Heavy-Duty Diesel Engine NO Reduction With Nitrogen-Enriched Combustion Air

Final Report:
Heavy-duty Diesel Engine NOX Reduction

with Nitrogen-Enriched Combustion Air
Contract # 02-VTCE-GS-009
As a CRADA Program with Mack Truck
Principal Investigator:
Steve McConnell
Research Engineer
Center for Transportation Research
Argonne National Laboratory
DOE monitor Gurpreet Singh

Program Office OHVT
Executive Summary
The concept of engine emissions control by modifying intake combustion gas
composition from that of ambient air using gas separation membranes has been
developed during several programs undertaken at Argonne. These have led to
the current program which is targeted at heavy-duty diesel truck engines. The
specific objective is reduction of NOx emissions by the target engine to meet
anticipated 2007 standards while extracting a maximum of 5 percent power loss
and allowing implementation within commercial constraints of size, weight, and
cost. This report includes a brief review of related past programs, describes work
completed to date during the current program, and presents interim conclusions.
Following a work schedule adjustment in August 2002 to accommodate problems
in module procurement and data analysis, activities are now on schedule and
planned work is expected to be completed in September, 2004. Currently, we
believe that the stated program requirements for the target engine can be met,
based upon extrapolation of the work completed. Planned project work is
designed to experimentally confirm these projections and result in a specification
for a module package that will meet program objectives.
Acknowledgements
The project: Heavy-duty Diesel Engine NOX Reduction with Nitrogen-Enriched
Combustion Air was sponsored by:
Program Office (OHVT)
DOE monitor(s) Gurpreet Singh
As a CRADA Program with Mack Truck
Contract # 02-VTCE-GS-009
Introduction
Previous programs investigating engine intake composition at Argonne have led
to identification of diesel engine operating parameters required to meet
particulate emissions for the target engine. Initially, oxygen enrichment was
found to provide some solutions. Increased oxygen in the engine intake gave
good reduction in unburned hydrocarbons and CO emissions at startup and
during the cold-phase (bag 1) of the FTP cycle. Unfortunately nitrogen
compounds formed with this oxygen-rich operation increased.
For example, a gasoline-fueled engine running with a 2% oxygen increase
showed a 19% drop in CO emissions during the first bag of the FTP. A 4%
increase in oxygen level yielded a 40% decrease in CO emissions during the first
bag of the FTP. The oxygen enrichment increased the NOx emissions 57% and
500% increase in NOx respectively. When applied to diesel engines, the same/
similar oxygen enrichment of the intake gas would result in lower CO, HC and
PM emissions under a standard FTP protocol. However, NOx emissions
remained at unacceptable levels for anticipated year 2007 standards of .1
gram/bhp-hr.
Utilizing an engine intake gas with reduced oxygen (enriched nitrogen) was found
to result in oxide of nitrogen emissions reductions for gasoline engines and for
diesel engines, substantial emission reduction was observed. Using an intake
gas of 82 % nitrogen, balance oxygen, over the diesel engine test protocol, tests
showed NOx emissions were reduced by up to 80%. Further tests showed when
using a practical range of nitrogen enrichment compositions, particulate
emissions in diesel engines could be controlled to desired levels by adjusting the
nitrogen composition. Among the various technologies available to produce
nitrogen enriched air, membrane separation was deemed the most likely to be
able to meet the constraints of commercial use with the target engines.
From this work, a set of requirements for implementation of a nitrogen-enriched
intake package were established in consultation with a manufacturer of the target
engines, Mack Trucks. These requirements became the basis of a CRADA
established September 2000, between Mack Trucks and DOE. Argonne was
selected to implement the program in conjunction with Mack Trucks and the
current program began October 2001.
Background
The idea of varying combustion air composition using oxygen enrichment
originated in late Sixties. In the late 80s and early 90s, Funding for research
into this technology at ANL had been modest. Initial support came from DOE/
Office of Industrial Technologies from 1989-1992. This study investigated
oxygen enrichment in industrial cogeneration. The next project was funded by
the Chicago Transit Authority. This study investigated smoke and particulate
reduction in city buses using oxygen enriching gas separation membranes. Then
the National Renewable Energy Laboratory (NREL) funded Light-duty Vehicle
Tests using a Chevy Lumina and oxygen enriching membranes.
A follow on to the NREL oxygen enrichment study was an oxygen enrichment
study of a M-85 (85% methanol) fueled Dodge Spirit which was funded by the
DOEs Alternative Fuels Office. This study proved that air toxics including
aldehides common in alcohol fueled vehicles can be significantly reduced with
oxygen enrichment. This study also showed that considerable reductions in
carbon monoxide and hydrocarbon emissions during cold start could be achieved
using oxygen enrichment.
From 1994 to 1999, ANL studied the engine out emissions benefit of varying the
intake air supplied to locomotive diesel engines. In 1999, nitrogen and oxygen
enrichment was applied to the CIDI program funded by DOE.
The CIDI program involved testing a Volkswagen 1.9L TDI engine with oxygen
and nitrogen enrichment. This study showed that significant improvements in
particulate reduction can be achieved using oxygen enrichment. These
reductions in PM emissions occurred without an increase in NOx emissions. The
nitrogen enrichment tests showed that nitrogen enrichment could be 30% more
effective in reducing NOx formation than cooled EGR.
Objectives
The primary technical objective of the program is to experimentally determine the
capacity of available membrane modules to perform to program target engine
requirements of nitrogen composition, intake flow, and power (maximum 23 hp).
The second objective is to establish the module parameters necessary to specify
the package size and form factor. Finally, these inputs will be provided to
prospective package manufacturers to obtain system cost estimates. To
accomplish these objectives, experimental data will be analyzed utilizing a
computational model of module behavior to specify necessary N2/O2 separation
factor and N2 and O2 permeation rates. The program objectives will be met
when experimentally-validated specifications for module packages can be
produced using the computational model.
The target diesel engine specifications are:
Displacement:
14 liter (Prototype twin turbocharger system)
Rated output:
460 hp (at 1800 rpm)
Commercial designation: Mack truck 460 hp engine
Fuel:
diesel #2
Intake flow, lb/min:
7 at idle
Optimized around 30 50
75 maximum
The module package requirements are:
Form factor: cylindrical (like that of milk can air cleaners that are seen
mounted on the side of an over-the-road tractor)
Size: 2 units maximum each approximately 18 inches in diameter and 2
feet long
Total power consumption: 23 hp
Intake gas composition: minimum 82% nitrogen
Technical Approach
The current program includes membrane material and module screening and
selection, laboratory evaluation of selected modules under simulated steady
state operating conditions, computational performance model development and
validation, and field testing with the candidate pre-production module elements
identified during the program operated on a representative target diesel engine.
Survey and Module Procurement
The programs approach to meeting the target engine nitrogen upgrading
requirements includes reviewing pertinent technical literature and recent
conference proceedings, survey of membrane/module manufacturers, and
compilation of resulting data to establish current system performance
capabilities. This data provides the foundation for selection of units for
procurement and lab testing at Argonne under simulated field conditions and
eventually at our industrial partners facilities with a target diesel engine.
Where data developed during this survey indicate a particular membrane
material or module will potentially meet the needs of the target engine intake flow
and composition, that material or module is sought from available suppliers.
Samples for testing are then obtained by out right purchase or some form of
evaluation agreement with the supplier. As the programs membrane parameters
are developed, suppliers will be contacted again as needed to update them with
the refined performance specifications.
Lab Testing
Membrane modules selected for testing are procured and installed in a reworked
test bench. This test bench is used to evaluate nitrogen upgrading performance
of modules over a range of throughputs and operating pressures. Data collected
from these runs with air and pure gases is analyzed to estimate membrane
selectivity and permeability using a computational model. From this analysis,
power consumption requirements are estimated and a permeation system is
specified that is anticipated to met target engine intake and load requirements.
Analysis of the data using the computational model will also identify process
options that may offer reduced energy consumption while maintaining target
nitrogen enrichment. Some of these options include use of single stage
compression of intake (standard turbocharger configuration), modulated
turbocharger compression of intake, addition of permeate suction blower, use of
sweep gases, and other process configurations. System configurations are
discussed in Appendix B.
Field Testing
One or more membrane systems will be assembled and tested at Argonne in
preparation for field test at a Mack Truck facility where the system under test will
be evaluated using a target diesel engine. These tests will include operation
over a standard 13 mode OICA test protocol. This protocol requires operation at
13 different steady state modes consisting of 4 loads at 3 different speeds and
idle.
The testing at Mack Truck will be conducted jointly by Argonne and Mack Truck
personnel and they will prepare a report on performance that will show the data
developed. From this, a report will be prepared by Argonne indicating the status
of the module performance and identify steps that might then be needed to
achieve target truck intake requirements. Engine performance criteria for the
target engine is provided in Appendix A along with test protocol details.
Module Procurement
Initially, contacts with prospective module suppliers were focused on those with
which Argonne had previous experience. They were asked to submit modules
that they felt would be their best units for our NEA application. Each of these are
described briefly below and further details are provided in the referenced
appendices.
Anticipating need for further improvement in module performance, we initiated
utilization of a computational model to guide our module specification. This
model is being validated during the laboratory tests to enable better specification
of permeation parameters to be provided to prospective suppliers.
To increase the range of available materials and systems, systematic literature
and patent searches are being conducted with the objective of identifying
additional, new candidates for our NEA modules.
Survey
Early on, Argonne identified candidate materials for NEA having a preferential
permeation of oxygen so that the product stream (retentate) would become
nitrogen rich. Details of these results were presented in earlier Argonne reports
and they are summarized here in tabular form (Table 1. Membrane Materials
Survey Results) and graphic format (Figure 1. Membrane Materials Availability
and Parameters). These results indicate that a rich field of potential membrane
materials exists having separation ability for nitrogen from air. For example,
material data reported from Air Products, Dow, and Monsanto shown and points
1, 10, and 31 respectively, suggest N2/O2 separation factors above 5 can be
achieved. In all, 31 candidates were identified having separation factors above
1.8. Further, several of these have reached a level of maturity, having been
fabricated as separation modules. The current project is now evaluating
materials that were found promising based upon this foundation survey work. As
the current project identifies the performance thresholds for the separation
parameters required to meet the target engine requirements, we expect to be
able to provide potential module suppliers with more precise specifications that
will enable them to work towards our applications goals.
#
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
Company
Air Products
Asahi Glass
Asahi Glass
Du Pont
Du Pont
Du Pont
Matsushita
Mitsubishi Chem
Mitsubishi Gass
Dow
Nippon Denso
Sagami
Sanyo
Sanyo
Sanyo
Sanyo
Shin Etsu
Shin Etsu
Teijin
Teijin
Teijin
Teijin
Teijin
Toray
Toray
Toray
Toray
Toray
Toray
Toyota
Monsanto
CMS
Material
PMSP
Silicone + Florinated Polimer (2 layers)
Aminosilicone + PMP (2 layers)
Fluorinated Polymide
Asymetric Polymide
Coated Polymide
Silicone
Asymetric Silicone
PPO + Silicone
Polycarbonate HF
Plasma Treated Silicone
Silicone
Asymetric Silicone
Silicone
PSMP
Silicone + substituted polyacetylene (2 layers)
Supported PMP
Supported PMP
Supported Silicone Hollow Fiber
Silicone + PMP (2 layers)
PMP
Cross Linked Amino Silicone
Supported PMSP coated with PMP
PS + Silicone
Cross Linked Silicone
Silicone Layer on Hollow Glass Fiber
Polysufone HF + Silicone Coating
PDD-2/CMS-7
Table 1: Membrane Materials Survey Results, highlighted membrane materials

meet desired specifications from past testing and the membrane model.
Source: Private Communication CMS, Inc.
Selected Membranes for Oxygen/Nitrogen Separation

1.00E-01
Region of Interest
15
32
13
14
1.00E-02
Permeability - cm^3(STP)/cm^2*sec*cmHG
Not Commercially Available Today
23
11
18
24
1.00E-03
17
20,25
12
29
28
26
19
4,21
30
2
Commercially Available Today
22
1
9
1.00E-04
6
3
31
1.00E-05
10
1.00E-06
1.8
2.8
3.8
4.8
5.8
Selectivity - oxygen/nitrogen
Figure 1: Membrane Materials Availability and Parameters

Source: Private Communication CMS, Inc.
6.8
Our ongoing survey effort is being carried out to maintain the currency of this
survey data as new materials and methods are developed in the
polymer/membrane/module industry.
Module Procurement
Three membrane suppliers were originally identified for the project from the
results of the Argonne membrane/module survey work. The three suppliers were
MEDAL, Prism Membrane and Compact Membrane Systems. A fourth module
candidate, manufactured by Avir, and available from a previous program at
Argonne, was used for first tests runs with our reworked test bench.
MEDAL
MEDAL is short for MEmbrane Separation Systems Dupont Air Liquide. MEDAL
is the membrane research division of Air Liquide. It was believed that MEDAL
may be a promising supplier of membranes because Renault has a relationship
with Air Liquide. After a number of attempts to obtain test modules from MEDAL,
they indicated that they were not going to participate in this program as a
consequence of their priorities for allocation of their resources. However, they
did provide an analysis of their system. We plan to contact this supplier in the
third quarter to determine if their priorities have changed and if they will be able
to participate in this program by supplying module samples. Details of our
contacts with MEDAL and a copy of their report are provided in Appendix D.
Avir
Modules made by Avir made for evaluation purposes and available from previous
program(s) at Argonne were selected for first tests under the current program.
These tests were conducted to verify our reworked test bench performance and
were not expected to yield acceptable nitrogen enrichment performance with
required throughput. However, acceptable nitrogen enrichment was achieved
but with unacceptable throughput and concomitant horsepower limits. Details of
the tests with the three module elements tested are given in Appendix E. Avir is
no longer in business so no further procurement efforts are planned with these
units.
Prism Membrane
Prism Membrane is the research division of Air Products and a manufacturer of
nitrogen generation systems. The Prism Membrane module was expected to
show promising results based upon data they had developed. Initial discussions
with Prism led to their offering Argonne a prototype with analysis that would be
purchased for this program. The procurement was completed and a sample was
received August 2002. This unit was tested at Argonne and showed adequate
nitrogen enrichment but the throughput at acceptable horsepower was deemed
marginal. However, the tests showed good agreement with the computational
model used during this program. Analysis of the data taken is continuing and is
planned for reporting in a subsequent report. Data and results to date for this
module are provided in Appendix F. Following review of their analysis and our
data, we plan to contact Prism to present them with our findings and the module
permeation specifications we are developing to determine if they will be able to
best the performance of the prototype they have supplied for these tests.
Compact Membrane Systems, Inc.
Compact Membrane Systems, Inc. (CMS) provides modules for gas
separations using technology licensed from Dupont Company for totally
fluorinated polymeric membrane applications. CMS was contacted because they
had worked with ANL in the past on both nitrogen and oxygen enriched air
applications and had made modules that were able to meet the current programs
nitrogen enrichment goals. However, data was not available to adequate
characterize their modules for the current program so tests with this module were
deemed necessary. Protracted negotiations with CMS finally resulted in an
agreement that will provide this program with a test module by fourth quarter.
Details of our contacts with CMS and the agreement reached are presented in
Appendix G.
Parker Hanifin
A module manufactured by Parker Hanifin was procured from them following
discussions with them providing our general requirements. The unit we received
will be characterized on our module test bench during the third quarter. The
available information on this module and its procurement is compiled in
Appendix G.
Model Development
As data was developed with modules over a range of conditions, it became
apparent that a computational model would be necessary to enable comparison
of the performance of the different types of modules and enable accurate
specification of performance levels necessary to achieve the target engine
requirements. Further, the model would enable projection of attainment of size
and energy consumption goals, both critical to bringing this technology into the
field. By August 2002, additional staff was brought on to the project with
expertise in membrane module development and modeling. This effort is
continuing and results to date are presented here.
Objective
There are two main objectives to achieve with the computational modeling work.
One is to provide a method for analyzing lab/field module data and planning
testing (e.g. model brings disparate data into a unified design tool). The other is
to enable evaluation of alternative process designs to meet project objectives
while avoiding extended test runs (e.g. evaluation of multi-pass and permeate
sweep schemes). These objectives come together to provide the primary goal of
the project- to specify a module system that will enable the target engine to meet
the prescribed 2007 emission specifications for NOx.
Model Description
The model is an iterative computational procedure derived from physical
principles. It has been used in studies covering feed pressures from near
atmospheric to 1000 psig, feed compositions from fractions of a percent to over
30 percent, permeate pressures from near vacuum to hundreds of psi, and a
wide range of through puts. It has performed well for both tubular and various
flat sheet materials. It is primarily a practical design tool that achieves its
greatest potential in evaluation of process configurations based on hard lab data,
allowing that data to provide effective selectivity that practically represents
achievable performance and enables assessment of process options. Given
good test data, it allows users to carry out what if studies with confidence that
results will be practically achievable.
Model Principles
The gas permeation through polymers is based upon the driving force (partial
pressure differential) of individual gases with relative smaller pressure losses
along module (feed to retentate ports). The membranes selectivity is defined as
the ratio of individual gas permeation rates.
Model Parameters
The model considers the membrane module to have three ports under typical
operating situations: feed, retentate, permeate. The primary parameters at these
ports are the absolute pressure (psia), flow (SCF/min), and composition, %
(nitrogen, balance oxygen)
The model incorporates the ratio of permeation rates as the selectivity
(unitless: O2 permeation rate / N2 permeation rate) which can be taken from
pure gas data as an idealized selectivity or from performance data as an
effective selectivity
The permeation rate (SCF/min/sqft-psi) is defined for the retained gas,
nitrogen. The module size is defined in terms of the membrane area (sqft).
Status
Initially, the model was used to examine the data on hand and identify trends to
follow in the refurbishing of the test bench and procurement of test modules.
Although most of the initial data was developed at relatively low pressure drops,
it was apparent that the model was working accurately. Subsequent laboratory
data taken was analyzed in comparison with the model and projected trends
were observed in the data. Further examination of the data and model validation
are in progress. As the validity is established, the model application work will
then focus on establishing module parameters necessary to meet target engine
requirements.
Consolidation of existing lab data
As used for this project to develop a preliminary analysis for the available data,
the model was run with all the lab data for a single module. Inputs were: feed
pressure, flow, and composition; retentate pressure and composition and
permeate pressure. Area and permeation were estimated from the module size
and flow data.
Module outputs were retentate flow, permeate flow, and permeate composition.
These were compared with the model predictions by calculation of average
deviations. An effective selectivity for a particular module was determined from
evaluation of permeate composition data. Note that the effective selectivity can
vary from the ideal selectivity due to different module behavior with mixed gases
(typically a minimal effect for nitrogen and oxygen) and leakage across the
surface of the module from feed-retentate side to the permeate side. Other
factors include flow patterns from feed to retentate and along the permeate
surface. The effect of water vapor was not included.
Specific results of the average deviations as a percentage of readings of module
data and model output for all 21 lab tests of Module 25 were: retentate flow 8%,
permeate flow 3%, and permeate composition nitrogen 2%.
Process Evaluation
Based upon the general agreement of model and available data for the two
modules, an example parametric study was run to illustrate a single stage
module using only feed pressure as the driving force. The analysis shows the
benefits that may be achieved by increased feed pressures while avoiding the
use of permeate suction.
Field Testing
Test Description:
An 8 cylinder 460 hp diesel engine manufactured by Mack Truck was tested
using the OICA 13 mode test. (See Figure 10) The 13 mode test is a widely
used certification test in Europe and has been recently accepted as part of the
certification process for heavy duty diesel engines in the United States. The 13mode test is a steady state test that examines the engine operation at three
different engine speeds. The engine speeds are defined as follows:
1. The high speed nhi is determined by calculating 70% of the declared
maximum net power. The highest engine speed where this power value
occurs (i.e. above the rated speed) on the power curve is defined as nhi.
2. The low speed nlo is determined by calculating 50% of the declared
maximum net power. The lowest engine speed where this power value
occurs (i.e. below the rated speed) on the power curve is defined as nlo.
3. The engine speeds A, B, and C to be used during the test are then
calculated from the following formulas:
A = nlo + 0.25(nhi - nlo)
B = nlo + 0.50(nhi - nlo)
C = nlo + 0.75(nhi - nlo)
There are 4 loads applied to the engine at these speeds, 100% load, 75% load,
50% load and 25% load. The final test operating point is idle.
First the engine is operated without Exhaust Gas Recirculation (EGR) at a given
mode, with advanced fuel injection timing. After the data for this point is
collected, the Beginning of Injection (BOI) timing is retarded and the fuel rate to
the engine is increased to bring the engines power back to the designated test
point and data is collected. This is repeated until there is four BOI timing data
points per test point. Next EGR is added and the process is repeated for three
EGR rates.
The engine data that was recorded was the engines speed, torque, BOI timing,
EGR flow, intake air flow, break mean effective pressure (BMEP), fuel flow,
turbine pressure and temperature, compressor pressure and temperature and
intercooler inlet and outlet temperatures and pressures.
13 Mode Test for a 460 hp Mack V8 Engine

1800.0
A-100 (Peak Torque)
1600.0
Engines performance curve
B-100
Intermediat
Speed
1400.0
A-75
B-75
1200.0
Torque (lb-ft)
C-100
(Peak Power)
C-75
1000.0
A-50
800.0
B-50
C-50
600.0
A-25
400.0
B-25
C-25
200.0
Idle
0.0
500
700
900
1100
1300
1500
Engine Speed (RPM)
Figure 2: Engine test points for the 13-mode test used in the baseline tests
1700
1900
Baseline Data Analysis:

For the purposes of this project, the engine performance and emissions data was
analyzed with respect to the change or reduction in the oxygen concentration of
the intake air charge. Why use the change in oxygen concentration as a gauge
of EGRs performance? Both EGR and Nitrogen Enriched Air (NEA) function by
acting as a diluent. EGR dilutes the intake air charge with carbon dioxide to
reduce the oxygen availability during combustion. NEA dilutes the intake air
charge with extra nitrogen to reduce oxygen availability during combustion.
Since diesel engines are lean burn engines, diesel exhaust gas has an over
abundance of oxygen. Therefore to reduce the oxygen concentration in the
intake air charge, a large quantity of exhaust gas is needed. At low loads,
greater than 50% EGR may be needed to reduce the intake oxygen content from
21% to 18%. That makes comparing EGR to NEA difficult because 3% NEA will
reduce the intake oxygen from 21% to 18% no matter what the load.
EGRs Effect on NOx (Baseline Data):
EGRs effect on NOx emissions were not unusual when the relationship between
brake specific NOx emissions were compared to oxygen concentration reductions
in the intake due to EGR. The less oxygen, (or the more EGR) the less NOx
formed. (see Figure 11).
An interesting trend was found when the change in emissions was examined.
(see Figure 12) The predictability of percent change in NOx emissions is
extremely strong with respect to oxygen levels. This trend can be used to show
the effectiveness of NEA. If the NEA data falls upon the same line, then the NEA
is as effective as EGR. If the NEA data falls above the trend line, then the NEA
is more effective than EGR.
BSNOx VS EGR
at 1800 RPM and 230 hp
(C-50)
7.000
6.000
For an APPROXIMATE intake manifold oxygen

concentration, subtract the oxygen concentration
reduction number from 21% which is approximately the
concentration of oxygen in the atmosphere.
BSNOx (g/bhp-hr)
5.000
4.000
3.000
-12 ATDC BOI
2.000
-9 ATDC BOI
1.000
-6 ATDC BOI
Less EGR
More EGR
0.000
0
0.5
1.5
2.5
Oxygen Concentration Reduction (%)
3.5
Figure 3: Brake specific NOx Emission with respect to EGR
Emissions Reduction with Respect to Oxygen Reduction and DI BOI Timing

(C-50)
100.0%
Negative emissions reductions are actually

and increase in emissions.
80.0%
60.0%

reduction number from 21% w hich is approximately the
Emissions Reduction (%)
40.0%
20.0%
Less EGR
More EGR
0.0%
0
0.5
1.5
2.5
3.5
-20.0%
-40.0%
% CO Reduction (-12 ATDC BOI)

-60.0%
% NOx Reduction (-12 ATDC BOI)

-80.0%
-100.0%

-120.0%
Figure 4: Percent Emissions reduction with respect to oxygen concentration reduction. Note that the percent change in
NOx emissions is independent of BOI timing, yet dependant on oxygen levels.
Survey Results
The model confirms that power requirements discussed in previous meetings are
valid. The model also confirms that the power requirements are high because
the membrane data was taken with 7 psig pressure on the feed side and 13
psig pressure on the permeate side. The model showed that if the feed pressure
was increased, the membranes separation would be more efficient and a
vacuum pump may not be needed. (see Figures 2 and 3) The model also
showed that power requirements could be lower with higher boost pressure on
the feed side and no vacuum on the permeate.
Along with the power requirements of the membrane, the model also shows the
effect of feed pressure on the membranes size, another key constraint to this
study.
Retentate Flow, SCF/min
Retentate Flow vs Feed Pressure

7
6
5
4
Selectivity 1.8
Retentate 85% N2
Permeate 15 psia
..
2
1
0
15
25
35
45
55
Feed Pressure, psia

Figure 5: This chart illustrates the typical behavior of a membrane module
operated at various feed pressures with fixed product composition (85% N2) and
fixed permeate pressure of 15 psia. The permeabilities are estimated from
module 25 giving a 1.8 N2/O2 selectivity.
Membrane Area, sqft
Retentate Flow vs Membrane Area

15000
Selectivity 1.8
Retentate 85% N2
Permeate 15 psia
10000
5000
0
15
25
35
45
55
Feed Pressure, psia

Figure 6: This chart illustrates the effect of pressure on membrane module size
operated at various feed pressures with fixed product composition (85% N2) and
fixed permeate pressure of 15 psia. . The permeabilities are estimated from
module 25 giving a 1.8 N2/O2 selectivity.
Future Directions
Now that the model has shown good correlation with the initial data, further
analysis of recent data using the model will be continued. Then, some
extensions will be incorporated to facilitate evaluation of alternate process
configurations and overall energy requirements. Finally, the model will be used
in the fourth quarter to generate membrane and module specifications that will be
checked during the field test of a selected module.
Membrane Test Bench Conclusions
Two prototype membrane bundles that were evaluated during laboratory tests,
one over a full range of actual operating pressures up to 50 psig. This unit
maintained excellent nitrogen enrichment. Both were characterized with a
computational membrane model. The results demonstrated that the full range of
test conditions could be simulated in the test bench, but the power requirement
for these modules exceeded the program target goal. However, we believe that
the stated program requirements for this target engine can be met, based upon
the test runs and analysis we have completed. Planned project work is designed
to experimentally confirm this and will result in a specification for a module
package that will meet the program objectives defined by our industrial partner,
Mack Truck. Using the membrane model and the performance data, we
20
determined that a feasible membrane bundle composed of the same polymer

could work with changes to its form factor-specifically a larger bore size. We are
providing this information to module manufacturers to enable them to provide
modified modules.
Engine Testing Results
During the first mode of the 13 mode test, the prototype twin turbocharger system
failed for the V8 diesel engine. A 440 Hp inline 6 cylinder diesel engine was
used in its place. It should be noted that the only matching airflow point was
C75.
At low loads, the driving pressure was not great enough to supply the
required NEA to match the EGR rates
At high loads, the turbocharger was unable to meet the air flow demands
of the NEA system (Retentate plus Permeate) for the engine.
At the NEA system design point (C75) the NEA system performed as well
as EGR.
Shows that the NEA system cannot be used as a plug and play system
but must be tailored to the overall engine intake air system.
NOx Comparison
6.00
Nox Baseline
5.00
NOx (g/bhp-hr)
Nox Membrane
4.00
3.00
2.00
1.00
0.00
A25
A50
B25
B50
C25
C50
A75
B75
C75
A25
Mode
Figure 7: Thirteen Mode Test comparing NOx emissions with and without NEA
Other comments
Mack Truck decided not to continue this avenue of testing for 2 reasons. The
first reason was that there currently no supplier capable of making 10,000 units
per year and they were unwilling to commit more time and money to a technology
that may never make it into production because of lack of supply. The second
reason was that Mack Truck was experiencing relative success with their EGR
system. Mack Truck had solved some of the major durability issues associated
with the use of EGR.
21
Appendix A Lab Test Facilities

Previous Test Benches
The project began anticipating using a membrane characterization bench used at
Argonne during previous studies of membrane modules for engine feed gas
enrichment. As modules were obtained for the current program, it became
apparent that higher operating pressures and increased accuracy in flow
measurement was needed to enable full-range measurements required by the
target engine operating parameters. To meet these needs, the test bench was
reworked following the program schedule adjustment made August 2002. Since
then, the bench has been proven out during numerous test runs with several
modules.
Original Configuration
The original membrane module characterization bench was designed to analyze
membrane performance in a low-pressure pressure driven configuration or a
vacuum driven configuration. The bench consisted of a positive displacement
blower for the membrane feed and a liquid ring blower to provide vacuum on the
permeate side of the membrane. A oxygen analyzer was used to measure the
oxygen content of the feed, permeate and retentate flows. Permeate, feed and
retentate pressures were also measured. Retentate and permeate flows were
measured using a relative flow measurement method- hot wire anemometry.
Current Configuration
The test bench was redesigned and rebuilt to achieve four goals:
Wider feed test pressure range
High accuracy flow measurement
Flexibility in accommodating a variety of modules
Facilitate tests with
pure gases to help
validate model
calculations
The reconstruction has
resulted in a robust bench
built in sections for
portability and flexibility.
The new flow
measurement equipment
now gives high accuracy
for both permeate and
retentate and is traceable
to NIST standards. Feed
air is now obtained from
house air compressor to
enable testing at higher
Figure 8: New redesigned membrane test bench 22
pressures. To accommodate this, extra filtration was included to assure removal

of vapors as well as particulates and droplets. Heavy-duty pressure regulators
enable accurate feed gas pressure control over a range of a few psig up to 50
psig. The test bench is designed to be operated at ambient or elevated
temperatures. An optional heater unit can provide feeds above ambient to about
250 deg. F. A new feed gas manifold facilitates tests with pure gases that are
done at zero retentate flow to help validate the membrane model. The retentate
and permeate flows are now entirely captured to positive displacement meters.
The positive displacement meters are generally used to calibrate most other
types of flow meters. The feed, permeate, and retentate piping systems have
been modified to accommodate change-out of a wider variety of module types
without complete reassembly of the bench piping. With the changes now
completed, the test bench can accommodate permeate flows up to about 100
SCF/min or retentate flows up to about 400 SCF/min. See Figure 6 for a
Schematic of the membrane test bench.
An extensive (20+page) safety plan was prepared to meet Argonne safety
requirements.
23
Figure 9: Schematic of the Current Membrane Characterization Bench
For these runs, feed air was obtained from a house air compressor at feeds up to
50 psig. The extra filtration provided by the bench was used to assure removal
of vapors as well as particulates and droplets. During the runs with the Prism
module, the test bench retentate flows ranged up to about 30 SCF/min. Pure gas
runs were made with pressurized cylinders containing industrial grade gases.
Appendix B. System Configurations

As the concept of adjusting engine inlet gas composition has developed, various
system configurations have been considered. These include variations in the
way the gas composition is adjusted with module configurations and ways of
positioning the modules with respect to the engine inlet, turbocharger, and
exhaust system. Some variations include the addition of a vacuum pump on the
permeate stream. Most system configurations that have been explored to date
are described here and others are under consideration and will be described in
later reports.
Preturbocharger Upgrading {Turbocharger After Module no vac pump}
In this NEA system, depicted in Figure 6, the turbocharger compressor is used to
pull the retentate first through a particulate filter and then through the membrane.
The permeate is vented directly to the atmosphere and the engine is otherwise
configured normally. A mechanical advantage to this scheme is that little
physical modification to the engine intake setup is required since the module and
filter can be plugged in ahead of the existing turbocharger compressor inlet. A
downside of this configuration is that the membrane module operating pressure
(feed to retentate) will not operate at the higher pressures that can be delivered
by the compressor at its discharge. For this reason, without enormous
improvement in module performance at low feed-retentate pressures, this
configuration is not likely to achieve practical application.
Figure 10: Preturbocharger Ungrading without a vacuum pump on the permeate side
Preturbocharger Upgrading With Vacuum Forced Permeate{Turbocharger

After Module}
The Preturbocharger Upgrading configuration shown in Figure 6 above can be
made more effective by introducing a vacuum pump on the permeate stream, as
shown in Figure 7, to effect reduced permeate pressure and thereby greater
feed-to-permeate pressure ratio. This benefit is likely offset by the additional cost
and operating power required by the vacuum pump. Further, even at very low
vacuum, the feed-to-permeate pressure ratio will remain below that required for
acceptable system performance without significant increase in module
membrane enrichment capability.
Figure 11: Preturbocharger Upgrading With Vacuum Forced Permeate
Post turbocharger Upgrading {Turbocharger ahead of module}

This NEA system uses a pressure driven system having the discharge of the
turbocharger compressor feed into the membrane module. Filtration for the
module can be placed ahead of the compressor as shown in Figures 6 and 7 or
after the compressor and just before the module. In either case, the module
benefits from having the full discharge pressure of the compressor available at
the module feed and the system performance is thereby improved substantially
due to the higher feed-to-retentate pressures available relative to the
approximate ambient pressure of the permeate. A mechanical downside to this
configuration is that the membrane module must withstand the full compressor
discharge pressure and also the module must cope with or be protected from the
wide range of compressor discharge pressures delivered over the operating load
range of the target engine.
27
Figure 12: Post turbocharger Upgrading with Permeate Vent to the Atmosphere
Postturbocharger Upgrading With Vacuum Forced Permeate{Turbocharger

Ahead of Module}
Figure 13: Postturbocharger Upgrading With Vacuum Forced Permeate
28
Figure 9 (Postturbocharger Upgrading With Vacuum Forced Permeate) shows a

variant of the system shown in Figure 8 that includes addition of a vacuum pump
on the permeate discharge of the module. This can improve the feed-reject to
permeate pressure ratio and thereby increase system separation performance.
These are realized at the additional cost of the vacuum pump and its driving
energy as well as possible increases in module costs due to the need to have
vacuum-capable porting installed at the module permeate port. Further, design
consideration must be given to the possible module failure mode of the module
resulting from high pressures in the feed-retentate zone of the module causing a
high pressure in at the module permeate side and on into the vacuum pump and
its lines. Note that this failure mode would not be significant using the system
shown in Figure 7 (Preturbocharger Upgrading With Vacuum Forced Permeate)
which has the feed-retentate pressures near ambient and consequently little
effect upon the permeate port or vacuum pump of that configuration.
29
Appendix C. Field Test Facilities

Test Unit Preparation At Argonne
The most effective module selected during the laboratory testing and modeling
activities will be prepared for installation in the engine test laboratory of the
industrial partner, Mack Truck. This will include the baseline performance testing
under steady and non-steady state conditions and corresponding model
validation. Subsequent to these tests, the module will be suitably mounted for
operation in the engine testing facility. Accurate flow and pressure
measurements will require some modification to the test setup to accommodate
the actual flows anticipated during dynamic testing on the target test engine.
Adequate filtration must be included and the set up must be prepared to handle
the range of ambient temperatures anticipated.
Test Plan At Industrial Partner
A full-scale target engine will be set up on the industrial partners engine test bed
by their staff. The Argonne nitrogen enrichment module assembly will be
integrated into the test engine using the system configuration selected during the
current programs bench scale testing.
The partners test cell meets all federal engine test standards for heavy duty
compression ignition engines and can perform the required protocols for the 13
mode test specified in the code of federal regulations, section 40, over all the
target engine operating ranges required for the tests. These modes include idle,
and four percentages at each of the peak torque, intermediate torque-power
point, and peak power (C100 point). A typical 13 mode test showing actual
typical target engine torque vs speed is depicted in Figure 10.
The tests, including the 13 mode test, will relate the NEA test assembly
performance to previous baseline tests performed by the partner on the target
engine using EGR. Analysis earlier during this project at Argonne of the baseline
data was used to guide selection of nitrogen enrichment criteria for the laboratory
bench scale tests. A summary of that analysis is presented below.
Baseline Data and Evaluation of Test Results
Test Description:
An 8 cylinder 460 hp diesel engine manufactured by Mack Truck was tested
using the OICA 13 mode test. (See Figure 10) The 13 mode test is a widely
used certification test in Europe and has been recently accepted as part of the
certification process for heavy duty diesel engines in the United States. The 13mode test is a steady state test that examines the engine operation at three
different engine speeds. The engine speeds are defined as follows:
4. The high speed nhi is determined by calculating 70% of the declared
maximum net power. The highest engine speed where this power value
occurs (i.e. above the rated speed) on the power curve is defined as nhi.
5. The low speed nlo is determined by calculating 50% of the declared
maximum net power. The lowest engine speed where this power value
occurs (i.e. below the rated speed) on the power curve is defined as nlo.
30
6. The engine speeds A, B, and C to be used during the test are then
calculated from the following formulas:
A = nlo + 0.25(nhi - nlo)
B = nlo + 0.50(nhi - nlo)
C = nlo + 0.75(nhi - nlo)
There are 4 loads applied to the engine at these speeds, 100% load, 75% load,
50% load and 25% load. The final test operating point is idle.
First the engine is operated without Exhaust Gas Recirculation (EGR) at a given
mode, with advanced fuel injection timing. After the data for this point is
collected, the Beginning of Injection (BOI) timing is retarded and the fuel rate to
the engine is increased to bring the engines power back to the designated test
point and data is collected. This is repeated until there is four BOI timing data
points per test point. Next EGR is added and the process is repeated for three
EGR rates.
The engine data that was recorded was the engines speed, torque, BOI timing,
EGR flow, intake air flow, break mean effective pressure (BMEP), fuel flow,
turbine pressure and temperature, compressor pressure and temperature and
intercooler inlet and outlet temperatures and pressures.
31
13 Mode Test for a 460 hp Mack V8 Engine

1800.0
A-100 (Peak Torque)
1600.0
Engines performance curve
B-100
Intermediat
Speed
1400.0
A-75
B-75
1200.0
Torque (lb-ft)
C-100
(Peak Power)
C-75
1000.0
A-50
800.0
B-50
C-50
600.0
A-25
400.0
B-25
C-25
200.0
Idle
0.0
500
700
900
1100
1300
1500
1700
1900
Engine Speed (RPM)
Figure 14: Engine test points for the 13-mode test used in the baseline tests
32
Data Analysis:
For the purposes of this project, the engine performance and emissions data was
analyzed with respect to the change or reduction in the oxygen concentration of
the intake air charge. Why use the change in oxygen concentration as a gauge
of EGRs performance? Both EGR and Nitrogen Enriched Air (NEA) function by
acting as a diluent. EGR dilutes the intake air charge with carbon dioxide to
reduce the oxygen availability during combustion. NEA dilutes the intake air
charge with extra nitrogen to reduce oxygen availability during combustion.
Since diesel engines are lean burn engines, diesel exhaust gas has an over
abundance of oxygen. Therefore to reduce the oxygen concentration in the
intake air charge, a large quantity of exhaust gas is needed. At low loads,
greater than 50% EGR may be needed to reduce the intake oxygen content from
21% to 18%. That makes comparing EGR to NEA difficult because 3% NEA will
reduce the intake oxygen from 21% to 18% no matter what the load.
EGRs Effect on NOx:
EGRs effect on NOx emissions were not unusual when the relationship between
brake specific NOx emissions were compared to oxygen concentration reductions
in the intake due to EGR. The less oxygen, (or the more EGR) the less NOx
formed. (see Figure 11).
An interesting trend was found when the change in emissions was examined.
(see Figure 12) The predictability of percent change in NOx emissions is
extremely strong with respect to oxygen levels. This trend can be used to show
the effectiveness of NEA. If the NEA data falls upon the same line, then the NEA
is as effective as EGR. If the NEA data falls above the trend line, then the NEA
is more effective than EGR.
33
BSNOx VS EGR
(C-50)
7.000
6.000
concentration, subtract the oxygen concentration reduction
number from 21% which is approximately the concentration
of oxygen in the atmosphere.
BSNOx (g/bhp-hr)
5.000
4.000
3.000
2.000
-12 ATDC BOI

-9 ATDC BOI
1.000
-6 ATDC BOI
Less EGR
More EGR
0.000
0
0.5
1.5
2.5
Figure 15: Brake specific NOx Emission with respect to EGR.
3.5
Emissions Reduction with Respect to Oxygen Reduction and DI BOI Timing

(C-50)
100.0%
Negative emissions reductions are actually

and increase in emissions.
80.0%
60.0%

reduction number from 21% w hich is approximately the
Emissions Reduction (%)
40.0%
20.0%
Less EGR
More EGR
0.0%
0
0.5
1.5
2.5
3.5
-20.0%
-40.0%

-60.0%

-80.0%
-100.0%

-120.0%
Figure 16: Percent Emissions reduction with respect to oxygen concentration reduction. Note that the percent change in NOx emissions is
independent of BOI timing, yet dependant on oxygen levels.
Appendix D. MEDAL Modules

Test Objectives
As MEDAL provided analysis of their system as presented below, but did not
provide hardware for test, no test program was undertaken to confirm their
reported results.
Suppliers Data
The following report was provided at Argonnes request for NEA membrane
modules. As indicated above, their design criteria differ from those provided
initially by Mack Truck and Argonne, but the report does provide useful
information on potential module configurations.
MEDAL
MEmbrane Separation Systems Du Pont Air Liquide
Jan 31, 2002

From: Ian Roman (MEDAL)
Greg Fleming (MEDAL)
To:
Steve McConnell (Argonne)

Raj Sekar (Argonne)
Air Separation Membranes for NOx Reduction in Diesel Engines.

Membrane Upstream of the Turbocompressor
Summary:
An earlier analysis by MEDAL defined the need and opportunity for a new
process for NOx reduction in heavy-duty diesel engines and mapped out the
requirements for a membrane-based solution. The membrane was placed
downstream of the turbocompressor. It enriched the air stream from the
turbocompressor to 81% inerts, with about 12% loss of the of the original feed
stream. The energy of compression was provided by simply sizing the
turbocompressor up by 12%. The analysis considered a broad range of
membrane performance, encompassing commercially available high-selectivity
membranes for industrial N2 as well as new high-permeance membranes under
development. Our conclusion was that a stand-alone membrane in this design
would not be efficient due to excessive membrane size, cost or percentage loss
of compressed air. Addition of a vacuum pump to help extract the permeate
would help, but the economics were still marginal.
An alternative design solution is examined in the present analysis. The
membrane is placed before the turbocompressor, with a dedicated vacuum pump
to extract the permeate. This design is easier to retrofit on existing engines and
provides a definite advantage in process control and a small advantage in energy
cost compared to the post-turbocompressor design with or without vacuum
pump. However, our analysis indicates that it would require markedly higher
membrane size and cost due to the smaller transmembrane pressure difference.
Even with fully-optimized high-permeance membrane, we estimate an
excessively large membrane area.
Background
24
Reduction of emissions of nitrogen oxides (NOx) from diesel engines is

required by increasingly stringent government regulations. This is forcing diesel
engine manufacturers to look for new technology to manage NOx. The most direct
solution is after-treatment of the exhaust to remove NOx. However, the cost of aftertreatment with available methods is too high. Another solution is to reduce the
formation of NOx in combustion, by slightly reducing the O2 concentration in the feed
air to reduce the combustion temperature. Replacing 21% O2 with 19% O2 reduces
NOx by as much as 70%. While this approach cannot eliminate all NOx, it is
attractive because it can be applied with available technology.
Exhaust gas recirculation, or EGR, reduces O2 concentration by adding a
fraction of the O2-poor exhaust to the feed. EGR is cost-effective in spark-ignition
engines but doesn't work as well in diesel engines, which produce dirtier exhaust.
EGR increases particulates, another emission that must be managed. EGR also
increases (and feeds back to the engine) carbon and sulfur, which may reduce
engine life. A further problem is that the exhaust gas is hot and must be cooled
before recirculation to avoid reducing engine power. In spite of the drawbacks, EGR
is the solution most widely studied for NOx reduction so far.
A cleaner method to reduce the O2 concentration in the feed air is to use a
N2-enrichment membrane. In this case, a small part of the feed air is forfeited as a
waste stream enriched in oxygen; the loss is equivalent to added compression cost.
Several other technical solutions to reduce NOx have been examined by the
diesel engine manufacturers, but none appear competitive with EGR or membrane
in the timeframe needed.
Membrane and process requirements
1. The size and weight of the membrane plus related system must be small. In
earlier discussions with diesel engine manufacturers, an informal first-cut target size
was set based on practical considerations: a single bundle, roughly the size of a 12"diameter Medal bundle. Small membrane size is made possible by a combination of
high O2 permeance, small fiber size, and adequately high feed/permeate pressure
difference.
2. The cost of the membrane plus related system must be low. A tentative figure
based on the estimated ultimate cost of the EGR system is $1500. This figure is only
a starting point, but it defines the order of magnitude.
3. The added cost of compression must be less than 12% of that of the
turbocompressor. Low cost is favored by high O2/N2 in the membrane and high
feed/permeate pressure ratio.
4. The loss of pressure from feed to N2-enriched nonpermeate in the membrane is a
compression cost and must be small. Low pressure loss is best achieved with shellside feed.
5. The membrane must have good resistance to fouling in service because the cost
and size requirements preclude membrane replacement and feed air pretreatment.
6. The membrane must have performance with shallow dependence on
temperature, since the feed is now uncompressed atmospheric air with
temperature in a broad range. This requirement will impose additional cost for
high-selectivity membranes.
Simulated performance of membranes
24
Criteria and assumptions

The performance of several air-separation membranes in the preturbocompressor design was simulated with a computer model. The results are for
first-cut analysis, and are limited by the following simplifying assumptions.
- The target purity was set at 19% O2, even though an advanced ultimate solution
may be to match the purity to the variable load and engine speed with an optimized
schedule. 19% has been determined a useful average concentration, and is a good
starting point.
3
- The full-load flowrate to the engine manifold was 1400 Nm /h. This is a
simplification, but it is the worst case. Moreover we do not have flowrate/speed/load
curves from the engine manufacturer to map the variable flowrate.
- The membrane must provide a range of flowrate of N2-enriched air to match the
engine requirement. Range is available by varying in concert the permeate pressure
(vacuum pump) and the nonpermeate flowrate (throttle valve) in parallel with air
demand. For example, for a moderate-selectivity membrane (5.0 O2/N2), varying the
permeate pressure from 0.5 to 0.25 bara provides an almost 3-fold range of flowrate
of 81% inerts, with limited effect on compression energy.
Perm press. Flowrate of 81% N2 Energy/mole of 81%
0.50 bara
1.00
1.0
0.40 bara
1.56
1.06
0.33 bara
1.79
1.25
0.25 bara
2.71
1.15
Note that compression energy actually decreases at low enough permeate pressure,
where the separation is more efficient. However, we do not know if that pressure
below 0.33 bara is practical. For a first-cut analysis, the permeate pressure for full
engine load (1400 Nm3/hr) was fixed at 0.33 bara.
- The following bundle dimensions were assumed: 30 cm diameter by 73 cm active
length.
- Shell-feed operation was preferred because it allowed minimizing pressure loss
from feed air to the nitrogen product. Tube feed is more widely used for industrial
nitrogen (95-99.5% N2) but is unsuited for 81% N2. The fiber was carefully sized to
provide the optimum balance of pressure loss and membrane surface area.
- Three membranes were considered. One was a state-of-the-art high-selectivity
membrane, with 5.0 O2/N2. The membrane (fiber and bundle dimensions) was
customized for this application. The second membrane was a high-permeability
membrane similar to the perfluorodioxole membrane described in Poola et al, with
O2/N2 of 2.5 and large fiber size (1100 m diameter) and assuming a practical 12
bundle geometry. The third membrane was a hypothetical case with intermediate
performance (800 GPU O2 permeance and 3.4 O2/N2), again carefully optimized for
this application.
Results
The critical parameters examined were the membrane size (target: one 12 bundle),
expressed as number of 12" bundles (target: one bundle), the compression energy
and the pressure loss in the membrane. One high value among these parameters is
enough to disqualify the membrane.
Even in early calculations, it was clear that the membrane area was too high.
The high-selectivity membrane fells far short 22 bundles at full load. This was
expected because of the earlier analysis with the post-turbocompressor design.
24
The high-permeability membrane required 3 bundles, still too high. Moreover,

the membrane also fell clearly short of the compression energy target, as in the
post-turbocompressor design. The compression energy per mole of 81% N2 was
50% higher than for the high-selectivity membrane.
Conclusions
Compared to the post-turbocompressor design, with or without vacuum
pump, the design placing the membrane before the turbocompressor with a
dedicated vacuum pump to extract the permeate provides gains in process
control and in energy cost, but the required membrane size and cost increases
significantly due to smaller transmembrane pressure difference. Neither design
meets our [MEDALs] criteria for an economically-viable membrane.
NOTE: This concludes MEDALs report
Module Procurement and Description
MEDAL is short for MEmbrane Separation Systems Dupont Air Liquide. MEDAL is
the membrane research division of Air Liquide and makes hollow fiber membrane
modules in a range of sizes of potential application to our target requirements. They
offered preliminary analysis of three possible prototypes based upon three variations
of sizes and material types. For one example, they suggested a membrane with a
5.0 O2/N2 selectivity. The bundle dimensions were assumed to be 30 cm diameter
by 73 cm active length. Shell-feed operation was preferred because it allowed
minimizing pressure loss from feed air to the nitrogen product. The second
membrane was a high-permeability membrane similar to a perfluorodioxole
membrane with O2/N2 of 2.5 and large fiber size (1100 m diameter) and assuming
a practical 12 bundle geometry. The third membrane was a hypothetical case with
intermediate performance (800 GPU O2 permeance and 3.4 O2/N2).
Although we considered their modules good candidates for our test program and
made a request to them to participate in our program. On December 6, 2001 a
meeting took place between MEDAL, Mack Truck and Argonne National
Laboratory. MEDALs major concern during this meeting was funding. They felt
that meeting this projects needs might be achieved with a significant effort on
their part. Their concern was recovering their costs. The outcome of the
meeting was that MEDAL would perform a membrane analysis on paper (copy in
Appendix D) to look at the feasibility of their membrane polymers. The analysis
would use a membrane bundle size of 12 inches in diameter and three feet in
length.
Even though the membrane analysis report was not favorable in their view- likely
due to their imposition of more stringent requirements than suggested by
Argonne, ANL requested a 100 SCFM membrane bundle for characterization in
an attempt to provide an indication of how much development is needed to meet
performance criteria.
Subsequently, Mr. Greg Flemming of MEDAL sent the following E-mail in
response to the request for a membrane bundle to characterize.
Unfortunately, I currently do not have the resources to provide you with
the requested membranes. Cost is not the key constraint. We currently
have a very full agenda. Sorry I can not help, I hope you understand.
24
Anticipated Performance
MEDALs analysis of the three different membranes (copy of report given in
Appendix D) was based on parameters from a previous study that were
considerably different than those proposed by Argonne and Mack Truck for the
current target engine. Consequently, their conclusions may not directly pertain to
our target engine requirements, but their report does provide significant
information on potential modules.
Because MEDAL chose not to supply modules for our testing, but did provide
some data and analysis, we have only reviewed this material to develop
anticipated performance. From the data provided, we concluded that some of
their membrane materials had good potential for high separation factors (up to 5
O2/N2), but that their gas permeation rates were apparently not adequate to
provide acceptable size or form factor without some further development or
modification. However, we still deem these of interest for testing because they
offer the possibility of higher separation factors than any of the other modules we
have tested to date.
Work in Prospect
Although MEDAL has not yet agreed to provide modules for our testing, we will
continue to evaluate the data provided in their report utilizing the results of our
computational modeling and attempt to maintain contact with them to assess
their recent developments and develop laboratory data with their modules.
24
Appendix E. Avir Modules

Test Objectives
We chose to make the first runs on our rebuilt bench using modules that, if
damaged during the bench shakedown tests, would not delay the program. The
modules selected were on hand and had been previously subjected to a variety
of tests. The test runs were selected to enable characterizing the modules over
the range of anticipated target engine pressures, but at ambient temperatures,
while establishing the correct operation of our reworked test bench.
Test Program and Parameters
During preliminary leak testing, limited test results were obtained for the (Avir)
module at pressures under the 15 psig level: 5, 10, and 15 psig feed. The feed
(house air) was thoroughly filtered of droplets and dusts as well as conditioned
with a vapor absorber. Retentate flows were varied from zero to about 20
SCF/hr.
Three modules were installed in a single chamber. These were manifolded at
each end with double end caps using O-rings. The permeate was manifolded
into the containing chamber and out a side port.
Suppliers Data
The three elements installed for these tests were marked as follows:
A/G Technology Corporation
Avir Gas Separation
US Patent 4,681,605
34 Wexford Street
Needham MA 02194
Model No. GSP-SEI-75M
Maximum Air Supply 110 psig
Maximum Air Temperature 40 C
Element 1
Serial No.
Element 2
Element 3
7A8A9A302201AL 2A3A4A302301AL 16A30123A301301JD
Nominal
Capacity,
SCFM
10
12
12
Oxygen, %
30
31
31
Temperature, C
24
24
22
Pressure, psig
60
60
60
Table 2: Avir Gas Separation Module specifications

Avir Gas Separation module elements (Figure 13) were available to this
project from previous programs at Argonne that focused on small engine intake
stream enrichment. They had been supplied by A/G Technology Corporation in
24
Needham MA. Three module elements were selected from this group of
modules and enclosed in a single chamber (Figure 14), in effect, operated in
parallel, with individual feeds, retentates, and permeates manifolded to result in a
three port module.
Each element measured 51 inches long (full diameter distance) and 3 inches in
diameter.
Feed
Retentate
3 inch dia.
51 inches long
Permeate
Figure 17: Single Avir membrane module element
24
Membrane Module Chamber, 7 inches in

diameter, 57 inches long
Retentate Port
Permeate Port
Feed Port
Figure 18: Membrane Module Chamber
24
The module elements utilized a tube and shell configuration with potted
ends leading to the feed and retentate ends. Permeate was released through
cross bores into the individual module elements as shown in Figure 15.
Permeate Vent Ports
Figure 19: Permeate Vent Crossbores
Module Mounting and Housing

The three modules selected for testing were mounted in an outer chamber that
measured 7-11/16 inches in diameter and was 52-3/4 inches long.
The module chamber provides manifolding of the feeds, retentates, and the
permeates into three separate ports. This is accomplished by mounting the
elements on flanges at the feed and retentate ends. These streams are kept
apart from the permeate with seals on the elements comprised of two O-rings at
each end (Figure 16) that snugly fit into the flange holes.
24
Dual O-rings
Figure 20: Dual O-rings on Module Elements
First, the feed end of each of the three module elements was inserted into
the feed Element Mounting Flange (Figure 17). That flange was bolted on to the
feed end of a cylindrical chamber which has a diameter of about 7 inches (Figure
18). Then the retentate ends of the elements were inserted into the retentate
Element Mounting Flange and that flange was bolted to the cylindrical chamber.
At this point, both feed and retentate ends of the elements are retained by the
Mounting Flanges. The mounting holes in the flanges also serve as seals
between the feed gas and permeate on the feed end of the chamber and the
retentate gas and permeate on the retentate end of the chamber.
24
O-ring groove for manifold cap
Element Mounting and

Sealing bores
Flange Bolt Circle
Figure 21: Element Mounting Flange
24
Retentate Discharge from

Elements
Flange bolted to chamber
Manifold cap o-ring
Manifold cap bolt circle
Figure 22: Module Flange Element Bolted to Membrane Module Chamber
Assembly was completed by installing Manifold Caps (Figure 19) at both

ends. At the feed end, .these distributed gas to the feed ports of the three
elements. Similarly, at the retentate end, the element retentate discharge was
collected under the retentate Manifold Cap and led to the retentate piping.
24
Chamber Side
Manifold Cap Bolt

Circle
Combined Flow Port
Figure 23: Manifold Cap
Results and Analysis

Initial analysis of the data confirms our suspicion that the Avir modules were
previously degraded during tests that allowed severely contaminated feeds to
enter the modules. In spite of this suspected degradation, the modules showed
some gas separation and the runs successfully demonstrated the operability of
the test bench.
Data indicated that in order to achieve separation, high permeate flows are
required. This is indicative of membrane elements had been subjected to feeds
containing significant contaminants which deteriorated the membrane walls.
Further, the elements had an oil-like odor similar to compressed air derived from
an oil lubricated compressor.
The Avir units were labeled at about 10 SCFM each providing 31 % oxygen. In
our runs, the three elements operating together as a unit showed permeate
compositions only as high as about 26%, indicating that they were well below
rated oxygen capacity.
24
From the retentate flow and composition data, the modules showed separation to
over 85 percent nitrogen, but the retentate flows were relatively small, being only
a few standard cubic feet per minute. As feed pressure was increased from 5 to
15 psig, the data show that for given retentate flow, the unit gave increased
nitrogen composition. However, as can be seen in the chart of retentate nitrogen
percentage and permeated gas flow, the units performance fell far short of the
program target of about 1000 SCF/min.
Avir Modules With Air Feed
Nitrogen %
Retentate Nitrogen and Retentate Flow
86
85
84
83
82
81
80
79
5 PSIG
10 PSIG
15 PSIG
10
15
Retentate, scf/min
Figure 24: The membrane achieved good separation at low retentate flows but
lost almost all of its ability to separate as the retentate flow increased.
24
Avir Modules With Air Feed
Nitrogen %
Retentate Nitrogen and Permeate Flow
86
85
84
83
82
81
80
79
5 PSIG
10 PSIG
15 PSIG
Permeate, scf/min
Figure 25: Note that if the permeate flow is compared to the retentate flow, it
becomes apparent that in order to achieve separation the permeate must be
greater than the retentate. This directly relates to large pumping losses.
Work in Prospect
We concluded that these units, originally developed for oxygen enrichment, are,
in their present condition, not useful for our program of nitrogen enrichment.
Consequently, the elements were removed from the test bench. Completion of
data analysis and computational modeling for these runs is pending.
24
Appendix F. Prism Modules

Test Objectives
The Prism unit was the first new module to be tested under the current program.
Although this unit appeared promising for our application, little data was available
from the manufacturer that directly pertained to the operating conditions we
anticipated for operation with the target engine. Consequently, a full mapping out
of the modules permeation parameters was planned so that the results could be
analyzed with the computational model. These would lead to an initial
specification of minimum performance needed to achieve target engine
objectives.
Test Program and Parameters
The tests were completed using as feed gas fully filtered house air at indoor
ambient temperatures from 5 psig up to 50 psig. Pure gas tests with nitrogen
and carbon dioxide from 5 to 10 psig feed were conduced at ambient
temperatures. Permeate was collected at ambient pressure (less than about an
inch of water column pressure resulting from pressure drop across exit lines and
flow meter) retentate flows were set from about 1 to 30 SCF/min.
Table 3 gives the run conditions and sequence in which they were conducted.
Due to a gage limitation of 5 psi from feed to retentate, runs at higher flows were
restricted to about 23 SCF/min retentate. Runs were limited to 50-psig feed, as
this was the manufacturers upper operating pressure. All runs were carried out
with the permeate at about ambient pressure.
Test runs were initiated at 5 psig with the fully filtered house air. Generally,
three retentate flows were set at a given pressure. Then, the feed pressure was
changed and another set of retentate flows was run. This continued over a
series of tests up to 50-psig feed designed to detect noticeable degradation in
performance related solely to exposure to pressure. Running the modules
through a cycle of increasing pressures and returning to lower reference
pressures accomplished this. Further, tests were made with pure gases at 5, 10,
15, and 10 psig to establish a base line of permeate flow. Higher pure gas test
runs were not made, as these were the maximum we could supply at the needed
flows. Pure gas runs at these pressures were deemed adequate for later data
analysis.
24
Table 3: Runs Completed With Prism Alpha 5 Module in Sequence of Testing*
Feed Pressure, psig
Feed Gas
5
5
5
10
10
10
AIR
AIR
AIR
AIR
AIR
AIR
10
10
5
AIR
NITROGEN
NITROGEN
5
10
15
20
20
20
AIR
AIR
AIR
AIR
AIR
AIR
5
10
15
15
20
5
20
30
NITROGEN
NITROGEN
NITROGEN
NITROGEN
NITROGEN
AIR
AIR
AIR
5
30
30
40
40
AIR
AIR
AIR
AIR
AIR
5
AIR
40
AIR
40
AIR
40
AIR
50
AIR
50
AIR
50
AIR
* All runs at indoor ambient temperature, approximately 72 F and
permeate at ambient pressure
Run Number
(Series 1)
1-1
1-2
1-3
1-4A
1-4B
1-5
(Series 2)
4-1
4-2
4-3
(Series 3)
5-2
5-3
5-4
5-5A
5-5B
5-6
(Series 4)
6-1
6-2
6-3
6-3A
6-3B
6-6
6-7A
6-7B
(Series 5)
7-1
7-2
7-3
7-4
7-5
(Series 6)
8-1
8-2
8-3
8-4
8-5A
8-5B
8-6
24
Suppliers Data
Prism Membranes is the research division of Air Products and a manufacturer of
nitrogen generation systems. Prism Membranes provides a wide variety of
designs, sizes, and membrane materials to OEM partners who are interested in
integrating membranes into their equipment to provide additional value to their
customers. It was thought that Prism Membranes would be a promising supplier
because of their previous work with nitrogen enriched intake air for diesel
engines. At our request, Prism membrane analyzed several membrane types
with 5.25 inch diameter membrane bundles.
Prisms review of our application led them to consider one of their low selectivity,
high permeability membranes and to resize it to meet flow requirements. Their
analysis concluded that If the low selectivity, high permeability prism membrane
modules are sized at 18 inches in diameter and the maximum flow is corrected to
70 lb/min, 2 membrane bundles are needed. With advanced coatings and
optimization this could be reduced to one bundle. (from email from Mr.Charles
Page of Prism Membranes in to Mr. McConnell of Argonne).
In the same email, Prism provided data with air feed at 40 psig, 70F, for
modules typical of the type described above. These data are presented in Table
4 for a 5.25 inch diameter by 24 inch long module. They qualified these
estimated data in the email, saying We assumed membrane permeation
properties which were much better than our existing product capability, but which
still might be achievable with a lot of development effort. Therefore, the
performance below is pretty optimistic given our current technology.
On the basis of their suggestion that this fiber in a more-developed stage might
meet the project requirements, Argonne requested that they provide a prototype
of the best fiber they currently could offer for testing during this program.
Table 4. Data Provided by Prism Membranes
Feed
lb/min
NEA
lb/min
0.984
1.08
1.22
1.429
1.766
2.399
Feed
NEA O2 % Permeate/ Permeate No. modules needed
Pres Drop from Mem Feed Ratio Pres
@ 70 lb/min
@ 40 lb/min @ 10 lb/min
psi
%
%
psig
NEA
NEA
NEA
0.468
0.563
0.701
0.907
1.239
1.863
0.23
0.27
0.31
0.38
0.49
0.69
17
17.5
18
18.5
19
19.5
52.7
48.2
42.8
36.8
30
22.5
-7.2
-7.2
-7.2
-7.2
-7.2
-7.2
150
125
100
78
57
38
86
72
58
45
33
22
22
18
15
12
9
6

The unit tested during this program was provided by Prism Membranes and
designated model Alpha, S/N 005, (Model #PP0376-DS). The case was rated at
50 psig maximum.
The unit was provided to Argonne at a cost of $3000.00. The Membrane bundle
was made from Prism membranes experimental polymers used in their internal
investigation of NEA for diesel engines.
24
The Prism Alpha 5 module was supplied in a single chamber with three ports.
Two of these ports were used to bring out connections to the modules feed and
retentate. The permeate was vented from the module through ports opening
from the circumference of the chamber. The unit measured 24-1/2 inches long
(full diameter distance) and about 4 inches in diameter. The module is shown in
Figure 22.
The module was obtained as the result of a meeting that took place between staff
of Prism Membrane, Mack Truck and Argonne National Laboratory. The major
concern expressed by Prism Membranes during this meeting was that if a
module supplied by Prism Membranes could not meet our requirements, they
would not be paid for their work. The solution to the funding problem was to
allow the development work to be rolled into the cost of the prototype. The final
outcome of the meeting was that Prism Membrane would perform analysis on
paper to look at the feasibility of their membrane.
Prism Membrane subsequently agreed to supply a prototype membrane module
for this study. The membrane fiber production line was shut down and enough
fibers for several prototype membrane modules were spun. After several
attempts to build a prototype, the sealing issues were solved and Prism
Membrane supplied a working membrane module to ANL for characterization.
This module is the best membrane they have for our application. They are
providing this membrane as an indicator of what Prisms membrane performance
is with respect to this application. It will also provide an indication of how much
development is needed to meet our performance criteria.
24
Permeate Ports
Figure 26: Prism Membrane prototype. Note the permeate ports have no provisions for
measuring permeate flow.
Module Mounting and Housing

.
The Prism Alpha 5 module was installed into an existing chamber (7-11/16
inches in diameter and 52-3/4 inches long) with three ports shown in Figure 27.
Two of these ports were used to bring out connections to the modules feed and
retentate. The permeate was vented from the module through ports opening
from its circumference into the chamber. On the side of the chamber was
located a single port which discharged to a permeate flow meter and then vented
to the atmosphere. This chamber had the same body and ends as used for
earlier tests with the Avir modules, but porting at the ends was significantly
modified to accommodate the Prism Alpha 5. The re-piped unit resembles the
setup of the Avir modules.
24
Figure 27: Prism Alpha 5 module installed in chamber
The feed and retentates ends were fitted with O-Ring tubing fittings for inch
tubing as shown in Figure 28.
24
O-ring tube fitting
Figure 28: Prism Alpha 5 module fitted with inch tubing fitting.
24
Bored through O-ring fitting
Figure 29: End flange with bored through o-ring fitting
As shown in Figure xx, the chamber end caps were fitted with inch bored
through O-ring tubing fittings. This allowed flexibility in locating the module
inside the chamber, minimizing any critical dimensions and allowing easy
assembly and disassembly. During assembly, a inch tube section was fitted to
each end of the module. These penetrated the end caps through these bored
through fittings and led to external feed and retentates lines. Figure 30 shows
the module in place just prior to installation of the end cap.
24
Port extension tube
Figure 30: Module in place prior to installation of an end cap

The data developed with the Prism module demonstrates that to achieve a given
nitrogen enrichment, higher feed pressures will yield higher throughputs. These
are substantially higher than those obtained from the likely-degredated Avir unit
previously tested.
The Prism unit was not labeled by the manufacturer with throughput performance
levels. We found that the unit could readily produce our target level of 81%
nitrogen in the retentate at pressures as low as 5 psig feed. However, as the
feed pressure was increased, we could achieve increasingly higher throughputs
while maintaining this target nitrogen concentration.
As feed pressure was increased from 5 to 50 psig, the data shown in Figure 2
indicate that for given retentate flow, the unit gave increased nitrogen
composition. For example, at 10 SCF/min the nitrogen composition in the
retentate was about 79.5 % at 5 psig feed while at 50 psig feed, for the same
retehtate flow, the retentate was over 85% nitrogen.
24
Prism Alpha 5 Modules With Air Feed

Retentate Nitrogen,% and Retentate Flow
89
Nitrogen %
5 PSIG
.
10 PSIG
84
50 PSIG
30 PSIG
20 PSIG
79
0
10
15
20
25
Retentate, scf/min
Figure 31: At each test pressure, the membrane achieved good separation at low retentate flows
and retained significant ability to separate as the retentate flow was increased.
As shown in Figure 3, retentate nitrogen percentage over most of the test

nitrogen composition test range resulted in a permeated gas flow primarily
related to module feed pressure. Pure gas tests with nitrogen showed a close
relationship to the feed pressure. This illustrates the trade off between
increasing feed pressure to gain higher nitrogen enrichment and increased power
required as a result of gas lost to permeate.
24

Retentate Nitrogen and Permeate Flow
91
Nitrogen, %
89
5 PSIG
87
10 PSIG
85
.
30 PSIG
83
50 PSIG
81
79
Pure N2 5#
0
Permeate, SCF/min
8
Pure N2 10#
Figure 32: As the operating pressure was increased, approximately proportional

increases in permeate flow were measured. This illustrates the trade-off
between the higher throughput gained (Figure 2) at higher pressures and
increased permeate flow (horsepower required).
Preliminary power estimates to achieve given levels of nitrogen enrichment are
shown in Figure 4. Here the pressure loses along the membrane (feed to
retentate) and across (feed to permeate) the membrane are combined to give a
horsepower requirement for operation at feed pressures of 5, 10, 20, 30, and 50
psig. For the target enrichment of 81% nitrogen in the retentate, the highest
throughput measured was about 23 SCF/min at about 1.4 horsepower. Further
analysis of the data and model development are expected to enable us to refine
the power required for target enrichment and estimate corresponding module
volumes needed.
Utilizing the data developed with the Prism module, estimates of form factors and
power requirements were made using feed to permeate and feed to retentate
flows and pressures using 70% isentropic compression. From analysis of the
power requirements of these two streams, we believe that module fiber
configuration could be modified to gain significant reduction in horsepower. We
expect to confirm these estimates as we continue lab tests with the next
modules.
24

Retentate Nitrogen,% and Total Horsepower
89
Nitrogen %
5 PSIG
10 PSIG
50 PSIG
84
.
30 PSIG
20 PSIG
79
0.0
0.5
1.0
1.5
Total Horsepower
Figure 4. Total Horsepower Calculated To Yield Given Percentage Of Retentate

N2
Our preliminary conclusion is that this type of module, if its performance is
extrapolated to the size and form factor required for the target engine, could
marginally meet the target requirements. Our continuing data analysis and
computer modeling are expected to confirm this conclusion and provide us with
benchmark membrane parameters that will need to be met to achieve the target
performance. We expect that these parameters of required separation factors
and permeation rates will guide our selection of future membrane materials and
predict module size and form factor combinations that would be needed to meet
the target goal.
Now that the second module testing has been completed, the test bench is being
readied for the next module, a Parker-Hanifin unit that is in hand. This unit has a
higher/lower total pressure rating compared with the Prism unit and the overall
volume is roughly the same. The form factors are quite different, the current
Prism being 3 times larger diameter and about a third of the length of the ParkerHanifin unit. As the re-piping is carried out for this third module, some
modifications will be made to the bench to allow a second module to remain in
24
position while a main module is tested. This will allow us to proceed with the new
module while holding the current module in place in the case some additional
tests are deemed necessary following complete data analysis. At this time,
however, we believe that the current Prism module has been completely
characterized to allow us to complete data analysis and determine the modules
anticipated suitability for the target 440 hp diesel truck engines.
During these tests, nitrogen concentrations upwards of 90% were achieved at
lower retentate flows and reject flows of over 20 SCF/min were sustained while
producing retentate having over 80% nitrogen. Initial horsepower estimates were
also made combining the pressure drop along the module (feed to retentate) and
across the membrane (feed to permeate). These results show this single module
element capable of providing about 3 % of the intake flow needs of the target
engine (1024 SCF/min) at upwards of 3% enrichment of nitrogen (82%)while
consuming 1.5 total horsepower. Preliminary extrapolation of these results to the
target engine requirements suggest that the performance of a cluster of these
modules would provide marginal performance. For example, if the module feedto-retentate pressure drop were reduced by fiber reconfiguration and the effective
separation factor was slightly increased, a target package of two 18 inch
diameter bundles of 24 inch length with a 50 psig feed could be achieved at an
estimated horsepower less than about 25. Our next module candidate (from
Parker-Hanifin) is expected, based upon data supplied by the manufacturer, to
exceed the parameters of the current Prism unit under test. These estimates and
actual performance levels will be confirmed during our next lab tests.
24
Appendix G. Parker Hanifin and CMS Modules

Test results with the Parker-Hanifin module showed generally improved
performance compared with the Avir and Prism modules tested previously. This
module provided more than adequate nitrogen enrichment with retentate flows
higher and permeate flows as low or lower than modules previously tested. With
the fibers and form factor as currently configured, we believe that this type of
module could meet the intake NEA requirements set by our CRADA partner,
Mack Truck, Inc. for a 440 hp diesel truck engine. However, as it is currently
configured, we deem it not acceptable to reach the target energy requirements
they specify. We believe this module does show promise because it was not
designed specifically for our application but was able to produce the required
target nitrogen level at the anticipated turbocharger operating pressures.
Since the last report, the second module type tested, a Prism module (Alpha 5)
manufactured by Air Products, was left in place in the test bench and decoupled
from the bench gas flow system. Then, the Parker-Hanifin module was installed
in the test bench and runs were carried out at feed pressures of 5, 20 and 30
psig. Following these tests, the module manifold was modified to allow
installation of a module from CMS, Inc. This was completed and initial runs were
undertaken at 5 and 10 psig feed pressure. All tests were conducted at indoor
ambient temperatures with the permeate at ambient pressure.
The experimental work planned for the Parker-Hanifin module has now been
sufficiently completed to allow data reduction and evaluation. These results are
presented in this report.
During these tests with the Parker-Hanifin module, nitrogen concentrations up to
almost 90% were achieved at retentate flows of over 2 SCF/min (20 psig feed).
At the target 83% nitrogen NEA the module produced over 16 SCF/min (30 psig
feed). The permeate flows ranged from about 1 to 5 SCF/min over the range of
testing (5 to 30 psig feed).
Initial horsepower estimates were made combining the pressure drop along the
module (feed to retentate) and across the membrane (feed to permeate). These
results show this single module element requires about 0.05 horsepower per
SCF/min of 83% nitrogen NEA produced at 5 psig feed pressure. Considering
that the target engine requires 866 SCF/min (70 lbs/min) of 83% NEA, this
module, if packaged in a set and run in its current configuration, would require
about 43 horsepower- a figure greater than the estimated acceptable horsepower
for the Mack Truck target engine allocation of 23 or less.
Because the permeation rate of the CMS, Inc. module exceeded the bench test
measuring capacity for the permeate flow when feed pressures exceeded 10
psig, the data set for this module was not completed. We plan to resize the flow
measurement equipment and then develop data at the higher feed pressures.
24
Test Bench With CMS, Inc. and Parker-Hanifin Modules

The bench was fitted with the CMS and Parker-Hanifin modules in parallel and a
set of valves enabling switching between the two for individual module testing.
To accomplish this, a set of manifolds was installed, one for each of the feed,
permeate, and retentate flow lines. The current configuration with the three
modules in place is shown in Figure 1.
Figure 33: Membrane test bench with Parker-Hanifin module
For these runs, feed air was obtained from a house air compressor at feeds up to
50 psig. The extra filtration provided by the bench assured removal of vapors as
well as particulates and droplets.
Test Program and Module
The second module set tested during this program, manufactured by Prism, was
kept on the bench but the piping to it was removed. A manifold was installed to
accommodate the immediate installation of the module procured from ParkerHanifin, (Parker Filtration Type ML, 71700-L1001, Serial number 070466). The
element measured 65 inches long and 3 inches in diameter except at the end
caps which were 4-1/2 inches in diameter. Permeate was withdrawn from two
side-arm ports manifolded together.
Following leak testing, test runs were initiated with the Parker-Hanifin module at
5 psig feed with fully filtered house air. All runs were conducted at indoor
ambient temperatures. Permeate was flowed through a flow meter and vented to
24
atmosphere. Generally, three retentate flows were set at a given pressure.

Then, the feed pressure was changed and another set of retentate flows was run.
This continued over a series of tests up to 30-psig feed designed to detect
noticeable degradation in performance related solely to exposure to pressure.
Running the modules through a cycle of increasing pressures and returning to
lower reference pressures accomplished this.
Subsequently, a module obtained from CMS, Inc. (Designated IMS #72, 591-76)
was mounted on the bench and plumbed up to the manifold. This module was
prepared and tested under similar conditions, but only at 5 and 10 psig feed
pressures. Because of the high permeate flows encountered as the feed
pressure was increased beyond 10 psig, we will need to modify the flow
measurement equipment to accommodate the higher flows.
Tests Completed
Tests with the Parker-Hanifin module were completed at pressures of 5, 20, and
30 psig with air at indoor ambient temperature. All runs were carried out with the
permeate at about ambient pressure (less than about an inch of water column
pressure resulting from pressure drop across exit lines and flow meter).
Tests with the CMS module were put on hold pending modification of the
permeate flow measurement equipment to accommodate the higher-thananticipated permeate flows encountered at feed pressures about 10 psig.
Test Results
Data analysis was completed for the Parker-Hanifin module. Results indicate
that the module could be repeatedly operated up to 30 psig feed pressure and up
to 5 psi feed to retentate pressure drop without indication of short term
deterioration of performance. During the test runs, air could be upgraded to near
90% nitrogen. Data analyzed to date are seen in graphic form in Figures 1 and
2.
24
Parker 070466 Module

Retentate Nitrogen, %
Permeate at Ambient Temperature and Pressure
89
Air Feed
Pressure
5 PSIG
20 PSIG
30 PSIG
84
79
0
10
15
20
Retentate Flow, scf/min
25
Figure 34: As feed pressure was increased, the membrane module produced increased
flow of retentate (NEA) for the same retentate composition.

The data developed with the Parker-Hanifin module demonstrates, as in the case
of the Prism module previously tested, that to achieve a given nitrogen
enrichment, higher feed pressures will yield higher throughputs. However, there
is a trade-off with the amount of horsepower estimated to be taken across and
through the module.
While the Parker-Hanifin unit was not labeled by the manufacturer with
throughput performance levels, we had been assured that it would meet our
nitrogen concentration requirements. As tested, it could readily produce our
target level of 83% nitrogen in the retentate at pressures as low as 5 psig feed.
However, as the feed pressure was increased, we could achieve increasingly
higher throughputs while maintaining this target nitrogen concentration.
As feed pressure was increased from 5 to 30 psig, the data shown in Figure 2
indicate that for our target nitrogen composition of 83%, the retentate flow
increased from about 2 to about 20 SCF/min as the feed pressure was increased
from 5 to 30 psig.
24
Over this same range of feed pressures, as shown in Figure 3, the permeate
flow increased from about 1 to about 5 SCF/min.

Retentate Nitrogen, %
89
Air Feed
Pressure
5 PSIG
84
20 PSIG
30 PSIG
79
0
Permeate Flow, scf/min

Figure 35: As the feed pressure was increased, approximately proportional increases in
permeate flow were measured. This illustrates the trade-off between the higher
throughput gained (Figure 2) at higher pressures and increased permeate flow
(horsepower required).
Power estimates to achieve given levels of nitrogen enrichment are shown in

Figure 4. Here the pressure loses along the membrane (feed to retentate) and
across (feed to permeate) the membrane are combined to give a horsepower
requirement for operation at feed pressures of 5, 20, and 30 psig. For the target
enrichment of 83% nitrogen in the retentate, the highest throughput measured
was about 15 SCF/min at about 0.1 horsepower (70% isentropic compression)
per scf/min of retentate NEA produced. Further analysis of the data and model
development are expected to enable us to further refine the power required for
target enrichment and estimate corresponding module volumes needed.
24

Retentate Nitrogen %
89
5 PSIG
84
20 PSIG
30 PSIG
79
0.0
0.1
0.2
0.3
Horsepower per unit NEA produced, hp per scf/min

Figure 4. Total horsepower calculated per unit retentate (NEA) flow to yield given
percentage of retentate N2
Tests in Prospect
After modification of our flow measurement for the permeate stream to
accommodate the higher flows we experienced with the CMS module, we plan to
complete testing of this module as scheduled.
24

Heavy-Duty Diesel Engine NO Reduction With Nitrogen-Enriched Combustion Air

Uploaded by

Copyright:

Available Formats

Heavy-Duty Diesel Engine NO Reduction With Nitrogen-Enriched Combustion Air

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Heavy-Duty Diesel Engine NO Reduction With Nitrogen-Enriched Combustion Air

Uploaded by

Copyright:

Available Formats

Final Report:

Heavy-duty Diesel Engine NOX Reduction

DOE monitor Gurpreet Singh

Table 1: Membrane Materials Survey Results, highlighted membrane materials

Selected Membranes for Oxygen/Nitrogen Separation

Not Commercially Available Today

Commercially Available Today

Figure 1: Membrane Materials Availability and Parameters

13 Mode Test for a 460 hp Mack V8 Engine

Engines performance curve

Engine Speed (RPM)

Baseline Data Analysis:

For an APPROXIMATE intake manifold oxygen

-12 ATDC BOI

Oxygen Concentration Reduction (%)

Figure 3: Brake specific NOx Emission with respect to EGR

Emissions Reduction with Respect to Oxygen Reduction and DI BOI Timing

Negative emissions reductions are actually

For an APPROXIMATE intake manifold oxygen

Emissions Reduction (%)

% CO Reduction (-12 ATDC BOI)

% NOx Reduction (-12 ATDC BOI)

% NOx Reduction (-9 ATDC BOI)

Retentate Flow, SCF/min

Retentate Flow vs Feed Pressure

Feed Pressure, psia

Membrane Area, sqft

Retentate Flow vs Membrane Area

Feed Pressure, psia

determined that a feasible membrane bundle composed of the same polymer

Appendix A Lab Test Facilities

Figure 8: New redesigned membrane test bench 22

pressures. To accommodate this, extra filtration was included to assure removal

Figure 9: Schematic of the Current Membrane Characterization Bench

Appendix B. System Configurations

Preturbocharger Upgrading With Vacuum Forced Permeate{Turbocharger

Figure 11: Preturbocharger Upgrading With Vacuum Forced Permeate

Post turbocharger Upgrading {Turbocharger ahead of module}

Postturbocharger Upgrading With Vacuum Forced Permeate{Turbocharger

Figure 13: Postturbocharger Upgrading With Vacuum Forced Permeate

Figure 9 (Postturbocharger Upgrading With Vacuum Forced Permeate) shows a

Appendix C. Field Test Facilities

13 Mode Test for a 460 hp Mack V8 Engine

Engines performance curve

Engine Speed (RPM)

-12 ATDC BOI

Oxygen Concentration Reduction (%)

Figure 15: Brake specific NOx Emission with respect to EGR.

Emissions Reduction with Respect to Oxygen Reduction and DI BOI Timing

Negative emissions reductions are actually

For an APPROXIMATE intake manifold oxygen

Emissions Reduction (%)

% CO Reduction (-12 ATDC BOI)

% NOx Reduction (-12 ATDC BOI)

% NOx Reduction (-9 ATDC BOI)

Appendix D. MEDAL Modules

Jan 31, 2002

Steve McConnell (Argonne)

Air Separation Membranes for NOx Reduction in Diesel Engines.