Palladium/Alloy-based Catalytic Membrane Reactor Technology Options for Hydrogen Production: A Techno-Economic Performance Assessment Study
Worcester Polytechnic Institute
Liang-Chih Ma
Ph.D. 2015
i
For my parents
ii
Acknowledgements
I wish to express my sincere gratitude to my advisors, Professor Nikolaos K. Kazantzis and Professor Yi Hua Ma, for their assistance, encouragement, gentle pushing, and support with their immense knowledge and infinite patience over the years. Their guidance and expertise helped me in all the time of research study and writing of my PhD dissertation, greatly influencing my thoughts in the way of creativity and versatility. More importantly, their hard working attitude and unlimited passion for research also affect my philosophy, which is the key to establish the brilliant milestone of chemical engineering education.
I would like to thank the members of my dissertation committee: Professor Bernardo Castro-Dominguez, Professor Michael T. Timko and Professor Huong N. Higgins, for their inspiring guidance, insightful comments and encouragement. In particular, Professor Castro- Dominguez not only gives me constant emotional support and helpful suggestions for my research study, but also teaches me how to work effectively and smart. His assistance is invaluable in accomplishing my dissertation. I would also like to thank Professor Ivan P. Mardilovich and Dr. Federico Guazzone for their emotional support and encouragement.
My sincere thank also goes to all of my colleagues in the Center for Inorganic Membrane Studies (CIMS), including Dr. Jacopo Catalano, Dr. Chao-Huang Chen, Dr. Reyyan Koc, Dr. Alexander S. Augustine, Pei-Shan Yen and Rui Ma. Sharing their expertise, experience and friendship makes my PhD life more interesting and enjoyable. I am grateful to the faculty of Department of Chemical Engineering at WPI and the administrative assistants Felicia Vidito, Tiffany Royal and Paula Moravek. The technical help and support provided by Doug White and Jack Ferraro is also acknowledged.
I would like to give my appreciation to a number of friends who helped to keep it all in perspective and made my PhD life more interesting and colorful, in particular Liuxi Chen, Xue (Sarah) Zhang, Minchao (Matt) Yin and Patrick D. O’Malley.
The financial support provided by the U.S. Department of Energy through Grant No. DE- FE0004895 as well as WPI are gratefully acknowledged.
i
Acknowledgements
Last but not least, I would like to appreciate my family: my parents, elder brother and elder sister, for supporting me spiritually throughout my life. You are always there when I need you. I love you all.
ii
Abstract
Hydrogen (H2) represents an energy carrier endowed with the potential to contribute to the design of a robust and reliable global energy system by complementing electricity as well as liquid fuels use in an environmentally responsible manner provided that the pertinent H2 production technologies (conventional and new ones) can reach techno-economically attractive performance levels in the presence of irreducible
(macroeconomic, fuel market, regulatory) uncertainty. Indeed, the role of H2 in the global energy economy is widely recognized as significant in light also of fast-growing demand in the petrochemical and chemical processing sector as well as future regulatory action on greenhouse gas emissions. Pd and Pd/Alloy-based catalytic membrane reactor (CMR) modules potentially integrated into H2 production (HP-CMR) process systems offer a promising technical pathway towards H2 production with enhanced environmental performance in a carbon-constrained world. However, the lack of accumulated operating experience for HP-CMR plants on the commercial scale poses significant challenges. Therefore, any preliminary attempt to assess their economic viability is certainly justified. A comprehensive techno-economic performance assessment framework has been developed for
HP-CMRs with CO2 capture capabilities. A functional Net Present Value (NPV) model has been developed first to evaluate the economic viability of HP-CMRs. The plant/project value of HP-CMR is compared to other competing technology options such as traditional coal-gasification and methane steam reforming- based hydrogen production plants with and without CO2 capture. Sources of irreducible uncertainty (market and regulatory) as well as technology risks are explicitly recognized and the effect of these uncertainty drivers on the plant’s/project’s value is taken into account using Monte Carlo techniques. Therefore, more realistic distribution profiles of the plant’s economic performance outcomes are generated rather than
single-point value estimates. It is shown that future regulatory action on CO2 emissions could induce appealing NPV-distribution profiles for HP-CMRs in the presence of uncertainty and technology risks. Finally, a creatively structured portfolio of technology-push and/or market-pull policy incentives could lead to more attractive profiles, suggesting possible means for accelerating the realization of demonstration projects of HP-CMRs at the commercial scale. Furthermore, the proposed research work aims at the development of a systematic methodological framework to assess the economic value of flexible alternatives in the design and operation of HP-CMR plants with carbon capture capabilities under the aforementioned sources of uncertainty. The main objective is to demonstrate the potential value enhancement associated with the long-term economic performance of flexible HP-CMR project investments by managing the uncertainty associated with future environmental regulations. Within the proposed context, promising design flexibility concepts for HP-CMR plants are introduced and operational as well as constructional flexibility options are identified and assessed. In particular, operational flexibility will be realized through periodic and temporary shutdowns of the carbon capture unit in response to regulatory uncertainties. Constructional flexibility will be realized by considering the installation of a carbon capture unit at three strategic periods: 1) installation in the initial design phase, 2) retrofitting at a later stage and 3) retrofitting with preinvestment. Monte Carlo simulations and financial analysis will be conducted in order to demonstrate that, in the presence of irreducible uncertainty, design flexibility options could lead to economic performance enhancement of HP-CMR plants by actively responding to the above sources of uncertainty as they get resolved over the plant’s lifetime.
iii
Executive Summary
Palladium/alloy-based catalytic membrane reactors (CMRs) provide a promising technical
pathway towards hydrogen production (HP), in which Pd-based membranes separate H2 with extremely high efficiency via the solution-diffusion mechanism. Integration of CMR modules into
H2 production (HP-CMR) process systems generates great potential for CO2 capture, H2 purification, process intensification, as well as the advancement of clean energy and environmental policy goals. Consequently, this work aims at examining the techno-economic performance of Pd/alloy-based CMR technology for hydrogen production through a comprehensive economic performance assessment and cost analysis framework based on experimental and simulation studies. There are three major areas covered by this research study: (i) estimating the costs of CMR modules, (ii) analyzing economic performance of HP-CMR plants, and (iii) studying engineering design flexibility-options applicable for CMRs.
At first, to estimate the costs of CMR modules, a comprehensive economic performance evaluation framework for CMR modules potentially integrated into H2 production via natural gas (methane) steam reforming (MSR) process systems has been developed. In particular, the development of detailed comprehensive baseline models for the Fixed Capital Investment (FCI), Total Capital Investment (TCI), and Total Product Cost (TPC) was pursued followed by an explicit recognition of various sources of uncertainty. The effect of these uncertainty sources on FCI, TCI and TPC has been taken into account through the integration of Monte Carlo simulation methods into the aforementioned cost models. As a result, insightful distribution profiles of TCI and TPC are derived rather than single-point value estimates and more realistic distributions of CMR economic performance outcomes have been generated. The results derived have shown that FCI, TPI and TPC profiles become economically appealing by reducing palladium thickness. In particular, when Pd thickness is below the 20 µm level, the TCI and TPC become more economically competitive compared to the conventional technology option. Moreover, as the Pd unit price has by far the most significant effect of cost, the reduction in the spread of various possible economic performance outcomes has been observed when the Pd thickness is reduced. This result can be attributed to the fact that thinner membranes require lower Pd amounts to achieve the same H2 recovery level.
iv
Executive Summary
The effect of membrane thickness on economic performance was further studied through the experimental analysis of the lifetime of Pd/Au membranes. The objective of this study was to explore the tradeoff between permeance-thickness and lifetime in terms of economic performance
outcomes. Four membranes were tested at 350 and 450°C under pure H2 with occasional He leak tests for a cumulative time of 19,200 hours or an equivalent time of 2.2 years. The experimental
results of the produced H2 purity and elapsed testing times were analyzed to estimate the membranes’ lifetime at different membrane thicknesses. Please notice that the lifetime of
membranes was defined to target a H2 purity equal or greater than 99%. The experimental results showed that the membrane lifetime was extended by 16% when the Pd layer thickness was increased from 2.7 to 4.6 µm while increasing the Pd layer thickness from 2.7 to 10.4 µm leads to an increase of the membrane lifetime by 152%. Through the proposed evaluation framework, it was shown that the expected values of the FCI/TCI increase as the Pd layer thickness increases, since more quantities of Pd are needed, leading to a higher capital investment. Nonetheless, the expected value of the TPC decreases with higher Pd layer thicknesses, indicating that prolonged membrane lifetimes effectively reduce the costs of membrane replacement. Finally, it was observed that the expected value of levelized cost (LC) as well as its spread/variability decreases as the Pd thickness is reduced; this effect is attributed to higher hydrogen production levels when using thinner membranes.
An actual large-scale catalytic membrane reactor module used for water-gas shift (WGS) reaction has been also assessed within the proposed evaluation framework. The results showed that the expected value of TCI for an actual large-scale CMR module is $22,418 K/m2 which is approximately 1,200 times higher than the expected value of TCI for an industrial-scale CMR module. This result implies that the TPC profiles become more economically appealing when the capacity of the CMR module increases. Additionally, it was shown that the capacity of purchased equipment is the most critical factor in the TCI performance. This work effectively validated previous techno-economic results on CMR technology, utilizing actual data and extrapolating the economics of scale. Furthermore, a TCI learning curve for the CMR modules has been generated to demonstrate the importance of technological progress over time on the economic features of this innovative technology.
v
Executive Summary
As previously mentioned, an economic assessment of coal-fired HP-CMR plants was developed. First, a detailed comprehensive Net Present Value (NPV) model was developed in order to assess the economic viability of HP-CMR plants, while irreducible sources of market and regulatory uncertainty are identified and their effect on the plant’s economic performance is explicitly taken into account through Monte Carlo techniques. In the study, the technical performance of HP-CMR plants shows an 16% reduction in the coal feed requirements to achieve
the same H2 production level (616.5 tonne per day) by the conventional HP plants that utilize
pressure swing adsorption (PSA) as the H2 purification unit (HP-PSA plants). In addition, when HP plants encompass carbon capture and sequestration (CCS) systems, the HP-PSA plant has an efficiency of 90% while the HP-CMR plant displays an efficiency of 98%. Considering the lower
CO2 production levels and the higher CO2 capture efficiency achieved by the HP-CMR technology option, the HP-CMR plant emits 0.07 Mtonnes of CO2 per Mtonnes of coal feed compared to a
value of 0.23 for the conventional HP plant, leading to an overall reduction in CO2 emissions of 70%. Furthermore, the HP-CMR technology option enables the elimination of traditional WGS
reactors, PSA and Selexol units in conventional H2 production plants, showing a 26% reduction in TCI costs for HP. The economic performance of the HP-CMR plant was comparatively assessed against the conventional HP-PSA plant. The analysis results showed that, in the absence of any
regulatory action on CO2 emissions HP-CMR could not be perceived as an economically viable
option; however, better prospects for HP-CMR arise if future regulatory action on CO2 emissions is introduced.
Finally, the development of a systematic framework for the economic performance assessment of various flexible design options for HP-CMR plants was pursued. The implementation of flexibility options provides a dynamic framework to pro-actively deal and manage irreducible uncertainties in the fuel market and CO2 emissions regulatory landscape. Therefore, the proposed approach explicitly accounts for and incorporates into its valuation framework managerial flexibility to respond as uncertainties are resolved as well as the inherent optionality element embedded in the investment decision. In particular, operational and constructional flexibility options for HP-CMR plants were first developed and then comparatively assessed in terms of their potential economic value-enhancing properties. In light of the results obtained, operational flexibility, designed to temporarily shut down the plant when the cash flows become negative in two consecutive years and restart when positive cash flows are generated,
vi
Executive Summary
suggest a value-enhancing flexible design option in response to uncertain carbon taxation patterns. Constructional flexibility that provides the HP-CMR system with the capacity of integrating a CCS system encompasses two options: (1) Inclusion of a CCS system in the initial design phase and (2) Inclusion of a CCS system at a later stage. For the inclusion in the initial design phase, a value- enhancing capacity is realized in response to a “carbon tax penalty” as the CCS system reduces
CO2 emissions. When the CCS system was subject to an operational flexibility option, an improved NPV distribution profile was observed. For the CCS-system inclusion at a later stage, a less favorable NPV-based performance evaluation emerged when retrofitting without a preinvestment option. However, when the preinvestment option is incorporated, improved economic performance could be attained. Other factors, such as the year of introducing the CO2 tax, the expected tax
growth rate and the initial CO2 tax rate, significantly influenced the economic performance characteristics associated with the different flexibility design options presented in this work. It is suggested that the appropriate implementation of operational and constructional flexibility options can significantly improve the economic performance of HP-CMR plants.
vii
Table of Contents
Acknowledgements ...... i
Abstract…….. …………………………………………………………iii
Executive Summary ...... iv
Table of Contents ...... viii
List of Figures ………………………………………………………...xii
List of Tables ……………………………………………………….xvii
Chapter 1 Introduction ...... 1
Chapter 2 Literature Review ...... 12
2.1 Palladium Membrane ...... 12
2.2 Hydrogen Transport Mechanism in Palladium Membranes ...... 15
2.3 Palladium Alloy Membrane ...... 25
2.4 Palladium/Alloy-based Catalytic Membrane Reactor ...... 30 2.5 Application of Palladium/Alloy-based Catalytic Membrane Reactor in the Generation of Hydrogen ...... 32 2.5.1 Hydrogen from Natural Gas ...... 32 2.5.2 Hydrogen from Coal ...... 37
2.5.3 Approaches for CO2 capture ...... 42 2.5.4 Kinetics of Methane Steam Reforming ...... 47 2.5.5 Kinetics of the Water-Gas Shift Reaction ...... 52 2.5.5.1 High Temperature Shift Reaction ...... 53 2.5.5.2 Low Temperature Shift Reaction ...... 57
2.6 Engineering Design Flexibility ...... 62
Chapter 3 Techno-Economic Performance Evaluation ...... 65
3.1 Technical Performance Evaluation Framework ...... 65
viii
Table of Contents
3.2 Economic Performance Assessment and Cost Analysis Framework ...... 74 3.2.1 Capital Investment and Total Product Cost Baseline Estimation ...... 74 3.2.1.1 An Industrial-Scale Pd/Alloy-based CMR Module for WGS Integrated into Natural Gas-based Hydrogen Production Plants ...... 74 3.2.1.2 An Actual Large-Scale Pd/Alloy-based Separation Module for Hydrogen Purification ...... 81 3.2.1.3 An Actual Large-Scale Pd-based CMR Module for Water- Gas Shift Reaction ...... 85 3.2.1.4 Coal-based Hydrogen Production Plants with Integrated Industrial-Scale Pd/Alloy-based CMR Modules...... 91 3.2.2 Economic Performance Assessment: Baseline Net Present Value Model ...... 96 3.3 Economic Performance Analysis and Evaluation under Uncertainty: Integration of Monte Carlo Simulation Methods .....99
3.4 Formulation of Engineering Design Flexibility Options ...... 110
3.5 Synthesis and Characterization of Pd/Au Membranes ...... 114
Chapter 4 Natural Gas in Hydrogen Production: A Cost Study……………………………………………...... 117
4.1 Introduction ...... 117
4.2 Results and Discussion ...... 120
4.3 Conclusion ...... 132
Chapter 5 An Integrated Experimental/Techno-Economic Analysis of the Lifetime of Pd/Au Membranes ...... 133
5.1 Introduction ...... 133
5.2 Results and Discussion ...... 137 5.2.1 Hydrogen Permeation and Helium Leak Tests ...... 137
ix
Table of Contents
5.2.2 Membrane Lifetime Estimation ...... 146 5.2.3 Economic Performance Assessment of Pd-based Separation Modules with Various Pd Layer Thickness Values ...... 150
5.3 Conclusion ...... 158
Chapter 6 A Cost Assessment Study for a Large-Scale Water- Gas Shift Catalytic Membrane Reactor Module in the Presence of Uncertainty ...... 159
6.1 Introduction ...... 159
6.2 Results and Discussion ...... 163
6.3 Conclusion ...... 171
Chapter 7 Integration of Membrane Technology into Hydrogen Production Plants with CO2 Capture: An Economic Performance Assessment Study ...... 172
7.1 Introduction ...... 172
7.2 Results and Discussion ...... 174 7.2.1 Capital Investment Cost Estimation under Uncertainty: the CMR Module Case ...... 174 7.2.3 Economic Assessment of Hydrogen Production Technology Options ...... 176 7.2.3 The Role of Discount-Rate in the Valuation of HP-CMRs: Sensitivity Analysis ...... 190
7.3 Conclusion ...... 194
Chapter 8 Design Flexibility for Hydrogen Production Plants Integrated with Membrane Technology: An Economic Performance Evaluation Study ...... 195
8.1 Introduction ...... 195
8.2 Results and Discussion ...... 197
x
Table of Contents
8.2.1 Operational Flexibility for the HP-CMR Plant in the Presence
of CO2 tax ...... 198 8.2.2 Constructional Flexibility−Inclusion of a CCS System in the Initial Design Phase ...... 200 8.2.3 Constructional Flexibility−Inclusion of a CCS System at a Later Stage ...... 203 8.2.4 Sensitivity Analysis ...... 207
8.3 Conclusion ...... 220
Chapter 9 Concluding Remarks and Suggestions for Future Work ...... 223
9.1 Concluding Remarks ...... 223
9.2 Suggestions for Future Work ...... 226
Appendix………………………………………………………...... 228
A.1 One-dimensional Model for Pd/alloy-based CMR at Steady State Conditions ...... 228
A.2 Historical Palladium Unit Price from 2010 to 2015 ...... 232
A.3 Historical Gold Unit Price from 2010 to 2015 ...... 233
A.4 Historical Labor Cost per Hour from 2010 to 2015 ...... 234
A.5 Historical Coal Price from 2010 to 2015 ...... 235
A.6 Historical Inflation Rate from 2004 to 2015 ...... 236
References…………………………………………………………… 237
xi
List of Figures
Figure 1-1. Distribution of the global hydrogen market. Data adapted from Linde Engineering [6]...... 2
Figure 1-2. Hydrogen consumption in the production of ammonia from 2004 to 2013. Data adapted from U.S. Geological Survey, Mineral Commodity Summary, 2006-2005 [7,8,10,11,12,13,14,15,16]...... 3
Figure 1-3. Share of primary H2 sources and technology pathways...... 4
Figure 1-4. Share of global recoverable coal resources in 2011. Data adapted from U.S. EIA [23]...... 6
Figure 2-1. Equilibrium solubility isotherms of PdHn for the Pd-H system [58,59,60].1...... 13
Figure 2-2. Schematic diagram of solution-diffusion mechanism...... 15
Figure 2-3. The n-value as a function of selectivity for various values of r (ratio of the contribution of the Knudsen diffusion to the overall leak). The figure was taken from Guazzone et al. [70]...... 23
Figure 2-4. Relative permeability between Pd alloy and Pd with various alloying metal contents.2 ...... 26
Figure 2-5. Schematic diagram of the Pd/alloy-based CMR used for H2 production...... 31
Figure 2-6. Process block flow diagram of the conventional natural gas-based H2
production plant with CO2 capture, based on the design of the DOE/NETL report [44]...... 34
Figure 2-7. Process block flow diagram of the conventional natural gas-based H2
production plant with CO2 capture, based on the design of the DOE/NETL report [44]...... 38
Figure 2-8. Global CO2 emissions from fossil fuel combustion from 1975 to 2010 [132]...... 43
xii
List of Figures
Figure 2-9. Equilibrium constant of the WGS reaction as a function of temperature...... 52
Figure 2-10. Rate constants (k) of the LST reaction at 180°C in terms of various Cu/Zn atomic ratios [153]...... 58
Figure 3-1. Schematic diagram of the Pd/alloy-based CMR used in the one- dimensional modeling framework...... 66
Figure 3-2. Schematic of a catalytic membrane reactor (CMR) containing three Pd/Au membranes for Water-Gas Shift reaction...... 72
Figure 3-3. Process block flow diagram of the natural gas-based H2 production plant integrated with membrane technology...... 76
Figure 3-4. Schematic of the actual large-scale Pd/alloy-CMR module built at Worcester Polytechnic Institute (WPI) [30]...... 86
Figure 3-5. Process block flow diagram of the coal-based H2 production plant integrated with membrane technology...... 92
Figure 3-6. Methodological steps in the Monte Carlo simulation procedure [147,]...... 102
Figure 3-7. Depiction of the Pd/Au membranes used in this study before and after test...... 115
Figure 4-1. Distribution profiles of fixed capital investment for various palladium and gold layer thickness values; CMR, catalytic membrane reactor; Pd, palladium; Au, gold...... 121
Figure 4-2. Distribution profiles of total capital investment for various palladium and gold layer thickness values; CMR, catalytic membrane reactor; Pd, palladium; Au, gold...... 122
Figure 4-3. Distribution profiles of total product cost for various palladium and gold layer thickness values; CMR, catalytic membrane reactor; Pd, palladium; Au, gold...... 123
xiii
List of Figures
Figure 4-4. P95, P5 and expected value lines of fixed capital investment for various Pd layer thickness values; Pd, palladium; CMR, catalytic membrane reactor...... 126
Figure 4-5. P95, P5 and expected value lines of total capital investment for various Pd layer thickness values; Pd, palladium; CMR, catalytic membrane reactor...... 127
Figure 4-6. P95, P5 and expected value lines of total product cost for various Pd layer thickness values; Pd, palladium; CMR, catalytic membrane reactor...... 128
Figure 4-7. Tornado diagram for total capital investment (Pd unit price: US$16·9– 24·4/g; WGS reactor cost: baseline ±10%; catalyst unit price: baseline ±10%; working capital to total capital investment ratio range 10–20%); Pd: palladium; WGS, water–gas shift...... 130
Figure 4-8. Tornado diagram for total product cost (Pd unit price: US$16·9– 24·4/g; palladium membrane lifetime: 1–5 years; WGS reactor cost: baseline ±10%; raw material cost per year: baseline ±10%; working capital to total capital investment ratio range: 10–20%; financing interest to total capital investment ratio range: 6–10%); Pd, palladium; WGS, water–gas shift...... 131
Figure 5-1. Hydrogen permeance and helium leak tests of MA-156...... 138
Figure 5-2. Hydrogen permeance and helium leak tests of MA-157...... 139
Figure 5-3. Hydrogen permeance and helium leak tests of MA-159...... 142
Figure 5-4. Hydrogen permeance and He leak tests of MA-160, and yellow lines indicating membrane oxidation stages...... 145
Figure 5-5. Experimental produced H2 purity for various Pd thickness values at 450°C and a retentate pressure of 2 bar...... 147
Figure 5-6. Average hydrogen permeance with the standard deviation for Pd-based separation modules using various palladium layer thicknesses...... 148
xiv
List of Figures
Figure 5-7. Cumulative probability distribution profile of levelized cost for Pd- based separation modules with various Pd layer thickness values at 450°C and a retentate pressure of 50 bar...... 152
Figure 5-8. P95, P5, and expected value lines of levelized cost for Pd-based separation modules with various palladium layer thickness values. ..154
Figure 5-9. Tornado diagram for the levelized cost of hydrogen produced via Pd- based separation modules using 2.7 µm thick membranes...... 156
Figure 5-10. Tornado diagram for the levelized cost of hydrogen produced via Pd- based separation modules using 10.4 µm thick membranes...... 157
Figure 6-1. Schematic of a CMR module for water gas shift reaction and photograph of the actual CMR rig built at WPI...... 160
Figure 6-2. Cost distribution of a WGS-CMR module...... 164
Figure 6-3. Normalized Total Capital Investment of WGS-CMR modules for various capacities...... 166
Figure 6-4. Tornado diagram for Total Capital Investment...... 167
Figure 6-5. Experience curve for Total Capital Investment of WGS-CMR modules...... 170
Figure 7-1. Cost distribution of an industrial scale CMR module...... 175
Figure 7-2. Coal feed, CO2 production, and CO2 emissions of hydrogen production plants for various technology options...... 177
Figure 7-3. Cumulative probability distribution profiles of the Net Present Value
for various technology options in (a) the absence of CO2 tax and (b)
under an initial tax rate of $30 per tonne of CO2...... 181
Figure 7-4. Tornado diagrams for the NPV of various technology options in the
presence of CO2 tax...... 183
Figure 7-5. Net present value cumulative probability distribution profiles for
various technology options in the absence of CO2 tax and their ENPV rates...... 189
xv
List of Figures
Figure 7-6. Net present value cumulative probability distribution profiles for HP- CMR plants under various discount rates...... 192
Figure 8-1. Cumulative probability distribution profiles of the Net Present Value (NPV) with the Expected Net Present Value (ENPV) for Case A and
Case B under an initial tax rate of $30 per tonne of CO2...... 199
Figure 8-2. Cumulative probability distribution profiles of the NPV with the ENPV for various constructional flexibility options with inclusion of a CCS
system in the initial design phase (a) in the absence of CO2 tax and (b)
under an initial tax rate of $30 per tonne of CO2...... 201
Figure 8-3. Cumulative probability distribution profiles of the NPV with the ENPV for various constructional flexibility options with inclusion of a CCS
system at a later stage (a) in the absence of CO2 tax and (b) under an
initial tax rate of $30 per tonne of CO2...... 205
Figure 8-4. ENPV difference of cases between with and without flexibility options
for various initial CO2 tax...... 208
Figure 8-5. ENPV difference of cases between with and without design flexibility options for various expected growth rates...... 212
Figure 8-6. ENPV difference of cases between with and without design flexibility
options at different year of introducing CO2 tax...... 215
Figure 8-7. Tornado diagrams for ENPV of HP-CMR plants with various flexibility options...... 219
Figure 9-1. Process block flow diagram of the natural gas-based H2 production plant with CMR technology for MSR...... 227
xvi
List of Tables
Table 2-1. Mechanism for catalytic reforming of methane...... 48
Table 2-2. Kinetic parameters for methane steam reforming over a Ni/MgAl2O4 catalyst [31,148,149]...... 51
Table 2-3. Possible mechanisms for the high-temperature WGS reaction over a Fe- based catalyst...... 54
Table 2-4. Activation energies and parameters for the empirical power-law rate equation of the HTS reaction over a Fe-based catalysts...... 56
Table 2-5. Possible mechanisms for the low-temperature WGS reaction over Cu- based catalyst [166]...... 60
Table 2-6. Activation energies and parameters for the empirical power-law rate equation over various Cu-based catalysts...... 61
Table 3-1. Parameters of Pd/Au alloy composite membranes and high-temperature shift catalysts used in the process modeling framework...... 71
Table 3-2. Feed specifications, reaction conditions, and H2 permeance of membranes...... 72
Table 3-3. Industrial scale CMR module specifications used for cost analysis...... 73
Table 3-4. Industrial scale membrane reactor module specifications, feed specifications and reaction conditions [176]...... 77
Table 3-5. Cost data used in the economic performance assessment framework (Cost base: 2012)...... 77
Table 3-6. Electricity, steam consumption and labor cost data (Cost base: 2012). .78
Table 3-7. Economic parameters used in cost analysis [177]...... 78
Table 3-8. Estimation of total capital investment for the membrane reactor module (Cost base: 2012)...... 79
xvii
List of Tables
Table 3-9. Estimation of total product cost (TPC) for the membrane reactor module (Cost base: 2012)...... 80
Table 3-10. Estimation of capital investment costs for the large-scale Pd-based separation module...... 82
Table 3-11. Large-scale Pd-based separation module specifications used for cost analysis...... 83
Table 3-12. Estimation of total product cost and levelized cost for the large-scale Pd-based separation module...... 84
Table 3-13. Estimation of production costs for a Pd composite membrane. † ...... 87
Table 3-14. Estimation of Capital Investment for a large-scale Pd-based CMR module. ‡ ...... 88
Table 3-15. Actual large-scale Pd/alloy-based CMR specifications used for cost analysis...... 90
Table 3-16. Estimation of the capital investment for an industrial scale Pd/alloy- based CMR module. a ...... 94
Table 3-17. Total product cost estimation for the HP-CMR plant. a ...... 95
Table 3-18. Uncertain cost model inputs and corresponding probability distributions for an industrial-scale Pd/alloy-based CMR module for WGS integrated into natural gas-based hydrogen production plants. 103
Table 3-19. Uncertain cost model inputs and corresponding probability distributions for an actual large-scale Pd/alloy-based separation module for hydrogen purification...... 104
Table 3-20. Uncertain cost model inputs and corresponding probability distributions for an actual large-scale Pd-based CMR module for Water-Gas Shift reaction...... 105
Table 3-21. Uncertain cost model inputs and corresponding probability distributions for coal-based hydrogen production plants with integrated industrial-scale Pd/Alloy-based CMR modules...... 106
xviii
List of Tables
Table 3-22. Uncertain cost model inputs and corresponding probability distributions for coal-based HP-CMR plants with integrated flexibility options...... 108
Table 3-23. Descriptions and assumptions for all cases with various engineering design flexibility...... 112
Table 3-24. Composition of the membranes shown in this study...... 114
Table 4-1. Fixed capital investment/total capital investment/total produce cost expected values for various palladium thickness values...... 124
Table 5-1. Long term tests presented in the literature...... 136
Table 5-2. List of characteristics of MA-159...... 141
Table 5-3. List of characteristics of MA-160...... 144
Table 5-4. Lifetime estimation for Pd-based separation modules with various palladium thickness values...... 149
Table 5-5. Pd-based separation module cost summary regarding fixed capital investment, total capital investment, and total product cost...... 150
Table 5-6. P95, P5, and the expected value of levelized cost for a separation module using various Pd layer thicknesses...... 154
Table 6-1. WGS-CMR module cost summary...... 164
Table 7-1. Industrial scale CMR module cost summary...... 175
Table 7-2. Hydrogen production plant specifications and capital investment costs for various technology options...... 178
Table 7-3. ENPV performance (B$) of technology options under various CO2 tax scenarios...... 187
Table 7-4. ENPV difference (B$) between two technology options under various
CO2 tax scenarios...... 187
Table 7-5. Sensitivity analysis results of the NPV-based economic performance assessment for HP-CMR plants with various discount rates...... 193
xix
List of Tables
Table 8-1. Comparison of financial metrics derived from simulation results between Case A and Case B...... 199
Table 8-2. Financial metrics derived from simulation results for various constructional flexibility options with inclusion of a CCS system in the initial design phase...... 202
Table 8-3. Financial metrics derived from simulation results for various constructional flexibility options with inclusion of a CCS system at a later stage...... 206
Table 8-4. Summary of ENPV difference (M$) of cases between with and without
flexibility options for various initial CO2 tax rates...... 209
Table 8-5. Summary of ENPV difference (M$) of cases between with and without
flexibility options for various expected growth rates of CO2 tax...... 213
Table 8-6. Summary of ENPV difference (M$) of cases between with and without
flexibility options at different year of introducing CO2 tax...... 215
Table 8-7. Quantitative summary of the NPV outcomes for all cases...... 221
Table 8-8. Quantitative summary of the valuable findings for all cases...... 222
xx
Chapter 1 Chapter 1 Introduction Introduction
Global energy demand growth patterns accompanied by climate change mitigation challenges have provided a significant stimulus to creative thought and attendant research activities aiming at the development and successful demonstration of advanced clean energy technology options worldwide. Indeed, according to the U.S. Energy Information Agency, energy demand in the U.S. increases by 0.4% annually [1], and in light of the current state of affairs in the field of energy production, energy supply security considerations, cost advantages as well as relatively low technology risks have traditionally favored the expansion of the use of fossil fuels worldwide to meet demand growth in the near/medium-term [2,3,4]. Within the above context, hydrogen represents an energy carrier with the potential to reliably contribute to a robust global energy system as a complementary option to electricity as well as liquid fuels use in an environmentally responsible manner provided that the pertinent hydrogen production technologies (both the conventional and the more advanced ones recently proposed) can reach appealing techno- economic performance levels in the presence of fuel market and regulatory uncertainties [2,3,4].
Apart from the scope of hydrogen energy, today’s global hydrogen market is already greater than 40 billion dollars [5]. As shown in Figure 1-1, its use in the chemical and petrochemical industry is essential with a global share of 54% in the production of ammonia, 35% for upgrading fuels and 6% for a range of applications in the electronics industry [6]. Furthermore, according to U.S. Geological Survey from 2006 to 2015 [7,8,9,10,11,12,13,14,15,16], the
1
Chapter 1 Introduction hydrogen consumption for global ammonia production has increased by 22% over the past decade (please see Figure 1-2). Along with the promising technology development of fuel cells in the transportation, the hydrogen market is expected to expand rapidly in the near-term. Therefore, a great potential market of hydrogen is also considered for the provision of the requisite incentives aiming at the development and successful demonstration of hydrogen economy from hydrocarbon economy.
Figure 1-1. Distribution of the global hydrogen market. Data adapted from Linde Engineering [6].
2
Chapter 1 Introduction
Figure 1-2. Hydrogen consumption in the production of ammonia from 2004 to 2013. Data adapted from U.S. Geological Survey, Mineral Commodity Summary, 2006-2005 [7,8,10,11,12,13,14,15,16].
Hydrogen is the most abundant element and freely spreads throughout the universe. However, hydrogen does not appear in nature as the fuel hydrogen, and thus it has to be produced from a natural resource. According to the report by Midilli et al. [17], the time period for introduction of a hydrogen economy can be split into three phases – involving the near-, the medium-, and the long-term. The primary source for hydrogen production will be changed from fossil fuels and biomass to water from the near/medium- to the long-term. Currently, hydrogen production technology pathways involve the direct chemical conversion of fossil fuel resources, i.e. steam reforming of natural gas (methane), partial oxidation of hydrocarbons, or coal gasification, and electro-chemical conversion, i.e. via water electrolysis using electricity, nuclear or renewables [2,18]. Focusing on the current state of affairs in the field of hydrogen production, supply stability considerations, cost advantages as well as low technology risks have traditionally
3
Chapter 1 Introduction
favored the expansion of natural gas- and coal–based hydrogen production capacity worldwide to meet demand growth in the near/medium-term [2,4,19]. It should be pointed out that recent developments and advancements in hydraulic fracturing and horizontal drilling technology induce the potential for a considerable expansion of shale gas supply in places such as the USA, and may also precipitate significant changes in hydrogen production [1]. Under current market conditions, natural gas (methane) steam reforming (MSR) represents the dominant technology pathway [2,4,19] for the production of hydrogen worldwide, contributing with 48% of total hydrogen supply with efficiencies ranging from 70% to 80% (please see Figure 1-3) [20].
Figure 1-3. Share of primary H2 sources and technology pathways.
4
Chapter 1 Introduction
The conventional hydrogen production system via the MSR technology pathway is composed of two basic process steps:
(i) The reforming step in which methane (CH4) reacts endothermically with excess
steam (H2O) at high temperature and in the presence of a nickel-based catalyst
producing carbon monoxide (CO), carbon dioxide (CO2), hydrogen (H2) and
unreacted CH4 (the mixture collectively known as syngas) [2,21,22]. This process step primarily involves three reactions, as shown in Equation 1-1 to 1-3 [22]. Typically desulfurization is pursued through absorption before the natural gas reaches the catalyst in the above reforming step [2,3,4].
MSR reaction I: CH + H O CO + 3H ΔH298K = 206.1 kJ/mol 1-1
4 2 ⇄ 2 WGS reaction: CO + H O CO + H ΔH298K = -41.15 kJ/mol 1-2
2 ⇄ 2 2 MSR reaction II: CH + 2H O CO + 4H ΔH298K = 165.0 kJ/mol 1-3
4 2 ⇄ 2 2 (ii) The water–gas shift (WGS) reaction step in which CO reacts exothermically with
steams producing CO2 and H2 (Equation 1-2). This step is followed by a separation
process in which CO2 is traditionally separated from the reactant mixture through chemical absorption with the aid of an alkaline-based solution, and a purification process step for the resulting hydrogen-rich gas attained by way of standard pressure swing adsorption (PSA) techniques [2,3,4].
Instead of MSR technology, there is an increasing interest in coal gasification technology for hydrogen production due to the abundance of coal reserves on Earth as well as the low and non-volatile price when compared to natural gas. Indeed, in light of the statistical analysis by U.S. Energy Information Administration (EIA), the amount of global recoverable coal (estimated in 2011) is about 980 billion short tons, which can supply the global coal market for more than 100 years at current consumption rates [23]. Focusing on the U.S. market in particular, coal gasification represents a quite economically attractive, reliable and mature hydrogen production technology
5
Chapter 1 Introduction
option given that the U.S. geologically possesses over one fourth of the global coal reserves (please see Figure 1-4) [24].
Figure 1-4. Share of global recoverable coal resources in 2011. Data adapted from U.S. EIA [23].
Coal gasification technology produces H2 by generating syngas from pulverized coal and
oxygen (O2) at 677°C and 5.6 MPa. Syngas is composed mainly by CO, CO2, H2 and H2O along with other particulate impurities which are treated with water wash in a scrubber. Conventionally,
syngas is sent to WGS reactors to convert CO to CO2 and simultaneously generate H2. The H2- rich gas is then cleaned from NH3, SO2, Hg and H2S followed by the H2 purification step performed by PSA [2,3,4].
As mentioned earlier, hydrogen production will mainly rely on fossil fuels in the near/medium-term and inevitably continue to be a source of significant greenhouse gas (GHG)
6
Chapter 1 Introduction
emissions. It is expected that the use of carbon capture and sequestration (CCS) systems in hydrogen production facilities will receive increasing attention due to future regulatory action on
CO2 emissions, such as the introduction of a carbon tax or market-inspired emission permit trading mechanisms. Indeed, carbon taxation levels, currently considered as appropriate to generate the “right” market price signals and thus facilitate their internalization by the pertinent markets, display variability thereby affecting the realization prospects of demonstration plants and the final adoption of new advanced clean energy technology options. Examples of various carbon tax levels include the $12 per tonne of CO2 in Boulder Colorado [25], the $30 per tonne of CO2 in Finland
[26], the $23 per tonne of CO2 in Sweden [26], the $30 per tonne of CO2 in British Columbia [27], and the $30 per tonne of CO2 in Australia [27]. Furthermore, carbon tax changes over time have been observed, such as in the Norwegian offshore which in 1991 was U.S.$50/tonne, but was reduced to U.S.$38/tonne in 2000.
However, from an economic viewpoint, the implementation of a CCS system requires substantial capital investment while incurring additional operation and maintenance (O&M) costs.
In particular, the CO2 capture process component is the most expensive one representing approximately 75% of the overall cost. Consequently, various CCS options have been investigated throughout the literature [28] relying on adsorption, absorption, chemical looping and membrane
separation. While amine-based sorption methods for CO2 capture are well known and considered mature technology options, their thermal and oxidative degradation characteristics could generate potentially hazardous and toxic substances and carcinogens [29]. Moreover, due to demonstrated enhanced capture efficiency levels and additional benefits, the use of membrane-based separation methods is now recognized as a potentially meritorious technology option [29].
In light of the above considerations, the production of hydrogen via the WGS reaction in catalytic membrane reactors (CMRs) integrated into a natural gas- or coal-based hydrogen production system represents an appealing technology option (HP-CMR) due to its great potential for CO2 capture, H2 purification, as well as process intensification. As shown throughout the pertinent literature [30,31,32] the constant removal of the product leads to higher conversion levels of the WGS reaction than the ones attainable under conventional conditions. Moreover, CMRs used in WGS process systems make possible the generation of pure H2 and clean pressurized CO2 realized in a single unit, and display higher product recovery and energy-efficiency potential than
7
Chapter 1 Introduction
pressure swing adsorbers of traditional H2 separation units [33]. In particular, Pd- and Pd/alloy- based membranes with stainless steel, Hastelloy or Inconel supports also exhibit significant
advantages including the capacity of yielding ultra-pure H2 at a stable high flux under high temperatures (400–600°C) and pressures (20–50 bar). Supplementary advantages include their long term durability, improved thermal, chemical and mechanical stability, cost-effective fabrication and maintenance, as well as easiness in scale up and practical assembly/disassembly capabilities for both small and large scale industrial applications [30,31,32,34,35,36,37,38,39]. Given the appealing performance characteristics that HP-CMR exhibits, it is expected that the HP- CMR technology option integrated into natural gas- or coal-based hydrogen production plants could potentially lead to:
(1) Enhanced environmental performance through a cost-effective design accommodation of a CCS system while maintaining competitive efficiency performance levels compared to the conventional technology option. (2) Attractive economic performance prospects, particularly in light of the anticipated
future regulatory action on CO2 emissions.
It should be pointed out that increasingly stringent environmental performance
requirements on CO2 emissions in natural gas- and coal-based hydrogen production plants provide ample motivation for the possibility of integrating CCS systems into their design [2,40,41,42,43,44]. However, despite advancements in our conceptual understanding of many potential design integration options of CCS systems into conventional hydrogen production plants, the lack of any accumulated operating experience even on the technology demonstration front poses significant challenges for HP-CMR plants. Within such a context, any preliminary research attempt to assess the economic viability and performance prospects of the new technology option is certainly justified, thus contributing to recently intensified efforts to provide the appropriate incentives and stimulate the realization of advanced technology demonstration projects at the commercial scale worldwide. Indeed, the use of advanced technology options such as the proposed one offer reasonable prospects of inducing appealing economic performance and valuation profiles in a carbon-constrained world [45].
Mueller-Langer et al. [2] offered a comprehensive evaluation of the techno-economic performance characteristics associated with conventional hydrogen production options (based on
8
Chapter 1 Introduction
methane steam reforming, coal and biomass gasification as well as water electrolysis) within the European Union (EU) context, and also pointed out the prospects of enhanced environmental and economic performance through the adoption of more advanced options such as HP-CMR with
CCS capabilities. Based also on attainable process efficiency levels, CO2 emissions and H2 production costs, it was concluded that natural gas and coal-based HP-CMR options could
represent important viable routes for clean H2 generation. Further economic studies for H2 production plants include those reported by Kreutz et al. [46,47] who evaluated the co-production
of H2 and electricity from coal using conventional technologies via the EPRI TAG methodology. Low operating costs were found to be the most important factor contributing to the overall economic competitiveness of this option, while H2 distribution and storage technologies acted as the biggest constraints. Van Vuuren et al. [48] developed interesting projected economic performance profiles of the deployment of CCS technologies taking into account various techno- economic parameter uncertainties, concluding that uncertainty in techno-economic parameters could have a severe impact on model-based process economics projections. Additionally, Saboohi et al. [49] developed a stochastic model for an energy supply system with uncertain fuel prices and applied it to a case where gas fired engines were competitive with renewable energy sources. The results were discussed in terms of expected value of perfect information and the value of a stochastic analytical framework. In the absence of any accumulated operating experience and associated data, an economic performance evaluation framework for a new technology option such as HP-CMR would be inevitably based on reasonable theoretical estimates (by drawing on the most comprehensive studies and expert opinion) while explicitly acknowledging irreducible uncertainty sources (market, regulatory, etc.) [50].
Engineering design flexibility encompasses with a dynamic pro-active methodological framework to analyze, respond and manage the effect of various sources of irreducible uncertainty on the system’s lifetime environmental performance by alleviating negative impacts arising from downside risks and/or enhancing benefits associated with upside opportunities [51,52,53]. In light of the research study by Zhang et al. [54], potentially value-enhancing flexibility options for Integrated Gasification Combined Cycle (IGCC) power plants with embedded membrane reactor modules (IGCC-MR) were proposed and evaluated under various uncertain CO2 tax scenarios. It was demonstrated that the economic performance of IGCC-MR plants can be greatly improved when flexibility options are implemented appropriately. For the study of economic performance
9
Chapter 1 Introduction
of hydrogen production plants, significant research work has been done on building comprehensive economic performance evaluation frameworks; nevertheless little effort has been put into potentially value-enhancing flexibility options for plant design and operations management. As
stimulated by uncertain regulatory actions on CO2 emissions, it is expected that the economic performance of HP-CMR plants can be improved by dealing with uncertainties encountered via embedded flexibility options.
The overall objective of this research study was to develop a methodological evaluation and analysis framework to assess the techno-economic performance and feasibility prospects of the Pd/alloy-based CMR technology option integrated into hydrogen production plants. Furthermore, identification, evaluation, and demonstration of potentially value-enhancing flexibility options for HP-CMR plants were conducted, as efforts to realize demonstration projects of HP-CMR plants at the commercial scale world-wide intensify. To sum up, the specific objectives of this research study were as follows:
1. Develop comprehensive baseline models for Fixed Capital Investment (FCI), Total Capital Investment (TCI), and Total Product Cost (TPC) to evaluate the economic performance of an industrial scale Pd-based membrane reactor module potentially integrated into a hydrogen production plant. In particular, recognize various sources of irreducible uncertainty (including raw material market prices, labor costs, membrane lifetime and maintenance costs, financing interest costs, etc.) and explicitly take into account their effect on TCI and TPC using Monte Carlo techniques.
2. Combine experimental long-term H2 permeance and He leak results of Pd/alloy membranes to predict the membrane lifetime and theoretical economic evaluation results to explore the tradeoff between permeance-thickness and lifetime in terms of economic performance outcomes. 3. Evaluate economic performance of the large-scale WGS-CMR module (built at WPI) reported by Catalano et al. [30] using the developed evaluation framework in order to obtain realistic profiles of economic outcomes and other valuable information by explicitly acknowledging irreducible uncertainty. 4. Structure a detailed comprehensive baseline Net Present Value (NPV) model to assess the economic viability of HP-CMR plants, while irreducible sources of market and regulatory
10
Chapter 1 Introduction
uncertainty (involving coal market price, nominal discount rate, H2 delivery cost, H2 selling
price, CO2 tax rate. etc.) are identified and their effect on the plant’s economic performance is explicitly taken into account through Monte Carlo techniques. 5. Design flexibility options for HP-CMR plants to deal with the intrinsic uncertainties
encountered in uncertain CO2 tax scenarios, and evaluate their potential economic value- enhancing properties by the developed evaluation framework that takes into consideration irreducible sources of market and regulatory uncertainty and integrates them through Monte Carlo techniques.
11
Chapter 2 Chapter 2 Literature Review Literature Review
2.1 Palladium Membrane
The permeation of hydrogen (H2) through transition metals, known as platinum metals, was first observed with the experiment on homogeneous plates of platinum (Pt) and iron (Fe) by Deville and Troost in 1863 [55]. Three years later, Graham (1866) discovered the phenomenon of
diffusion of H2 through palladium (Pd) at elevated temperature [56]. In light of his report, Pd could
absorb up to 935 times its own volume of H2 when it was cooled down from red heat. Since then, there has been a substantial expansion of research activity in the fields of Pd-H system and Pd- based membranes for H2 separation.
Diffusion of H2 into Pd lattices can form an alloy of Pd with metallic H2, usually called Pd hydrides. According to the Hume-Rothery rules, since the atomic radius of H (1.558 Å) and the atomic radius of Pd (1.375 Å) are very close, they are considered to form a solid solution [57]. In addition, atomic H is small enough to fit into interstitial positions of Pd, so that the Pd hydride is recognized as a kind of interstitial solid solution. To understand characteristics of the Pd-H system, studying the Pd-H phase diagram is a good way to illuminate them. Figure 2-1 shows the
equilibrium solubility isotherms of PdHn for the Pd-H system [58]. The original data was adapted from the research studies by Wicke and Nernst (1964) [59] and Frieske and Wicke (1973) [60] in
12
Chapter 2 Literature Review
which the molar H/Pd ratio (n) at different H2 partial pressures of different isotherms were determined by means of magnetic susceptibility.
1 Figure 2-1. Equilibrium solubility isotherms of PdHn for the Pd-H system [58,59,60].
1 In the original article by Shu et al. (1991) [58], the y-axis values of 10, 20, 30 and 40 in the inset are incorrect. The revised y-axis values of 1, 2, 3 and 4 were made by Guazzone (2005) [61] according to the research work by Wicke and Nernst (1964) [59].
H2 solubility, n (=H/Pd), is defined as the ratio of H atoms absorbed per Pd atom and is given as a function of H2 partial pressure and temperature, shown in Figure 2-1. Normally, Pd hydrides containing less H with a smaller lattice parameter is called α phase, while Pd hydrides possessing more H with a larger lattice parameter is called β phase. Both phases have a face
13
Chapter 2 Literature Review
centered cubic (FCC) crystalline structure. Gillespie and Galstaun (1936) [62] found that there is a critical H2 partial pressure (19.87 atm) and temperature (295.3°C) and corresponding n value (0.270) in the Pd-H phase diagram. Below the critical conditions, the Pd-H system exhibits a miscibility gap region where the two phases, α and β, would coexist with each other. In the miscibility gap region, H2 solubility of each isotherm is independent of H2 partial pressure, but dependent on the ratio of β phase to α phase, and as the H2 solubility is given, the proportion of α phase or β phase can be determined by the level rule. In addition, several phase transformations between α and β in the immiscibility gap region would lead to severe deformation of Pd that is
known as “H2 embrittlement” [63]. The mechanism of H2 embrittlement can be explained by lattice deformation after cycles of α and β phase transformation since the nucleation of the β phase with the larger lattice parameter from the α phase would induce lattice expansion and thereby impose strain in the lattice. Although the lattice of Pd would also expand beyond the miscibility gap region when loading H atoms, the change in lattice is gradual, rather than the abrupt change of phase transformation.
For the Pd membrane separation system, H2 embrittlement would create microcracks on the membrane surface to lower the separation performance. Therefore, care is necessarily taken
not to exceed the maximum H2 partial pressure (Pmax) of α phase at each given operating temperature in the miscibility gap region, above which the phase transformation would occur due
to nucleation of β phase. The Pmax of α phase depends on operating temperature, which can be described by Equation 2-1 [62]:
1877.82 log( max) = 4.6018 2-1
𝑃𝑃 − 𝑇𝑇 where P is in atm and T is in K. Alternatively, alloying Pd with other metals, such as Ag and Au, can reduce the critical temperature to diminish the area of the miscibility gap region due to higher
H2 solubility of Pd alloy, thereby reducing the probability of H2 embrittlement.
14
Chapter 2 Literature Review
2.2 Hydrogen Transport Mechanism in Palladium Membranes
The permeation behavior of H2 through Pd membranes has been intensively investigated for many years, and it follows the multistep process called the “solution-diffusion mechanism.”
The solution-diffusion mechanism, involved in H2 transport from the high H2 partial pressure side to the low H2 partial pressure side, is summarized in the following seven steps [58,64] and shown schematically in Figure 2-2:
(1) H2 transport from the bulk gas to the gas layer nearby the membrane surface,
(2) reversible dissociative chemisorption of H2 onto the membrane surface, (3) reversible dissolution of atomic H from the surface into the bulk membrane, (4) diffusion of atomic H through the bulk membrane,
(5) transition of atomic H from the bulk membrane to the surface on the low H2 partial pressure side, (6) recombinative desorption from the membrane surface,
(7) H2 transport away from the membrane surface to the bulk gas.
Figure 2-2. Schematic diagram of solution-diffusion mechanism.
15
Chapter 2 Literature Review
It is well known that Sieverts’ Law describes the solubility of a diatomic gas in metals which is proportional to the square root of the partial pressure of the gas in a thermodynamic
equilibrium [85]. Based on Sieverts’ Law, the H2 flux through the solution-diffusion mechanism
can also be derived as a function of the difference of square roots of the H2 partial pressure
(Sieverts’ equation), which is very significant for the estimation of the H2 flux in practical
applications. Therefore, to better understand the permeation behavior of H2 through Pd membranes, the derivation of the Sieverts’ equation may provide a good way to illuminate it.
First, H2 flux through Pd membranes, governed by a dynamic equilibrium between desorption and adsorption, can be expressed as the difference between the adsorption rate and the desorption rate, as shown in Equation 2-2:
J = k k P (1 ) 2-2 2 2 H2 d ∙ θ − a ∙ H2 ∙ − θ 3 2 where J stands for H2 flux at the exit of diffusing H2 with the unit of m /(m -h), P is the H2
partial pressureH2 with the unit of atm, and is the surface H2 coverage. k and k areH2 the rate 3 2 3 2 constants of desorption and adsorption θwith the units of m /(m -h-atm)d anda m /(m -h), respectively. In the adsorption-desorption process, due to the dissociative chemisorption of H2,
step (2), which enables only H2 molecules to be adsorbed onto the membrane surface and then to
be dissociated into H atoms, Pd membranes have potentially infinite selectivity to H2.
Likewise, in the dissolution and transition processes, the H2 flux through the Pd membrane at a dynamic equilibrium can also be expressed as the difference between the dissolution rate and the transition rate, as described in Equation 2-3:
J = k n / (1 ) k (1 n / ) 2-3
H2 o ∙ H pd ∙ − θ − i ∙ − H Pd ∙ θ where k and k are the rate constants for the dissolution into the membrane and the transition out 3 2 of the membraneo i respectively with the unit of m /(m -h), and n / is the atomic H/Pd ratio expressed as the solubility of Pd. With regard to the diffusion of HH atomsPd in the bulk membrane, the permeation flux at steady state can be depicted by Fick’s Law as seen in Equation 2-4:
16
Chapter 2 Literature Review
D J = (C C ) 2-4
2 H ∙ ret − per δ 2 where D is the H2 diffusivity in m /s, is the thickness of Pd membrane in m, and C and C
are the H2 concentration in the Pd sub-δlayers adjacent to the membrane surface at the retentateret sideper and permeate side in mol/m3, respectively. The diffusion behavior of H atoms into Pd membranes can be described by “jump model” that H atoms jump between octahedral interstitial sites in the FCC lattice of Pd [65]. The predictions of the jump model have been verified by quasielastic neutron scattering in which it was found that the average jump distance is a/√2 where a is the
lattice parameter of FCC Pd [65]. In Equation 2-3, the concentration of H2 can be expressed as the 3 product between the atomic H/Pd ratio and the H2 solubility coefficient ( , with the unit of mol/m ) as shown in Equation 2-5: κ
C = n / 2-5
κ ∙ H Pd
The physical meaning behind is the concentration of H2 in the Pd sub-layers as n / value is
equal to 1, and is not to be confusedκ with the H concentration. After substitution of HEquationPd 2-5 into Equation 2-4, Equation 2-6 can be obtained as shown in the following:
D J = (n / , n / , ) 2-6
2 ∙ κ H ∙ H Pd ret − H Pd per δ To study the rate-limiting step in the solution-diffusion model, Ward and Dao [66] observed that, in the absence of external mass transfer resistance, the diffusion-limited permeation is expected for clean Pd membranes with the thickness greater than 1 µm at operating temperatures above approximately 300°C. Under such conditions, the Pd structure is in α phase, so that the H/Pd
ratio can be assumed to be very small (n / 1), and thus the term (1 n / ) in Equation 2-3
can be neglected. Since diffusion is expectedH Pd ≪ to be the rate-limiting step,− HbothPd the adsorption- desorption process and the dissolution-transition process reach a dynamic equilibrium; thereby the overall rates of adsorption and dissolution (Equation 2-2 and 2-3) are approaching zero (please see Equation 2-7).
17
Chapter 2 Literature Review
J = k k P (1 ) = k n / (1 ) k = 0 2-7 2 2 H2 d ∙ θ − a ∙ H2 ∙ − θ o ∙ H Pd ∙ − θ − i ∙ θ After transposition of formulae, the ratio between and (1 ) can be expressed by
Equation 2-8: θ − θ
k P k n / = = 2-8 (1 ) k 2 k θ a ∙ H o ∙ H Pd � − θ d i
Solving Equation 2-8 for n / , the mathematical relationship between the H/Pd ratio and the H2
partial pressure can be obtainedH Pd as displayed in Equation 2-9, where n / is proportional to the
square root of the H2 partial pressure corresponding to the graph in theH Pdα region in Figure 2-1. Furthermore, it was also found that Equation 2-9 is equivalent to Henry’s Law and is only valid for the low atomic H/Pd ratio domain.
. k k . n / = P 2-9 k k 0 .5 i ∙ a 0 5 H Pd 0 5 H2 o ∙ d Based on Equation 2-9, the Sieverts’ constant ( K ) is defined as the equilibrium constant (with the 0.5 unit of atm ) for the adsorption-desorption processs and the dissolution-transition process (please see Equation 2-10). The value of K for various operating temperatures can be determined by
calculating slopes of isotherms in thes low atomic H/Pd ratio domain in Figure 2-1.
k k . K = 2-10 k k 0. 5 o ∙ d s 0 5 i ∙ a From the thermodynamics point of view, the Sieverts’ constant is dependent on temperature and can be expressed as a function of temperature in the Arrhenius relation [83] as shown in Equation 2-11:
18
Chapter 2 Literature Review
S H K = exp exp 2-11 2R 2RT ∆ H2 ∆ H2 s � � �− � where H denotes the heat of desorption in J/gmol, and S indicates the entropy of desorption
in J/gmol∆ -Hk.2 By the substitution of Equation 2-9 and 2-10 into∆ H 2Equation 2-6, the Sieverts’ equation can be derived as shown in Equation 2-12, which indicates that the permeation flux of H2 is linearly dependent on the difference of square roots of the H2 partial pressure at a given temperature. Please
notice that the Sieverts’ equation is only valid at low H concentration in the bulk Pd and H2 diffusion is expected to be the rate-limiting step in the whole permeation process. Based on
Equation 2-12, the H2 solubility is defined as the ratio of the solubility coefficient to the Sieverts’ constant ( /K ) with the unit of mol/(m3-atm0.5).
κ s D J = (P . P . ) 2-12 K , , ∙ κ 0 5 0 5 H2 ∙ H2 ret − H2 per s ∙ δ As mentioned earlier, both the H2 diffusivity and the H2 solubility of Pd membranes depend on operating temperatures, and can be described by Arrhenius equations as displayed in Equation 2-13 and 2-14:
E D = D exp 2-13 R T − D 0 ∙ � � H∙ S = S exp 2-14 R T ∆ s 0 ∙ � � ∙ where E is the activation energy for H2 diffusion in J/gmol, and H is the enthalpy of H2
dissolutionD in J/gmol. R stands for gas constant in J/(gmol-K), while∆ Ts indicates the operating temperature in K. D and S denote constants for the H2 diffusivity and the H2 solubility,
respectively. Furthermore,0 since0 the H2 permeability (Q) is defined as the product of H2 diffusivity
and the H2 solubility, a similar temperature dependent expression for Q can be derived as shown in Equation 2-15:
19
Chapter 2 Literature Review
E Q = D S = Q exp 2-15 R T − Q ∙ 0 ∙ � � ∙ where Q (as the product of D and S ) stands for the permeability constant, and E (as the sum
of E and0 H ) denotes the activation0 0 energy for H2 permeability in J/gmol. After substitutionQ of
EquationD 2∆-15s into Equation 2-12, the expression for J in terms of Q can be yielded (please see Equation 2-16). H2
Q . . J = (P , P , ) 2-16 0 5 0 5 2 H ∙ H2 ret − H2 per δ In light of Equation 2-16, the H2 flux is proportional to the reciprocal of membrane thickness as well as the difference of square roots of the H2 partial pressure. It should be pointed out that it is very hard to measure and obtain the accurate value of membrane thickness in practice due to non-
uniformity of support surface and membrane deposits. In addition, since the H2 partial pressure is determined by external operating conditions rather than properties of Pd membranes, the H2
permeance (F) is quite often used to indicate the H2 permeation performance of Pd membranes in practical cases for the purpose of comparing membrane properties (please refer to Equation 2-17).
Q J F = = . . 2-17 (P , H2P , ) 0 5 0 5 δ H2 ret − H2 per
Through H2 permeation experiments with Pd membranes, it has been found that the H2 pressure exponent (generally defined as the n-value) in Equation 2-16 may deviates from 0.5, making the Sieverts’ equation become invalid [67,68,69,70]. As mentioned earlier, the Sieverts’
equation is only valid at low H concentration in the bulk Pd and H2 diffusion is expected to be the
rate-limiting step in the whole permeation process. If the rate-limiting step is not H2 diffusion, the
term (1 n / ) cannot be neglected in Equation 2-3, and thus both the adsorption-desorption process −andH thePd dissolution-transition process will not reach a dynamic equilibrium to make Equation 2-7 valid. Consequently, the Sieverts’ equation cannot be derived. Collins and Way [68] observed that the n-value for the pure Pd film was not 0.5, and it would decrease from 0.602 to 0.566 with increasing temperature from 550 to 600°C under the pressure difference of 15 bar. It
20
Chapter 2 Literature Review
was supposed that H2 diffusion may not be the rate-limiting step since surface contaminants initially present on the membrane surface would affect H2 solubility of the membrane surface. In
addition, both H2 solubility and H2 diffusivity depend on the operating temperature, and thus the operating temperature is considered as a potential factor to cause the deviation of n-values by
affecting H2 solubility and H2 diffusivity. Morreale et al. [69] found that a Pd membrane that was characterized at the high pressure (27.6 bar) would have an n-value of 0.62 greater than the generally accepted value of 0.5. In light of their report, the deviation of n-value may be attributed
to the fact that H2 diffusivity would be largely enhanced under high pressure conditions. Furthermore, an increased concentration of H atoms within the Pd lattice under high pressure conditions may lead to a non-ideal solution in which attractive forces are exhibited between
dissolved H atoms, making the variation of the Sieverts’ constant to affect H2 solubility. As a
result, the assumption that H2 diffusion is the rate-limiting step in the H2 transport process may become invalid to make deviation of the n-value.
Another factor to change the rate-limiting step from H2 diffusion is membrane thickness.
When the thickness of Pd membranes is extremely thin, H2 diffusivity may approach or exceed the adsorption rate or absorption rate, allowing the adsorption or the absorption to become the rate- limiting step. Ward and Dao [66] proposed a comprehensive model of H2 permeation behavior to investigate the effect of external mass transfer, surface adsorption and desorption, transitions, and
diffusion on H2 permeation in Pd membranes. According to their research study, H2 adsorption would become the rate-limiting step in the permeation process when the thickness of Pd membranes was below 1 µm. It is because extremely high H2 flux of such thin membranes would
induce external mass transfer to become a significant resistance for H2 permeation. Thus, Sieverts’ equation would become invalid to cause the deviation of the n-value. Zhao et al. [71,72] observed
that the H2 flux was a function of transmembrane H2 partial pressure rather than the square root of
transmembrane H2 partial pressure as the Pd membrane thickness was below 1 µm, consistent with the calculation results by Ward and Dao [66]. Wu et al. [73] also reported that the n-value of extra- thin membrane (0.3-0.4 µm) would approach 1 since the H2 permeation was mainly determined
by the surface process instead of the H2 diffusion.
As the Pd membrane contains pinholes or cracks, the solution diffusion mechanism is no
longer the only possible way for H2 permeation through the membrane, since H2 would also pass
21
Chapter 2 Literature Review
through the membrane by these pinholes or cracks. The transport behavior of H2 trough pinholes or cracks can be describe by the dusty-gas model that consists of the Knudsen diffusion and the viscous flow, as shown in Equation 2-18 [35]:
2 8 d 1 d J = + P P = + P P 2-18 3 RTM 8 RT2 εµk εµv �α β ∙ avg� ∙ ∆ � ∙ � ∙ ∙ ∙ avg� ∙ ∆ π δ√ δη where P is the Knudsen diffusion, P P is the viscous flow, and are the Knudsen
and theα viscous∙ ∆ geometric factor coefficients,β ∙ avg ∙ respectively,∆ P is theµ kaverageµv pressure of the retentate side pressure and the permeate side pressure in bar, isavg the porosity, d is the pore diameter in m, M is the gas molecular weight in g/mol, and is the gasε viscosity in Pa·s. The more detailed study about the dusty-gas model was presented byη Mason et al. [74].
According to Equation 2-18, if the Knudsen diffusion or the viscous flow dominates the
H2 permeation process, the H2 flux of the Pd membranes will depend on the pressure difference rather than the difference of square roots of the pressure. Guazzone et al. [70] investigated the
relationship between the n-value and the H2/He selectivity of the Pd membrane and found that the n-value declines as the H2/He selectivity increases. In particular, the n-value would approach 0.5 when selectivity is higher than 400, as shown in Figure 2-3. The effect of selectivity on the n-value was evidenced by their experiment results that the n-value was as high as 0.75 at 500°C for the Pd
membrane as the H2/He selectivity was much lower than 400.
22
Chapter 2 Literature Review
Figure 2-3. The n-value as a function of selectivity for various values of r (ratio of the contribution of the Knudsen diffusion to the overall leak). The figure was taken from Guazzone et al. [70].
In the case of composite Pd membranes, the mass transfer resistance of the support may place significant impact on the H2 permeation behavior, causing the deviation of n-values. The reason can be explained by total mass transfer resistances ( ) throughout the composite Pd membrane, as described by Equation 2-19 [75]: 𝑅𝑅𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡
1 = = = + + + + + + 2-19 𝛿𝛿𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜 𝛿𝛿1 𝛿𝛿2 𝛿𝛿3 𝑅𝑅𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 𝑅𝑅1 𝑅𝑅2 𝑅𝑅3 ⋯ ≈ ⋯ 𝐹𝐹𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜 𝑄𝑄𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜 𝑄𝑄1 𝑄𝑄2 𝑄𝑄3 where subscripts 1, 2, and 3 refer to respective layers in series throughout the composite membrane. In light of Equation 2-19, it is clear that the overall mass transfer resistance for H2 permeation is the sum of the resistance of each layer throughout the composite membrane. When the layer has the extremely thin thickness or the significantly high permeability in relation to other
23
Chapter 2 Literature Review
layers, the mass transfer resistance contributed by this layer can be neglected in the overall mass transfer resistance. Typically, the thickness of the Pd membrane is much smaller than that of the porous support, but the mass transport resistance of porous support is quite often negligible since the permeability of porous support is extremely high when compared to that of the Pd membrane. However, in some cases, the significant support resistance by lower permeability may greatly contribute to the overall mass transfer resistance to affect the H2 permeation behavior. As the result, the Sieverts’ equation may become invalid to cause the deviation of the n-value. It is evidenced by Guazzone et al. [70] that the n-value of the composite Pd membrane was higher than
0.5 using the low-permeability support. Furthermore, the activation energy of the overall H2 permeability is recognized as another useful indicator to describe the support resistance. Collins and Way [68] reported that the deviation of the n-value from 0.5 to 0.573 may be attributed to the
higher activation energy for H2 permeation through the composite Pd membrane than the pure Pd foil. It implies that the significant support resistance existing in the composite membrane has great potential to cause the deviation of the n-value.
24
Chapter 2 Literature Review
2.3 Palladium Alloy Membrane
Alloying of Pd with other potential metals provides a promising way to modify physical and chemical properties and to reduce membrane cost for different demand purposes. To minimize
the probability of H2 embrittlement, Hunter [76] found that Pd/Ag alloys containing 20 wt% Ag
or more can prevent H2 embrittlement after 30 cycles of heating and cooling in H2 atmosphere. The superior mechanical stability of Pd/Ag alloys to withstand temperature cycling can be contributed to suppression of α to β phase transformation [77]. Pd/Ag alloys with higher H2 solubility can reduce the critical temperature and pressure for phase transformation and, thereby, inhibit α to β phase transformation during the heating and cooling. The lower critical temperature and pressure of Pd/Ag alloys can be illustrated by the equilibrium solubility isotherms in the Pd/Ag
alloy system [78]. Thus, alloying of Pd with Ag is an effective way to protect membranes from H2 embrittlement.
Another advantage of alloying Pd with other metals is to improve H2 permeability of Pd membranes. As shown in Figure 2-4, Pd alloys of Pd and other metals, including gold (Au), silver
(Ag), copper (Cu), platinum (Pt), and etc., shows higher H2 permeability than the pure Pd, and H2 permeability of the Pd alloys would change with alloying metal contents and operating
temperatures [79,80,81,82]. Based on the H2 transport mechanism, the H2 permeability is the
product of the H2 diffusivity and the H2 solubility. Both diffusivity and solubility are dependent on temperatures and can be described using Arrhenius equations. In the Pd/Ag alloy system, the diffusivity decreases when the Ag content increases [78,83,84]. However, the solubility of Pd alloys is enhanced by the Ag addition, and there is a peak for the solubility as the Ag content is about 30 wt% Ag [83,84,85]. Since the extent of the solubility improved is greater than the extent of the diffusivity declined by the Ag addition, the permeability of the Pd foil can be effectively improved by alloying with Ag. The maximum permeability of the Pd/Ag alloy foil was measured
with a composition of 27 wt% Ag [80]. Similarly, Pd/Au alloy foils display a superior H2 permeability performance than pure Pd foils due to large enhancement in the solubility by the Au addition [83,84,85]. In contrast to Pd/Ag and Pd/Au alloys, both the diffusivity and the solubility in the Pd/Cu system are reduced to a low value by alloying with Cu; however, it is interesting found that the permeability would suddenly increase and pass through a peak as the Cu content is
25
Chapter 2 Literature Review
near 40 wt% Cu [86]. This phenomenon can be attributed to the body centered cubic (BCC) β phase in Pd/Cu alloys. When the composition of Pd/Cu alloys contains about 40 wt% Cu, the disordered FCC α phase would be changed to form the ordered BCC β phase with the higher diffusivity, and thus the permeability of Pd/Cu alloys could be raised [87]. Consequently, phase transformation of Pd alloys for diffusivity improvement is considered as a potential approach to enhance permeability.
Figure 2-4. Relative permeability between Pd alloy and Pd with various alloying metal contents.2
2 Data at 350°C was adapted from the research work by McKinley [79,80] and Knapton [81], while data at 500°C was taken from the research study by Gryaznov [82].
26
Chapter 2 Literature Review
Apart from the H2 embrittlement and the H2 permeability, mechanical strength of Pd-based
membranes also plays a significant role for the H2 separation. Pd-based membranes with higher mechanical strength would possess superior mechanical resistance and durability under elevated temperature and pressure conditions, extending the membrane lifetime to reduce capital costs. In the Pd/Pt alloy system, it has been found that the ultimate tensile strength would be enhanced with increasing the content of Pt [88,89]. In light of the research study by Berseneva et al. [88], the
ultimate tensile strength of Pd/Pt alloys is proportional to the dissolution content of H atoms;
however, for Pd95Pt5 and Pd85Pt15 alloys, there is a drop in the ultimate tensile strength when the
dissolution content of H atoms exceeds a certain value. This is due to the fact that the α phase of Pd/Pt alloys has been completely transformed to the β phase with the higher lattice constant when enough H atoms are dissolved in Pd/Pt alloys. The significant lattice expansion would cause the drop in the ultimate tensile strength. Therefore, appropriate addition amounts of Pt should be carefully considered in order to effectively strengthen the ultimate tensile strength by means of alloying Pd with Pt.
Finally, Pd alloys have received significantly growing attention due to their superior
chemical stabilities to resist sulfur poisoning, in particular hydrogen sulfide (H2S). It should be
pointed out that H2S is a typical by-product in syngas (from natural gas reforming or coal gasification) along with H2, CO, CO2, and other impurities. It has been demonstrated that even a
trace amount of H2S (as low as 4 ppm) in the feed H2 can poison pure Pd membranes to cause drastic reduction in H2 permeance [79]. When a pure Pd membrane is exposed to a high
concentration of H2S, it may cause irreparable rupture on the membrane surface to severely lower
the separation performance of the membrane [90,91]. To resolve the H2S poisoning issue, the most promising approach in the present is to alloy Pd with other metals by which the surface reactivity and electronic structure of Pd could be modified [92,93].
Morreale et al. [94] found that Pd/Cu alloys with FCC crystalline phase have the capability
to resist H2S poisoning in the H2 atmosphere with 1000 ppm H2S. According to their research
study, FCC Pd70Cu30 membranes exhibited 0-10% declines in the H2 permeance under various
temperatures ranging from 623K to 1000K, while BCC Pd60Cu40 membranes show a steep drop in
the H2 permeance. It was suggested that the surface electronic structure or chemical property of
FCC crystalline phase is responsible for the H2S poisoning resistance [94,95]. To understand the
27
Chapter 2 Literature Review
H2 permeance behavior of BCC Pd/Cu alloys in response to H2S poisoning, O’Brien et al. [96] observed that an extra thin (< 3 nm) Pd-Cu-S terminal layer would rapid form at top surface when
BCC Pd47Cu53 membranes were exposed to 1000 ppm H2S in H2 atmosphere at 350°C. This
terminal layer would cause the inhibition for H2 dissociation or H atom permeation.
Pd/Au alloys also have been extensively studied for resistance of H2S poisoning due to their attractive chemical stability. McKinley [79] observed that a Pd/Au alloy membrane containing 40 wt% Au just has a 21% decline in H2 permeance when exposed to 4 ppm H2S in H2
atmosphere in 6 days, while a pure Pd membrane has 71% decline in H2 permeance. Way et al. [97] characterized a Pd/Au alloy membrane containing 15 wt% Au under the water-gas shift
(WGS) syngas (51 mol% H2, 26 mol% CO2, 21 mol% H2O, and 2 mol% CO) at 400°C, and
observed that Pd/Au alloy membranes can sustain 65% H2 permeability when exposed to 5 ppm
H2S in WGS syngas. Gade et al. [98] used the cold work method to fabricate Pd/Au alloy membranes with various Au content (7 wt% Au, 10 wt% Au, and 20 wt% Au), and tested them under a WGS syngas (51 mol% H2, 29 mol% CO2, 19 mol% H2O, and 1 mol% CO) with 20 ppm
H2S at 400°C. It was found that a Pd/Au alloy membrane with 20 wt% Au content just loses 40%
H2 permeability when exposed to the WGS syngas in the presence of 20 ppm H2S. Chen and Ma
[36] found that the H2 permeance Pd/Au alloy (8 wt%) membranes declined about 80% during the exposure of 54.8 ppm H2S at 400°C; however, it would almost recover in the pure H2 atmosphere at higher operating temperatures. It was supposed that the permeance loss is attributed to the
surface site blocking by the dissociative adsorption of H2S rather than bulk sufidation, and more permeance recovery achieved at higher operating temperature due to exothermic nature of the
dissociative adsorption of H2S on metals.
Platinum, as one of platinum group metals, is well known to have the hydrogen permeation
property as well as the strong resistance to H2S poisoning, and thus there have been research
activities in Pt and Pt composite membranes for H2 separation in the presence of H2S [99,100,101]. Recently, Pd/Pt alloy membranes and Pd/Pt composite membranes have received growing attention due to their outstanding chemical stability that can prevent the bulk sulfidation even
exposed in the pure H2S atmosphere at elevated temperatures [102,103]. Howard and Morreale
[102] prepared Pd4Pt alloy membranes and characterized them in the H2 atmosphere containing
1000 ppm H2S and 10 mol% helium (He). They reported that the H2 flux of membranes displayed
28
Chapter 2 Literature Review
a steep drop followed by a slow recovery over the 120 hours exposure without any treatment for characterized temperatures of 400 and 450°C, respectively. It was supposed that the Pd content in
the membrane surface would gradually reduce during the exposure of H2S due to the formation of
Pd4S along the grain boundaries. As a result, the relative content of Pt would gradually increase in
the membrane surface to improve the resistance to H2S poisoning due to the outstanding chemical stability of Pt. However, there is no proper explanation for the recovery behavior of Pd4Pt membranes that enables the flux to return the initial level before H2S poisoning, and thus further experimentation and investigation are needed.
In summary, Pd alloy membranes have many advantages over pure Pd membranes,
including minimization of the probability of H2 embrittlement, enhancement of H2 permeability,
improvement of mechanical strength, and exhibition of superior chemical stability to resist H2S poisoning. Therefore, the approach of alloying Pd with other potential metals indeed plays an
indispensable role in the development of Pd-based membrane technology for H2 separation, and more efforts are needed to put into Pd alloy research studies for production of high-performance membranes.
29
Chapter 2 Literature Review
2.4 Palladium/Alloy-based Catalytic Membrane Reactor
According to International Union of Pure and Applied Chemistry (IUPAC), a membrane reactor is a device which incorporates the membrane separation process and the chemical reaction step; thus, the separation and the chemical reaction can be carried out simultaneously in a single unit. Based on the membrane function, membrane reactors can be generally classified into several categories [58,104,105]. When a membrane reactor is used for performing a catalyst reaction and the membrane acts as a catalyst and a permselective barrier, this kind of reactor is referred to as the catalytic membrane reactor (CMR). If the membrane is only permselective to certain species without any catalyst function, it would be referred to as the inert catalytic membrane reactor (ICMR). In contrast, when the membrane only acts as a catalyst without any permselectivity, it would be referred to as the non-permselective catalytic membrane reactor (NCMR). According to different types of catalyst bed, both CMR and ICMR can be further categorized into the packed bed catalytic membrane reactor (PBCMR), the fluidized bed catalytic membrane reactor (FBCMR), the packed bed inert catalytic membrane reactor (PBICMR), and the fluidized bed inert
catalytic membrane reactor (FBICMR). The principle of a Pd/alloy-based CMR used for H2 production is illustrated in
Figure 2-5, in which the Pd/alloy composite membrane extends along the length of the packed bed reactor and is in contact with the syngas and catalysts. Since the Pd/alloy composite
membrane is only permselective to H2 and has no any catalyst function to reactions, so that the Pd/alloy-based CMR can be referred to the PBICMR. Today, Pd/alloy-based CMRs for hydrogen production have received a considerable interest due to its great potential for process intensification. Process intensification is one of significant objectives in the field of chemical engineering, since it can provide an approach to optimize capital investments, energy costs, and productivity of chemical plants [106].
30
Chapter 2 Literature Review
Figure 2-5. Schematic diagram of the Pd/alloy-based CMR used for H2 production.
In the natural gas- and coal-based hydrogen production processes, there is a limitation on the syngas conversion due to the thermodynamic equilibrium of the steam reforming reaction and WGS reaction. In practice, changing reaction conditions, such as operating temperature, and/or developing new catalysts can modify kinetics of reactions to improve the conversion. Alternatively, removing products from the reaction mixture during the reaction process can overcome the limitation of thermodynamics by continuously shifting the chemical equilibrium
towards the product, according to Le Chatelier’s principle [107]. Since chemical reactions and H2 separation are allowed to be carried out in a single Pd/alloy-based CMR, higher syngas conversion can be obtained by continuously shifting the chemical equilibrium to the product, endowing Pd/alloy-based CMRs with great potential for process intensification. The application of Pd/alloy- based CMR technology for the interest of shifting the chemical equilibrium towards the product was first proposed by Gryaznov and his coworkers in 1970 [108]. They conducted cyclohexane dehydrogenation using the Pd/alloy-based CMR, demonstrating that the higher overall conversion of cyclohexane could be obtained by the Pd/alloy-based CMR.
31
Chapter 2 Literature Review
2.5 Application of Palladium/Alloy-based Catalytic Membrane Reactor in the Generation of Hydrogen
2.5.1 Hydrogen from Natural Gas
Natural gas reforming also named methane steam reforming (MSR) is a well-established production method that currently generates 95% of hydrogen in the U.S. [21]. In the history of natural gas reforming, the process in which hydrocarbons were converted into H2 by passing over a catalyst of calcium oxide with the steam was first described by Tessie du Motay and Marechal in 1868 [109]. Since then, there has been a substantial expansion of research activity in the hydrocarbon reforming process. Mond and Langer claimed a two-step conversion process using nickel (Ni) as a catalyst in 1889. The first detailed study for the catalytic reaction between methane and steam was published by Neumann and Jacob in 1924 [110]. Shortly thereafter, the reforming process for industrial conversion of natural gas into syngas or hydrogen received significant attention and interest, resulting in numerous patents issued. In 1930, the first steam methane reformer for industrial application was installed and commissioned at Baton Rouge by Standard Oil of New jersey [111].
According to DOE/NETL report [44], a conventional natural gas-based hydrogen production plant with CO2 capture consists of four basic process steps (please see Figure 2-6):
(1) Methane steam reforming. This step is carried out at high temperatures (700- o 1000 C) and mild pressures (3-25 bar) in which methane (CH4) reacts
endothermically with excess steam (H2O) in the presence of a Ni-based catalyst to
produce carbon monoxide (CO), carbon dioxide (CO2), hydrogen (H2) and
unreacted CH4 (the mixture collectively known as syngas) [2,21,22]. Three major reactions in this step are shown in Equation 2-21 to 2-23 [22]. Typically, the use
of an excess of steam in this step can not only enhance CH4 conversion but also prevent coke formation via the Boudouard reaction (Equation 2-23) [44]. Furthermore, desulfurization is conducted using zinc oxide (ZnO) based sorbent before the natural gas reaches the catalyst in the above reforming step [2,44].
32
Chapter 2 Literature Review
MSR reaction I: CH + H O CO + 3H ΔH298K = 206.1 kJ/mol 2-20
4 2 ⇄ 2 WGS reaction: CO + H O CO + H ΔH298K = -41.15 kJ/mol 2-21
2 ⇄ 2 2 MSR reaction II: CH + 2H O CO + 4H ΔH298K = 165.0 kJ/mol 2-22
4 2 ⇄ 2 2 Boudouard reaction: 2CO CO + C ΔH298K = -172 kJ/mol 2-23
⇄ 2 (2) WGS reaction. The CO present in syngas exothermically reacts with steam to
further generate H2 and CO2 (Equation 2-23). Since the exothermic WGS reaction is favored at low temperature, the WGS reaction is typically performed in two reactors: high temperature shift (HTS) and low temperature shift (LTS) reactors. The HTS reactor is operated at 340 to 455°C using a Fe-based catalyst in order to take advantage of the high catalytic reaction rate, while the LTS reactor is operated at 200 to 215°C using a Cu-based catalyst to take advantage of high equilibrium conversion. Approximately 94% CO conversion can be achieved in the HTS reactor, and the converted syngas will be sent to the LTS reactor for completion of the WGS reaction.
(3) CO2 removal. An amine scrubber is used to perform chemical absorption to capture
CO2 in the product stream. According to DOE/NETL report [44,112], attainable
efficiency of CO2 capture by means of chemical absorption can reach a level as high as 95%.
(4) H2 purification. This step is achieved using the pressure swing adsorption (PSA) unit in which solid adsorbents, such as 5A zeolite, alumina, and activated carbon,
adsorb all gas species contained in the product stream, except for H2. The PSA unit
is typically operated at 26 bar to yield 99.99% pure H2 with a recovery efficiency of 80% [44,113].
33
Chapter 2 Literature Review
Figure 2-6. Process block flow diagram of the conventional natural gas-based H2 production plant with CO2 capture, based on the design of the DOE/NETL report [44].
Since Pd/alloy-based CMRs enable chemical reactions and H2 separation to be performed simultaneously in one single unit, the use of Pd/alloy-based CMRs can overcome the equilibrium limitation by continuously shifting the chemical equilibrium to the product, which allows to
achieve the higher CH4 conversion against conventional steam reformers. The application of Pd- based CMRs for MSR has been extensively studied and reported in the literature. Uemiya et al. [114] performed MSR over a Ni catalyst using a Pd-based CMR in which the thin Pd membrane
was supported on cylindrical porous glass. They reported that the CH4 conversion achieved in the
Pd-based CMR exceeds the equilibrium conversion. Furthermore, the level of CH4 conversion increases with increasing operating temperature since higher temperatures would lead to higher H2 permeability of the Pd membrane. Even though high operating pressures were thermodynamically
unfavorable for MSR, higher level of CH4 conversion in the Pd-based CMR can be achieved by
higher operating pressures. It may be attributed to the fact that H2 permeation through the Pd
membrane would increase with increasing the H2 partial pressures in the retentate side based on Sieverts’ law, enhancing the shifting of the chemical equilibrium to the product. Shu et al. [115]
carried out MSR over a Ni/Al2O3 catalyst using a Pd-based CMR in which the Pd membrane was
deposited on the porous stainless steel tube. In light of their studies, higher level of CH4 conversion
34
Chapter 2 Literature Review
can be obtained in the Pd-based CMR over the calculated equilibrium conversion; however, CH4 conversion decreases with increasing operating pressures, contradicting the result from Uemiya et al. [114]. It was supposed that the lower flow rate of the sweep gas used in the study may lead to
a low separation efficiency, resulting in the reduction of CH4 conversion.
To study the effect of H2 permeation of Pd-based membranes on CH4 conversion, Kikuchi
[116] found that the higher the value of H2 permeation, the higher the level of CH4 conversion. As
a result, reducing the membrane thickness and/or increasing the driving force for H2 permeation
and/or improving the permeability of Pd membranes can effectively enhance CH4 conversion, based on Sieverts’ equation (Equation 2-12). Focusing on the driving force for H2 permeation, it
could be effectively enhanced by the use of a sweep gas in the permeate side, and H2 permeation greatly depends on the flow rate of the sweep gas. Consequently, a higher flow rate of the sweep
gas would lead to a higher CH4 conversion, as demonstrated in the literature
[114,115,117,118,119]. To investigate the effect of gas species used as a sweep gas on CH4
conversion, Gallucci et al. [119] observed that that the highest CH4 conversion was obtained when
O2 was used as the sweep gas. This is because H2 would be consumed in the permeate side by the combustion of H2 and O2, thereby increasing the driving force for H2 permeation. However, from
an industrial point of view, the use of steam as a sweep gas is better than the use of O2 due to the
concern of H2 consumption.
The feed composition and the feed flow rate also impose great influence on the CH4
conversion. Nam et al. [118] showed that CH4 conversion declines with increasing the feed flow rate, in particular in the presence of sweep gas. This result was in agreement with the report by
Barbieri et al. [117]. The reason can be attributed to the fact that H2 permeation of Pd membranes would reduce when the feed flow rate increases, lowering the extent of equilibrium shift. Tong et
al. [120] observed that the CH4 conversion would monotonically increase with the value of
H2O/CH4 ratio in the traditional MSR reactor, in agreement with thermodynamic equilibrium
calculation. For Pd-based CMR, the same monotonic increase in the CH4 conversion was also found in the experimental studies [115,117,119,121]. However, an excess of steam would dilute the H2 concentration in the retentate side to lower the driving force for H2 permeation, pointed out by Aasberg-Petersen et al. [122]. In contrast, carbon may be formed in the system to reduce the reforming efficiency and the separation efficiency when the amount of steam in feed is insufficient.
35
Chapter 2 Literature Review
Therefore, appropriate H2O/CH4 ratios should be carefully taken into account in order to achieve high CH4 conversion. According to Laegsgaard Jorgensen et al. [121], carbon formation on the
Ni-based catalyst and the Pd membrane surface can be effectively prevented as the H2O/CH4 ratio is higher than 2.5.
36
Chapter 2 Literature Review
2.5.2 Hydrogen from Coal
Coal gasification is one of major technology pathways for generation of H2, contributing
approximately one-fifth of global H2, as shown in Figure 1-3. In light of DOE/NETL report [44],
a conventional coal-fired hydrogen production plant with CO2 capture consists of five basic process steps (please see Figure 2-10):
(1) Gasification. In this step, pulverized coal reacts with O2 (supplied by an air separation unit (ASU)) in the gasifier at elevated pressure and temperature (56 bar
and 677◦C) to produce syngas, primarily composed of CO, CO2, H2, H2O and other particulate impurities. (2) Particulate removal. Syngas is treated with water in a scrubber to remove chlorides, alkali, and particulate matter. (3) WGS reaction. The CO present in syngas reacts with steam in a series of WGS
reactors (high temperature and low temperature reactors) to convert CO and H2O
into H2 and CO2. (4) Gas clean up. This step involves three process units: the sour water stripper, the mercury removal, and the Selexol unit. First, the syngas passes through the sour
water stripper to remove NH3, SO2, and other impurities from the scrubber. Then,
95% of mercury is removed through a carbon bed. Finally, both H2S and CO2 removal is performed in the Selexol unit. The Selexol unit used in this study is a
dual stage one and the attainable efficiency of CO2 capture can reach a level as high as 90% [44,123].
(5) H2 purification. Solid adsorbents, such as 5A zeolite, alumina, and activated carbon, in the PSA unit adsorb all gas species contained in the syngas, except for
H2. Typically, the PSA unit is operated at 26 bar to yield 99.99% pure H2 with a recovery efficiency of 80% [44,113].
37
Chapter 2 Literature Review
Figure 2-7. Process block flow diagram of the conventional natural gas-based H2 production plant with CO2 capture, based on the design of the DOE/NETL report [44].
The use of Pd/alloy-based CMRs for performing WGS reaction have great potential to achieve higher CO conversion when compared to conventional packed bed reactors due to the shifting of the chemical equilibrium to the product in Pd/alloy-based CMRs. There has been a substantial expansion of research activity in the application of Pd/alloy-based CMRs for WGS reaction.
Kikuchi et al. [124] performed high-temperature WGS reaction over a Fe2O3/Cr2O3 catalyst using a Pd-based CMR in which the Pd membrane (20 µm thick) deposited on the porous glass tube by the electroless plating method. According to the Kikuchi et al. [124], the CO conversion achieved in the Pd-based CMR was as high as 98%, much exceeding 78% of equilibrium conversion. The optimal operating conditions reported in their report were as follows:
400°C system temperature, 5 atm reaction pressure, equimolar ratio of H2O to CO in the feed, 50 ml/min feed flow rate of H2O/CO mixture, 400 ml/min flow rate of sweep Ar in the permeate side, and 1 atm permeate pressure. In addition, it was observed that the CO conversion increased with increasing the value of the time factor (W/F), which was defined by the weight of catalyst (W) and
38
Chapter 2 Literature Review
the feed flow rate of CO (F). It can be attributed to the fact that higher value of W/F would lead to
higher H2 permeation by increasing contact between H2 and the Pd membranes. It was also found that enhancing H2 permeation by increasing the pressure difference between the retentate side and the permeate side can effectively improve the CO conversion due to more chemical equilibrium shifting to the product. Therefore, reducing the membrane thickness and/or increasing the driving
force for H2 permeation and/or improving the permeability of Pd membranes for improving H2 permeation provide an effective way to enhance CO conversion, as demonstrated in the literature [32,125,126,127,128,129,130].
Basile et al. [126] performed low-temperature WGS reaction over a Cu-based catalyst using a Pd-based CMR in which the Pd membrane (0.2 µm thick) was deposited on the porous ceramic tube (γ-Al2O3/α-Al2O3) by the co-condensation method. In light of their report, a maximum CO conversion of 98.89% in the Pd-based CMR can be achieved under 322°C and 1.12 bar, demonstrating that the use of ultrathin Pd membranes required milder operating conditions than the use of thick Pd membranes for achieving the same level of CO conversion reported by Uemiya et al. [125]. Moreover, it was pointed out that the maximum CO conversion was considered as a compromise between the thermodynamic and kinetic considerations, since higher equilibrium conversion of exothermic WGS reaction can be obtained at lower operating temperatures but the reaction rate would decrease with increasing operating temperatures. Therefore, an appropriate operating temperature is one of significant concerns to achieve high CO conversion.
To study the effect of molar ratio of H2O/CO on CO conversion, Uemiya et al. [125] found
that higher molar ratio of H2O/CO would lead to higher CO conversion in the Pd-based CMR. Furthermore, it was experimentally demonstrated that the use of Pd-based CMRs require less steam than the use of conventional WGS reactors to achieve the same level of CO conversion, providing the potential to lower the production cost of hydrogen.
Regarding the effect of feed composition on CO conversion in the Pd-based CMR, Basile et al. [126] showed that the presence of CO2 and N2 in the feed stream would lower the H2 partial
pressure in the retentate side to reduce the driving force for H2 permeation, resulting in lower CO
conversion. Also, the reduction of CO partial pressure by the presence of CO2 and N2 in the feed stream would lower the reaction rate and in turn reduce the CO conversion. Furthermore, Criscuoli
39
Chapter 2 Literature Review
et al. [32] observed that CO conversion greatly depends on the feed composition, and the highest CO conversion was achieved when the feed composition approached to the most thermodynamically favorable one.
To investigate the effect of H2 permeability of membranes on CO conversion, Basile et al. [127] experimented with cold rolled Pd membranes (70 µm) and the cold rolled Pd/Ag alloy membranes (50 µm) supported on porous ceramic tubes for low-temperature WGS reaction. They found that CO conversion obtained using Pd/Ag alloy membranes is higher than that using pure
Pd membranes. It may be attributed to the fact that higher H2 permeability of Pd/Ag alloy membranes would remove more H2 from the retentate side, and thus more chemical equilibrium can be shifted to the product to enhance CO conversion. In addition, Basile et al. [127] pointed out
that if the H2 permeation is lower than the reverse WGS reaction rate, H2 will be consumed rapidly and thus the CO conversion will drop suddenly, as evidenced in their report. Iyoha et al. [129,130] performed the WGS experiment using Pd/alloy-based CMRs with 125 µm thick Pd and Pd/Cu alloy tubes respectively at 900°C in the absence of catalysts, demonstrating that the operating temperature as high as 900°C could allow to carry out WGS reaction without any catalyst. Furthermore, it was observed that the use of Pd or Pd/Cu alloy membranes in CMRs can induce higher CO conversion than equilibrium conversion. However, both CO conversion and H2 recovery achieved by the use of Pd/Cu alloy membranes are lower than that by the use of Pd
membranes, mainly attributed to relative lower H2 permeability of Pd/Cu alloy membranes at 900°C. Iyoha et al. [129,130] also found that CO conversion in both Pd- and Pd/Cu alloy-based CMRs apparently increased when the residence time of the feed stream increase, supposing that the surface of Pd and Pd/Cu alloy membranes is catalytic active for WGS reaction at 900°C.
Studies of the Pd/alloy-based CMR worked at high operating pressures (> 14 bar) were also reported in the literature. Augustine et al. [37] conducted high-temperature WGS experiments
at 14.4 bar without any sweep gas over a Fe2O3/Cr2O3 catalyst in the Pd-based CMR in which Pd membranes were supported on porous Inconel tubes. According to their report, the maximum CO conversion of 98.2% and H2 recovery of 81.2% were obtained at 450°C with a H2O/CO molar ratio of 2.6, demonstrating that the high level of CO conversion can be also achieved under high pressure conditions without the aid of sweep gas. Catalano et al. [30] performed high-temperature WGS
reaction over a Fe2O3/Cr2O3 catalyst using a large-scale Pd-based CMR under high operating
40
Chapter 2 Literature Review
pressure conditions, in which Pd membranes with surface area of 0.02 m2 were supported on porous stainless steel tubes by the electroless plating method. The simulated syngas used in WGS reaction was composed of 20.5% CO, 19.5% H2, 8.6% CO2, and 51.4% H2O. In light of their report, the use of retentate pressure as high as 20 bar and operating temperature of 440°C can
achieve 98.1% CO conversion and 81.5% H2 recovery in the Pd-based CMR without any sweep gas in the permeate side. It was suggested that using high operating pressures in the Pd-based CMR can enable the hydrogen production system to become simpler and more economic due to
elimination of other separation units for mixtures of H2 and sweep gas. Within the above context, the use of high operating pressure provide an economic alternative to achieve high CO conversion.
41
Chapter 2 Literature Review
2.5.3 Approaches for CO2 capture
The major resource for hydrogen production will continuously rely on fossil fuels in the near- to medium-term, and thus the enormous quantities of CO2 emissions will become a significant issue under the raising environmental awareness and stringent environmental policies. According to Le Quere et al. [131], fossil fuel combustion by human activities contributed to 87% of the global CO2 emissions in 2011. In particular, the use of coal is the major contributor with a share of 43% for CO2 emissions. The amount of CO2 emissions from coal increased by 134% from 1975 to 2010 as shown in Figure 2-8 [132]. Furthermore, it is expected that a regulatory action on
CO2 emissions, such as the introduction of a CO2 tax, will be imposed in the U.S. as in many places, such like British Columbia with a tax rate of $30 per tonne of CO2 at 2012, $23 per tonne of CO2 at 2014 in Australia, $11.8 per tonne at 2005 in New Zealand, and $38 per tonne of CO2 at 2000 in Norwegian offshore [133]. Therefore, the global market price of hydrogen have great potential to be subject to a regulatory action on CO2 emissions.
42
Chapter 2 Literature Review
Figure 2-8. Global CO2 emissions from fossil fuel combustion from 1975 to 2010 [132].
Carbon capture and sequestration (CCS) systems are recognized as one practical approach to effectively reduce CO2 emissions [2,43,44]. Generally, carbon capture has fallen into four categories: absorption, adsorption, chemical looping, and membrane technology. Common absorbents used in the physical absorption process involve Selexol, Rectisol and Fluor solvents, which physically absorb CO2 under a condition with high pressures and low temperatures via physical affinities based on Henry’s law, and desorb CO2 at reduced pressures and elevated temperatures [44,112,134,135]. As opposed to physical absorption, chemical absorbents, such as monoethanolamine (MEA), methyl diethanolamine (MDEA) and N-methyldiethanolamine
(MDEA), chemically capture CO2 to form carbamates through a zwitterion mechanism in an absorber, and then are thermally regenerated by desorbing CO2 in a stripper for the cyclic use
[44,112,134,135]. Due to the high CO2 capture efficiency around 90-95% and the ease of
43
Chapter 2 Literature Review
retrofitting into existing chemical/power plants, absorption technology has been commercially mature for many decades [44,112,134,136].
Like adsorption, adsorption also can be distinguished into physisorption and chemisorption by various mechanisms of van der Waals and covalent bonding for interactions between CO2 molecules and the material surface [134,135]. The materials for physical adsorbents primarily involve activated carbons, zeolites, ordered mesoporous silica, and metal-organic frameworks (MOFs), and their adsorption capacities (usually described by adsorption isotherms) greatly depend on their chemical and physical properties as well as internal and external structures [134,135]. For example, adsorption capacities of commercially available zeolites (including 4A, 5A, and 13X) are greatly affected by pore diameter and volume, surface area, charge density, and chemical composition of cations inside their porous structures [135,137]. In the physical
adsorption process, CO2 in flue gas streams is trapped by entering in packed or fluidized beds filled with physical adsorbents, and then desorbed from adsorbents surfaces and concentrated using either vacuum, temperature or pressure swing adsorption cycles [137]. Chemical adsorbents used
for CO2 capture are generally categorized into metal-based adsorbents and hydrotalcite-like compounds in terms of their electronic properties and structures [135]. Metal oxides with a low
ratio for charge to radius (such as Na2O, K2O, CaO and MgO) chemically adsorb CO2 on their basic sites via the exothermic carbonation. Followed by the endothermic calcination, CO2 can be in situ removed from chemical adsorbents and then collected for achieving the capture purpose [138]. Hydrotalcite is a Mg2+ and Al3+ layered double hydroxide. Hydrotalcite-like compounds are well known as a class of anionic clays, which are composed of positively charged brucite-like layers resulting from the substitution of trivalent cations for divalent cations in the octahedral sites [135,137]. The charge-balanced framework of hydrotalcite-like compounds induced by the 2- positive charge compensation with species, such as CO3 anions, endows them with an unique
stable anion-exchange property to chemically adsorb CO2 [135,137]. However, adsorption processes are more expensive when compared to absorption processes due to the low CO2 loading, high-cost raw materials, and relative low kinetics, causing decreased interest in industrial application [136].
Chemical looping is known as a circulating fluidized bed process in which chemical intermediates are reacted and regenerated to split a given reaction into several sub-reactions [139].
44
Chapter 2 Literature Review
Chemical looping systems integrated into hydrogen production plants for CO2 capture have been
proposed and techno-economically evaluated, exhibiting appealing performance on CO2 capture
and H2 production with low capital costs [139]. Basically, there are two different types of chemical looping processes in terms of the particle reaction characteristics. The first type is based on cyclic redox reactions, which includes a reducer and an oxidizer [139,140]. The reducer allows the syngas
stream to convert to H2O and CO2 via the reduction of metal oxide particles, while the oxidizer is used to perform oxidation of reduced metal oxide particles with steam to produce hydrogen. The second type is associated with the exothermic carbonation and endothermic calcination, which involves a carbonation reactor and a calcination reactor [139,141]. In the adsorption process, the carbonation reactor carries out WGS reaction and carbonation reactions of CaO sorbents simultaneously to produce a hydrogen-enriched stream with CaCO3 particles. Subsequently, the
CaCO3 particles are calcinated in the calcination reactor for sorbent regeneration and pure CO2 stream production. Even though chemical looping technology is intensively investigated for many years, operating condition, sorbent stability, and cost of sorbent particles are still the crucial issues for commercialization.
The application of H2-selective membranes for CO2 capture in the hydrogen production system is a state-of-the-art technology. Typically, the CO2 capture process includes two steps
[136]. First, CO2 and H2O with a trace amount of other gas species are left in the retentate stream
from CMRs since the use of H2-selective membranes enable to achieve high level of reactant
conversion and H2 recovery. Subsequently, condensation is carried out in the cold trap to remove the H2O from the retentate stream, and thus a high purity CO2 stream can be obtained for achieving the CO2 capture purpose. Membrane technology for CO2 capture is recognized as a promising
technology option since it can not only theoretically achieve 100% efficiency of CO2 capture but
also have great potential for process intensification and H2 purification [136]. Typically, there are
five categories of H2-selective membranes: dense polymeric membranes, proton conducting membranes, microporous carbon membranes, microporous ceramic membranes and dense metallic membranes [142]. In the present, only microporous ceramic membranes and dense metallic
membranes are practically applied in CMRs due to their superior H2 separation performance as well as outstanding thermal stability [136]. The most common microporous ceramic membranes
are made of SiO2 and Al2O3, while Pd and Pd alloy membranes are the most typical dense metallic membranes used for hydrogen production. Molecular sieving mechanism is used to describe the
45
Chapter 2 Literature Review
transportation of H2 through microporous ceramic membranes, which is different from solution- diffusion mechanism for Pd and Pd alloy membranes [142]. Even though ceramic membranes show better chemical and thermal stability than Pd and Pd alloy membranes, Pd and Pd alloy
membranes possess an unparalleled H2 selectivity of 99.9999% by which the extremely high level of H2 purification performance and CO2 capture efficiency can be readily accomplished. Therefore, Pd/alloy-based CMR technology integrated into hydrogen production plants provides a
promising technology option for process intensification, purification, and CO2 capture in a cost- effective approach [143,144,145,146,147].
46
Chapter 2 Literature Review
2.5.4 Kinetics of Methane Steam Reforming
The model of reaction kinetics for methane steam reforming over a Ni-based catalyst was proposed by Rostrup-Nielsen [110], which is fairly well accepted in the literature and extensively studied. According to Rostrup-Nielsen [110], this proposed model is based a Langmuir- Hinshelwood mechanism by which the procedure of methane steam reforming can be described by following steps:
(1) Water vapor (H2O) reacts with Ni atoms on the catalyst surface, producing adsorbed O
atoms (O*) and H2.
(2) CH4 either reacts with O* or is dissociatively adsorbed on the Ni catalyst surface to form
chemisorbed radicals, such as CH4*, CH3* and CH2*. (3) The total active site concentration is much higher than chemisorbed radicals, so that
carbon-containing radicals (CH4* and CH3*) may react with active sites to generate H radicals (H*). (4) Chemisorbed radicals react with adsorbed O atoms to form carbon monoxide-containing
radicals, such as CO*, CHO* and CH2O*.
(5) H2 is generated by H* recombinative desorption from the catalyst surface. H2 generated by
step 1 and step 5 will be in equilibrium with H* and H2*.
(6) CO and CO2 are yielded by desorption of CO* and CO2*, respectively.
47
Chapter 2 Literature Review
Table 2-1 summarizes chemical reactions in the mechanism of methane steam reforming, in which reactions (g), (h), and (i) are considered as rate-determining steps and corresponding to the MSR reaction I, WGS reaction, and MSR reaction II, respectively [110].
Table 2-1. Mechanism for catalytic reforming of methane.
Chemical Reaction Denotation
H O + H + O (a) ∗ CH2 + ∗ ⟶ CH2 (b) ∗ CH4 + ∗ ⟶ CH4 + H (c) ∗ ∗ ∗ CH4 + ∗ ⟶ CH3 + H (d) ∗ ∗ ∗ CH3 + ∗O⟶ CH2 O + (e) ∗ ∗ ∗ CH2O + ⟶ CO2 + H∗ (f) ∗ ∗ ∗ CHO2 + ∗ ⟶CO + H (g) ∗ ∗ ∗ CO + O∗ ⟶ CO + (h) ∗ ∗ ∗ CHO + O⟶ CO2 +∗ H (i) ∗ ∗ ∗ ∗ 2H H +⟶ 2 (j) ∗ ∗ H ⟶H 2+ ∗ (k) ∗ CO2 ⟶ CO2 +∗ (l) ∗ CO ⟶ CO +∗ (m) ∗ 2 ⟶ 2 ∗
48
Chapter 2 Literature Review
In light of research study by Xu and Froment [148], rate equations for the MSR reaction I,
WGS reaction, and MSR reaction II for methane steam reforming over Ni/MgAl2O4 can be expressed as the following:
P P P P K ,3 MSR reaction I: k CO H2 2-24 R = . CH4 H2O P 1 � DEN− eq 1 � 1 2 5 2 H2 P P P P k K , WGS reaction: R = CO2 H2 CO H2O 2-25 P 3 � DEN− eq 2 � 2 2 H2 P P P P k K , 4 MSR reaction II: 2 CO2 H2 2-26 R = . CH4 H2O P 2 � DEN− eq 3 � 3 3 5 2 H2 P DEN = 1 + K P + K P + K P + K PH2O co CO H2 H2 CH4 CH4 H2O where � H2 �
• P = y P (i = CH , CO, CO , H O, H ), in which P is the partial pressure of species i in bar,
yi is thei gas-phase4 mole fraction2 2 of component2 i, iand P is the total pressure of the system;
• Ki (i = CH , CO, CO , H O, H ) is the adsorption constant of species i;
• Ri (m = 14, 2, 3) denotes2 2 the2 reaction rate for the MSR reaction I, WGS reaction, and 0.5 MSRm reaction II, respectively, with the unit of kmol-bar /kgcat.-h;
• k (m = 1, 2, 3) denotes the reaction rate constant for the MSR reaction I, WGS reaction,
andm MSR reaction II, respectively;
• K , (m = 1, 2, 3) stands for the reaction equilibrium constant for the MSR reaction I, WGSeq m reaction, and MSR reaction II, respectively.
It should be pointed out that the adsorption constant, reaction rate constant, and reaction equilibrium constant can be described using Arrhenius equations. Kinetic parameters for methane steam reforming over a Ni/MgAl2O4 catalyst are summarized in Table 2-2 [31,148,149]. Based on
49
Chapter 2 Literature Review
three rate-determining reactions as mentioned earlier, the formation rate of species ( ) can be described using Equation 2-27: 𝑖𝑖 𝑟𝑟𝑖𝑖
= = 1, 2, 3 2-27
𝑟𝑟𝑖𝑖 Σν𝑖𝑖𝑖𝑖𝑅𝑅𝑚𝑚 𝑤𝑤ℎ𝑒𝑒𝑒𝑒𝑒𝑒 𝑚𝑚 Please notice that is the stoichiometric coefficient of species . When corresponds to a reactant, it is negative,ν𝑖𝑖𝑖𝑖 while corresponds to a product, it would𝑖𝑖 be positive.ν𝑖𝑖𝑖𝑖 Therefore, the formation rate of species are νas𝑖𝑖𝑖𝑖 follows: 𝑖𝑖 = ( + ) 2-28
𝑟𝑟𝐶𝐶=𝐻𝐻4 ( − +𝑅𝑅1 +𝑅𝑅32 ) 2-29 𝑟𝑟𝐻𝐻2𝑂𝑂 =−3 𝑅𝑅1+ 𝑅𝑅2+ 4 𝑅𝑅3 2-30 𝑟𝑟𝐻𝐻2 𝑅𝑅=1 𝑅𝑅2 𝑅𝑅3 2-31 𝑟𝑟𝐶𝐶𝐶𝐶 = 𝑅𝑅1 −+𝑅𝑅2 2-32 𝑟𝑟𝐶𝐶𝐶𝐶2 𝑅𝑅2 𝑅𝑅3
50
Chapter 2 Literature Review
Table 2-2. Kinetic parameters for methane steam reforming over a Ni/MgAl2O4 catalyst [31,148,149].
Pre-factor Value Unit
1 1 : ( ) = , = , , , , ∆𝐻𝐻𝑖𝑖 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝐾𝐾𝑖𝑖 𝑇𝑇 𝐾𝐾𝑖𝑖 𝑇𝑇𝑟𝑟𝑟𝑟𝑟𝑟𝑒𝑒𝑒𝑒𝑒𝑒 � � − �� 𝑤𝑤ℎ𝑒𝑒𝑒𝑒𝑒𝑒 𝑖𝑖 𝐶𝐶𝐶𝐶4 𝐶𝐶𝐶𝐶 𝐶𝐶𝐶𝐶2 𝐻𝐻2𝑂𝑂 𝐻𝐻2 𝑅𝑅 𝑇𝑇𝑟𝑟𝑟𝑟𝑟𝑟 𝑇𝑇
, 40.91 1/bar
𝐾𝐾𝐶𝐶𝐶𝐶,648𝐾𝐾 2.960 × 10 1/bar −2 2 𝐾𝐾𝐻𝐻 648, 𝐾𝐾 1.791 × 10 1/bar −1 4 𝐾𝐾𝐶𝐶𝐶𝐶 ,823𝐾𝐾 4.125 × 10 dimensionless −1 𝐾𝐾𝐻𝐻2𝑂𝑂 823 𝐾𝐾 70.65 kJ/mol ∆𝐻𝐻𝐶𝐶𝐶𝐶 −82.90 kJ/mol ∆𝐻𝐻𝐻𝐻2 −38.28 kJ/mol ∆𝐻𝐻𝐶𝐶𝐶𝐶4 −88.68 kJ/mol ∆𝐻𝐻𝐻𝐻2𝑂𝑂 1 1 : ( ) = , = 1, 2, 3 𝑚𝑚 𝑚𝑚 𝑚𝑚 𝑇𝑇𝑟𝑟𝑟𝑟𝑟𝑟 𝐸𝐸 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝑘𝑘 𝑇𝑇 𝑘𝑘 𝑒𝑒𝑒𝑒𝑒𝑒 � � − �� 𝑤𝑤ℎ𝑒𝑒𝑒𝑒𝑒𝑒 𝑚𝑚 0.5 , 1.842 × 10 𝑅𝑅 𝑇𝑇𝑟𝑟𝑟𝑟𝑟𝑟 𝑇𝑇 kmol-bar /kgcat-h −4 𝑘𝑘1,648𝐾𝐾 7.558 kmol-bar/kgcat-h 0.5 𝑘𝑘2,648𝐾𝐾 2.193 × 10 kmol-bar /kgcat-h −5 𝑘𝑘3 648𝐾𝐾 240.1 kJ/mol 𝐸𝐸1 243.9 kJ/mol 𝐸𝐸2 67.1 kJ/mol 3 , 𝐸𝐸 : , ( ) = = 1, 2, 3 = ∆𝐺𝐺𝑅𝑅𝑅𝑅𝑅𝑅 𝑚𝑚 𝑜𝑜 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝐾𝐾𝑒𝑒𝑒𝑒 𝑚𝑚 𝑇𝑇 𝑒𝑒𝑒𝑒𝑒𝑒 �− � 𝑤𝑤ℎ𝑒𝑒𝑒𝑒𝑒𝑒 𝑚𝑚 𝑎𝑎𝑎𝑎𝑎𝑎 ∆𝐺𝐺𝑅𝑅𝑅𝑅𝑅𝑅 Σ 𝜈𝜈𝑚𝑚𝑚𝑚𝐺𝐺𝑖𝑖 = 75.262 + 7.5925 × 10 × (𝑅𝑅𝑅𝑅) + 1.87 × 10 × ( ) kJ/mol 𝑜𝑜 −2 −5 2 𝐺𝐺𝐶𝐶𝐶𝐶4 = −241.74 + 4.174 × 10 × 𝑇𝑇( 𝐾𝐾) + 7.428 × 10 × 𝑇𝑇(𝐾𝐾) kJ/mol 𝑜𝑜 −2 −6 2 𝐺𝐺𝐻𝐻2𝑂𝑂 = −109.885 9.2218 × 10 𝑇𝑇× 𝐾𝐾( ) + 1.4547 × 10 𝑇𝑇 ×𝐾𝐾 ( ) kJ/mol 𝑜𝑜 −2 −6 2 𝐺𝐺𝐶𝐶𝐶𝐶 = −0 − 𝑇𝑇 𝐾𝐾 𝑇𝑇 𝐾𝐾 kJ/mol 𝑜𝑜 𝐺𝐺𝐻𝐻2 = 393.36 3.8212 × 10 × ( ) + 1.3322 × 10 × ( ) kJ/mol 𝑜𝑜 −3 −6 2 𝐺𝐺𝐶𝐶𝐶𝐶2 − − 𝑇𝑇 𝐾𝐾 𝑇𝑇 𝐾𝐾
51
Chapter 2 Literature Review
2.5.5 Kinetics of the Water-Gas Shift Reaction
The WGS reaction is well known as a reversible exothermic reaction. High operating temperatures are quite often required to achieve a high level of reaction rates; however, the exothermic WGS reaction would be limited by thermodynamic equilibrium under high operating temperature conditions. In other words, when the operating temperature increases, the reaction rate would increase but the equilibrium conversion of CO would decrease since the value of the equilibrium constant (Keq) decreases, as shown in Figure 2-9. Consequently, the WGS reaction is typically carried out into two separate reactors, high temperature shift (HTS) and low temperature shift (LTS) reactors, with different catalysts in order to enhance the overall CO conversion. Please notice that the WGS reaction is independent of operating pressures since the total number of moles in the system does not change during the reaction process.
Figure 2-9. Equilibrium constant of the WGS reaction as a function of temperature.
52
Chapter 2 Literature Review
2.5.5.1 High Temperature Shift Reaction
The WGS reaction is carried out under higher temperature conditions (around 340 to 455°C) using a Fe-based catalyst in order to take advantage of the increase of the reaction rate for improving the CO conversion. According to DOE/NETL report [44], the CO conversion can reach
as high as 94% in the HTS reactor. The Fe-based catalyst is composed of Fe2O3, Cr2O3, and
promoter/stabilizers (e.g. CuO2, Al2O3, MgO, ZnO). Fe2O3 is the primary component in the Fe- based catalyst, and its content generally is higher than the minimum value of 86 wt%. If the content of Fe2O3 is lower than 86 wt%, Fe2O3 can form a homogeneous solid solution with Cr2O3. Cr2O3
and Al2O3 are two significant components that can help minimize the sintering of Fe2O3 to prevent
loss of surface area. Furthermore, Cr2O3 can act as a stabilizer to increase the intrinsic activity of
Fe2O3 [150,151].
For the HTS reaction over a Fe-based catalyst, the reaction rate has been extensively investigated for many years, and a variety of mechanisms has been proposed to describe the catalytic behavior during the reaction process. These mechanisms generally are based on two different models, namely the associative (adsorptive) model and the regenerative (oxidation- reduction) model, as shown in Table 2-3 [152,153,154,155,156]. For the Eley-Rideal and the Langmuir-Hinshelwood mechanisms, even though rate equations based on these mechanisms were reported to match the steady state data of water-gas shift reaction well, actual rate equations during the transient period were much different from that at steady state. This is because the dynamics of the H2 liberation is always slower than CO2 evolution and pretreatment of the catalyst with H2O
also affect the formation of H2 [153]. Hence, the mechanism that incorporates both the associative model and the regenerative model is considered to be more accurate to describe the kinetics in the transient stage. The mechanism proposed by Tinkle and Dumesic [155] is rather similar to that reported by Salmi et al. [156]. The main difference between them is the step for the interconversion of H2O, as seen the reaction (IV.d) and reaction (V.d) in Table 2-3.
53
Chapter 2 Literature Review
Table 2-3. Possible mechanisms for the high-temperature WGS reaction over a Fe-based catalyst.
Mechanism Model Denotation
I. Eley-Rideal mechanism [153] Associative model CO + CO (I.a) ∗ H O +∗ ⟶CO CO + H + 2 (I.b) ∗ II.2 Langmuir⟶-Hinshelwood2 2 mechanism∗ [153] Associative model H O + H O (II.a) ∗ CO2 + ∗ ⟶CO2 (II.b) ∗ CO +∗H⟶O CO + H + 2 (II.c) ∗ ∗ III. [154]2 ⟶ 2 2 ∗ Associative & Regenerative models CO + CO (III.a) r.d.s ∗ H O +∗ ⟶3 2H + O (III.b) CO + O CO ∗+ ∗ 2 ∗ ⟶ (III.c) ∗ ∗ ∗ CO CO⟶+ 2 ∗ (III.d) ∗ 2H2 ⟶ H 2+ 2∗ (III.e) r.d.s ∗ IV. [155⟶ ] 2 ∗ Associative & Regenerative models CO + O CO (IV.a) r.d.s ∗ ∗ CO CO⟶ + 2 (IV.b) r.d.s ∗ H O2 ⟶+ 2H ∗O (IV.c) ∗ H2O +∗ O⟶ 2 2OH (IV.d) ∗ ∗ ∗ 2OH2 2O⟶+ H (IV.e) r.d.s ∗ ∗ V. [156⟶] 2 Associative & Regenerative models CO + O CO (V.a) r.d.s 2 2 ∗ ∗ CO CO + 2 (V.b) r.d.s 2 ⟶ ∗ 2 H O2 ⟶+ 2 H∗ O (V.c) 1 1 ∗ 1 ∗ H2O + + 2 2H + O (V.d) r.d.s 1 ∗ ⟶ 1 2 ∗ 1 2 ∗ ∗ 2H2 2 + H (V.e) r.d.s 1 ∗ ∗ ⟶ ∗ 1 : active⟶ site;∗ : 2active site that selectively adsorbs H atom; : active site that1 selectively adsorbs O atom; r.d.s.: rate-determining step ∗ ∗ 2 ∗
54
Chapter 2 Literature Review
Presently, there are many rate equations derived from proposed mechanisms for the HTS reaction over a Fe-based catalysts reported in the literature; nevertheless, some of them are essentially different and even conflicting with each other. It is supposed that a general expression of kinetics for the HTS reaction does not exist, since there are too many factors to lead the discrepancy between them [150,157]. In contrast to kinetics derived from proposed mechanisms, the empirical power-law rate equation (Equation 2-33) is more acceptable and relatively simple to describe the kinetics of the HTS reaction. However, it should be pointed out that the empirical power-law rate equations are only valid for specific reaction conditions and catalysts.
r = k P P P P (1 ) 2-33 α β γ ϵ co CO H2O CO2 H2 where = P ∙ P ∙ /(P ∙ P ∙ K ∙ ) − Φ
CO2 H2 E CO H2O eq Φk = k exp∙ ( ) ∙ ∙ RTa 0 ∙ − In Equation 2-33, is the reaction rate of CO in gmol/gcat-s, is the partial pressure of species (α+β+γ+δ) in atm, is rate 𝑟𝑟constant,𝐶𝐶𝐶𝐶 denotes the pre-exponential factor𝑃𝑃𝑖𝑖 in gmol/gcat-atm , is 𝑖𝑖the activation𝑘𝑘 energy for the 𝑘𝑘HTS0 reaction in kJ/gmol, and , , , and are adjustable exponents𝐸𝐸𝑎𝑎 of CO, H2O, CO2, and H2, respectively. The activation energies𝛼𝛼 𝛽𝛽 𝛾𝛾and parameters𝜀𝜀 for the empirical power-law rate equation reported in the selective literature are summarized in Table 2-4. Please notice that catalyst particle sizes may affect the measurement of activation energy to obtain different value of activation energy. For example, the larger size of catalyst particles may lead to the lower measured activation energy since diffusion limitation could be effectively reduced in the large catalyst particles [158].
55
Chapter 2 Literature Review
Table 2-4. Activation energies and parameters for the empirical power-law rate equation of the HTS reaction over a Fe-based catalysts.
Catalyst [kJ/mol] ln ( ) (CO) (H2O) (CO2) (H2) Reference
𝑎𝑎 0 Fe2O3/Cr2O3 𝐸𝐸 114.7 8.45𝑘𝑘 𝛼𝛼 0.9 𝛽𝛽 0.25 𝛾𝛾 -0.6 𝛿𝛿 0 [159]
Fe2O3/Cr2O3 40.7 16.92 1 1 0 0 [160]
Fe2O3/Cr2O3 119 7.61 0.81 -0.024 -0.16 -0.044 [158]
Fe2O3/Cr2O3/ Al2O3 79.8 14.78 0.74 0.47 -0.18 0 [161]
Fe3O4/Cr2O3 95 26.1 1.1 0.53 0 0 [153]
Fe3O4/Cr2O3 112 11.7 1 1 0 0 [162] 1 Fe2O3/Cr2O3/CuO 111 6.55 1 0 -0.36 -0.09 [157] 2 Fe2O3/Cr2O3/CuO 88 1.52 0.9 0.31 -0.156 -0.05 [157]
1 Referred to HTC1 in the study by Hla et al. [157]. 2 Referred to HTC2 in the study by Hla et al. [157].
56
Chapter 2 Literature Review
2.5.5.2 Low Temperature Shift Reaction
The WGS reaction is conducted at lower temperatures (180 to 215°C) using a Cu-based catalyst in order to take advantage of equilibrium conversion. Due to the exothermic nature of the WGS reaction, the equilibrium conversion of CO would decrease with increasing operating temperatures. Currently, the most common Cu-based catalyst for industrial LTS reactors is
Cu/ZnO/Al2O3, in which the metallic Cu acts as the active phase while the ZnO and Al2O3 serve as structural promoters and stabilizers that improve Cu dispersion inside the catalyst and prevent Cu from sintering [151].
Uchida et al. [163] investigated the effect of Cu contents in ZnO on the reaction rate of the LTS reaction, and found that the catalytic activity of Cu-based catalysts is greatly dependent on the catalyst composition. Figure 2-10 shows rate constants (k) of the LTS reaction at 180°C in terms of various contents of Cu into ZnO. In light of Figure 2-10, it is clearly observed that the rate constant would reach a maximum value as the Cu/Zn atomic ratio is at 0.4. Furthermore, it has been reported that the catalyst activity is also greatly associated with the Cu dispersion and Cu surface area [164,165]. Even though the metallic Cu provides active sites for the catalytic activity, too high Cu concentration in the catalyst does not necessarily enhance the reaction rate. Uchida et al. [164] observed that the aggregation and the sintering of Cu particles would occur under reaction conditions as the Cu concentration is over a certain value, and thus the available Cu surface area would be reduced to lower the reaction rate. According to Uchida et al. [164], the optimal value of the Cu concentration for the balance between the negative impact of sintering and the positive impact of active sites is at approximately 29 mol% Cu.
57
Chapter 2 Literature Review
Figure 2-10. Rate constants (k) of the LST reaction at 180°C in terms of various Cu/Zn atomic ratios [153].
58
Chapter 2 Literature Review
Much research has been done with respect to the LTS reaction mechanism over a Cu-based catalysts, but arguments still exist in the literature regarding the mechanistic details. Currently, the two most accepted mechanisms are the “Redox mechanism” and “Carboxyl mechanism”, as show in Table 2-5 [166]. The main difference between them is the oxidation process of CO. According to the redox mechanism, CO is adsorbed on the metallic Cu surface first and then oxidized by the atomic O (O ). The atomic O is generated by either the dissociation step of adsorbed hydroxide ∗ (OH ), reaction (1.d), or disproportionation step of two hydroxides, reaction (1.e). However, based ∗ on the carboxyl mechanism, the adsorbed CO (CO ) is oxidized by the OH instead of the O to ∗ ∗ ∗ form the COOH . As a result, the interconversion and oxidation of COOH would generate CO ∗ ∗ ∗ and H for the production of CO2 and H2, as seen in reaction (2.e) and (2.f). 2 ∗ The empirical power-law rate equation (Equation 2-33) for the LTS reaction obtained from experimental results with numerical fitting is more acceptable in the literature when compared to that derived from proposed mechanism. As mention before, the empirical power-law rate equations are only valid for specific reaction conditions and catalysts; otherwise, it cannot be adopted. Table 2-6 lists the activation energies and parameters for the empirical power-law rate equation in terms of various Cu-based catalysts. Ovesen et al. [167] determined activation energies and parameters
of the empirical power-law rate equation over Cu/ZnO/Al2O3 and Cu/Al2O3 catalysts. In light of their study, the fudge factor (ε) was included in order to correct the pressure dependence, since the activation energy slightly changed with the total pressure, as seen in Table 2-6. In addition, it was suggested that the deactivation of catalyst would occur with time of stream, lowering the reaction rate. Thus, the correction of deactivation has been taken into account in parameters for the empirical power-law equation.
59
Chapter 2 Literature Review
Table 2-5. Possible mechanisms for the low-temperature WGS reaction over Cu-based catalyst [166].
Mechanism Denotation
1. Redox mechanism CO + CO (1.a) ∗ H O +∗ ⟶ H O (1.b) ∗ H2O +∗ ⟶ H2 + OH (1.c) ∗ ∗ ∗ OH2 + ∗ ⟶O + H (1.d) ∗ ∗ ∗ OH + ∗OH⟶ H O + O (1.e) ∗ ∗ ∗ ∗ CO + O ⟶CO 2+ (1.f) ∗ ∗ ∗ CO CO⟶+ 2 ∗ (1.g) ∗ H 2+⟶ H 2 H ∗+ 2 (1.h) ∗ ∗ ⟶ 2 ∗ 2. Carboxyl mechanism CO + CO (2.a) ∗ H O +∗ ⟶ H O (2.b) ∗ H2O +∗ ⟶ H2 + OH (2.c) ∗ ∗ ∗ CO2 + OH∗ ⟶ COOH + (2.d) ∗ ∗ ∗ COOH + ⟶CO + H ∗ (2.e) ∗ ∗ ∗ COOH + ∗OH⟶ 2CO + H O (2.f) ∗ ∗ ∗ ∗ CO CO + ⟶ 2 2 (2.g) ∗ H 2+⟶ H 2 H ∗+ 2 (2.h) ∗ ∗ : active ⟶site 2 ∗
∗
60
Chapter 2 Literature Review
Table 2-6. Activation energies and parameters for the empirical power-law rate equation over various Cu-based catalysts.
Catalyst [kJ/mol] ln ( ) (CO) (H2O) (CO2) (H2) Reference
𝑎𝑎 0 Cu/Zn/Cr 𝐸𝐸 - -𝑘𝑘 𝛼𝛼 0.8 𝛽𝛽 0.5 𝛾𝛾 0 𝛿𝛿-0.15 [159]
Cu/ZnO/Al2O3 52.8 15.2 1 1 0 0 [161] † Cu/ZnO/Al2O3 86.5 - 1 1.4 -0.7 -0.9 [167] * Cu/ZnO/Al2O3 78.2 - 1 1.5 -0.7 -0.7 [167]
Cu/Al2O3 59.3 - 1 1.9 -1.4 -0.9 [167]
Cu/ZnO/Al2O3 47.4 12.6 1 1 0 0 [168]
Cu/ZnO/Al2O3 79 9 0.8 0.8 -0.9 -0.9 [169]
† : 5 bar of total pressure; * : 20 bar of total pressure
61
Chapter 2 Literature Review
2.6 Engineering Design Flexibility
Accurate forecasting of the long-term benefit and costs of technological systems is not empirically feasible. What was advanced yesterday may become obsolete tomorrow. That is the reality that designers, investors, and analysts may encounter. To enhance the value of technology systems, we cannot only rely on forecasting to passively deal with uncertainties by robust design, but rather we should actively cope with uncertainties by flexible design.
In the traditional process for engineering design, the objective is to create robust designs that can perform satisfactorily under unexpected circumstances. The approach is to strengthen the design itself for specifications. However, the design itself might be changed in response to external market factors. Furthermore, changes in specifications with uncertainties over a period of time are usually ignored by system designers. General design and evaluation are just based on an unrealistically narrow range of conditions. For instance, the design of conventional coal-fired power plants may focus on: (1) converting coal into electricity in a cost-effective way, (2) matching the expected demand of electricity, (3) supplying and distributing electricity to customers in an economical way, and (4) extending lifetime of plants in order to optimize benefits. However, the design of plants cannot comprehensively cope with all possible changes in the future and deal with them. Any unexpected change may create benefits or losses. For example, state-of-the-art nuclear power plants meet the rapid growth in electricity demand and also have lower levelized costs of electricity when compared to conventional coal-fired power plants. This emerging technology option may change the marketing environment and in turn affect profits of coal-fired power plants even though coal-fired power plants possess robust design characteristics. In contrast, flexible design endows a technological system with an ability to adapt, change, and be reconfigured when facing uncertainties [51]. It can potentially alleviate negative impacts from downside risks and/or enhance benefits on upside opportunities. For instance, coal-fired power plants with flexible design may change their operation or sale mode to adapt current conditions, such as selling syngas, modifying the design of plants to produce chemicals, or shutting down plants until conditions appear favorable.
In economic performance assessment and analysis, financial evaluation methods are employed to determine whether to accept or reject an investment project by ranking possible
62
Chapter 2 Literature Review
choices. The standard methods are generally based on the cash flow of an investment, such as Net Present Value (NPV) and Internal Rate of Return (IRR). Cash flow represents the stream of benefits and costs that would occur in each period of the project under assumed conditions. For example in the NPV method, the evaluation process discounts each future cash flow in each period by a specified discount rate and brings it back to a reference time. Finally, it sums all discounted cash flows with the total capital investment to compute the NPV-metric. Unfortunately, traditional evaluation methods do not offer an explicit way to incorporate and quantify uncertain conditions that could have asymmetric impacts on investment performance into valuations [170]. It is because uncertain conditions with asymmetric impacts, such as product cost and market demand, constantly change with time. Furthermore, traditional evaluation methods do not account for benefits generated by managerial flexibility in response to new circumstances when uncertainties are resolved. As a result, traditional evaluation methods could lead to suboptimal and/or incorrect investment decisions if they are not used appropriately. Within the above context, to enhance evaluation methods with approaches that explicitly assess embedded flexibility into projects in response to uncertainties is certainly justified.
Flexibility for engineering design basically can be categorized into two classes: one is flexibility “on” engineering systems, while the other one is flexibility “in” engineering systems [51,52,54]. Typically, flexibility options “on” engineering systems are recognized as operational flexibility which governs decision rules for system operation and management to deal with uncertainties. Unlike flexibility options “on” engineering systems, flexibility options “in” engineering systems are referred to constructional flexibility that focuses on modification of system design to change system operation modes in response to uncertainties.
Operational flexibility in engineering systems can take various forms, primarily including fuel flexibility, product flexibility, and volume flexibility [53]. Fuel flexibility indicates the capacity to change the primary fuels used in production line in response to fuel market uncertainty. For example in coal-fired hydrogen production plants, the designed gasifier has the potential to handle great variety of primary fuels, involving coal, petroleum coke, biomass, and a blend of these fuels, to produce the desired syngas for hydrogen production. The selection of primary fuels is determined by decision rules in order to prevent or minimize the negative impacts of the fuel market variability. Product flexibility can be deemed as the ability to manufacture different
63
Chapter 2 Literature Review
products on the same production line. This kind of operational flexibility is used to cope with market uncertainty on product sales. For example, membrane reactor modules embedded into Integrated Gasification Combined Cycle (IGCC) power plants (IGCC-MR) [146] provide a promising technical pathway for the co-production of electricity and hydrogen. The option for the plant to sell electricity, hydrogen, or a portfolio of electricity and hydrogen can be governed by decision rules of product flexibility in response to the uncertain market demand, supply, and product unit price. Volume flexibility represents the capability to change total product volumes with respect to uncertain economic, market, and regulatory environments. From a different point of view, volume flexibility can be recognized as the capacity to adjust or switch the system operation modes, such as shutting down the system under an unfavorable economic condition and then reopening it when a favorable economic condition is presented.
Constructional flexibility focuses on changing the original design of the system, such as adding or deleting components [53]. Normally, there are two important concerns in constructional flexibility. One is timing for restructuring the system, while the other one is the initial system design for the ease of future system restructure [53,54]. Both of them are associated with capital investments for system restructure as well as system performance after restructure; they primarily determine the value-enhancing performance of constructional flexibility. Therefore, in order to maximize the value-enhancing performance of constructional flexibility, these two concerns should be carefully evaluated and balanced in response to various uncertainties. Certainly, constructional flexibility can be combined with operational flexibility, and the system performance may be further enhanced through the combination of these two different types of engineering design flexibility options [54].
64
Chapter 3
Chapter 3 Techno-Economic Performance Evaluation Techno-Economic Performance Evaluation
3.1 Technical Performance Evaluation Framework
The technical performance of industrial scale Pd/alloy-based CMR modules potentially integrated into hydrogen production (HP) plants is evaluated with the aid of a model accompanied by the following simplifying assumptions:
• All gas species follow the ideal gas law. • Isothermal conditions apply to the module’s operation. • Isobaric conditions apply to both retentate and permeate sides.
• The membrane exhibits theoretically infinite selectivity of H2 while the H2 permeation flux of the membrane follows Sieverts’ law as shown in Equation 3-1. • Plug flow conditions apply to both retentate and permeate streams. Note that radial and axial dispersions are neglected. • There are no mass transfer limitations or concentration polarization. • No poisoning of the catalyst occurs during the WGS reaction.
65
Chapter 3 Techno-Economic Performance Evaluation
The Pd/alloy-based CMR considered in the one-dimensional model is a shell and tube configuration as seen in Figure 3-1. The Pd/alloy composite membrane tube is placed at the center of a stainless steel shell casing, mounted with the cap in the end of shell casing. The entrance of membrane tube at the membrane side is blocked with a cap, while the other side with the tube exit is located at the end of reactor. The Pd/alloy composite membrane is on the outer surface of membrane tube, and the annular space between the shell casing and the membrane is filled with high temperature water-gas shift (WGS) catalyst. The remaining space of the reactor is filled with inert packing material, such as quartz sand. In Figure 3-1, radial and axial directions are denoted by r and L, respectively. In addition, r1 and r2 represent the outer radius of the membrane and inner radius of the shell casing, respectively.
Figure 3-1. Schematic diagram of the Pd/alloy-based CMR used in the one-dimensional modeling framework.
66
Chapter 3 Techno-Economic Performance Evaluation
The membranes used in the CMRs are Pd/Au alloy (12 wt% Au) composite membranes for
the purpose of retaining high permeance levels in the presence of trace amounts of H2S, as well as ensuring permeance recoverability whenever feed conditions with high sulfur concentration levels occur. The permeance of the membranes is affected by the presence of Au. Therefore, a quite 3 2 standard Equation 3-1 is used to describe the temperature-dependent H2 flux (J ) in m /m -h-
bar0.5: H2
Q ( P P ) per J = ret 3-1 ∙ � H2 − � H2 H2 ϕ HSS poisoning ∙ ϕAu alloying δ E where Q = Q exp R T − Q 0 ∙ � � ∙ In Equation 3-1, is the permeance decline coefficient due to exposure to 2
S ppm H2S at 400°C assumedϕH inS poisoning the present study, is the permeance decline coefficient due to the alloying of 12 wt% Au, δ is the Pd/AuϕAu alloying membrane thickness in μm, Q is the H2 3 2 0.5 3 permeability of pure Pd foils in m -μm/(m -h-bar ), Q0 is the H2 permeability constant in m - 2 0.5 μm/(m -h-bar ), EQ represents the activation energy for H2 permeation in kJ/gmol, R is the gas
constant in J/(gmol-K), T is the operating temperature in K, and P and P represent the H2 ret per partial pressure (bar) in the retentate side and permeate side respectively.H2 H 2Please notice that, membranes with thickness values ranging from 5 µm to 10 µm have been synthesized at WPI’s Center for Inorganic Membrane Studies and their satisfactory performance levels have also been experimentally confirmed [30,38,39,171,172]. Furthermore, the reaction temperature is considered as high as 400°C in this modeling framework for the purpose of maintaining high
efficiency of CMRs and prolonging the lifetime of membranes, even though H2 permeation through the membrane can be promoted by higher temperature conditions. It is because the pinhole formation would occur at temperature conditions higher than 400 to 450°C due to the incoherent sintering of the small Pd clusters and/or Pd crystallites [35].
Regarding the kinetics for the high temperature WGS reaction, the empirical power-law rate equation (Equation 2-33) is employed in this modeling framework. Activation energies and
67
Chapter 3 Techno-Economic Performance Evaluation
parameters for the empirical power-law rate equation are based on the commercial high temperature catalyst HTC1 (Fe2O3/Cr2O3/CuO) reported by Hla et al. [157]. Associated with
porosity of catalyst (θ) and void fraction catalyst bed (ε) in the CMR, the WGS reaction rate (RWGS) can be expressed by Equation 3-2 with the unit of gmol/m3-s. Please notice that, even though the
equilibrium constant (Keq) for the WGS reaction has been extensively studied in the past decades, the standard Arrhenius equation (Equation 3-3) is the most used commonly and is also used in this modeling framework [173].
111000 R = (1 ) d (1 ) 1000 10 . exp RT 2 845 − WGS − ε ∙ cat ∙ − θ ∙ ∙ P ∙ P � � 3-2 P P . P . 1 K PCO P −0 36 −0 05 CO2 H2 CO CO2 H2 ∙ ∙ ∙ ∙ ∙ � − 2 � eq ∙ ∙ H O
4577.8 K = exp( 4.33) T 3-3 eq − For each gas species involved in the WGS reaction, the reaction rate can be obtained by
multiplying the stoichiometric coefficient of gas species with RWGS, as displayed in the following:
r = ( 1) RWGS 3-4 rCO = ( 1) R H2O − ∙ WGS 3-5
rCO2 = (1−) R∙WGS 3-6 r = (1) R H2 ∙ WGS 3-7 ∙
where r , rH2O, rCO2, and rH2 are the reaction rates of CO, H2O, CO2, and H2, respectively, in gmol/gcatCO-s.
The modeling framework was realized by a comprehensive mathematical model which is composed of a system of mass balance equations for all gas species considered. Ordinary differential equations are used to describe steady-state mass balances for each species in both
68
Chapter 3 Techno-Economic Performance Evaluation
retentate and permeate sides, which can be derived from the principle of conservation of mass as shown in the following [174]:
(I) Mass balance equations for gas species along the length of the reactor in the retentate side:
dF _ = (r r ) r r J 3-8 dL ret H2 2 2 π ∙ 2 − 1 ∙ H2 − 2π 1 ∙ H2
dF _ = (r r ) r i = CO, H2O, CO2 3-9 dL ret i 2 2 π ∙ 2 − 1 ∙ H2 where F _ and F _ denote the molar flow rate of H2 and gas species considered in the retentate
side withret theH2 unit ofret gmol/si .
(II) Mass balance equations for H2 along the length of the reactor in the permeate side:
dF _ = r J 3-10 perdL H2 2π 1 ∙ H2
where F _ represents the molar flow rate of H2 in the permeate side with the unit of gmol/s.
2 Regardingper H the boundary conditions in the one-dimensional modeling framework, the molar flow rate of each gas species at the inlet in the catalytic zone remains constant for all times, as shown in Equation 3-11:
dF = 0 j = H2, CO, H2O, CO2 3-11 dLretj
At the exit of the catalytic zone, the molar flow rate of each gas species has no further change when it enters the interface between the inert packing and the catalytic zone, as described in Equation 3-12:
69
Chapter 3 Techno-Economic Performance Evaluation
dF = 0 j = H2, CO, H2O, CO2 3-12 dLretj
For the boundary condition in the permeate side, the molar flow rate of H2 at the inlet is zero
(Equation 3-13) since there is no H2 permeation through the Pd/alloy membrane.
F _ | = 0 3-13
per H2 L=0
The CO conversion (XCO) and the H2 recovery (R ) in the modeling framework are calculated
using Equation 3-14 and 3-15, respectively. H2
Fret_CO|L=0 Fret_CO|L=L XCO = 3-14 Fret_CO|L=0 −
F _ |L=L R = 3-15 F _ |L=0per +H2 F _ |L=0 H2 ret CO ret H2 Finally, the set of all equations in one dimensional modeling framework are numerically integrated with the aid of a Runge-Kutta-Fehlberg 56 (RKF56) integration algorithm in version 6.2 Polymath software package. The parameters of the membrane and high-temperature shift catalysts used in the aforementioned modeling framework are summarized in Table 3-1. Please notice that this modeling framework based on steady state mass balance equations is in very good agreement with the experimental results reported by Augustine et al. [37].
70
Chapter 3 Techno-Economic Performance Evaluation
Table 3-1. Parameters of Pd/Au alloy composite membranes and high-temperature shift catalysts used in the process modeling framework.
Pd/Au alloy membrane thickness, δ [μm] 6 3 2 0.5 a H2 permeability constant, Q0 [m -μm/(m -h-bar )] 6281.2 a Activation energy for H2 permeation, EQ [J/gmol] 15630 b Permeance decline coefficient due to exposure to 2 ppm H2S at 400°C, ϕH2S poisoning 0.6 b Permeance decline coefficient due to the alloying of 12 wt% Au, ϕAu alloying 0.58 3 c Solid catalyst density, dcat [kg/m ] 2628 Porosity of catalyst, θ [%]c 50 Void fraction of catalytic bed, ε [%]d 20
a Referred to the study by Ayturk et al. [31] b Referred to the PhD dissertation by Chen [175] c Referred to the study by Adam and Barton [173] d Referred to the study by Koc et al. [146]
Within this context, the CMR model was simulated for industrial conditions under the feed
specifications [146], reaction conditions, and H2 permeance of membranes listed in Table 3-2. Furthermore, the process system dimensions and the required inlet stream for CMR modules are
based on the H2 production target of 616.5 tonnes per day in conventional coal-fired H2 production plants for comparison purposes [44]. The CMR system consists of 10 bundles of CMRs in a parallel configuration with 4313 membrane tubes. This process system is capable of achieving
99% CO conversion and 96% H2 recovery, and its specifications used for cost analysis are
summarized in Table 3-3. A conservative 98% CO2 capture efficiency is assumed to avoid overestimating the associated economic performance of HP-CMR plants, even though in practical use, higher efficiencies can be currently achieved by membrane technology [146].
71
Chapter 3 Techno-Economic Performance Evaluation
Figure 3-2. Schematic of a catalytic membrane reactor (CMR) containing three Pd/Au membranes for Water-Gas Shift reaction.
Table 3-2. Feed specifications, reaction conditions, and H2 permeance of membranes.
Retentate pressure [bar]a 50 Permeate pressure [bar]a 1 Reaction temperature, T [°C] 400 Inlet stream [gmol/s] 8270 a Inlet composition [%] CO: 23.0 – CO2: 8.8 – H2: 22.1 – H2O: 46.1 Retentate outlet stream [gmol/s] 4702
Retentate outlet composition [%] CO: 0.4 – CO2: 55.6 – H2: 3.1 – H2O: 40.9 3 2 0.5 H2 permeance [m /(m -h-bar )] 22.3
a Referred to the study by Koc et al. [146]
72
Chapter 3 Techno-Economic Performance Evaluation
Table 3-3. Industrial scale CMR module specifications used for cost analysis.
Single membrane tube Industrial scale CMR module
Pd thickness [μm] 6 Total Pd weight [kg] 1260.6 Au thickness [μm] 0.51 Total Au weight [kg] 171.9
Outer radius of the membrane, r1 Number of bundle CMRs in the 2.54 10 [cm] module
Inner radius of the shell casing, r2 Number of membrane tubes in each 3.81 4313 [cm] CMR 3 Membrane tube length, L [m] 2.54 Total Vshell [m ] 499 Membrane area [cm2] 4052 Total membrane area [m2] 17474 3 3 Vannulus [cm ] 6432 Total Vannulus [m ] 277
Wcatalyst [kg] 6.76 Total Wcatalyst [tonne] 291.6
Finlet [gmol/s] 0.192 Total Finlet [gmol/s] 8270
Fper_H2 [gmol/s] 0.083 Total Fper_H2 [TPD] 616.5
Fret_CO2 [gmol/s] 0.061 Total Fret_CO2 [TPD] 9935.8
TPD = Tonne Per Day
73
Chapter 3 Techno-Economic Performance Evaluation
3.2 Economic Performance Assessment and Cost Analysis Framework
3.2.1 Capital Investment and Total Product Cost Baseline Estimation
3.2.1.1 An Industrial-Scale Pd/Alloy-based CMR Module for WGS Integrated into Natural Gas-based Hydrogen Production Plants
In the present study, the industrial-scale data presented in the report by Criscuoli et al. [176] are used in the proposed economic performance assessment and cost analysis framework for membrane reactor modules integrated in natural gas-based hydrogen production plants. Figure 3-3 presents the process block flow diagram of HP-CMR plant with CO2 capture, based on the design
of the DOE/NETL report [44]. In particular, a rather conservative 85% H2 recovery performance target level is considered and the pertinent set of feed and reaction conditions as well as the process system dimensions and the requisite flow rate values conforming to industrial scale load specifications are given in Table 3-4 [176]. The Pd-based industrial scale membrane reactor module considered in this study consists of 2790 membrane tubes capable of attaining the 85% H2 recovery target value. Furthermore, palladium/gold (Au) membranes are also considered with a fixed composition of 12% wt in gold.
Table 3-5, Table 3-6, and Table 3-7 encompass key model inputs. Cost figures involving the industrial-scale membrane reactor module and assorted equipment were adopted from the report by Criscuoli et al. [176] and the DOE/NTEL report by Chou and Kuehn [44]. Please note that equipment costs included in the present study do not take into account the cost-reducing potential of future technological advances and ‘learning’. It should also be pointed out that the cost figures obtained from the above bibliographical sources were updated with the aid of Chemical Engineering plant cost indexes [177] in order to obtain the most recent equivalent cost figures (Table 3-7).
The various cost components of the analysis conducted and all pertinent details of the associated baseline total capital investment (TCI) and total product cost (TPC) models can be found in Table 3-8 and Table 3-9, respectively. Standard practice in engineering economic analysis
74
Chapter 3 Techno-Economic Performance Evaluation requires the calculation of the fixed capital investment (FCI) cost first as the sum of all direct and indirect costs, followed by the calculation of the TCI as the sum of FCI and working capital [178]. Furthermore, the TPC is calculated by forming the sum of manufacturing cost, general expenses as well as membrane and catalyst replacement costs [178]. In particular, Table 3-8 and Table 3-9 encompass a detailed list of all cost components (FCI/TCI/TPC model inputs) and the displayed numerical estimates correspond to the average/expected values (representing the baseline case) of the probability distributions of the various model inputs considered as inherently uncertain in the sequel. Please note that a relatively thick palladium layer of 75 μm was used in the above baseline calculations in agreement with the study by Criscuoli et al. [176] for comparison purposes, even though further cost reduction would be possible by decreasing the palladium layer thickness. Please note that as key components of FCI (pertaining for example to installation and engineering and supervision) depend on equipment cost, the FCI cost of the membrane reactor module increases proportionally with the reactor equipment cost. Furthermore, membrane replacement costs are explicitly taken into account and calculated by dividing the membrane reactor module equipment cost by the membrane lifetime. It should be pointed out that membrane reactors have to be tested under actual reaction conditions within a long-term horizon to determine more accurately the membrane lifetime. Evidently a higher membrane lifetime would lead to a reduction in replacement costs. Finally, please note that neither the palladium recovery nor the membrane reactor leasing option was considered in the present study; these options, however, would almost certainly lead to additional economic benefits.
75
Chapter 3 Techno-Economic Performance Evaluation
Figure 3-3. Process block flow diagram of the natural gas-based H2 production plant integrated with membrane technology.
76
Chapter 3 Techno-Economic Performance Evaluation
Table 3-4. Industrial scale membrane reactor module specifications, feed specifications and reaction conditions [176].
Inlet stream [kgmol/h] 246
Inlet composition [%] CO: 11.23‒H2O: 22.46‒CO2: 9.88‒ H2: 55.39‒CH4: 1.04 Retentate outlet stream [kgmol/h] 108
Retentate outlet composition [%] CO: 1.292‒H2O: 26.864‒CO2: 46.775‒ H2: 22.701‒CH4: 2.368
Permeate H2 stream [kgmol/h] 138 Temperature [K] 595 Retentate pressure [atm] 16 Permeate pressure [atm] 1 Catalyst [kg] 1866 Permeabiliy [kgmol/m2-h-atm0.5] 1.87 × 10-5 Palladium thickness [µm] 75 Internal diameter of membrane tubes [m] 0.02 Length of membrane tubes [m] 2.6 Number of membrane tubes 2790 Total membrane area [m2] 456
Table 3-5. Cost data used in the economic performance assessment framework (Cost base: 2012).
Raw material [$/year] 2,088,854 Catalyst [$/kg] 5 Steam [$/ton year] 119,454 Compressor (from 1 to 16 atm) [$] 580,674
77
Chapter 3 Techno-Economic Performance Evaluation
Table 3-6. Electricity, steam consumption and labor cost data (Cost base: 2012).
Conventional CMR technology option Electricity consumption [kW-h] - 2,616,897 Steam consumption [Ton/year] 4.9 1.0 Labor hour [h] 8,844 8,844
Table 3-7. Economic parameters used in cost analysis [177].
Euro to US Dollar exchange rate In 2000 1.083 Chemical engineering plant cost indexes In 2000 394.1 In 2001 394.3 In 2002 395.6 In 2003 402.0 In 2004 444.2 In 2005 468.2 In 2006 499.6 In 2007 525.4 In 2008 575.4 In 2009 521.9 In 2010 550.8 In 2011 585.7 In 2012 584.6 In 2013 567.3 In 2014 576.1
78
Chapter 3 Techno-Economic Performance Evaluation
Table 3-8. Estimation of total capital investment for the membrane reactor module (Cost base: 2012).
Cost Component Cost: US$ I. Direct Costs A. Equipment + installation + instrumentation + piping + electrical + insulation + painting 1. Purchased equipment WGS reactor 525,497 HTS catalyst 9,675 Palladium membrane 8,948,562 Compressor 580,674 Total membrane module 10,064,408 2. Installation (40% of A1.) 4,025,763 3. Instrumentation and controls, installed (29% of A1.) 2,918,678 4. Piping, installed (45% of A1.) 4,528,984 5. Electrical, installed (25% of A1.) 2,516,102 Equipment total 24,053,935 B. Buildings, process and auxiliary (40% of A1) 4,025,763 C. Service facilities and yard improvements (70% of A1) 7,045,085 D. Land (6% of A1) 603,864 Subtotal 35,728,648 II. Indirect Costs A. Engineering and supervision (17.5% of direct cost) 6,252,513 B. Legal expenses (2% of fixed-capital investment) 1,150,169 C. Construction expense and contractor's fee (15% of fixed-capital 8,626,266 investment) D. Contingency (10% of Fixed-capital investment) 5,750,844 Subtotal 21,779,792 III. Fixed capital investment (= Direct costs + Indirect costs) 57,508,440 IV. Working capital ( 15% of V) 10,148,548 V. Total capital investment ( = III + IV) 67,656,988 HTS: high-temperature shift; WGS: water-gas shift
79
Chapter 3 Techno-Economic Performance Evaluation
Table 3-9. Estimation of total product cost (TPC) for the membrane reactor module (Cost base: 2012).
Cost Component Cost: US$ I. Manufacturing cost (= Direct production costs + Fixed charges + Plant overhead costs) A. Direct production costs 1. Raw materials 2,088,854 2. Operating labor 311,701 3. Direct supervisory and clerical labor (15% of operating labor) 46,755 4. Utilities 380,515 5. Maintenance and repair (6% fixed-capital investment) 3,450,506 6. Operating supplies (0.75% of fixed-capital investment) 431,313 7. Laboratory charges (15% of operating labor) 46,755 8. Patents and loyalties (3% of product cost) 878,225 Total direct production costs 7,634,626 B. Fixed charges 1. Depreciation (10% of fixed-capital investment) 5,750,844 2. Local taxes (2.5% of fixed-capital investment) 1,437,711 3. Insurance (0.7% of fixed-capital investment) 402,559 Total fixed charges 7,591,114 C. Plant overhead costs (10% of product cost) 2,927,418 Subtotal 18,153,158 II. General expenses A. Administrative costs (3.5% of product cost) 1,024,596 B. Distribution & selling costs (11% of product cost) 3,220,160 C. R&D costs (5% of product cost) 1,463,709 D. Financing interest (8% of total capital investment) 5,412,559 Subtotal 11,121,025 III. Total product cost A. Membrane replacement 2,982,854 B. HTS catalyst replacement 1,935 C. Product cost (= Manufacturing cost + General expenses) 29,274,183 Total 32,258,972 HTS: high-temperature shift
80
Chapter 3 Techno-Economic Performance Evaluation
3.2.1.2 An Actual Large-Scale Pd/Alloy-based Separation Module for Hydrogen Purification
In the present study, baseline capital investment models for FCI and TCI estimation of a large-scale Pd/alloy-based separation module has been developed to evaluate the effect of the lifetime and thickness of the membrane, following the standard practice in engineering system economic analysis [178]. Within the modeling framework, the FCI is calculated by adding direct costs with indirect costs while the TCI is computed as the sum of the FCI and the working capital (WC). All cost model inputs considered for the FCI/TCI estimation are listed in Table 3-10. Please notice that the cost figures of purchased equipment involving the Pd/alloy-based membrane module and assorted apparatus, such as pressure monitor system and the mass flow control system, are adopted from the large-scale Pd-based separation module built in Worcester Polytechnic Institute and specifications shown in Table 3-11 [30]. Note that the ratios with specific ranges and bases indicated in Table 3-10 are approximations that reflect the effect of many factors in uncertain future states on FCI/TCI, such as module location, type of process, and complexity of equipment. In addition, the Chemical Engineering Plant Cost Indexes (Table 3-7) were used to update all pertinent cost estimates by taking into account the time value of money for obtaining the most recent equivalent cost figures [177]. It should be pointed out that cost indexes are used only to offer general, rather conservative estimates without considering any reasonable and feasible cost- reducing possibility of future technological advances. Motivated by the fact that TPC and levelized cost LC are recognized as two significant indicators broadly applied in the economic performance analysis of chemical plants, the baseline TPC/LC models for hydrogen production via the large- scale Pd-based separation module are proposed. The various cost components and pertinent details of the associated baseline TPC/LC models are shown in Table 3-12. According to the standard practice in engineering economic analysis, the TPC is calculated by forming the sum of manufacturing cost, general expenses, and membrane replacement, while the LC is obtained from the TPC divided by the hydrogen production through the Pd-based separation module. Please notice that the cost of membrane replacement considered in the proposed baseline models is calculated by the Pd/Au alloy composite membrane cost over the membrane lifetime. Furthermore, the values of H2 production in Kg per year used for LC estimation are calculated based on H2 permeance of Pd/Au alloy composite membranes from the experimental data.
81
Chapter 3 Techno-Economic Performance Evaluation
Table 3-10. Estimation of capital investment costs for the large-scale Pd-based separation module.
I. Direct Costs A. Equipment + installation + instrumentation + piping + electrical + insulation + painting 1. Purchased equipment a. Pressure monitor system (1) Manometer (range: 25000 Torr) × 2 (2) Manometer (range: 5000 Torr) × 1 (3) Digital power supply and readout × 2 b. Mass flow control system (1) Mass flow controller × 4 (2) Mass flow indicator × 2 (3) Pressure controller × 2 (4) Flow-Bus interface box × 1 (5) Digital readout/control unit blind front × 2 c. Temperature control system (1) Temperature controller × 2 (2) Profile Probe × 2 (3) Thermocouple differential analog input module × 3 (4) CompactDAQ chassis × 1 d. Computer e. Preheater (1) Shell casing (2) Ceramic fiber heater f. Pd-based membrane reactor (1) Pd/Au alloy composite membrane i) Pd cost ii) Au cost iii) 316L SS support (2) Shell casing (3) Ceramic fiber heater 2. Installation (25-55% of purchased equipment) 3. Instrumentation and controls, installed (8-50% of purchased equipment) 4. Piping, installed (10-80% of purchased equipment) 5. Electrical, installed (10-40% of purchased equipment) B. Buildings, process and auxiliary (10-70% of purchased equipment) C. Service facilities and yard improvements (40-100% of purchased equipment) D. Land (4-8% of purchased equipment) II. Indirect Costs
82
Chapter 3 Techno-Economic Performance Evaluation
A. Engineering and supervision (5-30% of direct cost) B. Legal expenses (1-3% of fixed capital investment) C. Construction expense and contractor's fee (10-20% of fixed capital investment) D. Contingency (5-15% of fixed capital investment) III. Fixed Capital Investment ( = Direct Costs + Indirect Costs ) IV. Working Capital ( 10-20% of Total Capital Investment ) V. Total Capital Investment ( = Fixed Capital Investment + Total Capital Investment )
Table 3-11. Large-scale Pd-based separation module specifications used for cost analysis.
Outer diameter of the support tube [m] 0.025 Length of the support tube [m] 0.356 Length of the porous part [m] 0.254 Outer diameter of the shell casing [m] 0.051 Wall thickness of shell casing [m] 0.00305 Inner diameter of the shell casing [m] 0.045 Length of the shell casing [m] 0.914 Membrane area [m2] 0.0203 Volume for shell casing [m3] 0.00143
83
Chapter 3 Techno-Economic Performance Evaluation
Table 3-12. Estimation of total product cost and levelized cost for the large-scale Pd-based separation module.
I. Manufacturing Cost (= direct production costs + fixed charges + plant overhead costs) A. Direct production costs 1. Raw materials (10-20% of product cost) 2. Operating labor (10-15% of product cost) 3. Direct supervisory and clerical labor (10-20% of operating labor) 4. Utilities (10-15% of product cost) 5. Maintenance and repair (2-10% fixed capital investment) 6. Operating supplies (0.5-1% of fixed capital investment) 7. Laboratory charges (10-20% of operating labor) 8. Patents and loyalties (0-6% of product cost) B. Fixed charges 1. Depreciation (10% of fixed capital investment) 2. Local taxes (1-4% of fixed capital investment) 3. Insurance (0.4-1% of fixed capital investment) 4. Financing interest (6-10% of total capital investment) C. Plant overhead costs (5-15% of product cost) II. General Expenses A. Administrative costs (2-5% of product cost) B. Distribution & selling costs (2-20% of product cost) C. R&D costs (5% of product cost) III. Total Product Cost (= membrane replacement + product cost) A. Membrane replacement (= Pd/Au alloy composite membrane / membrane lifetime) B. Product cost (= manufacturing cost + general expenses) IV. Levelized Cost (= total product cost / H2 production)
84
Chapter 3 Techno-Economic Performance Evaluation
3.2.1.3 An Actual Large-Scale Pd-based CMR Module for Water-Gas Shift Reaction
In the baseline functional FCI/TCI model formulation, the FCI is calculated as the sum of all direct and indirect costs while the TCI is estimated by combining the FCI with the working capital (WC) in compliance with standard practices in economic analysis of engineering systems [178]. Table 3-13 and Table 3-14 shows the 21 cost model inputs considered in the FCI/TCI estimation along with their corresponding spread range. Please notice that the cost figures of purchased equipment involving the Pd-based CMR and assorted apparatus, such as the micro gas chromatography (GC) system, the high pressure liquid metering pump, and the pressure monitor system, are adopted from the actual large-scale CMR module built in the facilities of Worcester Polytechnic Institute (WPI) and shown in Figure 3-4 [30] with their specifications listed in Table 3-15. Furthermore, the range of percentages indicated in Table 3-13 and Table 3-14 are approximations, reflecting the effect of many factors in uncertain future states on the FCI/TCI, including the module location, type of process, and complexity of the equipment. Furthermore, the cost figures obtained from the actual large-scale CMR module are updated by means of the Chemical Engineering Plant Cost Indexes (Table 3-7) in order to obtain the most recent equivalent cost figures [177].
85
Chapter 3 Techno-Economic Performance Evaluation
Figure 3-4. Schematic of the actual large-scale Pd/alloy-CMR module built at Worcester Polytechnic Institute (WPI) [30].
86
Chapter 3 Techno-Economic Performance Evaluation
Table 3-13. Estimation of production costs for a Pd composite membrane. †
I. Electroless plating setup cost II. Production costs 1. Raw materials (a) Pd cost (b) 316L SS support 2. Operating labor (12.5-50% of raw materials) 3. Direct supervisory and clerical labor (10-20% of operating labor) 4. Utilities (12.5-50% of raw materials) 5. Maintenance and repair (2-10% of electroless plating setup cost) 6. Operating supplies (0.5-1% of electroless plating setup cost) 7. Laboratory charges (10-20% of operating labor) 8. Patents and loyalties (0-15% of raw materials)
† Please notice that the 316L SS support cost is updated to 2014 by Chemical Engineering Plant Cost Indexes.
87
Chapter 3 Techno-Economic Performance Evaluation
Table 3-14. Estimation of Capital Investment for a large-scale Pd-based CMR module. ‡
I. Direct Costs A. Equipment + installation + instrumentation + piping + electrical + insulation + painting 1. Purchased equipment a. Micro GC system (1) Micro GC × 1 (2) Trap (H2O) × 2 (3) Big universal trap (He) × 2 b. High pressure liquid metering pump c. Pressure monitor system (1) Manometer (range: 25000 Torr) × 2 (2) Manometer (range: 5000 Torr) × 1 (3) Digital power supply and readout × 2 d. Drain valve e. Mass flow control system (1) Mass flow controller × 4 (2) Mass flow indicator × 2 (3) Pressure controller × 2 (4) Flow-Bus interface box × 1 (5) Digital readout/control unit blind front × 2 f. Chiller g. Temperature control system (1) Temperature controller × 2 (2) Profile Probe × 2 (3) Thermocouple differential analog input module × 3 (4) CompactDAQ chassis × 1 h. Computer i. Preheater (1) Shell casing (2) Ceramic fiber heater j. Pd-based catalytic membrane reactor (1) Pd composite membrane (2) High temperature shit catalyst (3) Shell casing (4) Ceramic fiber heater 2. Installation (25-55% of purchased equipment) 3. Instrumentation and controls, installed (8-50% of purchased equipment) 4. Piping, installed (10-80% of purchased equipment) 5. Electrical, installed (10-40% of purchased equipment)
88
Chapter 3 Techno-Economic Performance Evaluation
B. Buildings, process and auxiliary (10-70% of purchased equipment) C. Service facilities and yard improvements (40-100% of purchased equipment) D. Land (4-8% of purchased equipment) II. Indirect Costs A. Engineering and supervision (5-30% of direct cost) B. Legal expenses (1-3% of fixed capital investment) C. Construction expense and contractor's fee (10-20% of fixed capital investment) D. Contingency (5-15% of fixed capital investment) III. Fixed Capital Investment ( = Direct Costs + Indirect Costs ) IV. Working Capital ( 10-20% of Total Capital Investment ) V. Total Capital Investment ( = Fixed Capital Investment + Total Capital Investment ) ‡ Please notice that all of costs are updated to 2014 by Chemical Engineering Plant Cost Indexes.
89
Chapter 3 Techno-Economic Performance Evaluation
Table 3-15. Actual large-scale Pd/alloy-based CMR specifications used for cost analysis.
Pd thickness [μm] 8.300
Outer diameter of the support tube [m] 0.025 Length of the support tube [m] 0.356 Length of the porous part [m] 0.254 Outer diameter of the shell casing [m] 0.051 Wall thickness of shell casing [m] 0.00305 Inner diameter of the shell casing [m] 0.045 Length of the shell casing [m] 0.914 Membrane area [m2] 0.0203 Volume for shell casing [m3] 0.00143 Annular volume for catalyst packing [m3] 0.00027 Void fraction of catalyst (assumed) 0.657 Catalyst dilution (assumed) 0.800 Solid catalyst density (calculation) [kg/m3] 2130.0 Catalyst bed density (calculation) [kg/m3] 1119.528 Weight of catalyst [kg] 0.302
90
Chapter 3 Techno-Economic Performance Evaluation
3.2.1.4 Coal-based Hydrogen Production Plants with Integrated Industrial-Scale Pd/Alloy-based CMR Modules
The process block flow diagram of HP-CMR plant with CO2 capture, based on the design of the DOE/NETL report [44], is presented in Figure 3-5. Conforming to standard practice in engineering system economic analysis, the calculation of the FCI is performed through the sum of direct and indirect costs, followed by the calculation of the total capital investment TCI as the sum of FCI and working capital WC [178]. In the TCI estimation, cost figures associated with the assorted process units/equipment such as the GE energy gasifier, the syngas scrubber, and the single stage Selexol unit are adopted from the DOE/NETL report [112]. Furthermore, the six- tenths factor rule [178] (Equation 3-16) is used to calculate equipment costs since pertinent cost
data are not available for the commercial scale HP-CMR plant under consideration (685 tonnes H2 per day [44]). The six-tenths factor rule represents a standard and widely used empirical set of practical guidelines used for designing chemical plants (sizing industrial process system units) on the commercial scale that has been proven quite satisfactory in practice while maintaining sound robustness margins [179]. The Chemical Engineering Plant Cost Indexes shown in Table 3-7 [177] are used to update all pertinent cost estimates by taking into account the time-value of money. It should be pointed out that cost indexes are used only to provide general, rather conservative estimates without incorporating the reasonable cost-reducing possibility of future technological advances mathematically represented by “learning curves”. The TCI for the HP-CMR technology option includes various components as shown in Table 3-16. Furthermore, the key cost components of the FCI, such as installation and engineering & supervision, depend on purchased equipment costs.
Capacity of Equipment A Cost of Equipment A = Cost of Equipment B × . 3-16 Capacity of Equipment B 0 6 � �
91
Chapter 3 Techno-Economic Performance Evaluation
Figure 3-5. Process block flow diagram of the coal-based H2 production plant integrated with membrane technology.
Motivated by the fact that the TPC represents another major indicator broadly used in the economic performance analysis of chemical plants, a detailed TPC model for coal-fired HP-CMR plants is proposed as shown in Table 3-17. Within the TPC model, the production cost consists of
several components including operating costs, CO2 transportation and storage, H2 delivery, carbon tax, insurance, patents and loyalties, as well as plant overhead-relevant costs. In particular, the overall operating cost is comprised of three key components: (1) fuel, (2) fixed O&M costs, and (3) variable O&M costs. In this study, bituminous coal is considered as fuel (IllinoisNo. 6), and pertinent costs were obtained from the United States Energy Information Administration (US EIA) [180]. The fixed and variable O&M costs were obtained from relevant DOE/NETL reports [44,112] and also updated utilizing the Chemical Engineering Plant Cost Indexes [177]. The annual operating labor costs for the CMR module are estimated by multiplying the annual working hours (h/year) by the labor cost rate ($/h). A working hour is calculated based on the assumption of a unit operation under 53 employee hours/day/processing step, with two major steps involved in the CMR module’s operating mode (heat transfer and reaction/separation) [178]. The operating labor
92
Chapter 3 Techno-Economic Performance Evaluation
requirement for large equipment was estimated based on the plant’s capacity as specified by Peters et al. [178]. Furthermore, membrane replacement costs are taken into account in the variable O&M
costs and calculated by dividing the membrane cost by the membrane lifetime. CO2 transport and
storage (T&S) costs are also included in the production cost. The estimation for CO2 T&S costs is
based on the cost assessment study in the DOE/NETL report [181]. The H2 delivery costs include
the transport and distribution expenses for compressed H2 (350 bar). Please notice that the H2
delivery costs vary depending on the dispensing pathways; for instance, the H2 delivery costs in 2013 by pipeline, pipeline-tube trailer, and tube trailer are $4.44, $3.16, and $3.00 per kg, respectively [182]. Additionally, a carbon price (tax) is explicitly taken into account within this framework. Finally, for the incorporation of all other production cost components and general expenses in the TPC model, the standard procedure presented in the book by Peters et al. [178] has been followed in Table 3-17.
93
Chapter 3 Techno-Economic Performance Evaluation
Table 3-16. Estimation of the capital investment for an industrial scale Pd/alloy-based CMR module. a
I. Direct Costs A. Equipment + installation + instrumentation + piping + electrical + insulation + painting 1. Purchased equipment a. WGS reactor b. HTS catalyst c. Pd/Au alloy composite membrane 2. Installation (25-55% of purchased equipment) 3. Instrumentation and controls, installed (8-50% of purchased equipment) 4. Piping, installed (10-80% of purchased equipment) 5. Electrical, installed (10-40% of purchased equipment) B. Buildings, process and auxiliary (10-70% of purchased equipment) C. Service facilities and yard improvements (40-100% of purchased equipment) D. Land (4-8% of purchased equipment) II. Indirect Costs A. Engineering and supervision (5-30% of direct cost) B. Legal expenses (1-3% of fixed capital investment) C. Construction expense and contractor's fee (10-20% of fixed capital investment) D. Contingency (5-15% of fixed capital investment) III. Fixed capital investment ( = I + II ) IV. Working capital ( 10-20% of V) V. Total capital investment ( = III + IV ) a Please notice that the cost proportions/ranges shown in this table are commonly used for the estimation of capital investment [146,178]
94
Chapter 3 Techno-Economic Performance Evaluation
Table 3-17. Total product cost estimation for the HP-CMR plant. a
I. Production cost A. Operating costs 1. Fuel 2. Fixed O&M costs a. Operating labor cost (0.8% of fixed capital investment) b. Maintenance labor cost (0.8% of fixed capital investment) c. Administrative & Support labor (25% of operating labor cost) d. Property taxes (2% of fixed capital investment) 3. Variable O&M costs a. Maintenance material cost (2% of fixed capital investment) b. Water c. Chemical (1) COS catalyst (2) WGS catalyst (3) Selexol solution (4) Waste disposal (5) PSA adsorbent d. Membrane replacement (= membrane cost/membrane lifetime) B. CO2 transport & storage C. H2 delivery D. Carbon tax levied in 2015 E. Insurance (0.4-1% of fixed capital investment) F. Patents and loyalties (0-6% of total product cost) G. Plant overhead costs (5-15% of total product cost) II. General expenses A. Administrative costs (2-5% of total product cost) B. Marketing costs (2-6% of total product cost) C. R&D costs (5% of total product cost) D. Financing interest (6-10% of total capital investment) III. Total product cost (=Production cost + General expenses) a Please notice that the cost proportions/ranges shown in this table are commonly used for the estimation of capital investment [146,178]
95
Chapter 3 Techno-Economic Performance Evaluation
3.2.2 Economic Performance Assessment: Baseline Net Present Value Model
The net present value (NPV) represents a measure of the increase or decrease in wealth associated with an investment in an engineering project/plant [183]. Moreover, the NPV metric (in particular the expected NPV (ENPV) in the proposed stochastic evaluation framework) represents a risk neutral valuation metric to assess lifecycle economic performance and a rather helpful one for ranking valuation outcomes of competing technology options [183]. The proposed economic performance assessment framework under uncertainty incorporates additional/complementary economic lifecycle performance metrics such as the “value at risk” and “value at opportunity” that are used from the perspective of risk-averse and risk-seeker analysts/decision makers respectively. In this study, the baseline NPV-model relies on the standard calculation by forming the difference of the gross present value (GPV) of future net cash flows and the initial TCI [183] (Equation 3-17):
NPV = GPV TCI 3-17
− It should be pointed out that the primary objective of the present study is the development and presentation of the structural characteristics of a methodologically sound economic performance assessment framework in the presence of irreducible uncertainty (rather than a detailed investment valuation method that would require data derived from accumulated plant operating experience). This assessment can potentially inform decision making regarding the realization of demonstration plants for a new, yet untested, technology option such as HP-CMR at the commercial scale. In light of the above objective, and given the fact that the latest Chemical Engineering Plant Cost Index needed to update all pertinent cost estimates of HP plants is available for the year 2012, in the present study all HP plants considered are assumed to start operations in the year 2013, the lifetime of each plant is 30 years, and regulatory action on CO2 emissions is introduced in the year 2015. Within this context, the GPV is defined as the sum of the discounted stream of net cash flows over 30 years [183]:
96
Chapter 3 Techno-Economic Performance Evaluation
CF CF CF CF GPV = 30 = + + + 3-18 (1 + tr) 1 +1r (1 + 2r) (1 + 30r) � t 2 ⋯ 30 t=1
where CFt is the nominal net cash flow in year t encompassing annual revenue generated by hydrogen sales and TPC, depreciation as well as the plant’s savage value. In the above equation r represents the nominal discount rate that is used in this assessment. In particular, the reportable
net income in year t (It) was estimated using Equation 3-19, where Rt is the revenue generated in year t through hydrogen sales in the pertinent market:
I = R TPC 3-19
t t − t Please notice that the levelized cost of hydrogen production (the minimum required hydrogen selling price to achieve a 10% annual rate of return over the life of the HP plant) changes since the selling price varies broadly within the pertinent market [184]. Therefore, to avoid a possible overestimation of the revenue stream, the H2 selling price was conservatively based on a figure of $10 per kg. In addition, the inflation rate (λ) used in the calculation of the above discounted cash flows was estimated using Equation 3-20, where CPI is the consumer price index [185]:
CPI = 1 3-20 CPI t t λ t−1 − The depreciation cost is deducted as a business expense, practically acting as a tax shield (β) and calculated as follows:
= FCI D ( + ) 3-21
βt ∙ t ∙ ΦState ΦFederal In Equation 3-21, Dt represents the depreciation rate in year t [186], and ΦState and ΦFederal the combined state and local sales tax rate and federal corporate tax rate respectively. Please notice
that the value of ΦFederal follows the federal corporate tax schedule defined by the Internal Revenue Service (IRS), U.S. Department of the Treasury that depends on the taxable income level [187],
97
Chapter 3 Techno-Economic Performance Evaluation
while the value of ΦState determined by the plant location. The salvage value (SV) of HP plants was computed using Equation 3-22 [183], where MV denotes the market value of the HP plants considered and BV is their book value:
SV = MV (MV BV) ( + ) 3-22
− − ∙ ΦState ΦFederal In the present study, since the recovery period is 20 years, the book value of the plant after 30 years is zero. The term (MV − BV), in Equation 3-22, is used to represent the taxable gain at the time the plant is sold.
In the evaluation study of engineering design flexibility, since plant retrofitting potentially would potentially cause the plant components to be mismatched as discussed in the later, a de-
rating factor (CFt) will be incorporated into the revenue calculation as shown in Equation 3-23:
R = SP CF PS (1 DF ) 1000 365 3-23
t t ∙ t ∙ ∙ − t ∙ ∙ where SPt is the unit selling price of hydrogen in US$/Kg in year t, CFt is the capacity factor in
year t, and PS represents the plant size which is equal to 685 tonnes H2 per day. Furthermore, the
HP-CMR plant is assumed to be built in 2014 and start operating in 2015, while the CO2 tax is
assumed to be introduced in 2017. To simulate the growth behavior of CO2 tax throughout the plant’s lifetime, a geometric Brownian motion model (or lognormally distributed Wiener process model) [50,188] is adopted by taking into account expected annual growth rate and volatility of
CO2 tax rate as shown in Equation 3-24:
( ) = = 3-24 Wk µ∙k+σ∙B k Υk Υ0 ∙ 𝑒𝑒 Υ0 ∙ 𝑒𝑒 where is the CO2 tax rate after k year of introducing tax with the unit of $/tonne; is the initial
CO2 taxΥk rate in $/tonne; W captures the Wiener process; represents the expected Υannual0 growth rate of CO2 tax (or draft kin the Wiener process); indicatesµ the volatility of CO2 tax rate (or ( ) volatility in the Wiener process); B k stands for theσ standard Brownian motion.
98
Chapter 3 Techno-Economic Performance Evaluation
3.3 Economic Performance Analysis and Evaluation under Uncertainty: Integration of Monte Carlo Simulation Methods
The starting point is the recognition that the TCI, TPC, LC and NPV models are inherently uncertain and driven by various irreducible market, regulatory and other uncertainties. Therefore, an economic performance evaluation framework of a technology option based on expected/average/baseline values for all the above uncertainty sources (the uncertain TCI/TPC/LC/NPV model inputs) leads to single-point estimates and quite often to unsatisfactory economic appraisals [50,189]. This is known as the “flaw of averages” according to which the economic performance evaluated at average/baseline conditions does not necessarily represent the average economic performance [50]. Furthermore, a Monte Carlo simulator integrated into the proposed framework can simultaneously take into account multiple uncertain model inputs associated with the future state of the economic, regulatory and process operating environment as opposed to the conventional sensitivity analysis where one input variable is allowed to vary while keeping all others fixed [50]. In light of the above, the present study’s key objective is the development of a methodologically sound economic performance evaluation framework for HP plants by explicitly taking into account various sources of uncertainty. Consequently, the derivation of TCI, TPC, LC and NPV distribution profiles is sought that allows the statistical characterization of “risks and opportunities/rewards” in the presence of uncertainty [50,189]. For the attainment of such a goal the following methodological path is pursued Figure 3-6:
(1) Step 1: Comprehensive baseline functional TCI/TPC/LC/NPV-models are developed first (please see previous sub-sections). (2) Step 2: Key uncertain model input variables are identified admitting reasonable probabilistic representations through appropriately selected distribution profiles. Using Monte Carlo techniques and random sampling from the above distributions, model input uncertainties are propagated through the TCI/TPC/LC/NPV-model, generating distribution pro-files for TCI/TPC/LC/NPV that can be characterized in a statistical/probabilistic sense. Please notice that the expected value of the above sample of TCI/TPC/NPV values generated by Monte Carlo simulations represents an
99
Chapter 3 Techno-Economic Performance Evaluation
unbiased estimator of the expected value of TCI/TPC/LC/NPV, and thus the aforementioned “flaw of averages” can be effectively overcome [50].
As mentioned above, integrating standard Monte Carlo simulation methods into the proposed framework allows the inherent uncertainty of key model inputs to be explicitly recognized. Indeed, the Monte Carlo simulator generates a broad range of economic performance outcomes, and therefore allows the statistical/probabilistic characterization of the impact of uncertain future states on the economic performance of the CMR module as well as the entire HP plant. In the present study, the software package XLSim was used for conducting the Monte Carlo simulation runs [50]. All uncertain model inputs considered are listed along with the corresponding probability distributions in Table 3-18, Table 3-19, Table 3-20, Table 3-21 and Table 3-22, respectively, in terms of different topic studies. Please notice that the lack of accumulated operating experience and the dearth of plant data at the commercial scale pose significant challenges to any economic performance evaluation of Pd/alloy-based CMR modules as well as HP-CMR plants. In this study, we considered four different types of probability distributions for the uncertain model input variables (please see Table 3-18, Table 3-19, Table 3-20, Table 3-21 and Table 3-22):
(i) In the presence of available historical data, the bootstrap distribution (BD) were developed using standard re-sampling techniques [50]. (ii) In the absence of any reliable data at the commercial scale, the simple uniform distribution (UD) and triangular distribution (TD) were considered, which span the range of estimates reported in recent comprehensive studies/reports (cited in the list of bibliographical references) [50]. (iii) The normal distribution (ND) was only used to capture the standard Brownian motion that represents the increment random variables in the Wiener process of the proposed geometric Brownian motion model [50,188].
A normal distribution (known as a Gaussian distribution) is a symmetrically continuous probability distribution with a bell-shaped probability density function, commonly used in the natural and social sciences to represent real-valued random variables [188]. A triangular distribution is a continuous probability distribution with a probability density function shaped like
100
Chapter 3 Techno-Economic Performance Evaluation
a triangle as defined by a model input’s minimum, maximum, and the mostly like values. It is generally applied for cost estimation and risk assessments with little knowledge about the model input outside an approximate estimate of its minimum, maximum, and the most likely values [50,147,196,190,191]. A simple uniform distribution (also called as a rectangular distribution) has a rectangular probability density function which is defined by only two parameters, a model input’s minimum and maximum values [50,147,196,192,193]. The purpose of using this distribution type is to estimate the value of a model input based on an appropriate and reasonable range that is determined by the model input’s minimum and maximum values, similar to the use of a triangular distribution. A bootstrap distribution, which is developed using standard re-sampling methods, is useful and commonly applied for cost estimation and risk assessments in the presence of available historical and associated data [50,147,196,194,195].
101
Chapter 3 Techno-Economic Performance Evaluation
Figure 3-6. Methodological steps in the Monte Carlo simulation procedure [147,196].
102
Chapter 3 Techno-Economic Performance Evaluation
Table 3-18. Uncertain cost model inputs and corresponding probability distributions for an industrial-scale Pd/alloy-based CMR module for WGS integrated into natural gas-based hydrogen production plants.
Uncertainty driver Minimum Most likely Maximum Installation ratio, UDa 25.0% 55.0% Instrumentation and controls ratio, UDa 8.0% 50.0% Installed piping ratio, UDa 10.0% 80.0% Installed electrical ratio, UDa 10.0% 40.0% Buildings, process and auxiliary ratio, UDa 10.0% 70.0% Service facilities and yard improvements ratio, UDa 40.0% 100.0% Land ratio, UDa 4.0% 8.0% Engineering and supervision ratio, UDb 5.0% 30.0% Legal expenses ratio, UDc 1.0% 3.0% Construction expense and contractor's fee ratio, UDc 10.0% 20.0% Contingency ratio, UDc 5.0% 15.0% Working capital ratio, UDd 10.0% 20.0% Direct supervisory and clerical labor ratio, UDe 10.0% 20.0% Maintenance and repair ratio, UDc 2.0% 10.0% Operating supplies ratio, UDc 0.5% 1.0% Laboratory charges ratio, UDe 10.0% 20.0% Patents and loyalties ratio, UDf 0.0% 6.0% Local taxes ratio, UDc 1.0% 4.0% Insurance ratio, UDc 0.4% 1.0% Financing interest ratio, UDc 6.0% 10.0% Plant overhead costs ratio, UDf 5.0% 15.0% Administrative costs ratio, UDf 2.0% 5.0% Distribution & selling costs ratio, UDf 2.0% 20.0% Pd unit price [$/g], RH Historical data from 2011-2014, www.kitco.com Au unit price [$/g], RH Historical data from 2011-2014, www.kitco.com Labor cost per hour [$/h], RH Historical data from 2010-2014, www.bls.gov Electricity price [$/kWh], RH Historical data from 2010-2014, www.eia.gov Pd membrane lifetime, TD 1 3 5 HTS catalyst lifetime, TD 4 5 6
RH = Resample historical data, TD = Triangular distribution, UD = Uniform distribution a Based on purchased equipment cost, b Based on direct cost, c Based on Fixed-capital investment d Based on total capital investment, e Based on operating labor, f Based on product cost.
103
Chapter 3 Techno-Economic Performance Evaluation
Table 3-19. Uncertain cost model inputs and corresponding probability distributions for an actual large-scale Pd/alloy-based separation module for hydrogen purification.
Uncertainty driver Minimum Maximum Pd unit price [$/g], RH Historical data from 2011-2015, www.kitco.com Au unit price [$/g], RH Historical data from 2011-2015, www.kitco.com 3 2 0.5 H2 permeance [Nm /m -h-bar ], RH Experimental data in this study Ratio for installation, UDa 25.0% 55.0% Ratio for instrumentation and controls, installed, UDa 8.0% 50.0% Ratio for piping, installed, UDa 10.0% 80.0% Ratio for electrical, installed, UDa 10.0% 40.0% Ratio for buildings, process and auxiliary, UDa 10.0% 70.0% Ratio for service facilities and yard improvements, UDa 40.0% 100.0% Ratio for land, UDa 4.0% 8.0% Ratio for engineering and supervision, UDb 5.0% 30.0% Ratio for legal expenses, UDc 1.0% 3.0% Ratio for construction expense and contractor's fee, UDc 10.0% 20.0% Ratio for contingency, UDc 5.0% 15.0% Ratio for working capital, UDd 10.0% 20.0% Ratio for raw materials, UDe 10.0% 20.0% Ratio for operating labor, UDe 10.0% 15.0% Ratio for direct supervisory and clerical labor, UDf 10.0% 20.0% Ratio for utilities. UDe 10.0% 15.0% Ratio for maintenance and repair, UDc 2.0% 10.0% Ratio for operating supplies, UDc 0.5% 1.0% Ratio for laboratory charges, UDf 10.0% 20.0% Ratio for patents and loyalties, UDe 0.0% 6.0% Ratio for local taxes, UDc 1.0% 4.0% Ratio for insurance, UDc 0.4% 1.0% Ratio for financing interest, UDd 6.0% 10.0% Ratio for plant overhead costs, UDe 5.0% 15.0% Ratio for administrative costs, UDe 2.0% 5.0% Ratio for distribution & selling costs, UDe 2.0% 20.0% RH = Resample historical data, UD = Uniform distribution a Based on purchased equipment cost, b Based on direct cost, c Based on fixed capital investment, d Based on total capital investment, e Based on product cost, f Based on operating labor.
104
Chapter 3 Techno-Economic Performance Evaluation
Table 3-20. Uncertain cost model inputs and corresponding probability distributions for an actual large-scale Pd-based CMR module for Water-Gas Shift reaction.
Uncertainty driver Minimum Maximum Pd unit price [$/g], RH Historical data from 2011-2015, www.kitco.com Electroless plating setup cost [$], UD 1000 2000 Ratio for Operating labor, UDa 12.5% 50.0% Ratio for Direct supervisory and clerical labor, UDb 10.0% 20.0% Ratio for Utilities, UDa 12.5% 50.0% Ratio for Maintenance and repair, UDc 2.0% 10.0% Ratio for Operating supplies, UDc 0.5% 1.0% Ratio for Laboratory charges, UDb 10.0% 20.0% Ratio for Patents and loyalties, UDa 0 15.0% Ratio for installation, UDd 25.0% 55.0% Ratio for instrumentation and controls, installed, UDd 8.0% 50.0% Ratio for piping, installed, UDd 10.0% 80.0% Ratio for electrical, installed, UDd 10.0% 40.0% Ratio for buildings, process and auxiliary, UDd 10.0% 70.0% Ratio for service facilities and yard improvements, UDa 40.0% 100.0% Ratio for land, UDd 4.0% 8.0% Ratio for engineering and supervision, UDe 5.0% 30.0% Ratio for legal expenses, UDf 1.0% 3.0% Ratio for construction expense and contractor's fee, UDf 10.0% 20.0% Ratio for contingency, UDf 5.0% 15.0% Ratio for working capital, UDg 10.0% 20.0% RH = Resample historical data, UD = Uniform distribution a Based on raw materials, b Based on operating labor, c Based on electroless plating setup cost, d Based on purchased equipment cost, e Based on direct cost, f Based on fixed capital investment, g Based on total capital investment.
105
Chapter 3 Techno-Economic Performance Evaluation
Table 3-21. Uncertain cost model inputs and corresponding probability distributions for coal-based hydrogen production plants with integrated industrial-scale Pd/Alloy-based CMR modules.
Uncertainty driver Minimum Most likely Maximum Pd unit price [$/g], RH Historical data from 2010-2014, www.kitco.com Au unit price [$/g], RH Historical data from 2010-2014, www.kitco.com Labor cost per hour [$/h], RH Historical data from 2010-2014, www.bls.gov Coal price [$/tonne], RH Historical data from 2010-2014, www.eia.gov Inflation rate, RH Historical data from 2010-2014 [185]
H2 delivery cost [$/kg], RH 350 bar H2 dispensing pathways (2013) [182] Pipeline Pipeline-tube Tube trailer trailer 4.44 3.16 3.00 316L PSS support price [$/cm2 for a 24 cm2 support 10.2 11.3 12.4 area], TD Pd/Au membrane lifetime, TDa 1 3 5 Nominal discount rate for HP-CMR plant, TDb 14.4% 16.0% 17.6%
CO2 transport & storage costs [$/tonne], TD 9 10 11
CO2 tax rate (start at 2015) [$/tonne], TD 27 30 33
Annual growth rate of CO2 tax, TD 5.4% 6.0% 6.6%
H2 selling price [$/kg], TD 9 10 11 Combined state and local sales tax rates, TD 0.0% 6.4% 9.5% Ratio for market value of plant after 30 years, TDc 13.5% 15.0% 16.5% Ratio for Installation, UDd 25.0% 55.0% Ratio for instrumentation and controls, installed, UDd 8.0% 50.0% Ratio for piping, installed, UDd 10.0% 80.0% Ratio for electrical, installed, UDd 10.0% 40.0% Ratio for buildings, process and auxiliary, UDd 10.0% 70.0% Ratio for service facilities and yard improvements, UDd 40.0% 100.0% Ratio for land, UDd 4.0% 8.0% Ratio for engineering and supervision, UDe 5.0% 30.0% Ratio for legal expenses, UDc 1.0% 3.0% Ratio for construction expense and contractor's fee, UDc 10.0% 20.0% Ratio for contingency, UDc 5.0% 15.0% Ratio for insurance, UDc 0.4% 1.0% Ratio for working capital, UDf 10.0% 20.0% Ratio for financing interest, UDf 6.0% 10.0% Ratio for plant overhead costs, UDg 5.0% 15.0% Ratio for patents and loyalties, UDg 0.0% 6.0%
106
Chapter 3 Techno-Economic Performance Evaluation
Ratio for administrative costs, UDg 2.0% 5.0% Ratio for marketing costs, UDg 2.0% 6.0%
RH = Resample historical data, TD = Triangular distribution, UD = Uniform distribution a Based on available lab-scale data and estimates reported in the literature [38], b Notice that the nominal discount rate used for the conventional plants was 12.95% [44], c Based on fixed capital investment, d Based on purchased equipment cost, e Based on direct cost, f Based on total capital investment, g Based on total product cost.
107
Chapter 3 Techno-Economic Performance Evaluation
Table 3-22. Uncertain cost model inputs and corresponding probability distributions for coal-based HP-CMR plants with integrated flexibility options.
Uncertainty driver Minimum Most likely Maximum Pd unit price [$/g], BD Historical data from 2011-2015, www.kitco.com Au unit price [$/g], BD Historical data from 2011-2015, www.kitco.com Labor cost per hour [$/h], BD Historical data from 2011-2015, www.bls.gov Coal price [$/tonne], BD Historical data from 2011-2015, www.eia.gov Inflation rate, BD Historical data from 2010-2014 [185]
CO2 transport & storage costs [$/tonne], BD Depend on location of basins [181] Illinois East Williston Powder Texas River 11 11 16 24
H2 delivery cost [$/kg], BD 350 bar H2 dispensing pathways (2013) [182] Pipeline Pipeline-tube Tube trailer trailer 4.44 3.16 3.00 316L PSS support price [$/cm2 for a 24 cm2 support 10.2 11.3 12.4 area], TD Pd/Au membrane lifetime, TDa 1 3 5 Nominal discount rate for HP-CMR plant, TD 14.4% 16.0% 17.6% Combined state and local sales tax rates, TD 0.0% 6.4% 9.5% Ratio for market value of plant after 30 years, TDb 13.5% 15.0% 16.5%
H2 selling price [$/Kg], TD 9 10 11
Initial CO2 tax rate (start at 2017) [$/tonne], TD 27 30 33
Expected growth rate of CO2 tax (μ), TD 5.4% 6% 6.6%
Volatility of CO2 tax rate (σ), UD 0 20% Random variable used in the standard Brownian motion, Mean Standard deviation ND 0 1 Capacity factor for HP-CMR plants without CCS, TD 80% 85% 90% Capacity factor for HP-CMR plants with CCS, TD 75% 80% 85% Extra O&M cost for starting up the plant [$], UD 4,000,000 5,000,000 Output de-rating factor without pre-investment, UD 5% 10% Output de-rating factor with pre-investment, UD 0 5% Ratio for Installation, UDc 25.0% 55.0% Ratio for instrumentation and controls, installed, UDc 8.0% 50.0% Ratio for piping, installed, UDc 10.0% 80.0% Ratio for electrical, installed, UDc 10.0% 40.0% Ratio for buildings, process and auxiliary, UDc 10.0% 70.0% Ratio for service facilities and yard improvements, UDc 40.0% 100.0% Ratio for land, UDc 4.0% 8.0%
108
Chapter 3 Techno-Economic Performance Evaluation
Ratio for engineering and supervision, UDd 5.0% 30.0% Ratio for legal expenses, UDb 1.0% 3.0% Ratio for construction expense and contractor's fee, UDb 10.0% 20.0% Ratio for contingency, UDb 5.0% 15.0% Ratio for insurance, UDb 0.4% 1.0% Ratio for working capital, UDe 10.0% 20.0% Ratio for financing interest, UDe 6.0% 10.0% Ratio for plant overhead costs, UDf 5.0% 15.0% Ratio for patents and loyalties, UDf 0.0% 6.0% Ratio for administrative costs, UDf 2.0% 5.0% Ratio for marketing costs, UDf 2.0% 6.0%
BD = Bootstrap distribution, TD = Triangular distribution, UD = Uniform distribution, ND = Normal distribution a Based on available lab-scale data and estimates reported in the literature, b Based on fixed capital investment, c Based on purchased equipment cost, d Based on direct cost, e Based on total capital investment, f Based on total product cost.
109
Chapter 3 Techno-Economic Performance Evaluation
3.4 Formulation of Engineering Design Flexibility Options
The present study focuses on potential regulations on CO2 emissions. Among various forms of operational flexibility, volume flexibility is more appropriate and practical to be integrated into coal-based HP-CMR plants. Based on the concept of volume flexibility, one flexibility option considered is to temporally shut down the plant for a year if the cash flows for the previous two consecutive years are negative. The plant will be reopened for a year when the estimated cash flows, which incorporate the loss of an extra O&M cost for starting up the plant, are positive in the previous two consecutive years. Case B is used to represent the plant that considers this kind
of flexibility option. This flexibility option is introduced at the third year after the CO2 tax was imposed.
For constructional flexibility, two strategies are proposed based on various design phases of a carbon capture and sequestration (CCS) system. The first one is built upon the inclusion of a CCS system in the initial design phase, while the other one focuses on the construction of a CCS system at a later stage. The capacity of the plant is assumed to reduce when CCS is performed, since an output de-rating and efficiency loss will result from operating condition requirements and the expenses for running the CCS system [54].
In the first strategy of constructional flexibility, two flexibility options are identified and introduced in Case C and Case D respectively. Case C represents the plant that considers the construction of a CCS system in the initial design phase and assumes a continuous operation of it throughout the plant’s lifetime, while Case D indicates the plant that includes a CCS system in the initial design phase but runs it for a year when the cash flows for operating the CCS system are higher than without operating the CCS system in the previous two consecutive years. The flexibility option considered in Case D is introduced after 2 years since the CO2 tax was imposed. For the second strategy, that of retrofitting with a CCS system at a later stage, several assumptions should be considered:
• The CCS system considered for the HP-CMR plant is associated also with the CO2 conditioning and compression unit.
110
Chapter 3 Techno-Economic Performance Evaluation
• The configuration of the plant has to be modified for plant retrofitting, inducing the an output de-rating on hydrogen production due to potentially mismatched plant components in light of the above modification. • Retrofitting investment is assumed to consist of the retrofitting fee and the output de-
rating cost [54]. The retrofitting fee is based on the FCI of the CO2 conditioning and compression unit, while the output de-rating cost is estimated by considering the building cost of a makeup plant for compensation of the output de-rating. The output de-rating cost is equal to the product of the de-rating factor and the FCI of the HP-
CMR plant with CO2 capture.
Within the above assumptions, there are two flexibility options proposed in the present study (Case E and Case F) by taking into account various capital investment methods as well as their relevant de-rating factors produced by plant retrofitting. In Case E, the constructional flexibility option is introduced at the third year after the CO2 tax was imposed, by which the plant is retrofitted with a CCS system when the estimated cash flows for the plant with CCS are higher than without CCS in the previous two consecutive years. The same operational flexibility option for CCS system operation used in Case D is also included in Case E, but the time of its introduction is at the third year since the plant retrofitting. Similarly, Case F considers both constructional and operational flexibility options based on exactly the same decision rules used in Case E. However, Case F additionally takes into account a preinvestment option for constructional flexibility which facilitates the plant restructure with a CCS system. As a result, Case F would have a lower value of de-rating factor than Case E, reducing the cost of retrofitting investment and the losses of hydrogen production by retrofitting. However, the TCI and its relevant costs in TPC before retrofitting would be higher for Case F. Table 3-23 summarizes the descriptions for all cases considered in this study.
111
Chapter 3 Techno-Economic Performance Evaluation
Table 3-23. Descriptions and assumptions for all cases with various engineering design flexibility.
Denotation Engineering design flexibility Descriptions Case A No engineering design • The HP-CMR plant does not consider any flexibility engineering design flexibility representing the baseline case in the present study. Case B Operational flexibility • The HP-CMR plant considers the operational flexibility option for plant operation. • The plant will be temporally shut down for a year if the cash flows for the previous two consecutive years are negative. • The plant will be reopened for a year if the estimated cash flows minus an extra O&M cost for starting up the plant are positive in previous two consecutive years. • The operational flexibility option is introduced in the third year after the CO2 tax was imposed. Case C Constructional flexibility • The HP-CMR plant includes a CCS system in the initial design phase as a constructional flexibility option. • The CCS system is kept operating throughout the plant’s economic life. Case D Constructional flexibility and • The HP-CMR plant includes a CCS operational flexibility system in the initial design phase as a constructional flexibility option. • Another operational flexibility option for CCS system operation is also considered, by which the CCS system is performed for a year when the cash flows for the plant with CCS are higher than without CCS in the previous two consecutive years. • The operational flexibility option is introduced after 2 years since the CO2 tax was imposed.
112
Chapter 3 Techno-Economic Performance Evaluation
Case E Constructional flexibility and • The HP-CMR plant considers operational flexibility constructional and operational flexibility options without preinvestment. • The constructional flexibility option considered is introduced in the third year after the CO2 tax was imposed, by which the plant is retrofitted with a CCS system when the estimated cash flows for the plant with CCS are higher than without the CCS system in the previous two consecutive years. • An operational flexibility option is introduced after 2 years since the plant was retrofitted with a CCS system, by which the CCS system is operated for a year when the estimated cash flows for the plant with CCS are higher than without the CCS system in the previous two consecutive years. Case F Constructional flexibility and • The HP-CMR plant considers operational flexibility constructional and operational flexibility options with preinvestment. • The constructional flexibility option considered is introduced in the third year after the CO2 tax was imposed, by which the plant is retrofitted with a CCS system via preinvestment when the estimated cash flows for the plant with CCS are higher than without the CCS system in the previous two consecutive years. • An operational flexibility option considered is introduced after 2 years since the plant was retrofitted with a CCS system, by which the CCS system is operated for a year when the estimated cash flows for the plant with CCS are higher than without the CCS system in the previous two consecutive years.
113
Chapter 3 Techno-Economic Performance Evaluation
3.5 Synthesis and Characterization of Pd/Au Membranes
Asymmetric Pd/Au membranes were synthesized [171] on porous stainless steel (PSS) tubular supports 316L obtained from MOTT Metallurgical with an outer diameter of 0.5” and a length of 6”, having a total porous surface area of 60 cm2. One end was welded to a dense nonporous tube capped at one end and to a nonporous tube on the other end. First, the supports were cleaned with acetone in an ultrasonic bath, followed by their oxidation at 600oC for 12 h under air [197]. An intermediate layer of activated Al2O3 particles was deposited on the pores of the oxidized supports and cemented with Pd, as previously described [198,199]. Afterwards, the
membranes were activated with SnCl2-PdCl2 and then a dense layer of pure Pd was formed via electroless plating [200,201]. The deposition of gold on the surface of the Pd layer was obtained through a conventional electro-deposition [202]. Please notice that the thickness of the membranes was estimated through gravimetric methods and thus was assumed that the Pd or Au distribution on the surface of the membrane was uniform. The composition of the four Pd- and Pd/Au membranes used in this study are shown in Table 3-24 while pictures are shown in Figure 3-7:
Table 3-24. Composition of the membranes shown in this study.
Nomenclature Composition (Initial) Other
MA-156 6.0 µm Pd PSS support media grade of 0.5 μm MA-157 2.7 µm Pd + 0.1 µm Au PSS support media grade of 0.5 μm MA-159 (5-15) µm Pd + 0.4 µm Au PSS support media grade of 1.0 μm MA-160 (4.6-10.4) µm Pd + 0.1 µm Au PSS support media grade of 0.5 μm
114
Chapter 3 Techno-Economic Performance Evaluation
Figure 3-7. Depiction of the Pd/Au membranes used in this study before and after test.
The membranes were characterized based on their hydrogen permeation and He leak features at different times. These tests were conducted using a system equipped with a data acquisition board to continuously log the hydrogen flux via mass flow meters, shell-and-tube-side pressures and the temperature of the system. It is important to mention that the accuracy of the measurements was determined to be within 1% as previously described [34]. Hydrogen flux was measured constantly at 350 and 450oC and 0.5 bar; while weekly, the membranes’ flux was measured at ten different pressures (0.5-6 bar) to corroborate Sieverts’ behavior. Please notice that helium leak was measured under pure He stream at different pressures (0.5-6 bar). The permeance of the membrane was estimated to follow Sieverts’ law as shown in Equation 3-25:
F F = . . 3-25 A P , �H2 P , H2 0 5 0 5 � H2 ret − H2 per� where F is the hydrogen permeance, A is the permeable surface area of the membrane, F is
H2 . . H2 hydrogen flow rate of the permeate, and P , and P , are the hydrogen partial pressure �at the 0 5 0 5 2 2 retentate and the permeate, respectively. Furthermore,H ret H theper ideal H2/He selectivity ( ) of the
𝐻𝐻2 𝑆𝑆 �𝐻𝐻𝐻𝐻
115
Chapter 3 Techno-Economic Performance Evaluation
membrane (Equation 3-26) was used to estimate the H2 purity produced by the membrane (Equation 3-27).
F S = F 3-26 �H2 H2� He �He 1 H Purity % = 1 + −1 3-27
2 � H2 � 𝑆𝑆 �He
116
Chapter 4
Chapter 4 Natural Gas in Hydrogen Production: A Cost Study Natural Gas in Hydrogen Production: A Cost Study
4.1 Introduction
The hydrogen purification step lies at the heart of a viable hydrogen production technology option. The specific hydrogen separation method is of course chosen based on the application under consideration as well as the desirable hydrogen purity target [44]. Currently, the available hydrogen purification technology options are: chemical or physical absorption; adsorption; cryogenic; and membrane-based processes [3,4]. Note also that a recently recognized promising technology option is the integration of palladium (Pd)- and palladium/alloy-based membrane reactor modules in power plants for the co-production of electricity and pure hydrogen [143,144,146,203]. Within such a context, palladium and palladium/alloy-based composite membrane reactor technology could provide the means for simultaneous carbon dioxide capture and extra-purity hydrogen production realized in a single process unit in coal-fired (integrated gasification combined cycle) as well as natural gas-fired (natural gas combined cycle) power plants with carbon dioxide capture and sequestration (CCS)-ready design [44,144]. The above technical pathway and electricity co-production exhibit enhanced environmental performance through a cost-effective design accommodation of a CCS system while still being comparable to the
117
Chapter 4 Natural Gas in Hydrogen Production: A Cost Study
conventional technology option efficiency performance standards, and appealing economic performance prospects if future regulatory action on carbon dioxide emissions is pursued [146].
In light of the above considerations, the production of hydrogen by means of methane steam reforming (MSR) in integrated palladium- and palladium/alloy-based catalytic membrane reactor (CMR) modules represents an attractive technology option (HP-CMR), which has generated further interest due also to its great potential for process intensification. In particular, palladium- and palladium/alloy-based membranes with stainless steel, Hastelloy or Inconel supports [31,34,36,204] exhibit significant advantages such as high hydrogen flux and permeability, stable high selectivity, improved thermal, chemical and mechanical stability, cost- effective fabrication and maintenance, resistance to usability at high temperature (400–600°C) and pressure (20–50 atm) [205], ease of scale-up as well as long-term durability (5 years) [3]. As a result, considerable attention is now increasingly concentrated on the use of membrane reactor modules for large-scale hydrogen production [40,41,42,122,206]. The advantages of palladium- based CMRs over the conventional packed bed reactors in the MSR process have been amply demonstrated in several experimental and theoretical studies [31]. Moreover, CMRs exhibit considerable advantages over traditional reformers including the elimination of high and low temperature shift reactors, pre-oxygenation and hydrogen separator, thus enabling reaction, separation and product concentration processes to take place in a single unit operation. In addition, the continuous removal of the product hydrogen during the MSR process by a CMR provides enhanced methane conversion beyond the thermodynamically determined equilibrium. Compared to the conventional process, higher attainable conversion levels in a CMR allow operation to be conducted at lower temperatures, thus providing a prolonged catalyst lifetime, lower production costs, reduced material costs as well as the production of high-pressure carbon dioxide readily available for sequestration [31,144]. In particular, HP-CMR with carbon dioxide capture may lead to a significant reduction in total capital plant cost due to the reduced equipment size resulting from the elimination of the traditional WGS reactors and amine scrubbing units as well as reduced compression-related costs due to the high-pressure state of the retentate stream [31].
However, the lack of significant accumulated operating experience for HP-CMR plants on the commercial scale poses significant challenges. Therefore, any preliminary attempt to assess HP-CMR’s economic viability and performance prospects is certainly justified as efforts to
118
Chapter 4 Natural Gas in Hydrogen Production: A Cost Study
stimulate the realization of technology demonstration projects at the commercial scale worldwide intensify. Within the above context, a natural starting point in this study is the development of comprehensive baseline models for total capital investment (TCI) and total product cost (TPC), in order to evaluate the economic performance of an industrial-scale membrane reactor module integrated into a hydrogen production plant. Various sources of uncertainty (raw material market prices, labor costs, membrane lifetime and maintenance costs, financing interest costs, etc.) are recognized and their effect on TCI and TPC is explicitly taken into account using Monte Carlo techniques. As a result, insightful distribution profiles of TCI and TPC are derived rather than single-point value estimates, and thus more realistic distributions of membrane reactor module economic performance outcomes are generated and characterized in the presence of uncertainty.
In the present research study an economic performance assessment and cost analysis of an industrial-scale membrane reactor module integrated into a MSR-based hydrogen production plant is conducted comparatively against the baseline traditional/conventional case [176] in the presence of uncertainty. Furthermore, palladium/gold (Au) membranes are also considered with a fixed composition of 12% wt in gold.
119
Chapter 4 Natural Gas in Hydrogen Production: A Cost Study
4.2 Results and Discussion
By the proposed assessment and cost analysis framework, the FCI, TCI, TPC cumulative probability distribution profiles for the industrial-scale membrane reactor module considered in this study are shown in Figure 4-1, Figure 4-2 and Figure 4-3, respectively, for various palladium layer and gold layer thickness values as well as the distribution profile of the associated costs in the traditional technology option case for comparison purposes. Please note that each of these distribution profiles and the corresponding statistical characterization offer valuable pieces of information regarding the economic performance of the technology options considered in the present study.
• The probability (vertical axis) that the FCI/TCI/TPC falls below a desirable cost target level (horizontal axis) as well as the complementary probability for the cost to be higher than the aforementioned target level can be easily deduced. For example, there is a 10% probability for TPC to be lower than US$8 million in the 20 μm palladium layer thickness case (Figure 4-3). • The expected values of the FCI, TCI, TPC distribution profiles (also tabulated in Table 4-1) as well as the associated value ranges (a measure of dispersion/spread of possible economic performance outcomes) can also be readily deduced. • The downside risk (also known as ‘value at risk’ [50]) can be quantified at a pre-specified level. For example, a FCI/TCI/TPC threshold value can be easily identified for which there
is a 10% probability of a higher cost to be realized (usually denoted as the P90 value; the 90th percentile). For example, there is a 10% probability for TPC to be higher than US$10 million (and a 90% probability to be lower than US$10 million) in the 20 μm palladium layer thickness case (Figure 4-3).
120
Chapter 4 Natural Gas in Hydrogen Production: A Cost Study
Figure 4-1. Distribution profiles of fixed capital investment for various palladium and gold layer thickness values; CMR, catalytic membrane reactor; Pd, palladium; Au, gold.
121
Chapter 4 Natural Gas in Hydrogen Production: A Cost Study
Figure 4-2. Distribution profiles of total capital investment for various palladium and gold layer thickness values; CMR, catalytic membrane reactor; Pd, palladium; Au, gold.
122
Chapter 4 Natural Gas in Hydrogen Production: A Cost Study
Figure 4-3. Distribution profiles of total product cost for various palladium and gold layer thickness values; CMR, catalytic membrane reactor; Pd, palladium; Au, gold.
123
Chapter 4 Natural Gas in Hydrogen Production: A Cost Study
Table 4-1. Fixed capital investment/total capital investment/total produce cost expected values for various palladium thickness values.
CMR with CMR with CMR with CMR with CMR with CMR with CMR with Conventional Pd thickness Pd thickness Expected Pd Pd Pd Pd Pd Technology 10 μm and 5 μm and Value thickness thickness thickness thickness thickness Option [176] Au thickness Au thickness 75 μm 50 μm 20 μm 10 μm 5 μm 0.85 μm 0.42 μm Fixed capital investment 10.64 57.81 30.22 10.34 6.29 4.69 7.16 5.07 [M$]
Total capital investment 12.54 68.07 35.59 12.18 7.41 5.52 8.43 5.97 [M$]
Total product cost 9.28 32.94 19.19 9.26 7.24 6.44 7.67 6.63 [M$]
CMR: catalytic membrane reactor; Pd: palladium.
From Figure 4-1, Figure 4-2, Figure 4-3 and Table 4-1 it can easily be inferred that the expected values of FCI, TCI and TPC obtained from the respective distribution profiles are higher in the CMR-based technology option than those associated with the conventional technology platform option for palladium thicknesses greater than 20μm. However, FCI, TCI and TPC distribution profiles for CMRs start shifting towards the left of the vertical axis and thus becoming economically appealing, especially when palladium thickness falls below the 20μm level (in agreement with observations reported by Criscuoli et al. [176]. This behavior also conforms to the intuitively expected behavior because lower palladium thickness values lead to lower membrane areas required for the attainment of the same hydrogen recovery target level, and consequently to lower costs. Please note that the observed shift of the distribution profile curves to the left with decreasing palladium thickness implies a higher chance of costs falling below a certain target level, and thus enhanced economic performance. It should also be noted that over the years thin membranes with palladium thicknesses within the 5–10μm range have been synthesized at Worcester Polytechnic Institute’s Center for Inorganic Membrane Studies, and the attainment of high performance levels has been experimentally validated and patented [30,38,39,171]. Finally, given the enhanced environmental performance prospects of the MSR-based HP-CMR technology
124
Chapter 4 Natural Gas in Hydrogen Production: A Cost Study
option, the results of the cost analysis under uncertainty also demonstrate a significant window of opportunity for the proposed technology option to compete favorably on the overall techno- economic performance front with the traditional/conventional option, and therefore to be seriously considered for the provision of the requisite incentives aimed at the acceleration of its demonstration on a commercial scale.
Within a range of palladium thickness values, a second set of figures (Figure 4-4, Figure 4-5 and Figure 4-6) aims at a comparative appraisal of FCI, TCI and TPC for CMRs and also against the conventional non-membrane-based technology option on the average/expected value level as well as at equal levels of probability associated with potential ‘risks and opportunities’
such as the P95 and P5 values (the 95th and 5th percentile, respectively). The upper P95 line (a possible means of quantifying ‘risks’) graphically shows that there is a 5% chance of incurring a cost higher than the value on the line (equivalently a 95% chance of a cost lower than the value on the line), and similarly, the lower P5 line graphically shows the 5% chance of incurring a cost lower than the value on the line (a possible means of identifying ‘opportunities’). The middle line provides the expected cost value for the various palladium thicknesses considered. Please note that as the palladium thickness decreases not only do the FCI, TCI and TPC values decrease as noted
earlier, but so does the spread/variability (quantified as P95–P5) of the various costs – that is, the dispersion of possible economic performance outcomes. As the palladium unit price has by far the most significant effect on cost (as will be established through tornado diagrams presented below), the observed reduction in the variability/spread of the various costs with decreasing palladium thickness can be attributed to the fact that thinner membranes require lower palladium amounts for the attainment of the same hydrogen recovery level.
125
Chapter 4 Natural Gas in Hydrogen Production: A Cost Study
Figure 4-4. P95, P5 and expected value lines of fixed capital investment for various Pd layer thickness values; Pd, palladium; CMR, catalytic membrane reactor.
126
Chapter 4 Natural Gas in Hydrogen Production: A Cost Study
Figure 4-5. P95, P5 and expected value lines of total capital investment for various Pd layer thickness values; Pd, palladium; CMR, catalytic membrane reactor.
127
Chapter 4 Natural Gas in Hydrogen Production: A Cost Study
Figure 4-6. P95, P5 and expected value lines of total product cost for various Pd layer thickness values; Pd, palladium; CMR, catalytic membrane reactor.
128
Chapter 4 Natural Gas in Hydrogen Production: A Cost Study
Next, the useful tool of a tornado diagram is employed in order to conduct a sensitivity analysis and draw useful insights into the sources of uncertainty with the highest impact on FCI, TCI and TPC [51]. A tornado diagram graphically summarizes the relative impact of variations of the various uncertain model inputs considered in the present study over their respective ranges (from the lowest to the highest value) under the assumption that all other model inputs remain at their baseline values (fixed at their average/expected values). Therefore, within such a context, a tornado diagram nicely illustrates which uncertain model inputs affect an economic performance index/metric the most (in this study the expected value of FCI, TCI and TPC). In a tornado diagram each model input is represented by a bar as it varies over its prescribed range depicting its impact on the above performance index, and all bars are sorted from the long bars at the top to the shorter bars at the bottom [51]. In this manner, a tornado diagram prioritizes the most consequential inputs (sources of uncertainty) in terms of bar length, and facilitates the visualization of their impact on economic performance primarily in cases of large-scale multi-input models. Figure 4-7 and Figure 4-8 provide representative tornado diagrams for TCI and TPC in the palladium-based CMR case. Palladium price, working capital, membrane reactor and catalyst cost constitute the set of model inputs (uncertainty sources) with the highest impact on TCI in descending order, whereas palladium price, membrane lifetime, financing interest rate, working capital, raw materials and reactor cost constitute the set of model inputs with the highest impact on TPC (also in descending order). In light of the tornado diagrams, the upper right zone of a TCI distribution profile curve (Figure 4-2) represents the ‘zone of adverse economic performance outcomes (risks)’ corresponding to relatively higher palladium prices, working capital levels, membrane reactor and catalyst costs, whereas the lower left zone represents the ‘zone of enhanced economic performance outcomes (opportunities)’ corresponding to lower palladium prices, working capital levels, membrane and catalyst costs. Similarly, the upper right zone of a TPC distribution profile curve (Figure 4-3) represents the ‘zone of adverse economic outcomes (risks)’ corresponding to relatively higher palladium prices, financing interest costs, working capital levels, raw materials costs and shorter membrane lifetimes, whereas the lower left zone represents the ‘zone of enhanced economic performance outcomes (opportunities)’ corresponding to lower palladium prices, financing interest costs, working capital levels, raw material costs as well as longer membrane lifetimes
129
Chapter 4 Natural Gas in Hydrogen Production: A Cost Study
Figure 4-7. Tornado diagram for total capital investment (Pd unit price: US$16·9–24·4/g; WGS reactor cost: baseline ±10%; catalyst unit price: baseline ±10%; working capital to total capital investment ratio range 10–20%); Pd: palladium; WGS, water–gas shift.
130
Chapter 4 Natural Gas in Hydrogen Production: A Cost Study
Figure 4-8. Tornado diagram for total product cost (Pd unit price: US$16·9–24·4/g; palladium membrane lifetime: 1–5 years; WGS reactor cost: baseline ±10%; raw material cost per year: baseline ±10%; working capital to total capital investment ratio range: 10–20%; financing interest to total capital investment ratio range: 6–10%); Pd, palladium; WGS, water–gas shift.
.
131
Chapter 4 Natural Gas in Hydrogen Production: A Cost Study
4.3 Conclusion
A comprehensive economic performance evaluation framework for membrane reactor modules potentially integrated into hydrogen production through MSR process systems has been developed. This new technology option offers the prospect of hydrogen production with enhanced environmental performance in a carbon dioxide-constrained world, and as no significant operating experience has been accumulated at the industrial/commercial scale a preliminary assessment of its economic performance offered significant motivation for the pursuit of this study. In particular, the development of detailed comprehensive baseline models for FCI, TCI and TPC was pursued first followed by an explicit recognition of various sources of uncertainty. The effect of these uncertainty sources on FCI, TCI and TPC has been taken into account through the integration of Monte Carlo simulation methods into the aforementioned cost models. As a result, insightful distribution profiles of FCI, TCI and TPC have been derived rather than single-point value estimates, and a more realistic distribution of CMR economic performance outcomes has been generated through which ‘risks and opportunities’ can be identified in the presence of uncertainty. Furthermore, the results derived have shown that FCI, TPI and TPC profiles become economically appealing by reducing palladium thickness, a reasonably attainable objective with the aid of current membrane synthesis methods. Finally, tornado diagrams have been developed to identify, evaluate and prioritize the impact of the above uncertainty sources on TCI and TPC.
132
Chapter 5
Chapter 5 An Integrated Experimental/Techno-Economic Analysis of the Lifetime of Pd/Au Membranes An Integrated Experimental/Techno-Economic Analysis of the Lifetime of Pd/Au Membranes
5.1 Introduction
Hydrogen purification can be achieved by different processes such as pressure swing adsorption (PSA), cryogenic distillation, and membrane separation [77,207,208]. Consequently,
Palladium and Palladium/alloy membranes have been widely studied due to their high H2 permeability, theoretical infinite hydrogen selectivity and chemical stability under hydrocarbon containing gas streams [209]. In specific, Pd/alloy composite membranes containing Au have a superior hydrogen flux than pure Pd membranes and enhanced stability and recoverability [39]. The performance of Pd/Au membranes has been successfully demonstrated under actual industrial settings by Ma et al. [171,172] where the performance of Pd, Pd/Au, Pd/Pt and Pd/Au/Pt membranes were tested under actual coal derived syngas for a cumulative time of 4275 hours.
Pd/Au membranes showed high H2 purity level of 99.89% at 450°C and 12.6 bar for over 200 h in syngas atmosphere. Additionally, Catalano et al. [30] demonstrated the application of membrane technology applicable for membrane reactors at a large-scale used for the generation and separation of H2. Composite palladium membranes increased the CO conversion and the H2
133
Chapter 5 An Integrated Experimental/Techno-Economic Analysis of the Lifetime of Pd/Au Membranes
3 -1 recovery in a Water-Gas Shift (WGS) reactor achieving a H2 production of 5.6 Nm day with a quality of 99.2-99.97%.
Although, significant technical developments on this field have occurred since the 80’s, different factors such as the lifetime of the membranes remain uncertain. For instance, the phrase “long-term test” of membranes has been widely used in the pertinent literature ranging from 80 h to 8,640 h [35,171,209] and therefore the lifetime of the membranes remain unidentified; Table 5-1 shows a literature review of different “long-term” hydrogen permeation test times for Pd and Pd/Alloy membranes. Since the formation of pinholes is related by the movements of Pd crystallites [35], it is expected that at prolonged times and high temperatures, which increases molecular mobility, the probability of pinhole formation increases, especially as the Pd layer thickness reduces. Furthermore, it has been shown that the reduction of H2 flux on thicker membranes is economically unfavorable [196]. Therefore, it is naturally possible to infer an economic tradeoff between lifetime and the H2 flux; both characteristics determined by the membrane thickness. Consequently, in order to accelerate the potential industrial application of membrane technology, it is necessary a genuine long-term study combined with the analysis of the economic characteristics that encompass the features of this technology. It is important to mention that the lifetime of Pd- and Pd-alloy membranes has been assumed in the pertinent literature to be of 5 years and that only four replacements are needed throughout a 25-year plant lifetime [210].
Due to the lack of operating experience, economic evaluations for a new technology such as Pd-based membranes have been based on theoretical estimates such as the after mentioned lifetime. In specific, the studies reported by Ma et al. [145,146,147,211] have used detailed comprehensive baselines of economic indicators, such as the Net Present Value (NPV) and Total Capital Investment (TCI), where various sources of uncertainty are identified and integrated into the framework of study. Their impact on the valuation profile has been taken into account through Monte Carlo simulation methods by which these uncertainties are propagated through the economic models. As a result, these economic evaluations allowed for the derivation of economic performance outcomes arranged as distribution profiles, rather than single-point value approximations which lead to errors in valuation assessments. Consequently, the objective of this work is to combine long-term H2 permeance results to explore the influence of membrane thickness on the lifetime of the membrane and theoretical economic evaluation results to explore
134
Chapter 5 An Integrated Experimental/Techno-Economic Analysis of the Lifetime of Pd/Au Membranes the tradeoff between permeance-thickness and lifetime in terms of economic performance outcomes.
135
Chapter 5 An Integrated Experimental/Techno-Economic Analysis of the Lifetime of Pd/Au Membranes Table 5-1. Long term tests presented in the literature.
Conditions/Preparation/ Membrane Testing Type of membrane Reference Application Thickness time
Pd–23%Ag/stainless steel H2/N2 mixture (50:50), 350- 7 µm 2400 h [212] composite membranes 450oC
o Pd–Ag foils H2/N2 mixture, 150-400 C 50-60 µm 12 months [213]
o Pd Ag rolled membranes Pure H2, 300–400 C 50 µm 1440 h [214]
Pd membrane/PSS Methanol Steam Reforming, 15 µm 900 h [215] 350oC
Pd membrane/ Porous Activation via photocatalytic 0.3-0.4 µm 160 h [73] titania ceramic deposition, 400-500oC Pd/Ag membranes Thermal cycling 25-400oC 5.5 µm 960 h [216] supported on Alumina
o Pd/Cu membranes Effect of H2S, 350-450 C 15 µm 1200 h [217]
o Palladium/PSS Pure H2, 350 C 19-28 µm 1100 h [34]
Palladium/ Silicon wafer Water gas shift, 320-380oC 1 µm 140 h [218]
Palladium Under heat treatments, 700- 4 µm 12‒446 h [219] membranes/YSZ/PSS 500oC
Pd/ceramic membranes Natural gas reformate 4 µm 3600 h [220] separation, 400oC Microstructured Pd/Ag Hydrogen at 450oC 4.7-13 µm 1200 h [221] membranes
Pd/Ag films Under H2 with oxidation (air 4 µm 2040 h [222] treatments) 350oC
136
Chapter 5 An Integrated Experimental/Techno-Economic Analysis of the Lifetime of Pd/Au Membranes 5.2 Results and Discussion
5.2.1 Hydrogen Permeation and Helium Leak Tests
This study contains the results of four membranes tested at 350 and 450°C under pure H2 with occasional He leak tests for a cumulative time of 19,200 hours or an equivalent time of 2.2 years to attain operating experience associated with this technology. In addition, as shown in Figure 3-7, the membranes used in this work showed an excellent physical integrity compared to those long-term reported in the pertinent literature [146,147]. The following subsections contain the detailed description of the testing procedures and results of each membrane.
(a) MA-156
A pure Pd membrane (MA-156) was synthesized displaying an initial He leak of 0.3
sccm/bar at room temperature. The H2 permeance test of MA-156 as a function of time is shown 3 -2 -1 -0.5 in Figure 5-1. The initial H2 permeance was of 8.2 and 14.8 Nm m h bar at 350 and 450°C, respectively. Please notice that during testing, the membrane was oxidized at 2 bar under air at 3 -2 -1 -0.5 300°C for 34 h. After oxidation the H2 permeance increased to 16.5 and 22.8 Nm m h bar at 350 and 450°C, respectively corresponding to an enhancement of 101% and 54%; while the He
leak did not seem to change after the oxidation stage. This increase in H2 permeance after oxidation is in agreement with previous studies [222,223]. During 850 h of further testing, the H2 permeance was stable with a permeance of 22.3 Nm3m-2h-1bar-0.5 at 450°C and an ideal selectivity of 490. Furthermore, the He leak after 2150 h was 6.9 sccm/bar.
137
Chapter 5 An Integrated Experimental/Techno-Economic Analysis of the Lifetime of Pd/Au Membranes
Figure 5-1. Hydrogen permeance and helium leak tests of MA-156.
138
Chapter 5 An Integrated Experimental/Techno-Economic Analysis of the Lifetime of Pd/Au Membranes (b) MA-157
Membrane MA-157 was synthesized as a thin Pd/Au membrane and was tested for 2200 h as shown in Figure 5-2. After 670 h of testing under hydrogen atmosphere, the permeance for this gold enriched membrane had an initial value of 38 Nm3m-2h-1bar-0.5 at 450°C with an ideal selectivity of 520. The permeance and He leak increased with time. MA-157 had an average permeance of 45 Nm3m-2h-1bar-0.5 at 450°C and a final He leak of 16.5 sccm/bar. It is important to mention that a long-term (>1000 h) test has never been reported for a membrane thickness thinner than 4 µm.
Figure 5-2. Hydrogen permeance and helium leak tests of MA-157.
139
Chapter 5 An Integrated Experimental/Techno-Economic Analysis of the Lifetime of Pd/Au Membranes (c) MA-159
MA-159 had a He leak after synthesis of 0.01 sccm/bar. This Au enriched membrane 3 -2 -1 -0.5 displayed an initial H2 permeance of 7.4 Nm m h bar and a He leak of 0.42 sccm/bar at 350°C. The temperature was increased to 450°C and consequently the permeance increased gradually to 42 Nm3m-2h-1bar-0.5 after 137.5 h due to in-situ alloying. The permeance remained stable; however, the He leak increased considerably (~35 sccm/bar) and the experiment was stopped after 800 h of testing for further investigation and characterization of the leak location. MA-159 was repaired and renamed as MA-159b. Gold rings were plated at both sides of the welding area, reducing the He leak to 0.3 sccm/bar at room temperature. The membrane was tested for 100 h at 450°C under pure H2. The He leak increased again to 45 sccm/bar and the test was terminated. MA-159b was subjected to a mild polish treatment to remove any Pd peel off. Palladium rings were plated in the welding area close to the cap, twice. This treatment reduced the overall He leak of the membrane to 30 sccm/bar. Subsequently, the first half of MA-159b was plated once, reducing the leak even further to 18 sccm/bar. Finally, a complete Pd plating treatment was conducted along the whole membrane. Undetectable He leak was found at pressure differences of 2 bar. At this point MA-
159b was renamed to MA-159c. MA-159c was tested for ~200 h showing a H2 permeance of 25 and 35 Nm3m-2h-1bar-0.5 at 350 and 450°C, respectively. He leak increased gradually to 8.3 sccm/bar. The membrane was removed from the module and plated once more to increase the Pd layer thickness to 11 µm; this membrane was renamed to MA-159d. MA-159d was tested for 700 3 -2 -1 -0.5 h showing a stable H2 permeance of 35 Nm m h bar , similar to the permeance displayed by MA-159c. Nevertheless, He leak appeared reaching 40 sccm/bar at room temperature. The membrane was replated to 12 µm, renamed as MA-159e and tested for 1000 h showing a He leak of 15 sccm/bar by the end of the test. The permeance of MA-159e, was stable with a value of 23 and 35 Nm3m-2h-1bar-0.5 at 350 and 450°C, respectively. Notice that the membrane remained under pure He for 20 h, as indicated in Figure 5-3, due to a problem with the H2 security system. Afterwards this period, the permeance of the membrane showed a steady decline at 350°C and stopped once the temperature was increased to 450°C. After MA-159e failed, it was repaired once
more and renamed as MA-159f. The membrane was tested for 1400 h showing a stable H2 permeance of 32 Nm3m-2h-1bar-0.5 at 450°C, while He leak increased gradually from >1 to 17 sccm/bar.
140
Chapter 5 An Integrated Experimental/Techno-Economic Analysis of the Lifetime of Pd/Au Membranes Notice that the cumulative test of MA-159 after ~4000 h of continuous test and 5 repair steps is shown in Figure 5-3. Overall, the membrane showed a stable permeance as shown in Table 5-2; however, it displayed a continuous He leak that seemed to decline after repair, but increase as the tests proceeded. It is important to mention that, for each test session, additional palladium was deposited to cover the defects of the membrane before the test was reinitiated; nevertheless, for MA-159b, only rings at the welding area were plated and consequently the increase in thickness was neglected. The He leak, shown in Table 5-2, represents the value obtained at 450°C before the
test was terminated; while the H2 permeance was taken as the average value displayed throughout the test at 450°C. Furthermore, it is important to mention that the media grade of the support influenced significantly the lifetime of the membranes. This is evidenced by the lower stability displayed by MA-159 where several repairs were needed.
Table 5-2. List of characteristics of MA-159.
* Dense Pd Final He leak* H2 Permeance Membrane Testing time [h] thickness [µm] [sccm-bar-1] [Nm3m-2h-1bar-0.5] MA-159 830 5 35 45 MA-159b 250 5 (rings) 45 40 MA-159c 210 8.5 8.3 36 MA-159d 700 11 20.5 34 MA-159e 1000 12 15 33 MA-159f 1350 15 17 31 * Hydrogen permeance and helium leak at 450oC
141
Chapter 5 An Integrated Experimental/Techno-Economic Analysis of the Lifetime of Pd/Au Membranes
Figure 5-3. Hydrogen permeance and helium leak tests of MA-159.
142
Chapter 5 An Integrated Experimental/Techno-Economic Analysis of the Lifetime of Pd/Au Membranes (d) MA-160
3 -2 -1 -0.5 This membrane had a stable H2 permeance of 50 Nm m h bar at 450°C. Furthermore, the He leak after 1200 h; nevertheless, after 1480 h, the membrane was accidentally exposed to He for 62 h as indicated by the yellow lines in Figure 5-4. There was no immediate change to the membrane’s He leak; but, the performance deteriorated gradually after the He exposure, increasing its He leak to a maximum leak of 14.5 sccm/bar. The H2 permeance dropped immediately by 12% to 44 Nm3m-2h-1bar-0.5 and remained constant until the end of this test; the total testing time was 2950 h. At this point the test was partially terminated to identify the location of the leak as well as repair the membrane. Gold rings were plated on both sides of the welding area, reducing the He leak to 0.5 sccm/bar. The membrane renamed as MA-160b and was tested for 1000 h. Notice that He leak increased to 45 sccm/bar by the end of the test and that due to technical problems, the membrane was oxidized with air for 10 h, after 400 h of this test or 3350 of cumulative testing time, causing a sudden He leak growth.
MA-160b was repaired by plating 5.8 µm of Pd and renamed as MA-160c. During this test, it can be observed that several oxidation stages due to a failure in the system. These oxidation stages occurred at the times where the temperature oscillated. These oxidations did not affect the permeance nor the He leak presented in the membrane. Subsequently, the temperature of the membrane was raised to 450°C and its permeance was tested for ~600 h; the He leak remained stable throughout this test. Unfortunately, the system suffered another error and a third oxidation stage occurred. The membrane was then tested at 350°C for 200 h followed by a continuous He test of 300 h. Afterwards, H2 was fed once more; nevertheless, the membrane’s permeance reduced by ~50% while the He leak remained unchanged. The membrane test was reinitiated after 5760 h of test showing a stable permeance of 20 Nm3m-2h-1bar-0.5 and a He leak of 1.5 sccm/bar. It is important to mention, that at the beginning of the test, the membrane suffered from several cooling- heating cycles due to failures of the electric system; nevertheless, the He leak remained unaffected. The hydrogen permeance, on the other hand, reduced from 27 to 20 Nm3m-2h-1bar-0.5 after the temperature fluctuation occurred at 6,630 h. Afterwards, the permeance showed a steady decline reaching a steady state of 15 Nm3m-2h-1bar-0.5 at 8,000 h until 9,300 h. The test was terminated for maintenance of the system and reinitiated to test the membrane for another 1,000 hours. Finally, the cumulative test of MA-160, shown in Figure 5-4, was 10,700 h of continuous test with 3 repair
143
Chapter 5 An Integrated Experimental/Techno-Economic Analysis of the Lifetime of Pd/Au Membranes steps. The membrane showed a reduction in permeance of 62% but an excellent separation performance. Please notice that some of this permeance test has been previously presented [224].
Table 5-3. List of characteristics of MA-160.
* Dense Pd Final He leak* H2 Permeance Membrane Testing time [h] thickness [µm] [sccm-bar-1] [Nm3m-2h-1bar-0.5] MA-160 2950 4.6 14.5 45 MA-160b 1000 4.6 (rings) 45 43 MA-160c 6750 10.4 4.5 18
* Hydrogen permeance and helium leak at 450oC
144
Chapter 5 An Integrated Experimental/Techno-Economic Analysis of the Lifetime of Pd/Au Membranes
Figure 5-4. Hydrogen permeance and He leak tests of MA-160, and yellow lines indicating membrane oxidation stages.
145
Chapter 5 An Integrated Experimental/Techno-Economic Analysis of the Lifetime of Pd/Au Membranes 5.2.2 Membrane Lifetime Estimation
In order to assess the effect of the membranes’ lifetime on the proposed TPC and the LC
models; the experimental H2/He selectivities of the membranes at different times and thicknesses were used to calculate the hydrogen purity produced, as shown in Figure 5-5. Notice that the
produced H2 quality decreases with time caused by the sintering of Pd crystalizes at high operating temperatures which accelerates the pinhole formation at membranes’ surface [35]. This aging behavior is strongly dependent on the thickness of the membrane; for instance, as shown in Figure 5-5, membranes with high values of Pd layer thickness have lower decreasing rates on hydrogen purity compared to that with low thickness values. For example, the membrane with a 10.4 µm Pd thickness can sustain the hydrogen purity higher than 99.8% over 4500 hours longer than that of 2.7 µm thick Pd with 1400 hours, indicating that longer membrane lifetime at the specific hydrogen purity can be achieved by increasing the Pd layer thickness. Moreover, the hydrogen purity at various testing times can be fitted well by polynomial equations as presented in Figure 5-5. Notice that even though thicker membranes seem to prolong the lifetime of the membrane; the capital investment will also increase due to the need of higher Pd quantities. Furthermore, since the H2 flux of Pd-based membranes is determined by the membrane thickness (according to Fick’s law);
the H2 production will be reduced using thicker membranes as demonstrated in Figure 5-6. Focusing on the above issue, one of the key objectives is to evaluate the economic performance of separation modules using various membrane thicknesses by taking into account membrane lifetime and hydrogen production.
146
Chapter 5 An Integrated Experimental/Techno-Economic Analysis of the Lifetime of Pd/Au Membranes
Figure 5-5. Experimental produced H2 purity for various Pd thickness values at 450°C and a retentate pressure of 2 bar.
147
Chapter 5 An Integrated Experimental/Techno-Economic Analysis of the Lifetime of Pd/Au Membranes
Figure 5-6. Average hydrogen permeance with the standard deviation for Pd-based separation modules using various palladium layer thicknesses.
148
Chapter 5 An Integrated Experimental/Techno-Economic Analysis of the Lifetime of Pd/Au Membranes The hydrogen purity chosen for this study was of 99.0% as the baseline purity to investigate the effect of membrane thickness on economic performance of separation modules. With the aid of polynomial equations fitting the hydrogen purity trajectory versus the elapsed testing time, the estimated membrane lifetime at the hydrogen purity of 99.0% was calculated as listed in Table 5-4. Please notice that the membrane lifetime is extended by 16% when the Pd layer thickness is increased from 2.7 to 4.6 µm while increasing the Pd layer thickness from 2.7 to 10.4 µm leads to an increase of the membrane lifetime by 152%. Following the long-term testing of hydrogen permeance and the membrane lifetime estimation, the economic performance assessment is conducted and the results are presented and discussed in the next subsection.
Table 5-4. Lifetime estimation for Pd-based separation modules with various palladium thickness values.
Lifetime Pd layer thickness [µm] Purity [%] [hour] [year] 2.7 99.0 3113 0.36 4.6 99.0 3597 0.41 10.4 99.0 7856 0.90
149
Chapter 5 An Integrated Experimental/Techno-Economic Analysis of the Lifetime of Pd/Au Membranes 5.2.3 Economic Performance Assessment of Pd-based Separation Modules with Various Pd Layer Thickness Values
Considering all uncertain model inputs listed in Table 3-19, the economic performance of Pd-based separation modules is evaluated to evaluate the effect of different Pd layer thicknesses. The economic simulation results regarding various Pd layer thickness values are summarized in Table 5-5 in terms of expected values of the FCI/TCI/TPC probability distribution profiles. Please notice that an expected value as a probability-weighted average of the probability distribution profile represents an unbiased estimator under conditions of uncertainty, offering valuable pieces of information for economic performance evaluation and analysis. In Table 5-5, the expected values of the FCI/TCI increase as the Pd layer thickness increases, conforming to the intuitively expected behavior that higher Pd layer thickness values lead to higher capital investment due to more Pd needed. However, the expected value of the TPC decreases with the increase in Pd layer thicknesses, indicating that prolonged membrane lifetime contributed by high values of Pd layer thickness can effectively reduce the TPC by lowering the cost of membrane replacement.
Table 5-5. Pd-based separation module cost summary regarding fixed capital investment, total capital investment, and total product cost. Expected fixed Expected total Expected total
Pd layer capital investment capital investment product cost thickness [µm] [K$] [K$/m2] [K$] [K$/m2] [K$] [K$/m2] 2.7 168.68 8,327 198.73 9,810 241.04 11,899 4.6 168.74 8,330 198.81 9,814 240.82 11,887 10.4 168.93 8,339 199.03 9,825 239.92 11,843
As mentioned before, H2 production is also affected by the Pd layer thickness, recognized as a cost component placing significant impact on economic performance of separation modules. To evaluate the economic performance of separation modules by taking into consideration the lifetime of the membranes and hydrogen production, the focus now is placed on the LC for further
150
Chapter 5 An Integrated Experimental/Techno-Economic Analysis of the Lifetime of Pd/Au Membranes detailed analysis and discussion. The LC cumulative probability distribution profile for Pd-based separation modules considered in this study at various Pd layer thickness values is shown in Figure 5-7. Please notice that the distribution profiles and the corresponding statistical characterization provide valuable pieces of information regarding the various Pd layer thickness values in the present study.
• The probability (vertical axis) that the LC falls below a desirable cost target level (horizontal axis) as well as the complementary probability for the cost to be higher than the aforementioned target level can be easily inferred. For example, there is a 30% probability for the LC to be lower than 120 $/Kg, while in the 10.4 µm Pd layer thickness case and a 10% probability for the LC to be higher than 100 $/Kg in the 2.7 µm Pd layer thickness case. • The “value at risk” and the “value at opportunity” can be quantified at pre-specified
probabilistic levels. In our case, the “value at risk” is defined as the threshold-value of P95 representing that there is a 5% probability of incurring a cost higher than this value, while the “value at opportunity” indicates that there is a 5% probability of incurring a cost lower than a
P5 threshold-value.
In Figure 5-7, notice that the 2.7 µm Pd layer thickness case has the most attractive LC distribution profile, indicating that the hydrogen production has the more significant impact on the LC performance compared to the lifetime even though the TPC can be effectively reduced by the prolonged membrane lifetime as mentioned before. In light of the above results derived, reducing Pd layer thickness not only lowers the FCI/TCI on separation modules but also decreases the LC by increasing hydrogen production, allowing economic performance of separation modules to become appealing. A similar behavior has been shown in polymeric membranes where high gas permeability values dominate the economic performance outcome [225].
151
Chapter 5 An Integrated Experimental/Techno-Economic Analysis of the Lifetime of Pd/Au Membranes
Figure 5-7. Cumulative probability distribution profile of levelized cost for Pd-based separation modules with various Pd layer thickness values at 450°C and a retentate pressure of 50 bar.
152
Chapter 5 An Integrated Experimental/Techno-Economic Analysis of the Lifetime of Pd/Au Membranes Furthermore, within the experimental range of Pd layer thickness values tested, Figure 5-8 shows a comparative appraisal of LC for the various Pd layer thickness values on the average/expected value level as well as equal levels of probability associated with potential “value
at risk” and the “value at opportunity”. In Figure 5-8, the upper P95 line (a possible means of identifying “risk”) graphically shows that there is a 5% probability of incurring a cost higher than
the value on the line, and similarly, the lower P5 line (a possible means of quantifying “opportunity”) graphically shows the 5% probability of incurring a cost lower than the value of the line. Besides, the middle line provides the expected LC values for the various Pd layer
thicknesses considered. Furthermore, the P95, P5, and the expected value of LC for the separation modules using various Pd layer thicknesses are summarized in Table 5-6. It should be pointed out that not only does the expected LC value decreases when the Pd layer thickness reduces as noted earlier, but so does the spread/variability (quantified as P95 minus P5) of the LC – that is, the dispersion of possible economic performance outcomes. The observed reduction in the spread/variability of the LC with decreasing Pd thickness can be attributed to the fact that the extent of improvement on the LC by hydrogen production using thinner membranes is higher than that by the TPC using thicker membranes, enabling the spread/variability of the LC to be effectively reduced by higher hydrogen production using thinner membranes.
153
Chapter 5 An Integrated Experimental/Techno-Economic Analysis of the Lifetime of Pd/Au Membranes
Figure 5-8. P95, P5, and expected value lines of levelized cost for Pd-based separation modules with various palladium layer thickness values.
Table 5-6. P95, P5, and the expected value of levelized cost for a separation module using various Pd layer thicknesses.
Levelized cost [$/Kg] Pd layer thickness [µm] P5 Expected value P95 2.7 29.7 57.6 107.6 4.6 33.0 62.7 115.7 10.4 79.8 157.4 290.2
154
Chapter 5 An Integrated Experimental/Techno-Economic Analysis of the Lifetime of Pd/Au Membranes Additionally, a sensitivity analysis on the effects of uncertainty on the expected LC-value is performed via Tornado diagrams [54], as shown in Figure 5-9 and Figure 5-10. Tornado diagrams represent the relative impact of variations of the uncertain inputs in the LC model over their respective ranges assuming that all other inputs remain fixed at expected values. In these diagrams, each input symbolizes a “bar” as it fluctuates over the prearranged array and thus depicting its impact. Please notice that all bars are arranged from the long bars to shorter ones at the bottom [51], in order to prioritize the inputs with the greatest impact of on the LC-value. In this work, the two extreme membrane thicknesses presented (2.7 and 10.4 µm) were studied and sized against other model inputs. For 2.7 µm membranes, the inputs and ranges considered were: Pd unit price: 18.6-28.2 $/g; Au unit price: 36.5-57.0 $/g, raw materials to product cost ratio range: 10-20%; operating labor to product cost ratio range: 10-15%; plant overhead costs to product cost ratio range: 5-15%; distribution & selling costs to product cost ratio range: 2-20%; membrane lifetime: baseline ± 20%; hydrogen permeance: 38.7-54.8 Nm3/m2-h-bar0.5. For 10.4 µm membranes the inputs and ranges were: Pd unit price: 18.6-28.2 $/g; Au unit price: 36.5-57.0 $/g, raw materials to product cost ratio range: 10-20%; operating labor to product cost ratio range: 10- 15%; plant overhead costs to product cost ratio range: 5-15%; distribution & selling costs to product cost ratio range: 2-20%; membrane lifetime: baseline ± 20%; hydrogen permeance: 20.3- 14.8 Nm3/m2-h-bar0.5.
The Tornado diagrams for both membrane thickness-values show that distribution and selling costs of H2 is the most influential uncertainty followed by the raw materials, plant overhead costs, labor and the hydrogen permeance of the membrane. It is important to mention that, in both cases, the H2 permeance of the membranes is significantly more influential than the membrane lifetime since the LC is normalized by the H2 production rate. Nevertheless, when comparing Figure 5-9 and Figure 5-10, it is possible to observe that the influence of the membranes’ lifetime is more critical for the case of the thinner membrane; while the thicker membranes’ vulnerability towards uncertainty is more influenced by the Pd and Au unit price. Therefore, one can conclude that, using thinner membranes not only yields an appealing LC performance by lowering the capital cost and enhancing the hydrogen production, but also reduces the dispersion of possible economic performance outcomes since the LC is predominated by the hydrogen production rather than the membrane lifetime.
155
Chapter 5 An Integrated Experimental/Techno-Economic Analysis of the Lifetime of Pd/Au Membranes
Figure 5-9. Tornado diagram for the levelized cost of hydrogen produced via Pd-based separation modules using 2.7 µm thick membranes.
156
Chapter 5 An Integrated Experimental/Techno-Economic Analysis of the Lifetime of Pd/Au Membranes
Figure 5-10. Tornado diagram for the levelized cost of hydrogen produced via Pd-based separation modules using 10.4 µm thick membranes.
157
Chapter 5 An Integrated Experimental/Techno-Economic Analysis of the Lifetime of Pd/Au Membranes 5.3 Conclusion
This work shows the combination of long-term H2 permeance and He leak test results to predict the membrane lifetime and theoretical economic evaluation results to explore the tradeoff between permeance-thickness and lifetime in terms of economic performance outcomes. Four o membranes were tested at 350 and 450 C under pure H2 with occasional He leak tests for a cumulative time of 19,200 hours or an equivalent time of 2.2 years. Notice that a long-term (>1000 h) test was reported for a membrane with a thickness of 2.7 µm, which is the first time reported in
the literature. The experimental results of the produced H2 purity and elapsed testing times were analyzed to estimate the membranes’ lifetime at different membrane thicknesses. The lifetime of
the membrane was defined to target a H2 purity equal or greater than 99%. The economic evaluation framework for H2 separation units was developed by taking in consideration various uncertainty factors via Monte Carlo simulations. The expected values of the FCI/TCI increase as the Pd layer thickness increases since higher Pd is needed leading to a higher capital investment. Nonetheless, the expected value of the TPC decreases with higher Pd layer thicknesses, indicating that prolonged membrane lifetime effectively reduces the cost of membrane replacement. Furthermore, it is observed that the expected LC value as well as its spread/variability decreases as the Pd thickness is reduced; this effect is attributed to the higher hydrogen production using thinner membranes. Additionally, a sensitivity analysis showed, via Tornado diagrams, that high
H2 permeance is more significant than lifetime on the LC performance. In summary, using thinner membranes yields an appealing LC performance by lowering the capital cost and enhancing the hydrogen production.
158
Chapter 6
Chapter 6 A Cost Assessment Study for a Large-Scale Water-Gas Shift Catalytic Membrane Reactor Module in the Presence of Uncertainty A Cost Assessment Study for a Large-Scale Water-Gas Shift Catalytic Membrane Reactor Module in the Presence of Uncertainty
6.1 Introduction
Palladium-based catalytic membrane reactors (CMRs) represent an efficient technology option that increases the CO conversion in water-gas-shift (WGS) reactors and accordingly improves H2 recovery levels [226,227]. Subsequently, the production of hydrogen in CMRs via natural gas steam reforming (MSR) and water–gas shift (WGS) of the coal-derived syngas generates strong interest as an alternative clean process system. In particular, the application of membrane technology to the WGS reaction has been shown to be efficient technically and economically [228]. In a WGS-CMR module, the catalyst is confined within the reactor (retentate) along with a tubular palladium membrane situated throughout the reaction zone (Figure 6-1). The membrane continuously removes “in-situ” the H2 generated, altering the thermodynamic composition; this effect, known as process intensification, allows higher CO conversions. Notice that this approach yields conversions that exceed the ones commonly attained in conventional packed bed reactors. Additionally, the higher retentate pressure facilitates the acquisition of clean
159
Chapter 6 A Cost Assessment Study for a Large-Scale Water-Gas Shift Catalytic Membrane Reactor Module in the Presence of Uncertainty pressurized CO2 and water, enabling the process of carbon capture. For all the aforementioned technical features, Pd-based CMRs have shown to be an important technology option for the development of H2 economy [229]
Figure 6-1. Schematic of a CMR module for water gas shift reaction and photograph of the actual CMR rig built at WPI.
160
Chapter 6 A Cost Assessment Study for a Large-Scale Water-Gas Shift Catalytic Membrane Reactor Module in the Presence of Uncertainty In CMRs, the reaction and separation processes are conducted simultaneously, and therefore, the high- and low-temperature shift reactors along with a hydrogen separation unit, such as pressure swing adsorption (PSA) in the traditional hydrogen production system [44,230] can be eliminated from the structure of the overall process system. Additionally, the simultaneous
separation of H2 and CO2 is particularly appealing due to increasing efforts focusing on the reduction of carbon dioxide emissions [5,231]. When compared to the conventional hydrogen production process, a CMR allows lower temperatures of operation, while still reaching higher conversion levels; this enables prolonged catalyst lifetime, lower production costs, reduced material costs for the reactor and a facilitated route for carbon capture. Moreover, process intensification in CMR technology provides compactness, modularity, reduced equipment size to production capacity ratio, practical assembly/disassembly capabilities and operational flexibility [31]. Additional advantageous features include the good allocation of material and energy resources, waste management and a superior environmental performance [147].
Many studies have been reported regarding the potential industrial applications of Pd-based CMR technology, but their results rely on a proof-of-concept-centered approach [232,233], giving rise to significant challenges for techno-economic performance evaluation in the absence of any accumulated operating experience at the commercial scale and reliable data. Furthermore, the few large-scale tests of CMRs under actual industrial conditions that have been published do not offer comprehensive sets of pertinent techno-economic data. The study by Catalano et al. [30] showed that H2 production of 1.2 lb/day with purity higher than 99% can be reached through a WGS-CMR using a gas mixture similar to an actual syngas composition. The surface area of the membrane used in this work was 200 cm2 with operating conditions corresponding to a temperature range of 420-440oC and a retentate pressure of 20 bar. In Figure 6-1, a photograph of the CMR rig shows its different components, including the mass flow and pressure controllers, the preheater, the heat exchangers, and back pressure regulators. Furthermore, Patrascu et al. [234] reported the use of a
large-scale CMR for methane steam reforming (MSR) capable of achieving a H2 permeate flux of 1.6 NL/min while utilizing a membrane with a surface area of 175 cm2 and being operated at temperatures within the 440-525oC range and at a pressure of 10 bar. Moreover, Ma et al. [171,172] demonstrated the satisfactory performance characteristics of different Pd and Pd/alloy membranes under actual coal-derived syngas and industrial scenarios but solely for H2 purification purposes.
161
Chapter 6 A Cost Assessment Study for a Large-Scale Water-Gas Shift Catalytic Membrane Reactor Module in the Presence of Uncertainty The present research study focuses on building upon the promising results reported in [146,147,196] in order to evaluate the profile of economic performance outcomes generated by the large-scale WGS-CMR module reported by Catalano et al. [30] under the aforementioned uncertainties. The objective of the present study is to obtain realistic profiles of economic outcomes and other valuable information by explicitly acknowledging irreducible uncertainty sources, and therefore, inform potential demonstration and deployment efforts of CMR technology in an industrial setting.
162
Chapter 6 A Cost Assessment Study for a Large-Scale Water-Gas Shift Catalytic Membrane Reactor Module in the Presence of Uncertainty 6.2 Results and Discussion
An economic performance assessment of an actual large-scale Pd-based CMR module has been conducted with the integration of standard Monte Carlo simulation methods that take into account inherent sources of uncertainty. Figure 6-2 shows cumulative probability distribution profiles as well as associated expected values for the FCI and TCI demonstrating the economic performance of this new technology option in a statistical and more realistic manner compared to traditional single-point-based cost studies. In particular, it is easy to observe that there is a 20% chance for the TCI to be lower than $400 K, while there is a 10% probability for the FCI to be higher than $450 K. In addition, it is useful to quantify the “value at opportunity” and the “value at risk” at a predetermined probability level for further study and discussion. In the present study, the “value at opportunity” is defined as P5, representing the 5 % probability for a capital investment cost to be lower than the P5 threshold-value; while the “value at risk” is characterized as P95, showing the probability for the capital investment cost to be higher than the 5% threshold-value. Furthermore, the expected values of the TCI and FCI (ETCI/EFCI) have an outcome frequency of 52% as shown in Figure 6-2. The features of the cumulative distribution profiles for EFCI, ETCI,
P95 and P5 for the CMR module are summarized in Table 6-1.
The module cost per membrane unit area for the actual large-scale system, P95, P5, and the expected values for the FCI/TCI shown in the present study surpass the industrial-scale CMR module cost previously reported [147]. The ETCI for an industrial-scale CMR module [147] is $18,805/m2 which is approximately 1,200 times lower than the ETCI delineated in this work or $22,418 K/m2. Based on the six-tenths factor rule [235],the cost reduction in large-scale units can be attributed to a scale-up effect, which induces reasonable cost-reducing prospects on equipment costs at higher capacities. Furthermore, the design of a multiple tube membrane module and the elimination of the potentially overlapping auxiliary process components and assorted apparatus contribute further to the discrepancies in economic outcomes between the industrial- and large- scale CMR modules. It should be pointed out that the key cost components of the FCI/TCI models are strongly dominated by the purchased equipment cost and consequently, the lower module cost per membrane area presented in an industrial-scale seems reasonable.
163
Chapter 6 A Cost Assessment Study for a Large-Scale Water-Gas Shift Catalytic Membrane Reactor Module in the Presence of Uncertainty
Figure 6-2. Cost distribution of a WGS-CMR module.
Table 6-1. WGS-CMR module cost summary.
Fixed capital investment Total capital investment [K$] [K$/m2] [K$] [K$/m2]
P95 472.4 23,319 560.4 27,664 Expected value 385.5 19,028 454.1 22,418 P5 305.3 15,070 357.5 17,649
164
Chapter 6 A Cost Assessment Study for a Large-Scale Water-Gas Shift Catalytic Membrane Reactor Module in the Presence of Uncertainty The costs associated with economies of scale were further studied by investigating the effect of capacity on the normalized TCI of WGS-CMR modules, as shown in Figure 6-3. The upscaling factor, previously computed as the six-tenths factor rule, was included in the stochastic model as one of the uncertain drivers with a reasonable range of 0.4 to 0.8 and a most likely value of 0.6 [178]. Changes in the upscaling factor resulted in an exponential decrease in the normalized TCI of CMR modules (Figure 6-3). For example, when the capacity is increased from 1.2 lb/day to 1000 lb/day, the normalized TCI is reduced by 92% from 378,458 $/(lb/day) to 29,916 $/(lb/day). Additionally, Figure 6-3 clearly shows that when the capacity is increased to the aforementioned industrial-scale CMR [147], the TCI of the module lays accordingly (shown as an individual red point). These results clearly validate the results of previously published work in CMRs [146,147,196,236], since actual techno-economic features were used in this study. It is suspected that the slight discrepancy between the expected value of this work and that reported previously [147] at the industrial scale is caused by the reduction of raw material costs and/or certain features associated with inventories of scale. Notice that the spread of the P95 and P5 values
(P95-P5) becomes more obvious as the capacity increases, implying that the risk increases accordingly, possibly due to unknown factors of upscaling such as safety, land, regulatory parameters and others.
In addition, the results presented in Figure 6-3 imply that scaling up the module can effectively reduce the levelized cost of hydrogen defined as the normalized total product cost (TPC) by the hydrogen production. Notice that the TPC greatly depends on the TCI as standard practice in the economic analysis of engineering systems suggests [50,196]. For instance, the financing interest in the TPC is about 6-10% of the TCI.
165
Chapter 6 A Cost Assessment Study for a Large-Scale Water-Gas Shift Catalytic Membrane Reactor Module in the Presence of Uncertainty
Figure 6-3. Normalized Total Capital Investment of WGS-CMR modules for various capacities.
166
Chapter 6 A Cost Assessment Study for a Large-Scale Water-Gas Shift Catalytic Membrane Reactor Module in the Presence of Uncertainty The effects of uncertainty on the ETCI value for the large-scale CMR module are analyzed through the Tornado diagram presented in Figure 6-4. In such a manner, a Tornado diagram graphically illustrates the relative impact on the TCI profile of variations of the different uncertain model inputs considered in the present study. Specifically, the change of the ETCI caused by the variation of each input over its prescribed range from the lowest to the highest value is depicted with a bar in the Tornado diagram, under the assumption that all other model inputs remain at their baseline values. Afterwards, all bars are sorted from long to short, in order to clearly distinguish between the relative impact of various uncertain model inputs on TCI-relevant performance characteristics [51]. Therefore, the major advantage of using a Tornado diagram in this study is to illustrate which uncertain model inputs have the greatest impact on the expected value of the TCI.
Figure 6-4. Tornado diagram for Total Capital Investment.
167
Chapter 6 A Cost Assessment Study for a Large-Scale Water-Gas Shift Catalytic Membrane Reactor Module in the Presence of Uncertainty The ETCI-Tornado diagram of the large-scale CMR module is based on the following values: Pd unit price: 17.7-28.2 $/g; electroless plating setup cost: $1000-2000; operating labor to raw materials ratio range: 12.5-50%; direct supervisory and clerical labor to operating labor ratio range: 10-20%; utilities to raw materials ratio range: 12.5-50%; maintenance and repair to electroless plating setup cost ratio range: 2-10%; operating supplies to electroless plating setup cost ratio range: 0.5-1%; laboratory charges to operating labor ratio range: 10-20%; patents and loyalties to raw materials ratio range: 0-15%; installation to purchased equipment ratio range: 25- 55%; instrumentation and controls, installed to purchased equipment ratio range: 8-50%; piping, installed to purchased equipment ratio range: 10-80%; electrical, installed to purchased equipment ratio range: 10-40%; buildings, process and auxiliary to purchased equipment ratio range: 10-70%; service facilities and yard improvements to purchased equipment ratio range: 40-100%; land to purchased equipment ratio range: 4-8%; engineering and supervision to direct cost ratio range: 5- 30%; legal expenses to fixed capital investment ratio range: 1-3%; construction expense and contractor’s fee to fixed capital investment ratio range: 10-20%; contingency to fixed capital investment ratio range: 5-15%; working capital to total capital investment ratio range: 10-20%; capacity of purchased equipment: ± 20%.
As shown in Figure 6-4, the capacity of purchased equipment has the most significant impact on the TCI performance, suggesting that the purchased equipment cost greatly determines the value of TCI. Please also notice that the TCI increases by 11% when the capacity of purchased equipment is increased by 20%, while the TCI decreases by 13% when the capacity of purchased equipment is reduced by 20%, which is indicative of the significant impact of the CMR module capacity on TCI performance. As a result, since the purchased equipment costs for a larger capacity, such as in an industrial-scale CMR module case, would be significantly reduced from those presented in the large-scale CMR module; a significant reduction in cost for an industrial- scale CMR seems indeed a valid conjecture. Additionally, membrane features such as Pd unit price and operating labor show a marginal effect on TCI performance when compared to the capacity of purchased equipment. Indeed, the potential economy of scale has shown to significantly influence the economic profile characteristics of CMR technology; however these features can also be the result from learning curves, spreading of set-up costs and certain stochastic processes associated with inventories [237].
168
Chapter 6 A Cost Assessment Study for a Large-Scale Water-Gas Shift Catalytic Membrane Reactor Module in the Presence of Uncertainty Scaling up the capacity of CMRs appears to be the next step towards the deployment of this innovative technology. Nonetheless, it is important to consider that the increase in size of a plant/technology is not totally influenced by capital costs, market growth or the magnitude of investments, but rather by a steady technological progress over time, as shown by Lieberman [238]. Technological progress over time is captured by learning or experience curves which describe the reduction of capital costs due to upscaling, mass production, process improvements, learning-by-doing, and reduction in raw material costs [239]. In this work, TCI learning curves of the CMR technology were generated taking into consideration the inherent uncertainties reflected in Equation 6-1 [240,241]:
TCI (q) = TCI × q 6-1 α 0 where TCI (q) is the total capital investment [K$] after learning a cumulative production of WGS- CMR modules (q), TCI is the initial total capital investment of the WGS-CMR module [K$] and
is the learning index.0 Notice that the learning index was estimated based on statistical progress
αratios reported by Christiansson [240], and assuming that the WGS-CMR module belongs to the category of “Modules.” The progress ratio was treated as one of uncertain drivers of the TCI estimation modeling framework with a maximum, average and minimum value of 0.7, 0.83, and 0.95, respectively. It is important to mention that the progress ratios considered in this work were assumed to be stable over time and that the phase of plant (start-up or production) did not influence the rate of learning.
The TCI learning curves, represented in Figure 6-5, depict exponential reductions as the cumulative production of the modules increase, indicating that process improvement and learning- by-doing significantly affect the economic features of a developing technology option such as WGS-CMR modules. For example, when the cumulative production of modules increases to 100; there is a 70% reduction in the TCI from $454,149 to $137,692. The factors lying behind a learning curve are associated with process improvement and learning-by-doing, such as labor efficiency, system redesign, improved manufacturing methods, new cost-effective materials, and improved construction efficiency [239]. Furthermore, although, experience in manufacturing reduces the risk in any given technology option; Figure 6-5 shows an increase in the spread of P95 and P5 as the
cumulative number of produced CMR module increases. The increase in risk (P95-P5) as more units
169
Chapter 6 A Cost Assessment Study for a Large-Scale Water-Gas Shift Catalytic Membrane Reactor Module in the Presence of Uncertainty are produced is caused by assuming constant progress ratios over time; in other words, it is assumed that the start-up and the production stage have the same learning curves.
The combination of manufacturing experience and the upscaling process will lead to significant cost reductions. Therefore, it is recommended that future research efforts focus on generating further operating experience and plant data at a pilot/commercial scale as well as studies on CMR manufacturing.
Figure 6-5. Experience curve for Total Capital Investment of WGS-CMR modules.
170
Chapter 6 A Cost Assessment Study for a Large-Scale Water-Gas Shift Catalytic Membrane Reactor Module in the Presence of Uncertainty 6.3 Conclusion
A comprehensive economic evaluation framework for an actual large-scale catalytic membrane reactor module used for water-gas-shift (WGS) reaction has been developed through detailed baseline models for the Fixed Capital Investment (FCI) and the Total Capital Investment (TCI) under various sources of uncertainty. The effect of these uncertainty sources has been taken into account through the integration of Monte Carlo simulation methods. Accordingly, distribution profiles of these economic indicators have been derived rather than single-point value estimates, and a more realistic distribution of CMR economic performance outcomes has been generated through which “risks” and “opportunities” were identified. The results shown for the actual large- scale CMR displays much higher expected values for the FCI/TCI than that of an industrial-scale, indicating that TPC profiles become economically appealing when the capacity of the CMR module increases. Additionally, it is shown that the capacity of purchased equipment is the most critical factor in the TCI performance. This work effectively validated previous techno-economic results on CMR technology, utilizing actual data and extrapolating the economics of scale. A Tornado diagram has been developed to identify the relative impact of the various uncertainty drivers on TCI, as well as elucidate the sensitivity of the ETCI with respect to equipment capacity. Finally, a TCI learning curve for the CMR modules has been generated to demonstrate the importance of technological progress over time on the economic features of this innovative technology.
171
Chapter 7
Chapter 7 Integration of Membrane Technology into Hydrogen Production plants with CO2 Capture: An Economic Performance Assessment Study Integration of Membrane Technology into Hydrogen Production Plants with CO2 Capture: An Economic Performance Assessment Study
7.1 Introduction
In the absence of any accumulated operating experience and associated data, an economic performance evaluation framework for a new technology option such as Pd and Pd/Alloy-based catalytic membrane reactor (CMR) modules potentially integrated into H2 production (HP-CMR) process systems (HP-CMR) would be inevitably based on reasonable theoretical estimates (by drawing on the most comprehensive studies and expert opinion) while explicitly acknowledging irreducible uncertainty sources (market, regulatory, etc.) [50]. In the present study such a methodological and analytical perspective is followed. First, a detailed comprehensive baseline net present value (NPV) model is structured in order to assess the economic performance characteristics of HP-CMR plants. Then various sources of uncertainty are identified and integrated into the proposed framework, while their impact on the HP-CMR plant’s valuation profile is explicitly taken into account through Monte Carlo simulation methods by which the above uncertainty “drivers” (model inputs) are propagated through the above NPV-model
172
Chapter 7 Integration of Membrane Technology into Hydrogen Production Plants with CO2 Capture: An Economic Performance Assessment Study [50,189]. As a result, the proposed economic performance evaluation framework for the HP-CMR plant allows the derivation of more realistic distribution profiles of economic performance outcomes rather than single-point value estimates that often lead to unsatisfactory valuation assessments by overlooking significant uncertainty effects over the plant’s lifetime [50,189]. Furthermore, a comparative economic performance assessment in the presence of uncertainty is conducted between the HP-CMR and the conventional hydrogen production options, demonstrating that appealing economic performance outcome distribution profiles of HP-CMR plants could emerge as regulatory action on CO2 emissions is introduced.
173
Chapter 7 Integration of Membrane Technology into Hydrogen Production Plants with CO2 Capture: An Economic Performance Assessment Study 7.2 Results and Discussion
7.2.1 Capital Investment Cost Estimation under Uncertainty: the CMR Module Case
A CMR module cost analysis in the presence of uncertainty is conducted with the integration of standard Monte Carlo simulation methods into the FCI/TCI model presented earlier. The cumulative probability distribution FCI and TCI profiles as well as the associated expected values for industrial scale CMR modules are presented in Figure 7-1. These profiles and the corresponding statistical characterization possibilities provide valuable pieces of information with respect to potential economic performance outcomes [196]. For example, it is possible to observe that there is a 10% probability for the FCI to be lower than 225 M$ and a 90% probability for the TCI to be higher than 400 M$. The “value at risk” and the “value at opportunity” are quantified at pre-specified probabilistic levels of 95% and 5% (P95 and P5) respectively. A threshold-value of
P95 represents that there is a 5% probability of incurring a cost higher than this value, while there is a 5% probability of incurring a cost lower than a P5 threshold-value. P95, P5, and expected values of FCI/TCI for the CMR module are summarized in Table 7-1.
174
Chapter 7 Integration of Membrane Technology into Hydrogen Production Plants with CO2 Capture: An Economic Performance Assessment Study
Figure 7-1. Cost distribution of an industrial scale CMR module.
Table 7-1. Industrial scale CMR module cost summary.
Fixed capital investment Total capital investment [M$] [$/m2] [M$] [$/m2]
P95 354.3 20,275 419.7 24,018 Expected value 279.1 15,969 328.6 18,805 P5 212.5 12,161 249.6 14,284
175
Chapter 7 Integration of Membrane Technology into Hydrogen Production Plants with CO2 Capture: An Economic Performance Assessment Study 7.2.3 Economic Assessment of Hydrogen Production Technology Options
The technical performance of HP-CMR plants is first comparatively evaluated in terms of coal feed requirements and CO2 production and emission levels against conventional HP plants that utilize PSA as a purification unit (HP-PSA plants). Pertinent results are graphically shown in Figure 7-2. Correspondingly, plant specifications and capital investment costs associated with all technology options considered in this study are tabulated in Table 7-2. These results show that the higher efficiency displayed by the HP-CMR plant reduces the coal feed requirements to achieve the same H2 production levels by 16% compared to the conventional HP-PSA plants. The reduction in the coal feed requirements is attributed to process intensification-relevant benefits attributed to CMRs in which the WGS reaction and separation are simultaneously carried out [242].
Furthermore, the lower CO2 production level is directly reflected on the lower carbon feed requirements in the HP-CMR case, as shown in Figure 7-2. Please also notice that HP plants with integrated CCS systems lead to a substantial reduction of the amount of CO2 emissions, but their degree of efficiency varies. The HP-PSA plant has an efficiency of 90% while the HP-CMR plant
displays an efficiency of 98%. Considering both the lower CO2 production levels and the higher
CO2 capture efficiency achieved by the HP-CMR technology option, the HP-CMR plant emits
0.07 Mtonnes of CO2 per Mtonnes of coal feed compared to a value of 0.23 for the conventional
HP plant, leading to an overall reduction in CO2 emissions of 70%. Other than higher levels of technical performance displayed by the HP-CMR plant, lower capital investment costs for the
same H2 production target level can be also achieved using the HP-CMR technology option as seen in Table 7-2. Notice that the benefits of the HP-CMR technology option lead to a reduction of the TCI costs for HP by 26%. Indeed, process intensification opportunities combined with the
simultaneous CO2 capture and extra purity H2 production enabled by CMRs allow a significant reduction in the pertinent TCI costs by reducing the size of the process equipment needed and replacing the traditional WGS reactors, PSA as well as Selexol units.
176
Chapter 7 Integration of Membrane Technology into Hydrogen Production Plants with CO2 Capture: An Economic Performance Assessment Study
Figure 7-2. Coal feed, CO2 production, and CO2 emissions of hydrogen production plants for various technology options.
177
Chapter 7 Integration of Membrane Technology into Hydrogen Production Plants with CO2 Capture: An Economic Performance Assessment Study Table 7-2. Hydrogen production plant specifications and capital investment costs for various technology options.
TCI [B$] Plant size H2 production Coal feed CO2 production CO2 captured CO2 emissions [TPD] [TPD] [TPD] [TPD] [TPD] [TPD]
HP-PSA plant w/o CCS sys.a 1.82 685 616.5 5,302 12,137 0 12,137 HP-PSA plant w CCS sys.a 2.14 685 616.5 5,302 12,137 10,954 1,183 HP-CMR plant w CCS sys. 1.58 685 616.5 4,384 9,936 9,737 199 a For the pertinent data sources of the HP-PSA plants, please see DOE/NETL report [44].
178
Chapter 7 Integration of Membrane Technology into Hydrogen Production Plants with CO2 Capture: An Economic Performance Assessment Study Considering all uncertain model inputs listed in Table 3-21, the economic performance of HP plants and range of potential outcomes are comparatively evaluated amongst the various technology options by conducting detailed simulation studies. As it will be demonstrated below, the inclusion of regulatory action on car-bon has a significant effect on the NPV distribution profiles of all technology options considered in the present study, generating comparative economic performance advantages and noticeably promising prospects in the HP-CMR case. Indeed, it is shown that HP-CMR could become an economically appealing technology option under a certain set of regulatory conditions, thus deserving attention as a potentially meritorious option whose realization prospects of an initial fleet of demonstration plants at the commercial scale ought to be seriously considered.
The simulation results of NPV distribution profiles in the absence and presence of regulatory action on carbon are presented in Figure 7-3 (a) and (b), respectively. It should be
pointed out that the initial CO2 tax rate of $30 per tonne of CO2 considered represents a reasonable first estimate based on current market conditions and average “best” estimates provided by experts
[243]. Furthermore, $30 per tonne of CO2 indicates an initial tax level higher than the CO2 abatement cost, thus providing an incentive for HP plant owners/operators to invest in CCS systems. At this point, it should be pointed out that a price on emissions (of any type) conforming to a simple stochastic model that makes use of an initial emissions price and an annual growth rate (such as the one presented) could have been considered within the context of the present research study. We chose the “carbon tax” term since such a type of emissions price is now widely acknowledged of generating the appropriate price signals that can be internalized by the pertinent markets in the most direct and seamless manner [25,48]. In Figure 7-3 (a), notice that in the absence of CO2 tax, the conventional HP-PSA plant without any CCS system installed has the most attractive NPV distribution profile, and therefore, the graphically depicted results naturally reproduce the current state of affairs in the field of hydrogen production. Moreover, it also confirms the fact that the capital investment and O&M costs associated with the installation and operation of a CCS system would result in less attractive economic outcome profiles in all cases
in the absence of a CO2 tax. However, under the conditions considered and in the presence of regulatory action on carbon the NPV distribution profile of HP-CMR emerges as a comparatively appealing one. Indeed, one observes that key performance assessment metrics deduced from the above NPV-distribution profiles such as ENPV, P95 and P5-threshold values (“value at
179
Chapter 7 Integration of Membrane Technology into Hydrogen Production Plants with CO2 Capture: An Economic Performance Assessment Study opportunity”, “value at risk”) are improved compared to the ones associated with the NPV distribution profiles of the traditional technology option with CCS. This outcome can be jointly attributed to the fact that superior efficiency gains in the HP-CMR case coupled with higher regulatory cost savings (versus CCS-operating costs) realized when compared to the traditional option result in enhanced economic performance in an uncertain carbon-constrained world over the plant’s lifetime. In light of these findings, a stronger justification basis could be built for the provision of the right incentives aiming at the realization of technology demonstration projects at the commercial scale for a technology option such as HP-CMR, thus supporting broader energy and environmental policy goals briefly presented in the Introduction section.
Finally, it would be worth noticing that the conventional HP-PSA plant without a CCS system appears to be the most sensitive to future regulatory action, while the HP-CMR plant appears to be less sensitive due to the fact that the degree of change of the NPV distribution profiles is mainly determined by the “carbon tax penalty” which is computed through the amount of CO2
emissions and the CO2 tax rate (thus allowing the HP-CMR plant to exhibit more appealing economic performance outcomes under the stated regulatory conditions).
180
Chapter 7 Integration of Membrane Technology into Hydrogen Production Plants with CO2 Capture: An Economic Performance Assessment Study
(a)
(b) Figure 7-3. Cumulative probability distribution profiles of the Net Present Value for various technology options in (a) the absence of CO2 tax and (b) under an initial tax rate of $30 per tonne of CO2.
181
Chapter 7 Integration of Membrane Technology into Hydrogen Production Plants with CO2 Capture: An Economic Performance Assessment Study The effects of uncertainty on the expected NPV-value (ENPV) for the various technology options integrated with CCS systems are also analyzed through Tornado diagrams and presented in Figure 7-4. A Tornado diagram summarizes graphically the relative impact of variations of the numerous uncertain NPV-model inputs over their respective ranges (from the lowest to the highest value) under the assumption that all other model inputs remain at their baseline values (fixed at their expected values). In a Tornado diagram each model input is represented by a bar as it varies over its prescribed range depicting its impact on the above performance indexes, and all bars are sorted from the long ones at the top to the shorter ones at the bottom in a decreasing impact order [51]. In this manner, a Tornado diagram prioritizes the most consequential inputs (sources of uncertainty) in terms of bar length and facilitates the visualization of their impact on economic performance, primarily in cases of large-scale multi-input models. Figure 7-4 (a) and (b) provide representative ENPV-Tornado diagrams for the case of the traditional HP as well as the HP-CMR option. The uncertain model inputs considered and their respective prescribed ranges used include the following: Pd unit price: 16.9–25.9$/g, Au unit price: 39.4–53.7$/g, coal price: 61.0–
48.5$/tonne, inflation rate: 1.46–3.16%, H2 delivery cost: 3.0–4.4$/kg, Pd/Au membrane lifetime:
1–5 years, CO2 transport & storage cost: 9–11$/tonne, annual growth rate of CO2 tax: 5.4–6.6%,
H2 selling price: 9–11$/kg, combined state and local sales tax rates: 0–9.5%, financing interest to total capital investment ratio range: 6–10%, and plant over-head costs to total product cost ratio range: 5–15%. Please notice that hydrogen selling price, hydrogen delivery cost, plant over-head costs, financing interest costs, taxation levels and inflation rate constitute the set of model inputs with the highest impact on ENPV.
182
Chapter 7 Integration of Membrane Technology into Hydrogen Production Plants with CO2 Capture: An Economic Performance Assessment Study
(a) HP-PSA plant w CCS sys.
(b) HP-CMR plant w CCS sys.
Figure 7-4. Tornado diagrams for the NPV of various technology options in the presence of CO2 tax.
183
Chapter 7 Integration of Membrane Technology into Hydrogen Production Plants with CO2 Capture: An Economic Performance Assessment Study As shown in Figure 7-4, hydrogen selling price has the most significant impact on NPV in both cases, suggesting that economic performance of HP plants greatly depends on the revenue generated through hydrogen sales. Additionally, since the H2 selling price is mainly determined by production costs [184], production cost disruption (referring to situations where production costs undergo significant departures from their expected values used to design the coordination scheme which aligns the plans and objectives of the project undertaken [244]) can significantly influence the H2 economy by shifting selling prices. However, in the short and medium term,
production cost disruptions in the H2 economy are unlikely to happen unless alternative energy
prices become increasingly competitive, H2 demand grows rapidly and/or a breakthrough in H2 production technology occurs. Finally, when comparing the two cases, it is found that HP-PSA plants appear to be more sensitive to the selling price than HP-CMR plants, implying that the effect of the selling price on NPV-based performance evaluation is jointly determined by production costs and the nominal discount rate of the various HP options.
H2 delivery cost is another uncertainty driver that exerts a significant effect on the NPV of HP plants. The delivery cost is determined by the geographic and market characteristics, such as transport distance, delivery chain, density demand, etc. Yang and Ogden [245] reported that compressed gas truck delivery is ideal for small stations with low demand, while cryogenic liquid
trucks of liquefied H2 is preferred for long distances and robust, healthy demand patterns. Within such a context, any geographic and market characteristics should be considered and carefully evaluated for customization of H2 delivery in order to enhance the economic performance of HP
plants. Furthermore, it should be pointed out that considerably high costs for H2 delivery would drive the ENPV value of HP-PSA plants to become negative, while HP-CMR plants can withstand the effect of delivery costs to maintain a positive ENPV due to their production cost.
The lifetime of Pd/Au membranes is an uncertain model input that affects the ENPV of HP-CMR plants. A long lifetime would reduce the total product cost by lowering the membrane replacement cost, thus increasing the plant’s ENPV. Since the replacement cost is jointly determined by the membrane cost and lifetime, the effect of lifetime is also correlated with the Pd unit price as well as the membrane thickness. When the membrane cost is relative low by using thinner membranes, the impact of the membranes’ lifetime on the ENPV could be higher than the Pd unit price (see Figure 7-4 (b)); in contrast, when the membrane cost dominates the membrane
184
Chapter 7 Integration of Membrane Technology into Hydrogen Production Plants with CO2 Capture: An Economic Performance Assessment Study replacement cost in cases of thicker membranes use, the Pd unit price could have a greater effect on ENPV than membrane lifetime [196]. It should be pointed out that currently the fabrication of thin membranes with high durability characteristics is attainable and has been demonstrated by Augustine et al. [38]. Moreover, lifetime extension can be also achieved by repairing defects on the membranes and/or through post-treatment for reactivation of contaminated membranes [223,246,247]. Given the current pace of advancements in membrane science and technology, it is reasonably expected that the effect of membrane lifetime could be further minimized in the future. Please notice that this feature was not included in the present evaluation framework.
Comparing the Tornado ENPV diagrams of the traditional HP-PSA plant to an HP-CMR plant, one infers that the HP-PSA plant appears to be more sensitive to financing interest costs, while HP-CMR plants appear to be more susceptible to combined state and local sales tax rates. Please notice that federal corporate tax rates, as mentioned before, depend on the reportable income and therefore are not shown as uncertainty drivers in the Tornado diagrams; however, they are expected to have a high impact on the ENPV of the plants. Indeed, higher capital investment costs for HP-PSA plants (Table 7-2) generate higher financing interest costs. In contrast relatively lower capital investment costs for HP-CMR plants could mitigate the effect of corresponding financing interest costs on ENPV. Please also notice that combined with a relatively low total product cost, tax rates become critical in the calculation of net cash flows. As a result, tax rates can have a more substantial impact on the ENPV than financing interest costs under low capital investment and total product costs. Focusing on the effect of taxation levels, tax incentives have been occasionally introduced by states to influence business relocation, expansion, or start-up formation-relevant decision-making [248]. However, tax incentives in isolation could not determine “optimal” technology choices and plant locations since they do not represent the only criterion in an inherently complex decision-making process (other factors include developments in hydrogen markets and distribution challenges, macroeconomic conditions, state of regulatory framework and geopolitical stability).
The effect of regulatory action (with an initial CO2 tax rate starting on 2015) on ENPV is further examined in the following three technology option cases: (i) Case A refers to an HP-PSA plant without any CCS system installed; (ii) Case B corresponds to an HP-PSA plant with a CCS system installed, and (iii) Case C corresponds to an HP-CMR plant with a CCS system installed.
185
Chapter 7 Integration of Membrane Technology into Hydrogen Production Plants with CO2 Capture: An Economic Performance Assessment Study
The ENPV in the three cases considered is computed and tabulated at different initial CO2 tax rates as shown in Table 7-3. From this table, it can be inferred that the ENPV of conventional
HP plants without CCS varies significantly as the initial CO2 tax rate increases, becoming
economically problematic when the rate reaches $65 per tonne of CO2 and the associated regulatory compliance costs become quite significant. However, conventional HP plants with CCS
systems do not display negative ENPVs within the range of CO2 tax rates considered since the regulatory compliance cost savings exceed the costs associated with the operation of the CCS system.
The technology options considered were then evaluated under a set of different tax rates in order to visualize their impact on the associated ENPVs as shown in Table 7-4. First an economic performance comparison involving Cases A and B was performed and labeled as the “Traditional technology option.” This case demonstrates the economic effect of having a CCS system installed in a traditional HP plant under future regulatory action on CO2 emissions. As shown in Table 7-4, the ENPV difference favors the use of CCS systems when the tax rate is higher than $25 per tonne
of CO2. Lower tax rates do not seem to incentivize the integration of CO2 capture technologies in HP plants. Similar conclusions have been previously drawn and discussed in the context of coal- fired power plants with CCS units installed [146,243,249].
186
Chapter 7 Integration of Membrane Technology into Hydrogen Production Plants with CO2 Capture: An Economic Performance Assessment Study
Table 7-3. ENPV performance (B$) of technology options under various CO2 tax scenarios.
Initial CO2 tax rate [$/tonne CO2] Case Case Explanation 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 Case A HP-PSA plant w/o CCS sys. 2.17 2.01 1.84 1.67 1.50 1.34 1.16 0.99 0.81 0.63 0.45 0.26 0.07 -0.13 -0.33 Case B HP-PSA plant w CCS sys. 1.32 1.31 1.29 1.27 1.26 1.24 1.23 1.21 1.19 1.18 1.16 1.14 1.13 1.11 1.09 Case C HP-CMR plant w CCS sys. 1.62 1.61 1.61 1.61 1.61 1.61 1.60 1.60 1.60 1.60 1.60 1.59 1.59 1.59 1.59
Table 7-4. ENPV difference (B$) between two technology options under various CO2 tax scenarios.
Initial CO2 tax rate [$/tonne CO2] ENPV difference 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70
Traditional technology -0.85 -0.70 -0.55 -0.40 -0.25 -0.09 0.06 0.22 0.38 0.54 0.71 0.88 1.06 1.24 1.43 option (Case B – Case A) Membrane technology -0.56 -0.39 -0.23 -0.06 0.10 0.27 0.44 0.61 0.79 0.96 1.15 1.33 1.52 1.72 1.92 option (Case C – Case A)
187
Chapter 7 Integration of Membrane Technology into Hydrogen Production Plants with CO2 Capture: An Economic Performance Assessment Study The second economic comparison, performed in this study, was between Case C and A, which was labeled as “Membrane technology option.” This option shows the economic effect of
integrating membrane technology into HP plants in the presence of future CO2 emission regulations. Based on the proposed NPV-based assessment framework, the analysis indicates that Case C, i.e. the HP-CMR option becomes comparatively economically appealing when the
regulatory tax rate reaches $15/tonne of CO2. Therefore, it is possible to conclude that, depending on the course of action of future regulatory policies on carbon, there is a window of opportunity for HP-CMR to be perceived as an economically robust, if not superior, competitor to the incumbent technology option exhibiting also enhanced environmental performance in a carbon- constrained world. However, in the absence of regulatory action, the conventional HP options would remain uncontested when economic metrics dominate decision making in selecting the best technology options. It would be also worth noting that the HP-CMR option, when compared to the traditional HP with CCS, attains a positive ENPV at lower regulatory tax rates.
The rate under which the ENPV changes as a function of the initial carbon tax (shown in Equation 7-1) is considered in all cases as a measure of sensitivity or robustness of economic
performance to future regulatory action on CO2 emissions:
r = - 7-1 ∆ ENPV ∆ Initial CO2 tax rate
Notice that the above ratio captures the degree of sensitivity of economic performance of an HP-relevant technology option (when evaluated through the ENPV-metric) with respect to a future carbon price. The calculated values of the rate r are: 0.035, 0.003 and0.0004 [B$] per $/tonne of CO2 for cases A, B and C, respectively. In Figure 7-5, NPV distribution profiles for all HP technology options considered are depicted in the absence of any regulatory action along with their respective rates. Furthermore, since the magnitude of the rate r is smaller in the HP-CMR case than in the traditional HP (Cases A and B), the NPV-profile of the HP-CMR finds itself located to the right of the NPV-profiles of Cases A and B (as also previously depicted in Figure 7-3 (b) under an initial CO2 tax rate of $30 per tonne of CO2), thus starting to become comparatively more
appealing under regulatory taxes greater than $15/tonne of CO2. It is important to mention that the
188
Chapter 7 Integration of Membrane Technology into Hydrogen Production Plants with CO2 Capture: An Economic Performance Assessment Study NPV profile of the HP-CMR plants under goes a smaller shift compared to the traditional technology option in response to the introduction of a carbon tax, thus demonstrating the relative robustness of its economic performance under regulatory uncertainties. Please notice that the arrows shown in Figure 7-5 represent the rate under which the NPV-curve of each technology option is displaced toward less attractive valuation zones when a carbon price is imposed.
Figure 7-5. Net present value cumulative probability distribution profiles for various technology options in the absence of CO2 tax and their ENPV rates.
189
Chapter 7 Integration of Membrane Technology into Hydrogen Production Plants with CO2 Capture: An Economic Performance Assessment Study 7.2.3 The Role of Discount-Rate in the Valuation of HP-CMRs: Sensitivity Analysis
The discount rate at which future cash flows are discounted represents a key determinant of any technology investment project valuation approach, particularly in cases involving capital intensive projects where their respective NPV distribution profiles depend crucially on the choice of the discount rate. It should be pointed out that the aforementioned discount rate depends essentially on two factors: (i) the perceived risk of the specific project (ii) as well as its financing mechanism [51,183,189]. Indeed, investment in advanced energy-relevant technology projects encompasses a diverse set of risks such as: macro-economic ones related to global economic growth and demand for hydrogen as well as availability of labor and capital, regulatory risks, technology reliability risks, price and volume risks in the hydrogen markets and fuel price (coal, natural gas) risks. Please notice that these project risks could affect different technology options in different ways. Focusing solely on capturing project risk, 100% equity financing is usually assumed, and the most common practice is to add a project specific risk premium to the risk free rate of return when estimating the discount rate (the risk free rate of return is generally determined by the long-term rates of return of government bonds in the United States) [51,183,189]. It should be pointed out that the estimation of the risk premium could become quite complex. If the risk of the project under consideration is similar to other ones undertaken by the firm, and if the firm’s common stock is traded on open markets then in principle the risk premium can be estimated using published historical stock price data [51,183,189].
As mentioned earlier, the nominal discount rate used by the DOE for evaluating investments in conventional hydrogen production is 12.95% [44]. When the financing structure of the project is also taken into account, the risk-adjusted nominal discount rate is a weighted average of the cost of funds obtained from shareholders (“cost of equity”) and borrowed from debt-holders (“cost of debt”) with relative amounts of equity and debt being the respective weights (known as the weighted average cost of capital) [183,189]. In competitive hydrogen markets, what matters to the investor is the profitability of the investment against the risk to the capital employed. In particular, the level of risk anticipated by an investor in a hydrogen production plant will be reflected on the level of return expected on that investment. Moreover, the greater the business and financial risks, the higher the overall return that will be demanded. Since it is quite difficult for the
190
Chapter 7 Integration of Membrane Technology into Hydrogen Production Plants with CO2 Capture: An Economic Performance Assessment Study traditional levelized cost-based methodology to incorporate risks and uncertainty effectively, the proposed approach using a detailed NPV model coupled with a Monte Carlo simulator represents a methodologically appropriate and sound approach to effectively cope with multiple risk and uncertainty sources by generating a range of economic performance assessment outcomes for a new technology option such as HP-CMR [51,183,189].
Within the above context, we complemented the valuation assessment with a sensitivity analysis conducted for different representative values of the discount rate that span a reasonable range (between 10% and 20% in nominal terms) associated with business and financing risks. Lower values of the discount rate could correspond to a more stable environment of the producing firm (low macro-economic risks, acceptable technology reliability levels, an available captive market and long-term contracts, availability of capital, low financing risks) whereas higher values of the discount rate could correspond to conditions of elevated risks in the firm’s (macroeconomic, operating, financing, market) environment. The simulation results derived are plotted in Figure 7-6 and tabulated in Table 7-5. Please notice the expected adverse impact of a higher discount rate on the economic performance and valuation profile of HP-CMR reflected on all indicators: a deteriorating ENPV, an increasing probability of loss, a decreasing range of upside opportunities and a deteriorating “value at risk” pattern (downside risk). It becomes therefore apparent that creatively structured financing mechanisms leading to a reduction of the cost of capital/discount rate could induce more appealing economic performance outcomes and valuation profiles, thus unlocking the investment potential of a clean hydrogen production project that utilizes the HP- CMR technology option.
191
Chapter 7 Integration of Membrane Technology into Hydrogen Production Plants with CO2 Capture: An Economic Performance Assessment Study
Figure 7-6. Net present value cumulative probability distribution profiles for HP-CMR plants under various discount rates.
192
Chapter 7 Integration of Membrane Technology into Hydrogen Production Plants with CO2 Capture: An Economic Performance Assessment Study Table 7-5. Sensitivity analysis results of the NPV-based economic performance assessment for HP-CMR plants with various discount rates.
Discount rate [%]
10 11 12 13 14 15 16 17 18 19 20
P5 [B$] 0.73 0.52 0.35 0.21 0.08 -0.04 -0.14 -0.24 -0.31 -0.38 -0.46 ENPV [B$] 3.56 3.11 2.72 2.39 2.10 1.84 1.60 1.42 1.24 1.08 0.94 P95 [B$] 5.87 5.22 4.64 4.16 3.75 3.35 3.04 2.77 2.51 2.27 2.06 Standard deviation [B$] 1.61 1.47 1.35 1.24 1.15 1.07 1.00 0.94 0.88 0.83 0.79 Probability of loss 2% 2% 3% 4% 5% 6% 8% 10% 12% 14% 17%
193
Chapter 7 Integration of Membrane Technology into Hydrogen Production Plants with CO2 Capture: An Economic Performance Assessment Study 7.3 Conclusion
Membrane reactor technology integrated into hydrogen production plants (HP-CMR) generates promising prospects of advancing energy and environmental policy goals. Technical
advantages of HP-CMR include a reduction in coal-feed requirements and consequently CO2 emission levels by means of process intensification and higher carbon capture efficiency. Indeed, the overall reduction of carbon emissions is 70% lower compared to the conventional technology option. Another technical and environmental performance-relevant advantage is the simultaneous
CO2 capture and generation of extra pure H2 through the use of CMRs. This new technology option enables the elimination of traditional WGS reactors, PSA and Selexol units in conventional H2 production plants, thus potentially enhancing the economic performance of an HP-CMR plant. Indeed, HP-CMR plants showed a 26% reduction in total capital investment costs and good valuation prospects even under higher nominal discount rates reflecting the added risks associated with a new technology option and the lack of relevant accumulated operating experience. Motivated by these preliminary realizations, a comprehensive economic performance assessment framework was developed in the presence of irreducible uncertainties through the integration of a detailed NPV-model with a Monte Carlo simulator. Within the proposed economic performance evaluation framework for a HP-CMR plant more realistic NPV-distribution profiles were derived rather than single-point value estimates that often lead to unsatisfactory valuation assessments by overlooking significant uncertainty effects over the plant’s lifetime. Furthermore, the economic performance of a HP-CMR plant was comparatively assessed against the conventional coal gasification-based hydrogen production plant. The analysis conducted also suggested that in the
absence of any regulatory action on CO2 emissions HP-CMR could not be perceived as an economically viable option. However, better prospects for HP-CMR arise if future regulatory action on CO2 emissions is introduced, and therefore, initiatives to stimulate the realization of demonstration projects of this innovative technology at the commercial scale might be warranted. Finally, it was found that the degree of sensitivity of economic performance (when evaluated through the ENPV-metric) with respect to a future carbon price conforms to the following inequality (in descending order amongst the technology options considered): HP-PSA without CCS system > HP-PSA with CCS system > HP-CMR.
194
Chapter 8
Chapter 8 Design Flexibility for Hydrogen Production Plants Integrated with Membrane Technology: An Economic Performance Evaluation Study Design Flexibility for Hydrogen Production Plants Integrated with Membrane Technology: An Economic Performance Evaluation Study
8.1 Introduction
Membrane technology integrated into coal-fired hydrogen production plants for CO2 capture is recognized as a promising technology option with the potential of improving the
hydrogen productivity and capturing CO2 in a cost-effective approach. In particular, catalytic
membrane reactors (CMRs), with the use of Pd/alloy membranes as H2-selective membranes, embedded into coal-fired hydrogen production facilities perform WGS reactions and hydrogen separation simultaneously, enabling higher attainable CO conversion, H2 recovery, and CO2 capture levels to be economically achieved in one single unit [143,144,145,146]. Indeed, in light of the research study by Ma et al. [196], Pd/alloy-based CMR modules integrated into coal-fired hydrogen production (HP-CMR) plants not only exhibit outstanding technological performance on the high-purity hydrogen production level but also appealing environmental performance on the
CO2 capture one. In particular, HP-CMR displays superior economic performance under various
CO2 tax scenarios. Within the above context, there is growing attention and interest focused on the
195
Chapter 8 Design Flexibility for Hydrogen Production Plants Integrated with Membrane Technology: An Economic Performance Evaluation Study deployment of Pd/alloy-based CMR modules for industrial scale hydrogen production due to their appealing techno-economic performance profile over other technology options [40,41,42,122,147]. Significant research work has been done on building comprehensive economic performance evaluation frameworks, in which various sources of inherent uncertainties in hydrogen production plants are identified and taken into account; nevertheless, little effort has been put into potentially on value-enhancing flexibility options for plant designs and operations management. General design and evaluation (that are based on an unrealistically narrow range of conditions) are inadequate, since conditions continually change with time. Moreover, traditional evaluation methods do not elucidate benefits generated by flexibility options in response to uncertainties. Therefore, to improve evaluation methods as well as economic performance of HP- CMR plants by allowing embedded flexibility into projects is certainly justified.
As stimulated by uncertain regulatory actions on CO2 emissions as well as the powerful virtues of engineering design flexibility, the primary objective of this research study is to develop a methodological assessment and analysis framework building upon the promising research work by Ma et al. [147] in order to assess potentially value-enhancing flexibility options for HP-CMR plants under uncertain CO2 tax scenarios, and draw useful information about their performance characteristics.
196
Chapter 8 Design Flexibility for Hydrogen Production Plants Integrated with Membrane Technology: An Economic Performance Evaluation Study 8.2 Results and Discussion
Potentially value-enhancing design flexibility options are proposed and classified under operational and constructional flexibility options respectively in Table 3-23. In this section, through the integration of standard Monte Carlo simulation methods, the cumulative probability distribution profiles of the Net Present Value (NPV) with corresponding Expected Net Present Values (ENPVs) for various flexibility options are presented for statistical characterization. In addition, standard financial metrics and tools derived from the NPV distribution profiles are used to evaluate the performance of various flexibility options in response to a potential future CO2 tax scenario. Finally, sensitivity analysis with respect to various model inputs in the above CO2 tax scenarios is conducted, generating additional pieces of valuable information about the performance characteristics of the proposed flexibility options.
197
Chapter 8 Design Flexibility for Hydrogen Production Plants Integrated with Membrane Technology: An Economic Performance Evaluation Study
8.2.1 Operational Flexibility for the HP-CMR Plant in the Presence of CO2 tax
The cumulative probability distribution profiles of the NPV with their corresponding ENPVs for Case A (the baseline case) and Case B (the case with consideration of the operational
flexibility option) with an initial tax rate of $30 per tonne of CO2 are presented in Figure 8-1 along
with their financial metrics summarized in Table 8-1. The initial CO2 tax rate of $30 per tonne of
CO2 considered is recognized as an appropriate and reasonable estimate used in the research study by Ma et al. [147], as it could provide an incentive for carbon capture and sequestration (CCS) system investment in solid-fueled plants. Notice that the lower level of the CO2 abatement cost can be currently achieved using more mature technology [243]. In Figure 8-1, both cases have similar NPV performance characteristics with a 30% probability for the NPV to be higher than 1.5 B$. Case A has an 11% probability to generate a NPV lower than -0.5 B$ while there is a 10% probability for Case B to generate a NPV lower than -0.5 B$, implying that the plant with the operational flexibility option has lower downside risks than the baseline case. From Table 8-1, the ENPV generated in Case B is higher than in Case A, demonstrating that the operational flexibility option has value-enhancing capacities in response to regulatory action on carbon. In addition, it can be inferred that Case B exhibits higher security in economic performance than Case A since the spread (defined as P95-P5), the probability of loss, as well as the standard deviation of the NPV distribution profile are lower in Case B. These traits are attributed to the fact that negative cash flows generated throughout the plant’s economic life can be effectively reduced by the operational flexibility option.
198
Chapter 8 Design Flexibility for Hydrogen Production Plants Integrated with Membrane Technology: An Economic Performance Evaluation Study
Figure 8-1. Cumulative probability distribution profiles of the Net Present Value (NPV) with the Expected Net Present Value (ENPV) for Case A and Case B under an initial tax rate of $30 per tonne of CO2.
Table 8-1. Comparison of financial metrics derived from simulation results between Case A and Case B.
Case A Case B
P95 [B$] 2.26 2.27 ENPV [B$] 0.86 0.88
P5 [B$] -0.91 -0.83 Standard deviation [B$] 1.01 0.98 Spread [B$] 3.17 3.11 Probability of loss 23% 22%
199
Chapter 8 Design Flexibility for Hydrogen Production Plants Integrated with Membrane Technology: An Economic Performance Evaluation Study 8.2.2 Constructional Flexibility−Inclusion of a CCS System in the Initial Design Phase
In this sub-section, a constructional flexibility design with the strategy of including a CCS system in the initial design phase is evaluated and discussed. Figure 8-2 shows the cumulative probability distribution profiles of the NPV for various constructional flexibility options in the absence and in the presence of CO2 tax, while Table 8-2 summarizes the financial metrics derived from simulation results.
As shown in Figure 8-2 and Table 8-2, Case A displays more appealing NPV performance than Case C and Case D in the absence of CO2 tax; however, in the presence of CO2 tax, the values
of the ENPV, P95 and P5 in Case A significantly decrease, accompanied by a higher probability of loss. In contrast to Case A, both Case C and Case D that consider the constructional flexibility
option are less sensitive to CO2 taxes, resulting in higher ENPVs, smaller spreads, and lower probability of loss. This outcome can be attributed to the fact that the NPV distribution profiles in
the presence of CO2 tax are mainly determined by the “carbon tax penalty” which is computed
through the CO2 tax rate and the amount of CO2 emissions. With the construction flexibility option for inclusion of a CCS system in the initial design phase, the “carbon tax penalty” can be alleviated
by running the CCS system to reduce the amount of CO2 emissions.
As shown in Figure 8-2 and Table 8-2, even though there is no apparent difference between
Case C and Case D in the presence of CO2 tax, Case D displays a 35% higher ENPV and a 33%
lower probability of loss than Case C in the absence of CO2 tax. It demonstrates that the operational flexibility option which manipulates the operation of the CCS system can effectively improve the economic performance in response to uncertain CO2 tax scenarios.
200
Chapter 8 Design Flexibility for Hydrogen Production Plants Integrated with Membrane Technology: An Economic Performance Evaluation Study
(a)
(b) Figure 8-2. Cumulative probability distribution profiles of the NPV with the ENPV for various constructional flexibility options with inclusion of a CCS system in the initial design phase (a) in the absence of CO2 tax and (b) under an initial tax rate of $30 per tonne of CO2.
201
Chapter 8 Design Flexibility for Hydrogen Production Plants Integrated with Membrane Technology: An Economic Performance Evaluation Study Table 8-2. Financial metrics derived from simulation results for various constructional flexibility options with inclusion of a CCS system in the initial design phase.
In the absence of CO2 tax In the presence of CO2 tax Case A Case C Case D Case A Case C Case D
P95 [B$] 2.88 2.21 2.56 2.26 2.19 2.21 ENPV [B$] 1.49 0.89 1.21 0.86 0.87 0.88 P5 [B$] -0.20 -0.73 -0.45 -0.91 -0.76 -0.76 Standard deviation [B$] 0.98 0.93 0.96 1.01 0.93 0.93 Spread [B$] 3.09 2.94 3.02 3.17 2.94 2.97 Probability of loss 8% 21% 14% 23% 22% 22%
202
Chapter 8 Design Flexibility for Hydrogen Production Plants Integrated with Membrane Technology: An Economic Performance Evaluation Study 8.2.3 Constructional Flexibility−Inclusion of a CCS System at a Later Stage
The strategy of retrofitting the plant with a CCS system at a later stage for the constructional flexibility option considered is evaluated and discussed in this sub-section. Figure 8-3 presents cumulative probability distribution profiles of the NPV for constructional flexibility options based on various strategies in the absence and in the presence of CO2 taxes, while Table 8-3 summarizes the financial metrics derived from the simulation results. Please notice that the expected service time for the CCS system (in Table 8-3) indicates how long the CCS system has been operational in the plant.
For the inclusion of a CCS system at a later stage without the preinvestment option (Case
E), the performance on the ENPV, P95, P5, spread, and probability of loss in the absence of CO2 tax is more attractive than in Case D (please see Figure 8-3 (a) and Table 8-3). This positive effect is attributed to the fact that retrofitting with a CCS system at a later stage reduces unnecessary capital investments and associated costs (such as fixed and variable O&M costs) when CO2 tax is
absent. However, in the presence of CO2 tax, Case D displays superior NPV performance than Case E even though both cases consider the operational and constructional flexibility options. It is because the de-rating factor, generated by retrofitting with a CCS system at a later stage, for Case E places negative impact on NPV performance by reducing the revenue from hydrogen sales. When comparing the variation in the NPV performance between the scenarios (in the absence of
CO2 tax and in the presence of CO2 tax), one notices that the ENPV decreases by 47% for Case E, which is much higher than the 26% found in Case D. Simultaneously, the probability of loss increases 16% for Case E while for Case D only 8%, indicating that Case E is more sensitive to
CO2 tax than Case D. In light of the above results, the inclusion of a CCS system at a later stage
should be more favorable at lower CO2 tax scenarios than in the initial design phase due to the sensitivity to CO2 tax.
Case E and Case F are now compared in the absence and the presence of CO2 to investigate the effect of the preinvestment option for retrofitting with a CCS system at a later stage. As presented in Figure 8-3 and Table 8-3, Case E has a superior NPV performance than Case F in the
absence of CO2 tax; however, the NPV performance for Case F becomes more attractive than Case
E in the presence of CO2 tax. In particular, considering the preinvestment option, the value for the
203
Chapter 8 Design Flexibility for Hydrogen Production Plants Integrated with Membrane Technology: An Economic Performance Evaluation Study ENPV increases 14% and the probability of loss reduces 3%, since the preinvestment option can provide insurance to the constructional flexibility for unfavorable scenarios, which results in a lower cost of retrofitting investment and a value of the de-rating factor.
Under the baseline scenario (the initial CO2 tax rate is $30 per tonne of CO2 and the
expected growth rate of CO2 tax is 6%), Case F shows the most appealing performance over other cases in terms of ENPV and probability of loss. It indicates that the combination of constructional flexibility and operational flexibility with the preinvestment option would provide the greatest improvement on NPV performance. Case F shows significant reduction in downside risks and
enhancement in upside opportunities to deal with uncertainties underlying in the baseline CO2 tax
scenario. However, for different CO2 tax scenarios, Case F may not exhibit the most appealing NPV performance over other cases, since cases have different responses to uncertain inputs (such
as initial CO2 tax rates) and thus produce different NPV performance profiles. In order to understand the potential characteristics of various flexible design options towards key uncertainties
(including initial CO2 tax rate, expected growth rate of CO2 tax, and year of introducing CO2 tax)
that underlie the various CO2 tax scenarios, a sensitivity analysis is conducted followed by a detailed discussion in the next sub-section.
204
Chapter 8 Design Flexibility for Hydrogen Production Plants Integrated with Membrane Technology: An Economic Performance Evaluation Study
(a)
(b)
Figure 8-3. Cumulative probability distribution profiles of the NPV with the ENPV for various constructional flexibility options with inclusion of a CCS system at a later stage (a) in the absence of CO2 tax and (b) under an initial tax rate of $30 per tonne of CO2.
205
Chapter 8 Design Flexibility for Hydrogen Production Plants Integrated with Membrane Technology: An Economic Performance Evaluation Study Table 8-3. Financial metrics derived from simulation results for various constructional flexibility options with inclusion of a CCS system at a later stage.
In the absence of CO2 tax In the presence of CO2 tax Case D Case E Case F Case D Case E Case F
P95 [B$] 2.56 2.88 2.85 2.21 2.18 2.25 ENPV [B$] 1.21 1.49 1.45 0.88 0.79 0.90 P5 [B$] -0.45 -0.20 -0.24 -0.76 -0.90 -0.77 Standard deviation [B$] 0.96 0.98 0.98 0.93 0.97 0.96 Spread [B$] 3.02 3.09 3.09 2.97 3.08 3.02 Probability of loss 14% 8% 9% 22% 24% 21% Expected service time of 30.0 0.0 0.0 30.0 12.9 19.8 the CCS system [year]
206
Chapter 8 Design Flexibility for Hydrogen Production Plants Integrated with Membrane Technology: An Economic Performance Evaluation Study 8.2.4 Sensitivity Analysis
A sensitivity analysis was conducted with respect to various initial CO2 tax rates in order to show the characteristics of various flexibility options in response to the intrinsic uncertainties
in the CO2 tax scenarios. Figure 8-4 presents the ENPV difference with and without flexibility
options for various initial CO2 tax rates (from 0 to $60 per tonne of CO2). The results of ENPV difference for various initial CO2 tax rates are summarized in Table 8-4.
The degree of the ENPV difference between Case B and Case A is positive within the
potential range of initial CO2 tax rates and gradually increases as the initial CO2 tax rate increases, indicating that the operational flexibility option is advantageous and its value enhancing properties become more pronounced under higher initial CO2 tax scenarios. For the inclusion of a CCS system in the initial design phase considered in constructional flexibility options (Case C and Case D), the
ENPV difference changes from negative to positive. When the initial CO2 tax rate is higher than
$25 per tonne of CO2, including a CCS system in the initial design phase is economically favorable. Furthermore, Case D can mitigate losses to produce higher ENPVs than Case C when the initial
CO2 tax is lower than $35 per tonne of CO2. Included a CCS system in the initial design phase
(Case C and Case D) is more economically favorable at higher initial CO2 tax rates, while
retrofitted with a CCS system at a later stage (Case E and Case F) is preferred at lower initial CO2
taxes. Specifically, when the initial CO2 tax rate is below $20 per tonne of CO2, higher ENPVs can be produced for the inclusion of a CCS system at a later stage rather than in the initial design phase. In contrast, when the initial CO2 tax rate is over $30 per tonne of CO2, higher ENPVs can be achieved by the inclusion of a CCS system in the initial design phase instead of at a later stage.
Notice that, when the initial CO2 tax rate is in the range from $20 to $30 per tonne of CO2, Case F displays the most appealing ENPV performance. The effect of the preinvestment option on the ENPV performance for the inclusion of a CCS system at a later stage is shown by the comparison of Case E and Case F (Figure 8-4 and Table 8-4). Case F generates higher ENPV than
Case E for initial CO2 tax rates above $15 per tonne of CO2, suggesting that the preinvestment option for the inclusion of a CCS system at a later stage is preferred to be introduced if the potential initial CO2 tax rate is higher than $15 per tonne of CO2.
207
Chapter 8 Design Flexibility for Hydrogen Production Plants Integrated with Membrane Technology: An Economic Performance Evaluation Study
Figure 8-4. ENPV difference of cases between with and without flexibility options for various initial CO2 tax.
208
Chapter 8 Design Flexibility for Hydrogen Production Plants Integrated with Membrane Technology: An Economic Performance Evaluation Study
Table 8-4. Summary of ENPV difference (M$) of cases between with and without flexibility options for various initial CO2 tax rates.
Initial CO2 tax rate [$/tonne CO2] ENPV difference 0 5 10 15 20 25 30 35 40 45 50 55 60 Case B - Case A 0 1 2 4 9 16 26 39 55 76 101 129 163 Case C - Case A -596 -498 -398 -297 -195 -92 12 118 225 335 446 559 675 Case D - Case A -281 -263 -237 -195 -136 -61 27 125 229 337 447 560 675 Case E - Case A 1 -5 -26 -59 -83 -85 -64 -23 33 101 180 266 357 Case F - Case A -39 -45 -60 -67 -51 -12 45 116 196 281 371 464 560
209
Chapter 8 Design Flexibility for Hydrogen Production Plants Integrated with Membrane Technology: An Economic Performance Evaluation Study The performance for the different flexibility options in response to uncertain expected
growth rates of CO2 tax is also characterized through a sensitivity analysis. Figure 8-5 shows the
ENPV difference between cases with and without flexibility options for the initial CO2 tax rate of
$30 per tonne of CO2 with various expected growth rates (please refer to Equation 3-24), while the results are summarized in Table 8-5.
In Figure 8-5 and Table 8-5, it is found that the degree of the ENPV difference between Case B and Case A is positive, and gradually increases with higher the expected growth rates. Indeed, the operational flexibility option for plant operation can effectively enhance NPV
performance for the potential range of expected growth rates of CO2 tax. For the constructional flexibility options which consider the inclusion of a CCS system in the initial design phase (Case C and Case D), the ENPV difference increases with higher growth rates, indicating that these options are more economically favorable to be introduced under higher expected growth rates. In particular, Case D shows higher ENPV than Case C and the ENPV difference increases as the growth rates reduce, implying that an additional operational flexibility option for CCS system
operation considered in Case D can further reduce the loss in unfavorable CO2 tax scenarios.
Regarding the inclusion of a CCS system at a later stage (Case E and Case F), there are different outcomes for cases with and without the preinvestment option within the potential range of expected growth rates. The ENPV difference between Case E and Case A has a minimum value when the expected growth rate is 5%. This outcome may be attributed to the fact that Case E is more sensitive than Case A when the expected growth rate is below 5%, and Case E become less sensitive than Case A when the expected growth rate is higher than 5%. In addition, the ENPV difference between Case E and Case A becomes positive when the expected growth rate is greater than 7.5%, indicating that the loss produced by the “carbon tax penalty” for Case A is greater than that produced by the de-rating effect in Case E.
The constructional flexibility with the preinvestment option shows that the ENPV difference between Case F and Case A increases at higher expected growth rates and it becomes positive when the expected growth rate is above 5% (Figure 8-5 and Table 8-5). Considering the preinvestment option (Case F) has lower sensitivity to the expected growth rate than Case A, and the gain generated by retrofitting can be greater than the loss produced by the “carbon tax penalty.” Moreover, Case F displays higher ENPV than Case E as the expected growth rate is above 2.5%
210
Chapter 8 Design Flexibility for Hydrogen Production Plants Integrated with Membrane Technology: An Economic Performance Evaluation Study since lower values of the de-rating factor and the retrofitting investment can further reduce downside risks to improve the NPV performance.
For the cases with the inclusion of a CCS system between the initial design phase and later stage, Case E has a superior ENPV performance profile than Case C and Case D when the expected growth rate is below 4%, but Case F exhibits a more appealing ENPV performance than Case C and Case D under the range of the expected growth rate considered (Figure 8-5 and Table 8-5). Even though the inclusion of a CCS system in the initial design phase is preferred under high expected growth rate scenarios, insurance provided by the preinvestment option makes the inclusion of a CCS system at a later stage more economically favorable.
211
Chapter 8 Design Flexibility for Hydrogen Production Plants Integrated with Membrane Technology: An Economic Performance Evaluation Study
Figure 8-5. ENPV difference of cases between with and without design flexibility options for various expected growth rates.
212
Chapter 8 Design Flexibility for Hydrogen Production Plants Integrated with Membrane Technology: An Economic Performance Evaluation Study Table 8-5. Summary of ENPV difference (M$) of cases between with and without flexibility options for various expected growth rates of CO2 tax.
Expected growth rate of CO2 tax [%] ENPV difference 0 1 2 3 4 5 6 7 8 9 10 Case B - Case A 13 5 6 8 11 17 26 41 65 102 160 Case C - Case A -209 -185 -156 -124 -86 -41 12 73 148 240 352 Case D - Case A -141 -129 -111 -88 -59 -21 27 84 156 245 355 Case E - Case A -20 -25 -34 -49 -67 -74 -64 -30 23 97 195 Case F - Case A -54 -49 -43 -36 -21 6 45 97 164 248 355
213
Chapter 8 Design Flexibility for Hydrogen Production Plants Integrated with Membrane Technology: An Economic Performance Evaluation Study
The year of introducing CO2 tax is another significant factor that affects the performance of the potentially value-enhancing flexibility options. Therefore, a sensitivity analysis was implemented to evaluate the performance of design flexibility options at different years of introducing a CO2 tax. Figure 8-6 presents the ENPV difference between with and without flexibility options under the baseline scenario (the initial CO2 tax rate is $30 per tonne of CO2 and the expected growth rate of CO2 tax is 6%) for different years of introducing the CO2 tax. The results are summarized in Table 8-6.
In Figure 8-6 and Table 8-6, it is found that the degree of ENPV difference between Case B and Case A is positive, and gradually decreases with shorter tax period. Indeed, the operational flexibility option for plant operation can provide insurance to lower the losses in the presence of
CO2 tax. For the constructional flexibility options which consider the inclusion of a CCS system in the initial design phase (Case E and Case F), the ENPV difference reduces with shorter tax period, showing that these options become economically unfavorable to be introduced in shorter tax periods. In particular, Case C displays lower ENPV than Case D, and the ENPV difference significantly decreases when the tax period is reduced. In addition, even though Case D considers an additional operational flexibility option for CCS system operation, the ENPV difference is still negative if the tax period is shorter than 22 years.
Case E and Case F are compared to study the effect of the preinvestment option on the
NPV performance for various years of introducing CO2 tax. As presented in Figure 8-6 and Table 8-6, the ENPV difference between Case E and Case A is higher than that between Case F and Case A as the tax period is shorter than 17 years. It implies that the preinvestment option is unnecessary to be introduced in the shorter tax period. This outcome may be attributed to the fact that the value enhancing property of the preinvestment option decreases as the tax period reduces, resulting in an inferior NPV performance for Case F. For the cases with the inclusion of a CCS system between the initial design phase and the later stage, Case E and Case F show superior ENPV performance than Case C and Case D when the tax period is shorter than 22 years (Figure 8-6 and Table 8-6). It indicates that cases with the inclusion of a CCS system in the initial design phase are more sensitive to the tax period. The effect of the year of introducing of CO2 tax on the NPV is mainly determined by the length of period for the inclusion of a CCS system, since the “carbon tax penalty” is reduced as the tax period decreases.
214
Chapter 8 Design Flexibility for Hydrogen Production Plants Integrated with Membrane Technology: An Economic Performance Evaluation Study
Figure 8-6. ENPV difference of cases between with and without design flexibility options at different year of introducing CO2 tax.
Table 8-6. Summary of ENPV difference (M$) of cases between with and without flexibility options at different year of introducing CO2 tax.
Year of introducing CO2 tax ENPV difference 3rd 8th 13th 18th 23th Case B - Case A 26 6 1 4 0 Case C - Case A 12 -328 -482 -554 -579 Case D - Case A 27 -208 -258 -277 -280 Case E - Case A -64 -102 -78 -35 -7 Case F - Case A 45 -57 -84 -76 -56
215
Chapter 8 Design Flexibility for Hydrogen Production Plants Integrated with Membrane Technology: An Economic Performance Evaluation Study Finally, the effect of various model inputs on the NPV performance for various flexibility options is analyzed through Tornado diagrams. As shown in Figure 8-7, the variation of ENPV is presented by a bar when the model input considered changes over its specified range under the assumption that all other model inputs remain the same. All bars in the figure are sorted from long at the top to short at the bottom, graphically illustrating the relative impact of uncertain model inputs on the ENPV [51]. The uncertain model inputs considered and their respective prescribed
ranges adopted include the following: initial CO2 tax rate: 0-$60 per tonne of CO2, expected growth
rate of CO2 tax: 0-10%, volatility of CO2 tax: 0-20%, financing interest to total capital investment ratio range: 6-10%, capacity factor without CCS: 80-90%, capacity factor with CCS: 75-85%, de- rating factor without preinvestment: 5-10%, de-rating factor with preinvestment: 0-5%.
In Figure 8-7, the initial CO2 tax rate has the most significant impact on the NPV performance in Case A, Case B, Case E, and Case F, while the capacity factor with CCS places the most significant impact on the NPV performance for Case C and Case D. This result suggests that the inclusion of a CCS system in the initial design phase can greatly reduce the negative impact of the initial CO2 tax rate on the NPV performance by preventing the loss from the “carbon tax penalty”, making the capacity factor with CCS predominates the NPV performance. Furthermore, cases considering the constructional flexibility options (Case C, Case D, Case E, and Case F) display shorter bars for initial CO2 tax rate and expected growth rate of CO2 tax on the left of the baseline value. This outcome may be attributed to the fact that these cases become less sensitive
to higher initial CO2 tax rates and expected growth rate of CO2 taxes, since the inclusion of a CCS
system can mitigate losses by the “carbon tax penalty” that increases as the initial CO2 tax rate and/or the expected growth rates of CO2 tax increases.
When comparing the cases without and with the operational flexibility option for plant operation (Case A and Case B), Case B has shorter bars for uncertain model inputs considered on the left of the baseline value even though the order of the relative impact of uncertain model inputs on NPV performance in both cases is same. This result implies that the operational flexibility option can effectively alleviate the downside risks.
To investigate the effect of the operational flexibility option for CCS system operation cooperating with constructional flexibility option, the comparison between Case C and Case D has
been conducted. Figure 8-7 shows that Case D has longer bars for the initial CO2 tax rate and the
216
Chapter 8 Design Flexibility for Hydrogen Production Plants Integrated with Membrane Technology: An Economic Performance Evaluation Study
expected growth rate of CO2 tax on the right of the baseline value, implying that the operational flexibility option for CCS system operation considered in Case D can effectively enhance upside opportunities at lower values of “carbon tax penalty.” Therefore, Case D could produce higher
ENPVs than Case C under lower initial CO2 tax rates and/or the expected growth rates.
The effect of the preinvestment option on the NPV performance is studied by comparing
Case E and Case F. Figure 8-7 shows that Case E has longer bars for the initial CO2 tax rate, the
expected growth rate of CO2 tax, the volatility of CO2 tax, and the capacity factor without CCS. This result may be attributed to the expected service time of the CCS system. In the present study, even though Case E and Case F have the same decision rules for retrofitting with a CCS system (Table 3-23); Case F performs the retrofitting earlier than Case E due to preinvestment.
As a result, the impact on the NPV performance is more pronounced in Case E (caused by the cash flow before retrofitting), which results in higher sensitivity for the capacity factor without CCS and other uncertain model inputs associated with the “carbon tax penalty.” Furthermore, Case F has a longer bar for the financing interest, indicating its significant impact on the NPV performance. Please notice that, the cost of financing interest between Cases F and Case E significantly depends on the retrofitting time and/or the existence of preinvestment.
217
Chapter 8 Design Flexibility for Hydrogen Production Plants Integrated with Membrane Technology: An Economic Performance Evaluation Study
(a) Case A
(b) Case B
(c) Case C
(d) Case D
218
Chapter 8 Design Flexibility for Hydrogen Production Plants Integrated with Membrane Technology: An Economic Performance Evaluation Study
(e) Case E
(f) Case F
Figure 8-7. Tornado diagrams for ENPV of HP-CMR plants with various flexibility options.
219
Chapter 8 Design Flexibility for Hydrogen Production Plants Integrated with Membrane Technology: An Economic Performance Evaluation Study 8.3 Conclusion
Hydrogen production plants operating with catalytic membrane reactors (HP-CMR) display promising environmental and economic performance profiles under specific market and regulatory conditions. The use of various engineering design flexibility options provide further performance gains over the plant’s lifetime since they allow irreducible uncertainties to be explicitly acknowledged and managed through the explicit integration of the inherent optionality element into the proposed valuation framework associated with managerial flexibility to respond to the above uncertainties as they get progressively resolved. Specifically, the present research work developed a methodological Net Present Value (NPV)-based analysis framework in which potentially value-enhancing flexibility options for HP-CMR plants were proposed and evaluated under various irreducible uncertainties. Table 8-7 and Table 8-8 provide a qualitative summary of the NPV outcomes and valuable findings for all cases with various flexibility options. In these studies, the flexibility design-options considered revealed the following features:
1) Operational flexibility. Designed to temporarily shut down the plant when the cash flow becomes negative and restart when positive. • Indicated a value-enhancing capacity in response to carbon taxation. • Exhibited higher security in its economic performance, since the spread
(P95-P5) and the standard deviation of the NPV-distribution profiles are reduced. 2) Constructional flexibility. Provided the HP-CMR system with the capacity of integrating carbon capture and sequestration (CCS) systems. a) Inclusion in the initial design phase • Indicated a value-enhancing capacity in response to a “carbon tax
penalty” as CCS systems reduce CO2 emissions • Improved NPV outcomes were presented when the CCS systems were subject to an operational flexibility option b) Inclusion in a later stage • Diminished NPV performance was found when retrofitting with CCS at a later stage without a preinvestment option
220
Chapter 8 Design Flexibility for Hydrogen Production Plants Integrated with Membrane Technology: An Economic Performance Evaluation Study • Improved performance for the ENPV and probability of loss was found when the preinvestment option is included in the flexibility option. This option displayed the most appealing performance when
subject to specific CO2 tax rates.
Other factors such as the year of introducing the CO2 tax, the expected tax growth rate and the initial CO2 tax rate significantly influenced the economic performance of the different flexibility design options presented in this work. Therefore, the appropriate implementation of operational and constructional flexibility options can significantly improve the economic performance of HP-CMR plants and therefore attract the investments needed for the deployment of this technology.
Table 8-7. Quantitative summary of the NPV outcomes for all cases.
In the presence of CO2 tax Case A Case B Case C Case D Case E Case F Maximum value [B$] 3.36 3.65 3.38 3.44 3.52 3.65 Minimum value [B$] -3.33 -1.73 -2.29 -2.26 -2.23 -2.14
P95 [B$] 2.26 2.27 2.19 2.21 2.18 2.25 ENPV [B$] 0.86 0.88 0.87 0.88 0.79 0.90
P5 [B$] -0.91 -0.83 -0.76 -0.76 -0.90 -0.77 Standard deviation [B$] 1.01 0.98 0.93 0.93 0.97 0.96 Spread [B$] 3.17 3.11 2.94 2.97 3.08 3.02 Probability of loss 23% 22% 22% 22% 24% 21% Expected service time of N/A N/A 30.00 30.0 12.9 19.8 the CCS system [year]
221
Chapter 8 Design Flexibility for Hydrogen Production Plants Integrated with Membrane Technology: An Economic Performance Evaluation Study Table 8-8. Quantitative summary of the valuable findings for all cases.
Denotation Case explanation Quantitative summary of performance
Case A • Baseline case Case B • Considered an operational • The highest NPV in the most favorable flexibility option for plant scenario operation • The highest NPV in the worst scenario • The highest value at opportunity
Case C • Included a CCS system in the • The highest value at risk initial design phase • The smallest standard deviation of the NPV distribution • The smallest spread of the NPV distribution Case D • Included a CCS system in the • The highest value at risk initial design phase • The smallest standard deviation of the • Considered an operational NPV distribution flexibility option for CCS system operation
Case E • Retrofitted with a CCS system at a later stage • Considered an operational flexibility option for CCS system operation Case F • Retrofitted with a CCS system • The highest NPV in the most favorable at a later stage by preinvestment scenario • Considered an operational • The highest ENPV flexibility option for CCS • The lowest probability of loss system operation
222
Chapter 9
Chapter 9 Concluding Remarks and Suggestions for Future Work Concluding Remarks and Suggestions for Future Work
9.1 Concluding Remarks
• A comprehensive economic performance evaluation framework has been successfully developed to evaluate industrial scale Pd/alloy-based catalytic membrane reactor (CMR) modules potentially integrated into natural gas-based hydrogen production (HP) facilities (HP-CMR). In this evaluation framework, detailed comprehensive baseline models for the Fixed Capital Investment (FCI), Total Capital Investment (TCI), and Total Product Cost (TPC) were developed to assess economic feasibility of CMR modules, while irreducible sources of market and regulatory uncertainty were identified and their effect on the economic performance was explicitly take into consideration through Monte Carlo techniques. • Through the developed evaluation framework, the FCI, TPI and TPC profiles of industrial scale CMR modules (with the aid of current membrane synthesis methods) are more economically appealing than conventional technology options. This result indicates that the CMR technology option generates promising prospects when integrated into HP plants and be recognized as the focus of a meaningful potential investment initiative in a pertinent technology demonstration project.
223
Chapter 9 Concluding Remarks and Suggestions for Future Work
• The use of thinner Pd/alloy membranes generates more appealing economic performance profile, even though the membrane lifetime can be extended by thicker Pd layer thickness (as demonstrated by experimental studies) to reduce the costs of membrane replacement. The economic performance outcome can be attributed to a lower capital investment and a
higher H2 production level that can be achieved when using thinner membranes. Therefore, when using the CMR technology option for HP, membrane thickness is a significant factor that should be carefully considered. • The module cost per membrane unit area for the actual large-scale WGS-CMR system built at WPI is much higher than the industrial scale CMR module, since the capacity of purchased equipment is the most critical factor in the TCI performance. This result implies that the TPC profiles become more economically appealing when the capacity of the CMR module increases. Therefore, the lower TCI and TPC achieved by the larger capacity of purchased equipment provides a stimulus to develop industrial scale modules. • A TCI learning curve for the CMR modules has been successfully generated, demonstrating the importance of technological progress over time on the economic features of this innovative technology. It is recommended that future research efforts focus on generating further operating experience and plant data at a pilot/commercial scale as well as studies on CMR manufacturing. • Through a comparative assessment of technical performance for HP-CMR modules integrated into coal-fired HP plants against conventional ones (HP-PSA plants), it was shown that HP-CMR plants integrated with carbon capture and sequestration (CCS)
systems could lead to an overall reduction in CO2 emissions of 70% while achieving the
same H2 production level (616.5 tonne per day). This result indicates that integration of
CMR modules into H2 production process systems generates promising prospects of advancing clean energy and environmental policy goals.
• CMR technology showed a 26% reduction in TCI costs for achieving the same H2 production level (616.5 tonne per day). Nevertheless, in the absence of any regulatory
action on CO2 emissions HP-CMR could not be perceived as an economically viable
option, and this technology becomes more appealing if future regulatory actions on CO2 emissions are introduced.
224
Chapter 9 Concluding Remarks and Suggestions for Future Work
• Operational and constructional flexibility options for HP-CMR plants under potential CO2 tax scenarios have been successfully developed and comparatively assessed within the proposed evaluation framework in terms of their potential economic value-enhancing properties. The results showed that the appropriate implementation of operational and constructional flexibility options can significantly improve the economic performance of HP-CMR plants. Value-enhancing capacities generated through the integration of the above flexible design options for HP-CMR plants have been successfully demonstrated, thus providing the intended stimulus for demonstration and deployment of the HP-CMR technology option.
225
Chapter 9 Concluding Remarks and Suggestions for Future Work
9.2 Suggestions for Future Work
• Pd/alloy-based CMR modules used for natural gas (methane) steam reforming (MSR) provide a promising technical pathway for HP (HP-MSRCMR), as demonstrated in the literature [31,114,115,116,117,118,119,250]. Figure 9-1 shows the proposed process block flow diagram of the HP plant with CMR technology integrated for MSR. Please notice that, the HP-MSR-CMR technology option enables the elimination of traditional steam reformers, WGS reactors, pressure swing adsorption (PSA) units and amine
scrubbers in conventional HP plants with CO2 capture creating great potential to generate appealing economic performance profiles. The economic performance evaluation and cost analysis frameworks for HP-MSR-CMR plants could be developed in a similar spirit to the ones presented in this Thesis. • In this study, the cost estimation of the assorted process units/equipment such as the GE energy gasifier, the syngas scrubber, and the single stage Selexol unit was based on the DOE/NETL report [112] with the aid of six-tenths factor rule [112]. A verification of the cost figures and the economic performance of HP-CMR plants can be performed by conceptually re-building and re-designing the system using process simulation packages such as Aspen Plus. • The multitube design and the operating conditions (involving temperature, pressure, and etc.) for industrial scale CMR modules used for MSR or WGS reactions significantly influence the technical performance of CMR modules. A computational fluid dynamics (CFD) model for the optimization of a multitube design (for industrial scale CMR modules) can be developed using COMSOL Multiphysics® in order to further improve the economic performance of the CMR technology option when integrated into HP systems. • Value-enhancing capacities of operational and constructional flexibility options considered for HP-CMR plants are potentially affected by their decision rules (such as the time of introducing flexibility options). Therefore, an optimization of decision rules for flexibility options can be pursued in order to increase their value-enhancing capability. • The CMR technology option integrated into HP systems has been demonstrated to provide promising technical and economic prospects of advancing clean energy and environmental
226
Chapter 9 Concluding Remarks and Suggestions for Future Work
policy goals. This innovative option can be applied in chemical production systems associated with HP (such as in the production of ammonia). The economic performance of such systems can be assessed through the proposed evaluation framework as initiatives to further expand the range of applications of CMR technology are sought.
Figure 9-1. Process block flow diagram of the natural gas-based H2 production plant with CMR technology for MSR.
227
Appendix
A.1 One-dimensional Model for Pd/alloy-based CMR at Steady State Conditions
# Gas constant
R = 8.314 # [m^3-kPa/kgmol-K] or [J/gmol-K]
# Operation conditions
T = 400+273 # reaction temperature [K] Pret = 50/1.01325*101.325 # retentate pressure [kPa] Pper = 1/1.01325*101.325 # permeate pressure [kPa]
# Feed specifications
F_dry = (0.4089+0.1562+0.392)*19445*1000/(60*60)/50000 # dry feed flow rate [gmol/s]
fr_co = 0.4089/(0.4089+0.1562+0.392) # mole fraction of CO in the dry feed
fr_co2 = 0.1562/(0.4089+0.1562+0.392) # mole fraction of CO2 in the dry feed
fr_h2 = 0.392/(0.4089+0.1562+0.392) # mole fraction of H2 in the dry feed
fr_h2o = 2*fr_co # mole fraction of H2O
Fi = F_dry*(fr_co+fr_co2+fr_h2+fr_h2o) # wet feed flow rate [gmol/s] Fi_co = F_dry*fr_co # initial wet feed flow rate of CO [gmol/s]
Fi_h2o = F_dry*fr_h2o # initial wet feed flow rate of H2O [gmol/s]
Fi_co2 = F_dry*fr_co2 # initial wet feed flow rate of CO2 [gmol/s]
228
Appendix
Fi_h2 = F_dry*fr_h2 # initial wet feed flow rate of H2 [gmol/s] xi_co = Fi_co/Fi # mole fraction of CO in the wet feed flow rate
xi_h2o = Fi_h2o/Fi # mole fraction of H2O in the wet feed flow rate
xi_co2 = Fi_co2/Fi # mole fraction of CO2 in the wet feed flow rate
xi_h2 = Fi_h2/Fi # mole fraction of H2 in the wet feed flow rate
P1 = Pret*(Fret_co/Fret_t) # partial pressure of CO in the retentate side [kPa]
P2 = Pret*(Fret_h2o/Fret_t) # partial pressure of H2O in the retentate side [kPa]
P3 = Pret*(Fret_co2/Fret_t) # partial pressure of CO2 in the retentate side [kPa]
P4 = Pret*(Fret_h2/Fret_t) # partial pressure of H2 in the retentate side [kPa] Fret_t = Fret_co+Fret_h2o+Fret_co2+Fret_h2 # wet feed flow rate in the retentate side [gmol/s]
# CMR dimensions
d1 = 2*0.0254 # outer diameter of the membrane [m] d2 = 3*0.0254 # inner diameter of the shell casing [m] r1 = d1/2 # outer radius of the membrane [m] r2 = d2/2 # inner radius of the shell casing [m] Am = 3.14*d1*L # membrane surface [m^2] # L is the length of the membrane Aa = 3.14*(r2^2-r1^2) # annular cross-section area [m^2]
# Catalyst specifications
dcat = 2628 # density of the HTC1 catalyst [kg/m^3] theta = 0.5 # void fraction of the HTC1 catalyst epsilon = 0.2 # void fraction of catalytic bed
229
Appendix
# Pd membrane properties
th = 6 # thickness of the Pd membrane [µm]
Qo = 6322.7 # H2 permeability constant [m^3-µm/m^2-h-atm^0.5] Ea = 15630 # activation energy for permeation [J/gmol] Q = Qo*exp(-Ea/(R*T)) # permeability [m^3- µm/m^2-h-atm^0.5] Kh2 = 0.6*0.58*Q/th*(Pper/(8.314*T)) # H2 permeance [kgmol/m^2-h-atm^0.5]
# Kinetics for WGS Reaction
Keq = exp(4577.8/T-4.33) # equilibrium constant Rwgs = ((1-epsilon)*dcat*(1-theta)*1000)*(10^2.845)*exp(-111000/R/T)*P1*P3^(-0.36)*P4^(- 0.09)*(1-(1/Keq)*(P3*P4/P1/P2)) # [gmol/m^3-s] # WGS reaction rate [gmol/m^3-s] r_co = (-1)*Rco # reaction rate of CO [gmol/m^3-s] r_h2o = (-1)*Rco # reaction rate of H2O [gmol/m^3-s]
r_co2 = (1)*Rco # reaction rate of CO2 [gmol/m^3-s]
r_h2 = (1)*Rco # reaction rate of H2 [gmol/m^3-s]
# Ordinary Differential Equations for Mass Balances
d(Fret_co) / d(L) = (Aa)*r_co # mass balance equation for CO in the retentate side Fret_co(0) = 0.0441726
d(Fret_h2o) / d(L) = (Aa)*r_h2o # mass balance equation for H2O in the retentate side Fret_h2o(0) = 0.0883451
d(Fret_co2) / d(L) = (Aa)*r_co2 # mass balance equation for CO2 in the retentate side Fret_co2(0) = 0.0168739
230
Appendix
d(Fret_h2) / d(L) = (Aa)*r_h2 - (3.14*d1)*Kh2*1000/(60*60)*((P4/101.325)^0.5- (Pper/101.325)^0.5) # mass balance equation for H2 in the retentate side Fret_h2(0) = 0.0423469
d(Fper_h2) / d(L) = (3.14*d1)*Kh2*1000/(60*60)*((P4/101.325)^0.5-(Pper/101.325)^0.5) Fper_h2(0) = 0 # mass balance equation for H2 in the permeate side
L(0) = 0 L(f) = 2.54 x_co = Fret_co/Fret_t # mole fraction of CO in the retentate side x_h2o = Fret_h2o/Fret_t # mole fraction of H2O in the retentate side
x_co2 = Fret_co2/Fret_t # mole fraction of CO2 in the retentate side x_h2 = Fret_h2/Fret_t # mole fraction of H2 in the retentate side
Rh2 = Fper_h2/(F_dry*fr_co+ F_dry*fr_h2) # H2 recovery Xco = ((F_dry*fr_co)-Fret_co)/(F_dry*fr_co) # CO conversion
231
Appendix
A.2 Historical Palladium Unit Price from 2010 to 2015
* The data is obtained from www.kitco.com.
232
Appendix
A.3 Historical Gold Unit Price from 2010 to 2015
* The data is obtained from www.kitco.com.
233
Appendix
A.4 Historical Labor Cost per Hour from 2010 to 2015
* The data is obtained from www.bls.gov.
234
Appendix
A.5 Historical Coal Price from 2010 to 2015
* The data is obtained from www.eia.gov.
235
Appendix
A.6 Historical Inflation Rate from 2004 to 2015
* The data is adapted from the CPI Detailed Report, U.S. Department of Labor [185].
236
References
[1] U.S. Energy Information Administration (U.S. EIA), 2014. Annual Energy Outlook 2014 with Projections to 2040, Available from: http://www.eia.gov/forecasts/aeo/pdf/0383(2014).pdf (accessed 20.09.2014). [2] Mueller-Langer, F., Tzimas, E., Kaltschmitt, M., Peteves, S., 2007. Techno-Economic Assessment of Hydrogen Production Processes for the Hydrogen Economy for the Short and Medium term. Int. J. Hydrogen Energy 32, 3797–3810. [3] U.S. Department of Energy (U.S. DOE). Hydrogen from Coal R&D Plan. Available from: http://www.netl.doe.gov/File%20Library/Research/Coal/energy%20systems/gasification/pubs/H YDROGEN-FROM-COAL4.pdf (accessed 20.09.2014). [4] U.S. Department of Energy (U.S. DOE). Hydrogen and Clean Fuels, System Studies. Available from: http://www.netl.doe.gov/research/coal/energy-systems/fuels/systems-studies (accessed 20.09.2014). [5] Balat, M., 2009, Possible Methods for Hydrogen Production. Energy Sources, Part A 31 (1), 39-50. [6] Linde Engineering. Industrial Hydrogen Production & Technology. Available from: http://www.hzg.de/imperia/md/content/gkss/institut_fuer_werkstoffforschung/wtn/h2- speicher/funchy/funchy-2007/5_linde_wawrzinek_funchy-2007.pdf (accessed 20.09.2014). [7] U.S. Department of the Interior, U.S. Geological Survey, 2006. Mineral commodity summaries 2006. Available from: http://minerals.usgs.gov/minerals/pubs/mcs/2006/mcs2006.pdf (accessed 11.09.2015). [8] U.S. Department of the Interior, U.S. Geological Survey, 2007. Mineral commodity summaries 2007. Available from: http://minerals.usgs.gov/minerals/pubs/mcs/2007/mcs2007.pdf (accessed 11.09.2015). [9] U.S. Department of the Interior, U.S. Geological Survey, 2008. Mineral commodity summaries 2008. Available from: http://minerals.usgs.gov/minerals/pubs/mcs/2008/mcs2008.pdf (accessed 11.09.2015). [10] U.S. Department of the Interior, U.S. Geological Survey, 2009. Mineral commodity summaries 2009. Available from: http://minerals.usgs.gov/minerals/pubs/mcs/2009/mcs2009.pdf (accessed 11.09.2015). [11] U.S. Department of the Interior, U.S. Geological Survey, 2010. Mineral commodity summaries 2010. Available from: http://minerals.usgs.gov/minerals/pubs/mcs/2010/mcs2010.pdf (accessed 11.09.2015).
237
References
[12] U.S. Department of the Interior, U.S. Geological Survey, 2011. Mineral commodity summaries 2011. Available from: http://minerals.usgs.gov/minerals/pubs/mcs/2011/mcs2011.pdf (accessed 11.09.2015). [13] U.S. Department of the Interior, U.S. Geological Survey, 2012. Mineral commodity summaries 2012. Available from: http://minerals.usgs.gov/minerals/pubs/mcs/2012/mcs2012.pdf (accessed 11.09.2015). [14] U.S. Department of the Interior, U.S. Geological Survey, 2013. Mineral commodity summaries 2013. Available from: http://minerals.usgs.gov/minerals/pubs/mcs/2013/mcs2013.pdf (accessed 11.09.2015). [15] U.S. Department of the Interior, U.S. Geological Survey, 2014. Mineral commodity summaries 2014. Available from: http://minerals.usgs.gov/minerals/pubs/mcs/2014/mcs2014.pdf (accessed 11.09.2015). [16] U.S. Department of the Interior, U.S. Geological Survey, 2015. Mineral commodity summaries 2015. Available from: http://minerals.usgs.gov/minerals/pubs/mcs/2015/mcs2015.pdf (accessed 11.09.2015). [17] Midilli, A., Ay, M., Dincer, I., Rosen, M.A., 2005. On Hydrogen and Hydrogen Energy Strategies: I: Current Status and Needs. Renew. Sustain. Energy Rev. 9 (3), 255-271. [18] Ewards, P.P., Kuznetsov, V.L., David, W.I.F., 2007. Hydrogen energy. Phil. Trans. R. Soc. A 365, 1043–1056. [19] U.S. Department of Energy (U.S. DOE), Hydrogen form Coal, 2014. Available from: http://www.energy.gov/fe/science-innovation/clean-coal-research/hydrogen-coal (accessed 18.12.2014). [20] Ewan, B.C.R., Allen, R.W.K., 2005. A Figure of Merit Assessment of the Routes to Hydrogen. Int. J. Hydrogen Energy 30 (8), 809-819. [21] U.S. Department of Energy (U.S. DOE). Hydrogen Production: Natural Gas Reforming. Available from: http://energy.gov/eere/fuelcells/hydrogen-production-natural-gas-reforming (accessed 31.1.2015).
[22] Froment, G.F., 2000. Production of Synthesis Gas by Steam- and CO2-Reforming of Natural Gas. J. Mol. Catal. A: Chem. 163, 147-156. [23] U.S. Energy Information Administration (U.S. EIA), International Energy Statistics. Available form: http://www.eia.gov/cfapps/ipdbproject/IEDIndex3.cfm?tid=1&pid=7&aid=6 (accessed 11.09.2015). [24] Turner, J.A., 2004. Sustainable hydrogen production. Science 305, 972–974. [25] Sumner, J., Bird, L., Dobos, H., 2011. Carbon Taxes: A Review of Experience and Policy Design Considerations. Clim. Policy 11, 922–943.
238
References
[26] Herzog, H.J., 2001. What Future for Carbon Capture and Sequestration. Environ. Sci. Technol. 35, 148A–153A. [27] Carbon Tax Center (CTC). Where Carbon is taxed. Available from: http://www.carbontax.org/where-carbon-is-taxed/ (accessed 11.09.2015). [28] Yang, H., Xu, Z., Fan, M., Gupta, R., Slimane, R.B., Bland, A.E., Wright, I., 2008. Progress in Carbon Dioxide Separation and Capture: A Review. J. Environ. Sci. 20, 14–27. [29] Goldbach, A., Bao, F., Qi, C., Bao, C., Zhao, L., Hao, C., Jiang, C., Xua, H., 2015. Evaluation of Pd Composite Membrane for Pre-combustion CO2 Capture. Int. J. Greenh. Gas Control 33, 69– 76. [30] Catalano, J., Guazzone, F., Mardilovich, I.P., Kazantzis, N.K., Ma, Y.H., 2013. Hydrogen Production in A Large Scale Water-Gas Shift Pd-based Catalytic Membrane Reactor. Ind. Eng. Chem. Res. 52, 1042–1055. [31] Ayturk, M.E., Kazantzis, N.K., Ma, Y.H., 2009. Modeling and Performance Assessment of Pd- and Pd/Au-based Catalytic Membrane Reactors for Hydrogen Production. Energy Environ. Sci. 2, 430–438. [32] Criscuoli, A., Basile, A., Drioli, E., 2000. An Analysis of the Performance of Membrane Reactors for the Water-Gas Shift Reaction using Gas Feed Mixtures. Catal. Today 56, 53–64. [33] Paglieri, S.N., Way, J.D., 2002. Innovations in Palladium Membrane Research. Sep. Purif. Rev. 31, 1–169. [34] Mardilovich, P.P., She, Y., Ma, Y.H., Rei, M.H., 1998. Defect-Free Palladium Membranes on Porous Stainless-steel Support. AIChE J. 44, 310–322. [35] Guazzone, F., Ma, Y.H., 2008. Leak Growth Mechanism in Composite Pd Membranes Prepared by the Electroless Deposition Method. AIChE J. 54, 487–494.
[36] Chen, C., Ma, Y.H., 2010. The Effect of H2S on the Performance of Pd and Pd/Au Composite Membrane. J. Membr. Sci. 362, 535–544. [37] Augustine, A.S., Ma, Y.H., Kazantzis, N.K., 2011. High Pressure Palladium Membrane Reactor for the High Temperature Water-Gas Shift Reaction. Int. J. Hydrogen Energy 36, 5350– 5360. [38] Augustine, A.S., Mardilovich, I.P., Kazantzis, N.K., Ma, Y.H., 2012. Durability of PSS- supported Pd-Membranes under Mixed Gas and Water-Gas Shift Conditions. J. Membr. Sci. 415– 416, 213–220. [39] Guazzone, F., Catalano, J., Mardilovich, I.P., Wu, T., Lambrecht, R.C., Datta, S., Kniep,J., Pande, S., Kazantzis, N.K., Ma, Y.H., 2013. Enhancement of the Long-Term Permeance, Selectivity Stability, and Recoverability of Pd–Au Membranes in Coal Derived Syngas Atmospheres. Energy Fuels 27, 4150–4160.
239
References
[40] Amelio, M., Morrone, P., Gallucci, F., Basile, A., 2007. Integrated Gasification Gas Combined Cycle Plant with Membrane Reactors: Technological and Economical Analysis. Energy Conversion and Management 48, 2680–2693. [41] Bracht, M., Alderliesten, P.T., Kloster, R., Pruschek, R., Haupt, G., Xue, E., Ross, J.R.H., Koukou, M.K., Papayannakos, N., 1997. Water Gas Shift Membrane Reactor for CO2 Control in IGCC Systems: Techno-economic Feasibility Study. Energy Convers. Mgmt. 38 (Suppl.), 159– 164. [42] Bredesen, R., Jordal, K., Bolland, O., 2004. High-Temperature Membranes in Power Generation with CO2 Capture. Chem. Eng. Process. 43, 1129–1158. [43] Kothari, R., Buddhi, D., Sawhney, R.L., 2008. Comparison of Environmental and Economic Aspects of Various Hydrogen Production Methods. Renew. Sustain. Energy Rev. 12, 553–563.
[44] Chou, V.H., Kuehn, N.J., 2010. Assessment of Hydrogen Production with CO2 Capture. Vol. 1: Baseline State-of-the-Art Plants. DOE/NETL, Available from: http://www.canadiancleanpowercoalition.com/pdf/SMR9%20-%20H2_Prod_Vol1_2010.pdf (accessed 20.09.2014). [45] Stiegel, G.J., Ramezan, M., 2006. Hydrogen from Coal Gasification: An Economical Pathway to a Sustainable Energy Future. Int. J. Coal Geol. 65, 173–190. [46] Chiesa, P., Consonni, S., Kreutz, T., Williams, R., 2005. Co-Production of Hydrogen, Electricity and CO2 from Coal with Commercially Ready Technology. Part A. Performance and Emissions. Int. J. Hydrogen Energy 30, 747–767. [47] Kreutz, T., Williams, R., Consonni, S., Chiesa, P., 2005. Co-Production of Hydrogen, Electricity and CO2 from Coal with Commercially Ready Technology. Part B. Economic Analysis. Int. J. Hydrogen Energy 30, 769–784. [48] Koelbl, B.S., van den Broek, M.A., van Ruijven, B.J., Faaij, A.P.C., van Vuuren, D.P., 2014. Uncertainty in the Deployment of Carbon Capture and Storage (CCS): A Sensitivity Analysis to Techno Economic Parameter Uncertainty. Int. J. Greenh. Gas Control 27, 81–102. [49] Mirkhani, S., Saboohi, Y., 2012. Stochastic Modeling of the Energy Supply System with Uncertain Fuel Price – A Case of Emerging Technologies for Distributed Power Generation. Appl. Energy 93, 668–674. [50] Savage, L.S., 2003. Decision Making with Insight. Brooks/Cole, Belmont, CA. [51] de Neufville, R., Scholtes, S., 2011. Flexibility in Engineering Design. MIT Press, Cambridge, MA. [52] Cardin, M.A., Kolfschoten, G.L., Frey, D.D., de Neufville, R., de Weck, O.L., Geltner, D.M., 2013. Empirical Evaluation of Procedures to Generate Flexibility in Engineering Systems and Improve Lifecycle Performance. Res. Eng. Design 24, 277-295.
240
References
[53] Saleh, J.H., Mark, G., Jordan, N.C., 2009. Flexibility: A Multi-Disciplinary Literature Review and a Research Agenda for Designing Flexible Engineering Systems. J. Eng. Design 20, 307-323. [54] Zhang, J., Cardin, M.A., Kazantzis, N.K., Ng, S.K.K., Ma, Y.H., 2015. Economic Evaluation of Flexibility in the Design of IGCC Plants with Integrated Membrane Reactor Modules. Systems Engineering 18 (2), 208-227. [55] Deville, H. É. S., Troost, L. J., 1863. Comptes Rendus 57, 965. [56] Graham, T., 1866. On the Absorption and Dialytic Separation of Gases by Colloid Septa. Phil. Trans. R. Soc. 156, 399-439. [57] Askeland, D.R., Fulay, P.P., Wright, W.J., 2011. The Science and Engineering of Materials. Cengage Learning, Stamford, CT. [58] Shu, J., Grandjean, B. P. A., Van Neste, A., 1991. Catalytic Palladium-Based Membrane Reactors: A Review. Can. J. Chem. Eng. 69, 1036-1060. [59] Wicke, E., Nernst, G. H., 1964. Phase Diagram and Thermodynamic Behavior of the Systems Pd/H2 and Pd/D2 at Normal Temperatures; H/D Separation Effects. Ber. Bunsen Ges. Phys. Chem. 68, 224-235.
[60] Frieske, H., Wicke, E., 1973. Magnetic Susceptibility and Equilibrium Diagram of PdHn. Ber. Bunsen Ges. Phys. Chem. 77, 48-52. [61] Guazzone, F., 2005. Engineering of Substrate Surface for the Synthesis of Ultra-Thin Composite Pd and Pd-Cu Membranes for H2 Separation. PhD Dissertation, Worcester Polytechnic Institute. [62] Gillespie, L. J., Galstaun, L. S., 1936. The Palladium-Hydrogen Equilibrium and New Palladium Hydrides. J. Am. Chem. Soc. 58, 2565-2573. [63] Lewis, F.A., 1967. The Palladium Hydrogen System. Academic Press, London, NY. [64] Ward, T.L., Dao, T., 1999. Model of Hydrogen Permeation Behavior in Palladium Membranes. J. Membr. Sci. 153, 211-231. [65] Skold, K., 1978. Quasielastic Neutron Scattering Studies of Metal Hydrides, in: Alefeld, G., Volkl, J., eds., 1978, Hydrogen in Metals I, Springer–Verlag, Berlin, 267–287. [66] Ward, T.L., Dao, T., 1999. Model of Hydrogen Permeation Behavior in Palladium Membranes. J. Membr. Sci. 153, 211-231. [67] Hurlbert, R.C., Konecny, J.O., 1961. Diffusion of Hydrogen through Palladium. J. Chem. Phys. 34, 655-658. [68] Collins, J.P., Way, J.D., 1993. Preparation and Characterization of a Composite Palladium- Ceramic Membrane. Ind. Eng. Chem. Res. 32, 3006-3013.
241
References
[69] Morreale, B.D., Ciocco, M.V., Enick, R.M., Morsi, B.I., Howard, B.H., Cugini, A.V., Rothenberger, K.S., 2003. The Permeability of Hydrogen in Bulk Palladium at Elevated Temperatures and Pressures. J. Membr. Sci. 212, 87-97. [70] Guazzone, F., Engwall, E.E., Ma, Y.H., 2006. Effects of Surface Activity, Defects and Mass Transfer on Hydrogen Permeance and n-Value in Composite Palladium-Porous Stainless Steel Membranes. Catal. Today 118, 24-31. [71] Zhao, H., Pflanz, K., Gu, J., Li, A.W., Stroh, N., Brunner, H., Xiong, G.X., 1998. Preparation of Palladium Composite Membranes by Modified Electroless Plating Procedure. J. Membr. Sci. 142, 147-157. [72] Zhao, H., Xiong, G., Baron, G. V., 2000. Preparation and Characterization of Palladium- Based Composite Membranes by Electroless Plating and Magnetron Sputtering. Catal. Today 56, 89-96.
[73] Wu, L., Xu, N., Shi, J., 2000. Preparation of a Palladium Composite Membrane by an Improved Electroless Plating Technique. Ind. Eng. Chem. Res. 39, 342-348. [74] Mason, E.A., Malinauskas, A.P., Evans, R. B., III, 1967. Flow and Diffusion of Gases in Porous Media. J. Chem. Phys. 46, 3199-3118. [75] Henis, J.M.S., Tripodi, M.K., 1981. Composite Hollow Fiber Membranes for Gas Separation: The Resistance Model Approach. J. Membr. Sci. 8, 233-246. [76] Hunter, J. B., 1960. A New Hydrogen Purification Process. Platin. Met. Rev. 4, 130-131. [77] Grashoff, G. J., Pilkington, C. E., Corti, C. W., 1983. The Purification of Hydrogen. A Review of the Technology Emphasizing the Current Status of Palladium Membrane Diffusion. Platin. Met. Rev. 27, 157-169. [78] Wicke, E., Brodowsky, H., Zuechner, H., 1978. Hydrogen in Palladium and Palladium Alloys, in: Alefeld, G., Volkl, J., eds., 1978, Hydrogen in Metals II, Springer–Verlag, Berlin, 73–151. [79] McKinley, D.L., 1967. Method for Hydrogen Separation and Purification. U.S. Patent 3,350,845. [80] McKinley, D.L., 1969. Method for Hydrogen Separation and Purification. U.S. Patent 3,439,474. [81] Knapton, A.G., 1977. Palladium Alloys for Hydrogen Diffusion Membranes. Platin. Met. Rev. 21, 44-50. [82] Gryaznov, V., 2000. Metal Containing Membranes for the Production of Ultrapure Hydrogen and the Recovery of Hydrogen Isotopes. Separ. Purif. Method 29, 171-187. [83] Holleck, G.L., 1970. Diffusion and Solubility of Hydrogen in Palladium and Palladium–Silver Alloys. J. Phys. Chem. 74, 503-511.
242
References
[84] Sakamoto, Y., Hirata, S., Nishikawa, H., 1982. Diffusivity and Solubility of Hydrogen in Palladium-Silver and Palladium-Gold Alloys. J. Less-Common Met. 88, 387-395. [85] Sieverts, A., Jurisch, E., Metz, A., 1915. Solubility of Hydrogen in the Solid Alloys of Palladium with Gold, Silver and Platinum. Z. Anorg. Allg. Chem. 92, 322-362. [86] Piper, J., 1966. Diffusion of Hydrogen in Copper-Palladium Alloys. J. Appl. Phys. 37, 715- 721. [87] Karpova, R.A., Tverdovskii, I.P., 1959. Hydrogen Sorption in the Dispersed Palladium- Copper Alloys. Zh. Fiz. Khim. 33, 1393-1400. [88] Berseneva, F.N., Timofeev, N.I., Zakharov, A.B., 1993. Alloys of Palladium with Metals of the Platinum Group as Hydrogen-Permeable Membrane Components at High Temperatures of Gas Separation. Int. J. Hydrogen Energy 18, 15-18. [89] Rakhtsaum, G., 2013. Platinum Alloys: A Selective Review of the Available Literature. Platin. Met. Rev. 57, 202-213. [90] Kajiwara, M., Uemiya, S., Kojima, T., 1999. Stability and Hydrogen Permeation Behavior of Supported Platinum Membranes in Presence of Hydrogen Sulfide. Int. J. Hydrogen Energy 24, 839-844. [91] Kulprathipanja, A., Alptekin, G.O., Falconer, J.L., Way, J.D., 2005. Pd and Pd–Cu Membranes: Inhibition of H2 Permeation by H2S. J. Membr. Sci. 254, 49-62. [92] Rodriguez, J. A., 1996. Physical and Chemical Properties of Bimetallic Surfaces. Surf. Sci. Rep. 24, 223-287. [93] Gao, H., Lin, Y.S., Li, Y., Zhang, B., 2004. Chemical Stability and its Improvement of Palladium-Based Metallic Membranes. Ind. Eng. Chem. Res. 43, 6920-6930. [94] Morreale, B.D., Ciocco, M.V., Howard, B.H., Killmeyer, R.P., Cugini, A.V., Enick, R.M., 2004. Effect of Hydrogen-Sulfide on the Hydrogen Permeance of Palladium–Copper Alloys at Elevated Temperatures. J. Membr. Sci. 241, 219-224. [95] Kamakoti, P., Morreale, B. D., Ciocco, M. V., Howard, B.H., Killmeyer, R.P., Cugini, A.V., Sholl, D.S., 2005. Prediction of Hydrogen Flux through Sulfur-Tolerant Binary Alloy Membranes. Science 307, 569-573. [96] O’Brien, C.P., Howard, B.H., Miller, J.B., Morreale, B.D., Gellman, A.J., 2010. Inhibition of Hydrogen Transport through Pd and Pd47Cu53 Membranes by H2S at 350°C. J. Membr. Sci. 349, 380-384. [97] Way, J.D., Lusk, M., Thoen, P., 2008. Sulfur-Resistance Composite Metal Membranes. U.S. Patent 20080038567.
243
References
[98] Gade, S.K., DeVoss, S.J., Coulter, K. E., Paglieri, S.N., Alptekin, G.O., Way, J.D., 2011. Palladium-Gold Membranes in Mixed Gas Streams with Hydrogen Sulfide: Effect of Alloy Content and Fabrication Technique. J. Membr. Sci. 378, 35-41. [99] Rodriguez, J.A., 1996. Physical and Chemical Properties of Bimetallic Surfaces. Surf. Sci. Rep. 24, 223-287. [100] Kajiwara, M., Uemiya, S., Kojima, T., 1999. Stability and Hydrogen Permeation Behavior of Supported Platinum Membranes in Presence of Hydrogen Sulfide. Int. J. Hydrogen Energy 24, 839-844. [101] Kajiwara, M., Uemiya, S., Kojima, T., Kikuchi, E., 2000. Hydrogen Permeation Properties through Composite Membranes of Platinum Supported on Porous Alumina. Catal. Today 56, 65- 73.
[102] Howard, B.H., Morreale, B.D., 2008. Effect of H2S on Performance of Pd4Pt Alloy Membranes. Energy Mater. 3, 177-185. [103] Edlund, D.J., Pledger, W.A., 1993. Thermolysis of Hydrogen Sulfide in a Metal-Membrane Reactor. J. Membr. Sci. 77, 255-264. [104] Dittmeyer, R., Hollein, V., Daub, K., 2001. Membrane Reactors for Hydrogenation and Dehydrogenation Processes Based on Supported Palladium. J. Mol. Catal. A: Chem. 173, 135-184. [105] Dixon, A.G., 2003. Recent Research in Catalytic Inorganic Membrane Reactors. Int. J. Chem. Reactor Eng. 1, 1542-6580. [106] Stankiewicz, A.I., Moulijn, J.A., 2000. Process Intensification: Transforming Chemical Engineering. Chem. Eng. Prog 96, 22-34. [107] Weyten, H., Luyten, J., Keizer, K., Willems, L., Leysen, R., 2000. Membrane Performance: The Key Issues for Dehydrogenation Reactions in a Catalytic Membrane Reactor. Catal. Today 56, 3-11. [108] Gryaznov, V.M., Polyakova, V.P., Savitskii, E.M., Frades,L., Krapova, E.V., Juarez, E., Shkola, G.V., 1970. Effect of the Nature and Size of the Second Component of Binary Palladium Alloys on their Catalytic Activity in Cyclohexane Dehydrogenation. Izv. Akad. Nauk SSSR, Ser. Khim. 11, 2520-2524. [109] Van Hook, J.P., 1980. Methane-Steam Reforming. Catal. Rev.-Sci. Eng. 21, 1-51. [110] Rostrup-Nielsen, J.R., 1984. Catalytic steam reforming, Catalysis: Science and Technology, Springer–Verlag, Berlin. [111] Adris, A.M., Pruden, B.B., Lim, C.J., Grace, J.R., 1996. On the Reported Attempts to Radically Improve the Performance of the Steam Methane Reforming Reactor. Can. J. Chem. Eng. 74, 177-186.
244
References
[112] Haslbeck, J.L., Kuehn, N.J., Lewis, E.G., 2013. Cost and Performance Baseline for Fossil Energy Plants. Vol. 1: Bituminous Coal and Natural Gas to Electricity. DOE/NETL, Available from: http://www.netl.doe.gov/File%20Library/Research/Energy%20Analysis/OE/BitBase_FinRep_Re v2a-3_20130919_1.pdf (accessed 20.09.2014). [113] Lopes, F.V., Grande, C.A., Rodrigues, A.E., 2011. Activated Carbon for Hydrogen Purification by Pressure Swing Adsorption: Multicomponent Breakthrough Curves and PSA performance. Chem. Eng. Sci. 66, 303–317. [114] Uemiya, S., Sato, N., Ando, H., Matsuda, T., Kikuchi, E., 1991. Steam Reforming of Methane in a Hydrogen-Permeable Membrane Reactor. Appl. Catal. 1991 67, 223-230. [115] Shu, J., Grandjean, B.P.A., Kaliaguine, S., 1994. Methane Steam Reforming in Asymmetric Pd- and Pd-Ag/porous SS Membrane Reactors. Appl. Catal., A 119, 305-325. [116] Kikuchi, E., 1995. Palladium/ceramic Membranes for Selective Hydrogen Permeation and their Application to Membrane Reactor. Catal. Today 25, 333-337. [117] Barbieri, G., Violante, V., Di Maio, F.P., Criscuoli, A., Drioli, E., 1997. Methane Steam Reforming Analysis in a Palladium-Based Catalytic Membrane Reactor. Ind. Eng. Chem. Res. 36, 3369-3374. [118] Nam, S.W., Yoon, S.P., Ha, H.Y., Hong S.A., Maganyuk, A.P., 2000. Methane Steam Reforming in a Pd–Ru Membrane Reactor. Korean J. Chem. Eng. 17, 288–291. [119] Gallucci, F., Paturzo, L., Fama, A., Basile, A., 2004. Experimental Study of the Methane Steam Reforming Reaction in a Dense Pd/Ag Membrane Reactor. Ind. Eng. Chem. Res. 43, 928- 933. [120] Tong, J., Matsumura, Y., Suda, H., Haraya, K., 2005. Experimental Study of Steam Reforming of Methane in A Thin (6 mm) Pd-based Membrane Reactor. Ind. Eng. Chem. Res. 44, 1454-1465. [121] Laegsgaard Jorgensen, S., Hojlund Nielsen, P.E., Lehrmann, P., 1995. Steam Reforming of Methane in a Membrane Reactor. Catal. Today 25, 303-307. [122] Aasberg-Petersen, K., Nielsen, C.S., Jorgensen, S.L., 1998. Membrane Reforming for Hydrogen. Catal. Today 46, 193-201. [123] Descamps, C., Bouallou, C., Kanniche, M., 2008. Efficiency of An Integrated Gasification Combined Cycle (IGCC) Power Plant including CO2 Removal. Energy 33, 874–881. [124] Kikuchi, E., Uemiya, S., Sato, N., Inoue, H., Ando, H., Matsuda, T., 1989. Membrane Reactor Using Microporous Glass-Supported Thin Film of Palladium. Application to the Water Gas Shift Reaction. Chem. Lett. 18, 489-492. [125] Uemiya, S., Sato, N., Ando, H., Kikuchi, E., 1991. The Water Gas Shift Reaction Assisted by a Palladium Membrane Reactor. Ind. Eng. Chem. Res. 30, 585-589.
245
References
[126] Basile, A., Criscuoli, A., Santella, F., Drioli, E., 1996. Membrane Reactor for Water Gas Shift Reaction. Gas Sep. Purif. 10, 243-254. [127] Basile, A., Chiappetta, G., Tosti, S., Violante, V., 2001. Experimental and Simulation of both Pd and Pd/Ag for a Water Gas Shift Membrane Reactor. Sep. Purif. Technol. 25, 549-571. [128] Tosti, S., Basile, A., Chiappetta, G., Rizzello, C., Violante, V., 2003. Pd-Ag Membrane Reactors for Water Gas Shift Reaction. Chem. Eng. J. 93, 23-30.
[129] Iyoha, O., Enick, R., Killmeyer, R., Howard, B., Ciocco, M., Morreale, B., 2007. H2 Production from Simulated Coal Syngas Containing H2S in Multi-Tubular Pd and 80wt% Pd- 20wt% Cu Membrane Reactors at 1173K. J. Membr. Sci. 306, 103-115. [130] Iyoha, O., Enick, R., Killmeyer, R., Howard, B., Morreale, B., Ciocco, M., 2007. Wall- Catalyzed Water-Gas Shift Reaction in Multi-Tubular Pd and 80wt%Pd-20wt%Cu Membrane Reactors at 1173K. J. Membr. Sci. 298, 14-23. [131] Le Quere, C., Andres, R.J., Boden, T., Conway, T., Houghton, R.A., House, J.I., Marland, G., Peters, G.P., van der Werf, G.R., Ahlstrom, A., Andrew, R.M., Bopp, L., Canadell, J.G., Ciais, P., Doney, S.C., Enright, C., Friedlingstein, P., Huntingford, C., Jain, A.K., Jourdain, C., Kato, E., Keeling, R.F., Klein Goldewijk, K., Levis, S., Levy, P., Lomas, M., Poulter, B., Raupach, M.R., Schwinger, J., Sitch, S., Stocker, B.D., Viovy, N., Zaehle, S., Zeng, N., 2013. The Global Carbon Budget 1959–2011. Earth Syst. Sci. Data 5, 165–185.
[132] International Energy Agency (IEA), 2014. CO2 Emission from Fuel Combustion Highlights 2014. Available from: https://www.iea.org/publications/freepublications/publication/CO2EmissionsFromFuelCombusti onHighlights2014.pdf (accessed 11.09.2015). [133] Carbon Tax Center (CTC). Where Carbon is Taxed. Available from: http://www.carbontax.org/where-carbon-is-taxed/ (accessed 11.09.2015). [134] D'Alessandro, D.M., Smit, B., Long, J.R., 2010. Carbon Dioxide Capture: Prospects for New Materials. Angew. Chem. Int. Ed. 49, 6058-6082.
[135] Yu, C.H., Huang, C.H., Tan, C.S., 2012. A Review of CO2 Capture by Absorption and Adsorption. Aerosol Air Qual. Res. 12, 745-769. [136] Damen, K., van Troost, M., Faaij, A., Turkenburg, M., 2006. A Comparison of Electricity and Hydrogen Production Systems with CO2 Capture and Storage. Part A: Review and Selection of Promising Conversion and Capture Technologies. Prog. Energy Combust. Sci. 32 (2), 215-246. [137] Choi, A., Drese, J.H., Jones, C.W., 2009. Adsorbent Materials for Carbon Dioxide Capture from Large Anthropogenic Point Sources. ChemSusChem 2, 796-854.
[138] Grasa, G.S., Abanades, J.C., 2006. CO2 Capture Capacity of CaO in Long Series of Carbonation/Calcination Cycles. Ind. Eng. Chem. Res. 45 (26), 8846–8851.
246
References
[139] Fan, L.S., Zeng, L., Wang, W., Luo, S., 2012. Chemical Looping Processes for CO2 Capture and Carbonaceous Fuel Conversion – Prospect and Opportunity. Energy Environ. Sci. 5, 7254- 7280. [140] Cormos, C.C., 2011. Hydrogen Production from Fossil Fuels with Carbon Capture and Storage based on Chemical Looping Systems. Int. J. Hydrogen Energy 36 (10), 5960-5971. [141] Blamey, J., Anthony, E.J., Wang, J., Fennell, P.S., 2010. The Calcium Looping Cycle for Large-Scale CO2 Capture. Prog. Energy Combust. Sci. 36 (2), 260-279. [142] Gallucci, F., Fernandez, E., Corengia, P., van Sint Annaland, M., 2013. Recent Advances on Membranes and Membrane Reactors for Hydrogen Production. Chem. Eng. Sci. 92, 40-66. [143] Dolan, M.D., Donelson, R., Dave, N.C., 2010. Performance and Economics of a Pd-Based Planar WGS Membrane Reactor for Coal Gasification. Int. J. Hydrogen Energy 35 (20), 10994– 11003.
[144] Scholes, C.A., Smith, K.H., Kentish, S.E., 2010. CO2 Capture from Pre-Combustion Processes—Strategies for Membrane Gas Separation. Int. J. Greenh. Gas Control 4 (5), 739–755. [145] Koc, R., Kazantzis, N.K., Ma, Y.H., 2011. Process Safety Aspects in Water-Gas-Shift (WGS) Membrane Reactors used for Pure Hydrogen Production. J. Loss Prev. Process Ind. 24 (6), 852–869. [146] Koc, R., Kazantzis, N.K., Ma, Y.H., 2014. Membrane Technology Embedded into IGCC Plants with CO2 Capture: An Economic Performance Evaluation under Uncertainty. Int. J. Greenh. Gas Control 26, 22–38. [147] Ma, L.C., Castro-Dominguez, B., Kazantzis, N.K., Ma, Y.H., 2015. Integration of Membrane Technology into Hydrogen Production Plants with CO2 Capture: An Economic Performance Assessment Study. Int. J. Greenh. Gas Control 42, 424-438. [148] Xu, J., Froment, G.F., 1989. Methane Steam Reforming, Methanation and Water-Gas Shift: I. Intrinsic Kinetics. AIChE J. 35, 88-96. [149] Branan, C. R., 1994. Rules of Thumb for Chemical Engineers. Gulf Publishing Company, Houston, TX. [150] Newsome, D.S., 1980. The Water-Gas Shift Reaction. Cat. Rev. - Sci. Eng. 21, 275-318. [151] Ratnasamy, C., Wagner, J.P., 2009. Water Gas Shift Catalysis. Cat. Rev. - Sci. Eng. 51, 325- 440. [152] Rhodes, C., Hutchings, G.J., Ward, A.M., 1995. Water-Gas Shift Reaction: Finding the Mechanistic Boundary. Catal. Today 23, 43-58. [153] Keiski, R.L., Salmi, T., Niemistoe, P., Ainassaari, J., Pohjola, V.J., 1996. Stationary and Transient Kinetics of the High Temperature Water-Gas Shift Reaction. Appl. Catal., A, 137, 349- 370.
247
References
[154] Oki, S., Mezaki, R., 1973. Identification of Rate-Controlling Steps for the Water-Gas Shift Reaction over an Iron Oxide Catalyst. J. Phys. Chem. 77, 447-452. [155] Tinkle, M., Dumesic, J.A., 1987. Isotopic Exchange Measurements of the Rates of adsorption/desorption and Interconversion of Carbon Monoxide and Carbon Dioxide over Chromia-Promoted Magnetite: Implications for Water-Gas Shift. J. Catal. 103, 65-78. [156] Salmi, T., Bostrom, S., Lindfors, L.E., 1988. A Dynamic Study of the Water-Gas Shift Reaction over an Industrial Ferrochrome Catalyst. J. Catal. 112, 345-356. [157] Hla, S.S., Park, D., Duffy, G.J., Edwards, J.H., Roberts, D.G., Ilyushechkin, A., Morpeth, L.D., Nguyen, T., 2009. Kinetics of High-temperature Water-Gas Shift Reaction over Two Iron- based Commercial Catalysts Using Simulated Coal-derived Syngases. Chem. Eng. J. 146, 148- 154. [158] Podolski, W.F., Kim, Y.G., 1974. Modeling the Water-Gas Shift Reaction. Ind. Eng. Chem. Process Des. Dev. 13, 415-421. [159] Bohlbro, H., Jorgensen, M.H., 1970. Catalysts for the Conversion of Carbon Monoxide. Chem. Eng. World 5, 46. [160] Moe, J.M., 1962. Design of Water-Gas Shift Reactors. Chem. Eng. Prog. 58, 33-36. [161] Keiski, R.L., Desponds, O, Chang, Y.F., Somorjai G.A., 1993. Kinetics of the Water–Gas Shift Reaction over Several Alkane Activation and Water–Gas Shift Catalysts. Appl. Catal., A 101, 317–338.
[162] Rhodes, C., Williams, B.P., King, F., Hutchings, G.J., 2002. Promotion of Fe3O4/Cr2O3 High Temperature Water Gas Shift Catalyst. Catal. Commun. 3, 381-384. [163] Uchida, H., Isogai, N., Oba, M., Hasegawa, T., 1967. Zinc Oxide-Copper Catalyst for Carbon Monoxide Shift Conversion. I. Dependency of the Catalytic Activity on the Chemical Composition of the Catalyst. Bull. Chem. Soc. Jpn. 40, 1981-1986. [164] Uchida, H., Oba, M., Isogai, N., Hasegawa, T., 1968. Zinc Oxide-Copper Catalyst for Carbon Monoxide-Shift Conversion. II. Catalytic Activity and the Catalyst Structures. Bull. Chem. Soc. Jpn. 41, 479-485. [165] Gines, M.J.L., Amadeo, N., Laborde, M., Apesteguia, C.R., 1995. Activity and Structure- Sensitivity of the Water-Gas Shift Reaction over Cu-Zn-Al Mixed Oxide Catalysts. Appl. Catal., A 131, 283-296. [166] Gokhale, A.A., Dumesic, J.A., Mavrikakis, M., 2008. On the Mechanism of Low- Temperature Water Gas Shift Reaction on Copper. J. Am. Chem. Soc. 130, 1402-1414. [167] Ovesen, C.V., Clausen, B.S., Hammershoi, B.S., Steffensen, G., Askgaard, T., Chorkendorff, I., Nørskov, J.K., Rasmussen, P.B., Stoltze, P., Taylor, P., 1996. A Microkinetic Analysis of the Water-Gas Shift Reaction under Industrial Conditions. J. Catal. 158, 170-180.
248
References
[168] Choi, Y., Stenger, H.G., 2003. Water Gas Shift Reaction Kinetics and Reactor Modeling for Fuel Cell Grade Hydrogen. J. Power Sources 124, 432-439. [169] Koryabkina, N.A., Phatak, A.A., Ruettinger, W.F., Farrauto, R.J., Ribeiro, F.H., 2003. Determination of Kinetic Parameters for the Water-Gas Shift Reaction on Copper Catalysts under Realistic Conditions for Fuel Cell Applications. J. Catal. 217, 233-239. [170] Banking Environment Initiative (BEI), 2012. An Options Approach to Unlocking Investment in Clean Energy. University of Cambridge, UK. Available from: http://www.cisl.cam.ac.uk/publications/sustainable-finance-publications/bei-options-approach- unlocking-investment-clean-energy (accessed 31.01.2015) [171] Guazzone, F., Catalano, J., Mardilovich, I.P., Kniep, J., Pande, S., Wu, T., Lamrecht, R.C., Datta, S., Kazantzis, N.K., Ma, Y.H., 2012. Gas Permeation Field Tests of Composite Pd and Pd– Au Membranes in Actual Coal Derived Syngas Atmosphere. Int. J. Hydrogen Energy 37, 14557– 14568. [172] Mardilovich, I.P., Castro-Dominguez, B., Kazantzis, N.K., Ma, Y.H., 2015. A Comprehensive Performance Assessment Study of Pilot-Scale Pd and Pd/alloy Membranes under Extended Coal-Derived Syngas Atmosphere Testing. Int. J. Hydrogen Energy 40, 6107–6117. [173] Adams II, T.A., Barton, P.I., 2009. A dynamic two-dimensional heterogeneous model for water gas shift reactors. Int. J. Hydrogen Energy 34, 8877–8891. [174] Bird, B., Stewart, W., Lightfoot, E., 2002. Transport Phenomena, 2nd ed. John Wiley & Sons, New York. [175] Chen, C., 2011. Sulfur Tolerance of Pd/Au Alloy Membranes for Hydrogen Separation from Coal Gas. PhD Dissertation, Worcester Polytechnic Institute. [176] Criscuoli, A., Basile, A., Drioli, E., 2001. An Economic Feasibility Study for Water Gas Shift Membrane Reactor. J. Membr. Sci. 181, 21-27. [177] Chemical Engineering Plant Cost Indexes. Available from: http://www.chemengonline.com/pci (accessed 31.01.2015). [178] Peters, M., Timmerhaus, K., West, R., 2003. Plant Design and Economics for Chemical Engineers. McGraw-Hill Education, New York. [179] Moore, F.T., 1959. Economies of Scale: Some Statistical Evidence. Q. J. Econ. 73,232–245. [180] U.S. Energy Information Administration (U.S. EIA). International Energy Statistics. Available from: http://www.eia.gov/cfapps/ipdbproject/IEDIndex3.Cfm?tid=1&pid=7&aid=6 (accessed 20.09.2014). [181] Grant, T., Morgan, D., Gerdes, K., 2013. Carbon Dioxide Transport and Storage Costs in NETL Studies. DOE/NETL, Available from: http://www.netl.doe.gov/File%20Library/Research/Energy%20Analysis/Publications/QGESS_C O2T-S_Rev3_20140514.pdf (accessed 20.06.2014).
249
References
[182] Sutherland, E., Elgowainy, A., Dillich, S., 2013. DOE Hydrogen and Fuel Cells Program Record H2 Delivery Cost Projections – 2013. DOE, Available from: http://www.hydrogen.energy.gov/pdfs/13013_h2_delivery_cost_central.pdf (accessed 20.06.2014). [183] Benninga, S., 2011. Principles of Finance with Excel. Oxford University Press, New York & Oxford. [184] U.S. Department of Energy (U.S. DOE). Fuel Cell Technologies Office Multi-Year Research, Development, and Demonstration Plan. Available from: http://energy.gov/eere/fuelcells/downloads/fuel-cell-technologies-office-multi-year-research- development-and-22 (accessed 20.06.2014). [185] U.S. Department of Labor (U.S. DOL), CPI Detailed Report Data for February 2014. Available from: http://www.bls.gov/cpi/cpid1510.pdf (accessed 31.01.2015). [186] Internal Revenue Service (IRS), How To Depreciate Property, U.S. Department of the Treasury. Available from: http://www.irs.gov/publications/p946/ (accessed 20/03/2014). [187] Internal Revenue Service (IRS), 2013. Instructions for Form 1120S, Department of the Treasury (2013). Available from: http://www.irs.gov/pub/irs-prior/i1120s--2013.pdf (accessed 20/03/2014). [188] Karatzas, I., Shreve, S.E., 1998. Brownian Motion and Stochastic Calculus, 2th ed. Springer, New York. [189] Brealey, R.A., Myers, S.C., Allen, F., 2008. Principles of Corporate Finance, 9th ed. McGraw-Hill, New York. [190] Chau, K.W., 1995. The Validity of the Triangular Distribution Assumption in Monte Carlo Simulation of Construction Costs: Empirical Evidence from Hong Kong. Constr. Manage. Econ. 13, 15-21. [191] Hoffman, F.O., Hammonds, J.S., 1994. Propagation of Uncertainty in Risk Assessments: The Need to Distinguish between Uncertainty due to Lack of Knowledge and Uncertainty due to Variability. Risk Anal. 14, 707-712. [192] Barmish, B.R., Lagoa, C.M., 1997. The Uniform Distribution: A Rigorous Justification for Its Use in Robustness Analysis. Math. Control Signals Syst. 10, 203-222. [193] Kuipers, L., Niederreiter, H., 2012. Uniform Distribution of Sequences. Dover Publications, Mineola, NY. [194] Glasserman, P., 2003. Monte Carlo Methods in Financial Engineering. Springer, New York. [195] Simon, J.L., 1997. Resampling: The New Statistics, 2nd ed. Resampling Stats, Arlington, VA.
250
References
[196] Ma, L.C., Kazantzis, N.K., Ma, Y.H., 2015. Natural Gas in H2 Production: A Cost Study of Membrane Modules. ICE Energy 168, 61–73. [197] Ma, Y.H., Mardilovich, P.P., She, Y., 2000. Hydrogen Gas-Extraction Module and Method of Fabrication. U.S. Patent 6,152,987. [198] Ma, Y.H., Mardilovich, I.P., Engwall, E.E., 2007. Composite Gas Separation Modules Having Intermediate Porous Metal Layers. U.S. Patent 7,175,694. [199] Ma, Y.H., Guazzone, F., 2010. Method for Fabricating a Composite Gas Separation Module. U.S. Patent 7,727,596. [200] Ma, Y.H., Mardilovich I.P, Engwall, E.E., 2007. Method for Curing Defects in the Fabrication of a Composite Gas Separation Module, U.S. Patent 7,172,644. [201] Ma Y.H., Mardilovich, P.P., Engwall, E.E., 2003. Thin Composite Palladium and Palladium/Alloy Membranes for Hydrogen Separation, Ann. N. Y. Acad. Sci. 984, 346-360. [202] Ma, Y.H., Mardilovich, I.P, Engwall, E.E., 2007. Composite Gas Separation Modules Having High Tamman Temperature Intermediate Layers, U.S. Patent 7,255,726. [203] Dijkstra, J.W., Pieterse, J.A.Z., Li, H., Boon, J., van Delft, Y.C., Raju, G., Peppink, G., van den Brink, R.W., Jansen, D., 2011. Development of Membrane Reactor Technology for Power Production with Pre-Combustion CO2 Capture. Energy Proc. 4, 715–722. [204] Ayturk, M.E., Mardilovich, I.P., Engwall, E.E., Ma, Y.H., 2006. Synthesis of Composite Pd- Porous Stainless Steel (PSS) Membranes with a Pd/Ag Intermetallic Diffusion Barrier. J. Membr. Sci. 285, 385-394. [205] Armor, J.N., 1998. Applications of Catalytic Inorganic Membrane Reactors to Refinery Products. J. Membr. Sci. 147, 217-233. [206] Wieland, S., Melin, T., Lamm, A., 2002. Membrane Reactors for Hydrogen Production. Chem. Eng. Sci. 57, 1571-1576. [207] Gupta, R.B., 2008. Hydrogen Fuel: Production, Transport, and Storage. CRC press, Boca Raton, FL.
[208] Ramasubramanian, K., Zhao, Y., Ho, W.S.W., CO2 Capture and H2 Purification: Prospects for CO2-Selective Membrane Processes. AIChE J. 59, 1033-1045. [209] Hatlevik, Ø., Gade, S.K., Keeling, M.K., Thoen, P.M., Davidson, A.P., Way, J.D., 2010. Palladium and Palladium Alloy Membranes for Hydrogen Separation and Production: History, Fabrication Strategies, and Current Performance. Sep. Purif. Technol. 73, 59-64. [210] Gazzani, M., Manzolini, G., 2015. Using Palladium Membranes for Carbon Capture in Integrated Gasification Combined Cycle (IGCC) Power Plants. In: Doukelis, A., Panopoulos, K., Koumanakos, A., Kakaras, E., 2015. Palladium Membrane Technology for Hydrogen Production, Carbon Capture and Other Applications. Woodhead Publishing Series in Energy, Cambridge, UK.
251
References
[211] Koc, R., Kazantzis, N.K., Nuttall, W.J., Ma, Y.H., 2013. An Economic Evaluation Framework for Membrane Reactor Modules in the Presence of Uncertainty: The Case for Process Safety Investment and Risk Reduction, J. Loss Prevent Proc. 26, 468-477. [212] Peters, T.A., Tucho, W.M., Ramachandran, A., Stange, M., Walmsley, J.C., Holmestad, R., Borg, A., Bredesen, R., 2009. Thin Pd–23%Ag/Stainless Steel Composite Membranes: Long- Term Stability, Life-Time Estimation and Post-Process Characterization. J. Membr. Sci. 326, 572– 581. [213] Tosti, S., Basile, A., Bettinali, L., Borgognoni, F., Chiaravalloti, F., Gallucci, F., 2006. Long- Term Tests of Pd–Ag Thin Wall Permeator Tube, J. Membr. Sci. 284, 393–397. [214] Tosti, S., Bettinali, L., 2004. Diffusion Bonding of Pd-Ag Rolled Membranes, J. Mater. Sci. 39, 3041 – 3046. [215] Lin, Y.M., Rei, M.H., 2001. Study on the Hydrogen Production from Methanol Steam Reforming in Supported Palladium Membrane Reactor. Catal. Today 67, 77–84. [216] Hou, K., Hughes, R., 2003. Preparation of Thin and Highly Stable Pd/Ag Composite Membranes and Simulative Analysis of Transfer Resistance for Hydrogen Separation. J. Membr. Sci. 214, 43–55.
[217] Pomerantz, N., Ma, Y.H., 2009. Effect of H2S on the Performance and Long-Term Stability of Pd/Cu Membranes. Ind. Eng. Chem. Res. 48, 4030–4039. [218] Dunbar, Z.W., 2015. Hydrogen Purification of Synthetic Water Gas Shift Gases Using Microstructured Palladium Membranes. J. Power Sources 297, 525-533. [219] El Hawa, H.W.A., Lundin, S.T.B., Paglieri, S.N., Harale, A., Way, J.D., 2015. The Influence of Heat Treatment on the Thermal Stability of Pd Composite Membranes. J. Membr. Sci 494, 113- 120. [220] Goldbach, A., Bao, F., Qi, C., Bao, C., Zhao, L., Hao, C., Jiang, C., Xu, H., 2015. Evaluation of Pd Composite Membrane for Pre-Combustion CO2 Capture, Int. J. Greenh. Gas Con. 33, 69- 76. [221] Peters, T.A., Stange, M., Sunding, M.F., Bredesen, R., 2015. Stability Investigation of Micro-Configured Pd–Ag Membrane Modules – Effect of Operating Temperature and Pressure. Int. J. Hydrogen Energy 40, 3497-3505. [222] Peters, T.A., Stange, M., Bredesen, R., 2011. On the High Pressure Performance of Thin Supported Pd–23%Ag Membranes—Evidence of Ultrahigh Hydrogen Flux after Air Treatment. J. Membr. Sci 378, 28–34. [223] Mejdell, A.L., Klette, H., Ramachandran, A., Borg, A., Bredesen, R., 2008. Hydrogen Permeation of Thin, Free-Standing Pd/Ag23% Membranes before and after Heat Treatment in Air. J. Membr. Sci. 307, 96–104.
252
References
[224] Mardilovich, I.P., Castro-Dominguez, B., Kazantzis, N.K., Ma, Y.H., 2015. Long-Term Testing and Recovery of Pd/Au Membranes. Annual Meeting of the North American Membrane Society, Boston, MA. [225] Castro-Dominguez, B., Leelachaikul, P., Messaoud, S.B., Takagaki, A., Sugawara, T., Kikuchi, R., Oyama, S.T., 2015. The Optimal Point within the Robeson Upper Boundary. Chem. Eng. Res. Des. 97, 109-119. [226] Brunetti, A., Caravella, A., Fernandez, E., Pacheco Tanaka, D.A., Gallucci, F., Drioli, E., Curcio, E., Viviente, J.L., Barbieri, G., 2015. Syngas Upgrading in a Membrane Reactor with Thin Pd-Alloy Supported Membrane. Int. J. Hydrogen Energy 40, 10883-10893. [227] Basile, A., Curcio, S., Bagnato, G., Liguori, S., Jokar, S.M., Iulianelli, A., 2015. Water Gas Shift Reaction in Membrane Reactors: Theoretical Investigation by Artificial Neural Networks Model and Experimental Validation, Int. J. Hydrogen Energy 40, 5897-5906. [228] Lin, Y.-M., Rei, M.-H., 2001. Separation of Hydrogen from the Gas Mixture out of Catalytic Reformer by using Supported Palladium Membrane. Sep. Purif. Technol. 25, 87–95. [229] Adams, B.D., Chen, A., 2011. The Role of Palladium in a Hydrogen Economy. Materials Today 14, 282–289. [230] Iaquaniello, G., Giacobbe, F., Morico, B., Cosenza, S., Farace, A., 2008. Membrane Reforming in Converting Natural Gas to Hydrogen: Production costs, Part II. Int. J. Hydrogen Energy 33, 6595-6601. [231] De Falco, M., Salladini, A., Palo, E., Iaquaniello, G., 2015. Pd-Alloy Membrane Reactor for Natural Gas Steam Reforming: An Innovative Process Design for the Capture of CO2. Ind. Eng. Chem. Res. 54, 6950−6958. [232] Saida, S.A.M., Simakov, D.S.A., Mokheimer, E.M.A., Habib, M.A., Ahmed, S., Waseeuddin, M., Román-Leshkov, Y., 2015. Computational Fluid Dynamics Study of Hydrogen Generation by Low Temperature Methane Reforming in a Membrane Reactor. Int. J. Hydrogen Energy 40, 3158-3169. [233] Cornaglia, L., Múnera, J., Lombardo, E., 2015. Recent Advances in Catalysts, Palladium Alloys and High Temperature WGS Membrane Reactors: A Review. Int. J. Hydrogen Energy 40, 3423-3437. [234] Patrascu, M., Sheintuch, M., 2015. On-Site Pure Hydrogen Production by Methane Steam Reforming in High Flux Membrane Reactor: Experimental Validation, Model Predictions and Membrane Inhibition. Chem. Eng. J. 262, 862-874. [235] Seider, W.D., Seader, J.D., Lewin, D.R., Widagdo, S., 2008. Product and Process Design Principles: Synthesis, Analysis and Evaluation, 2nd Edition. John Wiley and Sons, Inc. NY.
253
References
[236] Koc, R., Kazantzis, N.K., Ma, Y.H., 2011. A Process Dynamic Modeling and Control Framework for Performance Assessment of Pd/alloy-based Membrane Reactors used in Hydrogen Production. Int. J. Hydrogen Energy 36, 4934–4951. [237] Haldi, J., Whitcomb, D., 1967. Economies of Scale in Industrial Plants. J. Polit. Econ. 75, 373-385. [238] Lieberman, M.B., 1987. Market Growth, Economies of Scale and Plant Size in the Chemical Processing Industries. J. Ind. Econ. 36, 175-191. [239] Sjardin, M., Damien, K.J., Faaij, A.P.C., 2015. Techno-economic prospects of small-scale membrane reactors in a future hydrogen-fuelled transportation sector. Energy 31, 2523-2555. [240] Christiansson, L., 1995. Diffusion and Learning Curves of Renewable Energy Technologies. International Institute for Applied Systems Analysis, Laxenburg, Austria, Report no. WP-95-126. [241] Colpier, U.C., Cornland, D., 2002. The Economics of the Combined Cycle Gas Turbine–An Experience Curve Analysis. Energy Policy 30, 309-316. [242] Mendes, D., Mendes, A., Madeira, L.M., Iulianelli, A., Sousa, J.M., Basile, A., 2010. The Water-Gas Shift Reaction: From Conventional Catalytic Systems to Pd-Based Membrane Reactors – A review. Asia-Pac. J. Chem. Eng. 5, 111–137. [243] Al-Juaied, M., Whitmore, A., 2009. Realistic Costs of Carbon Capture, Energy Technology Innovation Policy. Kennedy School of Government, Harvard University, Cambridge, MA. [244] Xiao, T., Qi, X., 2008. Price Competition, Cost and Demand Disruptions and Coordination of a Supply Chain with One Manufacturer and Two Competing retailers. Omega 36, 741–753. [245] Yang, C., Ogden, J., 2007. Determining the Lowest-Cost Hydrogen Delivery Mode. Int. J. Hydrogen Energy 32, 268–286. [246] Roa, F., Way, J.D., 2005. The Effect of Air Exposure on Palladium–Copper Composite Membranes. Appl. Surf. Sci. 240, 85–104. [247] Mejdell, A.L., Chen, D., Peters, T.A., 2010. The Effect of Heat Treatment in Air on CO inhibition of a 3 µm Pd–Ag (23 wt.%) Membrane. J. Membr. Sci. 350, 371–377.
[248] Buss, T.F.,∼ 2001. The Effect of State Tax Incentives on Economic Growth and Firm Location Decisions: An Overview of the Literature. Econ. Dev. Q. 15, 90–105. [249] Patino-Echeverri, D., Fischbeck, P., Kriegler, E., 2009. Economic and Environmental Costs of Regulatory Uncertainty for Coal-Fired Power Plants. Environ. Sci. Technol. 43, 578–584. [250] Shirasaki, Y., Tsuneki, T., Ota, Y., Yasuda, I., Tachibana, S., Nakajima, H., Kobayashi, K., 2009. Development of Membrane Reformer System for Highly Efficient Hydrogen Production from Natural Gas. Int. J. Hydrogen Energy 34, 4482–4487.
254