC
designs for stationary power applications. Particular attention was given
to the thermal gradients and temperature extremes caused by different
producer gases during their electrochemical conversion in SOFCs. The
interactions between different biomass gasification processes, gas
processing technologies and SOFC concepts were also investigated
and the overall system efficiencies and power output as well as the re-
quirements for the balance of plant equipment resulting from the SOFC
cooling were determined. By relating the system efficiency to system
cost estimates, the power production costs were approximated and used
as basis for the economic comparison of the various B-IGFC systems.
6.1 Analysis constraints

6.1.1 SOFC simulation assumptions
6.1.1.1 Cell designs
A number of SOFC concepts are currently under development. However,
only few developers aim at stationary applications, Table 6-1.
Table 6-1: Cell de- Support Flow Operational

Developer
SOFC designs for sign design design temperature
stationary power
Delphi Corporation Planar Anode Co-current 650 to 750 °C
applications
Siemens AG Tubular Cathode Co-current 800 to 1000 °C
Co-, Counter-,
Haldor Topsoe A/S Planar Anode 700 to 850 °C
Cross-current
J-POWER Co. Ltd. Tubular Passive Co-current 900 to 1000 °C
Mitsubishi Heavy Industries Ltd. Planar Electrolyte Cross-current 900 to 1000 °C
Rolls-Royce International Ltd. Tubular Passive Cross-current 850 to 950 °C
ZTEK Corporation, Inc. Planar Electrolyte Co-current 900 to 1000 °C
To cover most of the SOFC designs summarized above, the SOFC cell
designs listed below were investigated in the system analysis.
235
6 System analysis
Cell design 1: Planar anode-supported counter-current cells with metal-

lic bipolar plates and 760 °C operational temperature.
Cell design 2: Planar anode-supported co-current cells with metallic bi-
polar plates and 760 °C operational temperature.
Cell design 3: Planar electrolyte-supported co-current cells with ce-
ramic bipolar plates and 950 °C operational temperature.
Cell design 4: Tubular cathode-supported co-current cells with air de-
livery tube, ceramic interconnects and 1000 °C operational temperature.
6.1.1.2 Operational parameters

The operational parameters of the SOFCs were defined based on the
results reported in [188].
Fuel utilization: The fuel utilization was set at 85 %.
Operational voltage: To obtain a high power output, the operational
voltage was fixed at 0.6 V. This voltage is comparably low but not en-
dangering the safe operation of SOFCs. Higher voltages yield higher
conversion efficiencies but less power output.
Current density: The current density, which together with the stack size
determines the fuel flow rate, was adjusted to satisfy the pre-defined fuel
utilization and operational voltage.
Gas temperature at cell inlet: The air and fuel gas temperature at the
cell inlet were determined for the reference system discussed in section
6.2.1 and each of the investigated SOFC designs, see above. Therefore,
an air-to-fuel ratio of 4.25 was assumed, which corresponds to an air
utilization of 20 % at 85 % fuel utilization. The gas inlet temperatures
were appraised to yield a mean cell temperature equal to the operational
temperature of the respective SOFC design. The thus determined gas
inlet temperatures were fixed for the simulation of the B-IGFC systems
presented in sections 6.2.2.2 to 6.2.2.4.
Air-to-fuel ratio: The air-to-fuel ratio was adjusted for the different B-
IGFC systems to yield a mean cell temperature equal to the operational
temperature of the respective SOFC design with the corresponding cell
inlet gas temperatures determined for the reference system. Fixing the
mean cell temperature has the advantage that the electrochemical con-
version of different fuel gases takes place on the same overall tempera-
ture level. This ensures the comparability of the performance achieved
by different B-IGFC systems because power output increasing effects
236
6 System analysis
induced by higher cell temperatures are excluded. For instance, fuel

gases containing considerable concentrations of hydrocarbons effec-
tively cool SOFCs, which leads to lower temperatures and thus in-
creased voltage losses and reduced power output. Without fixing the
mean cell temperature, the power output obtained from these gases is
systematically underestimated compared to fuel gases with lower hydro-
carbon content. Nevertheless, high temperature gradients and local
temperature extremes are possible despite a fixed mean cell tempera-
ture. This leaves the possibility of comparing different producer gases
with respect to thermal stress and sintering issues in the SOFC open.
Steam-to-carbon ratio: The power output of SOFCs usually decreases
with increasing water content of the fuel gas. Hence, the steam-to-
carbon ratio (SC) should be as low as possible without risking carbon
deposits. As the carbon deposition risk decreases with rising tempera-
ture, the SC was determined for the lowest temperature prevalent in the
different investigated B-IGFC systems, please refer to section 6.3.1.
6.1.1.3 Model parameters and settings

The parameters and model setting used for the simulation of the four in-
vestigated cell designs, see section 6.1.1.1, with the ATHENA SOFC
model are listed below. The macro cell geometry of the anode-
supported planar cell designs is given in Table 5-12. The corresponding
micro geometry is summarized in Table 6-2.
Table 6-2: Description Unit Value Source

Micro geometry
of planar anode-
supported cell Electrolyte thickness μm 10.0 [75]
design Cathode thickness μm 50.0
The heat conductivity coefficients of the ceramic materials used for the
simulation of planar cell designs can be found in Table 5-14. For the
heat conduction in the metallic bipolar plates, 18 W/m2 K was assumed.
Table 6-3 summarizes the parameters applied in the reproduction of the
electrochemical behavior of anode-supported cells. The data used for
the reproduction of planar electrolyte-supported cells is summarized in
Table 5-12, Table 5-14 and Table 5-20. Table 5-16 to Table 5-18 give
the geometry and model input data for the tubular cathode-supported
cell design. The general model settings are compiled in Table 6-4.
237
6 System analysis
Table 6-3: Description Unit Value Source

loss model input
data for simula- [174]
tion of anode- H2 oxidation pre-exponential factor A/m2 550000000.0
supported cell 100000.0
CO oxidation activation energy J/mol
designs [139]
CO oxidation pre-exponential factor A/m2 393000000.0
[174]
Ohmic polarization
Ionic conductivity of electrolyte 1/Ohm m T-dependent, Eq. 46 [191]
Electric conductivity of metallic IC 1/Ohm m T-independent [190]
Porosity of electrodes - 0.3
Tortuosity of electrodes - 6.0 assumed
Average pore diameter of electrodes m 1.0E-6
Table 6-4: Model settings Description

ATHENA SOFC
Electrochemical loss model settings
model settings
for system Considered electrochemical active species Hydrogen and Carbon monoxide
analysis Activation polarization According to Eq. 89
Ohmic polarization Contact resistance neglected
Diffusion polarization According to Ficks law, Eq. 111 to Eq. 115
Chemistry model settings
Methane steam reforming reaction According to Eq. 133
Water gas shift reaction At equilibrium
Ethane steam reforming reaction According to Eq. 134
Toluene steam reforming reaction According to Eq. 135
Thermal decomposition of acetic acid According to Eq. 137
Anisole steam reforming reaction According to Eq. 136
Film diffusion limitation for heterogeneous reactions Considered
Energy balance settings
Solid heat transfer mechanism Non-isothermal with solid heat conduction
Coupled heat and mass transport Considered
Heat loss mechanism at outer surface Neglected
d (n& ⋅ c p ⋅ T )
Definition of sensible heat gradient
dx
General model settings
Number of discretization points 40
Tolerance of mass balance calculations 0.5 %
Tolerance of energy balance calculations 1.0 %
238
6 System analysis
6.1.2 Overall system simulation assumptions

The aims of the overall system simulations were fourfold. Firstly to de-
termine the overall system efficiency, secondly to identify heat pinch
points, thirdly to determine the amount of useful heat at 200 °C which
can be recovered from the flue gas after all heat requirements were
covered and finally to estimate the size of the most important equipment
pieces, especially of the heat exchangers. It needs to be noted that an
optimization of the heat exchanger network was not part of the investi-
gation. Instead, the generalized heat integration network (HIN) dis-
cussed in section 5.2.3 was defined. The assumptions made for the
overall system simulations are summarized in Table 6-5.
Table 6-5: Parameter Unit Value

Overall system
Ambient temperature °C 25.0
simulation assump-
tions System pressure bara 1.01325
Wood humidity as delivered to plant % 42.0
Included in raw pro-
Gasifier heat losses -
ducer gas temperature
Heat losses in other equipment % 0.0
Desulfurization temperature, [149] °C 400
Post-combustor efficiency % 100.0
Useful heat temperature °C 200.0
Temperature approach of flue gas and SOFC cooling air in
K 20.0
recuperator
Pressure loss over system for blower power estimation bara 0.375
By employing measured producer gases temperatures at the outlet of

the corresponding gasification reactor, the resulting heat losses were in-
cluded in the system analysis. Heat losses of other equipment pieces
however were neglected. This assumption does not affect the electrical
system efficiency but yields slightly overestimated amounts of useful
heat. The useful heat temperature level was assumed at 200 °C to allow
the application of e.g. an organic rankine bottoming cycle. Because
pressure losses were neglected in the system simulations, an overall
system pressure loss of 375 mbar was assumed for the estimation of
the power requirement of blowers and pumps. The net AC power pro-
duced by the SOFC is calculated according to Eq. 199 based on the net
AC efficiency, see section 5.4.4, the wood mass flow and its lower heat-
ing value. The thermal and the overall system efficiency are computed
according to Eq. 200 and Eq. 201, where Pth represents the useful heat.
239
6 System analysis
Pnet , AC = η net , AC ⋅ m& feed ⋅ LHV feed Eq. 199
Pth
ηth = Eq. 200
Pnet , AC + Pth
ηtot = Eq. 201
6.2 System definition

6.2.1 Natural gas reference system
The development of standardized SOFC units operated with natural gas
is the primary goal of the American Solid State Energy Conversion Alli-
ance (SECA), [221]. The “mass customization” approach pursued by the
SECA should allow the linkage of these units to standardized and thus
cost efficient power plant size SOFC systems. Accordingly, a natural
gas fuelled SOFC stack with a size sufficient for a power output of
1 MWel was assumed as standard for the application in B-IGFC systems.
The natural gas reference system, shown in Figure 6-1, was defined to
determine for the four different cell designs the number of cells forming
the assumed standard stack and the gas inlet temperatures.
Figure 6-1: Water/ Steam

Outline of the Natural gas 25°C 90°C
natural gas refer- Desulfurizer Humidifier Reformer
ence system Sulfur
Air
Flue gas to HIN
Tin,SOFC
SOFC
Electricity
In a first step, the odorant of the natural gas, assumed as pure methane,
is removed. The gas is then humidified to yield a steam-to-carbon ratio
of 2, which was found sufficient to prevent carbon deposition in corre-
sponding equilibrium calculations, see section 6.3.1. The humidified
natural gas is introduced to an isothermal convective pre-reformer
where, depending on the reformer temperature, 25 to 30 % of the meth-
ane is converted to hydrogen and carbon monoxide, see Table 6-6. The
heat required for both, the steam generation and the endothermic re-
forming reactions, is taken from the hot flue gas. The temperature of the
pre-reformer was assumed 100 K below the operational temperature of
the various cell designs to ensure that enough heat at the required tem-
240
6 System analysis
perature level is available. After being heated to the cell design specific
temperature, the pre-reformed gas is fed to the SOFC, where its chemi-
cal energy content is partially converted to electricity. For heat recovery,
the depleted anode off gases are post-combusted and then conducted
through the heat integration network (HIN) discussed in section 5.2.3.
The electrical efficiency achieved with a fuel utilization of 85 % at a volt-
age of 0.6 V was 45 to 46 % depending on the pre-reforming degree
achieved at the cell design specific reformer temperature.
Table 6-6: Cell design

Species Unit
Gas compositions 1 and 2 3 4
of pre-reformed
H2 23.49 26.38 26.25
natural gas at the
cell inlet of the four CO 4.15 6.66 6.83
considered cell CO2 mol-% 2.76 1.60 1.44
designs
H2O 48.02 46.02 46.15
CH4 21.58 19.34 19.32
For the four investigated cell designs, Table 6-7 lists the number of cells
required to produce 1 MWel DC power and the gas inlet temperatures
needed for the accordance of the mean cell temperature and the opera-
tional temperature at an air-to-fuel ratio of 4.25. The established maxi-
mum temperature gradients and solid temperatures were adopted as
upper limit values for the safe operation of the various cell designs.
Table 6-7: Cell design

Description Unit
Stack sizes and 1 2 3 4
gas inlet tempera-
tures as well as Number of cells - 55164 51553 48057 6914
permitted maxi- Anode gas temperature 605.5 641.0 891.0 900.0
mum temperature °C
Cathode gas temperature 605.5 641.0 891.0 850.0
gradients and solid
temperatures for Mean current density A/m2 3037.1 3232.9 3468.1 2891.1
the four cell de- Max. temperature
signs determined K/mm 2.61 2.62 3.34 1.29
gradient
with the reference
natural gas system Max. solid temperature K 1094 1110 1341 1356
6.2.2 B-IGFC systems

6.2.2.1 Gasification processes
A fixed bed updraft and a fixed bed downdraft air gasification process as
well as a fluidized bed steam gasification process were investigated.
The composition of the producer gases and the main features of the
241
6 System analysis
gasification processes are compiled in Table 6-8. The functional princi-

ples of the investigated processes were outlined in the section 2.2.3.
Table 6-8: Updraft Downdraft Fluidized bed

Description Unit
Producer gas com- gasification gasification gasification
positions and prop-
General information
erties of the gasifi-
cation processes Babcock& Wil- HPR Tech-
Manufacturer Pyroforce AG
considered in the cox Vølund A/S nologies GmbH
system analysis - Own [33], own
Data sources [28], [29]
measurement measurement
Gasification agent Air Air Steam
Humidity of feed at
% 42.0 8.0 7.4
gasifier inlet
Gas temperature at
°C 75.0 650.0 756.0
gasifier outlet
Gas yield mn3 (dtf)/ kgdryfeed 2.054 2.902 1.073
LHV MJ/ mn3 5.193 4.797 7.288
Hydrogen equivalent mol H2/ mn3 22.30 18.96 31.17
Dry and tar free gas composition
H2 19.23 17.67 34.93
CO 20.53 19.41 28.08
CO2 13.52 10.32 19.18
mol-% (dtf)
CH4 5.51 1.63 9.59
C2H4 0.00 0.10 2.74
N2 41.21 50.87 5.48
Water load
H2O g/ mn3 (dtf) 493.0 43.3 626.7
Tar load
C7H8 (Toluene) 35330 1366 15430
C2H4O2 (Acetic acid) mg/ mn3 (dtf) 13608 0 98
C7H8O (Anisole) 5832 0 42
Sulfur load
C4H4S (Thiophene) 83 3 40
mg/ mn3 (dtf)
H2S 173 145 380
The water load of the producer gas originating from the updraft gasifica-
tion process was calculated to satisfy saturation at 75 °C, [28]. The gas
yield, which represents the producer gas volume on dry and tar free ba-
sis (dtf) per kg of dry wood, was computed to yield a cold gas efficiency
of 95 % including the heating value of tars, [4]. The producer gas com-
position as well as the tar load and its composition were reported in [28]
and [29].
242
6 System analysis
For the downdraft gasification process, the water load of the producer
gas was determined by means of mass balance, assuming a char flow
rate in form of carbon of 4 g/mn3 (dtf), [4]. The gas yield was calculated
to achieve a cold gas efficiency including tars of 80 %, [4]. The producer
gas composition was measured in the framework of a measuring cam-
paign conducted by fellow researchers of the author in 2003.
The water load of the fluidized bed steam gasification producer gas was
evaluated by application of mass balance calculations for a steam-to-
fuel ratio of 2.16, [33]. As no tar load and composition data could be
found for the investigated fluidized bed gasification process, values ob-
tained during a measuring campaign at a similar steam gasification plant
were employed. The gas yield was computed with a predefined cold gas
efficiency of 76 %, which corresponds to the development target of the
investigated fluidized bed gasification process, [33], increased by 10 %.
Measured data regarding the sulfur load and composition was not avail-
able for the three gasification processes. Hence, a rough estimation was
performed by hypothesizing that the sulfur present in the gasified wood
is entirely converted to either H2S or thiophene. This assumption yields
a conservative estimate due to the neglect of the amount of sulfur usu-
ally bound in particles, [222]. According to the DIN Plus norm, a sulfur
content of wood of 400 mg/kg on dry basis was assumed. The fractions
of thiophene and hydrogen sulfide were appraised via an analogy. It was
assumed that the ratio between organically bound sulfur and sulfur
bound in hydrogen sulfide equals the ratio between carbon bound in tars
and carbon bound in permanent gas species (CO, CO2, CH4, C2H4).
6.2.2.2 Downdraft gasification based B-IGFC systems

The producer gas obtained from the examined downdraft gasification
process has a temperature of 650 °C at the gasifier outlet, which forces
on the application of high temperature gas processing.
Due to the hydrocarbon content below 2 mol-%, the potential of the pro-
ducer gas for the chemical cell cooling via reforming reactions is low. In
this respect, a methanation unit in the gas processing could be favorable.
With roughly 1.4 g/mn3 (dtf), the tar load of the producer gas is very low.
Thus, the heating of the producer gas in heat exchangers should be un-
problematic regarding coking issues.
According to the assumption in section 6.2.2.1, sulfur is present in the
producer gas to over 99 % in form of hydrogen sulfide, which can be
243
6 System analysis
removed by employing zinc oxide adsorbents (ZnO). The sulfur bound in

organic species is unlikely to be an issue for stable SOFC operation.

Outline of the Wood
650°C 650°C
downdraft gasifi- Downdraft Particle filter Humidifier
Air
cation based Char/ Ash
"System 1"
Air Flue gas to HIN
400°C Tin,SOFC
Desulfurizer SOFC
Sulfur Electricity
Figure 6-2 shows the outline of the first downdraft gasification based
B-IGFC system configured for this analysis, named "System 1". In Sys-
tem 1, the humidity of the wood is decreased from 42 to 8 % in a con-
vective exhaust gas dryer. The dried wood is injected to the downdraft
gasifier where it is gasified with air. The gasification air is forced into the
gasifier by an electric air blower. The hot producer gas is de-dusted in a
ceramic particle filter operated at 650 °C and then humidified with wet
steam. This results in a cool down of the producer gas to 400 °C, which
corresponds to the operating temperature of the zinc oxide adsorbent
bed employed for the removal of hydrogen sulfide. Besides yielding the
desired gas cool down, the humidification is necessary to prevent car-
bon deposition at the temperature of the zinc oxide adsorbent bed tem-
perature. The de-dusted and desulphurized gas is heated to the cell de-
sign specific gas inlet temperature, see Table 6-7, and subsequently fed
to the SOFC, where the chemical energy of the producer gas is partially
converted to electricity and heat. The cooling air for the SOFC is pro-
vided by an electric air blower. The depleted electrode off-gases are
then guided through the heat integration network, explained in section
5.2.3, to cover the heat requirements resulting from the air and fuel
heating, steam generation and wood drying. Note that in System 1, or-
ganic sulfur species are not removed from the producer gas.
Figure 6-3 depicts the second downdraft gasification based B-IGFC sys-
tem configured for this analysis, named "System 2". In addition to Sys-
tem 1, an adiabatic methanation unit is placed downstream the ZnO
desulfurizer in System 2. The aim of the methanation step is to increase
the methane content of the producer gas, which should allow for more
SOFC internal reforming reactions and thus lower the required cooling
air flow rate.
244
6 System analysis

Outline of the Wood 650°C 650°C
downdraft gasifi- Downdraft Particle filter Humidifier
cation based Air Char/ Ash
"System 2"
Air
Tin,SOFC Flue gas to HIN
400°C 400°C
Desulfurizer Methanation SOFC
Electricity
Sulfur
6.2.2.3 Updraft gasification based B-IGFC systems

The investigated updraft gasifier yields a producer gas at 75 °C tem-
perature. Hence, both cold and hot gas processing may be applied.
With a hydrocarbon content over 5 mol-%, the potential for chemical cell
cooling through reforming reactions is indeed three times higher than
that of the downdraft gasification producer gas but still lower than that of
the pre-reformed natural gas assumed as reference fuel for SOFCs.
With almost 55 g/mn3 (dtf), the updraft gasification producer gas is highly
tar laden. Coking issues during heating and tar condensation problems
in pipes are therefore most probable. To prevent these issues, the tar
load should be reduced to moderate values as near as possible to the
gasifier outlet either by tar removal or by tar conversion to permanent
gas species.
Following the assumption in section 6.2.2.1, the sulfur signature of the
updraft producer gas strongly differs from that of the downdraft producer
gas. With almost 17 % of the sulfur present in the gasified wood being
bound in organic sulfur species, measures to degrade these species to
hydrogen sulfide for subsequent adsorption are inevitable to allow for
the effective operation of SOFCs.
Figure 6-4: Wood

Outline of the 75°C 45°C
Updraft Wet scrubber E-Precipitator
updraft gasifica- Air Char/ Ash Tars/ Char/ Ash
tion based
"System 3" Water/ Steam Air
45°C 300°C
Humidifier Desulfurizer SOFC
Sulfur Electricity
The scheme of the first updraft gasification based B-IGFC system de-
fined for this analysis is named "System 3" and depicted in Figure 6-4.
System 3 uses cold gas processing technology, thus representing
state-of-the-art. The wood with a humidity of 42 % is directly introduced
245
6 System analysis
to the updraft gasifier, where it is dried and converted to producer gas

using air as gasification agent. The gasification air is again forced
through the gasifier using an electric air blower. The comparably cold
and water saturated producer gas is coarsely de-dusted in a wet scrub-
ber of which the gas exits water saturated at a temperature of 45 °C.
The remaining particles together with the tars and organic sulfur species
are removed from the producer gas in an electrostatic precipitator. To
prevent carbon deposition in the zinc oxide adsorbent bed and the sub-
sequent gas heating to the cell inlet temperature, vapor is injected in a
humidifier. Hydrogen sulfide is adsorbed in a ZnO trap bed at a tem-
perature of 300 °C yielding concentrations around 0.1 ppmV. The de-
dusted and almost tar and sulfur free producer gas is super heated to
the cell inlet temperature and then fed to the SOFC. The heat recovery
is conducted in the same manner as in System 1 and 2.

updraft gasifica- Updraft Multiclone Humidifier
tion based Air Char/ Ash
"System 4"
470°C 85°C
Particle filter Methanation
Catalyst/ Ash Air
470°C 400°C
Cooler Desulfurizer SOFC
Sulfur Electricity
The second investigated updraft gasification based B-IGFC system is

named "System 4" and presented in Figure 6-5. Similar to System 3, the
wet wood is directly put into the updraft gasifier without external drying.
The producer gas is de-dusted in a multiclone and then humidified. In an
adiabatic fluidized bed methanation reactor, tars and organic sulfur spe-
cies are degraded to permanent gas species and hydrogen sulfide, re-
spectively. The methanation of tar laden producer gases using the fluid-
ized bed reactor concept was shown in [223]. It was found to be fairly
tolerant towards carbon depositions usually caused by higher hydrocar-
bons due to reactor internal carbon exchange processes, [224]. The
methane content of the producer gas is considerably increased, which
should be beneficial with respect to the cooling requirements of the em-
ployed SOFC. Simultaneously, the heat of the methanation reactions
yields an increase of the producer gas temperature from approx. 85 °C
at the reactor inlet to roughly 470 °C at the outlet. Ash and catalyst par-
246
6 System analysis
ticles are removed from the hot producer gas in a ceramic particle filter.
After a slight cool down, the hydrogen sulfide content is removed from
the producer gas in a ZnO trap bed at 400 °C. The subsequent heating
of the gas, conversion in the SOFC and heat recovery is carried out as
already mentioned above for the other investigated systems.
Figure 6-6 shows the third updraft gasification based B-IGFC system,
named "System 5". The producer gas originating from the air gasifica-
tion of the un-dried wood is again de-dusted in a multiclone.
Figure 6-6: Air

updraft gasifica- Updraft Multiclone CPO
Air
tion based Char/ Ash
"System 5" 500°C
Particle filter
Char/ Ash Air Flue gas to HIN
500°C 400°C Tin,SOFC
Sulfur Electricity
Tars and organic sulfur species are degraded in a catalytic partial oxida-
tion reactor (CPO) through combustion and reforming reactions. Thus, a
part of the chemical energy of the producer gas is converted to heat
which results in a considerable increase of the producer gas tempera-
ture. To minimize the loss of chemical energy, the air-to-fuel ratio is kept
as low as possible without risking carbon deposition in the equipment
downstream of the CPO. The catalytic partial oxidation of tar-laden pro-
ducer gases was shown and reported in [153]. The CPO air flow rate is
provided by an electric blower. Ash particles are removed from the par-
tially oxidized producer gas in a ceramic particle filter before it is cooled
to 400 °C for subsequent hydrogen sulfide removal in a ZnO trap bed.
The conversion of the gas to electricity and useful heat through the
SOFC and heat integration network is conducted as discussed above.
6.2.2.4 Fluidized bed steam gasification based B-IGFC systems

The producer gas originating from the investigated fluidized bed steam
gasification process has a temperature of 750 °C. Thus, high tempera-
ture gas processing is strongly recommended.
The potential of the steam gasification producer gas for chemical cell
cooling is higher than that of both other gases. Therefore, measures to
further increase the methane content were not investigated.
247
6 System analysis
Despite the elevated tar load around 16 g/mn3 (dtf), the gas handling
should be rather unproblematic. At the high temperature of the steam
gasification producer gas, all tar species are in the gas phase. In con-
trast, coking problems usually arise during the evaporation of con-
densed tars, [133]. Hence regarding coking issues, it should suffice to
ensure the gas phase state of the tars by restraining a minimum system
temperature of 400 °C, which is well above the dew point of most tars.
According to the assumption in section 6.2.2.1, the sulfur load consists
of hydrogen sulfide to more than 96 %. Hence, it is possible that a re-
moval of organic sulfur species is not necessary.

fluidized bed gasi- Fluidized bed Particle filter Humidifier
fication based Steam+Heat
Char/ Ash/ Bed material
"System 6"
Air
450°C 400°C
Sulfur Electricity
The first investigated fluidized bed steam gasification based B-IGFC

system is named "System 6" and depicted in Figure 6-7. For efficiency
reasons, the wood humidity is decreased from 42 to 7.4 % before the
wood is fed to the steam gasification reactor. The steam is produced us-
ing flue gas heat. The heat for the gasification reactions is generated via
combustion of unreacted char and transported into the gasification reac-
tor using heat pipes. For more information please see [168]. The quasi
nitrogen free producer gas has a temperature of 756 °C at which it is
de-dusted using a ceramic particle filter. To allow for the desulfurization
of the producer gas with ZnO, it is cooled and humidified by liquid water
injection thus also preventing carbon deposition. The temperature after
the water injection is approx. 450 °C thereby requiring further cool down
in a gas cooler. The heat is used for the pre-heating of the SOFC cool-
ing air. At 400 °C, H2S is removed from the gas, while organic sulfur
species remain. The de-dusted and partially desulphurized producer gas
is than heated to the cell inlet temperature and converted to heat and
electricity in the SOFC. The depleted off-gases are guided through the
heat integration network to cover all heat requirements of the system.
Figure 6-8 shows the second fluidized bed steam gasification based B-
IGFC system investigated in this work, which is named "System 7".
248
6 System analysis
Figure 6-8: Wood 756°C 756°C

Outline of the Fluidized bed Particle filter MeO
fluidized bed gasi- Steam+Heat Char/ Ash/ Bed material
fication based Water/ Steam
740°C
"System 7" Humidifier
Air Flue gas to HIN
440°C 400°C Tin,SOFC
Sulfur Electricity
The only difference between System 7 and System 6 is that a sulfur ad-
sorbent bed containing a metal oxide is placed downstream the particle
filter for coarse desulfurization and cracking of organic sulfur species.
According to [149], the metal oxide material has the potential to degrade
organic sulfur species. Accounting for the found influence of the metal
oxide on the producer gas composition, it was assumed that 10 % of the
producer gas reach equilibrium. The sulfur load in the raw producer gas
can thus be almost completely removed in the subsequent zinc oxide
trap bed, which should have a positive effect on the SOFC performance.
6.3 Technical evaluation

6.3.1 Simulation results of gas processing sections
Carbon deposition is an issue which is not limited to the SOFC but
which also has to be considered as potential failure source of all equip-
ment pieces in direct contact with producer gas at temperatures above
300 °C. Below 300 °C, carbon deposition can be assumed slow as long
as no catalyst is present. Note that carbon deposition results from the
decomposition of hydrocarbons or carbon monoxide, see section 2.4.7.3,
and should not be confused with tar condensation. Generally, the prob-
ability of carbon deposition decreases with increasing temperature and
atomic oxygen molar fraction of the producer gas. The latter may be in-
creased by adding water or oxygen. In turn, water removal and gas cool-
ing may cause carbon deposition issues.
The producer gases characterized in Table 6-8 exhibit water loads,
which are high enough to prevent carbon deposition at the operational
temperatures of the considered SOFC designs. However, this is not the
case at temperatures around 300 to 400 °C, which form the operational
temperature range of the ZnO adsorbents employed for H2S removal.
Except for System 5, where air is injected to the producer gas for its
subsequent catalytic partial oxidation, the atomic oxygen fraction of the
249
6 System analysis
producer gases was increased simply through addition of water or

steam in this analysis.
Table 6-9: Para- System

Unit
Steam-to-carbon meter NG 1 2 3 4 5 6 7
(SC) ratios of raw
producer gases, Min. T °C 600 400 400 300 400 400 400 400
minimum and safe SCraw 0.0 0.17 0.17 0.26 1.29 1.29 1.20 1.20
SC values SCmin - 1.5 1.25 1.25 1.48 1.4 0.09* 1.33 1.33
*: Air-to-fuel ratio of
CPO SCsafe 2.0 1.66 1.66 1.97 1.86 0.12* 1.78 1.78
Table 6-9 summarizes the steam-to-carbon ratios (SC) of the raw pro-
ducer gases, the minimum SC at which graphite does not exist at equi-
librium conditions and the SC used in this analysis, which is 33 % higher
than the minimum SC. Note that Table 6-9 gives the minimum and the
safe air-to-fuel ratio for the CPO performed in System 5.
Figure 6-9 shows the C-H-O ternary diagram with isotherm lines for
760 °C, 600 °C and 400 °C, which are the lowest operational tempera-
ture of the four investigated cell designs, the critical temperature for car-
bon deposition from methane and the operational temperature of the
ZnO trap beds in Systems 4 and 5, respectively.
Figure 6-9:
C-H-O molar frac-
tion ternary diagram
for graphite forma-
Raw gases
NG C
0.0
Equlibrium
temperatures
tion under equilib- S4
1.0 T=760°C
rium conditions. S5 T=600°C
T=400°C
The isotherm lines com- Processed gases
puted for different tem- NG 0.2
0.8
peratures divide the S4
diagram in a region S5
where graphite is formed Graphite
at equilibrium and a 0.4
graphite free region. The formation 0.6
inscribed points repre-
sent the raw fuel gases region CO
as received from the 0.6
gasifier or the NG grid 0.4
and the processed fuel
gases. The points are
CO2
only moved when addi-
CH4
tional carbon, hydrogen 0.8
0.2
or oxygen is added to the
gases. Thus, only the
humidification and the air
injection for the CPO 1.0
result in a shift of the 0.0
points representing the 0.0 0.2 0.4 0.6 0.8 1.0
different fuel gases. H H2O O
250
6 System analysis
The raw updraft producer gas used in System 4 and 5 is slightly above
the isotherm line of 400 °C, indicating that carbon deposition is likely to
occur in the ZnO trap bed. The water addition in System 4 shifts the rep-
resenting point out of the graphite formation region following the dashed
line. The air addition for the CPO in System 5 shifts the point represent-
ing the raw updraft producer gas towards the oxygen corner of the dia-
gram following the dotted line. Methane is clearly in the graphite forma-
tion region for all the displayed temperatures. In contrast to the updraft
producer gas, the carbon deposition risk for methane is not highest at
400 °C but at 600 °C. This is because methane is stable at tempera-
tures below 300 °C. The water addition to yield a SC of 2 follows the
dashed-dotted line.
Figure 6-10:
C-H-O molar frac-
tion ternary diagram
for graphite forma-
Raw gases
S1/S2 C
0.0
Equilibrium
temperatures
tion under equilib- S3 1.0 T=760°C
rium conditions. S6/S7 T=400°C
T=300°C
The isotherm lines com- Processed gases
S1/S2 0.2
puted for different tem- 0.8
peratures divide the S3
diagram in a region S6/S7
where graphite is formed
at equilibrium and a 0.4 Graphite 0.6
graphite free region. The
inscribed points repre- formation
sent the raw fuel gases
CO
as received from the
region
0.6
gasifier and the proc- 0.4
essed fuel gases. The CO2
points are only moved CH4
when additional carbon,
hydrogen or oxygen is 0.8
0.2
added to the gases.
Thus, only the humidifi-
cation results in a shift of
the points representing 1.0
the different fuel gases. 0.0
0.0 0.2 0.4 0.6 0.8 1.0
H H2O O
Figure 6-10 shows the C-H-O ternary diagram with isotherm lines for
760 °C, 400 °C and 300 °C, where the latter is the operational tempera-
ture of the ZnO trap bed used in System 3. The lower ZnO temperature
calls for more water addition in System 3 than in System 4. Together
with the removal of almost 75 % of the water content occurring in the
wet scrubber, the amount of additional water needed in System 3 is
three times higher than in System 4, see dashed line in Figure 6-10.
251
6 System analysis
The water load of the raw downdraft producer gas (S1/S2) is low, due to
the limited wood humidity that can be sustained by the downdraft gasifi-
cation process. For the desulfurization at 400 °C, the amount of addi-
tional water is comparable to that required in System 3, see dashed-
dotted line in Figure 6-10.
The raw producer gas originating from the investigated fluidized bed
steam gasification process (S6/S7) exhibits a comparably high water
content despite the low humidity of the gasified wood. This is due to
steam being used as gasification agent. Nevertheless, this producer gas
does not allow for desulfurization at 400 °C without water addition, see
dotted line in Figure 6-10.
The producer gas compositions at the cell inlet, including the above dis-
cussed additional water, obtained from the seven B-IGFC systems de-
fined in section 6.2.2, are compiled in Table 6-10. The models used to
simulate the methanation, catalytic partial oxidation and ZnO trap bed
desulfurization units were explained in the sections 5.2.4 to 5.2.6.
Table 6-10: System

Species Unit
Producer gas com- 1 2 3 4 5 6 7
positions a cell inlet
H2 11.56 19.26 10.81 7.87 17.15 16.17 17.93
CO 12.70 3.00 11.55 0.22 3.53 13.00 12.62
CO2 6.75 16.19 7.60 16.98 17.28 8.88 9.59
H2O mol-% 34.54 26.04 43.80 42.48 25.94 53.54 51.79
CH4 1.07 1.85 3.10 9.37 4.14 4.44 4.26
C2H4 0.07 0.00 0.00 0.00 0.00 1.27 1.13
N2 33.29 33.67 23.16 23.08 31.88 2.54 2.52
C7H8 217.5 0.0 0.0 0.0 459.7 1700.0 1600.0
C2H4O2 0.0 0.0 0.0 0.0 271.7 0.0 0.0
C7H8O ppmV 0.0 0.0 0.0 0.0 64.7 0.0 0.0
C4H4S 0.5 0.0 0.0 0.0 1.2 4.9 0.0
H2S 1.8 2.4 0.0 2.2 1.4 2.8 2.7
In summary, the use of ZnO trap beds for desulfurization requires an in-
crease of the water content of the producer gas to avoid carbon deposi-
tion in all investigated systems except System 5. One way to overcome
this requirement could be the application of adsorbent materials with
operational temperatures above 700 °C. However, presently such mate-
rials are not available. Regarding the system design, in Systems 1, 2
and 3 the extra water calls for additional equipment such as steam gen-
252
6 System analysis
erators and spray chambers as a matter of principle. In System 4, the

water content could be increased just by increasing the humidity of the
gasified wood to values higher than the standard 42 %. In System 6 and
7, an increase of the steam-to-fuel ratio to values higher than the inves-
tigated 2.16 would render supplementary gas humidification equipment
superfluous, however at the expense of the cold gas efficiency of the
steam gasification process.
6.3.2 Simulation results of SOFC operation with producer gases

The fuel gas composition plays a crucial role in the mass, heat and
charge transport processes taking place in SOFCs, section 5.3.5. The
impact of the producer gases compiled in Table 6-10 is discussed below,
focusing on not only the conversion efficiencies and expectable current
densities but also the maximum solid temperatures and temperature
gradients. The latter are critical for degradation processes, section
2.4.7.2, and structural stress induced cell failures, section 2.4.7.4.
6.3.2.1 Conversion efficiencies

Due to the predefined operational voltage and fuel utilization, the con-
version efficiency is the same for all investigated cell designs and only
depends on the fuel gas composition, Eq. 202, [70].
Eop ΔG
η SOFC , DC = UF ⋅ with E0 = − Eq. 202
E0 2⋅ F
Figure 6-11 shows the accordingly computed conversion efficiencies,

where System 4 yields the highest efficiency of all investigated gases.
Figure 6-11: 50
Conversion effi-
ciencies of the
different producer 40
gases and the
Conversion efficiency [%]
reference gas at
85 % fuel utiliza- 30
tion and 0.6 V
operational volt-
age 20
10
0
Reference System 1 System 2 System 3 System 4 System 5 System 6 System 7
253
6 System analysis
6.3.2.2 Current densities

Figure 6-12 compares the predicted average current density obtained
from the different cell designs and producer gases with 85 % fuel utiliza-
tion an operational voltage of 0.6 V.
Figure 6-12: 3500

Comparison of the Anode counter-current
predicted average Anode co-current
3000
current density ob- Electrolyte
tained from the Cathode
investigated cell 2500
2
designs and pro-

ducer gases 2000
1500
1000
500
0
None of the investigated producer gases yields current densities as high

as those obtained with the reference gas. The main reason for this is
that the reference gas is not diluted with nitrogen, such as the producer
gases in systems 1 to 5, which reduces activation losses, and is drier
than the producer gases, which yields higher Nernst voltages. The
trends regarding the current density values achieved with different pro-
ducer gases are quite similar in all four investigated cell designs.
Figure 6-13: 1.0000 5000

Voltage loss and
current density
distributions of the
electrolyte- sup-
ported cell pre-

2
dicted for the

Voltage [V]
reference gas 0.1000 2500
Hydrogen activation
CO activation
Oxygen activation
Ohmic
Current density
0.0100 0
0.00 0.02 0.04 0.06 0.08 0.10
Cell length [m]
254
6 System analysis
The electrolyte-supported cell produces the highest current density with

the reference gas. In this case, the dominating voltage losses are anode
activation and ohmic losses. The latter dominate at the cell inlet and de-
crease towards the cell end, Figure 6-13. With producer gas, the power
output of the electrolyte-supported cell is decreased by 50 % in average.
Figure 6-14: 1.0000 5000

Voltage loss and Hydrogen activation
current density CO activation
distributions of the Oxygen activation
electrolyte- sup- Ohmic

ported cell pre-
2
Current density
dicted for the

Voltage [V]
producer gas of 0.1000 2500

System 2
0.0100 0
0.00 0.02 0.04 0.06 0.08 0.10
Cell length [m]
Figure 6-14 shows the voltage loss and current density distributions of
the electrolyte-supported cell operated with the System 2 producer gas.
In contrast to the reference gas, the anode activation losses dominate
the ohmic losses along the entire cell length. Compared to the reference
gas, the activation losses are increased by 50 to 100 %, indicating that
the model parameters used for the electrolyte-supported cell describe
an anode catalyst with low activity in dilute gases.
Figure 6-15: 1.0000 5000

Voltage loss and Hydrogen activation
CO activation
current density Oxygen activation
Oxygen diffusion
distributions of the Ohmic
cathode- sup- Current density
2
ported cell pre-

dicted for the
Voltage [V]
0.0100 0
0.00 0.25 0.50 0.75 1.00 1.25 1.50
Cell length [m]
255
6 System analysis
The power output of the tubular cell is limited by ohmic losses, which are
at least one order of magnitude higher than the anode activation losses
for all the investigated gases, Figure 6-15. This explains the moderate
power output reduction of the tubular cell by 15 % in average when op-
erated with producer gases and indicates a very active anode catalyst.
Figure 6-16: 1.0000 5000

Hydrogen activation Hydrogen diffusion CO activation
Voltage loss and CO diffusion Oxygen activation Oxygen diffusion
current density Ohmic Current density
anode-supported

co-current cell
2
Voltage loss [V]
predicted for the

0.0100 0
0.00 0.02 0.04 0.06 0.08 0.10
Cell length [m]
Figure 6-16 shows the voltage loss and current density distributions of
the co-current anode-supported cell operated with the reference gas.
The power output is limited by the oxygen activation and to a lesser ex-
tent by the anode activation. The oxygen activation losses are approx.
twice as high as the anode activation losses. The anode diffusion losses
gain significance towards the cell end, which can be attributed to fuel
depletion and the long diffusion path through the thick anode.
Figure 6-17: 1.0000 5000

Hydrogen activation Hydrogen diffusion CO activation
Voltage loss and CO diffusion Oxygen activation Oxygen diffusion
current density Ohmic Current density
anode-supported
2
co-current cell
Voltage loss [V]
predicted for the

producer gas of 0.1000 2500
System 2
0.0100 0
0.00 0.02 0.04 0.06 0.08 0.10
Cell length [m]
256
6 System analysis
On average, the anode-supported cells yield 30 % less power output

with producer gas. Figure 6-17 depicts the voltage loss and current den-
sity distributions predicted for the co-current anode-supported cell oper-
ated with System 2 producer gas. Compared to the reference gas, the
anode activation losses increase by 50 to 100 % and reach values close
to the oxygen activation losses at the cell inlet and even higher values at
the cell outlet. Moreover, a considerable increase of the anode diffusion
losses can be observed, which is attributed to the high nitrogen content
of the System 2 producer gas. Given the almost 200 K lower operational
temperature of the anode-supported cells compared to the electrolyte-
supported cell, the anode catalyst described by the used model parame-
ters appears quite active in dilute gases. This advantage is however
cancelled by the lower operational temperature.
The power output differences between the anode-supported co- and
counter-current cells are comparatively small. Nevertheless an interest-
ing trend can be observed in Figure 6-12. With the reference gas, the
co-current cell yields an almost 10 % higher current density than the
counter-current cell. With the producer gases however, the co-current
cell always yields slightly lower current densities than the counter-
current cell. The reason for this can be found in temperature and current
density distributions resulting for the different flow patterns.
Figure 6-18: 6000 1200

Comparison of Current density Co-flow Current density Counter-flow
predicted solid Solid temperature Co-flow Solid temperature Counter-flow
5000 1150
temperature und
current density
distribution of
4000 1100
2
anode-supported
co- and counter-
3000 1050
current cells for
the reference gas
2000 1000
1000 950
0 900
0.00 0.02 0.04 0.06 0.08 0.10
Cell length [m]
Figure 6-18 compares the temperature and current density distributions

of the anode-supported co- and counter-current cells predicted for the
reference gas. The co-current cell yields a current density distribution
257
6 System analysis
with a wide and flat peak in the middle part of the cell. This is the result
of the low temperatures at the cell inlet, due to endothermic STR reac-
tions, which yield only low fuel conversion in this cell region leaving over
most of the fuel for the conversion in the latter parts of the cell.
Throughout the entire co-current cell, the anode and oxygen activation
are the dominant voltage losses.
In contrast, the counter-current cell features a plateau of very high cur-
rent densities due to high temperatures at the fuel inlet, despite endo-
thermic STR reactions taking place. The high temperatures are main-
tained by the cooling air, which is heated in the latter parts of the cell.
The fuel conversion at the cell inlet is mainly limited by the oxygen acti-
vation and diffusion. Towards the fuel outlet, the current density is in-
creasingly limited by anode activation losses, which result from the fast
fuel depletion in the front parts of the cell. In sum, the counter-current
cell yields a lower average current density with the reference gas than
the co-current cell due to the diffusion limitations at the cell inlet.
Figure 6-19: 6000 1200

Comparison of Current density Co-Flow Current density Counter-Flow
predicted solid Solid temperature Co-Flow Solid temperature Counter-Flow
5000 1150
temperature und
current density
distribution of
Sold temperature [K]

4000 1100
2
anode-supported
co- and counter-
3000 1050
current cells for
the producer gas
of System 2 2000 1000
1000 950
0 900
0.00 0.02 0.04 0.06 0.08 0.10
Cell length [m]
Figure 6-19 compares the temperature and current density distributions

of the anode-supported cells predicted for the System 2 producer gas.
The co-current cell yields a relatively even current density distribution
with a slightly decreasing trend towards the cell end. The fuel conver-
sion at the fuel inlet is fast due to the lack of STR reactions resulting in
approx. 40 K higher solid temperatures compared to the reference gas.
However, this is not beneficial to the total produced current as most of
the fuel is converted at a comparably low solid temperature of 1020 K,
258
6 System analysis
where the activation losses are relatively high. For the reference gas,
the maximum current density is reached in the second cell half at 1060
K. This is due to the low temperatures at the cell inlet resulting from the
STR inhibiting a fast fuel conversion.
The counter-current cell yields considerably higher current densities at
the fuel inlet than the co-current cell due to 50 K higher solid tempera-
tures. In contrast to the reference gas, with producer gas the counter-
current cell is not limited by oxygen diffusion but by the anode activation
like the co-current cell. Thus, the fast fuel conversion at the cell inlet
(and high temperatures) is beneficial to the current production of the
counter-current cell. In consequence, the total amount of current pro-
duced by the counter-current cell is slightly higher than that produced by
the co-current cell when operated with producer gases.
In summary it can be said that for a given operational voltage the anode
catalyst activity is the dominating factor with respect to the mean current
density obtained from different cell designs operated with producer
gases. Due to their high degree of nitrogen or water dilution, producer
gases require highly active anode catalysts to yield satisfying mean cur-
rent densities. It appears that an initial estimation of the catalyst activity
can be based on the cell design. Designs limited by ohmic losses, such
as the tubular and the planar electrolyte-supported cells, will probably
feature more active anode catalysts to compensate the ohmic losses.
The reason why this does not apply to the electrolyte-supported cell in-
vestigated in this analysis is because it was primarily developed as a
catalytic burner to provide heat to households. Thus, the anode catalyst
is most likely not tuned for high activity with dilute gases. Regarding dif-
fusion limitations resulting from the dilution quality of producer gases it
can be said that only the anode-supported cells showed a slight and
hence negligible response. Finally, it was found that cell designs with
counter-current flow pattern allow for high temperatures at the fuel inlet
and are thus beneficial to the conversion of producer gases.
6.3.2.3 Cooling requirements

The air-to-fuel ratio (AF) determines the power requirement of the cool-
ing air blower and the size of the heat exchangers required for air pre-
heating, hence making it significant in the overall system design. Figure
6-20 shows that the AF strongly varies for different cell designs and pro-
ducer gases. Regarding the latter, the general trend is that fuel gases
with high conversion efficiencies require lower AF due to the reduced
259
6 System analysis
amounts of waste heat. The differences between the AF required by the

different cell designs arise from the geometry dependent heat exchange
areas and cell internal heat transport mechanisms.
Figure 6-20: 15
predicted air-to- Anode co-current
fuel ratios re- 12 Electrolyte
quired to achieve Cathode
a mean cell tem-
Air-to-fuel ratio [-]
perature equal to 9
the operational
temperature of the
investigated cell 6
designs operated
with the different
producer gases 3
0
The electrolyte-supported cell requires the highest AF of all cell designs.

A strong increase of the AF can be observed with decreasing internal re-
forming potential of the different producer gases, which is a function of
the fraction of heating value provided by hydrocarbons.
Figure 6-21: 80 1400

Specie molar Hydrogen Carbon monoxide
fractions and solid Carbon dioxide Water
Methane Solid temperature
temperature dis-
tribution predicted 60 1250
for the electrolyte-
Mol fraction [%]
supported cell
operated with the 40 1100
reference gas and
1164 K cell inlet
temperature
20 950
0 800
0.00 0.02 0.04 0.06 0.08 0.10
Cell length [m]
The reason for this can be found in the cell inlet temperatures. Figure
6-21 shows the species molar fractions and the solid temperature distri-
butions predicted for the electrolyte-supported cell operated with the ref-
260
6 System analysis
erence gas. The methane STR is fast at the temperatures prevalent in

the electrolyte-supported cell. To prevent excessive cooling of the cell
inlet region and to reach a mean cell temperature of 1223 K for the AF
of the reference case, the cell inlet temperature of the fuel gas and the
cooling air was set to comparably high 1164 K, which is only 59 K below
the operational temperature. When operated with producer gases with
lower internal reforming potential than the reference gas, the amount of
heat that has to be removed from the cell increases. Consequently, to
preserve a temperature gradient between the solid structure and the
cooling air throughout the entire cell that is sufficient for the removal of
the increased heat amount, the air flow rate and thus the AF have to be
increased. Alternatively, the gas inlet temperatures can be reduced,
yielding a reduction of the required AF. In this analysis the gas inlet
temperatures were fixed.
The cathode-supported tubular cell features an air delivery tube in which
the cooling air is heated before it flows through the cathode gas channel.
In the reference case, the cooling due to the endothermic STR reactions
at the fuel inlet results in considerably increased temperatures in the lat-
ter parts of the cell. The cooling air in the air delivery tube is thus effec-
tively pre-heated and transports heat from the hot cell end region to the
STR region, Figure 6-22.
Figure 6-22: 1400

Solid structure Anode gas
and gas tempera- Solid structure

ture distributions Cathode gas
predicted for the 1300
Gas temperature in
cathode- air delivery tube
Temperatures [K]
supported tubular
cell with air deliv-
1200
ery tube operated
with the reference
gas
1100
1000
0.00 0.25 0.50 0.75 1.00 1.25 1.50
Cell length [m]
Once in the cathode gas channel, the temperature of the cooling air is
lowered due to the STR reactions. This results in a considerable in-
261
6 System analysis
crease of the temperature gradient between the solid structure and the
cooling air and accordingly allows a low AF.
When operated with producer gas with low internal reforming potential,
the cell inlet region is not sub-cooled yielding lower temperatures in the
latter parts of the cell. Figure 6-23 shows the temperature distribution
predicted for the tubular cathode-supported cell operated with the pro-
ducer gas of System 1, which has the lowest internal reforming potential
of all investigated gases. Compared to the reference gas, the solid tem-
perature in the later parts of the cell is approx. 50 K lower, while at the
fuel inlet it is almost 200 K higher. The cooling air is gradually heated
along the air delivery tube and the cathode channel and does not act as
heat source for the fuel inlet region. Hence, the AF has to be increased
to preserve a sufficiently high temperature gradient between the solid
structure and the cooling air for the removal of the cell waste heat.
Figure 6-23: 1400

Solid structure
and gas tempera-
ture distributions
cathode-
supported tubular
Temperatures [K]
cell with air deliv-

ery tube operated
1200
with the producer
gas of System 1
Anode gas
1100 Solid structure
Cathode gas
Gas temperature in
air delivery tube
1000
0.00 0.25 0.50 0.75 1.00 1.25 1.50
Cell length [m]
The anode-supported co-current cell is less sensitive than the electro-

lyte-supported cell towards the internal reforming potential of the fuel
gas despite the similar geometry and flow pattern. This can be attributed
to the cell inlet gas temperatures determined for the reference gas,
which are 119 K lower than the mean cell temperature compared to the
59 K determined for the electrolyte-supported cell. The lower cell inlet
temperatures are possible due to considerably slower STR reactions
and the enhanced heat conduction through the metallic bipolar plates
compared to the ceramic ones used in the electrolyte-supported cell.
262
6 System analysis
Figure 6-24 shows that in the anode-supported co-current cell the STR
reactions are distributed over a length roughly three times longer than in
the electrolyte-supported cell, Figure 6-21. Thus, most of the heat re-
quired for the endothermic STR reactions is provided by electrochemical
reactions and heat conduction through the metallic bipolar plates and
needs not to be provided in form of sensible heat of the inflowing fuel
gas and cooling air as is the case for the electrolyte-supported cell.
Figure 6-24: 80 1400

Specie molar Hydrogen Carbon monoxide
fractions and solid Carbon dioxide Water

Methane Solid temperature
temperature dis-
tribution predicted 60 1250
for the anode-

Mol fraction [%]
supported co-
current cell oper- 40 1100
ated with the ref-
erence gas and
914 K cell inlet
20 950
temperature
0 800
0.00 0.02 0.04 0.06 0.08 0.10
Cell length [m]
When operated with producer gas with lower internal reforming potential,
the anode-supported co-current cell requires lower AF than the electro-
lyte-supported cell due to more effective cooling resulting from the lower
cell inlet gas temperatures.
Among all investigated cell designs, the anode-supported counter-
current cell shows the lowest sensitivity of the AF towards the internal
reforming potential of the fuel gas. For gases with high internal reform-
ing potential, the heat amount produced in electrochemical reactions at
the cell inlet is high enough to cover the heat demand of the STR reac-
tions taking place. Further, the cooling air transports heat from the latter
parts of the cell to the cell inlet region where it is also consumed by the
STR reactions. When operated with fuel gas featuring low internal re-
forming potential, the heat amount produced in the latter cell parts is re-
duced due to fuel depletion resulting from the fuel conversion at the cell
inlet. The heat is then primarily transported from the fuel inlet region to
the latter cell parts through solid heat conduction. Thus, the counter-
current cell yields relatively constant AF for different fuel gases.
263
6 System analysis
Overall it was found that the AF increases with decreasing internal re-
forming potential of the fuel gas. The sensitivity of the AF towards the in-
ternal reforming potential depends on the cell design. Co-current cells
were found to be highly sensitive due to the constraint of the cell inlet
gas temperatures to the values determined for the reference case. In
contrast, the AF required by counter-current cells was established to be
almost independent of the internal reforming potential of the fuel gas.
This robustness makes this cell design appealing for the operation with
producer gases, whose temperatures and compositions often fluctuate.
6.3.2.4 Maximum solid temperatures and temperature gradients

Sintering of the nickel particles that are dissolved in the YSZ matrix of
Ni-YSZ anodes is one of the most important long-term degradation
mechanisms of SOFCs, [96], see section 2.4.7.2. Sintering takes place
at faster rates with rising temperature. This attributes special importance
to the maximum solid temperature caused by different fuel gases.
Figure 6-25: 1423

Comparison of the Anode counter-current Anode co-current Electrolyte Cathode
maximum solid 1373
Cathode supported
temperatures cell mean temperature
Maximum solid temperature [K]

investigated cell 1273
designs operated
with the reference 1223
gas and different Electrolyte supported
1173
producer gases Anode supported cell cell mean temperature
mean temperature
1123
1073
1023
973
Figure 6-25 compares the maximum solid temperatures predicted for the
investigated cell designs and producer gases. It can be seen that fuel
gases with high internal reforming potential tend to yield high maximum
solid temperatures, see "Reference" and "System 4". This can be attrib-
uted to the low solid temperatures resulting from endothermic STR reac-
tions at the fuel inlet, which are compensated by increased solid tem-
peratures in the latter cell parts to satisfy the accordance of the mean
cell temperature with the operational temperature of the respective cell
design.
264
6 System analysis
Figure 6-26: 150

Comparison of the
Temperature difference to design temperature [K]

Reference System 1
solid structure 100
System 2 System 3
System 4 System 5
temperature dis-
System 6 System 7 Over heating zone
tributions pre-
50
dicted for the
electrolyte-
supported cell 0
operated with the
reference gas and -50
the different pro- Sub cooling zone
ducer gases
-100
-150
0.0 0.2 0.4 0.6 0.8 1.0
Dimensionless cell length [-]
Figure 6-26 gives the comparison of the solid structure temperature dis-
tributions predicted for the electrolyte-supported cell when operated
with the reference gas and the different producer gases. Despite the
producer gas of System 4 having the highest internal reforming potential
of all investigated fuel gases, it does not yield the highest maximum
solid temperature. The reason for this is that the fuel inlet region is less
cooled compared to the reference case. For the System 4 producer gas,
the anode activation losses are dominant. Consequently, the maximum
current density is produced close to the fuel inlet where the reactant par-
tial pressures are comparably high. The released heat of the electro-
chemical reactions is sufficient to hinder a strong cooling of the fuel inlet
area. With the reference gas, the maximum current density is reached in
the second half of the cell as a consequence of the dominating ohmic
losses which only allow for high current densities at elevated tempera-
tures. The cooling of the fuel inlet area is more pronounced because,
compared to the heat required by the STR reactions, less heat is pro-
duced through electrochemical reactions. The more pronounced cooling
results in higher maximum solid temperatures at the cell outlet.
For the cathode-supported cell, Figure 6-27 shows that the maximum
solid temperatures are not reached at the fuel outlet but at around three
quarters of the cell length. This is due to the cooling effect stemming
from the fresh air flowing through the air delivery tube. The producer gas
of System 4 does not yield lower solid temperatures at the fuel inlet than
the reference gas despite its higher internal reforming potential. Never-
theless, the maximum solid temperature attained with the producer gas
of System 4 is higher than that obtained with the reference gas. The
reason for this is discussed based on Figure 6-28.
265
6 System analysis
Figure 6-27: 300

Comparison of the Reference System 1

solid structure System 2 System 3
200
temperature dis- System 4 System 5
tributions pre- System 6 System 7 Over heating zone
dicted for the 100
cathode-
supported cell 0
operated with the
reference gas and
the different pro- -100
ducer gases Sub cooling zone

-200
-300
0.0 0.2 0.4 0.6 0.8 1.0
The STR reactions taking place in the fuel inlet region with the reference
gas require considerably more heat than is provided by electrochemical
reactions and the pre-heated cooling air. This results in reduced solid
structure temperatures, which are again the reason for the ohmic limita-
tions hindering more electrochemical reactions. The producer gas of
System 4 yields current densities at the fuel inlet comparable to the ref-
erence gas, which in turn allow higher solid structure temperatures at
the cell inlet. Additionally, compared to the reference gas more heat is
provided to the fuel inlet by the pre-heated cooling air.
Figure 6-28: 1400 5000

Solid structure
temperature, gas
temperature and
current density 1300 3750
distributions pre-
2
Temperature [K]
dicted for the

cathode- 1200 2500
supported tubular
cell operated with
Cathode gas
the producer gas Gas temperature in air delivery tube
of System 4 and 1100 Solid structure
1250
the reference gas Solid structure - Reference gas
Current density
Current density - Reference gas
1000 0
0.00 0.25 0.50 0.75 1.00 1.25 1.50
Cell length [m]
The STR reactions are complete after roughly one tenth of the cell
length for both gases. For the reference gas, a very fast temperature in-
266
6 System analysis
crease occurs, caused by equally fast increasing current densities.

However for the producer gas of System 4, the current densities and
solid temperatures increase only slowly due to anode activation limita-
tions resulting from the dilution of the producer gas. Compared to the
reference gas, this shifts the maximum current density towards the cell
end, and yields higher temperatures in the latter parts of the cell. These
induce more effective cooling air pre-heating in the ADT, thus also in-
creasing the amount of heat supplied to the fuel inlet region.
Figure 6-29: 120

Comparison of the
Reference System 1
solid structure System 2 System 3
80
temperature dis- System 4 System 5
tributions pre- System 6 System 7 Over heating zone
dicted for the 40
anode-supported
co-current cell 0
operated with the
reference gas and
-40
the different pro- Sub cooling zone
ducer gases
-80
-120
0.0 0.2 0.4 0.6 0.8 1.0
As with the electrolyte-supported cell, the maximum solid temperatures

are achieved at the fuel outlet for the anode-supported co-current cell,
Figure 6-29. The low operational temperature slows down the STR reac-
tions considerably, which thus take place along almost half of the cell
length, Figure 6-24. Consequently, the fuel inlet region is noticeably less
cooled, which explains the maximum solid temperatures of 80 K over
the mean cell temperature compared to 120 K found for the electrolyte-
supported cell. In addition to the slower STR reactions, the temperatures
in the fuel inlet region are kept high due to enhanced heat conduction
through the metallic bipolar plates. Similar to the electrolyte-supported
cell, the producer gas of System 4 does not yield the highest maximum
solid temperature. The mechanism is the same as already discussed for
the electrolyte-supported cell.
The anode-supported counter-current cell reaches the maximum
solid temperature in the fuel inlet area, Figure 6-30. For decreasing in-
ternal reforming potential of the fuel gas, the maximum solid tempera-
ture moves towards the fuel inlet and vice versa. The magnitude of the
267
6 System analysis
maximum solid temperature depends on the ratio of the heat required by

STR and the heat supplied by electrochemical reactions, heat conduc-
tion through the metallic bipolar plates and heat convection from the pre-
heated cooling air which flows from the fuel outlet to the fuel inlet region.
Figure 6-30: 120

Comparison of the Temperature difference to design temperature [K]
solid structure
80
temperature dis-
tributions pre- Over heating zone
dicted for the 40
anode-supported
counter-current 0
cell operated with
the reference gas
-40
and the different Sub cooling zone
Reference System 1
producer gases
System 2 System 3
-80
System 4 System 5
System 6 System 7
-120
0.0 0.2 0.4 0.6 0.8 1.0
All investigated gases except the producer gas of System 4 yield maxi-
mum temperatures in the narrow range between 1084 and 1098 K. The
undiluted gases of the Systems 6 and 7 and the reference gas achieve
the highest maximum solid temperatures. This is because the current
densities are considerably higher in the fuel inlet region compared to the
other gases due to less activation polarization losses. Compared to the
reference case, the current densities reached with the diluted producer
gas of System 4 are considerably lower due to increased activation
losses. The heat amount released in the electrochemical reactions is
thus not sufficient to cover the heat demand of the STR reactions taking
place close to the fuel inlet. Consequently, the predicted solid tempera-
tures close to the fuel inlet for the System 4 producer gas are lower than
those obtained with all other gases. To satisfy the accordance of the
mean cell temperature and the design temperature, the lower solid tem-
peratures at the fuel inlet are compensated with higher temperatures
towards the fuel outlet region.
The role of tars in the cooling of the cell inlet area was investigated by
comparing the solid temperature distribution predicted for the electro-
lyte-supported cell operated with the tar free producer gas of System 3
with that predicted for the raw updraft producer gas, Table 6-8. Figure
6-31 depicts that the added tar load does contribute to the cooling of the
268
6 System analysis
cell inlet area. With a 6 K lower solid temperature at the cell inlet, the
impact of the tars is however small. Tar removal especially to prevent
excessive maximum solid temperatures or thermal gradients is therefore
not necessary.
Figure 6-31: 1300

Comparison of the With tars Without tars
solid structure
1275
temperature dis-
tributions pre-
dicted for the
1250
electrolyte-
supported cell
1225
operated with the
producer gas of
System 3 and the 1200
producer gas as
obtained from the 1175
updraft gasifier
1150
0.00 0.02 0.04 0.06 0.08 0.10
Cell length [m]
Besides the maximum solid temperatures, the maximum temperature

gradients are important with respect to the safe operation of SOFCs.
Figure 6-32 shows the maximum temperature gradients predicted for all
investigated fuel gases and cell designs.
Figure 6-32: 4
maximum tem- Anode co-current
Maximum temperature gradient [K/mm]
peratures gradi- Electrolyte

ents predicted for 3 Cathode
the four cell de-
signs operated
with the different
investigated fuel 2
gases
0
The reference gas yields the highest temperature gradients for all cell
designs and gases. Similar to the maximum solid temperatures, the
269
6 System analysis
temperature gradients increase with increasing internal reforming poten-

tial for the co-current cell designs. The counter-current cell design yields
quite similar temperature gradients for all investigated fuel gases. The
reason for this is that the maximum temperature gradient is reached in
the latter cell parts, where the cell is cooled by the inflowing cooling air.
Thus it is almost independent of the fuel gas composition, which mainly
determines the temperature level of the fuel inlet area.
In summary it can be stated that the conversion of the investigated pro-
ducer gases in SOFCs should cause fewer temperature-induced degra-
dation and failure issues than the conversion of pre-reformed natural
gas. The maximum solid temperatures established for the different pro-
ducer gases are considerably lower than those predicted for the refer-
ence gas. The only exception is the maximum solid temperature of the
cathode-supported cell operated with the producer gas of System 4,
which is 18 K higher than that of the reference case. Hence, it is possi-
ble that the cathode-supported cell degrades slightly faster when oper-
ated with System 4 producer gas compared to the reference gas. Re-
garding the maximum temperature gradients, all investigated gases
were found less problematic than the reference gas. For co-current cells,
the thermal stress usually increases with increasing internal reforming
potential. Exceptions from this mostly originate from gas dilution or high
water loads, which are both decisive for the importance of the anode ac-
tivation losses compared to the other voltage losses and may lead to
considerable changes in the current density distribution. This again af-
fects the temperature distribution and thus the thermal stress. In con-
trast, counter-current cells are, as a matter of principle, comparably in-
sensitive regarding the impact of different fuel gases on thermal stress.
6.3.3 System simulation results

As discussed above, considerable differences between the expectable
mean current densities and air-to-fuel ratios were found for the investi-
gated cell designs and producer gases. Due to the top-down approach
chosen for this analysis, the mean current density of the SOFC deter-
mines the overall system size. The air-to-fuel ratio defines the size of the
SOFC balance of plant equipment and the power requirements for the
SOFC cooling. The latter is important for the overall net system effi-
ciency. System simulations were performed to investigate the interac-
tions between the different gasification processes, gas processing sec-
tions and SOFC designs.
270
6 System analysis
6.3.3.1 Wood input

The stack sizes were determined for the different cell designs operated
with the reference gas to yield 1 MWel DC output at 85 % fuel utilization
and an operational voltage of 0.6 V, see Table 6-7. The basic idea be-
hind this top-down approach is to base the analysis on a standardized
SOFC stack of 1 MWel DC power output with methane as reference fuel.
The producer gas flow rates at the cell inlet were calculated through
Faraday's law, Eq. 203.
I tot ⋅ Aact
n&CG ,in =
2 F ⋅ UF ⋅ EqH 2 ,in
Eq. 203
with EqH 2 ,in = yH 2 ,in + yCO ,in + 4 ⋅ yCH 4 ,in + 6 ⋅ yC 2 H 4 ,in
+ 18 ⋅ yC 7 H 8 ,in + 4 ⋅ yC 2 H 4O2 ,in + 17 ⋅ yC 7 H 8O ,in
In Eq. 203, Itot stands for the mean current densities obtained for the
various producer gases and cell designs, see Figure 5-1, Aact repre-
sents the active area of the corresponding stacks, F is the Faraday con-
stant, UF defines the fuel utilization and EqH2,in is the hydrogen equiva-
lent coefficient of the respective producer gas.
The mass flow rates of wet wood were computed for all the investigated
systems and cell designs based on the raw gas yields, see Table 6-8, to
deliver the required producer gas flow rates at the cell inlet, Figure 6-33.
Figure 6-33: 1000

Mass flow of wet Anode counter-current
wood for the dif- 900 Anode co-current
ferent B-IGFC 800
Electrolyte
systems and cell Cathode
Wet wood mass flow [kg/h]
700
designs
600
500
400
300
200
100
0
System 1 System 2 System 3 System 4 System 5 System 6 System 7
6.3.3.2 Heat integration network sizing

The system heat flux depends on the producer gas flow rate, the air-to-
fuel ratio of the used SOFC and its respective cell inlet temperatures.
271
6 System analysis
For all investigated systems and cell designs, no heat pinch points were
found, indicating that the sensible heat of the post-combusted SOFC off-
gases is sufficient to cover all heat requirements.
Based on the predefined heat exchange coefficients, see section 5.2.3,
the required heat exchange areas were computed. Figure 6-34 shows
that the heat exchange areas of the systems employing anode-
supported cells are generally smaller than those of systems using elec-
trolyte- or cathode-supported cells. This can be traced down to the con-
siderably lower cell inlet temperatures of the anode-supported cells. De-
spite lower cell inlet temperatures, the electrolyte-supported cells require
larger heat exchange areas than the cathode-supported cells. This is the
consequence of the lower air-to-fuel ratios predicted for the cathode-
supported cells due to the cell internal air pre-heating through the air de-
livery tube. Regarding the different flow patterns of the anode-supported
cells, Figure 6-34 shows that the lower cell inlet temperatures and pre-
dicted air-to-fuel ratios of the counter-current cells result in considerably
smaller heat exchange areas compared to co-current cells.
Figure 6-34: 1500

Heat exchanger Fuel superheater
areas for the dif- Air superheater
Air recuperator
ferent B-IGFC 1200
systems and cell
Heat exchanger area [m ]
2
designs
900
600
300
0
Anode cc
Anode co
Electrolyte
Cathode
Anode cc
Anode co
Electrolyte
Cathode
Anode cc
Anode co
Electrolyte
Cathode
Anode cc
Anode co
Electrolyte
Cathode
Anode cc
Anode co
Electrolyte
Cathode
Anode cc
Anode co
Electrolyte
Cathode
Anode cc
Anode co
Electrolyte
Cathode
1 2 3 4 5 6 7
In conclusion it can be said that the heat exchange mechanisms in the

SOFC and the cell inlet temperatures, which mainly depend on the op-
erational temperature of the SOFC, are crucial for the required heat ex-
change areas. Especially the air pre-heating involves comparably large
heat exchangers.
272
6 System analysis
6.3.3.3 Efficiency losses

Figure 6-35 shows the efficiency loss fractions and the net AC electrical
efficiencies of the investigated B-IGFC systems and cell designs based
on their LHV input in form of wood.
Figure 6-35:
Efficiency loss Input LHV
Electricity AC
fractions and net
100%
10%
20%
30%
40%
50%
60%
70%
80%
90%
AC electrical effi-
0%
ciencies for the
different B-IGFC
systems and cell
Anode cc
designs Anode co
1
DC/AC losses
Electrolyte
Cathode
Anode cc
Anode co
Auxiliary power
Electrolyte
Cathode
Anode cc
Anode co
Sulfur losses
Electrolyte
Cathode
Anode cc
Anode co
SOFC losses
Electrolyte
Cathode
Anode cc
Fuel utilization
Anode co
5
Electrolyte
Cathode
Anode cc
Gas processing
Anode co
6
Electrolyte
Cathode
Anode cc
Gasification
Anode co
7
Electrolyte
Cathode
273
6 System analysis
The losses attributed to the gasification processes, the fuel utilization,

the electrochemical conversion and those originating from the sulfur
content of the different producer gases vary for the seven investigated
systems but are independent of the cell designs. The cell designs how-
ever differ in the auxiliary power requirements, which have to be cov-
ered by the power produced by the SOFC stacks. Thus, the net AC effi-
ciencies vary for the different cell designs. Despite the fuel utilization be-
ing fixed at 85 %, the corresponding loss fractions vary for the different
investigated systems. This is because the basis for the loss fractions
was defined as the total thermal input to the systems, while the 85 %
fuel utilization apply to the thermal input to the SOFCs. Thus, with in-
creasing efficiency of the gasification and the gas processing, the fuel
utilization losses increase in relative terms.
Downdraft gasification based B-IGFC systems: System 1 and 2
Both systems have a gross DC power efficiency of approx. 30 %. 20 %
of the total thermal input is converted to heat in the downdraft gasifier.
The adiabatic methanation employed in System 2 converts approx. 5 %
of the chemical energy of the producer gas into heat. This heat is how-
ever consumed again in the SOFC for the STR of the produced meth-
ane, allowing the reduction of the air-to-fuel ratio of the SOFC and low-
ering the auxiliary power requirements of System 2. For the co-current
cell designs, the cooling effect resulting from the adiabatic methanation
results in approx. 2 % lower auxiliary power requirements, thus increas-
ing the net AC efficiency. For the counter-current anode-supported cell,
which is comparably insensitive towards the producer gas composition
regarding the air-to-fuel ratio, the adiabatic methanation yields only a
negligible net AC efficiency increase.
The sulfur induced losses amount to approx. 2 % for both systems due
to similar total sulfur concentrations in both producer gases. The adia-
batic methanation of System 2 in fact converts the organic sulfur species
to hydrogen sulfide but also increases the water content of the producer
gas. Consequently, the hydrogen sulfide concentration leaving the zinc
oxide trap bed in System 2 is almost equal to the total sulfur concentra-
tion in the producer gas of System 1, where the organic sulfur species
were not removed.
The auxiliary power losses amount to a minimum of 5 % for the anode-
supported counter-current cell and to a maximum of almost 12 % for the
electrolyte-supported cell, the cathode-supported cell being in between.
274
6 System analysis
The losses attributed to the conversion of the direct current obtained

from the SOFCs to alternating current suitable for feed-in to the electri-
cal grid account for 1.5 to 2 % of the thermal input, depending on the net
DC electricity efficiency of the system.
The anode-supported counter-current cell achieves the highest net AC
efficiencies with 22.8 % and 23.1 % for System 1 and 2, respectively.
Updraft gasification based B-IGFC systems: System 3, 4 and 5
In all three systems, the gasifier converts 95 % of the thermal input into
the chemical energy of the producer gas and the remainder into low-
temperature heat. By entirely removing the tar load in System 3, approx.
20 % of the thermal input is lost yielding an effective cold gas efficiency
of approx. 75 %.
In System 4, 10 % of the LHV of the producer gas are converted to heat
in the adiabatic methanation which is, similar to System 2, consumed
again in the SOFC for the STR of the produced methane. Compared to
System 2, the amount of methane produced in the adiabatic methana-
tion is considerably higher due to the low temperature of the raw pro-
ducer gas at the methanation reactor inlet. This results in a heavy reduc-
tion of the required air-to-fuel ratios due to intensive chemical cooling of
the SOFCs. Further, instead of removing the tar load, their chemical en-
ergy is partially used for the fuel gas heat up in the adiabatic methana-
tion and partially converted the chemical energy bound in methane. The
steam reforming of the produced methane in the SOFC can thus be
seen as relocation of the final gasification reactions into the SOFC, di-
rectly using the SOFC waste heat. Such methanation-steam reforming
based processes to transport heat, referred to as "Chemical heat pipes",
were intensively studied in the nineteen eighties, [225].
The catalytic partial oxidation (CPO) employed in System 5 converts
approx. 45 % of the LHV of the tar load to high-temperature heat. This
heat is used for the conversion of the remaining tar load to hydrogen,
carbon monoxide but also methane due to comparably low process
temperatures of around 500 °C. Especially for the co-current cell de-
signs, the latter has a positive effect on the required air-to-fuel ratios.
For System 3, it was assumed that organic sulfur species are entirely
removed from the producer gas in the employed cold gas cleaning sys-
tem. The remaining H2S is adsorbed in the ZnO trap bed to levels below
0.1 ppmV. Consequently, the sulfur-induced losses of System 3 are
negligible. In contrast, the sulfur induced losses amount to approx. 2 %
275
6 System analysis
for Systems 4 and 5. Similar to System 2, the water produced during the
methanation in System 4 seriously obstructs the attainment of sulfur
concentrations below 2 ppmV. In System 5, total sulfur concentrations
around 2 ppmV are reached despite remaining parts of the organic sul-
fur species in the producer gas. The reason for this is the lower water
content of the producer gas after the CPO which allows achieving H2S
concentrations in the zinc oxide trap bed that are low enough to com-
pensate the organic sulfur species.
The auxiliary power losses resulting from the power consumption of fans
and blowers in System 4 are around 4 to 5 % and therewith the lowest in
this analysis. This is because the corresponding producer gas has the
highest internal reforming potential of all investigated gases and there-
fore requires the lowest air-to-fuel ratios. Due to the comparably high net
DC efficiencies, the DC/AC conversion losses are in the range of 2 to
3 % and thus are slightly higher than in the Systems 1, 2, 6 and 7.
Similar to systems 1 and 2, the anode-supported counter-current cell
yields the highest net AC efficiencies in the Systems 3 and 5 with
24.3 % and 26.5 %, respectively. Regarding System 4, the cathode-
supported cell designs benefits from the cooling air cool down at the fuel
inlet which allows a lower air-to-fuel ratio than that of the anode-
supported counter-current cell. The correspondingly lower auxiliary
power losses lead to a net AC efficiency of 32.1 % compared to 31.7 %
reached by the anode-supported counter-current cell. Overall it can be
stated that System 4 outperforms all other investigated systems be-
cause the combination of updraft gasification and adiabatic methanation
yields a producer gas with high internal reforming potential with a com-
parably high cold gas efficiency of approx. 85 %.
Fluidized bed steam gasification based systems: System 6 and 7
In the fluidized bed steam gasification process, 76 % of the chemical
energy of the feedstock is converted to chemical energy of the producer
gas and the remainder into high-temperature heat.
In System 7, the tar load and the hydrocarbon content are slightly de-
creased over the metal oxide bed used for the degradation of organic
sulfur species. The loss of internal reforming potential is however com-
pensated by a noticeable amount of carbon monoxide that is shifted to
hydrogen and carbon dioxide. If this shift occurs cell internally, the re-
leased heat of reaction adds to the SOFC waste heat thus lowering the
conversion efficiency.
276
6 System analysis
The sulfur induced losses of System 6 are 4.5 % compared with 1.9 %
of System 7. This shows that the neglect of the removal of almost
5 ppmV organic sulfur species in the system design may result in con-
siderable overestimation of the system efficiency. Compared to the Sys-
tems 1 to 5, the excessive water content of the fluidized bed producer
gases hinders the desulfurization in the zinc oxide trap beds more and
yields approx. 0.5 ppmV higher sulfur concentrations in the producer
gas at the cell inlet. A possibility to overcome this would be a reduction
in the amount of added water and thus a decrease of the steam-to-
carbon ratio. This however might cause carbon deposition problems.
The auxiliary power losses for the different cell types are almost similar
for both systems in the range between 4.5 to almost 9 %. The DC/AC
losses amount to roughly 2 %.
System 7 with the anode-supported counter-current cell reaches a net
AC efficiency of 23.7 %, which is slightly higher than the efficiencies
reached by the downdraft gasification based systems and close to the
efficiency of System 3, where 20 % of the thermal input was removed
from the producer gas in form of tars.
In sum it can be stated that the cold gas efficiency of the gasification
process and the auxiliary power requirements of the SOFC resulting
from the SOFC cooling are crucial for the overall system efficiency. In
both cases, the downdraft gasification process is not favorable. The air-
blown updraft gasification in contrast was found very promising for the
application in B-IGFC systems due to the high cold gas efficiency and
typically high hydrocarbon concentration in the producer gas. The latter
allow for the relocation of the final gasification reactions into the SOFC
using waste heat and thus lowering the cooling requirements and attrib-
uted auxiliary power requirements. The fluidized bed steam gasification
yields producer gas with high hydrocarbon content too, however with
lower cold gas efficiencies. The corresponding losses can barely be
compensated by the subsequent producer gas conversion to electricity.
6.3.3.4 Power and useful heat output

The amount of heat that can be recovered from the system flue gases
was determined through the generalized heat integration network dis-
cussed in section 5.2.3. Figure 6-36 shows that it strongly varies be-
tween the various cell designs. The electrolyte-supported cell generally
yields the lowest amount of useful heat. This is because the sensible
277
6 System analysis
heat below 200 °C of the system flue gases is not used. The heat loss
hence grows with increasing flue gas flow rate, which directly correlates
with the air-to-fuel ratio. The air-to-fuel ratio thus not only considerably
affects the net AC efficiency through the auxiliary power requirements
but also the thermal efficiency.
System 3 yields the smallest amount of useful heat because it was as-
sumed that the entire heating value of the tar load removed in the cold
gas cleaning system is required for the treatment of the wash water.
Figure 6-36: 100%

Net AC electrical Waste heat
efficiencies and 90% Useful heat
thermal efficien- Electricity AC
80%
cies for the differ-
ent B-IGFC sys- 70%
tems and cell 60%
Input LHV
designs
50%
40%
30%
20%
10%
0%
Cathode
Cathode
Cathode
Cathode
Cathode
Cathode
Cathode
Anode cc
Electrolyte
Anode cc
Electrolyte
Anode cc
Electrolyte
Anode cc
Electrolyte
Anode cc
Electrolyte
Anode cc
Electrolyte
Anode cc
Electrolyte
Anode co
Anode co
Anode co
Anode co
Anode co
Anode co
Anode co
1 2 3 4 5 6 7
With 75.7 %, System 4 with the cathode-supported cell yields the high-
est total system efficiency in this analysis. The corresponding thermal
efficiency is 43.6 %. The useful heat may be converted to additional
electricity through a bottoming cycle, thus covering the auxiliary power
requirements of the cooling air blower. Assuming a moderate electrical
efficiency of 12 % for an organic rankine cycle operated at 200 °C, [163],
the net AC system efficiency of System 4 could be increased to 37.3 %,
while also increasing the system power output by 16 %. This is sufficient
to be superior to state-of-the-art biomass gasification based gas engine
power plants featuring electrical efficiencies around 25 %, [226, 227].
Figure 6-37 depicts the net AC electrical power output of the investi-
gated B-IGFC systems and cell designs. The electrolyte-supported cell
features the lowest power outputs in all investigated systems. This re-
sults from the obtained low current densities when operated with pro-
ducer gas and the comparably small active area, which is a conse-
278
6 System analysis
quence of the reference gas based top-down approach chosen for this
analysis. The auxiliary power requirements are high due to the repeat-
edly discussed high air-to-fuel ratios (AF). The cathode-supported cell in
contrast yields the highest power outputs, due to the high mean current
densities achieved with the producer gases and the moderate AF.
Figure 6-37: 750

Net AC electrical Anode counter-current
power output of Anode co-current
the different B- 625 Electrolyte
Cathode
IGFC systems
and cell design 500
Net AC output [kWel]
375
250
125
0
Compared to the power output of the NG reference system of 820 kW

net AC, the operation with producer gases results in a power output de-
crease of a minimum 25 % for System 7 with the cathode-supported cell
and a maximum 70 % for System 1 with the electrolyte-supported cell.
6.4 Economic evaluation

The system efficiencies are related to the total investments costs (TI) to
asses the cost-effectiveness of the investigated B-IGFC systems, see
section 5.4. The latter is represented by the power production costs. The
TI linearly depend on the direct plant costs, which are estimated through
rough sizing of all major equipment pieces according to the mass and
energy fluxes determined in the system simulations. Note that the costs
were estimated using the same cost functions for all systems, hence
relative comparisons are considered valid. The absolute cost values
however should be understood as order of magnitude estimates.
6.4.1 Direct plant cost estimates

Figure 6-38 shows the direct plant cost estimates structured according
to the system modules discussed in section 5.4.2 for the different
279
6 System analysis
B-IGFC systems and cell designs. The wood handling module is a

considerable cost factor in downdraft and fluidized bed steam gasifica-
tion based systems. This can largely be attributed to the wood drying,
the costs of which strongly increase with decreasing air-to-fuel ratios
and flue gas temperatures.
Figure 6-38:
Direct plant cost Direct plant costs [2007-k€]
estimates struc-
1000
1500
2000
2500
3000
3500
500
tured according to
0
the system mod-
ules discussed in Anode cc
section 5.4.2 for
the different B- Anode co
Wood handling
1
IGFC systems Electrolyte

and cell designs Cathode
Anode cc
Anode co
2
Electrolyte
Gasifier
Cathode
Anode cc
Anode co
3
Electrolyte
Gas processing
Cathode
Anode cc
Anode co
4
Electrolyte
Cathode
FC Balance of Plant
Anode cc
Anode co
5
Electrolyte
Cathode
Anode cc
Anode co
Fuel cell
6
Electrolyte
Cathode
Anode cc
Anode co
Auxiliary
7
Electrolyte
Cathode
280
6 System analysis
Systems with anode-supported counter-current cells yield the highest

costs for wood drying, responsible for around one third of the direct plant
cost. Due to their high air-to-fuel ratios, electrolyte-supported cells have
clear cost advantages with respect to wood drying. Reducing the costs
for the wood drying is not possible in downdraft gasification based sys-
tems due to the stringent feed humidity requirements of the gasification
process. In contrast, the fluidized bed steam gasification can operate on
feedstock with higher humidity than that assumed in this analysis. Thus,
the cost for wood drying in System 6 and 7 could be reduced. The up-
draft gasification does not require wood drying and has thus consider-
able cost advantages regarding the wood handling module.
The gasification module comprises the gasification reactor and all ad-
ditional equipment required for its operation. In fixed bed gasification
based systems, the costs for the gasification reactors are rather small,
contributing around 10 % to the total system costs. Downdraft gasifiers
are slightly more expensive than updraft gasifiers due to their more
complex layout. It is difficult to further reduce the gasifier cost for the
fixed bed gasification based systems. In fluidized bed steam gasification
based systems, the gasification reactor costs amount to approx. one
fourth of the total system cost. As discussed in section 5.4.2, the inves-
tigated reactor concept uses heat pipes for the heat transfer between a
combustion chamber and a gasification chamber. The heat pipes and
the combustion chamber each account for approx. 25 % of the reactor
costs, the remainder is attributed to the reforming chamber and auxiliary
equipment required for e.g. steam generation, combustion air pre-
heating, etc. The costs incurred by the combustion chamber could be
saved by extracting the required heat directly from the SOFCs, thereby
also lowering the required air-to-fuel ratios and the cost attributed to the
fuel cell balance of plant equipment, [228]. The cost reduction potential
is considerable, however this option was not further investigated in this
work as the required SOFCs are presently not available and the costs of
these customized SOFCs are hence unknown.
The gas processing module includes all equipment pieces required for
the gas cleaning and conditioning, constituting around 5 to 15 % of the
total plant costs. In downdraft gasification based systems, the steam
generator required for the gas humidification is the most important cost
factor regarding the gas processing, accounting for approximately 50 %
of this cost module. In System 3, where most of the water initially pre-
sent in the producer gas originating from the updraft gasification process
281
6 System analysis
is removed in the employed cold gas cleaning system, the steam gen-
erator required for the re-humidification of the producer gas accounts for
almost 80 % of the gas processing module costs. This is because no ce-
ramic particle filter unit and zinc oxide trap bed are employed. The fluid-
ized bed systems have the advantage that the producer gas can be hu-
midified without the need for a steam generator by using the sensible
heat of the producer gas for the evaporation of the additional water.
Consequently, the gas processing module of these systems is compara-
tively inexpensive. The costs of the gas processing could be effectively
reduced in fluidized bed and updraft systems by increasing the wood
humidity at the gasifier inlet. This would not only reduce the cost in-
curred by the gas humidification but also the cost of wood drying as
mentioned above.
The fuel cell balance of plant module (FC-BoP) includes the heat ex-
changers required to adjust the producer gas and the cooling air tem-
perature to the predefined cell inlet values, the cooling air blower, the
DC/AC power converter and the anode off-gas combustor. Except for
the downdraft and fluidized bed steam gasification based systems em-
ploying anode-supported counter-current cells, the FC-BoP module is
the dominant cost factor in all the investigated systems, accounting for
25 to 50 % of the total system costs. In the systems employing electro-
lyte- or cathode-supported cells, the FC-BoP costs are generally higher
than in the corresponding systems based on anode-supported cells.
This is due to the considerably lower cell inlet temperatures and air-to-
fuel ratios required by the anode-supported cells.
The fuel cell module (FC) accounts for one eights to a quarter of the to-
tal system cost. The fuel cell costs are equal for all the systems due to
the top-down approach chosen for this analysis, which considers the FC
size as fixed. Note that the costs entailed by the catalysts employed in
systems 2, 4 and 5 were allocated to the FC module. The absolute
module cost in these systems can thus be increased by up to 25 %
compared to systems 1, 3, 6 and 7.
The auxiliary equipment module accounts for 7 to 12 % of the total
system costs. Half of these costs can be attributed to the heat ex-
changer required to recover useful heat from the system flue gases.
The direct plant costs are related to the net AC power output for a better
comparability of the systems. Figure 6-39 shows the specific direct plant
costs computed for the investigated B-IGFC systems and cell designs.
282
6 System analysis
Figure 6-39: 12,000

Specific direct Anode counter-current
plant costs for the Anode co-current

10,000 Electrolyte
different B-IGFC
Specific plant cost [2007-€/kWel net AC]

Cathode
systems and cell
designs 8,000
6,000
4,000
2,000
0
The downdraft gasification based systems yield comparably high spe-

cific costs in the range between 5587 and 10612 €/kWel net AC resulting
from their low power output and high system costs. The updraft gasifica-
tion based systems achieve the lowest specific plant costs, which are
between 2999 and 7156 €/kWel net AC. This can be attributed to their
satisfying power output and low system costs, which arise not only from
the low feed humidity requirements of the updraft gasification process
and the simple make of the corresponding reactors but also from the low
producer gas humidification requirements. The fluidized bed systems
are expensive but the undiluted producer gases allow relatively high
power outputs, which in sum leads to specific plant cost between those
of the downdraft gasification based systems and of the updraft gasifica-
tion based systems in the range of 4679 to 8445 €/kWel net AC. With
2999 €/kWel net AC, System 4 with the cathode-supported cell yields the
lowest specific plant costs of all the investigated systems.
6.4.2 Power production costs

The cost-effectiveness of the investigated B-IGFC systems is deter-
mined by relating the total investments for the investigated systems to
their net AC system efficiencies and power outputs. As discussed in
section 5.4.5, the time dependent worth of money is also taken into ac-
count by assuming a set planning horizon for which all expenses and
revenues are entered in a corresponding balance sheet.
Figure 6-40 shows the power production cost (PPC) fractions arising
from the depreciation of the different system modules, the feedstock
purchasing, the amortization of the indirect plant costs and the operation
283
6 System analysis
and maintenance expenses. Further, the revenues from heat sales per
sold kWhel net AC are depicted assuming 100 % heat sales. It can be
seen that for each of the systems and cell designs the sum of the PPCs
resulting from the equipment depreciation (D-costs) equals the capital
costs plus the costs for the operation and maintenance (C-costs).
Figure 6-40:
Power production PP cost and heat revenues [2007-€/kWhel net AC]
Heat revenues
costs structured
-0.15
-0.05
0.05
0.15
0.25
0.35
0.45
0.55
according to the
system modules
discussed in sec-
tion 5.4.2 includ- Anode cc
ing fuel, capital Anode co
Fuel
and operational Electrolyte

costs as well as
heat revenues for Cathode
Wood handling
the different B-
IGFC systems Anode cc
and cell designs
Anode co
2
Electrolyte
Cathode
Gasifier
Anode cc
Anode co
Gas Processing
Electrolyte
Cathode
Anode cc
FC Balance of Plant
Anode co
4
Electrolyte
Cathode
Anode cc
Anode co
Fuel cell
Electrolyte
Cathode
Auxiliary
Anode cc
Anode co
6
Electrolyte
Capital cost
Cathode
Anode cc
Anode co
Operation
Electrolyte
Cathode
284
6 System analysis
In Systems 1 and 2, the fuel costs (F-costs), the D-costs and the
C-costs each contribute to approx. one third to the overall PPCs. In all
other systems, the F-costs account for approx. 35 to 45 % of the PPCs.
This is mainly due to the lower specific plant costs. Regarding the D-
costs, the most important cost drivers are already discussed above.
However, it is worthwhile noting that the fuel cell stack accounts for
merely 5 to 15 % of the overall PPCs.
In the systems 1, 3, 5, 6 and 7, the anode-supported counter-current cell
design yields the lowest PPCs compared to all other cell designs. Sys-
tem 1 is the only system, where the lowest PPCs are reached by a cell
design that does not also yield the lowest specific plant costs. Com-
pared to the cathode-supported cell design, the 8 % (relative) efficiency
advantage of the anode-supported counter-current cell design is just
sufficient to compensate 4 % higher specific plant costs. The counter
example is System 2, where the cathode-supported cell design yields
lower PPCs than the anode-supported counter-current cell design de-
spite the latter featuring an almost 10 % (relative) higher net AC system
efficiency. The reason for this can be found in the 20 % lower specific
plant costs resulting from the higher current density produced by the
cathode-supported cell due to high catalyst activity in dilute gases.
This clearly shows that the PPCs are more sensitive towards the spe-
cific plant costs than towards the system efficiency. The latter is effec-
tive only on the F-cost share, which accounts for moderate 30 to 45 % of
the PPCs even though wood is an expensive feedstock. Hence, effi-
ciency maximization is limited regarding the reduction of the PPCs. In
contrast, power output increase and reduction of specific plant costs
through optimization of the SOFC operational parameters is effective for
55 to 70 % of the PPCs. This emphasizes once more the importance of
the power output of the cell, which mainly depends on the anode cata-
lyst activity especially for the operation with dilute producer gases.
Figure 6-41 depicts the net power production costs assuming that the
entire heat produced is sold. The heat revenues are sufficient to cover
most of the F-costs and are therefore crucial for the market competive-
ness of B-IGFC systems. Given the electricity price currently paid in
Germany for electricity produced from biomass using highly advanced
technologies (gray marked area), [229], the only competitive systems in-
vestigated in this analysis are the Systems 4 and 5. In System 4, the
cathode-supported cell designs yields the lowest PPCs with
0.1154 €/kWhel net AC. In System 5, the anode-supported counter-
285
6 System analysis
current cell design is superior to the cathode-supported cell design with

a PPC of 0.1409 €/kWhel net AC. Under subsidization, these systems
may be already financially viable.
Figure 6-41: 0.45

Effective power Anode counter-current
Net power production cost [2007-€/kWhel net AC]

0.40 Anode co-current
production costs Electrolyte
assuming 100 % 0.35 Cathode
heat sales for the
different B-IGFC 0.30
systems and cell
designs 0.25
0.20
0.15
0.10
0.05
0.00
6.4.3 Sensitivity analysis

The impact of the most important assumptions in the economic analysis
was assessed through the sensitivity analysis presented below. The
base case for the sensitivity analysis is System 4 with the cathode-
supported cell design.
6.4.3.1 Economic model factors

The economic model involves a variety of parameters whereby only a
limited number can be actively influenced by engineers. These parame-
ters include the plant availability and the lifetime of employed catalysts.
Figure 6-42: 150%

Sensitivity of the
power production
Power production cost [% of Base case]
costs towards im- Plant factor

125%
portant parameters Catalyst lifetime
of the economic Wood price
Heat price
model DPC
100%
75%
50%
-80% -60% -40% -20% 0% 20% 40% 60% 80%
Parameter variation [% of Base case]
286
6 System analysis
Figure 6-42 shows that the plant availability, represented by the plant
factor (PF), has considerable PPC reduction potential. Increasing the PF
by 10 % yields 7 % lower PPCs with values around 0.107 €/kWhel net
AC. Reduced plant availability leads to fast increasing PPCs. With a PF
of 63 %, which is 26 % (relative) lower than the base case value, the
PPCs are already increased by 30 % and reach the lower limiting price
of 0.15 €/kWhel net AC for biomass electricity from highly advanced
technologies in Germany. A further reduction of the PF to approx. 45 %,
which corresponds to 52 % (relative) of the base case, yields an in-
crease of the PPCs by 73 % to 0.2 €/kWhel net AC, which is the upper
limiting electricity price according to [229]. This shows that the plant
availability plays a crucial role for the competitiveness of the B-IGFC
technology due to the high involved capital costs.
Figure 6-42 further depicts that the PPCs increase fast with catalyst life-
times below 50 % of the base case. This indicates that the catalyst de-
velopment should focus on the achievement of catalyst lifetimes around
one to two years. Further improvements regarding the lifetime of the
catalysts employed in the adiabatic methanation and the catalytic partial
oxidation reactors in the investigated systems 2, 4 and 5, have almost
no impact on the PPCs.
The corresponding parameters in the economic model were varied to
evaluate the general importance of the plant cost estimate accuracy, the
wood price and the heat price or demand, respectively. The wood price
and the heat price or demand, respectively, are equally sensitive and
considerably affect the PPCs. However, the PPCs stay below the lower
limiting electricity price of 0.15 €/kWhel net AC despite either a heat price
or demand decrease or a wood price increase by 40 % compared to the
base case. The latter shows that a wood price of 117 € per dry ton of
wood would still allow profitable operation of System 4 with the cathode-
supported cell design. The same applies, when assuming 30 % higher
direct plant costs, which corresponds to the maximum error expected for
the plant cost estimates.
6.4.3.2 SOFC degradation and operational voltage

The cell degradation causes a gradual reduction of the power output of
the investigated B-IGFC systems and is thus a cost-effective factor.
Figure 6-43 shows that several stack exchanges may be necessary dur-
ing the planning horizon to reach the lowest possible PPCs. The number
of required exchanges strongly depends on the degradation rate. With a
287
6 System analysis
degradation rate of 0.8 % per 1000 h, which corresponds to the fourfold

of the SECA target, three stack exchanges are necessary to reach the
lowest PPCs, which are increased by approx. 20 % compared to the
base case where the stack has to be exchanged only once.
Figure 6-43: 140%

Sensitivity of the Degradation - 0.1 % per 1000 h
Degradation - 0.2 % per 1000 h
power production
Power production cost [% of Base case]
Degradation - 0.4 % per 1000 h
costs towards deg- 130% Degradation - 0.8 % per 1000 h
radation and num-
ber of FC stack
exchanges during 120%
the planning horizon

110%
100%
90%
0 1 2 3 4 5 6 7 8 9
Numer of FC exchanges during planning horizon [-]
It may be that the cathode-supported cell in System 4 degrades faster

than in other systems due to the maximum solid temperature which was
found higher than the permitted temperature determined with the refer-
ence gas. Assuming the degradation rate increases by 100 %, the cor-
responding cathode-supported stack would need to be exchanged twice
during the planning horizon to reach the lowest possible PPCs. Even
though, the resulting PPCs are lower than those achieved by the second
best cell design for System 4, namely the anode-supported counter-
current cell. Note that this calculation does not consider possible reve-
nues arising from the sale of used stacks.
The operational voltage was varied for the systems 2, 4 and 7 employ-
ing the cathode-supported cell design to investigate the corresponding
impact on the PPCs, see Figure 6-44. The net AC system efficiency in-
creases with the operational cell voltage which is mainly due to the in-
creasing conversion efficiency of the SOFC following from Eq. 202.
However, the efficiency increase is not linear with the operational volt-
ages because of rising auxiliary power requirements. The latter result
from the higher air-to-fuel ratios which are required to compensate the
increasingly inhomogeneous current density and temperature distribu-
tions predicted for higher operational voltages, see Figure 6-46. At an
operational voltage of 0.7 V, System 4 features a net AC system effi-
288
6 System analysis
ciency of 37.2 %. Despite this very high efficiency, the PPCs increase by
almost 70 % compared to the base case yielding 0.1912 €/kWhel net AC,
which renders profitable operation difficult.
Figure 6-44: 40 0.3

Sensitivity of the Efficiency - System 2 Efficiency - System 4 Efficiency - System 7
PPC - System 2 PPC - System 4 PPC - System 7
net AC system
efficiency and
PP cost [2007-€/kWhel net AC]

35
Net AC system efficiency [%]
0.25
power production
costs towards the
operational cell
voltage 30 0.2
25 0.15
20 0.1
0.6 0.625 0.65 0.675 0.7
Voltage [V]
The same trend applies to System 2, which confirms the initial assump-
tion that the operational voltage has to be as low as possible to yield
reasonable power outputs and specific plant costs when SOFCs are op-
erated with dilute fuel gases. For System 7, the PPCs initially decrease
with increasing operational voltage. The minimum PPCs are reached at
0.63 V. Further increasing the operational voltage leads to increasing
PPCs similar to Systems 2 and 4. The reason for the initial PPCs de-
crease is that the un-diluted producer gas of System 7 causes consid-
erably less activation polarization losses. This reduces the power output
decrease resulting from the higher operational voltages.
Figure 6-45: 1.2

Sensitivity of the System 2
maximum tem- System 4
Maximum temperature gradient [K/mm]
perature gradients System 7
towards the op- 1
erational cell volt-

age
0.8
0.6
0.4
0.6 0.625 0.65 0.675 0.7
Voltage [V]
289
6 System analysis
Thus up to a certain point, the efficiency surplus is sufficient to compen-

sate the higher capital cost arising from the increased specific plant
costs due to lower power output.
Figure 6-46: 3500

0.6 V 0.633 V 0.666 V 0.7 V
Sensitivity of the
current density
distribution pre- 2800
dicted for the
producer gas of
2
System 7 towards 2100

the operational
cell voltage
1400
700
0
0.00 0.25 0.50 0.75 1.00 1.25 1.50
Cell length [m]
However, the thermal stress imposed on the SOFCs employed in the in-
vestigated systems increases with higher operational voltages. Figure
6-45 exemplifies that the maximum temperature gradients increase with
increasing operational voltages. The reason for this lies in the current
density distributions, which become increasingly inhomogeneous with
rising operational voltages. Figure 6-46 shows for the producer gas of
System 7 that despite in relative terms more producer gas is converted
at the fuel inlet with higher operational voltages, the absolute amount
decreases. This leads to a drop of the solid temperatures at the fuel inlet
as the heat produced in the electrochemical reactions becomes insuffi-
cient to cover the heat demand of the STR reactions taking place. The
lower solid temperatures at the cell inlet imply higher temperatures in
the latter cell parts, which in turn lead to higher temperature gradients.
Consequently, the PPCs benefit resulting from higher operational volt-
ages for the producer gas of System 7 may be cancelled out by faster
degradation due to higher thermal stress.
In conclusion it can be stated that none of the parameters of the eco-
nomic models and important assumptions of the system analysis inves-
tigated in this sensitivity analysis change the relative differences be-
tween the considered B-IGFC systems. The absolute PPC estimates are
however in parts considerably changed, which could lead to a better or
worse competitiveness of the investigated systems.
290
291
292
7 Concluding remarks
Biomass has a large potential for electricity generation among the re-
newable energy sources. A promising approach to achieve high electri-
cal efficiencies is the application of fuel cells for the conversion of pro-
ducer gases originating from biomass gasification. This combination is
referred to as "Biomass Integrated Gasification Fuel Cell System"
(B-IGFC). The solid oxide fuel cell in particular has gained much interest
due to its ability of converting hydrocarboneous fuel gases and its com-
parably low fuel gas requirements. This thesis aimed at the demonstra-
tion of the B-IGFC technology on kW-scale and at the identification of
promising B-IGFC system concepts with power outputs around 1 MWel
by means of a systems analysis.
7.1 Conclusions
7.1.1 Experiments
The PSI B-IGFC system is based on the updraft gasification process.
The resulting producer gas is de-dusted and its tar load is reduced
through a catalytic partial oxidation (CPO) before being fed to the SOFC.
The starting point for the experimental part of the thesis was the
achievement of stable long-term operation of the in-house developed
lab-scale updraft gasifier. Diverse modifications of the gasifier and the
addition of a steam generator made a stable operation over 165 h non-
stop possible. Experiments with SOFC stacks showed that tar is to
some extent a fuel for SOFCs and that sulfur reduces the power output
but does not inhibit stable operation of SOFCs. Experiments with a com-
mercial CPO catalyst revealed satisfying conversion performance for
oxygenated tars and aromatics. The CPO catalyst was further found
suitable for the decomposition of organic sulfur compounds.
The PSI B-IGFC concept was first time demonstrated on kW-scale. The
SOFC delivered 40 % less current when operated with producer gas
compared to partially oxidized methane. The gasifier and the CPO were
operated without problems. Ash deposits in the SOFC system were
found. Overall the first-time demonstration of the PSI B-IGFC concept
was successful and possibilities for future improvements of the experi-
mental setup were identified.
293
7.1.2 System analysis

While most published B-IGFC systems analyses employ lumped SOFC
models and focus on idealized fuel gases, the system analysis in this
thesis is based on a locally resolved SOFC model aiming at investiga-
tions of measured producer gases.
The finite volume method based SOFC model was developed with re-
spect to fast applicability to various cell geometries, short calculation
times and high fuel flexibility. To allow the identification of critical operat-
ing conditions in SOFCs, the model describes all relevant charge, mass
and heat transport processes. Accounting for the hydrocarbon content of
producer gases, a Langmuir-Hinshelwood type applied kinetic model for
steam reforming reactions was included. The model was applied to the
planar anode-supported co- and counter-current cell designs, the planar
electrolyte-supported cell design and the tubular cathode-supported cell
design with air delivery tube. The validation against experimental and
benchmark test data proved that the model behaves physically correct
with mass and energy balance errors considerably smaller than compa-
rable state-of-the-art models. To assure the fuel flexibility of the model,
an extensive sensitivity analysis was conducted. It was found that the
impact of different fuel gases on the operational conditions of SOFCs
generally dominates geometrical and material induced phenomena de-
scribed by the corresponding model parameters.
The main objective of the conducted system analysis was to assess the
technical and economical feasibility of various B-IGFC systems with a
power output up to 1 MWel based on already or soon available biomass
gasification processes, gas processing technologies and SOFC designs
for stationary power applications. The analysis assumes a standardized
SOFC stack size. Based on the downdraft and updraft gasification proc-
esses as well as the fluidized bed steam gasification process, seven
B-IGFC systems were defined, mainly differing in the way sulfur and tar
removal or tar conversion are conducted. The producer gas composi-
tions at the cell inlet were computed with ASPEN PLUS. In a first step,
the power output, the thermal gradients and temperature extremes re-
sulting from the various producer gas compositions were investigated.
With producer gases, the power output of the investigated SOFCs is de-
creased by minimum 25 % for the tubular cell and by maximum 70 % for
the planar electrolyte-supported cell compared to pre-reformed natural
gas. The activation losses are pronounced and eventually put into the
294
dominant role due to the dilute character of producer gases. Thus, high
anode catalyst activity is essential for high power outputs. According to
the model parameters used in the analysis, the tubular cell features a
highly active anode catalyst. The anode-supported counter-current cell
partially balances out the lower catalyst activity with high temperatures
at the cell inlet. Diffusion limitations do not gain importance for any of
the investigated producer gases and cell designs.
The cooling requirements strongly depend on the SOFC flow pattern
and the internal reforming potential of the producer gases. The required
air-to-fuel ratios (AF) generally decrease with increasing internal reform-
ing potential. The co-current cells are considerably more sensitive to-
wards the internal reforming potential than the counter-current cell,
which requires the lowest AFs of all cell designs. Compared to pre-
reformed natural gas, all investigated producer gases cause lower tem-
perature gradients and temperature extremes, which may reduce tem-
perature induced cell degradation.
The interactions between the gasification processes, gas processing
technologies and SOFC designs were investigated with respect to the
net system efficiencies and balance of plant requirements resulting from
the SOFC cooling. A generalized heat integration network was defined.
The corresponding simulations were performed with ASPEN PLUS.
The water concentrations of the raw producer gases are insufficient to
prevent carbon depositions at the operational temperature of the gener-
ally employed zinc oxide sulfur trap beds. In all investigated systems,
except the one employing a CPO, considerable amounts of water have
to be added to the producer gases.
The net system efficiency is preeminently determined by the cold gas ef-
ficiency of the biomass gasification process and the auxiliary power re-
quirements. The latter directly correlate with the required air-to-fuel ra-
tios, which in turn are a consequence of the internal reforming potential
of the corresponding producer gases. The adiabatic methanation and
the low-temperature CPO are effective means to increase the internal
reforming potential of producer gases.
With 32.1 %, the highest system efficiency is reached by the combina-
tion of the updraft gasification with an adiabatic methanation and the tu-
bular cell design. The high system efficiency results from the high cold
gas efficiency of the updraft gasification process which already yields a
producer gas with considerable internal reforming potential. The latter is
295
further increased through the adiabatic methanation, which yields the

fuel gas with the highest internal reforming potential in the entire analy-
sis with an overall cold gas efficiency of 85 %. The formed methane is
subsequently reformed in the SOFC directly using waste heat, which
can be interpreted as the final gasification step, additionally lowering the
cooling requirements and thus increasing the net system efficiency.
The fluidized bed steam gasification process also yields a producer gas
with high internal reforming potential, however with low cold gas effi-
ciency. The corresponding losses can not be compensated in the SOFC.
Downdraft gasification producer gases are almost fully reformed. Con-
sequently, the amount of waste heat produced in the SOFC is increased
leading to higher air-to-fuel ratios. Combined with the moderate cold gas
efficiency of the downdraft gasification process, the corresponding
B-IGFC systems feature the lowest system efficiencies in this analysis.
With respect to sulfur induced power losses, the neglect of organic sul-
fur compounds may lead to a considerable overestimation of the net
system efficiencies. This applies especially to updraft and fluidized bed
steam gasification based B-IGFC systems.
Based on the results of the system simulations, an economic model was
developed to determine the profitability of the different B-IGFC systems.
Major equipment pieces were sized and the direct plant costs were es-
timated using pertinent cost functions. The indirect plant costs were
computed using factors relating the indirect to the direct plant costs.
For downdraft and fluidized bed steam gasification based systems, the
wood drying is identified as considerable cost factor. The gas process-
ing is a minor cost factor. Nevertheless, the producer gas humidification
bears potential for cost reductions especially for updraft and fluidized
bed steam gasification based systems, which allow an increase of the
water content of the producer gases simply by adjusting the operational
parameters of the gasification processes. The fuel cell balance of plant
equipment is determined as dominant cost driver, amounting for up to
50 % of the direct plant costs, while the fuel cell itself accounts for up to
25 %. This emphasizes the importance of properly set cell inlet gas
temperatures to yield low AFs and gas processing steps which increase
the internal reforming potential of producer gases.
Relating the direct plant costs to the power output yields the specific
plant costs. The downdraft gasification based systems feature the high-
est specific plant costs due to low power outputs, high wood drying and
296
fuel cell balance of plant costs. For the fluidized bed steam gasification
based systems, slightly lower specific plant costs are determined mainly
because of the comparably high power outputs as result of the undiluted
character of the corresponding producer gases. The lowest specific
plant costs are established for the updraft gasification based systems.
This is mainly because of the systems simplicity resulting in low direct
plant costs and minor auxiliary power requirements.
The plant costs were related to the corresponding system efficiencies to
yield power production cost (PPC) estimates. The basis for the calcula-
tion of the PPCs was a detailed cost analysis for a given planning hori-
zon including costs not only for operation and maintenance, feedstock
and utilities but also capital costs resulting from plant costs depreciation
and interests. The net value of all costs was evenly distributed through-
out the planning horizon using the annuity method.
The PPCs analysis revealed that the fuel costs amount to on average
40 % of the PPCs, the remainder being capital costs. Thus, specific
plant cost reduction is considerably more effective on reducing the
PPCs than increasing the system efficiency.
An increase of the operational voltage from 0.6 V to 0.7 V yields a sig-
nificant increase of the SOFC conversion efficiency which in turn entails
a gain of the net system efficiency from 32.1 to 37.1 % for the updraft
gasification based system with adiabatic methanation. However, the
PPCs increase from 0.1154 €/kWhel to 0.1912 €/kWhel too, mainly due to
considerably reduced power output of the SOFC.
The reduction of auxiliary power requirements in contrast leads to an in-
crease of the system efficiency and a decrease of the specific plant
costs, thus considerably lowering the PPCs.
Finally, the revenues from heat sales are crucial for the economical vi-
ability of the investigated B-IGFC systems without bottoming cycles.
7.2 Recommendations for future work

7.2.1 Experiments
The following recommendations are made for future long-term test:
Integration of a hot gas particle filtration unit
The removal of particles must be optimized to prevent fouling in the
cooling air pre-heater or other cold areas of the demonstration unit.
297
The currently installed cyclone is not sufficient and should be re-

placed by the hot gas particle filtration unit designed in the framework
of a master thesis co-supervised by the author, Figure 7-1.
Figure 7-1:
Hot gas filtration system
To achieve a minimum fly ash load in the filtered producer gas, the fil-
tration temperature must be kept below the ash melting point. Usually
temperatures around 600 to 650 °C are chosen for this purpose.
Means for active cooling may have to be added to the hot gas filtra-
tion unit.
Application of fine meshed CPO monoliths
Measurements of the producer gas composition downstream a CPO
monoliths featuring 400 channels per square inch (cpsi) via a capil-
lary showed that tar conversions over 90 % can be achieved. The
application of such fine meshed monoliths in long-term tests was
however not possible until now due to blocking issues resulting from
the moderate particle removal efficiency of the used cyclone. After
successful commissioning of the hot gas particle filtration unit, the
currently used 100 cpsi CPO monolith should be replaced by a mono-
lith with 400 cpsi.
298
Integration of a sulfur adsorption unit

The scope of future long-term tests should not merely be to demon-
strate stable operation but also high current output of the SOFC sys-
tem. Sulfur poisoning induced voltage losses should therefore be
minimized by integrating a sulfur adsorption unit into the demonstra-
tion unit. As a consequence of the maximum operational temperature
of 450 °C of state-of-the-art sulfur adsorbent materials, the integra-
tion of the sulfur adsorption unit includes a heat exchanger for gas
cool down and/or a steam/water injection unit for additional humidifi-
cation of the producer gas. The latter might be necessary to prevent
carbon deposition problems at temperatures as low as 450 °C.
7.2.2 Modeling and system analysis

The following recommendations are made for future extensions of the
SOFC model developed in this thesis:
Embedding of the model into flowsheeting packages such as ASPEN
PLUS, BELSIM or IPSEpro for simulation of B-IGFC systems with
energy and mass flow recycles.
Extension of the model to display the computed heats of reactions.
Implementation of an initial conjecture estimation subroutine based
on the database of already available solutions.
Extension of the model to the second dimension to allow the simula-
tion of SOFCs with cross-current flow patterns such as e.g. the Rolls
Royce cells and the investigation of boundary effects.
Implementation of a subroutine to compute the effective open circuit
voltage resulting from hydrocarbon containing fuel gases.
Implementation of steam reforming kinetics specifically determined
for important tar species such as naphthalene etc.
Implementation of Langmuir-Hinshelwood steam reforming kinetics
including a sulfur adsorption term to determine the impact of sulfur
poisoning on the mass and heat transfer processes taking place in
SOFCs.
Implementation of the Dusty-Gas diffusion model for higher accuracy
of the diffusion loss estimation.
Implementation of subroutine to compute chemical equilibria for the
estimation of the formation risk of undesired solid species.
Implementation of a subroutine for thermal stress approximations.
299
Following recommendations are made for future systems analysis:

Systems analysis with minimum possible air-to-fuel ratio for all inves-
tigated producer gases
In future system analyses, the cell inlet gas temperatures should be
varied to yield the lowest possible air-to-fuel ratio for a given fuel gas
composition and cell design dependent mean and maximum cell tem-
peratures and temperature gradients. This accounts for the found ma-
jor importance of the air-to-fuel ratio regarding the net system effi-
ciency and the power production costs.
Systems analysis including gas turbine and rankine bottoming cycles
Bottoming cycles may be a good possibility to cover the auxiliary
power requirements resulting from the SOFC cooling, simultaneously
generating surplus electricity from the high-temperature heat pro-
duced in the SOFC. This way both cost-effective factors, net system
efficiency and specific plant costs, could be lowered, provided that
the power surplus generated by the bottoming cycle is sufficient to
compensate the additional system costs.
300
301
302
Notation
Notation
Latin symbols
A Area [m2]
AmP Amortization period [a]
AN Annuity factor [%]
c Carbon mass fraction [-]
C Cost [€]
cp Heat capacity [J/(mol K)]
CP Length of construction period [a]
d Diameter [m]
D Diffusion coefficient [m2/sec]
DA Depreciation/Amortization costs [€]
DeP Depreciation period [a]
DK Knudsen diffusion coefficient [m2/sec]
Dm Molecular diffusion coefficient [m2/sec]
DP Dept portion [%]
DPC Direct plant costs [€]
DR Degradation rate [%/(1000 h)]
E Voltage [V]
Eact Activation energy for exchange current density [J/mol]
EqH2 Hydrogen equivalent molar flow [mol/sec]
F Faraday constant [C/mol]
G Gibbs enthalpy [J/mol]
h Height, Hydrogen mass fraction [m], [-]
H Enthalpy [J/mol]
I Current density [A/m2]
I0 Exchange current density [A/m2]
IF Inflation rate [%]
IPC Indirect plant costs [€]
IR Interest rate [%]
Kp Equilibrium constant [-]
L Length of cell-subunit [m]
303
Notation
l Length [m]
lc Length of planar cell [m]
lD8 Length of Delta8 cell [m]
LHV Lower heating value [J/mol]
lt Length of tubular cell [m]
M Molecular weight, Water load [kg/mol], [%]
m Mass [kg]
n Number of electrons, Nitrogen mass fraction [-]
nch Number of channels of planar cell [-]
NCIR Nominal capital interest rate [%]
ndot Molar flow [mol/sec]
nRE Number of repeating elements [-]
ntri Number of triangular tubes in Delta8 cell [-]
Nu Nusselt number [-]
o Oxygen mass fraction [-]
p Total or partial pressure [N/m2]
P Power [W]
p Price [€]
Pel Electrical Power (DC) [W]
PF Plant factor [%]
PH Planning horizon [a]
PPC Power production costs [€/kWHel]
Qdot Heat flux [W/m]
R Resistance or Ideal gas constant [Ω], [J/(mol K)]
r Radius [m]
rdl-reac Diffusion limited, length specific reaction rate [mol/(sec m)]
ri Inner radius [m]
rj Reaction rate of reaction j [mol/(sec m)]
rm Middle radius [m]
ro Outer radius [m]
ro Reaction order [-]
RoE Return on equity [%]
s Sulfur mass fraction [-]
SoP Start of production [-]
T Temperature [K]
TadtK Gas temperature in air delivery tube [K]
304
Notation
TaK Anode gas temperature [K]

Tamb Ambient temperature [K]
TI Total investment costs [€]
TRD Threshold degradation [%]
TsK Solid structure temperature [K]
UF Fuel utilization [-]
w Humidity [%]
wc Channel width of planar cell channels [m]
wce Width of planar cell [m]
wD8 Delta8 cell width [m]
WI Yearly wood price increase %
X Fraction [%]
xp Active fraction of area covered by rib [-]
xt Inactive fraction of tubular cell area [-]
y Molar fraction [-]
Greek symbols
α Upper triangular half angle [°]
α Convective heat exchange coefficient [W/(m2 K)]
αinsul Heat transfer coefficient through insulation [W/(m2 K)]
β Transfer coefficient [-]
βx,diff Mass transfer coefficient of specie x [m/sec]
βx,diff-reac Diffusion limited reaction conversion coefficient of specie x
βx,reac Reaction conversion coefficient of specie x
δ Thickness of component [m]
ΔH Heat of reaction [J/mol]
ε Porosity of porous media [-]
γ Pre-exponential factor [A/m2]
η Efficiency [%]
ηact Activation polarization voltage loss [V]
ηdiff Diffusion polarization voltage loss [V]
ηohm Ohmic voltage loss [V]
λ Air-to-fuel ratio [-]
λan Thermal conductivity of anode gas [W/(m K)]
λca Thermal conductivity of cathode gas [W/(m K)]
305
Notation
λs Solid structure heat conductivity coefficient [W/(m K)]

νav Average molecular speed [m/sec]
νij Stoiciometric coefficient [-]
ρ Specific conductivity [1/(Ω m)]
τ Tortuosity of porous media [-]
Subscripts and superscripts

0,in Inlet
act Active, Activation, Actual
ADT Air delivery tube
AEC Anode-electrolyte-cathode assembly
Am Amortization
an Anode gas channel, Anode
Ash Ash
Aux Auxiliary
av Average
blow Blower, Fan, Pump
ca Cathode gas channel, Cathode
Cap Capital
cat Catalyst
cat0 First catalyst batch in planning horizon
CG Clean gas, cold gas
chact Chemically active
circ Circumferential
conv Convective
cross Cross-sectional
cs Catalyst surface
d Dry
D8 Delta8
DC Direct current
Dep Depreciation
ed Educts
eff Effective
el Electrolyte
306
Notation
elact Electrochemically active

end End
equiv Equivalent
FC Fixed carbon, Fuel cell
FC0 First fuel cell in planning horizon
gasi Gasification
GP Gas processing
hloss Heat loss
hyd Hydraulic
ic Interconnect
IE Interest expense
Ins Insurance
Inv Inverter
mix Gas mixture
net Net
O+M Operation and maintenance
op Operational
oxi Oxidation
p Planar
PG Producer gas
prod Products
s Start
S Sulfur
SE Standard equipment
SH Sensible heat
SOFC Solid oxide fuel cell
start Start
stoic Stoiciometric
STR Steam reforming
sys System
t Tubular
th Thermal
tot Total
TPB Triple phase boundary
VM Volatile matter
WGS Water gas shift
307
Notation
Abbreviations
AC Alternating current
ADT Air delivery tube
AEC Anode-electrolyte-cathode assembly
AF Air-to-fuel ratio
AFC Alkaline fuel cell
APU Auxiliary power unit
ATR Autothermal reforming
BHPR Biomass Heatpipe Reformer
B-IGFC Biomass Integrated Gasification Fuel Cell System
BMT Benchmark test
BoP Balance of plant
BtL Biomass to Liquids
CECPI Chemical Engineering Magazine Plant Cost Index
CFD Computational fluid dynamics
CGO Gadolinium doped ceria
CHP Combined heat and power
CPO Catalytic partial oxidation
cpsi Channels per square inch
CSTR Continuous stirred tank reactor
CV Caloric value, Control volume
D8 Delta8 cell design by the Siemens AG
daf Dry and ash free
DC Direct current
DMFC Direct methanol fuel cell
dtf Dry, tar and ash free
FC Fuel cell
FEM Finite element method
FICFB Fast Internally Circulating Fluidized Bed
FVM Finite volume method
GC Gas chromatography
GT Gas turbine
HDP High power density
HGF Hot gas filtration
308
Notation
HHV Higher heating value

HIN Heat integration network
IC Interconnector, Bipolar plate
ICE Internal combustion engine
IEA International Energy Agency
LH Langmuir-Hinshelwood kinetics
LHV Lower heating value
LSCr Lanthanum strontium chromium perovskite
LSGM Lanthanum strontium magnesium gallate
LSM Lanthanum strontium manganite
MCFC Molten carbonate fuel cell
MFC Mass flow controller
MIEC Mixed ionic and electronic conductor
NVAM Numerical volume averaging method
OCV Open circuit voltage
PAFC Phosphoric acid fuel cell
PEMFC Polymer electrolyte membrane fuel cell
PG Producer gas
PL Power law kinetics
PO Partial oxidation
PPC Power production costs
PSI Paul Scherrer Institute
REV-WGS Reverse water gas shift
SAI Secondary air injection
SC Steam-to-carbon ratio
ScSZ Scandia stabilized zirconia
SECA Solid State Energy Conversion Alliance
SOFC Solid oxide fuel cell
STR Steam reforming
TPB Triple phase boundary
UI Voltage-current
WGS Water gas shift
WHSV Weight hourly space velocity
YSZ Yttria stabilized zirconia
309
310
References
References
[1] Novatlantis, Smarter living - Generating a new understanding for natural resources as
the key to sustainable development - the 2000-watt society, Dübendorf, Switzerland,
2005.
[2] A. Wokaun, Erneuerbare Energie, Teubner Verlag, Stuttgart, Leipzig, Germany, 1999.
[3] World Energy Outlook 2004, International Energy Agency, Paris, France, 2004.
[4] T. Nussbaumer, Technische Energienutzung von Biomasse, Swiss Federal Institute of

Technology, Zurich, Switzerland, 2004.
[5] M. Gronli, A theoretical and experimental study of the thermal degradation of biomass,
Norwegian University of Science and Technology, Trondheim, Norway, 1996.
[6] P. McKendry, Bioresour. Technol., 83 (2002) 37-46.
[7] G. W. Huber, S. Iborra, A. Corma, Chem. Rev., 106 (2006) 4044-4098.
[8] T. A. Hsu, M. R. Ladisch, G. T. Tsao, Chemtech, 10 (1980) 315-319.
[9] M. A. Serio, M. A. Wojtowicz, S. Charrenay, A. Bisio, S. Booty, Encyclopedia of En-

ergy Technology and the Environmen, Wiley, New York, United States of America, 1995.
[10] N. Arnstein, Experimental investigation of Solid Oxide Fuel Cells using biomass gasi-
fication producer gases, Norwegian University of Science and Technology, Trondheim,
Norway, 2005.
[11] B. M. Jenkins, L. L. Baxter, T. R. Miles, T. R. Miles, Fuel Process. Technol., 54

(1998) 17-46.
[12] L. L. Baxter, Biomass Bioenerg., 4 (1993) 85-102.
[13] J. R. Rostrupnielsen, L. J. Christiansen, Appl. Catal. A-Gen., 126 (1995) 381-390.
[14] K. Raveendran, A. Ganesh, K. C. Khilar, Fuel, 74 (1995) 1812-1822.
[15] K. Raveendran, A. Ganesh, K. C. Khilar, Fuel, 75 (1996) 987-998.
[16] H. Recknagel, E. Sprenger, W. Hönmann, Taschenbuch für Heizung und Klimatech-

nik, Universität Dortmund, Dortmund, Germany, 1994.
[17] User Documentation for Aspen Plus, Version 12.1, ASPEN Tech, Cambridge, United
States of America, 2002.
311
References
[19] A. Kaupp, J. Goss, Small scale gas producer-engine systems, German Appropriate
Technology Exchange (GATE), Eschborn, Germany, 1984.
[20] P. Morf, Secondary Reactions of Tar during Thermochemical Biomass Conversion,

Swiss Federal Institute of Technology, Zurich, Switzerland, 2001.
[21] J. Warnatz, U. Maas, R. W. Dibble, Verbrennung, Springer, Berlin, Germany, 1997.
[22] M. Bolhàr-Nordenkampf, H. Hofbauer, Biomass gasification combined cycle thermo-

dynamic optimisation using integrated drying, in: Proceedings of the ASME Turbo Expo
2004: Power for Land, Sea and Air, 2004, 14th to 17th June.
[24] J. C. Lee, R. A. Yetter, F. L. Dryer, Combust. Flame, 101 (1995) 387-398.
[25] A. Vogel, M. Bolhàr-Nordenkampf, M. Kaltschmitt, H. Hofbauer, R. Berger, A. Kather,

M. Kübel, B. Staiger, L. Wiese, Analyse und Evaluierung der thermo-chemischen Verga-
sung von Biomasse, Landwirtschaftsverlag GmbH, Münster, Germany, 2006.
[26] K. R. Cummer, R. C. Brown, Biomass Bioenerg., 23 (2002) 113-128.
[27] P. Morf, P. Hasler, T. Nussbaumer, Fuel, 81 (2002) 843-853.
[28] B. Teislev, Wood-chips gasifier combined heat and power, Babcock & Wilcox Volund
R&D Centre, Kolding, Denmark, 2001.
[29] B. Teislev, Harboore-woodchips gasifier and 1500 kW gas-engines operating at 32%

power efficiency in CHP configuration, in: Proceedings of the 12th European Conference
on Biomass, Amsterdan, 2002, 17th to 21st June.
[30] A. Beenackers, Renew. Energy, 16 (1999) 1180-1186.
[31] Website, http://www.pyroforce.ch, January 2008.
[32] Website, http://www.pyrocon.ch, January 2008.
[33] T. Metz, Allotherme Vergasung von Biomasse in indirekt beheizten Wirbelschichten,

Technische Universität München, Munich, Germany, 2006.
[34] D. J. Stevens, Hot Gas Conditioning: Recent Progress with Larger-Scale Biomass
Gasification Systems, National Renewable Energy Laboratory, Golden, United States of
America, 2001.
312
References
[35] R. Rauch, C. Pfeifer, K. Bosch, H. Hofbauer, D. Swierczynski, C. Courson, A. Kien-

nemann, Comparison of different olivines for biomass steam gasification, in: Proceedings
of the Science in Thermal and Chemical Biomass Conversion, Victoria, Canada, 2004,
30th August to 2nd September.
[36] Website, http://www.heatpipe-reformer.de, January 2008.
[37] R. Rauch, H. Hofbauer, Zweibett-Wirbelschichtvergasung in Güssing mit

2MWel/4,5MWth, in: Proceedings of the 7tes Holzenergiesymposion, Zurich, Switzerland,
2002, 18th October.
[38] E. Heinrich, The status of the FZK concept of biomass gasification, in: Proceedings
of the 2nd European Summer School on Renewable Motor Fuels, Warsaw, Poland, 2007,
29th to 31st August.
[39] O. Schulze, Advanced gas cleaning for biomass gasification, in: Proceedings of the
IEA Workshop 5, Dresden, Germany, 2006, 12th to 14th June.
[40] M. Seemann, Methanation of biosyngas in a fluidized bed reactor, Swiss Federal In-
stitute of Technology, Zurich, Switzerland, 2007.
[41] B. Lohrengel, Abgasreinigung/Immisionsschutz, FH Heilbronn, Heilbronn, Germany,

2004.
[42] W. Mojtahedi, E. Kurkela, M. Nieminen, J. Inst. Energy, 63 (1990) 95-100.
[43] D. C. Dayton, B. M. Jenkins, S. Q. Turn, R. R. Bakker, R. B. Williams, D. Belle-

Oudry, L. M. Hill, Energy Fuels, 13 (1999) 860-870.
[44] K. Salo, W. Mojtahedi, Biomass Bioenerg., 15 (1998) 263-267.
[45] S. Q. Turn, C. M. Kinoshita, D. M. Ishimura, T. T. Hiraki, J. Zhou, S. M. Masutani, Ind.

Eng. Chem. Res., 40 (2001) 1960-1967.
[46] N. J. J. Dekker, G. Rietveld, J. Fuel Cell Sci. Technol., 3 (2006) 499-502.
[47] A. Wojcik, H. Middleton, I. Damopoulos, J. Van herle, J. Power Sources, 118 (2003)
342-348.
[48] Website, http://www.ecn.nl/phyllis, December 2007.
[49] M. P. Cal, B. W. Strickler, A. A. Lizzio, Carbon, 38 (2000) 1757-1765.
[50] K. Kendall, Nature, 404 (2000) 233-235.
[51] T. Sanderson, An updated assesment of the prospects for fuel cells in stationary
power and CHP, DTI, Oxfordshire, England, 2005.
313
References
[52] L. Carrette, K. A. Friedrich, U. Stimming, Fuel Cells, 1 (2001) 5-39.
[53] R. Leinfelder, Reaktionskinetische Untersuchungen zur Methan-Dampf-

Reformierung und Shift-Reaktion an Anoden oxidkeramischer Brennstoffzellen, Universi-
tät Erlangen-Nürnberg, Erlangen-Nürnberg, Germany, 2004.
[54] R. Coll, J. Salvado, X. Farriol, D. Montane, Fuel Process. Technol., 74 (2001) 19-31.
[55] D. Wang, D. Montane, E. Chornet, Appl. Catal. A-Gen., 143 (1996) 245-270.
[56] S. Czernik, A. V. Bridgwater, Energy Fuels, 18 (2004) 590-598.
[57] D. N. Wang, S. Czernik, E. Chornet, Energy Fuels, 12 (1998) 19-24.
[58] J. Larminie, A. Dicks, Fuel Cell Systems Explained, John Wiley & Sons, Ltd., Chices-
ter, England, 2000.
[59] S. D. Vora, SECA Program at Siemens Westinghouse, in: Proceedings of the 6th
Annual SECA Workshop, Pacific Grove, United States of America, 2005, 18th to 21st
April.
[60] L. F. Brown, Int. J. Hydrog. Energy, 26 (2001) 381-397.
[61] S. Shaffer, Development Update on Delphi's Solid Oxide Fuel Cell System, in: Pro-
ceedings of the 6th Annual SECA Workshop, Pacific Grove, United States of America,
2005, 18th to 21st April.
[62] M. Klemm, Heissentteerung von Brenngas aus der Vergasung von Biomasse durch
katalytische partielle Oxidation, Fortschritt-Berichte VDI Reihe 6 Nr. 525, Düsseldorf,
Germany, 2005.
[63] U. Bartmann, Reaktionstechnische Untersuchungen zur katalytischen partiellen Oxi-

dation von Methan mit Sauerstoff zu Synthesegas in Festbettreaktoren, Ruhr-Universität
Bochum, Germany, Bochum, 1999.
[64] T. Miyazawa, T. Abe, K. Kunimori, K. Tomishige, J. Jpn. Pet. Inst, 48 (2005) 162-172.
[65] U. Bossel, The Birth of the Fuel Cell (1835-1845). Complete Correspondence be-
tween Christian Friedrich Schönbein and William Robert Grove, European Fuel Cell Fo-
rum, Lucern, Switzerland, 2000.
[66] Website, http://www.cogeneration.net/molten_carbonate_fuel_cells.htm, December

2006.
[67] R. Bosch, Perspectives on Fuel Cells vs. Incumbent Technologies, in: Proceedings
of the 2005 Fuel Cell Seminar, Palm Springs, United States of America, 2005, 14th to
18th November.
314
References
[68] L. J. Gauckler, D. Beckel, B. E. Buergler, E. Jud, U. R. Muecke, M. Prestat, J. L. M.

Rupp, J. Richter, Chimia, 58 (2004) 837-850.
[69] R. Kee, H. Zhu, D. Goodwin, Proceedings of the Combustion Institute, 30 (2005)

2379-2404.
[70] S. C. Singhal, K. Kendall, High-temperature Solid Oxide Fuel Cells: Fundamentals,

Design and Applications, Elsevier, London, Great Britain, 2004.
[71] Fuel Cell Handbook (Sixth Edition), DOE/NETL, Morgantown, West Virginia, United
States of America, 2002.
[72] Newsletter, FuelCell Japan, 1 (2004) 1-15.
[73] F. de Bruijn, Green Chem., 7 (2005) 132-150.
[74] A. Baker, Fuel Cell Today Market Survey: Large Stationary Applications,
www.fuelcelltoday.com, 2005.
[75] L. Blum, W. A. Meulenberg, H. Nabielek, R. Steinberger-Wilckens, Int. J. Appl. Ce-

ram. Technol., 2 (2005) 482-492.
[76] P. Bance, N. P. Brandon, B. Girvan, P. Holbeche, S. O'Dea, B. C. H. Steele, J.

Power Sources, 131 (2004) 86-90.
[77] B. Morel, J. Laurencin, Y. Bultel, F. Lefebvre-Joud, J. Electrochem. Soc., 152 (2005)

A1382-A1389.
[78] L. Ceres Power, DTI Project Summary No. PS210, (2005).
[79] S. M. Haile, Acta Mater., 51 (2003) 5981-6000.
[80] S. Elangovan, B. Heck, S. Balagopal, I. Bay, D. Larsen, Challenges in the Use of

Lanthanum Gallate Electrolyte for Intermediate TemperatureSOFCs, in: Proceedings of
the First International Conference on Fuel Cell Development and Deployment, Storrs,
United States of America, 2004, 7th to 10th March.
[81] O. Marina, J. W. Stevenson, Development of Ceramic Composites as SOFC Anodes,

in: Proceedings of the SECA Core Technology Program Review Meeting, Albany, United
States of America, 2003, 30th September.
[82] M. Hill, Ceramic Energy: Material Trends in SOFC Systems,

www.ceramicindustry.com, 2005.
[83] M. P. Brady, B. A. Pint, T. R. Armstrong, C. E. Milliken, E. D. Kreidler, Z. G. Lu, J. H.

Zhu, Oxidation behaviour and in-cell performance of developmental SOFC interconnect
315
References
alloys, in: Proceedings of the 2002 ASM Fall Meeting, Columbus, United States of Amer-
ica, 2002, 7th to 10th October.
[84] C. Voisard, Das planare Sulzer Hexis SOFC Konzept, in: Proceedings of the DGM-
Fortbildungsseminar Werkstofffragen der Hochtemperatur-Brennstoffzelle (SOFC), Jülich,
Germany, 2005, 20th to 22nd April.
[85] T. A. Morris, E. A. Barringer, S. C. Kung, R. W. McKain, MRS Bull., 30 (2005) 596-

600.
[86] L. Jones, Pressurisation of IP-SOFC technology for second generation hybrid appli-
cation, DTI, Oxfordshire, England, 2005.
[87] Development of innovative metal-supported IT-SOFC technology, DTI, Oxfordshire,

England, 2005.
[88] R. A. George, SECA Project at Siemens Westinghouse, in: Proceedings of the 3rd
Annual Solid State Energy Conversion Alliance (SECA) Workshop, Washington D.C.,
United States of America, 2002, 21st to 22nd March.
[89] N. Bessette, Status of the Acumentrics SOFC Program, in: Proceedings of the 5th
SECA Annual Workshop, Boston, United States of America, 2004, 11th to 13th May.
[90] P. Patel, Thermally Integrated High Power Density SOFC Generator, in: Proceed-
ings of the 6th SECA Annual Workshop, Pacific Grove, United States of America, 2005,
18th to 21st April.
[91] R. Steinberger-Wilckens, Recent Results of SOFC Stack Development at For-

schungszentrum Juelich, in: Proceedings of the 2004 Fuel Cell Seminar, San Antonio,
United States of America, 2004, 2nd to 5th November.
[92] N. Christiansen, J. B. Hansen, S. Kristensen, H. Holm-Larsen, S. Linderoth, P. V.

Hendriksen, P. H. Larsen, M. Mogensen, Status and Recent Progress in SOFC Devel-
opment at Topsoe Fuel Cell /Risø National Laboratory, in: Proceedings of the 2005 Fuel
Cell Seminar, Palm Springs, United States of America, 2005, 14th to 18th November.
[93] C. Wunderlich, M. Boltze, Auxiliary Power Unit Development - How the System Con-
cept interacts with the Stack Concept, in: Proceedings of the 2005 Fuel Cell Seminar,
Palm Springs, United States of America, 2005, 14th to 18th November.
[94] D. Norrick, Cummins Power Generation SECA Program Phase 1 Results and Ex-
perience, in: Proceedings of the 2005 Fuel Cell Seminar, Palm Springs, United States of
America, 2005, 14th to 18th November.
[95] S. C. Singhal, Solid State Ion., 135 (2000) 305-313.
316
References
[96] R. Stübner, Untersuchungen zu den Eigenschaften der Anode der Festoxid-

Brennstoffzelle (SOFC), Technische Universität Dresden, Dresen, Germany, 2002.
[97] S. P. Jiang, S. H. Chan, J. Mater. Sci., 39 (2004) 4405-4439.
[98] W. Z. Zhu, S. C. Deevi, Mater. Sci. Eng. A-Struct. Mater. Prop. Microstruct. Process.,
362 (2003) 228-239.
[99] E. Ivers-Tiffee, A. Weber, D. Herbstritt, J. European Ceram. Soc., 21 (2001) 1805-

1811.
[100] D. Simwonis, F. Tietz, D. Stover, Solid State Ion., 132 (2000) 241-251.
[101] J. Helbig, U. Schönholzer, Grundzüge der Keramik - Skript zur Vorlesung In-
genieurkeramik I, Swiss Federal Institute of Technology, Zürich, Switzerland, 2001.
[102] S. P. S. Badwal, Solid State Ion., 52 (1992) 23-32.
[103] S. Primdahl, M. Mogensen, J. Appl. Electrochem., 30 (2000) 247-257.
[104] E. Konysheva, J. Mertens, H. Penkalla, L. Singheiser, K. Hilpert, J. Electrochem.

Soc., 154 (2007) B1252-B1264.
[105] M. J. Jorgensen, P. Holtappels, C. C. Appel, J. Appl. Electrochem., 30 (2000) 411-

418.
[106] J. R. Rostrup-Nielsen, J. Sehested, Catalyst Deactivation 2001, Proceedings, 139

(2001) 1-12.
[107] M. V. Twigg, Catalyst Handbook, Wolfe Publishing Ltd., Frome, England, 1989.
[108] D. L. Trimm, Catal. Rev.-Sci. Eng., 16 (1977) 155-189.
[109] C. H. Bartholomew, Catal. Rev.-Sci. Eng., 24 (1982) 67-112.
[110] G. C. Bond, Appl. Catal. A-Gen., 149 (1997) 3-25.
[111] H. Kim, C. Lu, W. L. Worrell, J. M. Vohs, R. J. Gorte, J. Electrochem. Soc., 149

(2002) A247-A250.
[112] E. Achenbach, IEA Programm on R,D&D on advanced fuel cells: Final Report of
Activity A2, (1996).
[113] A. Selimovic, M. Kemm, T. Torisson, M. Assadi, J. Power Sources, 145 (2005) 463-
469.
[114] Structural limitations in the scale-up of anode supported SOFCs, Final report to
DOE NETL, Cambridge, United States of America, 2002.
317
References
[115] A. L. Dicks, J. Power Sources, 71 (1998) 111-122.
[116] U. Hennings, M. Brune, R. Reimert, GWF Gas Erdas, 145 (2004) 92-97.
[117] O. Marina, L. R. Pederson, D. J. Edwards, C. W. Coyle, J. Templeton, M. Engel-

hard, Z. Zhu, SOFC Operation on Hydrogen and Coal Gas in the Presence of Phospho-
rus, Arsenic and Sulfur Impurities, in: Proceedings of the 8th Annual SECA Workshop,
San Antonio, United States of America, 2007, 7th to 9th August.
[118] Y. Matsuzaki, Y. Isamu, Solid State Ion., 132 (2000) 261-269.
[119] L. Aguilar, S. Zha, Z. Cheng, J. Winnick, M. Liu, J. Power Sources, 135 (2004) 17-
24.
[120] R. H. Cunningham, M. Fowles, R. M. Ormerod, J. Staniforth, Sulphur poisoning of

the active materials used in SOFCs, DTI, Report F/01/00222/REP, 2004.
[121] H. Y. Tu, U. Stimming, J. Power Sources, 127 (2004) 284-293.
[122] W. C. Li, H. L. Bai, J. N. Hsu, S. N. Li, C. C. Chen, Ind. Eng. Chem. Res., 47 (2008)
1501-1505.
[123] R. Sime, S. Stucki, S. Biollaz, A. Wiasmitinow, Linking wood gasification with SOFC
hybrid processes, in: Proceedings of the 5th European SOFC Forum, Lucerne, Switzer-
land, 2002, 2nd to 5th July.
[124] R. Sime, A. Wiasmitinow, S. Biollaz, S. Stucki, Evaluation of Biomass Fuelled Solid

Oxide Fuel Cell - Gas Turbine Hybrid Process, in: Proceedings of the 2nd World Confer-
ence and Technological Exhibition on Biomass for Energy, Industry and Climate Protec-
tion, Rom, Italy, 2004, 10th to 14th May.
[125] F. P. Nagel, M. Künstle, S. Biollaz, Link-up of a SOFC to a Wood Gasifier, in: Pro-
ceedings of the 14th European Biomass Conference and Exhibition: Biomass for Energy,
Industry and Climate Protection, Paris, France, 2005, 17th to 21st October.
[126] T. Borowiecki, Pol. J. Chem., 67 (1993) 1755-1766.
[127] B. Coughlan, M. A. Keane, J. Mol. Catal., 71 (1992) 93-109.
[128] K. Eguchi, H. Kojo, T. Takeguchi, R. Kikuchi, K. Sasaki, Solid State Ion., 152
(2002) 411-416.
[129] D. Singh, E. Hernandez-Pacheco, P. N. Hutton, N. Patel, M. D. Mann, J. Power

Sources, 142 (2005) 194-199.
[130] K. Sasaki, Y. Teraoka, J. Electrochem. Soc., 150 (2003) A878-A884.
318
References
[131] K. Sasaki, Y. Teraoka, J. Electrochem. Soc., 150 (2003) A885-A888.
[132] A. O. Omosun, A. Bauen, N. P. Brandon, C. S. Adjiman, D. Hart, J. Power Sources,

121 (2004) 96-106.
[133] D. Wang, S. Czernik, D. Montane, M. Mann, E. Chornet, Ind. Eng. Chem. Res., 36
(1997) 1507-1518.
[134] C. H. Bartholomew, P. K. Agrawal, J. R. Katzer, Adv. Catal., 31 (1982) 135-242.
[135] C. H. Bartholomew, Appl. Catal. A-Gen., 212 (2001) 17-60.
[136] J. A. Rodriguez, J. Hrbek, Accounts Chem. Res., 32 (1999) 719-728.
[137] L. E. Amand, H. Kassman, M. Karlsson, B. Leckner, J. Inst. Energy, 70 (1997) 25-

30.
[138] L. de Sousa, Gasification of wood, urban wastewood (Altholz) and other wastes in
a fluidized bed reactor, Swiss Federal Institute of Technology, Zurich, Switzerland, 2001.
[139] E. Achenbach, J. Power Sources, 49 (1994) 333-348.
[140] C. Stiller, Design, Operation and Control Modelling of SOFC/GT Hybrid Systems,
Norwegian University of Science and Technology, Trondheim, Norway, 2006.
[141] A. Selimovic, Modelling of Solid Oxide Fuel Cells Applied to the Analysis of Inte-
grated Systems with Gas Turbines, Lund University, Lund, Sweden, 2002.
[142] W. Zhang, E. Croiset, P. L. Douglas, M. W. Fowler, E. Entchev, Energy Convers.

Manage., 46 (2005) 181-196.
[143] T. Kivisaari, P. Björnbom, C. Sylwan, B. Jacquinot, D. Jansen, A. de Groot, Chem.

Eng. J., 100 (2004) 167-180.
[144] S. Campanari, J. Power Sources, 92 (2001) 26-34.
[145] T. Reed, S. Gaur, A Survey of biomass gasification 2001 - Gasifier projects and
manufacturers around the world, Solar Energy Research Institute, Golden, United States
of America, 2001.
[146] F. Hebmüller, Einflussfaktoren auf die Kieselgurfiltration von Bier, TU Bergakade-

mie Freiberg, Freiberg, Germany, 2003.
[147] F. P. Nagel, M. Jenne, S. Biollaz, S. Stucki, Link-up of a SOFC with an updraft-

wood gasifier via hot gas processing, in: Proceedings of the 2006 Fuel Cell Seminar-30th
Anniversary, Honolulu, USA, 2006, 13th to 17th November.
319
References
[148] C. Pitta, Investigation of tar reforming employing a catalytic partial oxidation cata-
lyst, Swiss Federal Institute of Technology, Zurich, Switzerland, 2007.
[149] M. Damsohn, Heissgasentschwefelung im Biomass-Integrated-Gasification-Fuel-

Cell-System, Technische Universität Clausthal, Clausthal, Germany, 2006.
[150] Website, www.hexis.com, January 2008.
[151] A. L. Dicks, J. Power Sources, 61 (1996) 113-124.
[152] P. Hofmann, A. Schweiger, L. Fryda, K. D. Panopoulos, U. Hohenwarter, J. D.

Bentzen, J. P. Ouweltjes, J. Ahrenfeldte, U. Henriksen, E. Kakaras, J. Power Sources,
173 (2007) 357-366.
[153] F. P. Nagel, S. Biollaz, M. Jenne, S. Stucki, Link-up of a 1kW-SOFC with an up-

draft-wood gasifier via hot gas processing, in: Proceedings of the 15th European Bio-
mass Conference 2007: From Research to Market Deployment, Berlin, Germany, 2007,
7th to 11th May.
[154] J. Palsson, Thermodynamic Modelling and Performance of Combined Solid Oxide

Fuel Cell and Gas Turbine Systems, Lund University, Lund, Sweden, 2002.
[155] B. Thorud, Dynamic Modelling and Characterisation of a Solid Oxide Fuel Cell Inte-
grated in a Gas Turbine Cycle, Norwegian University of Science and Technology, Trond-
heim, Norway, 2005.
[156] T. Kivisaari, P. Björnbom, C. Sylwan, J. Power Sources, 104 (2002) 115-124.
[157] E. Liese, R. S. Gemmen, J. Eng. Gas Turb. Power, 127 (2005) 86-90.
[158] S. Ghosh, S. De, Energy, 31 (2006) 345-363.
[159] Y. Yi, A. D. Rao, J. Brouwer, G. S. Samuelsen, J. Power Sources, 144 (2005) 67-
76.
[160] J. Van herle, F. Maréchal, S. Leuenberger, Y. Membrez, O. Bucheli, D. Favrat, J.

Power Sources, 131 (2004) 127-141.
[161] R. Bove, P. Lunghi, N. M. Sammes, Int. J. Hydrogen Energy, 30 (2005) 181-187.
[162] R. Suwanwarangkul, E. Croiset, M. W. Fowler, P. L. Douglas, E. Entchev, M. A.

Douglas, J. Power Sources, 122 (2003) 9-18.
[163] T. Proell, R. Rauch, C. Aichering, H. Hofbauer, Coupling of biomass steam gasifi-

cation and SOFC - gas turbine hybrid system for highly efficient electricity generation, in:
Proceedings of the ASME Turbo Expo 2004: Power for Land, Sea and Air, Vienna, Aus-
tria, 2004, 14th to 17th June.
320
References
[164] E. Hernández-Pacheco, D. Singh, P. N. Hutton, N. Patel, M. D. Mann, J. Power

Sources, 138 (2004) 174-186.
[165] P. Costamagna et al., J. Power Sources, 96 (2001) 352-368.
[166] S. H. Chan, K. A. Khor, Z. T. Xia, J. Power Sources, 93 (2001) 130-140.
[167] T. Aloui, K. Halouani, Appl. Therm. Eng., 27 (2007) 731-737.
[168] K. D. Panopoulos, L. E. Fryda, J. Karl, S. Poulou, E. Kakaras, J. Power Sources,

159 (2006) 570-585.
[169] D. Froning, U. Reimer, A. Gubner, Combined Treatment of the Stack and the Sys-
tems Level in SOFC - Modelling, in: Proceedings of the 3rd Fuel Cell Research Sympo-
sium - Modelling and Experimental Validation, Dübendorf, Switzerland, 2006, 16th to
17th March.
[170] M. Roos, E. Batawi, U. Harnisch, T. Hocker, J. Power Sources, 118 (2003) 86-95.
[171] R. Bove, P. Lunghi, N. M. Sammes, Int. J. Hydrogen Energy, 30 (2005) 189-200.
[172] C. Stiller, B. Thorud, S. Seljebø, Ø. Mathisen, H. Karoliussen, O. Bolland, J. Power

Sources, 141 (2005) 227-240.
[173] T. Araki, T. Ohba, S. Takezawa, K. Onda, Y. Sakaki, J. Power Sources, 158 (2006)
52-59.
[174] P. Costamagna, A. Selimovic, M. Del Borghi, G. Agnew, Chem. Eng. J., 102 (2004)
61-69.
[175] E. Hernández-Pacheco, M. D. Mann, P. N. Hutton, D. Singh, M. K.E., Int. J. Hydro-

gen Energy, 30 (2005) 1221-1233.
[176] L. T. Lim, D. Chadwick, L. Kershenbaum, Ind. Eng. Chem. Res., 44 (2005) 9609-
9618.
[177] H. Timmermann, H. U., W. A., E. Ivers-Tiffée, R. Reimert, Development of a Fuel-

Flexible SOFC Stack Model, in: Proceedings of the 7th European SOFC Forum, Switzer-
land, Lucerne, 2006, 3rd to 7th July.
[178] R. Suwanwarangkul, E. Croiset, M. D. Pritzker, M. W. Fowler, P. L. Douglas, E.

Entchev, J. Power Sources, 154 (2006) 74-85.
[179] D. Larrain, J. Van Herle, F. Maréchal, D. Favrat, J. Power Sources, 131 (2004)
304-312.
321
References
[180] N. Autissier, D. Larrain, J. Van Herle, D. Favrat, J. Power Sources, 131 (2004) 313-
319.
[181] J. Van Herle, D. Larrain, N. Autissier, Z. Wuillemin, M. Molinelli, D. Favrat, J. Euro-

pean Ceram. Soc., 25 (2005) 2627-2632.
[182] P. Costamagna, K. Honegger, J. Electrochem. Soc., 145 (1998) 3995-4007.
[183] S. Cordiner, M. Feola, V. Mulone, F. Romanelli, Appl. Therm. Eng., 27 (2007) 738-
747.
[184] M. Sucipta, S. Kimijima, K. Suzuki, J. Power Sources, 174 (2007) 124-135.
[185] Website, http://www.athenavisual.com, June 2007.
[186] T. A. Milne, R. J. Evans, N. Abatzoglou, Biomass Gasifier "Tars": Their Nature,

Formation and Conversion, NREL, Golden, United States of America, 1998.
[187] VDI-Wärmeatlas. Berechnungsblätter für den Wärmeübergang, Verein Deutscher

Ingenieure. VDI-Gesellschaft Verfahrenstechnik und Chemieingenieurwesen, Düsseldorf,
Germany, 1977.
[188] F. Palazzi, N. Autissier, F. M. A. Marechal, D. Favrat, Appl. Therm. Eng., 27 (2007)

2703-2712.
[189] N. Autissier, F. Palazzi, F. Marechal, J. van Herle, D. Favrat, J. Fuel Cell Sci.
Technol., 4 (2007) 123-129.
[190] C. Rechenauer, E. Achenbach, Dreidimensionale mathematische Modellierung des

stationären und instationären Verhaltens oxidkeramischer Hochtemperatur-
Brennstoffzellen, RWTH Aachen, Jülich, Germany, 1992.
[191] U. Bossel, Facts & Figures, Berne, Switzerland, 1992.
[192] Website, http://dippr.byu.edu/, December 2007.
[193] F. A. Aly, L. L. Lee, Fluid Phase Equilib., 6 (1981) 169-179.
[194] B. E. Poling, J. M. Prausnitz, J. P. O'Connell, The properties of gases and liquids,

McGraw-Hill Professional, New York, United States of America, 2000.
[195] E. Schmidt, Einführung in die Technische Thermodynamik, Springer, Berlin, Göttin-

gen, Heidelberg, Germany, 1963.
[196] J. M. Bockris, S. Srinivasan, Fuel cells: Their Electrochemistry, McGraw-Hill, New

York, United States of America, 1969.
322
References
[197] N. Q. Minh, T. Takahashi, Science and technology of ceramic fuel cells, Elsevier
Science, Amsterdam, The Netherlands, 1995.
[198] K. Nisancioglu, Ohmic Losses, in: Proceedings of the Workshop on mathematical

modelling: "Natural gas fuelled solide oxide fuel cells and systems", Charmey, Switzer-
land, 1989, 2nd to 6th July.
[199] E. F. Sverdrup, C. J. Warde, R. L. Eback, Energy Convers., 13 (1973) 129-141.
[200] A. Solheim, K. Nisancioglu, Resistance and Current distribution in Fuel Cell Ele-
ments, in: Proceedings of the Second International Symposium on SOFC, Greece, Ath-
ens, 1991, 2nd to 5th July.
[201] N. P. Cheremisinoff, Handbook of Heat and Mass Transfer, Butterworth-

Heinemann, Houston-London-Paris, 1986.
[202] S. Campanari, P. Iora, J. Power Sources, 132 (2004) 113-126.
[203] S. Campanari, P. Iora, Fuel Cells, 5 (2005) 34-51.
[204] E. Achenbach, E. Riensche, J. Power Sources, 52 (1994) 283-288.
[205] K. Ahmed, K. Foger, Catal. Today, 63 (2000) 479-487.
[206] A. L. Lee, R. F. Zabransky, W. J. Huber, Ind. Eng. Chem. Res., 29 (1990) 766-773.
[207] I. Drescher, Kinetik der Methan-Dampf-Reformierung, Forschungszentrum Jülich,

Jülich, Germany, 1999.
[208] M. Pfafferodt, P. Heidebrecht, M. Stelter, K. Sundmacher, J. Power Sources, 149

(2005) 53-62.
[209] P. Costamagna, E. Arato, P. L. Antonucci, V. Antonucci, Chem. Eng. Sci., 51

(1996) 3013-3018.
[210] F. Lange, M. Ise, W. Kleinlein, B. Schricker, Delta 8 - A new cell design for cost op-
timized generators, in: Proceedings of the Fuel Cell Seminar & Exposition, San Antonio,
USA, 2007, 15th to 19th October.
[211] K. Huang, Development of Delta-Type SOFCs at Siemens Stationary Fuel Cells, in:
Proceedings of the Fuel Cell Seminar & Exposition, San Antonio, United States of Amer-
ica, 2007, 15th to 19th October.
[212] F. Marechal, F. Palazzi, J. Godat, D. Favrat, Fuel Cells, 5 (2005) 5-24.
323
References
[213] R. Turton, R. C. Bailie, W. B. Whiting, J. A. Shaeiwitz, Analysis, Synthesis and De-

sign of Chemical Processes, Pearson Education Inc., Prentice Hall PTR, Upper Saddle
River, United States of America, 2003.
[214] D. E. Garrett, Chemical Engineering Economics, Van Nostrand Reinhold, New York,
United States of America, 1989.
[215] A. V. Bridgwater, A. J. Toft, J. G. Brammer, Renew. Sust. Energ. Rev., 6 (2002)

181-248.
[216] V. Gnielinski, A. Mersmann, F. Thurner, Verdampfung, Kristallisation, Trocknung,

Friedr. Vieweg & Sohn Verlegsgesellschaft mbH, Braunschweig,Wiesbaden, Germany,
1993.
[217] Website, http://www.energychallenge.org/, January 2008.
[218] L. Schneider, Stromgestehungskosten von Grosskraftwerken, Öko-Institut e.V.,

Freiburg, Germany, 1998.
[219] Website, http://www.carmen-ev.de, January 2008.
[220] W. A. Surdoval, The U. S. Department of Energy Fossil Energy Fuel Cell Program -
Solid State Energy Conversion Alliance: Goals and Challenges, in: Proceedings of the
8th SECA Workshop, San Antonio, United States of America, 2007, 7th to 9th August.
[221] Website, http://www.netl.doe.gov/technologies/coalpower/fuelcells/seca.html,

January 2008.
[222] S. Jimenez, J. Ballester, Combustion Science And Technology, 178 (2006) 655-
683.
[223] M. Seemann, S. Biollaz, C. Aichernig, R. Rauch, H. Hofbauer, R. Koch, Methana-

tion of bio-syngas in a bench scale reactor using a slip stream of the FICFB gasifier in
Güssing, in: Proceedings of the 2nd World Conference on Biomass for Energy, Industry,
and Climate Protection, Rome, Italy, 2004, 10th to 14th May.
[224] M. C. Seemann, T. J. Schildhauer, S. M. A. Biollaz, S. Stucki, A. Wokaun, Appl.

Catal. A-Gen., 313 (2006) 14-21.
[225] R. E. Harth, U. Boltendahl, Interdiscip. Sci. Rev., 6 (1981) 221-228.
[226] C. Aichernig, H. Hofbauer, R. Rauch, R. Koch, RENET Report, (2006).
[227] Website, http://www.ctu.ch/front_content.php?idart=183, May 2008.
324
References
[228] J. Karl, Biocellus - Highly efficient SOFC systems with indirect gasification, in: Pro-
ceedings of the Bio-energy workshop EU-Russia, Moscow, Russia, 2004, 7th to 8th Oc-
tober.
[229] EEG, "Erneuerbare-Energien-Gesetz vom 21. Juli 2004 (BGBl. I S. 1918), zuletzt
geändert durch Artikel 1 des Gesetzes vom 7. November 2006 (BGBl. I S. 2550)".
325
326
Curriculum vitae
Personal Data
Surname Nagel
Given names Florian-Patrice
Date of birth 5th December 1977
Birthplace Berlin
Nationality German
Post-graduate experience
Researcher (LEM Laboratory for Energy and Material cy-

05/2004 – 06/2008
cles, PSI Paul Scherrer Institut, Villigen, Switzerland)
Assistant (Swiss Federal Institute of Technology (ETHZ),

10/2004 – 02/2006
Zurich, Switzerland)
Drive train development engineer (DaimlerChrysler Re-

10/2003 – 04/2004
search Center, Ulm, Germany)
Education
Doctorate studies (Swiss Federal Institute of Technology

05/2004 – 06/2008
(ETHZ), Zurich, Switzerland)
Diplom- Ingenieur studies (Department of Mechanical Engi-

10/1998 – 10/2003
neering, University of Stuttgart, Stuttgart, Germany)
Elementary and Grammar school (Hochrhein- Gymnasium,

08/1983 – 07/1996
Waldshut, Germany)
Under-graduate experience
01/2003 – 03/2003 Intern (LAE Lahmeyer Agua y Energia S.A., Lima, Peru)
12/2002 – 01/2003 Intern (Fundicion Las Rosas S.A., Malloco-Peñaflor, Chile)
11/2002 – 12/2002 Intern (EPS Ltda., Santiago de Chile, Chile)
12/2000 – 04/2002 Junior researcher (IPA Fraunhofergesellschaft, Stuttgart,

Germany)
06/1997 – 09/1998 Semi-skilled labor (AWW GmbH, Wutöschingen, Germany)
327
Publications
Conference pro- F. P. Nagel, M. Künstle, S.Biollaz: Link-up of a solid oxide fuel cell with a wood gasifier;
Proceedings of the 14th European Biomass Conference and Exhibition: Biomass for En-
ceedings th st
ergy, Industry and Climate Protection; Paris, France, 17 -21 October 2005
F. Nagel, M. Jenne, S. Biollaz, S. Stucki: Link-up of a SOFC with an Updraft Wood Gasifier
via Hot Gas Processing; Proceedings of the 2006 Fuel Cell Seminar: Fuel Cells in Para-
th th
dise - 30th Anniversary; Honolulu, HI, United States of America, 13 -17 November 2006
F. P. Nagel, S. Biollaz, M. Jenne, S. Stucki: Link-up of a 1 kW-SOFC with an updraft-wood

gasifier via hot gas processing; Proceedings of the 15th European Biomass Conference
th th
and Exhibition: From Research to Market Deployment; Berlin, Germany, 7 -11 May 2007
F. P. Nagel, S. Biollaz, M. Jenne, A. Schuler: Long-term tests of a complete biomass inte-

grated gasification fuel cell system (B-IGFC); Proceedings of the 16th European Biomass
nd
Conference and Exhibition: From Research to Industry and Markets, Valencia, Spain, 2 -
th
6 June 2008
S. Biollaz, F.P. Nagel, K. Dennerlein, H. Landes, W. Straub, J. Karl: Results from long term
testing of tubular SOFC with woodgas; Proceedings of the 16th European Biomass Con-
nd th
ference and Exhibition: From Research to Industry and Markets, Valencia, Spain, 2 -6
June 2008
S. Biollaz, F. P. Nagel, M. Jenne, A. Schuler: Long-term tests of a complete biomass inte-

grated gasification fuel cell system (B-IGFC); Proceedings of the 8th European SOFC
th th
Forum, Lucerne, Switzerland, 30 June-4 July 2008
Oral presentations "Link-up of a SOFC with an updraft-wood gasifier via hot gas processing (B-IGFC)" held at
th
the 2006 Fuel Cell Seminar in Honolulu, HI, USA, the 16 November 2006
"Link-up of a 1kW-SOFC with an updraft-wood gasifier via hot gas processing" held at the
th
15th European Biomass Conference and Exhibition in Berlin, Germany,the 8 May 2007
"Verstromung von Holz über die Hochtemperaturbrennstoffzelle" held at the Statusseminar

2008 of the Schweizerische Arbeitsgemeinschaft für Holzforschung (SAH) in Dübendorf,
th
Switzerland, the 19 March 2008
Scientific papers F.P. Nagel, T.J. Schildhauer, S.M.A. Biollaz, A. Wokaun: Performance comparison of pla-
nar, tubular and Delta8 solid oxide fuel cells using a generalized finite volume model; Jour-
(peer-reviewed) nal of Power Sources 184 (2008) 143–164
F.P. Nagel, T.J. Schildhauer, S.M.A. Biollaz, S. Stucki: Charge, mass and heat transfer
interactions in SOFCs operated with different fuel gases - A Sensitivity analysis; Journal of
Power Sources 184 (2008) 129–142
F.P. Nagel, T.J. Schildhauer, S.M.A. Biollaz, A.Schuler: The impact of sulfur on the per-
formance of a SOFC system operated with hydrocarboneous fuel gas; Journal of Power
Sources (2009), doi:10.1016/j.jpowsour.2008.12.092
F.P. Nagel, S. Ghosh, C. Pitta, T.J. Schildhauer, S.M.A. Biollaz: Biomass integrated gasifi-
cation fuel cell systems - Concept development based on experimental results and long-
term testing; Biomass and Bioenergy, in internal review process
F.P. Nagel, T.J. Schildhauer, S.M.A. Biollaz: Biomass-Integrated gasification fuel cell sys-
tems - Part 1: Definition of systems and technical analysis; International Journal of Hydro-
gen Energy, submitted
F.P. Nagel, T.J. Schildhauer, N. McCaughey, S.M.A. Biollaz: Biomass-Integrated gasifica-

tion fuel cell systems - Part 2: Economic analysis; International Journal of Hydrogen En-
ergy, submitted
F.P. Nagel, T.J. Schildhauer: Improving lumped SOFC models for system simulations by
parameterization of detailed model results; Journal of Power Sources, in preparation
Patent applications F.P. Nagel, T.J. Schildhauer, M. Jenne, S. Biollaz, Verfahren und Anlage zur Verstromung
fester Biomasse, WO 2008/055591 A2
328

Biomass Gasification and Solid Oxide Fuel Cell

Hochgeladen von

Copyright:

Verfügbare Formate

Biomass Gasification and Solid Oxide Fuel Cell

Hochgeladen von

Dokumentinformationen

Originaltitel

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Biomass Gasification and Solid Oxide Fuel Cell

Hochgeladen von

Copyright:

Verfügbare Formate

DISS. ETH NO.

Electricity from wood through the combination

for the degree of

accepted on the recommendation of

Villigen, im Juli 2008

Angesichts der wachsenden Weltbevölkerung, zahlreicher wirtschaftlich aufsteigernder

The on-call availability of energy was a prerequisite for the development

The gasification of wood instead of its combustion allows the application

The knowledge gained from the above mentioned experimental investi-

1.4 Organization of the thesis

experiments leading to the first-time long-term demonstration experi-

The combination of biomass gasification with SOFCs is a highly promis-

Figure 2-1: Biomass Gasification

In a first step the solid biomass is converted to a combustible gas mix-

 Long term growth (15-50 year); Short term single stem

 Herbaceous plants and grasses, i.e. switchgrass

 High sugar-/ starch-producing crops, i.e. sugarcane

 Oil-producing plants, i.e. rapeseed

 Aquatic plants, i.e. algae

The required energy is produced from light via photosynthesis based on

Cellulose, hemicellulose and lignin represent approx. 95-98 mass-%

Table 2-1: Dry fraction Eucalyptus

trinsic moisture content, also referred to as water load, is of greater in-

In contrast to the moisture content, the humidity is based on the wet

The moisture content can be converted to a humidity value via Eq. 4.

2.1.4.2 Proximate and ultimate analysis

The ash content is determined by heating a fuel sample up to 600 °C in

Table 2-2: M XVM XFC C H O N S Cl K Na Ash

Figure 2-5: 1.8

Atomic H/C Ratio

2.1.4.3 Caloric value

 The quantity of heat which is obtained by bringing all the products of

Figure 2-6: Thermochemical Physical / Chemical Biochemical

Figure 2-6 gives an overview of conversion technologies for heat and

2.2.2 Fundamentals of thermochemical conversion

All thermochemical conversion technologies involve all or some of the

Another important reaction in the reduction sub-process is the exo-

Table 2-3: Temperature Air-to-fuel ratio Product

The technical pyrolysis process comprises the sub-processes "drying",

peratures between 400 and 700 °C without addition of oxygen or steam,

2.2.3 Wood gasification technology overview

up to temperatures around 400 °C, e.g. the boiling point of benzan-

Figure 2-8: Fuel Fuel in fixed bed

Updraft Downdraft Stationary Circulating Entrained

both downwards in the gasifier running through the sub-processes of

Figure 2-9: Fuel

it is immediately mixed with a bed material such as sand. The portion of

The "Fast internally circulating fluidized bed" (FICFB) reactor depicted

duction" occur in the actual entrained flow gasification reactor. As the

Table 2-4: Down- Fluid. Circul. Entrain.

2.3 Gas processing

2.3.1 Gas cleaning

Electrostatic precipitators consist of two zones. In the first zone, the

Particle Raw gas

As a matter of principle, cyclones are relatively simple, robust and cheap

material of construction. Especially for the application of high tempera-

Hence they are predestined to the combination of biomass gasification

2.3.1.2 Alkali metals

plete tar removal to avoid condensation problems, while systems using

Long term growth (15-50 year); Short term single stem

Herbaceous plants and grasses, i.e. switchgrass

High sugar-/ starch-producing crops, i.e. sugarcane

Oil-producing plants, i.e. rapeseed

Aquatic plants, i.e. algae

The quantity of heat which is obtained by bringing all the products of

Geometry of the current paths

Current density and hence the electron concentration in the

Product (H2O, CO2) and educt (H2, CO) partial pressures

Size of the TPB

Formation of low conductive layers between the cathode and the