Communications

60
50

PD

PD

AcH

AcH

CO

CO

40

methanol as well as hydrocarbons. Also oligomerization

products could be detected. The highest selectivity to the
desired product 1,2-propandiol (45 % at approx. 40 %
conversion of propylene) was found at the lowest temperature
setting used (300 C).

S in % 30

Acknowledgement

20
10
0
0

10

20

30

40

50

60

X (C 3H6) in %

The authors express their thanks to BASF AG Ludwigshafen and the German Federal Ministry of Education,
Science, Research and Technology (BMBF) for the financial
support.

Figure 5. Selectivity-conversion-diagram for the parameter area: T = 300475 C;

p = 20 35 MPa; s = 0.8 7.2 s and x(C3H6)0/x(O2)0 = 1 32; open symbols:
subcritical conditions, closed symbols: supercritical conditions.

propylene > 45 % leads to losses in the selectivity of

1,2-propandiol. The best space-time yield of 1,2-propandiol
gained is with 36 mol h1 L1 in the range of the industrial
production (hydrolysis of propylene oxide) [10,11].
3.4 Activation Energy and Volume
In the temperature range 300 to 475 C the activation energy
is not constant. A positive value (approx. +20 kJ mol1) is
found below the critical temperature of water, a negative
above (approx. 30 kJ mol1). Above 425 C a normal
Arrhenius behavior is obtained. The values of the activation
energies calculated for the separate intervals are very low. The
broken partial reaction orders for propylene (a = 0.2) and
oxygen (b = 0.3) indicate a complex reaction mechanism [12].
The absolute amount of the activation volume at
incomplete oxygen conversion is considerably higher
(DV = 2540 cm3 mol1) than at complete conversion
(DV = 687 cm3 mol1). An acceleration of the reaction
by increasing pressure is therefore primarily possible at a high
oxygen feed concentration.

4 Summary
The partial oxidation of propylene was investigated in suband supercritical water with the help of an especially
developed continuous high-pressure plant. A differential loop
reactor was used in the form of a jet loop reactor. The test
parameters extended from 300 to 475 C, 20 to 35 MPa,
propylene feed concentration from 0.5 to 4 % (mol mol1) and
oxygen feed concentration from 0.2 to 2 % (mol mol1).
Depending on the other parameters the residence time ranged
between 0.8 and 7.2 s.
Due to the high reactivity of propylene (double bond and
three acidic hydrogen atoms) a multitude of products are
formed. The dosed oxygen is mostly converted completely.
The main products are 1,2-propandiol, acetaldehyde and CO.
The most important by-products are formaldehyde, CO2,
428

2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Received: November 27, 2002 [K2925]

References
[1] P. E. Savage, S. T. Gopalan, T. I. Mizan, C. J. Martino, E. E. Brock,
AIChE J. 1995, 41, 1723.
[2] D. Brll, C. Kaul, A. Krmer, P. Krammer, T. Richter, M. Jung, H. Vogel,
P. Zehner, Angew. Chem., Int. Ed. 1999, 111, 2998; Angew. Chem. 1999,
111, 3180.
[3] K. Weissermel, H.-J. Arpe, Industrial Organic Chemistry, 4th ed., VCH,
Weinheim 1994.
[4] H. Blenke, K. Bohner, S. Schuster, Chem. Ing. Tech. 1965, 37, 289.
[5] D. Brll, A. Krmer, I. Lappas, H. Fue, H. Vogel, Chem. Eng. Technol.
2000, 24, 142; Chem. Ing. Tech. 2000, 72, 377.
[6] I. L. C. Freriks, R. Bouwman, P. V. Geenen, J. Catal. 1980, 65, 311.
[7] W. M. H. Sachtler, Cat. Rev. 1970, 4, 27.
[8] D. Brll, Partialoxidationen in berkritischem Wasser mit molekularem
Sauerstoff Die Reaktionen von Methanol, Methan und Propylen mit
und ohne Silberkatalysatoren, Ph.D. Thesis, Darmstadt University of
Technology 2001.
[9] A. Krmer, Heterogen katalysierte Partialoxidationen in berkritischem
Wasser Die selektive Oxidation von Methanol, Methan und Propen an
Kupfer und Kupferlegierungen, Ph.D. Thesis, Darmstadt University of
Technology 2001.
[10] P. Kripylo, G. Landschtz, W. Pritzkow, Aufgaben zur Technischen
Chemie, 1st ed., VCH, Weinheim 1978.
[11] F. M. Dautzenberg, ACS Symp. Ser. 1989, 411, 99.
[12] M. Akimoto, K. Ichikawa, E. Echigoya, J. Catal. 1982, 76, 333.
This paper was also published in German in Chem. Ing. Tech. 2002, 74 (12), 81.

_______________________

Determination of New Group

Contributions to Calculate Limiting
Oxygen Concentrations
By Oliver Fu*, Maria Molnarne, Volkmar Schrder,
and Axel Schnbucher
A few years ago Subramaniam and Cangelosi developed a
group contribution method to predict limiting oxygen
concentration of combustible gases. The calculated values

[*]

Dipl.-Ing. O. Fu, Dr.-Ing. M. Molnarne, Dr. V. Schrder, Bundesanstalt

fr Materialforschung und -prfung, Abt. II: Chemische Sicherheitstechnik, D-12200 Berlin, Germany; Prof. Dr. A. Schnbucher, Universitt
Essen, Institut fr Technische Chemie I, Universittsstr. 5, D-45117 Essen,
Germany.

0930-7516/03/0404-0428
$ 17.50+.50/0
0930-7516/03/0404-0428
$ 17.50+.50/0

Chem. Eng. Technol. 26 (2003) 4

Communications
are compared to the recommended, experimental ones taken
from the database CHEMSAFE. The experimental data used
has been measured according to the German standard
DIN 51649, while the calculation method was developed on
the basis of experiments that had been carried out in the
Bureau of Mines. The evaluation of the results showed large
deviations between the measured and the estimated limiting
oxygen concentrations. Using the calculated values can yield a
mixture lying in the explosion area. The calculation method is
frequently applied in the chemical industry. This made it
necessary to define new group contribution values based on
the newly measured data material.

1 Introduction

Figure 1. Triangular diagram of fuel gas/inert gas/air mixture with the value of
the limiting air concentration (LAC).

A simple way to avoid explosive gas mixtures in chemical

industrial processes is the monitoring of oxygen content in the
respective gas mixture. The limiting oxygen concentrations
(LOC) of frequently used combustible gases and vapors in
mixtures with different inert gases, air and oxygen were
determined under atmospheric conditions. If the oxygen mole
fraction of a gas mixture is below the LOC value, even the
arbitrary addition of fuel gas cannot lead to explosive gas
mixtures.
Since 1997, the European standard EN 11271 [1] gives a
definition for the limiting oxygen concentration. Furthermore, a new European standard for the determination of
limiting oxygen concentrations of gases and vapors is being
developed by CEN/TC305/WG1.
In 1989, Subramaniam and Cangelosi published a group
contribution method for calculating the limiting oxygen
concentration of different fuel gas/nitrogen/oxygen mixtures
at ambient temperature and atmospheric pressure [2]. The
group contributions based on experimental data of flammability limits were measured by the Bureau of Mines and
collected by Zabetakis [3]. Recently measured, recommended
values of LOC for ambient temperatures and atmospheric
pressure are available in the database CHEMSAFE [4].
The purpose of this article is to compare the calculated
values with experimental ones taken from the database
CHEMSAFE. In case of deviations, the calculation method
should be adapted through determination of new group
contribution values based on the newer experimental data in
CHEMSAFE.

2 Basics
2.1 Limiting Oxygen Concentration
The limiting oxygen concentration is defined as the
maximum oxygen concentration in a mixture of a combustible,
air and inert gas in which an explosion will not occur,
determined under specified test conditions (see Fig. 1 for
illustration).
Chem. Eng. Technol. 26 (2003) 4,

The distance between upper and lower explosion limit

becomes smaller as the inert gas fraction increases. The region
of the maximum inert gas fraction is designated as the apex of
the explosion range.
The limiting oxygen concentration can be determined
graphically by drawing a tangent to the boundary curve of
the explosion range which is parallel to the inert axis. The
limiting air concentration (LAC) can be taken from the
intersection point of the parallel line and the air axis. In this
example the LAC value is approx. 41 mole percent. The
determined value must be converted to pure oxygen to receive
the LOC value according to the following equation1):
LOC = 0.209 LAC

(1)

Safety characteristics, such as limiting oxygen concentrations, are not intrinsic material properties. These values
depend on the test methods and apparatus parameters. Beside
the geometry of the ignition vessel, also the ignition energy
and the ignition source have an influence on LOC. Furthermore, LOC strongly depends on initial pressure and initial
temperature.
In the Anglo-Saxon literature different expressions can be
found for the limiting oxygen concentration. The expression
limiting oxygen concentration is useable even for dusts. For
a clearer definition of gas systems the term maximum oxygen
concentration (MOC) is used in CHEMSAFE. In spite of
that LOC is employed exclusively in this article, as the term
customary in CEN. In the Anglo-Saxon literature the
following terms are usual as well: maximum allowable
oxygen concentration, maximum safe percentage oxygen
concentration and minimum oxygen concentration.

1)

List of symbols at the end of the paper.

2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

0930-7516/03/0404-0429 $ 17.50+.50/0

429

Communications
2.2 Group Contribution Method of Subramaniam and
Cangelosi
The published calculation method [2] can be used for
different flammable gases and vapors (mentioned as fuel gases
in the following). In this case the inert gas is prescribed on
nitrogen and the oxidizer on oxygen. Furthermore, the
computed values are valid only for ambient temperature and
atmospheric pressure. The calculation occurs in three individual steps.
At first the stoichiometric number of oxygen moles required
per mole of the chosen fuel gas is determined. Basis of this
procedure is the general sum formula CcHhOoCldNnSs of a
fuel gas. The stoichiometric oxygen consumption (Os) can be
calculated by the following equation (2):
OS c n s

h a
4

o
2

(2)

The coefficient a shows the number of chlorine atoms in

relation to the number of hydrogen atoms (a = f(d,h)). If h is
less than d and no H2O is formed during reaction, then a is
equal to h. In case h is greater than d and no Cl2 is formed
during combustion, the coefficient a is equal to the number of
chlorine atoms. In case h is equal to d and only HCl and neither
Cl2 nor H2O is formed, a is equal to d as well.
In the second step the fuel gas molecule is divided into the
groups defined in the appendix of [2]. A numerical value is
assigned to every increment. The single values are summed up
and form the entire nitrogen requirement called NT.
Finally, the limiting oxygen concentration is calculated (in
[2] designated as Omax%) with Eq. (3):
Omax%

100 OS
1OS NT

(3)

It is indicated that the deviations expected by this method

are less than 2.5 % for hydrocarbon gases and liquids. The
maximum error is supposed to be smaller than 5 %.

3 Experimental
The test apparatus, described in the German standard DIN
51649 [5], can be applied for temperatures up to 200 C at
atmospheric pressure. The ignition vessel consists of a glass
tube with an inner diameter of 60 mm and a height of 300 mm.
The gas mixture is filled in at the bottom and let out at the top.
The ignition source is an induction spark between two
electrodes with a distance of 5 mm. The electrodes are located
60 mm above the bottom of the test vessel. For the generation
of the spark a high voltage of about 15 kV is necessary. The
spark duration is 0.5 s.
The ignition criterion is the visually observed flame
propagation from the electrodes. The step size for determining
the explosion limit is 0.1 mole percent, if the total fuel gas
amount is equal or less than 10 mol-%. If the total fuel gas
amount is greater than 10 mol-%, the step size is 0.2 mol-%.
430

2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

The explosion limit is defined as the fuel gas amount, at which

in the case of five experiments no flame propagation can be
observed.
The test vessel used in the Bureau of Mines is a 1500 mm
glass tube with an inner diameter of 50 mm. The glass tube is
closed at the bottom through a glass plate and can be
evacuated by a vacuum pump. Then, the gas mixture to be
tested is filled in and is mixed by circulation several minutes.
After mixing the glass plate is removed. The ignition occurs
directly after removing the plate either with a spark channel or
with a flame passed by at the now open, lower end of the tube.
Informations about the duration of ignition are not given. The
explosion limit is defined as the boundary line between the
flammable and non-flammable gas mixture. Information
about the step size between these points is not given.
The direction of flame propagation in both experimental
setups occurs in upward direction and thus in the same
direction as the convection of the hot gases arising from the
combustion. If the ignition would be introduced at the upper
end of the vessels, the flame had to expand downwards and
thus opposite to the convection of the hot gases. The result
would be a smaller explosion range, because for fulfilling the
explosion criterion the flame velocity must be higher than the
convection velocity, so that a flame can propagate.

4 Results and Discussion

In Tab. 1 the values for the limiting oxygen concentration
taken from CHEMSAFE as well as the values computed by
the group contribution method are shown. The relative
deviation of these values compared to the CHEMSAFE
values is shown in the last column. It was computed according
to the equation:
rel: deviation


LOCgroup contribution method LOCCHEMSAFE
LOCCHEMSAFE

100 (4)

A significant deviation of the predicted values from the

experimental ones according to DIN 51649 in CHEMSAFE
was found. The deviations for alkanes are between 15 % and
30 %. Alkenes show similar discrepancies between 12 % and
31 %. For the group of the alkines, the limiting oxygen
concentration is only available in CHEMSAFE for ethine.
The computed value is very close to the experimental one. For
the alcohols represented in the database deviations of 20 % up
to 38 % were observed. Also for the commercially important
gases hydrogen and carbon monoxide the deviations are
higher than 10 %.
The deviations of the rarely used gases 1,3-butadiene and
cyclopropane are below 5 %. The recommended LOC values
of both gases in the database CHEMSAFE are taken from
Zabetakis. The small deviations become understandable,
because Zabetakis values were used for the development of
0930-7516/03/0404-0430 $ 17.50+.50/0

Chem. Eng. Technol. 26 (2003) 4

Communications
Table 1. Comparison of the group contribution method of Subramaniam and Cangelosi with
recommended values taken from CHEMSAFE (version 1.4).
Sum
formula

LOC
(group contribution
method) [2]
[mol.-%]

LOC
(CHEMSAFE)
version 1.4 [4]
[mol.-%]

Methane

CH4

12.0

9.9

+21.2

Ethane

C2H6

11.4

8.8

+29.5

Propane

C3H8

11.5

9.8

+17.3

i-Butane

C4H10

12.2

10.3

+18.4

n-Butane

C4H10

11.6

9.6

+20.8

Hexane

C6H14

11.7

9.3

+25.8

Ethene

C2H4

10.0

7.6

+31.6

Propene

C3H6

11.0

9.3

+18.3

i-Butene

C4H8

12.0

10.7

+12.1

n-Butene

C4H8

11.2

9.8

+14.3

C2H2

6.4

6.3

+1.6

Ethanol

C2H5OH

10.8

8.6

+25.6

1-Propanol

C3H7OH

11.1

9.3

+19.6

2-Propanol

C3H7OH

12.0

8.7

+37.9

H2

5.0

4.3

+16.3

Substance name

Rel. Deviation
of the group
contribution method
[%]

Alkanes

Alkenes

Alkines
Ethine
Alcohols

Other Substances
Hydrogen
Carbon monoxide

CO

5.5

6.2

11.3

1,3-Butadiene

C4H6

10.8

10.4

+3.8

Cyclopropane

C3H6

11.6

11.7

0.9

Ethyl acetate

C4H8O2

11.2

9.8

+14.3

2-Butanone

C4H8O

11.4

9.5

+20.0

the Subramaniam and Cangelosi method. The values for

i-butane and i-butene are taken from Stickling [6], in which
limiting oxygen concentrations of alkanes and alkenes were
determined. The deviations of the computed LOC values are
about 11 % to 30 % higher than the experimental values of
Stickling.
In nearly all cases the estimated values are higher than the
experimental ones and thus can lead to accidental fires or
explosions. Only for carbon monoxide and cyclopropane the
estimated values are lower than the experimental ones.
The above-mentioned calculation method is frequently
used in the chemical industry and the applicability of the
method was tested within the framework of the European
standardization of the determination of limiting oxygen
concentrations as well. Therefore, it was decided that the
group contribution values of some compounds should be
redetermined from the recommended LOC values already
available in CHEMSAFE. The new values for the group
contributions are listed in Tab. 2.

Chem. Eng. Technol. 26 (2003) 4,

Table 2. New determined group contribution values in comparison to the values

published by Subramaniam and Cangelosi.
Group

Old value

New value
17.636

CH3

13.1169

CH2

11.1453

9.75

CH

6.31407

5.2

CH2=

13

17.756

CH=

9.11694

7.5

C=

3.76617

3.955

HC

17.7812

18.092

OH

0.426518

5.75

O=C

4.36966

6.4

O=CO

1.08435

13.67

17.202

Other substances
CH4
H2

8.5

10.127

CO

7.59

6.564

2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

0930-7516/03/0404-0431 $ 17.50+.50/0

431

Communications
The new calculated results for the limiting oxygen concentration are listed in Tab. 3. The relative deviations of those
gases used to develop the new group contribution values are
naturally very small, because for some specific structure
groups only a few LOC values were available. Furthermore,
measured values of the limiting oxygen concentration of fuel
gas mixtures cannot be found. Therefore, the applicability of
the method for mixtures could not be estimated at the present
state.
It can be seen in Tab. 3 that the computed values which were
not included in the determination of the new group contributions, show mostly small deviations from the values in
CHEMSAFE. But with increasing number of carbon atoms
in alkane molecules the prediction of the limiting oxygen
concentration leads to values which are located on the
dangerous side.

5 Conclusions
The great deviations of the computed LOC values from
Subramaniam and Cangelosi to the newer experimental
values in CHEMSAFE can presumably be explained by the
different test conditions. The experiments have been carried
out in completely different test vessels. For example, the
diameter of the tube according to DIN 51649 is wider and the
ignition sources are different.
Furthermore, a possible reason should be the different
definition of the explosion limits. In the German standard the
limit is defined as the concentration point at which flame
propagation just does not occur. For the development of the
group contribution values Subramaniam and Cangelosi used
experimental values, where the explosion limit was defined as
mean value between the reactive and the nonreactive gas

Table 3. Comparison of LOC values calculated with the new group contribution values and recommended data
taken from the database CHEMSAFE.
Substance name

Sum formula

Calculated LOC
(with new group
contribution values)
[mol-%]

LOC
(CHEMSAFE)
version 1.4 [4]
[mol-%]

Rel. deviation
of the group
contribution method
[%]

Note

Methane

CH4

9.9

9.9

0.0

Ethane

C2H6

8.8

8.8

0.0

Propane

C3H8

9.8

9.8

0.0

i-Butane

C4H10

9.9

10.3

3.8

(1)

n-Butane

C4H10

10.4

9.6

+8.7

(1)

n-Hexane

C6H14

11.2

9.3

+20.5

(1)

Ethene

C2H4

7.6

7.6

0.0

1-Propene

C3H6

9.3

9.3

0.0

i-Butene

C4H8

10.7

10.7

0.0

n-Butene

C4H8

10.0

9.8

+2.6

C2H2

6.3

6.3

0.0

Alkanes

Alkenes

(1)

Alkines
Ethine
Alcohols
Ethanol

C2H5OH

8.0

8.6

6.0

1-Propanol

C3H7OH

9.3

9.3

0.0

2-Propanol

C3H7OH

8.7

8.7

0.0

Hydrogen

H2

4.3

4.3

0.0

Carbon monoxide

CO

6.2

6.2

0.0

1,3-Butadiene

C4H6

9.6

10.4

7.2

(1), (2)

Cyclopropane

C3H6

12.9

11.7

+10.6

(1), (2)

Ethyl acetate

C4H8O2

9.8

9.8

0.0

2-Butanone

C4H8O

9.5

9.5

0.0

Other substances

(1) This substance was not used to determine the new group contribution values. The given LOC-value is a
prediction.
(2) The measured limiting oxygen concentration is taken from [3] and is not determined according to the German
standard DIN 51649.

432

2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

0930-7516/03/0404-0432 $ 17.50+.50/0

Chem. Eng. Technol. 26 (2003) 4

Communications
mixture. Therefore, possible deviations depend on the step
size.
The presented group contribution method is a useful tool to
predict limiting oxygen concentrations. The calculated values
achieved by the original method of Subramaniam and
Cangelosi compared to the experimental values according to
DIN 51649 however show high deviations. Furthermore, these
deviations lie on the unsafe side of the explosion range.
The purpose of the introduction of new group contribution
values was at least a better representation of the values
contained in CHEMSAFE. The predicted values correspond
well to the measured values. In order to achieve an extensive
validation of the method, more experimental LOC values
should be available.
Received: August 22, 2002, [K 2976]

Symbols used
LOC,
MOC,
Omax% [mol.-%]
LAC
[mol.-%]
Os
[]
NT

[]

[]

[]

[]

[]

[]

[]

[]

limiting oxygen concentration

limiting air concentration
stoichiometric number of oxygen moles
required per mole of combustible gas
total nitrogen required per mole of
combustible gas
number of chlorine atoms in relation
to the number of hydrogen atoms
number of the carbon atoms
in the fuel gas
number of the chlorine atoms
in the fuel gas
number of the hydrogen atoms
in the fuel gas
number of the nitrogen atoms
in the fuel gas
number of the oxygen atoms
in the fuel gas
number of the sulfur atoms
in the fuel gas

References
[1]
[2]
[3]
[4]
[5]
[6]

European Standard EN 11271: Explosive Atmospheres Explosion

Prevention and Protection Part 1: Basic Concepts and Methodology,
CEN, Brussels 1997, 6.
T. K. Subramaniam, J. V. Cangelosi, Chem. Eng. 1989, 108.
M. G. Zabetakis, Flammability Characteristics of Combustible Gases and
Vapors, Bureau of Mines Bulletin 627, U.S. Government Printing Office,
Washington D.C. 1965.
CHEMSAFE Database for Rated Safety Characteristics, Database
inhouse version 1.4, BAM, PTB, DECHEMA, Frankfurt/M. 2001.
DIN 51649 Teil 1: Bestimmung der Explosionsgrenzen von Gasen und
Gasgemischen in Luft, Beuth Verlag GmbH, Berlin 1987.
J. Stickling, Experimentelle und theoretische Bestimmung des Inertgaseinflusses auf die Explosionsgrenzen organischer Verbindungen, Abschlussbericht, Forschungsvorhaben AiF 10144, Paderborn 1997.

This paper was also published in German in Chem. Ing. Tech. 2002, 74 (6), 827.
Chem. Eng. Technol. 26 (2003) 4,

Dosing Pumps with Innovative Drive

Concept
By Bernd Eggert*

1 Introduction
Displacement pumps and particularly diaphragm pumps are
typically used for dosing in the low-pressure range up to
approximately 20 bar. However, the design and drive of
conventional, mechanically driven diaphragm pumps, results
in relatively narrow limits to their performance, in terms of
dosing and setting accuracy which affects the supply of
chemicals (see Fig. 1). In addition, solenoid-driven dosing
pumps produce due to their design strong pulsations, which
reduce the lifetime of both the pump and adjacent system
components. The integration of the pumps into central process
management systems is another aspect that requires improvement.

Figure 1. Unexpected operating cost due to inaccurate dosing.

2 Design of Dosing Pumps

In dosing applications, where dosing accuracy is usually the
most important requirement, the preferred pump type is a
displacement pump, e.g. diaphragm pump, piston pump and
piston diaphragm pump). In the low-pressure range up to
approximately 20 bar, mechanically actuated diaphragm
pumps are the most economical. The advantage over piston
pumps lies in their simple and leakage-free design, i.e. given
normal operating conditions there is no risk of contaminating
the environment with the dosing fluid. The high-pressure
range is the field of application for hydraulically driven piston
diaphragm pumps and process pumps. In applications with

[*]

Dipl.-Ing. B. Eggert, Product Manager Dosing Dumps, Grundfos GmbH,

D-23812 Wahlstedt, Germany.

2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

0930-7516/03/0404-0433
$ 17.50+.50/0
0930-7516/03/0404-0433
$ 17.50+.50/0

433