Cost and Performance For A Fueling Hydrogen Station

Cost and Performance Comparison Of
Stationary Hydrogen Fueling Appliances
Task 2 Report
April 2002
Prepared by:
Duane B. Myers, Gregory D. Ariff, Brian D. James, John S. Lettow,
C.E. (Sandy) Thomas, & Reed C. Kuhn
One Virginia Square

3601 Wilson Boulevard, Suite 650
Arlington, Virginia 22201
703/243-3383
Prepared for:
The Hydrogen Program Office

Office of Power Technologies
U.S Department of Energy
Washington, D.C.
Under
Grant No. DE-FG01-99EE35099
Cost and Performance Comparison Of Stationary Hydrogen Fueling Appliances Directed Technologies Inc.
FORWARD
This work was funded by the Hydrogen Program Office of the U.S. Department of Energy under
Grant No. DE-FG01-99EE35099 and represents the second task of three to be completed under
this contract. The first task presented a broad overview of the costs for creating infrastructures to
supply direct hydrogen, methanol, and gasoline to support fuel cell vehicles (FCV’s). A
conclusion of the report resulting from the first task was that “the costs of maintaining the
existing gasoline infrastructure per vehicle supported are up to two times more expensive than
the estimated costs of maintaining either a methanol or a hydrogen fuel infrastructure”.
The second task, as detailed in this report, was to provide a detailed analysis of the cost of
providing small-scale stationary hydrogen fueling appliances (HFA’s) for the on-site production
and storage of hydrogen from natural gas to fuel hydrogen FCV’s. Four potential reforming
systems were studied: 10-atmosphere steam methane reforming (SMR) with pressure-swing
adsorption (PSA) as gas cleanup, 20-atm SMR with metal membrane gas cleanup, 10-atm
autothermal reforming (ATR) with PSA gas cleanup, and 20-atm ATR with metal membrane gas
cleanup.
We acknowledge the support of the Department of Energy, especially Sig Gronich, DOE
Hydrogen Team Leader. We thank Phoebe Smith for the many aesthetic and instructive images
she created for this report.
i
Table of Contents
Page
FORWARD..................................................................................................................................... i
EXECUTIVE SUMMARY........................................................................................................... 1
1 INTRODUCTION.............................................................................................................. 5
2 DFMA METHODOLOGY................................................................................................ 6
2.1 Cost Estimation Using DFMA...................................................................................... 6
2.2 Production Volume Considerations .............................................................................. 7
2.3 Markup Rate Assumptions............................................................................................ 7
2.4 Design Considerations .................................................................................................. 9
2.5 Sample DFMA Cost Analysis..................................................................................... 10
3 REFUELING APPLIANCE HYDROGEN PRODUCTION RATE AND
MANUFACTURING QUANTITY................................................................................ 12
4 STEAM METHANE REFORMING VS. AUTOTHERMAL REFORMING ........... 13
4.1 Overview of SMR and ATR ....................................................................................... 13
4.1.1 Natural Gas Compression.................................................................................... 14
4.1.2 Natural Gas Purification ...................................................................................... 15
4.1.3 Catalytic Steam Reforming ................................................................................. 16
4.1.4 Water-Gas Shift Reaction.................................................................................... 18
4.2 SMR System Flowsheets, Cost and Performance....................................................... 18
4.2.1 Reactor and Vessel Sizing ................................................................................... 21
4.2.2 Heat Exchanger Sizing ........................................................................................ 23
4.2.3 Water Purification................................................................................................ 24
4.2.4 Other Equipment ................................................................................................. 24
4.2.5 Assembly ............................................................................................................. 26
4.2.6 SMR Reformer System Bill of Materials and Cost ............................................. 26
4.2.7 SMR Performance and Efficiency....................................................................... 26
4.3 ATR System Flowsheets, Cost and Performance ....................................................... 33
4.3.1 Reactor and Vessel Sizing ................................................................................... 33
4.3.2 Heat Exchanger Sizing ........................................................................................ 35
4.3.3 Water Purification................................................................................................ 36
4.3.4 Other Equipment ................................................................................................. 36
4.3.5 Assembly ............................................................................................................. 37
4.3.6 Alternate ATR Designs ....................................................................................... 37
4.3.7 ATR Reformer System Bill of Materials and Cost.............................................. 37
4.3.8 ATR Performance and Efficiency ....................................................................... 42
5 GAS CLEANUP TECHNOLOGIES.............................................................................. 42
5.1 Pressure Swing Adsorption (PSA).............................................................................. 42
5.1.1 PSA Gas Separation ............................................................................................ 42
5.1.2 PSA Design ......................................................................................................... 44
5.1.3 PSA System Cost................................................................................................. 49
5.2 Metal Membranes ....................................................................................................... 50
ii
5.2.1 Metal Membrane Gas Separation ........................................................................ 50

5.2.2 Hydrogen Permeability in a Reformer System Membrane Assembly ................ 56
5.2.3 Membrane Material Cost ..................................................................................... 58
5.2.4 Membrane Durability .......................................................................................... 64
5.2.5 Membrane System Total Cost ............................................................................. 67
5.3 Preferential Oxidation (PrOx)..................................................................................... 72
5.4 Gas Cleanup Summary Comparison........................................................................... 73
5.4.1 Effect on Hydrogen Compressor ......................................................................... 73
5.4.2 Gas Cleanup Selection and Cost Summary ......................................................... 74
6 HYDROGEN COMPRESSORS..................................................................................... 74
6.1 Existing Industrial Hydrogen Compressors ................................................................ 75
6.2 Small-Scale Innovative Natural Gas and Hydrogen Compressors ............................. 78
6.3 Service Station Hydrogen Compressor Design........................................................... 78
6.4 Compressor System Assembly and Test..................................................................... 90
6.5 Hydrogen Compressor Cost Summary ....................................................................... 93
7 STATIONARY STORAGE OF COMPRESSED HYDROGEN................................. 94
7.1 Compressed Hydrogen Storage Requirements ........................................................... 94
7.2 Types of Liners ........................................................................................................... 96
7.2.1 Metal Liners......................................................................................................... 97
7.2.2 HDPE Polymer Liners ......................................................................................... 97
7.2.3 Metallized Polymer Bladders .............................................................................. 98
7.3 Types of Fiber ............................................................................................................. 99
7.4 Stress Rupture and Safety Factor.............................................................................. 100
7.5 Current Pressure Vessel Pricing................................................................................ 101
7.6 Projected Carbon Composite Pressure Vessel Pricing.............................................. 102
7.7 Valving and Connection Hardware........................................................................... 106
7.8 Storage System Cost Summary................................................................................. 107
8 DISPENSERS ................................................................................................................. 108
9 TOTAL COST OF SMR- AND ATR-BASED STATIONARY FUELING
APPLIANCES ............................................................................................................... 109
9.1 Capital Cost Summary .............................................................................................. 109
9.1.1 Miscellaneous Capital Costs.............................................................................. 109
9.1.2 Total Capital Cost.............................................................................................. 110
9.2 Hydrogen Cost .......................................................................................................... 111
9.2.1 Method for Comparison of Cases...................................................................... 111
9.2.2 Operating Cost Assumptions ............................................................................. 113
9.2.3 Results ............................................................................................................... 115
10 CONCLUSIONS ............................................................................................................ 119
APPENDIX: Cost Estimate for Scale-up of Small-Scale HFA ............................................ 121
iii
List of Figures
Page
Figure 2-1. Approximate G&A, OH, and R&D Expenses of Prototypical HFA Company .......... 9
Figure 2-2. HFA Manufacturer Markup Rates............................................................................... 9
Figure 2-3. Detailed Cost Analysis of a Boiler Tube-Sheet......................................................... 10
Figure 2-4. Sample Assembly Analysis for a Boiler.................................................................... 11
Figure 2-5. Sample Bill of Materials for a Boiler ........................................................................ 11
Figure 3-1. HFA Production Parameters...................................................................................... 12
Figure 4-1. Natural Gas Composition .......................................................................................... 14
Figure 4-2. Two-Bed HDS System with Natural Gas Preheater (not to scale)............................ 15
Figure 4-3. HDS Pipe Section (not to scale)................................................................................ 16
Figure 4-4. SMR Process Conditions........................................................................................... 18
Figure 4-5. SMR Flowsheet ......................................................................................................... 20
Figure 4-6. HDS Preheater, with Tube-in-Tube Design (not to scale) ........................................ 21
Figure 4-7. SMR Reformer Vessel (not to scale)......................................................................... 21
Figure 4-8. Tubes and Tube-Sheets for Shell-and-Tube Vessel .................................................. 22
Figure 4-9. Proposed Layout of 10 atm SMR Reformer System ................................................. 25
Figure 4-10. Bill of Materials for 10 atm SMR RS...................................................................... 27
Figure 4-11. Bill of Materials for 20 atm SMR RS...................................................................... 30
Figure 4-12. ATR Process Conditions ......................................................................................... 33
Figure 4-13. ATR Flow Sheet ...................................................................................................... 34
Figure 4-14. ATR Vessel (not to scale) ....................................................................................... 35
Figure 4-15. Bill of Materials for 10 atm ATR RS ...................................................................... 38
Figure 4-16. Bill of Materials for 20 atm ATR RS ...................................................................... 40
Figure 5-1. PSA Flow Diagram, Showing Valve Configuration ................................................. 45
Figure 5-2. Batta Cycle Timing Chart.......................................................................................... 45
Figure 5-3. Parameters Used to Calculate Manufacturing Cost of the PSA systems .................. 49
Figure 5-4. Hydrogen Permeability of Selected Metals............................................................... 52
Figure 5-5. Neat Hydrogen Permeation Through Palladium Alloys............................................ 54
Figure 5-6. Hydrogen Permeation with Gas Mixtures, ................................................................ 55
Figure 5-7. Geometry for a Flat Metal Membrane Permeating Hydrogen from a Mixed Gas
Stream .................................................................................................................................... 56
Figure 5-8. Historical Prices for Platinum and Palladium ........................................................... 58
Figure 5-9. Estimated Material Costs of a 25-micron Palladium Membrane System to Cleanup
the Reformer Gas for a Stationary Fuel Cell System ............................................................ 60
Figure 5-10. Estimated Cost of a 12-micron Palladium Membrane ............................................ 61
Figure 5-11. Approximate Market Prices for Various Hydrogen-Permeable Metals .................. 64
Figure 5-12. Thermal expansion of Pd24Ag tubes with and without hydrogen .......................... 66
Figure 5-13. Proposed Membrane Assembly............................................................................... 70
Figure 5-14. Parameters Used to Calculate Manufacturing Cost of a Planar Membrane Gas
Cleanup System ..................................................................................................................... 71
Figure 5-15. Gas Cleanup Summary............................................................................................ 74
Figure 6-1. Hydrogen Compressor Attributes.............................................................................. 75
Figure 6-2. Typical Chemical Process Industry Compressor Application Ranges ...................... 76
Figure 6-3. Simplification of Typical Compressor Crankcase Elements..................................... 77
Figure 6-4. Norwalk 150 H2 Compressor..................................................................................... 77
Figure 6-5. Notional Hydrogen Compressor................................................................................ 81
iv
Figure 6-6. Bill of Materials - High Inlet Pressure Hydrogen Compressor ................................. 82
Figure 6-7. Bill of Materials - Low Inlet Pressure Hydrogen Compressor.................................. 84
Figure 6-8. Plate Valve................................................................................................................. 87
Figure 6-9. Cutaway of Damped Plate Valve .............................................................................. 88
Figure 6-10. Compressor and Blowdown Vessel Assembly........................................................ 90
Figure 6-11. Interstage Cooler Manufacture and Assembly ........................................................ 91
Figure 6-12. Skid Assembly and Mounting ................................................................................. 92
Figure 6-13. Compressor System Test ......................................................................................... 93
Figure 6-14. Hydrogen Compressor Cost Summary (4.8 kg/h rated capacity)............................ 94
Figure 7-1. Dispensable Hydrogen Storage Requirement Based on 69% Capacity Factor ......... 95
Figure 7-2. NGV2 Mandated Safety Factors for Composite Wrapped Pressure Vessels .......... 101
Figure 7-3. Summary of Pressure Vessel Price Quotes ............................................................. 102
Figure 7-4. Tank Cost Estimate Details ..................................................................................... 103
Figure 7-5. Connection Schematic for Large Steel Tanks ......................................................... 106
Figure 7-6. Connection Schematic for ASME Tanks and Composite Tanks............................. 107
Figure 7-7. Summary of Hydrogen Storage Systems................................................................. 108
Figure 9-1. HFA Costs for a 115 kg H2/day System.................................................................. 110
Figure 9-2. Discounted Cash Flow Calculation Assumptions ................................................... 111
Figure 9-3. Natural Gas Cost to Commercial Customers, 1983-2001. ...................................... 114
Figure 9-4. Cost of Hydrogen from Four HFA Options ............................................................ 116
Figure 9-5. SMR-PSA Cost of Hydrogen .................................................................................. 116
Figure 9-6. ATR-PSA Cost of Hydrogen................................................................................... 117
Figure 9-7. SMR-Membrane Cost of Hydrogen ........................................................................ 117
Figure 9-8. ATR-Membrane Cost of Hydrogen......................................................................... 118
Figure A-1. Scaleup Factors Applied to SMR-PSA System for 8-Times Capacity................... 120
Figure A-2. Estimated Cost of Hydrogen for 1464-Vehicle HFA ............................................. 121
v
EXECUTIVE SUMMARY
Over several studies, Directed Technologies, Inc. has analyzed the costs of representative
hydrogen fueling stations (HFA’s) to supply the early-introduction hydrogen powered fuel cell
vehicles (FCV’s) and the cost of hydrogen produced by these HFA’s. In previous studies we
evaluated the impact of fuel choice on FCV, the cost of other sizes and quantities of HFA’s, and
the infrastructure maintenance costs of various fuels. In this study we analyzed the costs for an
intermediate production rate (250/year) of HFA’s sized to support communities of 183 vehicles
each1 (about one-eighth the size of the current average gasoline station). This small HFA is
chosen to allow economical hydrogen production in the early years when there are low numbers
of FCV’s present in any geographical area. While the focus of this report is on the economics of
hydrogen production at this small unit size, it is noted that significant hydrogen cost reductions
can be achieved by scaling the HFA unit to a larger size. The Appendix to this report contains
details of an eight-fold capacity HFA that results in a 45% reduction in the cost of hydrogen.
For the baseline HFA, we compare the costs and efficiencies of two hydrogen-generation
technologies (steam methane reforming and autothermal reforming) and two hydrogen
purification technologies (pressure swing adsorption and metal membrane). Based on this study
we conclude that the most cost-effective option as determined by the wholesale cost of
hydrogen is steam methane reforming (SMR) with pressure swing adsorption (PSA) hydrogen
purification. The initial capital cost to install the preferred SMR-PSA to support 183 vehicles is
$253,014 per unit. The wholesale cost of hydrogen for this option including storage and
dispensing but excluding sales taxes2 and retail markup is $3.38/kg, or $1.55 per gallon of
gasoline equivalent.3 Autothermal reforming (ATR) of natural gas is a lower initial-cost option,
but the resulting cost of hydrogen is higher ($3.59/kg) because the ATR is less efficient than the
SMR. The capital costs for the four primary options studied, assuming a ten-year lifetime, are
listed in Figure 1. For reference, the maximum hydrogen production4 required to support 183
vehicles is 115 kg/day, or 2,000 standard cubic feet per hour (scfh).
The range of capital costs, $225,000 to $275,000 depending on the HFA option, corresponds to a
total annual investment of $56.25-$68.75 million per year to support the introduction of ~50,000
new FCV’s per year.
This study indicates that with current technology, pressure swing adsorption (PSA) is more
cost effective and reliable than metal membrane hydrogen purification. The higher costs of
membrane units relative to PSA’s are not justified by the potential for better hydrogen recovery
and smaller size.
1
The HFA is sized to permit refueling of approximately 20 fuel cell vehicles each day. Because the vehicles don’t need to refuel each day, each
HFA is able to support the entire hydrogen needs of a community of 183 vehicles.
2
“Sales taxes” refer to federal and state highway taxes.
3
The conversion to gasoline equivalent will depend on the efficiency benefit of FCV’s over conventional internal combustion vehicles at the time
of introduction. The initial efficiency benefit of a FCV is estimated to be roughly 2.2 times better than an internal combustion engine vehicle on
a lower heating value basis tank-to-wheels on a 1.25 accelerated combined EPA driving schedule, with the potential for realistic improvements in
the FCV design to increase the efficiency benefit to at least 2.67. The gasoline-equivalent cost represents the 2.2 efficiency benefit.
4
The HFA is sized to produce 115kg/day if run at full power for the full day. However, a utilization factor of 69% is used to adjust for seasonal,
and diurnal fluctuations in hydrogen demand. Thus the 183 vehicles supported by the station is a true measure of how many vehicles can be fully
accommodated by the HFA.
1
A Design for Manufacturing and Assembly (DFMA™) costing approach was used to obtain cost
estimates. This methodology is used extensively by industry for product cost estimation and to
compare the relative cost of competing manufacturing and assembly approaches. The DFMA
methodology is both a rigorous cost estimation technique and a method of product redesign to
achieve lowest cost. The DFMA approach used for this analysis provides a solid framework for
the cost study and is the only fair way to compare the cost of potential HFA configurations.
A breakdown of factors making up the capital costs is provided in Figure 1. The combined costs
for hydrogen compression, storage, and dispensing are roughly equal to the cost for the reformer
and purification system. Capital recovery (i.e., amortization of the initial investment over the life
of the HFA) accounts for ~48% of the cost of hydrogen for SMR and ~40% of the cost of
hydrogen for the ATR.
$300,000
$273,359
$253,014 $247,186
$250,000
$228,957
10% Contingency
$200,000 Miscellaneous
Compressor, Storage, Dispenser
Reformer System
Capital Cost
$150,000
$100,000
$50,000
$0
SMR + PSA ATR + PSA SMR + Membrane ATR + Membrane
Figure 1. Contribution of Subsystems to Capital Cost for 115 kg/day HFA’s. The “Miscellaneous” category
includes on-site installation, freight, taxes & insurance, and initial spares. The “Reformer System” category
includes the hydrogen production and gas cleanup subsystems.
The other contributors to the cost of hydrogen are the cost of natural gas, the cost of electricity,
operation and maintenance expenses (O&M), and taxes and insurance. The breakdown of the
cost of hydrogen with the contributions from each is given in Figure 2.
2
Costs in $/kg H2 SMR/PSA ATR/PSA SMR/Membrane ATR/Membrane

Hydrogen Cost5 $3.38 $3.59 $3.74 $4.28
Capital Recovery 1.66 1.50 1.78 1.62
Natural Gas 0.95 1.17 1.01 1.44
Electricity 0.23 0.41 0.37 0.68
O&M 0.33 0.31 0.33 0.33
Taxes & Insurance 0.23 0.20 0.24 0.22
Gasoline equiv.
$1.55 $1.65 $1.72 $1.96
($/gal)
HFA is assumed to run an average of 69% of capacity with 98% availability.
Capital Recovery assumes a 10% after-tax return on investment over its 10-year life. A 38% marginal tax rate (34%
federal, 4% state and local) is included in the return on investment calculation.
Natural gas price is based on the 19-year national average commercial rate of $5.34 per thousand scf.
Electricity price is based on the 10-year national average commercial rates of 7.5¢ per kW-hr.
The cost for water usage is negligible.
O&M includes yearly hydrogen desulfurization bed replacement and reformer and shift catalyst replacement after
five years. It also includes general maintenance for compressors, valves, etc.
Tax and Insurance costs refer to annual property taxes at 1.5% of capital investment and annual insurance premiums
at 1% of capital investment. Highway/road sales taxes are not included.
Gasoline equivalent price is based on an efficiency gain of 2.2 for hydrogen FCV’s over current gasoline ICEV’s.
Figure 2. Cost of Hydrogen Produced from the 2,000 scfh HFA Options
We conclude that the wholesale cost of hydrogen produced from early-year HFA’s (i.e. 2000scfh
HFA’s produced in 250 per year quantities) will be nearly competitive with the retail price of
gasoline on a per vehicle-mile basis, especially in regions with reformulated gasoline
requirements. We feel that this comparison of the untaxed cost of hydrogen with the taxed price
of gasoline is valid for the near to mid-term as hydrogen is unlikely to be taxed until it begins to
significantly displace gasoline road-tax revenues. When there are sufficient FCV’s to justify a
larger number of higher-volume stations, the cost of hydrogen is expected to decrease
significantly by taking advantage of economies of scale.
An appendix to this report describes a brief scale-up analysis6, detailing the reduced hydrogen
cost that results by increasing the size of the HFA from 2,000 scfh to 16,000 scfh. An HFA of
this size would support roughly 1464 vehicles, which is comparable to current gasoline stations.
A breakdown of the estimated cost of hydrogen for this HFA is given in Figure 3. Using scale-
up factors common to chemical processes, the capital cost of this 8x HFA was estimated to be
$1.16 million, resulting in a hydrogen cost of $1.87-$2.48/kg (dependent on assumptions about
utility discounts, natural gas feedstock cost, and equipment life). Thus, the 1x HFA derived
hydrogen cost of $3.38/kg is appropriate when discussing the early introduction of fuel cell
vehicles where many small stations need to be distributed over the country to support a sparse
FCV population, and the significantly lower hydrogen cost of $1.87-$2.48/kg, resulting from an
8x HFA, is appropriate for the latter years when the FCV population and population density is
much higher.
5
Subtotals may not add to exactly the listed total cost due to rounding.
6
For this scale-up analysis, production volume was maintained at 250 HFA units/year. It should be noted that for larger production runs, cost
reductions beyond those indicated here would be likely.
3
Costs in $/kg H2 16,000scfh SMR/PSA HFA

Hydrogen Cost7 $1.87
Capital Recovery $0.77
Natural Gas $0.59
Electricity $0.15
O&M $0.24
Taxes & Insurance $0.13
Gasoline equiv. ($/gal) $0.85
Estimates are based on a scaled-up version of a 2,000scfh HFA. Scale-up may not retain accuracy of
original analysis.
HFA is assumed to run an average of 69% of capacity with 98% availability.
Capital Recovery assumes a 10% after-tax return on investment for a 15-year life. A 38% marginal tax rate
(34% federal, 4% state and local) included in the return on investment calculation.
Natural gas price is based on the 19-year national average industrial rate of $3.30 per thousand scf.
Electricity price is based on the 10-year national average industrial rates of 4.65¢ per kW-hr.
The cost for water usage is negligible.
O&M includes yearly hydrogen desulfurization bed replacement and reformer and shift catalyst
replacement every five years. It also includes general maintenance for compressors, valves, etc.
Tax and Insurance costs refer to annual property taxes at 1.5% of capital investment and annual insurance
premiums at 1% of capital investment. Highway/road sales taxes are not included.
Gasoline equivalent price is based on an efficiency gain of 2.2 for hydrogen FCV’s over current gasoline
ICEV’s.
Figure 3. Cost of Hydrogen from 16,000 scfh (8x) SMR/PSA HFA with Optimistic Assumptions
7
4
1 INTRODUCTION
Over the next several decades, hydrogen fuel cell vehicles (FCV’s) are expected to increasingly
replace gasoline powered internal combustion engine vehicles (ICEV’s) as the primary
automotive platform. Whether this transition occurs rapidly or not will depend on many factors
including FCV cost, performance, safety and perceived environmental impact. However, a
critical factor in the successful transition is the reality and perception of hydrogen fuel
availability. Any successful transition pathway must achieve a cost effective route for the supply
of hydrogen fuel to FCV’s during this transition period when there are few FCV’s spread out
over a large geographic area.
The most promising hydrogen supply pathway, more fully described in multiple past Directed
Technologies Inc. reports8,9, consists of small scale natural gas reformation units producing pure
hydrogen gas which is then compressed to >5,000 psi for dispensing to FCV’s. These Hydrogen
Fueling Appliances, or HFA’s, consist of
• a natural gas reformer unit,
• a gas cleanup unit to purify the reformer outlet to a pure 99.99+% hydrogen stream,
• hydrogen compressor (to allow onsite storage)
• onsite storage of the hydrogen (for reformer unit load leveling)
• and hydrogen dispenser to allow dispensing of the hydrogen into vehicular high pressure
storage tanks at 5,000psi.
The HFA units are designed for a deliberately low hydrogen production rate, approximately
2,000 scfh (115 kg/day) of hydrogen, so as to support small communities of fuel cell vehicles.
Each HFA can support approximately 183 FCV’s which is approximately one sixth the vehicle
fleet supported by a typical gasoline fueling station. This small size is chosen to allow
economical hydrogen production in the early years when there are low numbers of FCV’s
present in any one geographical area. If there are more FCV’s than one Hydrogen Fueling
Appliance can handle, a second, third or fourth HFA unit can be placed next to the first so as to
increase capacity. Thus the hydrogen fueling appliances are modular to allow incremental
growth in demand. Similarly, single HFA hydrogen stations may be distributed across the
country providing a network of hydrogen refueling opportunities.
While the above described transition pathway is plausible, it will only be successful if the
hydrogen produced from the HFA’s is economical. Consequently, the focus of this report is on
the cost estimation of the HFA. Past studies examining the capital cost of HFA’s have been for
larger or small units or at either very low or very high annual unit production rates. While those
estimates are useful, they do not answer the question of what the projected cost of HFA’s would
be if produced in moderate (250 units/year) quantities consistent with annual FCV production
rates of 50,000/year.
8
“Hydrogen Infrastructure Report”, C.E. Thomas, Brian D. James, Ira F. Kuhn, Jr., Franklin D. Lomax, Jr.,.and George N. Baum, all of Directed
Technologies Inc., prepared for the Ford Motor Company as part of Ford’s “Direct-Hydrogen-Fueled Proton-Exchange-Membrane Fuel Cell
System for Transportation Applications” contract (Contract No DE-AC02-94CE50389) to the U.S. Department of Energy, Office of
Transportation Technologies, July 1997.
9
“Distributed Hydrogen Fueling Systems Analysis”, C.E. Thomas, John P. Reardon, Franklin D. Lomax, Jr., Jennifer Pinyan, Ira F. Kuhn, Jr.,
prepared for The Hydrogen Program Office, Office of Power Technologies, U.S. Department of Energy, under Grant No. DE-FG01-99EE35099,
October 2000.
5
Additionally, this report examines multiple natural gas reformation chemical pathways (Steam
Methane Reforming (SMR), Autothermal Reforming (ATR)) and multiple gas cleanup methods
(Pressure Swing Adsorption (PSA), membrane separation, Preferential Oxidation (PrOx)) to
determine the most cost effective approach. Each system requires a careful chemical and
mechanical engineering analysis to capture the appropriate performance parameters and cost
factors. Once moderately detailed mechanical designs and material and energy balances are
created for all system components, a complete system Bill of Materials is generated. This Bill of
Materials allows a line-by-line, element-by-element cost assessment to be conducted.
A Design of Manufacturing and Assembly (DFMA) costing approach is used to obtain cost
estimates. This methodology, described more fully in the next section, is used extensively by
industry for product cost estimation and to compare the relative cost of competing manufacturing
and assembly approaches. The DFMA methodology is both a rigorous cost estimation technique
and a method of product redesign to achieve lowest cost. The DFMA approach used within the
report provides a solid framework for this cost study and is the only fair way to compare the cost
of competing HFA approaches.
2 DFMA METHODOLOGY
2.1 Cost Estimation Using DFMA
The cost estimation methodology employed in this report is based on the Design for Manufacture
and Assembly (DFMA) techniques developed by Boothroyd and Dewhurst, described in Product
Design for Manufacture and Assembly, 2nd edition.10 The DFMA process has been formally
adopted by the Ford Motor Company (among others) as a systematic means for the design and
evaluation of cost-optimized components and systems. These techniques are powerful and are
flexible enough to incorporate historical cost data and manufacturing acumen that have been
accumulated by Ford since the earliest days of the company. Directed Technologies has adapted
and expanded the formal DFMA technique to include lessons from Ford and its own experience
to develop a system of tools and methods for cost estimation of engineering designs.
The DFMA methodology was used to estimate costs for equipment for which high-volume
manufacturing methods may not currently exist. Detailed manufacturing and assembly methods
were designed to “construct” the equipment from the ground up as much as possible rather than
using factored estimates that are common at this level of capital cost estimation. For off-the-
shelf equipment such as valves and instruments, a catalog or quoted price with volume discount
was used.
The cost of any component includes direct material cost, manufacturing cost, and assembly cost.
Direct material costs are determined from the exact type and mass of material employed in the
component. This cost is usually based on either historical volume prices for the material or
vendor price quotations. In the case of materials not widely used at present, the manufacturing
process must be analyzed to determine the probable high-volume price for the material. The
10
Boothroyd, Geoffrey, Peter Dewhurst, Winston Knight. Product Design for Manufacture and Assembly, Second Edition. Marcel Dekker, Inc.,
New York. 2002.
6
manufacturing cost is based on the required features of the part and the time required to generate
those features in a typical machine of the appropriate type. The cycle time can be combined with
the “machine rate” (the hourly cost of the machine based upon amortization of capital and
operating costs) and the number of parts produced per cycle to yield an accurate manufacturing
cost. The assembly costs are based upon the amount of time to complete the given operation and
the cost of either manual labor or of the automatic assembly process train. The piece cost
derived in this fashion is quite accurate as it is based upon an exact physical manifestation of the
part and the technically feasible means of producing it as well as the historically proven cost of
operating the appropriate equipment and amortizing its capital cost.
2.2 Production Volume Considerations
For each type of reforming system detailed in this report, a production volume of 250 identical
units per year was assumed. This assumption impacts batch size and equipment considerations,
as well as the amortization of equipment costs and the cost of purchased parts to the reformer
manufacturer. In addition, a production life of two years (500 units) was generally assumed,
where applicable, for life volume of equipment and tooling. After this time, equipment is
assumed to require replacing due to either wear or design changes. Naturally, increasing
production volume and/or life results in reduced costs. However, 250 units was assumed as an
appropriate production volume for initial introduction of the product, as discussed in Section 3.
2.3 Markup Rate Assumptions
Markup rate refers to the additional cost percentage to account for general and administrative
(G&A) expenses, material scrap, spending on research and development (R&D), and profit. In
some instances, markup rate also includes allowances for manufacturing and office building
expenses, but these facilities expenses are usually tabulated separately in standard cost
estimation practice.11 The markup rate is applied as a percentage increase to the material,
manufacturing and assembly cost. The value of markup rate varies depending on who is
performing the work.
In this analysis, two levels of markup may be applied to each component of the final system.
The lower level represents the markup applied by a vendor who sells a manufactured component
to the appliance manufacturer (the final assembler). The higher level is the markup applied by
the appliance manufacturer. The final resulting “cost” is thus actually a projected “price” to the
appliance purchaser (fueling station owner). In addition, the projected cost of hydrogen to the
consumer (potentially a FCV motorist) is provided in this report, with inclusion of operating
expenses to the reformer purchaser.
Because the production levels for the refueling appliance are low by mass-manufacturing
benchmarks, only final assembly of the refueling appliance is assumed to take place within the
HFA company’s facility. Manufacture and assembly of all subcomponents occurs at a vendor’s
facility where the machines used for those operations do not idle when not producing HFA
components but, rather, can be used to produce other non-HFA components. This method of
11
Standard automotive cost estimation, as practiced by Ford Motor Company and with which the authors are most familiar, typically does not
include facilities cost in the markup rates.
7
sub-contracting out the majority of the manufacturing work can be cost effective, particularly at
low production volumes, as it allows shopping around to find a vendor who has relatively high
machine utilization, and thus low price, but adequate excess capacity to undertake the HFA job.
Because markup rate will vary from vendor to vendor depending on specific circumstances and
machine utilization, a standard markup rate of 30% is used for all vendor manufacturing and
assembly operations as a reasonable upper limit. Typically automotive markup rates are 22% to
25% for highly utilized machines operating at 2 shifts. If all operations were to be brought in-
house but performed at very low production rates, required markup rates could easily soar to
100% or more.
Determining the appropriate markup rate for the HFA manufacturer is more difficult as it
requires amortizing the expenses of the business over a moderately low production rate.
Consequently, a prototype HFA company was analyzed to gauge the approximate operating
expenses. These expenses were then translated into markup rate percentages. Figure 2-1
outlines the General and Administrative (G&A) and Overhead (OH) expenses of a prototypical
HFA company.
Figure 2-2 details the markup rates applied to reflect the HFA company’s G&A, OH, R&D,
scrap, and profit for system assembly work conducted on the reformer system (the SMR or ATR
system including the PSA and abbreviated RS). A straight 31% markup rate is applied to both
the HFA company assembly work and to the components purchased by the HFA company from
the various vendors that manufacture the subassemblies. However, the hydrogen compressor,
storage, and dispenser subsystems are complete, largely off-the-shelf assemblies merely
assembled into the system. Thus a lower markup percentage of 15% is added to those
components. This is reasonable as these components are quite expensive and the HFA company
is merely packing these components into the final assembly, but at the same time there are real
costs associated with inclusion in the system.
8
General & Administrative (G&A) Expenses $2,297,500
Management Staff Direct Labor (11 staff) $945,000

Fringe Benefits @ 50% Direct Labor $472,500
Travel Budget $150,000
Office Supplies Misc. $30,000
Advertising Budget $500,000
Legal Budget $200,000
Overhead (OH) Expenses $1,908,000
Engineering Staff OH
Direct Labor
(10 engineer @ $75k/yr) $750,000
Fringe Benefits @ 50% Direct Labor $375,000
Facilities OH
Manufacturing Facility
(30,000 ft2 @ $12/ft2) $360,000
Office Facility
(15,000 ft2 @ $25/ft2) $375,000
Office Utilities $48,000
Research And Development (R&D) Budget $1,500,000

Figure 2-1. Approximate G&A, OH, and R&D Expenses of Prototypical HFA Company
Category Markup up on Markup on Pass

Manufactured/Assembled Through
Components Components
Profit 15% 5%
OH 3% 3%
G&A 7% 3%
R&D 4% 2%
Scrap 2% 2%
Total 31% 15%
Figure 2-2. HFA Manufacturer Markup Rates
2.4 Design Considerations
Before the cost analysis described above could be performed, a detailed design of each system
had to be created. This analysis examined both reforming (SMR and ATR) options from a
traditional process-engineering point of view. The unit operations that make up the reformer
systems in this study are similar to the industrial reforming operations used to produce hydrogen
at large scale (~ 450 kg/hr). Recent patents for small-scale reformer units were consulted for
background information and to find any process improvements that may be used in first-
generation hydrogen fueling stations.12,13
12
U.S. Patent 6,296,814
13
U.S. Patent 6,316,134
9
Once the necessary vessel volumes and conditions (pressure, temperature, gas composition, etc.)
were determined, the vessels and other components were designed. In this design process, an
effort was made to not only follow good DFMA practice by designing for ease of manufacture
and assembly, but also to ensure long life and good performance and to acknowledge existing
codes (i.e. ASME, TEMA) for pressure vessels and shell-and-tube heat exchangers. These codes
specify design criteria for diameters and thicknesses for various components. While the codes
were not strictly adhered to, they were generally taken as good practice for design. Tube and
shell thicknesses and tube-sheet thicknesses were calculated using the ASME standards as a
minimum, with thickness added to compensate for corrosion where appropriate. These
calculations take into consideration the properties of the material to be used, operating pressures
and temperatures, and the geometry of the design. The target life for the design of all
components (with the exception of some catalysts and adsorbents, as discussed in Section 4) was
a minimum of ten years. Materials were chosen based on cost as well as strength, corrosion
resistance, and machinability.
2.5 Sample DFMA Cost Analysis
While it is not feasible to demonstrate all levels of detailed cost analysis for all components in
this report, it is instructive to give samples of the various analyses performed. Figure 2-3
through Figure 2-5 detail the cost estimation for the inlet tube-sheet in the boiler of the 10-
atmosphere autothermal reforming (ATR) system, the assembly for the boiler, and a final bill of
materials for the boiler. Production of the boiler subcomponents (including the tube-sheet) and
assembly of the boiler are assumed to be performed by a vendor, and these activities incur the
vendor’s markup rate. In the bill of materials, an additional markup rate is applied by the system
manufacturer. This level of analysis is common to all components for which an off-the-shelf
analogy or price-quote is not available.
Tube-Sheet Material SS 316
Material cost/kg $/kg $8.82
Material density g/cm^3 8
Tube-Sheet Volume cm^3 411.851839
Tube-sheet Mass kg 3.294814712
Material cost $ $29.06
SETUP
Batch size 63
Setup rate $/min $3.00
Setup time minutes 92
Setup cost $ $276.00
Setup cost/unit $/unit $4.38
MACHINING tool change tool position machine time process time tool wear
Operation repeat s s/operation s/operation s rate, $/min $/operation Cost, $/unit
centerdrill 7 0 10.1 6.2 114.1 $3.00 0 $5.71
drill 7 12 10.1 10.06 153.12 $3.00 0 $7.66
countersink 7 12 10.1 4.884 116.888 $3.00 0 $5.84
end mill (periph.) - rough 1 12 20 225 257 $3.00 0 $12.85
remove piece 1 69.4 69.4 $1.00 $1.16
Total per header 710.508 $33.21
Machining Markup Factors

Vendor Manufactures 30.00%
Cost Summary
Material Cost $29.06
Machining Cost $37.59
Markup Cost $20.00
Total Cost pre-markup $ $66.65

Total Cost post-markup $ $86.65
Figure 2-3. Detailed Cost Analysis of a Boiler Tube-Sheet
10
Labor Rate $/min $1.75 additional 75cent/minute to account for assistant

Machine Rate $/min $0.50
Units/yr 250
Operation # repeats time, sec Machine Op.? Cost/Op.

Place Outlet Header 1 35.9 $1.05
Insert Tubes 7 44.1 $1.29
Weld tubes to header 7 1260 1 $47.25
Place Baffles 2 35.4 $1.03
Place Inlet Header 1 35.9 $1.05
Weld tubes to header 7 1260 1 $47.25
Ultrasound weld inspection 14 840 1 $31.50
Place Outlet end-dome 1 40.9 $1.19
Weld end-dome to header 1 1930 1 $72.38
Apply inner insulation 1 31 $0.90
Apply adhesive tape 1 40.6 $1.18
Place Shell 1 55.3 $1.61
Weld shell to outlet header 1 1930 1 $72.38
Inlet end-dome 1 40.9 $1.19
Weld end-dome to header 1 1930 1 $72.38
Ultrasound weld inspection 2 600 1 $22.50
Turn assembly on side 1 120 $3.50
Insert reformate ports 2 10.8 $0.32
Weld reformate ports 2 1930 1 $72.38
Apply outer insulation 1 31 $0.90
Apply adhesive tape 1 40.6 $1.18
Remove finished part 1 6 $0.18
Total 12248.4 $454.58
3.4 hrs
0.5 shifts
122 shifts/yr
Markup Rate 30.00%
Total Cost pre-Markup $454.58

Total Cost post-Markup $590.95
Figure 2-4. Sample Assembly Analysis for a Boiler
Component Name Part Name Quantity Unit Cost Cost Mfg Markup Cost w/ Markup
Boiler Inlet Header End-dome 1 $82.47 $82.47 31% $108.04
Tube-Sheet 1 $86.65 $86.65 31% $113.51
Boiler Outlet Header End-dome 1 $158.88 $158.88 31% $208.13
Tube-Sheet 1 $472.65 $472.65 31% $619.17
Boiler Core Boiler Shell 1 $2,156.78 $2,156.78 31% $2,825.38
Baffles 2 $30.17 $60.34 31% $79.05
Tubes 7 $63.37 $443.58 31% $581.09
Boiler Insulation 1 $27.03 $27.03 31% $35.41
Reformate inlet and outlet 2 $26.34 $52.68 31% $69.02
Boiler Assembly 1 $590.95 $590.95 31% $774.15
Total $4,132.02 $5,412.95

Figure 2-5. Sample Bill of Materials for a Boiler
11
3 REFUELING APPLIANCE HYDROGEN PRODUCTION RATE AND

MANUFACTURING QUANTITY
Figure 3-1 summarizes the hydrogen production rate parameters for the HFA and the parameters
for a single HFA service station and a larger station with six HFA units ganged together. Each
HFA produces a maximum of 115 kg H2 per day, sufficient to support 183 fuel cell vehicles if
each vehicle achieves 80 mpge and travels 12,000 miles per year. Thus is it appropriate for a
moderate to large fleet or a small public service station. This relatively small hydrogen refueling
plant is well-sized for the early days of fuel cell vehicle production when high geographic
concentrations of fuel vehicles are not expected.
Single HFA Service Six HFA Service

Station Station
Number of HVA’s 1 6
Peak Design H2 Production Rate 115 kg/day 690 kg/day
2,000 scfh 12,000 scfh
Fuel Cell Vehicles refueled each day 20 120
(Based on 80% tank refills)
Fuel Cell Vehicles supported by station 183 1,098
Fuel Cell Vehicles supported by 250 45,750
HFA’s
Assumptions/Basis
Station Capacity Factor14 69%
Station Availability Factor 98%15
Fuel Cell Vehicle fuel economy16 80 mpge17
Fuel Cell Vehicle mileage per year 12,000 miles/year
Fuel Cell Vehicle range 380 miles
Hydrogen stored onboard vehicle 4.76 kg H2
Figure 3-1. HFA Production Parameters
The parameters for a 6-HFA service station are also contained in Figure 3-1. The station is sized
to match the 125 vehicle refuelings conducted in the average US gasoline service station. The
station gangs together 6 hydrogen fueling appliances and can refuel 120 FCV’s per day and
support a community of 1,098 fuel cell vehicles.
An annual HFA production rate of 250 is assumed for the cost analysis. This level of HFA
production is sufficient to supply hydrogen for new FCV’s produced at a rate of 45,750 vehicles
per year. There are approximately 130 million automobiles in the United States with
approximately 7 million new gasoline automobiles sold in North America each year. Sales of the
top-selling single model automobile are about 250,000 each year. Thus this selection of HFA
14
H2 production de-rating factor to account for daily and seasonal fluctuations in demand.
15
Based on 22 days per year with one 8-hour shift of scheduled station down time.
16
Mileage based on the 1.25 accelerated combined federal drive schedule.
17
mpge = miles per gallon equivalent = FCV fuel economy with hydrogen consumptions translated into an equivalent amount of gasoline energy
(based on lower heating value of both hydrogen and gasoline).
12
manufacturing rate assumes only a modest annual production run of FCV’s and is consistent with
the hydrogen FCV startup scenario. Additionally, at this production rate, HFA hydrogen
production facilities sufficient for 457,000 FCV’s (approximately 8% of new vehicle sales) will
be in place at the end of a 10-year period (the estimated life of the HFA).
4 STEAM METHANE REFORMING VS. AUTOTHERMAL REFORMING

4.1 Overview of SMR and ATR
Both steam reforming and autothermal reforming are used industrially for converting
hydrocarbons to hydrogen. Methane-rich natural gas, propane, methanol, ethanol, naphtha
(gasoline), and biomass are all possible sources of hydrogen by reforming. Natural gas and
gasoline are widely available and would require minimal changes in distribution infrastructure to
supply the proposed hydrogen generating stations with fuel. However, natural gas is used as the
feedstock for this analysis because of its advantages over gasoline as a reforming fuel:
1. Methane has the highest possible ratio of hydrogen-to-carbon and results in the lowest
carbon dioxide emissions per kg of hydrogen generated.
2. Natural gas contains lower levels of sulfur and other impurities than does gasoline (~10
ppm vs. ~300 ppm). Sulfur can poison catalysts in the reforming process, so removing
sulfur from gasoline would be significantly more costly than from natural gas.
3. Natural gas reforming requires no on-site liquid storage tanks since the reforming process
is fed continuously from the natural gas pipeline.
4. Natural gas is produced from predominantly domestic supplies and would reduce the
need for imported crude oil.
Some localities may have an oversupply (i.e., a very inexpensive supply) of one of the other
fuels or lack a natural gas infrastructure, which could outweigh the inherent advantages of
natural gas reforming. However, for the vast majority of locations, natural gas will be the
preferred feedstock to the HFA.
The natural gas composition used for this study is shown in Figure 4-1. This is a normalized
version of the gas composition used in the 1997 DOE “Hydrogen Infrastructure Report”18 with
1% oxygen added. The composition is representative of an average U.S. city, but deviations
from the average are common.
Thermal efficiency is a measure of the performance of a reforming process and is defined as
H 2, LHV
ThermalEfficiency = (Equation 4-1)
Natural gas, LHV
18
Thomas, C.E., et al., Section 1, Hydrogen Infrastructure Report, 1997.
13
Component Volume %
Methane 90.7
Ethane 3.6
Propane 1.9
Carbon dioxide 1.0
Nitrogen 1.8
Oxygen 1.0
Water 0.01
Sulfur 10 ppm
Figure 4-1. Natural Gas Composition
The lower heating value (LHV) is the amount of heat evolved by full combustion of the fuel with
all combustion products remaining in the vapor phase. The natural gas rate counts all natural gas
fed to the system, including feed to the reformer and to any combustion devices used to generate
heat. Compressors and other motor-driven equipment were assumed to use electrical drives
rather than natural gas drives, so the motor power is not accounted for in the thermal efficiency
calculation. Calculating the thermal efficiency using the higher heating value (HHV) of
hydrogen and natural gas results in an efficiency about 5% higher than that defined in Equation
4-1.
Autothermal reforming (ATR) is generally considered the lower initial-cost option for hydrogen
generation because of a simpler reactor design, and steam methane reforming (SMR) is generally
the higher-efficiency option because of more complete methane conversion. The processing
options chosen for this comparison emphasize the relative strengths of each process, with the
result that there are many other potential variations that involve tradeoffs between capital cost
and efficiency. A few of these options are discussed in Sections 4.2 and 4.3, but the choices
covered in this analysis are not exhaustive.
The basic processing steps are common to both SMR and ATR:
1. Natural gas compression

2. Natural gas purification (i.e., sulfur removal)
3. Catalytic steam reforming of methane to hydrogen and carbon monoxide (CO)
4. Water-gas shift to convert carbon monoxide (CO) to carbon dioxide (CO2) and additional
hydrogen
Mass and energy balances and chemical equilibrium compositions were calculated in a
HYSYS™ process simulation.19
4.1.1 Natural Gas Compression
Natural gas is distributed by the utility at just slightly above atmospheric pressure and must be
compressed to achieve a more favorable combination of equilibrium, reaction rates, and
19
HYSYS is a chemical engineering simulation environment and is a product of AEA Technology
14
equipment sizing. The natural gas compressor design is identical for all of the considered
processing options, although the motor size is larger for higher discharge pressures.
4.1.2 Natural Gas Purification
The most common impurities in city natural gas are sulfur compounds, particularly mercaptan
compounds used as odorants. Sulfur poisons the reforming catalyst by forming nickel-sulfide, so
the natural gas sulfur level must be reduced to less than 1 ppm. The total sulfur content of the
natural gas was assumed to be 10 ppm. Typically there are several ppm of H2S naturally
occurring in and several ppm of mercaptan odorant added to distributed natural gas. Mercaptans
are converted irreversibly to hydrogen sulfide by the hydrodesulfurization (HDS) reaction over a
cobalt-molybdenum or nickel-molybdenum catalyst
R − SH + H 2 = R − H + H 2 S (Equation 4-2)
A small fraction of the product H2 from the reforming process is recycled to the HDS reactor to
facilitate this reaction. The reaction stoichiometry is one H2 per mercaptan molecule, but an
excess of H2 is added to ensure complete conversion of mercaptans to H2S.
The H2S is removed from natural gas by absorption in a bed of zinc oxide (ZnO). The H2S reacts
with ZnO in a 1:1 mole ratio to form zinc sulfide (ZnS) and water. In both the SMR and ATR
cases, the ZnO bed volume was sized to handle 10 ppm average sulfur content and 75%
conversion of ZnO to ZnS. The HDS catalyst and ZnO absorbent are contained in a single
vessel, which is an arrangement used for relatively low sulfur content gas streams.20 The
proposed parallel two-bed design is illustrated in Figure 4-2. One bed is sized to last 6 months,
and the valves are arranged so that the flow path can be switched without interrupting gas flow.
The vessel is constructed from standard flanged pipe sections to minimize material and assembly
costs, as shown in Figure 4-3.
Hot Reformate Manual

HDS Section Shut-off
Valve
NG
NG
HDS Preheater
Hot Reformate
Figure 4-2. Two-Bed HDS System with Natural Gas Preheater (not to scale)
20
Elvers, Barbara, et al., ed. Ullmann’s Encyclopedia of Industrial Chemistry, Volume A12, page 192, 1989.
15
Alumina 4" Pipe
Screen
HDS Catalyst
and Sulfur
Absorbent
Figure 4-3. HDS Pipe Section (not to scale)
Oxygen removal may be required as an additional natural gas pretreatment step in some locations
for the SMR. Peak shaving by adding propane and oxygen to the natural gas is a common
practice during times of high demand that can add significant oxygen to the natural gas, possibly
as much as 3%. Consistently high levels of oxygen in the natural gas can deactivate the
reforming catalyst. We assumed that the entrance region of the SMR reactor is filled with a
small fraction of palladium catalyst loaded onto inert aluminum oxide pellets. The palladium
will facilitate catalytic combustion to remove the oxygen. The ATR does not require an oxygen
removal step since it uses partial combustion to raise the process temperature, and all the oxygen
will be consumed prior to reaching the reforming catalyst.
4.1.3 Catalytic Steam Reforming
The difference between SMR and ATR is how heat is provided to activate the endothermic steam
reforming reaction
CH 4 + H 2O = CO + 3H 2 ∆Hrxn = 241 kJ/gmole (Equation 4-3)
In SMR, the catalyst is contained in tubes that are heated by an external burner. In ATR, a
portion of the natural gas is burned according to the following reactions to raise the temperature
of the process gas before it contacts the catalyst:
0.5CH 4 + O 2 = CO + H 2O (Equation 4-4)

CH 4 + 2O 2 = CO 2 + 2 H 2O (Equation 4-5)
A substoichiometric volume of oxygen is fed to the ATR to encourage partial combustion

(Equation 4-4) over full combustion (Equation 4-5). To supply this oxygen, air is fed along with
the natural gas and steam, causing a relatively large fraction of inert nitrogen gas to be carried
through the process.
The ratio of steam to carbon (S/C ratio) fed to the SMR typically ranges from 2.5 to 4 on a molar
basis. Lower S/C ratios tend to result in more methane decomposition to carbon (coking),
reducing the catalyst life and the reforming efficiency. Higher S/C ratios reduce the tendency for
coking and improve the conversion of methane, but more energy is required to generate the
16
steam than is available from other parts of the process. The S/C ratio chosen for any particular
reformer will depend on site-specific economic factors, especially on the price of natural gas.
Locations with higher natural gas price will realize a greater reduction in hydrogen cost by
increasing thermal efficiency than will locations with low natural gas price. We have chosen a
S/C ratio of 3 for this analysis (see Section 4.2 for details). The S/C ratio for ATR is typically
~1. The tradeoffs for higher S/C ratios in ATR are discussed in more detail in Section 4.3.6.
Traditional industrial reforming uses catalysts containing 10-20 weight percent nickel (Ni) as the
active ingredient. The nickel is dispersed on a high temperature-stable support, often the high-
temperature phase of aluminum oxide containing Group IIA elements (e.g., Mg) as stability and
anti-coking additives. The platinum group metals, particularly ruthenium (Ru) and rhodium
(Rh), are highly active catalysts for steam reforming21 but have not been in widespread industrial
use because of the added cost over nickel. Ruthenium is approximately 4 times as active as
nickel at 800ºC.22 The platinum group metals by themselves are not active for the water-gas shift
reaction, and loss of the shift reaction in the reformer would significantly increase the size of the
downstream water-gas shift reactor.
We propose to achieve the benefits of both metals by using a traditional Ni-based reforming
catalyst loaded with 0.5-1.0 weight percent ruthenium. The benefits of the proposed catalyst are
that platinum-group metals tend to avoid coking caused by methane decomposition to carbon,
that ruthenium is the most active platinum group metal catalyst, and that the price of ruthenium
over the past 10 years has averaged only 5-10% of the cost of rhodium.23 Ruthenium is more
stable than nickel at high temperatures and should remain better dispersed over several years of
catalyst use. Some platinum-group metal reforming catalysts are available commercially.
Haldor-Topsøe sells a reforming catalyst that contains a platinum-group metal and generally
costs about 50% more than the traditional nickel catalyst.24 Toyo Engineering (Japan) and
Synetix both advertise the availability of precious-metal steam reforming catalysts on their
websites.25,26
We used a reactor model taking into account intrinsic reaction kinetics and mass transfer
limitations to calculate the reformer reactor sizes. Kinetic expressions for the reforming reaction
were taken from the open literature27,28 and confirmed by comparisons to rule-of-thumb factors
for industrial reforming reactors.29 In general, the ATR requires more reforming catalyst than
the SMR. The higher rate of reaction at ATR temperatures is more than offset by the dilution
effect of inert nitrogen reducing the partial pressure of the reactants.
21
Rostrup-Nielsen, J.R. and J-H. Bak Hansen, “CO2-Reforming of Methane over Transition Metals.” J. Catalysis, 144, p 38-49 (1993).
22
Rostrup-Nielsen and Bak Hansen, ibid.
23
Johnson-Matthey Platinum Historical Price Report, http://www.platinum.matthey.com, 2002.
24
Personal communication with David Kelling, Haldor-Topsøe, January 2002.
25
Toyo Engineering website, http://www.toyo-eng.co.jp/Technology
26
Synetix website, http://www.synetix.com
27
Rostrup-Nielsen and Bak-Hansen, ibid.
28
Rostrup-Nielsen, Jen R. Catalytic Steam Reforming. Springer-Verlag, Berlin, 1984.
29
Ridler, D.E., and M.V. Twigg, “Steam Reforming” in Catalyst Handbook, Martyn V. Twigg, ed. Manson Publishing, London, 1996.
17
4.1.4 Water-Gas Shift Reaction
Additional hydrogen is produced by exothermic reaction of carbon monoxide with water by the
water-gas shift (WGS) reaction
CO + H 2O = CO2 + H 2 ∆Hrxn = -90 kJ/gmole (Equation 4-6)
In the presence of excess water some WGS reaction takes place in the reforming reactor,
however the equilibrium to hydrogen is much more favorable at temperatures below 450ºC. The
water-gas shift catalyst is chromium-promoted iron oxide, which is the traditional catalyst for the
high-temperature water gas shift process. The shift reactor was sized to provide an average of
three seconds of contact time between the flowing gas and the catalyst, which is in the mid-range
of size used in industrial systems.30
The proposed process does not include a separate low temperature water gas shift reactor since at
this relatively small scale the cost of hydrogen would increase by about 1¢/kg to pay for the cost
of an additional reactor (~$10,000-12,000), not including the increased cost for assembly,
foundation size, and building size. While 1-2¢/kg may seem a small effect, the point is that
inclusion of a low temperature water gas shift reactor increases the cost of hydrogen rather than
driving it down, thus the reactor is not cost effective.
4.2 SMR System Flowsheets, Cost and Performance
Steam reforming systems were designed for nominal 10 atm operation with pressure swing
adsorption (PSA) H2 purification and 20 atm operation with membrane H2 purification. The
hydrogen purification processes are described in detail in Section 5. The SMR processing
conditions are listed in Figure 4-4 and the flow sheet is shown in Figure 4-5.
Pressure
10 atm SMR 1000 kPa (nominal)
20 atm SMR 2000 kPa (nominal)
Temperatures
HDS/ZnO 400ºC
Reformer Inlet 430ºC
Reformer Outlet 800ºC
Shift Inlet 450ºC
Shift Outlet 250ºC
Condenser Outlet 55ºC
Miscellaneous
S/C 3
Figure 4-4. SMR Process Conditions
The steam-to-carbon (S/C) ratio of 3 was chosen to trade off the methane conversion and
likelihood of catalyst coking with the amount of heat available by burning the waste gas from the
30
Satterfield, Charles N., Heterogeneous Catalysis in Industrial Practice, 2nd ed. McGraw-Hill, New York, 1991.
18
hydrogen purification step. Slightly higher thermal efficiencies could have been achieved by
reducing the S/C ratio to less than 3, but the risk of catalyst coking would increase.
The proposed designs do not integrate heat transfer as much as theoretically possible so that the
process conditions are not over constrained. For example, if air used to cool the condenser is
used to cool the shift reactor, only one of the temperatures (condenser or shift reactor) may be set
independently without adding extra control valves and instrumentation. Additionally, if the air
pressure drop is relatively high due to a flow path through multiple units, the blower or fan cost
increases significantly. In the proposed design, a portion of the air used to cool the condenser is
routed to the burner to sustain the combustion. For optimum temperature control, independent
blowers or fans are used to provide the other cooling-air needs.
The steam boiler, reformer, reformate cooler, and shift reactor are all based on shell-and-tube
heat exchanger designs. A consistent manufacturing and assembly approach was used for all of
the shell-and-tube units.
A possible variation of the SMR is to operate at atmospheric pressure. The choice of materials
that are acceptable to use at reformer temperatures is dramatically increased at low pressure. For
example, the steam boiler design described in Section 4.2.2 has Incoloy 800H tubes with a wall
thickness of 0.90 mm to operate at 10 atm. If instead the pressure were 1 atm, the tubes could be
constructed of 316SS with a wall thickness of 0.25 mm or less. The potential cost savings at
lower pressure for units like the boiler, however, may be offset by larger reactor sizes for the
reformer and shift reactor. Both steam reforming and water gas shift reaction rates decrease with
decreasing pressure, leading to larger catalyst beds to achieve equivalent conversion.
19
Cost and Performance Comparison Of Stationary Hydrogen Fueling Appliances Directed Technologies, Inc.
Natural Gas
to Burner
Natural Gas
Main Shutoff Valve Filter
Natural Gas
to Reforming
Natural Gas
Flow Control Valve Nat. Gas (400C)
Burner
Exhaust (160C)
Relief Valve
Temp
Air (68C)
Natural Gas Steam (450C)
Compressor Burner
DI Water (25C)
HDS/ZnO
Boiler Burner Exhaust (506C)
Reformate
Reformate (760C)
F SMR Cooler
Water T
Reformate (800C)
P
Temp DI Bed
CIty
HDS Water
Preheater Nat. Gas (400C)
Reformate (800C)
Reformate (450C)
Relief Valve
Relief
Valve Raffinate (25C)
Hydrogen Raffinate
Storage Raffinate (55C) Surge Tank
Relief
Air (66C)
Valve
Vent
Burner Air
Control Valve
Dry Gas (55C)
HTS
Temp
Hydrogen
Hydrogen Purification
Compressor Water
P
Knockout Condenser
Shift Gas (250C)
Water (55C)
Hydrogen
Pressure-Reducing
Valve
Ambient air
Figure 4-5. SMR Flowsheet
20
4.2.1 Reactor and Vessel Sizing
Natural Gas Pretreatment
The natural gas feed is heated to 400ºC for operation of the HDS catalyst and ZnO absorption
bed. The proposed design uses the hot reformate stream to preheat the natural gas in a double-
pipe heat exchanger with a finned inner pipe, as shown in Figure 4-6. The reformate is inside the
inner pipe so that only the tube is designed for this high temperature stream, and the natural gas
is in the annular space. There is the potential for a small fraction of the natural gas to decompose
to carbon by thermal cracking on the hot surface of the inner tube, and the thermal cracking is
more likely to occur in natural gas with a high fraction of higher hydrocarbons. The preheater
could be moved downstream of the reformate cooler if coking is a problem without significantly
changing the total cost.
Figure 4-6. HDS Preheater, with Tube-in-Tube Design (not to scale)
The hydrodesulfurization (HDS) catalyst and zinc oxide (ZnO) absorbent are combined into a
single vessel, shown previously in Figure 4-3. The operating temperature of the HDS/ZnO bed
is 400ºC. The HDS catalyst volume is sized to provide a gaseous hourly space velocity (GHSV)
of 3000 hr-1, and the ZnO volume is designed to last 6 months. The vessels to hold the catalyst
and the absorbent are made from sections of 4-inch diameter stainless steel pipe with flanged
ends and screens to contain the catalyst and absorbent. Two such vessels are used in each
reforming system, so that one bed may be used in the process while the other is changed out,
giving the pair a one-year life.
Steam Reformer
The proposed reactor design is a shell-and-tube heat exchanger with the reforming catalyst in the
tubes and the burner exhaust in the shell (see Figure 4-7 and Figure 4-8). The natural gas and
steam mixture flows countercurrent to the combustion gas from the burner.
Outer Insulation
Burner
Shell
Inner Insulation
End-dome
NG & Steam Reformate
Reactor Tube Baffle Tube Sheet
Flue Gas
Figure 4-7. SMR Reformer Vessel (not to scale)
21
Figure 4-8. Tubes and Tube-Sheets for Shell-and-Tube Vessel

(Such a configuration is used, with varying numbers of tubes, for the SMR vessel, the HTS,
and most heat exchangers.)
Tubes are ½” (1.27 cm) outside diameter. Using fewer larger diameter tubes would simplify the
tube sheet construction, but heat transfer to the center of larger tubes would be insufficient to
make up for the endothermic reaction.
One of the biggest challenges to designing a steam reformer at this scale is the materials
selection for the reformer tubes. The tube metal is exposed to ~1000ºC combustion exhaust
(oxidizing atmosphere) on the outside and 800ºC reformate (reducing atmosphere) on the inside.
The alloy chosen for the tubes, the tube sheets, and the outlet end dome is Haynes 556. Other
possible alloys for the reformer tubes are other Haynes alloys such as 230 and Inconel alloys
such as 617 and 625. All of the potential high-temperature resistant alloys are a factor of five to
nine times more expensive than the common stainless steels and are difficult to work with. The
bulk cost of Haynes 556 was estimated at $60/kg, and DFMA techniques were used to estimate
the cost for tube forming. The cost for construction and assembly of the entire reformer was also
estimated using DFMA. Orbital welding was used to weld the individual tubes to the tube
sheets.
The tube sheets to which the tubes are welded at both ends of the reactor are significant
contributors to the reformer cost. As a result, the number of tubes in the reformer is one of the
most important factors in determining the overall cost. We have chosen a number of tubes that
result in a reasonable length (100 cm) to avoid excessive pressure drop and stress on the
unsupported tube metal. There are 180 tubes for the 10 atm SMR and 134 tubes for the 20 atm
SMR.
The 10-atmosphere steam-reforming reactor requires 16.7 L of catalyst to approach the

equilibrium conversion of methane at 800ºC (from the kinetic model). The amount of catalyst
needed to reach equilibrium methane conversion is 10.9 L for the 20 atm system due to faster
reaction rate at higher pressure. However, the equilibrium conversion of methane is only 73% at
20 atm compared to 84% at 10 atm.
22
The shell is constructed of Incoloy 800H. The shell is lined with 1” (2.54 cm) of insulation to
reduce heat loss to the surroundings while at the same time reducing the shell metal temperature.
Lowering the shell metal temperature allows the shell wall to be thinner than if the insulation
was only on the outside of the shell. An additional 5.1 cm (2 inches) is added to the shell
diameter to account for the insulation thickness. The outside of the shell is wrapped with ½” of
insulation. Incoloy 800H has excellent oxidation resistance at the temperatures to which it is
exposed through the insulation in the reformer shell (~500-600ºC). With an adequate insulation
liner, the reformer shell may be constructed of a more conventional stainless steel. For a safety
margin and to guard against oxidation at higher temperatures due to leaks in the insulation,
however, we chose the more resistant Incoloy 800H.
The burner is sized to provide enough heat for the reforming reaction and to generate the
superheated steam for the reformer (see Section 4.2.2). A nominal 150 kW (500,000 Btu/hr)
burner including flame detector and affiliated equipment is $700 (quote from Martin Controls).
Shift Reactor
The shift reactor is a shell-and-tube design similar in concept to the reforming reactor. The shift
catalyst is contained in the tubes, and the cooling air is in the shell. The process stream is cooled
from 450ºC at the reactor inlet to 250ºC at the reactor outlet. The rate of the shift reaction slows
considerably with the temperature drop, but the equilibrium shifts to 95% conversion of CO at
250ºC from 61% at 450ºC.
The 10-atmosphere shift reactor contains 130 kg of catalyst to approach the equilibrium
conversion of CO at 250ºC. The 20-atmosphere shift reactor contains 83 kg since the shift
reaction proceeds more quickly at higher pressure. The pressure does not affect the equilibrium
conversion of CO, since the water gas shift reaction has the same number of moles of products as
of reactants.
4.2.2 Heat Exchanger Sizing
Reformate Cooler
The reformate stream is cooled to 450ºC for the water-gas shift reaction. A variable speed
blower supplies the cooling air to the reformate cooler. The reformate cooler is a shell and tube
heat exchanger with the reformate inside finned tubes (Haynes 556 tubes with 316SS fins). The
outer shell is a thin metal jacket (316 SS) to contain the cooling air at essentially atmospheric
pressure.
Boiler
Deionized water is evaporated and superheated to 450ºC by heat transfer with the burner gases
leaving the reformer. The boiler is a shell and tube-type heat exchanger, although the shell is not
pressure rated because it contains atmospheric pressure burner exhaust. The steam is generated
and superheated inside finned tubes constructed of Incoloy 800H with 316SS fins.
23
Condenser
The condenser is a bank of stainless steel finned tubes with cross-flow of air as the cooling fluid.
The condenser design is similar to a commercially available room heater that uses condensing
steam to provide heat. The heat exchanger plus fan assembly can be purchased for about $1200
plus the cost of motor, speed controller, and motor starter (all off-the-shelf items). The heat
exchanger and fan assembly for the 20 atm system is about double (~$2500) the cost for the 10
atm system because of the higher pressure rating.
Water Knockout Vessel
A vessel is needed to separate condensed water before sending the reformate gas to the gas
cleanup system. The knockout vessel is designed to reduce the vapor velocity to 1 m/s or less so
that the water is removed by gravity separation. A one-stage separator provides a clean
separation since no other compounds are present in significant quantities in the liquid water
phase. The proposed design is a section of 4” steel pipe with welded end caps and welded inlet
tap. The knockout vessel is elevated so that water is returned to the deionization tank (Section
4.2.3) by gravity feed.
4.2.3 Water Purification
The water requirement for steam fed to the SMR systems is approximately 1-1.5 liters/minute.
Ions such as calcium and chloride common in city water can cause serious fouling and corrosion
problems at the high temperatures in the reforming system and can poison reforming and shift
catalysts. The water used to generate steam must be deionized before it is fed to the reformer
system. The proposed water purification system is a carbon filter followed by a deionization
(DI) tank. The deionization tank is a low-pressure rated vessel filled with deionization resin that
is changed on a regular schedule. US Filter provides a service that includes setup of the carbon
filter and deionization tank for approximately $1500 initially and resin replacement cost of
approximately $1500 per year.31 The deionization resin life is dependent on source water
quality, so the replacement cost listed here could vary in any specific situation.
The volume of water recovered from the condenser after the high-temperature shift reactor is
approximately 35% of the volume fed to the reformer. This water may contain trace levels of
metals picked up from the catalysts and may not be suitable for direct discharge to municipal
sewer. The proposed design mixes the condensate with fresh city water to make up the
necessary water rate.
4.2.4 Other Equipment
Fans and Blowers
Costs for cooling fans and blowers were estimated from catalog prices for standard items. The
fans with associated equipment (motors and motor starters) are included in the bills of materials
for the heat exchangers.
31
Personal communication with Randy Pittman, US Filter Corporation, February 2002.
24
Instruments
The instruments included in the design are only those necessary to assure proper operation for a
well-established process, such as thermocouples and pressure transducers to provide signals to
automatic control valves (see Figure 4-5). The first several installations will have a higher level
of instrumentation to verify the design conditions and to collect data for process improvements.
Instrument lists for the 10 atm and 20 atm systems are identical and are included with the system
bills of materials.
Pallet
Figure 4-9 shows a proposed layout of a 10-atmosphere SMR reformer system (RS) containing
the major components at their approximate relative sizes. The reforming components would fit
on an eight-by-thirteen foot pallet, with the hydrogen compressor, storage system, and dispenser
housed separately. This size of footprint offers generous room for the necessary vessels and to
allow easy access for servicing the components. The SMR and HTS reactors, and the HDS
system are placed near the edges of the footprint, making them especially easy to service and
remove for catalyst replacement. The 20-atmosphere SMR system would have a similar
floorplan, but would not require the bulky PSA system and 250-gallon raffinate surge tank,
reducing the necessary footprint to an eight-by-ten foot area. In both cases, all of the reformer
components would be secured to an appropriately sized, skid-mounted pallet for transportation
and installation purposes. This pallet would be placed directly onto a concrete slab at the
installation site.
PSA
Raffinate Water KO
Surge
Tank Condenser
HTS
NG 13'
Compressor
Boiler
SMR
HDS
8'
Reformate Cooler
Figure 4-9. Proposed Layout of 10 atm SMR Reformer System

(Cylindrical vessels are secured in racks (not shown), and all components are attached
to a skid-mounted pallet.)
25
4.2.5 Assembly
The cost to assemble the reformer system (RS) was calculated from the time to required to set
the vessels on the pallet, the materials and labor to install the necessary length of piping between
vessels, the material and labor to insulate the piping, and the labor to pressure test the completed
assembly. Eight man-days were allotted for this process. (With a team of two workers, each unit
takes four workdays, meaning that five such teams would be required to produce 250 units in one
220-day workyear.)
4.2.6 SMR Reformer System Bill of Materials and Cost
The overall SMR RS bill of materials is shown for a 10 atm system in Figure 4-10 and for a 20
atm system in Figure 4-11. The cost for each reforming system (RS) includes the gas cleanup
system, which is described in detail in Section 5. The initial capital cost for the 10atm SMR-
PSA RS is $123,545, while that for the 20atm SMR-membrane RS is $137,036.
4.2.7 SMR Performance and Efficiency
The thermal efficiency of the 10 atm SMR is 70% (LHV basis), corresponding to a hydrogen
production of 0.30 kg per kg of natural gas fed (including fuel to the burner). The thermal
efficiency of the 20 atm reformer is 66%, with a hydrogen production of 0.26 kg per kg of
natural gas fed (including fuel).
26
Total Sub System Cost Total Sub System Cost

Sub-system Name Component Name Part Name Material Basic dimensions Quantity Unit Cost Cost w/o Mfg Markup Mfg Markup Cost w/ Markup w/ Mfg Markup
Natural Gas Feed

Natural Gas Compressor 1 $1,225.00 $1,225.00 $2,260 31.00% $1,604.75 $2,961
System
Natural Gas Compressor
Explosion proof 5 kW 1 $900.00 $900.00 31.00% $1,179.00
Motor
Manual NG Shut-off Valve 1 $15.00 $15.00 31.00% $19.65
Natural Gas Control Valve 1 $120.00 $120.00 31.00% $157.20
Sulfur Removal Heat Exchanger Outer pipe SS 316 2.5" Sched 40, 12"L 1 $38.00 $38.00 $3,389 31.00% $49.78 $4,439
Inner tube Haynes 556, SS fins 1"Diam, 23"Length 1 $98.25 $98.25 31.00% $128.71
Reducers Incoloy 800H 2.5"Diam 2 $65.80 $131.60 31.00% $172.39
NG inlet/outlet Incoloy 800H 2 $12.14 $24.28 31.00% $31.81
Insulation CCMax4 1"Thick 1 $83.20 $83.20 31.00% $109.00
HX Assembly 1 $139.92 $139.92 31.00% $183.29
Hydro-desulfurization Bed HDS Vessel SS 316, 4" Pipe 30"Length 2 $268.30 $536.59 31.00% $702.94
Flanges SS 316, 150lb 4 $107.71 $430.82 31.00% $564.37
HDS Insulation CCMax4 2 $193.08 $386.16 31.00% $505.87
HDS catalyst Co_W 1.7 L 2 $44.74 $89.48 31.00% $117.22
Zinc Oxide Adsorbent Zinc oxide 3.95 L 2 $46.84 $93.68 31.00% $122.72
Vessel Assembly 2 $120.75 $241.51 31.00% $316.38
Flanges SS 316 4 $107.71 $430.82 31.00% $564.37

Flange gaskets 2 $6.50 $13.00 31.00% $17.03
Nuts/Bolts 5/8"Length 16 $6.50 $104.00 31.00% $136.24
Manual shutoff valves 4 $40.00 $160.00 31.00% $209.60
Pipe Tees SS 316 2 $30.00 $60.00 31.00% $78.60
Pipe elbows SS 316 4 $22.00 $88.00 31.00% $115.28
Pipe SS 316 1"pipe, Sched40 2.7 $12.00 $32.40 31.00% $42.44
Subsystem Assembly 1 $206.92 $206.92 31.00% $271.06
Boiler/Steam Super
Water Inlet Header End-dome SS 316 16.5"Diam, 0.125"Thick 1 $136.04 $136.04 $10,392 31.00% $178.21 $13,613
Heater
Tubesheet SS 316 16.5"Diam, 0.9"Thick 1 $528.74 $528.74 31.00% $692.65
Steam Outlet Header End-dome Inc 800H 16.5"Diam, 0.2"Thick 1 $407.50 $407.50 31.00% $533.83
Tubesheet Inc 800H 16.5"Diam, 1.1"Thick 1 $1,606.25 $1,606.25 31.00% $2,104.19
Boiler Core Boiler Shell SS 316 16.5"Diam, 30"Length 1 $389.25 $389.25 31.00% $509.91
Baffles SS 316 3/16" thick 2 $58.63 $117.26 31.00% $153.61
Tubes Inc 800H, SS fins 1"Diam, 30"Length 64 $85.22 $5,454.36 31.00% $7,145.21
Boiler Insulation ceramic fiber 2" thick 1 $58.59 $58.59 31.00% $76.76
Flue gas inlet and outlet SS 316 2 $10.51 $21.02 31.00% $27.54
Boiler Assembly 1 $1,672.89 $1,672.89 31.00% $2,191.49
Steam Reforming
SR Reactor Inlet Header End-dome SS 316 15.25"Diam, 0.37"Thick 1 $179.14 $179.14 $23,877 31.00% $234.68 $31,279
Reactor
Tubesheet Incoloy 800H 15.25"Diam, 1.25"Thick 1 $1,940.56 $1,940.56 31.00% $2,542.13
Steam/NG Mixing Port SS 1 $200.00 $200.00 31.00% $262.00
SR Reactor Outlet Header End-dome Haynes 556 15.25"Diam, 1.05"Thick 1 $2,740.96 $2,740.96 31.00% $3,590.65
Tubesheet Haynes 556 15.25"Diam, 2.18"Thick 1 $4,852.09 $4,852.09 31.00% $6,356.24
SR Reactor Core SR Reactor Shell Incoloy 800H 15.25"Diam, 40"Length 1 $726.43 $726.43 31.00% $951.62
SR Reactor Baffles Incoloy 800H 3/16" thick 4 $160.46 $641.83 31.00% $840.80
SR Reactor Tubes Haynes 556 1/2"Diam, 39.4"Length 180 $21.68 $3,902.99 31.00% $5,112.92
SR Catalyst Ru/Ni-AlO 16.7 L 1 $1,302.60 $1,302.60 31.00% $1,706.41
SR Reactor Insulation CCMax4 internal,external,header 1 $1,529.18 $1,529.18 31.00% $2,003.23
Flue gas inlet and outlet Haynes 556 2 $16.27 $32.53 31.00% $42.62
SR Reactor Assembly 1 $4,260.97 $4,260.97 31.00% $5,581.87
Burner Main Burner Assembly 1 $682.00 $682.00 31.00% $893.42
Natural Gas/Air Control
1 $886.00 $886.00 31.00% $1,160.66
Valve
Figure 4-10. Bill of Materials for 10 atm SMR RS

(1 of 3)
27

High Temperature
Reformate cooler HEX Shell SS 316 8"Diam, 18"Length 1 $284.82 $284.82 $15,043 31.00% $373.12 $19,707
Shift
Baffles SS 316 1/8" thick 2 $12.66 $25.32 31.00% $33.17
Tubes Haynes 556 1"Diam, 18"Length 13 $64.92 $843.96 31.00% $1,105.58
Inlet Tubesheet Incoloy 800H 8"Diam, 1.25"Thick 1 $427.09 $427.09 31.00% $559.49
Inlet End-dome Incoloy 800H 8"Diam, 0.2"Thick 1 $161.87 $161.87 31.00% $212.05
Outlet Tubesheet SS 316 8"Diam, 0.75"Thick 1 $161.87 $161.87 31.00% $212.05
Outlet End-dome SS 316 8"Diam, 0.15"Thick 1 $86.03 $86.03 31.00% $112.70
Air inlet/outlet SS 316 2 $10.51 $21.02 31.00% $27.54
External insulation ceramic fiber 1"Thick 1 $18.42 $18.42 31.00% $24.13
Blower Expl. proof motor 1 $1,200.00 $1,200.00 31.00% $1,572.00
Motor Starter 1 $250.00 $250.00 31.00% $327.50
Speed Controller 1 $200.00 $200.00 31.00% $262.00
HEX assembly 1 $699.39 $699.39 31.00% $916.20
HTS Reactor Inlet Header Tubesheet SS 316 15.25"Diam, 1.5"Thick 1 $803.14 $803.14 31.00% $1,052.11
End-dome SS 316 15.25"Diam, 0.16"Thick 1 $111.68 $111.68 31.00% $146.30
HTS Reactor Outlet Header Tubesheet SS 316 15.25"Diam, 1.25"Thick 1 $745.41 $745.41 31.00% $976.49
HTS Reactor Core HTS Reactor Shell SS 316 15.25"Diam, 1/16"Thick 1 $646.64 $646.64 31.00% $847.10
HTS Reactor Baffles SS 316 3/16" thick 6 $42.70 $256.22 31.00% $335.64
HTS Reactor Tubes SS 316 1"Diam, 90"Length 106 $14.99 $1,588.54 31.00% $2,080.99
HTS Reactor Catalyst Fe/Cr oxide 130 kg 1 $1,335.10 $1,335.10 31.00% $1,748.98
HTS Reactor Insulation ceramic fiber 1"Thick 1 $144.88 $144.88 31.00% $189.80
Air flow ports 2 $10.51 $21.02 31.00% $27.54
Blower Blower Expl. proof motor 1 $1,200.00 $1,200.00 31.00% $1,572.00
Motor Starter 1 $250.00 $250.00 31.00% $327.50
HTS reactor assembly 1 $3,249.04 $3,249.04 31.00% $4,256.25
Condenser Assembly Heat Exchanger + fan 1 $1,200.00 $1,200.00 $2,280 31.00% $1,572.00 $2,987
Motor + motor starter 1 $600.00 $600.00 31.00% $786.00
Water Knock-out Tank 1 $280.30 $280.30 31.00% $367.19
Water Purification Main Water Shut-off Valve 1 $50.00 $50.00 $2,450 31.00% $65.50 $3,210
Water Pump Positive displacement 1.5 L/min, 150 psig 1 $900.00 $900.00 31.00% $1,179.00
Deionization Bed DI Canister 1 $1,500.00 $1,500.00 31.00% $1,965.00
Pressure Swing
Adsorption Vessels Top Header SS 316 12"Diam, 200 psig cyclic load 4 $121.56 $486.23 $13,798 31.00% $636.97 $18,075
Adsorption
Bottom Header SS 316 12"Diam, 200 psig cyclic load 4 $119.63 $478.54 31.00% $626.89
Vessel Body SS 316 12"Diam, 72"Length, 200 psig 4 $593.54 $2,374.16 31.00% $3,110.15
Pressure Relief Valves 200 psig 4 $100.00 $400.00 31.00% $524.00
Activated Carbon
4 $176.70 $706.80 31.00% $925.91
Adsorbent
Zeolite 5A Adsorbent 4 $273.48 $1,093.93 31.00% $1,433.05
Silica gel 4 $153.92 $615.68 31.00% $806.54
Vessel assembly 4 $238.59 $954.35 31.00% $1,250.20
PSA Valves PSA Valves 27 varies $3,556.00 31.00% $4,658.36
Piping SS 316 3/4" pipe, sched40 $803.20 31.00% $1,052.19
Raffinate Surge Tank Vessel 250 gal, 150psi 27"Diam, 8'Length 1 $500.00 $500.00 31.00% $655.00
Pressure Relief Valve 1 $100.00 $100.00 31.00% $131.00
Regulator SS 316 5psi 1 $150.00 $150.00 31.00% $196.50
PSA Vessel Rack Stainless Steel 4.5'x1.5'x8' 1 $400.00 $400.00 31.00% $524.00
Subsystem Assembly 1 $1,178.88 $1,178.88 31.00% $1,544.33
Figure 4-10. Bill of Materials for 10 atm SMR RS (2 of 3)
28

Structural Supports Skid-mounted floor 1 $3,000.00 $3,000.00 $4,400 31.00% $3,930.00 $5,764
Vessel stands/frames 7 $200.00 $1,400.00 31.00% $1,834.00
Controls System Thermocouples 5 $40.00 $200.00 $10,750 31.00% $262.00 $14,083
Mass Flow Meter 1 $2,000.00 $2,000.00 31.00% $2,620.00
Pressure Transducers 3 $400.00 $1,200.00 31.00% $1,572.00
Computer 1 $800.00 $800.00 31.00% $1,048.00
I-P Converters 5 $300.00 $1,500.00 31.00% $1,965.00
PLC 1 $1,000.00 $1,000.00 31.00% $1,310.00
I/O Cards 2 $600.00 $1,200.00 31.00% $1,572.00
Pneumatic Control Valves 3 $400.00 $1,200.00 31.00% $1,572.00
Instrument Air System 1 $300.00 $300.00 31.00% $393.00
Controls Software 1 $600.00 $600.00 31.00% $786.00
Control Cabinet 1 $600.00 $600.00 31.00% $786.00
Assembly 1 $4,360.00 $4,360.00 $5,670 31.00% $5,711.60 $7,428
Piping (incl. insulation) 1 $1,310.00 $1,310.00 31.00% $1,716.10
10atm SMR RS Total $123,545
29

Natural Gas Feed

Natural Gas Compressor 1 $1,225.00 $1,225.00 $2,560 31.00% $1,604.75 $3,354
System
Explosion proof 1 $1,200.00 $1,200.00 31.00% $1,572.00
Motor
Heat Exchanger (to heat gas

Sulfur Removal Outer pipe SS 316 2.5"Diam, Sched40, 12"Long 1 $39.53 $39.53 $4,028 31.00% $51.79 $5,277
going into HDS)
Inner tube Haynes 556, SS fins 1"Diam, 23"Long 1 $104.79 $104.79 31.00% $137.27
Reducers Incoloy 800H 2.5"Diam 2 $65.80 $131.60 31.00% $172.39
NG inlet/outlet Incoloy 800H 2 $12.50 $25.00 31.00% $32.75
Insulation CCMax4 1" Thick 1 $86.17 $86.17 31.00% $112.89
HX assembly 1 $139.92 $139.92 31.00% $183.29
Hydro-desulfurization Bed HDS Vessel SS 316, 4" Pipe 32"Long 2 $286.17 $572.34 31.00% $749.77
Flanges SS 316, 300lb 4 $161.56 $646.23 31.00% $846.56
HDS Insulation CCMax4 1/4" Thick 2 $205.15 $410.29 31.00% $537.48
HDS catalyst Co_W 1.8 L 2 $47.37 $94.74 31.00% $124.11
Vessel Assembly 2 $120.75 $241.51 31.00% $316.38
Flanges SS 316, 300lb 4 $161.56 $646.23 31.00% $846.56

Flange gaskets 2 $6.50 $13.00 31.00% $17.03
Nuts/Bolts 1.25"Long 16 $9.75 $156.00 31.00% $204.36
Pipe Tees SS 316 2 $45.00 $90.00 31.00% $117.90
Pipe Elbows SS 316 4 $33.00 $132.00 31.00% $172.92
Pipe SS 316 1"pipe, Sched40 2.7 $12.00 $32.40 31.00% $42.44
Boiler/Steam Super
Boiler Water Inlet Header End-dome Incoloy 800H 16.5"Diam, 0.3"Thick 1 $562.34 $562.34 $14,057 31.00% $736.66 $18,415
Heater
Tubesheet Incoloy 800H 16.5"Diam, 0.5"Thick 1 $848.90 $848.90 31.00% $1,112.06
Boiler Steam Outlet Header End-dome Incoloy 800H 16.5"Diam, 0.35"Thick 1 $658.31 $658.31 31.00% $862.39
Tubesheet Incoloy 800H 16.5"Diam, 2"Thick 1 $2,760.80 $2,760.80 31.00% $3,616.65
Boiler Core Boiler Shell SS 316 16.5"Diam, 32"Long, 1/16"Thick 1 $398.99 $398.99 31.00% $522.67
Baffles SS 316 1/8" thick 2 $58.63 $117.26 31.00% $153.61
Tubes Incoloy 800H, SS fins 1"Diam, 32"Long 64 $108.66 $6,954.26 31.00% $9,110.08
Boiler Insulation ceramic fiber 2" thick 1 $62.14 $62.14 31.00% $81.41
Flue gas inlet and outlet SS 316 2 $10.51 $21.02 31.00% $27.54
Boiler Assembly 1 $1,672.89 $1,672.89 31.00% $2,191.49
Steam Reforming
SR Reactor Inlet Header End-dome SS 316 14"Diam, 0.61"Thick 1 $234.73 $234.73 $21,943 31.00% $307.50 $28,746
Reactor
Tubesheet Inc 800H 14"Diam, 1.52"Thick 1 $1,818.44 $1,818.44 31.00% $2,382.15
Steam/NG Mixing Port SS 1 $200.00 $200.00 31.00% $262.00
SR Reactor Outlet Header End-dome Haynes 556 14"Diam, 1.94"Thick 1 $2,138.36 $2,138.36 31.00% $2,801.25
Tubesheet Haynes 556 14"Diam, 2.85"Thick 1 $5,126.23 $5,126.23 31.00% $6,715.36
SR Reactor Core SR Reactor Shell Incoloy 800H 14"Diam, 40"Long, 1/16"Thick 1 $671.11 $671.11 31.00% $879.15
SR Reactor Baffles Incoloy 800H 1/8" thick 4 $94.46 $377.84 31.00% $494.97
SR Reactor Tubes Haynes 556 1/2"Diam, 39.4"Long 134 $29.75 $3,986.73 31.00% $5,222.61
SR Catalyst Ru/Ni-AlO 10.9 L 1 $850.20 $850.20 31.00% $1,113.76
SR Reactor Insulation CCMax4; Fiber board internal,external,header 1 $1,380.97 $1,380.97 31.00% $1,809.07
Flue gas inlet and outlet Haynes 556 2 $16.27 $32.53 31.00% $42.62
SR Reactor Assembly 1 $3,558.28 $3,558.28 31.00% $4,661.34
Burner Main Burner Assembly 1 $682.00 $682.00 31.00% $893.42
Natural Gas/Air Control
1 $886.00 $886.00 31.00% $1,160.66
Valve
Figure 4-11. Bill of Materials for 20 atm SMR RS

(1 of 3)
30

High Temperature
Reformate cooler HEX Shell SS 316 10"Diam, 18"Long 1 $284.82 $284.82 $15,720 31.00% $373.12 $20,593
Shift
Baffles SS 316 1/8" thick 2 $12.66 $25.32 31.00% $33.17
Tubes Haynes 556, SS fins 1"Diam, 18"Long 13 $64.60 $839.76 31.00% $1,100.08
Inlet Tubesheet Haynes 556 10"Diam, 1.14"Thick 1 $672.65 $672.65 31.00% $881.17
Inlet End-dome Haynes 556 10"Diam, 0.38"Thick 1 $340.16 $340.16 31.00% $445.61
Outlet Tubesheet Incoloy 800H 10"Diam, 0.76"Thick 1 $321.79 $321.79 31.00% $421.54
Outlet End-dome Incoloy 800H 10"Diam, 0.19"Thick 1 $141.61 $141.61 31.00% $185.51
Air inlet/outlet SS 316 2 $10.51 $21.02 31.00% $27.54
External insulation Ceramic fiber 2" Thick 1 $20.41 $20.41 31.00% $26.74
Blower Expl. proof motor 1 $1,200.00 $1,200.00 31.00% $1,572.00
Motor Starter 1 $250.00 $250.00 31.00% $327.50
HEX assembly 1 $699.39 $699.39 31.00% $916.20
31.00% $0.00
HTS Reactor Inlet Header Tubesheet SS 316 15.25"Diam, 1.86"Thick 1 $902.59 $902.59 31.00% $1,182.39
HTS Reactor Outlet Header Tubesheet SS 316 15.25"Diam, 1.72"Thick 1 $873.50 $873.50 31.00% $1,144.29
HTS Reactor Core HTS Reactor Shell SS 316 15.25"Diam, 75" Long, 1/16"Thick 1 $579.24 $579.24 31.00% $758.80
HTS Reactor Baffles SS 316 3/16" thick 6 $56.24 $337.44 31.00% $442.05
HTS Reactor Tubes SS 316 1"Diam, 75"Long 106 $17.11 $1,813.22 31.00% $2,375.31
HTS Reactor Catalyst Fe/Cr oxide 83 kg 1 $854.46 $854.46 31.00% $1,119.35
HTS Reactor Insulation Ceramic Fiber 2" Thick 1 $128.89 $128.89 31.00% $168.84
Air flow ports SS 316 2 $10.51 $21.02 31.00% $27.54
Blower Blower Expl. proof motor 1 $1,200.00 $1,200.00 31.00% $1,572.00
Motor Starter 1 $250.00 $250.00 31.00% $327.50
HTS reactor assembly 1 $3,249.04 $3,249.04 31.00% $4,256.25
Condenser
Heat Exchanger + fan 1 $2,500.00 $2,500.00 $3,610 31.00% $3,275.00 $4,729
Assembly
Motor + motor starter 1 $600.00 $600.00 31.00% $786.00
Water Pump Positive displacement 1.5 L/min, 300 psig 1 $900.00 $900.00 31.00% $1,179.00
Membrane Membrane Preheater Membrane Preheater 1 $2,500.00 $2,500.00 $18,764 31.00% $3,275.00 $24,581
Membrane Elements Active Membrane Pd/Cu, SS 8" discs 46 $312.97 $14,396.81 31.00% $18,859.82
Gaskets Grafoil 0.5" ID, 1" OD, 1/16" thick 141 $0.44 $61.45 31.00% $80.50
Baffles SS 316 8.5"Diam 47 $6.49 $304.83 31.00% $399.33
Disc spacers SS 316 94 $2.60 $244.40 31.00% $320.16
H2 collection tube porous sintered 316SS 0.5" OD 1 $20.00 $20.00 31.00% $26.20
Outside cylinder SS 316 8.5"Diam, 3'Long 1 $159.05 $159.05 31.00% $208.35
Inlet end dome SS 316 8.5"Diam 1 $71.35 $71.35 31.00% $93.47
Outlet end dome SS 316 8.5"Diam 1 $71.35 $71.35 31.00% $93.47
Inlet/outlet ports SS 316 1 $11.58 $11.58 31.00% $15.17
Assembly 1 $173.34 $173.34 31.00% $227.07
Raffinate Surge Tank Vessel 1 $500.00 $500.00 31.00% $655.00
Regulator SS 316 5 psi 1 $150.00 $150.00 31.00% $196.50
31

Mass Flow Meter 1 $2,000.00 $2,000.00 31.00% $2,620.00
Computer 1 $800.00 $800.00 31.00% $1,048.00

I-P Converters 5 $300.00 $1,500.00 31.00% $1,965.00
PLC 1 $1,000.00 $1,000.00 31.00% $1,310.00
I/O Cards 2 $600.00 $1,200.00 31.00% $1,572.00
Pneumatic Control Valves 3 $400.00 $1,200.00 31.00% $1,572.00
Control Cabinet 1 $600.00 $600.00 31.00% $786.00
Assembly 1 $4,360.00 $4,360.00 $6,325 31.00% $5,711.60 $8.285
20atm SMR RS Total $137,036
32
4.3 ATR System Flowsheets, Cost and Performance
Autothermal reforming systems were designed for nominal 10 atm operation with PSA H2
separation and 20 atm operation with membrane H2 separation. The hydrogen purification
processes are described in detail in Section 5. The ATR processing conditions are listed in
Figure 4-12 and the flow sheet is shown in Figure 4-13. The steam to carbon ratio for the ATR
design is set so that the heat duty to generate steam matches the heat duty to cool the reformate
to HTS temperature. In both systems the S/C ratio is approximately 1 (see Section 4.3.2). The
fuel-to-air ratio was set to allow enough combustion to result in an 800ºC reformate temperature
after the steam reforming section.
Pressure
10 atm ATR 1000 kPa (nominal)
20 atm ATR 2000 kPa (nominal)
Temperatures
HDS/ZnO 400ºC
Reformer Inlet 407ºC
Reformer Outlet 800ºC
Shift Inlet 450ºC
Shift Outlet 250ºC
Condenser Outlet 55ºC
Miscellaneous
S/C ratio 0.97-1.1
Oxygen/fuel ratio 0.56
Figure 4-12. ATR Process Conditions
4.3.1 Reactor and Vessel Sizing
The methodology used to size the SMR vessels was also used to design the ATR vessels. In
most cases, the ATR reactors are larger than the corresponding SMR reactors because the
dilution effect of nitrogen reduces the effective rate of reaction.
Natural Gas Pretreatment
The sulfur removal process design for the ATR is identical to that for the SMR (see Figure 4-2
and Figure 4-3). The ATR HDS and ZnO beds are larger sizes for the ATR because of the
higher natural gas rates needed to achieve the equivalent hydrogen production.
Reformer
The reformer reactor in the ATR system is a single chamber, unlike the multiple-tube design in
the SMR. The process stream first passes through a Pt-coated ceramic monolith to catalyze
partial combustion of the natural gas and generate heat. The process stream then proceeds
through a packed bed of catalyst similar to the SMR catalyst but having lower surface area to
better withstand the higher temperatures that result from combustion at the bed inlet. Figure
4-14 shows the conceptual design for the ATR vessel.
33
Relief Valve
Natural Gas
Main Shutoff Valve CIty
Water Boiler
DI Water (25C)
Filter
Steam (450C)
Temp
Natural Gas
to Reforming
Natural Gas
Flow Control Valve Nat. Gas (400C)
Natural Gas
Compressor Air
Compressor
HDS/ZnO
Reformate (760C)
F ATR
Ambient air
Water
Reformate (800C)
P
Temp DI Bed
HDS
Preheater Nat. Gas (400C)
Vent
Reformate (800C)
Reformate
Relief Valve
Cooler
Burner
Hydrogen Relief T
Valve Dry Gas
Reformate (450C)
Storage
(55C)
Air (66C)
Raffinate Vent Burner Air

(55C) Control Valve
Water
Knockout HTS
Temp
P Hydrogen
Condenser
Compressor Shift Gas (250C)
Water (55C)
Hydrogen PSA
Pressure-Reducing
Valve
Ambient air
Figure 4-13. ATR Flow Sheet
34
Alumina
Combustion
Monolith
Reforming
Catalyst
Insulation
Figure 4-14. ATR Vessel (not to scale)

(External insulation not shown.)
Since the endothermic heat of reaction is supplied by combustion in the process stream, one
design objective for the ATR reformer is to minimize the external surface area per unit volume
subject to the constraint that increasing reactor diameter increases fabrication cost. Reducing the
surface-to-volume ratio is desirable because it decreases the area available for heat loss to the
surroundings. The proposed design is a 25.4 cm (10 in.) inner diameter shell. A 3.8 cm layer of
low thermal conductivity insulation lines the inside of the shell to reduce heat flux from the hot
reformer to the shell metal. The reduced heat flux allows the shell to be made from thinner
metal, which results in significant cost savings.
Shift Reactor
The ATR shift reactor design is identical to the SMR design, except in size. The ATR requires a
larger shift reactor than the SMR because of the dilution effect of nitrogen on the reaction rate.
The 10 atm ATR shift reactor is 45% larger than the 10 atm SMR shift reactor, and the 20 atm
ATR shift reactor is 67% larger than the 20 atm SMR shift reactor.
4.3.2 Heat Exchanger Sizing
HDS Preheater
The design basis for the ATR HDS preheater is the same as for the SMR (see Figure 4-6). The
heat exchanger surface areas for both 10 atm and 20 atm are approximately 30% larger than the
SMR areas to account for higher rates of natural gas use.
35
Boiler/Reformate Cooler
The reformate stream is cooled to 450ºC for the water-gas shift reaction by giving up heat to
generate steam for the ATR reformer. The steam feed rate was chosen so that the heat duties for
these two processes are equal. The resulting S/C ratios are 1.08 for 10 atm and 0.97 for 20 atm.
The boiler/reformate cooler is a multiple finned-tube, tube and shell design.
Condenser
The condenser design is the same as for the SMR system (see Section 4.2.2). The condenser
duty in the ATR systems is lower than in the SMR systems because the low S/C ratio in the ATR
leaves less excess water to be removed.
Water Knockout Vessel
A vessel is needed to separate condensed water before sending the reformate gas to the gas
cleanup system. The knockout vessel is designed to reduce the vapor velocity to approximately
1 m/s so that the water is removed by gravity separation. A one-stage separator provides a clean
separation since none of the other components are present in significant quantities in the water
phase. The proposed design is section of 4” steel pipe with welded end caps and welded inlet
tap.
4.3.3 Water Purification
The water requirement for steam fed to the ATR systems is approximately 0.5-1 liters/minute.
The basic design for the ATR deionized water system is the same as for the SMR. The initial
cost is $1500, and the resin replacement cost is approximately $700 per year. The resin
replacement cost is less for the ATR than for the SMR because of the lower water requirement.
The volume of water recovered from the condenser after the high-temperature shift reactor will
be approximately 27% of the volume fed to the reformer. This water may contain trace levels of
metals picked up from the catalysts and my not be suitable for direct discharge to municipal
sewer. The proposed design recycles the condensate and mixes it with fresh city water to feed
the DI bed with the necessary water rate.
4.3.4 Other Equipment
Fans and Blowers
Costs for cooling fans and blowers were estimated from catalog prices for standard items. The
fans with associated equipment (motors and motor starters) are included in the bills of materials
for the heat exchangers.
36
Instruments
The instruments included in the design are only those necessary to assure proper operation for a
well-established process, such as thermocouples and pressure transducers to provide signals to
automatic control valves (see Figure 4-13). The first several installations will have a higher level
of instrumentation to verify the design conditions and to collect data for process improvements.
Instrument lists for the 10 atm and 20 atm systems are identical and are included with the system
bills of materials.
Pallet
The ATR RS layout would be similar to that of the SMR, shown in Figure 4-9. While the ATR
system does not contain a separate reformate cooler (as does the SMR), the PSA system in the
10atm ATR is significantly larger, resulting in a similarly sized footprint to the 10atm SMR.
Likewise, the 20atm ATR is similarly sized to the 20atm SMR.
4.3.5 Assembly
Like the SMR Assembly (see Section 4.2.5), eight man-days are required for assembly and
pressure testing of the reformer system.
4.3.6 Alternate ATR Designs
A sensitivity analysis of the effect of S/C on thermal efficiency was conducted using the
HYSYS™ simulation. Thermal efficiency decreases for lower S/C ratios because all or part of
the raffinate heat value is unrecoverable. (Raffinate is the waste stream from gas cleanup.)
Increasing the S/C ratio to 3 increases the thermal efficiency by ~3.5% but requires the use of a
separate burner and boiler to create heat for steam generation. The increase in efficiency is a
result of higher methane conversion caused by a shift in equilibrium and recovery of the raffinate
heat value in the burner. The boiler and the reformate cooler would be two separate heat
exchangers similar to the SMR design. The maximum cost differential for the added equipment
that would return the cost of capital based on the natural gas savings is ~$4700, but the estimated
increase in capital cost for the separate boiler and reformate cooler is $7500 (from the SMR
design). The sizes of both the catalytic reforming and the shift reactors would also increase in
size due to the higher total flow rate.
4.3.7 ATR Reformer System Bill of Materials and Cost
The overall ATR RS bill of materials is shown for a 10 atm system in Figure 4-15 and for a 20
atm system in Figure 4-16. The cost for each reforming system (RS) includes the gas cleanup
system, which is described in detail in Section 5. The initial capital cost for the 10atm ATR-PSA
RS is $102,384, while that for the 20atm ATR-membrane RS is $115,499.
.
37

Natural Gas Feed System Natural Gas Compressor 1 $1,225.00 $1,225.00 $2,260 31.00% $1,604.75 $2,961
Explosion proof 5 kW 1 $900.00 $900.00 31.00% $1,179.00
Motor
Heat Exchanger (to heat 2.5" Diam, Sched 40,

Sulfur Removal Outer pipe SS 316 1 $50.19 $50.19 $3,800 31.00% $65.75 $4,978
gas going into HDS) 16"Long
Inner tube Haynes 556, SS fins 1"OD, 21"Long 1 $90.16 $90.16 31.00% $118.11
Reducers Inc 800H 2.5"Diam 2 $65.80 $131.60 31.00% $172.39
NG inlet/outlet Inc 800H 2 $11.31 $22.63 31.00% $29.64
Insulation CCMax4 1" thick 1 $106.97 $106.97 31.00% $140.12
HX assembly 1 $127.98 $127.98 31.00% $167.65
Hydro-desulfurization Bed HDS Vessel SS 316, 4" Pipe 40" length 2 $358.05 $716.10 31.00% $938.09
Flanges SS 316, 150lb flange 8 $107.71 $861.64 31.00% $1,128.75
HDS Insulation CCMax4 1/4"Thick 2 $253.41 $506.81 31.00% $663.93
HDS catalyst Ni_W 2.3 L 2 $60.53 $121.06 31.00% $158.59
Vessel Assembly 2 $120.75 $241.51 31.00% $316.38
Flange gaskets 4 $6.50 $26.00 31.00% $34.06
Nuts/Bolts 16 $6.50 $104.00 31.00% $136.24
Pipe Tees 2 $30.00 $60.00 31.00% $78.60
Pipe elbows 4 $22.00 $88.00 31.00% $115.28
Pipe SS 316 pipe 1"pipe, Sched40 2.7 $12.00 $32.40 31.00% $42.44
Boiler/Steam Super Heater Boiler Inlet Header End-dome SS 316 8"Diam 1 $82.47 $82.47 $4,132 31.00% $108.04 $5,413
Tube-Sheet SS 316 8"Diam, 1/2"Thick 1 $86.65 $86.65 31.00% $113.51
Boiler Outlet Header End-dome Inc 800H 8"Diam 1 $158.88 $158.88 31.00% $208.13
Tube-Sheet Haynes 556 8"Diam, 1"Thick 1 $472.65 $472.65 31.00% $619.17
Boiler Core Boiler Shell Haynes 556 8"Diam, 21.3"Long 1 $2,156.78 $2,156.78 31.00% $2,825.38
Baffles Inc 800H 1/8"Thick 2 $30.17 $60.34 31.00% $79.05
Tubes Inc 800H 1"OD, 21.3"Long 7 $63.37 $443.58 31.00% $581.09
Boiler Insulation CCMax4; Ceramic Fiber 1 $27.03 $27.03 31.00% $35.41
Reformate inlet and outlet 2 $26.34 $52.68 31.00% $69.02
Boiler Assembly 1 $590.95 $590.95 31.00% $774.15
Autothermal reactor AT Reactor Inlet Header End-Dome Inc 800H 10"Diam 1 $101.11 $101.11 $11,280 31.00% $132.46 $14,777
Steam/NG/Air Mixing
1 $200.00 $200.00 31.00% $262.00
Port
AT Reactor Outlet Header End-Dome Haynes 556 10"diam 1 $387.93 $387.93 31.00% $508.19
10"Diam, 48"Long,
AT Reactor Core AT Reactor Shell Incoloy 800H 1 $2,055.50 $2,055.50 31.00% $2,692.70
0.25"Thick
ATR Inside Insulation CCMax4 1.5" thick 1 $860.14 $860.14 31.00% $1,126.79
Pt/Al2O3 washcoated
Combustion Monolith 1.5 L 1 $118.75 $118.75 31.00% $155.57
coerdierite
Reforming Catalyst Ru/Ni-Al 32.4 L 1 $2,527.20 $2,527.20 31.00% $3,310.63
ATR Outside Insulation Ceramic fiber 1/2" thick 1 $236.34 $236.34 31.00% $309.60
ATR Assembly 1 $277.39 $277.39 31.00% $363.38
ATR Thermocouples 2 $40.00 $80.00 31.00% $104.80
Flame starter Electric heat tube 1 $36.00 $36.00 31.00% $47.16
Air System Air Compressor 47 scfm @ 150 psig 1 $4,000.00 $4,000.00 31.00% $5,240.00
Air Control Valve 1 $400.00 $400.00 31.00% $524.00
Figure 4-15. Bill of Materials for 10 atm ATR RS

(1 of 2)
38

High Temperature Shift HTS Inlet Header End-dome SS 316 17.25" Diam 1 $53.10 $53.10 $10,948 31.00% $69.57 $14,341
Header Plate SS 316 17.25"Diam, 1.5"Thick 1 $1,055.34 $1,055.34 31.00% $1,382.50
HTS Bottom Header End-dome SS 316 17.25" Diam 1 $128.05 $128.05 31.00% $167.74
Header Plate SS 316 17.25"Diam, 1.5"Thick 1 $1,079.69 $1,079.69 31.00% $1,414.40
HTS Reactor Core HTS Reactor Shell SS 316 17.25"Diam, 93"Long 1 $713.91 $713.91 31.00% $935.23
HTS Reactor Baffles SS 316 3/16" Thick 6 $54.89 $329.34 31.00% $431.43
HTS Reactor Tubes SS 316 1"Diam, 93"Long 147 $12.13 $1,783.15 31.00% $2,335.92
HTS Reactor Catalyst Fe/Cr oxide 151 L (188kg) 1 $1,938.46 $1,938.46 31.00% $2,539.39
HTS Reactor Insulation Ceramic Fiber 2"Thick 1 $176.30 $176.30 31.00% $230.96
Air inlet/outlet ports SS 316 2 $10.51 $21.02 31.00% $27.54
HTS Assembly 1 $3,669.17 $3,669.17 31.00% $4,806.61
Heat Exchanger + fan +

Condenser Assembly 1 $3,000.00 $3,000.00 $3,366 31.00% $3,930.00 $4,409
variable speed motor
Water Pump Positive displacement 1.5 L/min @ 150 psig 1 $400.00 $400.00 31.00% $524.00
Pressure Swing Adsorption Adsorption Vessels Top End-Dome SS 316 16"Diam 4 $142.05 $568.20 $20,410 31.00% $744.34 $26,738
Bottom End-Dome SS 316 16"Diam 4 $142.05 $568.20 31.00% $744.34
Vessel Shell SS 316 16"Diam, 108"Long 4 $1,013.11 $4,052.44 31.00% $5,308.69
Pressure Relief Valves 4 $100.00 $400.00 31.00% $524.00
Activated Carbon
162L 4 $406.41 $1,625.65 31.00% $2,129.60
Adsorbent
Zeolite 13X Adsorbent 162L 4 $629.01 $2,516.05 31.00% $3,296.02
Silica gel 34L 4 $354.02 $1,416.06 31.00% $1,855.04
Vessel assembly 4 $468.95 $1,875.80 31.00% $2,457.30
PSA Valves PSA Valves 27 varies $3,556.00 31.00% $4,658.36
Piping SS 316 3/4" pipe, sched 40 1 $803.20 $803.20 31.00% $1,052.19
Raffinate Surge Tank Vessel 250 gal, 150psi 1 $500.00 $500.00 31.00% $655.00
Raffinate Burner Main Burner Assembly 1 $700.00 $700.00 31.00% $917.00
Regulator 316 SS 1 $150.00 $150.00 31.00% $196.50
PSA Vessel Rack Stainless Steel 4.5'x1.5'x8' 1 $400.00 $400.00 31.00% $524.00
Subsystem Assembly 1 $1,178.88 $1,178.88 31.00% $1,544.33

Mass Flow Meter 1 $2,000.00 $2,000.00 31.00% $2,620.00
Computer 1 $800.00 $800.00 31.00% $1,048.00
I-P Converters 5 $300.00 $1,500.00 31.00% $1,965.00
PLC 1 $1,000.00 $1,000.00 31.00% $1,310.00
I/O Cards 2 $600.00 $1,200.00 31.00% $1,572.00
Pneumatic Controls 3 $400.00 $1,200.00 31.00% $1,572.00
Pressure Relief Valves 3 $50.00 $150.00 31.00% $196.50
Control Cabinet 1 $600.00 $600.00 31.00% $786.00
Assembly 1 $4,360.00 $4,360.00 $5,670 31.00% $5,711.60 $7,428

10atm ATR RS Total $102,384
Figure 4-15. Bill of Materials for 10 atm ATR RS (2 of 2)
39

Natural Gas Feed

Natural Gas Compressor 1 $1,225.00 $1,225.00 $2,560 31.00% $1,605 $3,354
System
1 $1,200.00 $1,200.00 31.00% $1,572
Motor
Manual NG Shut-off Valve 1 $15.00 $15.00 31.00% $20
Natural Gas Control Valve 1 $120.00 $120.00 31.00% $157
Heat Exchanger (to heat $5,777

Sulfur Removal Outer pipe SS 316 2.5"Diam, 12.5"Long 1 $39.53 $39.53 $4,410 31.00% $52
gas going into HDS)
Inner tube Haynes 556, SS fins 1"Diam, 12"Long 1 $104.79 $104.79 31.00% $137
Reducers Inc 800H 2.5"Diam 2 $65.80 $131.60 31.00% $172
NG inlet/outlet SS 316 2 $12.50 $25.00 31.00% $33
Insulation CCMax4 1"Thick 1 $86.17 $86.17 31.00% $113
HX assembly 1 $139.92 $139.92 31.00% $183
Hydro-desulfurization Bed HDS Vessel SS 316, 4" Pipe 40"Long 2 $446.28 $892.57 31.00% $1,169
Flanges SS 316, 600lb 8 $140.02 $1,120.13 31.00% $1,467
HDS Insulation CCMax4 1/4"Thick 2 $313.73 $627.46 31.00% $822
HDS catalyst Ni_W 2.8 L 2 $73.69 $147.38 31.00% $193
Zinc Oxide Adsorbent Zinc oxide 6.6 L 2 $78.26 $156.52 31.00% $205
Vessel Assembly 2 $120.75 $241.51 31.00% $316
Flange gaskets 4 $6.50 $26.00 31.00% $34
Nuts/Bolts 16 $6.50 $104.00 31.00% $136
Manual shutoff valves 4 $40.00 $160.00 31.00% $210
Pipe Tees 2 $30.00 $60.00 31.00% $79
Pipe elbows 4 $22.00 $88.00 31.00% $115
Pipe SS 316 pipe 1"pipe 2.7 $12.00 $32.40 31.00% $42
Subsystem Assembly 1 $227.17 $227.17 31.00% $298
Boiler/Reformate $6,736
Boiler Inlet Header End-dome Haynes 556 8"Diam 1 $393.35 $393.35 $5,142 31.00% $515
Cooler
Tube-Sheet Haynes 556 8"Diam, 1.7"Thick 1 $952.36 $952.36 31.00% $1,248
Boiler Outlet Header End-dome Inc 800H 8"Diam 1 $312.60 $312.60 31.00% $410
Tube-Sheet Inc 800H 8"Diam, 2.1"Thick 1 $676.77 $676.77 31.00% $887
Boiler Core Boiler Shell Inc 800H 8"Diam, 21.3"Long 1 $1,306.78 $1,306.78 31.00% $1,712
Baffles Inc 800H 1/8"Thick 2 $30.10 $60.19 31.00% $79
Tubes Haynes 556, SS fins 1"OD, 21.3"Long 7 $114.33 $800.33 31.00% $1,048
Boiler Insulation Ceramic Fiber 2"Thick 1 $21.31 $21.31 31.00% $28
Reformate inlet and outlet Inc 800H 2 $13.53 $27.05 31.00% $35
Boiler Assembly 1 $590.95 $590.95 31.00% $774
Autothermal reactor AT Reactor Inlet Header End-Dome Inc 800H 10"diam 1 $222.91 $222.91 $10,410 31.00% $292 $13,637
Steam/NG/Air Mixing
SS 1 $200.00 $200.00 31.00% $262
Port
AT Reactor Outlet Header End-Dome Haynes 556 10"diam 1 $869.37 $869.37 31.00% $1,139
10"Diam, 28.5"Long,
AT Reactor Core AT Reactor Shell Incoloy 800H 1 $2,296.38 $2,296.38 31.00% $3,008
0.5"Thick
ATR Inside Insulation CCMax4 1"Thick 1 $500.17 $500.17 31.00% $655
Pt/Al2O3 washcoated
Combustion Monolith 1.4 L 1 $118.75 $118.75 31.00% $156
cordierite
Reforming Catalyst Ru/Ni-Al 16.2 L 1 $1,263.60 $1,263.60 31.00% $1,655
ATR Outside Insulation CCMax4 1/4"Thick 1 $145.19 $145.19 31.00% $190
ATR Assembly 1 $277.39 $277.39 31.00% $363
ATR Thermocouples 2 $40.00 $80.00 31.00% $105
Flame starter Electric Tube Heater 1 $36.00 $36.00 31.00% $47
Air System Air Compressor 47 scfm @ 300 psig 1 $4,000.00 $4,000.00 31.00% $5,240
Air Control Valve 1 $400.00 $400.00 31.00% $524
High Temperature $13,813

End-dome SS 316 15.25"Diam 1 $161.20 $161.20 $10,544 31.00% $211
Shift HTS Inlet Header
Tube-Sheet SS 316 15.25"Diam, 1.9"Thick 1 $904.75 $904.75 31.00% $1,185
HTS Outlet Header End-dome SS 316 15.25"Diam 1 $146.79 $146.79 31.00% $192
Tube-Sheet SS 316 15.25"Diam, 1.7"Thick 1 $890.59 $890.59 31.00% $1,167
HTS Reactor Core HTS Reactor Shell SS 316 15.25" ID 1 $666.04 $666.04 31.00% $873
HTS Reactor Baffles SS 316 3/16" Thick 1 $341.50 $341.50 31.00% $447
HTS Reactor Tubes SS 316 1"Diam, 94"Long 106 $27.68 $2,934.17 31.00% $3,844
HTS Reactor Catalyst Fe/Cr oxide 111 L (139kg) 1 $1,427.53 $1,427.53 31.00% $1,870
HTS Reactor Insulation Ceramic Fiber 2"Thick 1 $160.83 $160.83 31.00% $211
Air inlet/outlet ports SS 316 2 $10.51 $21.02 31.00% $28
HTS Assembly 1 $2,889.92 $2,889.92 31.00% $3,786
Condenser Heat Exchanger + fan + $4,409

1 $3,000.00 $3,000.00 $3,366 31.00% $3,930
Assembly variable speed motor
Water Knock-out Tank 1 $365.56 $365.56 31.00% $479
Water Purification Main Water Shut-off Valve 1 $40.00 $40.00 $1,940 31.00% $52 $2,541
Water Pump Positive displacement 1.5 L/min @ 300 psig 1 $400.00 $400.00 31.00% $524
Deionization Bed DI Canister 1 $1,500.00 $1,500.00 31.00% $1,965
Figure 4-16. Bill of Materials for 20 atm ATR RS

(1 of 2)
40

Membrane Membrane Preheater Membrane Preheater 1 $4,000.00 $4,000.00 $28,521 31.00% $5,240 $37,362
Membrane Elements Active Membrane Pd/Cu alloy, 316SS 8" discs 84 $261.66 $21,979.43 31.00% $28,793
0.5" ID, 1" OD, 1/16"
Gaskets Grafoil 168 $1.07 $179.67 31.00% $235
thick
Baffles 316SS 8.5" OD 84 $6.49 $544.80 31.00% $714
Disc spacers 316SS 168 $2.60 $436.80 31.00% $572
H2 collection tube porous sintered 316SS 0.5" OD 1 $45.00 $45.00 31.00% $59
Outside cylinder 316SS 8.5" ID 1 $232.08 $232.08 31.00% $304
Inlet end dome 316SS 1 $71.35 $71.35 31.00% $93
Outlet end dome 316SS 1 $71.35 $71.35 31.00% $93
Inlet/outlet ports 316SS 2 $5.79 $11.58 31.00% $15
Assembly 1 $198.89 $198.89 31.00% $261
Raffinate Surge Tank Vessel 1 $500.00 $500.00 31.00% $655
Pressure Relief Valve 1 $100.00 $100.00 31.00% $131
Regulator SS 316 5 psi 1 $150.00 $150.00 31.00% $197
Controls System Thermocouples 5 $40.00 $200.00 $10,750 31.00% $262 $14,083

Mass Flow Meter 1 $2,000.00 $2,000.00 31.00% $2,620
Pressure Transducers 3 $400.00 $1,200.00 31.00% $1,572
Computer 1 $800.00 $800.00 31.00% $1,048
I-P Converters 5 $300.00 $1,500.00 31.00% $1,965
PLC 1 $1,000.00 $1,000.00 31.00% $1,310
I/O Cards 2 $600.00 $1,200.00 31.00% $1,572
Pneumatic Control Valves 3 $400.00 $1,200.00 31.00% $1,572
Instrument Air System 1 $300.00 $300.00 31.00% $393
Pressure Relief Valve 3 $50.00 $150.00 31.00% $197
Controls Software 1 $600.00 $600.00 31.00% $786
Control Cabinet 1 $600.00 $600.00 31.00% $786
Assembly 1 $4,360.00 $4,360.00 $6,325 31.00% $5,712 $8,286

Piping (incl. insulation) 1 $1,965.00 $1,965.00 31.00% $2,574
20atm ATR RS Total $115, 499
Figure 4-16. Bill of Materials for 20 atm ATR RS (2 of 2)
41
4.3.8 ATR Performance and Efficiency
The thermal efficiency of the 10 atm ATR is 56%, corresponding to a hydrogen production of
0.22 kg per kg of natural gas fed. The thermal efficiency of the 20 atm reformer is 46%, with a
hydrogen production of 0.17 kg per kg of natural gas fed.
5 GAS CLEANUP TECHNOLOGIES

5.1 Pressure Swing Adsorption (PSA)
5.1.1 PSA Gas Separation
Pressure swing adsorption (PSA) is a commonly used industrial process for the purification of
gas streams. The most common separation processes employing PSA are the purification of
either oxygen or nitrogen from air and the purification of hydrogen from sources such as
catalytic reformer off gas, coke oven gas, and ethylene plant effluent gas. Commercial
application of PSA for hydrogen purification began in 1966,32 and hydrogen production is
currently the largest use of PSA in terms of both cumulative capacity and number of installed
units, with the total hydrogen output from PSA being greater than 1 billion scf/day worldwide.33
In addition to this vast industrial experience with PSA hydrogen purification processes, pressure
swing adsorption units have been built in a wide variety of capacities (ranging from 100,000
Nm3/h hydrogen production units operating at large refineries,34 to small medical oxygen units
producing ~ 1Nm3/h35) testifying to the economic viability of the technique for the relatively
small production rate required for the hydrogen fueling station application.
Pressure swing systems are based on selective adsorbent beds.36 A gas mixture is introduced to
the bed at an elevated pressure and the solid adsorbent selectively “adsorbs” certain components
of the gas mixtures, allowing the unadsorbed components to pass through the bed as purified
product gas. There are two principle mechanisms for separating components of a gaseous
mixture by using a solid material: surface adsorption (in which certain gas molecules stick to the
solid surface due to preferential chemical interaction) and size selectivity (in which the
“adsorber” is a porous solid with pores just large enough to allow certain components into the
pellet, while larger molecules must bypass the pellet, thus larger molecules are the first to elute).
For the surface adsorption of a binary gas mixture, the selectivity of an adsorbent can be
expressed as separation factor, αij
X i / Yi
α ij = (Equation 5-1)
X j /Yj
32
H. A. Stewart and J. L. Heck, Chem. Eng. Prog., 65, 78 (1969).
33
R. T. Cassidy and E. S. Holmes, AIChE Symp. Ser. 80, 68 (1984).
34
Linde AG of Höllriegelskreuth Germany has hydrogen PSA plants running at over 100,000 Nm3/h.
35
CarboTech of Essen, Germany and QuestAir of Vancouver, Canada manufacture small nitrogen and oxygen PSA units.
36
R. T. Yang, Gas Separation by Adsorption Processes, Imperial College Press:London, 1997. provides a excellent introduction to pressure swing
adsorption.
42
in which Xi and Yi are the mole fractions of component i on the solid surface and in the gas
phase, respectively. Thus, a high value of αij means that species i is strongly adsorbed and
component j could be produced as a high-pressure gaseous product.
In the simplest case, the amount of gas adsorbed on a solid surface is proportional to the pressure
of the gas, and so at high pressure the molecules are strongly adsorbed while decreasing the
pressure removes the gas from the surface, thus “cleaning” the adsorbent. This process of
increasing pressure to increase adsorption of an impurity and decreasing pressure to renew the
adsorbent surface constitutes a “pressure swing”. The amount of gas adsorbed also decreases
with increasing temperature and so a temperature swing can be used to construct an adsorption
cycle. However, increasing the temperature of a bed is a long and energy intensive process, so
pressure swing cycles are preferred.
There are three basic quantities used to define the performance of a PSA process: product purity,
product recovery, and adsorbent productivity/utilization. Product purity is a volume-averaged
quantity since the impurity concentration in the effluent stream varies with time. The product
recovery is defined as the amount the desired species in the product stream divided by the total
amount of that component in the process feed stream. Finally, the adsorbent productivity is
measured by the amount of product processed per unit of adsorbent per unit time.
A plethora of complex PSA cycles have been developed to maximize product purity, product
recovery and adsorbent productivity, but almost all of these cycles are composed of a small set of
standard steps that include adsorption-production, cocurrent depressurization, countercurrent
depressurization, pressure equalization, and purge. Gas is fed to the PSA bed at elevated pressure
and the strongly adsorbed species is retained as the high-pressure product passes through the bed.
As the concentration of the adsorbed species increases, the adsorbent becomes saturated and the
bed must be regenerated before more gas can be processed. A cocurrent depressurization step, in
which the bed is blown-down through the product end, can be used to remove the adsorbed
species. An immediate cocurrent depressurization leaves residual contaminant near the product
end of the bed. This contaminant is swept into the product when the bed is cycled to the
adsorption-production step. Also, the cocurrent depressurization sends the product gas remaining
in the void space of the bed to the waste stream, significantly reducing product recovery. In order
to recover some of the product gas from the void spaces of the bed, the bed can be depressurized
in stages. Because the gas released in the early stages of depressurization contains very little
contaminant, it can be used to repressurize other beds in multi-bed systems. To simplify control
strategies, the gas exchange between beds is generally run until the pressure in both beds
equilibrates and so these steps are called “pressure equalizations”. To improve product purity,
the final depressurization of a bed is usually carried out in a countercurrent fashion (i.e. the gas
exits from the feed end of the bed). This forces remaining adsorbed species to the beginning of
the adsorbent bed, leaving the product end of the bed contaminant free. This effect can be
amplified by purging the depressurized bed with high purity product immediately prior to
repressurization, thus insuring that as much contaminant as possible has been removed.
43
5.1.2 PSA Design
PSA Cycle Selection
In large PSA systems, adsorbent cost is a significant factor, despite the relatively low cost of the
material (~ $2/lb), since tons of adsorbent are used in a single plant. Also, hydrogen recovery is
of paramount importance given the high production rates of the large plants. Sizable hydrogen
PSA units are therefore optimized for product recovery and adsorbent utilization. The resulting
PSA cycles, such as the Fuderer cycle,37 employ ten adsorption beds or more to allow for
multiple pressure equalization steps and to therefore improve product recovery. In smaller PSA
systems, however, the costs are driven by valve and vessel costs, making complex cycles too
capital intensive for the small gains in product recovery and adsorbent utilization that they offer.
Thus the simplest cycle that provides reasonable recovery is preferred for our low H2 production
rate of 2000 scf/h.
Aside from product recovery and adsorbent utilization, there are a number of other issues that we
have considered in selecting the PSA cycle, including product pressure and cyclic versus
continuous production. Two-bed pressure swing adsorption cycles offer the lowest level of
mechanical complexity and cost, but in order to produce hydrogen continuously, the two bed
cycles cannot be run with a pressure equalization stage, leading to unacceptable hydrogen
recovery. If the two-bed cycle is to operate with a pressure equalization step to improve
recovery, hydrogen will not be produced continuously and a surge tank will be required to create
a constant pressure reservoir from which the hydrogen compressor may draw. Also, cycles that
produce at pressures significantly lower than inlet (i.e. reformer) pressure were not considered as
low pressure production adds an unnecessary burden on the hydrogen compressor and therefore
decreases overall system efficiency. Given the considerations mentioned above, we have selected
the “Batta” pressure swing adsorption cycle as the baseline design for our units.38
The Batta cycle is a four-bed process that in its traditional form utilizes 35 different valves.
Wherever possible we have replaced two independent 2 way valves with a three way valve as we
have determined that a single three-way valve is not only less expensive than two two-way
valves but it also saves on the amount of piping and welding that must be done during system
assembly (see Section 5.1.3). Our revised Batta cycle valve diagram is shown in Figure 5-1.
37
Andrija Fuderer, Union Carbide, US patent # 3986849.
38
Louis B. Batta, Union Carbide, US patent # 3,564,816.
44
Figure 5-1. PSA Flow Diagram, Showing Valve Configuration
Figure 5-2. Batta Cycle Timing Chart
45
The Batta cycle utilizes two pressure equalizations, a cocurrent depressurization, a

countercurrent depressurization and a purge step to achieve over 75% hydrogen recovery for
properly sized beds. The cycle steps and timing are summarized in Figure 5-2. Using Bed A as
an example of the cycle, the process gas enters the bed from valve V1A in the feed manifold
(light blue in Figure 5-1). As the dry reformate is fed to the bed, the contaminants are adsorbed
and pure hydrogen flows out of the tank through valve V2A in the product manifold (red). When
hydrogen production stops in bed A, the contaminant front is well away from the product end of
the bed. The bed then undergoes a pressure equalization with bed B through valves V4AB and
V4T1 in the “pressure equalization I” manifold (green). Valve V4T1 (and V4T2) are throttling
valves that are manually preset to regulate the gas flow during pressure equalization and thereby
to prevent fluidization of the adsorbent bed. More of the hydrogen trapped in the voids of the bed
is recovered in the subsequent cocurrent depressurization step. The pure hydrogen gas released
during the cocurrent depressurization through valves V5AB, V5R1 and V5T1 is used to purge
bed C but by the end of this step, the contaminant front is near the end of the adsorbent bed.
Valve V5R1 is a back-pressure regulator that ensures the bed is only depressurized to the
specified level since the pressure in beds A and C is not equalized during this step. In the second
pressure equalization step, some of the contaminants are released from the product end of the
bed. This slightly impure gas is sent to the inlet of bed C to begin the repressurization of this bed.
Allowing some break-through of contaminants results in higher hydrogen recovery, since the gas
fed to bed C contains mostly hydrogen, and by feeding the pressure equalization gas to the inlet
of bed C, the cycle ensures that the outlet end of the bed is not contaminated immediately prior
to production. Following pressure equalization II, bed A is blown-down countercurrently to near
atmospheric pressure. This step removes a large amount of the adsorbed contaminant through
valve V3AB and into the waste manifold (dark blue). The waste gas is sent to a surge tank that
then feeds the burner (in the SMR case) to provide heat for the steam reforming reaction. Next,
bed A is countercurrently purged with pure hydrogen from bed D to remove additional adsorbed
contaminants and to force the rest of the contaminants to the feed end of the bed. The purge gas
is mixed with the countercurrent depressurization gas in the waste surge tank to provide a stream
of constant pressure and composition to the burner. After the purge step, bed A is repressurized
by gas from bed D, bed B, and from the product line (manifold 6, yellow in Figure 5-1),
respectively.
PSA Adsorbent Selection
The large amount of industrial experience with hydrogen purification via pressure swing
adsorption has lead to a widely accepted multi-adsorbent bed design. In hydrogen separation
from steam reformer gas, the feed end of the bed must contain a small layer of silica gel or other
robust desiccant in order to remove residual water from gas stream. The next layer of the
adsorbent bed contains activated carbon. The second half of the adsorbent bed is different
depending on if the PSA is to be used in a steam reforming or in an autothermal reforming
system. For SMR gas, the two most difficult impurities to remove are CO and CH4 and so zeolite
5A is used as the adsorbent. In ATR systems, nitrogen is the most difficult impurity to remove
and so zeolite 13X (also called 10A) is used.
In the SMR case, the activated carbon adsorbs more methane and CO2 than CO, and so the
carbon monoxide is the first impurity to reach the zeolite 5A segment of the bed. Since 5A is a
46
good adsorbent of CO, the carbon monoxide front stops at near the beginning of the zeolite bed.
As more gas is fed to the bed, the activated carbon eventually becomes saturated with CH4 and
the methane enters the 5A section of the bed, through which it travels more quickly than the CO
front. In the ideal case, the amounts of activated carbon and zeolite are balanced such that the
CO and CH4 contaminant fronts breakthrough at the same time and therefore neither segment of
the bed is under or over utilized.
In the ATR case, the activated carbon is again primarily used as a methane and carbon dioxide
adsorbent. Zeolite 13X adsorbs nitrogen well but is also a good carbon monoxide adsorbent. Due
to the large amount of nitrogen in the ATR reformate stream, the ratio of zeolite to activated
carbon should be greater than in the SMR PSA unit.
PSA Bed Sizing
The sizing of the pressure swing adsorption beds was based on experimental and theoretical
work published by Jee et al.39 Although the equilibrium adsorption capacity of a given adsorbent
for an individual contaminant is part of the sizing equation, determining the necessary amount of
adsorbent is also dependent on the adsorption kinetics and gas diffusion rates (both diffusion of
the gas phase species to the adsorbent particle and diffusion of the species within the pores of the
adsorbent), as well as competitive adsorption phenomenon between the various contaminant
species. Temperature transients and heat transfer to the walls of the vessel are also important
factors in the design of PSA beds. Large temperature excursions of greater than 20ºC are not
uncommon in hydrogen purification processes due to the heat of adsorption. Adsorption of all
the contaminants considered here is a exothermic process, therefore, the temperature in the
adsorption bed increases as the impurities adsorb (and the adsorbent capacity decreases with
increasing temperature) and decreases as the contaminants desorb (thus increasing the adsorbent
capacity and making it more difficult to clean the adsorbent surface).
The model developed by Jee et al. assumes that the gas phase behaves as an ideal gas, that there
are no radial concentration gradients, that the gas and the adsorbent are in local thermal
equilibrium, and that the mass transfer rate is represented by a linear driving force model.
Despite these simplifying assumptions, the model is quite complex. The temperature dependent
multicomponent adsorption equilibrium for the adsorbents are fit to the Langmuir-Freundlich
model,
q mi Bi Pi ni
qi = n
(Equation 5-2)
1+ ∑ BjP
nj
j
j =1
in which qi is the amount of component i adsorbed (mol/g adsorbent), Pi is the partial pressure of
component i, qm = k1 + k2T, B = k3exp(k4/T), and ni = k5 + k6/T; where T is the temperature and
k1 – k6 are experimentally determined parameters for each combination of contaminant i and
adsorbent (either activated carbon or zeolite).
39
J.-G. Jee, M.-B. Kim and C.-H. Lee, Ind. Eng. Chem. Res., 40, 868 (2001).
47
The end result of the model is a series of coupled, transient partial differential equations for mass
transfer and heat transfer within the adsorption bed that are given below.
∂ 2 P ∂P ∂u ∂P  ∂2  1  ∂  1  ∂  1   1 − ε  n ∂qi
− DL + + P + u + PT  − D   +   + u    + ρ RT  ∑ =0
∂z 2 ∂t ∂z ∂z ∂z 2  T  ∂t  T  ∂z  T   ε  i =1 ∂t
L p

(Equation 5-3)
in which DL is the diffusion constant, P is the system pressure, u is the gas velocity, ρp is the
adsorbent particle density, R is the gas constant, and ε is the void fraction of the adsorbent bed.
A similar mass balance must be calculated for each component.
The energy balance equation is,
∂ 2T  ∂u ∂T  ∂T ∂q i 2h
( )
n
− KL + ερ C  T + u  + ερ C + ρ C − ρ (− ∆H )∑ + (T − Tw ) = 0
∂z  ∂z ∂z  ∂t i =1 ∂t
2 p pg g pg b pg b
R
(Equation 5-4)
in which KL is axial thermal conductivity, ρg and ρb are the density of the gas and the bulk, Cpg is
the heat capacity of the gas, ∆H is the average heat of adsorption, h is the heat transfer
coefficient, and R is the diameter of the bed.
To fully model the Batta cycle would require solving this series of coupled partial differential
equations over 12 transient steps for all four beds and repeating the process until “steady state”
cycle conditions have been reached. To simplify the solution and yet still provide reasonably
accurate bed sizes we have calculated the break-though times for a multicomponent mixture
containing the appropriate concentrations of H2, CO, CO2, CH4 and N2 in a single bed. The
break-though time was then equated to the break through point in the Batta cycle (i.e. the
beginning of the second pressure equalization step). This process was repeated until reasonable
bed sizes and cycle times were reached.
Another important consideration in determining the size of the PSA beds and the cycle time is
adsorbent fluidization. High gas velocities through packed beds will tend to lift or “fluidize” the
particles in the bed. The fluidized particles will chaffe and break apart, eventually destroying the
bed and clogging the PSA valves. Baeyens and Geldart have developed an equation to predict the
minimum velocity (Umf) at which a bed begins to fluidize40
0.0009(ρ p − ρ g )
0.934
g 0.934 d
U mf = (Equation 5-5)
0.87 µ 0.006 ρ g
in which µ is the gas viscosity, ρg and ρp are the densities of the gas and the particles,
respectively, g is the gravitational constant, and d is the average particle diameter (all variables
40
Baeyens and Geldart, Fluidization and Its Applications, Proc. Int. Symp. Toulouse, 263 (1973).
48
should be entered in SI units). The maximum gas velocity in each PSA step was calculated to be
well below the minimum fluidization velocity.
Using the above sizing methodology, the vessel sizes for the SMA PSA system are calculated to
be 7 feet long and 12 inches in diameter with a cycle time of ~ 2 minutes. The ATR vessels are 9
feet long and 16 inches in diameter also with a cycle time of 2 minutes due to the higher gas flow
rates and higher percentages of impurities in the ATR reformate stream.
5.1.3 PSA System Cost
The cost of each PSA system is estimated based on the flow diagram of Figure 5-1 and vessel
dimensions. PSA vessels are 316 stainless steel cylinders containing the dessicant, activated
carbon, and zeolite adsorbent. The material and manufacturing cost for each component of the
vessel was calculated, and the assembly time and cost for each vessel was estimated.
Likewise, a system assembly time and cost was estimated based on the methods for placing and
welding each of the several hundred components (pipe lengths, valves, tees, elbows, and crosses)
which comprise the pipework for the system. Pipes are ¾”-diameter stainless steel, and all
connections are welded using an orbital welding machine.
Figure 5-3 outlines the PSA design parameters and system costs, which include the HFA
manufacturer’s markup. While the PSA vessels are purchased from a vendor, the system
assembly is performed by the HFA manufacturer, with appropriate markups applied for each.
For a more detailed cost breakdown, refer to the bills of materials for the 10 atm SMR and ATR
reformer systems in Section 4.2.6 and 4.3.7, respectively.
SMR PSA System ATR PSA System

Targets
System Pressure (atm) 10 10
H2 Recovery 75% 75%
H2 Product Purity 99.999% 99.999%
CO Concentration in Product < 10 ppm < 10 ppm
Results
Bed Height 7 feet 9 feet
Bed Diameter 12 inches 16 inches
Total Volume of Adsorbent* 623 liters 1423 liters
Cycle Time 120 sec 130 sec
Total PSA System Cost $18,075 $26,738
* The total volume of adsorbent includes both activated carbon and zeolite for all four beds. All adsorbents were assumed to cost ~ $2/lb based on
quotes from GraceDavidson and Atofina.
Figure 5-3. Parameters Used to Calculate Manufacturing Cost of the PSA systems
49
5.2 Metal Membranes
5.2.1 Metal Membrane Gas Separation
Hydrogen can be separated from other gases by passing the gas mixture over a heated metal
membrane such as palladium or palladium alloys at high pressure. The hydrogen molecule is
first adsorbed on to a palladium site, and then is dissociated into hydrogen atoms. The hydrogen
atoms then diffuse through the membrane at a speed determined by the hydrogen front-to-back
partial pressures and metal temperature, and eventually these atoms recombine to form hydrogen
molecules at the back surface of the metal membrane. As long as there are no leaks through or
around the membrane, the output hydrogen is 100% pure, since no other gas can pass through a
solid metal membrane.
Hydrogen Permeability
In theory the amount of hydrogen diffusing through a metal membrane is given by:

QH ( slpm) = K 
mols  A(m 2 )
×
 m − s − Pa  t (m)
× 22.4
liters
mol
× 60
sec
min ute
× Pfh0.5 − Pp0.5 [ ] (Equation 5-6)
where QH = hydrogen flow through the membrane (standard liters per minute),
K = the metal membrane permeability (mols/m-s-Pa0.5),
A = the membrane area (m2),
t = the membrane thickness (m),
Pfh = the input feed hydrogen partial pressure (Pascal), and
Pp = the output permeate gas pressure (Pascal).
The square root dependence on pressure is known as Sieverts law. While metal membranes
theoretically should permeate hydrogen according to the square root of the difference in partial
pressures, experimental results often lead to a higher exponent of pressure dependence:
QH =
1344 × KA
t
[
× Pfhn − Ppn ] (Equation 5-7)
where “n” is often closer to 0.7 than to 0.5, but can vary from 0.5 to 1.0. If the limiting rate is
due to hydrogen dissociation at the front surface of the membrane rather than diffusion through
the membrane, then the flow would be directly proportional to pressure. Deviations from the
square root dependence may also be due to some hydrogen diffusing through micro-cracks (or
around edge seals) in the thin metal membrane as molecular hydrogen, where one would expect
hydrogen flow to be proportional to the hydrogen partial pressure difference (n =1.0), not the
square root of partial pressure difference (n = 0.5). In addition, metals that form hydrides have
been noted to deviate from the square root dependence.41
41
Islam, M.T., S. Roy, B.B. Pruden and R. E. Buxbaum, “Palladium coated high-flux membranes for hydrogen separation at high temperatures
and different pressures,” submitted to Canadian Journal of Chemical Engineering, November 1998.
50
Group Vb Metals
The hydrogen permeability of several metals is shown in Figure 5-4 as a function of inverse
temperature.42 Palladium has relatively high permeability that increases with increasing
temperature. However, the Group Vb metals (niobium, vanadium and tantalum) have even
higher permeability, with the interesting characteristic of increasing permeability with lower
temperatures in the 352oC to 727oC range shown. Zirconium (not included in Figure 5-4) also
has very high hydrogen permeability, similar to that of niobium.43 The Group Vb metals are also
much less costly than palladium (see subsection on Group Vb metals in Section 5.2.3, below),
and palladium prices have been very volatile in the last few years as demand has increased for
auto catalytic converters and supplies from Russia have been erratic.
Given these disadvantages of palladium relative to vanadium or other Group Vb metals (lower
permeability by a factor of ten or more, 50 times lower bulk cost per unit volume, and high price
volatility), one would expect that the Group Vb metals would make much better metal membrane
candidates. But these metals do not have the ability to readily dissociate hydrogen, and most
form tenacious oxide layers on the outer surface. To overcome this deficiency, Markides et al.
first coated palladium on both sides of a niobium membrane by vapor deposition.44 The thin
palladium layer serves to protect the niobium surface from oxygen, and also serves to dissociate
the hydrogen molecules at the front surface and to recombine those atoms at the back surface.
Buxbaum and Marker later developed an electroless deposition technique for coating palladium
on Group Vb metals, allowing them to coat both sides of tubular membranes.45,46 They
demonstrated a palladium-coated zirconium membrane with 100 times higher permeability than
pure palladium of the same thickness over a temperature from 323 to 423oC.
42
Buxbaum, R.E. and A. B. Kinney, “Hydrogen Transport through Tubular Membranes of Palladium-Coated Tantalum ad Niobium,” Ind. Eng.
Chem. Res. 35, 530-537 (1996). See also the web page of Robert E. Buxbaum, REB Research & Consulting, 3259 Hilton Road, Ferndale,
Michigan, www.rebresearch.com.
43
J. N. Armor, “Membrane Catalysis: where is it now, what needs to be done?”, Catalysis Today, 25, p. 199-207, 1995.
44
US Patent 3,350,846 by Markides, A.C., M.A. Wright and D. N. Jewett, “Separation of Hydrogen by permeation,” 1967.
45
Buxbaum, R. E. and T. L. Marker, “Hydrogen Transport through Non-Porous Membranes of Palladium-Coated Tantalum and Niobium,” J.
Membrane. Sci. 85, 29-38 (1993); US Patent 5,108,724 by Buxbaum, R. E., “Composite Metal Membranes for Hydrogen Extraction,” 1991 and
US Patent 5,149,420 by Buxbaum, R. E. and P. C. Hsu, “Method for Plating Palladium,” 1992.
46
Buxbaum, Robert E., Method for Plating Palladium, US Patent#5,149,420, September 22, 1992.
51
Figure 5-4. Hydrogen Permeability of Selected Metals
Moss and Dye from Los Alamos National Laboratory have demonstrated a composite membrane
by depositing thin layers of palladium on both sides of Group Vb metals. They use an Argon ion
gun to clean off the surface oxide layer in the vacuum chamber immediately prior to depositing
the thin Pd out layers. The palladium performs the disassociation role, while the bulk of the
metal such as vanadium provides higher permeability per unit thickness than pure palladium.
While palladium-coated Group Vb membranes have much higher permeability and lower cost
than pure palladium or palladium-alloy membranes, their durability has not been proven.
Buxbaum and Kinney observed 15% decay in the permeation of their palladium-coated tantalum
membrane after continuous exposure at high temperature over 31 days. Edlund suggested that
this deterioration may be due in part to intermetallic diffusion.47 That is, the palladium diffuses
at high temperature into the tantalum metal, while the tantalum diffuses into the palladium outer
surface. The net result is a gradual reduction over a few tens of hours (depending on
temperature) of palladium sites at the outer surface to dissociate the hydrogen, and possibly a
buildup of an oxide layer as tantalum atoms migrate to the surface. Other experiments have
shown even faster decrease in permeability of palladium-coated Group Vb metals.
47
Edlund, D., “Metal Membranes for High Temperature Gas Separations,” Proceedings of the 1990 Membrane Conference, Business
Communications, Norwalk, Connecticut, p. 77, 1991.
52
The Group Vb metals also suffer from the formation of hydrides in the presence of hydrogen at
temperatures below the normal operating temperatures for metal membranes. Thus during
warm-up and cool-down of the membranes in the presence of hydrogen, the metals expand as
hydrides are formed, reducing mechanical strength. Spitzig et al. have demonstrated a sharp
decline in the ductility of V, Nb and Ta in the presence of hydrogen in the temperature ranges
from 80oK to 260oK.48 Ductility dropped from 95% reduction in area without hydrogen to the
15% to 20% range with the addition of just 0.27% to 0.76% hydrogen to the metals at these low
temperatures. There was negligible reduction in metal ductility at temperatures above 300oK in
these experiments.
Buxbaum and Marker observed that tantalum was less susceptible to hydrogen embrittlement
than niobium or vanadium. However, all three do form hydrides at temperatures in the range
between room temperature and the typical 350 to 450oC operating temperatures for these
membranes. The membranes can be heated and cooled in an inert atmosphere to minimize
hydriding. But, particularly during cool-down after exposure to hydrogen during operation, there
will be some hydrogen trapped inside the membrane. It may require very long cool-down times
to maintain the integrity of membranes, especially if they are made very thin to reduce costs.
Until these degradation mechanisms such as intermetallic diffusion and hydriding can be
avoided, composite membranes will not be suitable for industrial hydrogen purification systems.
Permeability of Pd Alloys
If composite membranes with Group Vb alloys are not sufficiently robust, then membrane
separation systems must rely on palladium or its alloys. However, pure palladium is not stable in
the presence of hydrogen,49 becoming brittle and rupturing with a few heating cycles. Palladium
has therefore been alloyed with other metals such as silver or copper to improve durability and
also to reduce susceptibility to poisoning. Alloying Pd with metals such as copper, silver and
yttrium can also increase hydrogen permeation as shown in Figure 5-5. For example, palladium
and copper alloys have a maximum permeability at 40% by weight copper to palladium; 23%
silver to palladium; and 10% yttrium to palladium at 350oC.50 Both copper and silver alloys
have been used in commercial hydrogen membranes. Chen et al. have demonstrated even greater
hydrogen permeability by adding gadolinium or yttrium to palladium.51 With pure hydrogen
feed, permeation of hydrogen with alloyed membranes can increase by a factor of four over pure
palladium. Adding yttrium can also strengthen the alloys compared to the commercial
membranes such as Pd23Ag.52
48
Spitzig, W.A., C.V. Owen, T.E. Scott, “The Effects of Interstitials and Hydrogen-Interstitial Interactions on Low Temperature Hardening and
Embrittlement of V, Nb and Ta,” Metallurgical Transactions A, Vol. 17A, July 1986.
49
E. J. Serfass and H. Silman, “Hydrogen Production and Purification by the Diffusion Process,” The Chemical Engineer, p. 266, October 1965.
50
J. Shu, B.P.A. Grandjean, A. Van Neste and S. Kaliaguine, “Catalytic Palladium-Based Membrane Reactors: A Review,” Canadian J. of
Chemical Engineering, Vol. 69, p. 1036, October 1991.
51
F. L. Chen, Y. Kinari, F. Sakamoto, Y. Nakayama and Y. Sakamoto, “Hydrogen Permeation through Palladium-Based Alloy Membranes in
Mixtures of 10% Methane and Ethylene in the Hydrogen,” Int. J. of Hydrogen Energy, Vol. 21, No. 7, p. 555-561, 1996
52
D. Fort and I. R. Harris, “The physical properties of some palladium alloy hydrogen diffusion membrane materials,” J. of the Less-Common
Metals, 41 (1975) 313-327.
53
Figure 5-5. Neat Hydrogen Permeation Through Palladium Alloys

as a Function of Alloy Concentration in Weight Percent50
Effects of Other Gases on Permeability
However, researchers at Nagasaki University have also shown that hydrogen permeation
decreases significantly with the addition of 10% nitrogen,53 10% methane, 10% ammonia or 10%
ethylene to the hydrogen feed. Some of their results are summarized in Figure 5-6 that plots
hydrogen permeation relative to pure hydrogen permeation in pure palladium. For pure
palladium (bottom set of three bars), adding 10% nitrogen or 10% methane to the hydrogen feed
does not affect permeation significantly. For the other alloys with much higher neat hydrogen
permeation, however, adding 10% diluent gases significantly degrades permeation. Thus the
gain in hydrogen permeability for Pd8Gd drops from a factor of 4.3 improvement for neat
hydrogen to only 1.8 with a 10% mixture of methane (90% hydrogen). The authors speculate
that the other gases are chemisorbed on the palladium surface, inhibiting hydrogen dissociation.
53
F. Sakamoto, Y. Kinari, F. L. Chen and Y. Sakamoto, “Hydrogen Permeation through Palladium Alloy Membranes in Mixture Gases of 10%
Nitrogen and Ammonia in the Hydrogen, Int. J. of Hydrogen Energy, Vol. 22, No. 4, pp. 369-375, 1997.
54
Pd 6.3Y 5Ag
Pd6.3Gd 5Ag
Pd 5.8Y 2.5Pb
Pd 8 Gd
Pd 8Y
Pd24Ag H2+10%CH4-450C
H2+10%N2-450C
Pd H2-450C
0 1 2 3 4 5
Ratio of H2 Permeation to Neat H2 & Pure Pd
DT I : P al l adi um-al l oy . X LS; Q53 - 10 / 19 / 2001
Figure 5-6. Hydrogen Permeation with Gas Mixtures,

all Relative to Pure Hydrogen with Pure Palladium
Since the reformate from fuel processors will typically contain at best 80% hydrogen from steam
methane reformers or 35% to 40% from autothermal reformers, one might expect even worse
performance. Booth at al. showed data for the permeation of pure hydrogen and a 68% hydrogen
mixture (methanol reformate) through a Pd25Ag membrane.54 The neat hydrogen experiment
used a 7.5-micron thick Pd25Ag membrane with a surface area of 140 cm2, with a permeation
rate of 23,000 sccm at 10 atmospheres and 350oC. The hydrogen permeation dropped to 1,300
sccm for the 68% hydrogen reformate stream for a membrane with 22 cm2 active area under the
same conditions. Based on a hydrogen partial pressure drop from 10 atmospheres to 6.8
atmospheres on the reformate side, one would expect hydrogen flow of
A2 × ( P20.5 − P00.5 )
Q= × 23,000 = 2,677 (Equation 5-8)
A1 × ( P10.5 − P00.5 )
The measured flow was only 1,300 sccm, or approximately 50% of the flow expected based on
neat hydrogen measurements.55 The authors speculate that some of this loss may be due to
blanketing of the membrane surface by the depleted hydrogen gas mixture. If so, then the
hydrogen permeation would be improved by better gas mixing in the reformate side of the
membrane assembly. However, other experiments with adequate turbulent gas mixing have also
shown drops in permeability on this same order of 50% with reformate mixtures. For the
purposes of cost estimation in this report, we therefore assume that the actual hydrogen
permeability with real reformate gases will be 50% of the measured permeability with pure
hydrogen gas to account for the various degradation mechanisms.
54
J.C.S. Booth, M. L. Doyle, S. M. Gee, J. Miller, L-A. Scholtz and P.S. Walker, “Advanced hydrogen separation via thin supported Pd
membranes, Proceedings of the 11th World Hydrogen Energy Conference, Vol. 1, Stuttgart Germany, pg. 867-878, 23-28 June, 1996
55
The authors claimed that the reduction in measured flow amounted to a drop of only 25%, but their own data do not seem to substantiate this
value.
55
5.2.2 Hydrogen Permeability in a Reformer System Membrane Assembly
So far we have only considered the permeability of a metal membrane to pure or nearly pure
hydrogen gases under laboratory conditions. In an actual reformer application, the hydrogen will
be diluted by other gases (including CH4, CO, H2O, and, for partial oxidation (POX) orATR
systems, N2), and the reformate gas stream will flow through a membrane assembly, typically
either a plate-frame assembly or through a set of hollow palladium or palladium alloy tubes. In
either case, the hydrogen content in the feed stream will decrease as the gas moves through the
membrane assembly. The permeation will also decrease along the length of the membrane
assembly, since permeation depends on the hydrogen partial pressure at any point along the
membrane and that pressure will fall as hydrogen is removed from the gas stream. In particular,
the hydrogen partial pressure will be very low at the exit from the membrane assembly. For any
practical membrane system, the hydrogen partial pressure will not be zero at the exit, meaning
that some hydrogen must leave the membrane system without passing through the membrane.
Not all the hydrogen can be recovered from the reformate stream. The hydrogen recovery rate
will generally depend on the economics of the system. High recovery will require very large
membranes, which will be very costly.
For steam reformers, this wasted hydrogen (along with CO and CH4) can be burned to raise
steam for the main reactor. For an ATR reactor, the heat from burning the membrane raffinate
stream is not needed except to preheat the membrane feed, and system efficiency would suffer if
the heat from this stream cannot be efficiently utilized.
To evaluate the relationships between hydrogen recovery, feed gas hydrogen content flow rate
and system pressures, consider the geometry for a flat membrane as shown in Figure 5-7.
Feed Retentate
Pf (kPa) Pr (kPa)
Membrane
QHf (l/min) QHr (l/min)
QOf (l/min) QOr (l/min)
Permeate
Pp (kPa)
QHp (l/min)
Figure 5-7. Geometry for a Flat Metal Membrane Permeating Hydrogen from a Mixed Gas Stream
Where Pf = the input feed gas total absolute pressure (kPa),

QHf = the hydrogen flow rate (liters/minute),
QOf = feed flow rate of all other gases (liters/minute),
Pr = output (retentate) total pressure (kPa):[= Pf without any pressure drop along
membrane feed to retentate plenum],
56
QHr = the hydrogen retentate (exit gas) flow rate (slpm),

QOr = the retentate flow of all other gases (slpm) –[ = QOf without any leaks],
Pp = the permeate pressure (presumably all hydrogen in kPa), and
QHp = the permeate hydrogen flow (slpm).
As the mixed gas flows along this membrane, hydrogen is extracted, and the driving force
(difference in the square root of the hydrogen partial pressures across the membrane) decreases.
The hydrogen passing through a given size of membrane can be determined by integrating along
the membrane. Alternately, the average hydrogen flow through the membrane can be estimated
very accurately using the log mean driving force equation.56
DF f − DFr
LMDF = (Equation 5-9)
 DF f 
ln  
 DFr 
where DFf = the input (feed gas) driving force and

DFr = the output retentate (exit) gas driving force.
0.5
 QHf × Pf 
( ) − (P ) ( )
0.5 0.5 0.5
DF f = PHf =   − Pp (Equation 5-10)
 QHf + QOf
p

where PHf = the input partial pressure of hydrogen at the leading edge of the membrane.
Similarly, the driving force at the exit edge of the membrane is given by:
0.5
 Q × P 
DFr = ( PHr ) ( ) ( )
0.5 0.5 0.5
− Pp =  Hr r
 − Pp
 (QHr + QOr ) 
(Equation 5-11)
The previous two equations can be rewritten in terms of the hydrogen recovery fraction, R =
QHp/QHf and the input hydrogen feed fraction F = QHf/(QHf+QOf):
( ) − (P )
0.5 0.5
DF f = F × Pf p (Equation 5-12)
and
0.5
 F × (1 − R) × Pf 
( )
0.5
DFr =   − Pp (Equation 5-13)
 (1 − RF ) 
56
J.C. Amphlett, L. M. Kearns, R. F. Mann, B. A. Peppley and J. P. Salvador, “The Simulation of a Membrane Reactor for Generating Hydrogen
from Methanol for a PEM Fuel Cell Power System,” Proc. Of the 12th World Hydrogen Energy Conference, Vol. 3, p. 1977, 21-26 June 1998.
57
Combining the above equations with Equation 5-2 yields the expected hydrogen flow from a
membrane assembly, where we have added a reformate derating factor, kR, to account for the
reduced hydrogen permeation in the presence of reformate gases.57
1344 × K × k R × A × ( LMDF )
QH = (Equation 5-14)
t
5.2.3 Membrane Material Cost
The price of most precious metals including palladium has been extremely volatile over the last
few years, primarily due to the increased demand of automotive catalytic converters combined
with instabilities in supply from the former Soviet Union. As automobile catalytic converters
switched from very high cost rhodium to more use of the relatively less expensive palladium,
prices rose dramatically. As shown in Figure 5-8, the price of palladium has historically been
half the price of platinum prior to 1998. Since 1998, however, palladium prices have escalated,
reaching more than $1,000/troy ounce in January 2001, before falling back to the $400/troy
ounce range in October 2001.
1200
1000
Metal Price (US$/troy ounce)
800
600
Platinum
400
Palladium
200
0
Jan-93 Jan-94 Jan-95 Jan-96 Jan-97 Jan-98 Jan-99 Jan-00 Jan-01
DTI: metalcost.XLS, Y22 12/27/2001
Figure 5-8. Historical Prices for Platinum and Palladium
57
Various experiments have shown that hydrogen permeability decreases significantly with the addition of virtually any gases, as shown in
Figure 5-6 above. Even a few percent of helium, an inert gas, can reduce hydrogen permeability below the level calculated by theoretical
hydrogen partial pressure calculations.
58
Pure Palladium Membranes
The cost of the membrane is given by
C M = C P × A × t = 658,333 × Cto × A × t (Equation 5-15)
where CP = the metal membrane cost ($/m3) and

Cto = the cost of the metal in $/troy ounce, the usual market price units.
For palladium, CP = $395 million/m3 at a palladium market price of Cto = $600/troy ounce from
the table above. The required area in turn can be determined from the hydrogen flow
requirements required. For a fuel cell that must deliver “P” kW, the hydrogen flow required is
given by:
6P
QH = (Equation 5-16)
η
where η = the fuel cell efficiency and

P = required fuel cell output power in kW.
The cost of the metal membrane per kW of fuel cell power (exclusive of the pressure vessel and
plumbing to hold the membrane, heater and a reformate compressor) is then given by:
CM 2,940 × Cto × t 2
= (Equation 5-17)
P η × K × k R × ( LMDF )
Thus the cost per kW of the membrane varies as the square of the membrane thickness –
reducing thickness both increases permeability and reduces material cost.58 An example of the
effects of hydrogen content in the input feed on cost is demonstrated in Figure 5-9 for a pure
palladium 25-micron thick membrane operating at 400oC for various hydrogen recovery ratios.
(The curves in Figure 5-9 are examples for comparison only; the costs for the specific reformer
designs are explained in detail in Section 5.2.5.) We have assumed here that the reformate gas
decreases the effective hydrogen permeability by 50% (kR=0.5). The two upper curves represent
the likely membrane material cost in $/kW of fuel cell output power for autothermal reformers
(ATR) with a 25% hydrogen content in the input feed gas to the membrane system. The upper
ATR assumes 20-atmosphere feed pressure, and the lower curve 10-atm pressure. In all cases
the permeate hydrogen stream from the membrane system is assumed to be at one atmosphere.
For an ATR system with 25% hydrogen in the reformate, at least 20 atmosphere pressure is
essential to reduce membrane material costs below $1,500/kW. Even then, recovery fractions of
more than 75% are not economically feasible.
58
In reality the membrane manufactured cost does not decrease linearly as the thickness is reduced below a certain minimum, since fabricating
very thin membranes has low yield. Although the bulk material cost may decrease, the manufacturing costs increase with reduction in thickness.
This effect becomes significant for membranes thinner than 15 to 20 microns, where pinholes begin to dominate the production process and
reduce effective yield, particularly for large area membranes.
59
3,000
Pd =$500/troy ounce;
Membrane Thickness =25 microns;
FC Efficiency=50%:
K=12.5nanomols/m-sPa^0.5;
Membrane Material Cost ($/kW) 2,500 SMR H2 Content=70%;
ATR H2 Content=25%;
Reformate Degradation Factor=0.5
2,000
1,500
ATR - 10 atm
ATR - 20 atm
1,000
SMR-10 atm
SMR - 20 atm
500
-
0.4 0.5 0.6 0.7 0.8 0.9 1
Hydrogen Recovery, R
DTI: DOE membrane costs.XLS, X32;12/27/2001
Figure 5-9. Estimated Material Costs of a 25-micron Palladium Membrane System to Cleanup the Reformer
Gas for a Stationary Fuel Cell System
The membrane costs for steam methane reformers with 70% hydrogen content in the reformate
are more favorable, and the recovery fraction can exceed 80% without excessive cost. However,
bulk palladium material cost is still in the $500/kW to $800/kW range depending on pressure.
Given that the capital cost of a complete gas turbine generation system is in the range of
$300/kW to $400/kW, a stationary fuel cell system could not afford to pay even $500/kW for the
metal content of the gas cleanup system.
In theory, decreasing the membrane thickness can reduce the material costs. Cutting the
thickness in half would reduce costs by a factor of four. However, thinner membranes are more
difficult to manufacture and are more susceptible to formation of pinholes. For example,
Goodfellow, one supplier of palladium membranes from the UK, cannot supply large area,
pinhole-free membranes in thicknesses below 10 to 15 microns. Handling, sealing and
supporting these thin membranes at high temperature and pressure also increases the complexity
and cost of the membrane system. However, if a manufacturer could develop methods for
handling and mounting 12-micron membranes, then the material costs could be cut substantially
as shown in Figure 5-10. Total metal membrane material costs could be reduced to less than
$120/kW with 90% recovery for a 20 atmosphere SMR, or to $250/kW for a 10 atmosphere
SMR. The autothermal reformers would still require membrane materials costing at least
$450/kW with 80% hydrogen recovery at 20 atmosphere’s pressure. Lower pressure operation
(10 atmospheres or less) with an ATR would still be unrealistic due both to low hydrogen
recovery and high membrane material cost.
60
800
Pd =$500/troy ounce;
Membrane Thickness =12 microns;
FC Efficiency=50%:
700
K=12.5nanomols/m-sPa^0.5;
Membrane Material Cost ($/kW) SMR H2 Content=70%;
ATR H2 Content=25%;
600 Reformate Degradation Factor=0.5
500
400
ATR - 10 atm
300 ATR - 20 atm
SMR-10 atm
200 SMR - 20 atm
100
-
0.4 0.5 0.6 0.7 0.8 0.9 1
Hydrogen Recovery, R
DTI: DOE membrane costs.XLS, X32;12/27/2001
Figure 5-10. Estimated Cost of a 12-micron Palladium Membrane

for a Stationary Fuel Cell Gas Cleanup System
Supported Metal Membranes
Another approach to reduced precious metal content is to coat a porous metal59 or porous
ceramic with a very thin layer of palladium or its alloys. Depositing palladium and its alloys on a
porous metal substrate such as stainless steel can produce an effective membrane initially, but
the metal ions can diffuse through the palladium outer layer at high temperature, eventually
reducing permeability. For example, Russian scientists deposited a 10-micron layer of palladium
and 6% ruthenium on a porous steel substrate.60 But the hydrogen permeability decreased by
50% after 1,000 hours at 800oC. The XPS spectra after the test showed that various steel
components had migrated to the palladium surface. They then succeeded in reducing this
intermetallic diffusion by placing a 0.08-micron layer of molybdenum between the porous steel
and the outer palladium alloy. The hydrogen permeability of this composite membrane
decreased by only 30% over 1,000 hours at 800oC. Subsequent thicker intermediate layers of 0.8
microns of tungsten, 0.1 microns of tantalum oxide, 0.5 microns of magnesia and 1 micron of
zirconia prevented any intermetallic diffusion over 1,000 hours, but the authors did not report the
actual hydrogen permeability, which was undoubtedly much lower with this relatively thick
intermediate layers.
59
J. Shu, B.P.A. Grandjean, E. Ghali and S. Kaliaguine, “Simultaneous deposition of Pd and Ag on porous stainless steel by electroless plating,”
J. of Membrane Science, 77 (1998) 181-195.
60
V. M. Gryaznov, O. S. Serebryannikova, Y. M. Serov, M. M. Ermilova, A. N. Karavanov, A. P. Mischenko and N. V. Orekhova, “Preparation
and catalysis over palladium composite membranes,” Applied Catalysis A: General, 96 (1993) 15-23.
61
Intermetallic diffusion can be avoided by coating palladium on porous ceramic substrates.

Morooka et al. have demonstrated reasonable performance by using CVD (chemical vapor
deposition) to coat a porous α-alumina tube with pure palladium.61 The average Pd thickness
was only 4.4 microns, but the Pd was deposited over a very rough surface that presumably had
many thinner sections. In fact this coated porous membrane did transmit nitrogen. The
hydrogen-to-nitrogen selectivity was 1,000 to one after four CVD layers were built up on the
tubular substrate. However, a selectivity of 1,000 to one for CO would be marginally acceptable
for a reformer gas cleanup system. With 4% CO in the reformate, then a 1,000 to one selectivity
would reduce the CO to 40 ppm, compared to the10 ppm CO goal for PEM fuel cells. They
were able to cycle this membrane between 100oC and 300oC in the presence of hydrogen several
times without any hydrogen embrittlement.
Collins and Way also coated 11.4 microns of palladium on a porous ceramic substrate.62 They
measured hydrogen selectivity to nitrogen of only 380 to one at 550oC, and only 250 to one
selectivity of hydrogen with respect to helium. However, they believe that much of this leakage
was due to poor sealing of the membrane to its fixture (using Grafoil seals), not leakage through
the membrane itself. The selectivity ratios for nitrogen and helium do support the notion that
these gases are leaking through small holes. Thus nitrogen molecules have a kinetic diameter of
3.64 Å while helium has a kinetic diameter of only 2.6 Å. On this basis one would expect
helium to leak through small pinholes at a rate 3.64/2.6 = 1.4 times more than nitrogen, vs. a
measured selectivity ratio of 380/250 = 1.5. But pinholes in the membrane would also produce
this ratio of helium to nitrogen leaks.
Researchers at Waseda University in Japan claimed that they achieved 100% hydrogen
selectivity with a 5.8-micron Pd23Ag coating on a porous alumina cylinder.63 However, another
group from Kyushu University later reported that the Waseda U. membranes still had 0.01%
nitrogen in the permeate, or one part in 10,000, far above what would be needed in a membrane
to remove CO.64 In addition, the Waseda membrane cylinder was rather complex, made of two
layers—an inner ceramic layer with 10-micron pore size, coated with a ceramic layer with only
200 nm pore size. They deposited layers of Pd and Ag, and heated the composite membrane to
900oC to form a Pd23Ag alloy on the porous glass surface. The authors did not present any
cycling data nor data of permeation with mixed gases, but did report that this membrane could
survive down to 200oC in the presence of hydrogen without embrittlement or leaking. Another
membrane with pure palladium developed pinholes at temperatures below 300oC as evidenced by
nitrogen leakage, presumably due to the formation of β-phase hydrides at these lower
temperatures. Adding silver reduced hydride formation at the lower temperatures. The Pd23Ag
membrane was tested in a 50/50 mixture of hydrogen and nitrogen without detecting any
nitrogen in the permeate. If this technology could be proven durable, it would offer raw Pd
material costs lower than those shown above for a 12-micron thick Pd membrane.
61
S. Morooka, S. Yan, S. Yokoyama, and K. Kusakabe, “Palladium membrane formed in macropores of support tube by chemical vapor
deposition with crossflow through a porous wall, Separation Science and Technology, 30 (14) pp. 2877-2889, 1995.
62
J. P. Collins and J. D. Way, “Preparation and characterization of a composite palladium-ceramic membrane,” Ind. Eng. Chem. Res. 1993, 32,
3006-3013.
63
S. Uemiya, T. Matsuda and E. Kikuchi, “Hydrogen permeable palladium-silver alloy membrane supported on porous ceramics,” J. of
Membrane Science, 56 (1991) 315-325.
64
S. Yan, H. Maeda, K. Kusakabe, and S. Morooka, “Thin palladium membrane formed in support pores by metal-organic chemical vapor
deposition method and application to hydrogen separation,” Ind. Eng. Chem. Res., Vol. 33, No. 3, 1994.
62
Johnson Matthey has also developed a thin 7.5-micron PdAg coating on ceramic tubes.65 While
these supported membranes are more durable than freestanding membranes and particularly
freestanding pure palladium membranes that frequently fail after one or two cycles of heating,
they still leak some CO and CO2. Leakage rates of 25 ppm CO and 100 ppm CO2 were
measured on one 7.5-micron membrane.66 Furthermore, the leakage rate increased gradually
with thermal cycling of only 10 cycles, suggesting that more microcracks were opened up with
each cycle.
We conclude that thin palladium alloys coated on porous substrates may provide an economic
membrane in the future. However, to date no organization has shown that these thin membranes
will provide low enough CO leakage over the necessary lifetime for a commercial steam
reformer system. We suspect that the palladium alloy thickness will probably have to be
increased to the 12-micron range to provide a durable membrane system. Hence we use 12
microns as the likely material thickness for membrane cost assessments below.
Palladium Alloy Costs
Adding certain alloys such as copper and silver to palladium increases durability and
permeability, while reducing membrane cost by diluting palladium with less expensive metals.
The two most common alloys are Pd40Cu (60% palladium and 40% copper by weight), and
Pd23Ag. Each of these alloys has higher permeability than pure palladium, at least with a nearly
pure hydrogen feed gas. The volume fraction of palladium in these alloys can be determined
approximately from:
Vp 0.6ρ c
= = 0.524 (Equation 5-18)
VT 0.6ρ c + 0.4 ρ p
for Pd40Cu, where Vp = volume of palladium,

VT = total membrane volume,
ρc = density of copper (8.96 g/cc), and
ρp = density of palladium (12.2 g/cc).
Since the cost of copper is negligible compared to palladium, the effective cost of Pd40Cu is
52% of the cost of Pd for the same thickness membrane. In addition, the hydrogen permeability
of Pd40Cu is very similar to that of pure palladium (see Figure 5-5).
Similarly for Pd23Ag, the volume fraction of palladium is:
Vp 0.77 ρ a
= = 0.742 (Equation 5-19)
VT 0.77 ρ a + 0.23ρ p
65
Ibid, et al..
66
The reformate included 24% CO2 and 4.7% CO. In addition the kinetic diameter of CO2 (3.3 Å) is smaller than the diameter of CO (3.76 Å), so
one would expect more leakage of CO2 than CO through microcracks.
63
Since silver is selling at 100 times less than palladium, the material cost of Pd23Ag would be
approximately 74% that of pure palladium. But Pd23Ag has higher hydrogen permeability than
pure palladium. As shown in Figure 5-5, Pd23Ag has approximately 50% higher permeability
with gas mixtures than pure palladium, so the Pd23Ag area can be even less than that indicated
by Equation 5-9. However, Pd23Ag membranes are reportedly susceptible to degradation in the
presence of CO and H2S, while Pd40Cu is supposedly less susceptible to these gases. We
therefore use 12 microns of Pd40Cu as the basis for membrane material costs in the system cost
estimates below, with a 48% reduction in membrane material volume cost estimates compared to
pure palladium.
Group Vb Alloy Costs
The Group Vb metals have another significant advantage over palladium and its alloys: low cost.
As shown in Figure 5-11, the cost of palladium per unit weight was 27 times more than the next
most costly hydrogen-permeable metals at its $600/troy ounce price early in 2001. Assuming the
markups to make the metals into foils and the markups for handling, scrap, etc., the final
estimated cost per unit volume ranges from $395/cc for palladium to only $7.27/cc for vanadium.
For a given thickness and area, a vanadium membrane would therefore cost 54 times less than an
equivalent palladium membrane. But with higher permeability, the vanadium membrane could
have significantly less surface area than a palladium membrane to achieve the same hydrogen
throughput. And, as shown in Figure 5-4, the vanadium membrane could also operate at lower
temperature than palladium. However, to date no organization has been able to develop
membranes based on Group Vb alloys that have sufficient durability for even a few thousand
hours operation under realistic reformate conditions.
B u lk P rice $ /gra m M a rku p M a rg in D ensity C o st p e r

to fo rm (g/cc) u n it vo l.
fo il ($ /cc)
Pd $ 6 00 /tro y 1 2 .22 1 .2 1 .4 1 2 .2 395
o z.
Ta $ 2 0 0 /lb 0 .4 4 2 1 .4 1 6 .6 2 0 .45
V $ 2 0 0 /lb 0 .4 4 2 1 .4 5 .9 7 .2 7
Nb $ 7 0 /lb 0 .1 5 2 1 .4 8 .4 3 .5 3
Zr $ 2 0 0 /lb 0 .4 4 3 1 .4 6 .4 7 .8 8
Figure 5-11. Approximate Market Prices for Various Hydrogen-Permeable Metals
5.2.4 Membrane Durability
Metal membranes can fail or degrade in performance due to temperature cycling, surface
oxidation, poisoning or chemisorption of other gas species on the membrane surface.
Temperature Cycling
For a practical gas cleanup system, the metal membrane must survive repeated cycles of heating
and cooling. Heating pure palladium from 25oC to 400oC in the presence of hydrogen is not
practical due to the varying degree of expansion of the alpha and beta phases in the
64
neighborhood of 100oC. Palladium membranes exposed to hydrogen undergo an α- to β-phase

transition at temperatures below 300oC.67 Formation of the β-phase hydride forms of palladium
is thought to cause microcracks and embrittlement of the membrane. The alpha- and beta-
hydride phases exist at low temperature, but only the alpha-phase is thought to exist at and above
300oC. Therefore the operating system for a pure palladium membrane should include some type
of heating and cooling protocol in the presence of an inert gas, adding further cost and
complexity to the system. And even purging with an inert gas during startup and shutdown may
not be sufficient to prevent embrittlement, since hydrogen will be present inside the palladium
membrane at the time shutdown begins. The amount of time for the hydrogen to diffuse out of
the membrane will depend on membrane thickness and the hydrogen partial pressures on both
sides of the membrane during the shut-down procedure.
To avoid or minimize the effects of hydrogen embrittlement, alloys of palladium have been
shown to be more durable. For example, Shu has reported in his summary article68 the following
results of thermal cycling tests by Rodina et al.69 They cycled various Pd alloys between 20oC
and 520oC in the presence of one atmosphere of hydrogen, and noted the number of cycles
“before the outset of visible deformation of the membrane.” The results were as follows:
Pure Pd: 20 to 30 cycles

Pd20Ag: 50 to 70 cycles
Pd8Au: 200 cycles
Pd24Ag3Au: 300 cycles
Pd20Ag5Au1.26Ru: 2,000 cycles
Pd5.5Ni 3,000 cycles
They did not report any cycle measurements for Pd40Cu.
While adding silver and other alloys does tend to stabilize palladium membranes, they still suffer
stresses from hydriding during heating and cooling in the presence of hydrogen. Booth et al. of
Johnson Matthey (UK) have measured the thermal expansion of Pd24Ag membranes in both an
inert gas and in hydrogen as a function of temperature.70 As shown in Figure 5-12, their 65
micron thick PdAg tube expanded linearly by 2% in the inert gas when heated to 440oC. They
then added hydrogen and cooled the membrane. The membrane expanded over 5% as it cooled
below 150oC, presumably as hydrogen formed hydrides in this region. The sample was then
reheated in the presence of hydrogen, and it expanded by 6.5% before approaching the non-
hydrided behavior at temperatures above 200oC. The dashed line shows the expansion with 70%
hydrogen reformate gas.
67
K. Aoki, Y. Ogata, K. Kusakabe, and S. Morooka, “Applicability of palladium membrane for the separation of protium and deuterium,” Int. J.
Hydrogen Energy, Vol. 23, No. 5, pp. 325-332 1998.
68
J. Shu, B.P.A. Grandjean, A. Van Neste, and S. Kaliaguine, “Catalytic Palladium-based Membrane Reactors: A Review,” The Canadian
Journal of Chemical Engineering, Vol. 69, p. 1036, October 1991.
69
A. A. Rodina, M. A. Gurevich, and N. I. Dornonicheva, “The Interaction of Hydrogen with Certain Palladium-Gold and Palladium-Silver-Gold
Alloys, Russ. J. Phys. Chem. Vol. 45, p. 621, 1971.
70
Ibid., Booth et al.
65
Figure 5-12. Thermal expansion of Pd24Ag tubes with and without hydrogen
Hydrogen embrittlement can also be retarded by coating pure palladium on porous substrates, the
same method referenced above to cut down the amount of palladium required. Aoki et al. from
Kyushu University in Japan reported cycling a 3-micron Pd membrane coated on a porous
substrate 35 times without any degradation.71 They speculate that the palladium is mechanically
constrained within the small pores, preventing any significant embrittlement.
Chemisorption
Hydrogen permeation through palladium or palladium alloy membranes is strongly affected by

the presence of other gases. This decrease in permeability could be due to gases adsorbed on the
Pd sites inhibiting the dissociation of the hydrogen molecules. Alternately, chemisorption of
contaminates on the catalyst surface could lead to the formation of palladium compounds. Chen
et al. speculate that at lower temperatures gases may adsorb on the palladium, blocking sites for
hydrogen dissociations, while at higher temperatures palladium compounds may be formed.72
Two gases present in the reformate stream, CO and CO2, both have slightly higher heats of
adsorption on Pd than hydrogen.73 Hence hydrogen should not be able to dislodge either CO or
CO2 from a Pd site.
Hydrogen permeation can also be decreased by the blanketing of the membrane surface by other
gases in the reformate stream. As the hydrogen is withdrawn, a thin boundary layer may build
up that contains mostly the other gases in the stream. The effects of blanketing can be
minimized by controlling the flow of the reformate gases over the membrane surface with a goal
of instigating turbulent flow in the region of the membrane surface.
71
Ibid., Aoki et al.
72
Ibid. Chen, pg 557.
73
Hydrogen has been measured at 20 to 22 kcal/mole, compared to 27 to 40 kcal/mole for CO and a calculated value of 49 kcal/mole for CO2
[Ref: Musket, pg. 180].
66
Surface Poisoning
Once an element has chemisorbed on a Pd surface, it can form compounds that more or less
permanently poison the surface. Palladium is thought to be poisoned by sulfur, arsenic, chlorine,
mercury vapor, bromine and zinc.74 These elements tend to form tenacious compounds with
palladium, either permanently degrading permeation of hydrogen or requiring very aggressive
actions such as heating above 800oC to remove the compounds from the surface. Exposure to
hydrocarbons has been reported to cause degradation in some cases, but not others. Methane,
which will be present in the reformate stream, has been shown to cause hydrogen permeation
reduction in the 400oC to 600oC range, but not in the 300oC to 400oC range (see Figure 5-6,
above, for example, which shows significant loss of hydrogen permeation at 450oC in the
presence of methane).
H2S at 50 ppm in hydrogen caused a drop in permeation of a 24-micron thick palladium

membrane at 350oC to 50% of the neat hydrogen level in five minutes and to 20% after one
hour.75 Thiophene attacks PdAg membranes even faster than H2S.
As a result of these chemisorption and poisoning effects, the actual hydrogen permeation
measured in the laboratory with real reformate has consistently been less than the theoretical
values projected from neat hydrogen measurements. The hydrogen permeation drops more than
would be calculated from the log mean driving force based on actual hydrogen partial pressures
throughout the membrane assembly. As a result, the actual membrane area should be at least
twice the area suggested by neat hydrogen measurements.
5.2.5 Membrane System Total Cost
While the metal membrane is the most expensive material cost of a total membrane gas cleanup
system, the fabrication costs can also be substantial. The metal membranes must be mounted
and sealed in a pressure vessel with provisions for the reformate gases to pass by the membrane
surfaces. On the opposite (low pressure) side of the membrane, the hydrogen permeate gas must
be collected. Total manufacturing cost will depend on the mechanical design of the system. The
two most common designs are flat membranes and tubular membranes.
Tubular Membrane System Cost Estimates
Several organizations have experimented with tubular membrane gas cleanup systems. These
tubular membranes can be fabricated from metal alloys [e.g., Buxbaum], or they can be
fabricated by coating porous ceramic or metal cylinders with palladium or its alloys [e.g.,
Morooka, Uemiya, Yan]. Both options could entail considerable manufacturing expense. The
metal tubes must be carefully formed and then sealed to header plates to allow the reformate
gases to pass through the tubes or to allow the collection of the hydrogen permeate gases. The
porous ceramic or metal tubes might be less expensive to fabricate, but they too must be sealed
into header plates, and have the added complexity of requiring very lengthy deposition processes
74
R. G. Musket, “Effects of contamination on the interaction of hydrogen gas with palladium: a review,” J. of the Less-Common Metals, 45
(1976) 173-183.
75
Ibid., pg. 178.
67
to coat the porous tubes with palladium. For example, Uemiya et al describe the following
processes to form their coating of a palladium alloy on a ceramic tube with 300 nm average pore
size:
1. Pre-treatment of the tube with stannous chloride and palladium chloride solutions to form
palladium nuclei on the outer surface of the tube. Process was repeated 10 times.
2. Electroless plating of Pd in a bath of tetraammine complex of palladium, EDTA,
ammonium hydroxide and hydrazine.
3. Electroless plating of copper or silver, forming a second layer. Total plating time was 17
to 25 hours.
4. Heat treating of the Pd-Cu or Pd-Ag layers for 12 hours at 500°C for 12 hours to form a
PdCu or PdAg alloy on the tube surface. Later they reported heat treating at 900oC for
12 hours to obtain higher hydrogen permeability, presumably because a better PdAg alloy
was formed.
Alternately, Yan et al. used a metal-organic chemical vapor deposition (MOCVD) process to
coat their porous ceramic tubes with 150 nm average pore size with palladium. They described
the following process. The ceramic tube was sealed inside a high temperature reactor with a
glass sealant paste. The inside of the tube and the reactor were connected to separate vacuum
pumps. The tube was then subjected to the following steps:
1. Dry the ceramic tubes for 20 minutes at 120oC.

2. Heat treat support tubes for one hour at 750oC.
3. With a palladium (II) acetate target inside the reactor, flow argon gas into the reactor side
at 100-200 Pa, with the inside of the tube maintained at 6-9 Pa. Slowly heat to500oC to
pyrolyze the palladium acetate. Continue the pyrolysis for 2 hours in flowing argon at
500oC.
Planar Membrane System Cost Estimates
Process Options
Two process options were considered in the design of membrane gas separators:
1. Cool the product gas from the shift reactor to condense and remove the residual water.
This option concentrates the hydrogen, providing an increased pressure driving force for
hydrogen permeation. A condenser, a water separation vessel, and a heater to raise the
gas temperature to 300ºC are required between the shift reactor and the membrane. A
burner to recover heat from the raffinate for the heatup to 300ºC is also required.
68
2. Send the gas from the shift reactor directly to the membrane. More membrane surface
area is required than for Option 1 because the gas stream is diluted with water and the
membrane is less active at 250ºC than at 300ºC. The benefit is reduced equipment cost
by eliminating the condenser, the water separation, and the membrane preheater.
However, the condenser and the water separation would still be required downstream of
the membrane before the raffinate could be burned. (The raffinate gas could not be
vented directly to the atmosphere because of the residual levels of methane and carbon
monoxide.)
Just the added palladium cost for Option 2 is estimated to be 50% higher ($13,000) than for
Option 1, far more than the cost of the burner and heat exchanger (~$5,000).
The pressure vessel must be heated to the 400oC range initially for the membranes to be
effective. This heating operation should be carried out with inert gas to prevent the formation of
β-phase hydrides at low temperature in the presence of hydrogen. Once the vessel and the
membranes have been heated above 300oC, then the reformate gas from the reactor can be
flowed into the membrane vessel. We assume that the initial heating of the membrane will take
place during startup and may require a supplemental heater (not included in the cost) until
sufficient heat is available from the process.
System Design and Assembly
Oremet-Wah Chang of Albany, Oregon offers a Pd40Cu planar membrane system for
commercial sale.76 Solid circular Pd40Cu membranes are attached and sealed on two sides of a
circular porous frit with a central hole. The hydrogen passes through the membranes on both
sides of the frit, and the hydrogen permeate travels radially toward the center of the disc and is
collected along a central plenum. Multiple discs are stacked together in a pressure vessel,
alternating with gas distribution baffles such that the reformate gas introduced at the top of the
vessel passes back and forth radially from one membrane disc to the next. Hydrogen is removed
with each pass of the reformate gas, such that the hydrogen content at the bottom of the disc
stack is very low. Although the reformate path is rather tortuous through this maze of membrane
discs, it can be approximated by the flow of gas along a single planar membrane with the same
effective membrane area. Single membrane discs are rated to 400ºC, but the maximum
temperature for multiple-disc stacks is listed as 300ºC on the Oremet-Wah Chang website. We
assume that the membrane operating temperature will be 300ºC in the proposed system. The
hydrogen flux across the membrane decreases by about 14% by dropping from 400ºC to 300ºC,
increasing the number of discs required for a given hydrogen production.
We base our cost estimates for a planar membrane system on this commercial system, assuming
that the membranes are 12 microns thick Pd40Cu and that a 300ºC operating temperature is
feasible. We assume that the Pd40Cu is available in a foil that is adhered to a ¼”-thick porous
stainless steel disc. The production cost of these membrane discs is difficult to estimate without
knowing the exact fabrication process. The thin Pd40Cu membranes must be carefully sealed on
both the inner and outer radii of the disc such that they can survive without leakage at
temperatures up to 300oC and pressures up to 20 atmospheres.
76
www.alleghenytechnologies.com/wahchang/pages/products/hmm/hmm.htm
69
From the picture on the Oremet-Wah Chang web page, their four-inch disc has an active area of
about 4.8 x10-3 m2 on each side, or 9.6 x10-3 m2 total. For the 20 atm SMR system with a
requirement of 2.52 m2 active membrane area, we would need 262 discs. Alternately, Oremet-
Wah Chang offers an eight-inch membrane. If we assume that the inner circle and outer rims of
the eight-inch disc (not shown on the web page) have the same differential radii as the four-inch
discs, then the eight-inch disc would have an active area of approximately 0.0538 m2 (both
sides). The 20-atm SMR would then require only 47 of the eight-inch discs, and the 20-atm
ATR would require 84 discs.
The discs must be assembled into a pressure vessel that can be heated to the 300oC operating
temperature. Fittings must be provided on the top and bottom to feed in the reformate gas and to
extract the retentate gas and the hydrogen permeate. High temperature seals are required to
prevent leakage from the reformate/retentate high-pressure region to the permeate low-pressure
zone. Presumably, costs will vary depending on the number of discs required.
To estimate the membrane unit assembly costs, we devised a manufacturing process that is
optimistic about eliminating leakage around seals and through pinholes in the 12 µm-thick PdCu
alloy and may not represent the actual commercial process. We used DFMA techniques to
develop an assembly procedure that relies on several common off-the-shelf parts that can be
purchased in bulk quantities. The proposed membrane design is illustrated in Figure 5-13 and
uses the following steps:
1. Fabrication of a nominal 8” diameter pressure cylinder of the required length with one
end uncapped to allow insertion of the components.
2. Insertion of a ½” diameter porous stainless steel tube in the center
3. Insertion of alternating layers of components: gasket, spacer, gasket, flow baffle, gasket,
spacer, gasket, and active membrane; repeat.
4. Welding of the remaining endcap.
Membrane Disc
Baffle
Spacer
Gasket
Hydrogen
Collection Tube
Figure 5-13. Proposed Membrane Assembly
70
The membrane unit costs presented here do not include some significant items that are not easily
approached by DFMA techniques. We have not included costs for alloying the palladium and
copper and for forming the alloy into a 12-µm foil. Both procedures could add significantly to
the cost of a real system.
Cost Summary
The assumptions and total cost summary for the membrane units are listed in Figure 5-14, and
the detailed bills of materials for each unit are listed in Figure 4-11 and Figure 4-16 with the rest
of the RS. The hydrogen recovery for the ATR membrane is reduced to 75% to account for the
higher fraction of impurities in the reformate stream compared to the SMR. Because the
manufacturing and assembly costs were not investigated in detail and because potentially
significant cost elements (such as membrane alloying) are not represented, the total membrane
system cost estimate should be considered a lower limit rather than a realistic approximation.
SMR ATR
System Pressure (atm) 20 20
H2 Recovery 80% 75%
H2 Content in Feed (mole basis) 73% 43%
2
Pd40Cu Membrane Active Area (m ) 2.52 4.50
2
Total Membrane Area required (m ) 4.02 7.15
Total Membrane System Cost $24,581 $37,362
(Note: Potentially significant cost elements
are not included in this estimate.)
Assumptions: Palladium cost: $500/troy ounce; Membrane thickness = 12 microns; Neat hydrogen permeability = 1.25 nanomols/m-s-Pa0.5;
Reformate permeation degradation factor = 50%; Permeate pressure = 1 atm; Fraction of active membrane area to total area = 0.628 (including
π/4).
Figure 5-14. Parameters Used to Calculate Manufacturing Cost of a Planar Membrane Gas Cleanup System
If the proposed membrane system performs as projected, using a membrane instead of a PSA
would increase the recovery of hydrogen by 5%. In the best case this increase in hydrogen
recovery would improve the thermal efficiency of the overall process by 5%. In reality some
additional natural gas would be needed to fire the reformer burner to replace “lost” raffinate,
reducing the net efficiency gain to less than 5%. The maximum cost differential between the
membrane and PSA that would return the cost of capital (10%) is $5,400. The cost of a
membrane cleanup system at 10 atm is estimated to cost more than the membrane at 20 atm, or a
minimum of $25,000. With a PSA estimated to cost approximately $18,000, the additional cost
for the membrane is not justified, and we did not include an analysis of membrane cost for the 10
atm systems for this reason.
In-Situ Metal Membrane Reactors
The previous sections assumed that the metal membrane is used as a separate gas cleanup system
placed downstream of the steam methane reformer. Several organizations have placed metal
71
membranes inside the reformer to extract pure hydrogen directly. This arrangement eliminates
the need for separate water gas shift reactors (usually one at high temperature and a second shift
reactor at low temperature) and either a preferential oxidizer or PSA for final removal of CO
down to less than 10 ppm. In addition, by pulling the hydrogen directly out of the reformate in
the reactor, the steam reforming reaction is driven to completion at much lower temperatures.
Researchers at Los Alamos National Laboratory have demonstrated such an in-situ palladium
membrane reactor operating successfully on methane, octane (gasoline surrogate), ethanol and
methanol.77 They operated their membrane reactor for 215 days, 24 hours per day on methane at
500oC, much lower than the typical 700 to 900oC required reformer temperature without the in-
situ membrane. They also successfully ran their reactor on both California reformulated gasoline
and gasoline from the local Shell station that had 100 ppm of sulfur.78 Instead of the 900 to
1,000oC range typically required for gasoline, they were able to operate at 550oC with their in-
situ membrane system.
Despite the better operating conditions and less complex equipment of this membrane reactor,
the single tube membrane of the Los Alamos system would be much too expensive for any
commercial system. Their membrane tube was made of PdAg with a wall thickness of 178
microns. The membrane material cost alone for an 80-kW fuel cell reformer system operating at
10 atm would be $3.2 million, or $40,000/kW.
Haldor-Topsøe, a major catalyst and fuel reformer developer from Denmark, analyzed and
compared the cost of an in-situ membrane reactor with a conventional SMR plus PSA.79 For
their economic analysis, they assumed that a large industrial SMR could be built with an integral
2-micron thick Pd membrane on a ceramic substrate, something that has not to our knowledge
been demonstrated in a stand-alone laboratory system let alone incorporated into a SMR
operating at temperatures above 650oC. Given that 4 to 5 microns of palladium on porous
ceramic tubes operating at 300 to 400oC have been too leaky in laboratory experiments to keep
CO concentrations below 10 ppm, we doubt that a 2-micron coated tube array could survive very
long at 650oC in a harsh SMR environment. However, even with this extremely optimistic
assumption, Haldor-Topsøe concluded that a membrane reactor could not compete economically
with a conventional SMR with a separate down-stream PSA cleanup system. The membrane
reactor itself would have been lower cost with a 2-micron Pd layer in-situ membrane system,
according to their estimates, but the operating costs to drive the membrane system shifted the
economics in favor of the conventional SMR plus PSA system. As a result of this analysis, we
considered only separate membrane gas cleanup systems in this report.
5.3 Preferential Oxidation (PrOx)
Preferential oxidation (PrOx) is a gas-cleanup technology that can be used when the fuel cell is
designed to operate with less than pure hydrogen. Carbon monoxide (CO), if present in the
hydrogen stream at levels above 10 ppm, is a poison to the precious metal electrodes in fuel
77
S. A. Birdsell, R. S. Willms and R. C. Dye, “Pure hydrogen production from octane, ethanol, methanol, and methane reforming using a
palladium membrane reactor,” Proceedings of the Thirty Second Intersociety Energy Conversion Engineering Conference, Honolulu, Hawaii,
July 27-August 1, 1997, pp. 1942-6.
78
S. A. Birdsell and R. S. Willms, “Pure hydrogen production from gasoline reforming using a palladium membrane reactor,” Presentation
Record of the 1998 Annual Meeting of the American Institute of Chemical Engineers, Miami, Florida, November 15-20, 1998.
79
K. Aasberg-Petersen, C. S. Nielsen and S. L. Jorgensen, “Membrane reforming for hydrogen, Catalysis Today 46 (1998) 193-201.
72
cells. In the PrOx process the CO is selectively converted to carbon dioxide (CO2) without
significant oxidation of the hydrogen.
A PrOx reactor for the proposed hydrogen production is a single bed of approximately 5 L. The
exotherm from the CO oxidation is small enough that cooling is not required on the reactor. A
small radiator after the reactor cools the hydrogen before compression.
PrOx gas cleanup is not comparable to the PSA and membrane systems examined above. The
maximum hydrogen purity attainable from the SMR with PrOx is 76% (dry basis) with
approximately 20% CO2, 1% water, and 3% methane remaining in the product. Impurity levels
in the ATR product are even higher (>50%). The impurities increase the sizes of the hydrogen
compressor and the storage tanks and reduce the fuel cell vehicle efficiency and range. Fuel cell
vehicles operating with 76% hydrogen instead of nearly pure hydrogen suffer a power reduction
of ~25% for a given fuel cell size and have a range reduced by ~25% for a given on-board
hydrogen storage volume. PrOx may have an advantage over the other cleanup technologies in
mobile (i.e., on-vehicle) reformers because of its compact size and relatively light weight. In
stationary applications, however, the weight of the cleanup equipment is not a design factor, and
the footprint of the equipment is only a small driver of the overall cost.
Since we are designing a system for pure (99.99%) hydrogen fuel, the PrOx reactor does not
replace PSA or membrane gas cleanup and will only add to total cost of the system without any
increase in hydrogen production.
5.4 Gas Cleanup Summary Comparison
5.4.1 Effect on Hydrogen Compressor
More details on hydrogen compressors are provided in Section 6 below, but a brief discussion of
the influence of hydrogen cleanup method on compressor costs is appropriate here.
The output pressure from the gas cleanup system (membrane or PSA) will affect the cost and
energy losses in the downstream hydrogen compressor for a hydrogen fueling station application.
This can have a small impact on the cost of a high-pressure hydrogen fueling system. The lower
efficiency of the hydrogen compressor from a membrane system will also add to the electricity
cost of the fueling station.
A metal membrane system requires a large pressure differential across the membrane. As shown
above, differential pressures of at least 10 atm are required for SMRs, and at least 20 atm for
ATR systems, assuming in both cases that the hydrogen is withdrawn from the membrane system
at atmospheric pressure. Similar feed pressures are required with PSA gas cleanup systems, but
with one major difference: the output pressure from a PSA system is typically very near the feed
pressure – hydrogen passes through the PSA absorption beds with very little pressure drop.
Therefore a PSA system with a 20 atm reformate pressure can deliver 16 to 18 atm pressure to
the hydrogen compressor. A PSA hydrogen compressor can be smaller and operate with higher
efficiency than a metal membrane system compressor that must start with atmospheric pressure
hydrogen. A 3- or 4-stage compressor with 65% of its theoretical isentropic efficiency would
73
require energy equivalent to approximately 7% of the lower heating value of hydrogen to

compress that hydrogen to 10,000 psi, assuming an 18 atm input pressure from a PSA system.
The required compressor energy would double to 14% from a 1 atm metal membrane gas
cleanup system. Similarly, the compressor size would have to double from 14.3 hp for a PSA
hydrogen compressor to 28.8 hp for a metal membrane hydrogen compressor, both sized to
compress 2,000 scfh, or enough hydrogen to support an 80-kW fuel cell.
5.4.2 Gas Cleanup Selection and Cost Summary
Based on the evaluation of gas cleanup technologies, it was decided that a metal membrane gas
cleanup was only feasible with a 20 atm reformer system. As a result, both the ATR and SMR
10 atm RS’s used a PSA cleanup, while the 20 atm RS’s used a membrane. The estimated cost
of the gas cleanup (including HFA manufacturer’s markup) for each RS is summarized in Figure
5-15. For a detailed cost breakdown for each cleanup system, refer to the RS bills of materials in
Sections 4.2.6 and 4.3.7.
Reforming System 10 atm SMR 20 atm SMR 10 atm ATR 20 atm ATR
Gas Cleanup Method PSA Metal membrane PSA Metal membrane
Gas Cleanup Cost $18,075 $24,581 $26,738 $37,362
Figure 5-15. Gas Cleanup Summary
6 HYDROGEN COMPRESSORS
Current natural gas vehicles typically store NG in compressed gas form at pressure from 20.7
MPa to 24.8 MPa ( 3,000 psia to 3,600 psia). However, due to the lower density of hydrogen
gas, future hydrogen fuel cell vehicles must store compressed hydrogen gas at even higher
pressures, 34.5 MPa (5,000 psia) or greater so that the storage system volume may more easily
be packaged in the vehicle. Thus the relatively low hydrogen exit pressure from the gas
purification unit (1 atm for membrane separators as described above and 10-20 atm for pressure
swing adsorption (PSA) units) must be increased to at least the onboard storage pressure. In
practice, an even higher pressurization capacity is desired to allow full pressurization of the
target tank and fast filling from stationary storage tanks.
Directed Technologies, Inc. has previously examined the mass production cost of service station
stationary hydrogen compressors produced in large quantities (10,000 units per year). The
results of that study, conducted for The National Renewable Energy Laboratory,80 appear in
“Integrated Analysis of Hydrogen Passenger Vehicle Transportation Pathways, Final Report
(Revision 2), March 1998”. While still valid, the 1998 report is based on a hydrogen compressor
design that is a lower flow rate (2kg/hr vs. 4.8kg/hr) and is produced in larger quantities
(10,000/yr vs. 250/yr) than is of interest here. Since the design philosophy and outline are
fundamentally the same between the earlier cost study and the current study, the hydrogen cost
study that appears below is an updated and expanded version of the 1998 report. Much of the
explanatory text has been duplicated, in some cases verbatim.
80
Conducted under NREL subcontract No. AXE-6-16685-01.
74
For this cost study, two compressors systems are analyzed: a low inlet pressure system to
correspond to the output of a membrane gas cleanup system and a high inlet pressure system to
correspond to output from a PSA system. Features of the systems are listed in Figure 6-1. Both
designs have similar compression ratios per stage and are run at the same rotary speed.
Low Inlet Pressure High Inlet Pressure

Compressor Compressor
GAS CLEANUP SYSTEM Membrane PSA
FEED FROM:
Inlet Pressure 1.3 atm (5psig) 7.8 atm (100psig)
Output Pressure 476 atm (7,000psia) 476 atm (7,000psia)
Number of Stages 9 stages in 10 cylinders 6 stages
Compression Ratio per stage 1.92 1.99
Total Compression Ratio 355 61
Design Peak H2 Flow Rate 4.8 kg/hr (115 kg/day) 4.8 kg/hr (115 kg/day)
Design Peak RPM 535 rpm 535 rpm
Figure 6-1. Hydrogen Compressor Attributes
6.1 Existing Industrial Hydrogen Compressors
Three types of hydrogen compressors are used within the chemical process industry:81
reciprocating (piston and diaphragm), turbo (centrifugal and axial), and rotary flow. Figure 6-2
graphs the general flow capacity and discharge pressure for each class of compressor. Of the
three types, piston compressors are most commonly used for high discharge pressures within the
chemical industry.
81
“Build Your Working Knowledge of Process Compressors,” E. H. Livingston, Chemical Engineering Progress, February 1993.
75
Livingston, E. H., Burton Corblin North America, Inc., “Build Your Working Knowledge of Process Compressors,” Chemical Engineering
Progress, February 1993. Reproduced with permission of the American Institute of Chemical Engineers. Copyright © 1993 AIChE. All rights
reserved.
Figure 6-2. Typical Chemical Process Industry Compressor Application Ranges
For chemical industry reciprocating compressors, compression ratios per stage are typically
limited to 3 or less. However, for hydrogen service, compression ratio per stage is limited to
about 2 since an American Petroleum Institute guideline82 limits discharge temperature to 135°C.
Hydrogen cylinders usually utilize Teflon piston rings to prevent leakage and are typically
lubricated, thereby requiring filtration and careful design to avoid contamination. To increase
seal life, industrial piston compressors typically utilize cross-head elements to reduce side forces
and run at relatively low rpm (400-600 rpm). Active venting of the distance83 pieces (see Figure
6-3) is required to prevent flammable gas mixtures. Hydrogen-water separation knock-out drums
are required at the suction of every stage to ensure liquid free operation.84 For refinery usage,
high reliability is essential. Consequently, multiple compressors are operated at de-rated
throughputs (i.e., four machines at 33% or three machines at 50%.85). These factors, when
coupled with the low manufacturing production rates of the machines, result in what is generally
considered the high cost of hydrogen compressors.
82
API 681 as reported in “The Cost of Hydrogen Compression,” B. Cox, Imperial Oil Resources Ltd, B. Pruden, Industrial Hydrogen Chair,
University of Calgary, Report 96-02, Industrial Hydrogen Chair Research Program, University of Calgary, September 1996.
83
Distance pieces are spacers between the compression cylinders and crankcase which provide a chamber for active venting of hydrogen that may
have escaped the piston chamber.
84
Cox and Pruden.
85
Cox and Pruden.
76
Livingston, E. H., Burton Corblin North America, Inc., “Build Your Working Knowledge of Process Compressors,” Chemical Engineering
Progress, February 1993. Reproduced with permission of the American Institute of Chemical Engineers. Copyright © 1993 AIChE. All rights
reserved.
Figure 6-3. Simplification of Typical Compressor Crankcase Elements
For reference, two industrial hydrogen compressors are briefly outlined. Figure 6-4 outlines the
basic characteristics of the Norwalk Century 150 hydrogen compressor. This 5-stage, lubricated
piston compressor achieves the approximately desired total compression ratio (408 vs. our
requirement of 355) but at a mass flow 5.7 times greater than desired (27.2 kg/hr vs. 4.8 kg/hr).
Price of the unit86 (not including drive motor) is set at approximately $200,000 for quantities of
1-2, $160,000 for quantities of 20, and about $120,000 in large production (50-100). Thus price
drops almost in half as production volume increases to 100 units.
Norwalk Century 150 Based

H2 Compression Unit
5 stage, Intercooled, belt driven
Initial Pressure: 15 psia
Discharge Pressure: 6,000 psia
Max. Horsepower: 125 hp
H2 Prod. Rate: 192 cfm (60 lbs/hr)
Overall Efficiency: 70%
Approx. Weight: 15,000 lbs
Approx. Dimensions:
length 12 ft
width 8 ft
height 7.5 ft
Norwalk Company, Inc., Norwalk Process Compressors Brochure, Bulletin N-206F 1090-2, permission to reproduce Century-150 Compressor
from Brochure granted by Norwalk Company, Inc., Mario Perrotta.
Figure 6-4. Norwalk 150 H2 Compressor
The second industrial compressor of note is a 4-stage, air-cooled, piston compressor from
Greenfield Compression Inc. (formerly Henderson International Technologies, Inc.). This
compressor can accommodate 4.8 kg/hr flow and compression from 1 MPa (150 psia) to 48.3
MPa (7,000 psia). Thus it is appropriately sized and operated at a moderate speed (for long life).
The unit is a fully functional system, containing a drive motor, knock-out drums, and interstage
coolers (air cooled). Its price in small quantities with basic instrumentation is approximately
$60,000 with an approximate price of $40,000-$45,000 in larger quantities (100+ units). This
unit thus is well positioned as a benchmark and the following DFMA study will determine
86
Private communication with Ken Bird, Norwalk Compressors, Norwalk CT, March 1992.
77
whether design changes and a higher annual production rate might realistically achieve a lower
cost hydrogen compressor.
6.2 Small-Scale Innovative Natural Gas and Hydrogen Compressors
Fuel Maker Corp. of Salt Lake City, working with The Electrolyser Corporation, Ltd. of
Toronto, Canada, has modified an existing small-scale natural gas compressor for hydrogen
operation. The baseline natural gas compressor has been marketed for a number of years as a
garage-based appliance for the refilling of natural gas vehicles. The four-stage piston unit
compresses 2.08 kg/hr of natural gas from line pressure (~1 atm) to 20.1 MPa (2,915 psia). The
unit is entirely air-cooled and incorporates a unique safety design: the electric drive motor is
contained within the low pressure natural gas blow down vessel. Because the motor is immersed
in a positively pressurized natural gas environment, there can be no question of the electric motor
sparking and causing a fire. Hydrogen testing is currently underway at The Electrolyser Corp.
Because of the much lower density of hydrogen than natural gas, the modified compressor is
projected to compress only 0.57 kg/h of hydrogen gas. Thus it is a factor of 8.4 lower mass flow
rate than desired.
Another innovative compressor comes from Hurricane Compressors, formerly the High Pressure
Division of Grimmer Schmidt, Inc. The CNG 90 natural gas filling station compressor has a
natural gas capacity of 104 kg/hr with a compression ratio of up to 254:1 (5 psig to 5,000 psig).
The heart of the compressor system is a 460 CID, Ford LSG-875 natural gas engine, operating at
2,100 rpm. This V-8 engine is modified by replacing the head on one side to allow sequential
natural gas compression with intercooling. The other head remains configured for internal
combustion engine operation. Thus the four ICE pistons drive the opposing four compression
pistons. The compressor requires no electrical supply and the engine block is water-cooled.
Overall the compressor is compact (267 cm long x 127 cm wide x 123 cm high) and is available
skid-mounted or trailer-mounted. While not designed for hydrogen operation, the CNG 90’s
hydrogen throughput would be approximately 13 kg/hr, about 2.7 times the target flow-rate.
Cost of the CNG-90, of which 30 to 40 are produced each year, is about $70,000 to $100,000
depending on the level of upgrades to the basic compressor.87 As currently designed, the CNG
90 is not appropriate as a hydrogen compressor as the design of the upper stages piston seals
would result in major hydrogen leakage. However, this could potentially be fixed with minor
design changes.
6.3 Service Station Hydrogen Compressor Design
The hydrogen compressor design selected for cost estimation deviates from industrial
compressors in several key attributes. First, the hydrogen mass flow is chosen to be 4.8 kg/hr
(115 kg/day) which is significantly lower than the 50-100 kg/hr throughputs of industrial
compressors and the 13 kg/hr of the Hurricane Compressor but higher than the 0.57 kg/hr
throughput of the FuelMaker garage-based compressor (as modified to operate on hydrogen).
When combined with a reformer/purification unit operating 24 hours per day, a 115 kg/day
compressor is adequate to do 80% refills of approximately 26 fuel cell vehicles per day and can
87
Personal Communication with Hurricane Compressors, February 2002.
78
support approximately 265 FCV’s operating at full capacity.88,89 Vehicle fleet size varies widely
in the United States, but a great many fleets would need only one hydrogen compressor of this
size. The average gasoline service station in the United States refuels 125 vehicles per day: thus
4 hydrogen compressors of this size would be required for a mature hydrogen market. Or more
likely, larger capacity compressors would be designed for this larger application. However, we
are concerned here primarily with the early introduction of fuel cell vehicles that will require
smaller fueling capacity. Thus a 115 kg/day (maximum) compressor (running at low to
moderate rpm for reliability) supporting 265 fuel cell vehicles is an appropriately sized
compressor.
Because the compression requirements differ substantially between the low and high inlet
pressure operating requirements (compression ratio of 355 vs. 61), two compressor systems are
used as the basis for the this cost report. Both assume compressor blocks based on automotive
internal combustion engines used in combination with a custom crank-shaft and head to allow
different throws and bore diameters for each stage. The low pressure inlet compressor is based
on a V-10 engine block90 where the two cylinders are used in parallel for the first stage to allow
greater compressor mass flow: 9 stages and 10 cylinders. The high pressure inlet compressor is
based on a V-6 engine:91 6 stages and 6 cylinders. Both are oil-lubricated, and water-cooled.
Hydrogen flows sequentially through each of the stages and requires only a 1.92:1 or 1.98:1
compression ratio in each stage. This low compression ratio lowers hydrogen discharge
temperature and extends piston ring life.92 Interstage cooling is accomplished through a gas-to-
liquid tube-in-shell heat exchanger. A liquid-to-air radiator rejects the coolant water heat load to
the atmosphere. The pistons are oil-lubricated and rely on Teflon piston rings and dual Teflon
piston riders for low leakage and long life. Given the low operating temperature of each piston,
the low operating speed (535 rpm), and the low side-forces experienced by the piston, piston ring
lifetime is estimated at 4,000 hrs, a reasonable lifetime for low rpm lubricated industrial
compressors using Teflon piston rings.93
A key feature of the baseline compressor system is cost reduction through parts commonality
with existing internal combustion engines (ICE’s). Specifically, adaptation of an existing V-6 or
V-10 engine into a compressor is proposed. This concept has already been successfully
demonstrated by the CNG 90 compressor from Hurricane Compressors. The cylinder block and
crankcase/oil pan are retained and combined with custom cylinder heads, valving, pistons, and a
crank shaft. Interstage cooling and electric drive motor activation of the crank-shaft are added.
Finite element analysis has not been conducted on the engine block or head but structural
88
Based on fuel cell vehicles averaging 80 mpge, 380 vehicle range, and 12,000 miles traveled each year. Vehicles are assumed to require 80%
fuel refills every 9 days i.e. 3.8kg H2 refills (4.76kg x 80%) every 9 days.
89
The number of fuel cell refueled/supported by the hydrogen compressor differ slightly from those presented for the HFA in Section 3 because
the average output from the HFA must be adjusted for daily and seasonal demand variations whereas the compressor is sized for its peak flow
rate.
90
The bore and stroke specifications for the baseline V-10 are based on a Chrysler Viper engine.
91
The bore and stroke specifications for the baseline inline-6 are based on a BMW 325i.
92
While the Hurricane natural gas compressor achieves adequate lifetime with a 4:1 compression ratio per stage, the lower compression ratio for
the postulated hydrogen compressor is deliberately selected to more closely match standard industrial practice.
93
Experts from Greenfield Compression Inc. estimate ring life of up to 5000 hours for the lower stages and typically 2000-3000 hours for the
high pressure stages. However, these estimates assume an 800 rpm operating speed and a per stage compression ratio of about 2.6. By reducing
rpm to 535 and lowering compression ratio to 2.0, we consider 4000 hour piston life for all stages a realistic estimate.
79
integrity94 is considered adequate due to the low temperature of operation and the low cylinder
volumes involved (and thus low loads).
Figure 6-5 outlines the entire compressor system. Typically piston packings and vents are used
to scavenge and dispose of any hydrogen gas that leaks from the compression chamber around
the pistons and towards the crankshaft cavity. To avoid these packings, vents, and distance
pieces, the baseline compressor borrows a design feature from the Fuel Maker natural gas
compressor: the entire drive motor is mounted within a hydrogen filled blow down vessel so that
the entire blow-down vessel, crankshaft, and piston rod are immersed in a hydrogen
environment. Thus leakage around the piston, while still an efficiency loss, is not a safety issue.
Additionally, the electric drive motor need not be hydrogen explosion proof rated (Class 1,
Group B), significantly reducing cost. Of course, the reduced cost of the drive motor is traded
off against the increased cost of the custom bell dome covering the electric motor. But this trade
has been conducted, and the selected configuration is least costly.
Figure 6-6 and Figure 6-7 represent the bills of materials for the entire low pressure inlet and
high pressure inlet compressor systems: each system element is defined along with its material
and manufacturing cost. While the desire was to create a bottoms-up cost assessment where each
element of each subsystem was examined in detail, time and funding constraints as well as
common sense led some subsystems to be costed by price quote from existing suppliers: there
appeared no need to exhaustively dissect certain subsystems (radiators, electric motors, pressure
transducers) when reliable mass-manufactured costs were already available. Thus the “easy”
components appear simply as marked up costs and more detail is given to the “difficult”
components where mass-manufactured cost is not readily available. To more fully explain the
costing basis for the compressor system, selected components are discussed below.
94
Hydrogen embrittlement may lead to some strength degradation. However, the overall concept feasibility has been proven by the CNG 90
compressor.
80
Top View
Mounting Rods
Coolant Line Radiator
Interstage
Cooler Fan
0.91 m
(36 ")
I-Beam Blow Down Compressor
Supports Vessel
Fan Motor
Control Box
1.9 m
Side View
Interstage Cooler
Blow Down Compressor 0.86 m

Vessel (34 ")
Motor Mounts
Figure 6-5. Notional Hydrogen Compressor
81
Unit Unit
Make(M) Approximate Finished Part Dimensions Part Material Unit Tooling Tooling Manufac- Unit Cost Manfact.'s Integrators
Component Usage Material or Inner Outer Inner Outer Outer Outer Thick- Characteristics Cost Material % Cost Life- turing before Markup Markup Total
Buy(B) Diameter DiameterLength Length Width Height ness Specific Mass Volume per Mass Cost Scrap $ Time Cost Markup Cost
cm cm cm cm cm cm cm Gravity kg Liters $/kg $ units/life $ $ % %
Components Common with V8 ICE: $900.00
Crank Shaft Case included above

Bolts (Block to Crank Case) included above
Crank Shaft Case Gasket included above
Head Gasket included above
Oil Case included above
Main Bearing Caps (rear, rear intermediate, front, front interm.) included above
Cap Bolts included above
Crankshaft Main Lower Thrust Bearing included above
Crankshaft Main Upper Thrust Bearing included above
Crankshaft Lower Bearings included above
Crankshaft Upper Bearings included above
Connecting Rod Bearings included above
Connecting Rod Nuts included above
Connecting Rod Bolts 1.79787 1.798661 1.2007 1.2 0.89359 included above
Crankshaft Rear Seal included above
Unique/Additional Compenents:
Cylinder Head (cylinders 1,3,5) 1 Cast Iron B 24.384 13.97 10.16 7.86 16.3218 2.076569 $0.66 $10.77 50% $20,000 500 $97.17 $113.33 35% 62% $247.86
Cylinder Head (stages 2,4,6) 1 Cast Iron B 24.384 13.97 10.16 7.86 16.3218 2.076569 $0.66 $10.77 50% $20,000 500 $97.17 $113.33 35% 62% $247.86
Cylinder Liners
Cylinder 1 1 Forged Steel B 6.5532 8.4074 15.113 7.86 2.58803 0.329265 $0.66 $1.71 50% $10,000 500 $24.64 $27.20 35% 62% $59.49
Cylinder 2,3 2 Forged Steel B 4.572 8.4074 15.113 7.86 4.64439 0.590889 $0.66 $3.07 50% $10,000 1000 $14.64 $19.24 35% 62% $84.15
Inlet Valve Assy

Reed Valve, Stage 1 1 B $82.89 0% 62% $134.28
Suction Valve, Stage 2 1 B $78.99 0% 62% $127.96
Outlet Valve Assy

Reed Valve, Stage 1 included with Inlet Reed Valve above
Discharge Valve, Stage 2 1 B $80.99 0% 62% $131.20
Piston Assemblies
Piston - Cylinder 1 1 Forged Steel B 6.5532 10.16 7.86 2.69348 0.342681 $2.21 $5.94 15% $50,000 500 $106.64 $113.47 35% 62% $248.16
Piston - Cylinder 2,3 2 Forged Steel B 4.572 7.3152 7.86 0.94395 0.120096 $2.21 $2.08 15% $50,000 1000 $56.64 $59.03 35% 62% $258.21
Piston Ring Sets 1a,1b 2 3 4 5 6 7 8 9

For Cylinders 1 1 B Area 4.0001 4.003 2.88829 2.08396 2.084 1.08 1.0849 0.56478 0.56758 $20.00 0% 62% $32.40
For Cylinders 2,3 2 B Stroke 3.7322 3.88 3.88 3.88 2.02 3.88 2.02 3.88 2 $20.00 0% 62% $64.80
For Cylinders 4,5 2 B $20.00 0% 62% $64.80
For Cylinders 6 1 B $20.00 0% 62% $32.40
Connecting Rods
For Cylinders 1,2,3 3 Forged Steel B 12.7 1.905 1.27 7.86 0.2415 0.030726 $2.21 $0.53 15% $40,000 1500 $33.31 $33.92 35% 62% $222.54
For Cylinders 4,5,6 3 Forged Steel B 12.7 1.905 1.27 7.86 0.2415 0.030726 $2.21 $0.53 15% $40,000 1500 $33.31 $33.92 35% 62% $222.54
Crank Assy
10-throw Crank Shaft 1 Forged Steel B 3.81 27.432 7.86 4.91643 0.6255 $2.21 $10.84 15% $125,000 500 $414.40 $426.87 35% 62% $933.56
Figure 6-6. Bill of Materials - High Inlet Pressure Hydrogen Compressor
82
Unit Unit
Electric Motor Assy

Electric Motor (15 kW) 1 B $750.00 0% 62% $1,215.00
Shaft Spider Coupling 1 $17.00 0% 62% $27.54
Explosion proof Line and glands 1 B $175.00 0% 62% $283.50
Gear Reduction Assy 1 Steel B $50.00 0% 62% $81.00
Blow Down Vessel

Mating Bell and Shroud 1 Cast Aluminum B 27.94 7.86 35 $1.54 $54.02 50% $30,000 500 $84.38 $165.41 50% 62% $401.96
Gasket (Bell to Cylinder Block) 1 Teflon $4.00 35% 62% $8.75
End Plate 1 Aluminum $3.00 35% 62% $6.56
Bolts (Mating Bell to Cylinder Block) 8 Aluminum $0.30 0% 62% $3.89
Bolts (Mating Bell to End-Plate) 8 $0.30 0% 62% $3.89
Electrical Line Fitting/Gland 1 $40.00 0% 62% $64.80
Engine Mount Bushings 4 B $5.00 0% 62% $32.40

Pads for Engine Mount 4 B $6.95 0% 62% $45.04
Mounting Frame
I-Beam 2 B 198.12 10.16 10.16 28.00 $0.66 $18.53 $18.53 35% 62% $81.03
I-Beam (cross-supports) 4 B 91.44 10.16 10.16 12.93 $0.66 $8.55 $8.55 35% 62% $74.80
Mounting Arms (interstage cooler) 4 M 2.54 15.24 7.8 0.60233 0.077222 $0.66 $0.40 $4.00 $4.40 35% 0% $23.75
Mounting Rods (radiator, controls) 4 M 2.54 91.44 7.8 3.614 0.463333 $0.66 $2.39 $4.00 $6.39 35% 0% $34.51
Interstage Cooler Assy

Cylindrical Water Jacket 1 M 27.94 146.667 0.08 2.7 2.75902 1.02186 $1.59 $4.38 $12.00 $16.38 35% 0% $22.11
Endplates 2 M 27.94 0.17971 $1.59 $0.29 52% $9.80 $10.23 35% 0% $27.63
Water Fittings 2 B $0.50 35% 62% $2.19
Hydrogen Cooling Assy
Hydrogen Tubes Stages 1-7 7 B 1.27 146.667 0.17 $14.82 0% 62% $168.07
Stage 8,9 2 B 0.95 146.667 0.23 $16.51 0% 62% $53.48
H2 Line Fittings 18 B $5.00 0% 62% $145.80
Inlet Lines
Stage 1-7 7 B 1.27 75 0.17 $7.58 0% 62% $85.95
Stages 8,9 2 B 0.95 75 0.23 $8.44 0% 62% $27.35
Outlet Lines
Stage 1-7 7 B 1.27 75 0.17 $7.58 0% 62% $85.95
Stages 8,9 2 B 0.95 75 0.23 $8.44 0% 62% $27.35
Radiator (Water To Air) Assy 1

Radiator 1 B $250.00 0% 62% $405.00
Inlet Line 1 B $2.00 35% 62% $4.37
Outlet Line 1 B $2.00 35% 62% $4.37
Water Pump 1 B $150.00 0% 62% $243.00
Haz. Env./Expl. Proof elect mtr- water 1 B $400.00 0% 62% $648.00
Cooling Fan Assy

Cooling Fan (2000 scfm) 1 B $196.00 0% 62% $317.52
Haz. Env./Expl. Proof electric motor 1 B $500.00 0% 62% $810.00
Explosion proof cable glands 1 B $35.00 0% 62% $56.70
Explosion proof electrical lines 1 B $300.00 0% 62% $486.00
Pressure Transducer 2 B $85.00 0% 62% $275.40
Electronic Controller 1 B $200.00 0% 62% $324.00
Motor Drives (electronic controllers)

For compressor motor 1 B $1,150.00 0% 62% $1,863.00
For water pump 1 B $600.00 0% 62% $972.00
For fan 1 B $600.00 0% 62% $972.00
Temperature sensors 7 B $25.00 0% 62% $283.50
Total Cost = Total = $15,868.18
Figure 6-6. Bill of Materials - High Inlet Pressure Hydrogen Compressor (Continued)
83
Unit Unit
Components Common with V8 ICE: $1,050.00
Crank Shaft Case included above

Bolts (Block to Crank Case) included above
Crank Shaft Case Gasket included above
Head Gasket included above
Oil Case included above
Main Bearing Caps (rear, rear intermediate, front, front interm.) included above
Cap Bolts included above
Crankshaft Main Lower Thrust Bearing included above
Crankshaft Main Upper Thrust Bearing included above
Crankshaft Lower Bearings included above
Crankshaft Upper Bearings included above
Connecting Rod Bearings included above
Connecting Rod Nuts included above
Connecting Rod Bolts included above
Crankshaft Rear Seal included above
Unique/Additional Compenents:
Cylinder Head (cylinders 1a,1b,3,5,7,9) 1 Cast Iron B 40.64 13.97 10.16 7.86 27.2031 3.460948 $0.66 $17.95 50% $25,000 500 $107.17 $134.10 35% 62% $293.29
Cylinder Head (stages 2,4,6,8) 1 Cast Iron B 40.64 13.97 10.16 7.86 27.2031 3.460948 $0.66 $17.95 50% $25,000 500 $107.17 $134.10 35% 62% $293.29
Cylinder Liners
Cylinder 1a,1b 2 Forged Steel B 10.16 10.414 17.4752 7.86 0.56375 0.071724 $0.66 $0.37 50% $10,000 1000 $14.64 $15.20 35% 62% $66.48
Inlet Valve Assy

Outlet Valve Assy

Piston Assemblies
Piston - Cylinder 1a,1b 2 Forged Steel B 10.16 10.16 7.86 6.47431 0.823704 $2.21 $14.28 15% $50,000 1000 $56.64 $73.06 35% 62% $319.55
Piston Ring Sets 1a,1b 2 3 4 5 6 7 8 9

For Cylinders 1a,1b 2 B Area 4.0001 4.003 2.88829 2.08396 2.084 1.08 1.0849 0.56478 0.56758 $20.00 0% 62% $64.80
For Cylinders 2 1 B Stroke 3.7322 3.88 3.88 3.88 2.02 3.88 2.02 3.88 2 $20.00 0% 62% $32.40
For Cylinders 3,4 2 B $20.00 0% 62% $64.80
For Cylinders 5,6 2 B $20.00 0% 62% $64.80
For Cylinders 7,8 2 B
Connecting Rods
For Cylinders 1a,1b,2,3,4,5 6 Forged Steel B 12.7 1.905 1.27 7.86 0.2415 0.030726 $2.21 $0.53 15% $40,000 3000 $19.97 $20.59 35% 62% $270.12
For Cylinders 6,7,8,9 4 Forged Steel B 12.7 1.905 1.27 7.86 0.2415 0.030726 $2.21 $0.53 15% $40,000 2000 $26.64 $27.25 35% 62% $238.40
Figure 6-7. Bill of Materials - Low Inlet Pressure Hydrogen Compressor
84
Unit Unit
Crank Assy
10-throw Crank Shaft 1 Forged Steel B 3.81 45.72 7.86 8.19405 1.0425 $2.21 $18.07 15% $150,000 500 $464.40 $485.18 35% 62% $1,061.08
Electric Motor Assy

Electric Motor (20 kW) 1 B $900.00 0% 62% $1,458.00
Shaft Spider Coupling 1 $17.00 0% 62% $27.54
Explosion proof Line and glands 1 B $175.00 0% 62% $283.50
$50.00 0% 62% $81.00

Gear Reduction Assy 1 Steel B
Blow Down Vessel

500 $84.38 $165.41 50% 62% $401.96
Mating Bell and Shroud 1 Cast Aluminum B 27.94 7.86 35 $1.54 $54.02 50% $30,000
$4.00 35% 62% $8.75
Gasket (Bell to Cylinder Block) 1 Teflon
$3.00 35% 62% $6.56
End Plate 1 Aluminum
$0.30 0% 62% $3.89
Bolts (Mating Bell to Cylinder Block) 8 Aluminum $0.30 0% 62% $3.89
Bolts (Mating Bell to End-Plate) 8 $40.00 0% 62% $64.80
Electrical Line Fitting/Gland 1
$5.00 0% 62% $32.40
Engine Mount Bushings 4 B $6.95 0% 62% $45.04
Pads for Engine Mount 4 B
Mounting Frame $18.53 35% 62% $81.03

I-Beam 2 B 198.12 10.16 10.16 28.00 $0.66 $18.53 $8.55 35% 62% $74.80
I-Beam (cross-supports) 4 B 91.44 10.16 10.16 12.93 $0.66 $8.55 $4.00 $4.40 35% 0% $23.75
Mounting Arms (interstage cooler) 4 M 2.54 15.24 7.8 0.60233 0.077222 $0.66 $0.40 $4.00 $6.39 35% 0% $34.51
Mounting Rods (radiator, controls) 4 M 2.54 91.44 7.8 3.614 0.463333 $0.66 $2.39
Interstage Cooler Assy $12.00 $17.97 35% 0% $24.26

Cylindrical Water Jacket 1 M 27.94 200 0.08 2.7 3.7623 1.393446 $1.59 $5.97 $9.80 $10.23 35% 0% $27.63
Endplates 2 M 27.94 0.17971 $1.59 $0.29 52% $0.50 35% 62% $2.19
Water Fittings 2 B
Hydrogen Cooling Assy $20.21 0% 62% $229.19
Hydrogen Tubes Stages 1-7 7 B 1.27 200 0.17 $22.51 0% 62% $72.92
Stage 8,9 2 B 0.95 200 0.23 $5.00 0% 62% $145.80
H2 Line Fittings 18 B
Inlet Lines $7.58 0% 62% $85.95
Stage 1-7 7 B 1.27 75 0.17 $8.44 0% 62% $27.35
Stages 8,9 2 B 0.95 75 0.23
Outlet Lines $7.58 0% 62% $85.95
Stage 1-7 7 B 1.27 75 0.17 $8.44 0% 62% $27.35
Stages 8,9 2 B 0.95 75 0.23
Radiator (Water To Air) Assy 1 $300.00 0% 62% $486.00

Radiator 1 B $2.00 35% 62% $4.37
$2.00 35% 62% $4.37
Inlet Line 1 B
$150.00 0% 62% $243.00
Outlet Line 1 B
$400.00 0% 62% $648.00
Water Pump 1 B
Haz. Env./Expl. Proof elect mtr- water 1 B
$196.00 0% 62% $317.52
Cooling Fan Assy
$500.00 0% 62% $810.00
Cooling Fan (2000 scfm) 1 B $35.00 0% 62% $56.70
Haz. Env./Expl. Proof electric motor 1 B $300.00 0% 62% $486.00
Explosion proof cable glands 1 B
Explosion proof electrical lines 1 B $85.00 0% 62% $275.40
Pressure Transducer 2 B $200.00 0% 62% $324.00
Electronic Controller 1 B
$1,400.00 0% 62% $2,268.00
Motor Drives (electronic controllers) $600.00 0% 62% $972.00
For compressor motor 1 B $600.00 0% 62% $972.00
For water pump 1 B
For fan 1 B $25.00 0% 62% $445.50
Temperature sensors 11 B Total Cost = $18,312.06
Figure 6-7. Bill of Materials- Low Inlet Pressure Hydrogen Compressor (continued)
85
Cylinder Block: To minimize cost of this otherwise expensive specialty casting, the compressor
block utilizes the engine block of an existing mass-produced internal combustion inline-6 and V-
10 engines (ICE). All stages of the compressors have bore strokes and bore diameters equal to or
less than those used in the ICE configuration. Consequently, cylinder bore is reduced to the
proper dimensions by insertion of appropriate wall thickness cylinder liners and correct stroke is
assured by design of a custom crankshaft (as discussed below). All engine block coolant and
lubrication passageways typical to an ICE are carried over to the compressor.
Costs of the cast iron block, crankshaft case, crankshaft bearings and miscellaneous components
(as listed in Figure 6-6 and Figure 6-7) are estimated at $850 and $1000 respectively. Complete
new, fully assembled V-8 engines, including heads and valves can be purchased in quantities of
one for as low as $1,900.95 Thus, $850-$1000 is deemed a reasonable, and perhaps conservative,
price for fewer components, no assembly, and orders of 250 engines/year. An additional $50
machining allowance is added for sealing surface preparation or minor bored hole changes.
Lubrication System: To reduce wear, sliding surfaces within the compressor are oil lubricated
using the lubrication features (oil sump, galleries, channels) already cast into the cylinder block
for lubrication of the internal combustion engine. The lubricant is assumed to be a low vapor
pressure fluorocarbon oil so that very little oil is contained within the hydrogen process gas. An
oil filter is placed just upstream of the outlet to trap residual oil particulates. Unlike other
applications where the hydrogen stream must be 5 nines (99.999%) pure, it is assumed the fuel
cell can operate on ppm - level inert-oil contaminated hydrogen with no degradation.
Cylinder Heads: The compressor cylinder heads are sand-casted iron and contain boltholes for
mounting to the block. Compared to cylinder heads on internal combustion engines, the
compressor heads are relatively simple. They need not contain ports for air injection or spark
plugs and do not have features for cam actuated valving. Rather, the heads simply hold the plate
type hydrogen valves (described below) and attachment ports for the intercooler lines. While it
may be possible for both heads of an engine block to be identical, it is assumed here that each
head is different, thereby incurring increased tooling cost and resulting is a conservative cost
estimate.
Cylinder head cost is based on assumed $25k tooling cost, 7 heads produced per hr, $10/min
total casting cost, $10 additional machining, 3 hr setup, 3 hr takedown, 500 head volume
lifetime, and batch size of 417 units. This results in an approximate head cost of $181/head
before integrator markup and compares reasonable to an estimated price of $100 each based on a
cost quote from American Precision Casting,96 a specialist in prototype and low production run
castings, for production runs of several thousand per year. Additionally, the cost quote is based
on aluminum heads rather than iron heads and thus should be conservative.
Valves: Virtually all-industrial process gas compressors use spring-loaded, gas actuated valves.97
Thus to limit deviation from proven design practice and to eliminate the complexity of cam
95
J. C. Whitney Catalog No. 609C, prices guaranteed through March 1998, LaSalle, IL.
96
Private communication with Robert Kowalczyk, American Precision Castings, Inc., Chicago, IL, December 1997. Mr. Kowalczyk’s quote was
$80-$100 each for a 5,000 unit run of 4”x4”x16” aluminum blocks.
97
A Practical Guide to Compressor Technology, Heinz P. Bloch, McGraw-Hill, New York, 1996, page 52.
86
actuated automotive valving, the compressor is designed using similar industrial valving.98
Cross sections of typical plate valves are shown in Figure 6-8 and Figure 6-9. Design work
conducted by Hoerbiger Corporation99 of America for this specific application resulted in a
recommendation of reed valves for stages 1-3 and lubricated single-disc type valves for the
remaining stages. Pressure drop across the valves varied from 1.1 psi (stage 1) to 8.7 psi (stage
6). Prices at a production quantity of 1000 varied from $25 to $50 per valve. However, we are
concerned here with 500 units lifevolumes100 where a price quote was not readily available.
Thus to approximate the valving cost at lower production rates, the Hoerbiger values were
increased by 50% based on the assumption that half the cost of the valves is due to tooling and
that a halving of lifetime volume doubles the per valve tooling cost. Overall, valving costs sum
to $1425 making them one of the most costly set of components.
Bloch, H. P., “A Practical Guide to Compressor Technology,” (1966), The McGraw-Hill Companies, permission to reproduce granted by The
McGraw-Hill Companies.
Figure 6-8. Plate Valve
Pistons: The pistons are forged steel. Cost is based on material usage, an assumed $50,000
tooling cost amortized over a two year total lifetime production of 500 compressors per hear,
production rate of 60 units/hour, $3/min. forging cost, $4/piston trimming, washing, and
machining cost, 3 hour setup, 3 hour takedown, and batch size of 417 units. Care was taken in
the piston design to maximize the number of stages with identical pistons so as to reduce tooling
cost.
98
Automotive style cam-actuated valving for hydrogen compressor potentially offers cost advantages and thus is an intriguing concept.
99
Hoerbiger Corporation of America, Inc. of Fort Lauderdale, Florida is one of the largest producers of compressor valves in the world.
100
Based on 250 units per year and a two-year design lifetime. Some cylinders use identical valve designs, thus some valves have higher life-
volumes than others.
87
Bloch, H. P., “A Practical Guide to Compressor Technology,” (1966), The McGraw-Hill Companies, permission to reproduce granted by The
McGraw-Hill Companies.
Figure 6-9. Cutaway of Damped Plate Valve
Crankshaft: The crankshaft is a custom forged and polished crankshaft. Previous DTI
compressor studies have suggested a dual crank/cam arrangement to reduce piston wear on the
upper stages and thereby increase maintenance intervals,101 but that approach is not implemented
here. Cost of the custom cam/crank-shaft is based on forging and heat treating with ground
bearing surfaces, $150,000 tooling cost, 20 minute per crankshaft production time at $5/min, 3
hour setup and takedown, 125 unit batch size, and $5/connecting surface machining/polished
surface.
Crank Case: The compressor utilizes the cast iron crank case which normally would be mated
to the IC engine block. Since the block/crankcase cavity will be filled with hydrogen (and oil),
hydrogen-proof sealing mechanisms are needed for each end of the crankshaft. At the free end
of the crankshaft, after any counter weights needed for proper balance, an end-bell is bolted to
the crankcase such that the end of the crankshaft is encased. The other end of the crankcase has
mating features to allow a hydrogen-tight seal to the blow down vessel.
Electric Motor and Reducing Gear: The cam shaft is driven by a commercial three phase,
drip-proof, cast iron, 15-30 kW industrial electric motor with an approximate sales price of $750-
101
Electrolyser Corp. has experienced substantially decreased piston seal lifetimes on the final stage (highest pressure) of their hydrogen-
modified Fuel Maker garage compressor.
88
$1400 in quantities of 250/year. As discussed previously, the electric drive motor need not be
hydrogen explosion proof (Class 1, Group B) because it is mounted in a non-flammable
environment, namely a slightly pressurized, pure hydrogen environment.102 The motor is either a
6-pole (1200rpm) or 8-pole (900rpm) motor directly linked to the compressor shaft without rpm
step-down. The motor is cooled by the suction hydrogen flow around and through the motor and
by heat conduction to the bell housing. A spider coupling is used to the link the motor and
crankshafts.
Blow-Down Vessel: A blow-down vessel encases the electric motor and mates to the
compressor block and crank case. The blow-down vessel is constructed in two pieces. A sand-
cast aluminum shroud provides a mating surface to the cylinder block/crank case and encases the
electric drive motor. Coolant galleries are cast in the structure to remove heat generated by the
motor. The shroud snugly encases the electric motor so that good thermal contact results, aided
by thermal grease such that heat is transferred from the motor to the blow-down vessel. A flat
aluminum endplate bolts to the blow-down vessel to complete the hydrogen seal.
Because the electric motor is large (33 cm diameter, 56 cm long), the aluminum blow-down
vessel shroud must also be large (33 cm diameter, 78 cm long to accommodate motor, gearbox,
and flexible coupling). The cost of the shroud after markup is estimated at $536 based on a
preliminary quote from American Precision Castings, Inc.103 and $4 machining cost (drilling and
tapping).
Interstage Cooler: Interstage cooling of the hydrogen gas is accomplished by a tube-in-shell

heat exchanger. For each stage, the compressed hydrogen is cooled from 178°C to 68°C after
traversing two passes in the 1.5 m long heat exchanger. Water is the cooling medium. The first
seven heat exchange stages use 304 seamless stainless steel 0.5 inch OD schedule 10 piping to
contain the hydrogen gas. DMV Stainless USA prices this piping at $3.08/ft in quantities of
approximately 23,000 lbs. While thinner walled piping for some of the stages is possible, use of
a standard pipe allows bulk purchases and possible cost reduction. A more thorough cost
examination than was conducted here would need to be performed to determine optimal sizing.
The last heat exchange stage uses 304 seamless stainless steel 3/8 inch OD schedule 40 pipe
which DMV Stainless USA prices at $3.43/ft in quantities less than 3,000 lbs.
Radiator: An industrial finned tube radiator is used to reject heat from the interstage cooling
water to the environment. The thermal load is 14 kW and its cost is estimated at $300, after
markup, including frame and piping.
Coolant Pump: A water coolant pump is required to circulate cooling water from the interstage
heat exchanger to and from the radiator. The cost of this electric motor driven, 20 gallon per
minute, low pressure head pump is estimated at $150 in low quantities.
Skid Mounting: The entire compressor system is mounted on a skid at the factory to facilitate
transport and reduce installation cost. The skid consists of multiple painted I-beam members
102
The slight hydrogen over-pressure within the blow-down vessel ensures that air cannot leak into the vessel to create a combustible gas
mixture.
103
Private communication with Robert Kowalczyk, American Precision Castings, Inc., Chicago, IL, December 1997. Mr. Kowalczyk’s quote
was for $300-$600 each for 78 cm x 33 cm ID, 0.5 inch wall thickness aluminum castings with coolant galleries in a 5,000 unit production run.
89
welded together to form a mounting base. Material cost for the beams is based on a steel cost of
$0.66/kg for a total of $74. Welding, assembly, and painting costs are discussed in Section 6.4.
Safety and Control System: The compressor is controlled primarily by an rpm signal sent from
the main reformer controller to the electric drive controller and then to the electric motor
attached to the crankshaft. The coolant pump is run at constant (full) speed at all times.
Temperature sensors (thermocouples) measure the compressor inlet hydrogen temperature and
each stage’s exit H2 temperature. Should there be a substantial leak or coolant system failure, at
least one of the temperature sensors will give warning and the compressor will be shut down.
6.4 Compressor System Assembly and Test
Both low and high inlet pressure compressor systems are assembled in virtually identical steps.
Therefore, for analysis simplification, no cost distinction is made between the two compressor
systems. Assembly cost of the compressor system is detailed in Figure 6-10 through Figure
6-12. As outlined in Figure 6-10, only a preliminary assembly cost is estimated for the
compressor unit itself based on the time it takes a trained mechanic to assemble a V-8 internal
combustion engine. Thus assuming 2 hours assembly time for the compressor (including
cylinder block, crank shaft, bearings, crank case, connecting rods, cam followers, pistons, heads,
etc.) and an additional 1.5 hours for mounting the blowdown vessel components (blowdown
vessel, electric motor, gear box, drive coupling, gasket), a $1/min. labor rate, and a $0.5/min.
equipment rate yields an assembly cost of $502 after markup.
Compressor assembly time $1.0/min Labor Rate

$0.5/min Equipment Rate
$1.5/min Total Station Rate Station
Cost
# of Operation Markup After
Operations Time Total time (min) Factor Markup
1 210 min assemble compressor/l 210 min
blowdown vessel
Station Cost = $310.00 1.62 $502.20
Total Assembly Cost (marked up) = $502.20

Figure 6-10. Compressor and Blowdown Vessel Assembly
Markup rate for the manufacture and assembly operations is estimated at 62% for the compressor
manufacturer. This high markup rate is consistent with production of only 250 units per year
with each unit selling for only $20,000. The reformer manufacturer is able to achieve a lower
markup rate on the same unit sales because each HFA unit sells for over $250,000. Additionally,
the assumption is made that the compressor company is only producing a single product (i.e. just
the hydrogen compressor). If the compressor manufacturer sold many products, as would be the
case if an established compressor company added this particular hydrogen compressor to its line
of products, then the markup rate might be significantly reduced.
90
Manufacture and assembly of the interstage cooler occurs in six major steps as detailed in Figure
6-11. These steps are: cutting and bending the tubes, cutting heads from sheet stock, machining
top head, weld bottom to cylindrical tube, braze tubes to top head, paint top head, attach fittings,
pressure test completed interstage cooler. The assembly is assumed to occur at an outside
vendor’s facility. The cost of these steps is approximately $101, after markup.
Station Markup Station Station Markup Station
Cost Factor Cost Cost Factor Cost
Before After Before After
Markup Markup Markup Markup
Station 1 Station 6
Cut pipes to length 8 pipes at $0.30 each = $2.40 1.62 $3.89 Braze pipes to header
$0.5/min Labor rate
Station 2 $1.5/min Equipment rate
Bend pipes $1.0/min Labor rate $2.0/min Total Station Rate
$1.0/min Equipment rate
$2.0/min Total Station Rate # of Operation
# of Operation Operations Time Total time (s)
Operations Time Total time (s) 1 4.0 sec acquire header 4.0 sec
8 2.0 sec pick up pipes 16.0 sec 1 2.5 sec load header 2.5 sec
8 2.5 sec load pipes in fixture 20.0 sec 1 1.0 sec close header fixture 1.0 sec
8 1.0 sec close fixture 8.0 sec 8 2.0 sec acquire tube 16.0 sec
8 5.0 sec bend 40.0 sec 8 4.0 sec apply braze 32.0 sec
8 1.0 sec open fixture 8.0 sec 8 7.5 sec load tube 60.0 sec
8 2.0 sec remove pipe 16.0 sec 1 7.5 sec load in furnace 7.5 sec
8 3.0 sec transfer pipe to next statio 24.0 sec 1 2.0 sec close furnace 2.0 sec
132.0 sec 1 240.0 sec heat in furnace 240.0 sec
1 2.0 sec open furnace 2.0 sec
Station Cost = $4.40 1.62 $7.13 1 7.5 sec remove workpiece 7.5 sec
Station 3 16 2.0 sec disengage fixture 32.0 sec
CNC laser cut heads from sheet stock 1 7.5 sec remove workpiece 7.5 sec
$0.5/min Labor rate 414.0 sec
$1.5/min Total Station Rate Station Cost = $13.80 1.62 $22.36
# of Operation Station 7
Operations Time Total time (s) Paint header
2 4.0 sec acquire sheet 8.0 sec $0.5/min Labor rate
2 2.5 sec load sheet 5.0 sec $0.0/min Equipment rate
2 1.0 sec close fixture 2.0 sec $0.5/min Total Station Rate
2 106.0 sec laser cut 212.1 sec
2 1.0 sec open fixture 2.0 sec # of Operation
2 4.0 sec remove head 8.0 sec Operations Time Total time (s)
2 4.0 sec remove border 8.0 sec 1 240.0 sec paint header 240.0 sec
245.1 sec 240.0 sec
Station Cost = $6.13 1.62 $9.92 Paint consumables $2.00

Station 4
Weld head to cylindrical shell Station Cost = $4.00 1.62 $6.48
$1.0/min Labor rate Station 8
$0.5/min Equipment rate Attach H2 line and coolant fittings
$1.5/min Total Station Rate $1.0/min Labor rate
# of Operation $0.0/min Equipment rate
Operations Time Total time (s) $1.0/min Total Station Rate
1 4.0 sec acquire shell 4.0 sec # of Operation
1 4.0 sec acquire head 4.0 sec Operations Time Total time (s)
1 1.0 sec close fixture 1.0 sec 18 1.8 sec acquire fitting 32.4 sec
1 106.0 sec weld 106.0 sec 18 6.0 sec screw in 108.0 sec
1 1.0 sec open fixture 1.0 sec 18 3.0 sec recheck fittings 54.0 sec
1 4.0 sec remove workpiece 4.0 sec 194.4 sec
120.0 sec
Station Cost = $3.24 1.62 $5.25
$0.02/inch 28.3 inches welding consumables $0.57 Station 9
Pressure test
Station Cost = $3.57 1.62 $5.78 $1.0/min Labor rate
Station 5 $1.0/min Equipment rate
Machine header $2.0/min Total Station Rate
# of Operation # of Operation
Operations Time Total time (s) Operations Time Total time (s)
8 - clamp - 18 2.0 sec acquire lines 35.1 sec
8 - centerdrill - 18 2.0 sec attach lines 36.0 sec
8 - drill - 1 180.0 sec pressure test 180.0 sec
8 - chamfer - 18 2.0 sec detach lines 36.0 sec
8 - drill - 287.1 sec
8 - unclamp -
8 - flip - Station Cost = $9.57 1.62 $15.50
8 - clamp -
8 - centerdrill -
8 - chamfer -
8 - unclamp -
Station cost based on DFMA software analysis= $15.28 1.62 $24.75 Total Assembly Cost (marked-up) = $101.06
Figure 6-11. Interstage Cooler Manufacture and Assembly
The cylinder block and blow down vessel are attached to the skid through four rubber bushings
on mounting pads to reduce vibration. Each mount is estimated at $17.64, after markup,
including cost of the mounting pads.104 Other system components (interstage cooler, radiator,
fan, controls) are bolted to mounting bars that are in turn welded to the skids.
104
Engine mount and pad cost based on list price in J. C. Whitney catalog.
91
Figure 6-12 details the cost of assembling the skid and mounting and wiring components to it.
All work is assumed done by the compressor manufacturer and thus is marked up by 62%. Total
assembly cost for these steps is $111, after markup.
Station Markup Station Station Markup Station

Cost Factor Cost Cost Factor Cost
Before After Before After
Labor rate Markup Markup Mount radiator assembly $1.0/min Labor rate Markup Markup
Equipment rate $0.3/min Equipment rate
Total Station Rate $1.3/min Total Station Rate
# of Operation
Total time (s) Operations Time Total time (s)
I-beams 18.0 sec 1 5.0 sec acquire radiator assy 5.0 sec
ams 18.0 sec 1 7.5 sec position assy 7.5 sec
s-beams 36.0 sec 1 2.0 sec acquire screwdriver 2.0 sec
ss-beams 36.0 sec 8 5.0 sec tighten screws 40.0 sec
welder 2.0 sec 1 2.0 sec put down screwdriver 2.0 sec
s to cross-bea 180.0 sec 56.4 sec
G welder 2.0 sec
291.9 sec $0.02/inch 0.0 inches welding consumables $0.00
sumables $0.96 Station Cost = $1.18 1.62 $1.90

Mount fan assembly $1.0/min Labor rate
Station Cost = $8.26 1.62 $13.38 $0.3/min Equipment rate
$1.3/min Total Station Rate
Labor rate # of Operation
Equipment rate Operations Time Total time (s)
Total Station Rate 1 5.0 sec acquire fan assy 5.0 sec
1 7.5 sec position assy 7.5 sec
1 2.0 sec acquire screwdriver 2.0 sec
Total time (s) 8 5.0 sec tighten screws 40.0 sec
ooler supp. rod 7.8 sec 1 2.0 sec put down screwdriver 2.0 sec
s 36.0 sec 56.4 sec
welder 2.0 sec
45.0 sec Station Cost = $1.18 1.62 $1.90
G welder 2.0 sec Mount control box
92.7 sec $1.0/min Labor rate
sumables $0.24 $1.0/min Total Station Rate
# of Operation
Station Cost = $2.56 1.62 $4.14 Operations Time Total time (s)
1 5.0 sec acquire control box 5.0 sec
Labor rate 1 7.5 sec position assy 7.5 sec
Equipment rate 1 2.0 sec acquire screwdriver 2.0 sec
Total Station Rate 8 5.0 sec tighten screws 40.0 sec
1 2.0 sec put down screwdriver 2.0 sec
56.4 sec
Total time (s)
ator supp. rods 11.7 sec Station Cost = $0.94 1.62 $1.52
s 54.0 sec Electrically wire system
welder 2.0 sec $1.0/min Labor rate
45.0 sec $0.0/min Equipment rate
G welder 2.0 sec $1.0/min Total Station Rate
114.6 sec # of Operation
Operations Time Total time (s)
sumables $0.24 1 300.0 sec wire system 300.0 sec
300.0 sec
Station Cost = $3.11 1.62 $5.03
Station Cost = $5.00 1.62 $8.10
Labor rate Interstage cooler to compressor pipes
Equipment rate
Total Station Rate Cut pipes to length 16 pipes at $0.30 each = $4.80 1.62 $7.78
Bend pipes $1.0/min Labor rate

Total time (s) $1.0/min Equipment rate 16 pipes with 2 bends each
osition HX 120.0 sec $2.0/min Total Station Rate
7.8 sec
7.8 sec # of Operation
7.8 sec Operations Time Total time (s)
7.8 sec 32 2.0 sec pick up pipes 64.0 sec
et 7.8 sec 32 2.5 sec load pipes in fixture 80.0 sec
nch 7.8 sec 32 1.0 sec close fixture 32.0 sec
20.0 sec 32 5.0 sec bend 160.0 sec
tchet 2.0 sec 32 1.0 sec open fixture 32.0 sec
ench 2.0 sec 32 2.0 sec remove pipe 64.0 sec
190.7 sec 32 3.0 sec transfer pipe to next statio 96.0 sec
528.0 sec
Station Cost = $3.97 1.62 $6.44 Station Cost = $17.60 1.62 $28.51
Weld pipes to interstage cooler & compressor
Mount compressor/blow-down vessel $1.0/min Labor rate
$1.0/min Labor rate $0.5/min Equipment rate
$0.3/min Equipment rate $1.5/min Total Station Rate
$1.3/min Total Station Rate # of Operation
Operations Time Total time (s)
# of Operation 16 3.0 sec acquire pipe 48.0 sec
Operations Time Total time (s) 16 5.5 sec position pipe 88.0 sec
1 180.0 sec acquire & position compr/b 180.0 sec 1 2.0 sec acquire MIG welder 2.0 sec
4 2.0 sec acquire bolt 7.8 sec 32 7.5 sec weld 240.0 sec
4 2.0 sec position bolt 7.8 sec 32 5.0 sec reposition pipe & welder 160.0 sec
4 2.0 sec acquire nut 7.8 sec 1 2.0 sec put down MIG welder 2.0 sec
4 2.0 sec position nut 7.8 sec 539.9 sec
4 2.0 sec acquire rachet 7.8 sec
4 2.0 sec acquire wrench 7.8 sec $0.02/inch 64.0 inches welding consumables $1.28
4 5.0 sec tighten nut 20.0 sec
1 2.0 sec put down ratchet 2.0 sec Station Cost = $14.78 1.62 $23.94
1 2.0 sec put down wrench 2.0 sec
250.7 sec
Station Cost = $5.22 1.62 $8.46 Total Assembly Cost (marked-up) = $111.11
Figure 6-12. Skid Assembly and Mounting
Figure 6-13 outlines the testing cost of the entire system. One test engineer working for 30
minutes is dedicated to this task although in reality the system checkout might be several hours
92
with the engineer simultaneously monitoring multiple systems. Hydrogen gas would be
circulated through a low pressure reservoir thus incurring only negligible feedstock costs.
Electricity to operate the system is also negligible (3 hours x 9 kW x $0.1/kWh = $2.70).
Compressor test $1.0/min Labor Rate

$1.0/min Equipment Rate
$2.0/min Total Station Rate Station
Cost
# of Operation Markup After
Operations Time Total time (s) Factor Markup
1 1800.0 sec test system 1800.0 sec
1800.0 sec
Station Cost = $60.00 1.62

$97.20
Total Test Cost (marked up) = $97.20

Figure 6-13. Compressor System Test
6.5 Hydrogen Compressor Cost Summary
Figure 6-14 summarizes the projected cost of a 9-stage low inlet pressure hydrogen compressor
and the 6-stage high inlet pressure hydrogen compressor. Both are designed to compress 4.8
kg/h to 7,000 psi from either 1 atm(low inlet pressure) or 150psi (high inlet pressure). However
it should be noted that the compressors operate at a conservative 535 rpm. Should rpm be
increased to 2,100, as currently used in the Hurricane Compressor CNG 90 natural gas
compressor, hydrogen flow rate would increase by a factor of 3.9 for only slightly greater system
cost.105
To ensure reliability of service, it is industrial practice within the chemical industry to utilize
several compressors in parallel, each operating at a reduced throughput. Thus when one
compressor is not operational due to scheduled maintenance or unscheduled breakdown, the
other compressor can increase duty load without compromising overall compression
requirements. It is, however, not industrial practice within the natural gas vehicle refueling
industry. Natural gas refueling stations typically use only one compressor to boost line pressure
natural gas to 24.8 MPa (3,600 psi). Remarkably, natural gas compressor utilization is reported
at 99.5% with downtime only an hour or two during those infrequent breakdowns.106 The total
cost figure presented in Figure 6-14 assumes only one compressor, and does not include the 15%
markup applied by the HFA manufacturer.
105
Compressor block and head cost would largely be unaffected. However, electric motor and intercooler cost would rise appreciably.
106
Case Studies of Cost Effective Natural Gas Fueling Stations, Gas Research Institute Publication, Undated.
93
Low Inlet High Inlet

Pressure Pressure
Compressor Compressor
Materials/Manufacture $18,312.06 $15,868.18
Compressor/Blowdown $502.20 $502.20
Vessel Assy
Interstage Cooler $101.06 $101.06
Manufac./Assy
Skid $111.00 $111.00
Assembly/Mounting
System Test $97.20 $97.20
15% Miscellaneous $2,868.53 $2,501.95
Allowance
Total Estimated Cost $21,992.05 $19,181.59
Figure 6-14. Hydrogen Compressor Cost Summary (4.8 kg/h rated capacity)
7 STATIONARY STORAGE OF COMPRESSED HYDROGEN

The storage of compressed, pure hydrogen in high pressure cylinders is the least complex storage
mechanism currently available for on-site bulk storage at local or regional hydrogen generation
and dispensing stations. Unlike liquefied hydrogen (LH2), compressed hydrogen (CH2) does not
require energy and capital intensive liquefaction equipment or suffer from hydrogen losses due
to boil-off. Unlike metal hydride storage, CH2 systems are not cycle-life limited or susceptible
to gas contaminants. However CH2 storage systems have perceived weaknesses in three areas:
1. system cost
2. compression energy requirements to reach storage/dispensing pressure
3. land use/safety fears related to large compressed hydrogen pressure vessels
The first issue, system cost, is the primary topic to be discussed below with the other two topics
discussed as they relate to system cost.
7.1 Compressed Hydrogen Storage Requirements
Figure 7-1 summarizes the hydrogen storage requirements for two sizes of service station
considered here. The first service station is appropriate for small fleets or very small public
stations. The station is assumed to only utilize one hydrogen reformer appliance and thus refuels
only 20 automobiles per day107. This is equal to or larger than most vehicle fleets in the United
States. Additionally, it is adequate to support 183 fuel cell vehicles in the community
surrounding the service station. Thus this relatively small hydrogen refueling plant is well sized
for the early days of fuel cell vehicle production when high geographic concentrations of fuel
vehicles are not expected. The other type of station considered is defined as an average public
station that gangs together 8 hydrogen fueling appliances and can refuel 160 vehicles per day and
107
As discussed later in the report, refueling is defined as a refilling of 80% of the fuel tanks total capacity.
94
support a community of 1,464 fuel cell vehicles. The station is sized to approximately match the
175 vehicle refuelings conducted in the average US gasoline service station.
The on-site hydrogen dispensable storage capacity is set at 40% of daily maximum production108
capacity for two purposes: 1) to serve as an overnight storage reservoir when vehicle refueling
may not take place and 2) to allow for demand surges, i.e., multiple cars arriving simultaneously
at the service station. Each fuel cell vehicle (FCV) is assumed to achieve 80 mpgequivalent fuel
economy and thus requires 4.76 kg of hydrogen to travel 611 km (380 miles) on the 1.25 times
accelerated combined federal drive schedule. Average annual mileage of each FCV is assumed to
be 19,311 km (12,000 miles).
Dispensable Total H2
Average Maximum Average H2 Storage Storage
Number of H2 Vehicle No. of Requirement Req.
Application Reformer Production Refuels per FCV’s (40% of max (at 58% H2
Appliances Capacity day110 Supported daily capacity) recovery
109
(80% (kg) factor)
(kg/day) refills) (kg)
Small Public
Station /
1 75.4 20 183 46 79
Medium-
Large Fleet
Average
Public
Service 8 603 160 1464 368 634
Station111
Figure 7-1. Dispensable Hydrogen Storage Requirement Based on 69% Capacity Factor112
Praxair Inc. has conducted a detailed analysis of gaseous hydrogen service station storage
requirements and dispensing options.113 Within the report, two main storage and dispensing
options are examined: cascade storage/filling and booster/compression storage/filling. Cascade
filling is currently utilized by natural gas vehicle service stations and consists of charged banks
of pressure vessels sequentially filling the vehicle tanks. Typically, three or four cascades are
used. The advantage of this approach is preservation of high-pressure gas in at least one of the
storage vessels so that the vehicular tank can be filled to design pressure (34.5 MPa/5,000 psia)
108
This percentage is based on a 15 hour dispensing period and the fuel cell vehicle load profile detailed in “Distributed Hydrogen Fueling
System Analysis” by Thomas, et.al.
109
Based on 69% station capacity factor at 95% station utilization.
110
Based on 80mpgequivalent fuel cell vehicle mileage and 380 miles maximum vehicle range.
111
The average gasoline service station in the United States refuels approximately 175 vehicles per day.
112
Due to fluctuation in hydrogen demand, the average capacity factor for these very small fueling stations is likely to be as low as 50% of this
maximum capacity. Previous Ford/DTI hydrogen infrastructure studies have assumed 69% average capacity factor averaged over a year for on-
site hydrogen production.
113
Hydrogen Production and Fueling System Infrastructure for PEM Fuel Cell-Powered Vehicles, Final Report, T.G. Halvorson,. C.E. Terbot,
M.W. Wisz, Praxair, Inc. April 12, 1996, prepared for The Ford Motor Company under Ford Subcontract No. 47-2-R31157 of the Ford Motor
Company/Department of Energy Contract entitled “Research and Development of Direct Hydrogen Fueled Proton Exchange Membrane (PEM)
Fuel Cell System For Transportation Applications”.
95
efficiently. A disadvantage of the approach is low hydrogen recovery factor: only 58% of the
hydrogen in each tank is cycled in and out of the tank114. Thus storage of 1 kg of dispensable
hydrogen requires storage of 1.7 kg of total hydrogen.
In the booster/compressor concept, hydrogen is stored in pressure vessels at an intermediate

pressure (24.8 Mpa, 3,600 psia) and the compressor is used to boost the hydrogen from this
intermediate pressure to full pressure directly into the vehicle. The compressor must be sized for
a higher mass flow to allow fast fill of the vehicle tanks. Additionally the compressor requires
more complex routing since the compressor is fed by both the reformer and the intermediate
pressure storage tanks. To conserve energy, vehicle tanks are first filled from the intermediate
pressure vessels to the extent possible, then filled by the booster compressor to top them off.
Advantages of the booster compressor concept are potentially lower storage tank cost (since the
storage pressure is only 3,600 psia instead of 7,000 psia), and lower energy cost (since not all
hydrogen gas is pressurized to maximum pressure). Additionally, because the intermediate
storage pressure fluctuates between 450 psia and 3,600 psia, tank recovery factor is much higher,
reaching 95%. Praxair estimates storage tank cost savings of 50% compared to cascade system
tankage.
Figure 7-1 lists the two categories of refueling stations, their hydrogen production capacity, and
stationary storage requirement. Based on a detailed examination accounting for daily and
seasonal dispensing fluctuations, DTI estimates that refueling station dispensable storage should
be approximately 40% of maximum daily production (or 58% of average daily production
assuming a 69% capacity factor). Consequently, 46 kg of dispensable storage is required for a
Small Public Station and 368 kg for an Average Public Station. However, only 58% of the
hydrogen is recoverable in the cascade mode.
While the Praxair analysis indicates that substantial savings can accrue by using the booster
compressor refueling strategy, more recent analysis115 by Directed Technologies Inc. indicates
that cascade storage and booster compressor strategy are approximately equivalent in cost.
Consequently the cascade storage strategy is adopted in this report for the storage cost estimate
as it is the simplest technique and does not require a high flow-rate booster compressor which
may not have suitable reliability given the intermittent nature of operation.
7.2 Types of Liners
The most prevalent hydrogen storage tank in use today is the solid-walled steel tank. Steel tanks
have relatively thick steel walls to contain the tank internal pressure and maintain serviceability
over many cycles and very long lifetimes in a hydrogen environment. Built to ASME standards
for metal pressure vessels, the tanks have a safety factor on burst pressure of 4 resulting in high
weight and high cost.
114
Based on cascade dispensing to a 5,000 psig receiver cylinder as calculated using rational-function curve fits to data provided by Tom
Halverson of Praxair based on his MathCAD model described in Halvorsen(1996).
115
“Distributed Hydrogen Fueling Systems Analysis”, C.E. Thomas, John P. Reardon, Franklin D. Lomax, Jr., Jennifer Pinyan, Ira F. Kuhn, Jr.,
prepared for The Hydrogen Program Office, Office of Power Technologies, U.S. Department of Energy, under Grant No. DE-FG01-99EE35099,
October 2000.
96
Composite wrapped tanks offer an intriguing alternative to the standard metal tanks. Composite
tanks utilize high strength carbon or glass fibers as strength members to contain the hydrogen gas
internal pressure. However since fiber/resin composites are porous to hydrogen molecules, an
impermeable liner is constructed within the tanks to serve as both a hydrogen barrier and as a
wrapping mandrel for the fiber/resin. Three main options exist for this liner:
1. a metal liner, typically 0.095 inches 6061-T61 aluminum,
2. a polymer liner, typically 0.25 inches high density polyethylene (HDPE),
3. a metallized polymer bladder, typically 1,000 angstrom silver film sandwiched between
0.001” thick layers of low density polyethylene (LDPE).
7.2.1 Metal Liners
There are no unresolved longevity issues associated with metal liners as they have been in use
for many years. Typically the liners are made starting from extruded seamless aluminum tube
and the production process consists of 6 major additional steps.116
1. Drawing: Starting from 0.25-0.5” thick extruded seamless 6061 aluminum tube, the tube
is drawn to the desired thickness and length. Liner thickness is based on cycle life with
0.095” considered appropriate for 10,000 cycles as an automotive storage tank.117,118
Since the stationary storage tank application will incur more frequent shallow pressure
cycles ( daily or perhaps many cycles per day) and the expected tank lifetime may be
longer than the nominal 10 year lifetime of an automotive system, the aluminum liner
thickness is arbitrarily increased to 0.15”.
2. Hot Metal Spinning: To form the oblate ellipsoid end-caps of the liner, the drawn
aluminum tubing is hot metal swaged to the appropriate shape.
3. Heat Treatment: The formed liner is heat treated to the appropriate temper, typically T61
for 6061 aluminum.
4. Drilling/Tapping: Each end of the liner is drilled and then tapped to form the appropriate
diameter threaded openings for the solenoid valve at one end and a plug at the other.
5. Cleaning: The liner is pressurized hot water spray cleaned.
6. Grinding: The liner is ground to a smooth finish to both provide a smooth surface upon
which the fibers will evenly grip and to allow the visual detection of flaws.
An approximate price quote of $4,000-$5,000 per aluminum liner was obtained from Hydrospin
Corp. based on a 20” diameter, 23’ long aluminum liner of indeterminate wall thickness and an
annual production run of 2-3 thousand. As shall be shown, this price is substantially higher than
the HDPE liner cost. Consequently, metal liners are dropped from further consideration and cost
estimation focuses on polymer liners.
7.2.2 HDPE Polymer Liners
HDPE polymer liners are in extensive use for natural gas storage for vehicles. While currently
not used for stationary hydrogen tanks, the extrapolation to hydrogen storage tanks is straight-
forward. The only uncertainty is lifetime of the polymer in a hydrogen environment. However,
116
Private communication with Peter Lindgren of Hydrospin, Huntington Beach CA, June 1997.
117
Private communication with Rex Haddock of EDO Fiber Sciences, Salt Lake City, Utah, March 1995.
118
If a car travels 12,000 miles/year over a ten year lifetime, and refills 80% of the tank each refueling, then only 400 refueling cycles are
experienced. Thus, designing for 10,000 refill cycles is perhaps excessively conservative.
97
preliminary indications are HDPE is sufficiently stable to provide adequate lifetimes.

Consequently, polymer liner fabrication cost is explored in some detail.
The permeation rate of hydrogen through HDPE is approximately 3 orders of magnitude higher
than pure silver metal.119 Consequently, a liner thickness of 0.25” is selected to achieve a
leakage rate of much less than 0.5%/year (through the liner walls) even at 7,000 psia service
pressure.
The manufacturing method for high volume polymer liner tank production includes three
steps.120
1. HDPE extrusion of the cylindrical portion of the tank and subsequent cut-off to proper
length
2. Insertion molding: Stainless steel bosses are customarily embedded within the liner end-
domes to provide a secure threaded attachment point for the solenoid valve and end plug.
Consequently, an insertion molding process is appropriate to mate the bosses to the
HDPE end-dome.
3. Friction/Vibration mating of the end-domes to the cylindrical center section. Since
friction bonding locally melts the polymer to join the two pieces, there should be no local
structural strength reduction at the mating seam as there would be in an aluminum welded
seam.
HDPE liners with the above described fabrication method are adopted as the baseline liner for
cost analysis purposes.
7.2.3 Metallized Polymer Bladders
A tank liner innovation comes from a solar power regenerative fuel cell aircraft design by
Lawrence Livermore National Laboratory (LLNL). Aircraft are extremely weight sensitive and
therefore, to reduce hydrogen tank weight, the metal liner of the compressed hydrogen vessel
was replaced with a laminated metallized polymeric bladder to function as the gas barrier. The
design calls for 1,000 angstrom silver films to be deposited by sputtering onto 0.0025 cm thick
polyethylene terephthalate (PET) and latex bonded on both sides with 0.0025 cm of 5% by
weight ethylene vinyl acetate (EVA) and 95% by weight low density polyethylene (LDPE)
copolymer.121 A derivative of this method was experimentally demonstrated by a team of
researchers from LLNL, EDO Fiber Science Division, and Aero Tec Laboratories, Inc. under
Department of Energy and Ford Motor Company sponsorship.122 To simplify metallized liner
fabrication, only the cylinder portion of the liner was metallized bladder: the end-domes were
roto-molded HDPE polymer. Mating of the bladder to the HDPE end-domes proved problematic
and thus while the liner was fabricated and successfully carbon fiber overwrapped
(demonstrating use of an inflatable bladder as a wrapping mandrel), leakage occurred at the
119
Personal communication, Fred Mitlitsky, Lawrence Livermore National Laboratory, June 1997.
120
Personal communication, Tammy Siama, Lincoln Composites, Lincoln, Nebraska, October 1997.
121
“Regenerative Fuel Cells for High Altitude Long Endurance Solar Powered Aircraft,” Fred Mitlitsky, et. al. Lawrence Livermore National
Laboratory, UCRL-JC-113485 Preprint, July 28, 1993.
122
DOE Contract No. DE-ACO2-94CE50389 to Ford Motor Company, Detroit, Michigan. Tank liner work summarized in “Final Report for
Hydrogen Storage”, Fred Mitlitsky, Lawrence Livermore National Laboratory, March 1997.
98
cylinder/end-dome seams during hydrotesting. Researchers are confident that the problems
could be corrected given minor design changes.
For stationary storage tank application, liner weight is not a vital issue (as it is for automotive
tank applications). Consequently, cost would be the primary reason to select metallized bladders
over aluminum or polymer liners. Future embodiments of metallized bladders will be either
entire metallized bladders (i.e., cylinder plus end-domes) or will correct the early developmental
problems and be metallized bladder cylinders mated to polymer end-domes. In the first instance,
some mechanism is required to securely mate the bladder to the end-dome bosses. A very low
cost, leak proof method has not yet been proposed so consideration shifts to the second option.
For bladder cylinder/HDPE end-dome construction, the end-domes will be virtually identical to
the full polymer liner end-domes. Thus, end-dome cost would be the same and all cost
differential between the liner types stems from the cylinder section fabrication and mating to the
end-domes. As shall be shown later, cylinder materials, fabrication, and mating cost sum to less
than $50 (before markup) and thus, metallized polymer liners could at most save that much
money. Consequently, detailed costing will focus on HDPE liners, with the knowledge that
should metallized polymer liners prove technologically adequate, they offer a slight potential
cost savings.
7.3 Types of Fiber
Two general classes of fibers are primarily considered for pressure vessels: carbon fibers and
glass fibers. Of these, carbon fibers used with an epoxy resin currently achieve the highest
strength. The carbon fibers under consideration for tanks are all polyacryolonitrile (PAN) fibers,
and are primarily differentiated by tensile strength. Thus, T-1000 fiber from Toray nominally
has a tensile strength of 1,000,000 psi, and T-700 fiber, also from Toray, a tensile strength of
700,000 psi. (In practice, however, a tensile strength of 940-950 kpsi more accurately represents
T-1000G, and 685 kpsi more accurately represents T-700. Zoltec, a carbon fiber manufacturer
based out of St. Louis, MO, markets a lower strength fiber named Panex-33123 that achieves
approximately 525 kpsi tensile strength.)
All three carbon fibers (T-1000G, T-700, and Panex-33) are derived from a polyacryolonitrile
(PAN) precursor which sells for $6.6-$8.8/kg from its Japanese supplier.124 The PAN is
purchased by carbon fiber manufacturers who process it into carbon fibers in five steps: (1)
spinning the PAN to form a fiber, (2) stretching, (3) oxidation and stabilization at 220°C under
tension, (4) carbonization in inert atmospheres up to 1,600°C, and (5) graphitization up to
3,000°C. The specific heat treatment regimen determines the ultimate fiber properties and thus
product classification (T-1000, T-700, Panex-33). Additionally the higher strength fibers
undergo further, and more costly, processing.
Cost of carbon fibers differs greatly but is a reflection of processing cost differences, differing
material selectivity, and economies of scale rather than inherent precursor cost differentials. Of
these, production volume is the primary reason for the large cost disparity between the T-1000G
123
The 33 in the material name refers to a fiber modulus of 33 Msi.
124
Although the T-1000G fibers are made from a more selective and thus higher cost precursor.
99
sales price of approximately $154/kg125 and a Panex-33 price of $18/kg.126 While Panex-33 is
becoming a commodity item used in many consumer applications with annual production of 3.8
million kg per year,127 T-1000G continues to be used primarily in military and aerospace
applications where performance demands outpace cost concerns with production at
approximately 45,000 kg/year.
Panex-33, from Zoltec Corp., is a lower strength and lower cost carbon fiber. However, it is
unproven for continuous filament wound pressure vessels and thus is not carried as a serious
fiber for station hydrogen tanks.
Structural glass, or S-glass, is the primary option for glass fibers. Although S-glass has a lower
tensile strength than the carbon fibers, S-glass is dramatically cheaper, typically selling for $4.4
to $6.6/kg. Thus, for applications such as stationary storage where high system weight is not a
problem, S-glass might be a reasonable candidate material were it not for stress rupture concerns.
7.4 Stress Rupture and Safety Factor
Safety factors are used in mechanical designs to prevent component failure if subjected to greater
than design stress and to account for material property uncertainty. In the case of fiber wrapped
pressure vessels, safety factors are primarily used as a defense against stress rupture. Stress
rupture refers to the spontaneous failure of composite fibers when subjected to loads for an
extended period of time. The loads do not have to approach fiber yield stress for failure to incur.
Rather for each type of fiber, empirical data show a clear relationship between fiber stress and
time to failure. Glass fibers suffer most severely due to creep and thus require a safety factor of
three to four to reduce fiber stress sufficiently to achieve nearly infinite service life. Carbon
fibers are not impacted as greatly and only require a safety factor of 1.2 to 1.3 to achieve long
life.
While ASME pressure vessel code dictates a safety factor of 4 for solid metal pressure vessels,
the code does not address composite pressure vessels. However, composite pressure vessels for
natural gas vehicles (NGV) are in extensive use. Since NGV tanks are directly analogous to
stationary hydrogen storage tanks, the NGV tanks provide a relevant set of safety factors. Safety
factors for NGV automotive pressure vessels mandated by the NGV-2 code128,129 are shown in
Figure 7-2.
125
Private communication, Jane Schoeken, Toray-US, for T-1000G in orders over 534 kg..
126
Private communication, Joan Schumacher, Zoltec Corporation, St. Louis, MO, September 1997.
127
Private communication, Peter Hurwitch, Zoltec Corp., St. Louis, MO, December 1997. Panex-33 production capacity is expected to increase
to 8.8M kg/year when a new Texas production facility comes on line in 1998.
128
American National Standard for Basic Requirements for Compressed Natural Gas Vehicle (NGV) Fuel Containers, ANSI/AGA, NGV2-1992,
First Edition 1992 Approved August 6, 1992.
129
In May of 1996, a draft proposed revision of the 1992 NGV2 standards was circulated by ANSI for review and comment. However, no
changes to the safety factors were proposed.
100
Waist Wrapped Full Wrap

(Type 2) (Types 3 and 4)130
S-Glass 2.65 3.5
Carbon 2.25 2.25

Figure 7-2. NGV2 Mandated Safety Factors for Composite Wrapped Pressure Vessels
Only fully fiber wrapped pressure vessels are considered here for hydrogen stationary storage
and thus the full wrap NGV safety factors are most appropriate. However, for cost conservatism
a safety factor of 3.0 is adopted for the cost analysis presented below.
7.5 Current Pressure Vessel Pricing
Directed Technologies has previously examined hydrogen storage vessel cost for stationary
applications.131,132 Those findings are summarized and updated below and in Figure 7-3.
Steel Tanks
Current stationary storage tank costs were obtained from price quotes133 from CP Industries of
McKeesport, Pennsylvania. The tanks are 41.4 MPa (6,000 psia) ASME certified steel tanks134
with 41 cm outer diameters, length of 9.14 m (30 feet), and hold 765 liters (27 cubic feet) water
volume. Approximate price for a bare tank in small quantities is $13,000 yielding $642 per kg
hydrogen stored. For reference, larger 5,000psi rated pressure vessels,135 primarily for CNG
stationary storage, are produced at hundreds to a thousand per year quantities and sell for
approximately $500 per kg hydrogen stored. CP Industries feels that only minor cost reduction
is achievable with increased manufacturing volume due to high number of processing steps that
conducted one pressure vessel at a time. Consequently, the authors estimate the price of 7,770
psi rated vessels, to be used for 7000psi hydrogen storage, manufactured at approximately 1,500
vessels per year to support 250 systems per year, at approximately $642/kg- the increase in
pressure is offset by the increase in production rate.
Additionally, price quotes were obtained from the Norris Cylinder Co. in 100+ tank purchase
quantities for an ASME approved 7,000psi vessel holding 1.33kg of hydrogen. The tank price is
approximately $726 per cylinder resulting in a vessel price of $545.80/kg hydrogen stored. A
Department of Transportation (DOT) approved 6,000psi K-cylinder is also available priced at
130
Both Type 3 and Type 4 tanks are fully composite wrapped. However, Type 3 tanks have metal liners and Type 4 tanks have non-metal liners.
131
Distributed Hydrogen Fueling System Anlaysis, C.E. Thomas, J.P. Reardon. F.D. Lomax, Jr., J. Pinyan, and I. F. Kuhn, Jr., October 2000.
Prepared for The Hydrogen Program Office, Office of Hydrogen Technologies, US Department of Energy under Grant No. DE-FG01-
99EE35099.
132
“Integrated Analysis of Hydrogen Passenger Vehicle Transportation Pathways”, C.E. Thomas, B.D. James, F.D. Lomax, Jr., I.F. Kuhn, Jr.,
prepared for The National Renewable Energy Laboratory, U.S. Department of Energy, March 1998.
133
Personal Communication with Jim Celinski of CP Industries, February 2002.
134
These vessels have a safety factor of 3 and a rated operating pressure of 6,667psi. However, standard practice is to set the maximum operating
pressure 10% below rated pressure. Thus these 6,667 psi rated vessels are referred to as 6,000psi vessels.
135
20 inch diameter, 23 feet length, 34.2 cubic foot water volume, approximately $11,000 for the bare tank.
101
approximately $671 per cylinder or $632.80/kg hydrogen stored in 100+ quantities. The DOT
approved tanks meet DOT3AA6000 specifications.
Composite Tanks
NGV Systems, Inc. of Long Beach, California produces 24.8 MPa (3,600 psia) composite
reinforced aluminum cylinders for natural gas vehicle fuel storage. These composite reinformed
aluminum tanks range in price from $331 to $474 per kg of hydrogen stored. An additional
storage tank data point is obtained from Aerojet of Sacramento, California which recently
completed a price projection for large carbon/fiberglass wrapped, aluminum lined pressure
vessels. The tanks are rated for 26.2 MPa (3,800 psia) and measure 7 m long with a 50.8 cm
internal diameter. Though designed for natural gas storage, they could just as easily store
hydrogen gas. Ten tanks are stacked in parallel on a wheeled rack so that the entire ten tank
assembly can be easily rolled into a truck for transportation and rolled off the truck at the
destination site. Aerojet conducted only a cost analysis and is not currently producing the tanks.
Estimated price for the ten tank assembly including wheeled storage rack was $150,000.136 Thus
each tank was approximately $14,850 assuming the wheeled rack cost $1,500. This price is
based on 1,800 tanks produced yearly, and results in a price of $604/kg of stored hydrogen.
Type of Tank Source/Purchase Kg H2/vessel Price per kg H2

Quantity
6,000 psi Steel CP Industries 20.2 kg $642/kg
H2/vessel
ASME 7,000psi Norris Cylinder 1.33 kg $548/kg
Steel cylinders (100+ cylinders) H2/vessel
(K cylinders)
DOT 6,000psi Norris Cylinder 1.06kg H2/vessel $633/kg
Steel Cylinders (100+ cylinders) (K cylinders)
3,600 psi NGV Systems 3.35kg H2/vessel $331/kg
Composite Inc. (low
reinforced Alum. quantity)
3,600 psi Aerojet estimate 24.5 kg $604/kg
Carbon/Fiberglass (1800 tanks/yr) H2/vessel
alum liner
Figure 7-3. Summary of Pressure Vessel Price Quotes
7.6 Projected Carbon Composite Pressure Vessel Pricing
Previous DTI reports have examined in detail the cost of fully carbon fiber/resin wrapped,
polymer lined pressure vessels for both stationary and automotive hydrogen storage. These
previously generated cost models were updated to reflect recent raw material changes, a 250
systems per year manufacturing rate, a 7,000 psi nominal storage pressure, a safety factor of 3.0
and the appropriate tank dimensions based on station size. The DFMA methodology was applied
to each manufacturing step as described below and in Figure 7-4.
136
Personal Communication with Nancy Shrimp of Aerojet, June 1997.
102
9.00 R, inch 3.00 Safety Factor

71.50 L cylinder, inch 7000 Operation pressure, psia
83.50 Total Length T-700 Fiber Type
4.64 L/D 1.67E+06 Fiber/resin PF
1.5 Endcap Ratio of Semi-Major 12.30 Eff. Int Vol.
0.75 eccentricity 267.31 Fiber/Resin Wt, lbs
10.53 Cylindrical Vol, ft3
1.18 Endcap Volume, ft3 32.558 % resin
11.71 Total Internal Volume, ft3 67.44186 % fiber
0.25 Cyliner Liner wall thickness, i 85.91 Ext. Length, in
0.25 Aver. Endcap Liner wall thick 20.41 Ext. Diam., in
0.59 Cyliner Liner Volume, ft3
0.12 Endcap Liner Volume, ft3 42.43 Liner Wt., lbs
59.65 Liner Density, lb/ft3 87.03 Resin Wt, lbs
42.43 Liner Wt, lbs 180.28 Fiber Wt., lbs
96.60 Fiber/Resin Density, lb/ft3 3.04 Metallic Boss, lbs
2.77 Fiber/Resin Volume, ft3 3.1 Solenoid Valve & PRD
0.95 Fiber/Resin Thickness, inche 315.88 TOTAL TANK WT, lbs
2.41 Cylindrical Vol, ft3 1.34E+06 Total PF
1.66 Endcap Volume, ft3 16.49 Total Volume, ft3
1.863 H2 Density, lb/ft3
9.89 H2 Stored at 7 kpsi, kg
Materials Cost:
0.5 liner cost, $/lb
1.5 Resin Cost, $/lb
9 Loaded Fiber Cost, $/lb
2 Abrasion Fiber Cost, $/lb
10% % of loaded fiber weight that is abrasion fiber weight
Manufacturing Cost:
$2.00 $/min of fiber wrapping machine
3 number of spindles
6.562 Average Tape laydown rate, tape feet/sec
4.92E-03 lbs of dry load fiber per foot of tape
36620 tape feet for tank
1.1 mark-up Factor in wrapping time for abrasion fiber
102.31 wrapping time, minutes
$68.21 wrapping cost ($/min*min/#ofspindles)
10 Tank set-up and fiber change out time, minutes
Labor Cost
$1.00 Labor Rate, $/min ($60/hr but can obs./oper. multiple spindles at once)
3 Number of spindles observe/operated by one person
$0.33 effective $/min labor rate ($/min divide by number of spindles obs./oper.)
112.31 time, minutes (wrapping time plus set-up/take-down time)
$37.44 labor cost
$0.05 Gell Cost, $/min (gel at 100deg F)

360 Gelling time, minutes
$30.60 Labor cost to mind the gelling/curing process
(modeled as 1 person to mind all cured tanks)
$48.60 Gelling Cost, $
$0.05 curing cost, $/min

150 Curing time, minutes
$7.50 Curing Cost, $
$72.18 Liner Fabrication Cost, $

$34.57 Boss Cost, $ (materials and manufacture)
Manufacturing Cost Summary (pre-tank mfc markup)

$68.21 wrapping cost ($/min*min/#ofspindles)
$37.44 labor cost
$48.60 Gelling Cost, $
$7.50 Curing Cost, $
$72.18 Liner Fabrication Cost, $
$233.92 Total Manufacturing Cost, $ (post-markup)
Figure 7-4. Tank Cost Estimate Details
103
Markup Factors
1.50 liner material cost
1.50 structural fiber material cost
1.50 structural resin material cost
1.50 abrasion fiber material cost
1.50 abrasion resin material cost
1.50 boss material cost
1.50 manufacturing cost
$31.82 liner material cost

$2,433.78 structural fiber material cost
$195.82 structural resin material cost
$54.08 abrasion fiber material cost
$13.21 abrasion resin material cost
$51.86 boss material &manufacture cost
$350.89 manufacturing cost
$3,131.46 Tank Cost, $ (including markups)

$0.00 Solenoid Cost (including markup), $
$50.00 Transportation
$20.00 Inspection
$3,201.46 Subtotal
$320.15 10% Cost Contingency
$3,521.61 Total Cost, $ (including markups and contingency)
$356.02 Cost/kg of H2
Figure 7-4. Tank Cost Estimate Details (continued)
The first step in tank cost estimation is definition of the tank. Based on internal dimensions of
the tank, the fiber tensile strength, the tank safety factor, the mass ratio between fiber and resin,
and the tank liner material and thickness, the weight of each tank element (fiber, resin, liner) is
calculated. The mass of each tank element and the cost per mass are combined to yield material
cost. A T-700 carbon fiber cost of $20/Kg is assumed for all tanks.
Once fiber weight is determined, the tank winding time is estimated based on the weight of fiber
per length and typical winding machine laydown speeds. For T-700 fiber, a 7.4-micron fiber
diameter, 45,714 monofilaments per tow (48K tows), and four tows per strand were assumed for
a total fiber mass of 1.4 x10-2 kg/m. Laydown speed is assumed to be 0.61 m/s consistent with
wet winding.
A machine cost of $2/minute is used for a 4-axis, three spindle fiber winding machine. This cost
is based on a purchase price of approximately $200,000137 and the assumption of relatively low
operating cost -- probably a good assumption since electrical requirements are low. The machine
cost rate is deliberately high to cover potentially higher than expected winding machine capital
cost.138 Use of two spindle winding machines allows two tanks to be simultaneously wound.
Winding machines for golf clubs have up to 20 spindles. But since the stationary storage tanks
are so much larger, only three spindles are postulated.
Current winding machines have the capacity to lay down multiple tows at once, and therefore
require multiple spool kreels. Current spools hold 4.5-6.3 kg of fiber, however, some companies
137
Personal communication, Joe Jansen, McLean-Anderson.
138
Bill Rose of Entek thinks $200K for a 4-axis, 3 spindle winding machine may be a little low.
104
are considering increasing spool fiber weight to 45 kg to decrease spool change out
requirements.139 For this cost study, these larger spools and four tow simultaneous lay down is
assumed. Spool change out time is estimated at two minutes. Since carbon fibers are more
susceptible to impact and abrasion damage than glass fibers, leading pressure vessel
manufacturers routinely blend glass fibers into their ostensibly all-carbon tanks to achieve
improved damage resistance.140 For simplicity, the cost estimate assumes that an additional
quantity of fiberglass (10% of the carbon fiber mass) is added to the carbon tanks. Fiber
wrapping time is also increased 10%.
The tanks are set to gel for three hours at 39°C and cure for two hours at 121°C. Gel duration is
based on detailed tank manufacturing procedures from EDO Fiber Sciences141 and cure time is
based on an SCI carbon tank procedure.142 However in general, gel and cure time are greatly
affected by temperature, and thus can vary widely. The times and temperature are meant to be
representative and at a cost of $0.05/minute, do not amount to much money ($18). The cost of
$0.05/minute is loosely based on a firing cost of $5/ft3 for ceramic parts that occurs at a much
higher temperature and typically lasts 3 days.143
The manufacturing method for high volume polymer liner tank production involves several
steps, described in Section 7.2.2.
In addition to insertion molding of the end-domes and ultrasonic welding to an extruded

cylindrical section, the formed liner undergoes a surface preparation to ensure even adhesion of
the fiber and a hydrostatic leak test. This leak test is intended to identify gross flaws in the liner
prior to wrapping of the expensive carbon fiber. After wrapping, an additional pressure test is
performed with helium gas to verify the integrity of the entire tank system.
A 10% cost contingency is added to ensure cost conservatism and cover incidentals. The costs
presented here are based on tank production rates of approximately 2,750 per year. This is based
on a single three-spindle winding machine producing 12 tanks per day, one shift per day, and 220
work-days per year.144 Labor for two people is allotted: one person to operate the winding
machine and one to tend the gel/cure heat treatments. The winding machine operator is skilled
labor and conservatively priced at $1/min. Because the 7,000 psia tank takes about two hours to
wind, the operator can easily observe all 3 tanks simultaneously. The heat treatment tender is
relatively unskilled labor and is priced at $0.75/min. Since the tanks can be gathered and heat-
treated in one batch per day, the heat treatment tender can also serve as an assistant to the
winding machine operator.
A pressure vessel manufacturer markup rate of 50% is applied. This markup is judged to be
appropriate for a relatively small business ($11M + revenues) that has a moderate product base
139
“Filament Winding with Heavy Towpreg,” Karen Fischer, High Performance Composites, May/June 1996.
140
Lincoln Composites product brochure "All-Composite NGV Fuel Tanks by Lincoln Composites", 1996.
141
“Final Report for Hydrogen Storage”, Fred Mitlitsky, Lawrence Livermore National Laboratory, March 1997.
142
“Lighter Weight Fiber/Metal Pressure Vessels Using Carbon Overwrap”, E.E. Morris, et al., Structural Composites Industries, Pomona,
California, American Institute of Aeronautics and Astronautics, AIAA-86-1504, June 1986.
143
Ceramic firing cost estimate from Sigi Mundhinki at Coors.
144
With the winding machine laydown rates assumed, a set of three tanks requires 2.13 hours to complete. Thus the shift needs to be 9 hours long
to achieve 4 complete cycles. This is longer than the standard 7-hour productive work day but is not expected to have an appreciable affect on
pressure vessel cost.
105
over which the administrative and managerial costs of advertising, accounting, and corporate
planning can be amortized.
7.7 Valving and Connection Hardware
For systems with many tanks, the valving and interconnections between tanks can be quite
costly. Figure 7-5 and Figure 7-6 detail the piping schematics assumed for the large steel tanks,
7000psi ASME tanks, and 7000 psi composite tanks. In all arrangements, 4 stage cascades are
assumed with each cascade controlled by a pneumatic metering valve, a manual shut-off valve,
and a safety pressure relief valve. Flex line is used to connect the tanks to facilitate in-field
assembly of the cascade. Additionally, welding connections are assumed everywhere except the
cylinder fittings and the pneumatic valve connections as these may need to be replaced for
servicing during the lifetime of the unit. Additional pneumatic and manual valves are placed at
the inlet and outlet of the system to allow appropriate charging and discharging of the entire
system. The entire valve train is assembled at the factor. Welded connections are made by
orbital welders to maintain quality control and rapid assembly. The four very large steel tanks
are assumed to be transported individually and assembled on-site. The smaller steel K-cylinders
and the moderate size composite tanks are assumed to be factory assembled into modular racks
so that field installation is easy and rapid.
Flow control of the cascade can be affected by placing the control valves at either the storage
tanks or within the dispenser. When placed at the storage tanks, only a single hydrogen line
needs to run out the dispenser. When placed within the dispenser, 4 pressurized hydrogen
lines(one for each stage of the cascade) plus a fifth control line need to run between the tanks and
the dispenser. While the cost estimates presented here assume placement of the control valves at
the storage tanks, this arrangement is only practical for use with a single dispenser, since a set of
control valves is needed for each simultaneous dispensing procedure. Control valving is further
discussed in section 8.0.
PNEUMATIC VALVE
MANUAL SHUT-OFF
PNEUMATIC VALVE
MANUAL SHUT-OFF
PRESSURE RELIEF
DEVICE
TANK
Figure 7-5. Connection Schematic for Large Steel Tanks
106
PNEUMATIC VALVE
MANUAL SHUT-OFF
TANK
FLEXIBLE LINE
PRESSURE RELIEF
DEVICE
Figure 7-6. Connection Schematic for ASME Tanks and Composite Tanks
7.8 Storage System Cost Summary
A summary of total system cost appears in Figure 7-7 for both the Small Public Station and the
Average Public Station scenario. Composite pressure vessels are estimated to be the lowest cost
system primarily due to the lower cost of the tanks themselves. The composite pressure vessel
cost of $356/kg H2 is judged to be achievable due to a moderately high rate of manufacture, use
of a HDPE plastic liner that is much cheaper to form than a metal liner, and use of mid-grade
composite fiber that has been steadily decreasing in price due to economies of scale. Additional
cost reduction can be achieved by producing larger tanks (and thus lowering connection costs)
but a moderate size tank has been chosen to allow tank stacking to maintain a small overall
hydrogen storage station footprint. The costs presented in Figure 7-7 do not include the assumed
15% HFA manufacturers pass-through markup rate. With that markup, the HFA manufacturers
component price for the 7,000psi composite system is $40,083.
107
Small Public Small Public

Station/Med- Station/Med-
Application Large Fleet Large Fleet
(79 kg H2 storage (634 kg H2 storage
requirement) requirement)
Large Steel Tank

7000psi
Tanks (4 tanks) (at $642/kg) $50,718 $407,028
Conn. Hardware $4,854 $8,319
Assembly $740 $1,887
Total $56,312 $417,234
ASME(K)7000psi
Tanks (91 tanks) (at $545.8/kg) $43,118 $346,037
Assembly $1,360 $4,905
Total $60,024 $445,614
Composite-7000psi
Tanks (11 tanks) (at $356/kg) $28,124 $225,704
Assembly $1,204 $3,706
Total $34,855 $244,224
Figure 7-7. Summary of Hydrogen Storage Systems
8 DISPENSERS
A hydrogen dispensing system is required to meter hydrogen gas from the stationary hydrogen
pressure vessels to the waiting fuel cell vehicles. The system must have appropriate logic to
control the flow of hydrogen from the appropriate stage of the four-stage cascade storage
system(unless flow control is conducted external to the dispensing head), record the actual mass
of hydrogen transferred, and maintain safe conditions during all phases of the transfer. In
addition, it must have breakaway refueling hoses to prevent significant amounts of hydrogen
discharge should a fuel cell vehicle pull away from the dispenser with the hose still attached, and
should allow for a pressure compensated hydrogen fill so as to achieve a rapid and full refueling.
The dispensing equipment chosen to accomplish these goals is based on prototype fast-fill
hydrogen dispensing equipment from Kraus Group Inc. The Kraus SHM 1CLG system has the
following features:
• system components sized (where possible) for a 5,000 psig service pressure
• 500 scfm nominal flow capacity
• complete dispenser unit housed in a stainless steel enclosure
• overhead hose retractor
• custom fitted nozzle holster
• a breakaway coupling assembly
108
• full control through use of a mass flow meter

• electronic pressure-temperature compensation
• use of Kraus InfoPac transceiver allowing price setting and totalizer access through a
hand held infra-red transmitter unit
• substantial safety features and annunciation technology
Due to the electronic complexity of the dispensing unit, a detailed DFMA analysis was not
conducted. Instead, a price quote of $24,370 was obtained for the above-described units in small
quantities.145
Since hydrogen dispensing equipment will be very similar to compressed natural gas dispensing
equipment, a price quote was also obtained from Tulsa Gas Technologies Inc.(TGT) of Tulsa,
Oklahoma. TGT sells a Model 7103-6-CNG50 single hose, fast fill, 3-stage cascade sequencing,
fully automated, CNG dispenser including credit card reader for a list price146 of $25,452.
Control valving for the 3-stage cascade is estimated to compromise approximately $7,500 of the
total price. A dual hose CNG dispenser, including redundant flow control to allow two
simultaneous independent refueling has a list price of $36,456. These dispensers are currently
produced in quantities of 75-100 per year and only a modest price reduction is anticipated by
TGT for volume sales of 250 units per year.
Neither the Krauss nor TGT dispensing units meet the 7,000 psig hydrogen dispensing pressure
requirement and both assume a 3-stage rather than 4 –stage cascade. Adjusting for increased
pressure capability, subtracting $5,000 for the control valving since these controls are assumed to
be mounted with the storage tanks, and applying a 10% price discount for volume purchases, the
dispensing unit is expected to cost approximately $18,000 in annual production quantities of 250.
With the HFA manufacturer’s 15% markup applied, the sale price of the dispenser is $20,700.
9 TOTAL COST OF SMR- AND ATR-BASED STATIONARY FUELING

APPLIANCES
9.1 Capital Cost Summary
9.1.1 Miscellaneous Capital Costs
Capital costs that are not specific to a particular subsystem are listed below. Labor rates were
assumed to be $60/hour.
Freight: A quoted price for cross-country truck shipping is $0.40/ton-mile. The average
shipping cost per HFA was calculated by assuming a 1500-mile total trip length for an
approximate weight of 3 tons.
Taxes and Insurance: Typically taxes and insurance are a few percent of the cost of the
equipment. We have assumed a taxes and insurance rate of 2% of installed equipment.
145
Personal communication with Andy Perreault, Kraus Group Inc., March 2001.
146
Personal communication with Tom Sewall, Tulsa Gas Technologies, Inc. March 2002.
109
Spares: An allowance for an initial set of spare parts is typically included in the capital cost. We
have assumed that the spares are (as a percentage of the installed equipment cost) 10% for
compressors and 2% for process and storage equipment.
Installation: The time to set the HFA system on its foundation and to make connections to
utilities was estimated to require two days for two laborers. We have assumed that the utility
connections are easily accessible and require only a short pipe run or length of wire to complete
the installation.
Contingency: A 10% contingency was applied to the final capital cost estimate to account for
incidental expenses and unforeseen costs. This level of contingency is appropriate based on the
typical results from a DFMA analysis.
The following are not explicitly included in the capital cost:
1) Land, which will be site-specific.

2) Long pipelines or wire runs to connect to utilities.
3) Working capital, since the value of the hydrogen inventory at any time is relatively low
(<$1,000) and there is no storage of raw materials.
4) Freeze protection for the deionized water system which may be required in some
climates.
9.1.2 Total Capital Cost
The total capital cost and the contributions of each equipment category to the total are listed in
Figure 9-1.
SMR/PSA ATR/PSA SMR/Membrane ATR/Membrane

Total $253,014 $228,957 $273,359 $247,186
Reformer $123,545 $102,384 $137,036 $115,499
System147
Hydrogen $20,780 $20,780 $25,424 $25,424
Compressor
Hydrogen Storage $43,583 $43,583 $43,583 $43,583
Dispenser $20,700 $20,700 $20,700 $20,700
Miscellaneous148 $21,404 $20,695 $21,765 $19,507
10% Contingency $23,001 $20,814 $24,851 $22,471
$/kW $2,279 $2,063 $2,463 $2,227
Figure 9-1. HFA Costs for a 115 kg H2/day System149
The combination of the reformer and the hydrogen purification is the largest single contributor to
the capital cost. The combined cost of hydrogen compressor, storage, and dispenser, however, is
nearly identical to or greater than the cost of the rest of the system. Both the reforming system
147
The reformer system includes the natural gas compressor, the reformer and all associated processing, and the hydrogen purification system.
148
“Miscellaneous” includes on-site installation, foundation and building, taxes and insurance, spares, and freight.
149
Costs include a 31% vendor markup, except for 15% markup on pass-through items.
110
and the compression/storage/dispensing system deserve equal attention as targets for cost
reduction efforts.
The capital costs are consistent with previous one-of-a-kind SMR estimates150 when a progress
ratio of 86% is applied for higher-volume HFA manufacturing.
9.2 Hydrogen Cost
Total hydrogen costs are calculated to facilitate comparison of the different options with varying
capital and operating costs. The cost of hydrogen depends to a significant degree on the
assumptions, and this analysis used assumptions that are typical for a large corporation
evaluating a new chemical process installation.
9.2.1 Method for Comparison of Cases
A discounted cash flow (DCF) analysis was used to compare reforming options with differing
initial investments and operating expenses. The cost of hydrogen was determined by calculating
the cost of hydrogen that results in a net-present value (NPV) of $0 for the HFA over ten years of
operation. Solving for the cost of hydrogen at $0 NPV yields the “wholesale price” at which the
reformer operator realizes an after-tax return on investment equal to the cost of capital (i.e., the
discount rate). The cost of hydrogen calculated in this analysis is somewhere between the
wholesale and retail level. We have included the capital cost for storage and dispensing, which
would not typically be reflected in the wholesale costs of other fuels but have excluded retail
markup and profit for the HFA operator. Assumptions used for the DCF calculations are listed
in Figure 9-2.
Discount rate 10%

Inflation rate 3.2%151
Depreciation schedule MACRS152
Marginal tax rate 38%
Capacity factor 69%
Availability factor 98%
Figure 9-2. Discounted Cash Flow Calculation Assumptions
Discount Rate
The discount rate (or cost of capital) will depend on how a company finances construction and
installation of the reformers. Small startup-type companies rely heavily on equity financing and
may have a cost of capital greater than 10%. Large corporations have access to a mix of debt
150
DOE Hydrogen Infrastructure Report, 1997.
151
Bureau of Labor Statistics average for 1982-2001, obtained from BLS website.
152
The Modified Accelerated Cost Recovery System (MACRS) is the IRS depreciation schedule that applies to chemical processing equipment.
MACRS allows for annual depreciation of the following percentages of the capital investment over 6 years: 20.0%, 32.0%, 19.2%, 11.5%,
11.5%, and 5.8%.
111
(bond) and equity financing, and their cost of capital could be less than 10%. Corporations
sometimes include a risk factor in the cost of capital, which raises the effective discount rate.
We have assumed a 10% discount rate to be consistent with previous studies of hydrogen
infrastructure153 and common engineering practice.
Tax Rate
A corporation with more than $335,000 in taxable income will pay an effective marginal federal
tax rate of at least 34%. The 38% marginal tax rate used for the hydrogen cost calculations
assumes that the state and local taxes add on average an additional 4%. The tax rate has a
moderate effect on the cost of hydrogen needed to generate a positive return on investment.
Complete elimination of the income tax decreases the cost of hydrogen needed to generate a 10%
return by $0.17/kg, or roughly 5%.
Capacity and Availability Factors
The capacity factor is the fraction of rated process throughput that is used on an average basis. A
capacity factor of 69% was determined from previous studies to provide the required flexibility
between peak and slack demand154.
The availability factor is the fraction of hours in a year that the HFA is operational. The HFA
cannot afford to be out of operation for extended periods of time because the sparse distribution
of stations limits the refueling choices for FCV operators. We have assumed that routine
maintenance tasks can be completed during frequent 8-hour shutdowns rather than less frequent
multi-day shutdowns. An availability factor of 98% allows for 22 such shutdowns annually. We
reduced the availability factor to 95% for the year in which the catalyst change outs take place
due to the extensive work required for the procedure.
Years of Analysis
An eleven-year period is assumed for the cost analysis. All capital expense takes place in year
one, and the reformer is assumed to have a ten-year operational lifetime. The zinc oxide
absorbent and deionization resin are replaced every six months, and the reforming catalyst is
replaced after five years. The life of the reformer could be extended by an equipment overhaul at
the end of ten years. The cost of hydrogen decreases with increasing number of years of
operation. Because we are calculating the cost of hydrogen to result in $0 NPV in the final year
of operation, increasing the number of years divides the capital recovery into a smaller annual
amount.
153
154
112
9.2.2 Operating Cost Assumptions
Catalyst Life and Replacement
The zinc oxide sulfur absorption system is designed for one year of operation before change out.
Each parallel bed provides six months of service, and opening and closing two valves can switch
flow to the other bed.
We assumed a five-year lifetime for the reformer and shift catalysts. The conceptual catalyst
replacement method would use a central repair shop, with ten spares of each reactor. The cost
for purchasing the spares is amortized over the 250 stations and is included in the initial capital
costs (see Section 9.1.1). The cost for changing the catalysts includes 1) freight for shipping
replacement reactors to the site and the spent reactors to the shop, 2) labor for removing the
reactors from the station, opening the reactors, changing the catalysts, and re-welding the
reactors, and 3) the cost of a replacement catalyst charge.
Utilities
Natural Gas
The price of natural gas is a significant factor in determining the cost of hydrogen, but natural
gas costs vary by state, by season, and by market conditions of other related energy sources. We
assumed that the natural gas supplied to the reforming stations would fall into the “commercial
use” category and would remain at the historical average cost. The monthly average commercial
natural gas price in the U.S. for the period 1983-2001 was $5.34 per thousand standard cubic feet
(scf), (see Figure 9-3).155
155
DOE Energy Information Administration website, http://www.eia.doe.gov/oil_gas/natural_gas/info_glance/natural_gas.html
113
$10
$8
Natural Gas Cost ($/1000 scf)
$6 Commercial customers
Average = $5.34
$4
$2
(Monthly averages since Oct. 1983)

$0
Oct-83 Oct-85 Oct-87 Oct-89 Oct-91 Oct-93 Oct-95 Oct-97 Oct-99 Oct-01
Date
Figure 9-3. Natural Gas Cost to Commercial Customers, 1983-2001.
If in a fully developed hydrogen market the reformer operators could arrange volume purchase
contracts for natural gas, the cost of gas could fall into the range of the “industrial-use” category,
with an average purchase price of $3.30/thousand scf from 1983-2001. Reducing the natural gas
cost to the industrial level results in a reduction in hydrogen cost of 11% (36¢/kg) for the 10 atm
SMR and 12% (44¢/kg) for the 10 atm ATR.
Electricity
The average cost of electricity purchased by commercial customers in the U.S. for the period
1990-2000 was 7.5¢/kW-hr156. It is possible that purchasing some electricity at off-peak rates,
especially during the night when other demand for electricity is low, could reduce the average
cost of electricity.
As with natural gas, moving into the “industrial” category for electricity would lower costs. The
average cost to industrial customers for the period 1990-2000 was 4.65¢/kW-hr. Decreasing the
electricity cost to the industrial level reduces the hydrogen cost by 3% (9¢/kg) for the 10 atm
SMR and 4% (15¢/kg) for the 10 atm ATR.
Water
The cost of city water is negligible compared with the other operating costs and is excluded from
this analysis, although the cost of deionizing the water is included. Deionization bed
156
DOE Energy Information Administration website, http://www.eia.doe.gov/fuelelectric.html
114
replacement frequency would be greater in locations with poor water quality, including high
hardness or salt content.
Startup and Shutdown Accessories
Prior to startup and before shutdown the reformer system is flushed with nitrogen gas to
minimize the chance of forming an explosive mixture of natural gas and air. The nitrogen is
supplied from two high-pressure cylinders that are located next to the deionization bed on the
reformer platform, and the cost of nitrogen is included in the general Operation and Maintenance
budget. The nitrogen cylinders are installed so that they can be automatically opened to the
process in the event of an emergency shutdown. Emergency shutdowns would be activated by
the control system according to a set of preprogrammed conditions in the control logic.
Operating Labor
We assumed that the reformer would operate unattended except for maintenance and emergency
repairs. The process control computer would on a routine basis electronically transfer operating
data to a central monitoring station that has responsibility for multiple reformers.
There may initially be a need to operate the hydrogen fuel stations in a “full-serve” mode until
the public is sufficiently comfortable with the dispenser operation. Costs for a station attendant
are not included in the operating costs and would be accounted for in the retail markup. If
hydrogen fuel stations were co-located with existing gasoline stations, service stations, or
convenience stores, the cost of the attendant would be spread over the entire operation.
Station Owner Corporate Overhead
Markup to cover corporate overhead expenses and additional profit is not included in the
operating cost. (This is separate from the reformer manufacturer overhead, which is included in
the equipment markup rates.) Typically these costs would include sales and marketing expenses,
corporate staff salaries, research and development expenses, insurance, and other miscellaneous
expenses. The overhead rate will depend significantly on what type and size of company is
operating the reformer stations. Small businesses have a relatively low overhead charge, and
large multinational corporations can have overhead expenses in excess of 25% of revenues.
9.2.3 Results
The costs of hydrogen to provide a 10% after-tax return to the HFA operator for each of the four
options detailed above are listed in Figure 9-4. Supporting calculations for each option are
included as Figure 9-5 through Figure 9-8. The contributions of each of the operating cost
categories are included to highlight the most important drivers of the hydrogen cost. The costs
do not include any profit or markup beyond returning the cost of capital.
115
Costs in $/kg H2 SMR/PSA ATR/PSA SMR/Membrane ATR/Membrane

Hydrogen Cost157 $3.38 $3.59 $3.74 $4.28
Capital Recovery 1.66 1.50 1.78 1.62
Natural Gas 0.95 1.17 1.01 1.44
Electricity 0.23 0.41 0.37 0.68
O&M 0.33 0.31 0.33 0.33
Taxes & Insurance 0.23 0.20 0.24 0.22
Figure 9-4. Cost of Hydrogen from Four HFA Options
The cost of hydrogen from an SMR is lower than from an ATR over a ten-year lifetime, but the
lower capital cost of the ATR makes it more favorable in the initial years of operation. The costs
equalize by the fourth year of operation, and the gap between SMR and ATR widens each year
thereafter. Any of the options we considered would effectively lose money if operated for less
than ten years at the costs listed in Figure 9-4, but the ATR-PSA system would be the most
economical if the station were to stay in operation for four years or less.
SMR-PSA (10 atm) Overall System Economics
Basic Assumptions Results

Discount Rate 10.0% Hydrogen Price $3.38 /kg
Inflation Rate 3.2% NPV $0
Marginal Tax Rate 38.0% Payback Period 6 years, nondiscounted
Natural Gas Cost $5.34 /thousand scf
Electricity Cost $0.075 /kW-hr *Inflation applied to O&M only.
Utilization Factor 0.69 *Additional downtime is assumed for the catalyst changeout year.
Availability Factor 0.98 *Natural gas and electricity use and hydrogen production reduced by utilization and availability factors
Natural Gas LHV 927 Btu/scf
Year
Investments Basis 1 2 3 4 5 6 7 8 9 10 11
Initial capital $253,014
Expenses
Natural gas cost $26,867 $26,867 $26,867 $26,867 $26,044 $26,867 $26,867 $26,867 $26,867 $26,867
Electricity cost $6,522 $6,522 $6,522 $6,522 $6,322 $6,522 $6,522 $6,522 $6,522 $6,522
HDS/ZnO bed replacement $1,000 $1,032 $1,065 $1,099 $1,134 $1,171 $1,208 $1,247 $1,287 $1,328 $1,370
Catalyst Replacement $13,423
DI bed replacement $1,500 $1,548 $1,598 $1,649 $1,701 $1,756 $1,812 $1,870 $1,930 $1,992 $2,055
Operating & Maintenance $4,210 $4,345 $4,484 $4,627 $4,775 $4,928 $5,086 $5,249 $5,417 $5,590 $5,769
Insurance $2,530 $2,530 $2,530 $2,530 $2,530 $2,530 $2,530 $2,530 $2,530 $2,530
Property Taxes $3,795 $3,795 $3,795 $3,795 $3,795 $3,795 $3,795 $3,795 $3,795 $3,795
Revenues/Cost Savings
Hydrogen Production (kg/year) 28,418 28,418 28,418 28,418 27,548 28,418 28,418 28,418 28,418 28,418
Hydrogen Sales $95,981 $95,981 $95,981 $95,981 $93,042 $95,981 $95,981 $95,981 $95,981 $95,981
Pre-tax cash flow -$253,014 $49,342 $49,121 $48,892 $48,656 $33,073 $48,161 $47,902 $47,634 $47,358 $47,073
Depreciation $0 $50,603 $80,964 $48,579 $29,147 $29,147 $14,574 $0 $0 $0 $0
Income tax $0 -$479 -$12,101 $119 $7,413 $1,492 $12,763 $18,203 $18,101 $17,996 $17,888
After-tax cash flow -$253,014 $49,821 $61,221 $48,773 $41,243 $31,581 $35,398 $29,699 $29,533 $29,362 $29,185
Cumulative Net Cash Flow -$253,014 -$203,193 -$141,971 -$93,199 -$51,956 -$20,374 $15,023 $44,722 $74,255 $103,617 $132,802
Discounted Cash Flow -$253,014 $45,292 $50,596 $36,644 $28,169 $19,610 $19,981 $15,240 $13,777 $12,452 $11,252
Cumulative DCF -$253,014 -$207,722 -$157,126 -$120,482 -$92,313 -$72,703 -$52,722 -$37,482 -$23,704 -$11,252 $0
Calculated Values
Inflation multiplier 1.00 1.03 1.07 1.10 1.13 1.17 1.21 1.25 1.29 1.33 1.37
Natural gas price ($/million Btu) $5.76 $5.76 $5.76 $5.76 $5.76 $5.76 $5.76 $5.76 $5.76 $5.76 $5.76
Natural gas feed (kgmole/hr) 1.018 0.000 0.703 0.703 0.703 0.703 0.703 0.703 0.703 0.703 0.703 0.703
Hydrogen sales price ($/kg) $3.38 $3.38 $3.38 $3.38 $3.38 $3.38 $3.38 $3.38 $3.38 $3.38 $3.38
Electricity usage (kW) 14.68 0.00 10.13 10.13 10.13 10.13 10.13 10.13 10.13 10.13 10.13 10.13
Electricity cost ($/kW-hr) $0.075 $0.075 $0.075 $0.075 $0.075 $0.075 $0.075 $0.075 $0.075 $0.075 $0.075
Depreciation schedule 0.0% 20.0% 32.0% 19.2% 11.5% 11.5% 5.8% 0.0% 0.0% 0.0% 0.0%
Annual hours of operation 0 8585 8585 8585 8585 8322 8585 8585 8585 8585 8585
Figure 9-5. SMR-PSA Cost of Hydrogen
157
116
ATR-PSA (10 atm) Overall System Economics

Inflation Rate 3.2% NPV ($0)
Year
Expenses
Natural gas cost $33,339 $33,339 $33,339 $33,339 $32,319 $33,339 $33,339 $33,339 $33,339 $33,339
Insurance $2,290 $2,290 $2,290 $2,290 $2,290 $2,290 $2,290 $2,290 $2,290 $2,290
Property Taxes $3,434 $3,434 $3,434 $3,434 $3,434 $3,434 $3,434 $3,434 $3,434 $3,434
Hydrogen Sales $102,121 $102,121 $102,121 $102,121 $98,994 $102,121 $102,121 $102,121 $102,121 $102,121
Pre-tax cash flow -$228,957 $44,898 $44,691 $44,478 $44,259 $27,940 $43,798 $43,556 $43,307 $43,050 $42,784
Depreciation $0 $45,791 $73,266 $43,960 $26,376 $26,376 $13,188 $0 $0 $0 $0
Tax $0 -$340 -$10,858 $197 $6,795 $594 $11,632 $16,551 $16,457 $16,359 $16,258
Cumulative DCF -$228,957 -$187,832 -$141,923 -$108,654 -$83,066 -$66,087 -$47,930 -$34,072 -$21,546 -$10,227 $0
Calculated Values
Figure 9-6. ATR-PSA Cost of Hydrogen
SMR-Membrane (20 atm) Overall System Economics

Year
Expenses
Natural gas cost $28,706 $28,706 $28,706 $28,706 $27,827 $28,706 $28,706 $28,706 $28,706 $28,706
Insurance $2,734 $2,734 $2,734 $2,734 $2,734 $2,734 $2,734 $2,734 $2,734 $2,734
Property Taxes $4,100 $4,100 $4,100 $4,100 $4,100 $4,100 $4,100 $4,100 $4,100 $4,100
Hydrogen Sales $106,409 $106,409 $106,409 $106,409 $103,151 $106,409 $106,409 $106,409 $106,409 $106,409
Pre-tax cash flow -$273,359 $53,029 $52,792 $52,547 $52,294 $38,155 $51,764 $51,487 $51,200 $50,904 $50,599
Depreciation $0 $54,672 $87,475 $52,485 $31,491 $31,491 $15,745 $0 $0 $0 $0
Tax $0 -$624 -$13,180 $23 $7,905 $2,532 $13,687 $19,565 $19,456 $19,344 $19,228
Cumulative DCF -$273,359 -$224,583 -$170,062 -$130,600 -$100,282 -$78,163 -$56,670 -$40,289 -$25,480 -$12,095 $0
Calculated Values

Figure 9-7. SMR-Membrane Cost of Hydrogen
117
ATR-Membrane (20 atm) Overall System Economics

Year
Expenses
Natural gas cost $40,798 $40,798 $40,798 $40,798 $39,549 $40,798 $40,798 $40,798 $40,798 $40,798
Insurance $2,472 $2,472 $2,472 $2,472 $2,472 $2,472 $2,472 $2,472 $2,472 $2,472
Property Taxes $3,708 $3,708 $3,708 $3,708 $3,708 $3,708 $3,708 $3,708 $3,708 $3,708
Hydrogen Sales $121,501 $121,501 $121,501 $121,501 $117,781 $121,501 $121,501 $121,501 $121,501 $121,501
Pre-tax cash flow -$247,186 $48,350 $48,129 $47,900 $47,665 $31,196 $47,171 $46,912 $46,645 $46,369 $46,085
Depreciation $0 $49,437 $79,100 $47,460 $28,476 $28,476 $14,238 $0 $0 $0 $0
Tax $0 -$413 -$11,769 $167 $7,292 $1,034 $12,515 $17,827 $17,725 $17,620 $17,512
Cumulative DCF -$247,186 -$202,856 -$153,354 -$117,492 -$89,916 -$71,188 -$51,625 -$36,700 -$23,208 -$11,016 $0
Calculated Values
Figure 9-8. ATR-Membrane Cost of Hydrogen
Comparison with Previous Cost Estimates
A comparison of hydrogen costs with previous studies is instructive to demonstrate the influence
of assumptions on the final result. The 1997 Hydrogen Infrastructure Report examined a range
of options for early-generation HFA’s.158 The cost of hydrogen from a factory-built, on-site,
low-pressure SMR with PSA gas cleanup was estimated to be ~$2/kg. The capital cost for the
factory-built HFA was a factored estimate and was not subjected to the same level of DFMA
analysis as the systems analyzed in this study, so it was less certain than the current estimates.
Hydrogen storage costs also have been updated since 1997 and may alone add up to 5%
additional cost per kg of hydrogen.
Applying the economic parameters, the cost of natural gas, the cost of electricity, and the 15-year
lifetime from the previous study to the SMR/PSA HFA from this report results in a hydrogen
cost of $3.04/kg, or 10% less than our estimate. This comparison highlights how dependent the
cost of hydrogen is on the set of economic assumptions and how from a given capital cost and
system efficiency different parties could arrive at disparate conclusions.
Any of the options considered in this study is favorable compared to liquid hydrogen delivered
by truck. The cost of liquid hydrogen delivered to HFA’s serving 183 cars has been estimated at
$6/kg. Liquid hydrogen becomes more competitive as the stations become larger but is far more
expensive than any of the small-scale HFA’s for the introductory period.
158
118
Comparison with Cost of Gasoline
Hydrogen cost relative to the cost of gasoline on a per-mile basis is the critical factor in
acceptance of hydrogen as a vehicle fuel. Since the costs from this report represent the early
years of the hydrogen infrastructure, a somewhat higher cost for hydrogen may be tolerated than
would be acceptable long-term. The fuel economy for a gasoline internal combustion engine
vehicle (ICEV) is assumed to be 36 miles per gallon. The assumed energy efficiency of
hydrogen FCV’s is 80 miles per gallon equivalent, which results in an efficiency factor of 2.2.
The wholesale cost of hydrogen for the SMR-PSA option is then $1.55 per gallon equivalent,
normalized for energy content of each fuel and adjusted for efficiency differences. This
hydrogen cost is only modestly higher than recent retail gasoline prices. The average standard
regular grade retail price of gasoline from 1995 to the present has been $1.20/gallon and for
reformulated regular grade $1.28/gallon.159 During the years 2000 and 2001 there were several
price spikes above $1.50/gallon for standard regular grade. In the regions where gasoline prices
are significantly higher than the averages, the cost of hydrogen could be competitive as long as
there is an efficiency benefit for FCV’s over ICEV’s.
Gasoline-powered hybrid electric vehicles are more efficient than conventional ICEV’s and
reduce the advantage of hydrogen-powered FCV’s. A best-case scenario for gasoline hybrids
would reduce the FCV efficiency advantage to 1.5. The resulting cost of hydrogen would be
$2.29/gallon equivalent.
10 CONCLUSIONS
There are two general conclusions that can be taken from this analysis:
1. Steam methane reforming is more efficient than autothermal reforming, and the
efficiency benefit results in a lower cost of hydrogen over a ten-year system lifetime even
with a slightly higher initial capital cost ($253,014 SMR vs. $228,957 ATR). The
difference between the SMR and ATR costs of hydrogen shrinks as the cost of the natural
gas feedstock decreases, but only with zero-cost natural gas (i.e. free) does the ATR
match the SMR. For a given cost of hydrogen the SMR and ATR economic returns are
equal by the fourth year of operation, with the SMR advantage increasing every year
thereafter.
2. Pressure-swing absorption (PSA) is a more economical and reliable option than any other
hydrogen cleanup system at this time. The SMR and ATR systems with PSA hydrogen
purification do not rely on any technologies that are not currently available. Significant
reductions in the cost and reliability of membrane purification systems are required to
make them competitive with PSA.
The most economical source of hydrogen for first-generation FCV’s as judged by the wholesale
cost of hydrogen is steam methane reforming with pressure swing adsorption hydrogen
purification (SMR-PSA). The annual investment to add 250 new HFA’s of the favored SMR-
PSA design would be approximately $63.25 million. The wholesale cost of hydrogen from the
SMR-PSA HFA is $3.38/kg, which is the equivalent of $1.56/gallon of gasoline for an ICEV
159
DOE Energy Information Administration website, http://www.eia.doe.gov/
119
after adjustment for the higher efficiency of the fuel cell engine. This equivalent gasoline cost is
at the upper end of current retail gasoline costs160. When there are sufficient FCV’s to justify a
larger number of higher-volume stations, the cost of hydrogen should decrease by taking
advantage of economies of scale in both HFA manufacture and the reforming process.
160
Of course this comparison compares the taxed price of gasoline to the untaxed equivalent price of hydrogen. But hydrogen is not expected to
be taxed in the early/mid years of its introduction as a motor fuel.
120
APPENDIX: Cost Estimate for Scale-up of Small-Scale HFA

The HFA’s analyzed in the main sections of this report are sized to service 183 FCV’s (2000
scf/h), an appropriate number for the early years of FCV introduction when there are relatively
few vehicles within geographic range of the station. For comparison, typical high-volume
gasoline stations service 6-10 times as many vehicles. An obvious question is, what would be
the capital cost and cost of hydrogen for an HFA of comparable size to a gasoline station? In
this appendix we have estimated the costs for a 1464-supported vehicle HFA (~16,000 scf/h)
with the SMR-PSA option by scaling-up the costs from the 183-supported vehicle design.
A.1 Capital Cost
The cost for the larger HFA was not estimated by the same detailed method described in Sections
2 and 3, but instead by applying scaling factors to the components of the smaller HFA. As a
result, the cost for the scaled-up HFA is subject to larger uncertainties. In fact, it is not at all
certain that the reformer design for the 183-supported vehicle system would be chosen for the
1464-supported vehicle HFA.
Most of the HFA subsystem costs were scaled up using scale factors common in the chemical
process industries. These scale factors reflect the affects of increased capacity only, not
increased production rate. Thus the resulting scaled capital costs are consistent with a production
rate of 250 HFA’s per year and additional cost reductions would presumably result if production
rate were further increased. The cost scale factor was calculated by the equation
Scale exponent
 Capacity , scaledup 
Cost , scaledup = (Cost , original ) x 
 Capacity, original 
The scale exponents used for each subsystem are listed in Figure A-1.
Scaleup Cost Multiplier

Exponent for 8x Capacity
Compressors 0.75 4.76
HDS 0.60 3.48
Reformer 0.60 3.48
Shift Reactor 0.60 3.48
Boiler 0.60 3.48
Condenser 0.60 3.48
PSA 0.95 7.21
Storage 0.95 7.21
Dispenser 0.50 2.83
Structural supports and building 0.70 4.29
Freight 0.70 4.29
Installation 0.70 4.29
Instruments -- 2.00
Figure A-1. Scaleup Factors Applied to SMR-PSA System for 8-Times Capacity.
121
The scale exponents for compressors (natural gas and hydrogen) and heat exchangers (sulfur
removal, reformer, shift reactor, boiler, and condenser) were taken from the text Plant Design
and Economics for Chemical Engineers.161 The PSA and hydrogen storage designs do not
significantly benefit from increases in scale, so we have assumed the costs to increase nearly
linearly with HFA capacity. Because the building size, installation, and total HFA weight (for
freight purposes) are determined by a combination of all components, we assumed a scale factor
between the extremes for these items. The dispenser costs are not linear with scale since
multiple fueling ports can be linked to a common dispensing unit and the vehicle turnover
frequency can be increased. The instrument needs will not increase significantly with scaleup,
but the connections to process lines and vessels will generally be larger. We assumed a simple
doubling of instrument cost for the 8x HFA.
Since the SMR-PSA HFA was determined to be the optimal reforming method, the scaling
procedure was applied to this design. The total capital cost estimated using these scaleup factors
for a 16,000 scf/h HFA is $1.16 million (compared to $253,014 for a 2,000 scf/h HFA).
A.2 Hydrogen Cost
Despite the uncertainties in the capital cost, it is instructive to demonstrate how economies of
scale may result in a significant reduction in the cost of hydrogen when the number of FCV’s is
high enough to justify larger refueling stations.
The hydrogen wholesale cost, including storage and dispensing, was calculated using the same
discounted cash flow (DCF) method described in Section 9.2. If the same assumptions are used
for the 1464-supported vehicle HFA as were used for the original HFA, the cost of hydrogen is
$2.48/kg. Since our assumptions tend to be conservative compared with previous hydrogen cost
estimates, we show in Figure A-2 how progressively more optimistic assumptions decrease the
cost of hydrogen.
Wholesale
Hydrogen Cost
1. 183-FCV HFA (baseline from report) $3.38/kg
2. 1464-FCV HFA with same assumptions as #1 $2.48/kg
3. #2 with “industrial” natural gas and electricity costs $2.05/kg
4. #3 with 15 year lifetime $1.87/kg
Figure A-2. Estimated Cost of Hydrogen for 1464-Vehicle HFA
The (untaxed) gasoline cost equivalent to $1.87/kg of hydrogen is $0.85/gallon, assuming a FCV
fuel efficiency of 80 mpge vs. 36 mpge for an ICEV. At this level, the cost of hydrogen is at
least competitive with the untaxed cost of gasoline on a per-mile-driven basis and may be less
than the cost of gasoline in many locations.
The “industrial” utility assumptions and 15-year lifetime are feasible considering the scale of the
HFA, which is more like a small chemical plant than a commercial business. The combined
161
Peters, Max S. and Klaus D. Timmerhaus. Plant Design and Economics, 4th ed., McGraw-Hill, New York, 1991.
122
natural gas and electric cost for HFA operation at the historical industrial prices would be
approximately $180,000 per year. Fundamentally, there is no reason why an HFA could not last
for 15 years, although additional investments or maintenance costs not included in this analysis
may be necessary to extend the lifetime beyond 10 years. Presumably the larger 1464-supported
vehicle HFA’s will be operated for a longer lifetime than the 183-supported vehicle HFA’s since
the larger units represent the scale required for a mature FCV infrastructure as opposed to an
introductory phase.
123

Cost and Performance For A Fueling Hydrogen Station

Uploaded by

Document Information

Original Description:

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

Cost and Performance For A Fueling Hydrogen Station

Uploaded by

Copyright:

Available Formats

Cost and Performance Comparison Of

Stationary Hydrogen Fueling Appliances

One Virginia Square

The Hydrogen Program Office

5.2.1 Metal Membrane Gas Separation ........................................................................ 50

Costs in $/kg H2 SMR/PSA ATR/PSA SMR/Membrane ATR/Membrane

Costs in $/kg H2 16,000scfh SMR/PSA HFA

2.2 Production Volume Considerations

2.3 Markup Rate Assumptions

General & Administrative (G&A) Expenses $2,297,500

Management Staff Direct Labor (11 staff) $945,000

Overhead (OH) Expenses $1,908,000

Research And Development (R&D) Budget $1,500,000

Category Markup up on Markup on Pass

2.4 Design Considerations

2.5 Sample DFMA Cost Analysis

Machining Markup Factors

Total Cost pre-markup $ $66.65

Figure 2-3. Detailed Cost Analysis of a Boiler Tube-Sheet

Labor Rate $/min $1.75 additional 75cent/minute to account for assistant

Operation # repeats time, sec Machine Op.? Cost/Op.

Markup Rate 30.00%

Total Cost pre-Markup $454.58

Total $4,132.02 $5,412.95

3 REFUELING APPLIANCE HYDROGEN PRODUCTION RATE AND

Single HFA Service Six HFA Service

4 STEAM METHANE REFORMING VS. AUTOTHERMAL REFORMING

Thermal efficiency is a measure of the performance of a reforming process and is defined as

1. Natural gas compression

4.1.1 Natural Gas Compression

4.1.2 Natural Gas Purification

Hot Reformate Manual

Alumina 4" Pipe

Figure 4-3. HDS Pipe Section (not to scale)

4.1.3 Catalytic Steam Reforming

CH 4 + H 2O = CO + 3H 2 ∆Hrxn = 241 kJ/gmole (Equation 4-3)

0.5CH 4 + O 2 = CO + H 2O (Equation 4-4)

A substoichiometric volume of oxygen is fed to the ATR to encourage partial combustion

4.1.4 Water-Gas Shift Reaction

CO + H 2O = CO2 + H 2 ∆Hrxn = -90 kJ/gmole (Equation 4-6)

4.2 SMR System Flowsheets, Cost and Performance

Figure 4-5. SMR Flowsheet

4.2.1 Reactor and Vessel Sizing

Natural Gas Pretreatment

Figure 4-6. HDS Preheater, with Tube-in-Tube Design (not to scale)

NG & Steam Reformate

Reactor Tube Baffle Tube Sheet

Figure 4-8. Tubes and Tube-Sheets for Shell-and-Tube Vessel

The 10-atmosphere steam-reforming reactor requires 16.7 L of catalyst to approach the

4.2.2 Heat Exchanger Sizing

Water Knockout Vessel

4.2.3 Water Purification

4.2.4 Other Equipment

Fans and Blowers

Figure 4-9. Proposed Layout of 10 atm SMR Reformer System

4.2.6 SMR Reformer System Bill of Materials and Cost

4.2.7 SMR Performance and Efficiency

Total Sub System Cost Total Sub System Cost

Natural Gas Feed

Flanges SS 316 4 $107.71 $430.82 31.00% $564.37

Figure 4-10. Bill of Materials for 10 atm SMR RS

Total Sub System Cost Total Sub System Cost