TOWARD ENVIRONMENTALLY CONSCIOUS PROCESS SYSTEMS ENGINEERING VIA JOINT THERMODYNAMIC ACCOUNTING OF INDUSTRIAL AND ECOLOGICAL SYSTEMS
DISSERTATION
Presented in Partial Fulfillment of the Requirements for
the Degree of Doctor of Philosophy in the Graduate
School of The Ohio State University
By
Jorge Luis Hau, M.S.
* * * * *
The Ohio State University 2005
Dissertation Committee:
Dr. Bhavik R. Bakshi, Adviser Approved by Dr. Isamu Kusaka
Dr. Michael J. Moran ______Adviser Dr. James F. Rathman Graduate Program in Chemical Engineering
ABSTRACT
Industrial societies are increasingly recognizing the need for shifting to more
environmentally conscious activities. Consequently, there is a need for modifying
existing processes and developing new technologies that minimize environmental impact
while providing stimulating economic value to businesses. However, this task poses
formidable challenges to the chemical engineering community because traditional tools are inadequate due to their primary emphasis on economic objectives and narrow view
that ignores the life cycle, economy and environment. Methods like Life Cycle
Assessment have broader views and focus more on ecological aspects—impact of emissions and resource consumption—but they lack a rigorous thermodynamic framework, and may even violate thermodynamic laws. In addition, like traditional engineering methods, they ignore the crucial role that ecosystems play in sustaining all industrial activity. Decisions based on approaches that take nature for granted continue to cause significant deterioration in the ability of ecosystems to provide goods and services that are essential for every human activity. In contrast, emergy analysis, a thermodynamic method from systems ecology, does account for ecosystems, but has encountered a lot of resistance and criticism, particularly from economists, physicists and engineers.
ii This dissertation develops a thermodynamic framework for evaluating ecological objectives in traditional process engineering. It expands the engineering concept of
Cumulative Exergy Consumption (CEC) analysis to include the contribution of ecosystems, which leads to the concept of Ecological CEC (ECEC). Practical challenges in computing ECEC are identified and a formal algorithm based on network algebra is proposed. ECEC is shown to be closely related to emergy, and both concepts become equivalent for certain conditions. This insight permits combination of the best features of emergy and exergy analysis, and shows that most of the controversial aspects of emergy analysis need not hinder its use for including the exergetic contribution of ecosystems.
Adopting a broader view requires expanding the analysis boundaries. Defining the system boundaries by including only the relevant processes may result in large truncation errors, while expanding it to include all interactions is computationally intractable. In practice, data and models are available at multiple spatial scales ranging from individual equipment and processes, to the supply and demand chains, to the economy and ecosystem. This work introduces a hierarchical approach that utilizes available information at all these scales and determines the trade-off between economic and ecological objectives of the process life cycle at multiple scales. A hierarchical approach was also developed for handling and representing ecological metrics at multiple spatial scales and degrees of aggregation. Examples illustrate the approaches presented in this dissertation and highlight the potential benefits of using thermodynamic principles to account for ecological aspects. An approach for enhancing the quality of life cycle inventories by reconciling it with the laws of thermodynamics was also developed.
iii DEDICATION
Dedicated to Gaia
iv
ACKNOWLEDGEMENTS
In the first place, I would like to acknowledge my parents, Eunice and Jorge Hau, for their unconditional support. Their guidance, wisdom and attitude toward life were essential for my formation as a professional, as well as a person. I would like to thank my sister, Siuyi Hau, for all her emotional and spiritual support. Thank to them, I have reached this far.
I would like to express my deepest gratitude to Dr. Bhavik R. Bakshi, my adviser, for he has made possible that this work be carried out through completion with his invaluable expertise and his unlimited interest, support, encouragement and above all, patience. Not only has he influenced the technical aspects of my research, but also infused me with the desire of making society more environmentally conscious and sustainable.
Partial financial support for this research from the National Science Foundation via Grants, BES-9985554 and DMI-0225933, is gratefully acknowledged. I would like to thank the Department of Chemical and Biomolecular Engineering for providing me with all kinds of support throughout my graduate studies. I would like to thank Professors
Isamu Kusaka, James F. Rathman, David L. Tomasko, Michael J. Moran and Michael R.
v Overcash for providing positive suggestions and insightful feedback in the various projects I undertook during my tenure as a graduate student. Constructive comments from
Mark T. Brown, Charles A.S. Hall, Sergio Ulgiati and various anonymous reviewers were very helpful in improving the clarity of my work.
I would like to gratefully acknowledge my colleagues Nandan Ukidwe, Heui-
Seok Yi, Wen-Shiang Chen, Ramon and Magaly Strauss, Oscar Lara, Hongshu Chen,
Daniel Arthur, Yi Zhang and Vikas Khanna for creating a great work environment at the
Sustainable and Efficient Process Systems Engineering group. I am especially thankful to
Mr. Nandan Ukidwe, with whom I had great discussions about thermodynamics, sustainability, economics, politics, society, etc. His viewpoints helped nurture my understanding of aspects in our group’s research.
I am thankful to my ex-roommates, Yonatan Necoechea, Juan Ignacio Sanz
Valero and Pete Cammarata. Their energetic, hardworking and determined personality made it easier for me to balance both my academic and personal life. I would also like to thank all my friends—especially Pierre Quet, Raymond Diono, Beatrice Lado, Jesus
Caraballo and Pierluigi Pisu—for making me feel like at home.
Last but not least, I want to thank my beloved Ewa Bakun, whose only presence is exciting, comforting, and supportive. Her contagious passion for knowledge and life were instrumental in giving me the energy and environment to complete my degree.
vi
VITA
July 3, 1976 Born – Valencia, Venezuela
January 1999 B.S. Chemical Engineering Universidad Simón Bolívar Sartenejas, Venezuela
January 2000 – June 2002 M.S. Chemical Engineering The Ohio State University Columbus, OH
September 2001 – present Graduate Research Associate The Ohio State University Columbus, OH
PUBLICATIONS
1. Hau, J.L., Bakshi, B.R., 2004. Expanding Exergy Analysis to account for Ecosystem Products and Services. Environmental Science & Technology 38(13), 3768- 3776.
vii 2. Hau, J.L., Bakshi, B.R., 2004. Promise and Problems of Emergy Analysis. Ecological Modelling 178(1-2), 215-225.
3. Yi, H.-S., Hau, J.L., Ukidwe, N.U., Bakshi, B.R., 2004. Hierarchical Thermodynamic Method for Evaluating the Sustainability of Industrial Processes. Environmental Progress 23(4), 302-314.
FIELDS OF STUDY
Major Field: Chemical Engineering
Studies in: Engineering Thermodynamics Process Systems Engineering Environmentally Conscious Decision Making
viii
TABLE OF CONTENTS
Page
Abstract...... ii
Dedication...... iv
Acknowledgements...... v
Vita...... vii
List of Tables ...... xiv
List of Figures...... xvi
Nomenclature...... xx
Chapters:
CHAPTER 1: Introduction ...... 1
1.1 Sustainability...... 10 1.2 Using Thermodynamics for Environmentally Conscious Decision Making .... 11 1.3 Dissertation Statement ...... 13 1.4 Structure of Dissertation ...... 14 1.5 Summary of Contributions...... 18
CHAPTER 2: Background...... 20
2.1 Brief History of Thermodynamics...... 21 2.2 First Law of Thermodynamics and Data Rectification...... 23 2.2.1 Data Rectification...... 26 ix 2.2.1.1 Solution to the Data Reconciliation Problem...... 27 2.2.1.2 Gross Error Detection...... 32 2.3 Second Law of Thermodynamics and Exergy...... 33 2.3.1 Exergy...... 33 2.3.1.1 The Reference State ...... 34 2.3.1.2 Maximum Work...... 36 2.3.1.3 Exergy Calculation...... 37 2.3.1.4 Standard Chemical Exergy ...... 44 2.3.1.5 Exergy vs. Energy...... 46 2.3.2 Exergy Applications...... 47 2.3.2.1 The Price of Steam...... 47 2.3.2.2 Exergy of Heat Transfer...... 48 2.3.2.3 Exergy and Ecosystems ...... 55 2.4 Thermodynamic Methods for Environmentally Conscious Decision Making . 56 2.4.1 Life Cycle Assessment...... 57 2.4.1.1 Methodological Framework...... 58 2.4.1.2 Major Issues and Shortcomings...... 60 2.4.2 Industrial Cumulative Exergy Consumption ...... 64 2.4.3 Thermoeconomics...... 66 2.4.4 Thermodynamic Input-Output Analysis ...... 67 2.4.5 Emergy Analysis...... 71 2.4.5.1 Historical Background ...... 72 2.4.5.2 Basic Principles...... 73 2.4.5.3 Emergy Algebra...... 76 2.4.5.4 Emergy Analysis of the main Earth Processes ...... 79 2.4.5.5 Metrics ...... 85
CHAPTER 3: Promise and Problems of Emergy Analysis ...... 87
3.1 Attractive Features of Emergy Analysis...... 88 3.2 Criticisms ...... 91 3.2.1 Emergy and economics...... 91 3.2.2 Maximum Empower Principle...... 93 3.2.3 Relation with other thermodynamic quantities...... 95 3.2.4 Combining disparate time scales ...... 97 3.2.5 Representing global energy flows in solar equivalents...... 98 3.2.6 Problems of quantification...... 99 3.2.7 Problems of allocation...... 100 3.3 Summary and challenges for the future ...... 101
CHAPTER 4: Ecological Cumulative Exergy Consumption ...... 104
4.1 Ecological Cumulative Exergy Consumption...... 105 4.2 Relation between ECEC and Emergy...... 108 4.3 ECEC Computation...... 111 x 4.3.1 Network Representation and Algebra...... 112 4.3.2 Allocation...... 116 4.3.2.1 Allocation in Fully Defined Networks...... 116 4.3.2.2 Allocation in Partially Defined Networks...... 118 4.3.3 Algorithm for ECEC Analysis...... 120 4.3.4 Illustrative Example...... 123 4.4 Examples...... 125 4.4.1 Chlor-Alkali Process by Mercury Cell ...... 125 4.4.2 Electricity from Coal versus Solar Energy ...... 128
CHAPTER 5: Enhancing Life Cycle Inventories via Reconciliation...... 133
5.1 Methodology...... 137 5.1.1 Process Inventory...... 140 5.1.2 Calculation of Material Balances...... 142 5.1.2.1 External Data and Assumptions...... 144 5.1.2.2 Verifying the Law of Mass Conservation...... 145 5.1.3 Data Rectification...... 146 5.1.3.1 Validating the Reconciled Inventory ...... 148 5.1.4 Exergy Analysis...... 148 5.1.4.1 Verifying the Second Law of Thermodynamics...... 149 5.1.5 Inventory Alternative Formulation ...... 149 5.1.5.1 Chemical Composition of the Streams ...... 150 5.1.5.2 Chemical Reactions...... 150 5.1.5.3 Measurement Units Conversion Factors...... 151 5.2 Case Study: The Chlorine/Caustic Soda Production Inventory...... 152 5.2.1 Calculation of Material Balances...... 153 5.2.2 Data Rectification...... 154 5.2.2.1 Alternative 1 – Single Black Box ...... 155 5.2.2.2 Alternative 2 – Separate Energy Sub-module...... 159 5.2.2.3 Alternative 3 – Known Reactions...... 162 5.2.2.4 Alternative 4...... 166 5.2.3 Exergy Analysis...... 168 5.3 Summary...... 170
CHAPTER 6: Evaluating Ecological and Economic Trade-Off of Processes...... 172
6.1 System Boundaries and Scales of Analysis ...... 174 6.2 Methodology...... 179 6.2.1 Level 1 – Manufacturing Scale: Traditional Exergy Analysis...... 180 6.2.2 Level 2 – Value Chain Scale: Cumulative Exergy Consumption Analysis 183 6.2.2.1 Interface between Manufacturing and Value Chain Scales ...... 184 6.2.2.2 Cumulative Exergy of Capital ...... 186 6.2.3 Level 3 – Economy Scale: Thermodynamic Input-Output Analysis...... 186 6.2.3.1 Cumulative Exergy of Capital ...... 190 xi 6.2.4 Level 4 – Ecosystem Scale: Thermodynamic Input-Output Analysis...... 190 6.2.5 Evaluating the Ecosystem-Economy Trade Off ...... 193 6.3 Case Studies...... 196 6.3.1 The Heat Exchanger...... 197 6.3.1.1 Hierarchical Analysis...... 199 6.3.1.2 Results...... 201 6.3.2 The CGAM Cogeneration System...... 205 6.3.2.1 Multiscale Analysis...... 208 6.3.2.2 Results...... 210 6.4 Discussion...... 215 6.5 Future Work...... 219
CHAPTER 7: Hierarchical Thermodynamic Metrics...... 221
7.1 Exergy of Output Streams and their Impact ...... 227 7.2 Illustrations of the Benefits of Thermodynamic Metrics...... 229 7.2.1 Energy versus Exergy Metrics for Hot Water...... 229 7.2.2 Maleic Anhydride Processes...... 230 7.2.3 Thermodynamic Metrics for Processes...... 231 7.3 Hierarchical Metrics...... 232 7.3.1 Multiscale System...... 233 7.3.2 Aggregation Hierarchy...... 235 7.3.3 Spatial Hierarchy...... 237 7.4 Case Study – Ammonia Process ...... 241 7.4.1 Process Scale...... 241 7.4.2 Life Cycle Scale...... 242 7.4.3 Economy Scale...... 245 7.4.4 Ecosystem Scale...... 249 7.5 Discussion...... 250
CHAPTER 8: Conclusions ...... 253
8.1 Concluding Remarks...... 253 8.1.1 Evaluation of ecosystem goods and services...... 254 8.1.2 Combination of data of different quality ...... 256 8.1.3 Evaluation of Ecological and Economic Trade-off ...... 256 8.1.4 Evaluation of Environmental Performance...... 257 8.2 Future Work...... 259
Appendices:
Appendix A: Details for ECEC Analysis of Example in Section 4.3.4...... 261
xii Appendix B: Details for ECEC Analysis of Chlor-Alkali Case Study in Section 4.4.1………...... 264
Appendix C: Details for the Case Study of the Caustic/Chlorine Process Inventory in Section 5.2…...... 266
Appendix D: Details for the Case Study of the Heat Exchanger Problem in Section 6.3.1………...... 284
Bibliography ...... 290
xiii
LIST OF TABLES
Table Page
Table 2.1: Conventional reference state for the atmosphere at 298.15 K and 101.325 kPa (Szargut et al., 1988)...... 36
Table 2.2: Price per unit of process steam at different conditions...... 48
Table 2.3: Specification of the heat streams required for the design of a heat transfer scheme...... 50
Table 2.4: General methodology of Life Cycle Assessment...... 59
Table 2.5: Global emergy budget of the Earth...... 83
Table 4.1: Tabular representation of Figure 4.2b...... 113
Table 4.2: Transaction coefficients and CEC for Figure 4.2b...... 114
Table 4.3: General ECEC Analysis Algorithm...... 122
Table 4.4: Subprogram for avoiding double counting...... 123
Table 4.5: ECEC of mercury cell process...... 128
Table 4.6: ICEC analysis of solar and coal-based power plants...... 131
Table 4.7: ECEC analysis of solar and coal-based power plants...... 131
Table 5.1: Process Inventory of the Chlorine/Caustic Soda Production...... 156
Table 5.2: Composition of the Coal and Residual Oil streams...... 157
Table 5.3: Component Material Balance (for each chemical element present)...... 159 xiv Table 5.4: Chemical Reactions used in this case study...... 161
Table 5.5: Reconciled and estimated values of the process variables...... 164
Table 5.6: Values of the gross error detection test and the optimized objective function...... 167
Table 5.7: Values of exergy loss and exergetic efficiency of the process inventory alternatives...... 169
Table 6.1: CEC-to-money ratios for the required economic sectors (Ukidwe and Bakshi, 2004)...... 200
Table 6.2: Economic optimum values for the CGAM Problem (Valero et al. 1994)..... 208
Table 7.1: Comparison of material intensity and exergy for two maleic anhydride processes...... 232
Table A.1: Data required for starting ECEC analysis algorithm.……………………… 262
xv
LIST OF FIGURES
Figure Page
Figure 2.1: General Open Thermodynamic System...... 25
Figure 2.2: Example of construction of the incidence, balance and master stoichiometric matrices...... 29
Figure 2.3: Model for calculating exergy of a material stream...... 39
Figure 2.4: Composite Curves for the example specified by Table 2.3...... 53
Figure 2.5: Exergy Composite Curves for the example specified by Table 2.3...... 54
Figure 2.6: Industrial Cumulative Exergy Consumption (ICEC) analysis...... 65
Figure 2.7: Integrated economic-ecological-human resource system (solid lines represent tangible interactions and dotted lines represent intangible interactions occurring as a consequence of emissions) (Ukidwe and Bakshi, 2004)...... 69
Figure 2.8: Emergy flow through network branches (Odum, 1996)...... 77
Figure 2.9: Emergy flow with network joints (Odum, 1996)...... 78
Figure 2.10: Emergy Flow with feedback...... 78
Figure 2.11: Main energy flows on Earth...... 80
Figure 4.1: Ecological Cumulative Exergy Consumption (ECEC) Analysis ...... 106
Figure 4.2: Example of a process network: (a) Original flowchart; and (b) Generic representation...... 113
xvi Figure 4.3: Allocation in industrial systems: (a) for splits; (b) for joints...... 117
Figure 4.4: Allocation in partially known systems: (a) for splits; (b) joints...... 119
Figure 4.5: Illustrative Example of ECEC Analysis...... 124
Figure 4.6: Flow diagram of the mercury cell process...... 127
Figure 4.7: (a) Exergy Flow diagram for a Coal-driven Power Plant; (b) Exergy Flow diagram for a Thermal Solar Power Plant...... 129
Figure 5.1: Feasibility of the process inventory, where W&lost,INE is the inevitable exergy
loss and E(W&lost ) is the expected value of the actual exergy loss...... 137
Figure 5.2: Algorithm for the LCI Process Data Rectification...... 141
Figure 5.3: Material Flows in Process Inventories...... 142
Figure 5.4: Model for setting up the LCI data reconciliation problem...... 147
Figure 6.1: Life Cycle of an industrial process...... 176
Figure 6.2: Characteristics of data and models at different scales of analysis...... 177
Figure 6.3: Network representation of a system at the Manufacturing Scale (B is exergy and C is cumulative exergy)...... 181
Figure 6.4: Network representation of a system at the Value Chain Scale...... 184
Figure 6.5: Disaggregation of exergy streams resulting from the expansion of system boundaries...... 185
Figure 6.6: Exergy exchange between an economic sector, ξk and the system at the value chain scale, S1...... 188
Figure 6.7: Flow Diagram of the Heat Exchanger system...... 198
Figure 6.8: Network representation of the Heat Exchanger system at multiple scales... 201
Figure 6.9: Heat Exchanger – Optimal Pareto Curve at the manufacturing scale...... 202
Figure 6.10: Heat Exchanger – Optimal Pareto Curve at the Economy scale...... 203
Figure 6.11: Heat Exchanger – Optimal Pareto Curve at the Ecosystem scale...... 204
Figure 6.12: Flow Diagram of the CGAM cogeneration system (Valero et al., 1994). . 206 xvii Figure 6.13: Network representation of the CGAM cogeneration system at multiple scales...... 209
Figure 6.14: CGAM Problem – Optimal Pareto Curve at the Equipment level...... 211
Figure 6.15: CGAM Problem – Optimal Pareto Curve at the Life Cycle level...... 212
Figure 6.16: CGAM Problem – Optimal Pareto Curve at the Economy level...... 214
Figure 6.17: CGAM Problem – Optimal Pareto Curve at the Ecosystem level...... 215
Figure 6.18: Optimal Pareto Surfaces at different levels of analysis...... 217
Figure 7.1: Environmental performance metrics of AIChE-CWRT (Schwarz et al., 2002)...... 223
Figure 7.2: Energy flow diagram at multiple spatial scales. (a) Flow diagram for economy and ecosystem scales, (b) Industrial processes considered for process and life cycle scales...... 234
Figure 7.3: Hierarchical structure of environmental performance metrics for a selected system...... 236
Figure 7.4: Mass flow diagram for an ammonia process. The energy value for 12 t/h of fuel is 160 MW...... 243
Figure 7.5: Exergy flow diagram for an ammonia process...... 244
Figure 7.6: Hierarchy of metrics for an ammonia process. The values of metrics are normalized by the production rate of ammonia...... 247
Figure 7.7: Mass and exergy flow for an ammonia plant and selected processes in its supply chain...... 248
Figure 7.8: Multiscale hierarchy of environmental performance metrics for ammonia process. The values are normalized by the production rate of ammonia...... 252
Figure C.1: 1st to 14th Columns of the Balance Matrix, B, for Alternative 1………….. 267
Figure C.2: 15th to 30th Columns of the Balance Matrix, B, for Alternative 1………… 268
Figure C.3: 31st to 46th Columns of the Balance Matrix, B, for Alternative 1………….269
Figure C.4: Transpose of Master Stoichiometric Matrix, ST, for Alternative 1……….. 270
Figure C.5: Balance Matrix, B, for the Main Process Module of Alternative 2……….. 271
xviii Figure C.6: Transpose of Master Stoichiometric Matrix, ST, for Main Process Sub- Module of Alternative 2………………………………………………………….. 272
Figure C.7: 1st to 18th columns of the Balance Matrix, B, for the energy sub-module of Alternative 2………………………………………………………………………273
Figure C.8: 19st to 32nd columns of the Balance Matrix, B, for the energy sub-module of Alternative 2………………………………………………………………………274
Figure C.9: Transpose of Master Stoichiometric Matrix, ST, for energy sub-module of Alternative 2………………………………………………………………………275
Figure C.10: 1st to 6th Columns of Balance Matrix, B, for the Main Process Module of Alternative 3………………………………………………………………………276
Figure C.11: 7th to 13th Columns of Balance Matrix, B, and Transpose of Master Stoichiometric Matrix, ST, for the main process module of Alternative 3………. 276
Figure C.12: 1st to 13th Columns of the Balance Matrix, B, for the Energy Sub-Module of Alternative 3………………………………………………………………………277
Figure C.13: 14th to 30th Columns of the Balance Matrix, B, for the Energy Sub-Module of Alternative 3…………………………………………………………………... 278
Figure C.14: Transpose of Master Stoichiometric Matrix, ST, for Energy Sub-Module of Alternative 3……………………………………………………………………....279
Figure C.15: 1st to 6th Columns of the Balance Matrix, B, for the Main Process Sub- Module of Alternative 4………………………………………………………….. 280
Figure C.16: 7st to 13th Columns of the Balance Matrix, B, and Transpose of Master Stoichiometric Matrix, ST, for Main Process Sub-Module of Alternative 4……... 280
Figure C.17: 1st to 13th Columns of the Balance Matrix, B, for the Energy Sub-Module of Alternative 4………………………………………………………………………281
Figure C.18: 14th to 30th Columns of the Balance Matrix, B, for the Energy Sub-Module of Alternative 4…………………………………………………………………... 282
Figure C.19: Transpose of Master Stoichiometric Matrix, ST, for Energy Sub-Module of Alternative 4………………………………………………………………………283
xix
NOMENCLATURE
0 Null matrix A Incidence matrix b Specific exergy (the caret designates mass basis) B Exergy B Vector of exergies or Balance matrix B0 Balance sub-matrix of B related to the exactly known variables B1 Balance sub-matrix of B related to the measured variables B2 Balance sub-matrix of B related to the unmeasured variables Bch Chemical exergy o bch Standard chemical exergy Bk Kinetic exergy Bp Potential exergy Bph Physical exergy Bth Thermal exergy C Cumulative exergy c Vector of exactly known variables C Vector of cumulative exergies Cp Constant-pressure specific heat capacity (the caret designates mass basis) d Vector of normalized errors Ec Sub-matrix Ef Ef Square matrix containing the permuted rows of the identity matrix Eu Sub-matrix Ef Ex Sub-matrix Ef f Vector of process variables g Specific Gibbs free energy (the caret designates mass basis), acceleration of gravity or process constraint h Specific enthalpy (the caret designates mass basis) I Identity matrix M Emergy M Vectors of emergies m& Mass flow MW Molecular weight xx P Projection matrix P Pressure Q Orthogonal matrix Q& Heat flow
Q1 Sub-matrix of Q Q2 Sub-matrix of Q R1 Upper triangular matrix s Specific entropy (the caret designates mass basis), unit process or single step energy transformation S Entropy or set of process units for a certain spatial scale of analysis S Master stoichiometric matrix Sgen Rate of internal generation of entropy within the system T Temperature u Specific internal energy (the caret designates mass basis) or unmeasured variable U Internal energy u Vector of unmeasured variables w Mass fraction or weighting factor W Waste or non market-valued output W Covariance matrix W& Net work done by the system
W&lost Exergy lost due to process irreversibilities.
W&lost,AVO Avoidable exergy loss
W&lost,INE Inevitable exergy loss x Molar composition or reconciled measured variable x Vector of measured variables X Vector of decision variables y Molar composition or measured variable y Vector of non-reconciled measured variables z Height Z Monetary value, cost or price zd Power measurement test statistic
Greek letters
γ Allocation coefficient γ Matrix of γij’s Γ Allocation matrix γp Diagonal matrix of γp,k’s o ∆g f Gibbs free energies of formation o ∆h f Standard molar enthalpy of formation xxi ε Threshold or Data Envelopment Analysis non-Archimedean infinitesimal η Diagonal matrix of CDP’s η Exergetic efficiency or Cumulative degree of perfection µ Data Envelopment Analysis multiplier µ + Data Envelopment Analysis scalar variable 0 µ − Data Envelopment Analysis scalar variable 0 ν Data Envelopment Analysis multiplier ξ Extent of reaction or economic sector ρ Ratio of cumulative exergy to monetary value Σ Covariance matrix τ Transformity Τ Matrix of transformities υ Velocity or number of atoms/stoichiometric coefficient
Sub-indices
0 Surroundings or manufacturing scale of analysis 0,r r-th reference species 1 Value chain scale of analysis 2 Economy scale of analysis 3 Ecosystem scale of analysis ch Crustal heat e Global energy inputs or ecological processes e,k Global energy input to unit k F Economic resource i Process stream or mearurement i Industrial processes ij from i-th to j-th unit or i-th component in the j-th stream j Process stream or unit k Process stream n Natural resources or inputs n,k External input to unit k NR Nonrenewable resource p Final products or outputs P Product p,k External output from unit k RR Renewable resource si Solar insolation te Tidal energy
xxii
CHAPTER 1
INTRODUCTION
The time to repair the roof is when the sun is shining.
–John F. Kennedy
An extraordinary characteristic of History is that by studying the past one can predict the future. On the current events of increasing anthropogenic pressure inflicted on the environment, many believe that society will always find its way out as proven in the past, i.e. there has always been an alternative energy source, a new technology or a relief valve. However, governments and societies have also collapsed by repeating the same mistakes of their predecessors.
Along the lines of environmental cataclysms, one cannot but find striking the chronicle of Easter Island as narrated by Diamond (1995). Easter Island is best known for its impressive stone statues; more than 200 gigantic figures—as much as 65 feet tall and weighting up to 270 tons—that once stood lining the coast. At least another 700 statues were left in progress as if their makers had quit their jobs all at once. When discovered, this isolated island was a grassland without a single tree, with a natural habitat of only insects and an impoverished population of about 2,000 people. For more than 2 centuries, 1 Easter Island was the source of entrancing mystery because it was impossible that its inhabitants—lacking resources, draft animals and any source of energy but themselves— may have erected these statues. Later research revealed that Easter Island, as Diamond perfectly described (1995, pp. 68), was once “an island with fertile soil, abundant food, bountiful building materials, ample lebensraum, and all the prerequisites for comfortable living.” Its society flourished to such a level of wealth and power that majestic statues started proliferating in the island, possibly as a show off to rival clans. Easter’s growing population was gradually consuming its resources faster than they could be replenished.
In just a few centuries, the people of Easter Island exhausted their forest and brought their plants and animals to extinction. Starvation submerged Easter’s society into a state of anarchy and cannibalism, where the once prosperous clans were now smashing down their rivals’ statues. One cannot help but wonder when Easter’s people started realizing the health of their environment and whether someone raised concerns at some point, like when their statues production came to a halt or when they cut down the last tree. It will probably remain unknown if they ever tried to take action but it was too late.
Today’s society is witnessing sharp changes in the environment. 33% decline in natural ecosystems with a 50% increase in human pressure (WWF, 2000). 25% of the world’s most important fish stocks are depleted or overexploited; and another 44% are being harvested at their biological limit. Forest cover has been reduced by 20% worldwide, with some forest ecosystems virtually gone. More than 50% of the mangroves in many countries have been reclaimed for human settlements. Grasslands have been reduced by more than 90% in many areas. 50% of world’s wetlands have been lost in the
2 last half century (WRI, 2000). All these figures suggest that our society is following
Easter Island’s path. But with history in our hands, it is up to us to decide when to take
action; hopefully not when our factories come to a halt or when we cut down the last tree.
Fortunately, there is indication of global concerns about the health of the
environment. Industrial societies are increasingly recognizing the need for shifting to more environmentally conscious activities. Business leaders have begun to realize that such changes are not only essential to prevent adverse social and environmental impacts, but also to assure a long term survival and success of their enterprises (Holliday, 1999).
From an economic perspective, environmentally conscious practices can benefit a business directly by increasing its tangible financial value through costs and risk reduction, increasing efficiency and capital conservation; or by building up intangible value through reputation, strategic relationships, human capital, and innovation (Bakshi
and Fiksel, 2003).
Chemical engineering has played an important role in human prosperity,
providing society with energy, technology, medical products, new materials, etc. But it
has also contributed in generating vast amounts of hazardous waste. Allen and Shonnard
(2002) estimate that over 100 kg/person of hazardous waste is produced daily, which is
60 times higher than household waste. More than 75,000 chemicals have been developed
in the last 50 years and very little is known about the impact that most of them can cause
to the environment. Clearly, chemical engineers hold a special responsibility for the
ecological damage of industrial activities and the obligation to shift toward
environmentally conscious decision making. Nevertheless, this quest poses many new
3 and formidable challenges to the chemical engineering community because traditional
engineering tools and techniques are inadequate due to their primary emphasis on
economic aspects and profitability (Cano-Ruiz and McRae 1998), and narrow view that
ignores the life cycle, economy and environment (Bakshi and Fiksel, 2003).
In the current free market economic system, prices often fail to account for goods
and services that lie outside the market. This includes the impact of many emissions from manufacturing processes, as well as the contribution of ecosystem goods and services
(natural capital) to industrial and economic activity. Examples of ecosystem goods include water, fertile soil, wood, and coal, while examples of ecosystem services include, rain, pollination, carbon sequestration, wind and other biogeochemical cycles. The importance of these goods and services has been widely recognized in many studies
(Arrow et al., 1995; Costanza et al., 1997; Daily, 1997; Dasgupta, 2002; Tilman et al.,
2002; WRI, 2000; UNEP, 2000). Costanza et al. (1997) estimate them to be about twice as valuable as the global gross national product. It is well understood that in a free market economy, ignoring such external social costs or externalities encourages overconsumption resulting in deterioration of the unaccounted goods and services
(Chapman, 2000). Consequently, a pivotal requirement for environmentally conscious decision making is the ability to account for the broader environmental impacts of engineering activities.
Environmental economists have developed approaches for valuing externalities by methods, such as travel cost, hedonic pricing and contingent valuation (Bockstael et al.,
2000; Farber et al., 2002; Nordhaus and Kokkelenberg, 1999). Other methods, such as
4 Total Cost Assessment (AIChE, 2000), also try to capture environmental impact in
monetary terms. However, proper use of these methods requires knowledge about emissions and their impact, and about the contribution of ecological systems to the life cycle of industrial activities and to human well-being. Obtaining such knowledge and incorporating it in environmental economics is an area of active research.
Engineering research has responded to environmental challenges for many decades. With greater understanding of the nature of these challenges, systems engineering has gradually broadened its analysis boundary from the equipment and manufacturing scale to the life cycle or value chain scale, and more recently, to the economy and ecosystem scales. At the manufacturing and equipment scale, thermodynamic methods such as mass and energy integration (Seider et al., 2003;
Linnhoff, 1997) and exergy analysis (Szargut et al., 1988; Bejan et al., 1996) have been quite popular. Exergy or available energy is the amount of work obtainable when matter is brought to a state of thermodynamic equilibrium with the state of the surroundings by means of reversible processes (Szargut et al., 1988). Exergy is lost or consumed in all processes, making it the ultimate limiting resource for the functioning of all systems.
Exergy analysis is useful for improving process thermodynamic efficiency by minimizing exergy losses. However, increasing the thermodynamic efficiency at this scale cannot only cost more, but need not be good for the environment either. This is because such narrowly focused analysis simply shifts the environmental problem outside the analysis boundary by actions such as preferring a higher quality fuel that burns more efficiently, or increasing the heat transfer area to reduce external utilities.
5 More recent engineering responses to the challenge of environmentally conscious
decision making have focused on the value chain or life cycle scale. These methods aim to focus on the entire supply and demand chains or life cycle of a selected product or process, and evaluate the impact of emissions (Azapagic, 1999; Burgess and Brennan,
2001). Life Cycle Assessment (LCA) is a popular approach for estimating the direct and
indirect environmental effects of emissions and other impact sources (ISO 14040, 1997),
and many tools and databases are available (Goedkoop and Spriensma, 2000; Bare et al.,
2000; Mallick et al., 1996; Shonnard and Hiew, 2000; NREL, 2005; GREET, 2005).
Most of these methods focus on identifying emissions and their impact during the life
cycle. Thus, these methods rely mainly on output-side information and are impact
oriented.
Methods at the value chain scale that rely on input-side information have also been developed, usually based on mass (Adriaanse et al., 1997; Matthews et al., 2000), energy (Spreng, 1988) and exergy (Szargut et al., 1988) inputs. Such methods are particularly appealing when information about emissions and their impact are not available, as is often the case in early stages of product and process design and assessment and for emerging technologies such as nanomanufacturing, new materials and green chemistry. Furthermore, even for more mature technologies, a traditional LCA with strong reliance on output-side information is often quite time consuming and expensive, making streamlined LCA methods appealing (Graedel, 1998). Methods that rely on input- side information have been suggested for overcoming these shortcomings of output-side
LCA methods. These methods rely on information about material and energy inputs,
6 which are often easier to obtain. They do not require information about emissions and
their impact, but may be able to use such information when available. Examples of input-
side methods include Material Intensity per Service Unit (MIPS) (Hoffmann et al., 2001)
and thermodynamic methods such as exergy (Szargut et al., 1988) and emergy analysis
(Odum, 1996). MIPS is a mass-based measure that does not account for the second law of
thermodynamics. In contrast, exergy analysis accounts for the first and second laws and
has a along history in the assessment of industrial products and processes, mainly at the
manufacturing scale. More recently, the benefits of exergy analysis have also been
demonstrated for assessing systems at coarser scales including life cycle assessment
(Cornelisen and Hirs, 2002) and input-output analysis of economic sectors (Ukidwe and
Bakshi, 2003, 2004).
Environmentally conscious decision making requires joint consideration of the
life cycle environmental and economic aspects. Consequently, input- and output-side
LCA methods have been combined with economic analysis, via combining both
objectives into one or methods for multicriteria decision making. Methods such as
Optimum LCA Performance (OLCAP) by Azapagic and Clift (1999) and Methodology
for Environmental Impact Minimization (MEIM) by Stefanis et al. (1995) are based on
multiobjective optimization of various environmental impact functions. OLCAP also considers economic objectives. Diwekar (2003) goes a step further by proposing a
framework for multiobjective optimization under uncertainty. Chain Management and
Products (CHAMP) is a methodology that also encircles the notion of industrial ecology
by seeking alternatives for reducing resource consumption and waste via modeling the
7 flow of materials in the system (Mellor et al., 2002). Chen and Shonnard (2004) present a
systematic and hierarchical approach for environmentally conscious chemical process
design based on methods such as Analytic Hierarchy Process (AHP). Hungerbuhler and coworkers have developed input-side indices based on mass loss (Heinzle et al., 1998;
Fischer and Hungerbuhler, 2000) and MIPS (Hoffmann et al., 2001) and combined them with economic factors via multiobjective optimization without and with uncertainty
(Hoffmann et al., 2004). Such mass-based indices are of limited use and can be misleading because they do not account for the energy content and quality of the
materials, and only ensure satisfaction of the law of mass conservation. In contrast,
exergy accounts for the first and second laws and is the ultimate limiting resource since it
is lost in all processes. Economic factors have been combined with exergy analysis via
Thermoeconomics (Valero and Lozano, 1997; Bejan et al., 1996; Bejan, 1997) as well as
multiobjective optimization (Lazzaretto and Toffolo, 2004). Thermoeconomics combines
exergetic and economic objectives into a single objective function by using an exergy to
money conversion factor. Other extensions of exergy analysis consider bigger spatial
scales. Cumulative Exergy Consumption (CEC) (Szargut et al., 1988) focuses on the
supply chain, Extended Exergy Accounting (EEA) (Sciubba, 2001) considers the supply
chain along with contributions from labor and capital, and Exergetic Life Cycle
Assessment (ELCA) (Cornelissen and Hirs, 2002) combines exergy and LCA.
While expansion from the manufacturing to the value chain scale is better for environmentally conscious decision making, the latter approach still relies on selecting the most relevant processes of the value chain. Such selection is often quite arbitrary and
8 subjective because there is no reliable way of screening out units with negligible effects
and because the sum of all these small contributions amounts to a large portion that may lead to errors higher than 50% (Treloar, 1997; Lenzen, 2001). Such partial selection is also vulnerable to convenient drawings of the boundaries to delude the outcome. On the other hand, expanding the system boundaries to include all interactions is computationally intractable because it demands huge amounts of information, some of which is inaccessible because of proprietary issues or unknown due to complexity in the system. However, comprehensive data for the entire economic network are available from government agencies in the form of economic input-output models (bea.gov), toxic release inventory (epa.gov/tri) and resource consumption (usgs.gov). These data do not require arbitrary boundary selection but are less detailed than data used in Process LCA or obtained from process engineering. LCA based on such data has been developed at the
economy scale (Lave et al., 1995). Most LCA methods fail to account for the contribution
of ecosystem goods and services—the very services that need to be sustained. Emergy analysis is a thermodynamic approach analogous to LCA at the value chain scale that
accounts for natural and economic capital. Emergy or embodied energy is a measure of
the available solar energy that has been used directly or indirectly to make a product
(Odum, 1996). It has been widely used in ecological analysis and modeling to determine
thermodynamic performance of ecosystems, but it has encountered a lot of resistance and
criticism from economists, physicists and engineers. Incorporation of natural capital
would represent analysis at the ecosystem scale. Finally, connecting all these spatial
scales must be carried on carefully because each scale provides data with different levels
9 of aggregation and uncertainty. Recent work on hybrid LCA focuses on ways of
combining data at different scales for LCA (Suh et al., 2004).
1.1 Sustainability
Sustainable development (WCED, 1987) was first defined as “development that
meets the needs of the present without compromising the ability of future generations to
meet their own needs.” Although this concept was embraced with great interest, it was
impractical because the meaning of the term “needs” is too vague and the needs of future
generations are simply unknown. Since its first appearance, sustainable development has
been defined in many other ways that allow clearer connections with more tangible goals.
An increasingly used concept is Sustainability, which is defined as the ability of a system
to function indefinitely. As opposed to sustainable development, sustainability does not
require defining needs, but it rather implies not endangering the systems that support
human kind's existence. A general sustainability strategy has often been described as
satisfying the triple bottom line, which means simultaneously enhancing economic
prosperity, environmental protection and social equity (Elkington, 1999).
Differences in the philosophical and ethical interpretation of sustainable
development have resulted in the emergence of the concepts of weak and strong
sustainability (Hediger, 1999). Weak Sustainability is an economic value principle that
states that the sum of economic, natural and social capital must be increasing or
maintained constant over time. This concept considers the three types of capital to be
mutually substitutable. Strong Sustainability is an ecologically-based physical principle
10 that states that natural capital must at least be maintained constant over time. It rejects the
notion of substitutability of natural capital, since many ecosystem goods and services like biodiversity and biogeochemical cycles cannot be replaced by other ecosystem, economic
or social processes. Both concepts can benefit from the quantification of natural capital
and its contribution to economic activity, which is one of the focuses of this dissertation,
as described in more detail in Section 1.3. The term “environmental sustainability” is
used in this dissertation to refer to the maintenance of natural capital.
1.2 Using Thermodynamics for Environmentally Conscious Decision Making
With increasing recognition of the importance of ecological products and services for industrial processes and human well-being, some approaches have been suggested to account for their contribution. These approaches are usually based on either economics or thermodynamics. Techniques from environmental economics attempt to assign a monetary value to ecological inputs (Bockstael et al., 2000; Nordhaus and Kokkelenberg,
1999). Methods based on thermodynamic principles rely on material and energy flow to account for ecological inputs. Material Flow Analysis (MFA) (Adriaanse et al., 1997;
Matthews et al., 2000) accounts for the flow of materials from the ecosystem to the economy, but ignores the inputs of ecological services. Energy flow analysis (Spreng,
1988) and its variations are promising due to their ability to objectively value all types of material and energy flows without violating physical laws, as methods like LCA often do.
The use of thermodynamic methods for evaluating environmentally conscious decision making of industrial products and processes is motivated by the fact that all
11 activities on earth rely on the availability of energy and its conversion to various goods
and services. Odum (1971, pp. 27) once wrote that “all progress is due to special power subsidies, and progress evaporates whenever and wherever they are removed.” He
suggested that it is ultimately the availability of energy, and not human creativity, what
guarantees survival of societies. Rifkin (2003) argues that slavery sustained ancient
civilizations just as fossil fuels maintain ours. Ancient technology developed
sophisticated weaponry to capture more land and slaves, just as modern technology
develops more advanced machinery to extract more fossil fuel. Consequently, it is in the preservation of our energy sources where really lies the key to progress and self-
preservation; and it is from ecosystem goods and services where our society finds its
power subsidies. Ultimately, all planetary activities depend on this available energy or
exergy (Ayres, 1994; Odum, 1996; Szargut et al., 1988), making it the ultimate limiting resource. Exergy provides a scientifically rigorous way to compare and combine streams of material and energy. Since exergy is lost or consumed in all processes for their functioning, it has been useful for improving process efficiency (Szargut et al., 1988).
Cumulative Exergy Consumption (CEC) analysis expands the analysis boundary by considering all industrial processes needed to convert natural resources into the desired industrial products or services. Many recent extensions of exergy analysis have focused on methods for environmentally conscious decision making and LCA (Ayres et al., 1998;
Sciubba, 2001; Wall, 2002; Connelly and Koshland, 2000; Cornelissen and Hirs, 2002).
The combination of exergy-based methods and LCA is attractive since exergy can provide a common ground for ecological and industrial processes, in which all types of
12 material and energy streams can be fairly assessed, or valued. In addition, exergy may be
related to some environmental impacts of emissions (Seager and Theis, 2002), may
quantify the sustainability of processes (Dewulf and Van Langenhove, 2002), and
characterize self-organized systems (Fath et al., 2001). Although these characteristics
make thermodynamics, especially exergy, ideal for the analysis of industrial and
ecological systems, the abovementioned efforts ignore the contribution of ecological products and services, thus limiting their ability to evaluate the “full cost” of industrial
activities.
1.3 Dissertation Statement
This dissertation focuses on the development of systematic approaches for
integrating economic and ecological considerations in process engineering decision making, recognizing that the holistic and multidisciplinary nature of this task demands unifying theories that can be easily assimilated outside the engineering community.
Challenges associated with this task are
• Evaluation of ecosystem goods and services or natural capital.
• Expansion of the analysis boundaries beyond the reaches of traditional process
systems engineering.
• Combination of data of different quality.
• Diversity of perspectives regarding the correct balance between economic
prosperity and environmental protection, i.e. the classical economic and
13 ecological trade-off.
This dissertation undertakes these challenges with the use of exergy and
thermodynamic principles because of all the promising characteristics discussed in the
previous section.
1.4 Structure of Dissertation
This dissertation is organized as a chapter containing the theoretical background
(Chapter 2), five chapters containing the main body of this research (Chapters 3 to 7) and
a chapter containing the general conclusions and future work (Chapter 8). A more
detailed description of each chapter is provided below.
Chapter 2 starts with a brief historical review of thermodynamics and the
introduction of the laws of thermodynamics. The methodologies developed and presented
in this work strongly rely on the use of collected data. As a result, this section describes a
technique for rectification of empirical data. This chapter introduces the concept of exergy, describes how to calculate it and shows many of its applications. This chapter closes with a section describing thermodynamic methods for environmentally conscious decision making that are relevant to this research.
An important and powerful feature of emergy analysis is that it accounts for the
contribution of ecosystem goods and services, which is a desired attribute in any method
for environmentally conscious decision making. Unfortunately, emergy has encountered
a lot of resistance and criticism, particularly from economists, physicists and engineers.
Some critics have focused on detailed practical aspects of the approach, while others have
14 taken issue with specific parts of the theory and claims. Chapter 3 discusses the main
features and criticisms of emergy and provides insight into the relationship between
emergy and concepts from engineering thermodynamics such as exergy and cumulative
exergy consumption. By identifying the main points of criticisms of emergy, this chapter
attempts to clarify many of the common misconceptions about emergy, inform the
community of emergy practitioners about the aspects that need to be communicated better
or improved, and suggest solutions.
Due to the importance of accounting for the contribution of ecosystems and the
obstacles hindering broader use of emergy, Chapter 4 starts with traditional or Industrial
CEC (ICEC) analysis, and expands it to include the contribution of ecosystems. The
approach proposed in this chapter determines the Ecological Cumulative Exergy
Consumption (ECEC) which also includes the exergy consumed by ecological processes
to produce the raw materials, dissipate the emissions, and functioning of industrial
processes. A systematic algorithm is presented for ECEC computation. Comparison of
ECEC with emergy indicates that both concepts are closely related. Such explicit
identification of the link between exergy and emergy should clear much of the confusion
and misunderstanding about both concepts, and enable combination of the best features
of both methods. This insight is used in this chapter to devise practical ways of combining emergy and exergy analyses to include the contribution of ecological inputs to
industrial processes, without being handicapped by the controversial aspects of emergy
analysis. The general approach in this chapter is illustrated with examples of the chlor-
alkali process and solar and coal-based processes for generating electricity. These
15 examples demonstrate the benefits of accounting for nature's inputs, and the unique
insight about environmental performance that may result from such accounting.
Expanding the analysis boundaries requires the additional effort of collecting all data and information of each process included in the life cycle. The performance of a life cycle analysis will strongly depend on the quality of this data. Since Life Cycle
Assessment (LCA) is widely used for environmentally conscious decision making, it is
the focus of Chapter 5. However, the challenges and solutions faced by LCA are
common to other life cycle analyses, such as ICEC and ECEC. Process data collected
from diverse sources may be outdated, unreliable, incomplete and/or unverifiable (Ayres,
1995). Frequently, material and energy balances are violated. Chapter 5 presents an
approach for data rectification of collected process data to enforce satisfaction of material
balance and the laws of thermodynamics. This approach uses the well known techniques
for data rectification and exergy analysis. The strength and challenges of implementing
such techniques are illustrated via the rectification of life cycle inventory data collected
for the production of caustic soda and chlorine.
Chapter 6 provides unique insight into the trade-off between ecological and
economic aspects of manufacturing processes when they are analyzed at multiple spatial
scales. Economic aspects are accounted for via traditional cost analysis and do not change
at different scales. Ecological factors are considered via exergy analysis of the inputs at
each scale, and depend on the selected processes. The scales considered in this analysis
correspond to existing methods at the manufacturing, value chain, economy and
ecosystem scales. The proposed methodology considers two objective functions:
16 economic cost and exergetic efficiency of the process life cycle at multiple scales. Trade- off between these objectives is represented via a series of Pareto optimal surfaces at various scales, thus avoiding arbitrary combinations until the final stages of decision making. Case studies of a heat exchanger and the CGAM cogeneration system (Valero et al., 1994) compare the proposed approach with existing methods, and highlight the benefits of adopting a multiscale and multiobjective view.
Industrial progress towards environmentally conscious decision making requires meaningful, practical and scientifically sound metrics. Most existing metrics rely on information about material and energy inputs and emissions from the main process and selected processes in its life cycle. Such metrics often result in multiple conflicting variables, making it difficult to use them for decision-making. Furthermore, metrics for environmentally conscious decision making need to be scientifically rigorous and capable of evaluating the broader economy and ecosystem scale impacts of selected processes and products. Chapter 7 proposes a framework for evaluating the environmental performance of industrial processes that satisfies these needs. This framework uses exergy analysis to combine different types of material and energy streams in a thermodynamically sound manner. Exergy analysis is also combined with end-point life cycle impact assessment
methods for evaluating the impact of emissions. This results in metrics for a selected
system with different levels of aggregation ranging from multiple to single dimensions.
The challenge of analyzing a process at life cycle and coarser spatial scales is met by
combining exergy analysis, life cycle assessment, input-output analysis, economic and
ecological aspects. The result is a doubly nested hierarchy, which analyzes processes at
17 multiple spatial scales of process, life cycle, economy and ecosystem. Each scale
contains another hierarchy based on the degree of aggregation of the metrics. A case
study of the ammonia process illustrates the characteristics of the proposed approach.
Chapter 8 provides concluding remarks on the achievements of this research. The
chapter closes with recommendations for implementing and enhancing the approaches
presented in this dissertation with the vision that someday they will be taught in chemical
engineering courses and become a common tool for evaluating industrial processes.
1.5 Summary of Contributions
The following list contains the major contributions made to connect traditional process engineering to environmentally conscious decision making.
• Detailed discussion of the main features and criticisms of emergy analysis needed
to identify the aspects that should be communicated better or improved, and to
bring this important tool into the mainstream of environmentally conscious
engineering decision making.
• Rigorous analysis of the relation between cumulative exergy consumption and
emergy analysis that demonstrates that they are conceptually analogous and
become identical under certain conditions.
• Development of a methodology for extending cumulative exergy consumption
analysis to include ecosystem goods and services. The framework is flexible to
the analyst’s allocation preferences.
18 • Explicit use of data rectification and exergy analysis as a means to cope with
inconsistent information of commercial and public process databases.
• Development of a hierarchical thermodynamic approach for utilizing available
information at various spatial scales of processes’ life cycle to evaluate ecological
objectives.
• Unique insight into the trade-off between ecological and economic aspects of
manufacturing processes when they are analyzed at multiple special scales.
• Use of thermodynamic methods at multiple spatial scales to overcome the
shortcomings that metrics for environmentally conscious decision making
commonly face.
• Demonstration of the features and benefits of exergy analysis and multiscale
hierarchical metrics.
• Demonstration of the application of the approaches developed in this research
through well known case studies.
19
CHAPTER 2
BACKGROUND
A theory is more impressive the greater is the simplicity of its premises, the more
different are the kinds of things it relates and the more extended its range of applicability.
Therefore, the deep impression which classical thermodynamics made on me.
–Albert Einstein
Everyday life circles around the belief that energy is consumed. For instance, householders paying for their utilities, classifying energy sources in renewable and nonrenewable, and global concerns about energy crises, only reinforce this idea.
Paradoxically, the first law of thermodynamics states that energy can neither be created nor destroyed. Then, how can energy be consumed or in short supply? Or why would energy sources be nonrenewable? The apparent contradiction originates from a misuse of terms. While energy cannot be consumed, it is constantly changing from available to unavailable, as a consequence of the second law of thermodynamics. In fact, entropy can be interpreted as the measure of the extent to which available energy in the universe is consumed (Rifkin, 2003). Consequently, it is available energy what householders are really paying for, what is harnessed from energy sources or what is constantly in short
20 supply. This available energy is also known as availability, essergy and exergy (Szargut
et al., 1988). Exergy is the term that is gaining more recognition worldwide. The concept
and applications of exergy are central to the work presented here, thus they are discussed
in detail in this chapter.
Since the reach of this work surpasses the limits of chemical engineering, this
chapter starts with a brief historical review of thermodynamics and the introduction of the
laws of thermodynamics. The methodologies developed and presented in this work
strongly rely on the use of collected data. The quality of this data is crucial to the
reliability of the methodologies’ outcome. Many of these databases incur in violations of
the material balance and the laws of thermodynamics. Consequently, the section of the
first law of thermodynamics also describes a technique for data reconciliation and gross
errors detection of empirical collected data. It appears in this section because it is based
on the enforcement of the material balance and the first law of thermodynamics. The
section of the second law of thermodynamics introduces the concept of exergy, describes
how to calculate it and shows many of its applications. This chapter closes with a section
describing thermodynamic methods for environmentally conscious decision making that
are relevant to this work.
2.1 Brief History of Thermodynamics
Thermodynamics is the branch of science concerned with the nature of heat and
its conversion to other types of energy. The laws of thermodynamics are the result of several thousand years of discoveries in mechanics and in the study of heat (Goldstein
21 and Goldstein, 1993). The ancient Greeks already had a good understanding of the
relation between force and work that are central to mechanics. For instance, they knew
how to produce a large force by the application of a smaller force with the help of a lever.
They were also aware of the existence of heat, though its relation to work was not clear.
The great breakthrough for mechanics came around 1666, when English scientist Isaac
Newton came up with the three laws of motion. Advances in the study of heat started with the advent of the steam engine and development of the caloric theory by the end of the eighteenth century. When it came to heat, there were disagreements between mechanics and the caloric theory; to a large extent because supporters of the caloric theory believed that heat was a material substance. Throughout the years, heat and other phenomena of nature such as kinetic and potential energy, electricity, and chemical reactions, were regarded as unconnected. They were developed independently and even measured in different units. In the early nineteenth century, it became apparent that these phenomena were interchangeable, i.e. that it was possible to produce one out of the other.
This raised the question of whether there was something that did not change in the intercourse of all these transformations.
It was between 1840 and 1850 that English physicist James Joule found a quantitative relation between heat and work, showing that they are just examples of energy and that heat was not a substance (Goldstein and Goldstein, 1993). Joules’ work led to the first law of thermodynamics, which states that the total energy of a system is conserved, regardless of the nature and number of transformations that the system undergoes. At this point, it was clear that the previously studied phenomena of nature
22 were interchangeable, measurable in the same units and that it is their total amount what
is conserved. Nowadays, the use of the first law is indispensable for the design and
operation of virtually every industrial process.
The first law does not rule out the possibility of having a machine that takes in one type of energy, transforms it into heat and back again into its original form in a perpetual cycle without any need for additional energy. If electricity could be recycled this way, there would be no need for fossil fuels. Since the middle ages, scientists have intuitively known that such machines cannot exist – yet there have always been attempts to create them. In 1824, French engineer Sadi Carnot conjectured that there was an unknown law of nature that did not allowed converting thermal energy into work without wasting some heat to the lower temperature heat reservoir (Goldstein and Goldstein,
1993). Later in 1850, German physicist Rudolf Clausius introduced the entropy property
of matter and stated the second law by saying that the entropy of the universe tends to a
maximum (Goldstein and Goldstein, 1993). The consequences of the second law are
tremendous because it imposes limits on how much work can be performed by a
thermodynamic system and explains why processes occur naturally in a certain direction.
2.2 First Law of Thermodynamics and Data Rectification
Since the law of mass conservation is central to the laws of thermodynamics, it is
briefly introduced in this section. According to the laws of physics, the mass flows
entering and leaving a system operating at steady state must be equal. This must apply at
any scale in industrial processes, from a piece of equipment through the whole plant, to
23 the entire value chain. Figure 2.1 shows an open thermodynamic system operating at steady state, such as a heat exchanger, a pump or a power plant. There are mass flows, denoted m& j , circulating across the system. By convention, mass flows entering the system have positive sign and those leaving it have negative sign. For this system, the law of mass conservation can be formulated as
∑ m& j = 0 (2.1) j
This equation must be true not only for each process unit, but also for any system enclosure. For non-reactive systems, the mass balance for each substance must hold. For processes with chemical reactions, the mass balance for each chemical element present must be satisfied.
The first law of thermodynamics—also referred as law of energy conservation— states that the energy flows entering and leaving a system operating at steady state must also be equal. Energy flows can be in the form of heat (e.g. solar radiation) or work (e.g. shaft work). The open thermodynamic system, shown in Figure 2.1, performs a certain amount of work W& . There are mass and heat flow streams circulating across the system
denoted m& j and Q& i , respectively. Ti is the temperature of the source interacting with the
system through the heat stream Q& i . By convention, mass and heat flows entering the system have positive sign and those leaving it have negative sign. On the other hand, work flow W& is considered positive if it is preformed by the system. The system travels
at a velocity υ and is located at a height z. Q& 0 represents the heat transferred between the
24 system and surroundings. The surroundings is at a temperature T0 and a pressure P0, it travels at a velocity υ0 and is located at a height z0. In the absence of nuclear, magnetic, electrical and interfacial effects, the first law of thermodynamic for the system can be written as (Sandler, 1989)
d 2 ˆ 2 ()U + m()ν 2 + zg = ∑ m& j (h j +ν j 2 + z j g)+ ∑Q& i + Q& 0 −W& (2.2) dt j i
Ti
T1 W& Q&1 Q&i υ
m&1 m& 2
m& j Thermodynamic System z
υ0 Q&0 z Surroundings (T ,P ) 0 0 0
Figure 2.1: General Open Thermodynamic System.
25 ˆ where U is internal energy of the system. h j , υj and zj are enthalpy per unit mass (the caret designates mass basis), velocity and height of the j-th stream, respectively. The term on the left hand side describes the rate of change of internal, kinetic and potential energy in the system. The first term on the right hand side represents the internal, kinetic and potential energy of the material streams. The second and third terms on the right hand side represent heat flow streams; and the fourth term is the flow of work performed by the system.
2.2.1 Data Rectification
Data rectification deals with the problem of improving quality of data to enhance understanding and interpretation of the operating status of the process. It develops on the premise that all measurements contain some type of error, which is why it is necessary to correct their values (Romagnoli and Sánchez, 2000). There are two types of error present in process data: random and systematic errors. Random errors are caused by small fluctuations in the properties and condition of the process or its surroundings. This type of errors is inherent in the measurement process. Systematic errors, also called gross error, are caused by incorrect calibration or malfunction of the instruments, process leaks, etc. As opposed to random errors, these errors are larger, occur occasionally, and should be identified and corrected. Because physical laws must be satisfied, they serve as hard constraints to the process of data rectification. There are three phases of the Data
Rectification process: Data Reconciliation, Gross Error Detection and Gross Error
Compensation. Data Reconciliation adjusts the measurements to satisfy the physical
26 constraints while minimizing the deviation from their original value. Gross Error
Detection determines which measurements contain gross errors. Gross Error
Compensation corrects the variables with gross errors.
In general, data reconciliation can be formulated by the following constrained weighted least-squares optimization problem (Narasimhan and Jordache, 2000)
Find xi ,u j for i = 1...n and j = 1...p n 2 Minimize ∑ wi ()yi − xi (2.3) i=1
S.t. g k ()x1 , x2 ,...xn ,u1 ,u2 ,...u p = 0 for k = 1...m
where xi is the reconciled estimate for the measured variable yi, uj is the estimate of the j- th unmeasured variable, wi is the weight for the square error of the i-th measurement and gk is a process constraint. There are n measured variables, p unmeasured variables and m process constraints. In general, the process is expected to have variables that are not measured due to economic or technical limitations.
2.2.1.1 Solution to the Data Reconciliation Problem
The solution described here was originally developed by Crowe et al. (1983) and has been used successfully for the last two decades. This solution applies to a linear steady state process with chemical reactions. Other cases have been studied in detail and their solutions can be found elsewhere (Crowe, 1986; Romagnoli and Sánchez, 2000;
Narasimhan and Jordache, 2000). Let K, J and R be respectively the number of process units, streams and chemical reactions in a process. Let Ck and Cj be the number of
27 components in process unit k and stream j, respectively. It is possible to describe the structure of a process through a K×J incidence matrix A, where Akj is 1 if the j-th stream enters the k-th process unit, -1 if the j-th stream leaves the k-th unit and 0 otherwise. A block matrix can be constructed by replacing each ±1 and 0 in A, by ±I (identity matrix)
and 0 (null matrix), respectively. Each matrix introduced has a size Ck × Ck , which
K J represents the number of components present in the k-th process unit. The ∑Ck × ∑C j k =1 j=1 balance matrix B is obtained by removing the columns of the block matrix where their pivotal ±1 connect a component that is present in a process unit but absent in the
K corresponding stream. Finally, a R × ∑Ck master stoichiometric matrix S is obtained by k=1 writing the stoichiometric coefficients in their corresponding place. Each row represents a chemical reaction occurring in a particular process unit and each column represents a component present in certain process unit. If the variables are measured in mass units, then each stoichiometric coefficient is divided by the molecular weight of its corresponding species before inserting them into the master stoichiometric matrix. Figure
2.2 shows an example of how the incidence, balance and master stoichiometric matrices are calculated.
The process variables that are linked to these structure matrices are the Cj component flow rates of all J streams and the extent of reaction ξr of the R reactions.
Accordingly, the vector of process variables f can be written as
28 4 (P) Stream 1 2 3 4 Unit 1 1 -1 0 1 1 Unit 2 Unit 3 Unit 2 0 1 -1 -1 (P) 1 (P, Q) 2 (Q) ⎡1 −1 0 1 ⎤ A = ⎢ ⎥ Reaction 1: P → Q ⎣0 1 −1 −1⎦
Stream 1(P) 2(P) 2(Q) 3(Q) 4(P) ⎡1 −1 0 0 1 ⎤ Unit 1 (P) 1 -1 0 0 1 ⎢0 0 −1 0 0 ⎥ (Q) 0 0 -1 0 0 B = ⎢ ⎥ ⎢0 1 0 0 −1⎥ Unit 2 (P) 0 1 0 0 -1 ⎢ ⎥ (Q) 0 0 1 -1 0 ⎣0 0 1 −1 0 ⎦
Unit 1 Unit 2 (P) (Q) (P) (Q) S = [−1 1 0 0] Reaction 1 -1 1 0 0
Figure 2.2: Example of construction of the incidence, balance and master stoichiometric matrices.
T T T T f = []f1 f 2 L f J ξ1 ξ 2 L ξ R (2.4)
where fj is the vector of component flow rates in stream j. The problem that ensures conservation of mass is formulated as
[]B ST f = 0 (2.5)
29 Considering that some of the process variables are unmeasured and others are known with certainty, the theory suggests classifying these variables into exactly known, measured and unmeasured (Crowe et al., 1983). Equation 2.5 can be rewritten as
B 0c + B1x + B 2u = 0 (2.6) where c, x and u are respectively the vectors of exactly known, measured and unmeasured variables. These vectors result from a reorganization of the variables in vector f. This operation can be represented in matrix algebra as
⎡c⎤ ⎡Ec ⎤ ⎢x⎥ =E f = ⎢E ⎥f (2.7) ⎢ ⎥ f ⎢ x ⎥ ⎣⎢u⎦⎥ ⎣⎢Eu ⎦⎥
where Ef is a square matrix containing the permuted rows of the identity matrix. Ec, Ex and Eu are partitions of Ef. B0, B1 and B2 from equation 2.6 can be calculated as
T T T T T T B 0 = []B S Ec , B1 = []B S Ex and B 2 = [B S ] Eu (2.8)
The solution of equation 2.6 involves two tasks: reconciling the measured variables and estimating the unmeasured variables. These tasks correspond to solving for x and then for u. Crowe et al. (1983) suggests multiplying equation 2.6 by a projection matrix P that satisfies
PB 2 = 0 (2.9)
So that equation 2.6 is reduced to
30 P[]B 0c + B1x = 0 (2.10)
A projection matrix can be found by applying a QR factorization in B2
(Narasimhan and Jordache, 2000), this is
B 2 = QR 1 = []Q1 Q 2 R1 (2.11)
K K K where Q is a ∑Ck × ∑Ck orthogonal matrix and R1 is a ∑Ck × n upper triangular k=1 k=1 k=1 matrix. The column vectors of B2, which belong to the subspace generated by the column vectors of Q1, are orthogonal to the column vectors of Q2. As a consequence,
T T Q 2 B 2 = 0 and P = Q 2 (2.12)
Crowe et al. (1983) derived the solution of equation 2.6 with the use of matrix algebra and Lagrange multipliers. The resulting equation for the vector of measured variables is
−1 T T T T T T x = y − Σ()Q 2 B1 []Q 2 B1 Σ ()Q 2 B1 Q 2 []B0c + B1y (2.13) where y is the vector of non-reconciled measured variables. Finally, the vector of unmeasured variables can be calculated as
T −1 T u = −()B 2 B 2 B 2 []B 0c + B1x (2.14)
There are cases in which some of the variables cannot be estimated. If a variable can be estimated from the measurements and the process constraints, then it is said to be 31 observable. A measured variable is redundant if it is observable even when its measurement is removed (Narasimhan and Jordache, 2000). Data reconciliation is not possible if there are not redundant variables. In this case, the problem is reduced to the estimation of unmeasured observable variables.
2.2.1.2 Gross Error Detection
The solution for data reconciliation implicitly assumes that only random errors are present. If there are gross errors in the measurements, then the resulting estimates may be inaccurate. Gross error detection involves strategies to identify the measurements with gross errors. Gross error compensation is the process of correcting these gross errors.
There are various methods for detecting gross errors (Narasimhan and Jordache,
2000). This paper will only describe the Maximum power measurement test, which is the one relevant to this work. A vector d can be calculated as d = Σ −1 ()y − x (2.15)
The vector d is normally distributed with a covariance matrix
−1 T T T T T T W = cov()d = ()()Q 2 B1 (Q 2 B1 Σ (Q 2 B1 ) ) (Q 2 B1 ) (2.16)
The following test statistics must hold for all measurements (Narasimhan and
Jordache, 2000)
d j zd , j = ≤ 1.96 j = 1,2,...,n (for a 5% level of confidence) (2.17) W jj 32 The measurements that do not satisfy equation 2.17 contain gross errors and should be compensated. Narasimhan and Jordache (2000) describe various methods for compensation.
2.3 Second Law of Thermodynamics and Exergy
The second law of thermodynamics establishes that the overall entropy in the universe increases inexorably. The entropy balance for a system such as the one in Figure
2.1 is (Sandler, 1989) dS Q& Q& ˆ i 0 = ∑ m& j s j + ∑ + + S&gen (2.18) dt j i Ti T0
where S and sˆ j are entropy of the system and entropy per unit mass of the j-th stream, respectively. Sgen is the rate of internal generation of entropy within the system. Sgen is zero for reversible processes. The absence of the work flow term indicates that work does not transfer entropy.
2.3.1 Exergy
An alternative way to interpret the second law is that although energy is neither created nor destroyed, its ability to do work decreases in real processes. For instance, friction in pipelines converts kinetic energy of the transported fluids into waste heat, which is energy with less capacity to do work (Hau and Bakshi, 2004b). This interpretation leads to the concept of exergy, which is formally defined as the useful work
33 performed by a system—at some arbitrary initial equilibrium state—when it is brought into thermodynamic equilibrium with the surroundings by means of a reversible process, i.e. a reversible path (Szargut et al., 1988). This final equilibrium state is achieved when the system reaches the same temperature, pressure, and composition as the surroundings
(since the surroundings are assumed to be an infinite reservoir, their thermodynamic state remains unaltered when interacting with the system). The condition of same composition implies that components of the system that do not exist in the surroundings cannot be present in the final equilibrium state. The temperature, pressure, and composition of the surroundings must be explicitly defined for this calculation, and are specified so that they represent a reasonable model of the environmental surroundings of the process of interest. The thermodynamic state of this model is called the reference state; and the chemical substances present in the reference state are referred to as reference species.
2.3.1.1 The Reference State
A distinguishing difference between exergy and other thermodynamic properties is that the reference state for exergy is determined by the surroundings. In fact, exergy of matter will change if the state of the surroundings changes, even when no changes occur in the system itself. Consequently, knowledge of the state of the surroundings is necessary for exergy calculations. Local surroundings can be considered, for which reference states for physical, kinetic and potential exergies should not be hard to get. A temperature of 25 °C (298.15 K), a pressure of 101.323 kPa and zero values for the height z0 and velocity υ0 of the earth surface are typically taken as reference state.
34 Defining the reference state for calculating chemical exergies is more complicated because the exact composition of the surroundings is usually unknown and because the state of the surroundings is generally not in thermodynamic equilibrium.
In defining the reference state, Szargut et al. (1988) considered three different natural surroundings, namely the Earth’s atmosphere, hydrosphere and lithosphere. As expected, the assumption of thermodynamic equilibrium suits better for the atmosphere and second for the hydrosphere. Thermodynamic calculations are also easier for the atmosphere because, for most cases, the properties of ideal gas can be applied. Using reference states based on projections of how the Earth would be if it was at thermodynamic equilibrium has been investigated (Szargut et al., 1988). However, for the reasons discussed by Szargut et al. (1988), it is more convenient and advantageous to adopt the reference states given by the current average compositions of the Earth’s atmosphere, hydrosphere and lithosphere.
The atmosphere provides an appropriate reference state for 10 chemical elements.
Other elements are represented by solid species, in the hydrosphere and lithosphere, in which they are most commonly present (see Szargut et al., 1988). Table 2.1 shows the elements present in the atmosphere in their most stable form and in the form in which they are present in the reference state (Szargut et al., 1988). According to Table 2.1, if a system containing NO, N2, and O2 is brought into equilibrium with its surroundings, the
1 1 final equilibrium state will be 298.15 K, and N2 and O2 at 7.578×10 and 2.039×10 kPa, respectively. Since NO is not a reference species, the system can perform work by converting NO into N2 and O2.
35 Element in Reference Species Standard most stable Mean Pressure, Standard Chemical Chemical Exergy, r o a o b form, i P0,r (kPa) Exergy, bch,r (kJ/mol) bch,i (kJ/mol) Ar Ar 9.060×10-1 11.69 11.69 -2 C (graphite) CO2 3.350×10 19.87 410.26 -4 D2 D2O g 3.420×10 31.23 263.79 0 H2 H2O g 2.200×10 9.49 236.09 He He 4.850×10-4 30.37 30.37 Kr Kr 9.700×10-5 34.36 34.36 1 N2 N2 7.578×10 0.72 0.72 Ne Ne 1.770×10-3 27.19 27.19 1 O2 O2 2.039×10 3.97 3.97 Xe Xe 8.700×10-6 40.33 40.33 a Equation 2.35; b Equation 2.36.
Table 2.1: Conventional reference state for the atmosphere at 298.15 K and 101.325 kPa (Szargut et al., 1988).
2.3.1.2 Maximum Work
The amount of work performed by the system, such as the one in Figure 2.1, can be derived from the laws of thermodynamics. Equation 2.18 can be rewritten—to get an expression for heat transferred between the system and surroundings or waste heat, as commonly referred to—as
dS T ˆ 0 Q& 0 = T0 − T0 ∑ m& j s j − ∑ Q& i − T0 S& gen (2.19) dt j i Ti
36 Notice from equation 2.19 that entropy generated by the system is not the only factor contributing to waste heat. An entropy decrease in the system, incoming material and heat flows also produce waste heat. An expression for the amount of work W& performed by the system can be obtained by plugging equation 2.19 into equation 2.2 and rearranging terms to give
d W& + T S& = − ()U − T S + m(ν 2 2 + zg) 0 gen dt 0 (2.20) ⎛ T ⎞ ˆ ˆ 2 ⎜ 0 ⎟ + ∑ m& j ()h j − T0 s j +ν j 2 + z j g + ∑⎜1− ⎟Q& i j i ⎝ Ti ⎠ where the second term on the left hand side represent lost work, i.e. mechanical energy converted into thermal energy by system irreversibilities such as friction. It can be observed from equation 2.20 that if the system operates reversibly, then entropy
generated Sgen is zero and maximum work W&max can be performed. This is
d W& = − ()U − T S + m()ν 2 2 + zg max dt 0 (2.21) ⎛ T ⎞ ˆ ˆ 2 ⎜ 0 ⎟ + ∑ m& j ()h j − T0 s j +ν j 2 + z j g + ∑⎜1− ⎟Q& i j i ⎝ Ti ⎠
2.3.1.3 Exergy Calculation
To calculate the exergy of a material stream let consider an open thermodynamic system operating at steady state as illustrated in Figure 2.3. The system is composed of a series of reversible thermal machines and frictionless semi-permeable membranes.
Nuclear, magnetic, electrical and interfacial effects are neglected. According to the above 37 definition, exergy of a substance can be determined by calculating the work performed by the system in Figure 2.3, when the state of the substance is brought to equilibrium with the state of its surroundings. At first, the substance is brought to the height of the surroundings. The work performed by this process is known as potential exergy Bp and is equivalent to potential energy. Then, the substance is brought to the same velocity of the surroundings. The work performed from this change is called kinetic exergy Bk and is also equivalent to kinetic energy. The third thermal machine calculates the physical exergy Bph by bringing the substance to the temperature and pressure of the surroundings—a state symbolized with the sub-script 0—by reversible means. The rest of the system described by Figure 2.3 calculates the chemical exergy Bch. Once the substance under consideration is at the temperature and pressure of the surroundings, frictionless semi-permeable membranes separate its components. There are two possible cases regarding the components of the substance under consideration: (i) the species are present in the surroundings—they are part of the reference species of the surroundings— but at different concentrations; or (ii) the species are not present in the surroundings. In the first case, work can be performed by bringing the concentrations of the substance’s components to that of the surroundings. In the second case, each component that is not present in the surroundings undergoes a reference chemical reaction where only reference species are involved. Reference species may be used as reactants. The final products of the reaction must be reference species. Work can be performed from the heats of reaction and from bringing the concentration of the reaction effluents to that of its surroundings.
The stream Q0 represents the total heat transferred between the system and surroundings.
38
Figure 2.3: Model for calculating exergy of a material stream.
39 Exergy of a material stream can be calculated by applying equation 2.21 on the system in Figure 2.3. This is
Actual State ˆ 2 B& j = W&max = m& j ()h j − T0 sˆ j +ν j 2 + z j g , ∀j ∈{}Set of mass streams Reference State
(2.22)
When identifying the input and output streams, equation 2.22 becomes
ˆ 2 B = m(h()T, P − T0 sˆ ()T, P +ν 2 + zg) ˆ ˆ 2 + ∑ m0,r ()h0,r ()T0 , P0,r − T0 s0,r ()T0 , P0,r +ν 0 2 + z0 g (2.23) r∈Effluents ˆ ˆ 2 − ∑ m0,r ()h0,r ()T0 , P0,r − T0 s0,r ()T0 , P0,r +ν 0 2 + z0 g r∈Influents where Effluents and Influents are reference species. For example, let consider a vessel filled with pure methane (CH4) at some temperature T and pressure P. The exergy is the useful work performed by the system when it is brought to a final state with temperature
T0, pressure P0 and containing only CO2 and H2O, each species at the same partial pressure as in the surroundings (see Table 2.1). These reference species can be connected to CH4 though the following reaction
CH 4 + 2O2 → CO2 + 2 H 2O (2.24)
where CO2 and H2O are effluents and O2 is an influent. The conceptual semi-permeable membrane is needed to allow influx of O2 for the reaction of CH4 to their reference species, and to allow influx and efflux of any reference species to bring them to the same
40 concentrations as in the surroundings. The exergy of CH4 can be estimated by using
Figure 2.3 and equation 2.23, which is reduced to
⎡ MW ˆ CO2 ˆ BCH = mCH ⎢()hCH ()T, P − T0 sˆCH ()T, P + ()hCO ()()T0 , P0,CO − T0 sˆCO T0 , P0,CO 4 4 4 4 MW 2 2 2 2 ⎣⎢ CH 4 2MW H 2O ˆ + ()hH O ()()T0 , P0,H O − T0 sˆH O T0 , P0,H O MW 2 2 2 2 CH 4 2MW ⎤ O2 ˆ − ()hO ()()T0 , P0,O − T0 sˆO T0 , P0,O ⎥ MW 2 2 2 2 CH 4 ⎦⎥
where MWi represents the molecular weight of the corresponding i-th species and P0,i is the partial pressure of the corresponding i-th reference species in the surroundings.
Calculation of equation 2.23 requires a fully defined state of the surroundings, specifically, height z0, velocity υ0, temperature T0, pressure P0 and composition of the r reference species x0,i (for i = 1 to r-1). Since the system involves chemical reactions, the choice of the reference state—for the calculation of the thermodynamic properties—must be done carefully. To be consistent with the heat of reaction data of all chemical reactions, the zero value of enthalpy can be set only once for each element (Sandler,
o 1989). Standard molar enthalpies of formation ∆h f and Gibbs free energies of formation
o ∆g f are used for this purpose. Their values for common species are available in most chemistry (Chang, 2001) and thermodynamics books (Sandler, 1989). The convention is
o o that ∆h f and ∆g f are set to zero for the most stable form of each element. At a standard temperature of 298.15 K and pressure of 101.3 kPa, for instance, the most stable forms of elemental carbon, oxygen and hydrogen are respectively graphite (C), molecular oxygen
41 o o (O2) and molecular hydrogen (H2). Therefore their values of ∆h f and ∆g f are set to
o zero. These standard properties of formation allow to calculate the heat of reaction ∆H rxn
o and the free Gibbs energy of reaction ∆Grxn . For example the standard heat and free
Gibbs energy of reaction for the reaction of CH4 shown in equation 2.24 are
∆H o = ∆ho + 2∆ho − ∆ho − 2∆ho and rxn f ,CO2 f ,H 2O f ,CH 4 f ,O2
∆G o = ∆g o + 2∆g o − ∆g o − 2∆g o rxn f ,CO2 f ,H 2O f ,CH 4 f ,O2
The exergy of methane can be expressed in terms of the standard free Gibbs energy of reaction by further rearranging equation 2.23. This is
ˆ ˆ o BCH = mCH [()()hCH ()T, P − T0 sˆCH ()T, P − (hCH ()T0 , P0 − T0 sˆCH ()T0 , P0 ) − ∆Grxn 4 4 4 4 4 4 MW CO2 ˆ ˆ + ()()hCO ()T0 , P0 − T0 sˆCO ()T0 , P0 − ()hCO ()()T0 , P0,CO − T0 sˆCO T0 , P0,CO MW 2 2 2 2 2 2 CH 4 2MW H 2O ˆ ˆ + ()()hH O ()T0 , P0 − T0 sˆH O ()T0 , P0 − ()hH O ()()T0 , P0,H O − T0 sˆH O T0 , P0,H O MW 2 2 2 2 2 2 CH 4 2MW ⎤ O2 ˆ ˆ − ()()hO ()T0 , P0 − T0 sˆO ()T0 , P0 − ()hO ()()T0 , P0,O − T0 sˆO T0 , P0,O ⎥ MW 2 2 2 2 2 2 CH 4 ⎦⎥
By assuming that the system and surroundings behave like ideal gases, the above equation is reduced to
⎡ ⎛ T ⎛ T C p,CH R ⎛ P ⎞⎞⎞ B = m ⎢⎜ C dT − T ⎜ 4 dT − ln⎜ ⎟⎟⎟ − ∆G o CH 4 CH 4 p,CH 4 0 ⎜ ⎟ rxn ⎜ ∫T0 ⎜ ∫T0 T MW P ⎟⎟ ⎢ ⎝ CH 4 ⎝ 0 ⎠⎠ ⎣⎝ ⎠ RT ⎛ P ⎞ 2RT ⎛ P ⎞ 2RT ⎛ P ⎞⎤ − 0 ln⎜ 0 ⎟ − 0 ln⎜ 0 ⎟ + 0 ln⎜ 0 ⎟⎥ MW ⎜ P ⎟ MW ⎜ P ⎟ MW ⎜ P ⎟ CH 4 ⎝ 0,CO2 ⎠ CH 4 ⎝ 0,H 2O ⎠ CH 4 ⎝ 0,O2 ⎠⎦⎥
42 For convenience, exergy of a substance is often partitioned into the following four terms to distinguish between the nature of exergy (Szargut et al., 1988)
B = Bch + B ph + Bk + B p (2.25)
where as defined previously, Bch, Bph, Bk and Bp are respectively chemical, physical, kinetic and potential exergies. The exact value for these exergies can be calculated by applying 2.22 on the corresponding subsystems shown in Figure 2.3. From Figure 2.3, expressions for physical, kinetic and potential exergies are readily available. They are
B p = m()z − z0 g (2.26)
2 2 Bk = m(ν −ν 0 ) 2 (2.27)
B =m hˆ()()T, P, x , x ,..., x − hˆ T , P , x , x ,..., x ph [()1 2 n−1 0 0 1 2 n−1 (2.28)
− T0 ()sˆ()()T, P, x1 , x2 ,..., xn−1 − sˆ T0 , P0 , x1 , x2 ,..., xn−1 ]
Equation 2.28 can also be written for molar thermodynamic properties as
B = n[()h()T, P, x , x ,..., x − h(T , P , x , x ,..., x ) ph 1 2 n−1 0 0 1 2 n−1 (2.29) − T0 ()s()()T, P, x1 , x2 ,..., xn−1 − s T0 , P0 , x1 , x2 ,..., xn−1 ]
Chemical exergy can be calculated by adding the remaining three work terms in
Figure 2.3. This is
43 ⎛ ⎞ o Bch = n⎜ g()T0 , P0 − ∑ xn g n ()T0 , P0 ⎟ + n∑ xn ()− ∆Grxn n ⎝ n ⎠ n
+ ∑ n0,r []g 0,r ()T0 , P0,r − g 0,r ()T0 , P0 (2.30) r∈Effluents
− ∑ n0,r []g 0,r ()T0 , P0,r − g 0,r ()T0 , P0 r∈Influents
Quite often, the term thermal exergy, denoted Bth, is used to refer to the sum of both the physical and chemical exergies. This is
Bth = Bch + B ph (2.31)
2.3.1.4 Standard Chemical Exergy
The standard chemical exergy of a substance is its specific exergy at standard temperature and pressure (Szargut et al., 1988). From equation 2.30, it is clear that the
o standard exergy of a reference species bch,r is
o o Bch,r bch,r = = g 0,r ()T0 , P0,r − g 0,r ()T0 , P0 (2.32) n0,r
By using equation 2.32, equation 2.30 can be rewritten in terms of standard chemical exergy of the reference species as
o Bch = n∑ xn []g n ()()T0 , P0 − g n T0 , Pn + n∑ xn (− ∆Grxn )n n n (2.33) o o + ∑ n0,r bch,r − ∑ n0,r bch,r r∈Effluents r∈Influents
44 In the case of ideal gases and by assuming constant heat capacity Cp, the isothermal change of Gibbs free energy for a substance can be calculated as follows
(Sandler, 1989)
⎛ P ⎞ IG IG ⎜ ⎟ g ()T0 , P − g (T0 , P0 )= T0 R ln⎜ ⎟ (2.34) ⎝ P0 ⎠
In the atmosphere, the properties of ideal gas can be assumed. Therefore, by plugging equation 2.34 into equation 2.32, the standard chemical exergy of the reference species in the atmosphere can be calculated as
⎛ P ⎞ bo = T R ln⎜ 0 ⎟ (2.35) ch,r 0 ⎜ ⎟ ⎝ P0,r ⎠
where P0,r is the partial pressure of the reference species in the atmosphere and are shown in Table 2.1. The standard chemical exergy of other substances can be calculated from equation 2.33 as
o o o o bch,n = −∆Grxn + ∑ν r bch,r − ∑ν r bch,r (2.36) r∈Effluents r∈Influents
For instance, molecular hydrogen can be converted to species in the reference state through the following chemical reaction
H + 1 O → H O (g) ∆G o = −228.6kJ / mol (2.37) 2 2 2 2 rxn
Then, the exergy of molecular hydrogen can be determined with equation 2.36
45 bo = −∆G o + bo − 1 bo = 236.1kJ / mol (2.38) ch,H 2 rxn ch,H 2O 2 ch,O2
2.3.1.5 Exergy vs. Energy
Exergy does not bring new information to what is already known through the laws of thermodynamics. Nevertheless, it offers a different perspective and represents a more convenient concept because it is very similar to enthalpy and Gibbs free energy, in contrast to entropy, and has the same units of energy but reflects the constraints of the second law, in contrast to energy.
An exergy balance equation can be obtained by combining first and second laws of thermodynamics. With the right reference state, equation 2.20 can be rewritten in exergy terms as
d ⎛ T ⎞ ⎜ 0 ⎟ W&lost = T0 S& gen = − ()B − PV + ∑ m& j b j + ∑⎜1− ⎟Q& i −W& (2.39) dt j i ⎝ Ti ⎠
where W&lost represents work or exergy lost due to process irreversibilities. On the far right hand side of equation 2.39, the first term accounts for exergy accumulation in the system, the second, third and fourth terms account for the exergy of all mass, heat and work flows, respectively, entering or leaving the system. The energy balance equation representing the first law, as shown by equation 2.2, can be rearrange as
d 2 ˆ 2 0 = − ()U + m()ν 2 + zg + ∑ m& j (h j +ν j 2 + z j g)+ ∑Q& i + Q& 0 −W& (2.40) dt j i
46 Comparing equations 2.39 and 2.40 shows that, (i) Exergy is lost, while energy is conserved; (ii) Only a fraction of heat, which is determined by the Carnot efficiency factor, is useful energy. This fact is captured by the exergy but not by the energy balance equation; (iii) Work is pure useful energy, therefore its term in the exergy balance does not differ from that in the energy balance.
2.3.2 Exergy Applications
This section illustrates how typical situations in everyday life and in process engineering can be treated from an exergy perspective. The examples provided below demonstrate the convenience of using exergy as a way of interpreting the second law of thermodynamics.
2.3.2.1 The Price of Steam
A remarkable feature of exergy is that it offers a measure of energy quality. For example, Table 2.2 shows the price per unit of energy of process steam at three different conditions. At first, it seems that low pressure steam is a better deal because 1 GJ of it is cheaper than 1 GJ of medium or high pressure steam. By calculating their price per unit of exergy, these types of steam can be compared based on their ability to do work. Not surprisingly, these prices are now comparable. Such result suggests that it is exergy, and not energy, what is valued by the economy. This, however, does not imply that there is an exact relation between economic value and exergy. As opposed to exergy, economic valuation involves human perception and willingness to pay.
47 Process Steam Conditions Price per unit Exergy Mass Energy Exergy Energy ($/t)a ($/GJ)b ($/GJ)c (%) Low Pressure 5 barg , 160°C 6.62 2.93 8.54 34.35 Medium Pressure 10 barg , 184°C 7.31 3.63 8.38 43.30 High Pressure 41 barg , 254°C 8.65 5.05 8.22 61.36 a From Turton et al. (1998); b Price per unit mass divided by latent heat; c Price per unit mass divided by thermal exergy from equation 2.31.
Table 2.2: Price per unit of process steam at different conditions.
Finally, the last column of Table 2.2 shows the ratio of exergy to energy content of the streams. This ratio can be regarded as an indicator of energy quality. It can be observed from Table 2.2 that higher pressure steam provides energy of higher quality.
2.3.2.2 Exergy of Heat Transfer
The exergy of a heat stream, B&i , can be calculated from equation 2.21 as
⎛ T ⎞ ⎜ 0 ⎟ B&i = W&max = ⎜1− ⎟Q& i , ∀i ∈{}Set of heat streams (2.41) ⎝ Ti ⎠ where the term surrounded by parentheses is well known as the Carnot efficiency factor.
Equation 2.41 implies that only a fraction of the energy content in a heat stream can be converted into work and this depends on the temperature of the source and the surroundings.
48 The study of heat transfer offers an interesting perspective on the inexorably production of entropy and loss of exergy. The loss of exergy that results from the transfer of heat from a state A to a state B at lower temperature can be calculated from equation
2.39. This is
⎛ 1 1 ⎞ ⎜ ⎟ W&lost = ⎜ − ⎟T0Q& (2.42) ⎝ TB TA ⎠
Equation 2.42 implies that there is an exergy loss inherent to heat transfer. This is a striking result because it says that no matter how efficient the heat transfer is carried on, the mere fact of having a temperature gradient is enough to incur in an exergy loss.
Energy integration is a popular practice in industrial processes for minimizing the cost of utilities by maximizing the utilization of heat streams in the plant (Douglas,
1988). Pinch analysis is the most accepted method for this purpose (Linnhoff, 1997). It implicitly minimizes entropy generation and exergy losses in the heat exchangers network (HEN). To illustrate the potential of pinch analysis, let assume that it is needed to cool down two hot streams with two cold streams with the minimum use of additional utilities. Specifications of the four streams are shown in the third and fourth columns of
Table 2.3. Let assume that an additional supply of heat—hot utility—is available at 300ºF and an additional heat sink—cold utility—is available at 80ºF. Temperature of the surroundings is 77ºF. Heat transferred from/to a stream that goes from state A to state B
(Douglas, 1988) is
ˆ Q& i = m& iC p,i ()Ti,B − Ti,A (2.43)
49 ˆ where m& i is the mass flow and C p,i is the constant-pressure specific heat in mass units of
ˆ the i-th stream. The product of m& i and C p,i , and the heat transferred for each stream in the example are shown in the second and fifth columns of Table 2.3, respectively.
Negative values of heat transferred indicate that heat is leaving the stream. The exergy transferred from/to a stream that goes from state A to state B can be calculated from the differential form of equation 2.41, which is
⎛ T ⎞ ⎜ 0 ⎟ dB&i = ⎜1− ⎟dQ& i (2.44) ⎝ Ti ⎠
where dQ& i is the differential form of equation 2.43, i.e.
ˆ m C T T Q& B& i & i p,i i,A i,B i i (Btu/hr·ºF) (ºF) (ºF) (Btu/hr) a (Btu/hr) b 1 (hot) 1,000 250 120 -130,000 -21,410 2 (hot) 4,000 200 100 -400,000 -47,102 3 (cold) 3,000 90 150 180,000 13,203 4 (cold) 6,000 130 190 360,000 47,975 c Energy Balance:10,000 W&lost = 7,334 a Equation 2.43; b Equation 2.46; c W&lost = −∑ B&i
Table 2.3: Specification of the heat streams required for the design of a heat transfer scheme.
50 ˆ dQ& i = m& iC p,i dTi (2.45)
Integrating equation 2.44 from state A to state B gives
B⎛ T ⎞ ⎛ ⎛ T ⎞⎞ 0 ˆ ˆ ⎜ ⎜ i,B ⎟⎟ B&i = ⎜1− ⎟m& iC p,i dT = m& iC p,i Ti,B − Ti,A − T0 ln (2.46) ∫A ⎜ T ⎟ ⎜ ⎜ T ⎟⎟ ⎝ i ⎠ ⎝ ⎝ i,A ⎠⎠
Exergy transferred for each stream, in the example, is shown in the sixth column of Table 2.3, where the negative values indicate that exergy is leaving the stream. A balance of the energy transferred by the streams, shown at the bottom of the fifth column of Table 2.3, indicates a heat deficit of 10 thousands Btu/hr. An uneducated conclusion is that an additional heat supply of 10 thousands Btu/hr is enough to satisfy the specifications in the example. However, this may result in a violation of the second law because heat can only be transferred from a source at a higher temperature than that of its destination. Therefore, there is a minimum additional energy that must be supplied in order to satisfy both the specifications and the second law. Similarly, there is a minimum exergy loss that is inevitable. This exergy loss is 7,334 Btu/hr for the streams of the example, as shown at the bottom of the sixth column of Table 2.3. This value is clearly not the minimum loss because there is at least a heat deficit of 10 thousands Btu/hr.
A well-known contribution of pinch analysis is that it ensures satisfaction of the second law of thermodynamics. Figure 2.4 shows the heat transfer scheme requiring minimum additional utilities determined through pinch analysis. All the hot streams are grouped together to form the hot composite curve. The cold composite curve is formed similarly with the cold streams. Refer to Douglas (1988) for more information on how to
51 obtain the composite curves. Results shown in Figure 2.4 indicate that 70 thousands
Btu/hr of additional heating and 60 thousand Btu/hr of additional cooling are required to satisfy both the specifications of the example and the second law. The exergy of these utilities can be calculated with equation 2.41. Updating the previous value of exergy loss by including the exergy transferred by the utilities gives a new value of 27,548 Btu/hr.
Satisfaction of the second law is ensured by specifying a minimum temperature approach prior to constructing the composite curves, for instance, it is 10ºF as shown in
Figure 2.4. There would be a second law violation if the composite curves were allowed to cross one another. Exergy losses can be illustrated through an exergy composite curve, which can be obtained by replacing temperature by Carnot efficiency in Figure 2.4. The exergy composite curves are shown in Figure 2.5. The area enclosed by the two composite curves represents exergy loss. Figure 2.5 allows identifying the areas of larger exergy loss which opens possibilities for improvements. One possible improvement is to use a hot utility at a lower temperature.
As a concluding remark, minimizing exergy losses in a process will result in better utilization of energy inputs. Nevertheless, it is not possible to eliminate the total amount of exergy losses in a process because there is a minimum amount of exergy loss that is required as a driving force to operate the process at the desired throughput, as illustrated by the abovementioned example of the heat streams. Feng et al. (1996) and
Feng and Zhu (1997) decomposed exergy losses into two parts: avoidable exergy loss and inevitable exergy loss. Inevitable exergy loss is the minimum exergy loss required to operate a process at a specific rate. Avoidable exergy loss is the loss that occurs due to
52 imperfections and inefficiencies in the process that derive from bad design, indiscriminate use of resources, leaks, emissions, etc. Avoidable exergy loss can be avoided by improving the design and operation of the process. Summarizing, any total exergy losses can be expressed as follows
350
300 Minimum Temperature 250 Approach of 10°F Hot Side 200
150 Cold Side
100 Temperature (°F)
50
0 0 100 200 300 400 500 600 Heat Transferred (Thousands Btu/hr)
Figure 2.4: Composite Curves for the example specified by Table 2.3.
53 30
25
20 Shaded area between curves Hot Side represents Exergy Loss
15 Cold Side 10
Carnot Efficiency (%) 5
0 0 100 200 300 400 500 600 Heat Transferred (Thousands Btu/hr)
Figure 2.5: Exergy Composite Curves for the example specified by Table 2.3.
W&lost = W&lost,INE +W&lost,AVO (2.47)
where W&lost,INE is inevitable exergy loss and W&lost,AVO is avoidable exergy loss. Techniques and correlations for calculating inevitable exergy losses for different units can be found elsewhere (Feng et al., 1996; Bejan, 1997). Alternatively, the concept of exergetic efficiency can be used. Exergetic efficiency is defined as
54 ∑ B& j η = j , ∀j ∈{}{}Set of product streams ∧ ∀n ∈ Set of input streams (2.48) ∑ B& n n
Since all real processes experience an inevitable exergy loss, they can only operate at an efficiency that is lower to the maximum achievable efficiency, ηmax, which can be determined as
∑ B& n −W&lost,INE n η max = , ∀n ∈{}Set of input streams (2.49) ∑ B& n n
2.3.2.3 Exergy and Ecosystems
The second law predicts that all isolated systems will inevitably tend to an equilibrium that represents a state of maximum disorder, in which no further changes will occur. When first stated, it brought a lot of controversy in the field of biology. Evolution did not seem to obey the second law, since life itself was far from tending to a state of total disorder. Progress in both physics and biology helped understand how evolution is possible despite the second law. In 1905, Austrian Physicist Ludwig Boltzmann suggested that the struggle for existence is a struggle for free energy available for work
(Boltzmann, 1886). Boltzmann’s statement entails that an external source of exergy is required for life to exist. US mathematician and statistician Alfred Lotka pointed out that, systems that prevail develop designs that maximize the flow of useful energy (Lotka,
1922). Lotka’s idea implies not only that living organisms require a source of exergy to exist, but also that they need to evolve in ways that allow them to maximize this exergy 55 inflow. Austrian Physicist Erwin Schrödinger held that organization is maintained by extracting order from the environment (Jørgensen, 2002). In general, all these statements reinforce the idea that life is possible because the sun provides a source of exergy that allows ecosystems to stay away from equilibrium.
Living organisms are part of a group known as self-organized systems. Such systems constantly restructure themselves to optimize their inflow of exergy. Self- organization is created in the presence of sustained energy or material gradients. Self- organized systems tend to stay far from equilibrium, minimize their entropy content
(disorder), maximize their exergy and possibly maximize the rate of exergy consumption for themselves and systems they depend on (Fath et al., 2001). As a consequence, development and growth are limited by the availability of resources and the ability to exploit them. Rifkin (2003) even relates the rise and fall of civilizations to their capacity for harnessing exergy from the environment.
2.4 Thermodynamic Methods for Environmentally Conscious Decision Making
Thermodynamic methods provide a solid and rigorous foundation for valuing ecosystem products and services. The use of these methods for environmentally conscious decision making started to proliferate during the energy crisis of the seventies with the purpose of minimizing the consumption of fossil fuels, whose high prices were wreaking havoc in the world economy. In general, these methods focused on energy and materials consumption. These methods included Net Energy Analysis (Spreng, 1988),
Thermoeconomics (Bejan et al., 1996), Cumulative Exergy Consumption (Szargut et al.,
56 1988) and Emergy Analysis (Odum, 1996). The second boost of thermodynamic methods for environmentally conscious decision making occurred in the early nineties, as a result of increasing environmental awareness. This second generation of methods for environmentally conscious decision making gives importance to the environmental impact of emissions and waste disposal. These methods include Life Cycle Assessment,
Exergetic Life Cycle Assessment (Cornelissen and Hirs, 2002), Thermodynamic Input-
Output Analysis (Ukidwe and Bakshi, 2004) and Extended Exergy Accounting (Sciubba,
2001).
The following sections describe the thermodynamic methods that are most relevant for this work. They are Life Cycle Assessment, Cumulative Exergy
Consumption, Thermoeconomics, Thermodynamic Input-Output Analysis and Emergy
Analysis.
2.4.1 Life Cycle Assessment
Life cycle assessment (LCA) evaluates the environmental impact of emissions released by all the systems involved in the life cycle of a product or process (Guinée et al., 1993a,b). A product-based LCA includes all the processes involved in the value chain of the product under consideration, e.g. extraction and processing of raw materials; processing, use, reuse and recycle of the products and waste disposal. A process-based
LCA includes planning, construction, operation and dismantling of the process under study (Lombardi, 2001). Product-based LCAs are much commonly used than Process- based LCA (Guinée et al., 1993a,b; Jensen et al., 1998; Graedel, 1998). LCA is still a
57 young and evolving tool. At its current status, there is a lot of controversy and uncertainty regarding feasibility of the study and accuracy and reliability of the results. Despite this fact, several companies and organizations continue using LCA for product improvement, design of new products, product information, ecolabelling, and the exclusion or admission of products from or to the market (Guinée et al., 1993a,b). LCA is also part the norms ISO 14000. Although it is not clear how, LCA promises to play an important role in policy making (UNEP, 1999) and process selection, design and optimization
(Azapagic, 1999; Burgess and Brennan, 2001).
2.4.1.1 Methodological Framework
Traditional LCA consists of four stages: goal definition and scoping, inventory analysis, impact assessment and improvement (Schaltegger et al., 1996). As shown in
Table 2.4, Goal definition and scoping consist of defining the goals of the analysis, setting up the system boundaries and validating the data. Inventory analysis has the steps of recording and allocation. Recording consists of collecting information and data, refining the system boundaries, and validating the data. Allocation consists of assigning a fraction of inputs and by-products to main product and co-product based on some rule.
Impact assessment has the steps of classification, characterization and valuation.
Classification consists of assigning the inventory input and output data to potential environmental impacts. Characterization consists of combining different stressor-impact relationships into a common framework. Valuation consists of assigning weighting factors to the different impact categories. Improvement has the steps of interpretation and
58 prevention activities. Interpretation consists of identifying the ecological weaknesses and potential improvements. Prevention activities consist of analyzing the improved situation.
Phase Step Description Goal definition To define goals of the analysis Goal Definition Scoping To set up the system boundaries and and Scoping functional unit Recording To collect information and data, refine the Inventory system boundaries, and validate the data Analysis Allocation To allocate inputs and by-products to main product and co-products Classification To assign the inventory input and output data to potential environmental impacts Impact Characterization To combine different stressor-impact Assessment relationships into a common framework Valuation To assign weighting factors to the different impact categories Interpretation To identify the ecological weaknesses and Improvement potential improvements Preventive Actions To analyze the improved situation
Table 2.4: General methodology of Life Cycle Assessment.
59 2.4.1.2 Major Issues and Shortcomings
Shortcomings of LCA have been motivation of many discussions and publications. Burgess and Brennan (2001) offer a concise and complete review of these problems. Many of these shortcomings are associated not only with LCA, but with any approach that expands the boundaries of the system under study to include indirect effects. This emerging kind of holistic methodologies takes in consideration the entire system to keep its properties of a whole. The potential of such holistic methods is that they focus on the impact created by the system as a whole and not by each of its components. When the system is broken down in parts, each subsystem is restricted to its own set of variables or parameters. The sum of all these subsystems will not necessarily capture the overall behavior of the larger system. Other examples of holistic concepts and approaches, besides LCA, are Industrial Ecology (Graedel and Allenby, 1995), Triple
Bottom Line (Elkington, 1999) and Ecosystem Ecology (Jørgensen, 2002). Jørgensen et al. (1992) have addressed that irreducible complex systems such as the ecosystem, require holistic approaches.
As disciplines advance toward a region where subsystems cannot be studied in isolation from one another, science itself is forced to abandon its traditional mechanistic reductionism and evolve to holistic approaches. Life Cycle thinking is a consequence of this evolution. In the same way, many of the problems that now face LCA are characteristic of this concept of a whole. It is important to identify them because the progress of methods for environmentally conscious decision making depends on solving these problems.
60 Setting the boundaries of the system can be a problem. Ideally, all units involved directly or indirectly in the production chain should be included. However, including more units in the system involves collecting more data, spending more money and increasing the complexity of the system. Besides, often some units will play a less important role than others and therefore they might be excluded without affecting the results. It might also happen that the data is just not available forcing the exclusion of some units. On the other side, excluding too many processes oversimplify the study and underestimate the results. Sensitivity analyses can help to determine whether a unit has to be included or could be excluded. If the analysis attempts to be used for design purposes, the units included should be those that possess properties that can be manipulated or at least influenced by the analyst. Otherwise, the unit would add complexity but no degree of freedom and that is not very practical.
Allocation has been one of the most discussed difficulties in LCA. Allocation may be defined as the act of proportionally distributing the responsibility for inputs used (i.e. resource consumption) and undesired outputs (i.e. emissions and waste streams) of processes in the life cycle (Stromberg et al., 1997). Allocation is a consequence of breaking down a network in subsystems. Deciding allocation becomes critical when two systems with strong interaction are studied. Then, the rules of allocation chosen for the
LCA of one subsystem can strongly affect the results of the other. Physical parameters have generally been discredited for not being able to represent the economic reality
(Burgess and Brennan, 2001; Weidema, 2001). According to Stromberg et al. (1997) and
Huppes (1994), economic value of products should be used as a basis for allocation
61 because they justify the existence of the industrial activity. Lee et al. (1995) point out that the major difficulty in assigning monetary values to environmental costs is that it is difficult to place causality on environmental effects. In general, there is no agreement on which allocation method to use. Guinee et al. (1993a) have proposed to apply sensitivity analyses to all significant allocation methods in future case studies.
Obtaining quantitative data is very often limited by its availability. The performance of the LCA will strongly depend on the quality of the data. Data that is too old, too sparse, too averaged may not be trustworthy. The costs of collecting data can increase at a level where it is not feasible to run a LCA. It is sometimes possible to reduce these costs by using general publicly available databases. To get some good quality data requires working in collaboration with the suppliers, distributors, etc.
Another situation is that the data obtained does not include some emissions or streams that are considered unimportant. Therefore, collecting such data often leads to inconsistencies like disappearance or creation of mass and energy. In such cases, LCA has no utility if physical data is wrong with respect to critical pollutants (Ayres, 1995).
Ayres (1995) argues that most of the recent literature focuses on developing or finding an acceptable way to model environmental impact, i.e. to select, evaluate and compare different categories. Seldom one alternative is clearly preferable than others; they just vary from one category to another. Heiskanen (2000) points out that LCA’s results may confuse rather than enlighten the managers and therefore could make decision-making harder. Hofstetter et al. (2000), on the other side, have presented a way to group all possible environmental impacts into a few categories, namely damage to
62 human health, ecosystem quality and resources. Moreover they offer three different perspectives that allow the analysts to evaluate those damages according to their priorities. These are Hierarchist, egalitarian and individualist.
Schaltegger (1997) argues that the global aggregation of interventions with local impact at different places provides very questionable results and may lead to wrong decisions. Later in his paper, he suggests a site-specific LCA for local impacts.
Schaltegger (1997) argues that from an economic point of view today’s LCA provides a small potential benefit given the high probability of potentially wrong decisions (because they are based on background inventory, unrepresentative, low quality and aggregated data) and high costs. Moreover, Heinskaken (2000) questions whether LCA’s results may be used to alleviate the pressure by spreading the impact to share it with the broader system, instead of creating a sense of responsibility. Other point of criticism in LCA is that the methodology makes the user think that it could influence environmental aspects outside their own organization, when in reality, the range of influence or decision-making potential is limited to the physical constraints of its organization (Heinskaken, 2000).
In general, the use of LCA as a decision-making tool is questionable given the facts that it can rarely point to the best technological choice and does not consider economic aspects. Moreover, LCA does not offer a compatible way to assist traditional cost-benefit analysis for decision-making. Huppes (1994) argues that the “main option for expanding the domain of LCA seems to be in the combined analysis of environmental effects and costs.”
63 2.4.2 Industrial Cumulative Exergy Consumption
Traditional or Industrial Cumulative Exergy Consumption (ICEC) is an extension of exergy analysis. It accounts for the exergy of all the natural resources used directly or indirectly to make a product (Szargut et al., 1988). Figure 2.6 depicts an ICEC analysis.
A stream is considered to be a natural resource if it is a direct product from ecological processes and a raw material for human activities, for example, coal, iron and fresh water.
Industrial Cumulative Exergy Consumption (ICEC) of a process is the sum of the exergy of all the natural resources consumed in all steps of the process and previous processes in the production chain. In general, ICEC of the production chain, Cp, is
Ni C p = Cn = ∑Cn,k (2.50) k =1
where, Ni denotes the number of process units included in the industrial production chain.
Cn,k and Cp,k are respectively the cumulative exergy of the natural resource entering and of the product leaving the k-th process unit. To apply the input-output network algebra developed in Chapter 4, each unit in the network is considered to have only one external input and output. These “final demand” and “value added” streams represent the sum of the exergy of all natural resources entering or final products leaving a single unit. ICEC analysis considers exergy and cumulative exergy of natural resource inputs to be equal, that is,
Cn,k = Bn,k (2.51)
64 Industrial Natural Processes, Products, Resources, Γ Bp,k, Cp,k Bn,k, Cn,k i
Figure 2.6: Industrial Cumulative Exergy Consumption (ICEC) analysis.
Industrial Cumulative Degree of Perfection (ICDP), η , is the ratio of the exergy of the final product(s) to the ICEC of the product(s). This is
Ni ∑ B p,k k =1 B p B p,k η p = = ; η p,k = (2.52) Ni C p C p,k ∑Cn,k k =1
where, η p and η p,k represent the ICDP of the production chain, and the k-th product, respectively. Equation 2.52 may also be written as,
−1 Cp = ηp ⋅ Bp (2.53)
where, Bp is the vector of product exergies, Bp,k, ηp is the N i × N i diagonal matrix with
ηp,k forming the diagonal terms, and Cp is the vector of product CEC, Cp,k. The approach for computing CEC of each product stream is discussed in detail in Chapter 4. In general, 65 the relationship between CEC for each product, Cp,k, and CEC of the inputs, Cn,k, may be written as,
Cp = Γi ⋅Cn (2.54)
where, Cn is the vector of input CEC, Cn,k, and Γi is the N i × N i allocation matrix. This matrix represents the exergy flow network and the selected allocation method. More details about allocation are in Chapter 4.
ICEC analysis shares some features of LCA since both methods consider to some extent, the life cycle of the product. Unlike LCA, ICEC analysis ignores emissions and their impact. ICEC analysis has been used widely and calculations for many industrial processes are available (Szargut et al., 1988).
2.4.3 Thermoeconomics
Thermoeconomics combines exergy analysis with economic cost analysis (Bejan et al., 1996). Nevertheless, although exergy analysis is incorporated, the thermoeconomic objective function remains purely economic as it is the total cost of the process Zp, this is
Z p = Z F ()ε ()X + Z K ()X + Z OM (X) (2.55) where X represents the vector of decision variables. These variables can be operating conditions, equipment specification, etc. ZF, ZK and ZOM are the monetary costs of fuel,
Capital, and Operation and Maintenance, respectively. The cost of fuel, which includes the cost of raw material and utilities, can be formulated as a function of the exergetic
66 efficiency. Thermoeconomic analysis was developed on the premise that it is vastly and unnecessarily challenging to find the optimal design—that of minimal cost—for a process. Thermoeconomics provides techniques to progressively improve a current design by using cost- and exergy-based metrics. In a way, it enhances the optimization iterative procedure by improving the utilization of resources in the process via increasing exergetic efficiency. The problem with Thermoeconomics is that it presupposes that minimal cost is the best design, which it may not be true from a sustainability viewpoint.
2.4.4 Thermodynamic Input-Output Analysis
Thermodynamic Input-Output Analysis recognizes the network structure of the integrated Economic-Ecological-Social (EES) system shown in Figure 2.7. Such a system is an open thermodynamic system with energy inputs from the three fundamental sources of energy, namely sunlight, geothermal heat and tidal or gravitation forces. The fourth fundamental source, namely nuclear energy, has not been considered as it does not appear naturally in ecosystems. In addition, internal energy storages such as petroleum reservoirs, coal stocks and metallic and non-metallic mineral deposits have been considered in the proposed approach. Material may also enter the EES system in the form of national imports and exit in the form of national exports. Imports and exports, however, have not been considered in this analysis as their inclusion would require knowledge about global economy that is beyond the scope of this work.
Thermodynamic Input-Output Analysis focuses on the economic system which is divided into smaller functional units called industry sectors. In the U.S., this task is
67 accomplished by the Bureau of Economic Analysis that defines industry sectors according to Standard Industrial Classification (SIC) or North American Industrial
Classification System (NAICS) codes. Ecological system, on the other hand, is divided into four conceptual ecospheres that encompass land (lithosphere), water (hydrosphere), air (atmosphere) and living flora and fauna (biosphere). Such classification assists categorization of vast number of ecological resources into smaller groups that are easier to work with, and is not critical to the applicability of TIOA. Any other user-defined classification scheme would also work as long as renewable and non-renewable resources are distinguished.
Figure 2.7 also shows interactions between economic, ecological and social systems. Interactions represented by solid lines arise on account of resource consumption and emissions, whereas those represented by dotted lines are intangible interactions indicating impact of emissions on human and ecosystem health. For instance, the dotted arrow between the economy and the ecosystems represents ecological services required for dissipating industrial emissions and their impact on ecosystem health. The solid arrow from the ecosystems to the economy, on the contrary, represents tangible interactions that include consumption of ecological resources as raw materials by the economic activity.
68
Sun Geothermal Tidal Energy Energy
Natural Resources as Ecosystems raw materials Economy
Litho. Atmos. Ecosystem services for dissipation, ecosyst em impact
Bio. Hydro.
Consumpt ion of Impact of emission Ecosystem Emissions on human health impact due to (CO2 in natural resources anthropogenic respirat io n) (O2 in air) emissions Final Demand Value Human Resources Added
Figure 2.7: Integrated economic-ecological-human resource system (solid lines represent tangible interactions and dotted lines represent intangible interactions occurring as a consequence of emissions) (Ukidwe and Bakshi, 2004).
The network structure of the economic system and monetary interactions between industry sectors are typically well-known. They are also the primary subjects of analysis in economic input-output literature (Miller and Blair, 1985; Leontief, 1986). Conversely, the network structure of ecological system need not be completely known as the underlying Ecological Cumulative Exergy Consumption (ECEC) analysis—as it will be explained in Chapter 4—can deal with partially-known ecological networks using
69 appropriate allocation rules. ECEC analysis also provides a common unit to compare economic and ecological resources, as any system, economic or ecological can be considered as a single network of energy flows (Odum, 1996). The emphasis of this paper is not on predicting how a complex, holarchic and chaotic system such as the EES system would evolve under the influence of external energy sources (Kay and Reiger, 2000), but to analyze available resource consumption and emissions data to understand how different industry sectors rely on ecosystems for their operations. In other words,
Thermodynamic Input-Output Analysis does not attempt to forecast emergent, non-linear, non-equilibrium and self-organizing properties of the EES system, but assumes that these properties are manifested in the measured material and energy flows. The algorithm of
Thermodynamic Input-Output Analysis can be summarized in the form of following three tasks.
1) Identify and quantify ecological and human resource inputs to the economic
system. Ecological inputs include ecosystem products such as crude oil, metallic
and non-metallic minerals and atmospheric nitrogen, and ecosystem services such
as wind and fertile soil. Human resources include employment of labor for
economic activities. Emissions and their impact on human and ecosystem health
are also included.
2) Calculate ECEC of ecological inputs using transformity values from systems
ecology. These inputs are classified as additive or non-additive to be consistent
with the network algebra rules used in emergy analysis (Odum, 1996). In general,
70 non-renewable resources are additive, while renewable resources are non-
additive.
3) Allocate direct inputs to economic sectors using input-output data and the
network algebra of ECEC analysis (Hau and Bakshi, 2004a). The network algebra
of ECEC analysis is based on a static input-output representation of the economic
system. Dynamic versions of input-output analysis that consider temporal changes
in the economic network are also available, and are currently being explored.
Also, use of monetary data for allocation is not a limitation of the approach, but is
rather caused by lack of comprehensive material or energy accounts of inter-
industry interactions.
2.4.5 Emergy Analysis
Emergy, specifically Solar Emergy, is “the available solar energy used up directly and indirectly to make a service or product” (Odum, 1996). Emergy analysis considers all systems to be networks of energy flow and determines the emergy value of the streams and systems involved. “Emergy, spelled with an "m", is a universal measure of real wealth of the work of nature and society made on a common basis” (Odum et al., 2000).
Emergy analysis presents an energetic basis for quantification or valuation of ecosystems goods and services. Valuation methods in environmental and ecological economics estimate the value of ecosystem inputs in terms that have been defined narrowly and anthropocentrically, while emergy tries to capture the ecocentric value. It attempts to assign the “correct” value to ecological and economic products and services based on a
71 theory of energy flow in systems ecology and its relation to systems survival. A fundamental principle of emergy analysis is the Maximum Empower Principle. It states that “systems that will prevail in competition with others, develop the most useful work with inflowing emergy sources by reinforcing productive processes and overcoming limitations through system organization” (Brown and Herendeen, 1996). Odum (1996) states that this principle determines which systems, ecological and economic, will survive over time and hence contribute to future systems.
2.4.5.1 Historical Background
Emergy analysis is a part of a much larger theory developed by H.T. Odum about the functioning of ecological and other systems. This theory explains how systems survive and organize in hierarchies by using energy at the efficiency that generates the most power (Odum and Odum, 1981). As pointed out by Hall (1995b), the hypothesis about the role of energy in survival and evolution of systems “has roots in the 19th century and was first stated explicitly by the biologist Alfred Lotka, who called the maximum power principle the fourth law of thermodynamics (1922).” Then he adds that,
“Odum has both used and expanded the maximum power concept as a general systems hypothesis throughout his career.”
H.T. Odum started developing the roots of the concept of Emergy probably in the
1950s when he and E.P. Odum identified the importance of energetics to ecology (Odum,
1953). The brothers subsequently realized the importance of the quality of energy and the necessity of using a “common denominator for energy flows of different kinds” (Hall,
72 1995b). From this concept, H.T. Odum extended the original concept as the maximum empower principle, and developed an energy systems language for the thermodynamics of open systems. Over the years, emergy became the dominant concept of this work. By the late 1970’s, during the energy crisis, and as humankind became more aware of the negative impact of industrial activities on ecosystems, H.T. Odum had already recognized the critical role that ecosystems play in the global economy, and that economic activities were shaped not only by economic rules, but also by ecosystem constraints. He also developed the concept that energy offered a common ground for integrating economic and ecosystems sciences.
2.4.5.2 Basic Principles
Emergy is measured in solar embodied joules, abbreviated sej. Emergy analysis characterizes all products and services in equivalents of solar energy, that is, how much energy would be needed to do a particular task if solar radiation were the only input. It considers the Earth to be a closed system with solar energy, deep earth heat and tidal energy as major constant energy inputs and that all living systems sustain one another by participating in a network of energy flow by converting lower quality energy into both higher quality energy and degraded heat energy. Since solar energy is the main energy input to the Earth, all other energies are scaled to solar equivalents to give common units.
Other kinds of energy existing on the Earth can be derived from these three main sources, through energy transformations. Even the economy can be incorporated to this energy flow network as, “wealth directly and indirectly comes from environmental resources
73 measured by emergy” (Odum, 1996). Examples are elevated and purified water, timber and oil. Therefore, the circulation of money is related to the flow of emergy.
An important concept in emergy analysis is Solar Transformity, defined as “the solar emergy required to make one Joule of a service or product” (Odum, 1996). Solar transformity is measured in sej/J. The solar transformity of a product is its solar emergy divided by its available energy, that is,
M =τ B (2.56) where M is emergy, τ is transformity and B is available energy. Since solar energy is the baseline of all emergy calculations, transformity of solar energy is unity.
Odum argues that the “energy flows of the universe are organized in an energy transformation hierarchy” and that “the position in the energy hierarchy is measured with transformities” (Odum, 1996). Therefore, transformity is regarded as a measure of energy quality. From a practical point of view, transformity is useful as a convenient way of determining the emergy of commonly used resources and commodities. Most case studies in the literature rely on the transformities calculated by Odum and co-workers to calculate the emergy of their inputs.
Most transformities are calculated from the yearly emergy flow to the Earth
(Odum, 2000). The total emergy input to the Earth is the sum of the emergy of solar insolation, deep earth heat and tidal energy. However, even these inputs are not added directly due to their different abilities to do work. The emergy of deep earth heat and tidal energy are calculated by comparing their energy quality to that of solar insolation. The
74 detailed calculations are based on energy balance equations for the earth, and are described by Odum (2000). These global emergy inputs are the driving force for all planetary activities. Determining their contribution to ecological goods and services is essential for further analysis.
Odum and coworkers have determined the emergy of the earth’s main processes such as, the total surface wind, rain water in streams, Earth sedimentary cycle, and waves absorbed on shore, to be that of the total emergy input to the Earth (Odum, 1996). Each of these processes is assigned the total value because they are considered co-products of the global geological cycle and cannot be produced independently with less amount of the total emergy. Furthermore, detailed knowledge about the underlying network and all the outputs from these earth processes is usually not available.
In the case of the Earth sedimentary cycle, Odum (1996) calculates the emergy per gram of sediment by estimating that a layer of nearly an inch of thickness of soil is removed from the continental land by erosion and replaced by earth uplift in a period of a thousand years. The flux of sediments is calculated by taking the product of the annual replaced layer and the average density of rocks. The emergy per gram of sediment is the global emergy budget divided by the flux of sediments. In this case, emergy is allocated according to the mass of sediments.
Conceptually, determining the emergy of non-renewable resources such as coal and petroleum requires accounting for solar inputs over geological time scales. Odum suggests using the replacement time of such material to estimate their historic emergy.
However, the transformity of non-renewable fuels used in most applications focuses only
75 on the current emergy from the sedimentary cycle. For example, the transformity of coal is the emergy per gram of sediment from the Earth sedimentary cycle divided by the
Gibbs free energy of a gram of coal (Odum, 1996). The transformity of other fuels are approximated based on their relative efficiency obtained in combustion chambers.
The emergy of economic inputs measured in terms of money is determined by multiplying the input in monetary units by the ratio of the nation’s total emergy to its economic gross national product, i.e.
⎛ M nation ⎞ M = Z⎜ ⎟ (2.57) ⎝ Z nation ⎠
where Z represents a particular economic input, Mnation is the total nation’s emergy and
Znation is the gross national economic product.
2.4.5.3 Emergy Algebra
Determining the emergy flow in a network requires techniques for distributing or allocating it to the nodes and edges. Figure 2.8 illustrates how emergy flows through energy branches in a systems network (Odum, 1996). Emergy is split according to the available energy of the outputs when a stream is divided without any energy transformation. In this case, transformity does not change its value. If there is an energy transformation, emergy is not distributed, but assigned entirely to each stream. This is justified based on the fact that they are co-products, meaning that none of the streams can be produced independently from the other. In this case transformity does change its value. 76 Split of a) exergy of one Emergy Transformity kind (J/t) (sej/t) (sej/J) 40 400 10 100 1000 10 60 600 10 Exergy of b) different Emergy Transformity kinds (J/t) (sej/t) (sej/J) 1000 250 100 4 1000 10 6 1000 167
Figure 2.8: Emergy flow through network branches (Odum, 1996).
The situation shown in Figure 1b does not conserve emergy through the network.
This adds complexity to the algebra since extra care is necessary to avoid double counting when emergy is combined. Figure 2.9 shows how emergy is determined for stream intersections (Odum, 1996). Emergy is added when available energy of the same kind is joined. Emergy is also added if available energy of different kinds and sources are joined. However, if available energy of different kinds but same sources are joined, then the higher value of emergy is assigned.
Figure 2.10 illustrates how emergy is determined in stream intersections with feedback loops. Emergy of loops are not accounted for in calculations to avoid double counting (Brown and Herendeen, 1996).
77 JointSplit of exergy of one Emergy Transformity a) kind (J/t) (sej/t) (sej/J) 40 400 10 100 1000 10 60 600 10
Exergy of different Emergy Transformity b) kinds (J/t) (sej/t) (sej/J) 3 1000 333 10 2000 200 97 1000 10
Separate sources 1000 333 1000 100 Same source 1000 10
Figure 2.9: Emergy flow with network joints (Odum, 1996).
240 sej/yr
400 400 160 sej/yr sej/yr sej/yr
Figure 2.10: Emergy Flow with feedback.
78 Based on this algebra, Odum and coworkers have determined the emergy of the earth main processes, such as the total surface wind, rain water streams, Earth sedimentary cycle and waves absorbed on shore, is that of the total emergy input to the
Earth (Odum, 1996). Each of these processes is assigned the total value because they are considered co-products of the global geological cycle and cannot be produced independently with less amount of the total emergy.
2.4.5.4 Emergy Analysis of the main Earth Processes
This section is adapted from Odum’s book and recent Emergy folios (Odum,
1996; Odum, 2000), and should be referred for detailed information. Figure 2.11 shows a diagram of the main energy flows on the Earth. The total emergy input to the Earth, often called global emergy budget, is the sum of the emergy of solar insolation, crustal heat and tidal energy. Since they are independent, they are added. Other energy sources are assumed to be generated from these three sources, e.g. hemisphere general circulation, ocean circulation, global sedimentary cycle, etc. Most of them are considered co-products and therefore, their emergy equals the global emergy budget.
Solar insolation is the solar energy that reaches the Earth by radiation. Its exergy,
Bsi, has been calculated by multiplying the solar constant (2 Langley per minute or
4.4x1010 J/m2-yr), the fraction of solar irradiation absorbed by the Earth (70 %) and the
Earth cross section facing the Sun ( 1.27x1014 m2)
⎛ 10 J ⎞ 14 2 24 J Bsi = ⎜4.4 ×10 2 ⎟()0.7 (1.27 ×10 m )= 3.93×10 ⎝ m ⋅ yr ⎠ yr
79 Solar insolation 3.93x1024 J/yr Atmosphere
Tidal energy Hydrosphere 5.2x1019 J/yr Litosphere
Crustal heat Sources from above 6.72x1020 J/yr 20 6.49x10 J/yr Radioactivity generation Deep heat 1.98x1020 J/yr 4.74x1020 J/yr Mantle Crust
Figure 2.11: Main energy flows on Earth.
As already mentioned, transformity of solar insolation, τsi, is unity by definition.
Thus, emergy of solar insolation is 3.93x1024 sej/yr.
The energy of crustal heat has been calculated by adding the total heat released by crustal radioactivity (1.98x1020 J/yr) and heat flowing up from mantle (4.74x1020 J/yr).
Thus, the available energy of crustal heat is 6.72x1020 J/yr. The total heat outflow from the surface is 13.21x1020 J/yr. The additional contribution of total heat outflow (6.49x1020
J/yr) is contributed from the above sources, i.e. solar insolation and tidal energy, by
80 passing energy downward as compression and chemical potentials. In Figure 2.11, the heat from the above sources is of the same quality than the heat from radioactivity generation and deep heat; therefore they have all the same transformity, τch. An emergy balance equation gives
24 19 20 (3.93 × 10 J / yr)⋅ ()1sej / J + (5.2 × 10 J / yr)⋅τ te = (6.49 × 10 J / yr)⋅τ ch
(2.58)
20 where τte is the transformity of tidal energy. The oceanic geopotential energy (2.14x10
J/yr) is contributed from solar insolation, crustal heat and tidal energy. Oceanic geopotential energy is of the same quality than tidal energy, therefore they have the same transformity. An emergy balance equation gives
(3.93 × 10 24 J / yr)⋅ ()1sej / J + (6.72 × 10 20 J / yr)⋅τ + (5.2 × 1019 J / yr)⋅τ = ch te 20 ()2.14 × 10 J / yr ⋅τ te
(2.59)
Transformity of crustal heat and tidal energy are obtained by simultaneously solving equations 2.58 and 2.59. The results are
τ ch = 11,981sej / J and τ te = 73,923sej / J
Emergy of crustal heat and tidal energy are obtained from eq 6, i.e.
20 24 M ch = (6.72 ×10 J / yr)⋅ ()11,981sej / J = 8.06 ×10 sej / yr and
81 19 24 M te = (5.2 ×10 J / yr)⋅ ()73,923sej / J = 3.83×10 sej / yr
These three main sources add the global emergy budget. Table 2.5 shows a summary of all the values calculated above for the global emergy budget. These values and their respective calculation were taken from the emergy folios (Odum, 2000). The energy sources generated from these three main inputs to the Earth are classified in three main process systems, named atmospheric circulation, ocean processes and Earth processes. They are all co-products and therefore, their emergy equals the global emergy budget.
The rest of this section is adapted from Odum’s book (1996), and therefore, it contains outdated values. In the case of the Earth sedimentary cycle, the emergy per gram of sediment is calculated. An estimate of a layer of 2.4 cm of thickness of soil is removed from the continental land by erosion and replaced by earth uplift in a lapse of 1000 years.
By considering an average density of rocks of 2.6 g/cm3 and a continental area of
1.5x1014 m2, the flux of sediments is
()2.4cm ⋅ 2.6 g / cm3 ⋅ 1.5 × 1018 cm2 ( ) ( ) = 9.36 × 1015 g / yr 1000 yr
The emergy per gram of sediment is the global emergy budget divided by the flux of sediments,
9.44 × 10 24 sej / yr = 1.01× 10 9 sej / g 9.36 × 1015 g / yr
82 Exergy (J/yr) Transformity (sej/J) Emergy (sej/yr) Solar insolation 3.93x1024 1 3.93x1024 Crustal heat 6.72x1020 11,981 8.06x1024 Tidal energy 0.52x1020 73,923 3.83x1024 Global budget 15.83x1024
Table 2.5: Global emergy budget of the Earth.
Notice that the value of the global emergy budget is 9.44x1024 sej/yr, which is the old value from Odum’s book. As determined before, the updated value of the global emergy budget is 15.83x1024 sej/y.
Conceptually, determining the emergy of non-renewable resources such as coal and petroleum would require accounting for solar inputs over geological time scales. This can be problematic since it is difficult, if not impossible to know the inputs and processes over such a long time. That is not the approach taken for non-renewable fuels (Odum,
1996). For example, coal is considered a material in the earth crust that becomes accessible for humans through the earth sedimentary cycle. Then the emergy of coal may be approximated based on the earth sedimentary cycle. The Gibbs free energy of coal is
29,302 J/g. Then the transformity of coal is,
1.01× 10 9 sej / g τ = = 3.4 × 10 4 sej / J coal 29,302J / g coal
83 To calculate the transformity of other fuels, an equivalence factor between fuels is used. Considering that the work generated from 1.65 J of coal is equivalent to that generated from 1 J of liquid motor fuel, the transformity of liquid motor fuel can be calculated as
4 ⎛ J coal ⎞ τ motor _ fuel = ()3.4 × 10 sej / J coal ⋅ ⎜1.65 ⎟ ⎝ J motor _ fuel ⎠ 4 = 5.6 × 10 sej / J motor _ fuel
Considering that 19% of crude oil is used in refining and transport, which means that the remaining 81% is used as liquid motor fuel, the transformity of crude oil can be calculated as
4 ⎛ J motor _ fuel ⎞ τ crude _ oil = ()5.6 × 10 sej / J motor _ fuel ⋅ ⎜.81 ⎟ ⎝ J crude _ oil ⎠ 4 = 4.5 × 10 sej / J crude _ oil
Considering that natural gas is 20% more efficient in boilers than coal, the transformity of natural gas can be calculated as
4 ⎛ J coal ⎞ τ natural _ gas = ()3.4 × 10 sej / J coal ⋅ ⎜1.20 ⎟ ⎝ J naural _ gas ⎠ 4 = 4.1× 10 sej / J natural _ gas
84 2.4.5.5 Metrics
The manufacture of a product P typically requires inputs of renewable, non- renewable and economic resources denoted RR, NR and F, respectively. Emergy analysis has metrics similar to those used in economic analysis to evaluate the performance of industrial activity. Since money is an incomplete measure of ecological goods and services, economic indicators underestimate the real contribution of ecological activity.
Emergy metrics intend to overcome flaws of economic indicators.
Net Emergy, Mnet, is the emergy gained by the economy in exchange for providing its services. It is analogous to the economic profit from a process.
M net = M P − M F (2.60)
The Emergy Yield Ratio of products is “the ratio of the yield output emergy flow to the sum of the feedback emergy from the economy” (Odum, 1996).
M EYR = P (2.61) M F
EYR is analogous to the economic return on investment. Emergy metrics have also been defined to evaluate the performance of industrial activity (Brown and Ulgiati,
1997). The Environmental Loading Ratio is defined as the sum of the feedback emergy from the economy and emergy from non-renewable resources divided by the emergy from renewable resources. This index is an indicator of the stress on the local environment. This is
85 ()M + M ELR = F NR (2.62) M RR
Brown and Ulgiati (1997) have modified this metric to consider the ecosystem services needed to dissipate the emissions. However, it ignores the actual impact due to emissions such as dead trees due to acid rain (Bakshi, 2002).
The Sustainability Index is defined as the ratio of the emergy yield ratio to the environmental loading ratio. This index evaluates the ecological-economic-integrated performance of the activity.
EYR SI = (2.63) ELR
86
CHAPTER 3
PROMISE AND PROBLEMS OF EMERGY ANALYSIS
Beliefs are what divide people. Doubt unites them.
–Peter Ustinov
Emergy analysis is perhaps H.T. Odum’s masterpiece. It is “the most exciting, important, far-out, comprehensive, crazy, unsubstantiated, and/or necessary…” of all of the ideas of H.T. Odum (Hall, 1995a). Emergy analysis presents an energetic basis for quantification or valuation of ecosystems goods and services. Valuation methods in environmental and ecological economics estimate the value of ecosystem inputs in terms that have been defined narrowly and anthropocentrically, while emergy tries to capture the ecocentric value. It attempts to assign the “correct” value to ecological and economic products and services based on a theory of energy flow in systems ecology and its relation to systems survival. A fundamental principle of emergy analysis is the Maximum
Empower Principle. It states that “systems that will prevail in competition with others, develop the most useful work with inflowing emergy sources by reinforcing productive processes and overcoming limitations through system organization” (Brown and
87 Herendeen, 1996). Odum (1996) states that this principle determines which systems, ecological and economic, will survive over time and hence contribute to future systems.
Since the early 1980s, emergy and emergy analysis have been used widely to analyze systems as diverse as ecological, industrial, economic, and astronomical (Odum,
1995a,b, 1996; Brown and Ulgiati, 1997, 2002; Lagerberg and Brown, 1999).
Unfortunately, like many groundbreaking ideas, emergy has encountered a lot of resistance and criticism, particularly from economists, physicists, and engineers.
Consequently, it has not been used much outside a small circle of researchers. This limited use of emergy analysis despite its broad relevance may be due to inadequate attention to details, poor communication of its potential importance, and lack of clear links with related concepts in other disciplines. The publication of Odum’s “how to” book (Odum, 1996) and the more recent emergy folios (Odum et al., 2000; Odum, 2000;
Brown and Bardi, 2001; Brandt-Williams, 2001) are important steps in making emergy more accessible. However, much more work is needed to connect emergy with concepts in other disciplines and to overcome a preconceived negative notion of emergy that is prevalent among many researchers outside of systems ecology.
3.1 Attractive Features of Emergy Analysis
Emergy analysis overcomes the inability of many existing approaches to adequately consider the contribution of ecological processes to human progress and wealth. A large range of ecological products and services do not receive any value from conventional economic approaches despite the fact that they are used and spent for the
88 making of economically valuable products, or indeed may be essential for life. The importance of accounting for nature’s services is gaining wide acceptance (Daily, 1997;
Holliday et al., 2002; Arrow et al., 1995), although the methods remain controversial.
Through the last two decades, economists have developed techniques to assign monetary values to ecological products and services. However, this assignment typically relies on consensus of boards of experts, often with tenuous physical and biological foundations, and generally scaled to some market-derived values that may be, for example, highly skewed by advertising. In contrast, emergy analysis is meant to be independent of human valuation, but based on the principles of thermodynamics, system theory, systems ecology and, ultimately contribution to survival. Among the most attractive characteristics of emergy analysis are:
• It provides a bridge that connects economic and ecological systems. Since emergy
can be quantified for any system, their economic and ecological aspects can be
compared on an objective basis that is independent of their monetary perception.
• It compensates for the inability of money to value non-market inputs in an
objective manner. Therefore, emergy analysis provides an ecocentric valuation
method.
• It is scientifically sound and shares the rigor of thermodynamic methods.
• Its common unit allows all resources to be compared on a fair basis. Emergy
analysis recognizes the different qualities of energy or abilities to do work. For
89 example, emergy reflects the fact that electricity is energy of higher quality than
solar insolation.
• Emergy analysis provides a more holistic alternative to many existing methods for
environmentally conscious decision making. Most existing methods, such as life
cycle assessment and exergy analysis, do expand the system boundary beyond the
scope of a single process so that indirect effects of raw material consumption,
energy use and pollutant emissions can be taken into account. However, these
methods focus more on emissions and their impact, while ignoring the crucial
contribution of ecosystems to human well being. The concept of critical natural
capital and a framework to account for have been suggested recently (Ekins et al.,
2003). Emergy analysis can quantify the contribution of natural capital for
sustaining economic activity (Bakshi, 2002).
These features of emergy analysis are particularly impressive since emergy was developed many decades before the more recent engineering and corporate interest in life cycle assessment, industrial ecology, and sustainability. Partly due to being a theoretical concept whose application posed significant demands on data requirements, lack of adequate details about the underlying methodology, and sweeping generalizations that still remain unproven, emergy has encountered a lot of criticism, and has not been used much outside a small circle of researchers. However, there is no doubt that as an idea, it was truly revolutionary and is expected to have a huge impact.
90 3.2 Criticisms
Emergy theory has been characterized as simplistic, contradictory, misleading, and inaccurate (Ayres, 2000; Cleveland et al., 2000; Mansson and McGlade, 1993;
Spreng, 1988). Rebuttals to many critiques have also been published (Patten, 1993;
Odum, 1995a,b). However, much of the persistent skepticism seems to stem from the difficulty in obtaining details about the underlying computations, and a lack of formal links with related concepts in other disciplines. Odum’s book (Odum, 1996), emergy folios (Odum et al., 2000; Odum, 2000; Brown and Bardi, 2001; Brandt-Williams, 2001), and plans for an emergy handbook are important and essential steps to provide greater insight and understanding about emergy.
The major criticisms of emergy analysis are discussed below. It is important to note that many criticisms are also valid for other methods that are popular for joint analysis of industrial and environmental systems, including, Life Cycle Assessment,
Cumulative Exergy analysis, Exergetic Life Cycle Assessment, and Material Flow analysis.
3.2.1 Emergy and economics
Odum (1988) argues that “money cannot be used directly to measure environmental contributions to the public good, since money is paid only to people for their services, not to the environmental service generating resources or assimilating wastes. Price is often inversely related to the contribution of a resource, because it contributes most to the economy when it is easily available, requiring few services for
91 delivery.” Brown et al. (1995) also argue that price does not determine value, giving the example that “a gallon of gasoline will power a car the same distance no matter what its price; thus, its value to the driver is the number of miles (work) that can be driven.” Since emergy does consider all contributions to the public good and truly measures value, it is suggested as a complete measure of wealth and a substitute for money (Odum, 1984).
Moreover, Odum (1996) considers transformity of a product as an indicator of its economic usefulness as “transformity increases in ecological and economic energy transformation chains.”
These claims are among the most controversial aspects of emergy analysis and have been most widely criticized (Ayres, 2000; Cleveland et al., 2000; Spreng, 1988).
The emergy theory of value, as other theories of value based on energy and exergy
(Cleveland et al., 2000; Spreng, 1988), focuses on the supply side and ignores human preference and demand. Modern economics, which is focused on humans and their values and not the biophysical world, has doubted the ability of all such theories to capture the value of products to humans. Some common arguments are that the emergy of a gallon of oil from whales has not changed, while its value to humans has. In addition, two paintings with similar emergies can have drastically different values, especially if one of them is by a renowned painter. Consequently, all of the thermodynamic theories of value have been rejected by economists over the last several decades. What most critiques about emergy-based valuation seem to miss is that emergy aims to provide an ecocentric value of ecological and industrial products and processes. This is in direct contrast to the economic view, which is anthropocentric. Clearly, the latter view is dominant today, but
92 the emergy view can still provide invaluable information that can be used for combining natural and economic capital, which is necessary for identifying both weak and strong sustainability. Eventually, economic valuation will have to adopt a more ecocentric view if it intends to guide humanity to its survival.
In engineering analysis of industrial systems, thermodynamics-based methods, such as pinch analysis and exergy analysis (Seider et al., 2003; Bejan et al., 1996), are commonly used along with cost criteria. Although the final decision is based mainly on economic criteria, thermodynamic methods are crucial for constraining the search space and for directing the decisions. In the same way, emergy analysis of industrial systems may be able to coexist with economic analysis, with emergy providing the supply side information, and economics capturing human demand and values. Such approaches are very likely to direct decisions towards more environmentally conscious industrial practices and enhance the understanding of the sustainability principles.
3.2.2 Maximum Empower Principle
This optimizing principle is one of the most daring aspects of emergy analysis.
Having its roots in work done by Boltzmann (1886) and Lotka (1922), the Maximum
Empower Principle claims that all self-organizing systems tend to maximize their rate of emergy use or empower (Odum, 1988, 1996). That is, “ecosystems, Earth systems, astronomical systems, and possibly all systems are organized in hierarchies because this design maximizes useful energy processing.” Thus, this principle can determine which species or ecosystems or any system will survive.
93 While some self-organizing systems have been shown to follow this principle
(Odum, 1995b), claiming its general applicability to all systems implies that the
Maximum Empower Principle can explain the order of the universe, and is akin to the unified theory that physicists have been piecing together. Not surprisingly, such broad, as yet unsubstantiated claims have made this principle extremely controversial (Ayres,
2000). Mansson and McGlade (1993) argue that the behavior of complex systems cannot be described with a one-dimensional optimizing principle categorizing this principle as misleadingly simplistic. They also claim to have invalidated this principle. However, the validity of their proof has been questioned (Odum, 1995a; Patten, 1993).
The Maximum Empower Principle seems to be one of Odum’s contributions that may be ahead of its time. Consequently, it will continue to be a cause of arguments and further scientific exploration, until it is scientifically proven or unproven. Recent results on maximum entropy production in self-organized systems indicate that some systems do tend to maximize power (Lorenz, 2003; Dewar, 2003). In addition, Giannantoni (2003) proposes a mathematical formulation of the Maximum Empower Principle, which may be essential for addressing questions about the validity of this principle, and for providing a general proof.
For engineering applications, agreement or disagreement with the Maximum
Empower Principle is not essential for using emergy analysis for gaining insight into the contribution of ecosystems. Differences between various valuation techniques lie in the diversity of perceptions and priorities among decision makers. Regardless these differences, all these methods value goods and services based on transactions of
94 representative attributes. Emergy analysis can be thought of a valuation technique based on energy transactions. Since the Maximum Empower Principle is irrelevant to the accounting preference, the use of emergy analysis should not be affected by the validity of this principle. Thus, emergy analysis can still provide valuable information about the contribution of ecosystems to engineering design and assessment of industrial systems.
3.2.3 Relation with other thermodynamic quantities
There seems to be much confusion about the relationship between emergy and other thermodynamic properties, such as energy, exergy, enthalpy, etc. The qualitative difference, as pointed out by Odum and coworkers, is that unlike emergy, these thermodynamic quantities do not recognize the difference in quality of various energy sources. A common example is that “a joule of sunlight is not equivalent to a joule of fossil fuel…” in the sense that they cannot do the same kind of work (Brown et al., 1995).
However, formal quantitative links are missing. This leads to impressions that emergy analysis is a “very different approach” from exergy analysis (Emblemsvag and Bras,
2001). Similarly, Ayres (2000) questions the need for emergy as opposed to “standard variables of thermodynamics, namely, enthalpy and exergy.”
There is also some confusion about the exact definition of available energy, denoted B in equation 2.56. It is certainly not Gibbs free energy because not all of it is available for work. Odum (1995a,b) argues that neither is it exergy because “exergy is defined to include only energy flows of similar qualities, that of mechanical work,” while available energy as defined in emergy analysis also considers “important inflows, such as
95 human services that require very large energy flows to maintain.” On the other hand,
Odum (2000) and Campbell (2001) define available energy in emergy analysis as exergy or energy with the potential to do work. Scrutiny of transformity calculations indicates that available energy as used in emergy and exergy may indeed be equivalent. For example, for heat engines the available energy of the system is the same as exergy since it is obtained by multiplying its heat content or flow by the Carnot factor (Odum, 1996).
The relationship of the transformities of fuels to their combustion efficiencies may be easily justified if available energy and exergy are equivalent. Odum uses the heat of combustion to determine available energy, which is shown to be close to exergy for fuels
(Szargut et al., 1988). Moreover, the use of exergy justifies why dissipated heat carries no emergy value. This lack of formal links between emergy and other thermodynamics quantities is a significant cause of skepticism about emergy among engineers. Some efforts have been made to connect emergy with exergy (Ulgiati, 1999).
Improved understanding of the relationship between emergy and exergy is essential for constructive cross-fertilization between these areas. Such insight is essential for greater use of the data and concepts of emergy analysis in evaluating the life cycle of engineering products and processes. A strong link between engineering thermodynamic concepts and emergy will help proving that many criticisms of emergy, such as its connection with economic value or the Maximum Empower Principle, are not relevant to using emergy to capture the thermodynamic aspects of ecological goods and services.
More importantly, it will clear up much of the confusion regarding the relation of emergy to other thermodynamic properties.
96 3.2.4 Combining disparate time scales
Conceptually, the calculation of emergy of some stored natural resources, such as metals, coal, and fossil fuels, would require knowing all the solar energy that was required to make it. Accounting for solar inputs over geological time scales is problematic since it is difficult, if not impossible to know the inputs and processes over such a long period (Ayres, 2000; Cleveland et al., 2000). Some common questions concern how to account for the emergy of metals that existed from the formation of the
Earth and whether the emergy of fossil fuels includes the emergy of the living systems from where they are derived.
Odum does distinguish between the emergy of storage and the emergy necessary for making the storage available for human use. The emergy of stored resources, such as fresh water, glaciers on land, atmosphere, and continents, is calculated by multiplying the global emergy budget by their respective replacement or turnover time (Odum, 1996).
The emergy for concentrating natural resources in the Earth’s crust so that they are available for human and other use is determined based on the Earth sedimentary cycle. In fact, the transformities of many resources, such as coal and oil, that are used in applications of emergy analysis consider only their current emergy required to concentrate these resources in the ore, and not that stored since prehistory. Ultimately this is a matter of selecting the appropriate temporal boundary for a given problem. Decisions about temporal and spatial boundaries are necessary for most holistic approaches, including life cycle assessment (LCA) (European Environment Agency, 1997).
Therefore, this kind of criticism applies not only to emergy analysis but also to the other
97 approaches. Greater interaction between emergy analysis and LCA may be useful for clarifying this issue.
3.2.5 Representing global energy flows in solar equivalents
Emergy analysis represents all energy flows in solar equivalents. This requires conversion of planetary energy inputs, such as tidal energy and crustal heat, into solar equivalents. Ayres (2000) questions such conversion since “there is no simple way to discover how much of any one form of energy might have been needed to produce another in the distant past.” Calculation of the emergy of deep Earth heat and tidal energy inherently carries some assumptions regarding the efficiency with which they are carried to their point of application. Although not explicitly stated, this is derived from the
Maximum Empower Principle. For example, the emergy of deep Earth heat is calculated by assuming that its transformity is equal to the transformity of the heat outflow contributed by the Earth sedimentary cycle that passes through the surface of the Earth.
This assumption may be justified since the Maximum Empower Principle and evolutionary pressures may have caused both processes to operate in a similar manner to result in heat flows of identical quality.
Another way of looking at the transformity of global energy flows in solar equivalents is as algebraic coefficients based on a global energy balance of current flows.
This view does not imply that solar energy is being converted into tidal energy or deep
Earth heat. The primary benefit of this approach is that it allows a fair comparison of the concentration of different kinds of energy.
98 Such conversion factors are commonly used in many techniques. Transformity of fuels has been calculated by comparing their qualities based on their efficiencies in combustion chambers. LCA uses conversion factors to add the environmental impact of various substances, e.g. greenhouse gases are typically expressed in equivalents of carbon dioxide to quantify the impact by Global Warming. Calculation of chemical exergies requires defining reference reactions to convert substances absent in the surroundings into components of the reference state or state of the surroundings.
3.2.6 Problems of quantification
Emergy analysis has not considered the uncertainty in many of the numbers used to calculate the transformities. Averaged transformity of industrial and geological processes are frequently used in specific case studies with no knowledge of the degree of certainty of the resulting output. For example, the transformity of natural gas is calculated based on its average efficiency relative to coal in boilers (Odum, 1996), but this efficiency depends strongly on the type of coal and natural gas as well as the characteristics of the boiler. Similarly, calculating the emergy of economic inputs via the emergy to money ratio may also be inaccurate and may involve double counting (Ayres,
2000; Cleveland et al., 2000). This approach also seems to counter emergy analysis’ argument that money is an incomplete measure of wealth.
H.T. Odum does recognize that there is no single transformity for any class of products or processes and that when viewed in greater detail, each production pathway for a given product (rain, wind, waves, oil, etc.) represents a unique transformation
99 process that will result in a different transformity. Nevertheless, it has been assumed that generalized transformities do not differ significantly from any specific case.
This criticism is also shared by most other approaches. Exergy analysis uses an average reference state for the Earth’s crust without addressing the resulting uncertainty.
Similarly, LCA tends to ignore the effect of emissions in local environments and errors in inventory data. Efforts are being made in each approach to address these criticisms and more interaction and exchange of ideas between these fields can be helpful. For example, the use of the emergy-to-money ratio to account for economic inputs may be addressed via systematic methods to consider a larger boundary consisting of the main processes selected for emergy analysis (Ukidwe and Bakshi, 2003, 2004). Such techniques have been studied in other fields, including LCA (Lave et al., 1995), and may be adopted by emergy analysts. More research to verify the numbers used in emergy analysis and, their uncertainty and assumptions should also help. Techniques, such as sensitivity and uncertainty analysis, must also become an integral part of emergy analysis.
3.2.7 Problems of allocation
The method used for partitioning or allocating inputs between multiple outputs makes the emergy algebra quite challenging. Allocation is probably the most confusing aspect of emergy analysis, particularly to engineers who are used to conservation equations, even for systems with recycle. Emergy algebra can be very sensitive to the level of knowledge of the system under study. The decision of whether multiple outputs are co-products or splits may not always be obvious either. For example, the outputs of a
100 crude oil distillation column may raise arguments about whether they should be treated as co-products or splits. Similarly, the results of considering different types of rocks as co- products (Odum et al., 2000) in the Earth sedimentary cycle are quite different from considering them to be splits (Odum, 1996).
Allocation is an important practical issue encountered by many techniques, including Cost Accounting, Cumulative Exergy Consumption, and Life Cycle
Assessment. Current consensus in the LCA community is to avoid allocation as far as possible (ISO 14040, 1997). When avoiding allocation is not possible, the sensitivity of the results to the allocation procedure should be evaluated. Hau and Bakshi (2004a) have proposed a formal algorithm based on network algebra that can be used for emergy analysis. Their approach prefers allocation that conserves emergy if information about the network and all its products is available. This is usually the case for industrial systems.
However, if such information is not available, as it is usually the case for ecological systems, allocation is avoided by assigning the same emergy to all the outputs. In the latter case, double counting must be avoided.
3.3 Summary and challenges for the future
Emergy is potentially one of the most groundbreaking contributions of H.T.
Odum. It provides an ecocentric view of ecological and human activities, which can be used for evaluating and improving industrial activities. Such techniques are crucial for appreciating the contribution of ecosystems to all human activities and meeting the challenges of environmentally conscious decision making. The Maximum Empower
101 Principle attempts to explain the behavior of self-organized systems based on thermodynamics. These concepts were put forth many decades before the current interest in life cycle analysis, industrial ecology and sustainability, and before adequate data and techniques were available.
Like many new ideas, Odum’s work on emergy has been controversial and is often criticized. Much of the criticism is directed towards the details of emergy analysis, or towards the link between emergy and money and the Maximum Empower Principle.
New ideas, as profound as emergy, require much work to iron out the details before it can be widely accepted and used. Odum’s book (Odum, 1996) and recent folios (Odum et al.,
2000) are important steps in this direction. However, more needs to be done to link emergy with other thermodynamic concepts and with other related techniques. Criticisms pertaining to uncertainty, sensitivity, and quantification apply not just to emergy analysis but to all methods that focus on a holistic view of industrial activity. These include life cycle assessment, material flow analysis, and exergy analysis. Research in all these areas can greatly benefit from one another. For example, more interactions between emergy analysis and LCA may permit LCA to account for the contribution of ecosystems, and may permit emergy analysis to avoid allocation. An input–output framework has been used in many disciplines, including economics and LCA, and may provide a formal way of representing information about the ecological and industrial network.
Establishing the links between emergy analysis and other thermodynamic concepts is essential for widening the use of emergy analysis and for removing the resistance to emergy among many engineers, physicists, and economists. These efforts
102 will not diminish the unique contribution of emergy, but will allow it to become more of a mainstream concept and have even greater impact. The complementary nature of emergy with other disciplines also needs to be explored. For example, human valuation of ecological goods and services requires information about the role of ecosystems, which may be provided via emergy analysis.
Controversy surrounding the Maximum Empower Principle reflects our limited understanding of the behavior of complex systems. Detailed studies of the Maximum
Empower Principle and its connection with other concepts governing the behavior of self- organized systems are necessary. The recent work by Giannantoni (2003) may be an important step in this direction. However, as discussed in this article, the Maximum
Empower Principle need not hinder the application of emergy analysis. Ultimately, the biggest challenge facing emergy analysis may be that of overcoming the preconceived misunderstandings about emergy to accept it as a legitimate and useful thermodynamic approach. These and many other challenges emanating from Odum’s deep insight into the workings of complex systems, his creativity and limitless energy should keep researchers busy for many years to come, and will continue to play an important role in the evolution of human activity towards sustainability.
103
CHAPTER 4
ECOLOGICAL CUMULATIVE EXERGY CONSUMPTION
A cynic is one who knows the price of everything and the value of nothing.
–Oscar Wilde
Exergy analysis has been extended for life cycle assessment and environmental performance evaluation of industrial products and processes. Although these extensions recognize the importance of capital and labor inputs and environmental impact, most of them ignore the crucial role that ecosystems play in sustaining all industrial activity.
Decisions based on approaches that take nature for granted continue to cause significant deterioration in the ability of ecosystems to provide goods and services that are essential for every human activity. Accounting for nature’s contribution is also important for determining the impact and environmental performance of industrial activities. In contrast, emergy analysis, a thermodynamic method from systems ecology, does account for ecosystems, but has encountered a lot of resistance and criticism, particularly from economists, physicists and engineers. This chapter expands the engineering concept of
Industrial Cumulative Exergy Consumption (ICEC) analysis to include the contribution of ecosystems, which leads to the concept of Ecological Cumulative Exergy 104 Consumption (ECEC). Practical challenges in computing ECEC for industrial processes are identified and a formal algorithm based on network algebra is proposed. ECEC is shown to be closely related to emergy, and both concepts become equivalent if the analysis boundary, allocation method, and approach for combining global energy inputs are identical. This insight permits combination of the best features of emergy and exergy analysis, and shows that most of the controversial aspects of emergy analysis need not hinder its use for including the exergetic contribution of ecosystems. Examples illustrate the approach and highlight the potential benefits of accounting for nature’s contribution to industrial activity.
4.1 Ecological Cumulative Exergy Consumption
Ecological processes convert global exergy inputs into ecological goods and services that are converted into economic goods and services by industrial processes.
Including ecological processes requires expansion of the system boundaries of ICEC analysis. Thus, Figure 2.6 needs to be expanded by including the exergy consumption of ecological processes, as shown in Figure 4.1. The exergy and cumulative exergy of inputs that drive ecological processes are represented as, Be,k, and Ce,k, respectively.
Equation 2.51 does not hold anymore for Figure 4.1. In fact, exergy and CEC of natural resources, Bn and Cn respectively, can be related through an equation similar to equation 2.53
−1 Cn = ηn ⋅ Bn (4.1)
105
Global Ecological Natural Industrial Products, Exergy, Processes, Resources, Processes, Bp,k, Cp,k Be,k, Ce,k Γe Bn,k, Cn,k Γi
Figure 4.1: Ecological Cumulative Exergy Consumption (ECEC) Analysis
where ηn is the ()N i + N e × (N i + N e ) diagonal matrix with ηn,k forming the diagonal terms. Ne denotes the number of units included in the ecological supply chain. As mentioned in Section 2.4.2, the number of inputs and outputs is equal to the total number of units because each unit has one external input and output. Variable ηn,k represents the efficiency with which ecological processes create the natural resource entering the k-th process unit from global exergy inputs. Clearly, as indicated by equation 2.51, ICEC analysis implicitly assumes that these efficiencies are unity, consequently ignoring ecological processes. Based on Figure 4.1, the exergy consumed in ecological processes to produce the natural resources and that for converting natural resources to industrial products may be written as
Cn = Γe ⋅Ce and Cp = Γi ⋅Cn (4.2)
106 where Γe and Γi are the allocation matrices for mapping ecological inputs to natural resource outputs and natural resources to industrial products, respectively. The cumulative exergy consumption in ecological and industrial processes (ECEC) to create each product may be written as
Cp = Γ ⋅Ce (4.3) where Γ represents the overall allocation matrix for ecological and industrial processes together. Whether Γ is equal to the product of Γe and Γi or not depends on the allocation method, as elaborated in Section 4.2. Alternate equations for ECEC may also be written as follows by combining equations 4.1 and 4.2
−1 Cp = Γi ⋅ ηn ⋅ Bn (4.4)
The total ECEC for the ecological-industrial production chain in Figure 4.1 may be written as
Ni +Ne C p = Cn = Ce = ∑Ce,k (4.5) k =1
Equations 4.3, 4.4 and 4.5 indicate that determining the total ECEC, Cp, requires knowledge of Bn,k, and ηn,k, while determining the ECEC of each product, Cp,k requires the allocation matrix, Γ. Similarly, determining the CDP of ecological processes requires the allocation matrix, Γe and the ecological inputs, Be,k. The allocation matrix, Γ, depends
107 on the network and the selected allocation method, for partitioning cumulative exergy between multiple outputs.
The ecological inputs, Be,k, represent global inputs such as solar, tidal and deep earth exergy. Equations analogous to equations 2.53 and 4.1 may calculate the CEC of global inputs as
−1 −1 Ce = ηe ⋅ Be or Ce,k = ηe,k Be,k (4.6)
Equation 4.6 is important for connecting exergy and emergy, as shown in Section
3.2. Here, ηe,k may equal unity if such inputs are assumed to be directly available without any previous transformation, or if they represent exergy of the same type (quality). This assumption does not ignore any known processes, unlike the assumption of ICEC analysis represented by equation 2.51. Alternatively, proportionality constants may be assigned to ηe,k if one global input is to be expressed in equivalents of another, as done in emergy analysis. Equations 4.3, 4.4 and 4.5 provide alternate ways of estimating the
ECEC of products from any production chain.
4.2 Relation between ECEC and Emergy
Deriving the exact relationship between ECEC and Emergy and conditions for their equivalence relies on writing matrix equations for emergy analysis of a network followed by comparing the equations to those derived for ECEC in Section 3.1. Equation
2.56 relating emergy and exergy may be written in matrix form as
Mp = Τp ⋅ Bp (4.7) 108 where Mp and Bp are vectors of emergy and exergy and Τp is the diagonal matrix of transformities. For a network similar to that considered for ICEC analysis, Mp , may be calculated as
Mp = Γ'i ⋅Mn (4.8)
where Mn is the emergy vector of the natural resources and Γ'i is the allocation matrix for emergy analysis. Like ECEC analysis, Γ'i contains information about the allocation rule for emergy, that is, how emergy is assigned among splits, co-products and joints.
Similarly, the emergy of natural resources, Mn , can be calculated as
Mn = Γ'e ⋅Me (4.9)
Equations 4.8 and 4.9 are analogous to equation 4.2 for ECEC analysis. As shown in equation 2.56 and 4.7, the emergy and exergy of global inputs are related as
Me = Τe ⋅ Be (4.10)
where, Τe represents the solar transformities of global inputs. Combining equations 4.8,
4.9 and 4.10, and using an overall allocation matrix, Γ' analogous to that in equation 4.3
Mp = Γ'⋅Τe ⋅ Be (4.11)
For ECEC and emergy to be equivalent, equation 4.12 must be satisfied
Cp = Mp (4.12)
109 Equations 2.53, 4.7 and 4.12 show that transformity is the reciprocal of the cumulative degree of perfection. This is
−1 Τp = ηp (4.13)
Furthermore, equations 4.3, 4.6, 4.11 and 4.12 imply that
−1 Γ ⋅ ηe = Γ'⋅Τe (4.14)
For a fair comparison, it is essential for both, ECEC and emergy to have the same analysis boundary that considers the same network of processes. Secondly, if the allocation rule used by emergy and cumulative exergy analysis is identical, then Γ = Γ'.
Under these conditions, equation 4.14 reduces to
−1 ηe = Τe (4.15)
This analysis indicates that ECEC and emergy are identical if cumulative exergy and emergy use the same approach for combining global energy inputs. Alternate approaches include, directly adding global energy inputs (using a unit transformity), or representing global energy inputs in solar equivalents using the transformities estimated in emergy analysis. Thus, the condition for equivalence between emergy and ecological cumulative exergy consumption is as follows.
Ecological cumulative exergy consumption and emergy are equivalent if the following are identical,
110 1) Analysis boundary.
2) Allocation method.
3) Approach for combining global energy inputs.
This condition shows that ecological cumulative exergy consumption and emergy are very closely related. Moreover, it justifies the use of the reciprocal of transformity to estimate the CDP of natural resources, as in equation 4.13. The illustrations in Section 5 are based on this insight.
There remain conceptual differences between emergy and ECEC analyses. ECEC analysis does not imply any relationship with economic value. In fact, ECEC analysis can complement economic analysis. Legitimacy of the Odum’s maximum empower principle is irrelevant for the applicability of ECEC analysis. There are clear links between ECEC and other thermodynamic quantities. Representing global exergy inputs in equivalents of solar energy is not necessary albeit convenient. ECEC faces similar quantification challenges as Emergy, but these challenges are no different from those faced by any holistic approach including life cycle assessment, as discussed in Chapter 3.
4.3 ECEC Computation
The equations for ECEC analysis given in Section 4.1 do not provide adequate details about how ECEC may be computed in practice. This section addresses such practical issues as allocation and network algebra followed by a formal algorithm for
ECEC analysis.
111 4.3.1 Network Representation and Algebra
The network algebra of input-output analysis provides a convenient and rigorous way of analyzing flow in any network. Any network may be represented as shown in
Figure 4.2. Figure 4.2a shows a dummy unit (unit 3) being used to separate the information of two final products leaving the same unit (unit 2). Figure 4.2b shows a generic representation of a three-unit system, such as the one represented by Figure 4.2a.
Figure 4.2b can be represented in a tabular form, as shown in Table 4.1. Cn,i represents the input to the system received by the i-th unit. Cij represents a direct transaction from the i-th to the j-th unit. Input-output analysis requires that the attribute being analyzed is conserved, for example mass, energy, cumulative exergy, money, etc. Therefore, Ci is the sum of all the direct inputs to the i-th unit, which also equals the sum of all the outputs from the i-th unit. The transaction coefficient, γij, represents the fraction of Ci that is transferred to the j-th unit. γp,i is the fraction of Ci leaving the system. Transaction coefficients, γij’s and γp,i’s, and the system inputs, Cn,i, are known variables. Transaction coefficients carry information about the allocation method used. In ICEC analysis, they are functions of the unit outputs exergy. Table 4.2 shows the values of the process network in Figure 4.2a.
The fifth column of Table 4.1 can be formulated as a system of equations. For a generic system this would look like
112 Cn,1
Cp,1 C11 B Bp,1 n,1 1 1 Cn,1 C13 Cp,3 B12 C31 Bp,2 C 3 C21 12 3 2 C23 Bn,2 C22 Bp,3 Cn,2 C32 2 Cn,3 B 22 C33
Cn,2
Cp,2
Figure 4.2: Example of a process network: (a) Original flowchart; and (b) Generic representation.
Units 1 2 3 Output Total 1 C11 = γ11C1 C12 = γ12C1 C13 = γ13C1 Cp,1= γp,1C1 C1 2 C21 = γ21C2 C22 = γ22C2 C23 = γ23C2 Cp,2= γp,2C2 C2 3 C31 = γ31C3 C32 = γ32C3 C33 = γ33C3 Cp,3= γp,3C3 C3 Input Cn,1 Cn,2 Cn,3 Total C1 C2 C3
Table 4.1: Tabular representation of Figure 4.2b.
113 Units 1 2 3 Output Total 1 0 γ12 0 γp,1 C1 2 0 γ22 γ23 γp,2 C2 3 0 0 0 γp,3 C3 Input Cn,1 Cn,2 0 Total C1 C2 C3
Table 4.2: Transaction coefficients and CEC for Figure 4.2b.
C p,1 = γ p,1C1 C = γ C p,2 p,2 2 (4.16) M
C p,n = γ p,nCn
For better mathematical manipulation, equation 4.16 is represented in matrix form as
Cp = γ p ⋅C (4.17)
where γp is the diagonal matrix with the coefficients γp,i’s along the diagonal and C is the vector formed by the coefficients Ci’s.
Another system of equations can be formulated by adding the elements of a column of Table 4.1 corresponding to a system unit. For a generic system this would look like
114 γ 11C1 + γ 21C2 +K+ γ s1Cs + Cn,1 = C1 γ 12C1 + γ 22C2 +K+ γ s2Cs + Cn,2 = C2 (4.18) M γ 1sC1 + γ 2sC2 +K+ γ ssCs + Cn,s = Cs
In matrix form, equation 4.18 can be written as
T γ ⋅C + Cn = C (4.19)
where γ is the square matrix formed with the transaction coefficients, γji’s. γ represents the interaction between system units. Cn is the vector of cumulative exergy of system inputs. Equation 4.19 can be rearranged, to solve for C, as follows
T −1 C = ()I − γ ⋅Cn (4.20)
The vector of cumulative exergy of the products, Cp, can be calculated by combining plugging equation 4.20 into 4.17. This is
T −1 Cp = γp ⋅ ()I − γ ⋅Cn (4.21)
The form of the generic allocation matrix Γ is then
T −1 Γ = γp ⋅ ()I − γ (4.22)
This network representation and algebra can be used for ICEC analysis. However, whether it can be used for computing ECEC depends on the allocation approach, as discussed next.
115 4.3.2 Allocation
Since most industrial and ecological processes have multiple outputs, it often becomes necessary to allocate or partition the inputs between multiple outputs. Due to its subjective character, a variety of methods has been suggested for allocation. These are based on the market value, mass, energy or exergy content, and energy quality of the outputs. Techniques for avoiding allocation by modifying the system have also been suggested (Weidema, 2001) and are recommended in the ISO 14000 standards (ISO
14040, 1997).
4.3.2.1 Allocation in Fully Defined Networks
Allocation according to the exergy of output streams is popular in ICEC analysis
(Szargut et al., 1988; El-Sayed and Gaggioli, 1989). In this approach, the cumulative exergy of an output stream from the i-th to the j-th unit, Cij, is
Cij = γ ijCi (4.23)
where γ ij can be calculated as
Bij γ ij = (4.24) ∑ Bij + B p,i j
where γij is the transaction coefficient from Unit i to Unit j, Bij is the exergy delivered from Unit i to Unit j, and Bp,i is the product stream from Unit i. Product streams, Bp,i, are
116 output streams that leave the system. Figure 4.3a shows the allocation based on equation
4.24. As shown in Figure 4.3b, when the streams allocated according to this scheme are combined, their cumulative exergy can be added. Equations 2.54 and 4.21 show that the allocation matrix for all industrial processes, Γi, is that given by equation 4.22. This is
T −1 Γi = γp ⋅ ()I − γ (4.25)
(a) Splits
Exergy CEC Algebra (J/t) (J/t) Operation 20 400 100 1000 ⎡0.4⎤ C = ⋅ 1000 30 600 output ⎢ ⎥ [] ⎣0.6⎦
(b) Joints 20 400 10 1000 400 ⎡ ⎤ 30 600 Coutput = []1 1 ⋅ ⎢ ⎥ ⎣600⎦
Figure 4.3: Allocation in industrial systems: (a) for splits; (b) for joints.
117 This allocation approach relies on detailed knowledge of the network and outputs for allocation. Its benefits are that cumulative exergy follows laws of conservation, making the algebra quite straightforward, intuitive, and consistent with widely used network algebra.
4.3.2.2 Allocation in Partially Defined Networks
If knowledge about the network structure and its outputs is not available, it is not possible to use equation 4.24 for allocating the cumulative exergy consumption to the outputs. For instance, Figure 4.4a shows a system where only two outputs are fully defined, whether there are more outputs or not is unknown, portrayed by the triple dots between the known outputs. Even if the existence of additional outputs was known, it is often not possible to know their exergy content or network. This is usually the case with ecosystems since complete knowledge about the ecological network and its goods and services is not available. One strategy for such partially defined systems is to avoid allocation entirely, and consider the exergy consumption of the process to be essential for making each product. Figure 4.4 illustrates this allocation approach. The main advantage of this approach is that the transaction matrix, γ, can be defined by ignoring the unknown streams without losing information. However, since this allocation scheme violates conservation, special care is needed to avoid double counting when outputs from such systems are combined. If the input streams originate from a partially known system like that in Figure 4.4a, adding their cumulative exergy consumption will result in double counting. If the streams are known to follow the allocation scheme shown in Figure 4.4a,
118 then the approach shown in Figure 4.4b, referred as maximum criterion, is used for combining streams. This avoids double counting by considering only the largest cumulative exergy of the set of inputs from a system under allocation scheme described in Figure 4.4. However, if the combined streams represent cumulative exergy over different temporal horizons, they may be added without double counting.
(a) Splits Exergy CEC Algebra (J/t) (J/t) Operation 20 1000 100 . 1000 . ⎡1⎤ : : Coutput = ⎢ ⎥ ⋅[]1000 30 1000 ⎣1⎦ (b) Joints
10 500 ⎡ 500 ⎤ 10 1000 C = 0 1 ⋅ 30 output []⎢ ⎥ 1000 ⎣1000⎦ Maximum criterion
Figure 4.4: Allocation in partially known systems: (a) for splits; (b) joints.
119 A similar allocation approach is used in emergy analysis (Odum, 1996), for determining the transformities of many ecological products and services as well as for allocation between “co-products”, even in fully defined networks. Odum’s justification for using this allocation approach is that inputs cannot be allocated among co-products since they cannot be produced independently by using a fraction of the process’ exergy consumption. Emergy analysis also selects the allocation approach depending on whether the products are of same or different energy quality, which is reflected in the transformities, and whether they are produced over different time horizons. Thus, in general, “renewable” resources are considered to be non-additive, while “non-renewable” resources are additive. While this is a legitimate and appealing approach, it has been the source of much confusion since it can cause the results of emergy analysis to change with the selected analysis boundary, and as more details become available (Hau and Bakshi,
2004b). In this Chapter, the allocation approach depicted in Figure 4.4 is used only for those ecological goods and services where details about the network and products are unknown. The sensitivity of the results to the allocation method, and techniques for avoiding allocation altogether are subjects of on-going research.
4.3.3 Algorithm for ECEC Analysis
Given an industrial network consisting of the main process and relevant processes in the supply chain, an exergy flow diagram can be derived by considering the main process units and calculating the exergy of each stream. The algorithm shown in Table
4.3 is for ECEC of a network of N units, where every unit delivers no more than one
120 stream to another unit and has only one input stream and one output stream crossing the system boundaries. The approach also requires values of CDP or transformity of the relevant inputs from the ecosystem.
The ECEC algorithm can use one or both allocation methods illustrated in Figure
4.3 and Figure 4.4. When ECEC of natural resource streams can be added, the allocation matrix and ECEC of products are calculated with equations 4.25 and 4.4, respectively.
When natural resource inputs cannot be added, the maximum criterion described in
Section 4.2 is applied to decide the CEC of the products. The algorithm is shown in Table
4.4. The j-th column of the allocation matrix contains the fraction of CEC of the j-th natural resource assigned to each product. The algorithm multiplies each column of the allocation matrix by the ECEC of its corresponding natural resource. Then, all numbers of the set of non additive inputs in each row, except the maximum, are set to zero. This algorithm is also equivalent to doing separate ECEC analyses for each natural resource input to obtain multiple ECEC values at each network edge corresponding to each ecological input. The ECEC values at each edge are added for additive ecological inputs, or the maximum value is taken for non-additive natural resources. The allocation methods and formal algorithm presented in this section avoid the confusing algebra that has plagued emergy analysis, without sacrificing the ability of emergy analysis to account for ecological inputs.
121 1 MAINPROGRAM, ECEC_Analysis 2 FOR i = 1 TO N 3 INPUT Bn,i 4 INPUT Bp,i 5 INPUT ηn,i 6 FOR j = 1 TO N 7 INPUT Bij 8 END 9 END 10 FOR i = 1 TO N 11 FOR j = 1 TO N Bij 12 γ ij = ∑ Bij + B p,i j 13 END
B p,i 14 γ p,i = ∑ Bij + B p,i j
15 END 16 γ = MATRIX (γij) 17 Bn = VECTOR (Bn,i) 18 Bp = VECTOR (Bp,j) 19 ηn = DIAGONAL MATRIX (ηn,i) 20 γp = DIAGONAL MATRIX (γp,i) T −1 21 Γi = γp ⋅ ()I − γ −1 22 Cn = ηn ⋅ Bn 23 PRINT “Can natural resource streams be added (system is fully specified)?” 24 INPUT Q 25 IF Q = “no” THEN 26 GO TO MaxSelect 27 END 28 Cp = Γi ⋅Cn
29 PRINT Cp 30 END
Table 4.3: General ECEC Analysis Algorithm.
122 31 SUBPROGRAM, MaxSelect 32 FOR i = 1 TO N 33 FOR r = 1 TO (N-1) 34 k = r
35 aik = ()Γi ik ⋅Cn,k 36 FOR j = (r+1) TO N 37 PRINT “Are inputs ” k “ and ” j “ additive?” 38 INPUT Q 39 IF Q = “yes” THEN
40 aij = (Γi )ij ⋅Cn,j
41 IF aij > aik DO 42 ()Γi ik = 0 43 k = j 44 END
45 OTHERWISE (Γi )ij = 0 46 END 47 END 48 END 49 END 50 END
Table 4.4: Subprogram for avoiding double counting.
4.3.4 Illustrative Example
The ECEC approach is illustrated via a simple network shown in Figure 4.5 (see
Appendix A for details about application of the algorithm). The two natural resource
123 inputs are assumed to be from the same ecological processes, and cannot be added. The
ECEC for each natural resource is shown in parentheses below each network edge in
Figure 4.5, and is propagated independently through the network. The input ECEC is allocated based on exergy in Unit 1. Since the ECEC values from the two resources cannot be added, the ECEC at each edge is the maximum value that is in the box.
However, if the inputs were additive, the result at each edge would be the sum of the
ECEC values in the corresponding parentheses.
Ecological Industrial Processes Process
30 1 10 ( 1000 / 0 ) ( 500 / 0 ) Be,j 10
20 ( 500 / 0 ) 2 10 ) ( 500 / 400 ) ( 0 / 400
Bn,k C 1 / C 2 ( n,k n ,k ) Box indicates largest ECEC value
Figure 4.5: Illustrative Example of ECEC Analysis.
124 4.4 Examples
These examples illustrate the application of ECEC analysis to processes after completing their ICEC analysis. The first example illustrates the large contribution from ecosystems to a typical chemical process that is ignored by ICEC analysis. The second example compares electricity generation by solar thermal versus coal thermal power plants. Both examples consider a narrowly defined boundary, and ignore labor and capital requirements, and the impact of emissions. Consequently, these analyses cannot be used for decision or policy making, but simply serve to illustrate the direct extension of ICEC to ECEC via transformities. Furthermore, since standard emergy analysis accounts for labor and capital requirements, equivalence between standard emergy analysis and the
ECEC results of this section requires a broader boundary. More holistic analysis of these processes along with metrics for comparing the impact and environmental performance of industrial processes are topics of on-going research.
4.4.1 Chlor-Alkali Process by Mercury Cell
A simplified flow diagram of selected processes from the extraction of natural resources to the three products, sodium hydroxide, hydrogen and liquid chlorine, and a by-product, dilute sulfuric acid is shown in Figure 4.6 (Morris, 1991). This process has four inputs: water, salt, coal, and sulfur. The exergy and ECEC are indicated on each stream of Figure 4.6, with ECEC surrounded by parentheses (see Appendix B for details about applying ECEC analysis). The exergy, ECDP or reciprocal of transformity, and
ECEC calculated via equation 4.1 are listed in Table 4.5. All the inputs are derived from
125 the earth main sources, namely solar insolation, crustal heat and tidal energy. Water, salt and sulfur are considered to originate from the sedimentary earth and hydrological cycles, calculated on a yearly basis by Odum (2000). Since they follow the allocation for partially defined networks described in Section 4.2, their ECEC cannot be added when combined. However, ECEC of coal can be added because it belongs to a different temporal horizon. With different approaches for obtaining the transformities of these natural resources, alternate methods could be used for their combination. For example, all four inputs should be added if they are partitioned as in fully defined networks, as done by Odum (1996) for geological products, or come from different temporal horizons.
Ideally, the sensitivity of the results to these variations should be evaluated. Results of the
ECEC analysis are provided in Table 4.5.
The overall analysis shows that ICEC is 66.68 MJ for the mercury cell process, resulting in a ICDP of 10.5 %. In contrast, the ECEC is 247.62 ×1010 sej, and the ECDP is 2.82 ×10 −6 J/sej. This example shows that accounting for the exergy consumed in ecological processes can change the numbers by as much as five orders of magnitude, which confirms the huge contribution of ecosystems due to ecological processes that convert low quality energy into high quality raw materials. It indicates that focusing only on the industrial processes from resource extraction onwards may be too narrow for life cycle assessment or environmental performance evaluation of industrial processes.
126 Solar Insolation Crustal Heat Tidal Energy
Ecological Processes
Water Salt Coal Sulfur 0.098 (0.26) 3.08 (3.08) 60.31 (241) 3.19 (6.38) 1 2 3 4 Water Salt Mine AC H2SO4 Reservoir Production Production
0.098 (0.26) 0.40 (3.08) 14.48 (241) 0.58 1.26 (21) (6.38) 0.008 13.22 (220) (0.02) Brine Preparation 5 Rectifier . 6 Plant
3.24 (92) 12.79 (220) Electrochemical Cells 7
3.24 (89) 7.86 1.78 0.09 (0.24) (215) (49) 9 8 Cl2 Cooler / Denuder Drier
1.14 (41) 1.53 3.35 (120) 1.74 (55) (44) 0.24 (4)
Cl2 10 Legend: Compressor Exergy in MJ (ECEC in 1010sej) 1.82 (48) # Unit number 0.23 (4) 0.50 (8) 0.28 (5) 11 Cooler Cooler / . 12 Cooler 13 Drier 1.812 0.42 1.455 (58) 3.305 (127) (46) (10) NaOH sol. Hydrogen Liquid 70% Chorine H2SO4
Figure 4.6: Flow diagram of the mercury cell process. 127 a -5 10 i Stream Material Bi (MJ) ηi=1/τi (10 J/sej) Ci (10 sej) n,1 Water Reservoir 0.098 3.717 b 0.26 c n,2 Salt Mine 3.08 10.000 b 3.08 c n,3 Coal (for electricity) 60.31 2.500 b 241.24 c b c n,4 Sulfur (for H2SO4) 3.19 5.000 6.38 n Overall 66.68 247.62 p,11 NaOH sol 1.455 0.248d 58.60e p,12 Hydrogen 3.305 0.258d 128.25e p,13 Liquid Chlorine 1.812 0.345d 52.58e d e p,9 70% H2SO4 (waste) 0.420 0.396 10.60 p Overall 6.992 0.282d 247.62 a Morris (1991); b Odum (1996, 2000); c Equation 4.1; d Equation 2.52; e Algorithm in Table 4.3.
Table 4.5: ECEC of mercury cell process.
4.4.2 Electricity from Coal versus Solar Energy
This example compares electricity generation via solar-based versus coal-based
thermal processes. It relies on ICEC analysis data provided by Szargut et al. (1988) and
Horlock (1992). Like ICEC analysis, this example also ignores emissions and their
impact, capital inputs such as equipment and land, and human resource inputs such as
labor. Consequently, this analysis is not holistic enough to permit decisions about either
approach. However, it does illustrate the approach developed in this Chapter.
Figure 4.7a shows the exergy flow diagram of a coal-driven steam power plant.
An additional 7.05 kW of exergy from fuel oil is required to extract 141.95 kW of coal
from the ground. Coal is mixed with air in a combustion chamber to heat the steam that 128 moves the turbine that produces the electricity. To complete the Rankine cycle, the partially condensed steam is recycled. Using equations 2.50 and 2.52, the ICEC of the process is 149.00 kW and ICDP of the process is 23.2%.
Air Coal Fuel oil Solar radiation 0 kW 141.95 kW 7.05 kW 270.82 kW
Coal Extraction Parabolic through collectors
141.95 kW 53.77 kW Oil Furnace Boiler heat exchanger
Steam Steam 51.84 kW Steam Turbine Steam Turbine
Condenser Condenser
Compressor Compressor 34.51 kW 34.51 kW Exhausted Electricity gases Electricity
Figure 4.7: (a) Exergy Flow diagram for a Coal-driven Power Plant; (b) Exergy Flow diagram for a Thermal Solar Power Plant.
129 Figure 4.7b shows the exergy flow diagram of a photothermal steam power plant.
A network of parabolic through collectors receives exergy of 270.82 kW in the form of solar radiation. The collectors concentrate the solar radiation on the receivers to heat the working fluid, typically oil. The heat content of the oil is transferred to the steam, in the
Boiler heat exchanger. Data for this process was obtained from (Singh et al., 2000). From equations 2.50 and 2.52, ICEC of the process is 270.82 kW and ICDP of the process is
12.7%.
Table 4.6 summarizes the results of the ICEC analysis. Exergy of exhausted gases has been neglected for the case of the coal-driven power plant so exergy of natural resources are all allocated to electricity. The coal-driven power plant is more efficient as evidenced by a higher ICDP, and from a traditional thermodynamic viewpoint, generating electricity from coal seems to be more efficient due to its higher ICDP, than solar-based electricity.
For ECEC analysis, the ECDP or reciprocal of transformity of solar radiation, coal and fuel oil are 1 J/sej, 2.50×10 −5 J/sej and 1.85×10 −5 J/sej, respectively. For the coal-based power plant, the ECEC of coal and fuel oil can be added. Table 4.7 summarizes the results after accounting for ecological goods and services. Due to the unit
ECDP or transformity of sunlight, the ECDP of the solar plant is equal to its ICDP.
However, the ECDP of the coal-driven power plant is significantly lower due to the exergy invested by ecological services in coal and oil for converting it into a more concentrated and higher quality source of energy. ECEC analysis now shows that photothermal electricity may be overwhelmingly thermodynamically superior to coal-
130 based electricity. However, inclusion of the exergy consumption due to economic and capital inputs and the impact of emissions may have a large effect on these numbers, and is necessary before reaching any conclusions about comparing these technologies.
Further extensions to include the contribution of indirect activity in the economic network are also essential for improving the accuracy of the results. A variety of existing methods may be useful for meeting these challenges (Sciubba, 2001; Odum, 1996;
Ukidwe and Bakshi, 2004).
a b Electricity from Cp (kW) ηp (%) Coal 149.00 23.2 Solar Energy 270.82 12.7 a Equation 2.50; b Equation 2.52.
Table 4.6: ICEC analysis of solar and coal-based power plants.
3 a 2 b Electricity from Cp (10 sej/s) ηp (10 J/sej,%) Coal 6,058,624.40 0.0006 Solar Energy 270.82 12.7 a Equation 4.4; b Equation 2.52.
Table 4.7: ECEC analysis of solar and coal-based power plants.
131 This work is expected to help “bridge the gap between Ayres’ industrial ecology and Odum’s systems ecology” (Anonymous Reviewer, 2003) and lead to new methods and insight for understanding the principles of sustainability. Many opportunities are available for further work. The challenge of combining resources over multiple temporal and spatial scales plagues many holistic techniques, including the ones discussed in this
Chapter. The transparency and utility of existing methods could be improved by developing a tiered system which distinguishes between resources according to their replenishment time. Instead of categorizing resources as renewable or non-renewable, this system could separate resources according to their renewability over daily, short- cycle, long-cycle, or cosmological time scales. A similar spatial hierarchy could also be defined. Ideally, a systematic multiscale statistical framework is needed that considers differences in the quality (uncertainty) of data at multiple temporal and spatial scales, and combines these data in an “optimal” and transparent manner. Concepts of “opportunity” and “sunk” costs from economics could also be useful for considering opportunities and alternatives that may be lost due to a decision. Research in these and other related areas is necessary for recognizing the full potential of thermodynamic methods for environmentally conscious decision making.
132
CHAPTER 5
ENHANCING LIFE CYCLE INVENTORIES VIA RECONCILIATION
We must as second best…take the least of the evils.
–Aristotle
Collecting all data and information of each process included in the LCA study is part of the recording step in the inventory analysis phase of the LCA methodology. The performance of the LCA will strongly depend on the quality of this data. Obtaining good quality data requires working in collaboration with the suppliers, distributors, etc., which can be quite expensive and even inaccessible if the company declines to disclose the required information. Consequently, process inventories are very often limited by their availability. Quite often, companies solve this shortcoming by using general publicly available databases. Since these databases collect information from different sources, disagreement is very common. Process inventories may be outdated, unreliable, incomplete and/or unverifiable (Ayres, 1995). Data that are too old are not representative for processes that change in the lapse of a few years. Collecting data that is too sparse and averaged may not be trustworthy, often leading to inconsistencies like disappearance or
133 creation of mass and energy. Frequently, process data does not include some emissions or streams that are considered unimportant, further increasing the gap in the mass balance.
In such cases, LCA has no utility if physical data is wrong with respect to critical pollutants (Ayres, 1995). An imbalance in which there is more mass in the output side than in the input side underestimates the amount of emissions associated with the supply chain of the missing mass. An imbalance in which there is more mass in the input side than in the output side also underestimates the amount of emissions, associated with the demand chain of the missing mass. Schaltegger (1997) argues that from an economic point of view current LCAs provide a small potential benefit given the high probability of potentially wrong decisions—due to their reliance on background inventory, unrepresentative, low quality and aggregated data—and high costs.
Techniques for data rectification have been used in Process Systems Engineering for more than two decades. Such techniques have been quite successful in dealing with inconsistent data. Applying data rectification to process inventories is an inexpensive and easy way to overcome the shortcomings discussed earlier in this Section. Moreover, it can be incorporated as a systematic approach for data validation in the recording step, thus ensuring that mass and energy balances will always agree in the final datasheet.
If the material flows in the system are not balanced, then it will not be possible to balance the energy flows. In satisfying the material balance, the total mass of each chemical element present must be equal between the inputs and the outputs. The first law or law of energy conservation is another condition that can be used to reconcile the process inventory. However, this additional constraint is hard to apply because the
134 reported energy data is usually not enough to be used for reconciliation. Moreover, the energy balance can always be compensated with dissipated heat or other low temperature energy that is never reported. Applying Data rectification techniques provides the most rigorous way to rectifying empirical data of LCA process inventories and estimating the values of missing data. Since these inventories typically present industrial processes as black boxes, data rectification may not be used at its full potential, such as applying its powerful statistic tests. Despite these limitations, data reconciliation techniques guarantee improvement of inconsistent LCA process inventories.
Using the second law for reconciling LCI databases is not possible because this law does not impose an equality constraint. Nevertheless, the second law constraint demands a minimum of energy required to operate any process. More specifically, if a process inventory does not meet the energy requirement imposed by the second law, then it is unfeasible, even when satisfying the mass and energy balances. In a LCA, neglecting this additional energy requirement translates into underestimating the amount of emissions. Consequently, enforcing the second law in the data validation procedure is a must-do.
Exergy analysis is a tool that has been successfully used for identifying the major causes of thermodynamic imperfection of thermal and chemical processes and quantifying the portion of energy that can be practically recovered. Because exergy analysis is based on thermodynamics, it is a great tool for detecting second law violations. All industrial processes incur in exergy losses that are quantifiable. For a specific process, the most thermodynamically efficient plants will experience lower
135 exergy losses than the average. Nevertheless, no matter how efficient a plant may be, its exergy losses cannot be lower than the minimum, inevitable exergy loss. A potential use of exergy analysis for the data validation of process inventories is calculating the exergy loss of the process based on the process inventory and comparing its value to a feasibility chart like the one shown by Figure 5.1. If the calculated exergy loss is lower than the
minimum possible, W&lost,INE , then the inventory represents a process that cannot exist because of the second law of thermodynamics. Sometimes, inevitable exergy losses may not be available, in such cases, it may not be possible to validate the inventory data.
Nevertheless, it is still possible to detect second law violations in some process inventories if their exergy loss is lower than zero (a negative value). Finally, for well known processes, it is possible to compare the exergy loss of the plant described by the available process inventory with the average for that process, this is the expected value
E(W&lost ).
136 Feasibility Of Process Inventory (%)
0 W&lost ,INE E(W&lost ) W&lost
Figure 5.1: Feasibility of the process inventory, where W&lost,INE is the inevitable exergy loss and E(W&lost ) is the expected value of the actual exergy loss.
5.1 Methodology
The proposed approach improves the quality of process inventories by adjusting the values of the reported variables to satisfy the mass balance. This is carried on using techniques for data rectification. These techniques also provide the means for estimating missing data. Enforcing the law of conservation of energy in the reconciliation exercise is quite challenging if the process inventory reports insufficient data for the energy balance constraint. Moreover, the energy balance constraint can always be satisfied by adjusting
137 the value of dissipated heat, which is always unknown. What is more, energy is required for meeting operating conditions in the process units. Thus, there is no way to know how much energy is required for the process unless these operating conditions are known, which is highly unlikely from the black box style of LCA process inventories. Therefore, data rectification with the energy balance constraints is not treated here; but the theory for this case can be found elsewhere (Romagnoli and Sánchez, 2000; Narasimhan and
Jordache, 2000). This methodology also performs exergy analysis to test for second law violations. This is applied over the reconciled data and primarily verifies that the exergy inputs are sufficient to drive the process. Data rectification with the second law constraint can be achieved by increasing the input flows of energy.
The methodology for the proposed approach is illustrated through the algorithm shown in Figure 5.2. Based on the process data, the first step is to calculate the material balances, i.e. the total and for each component. Since there is no assurance that the chemical reactions involved in the process will be known, a mass balance for each chemical element is suggested instead. The material balances are verified to assure that the mass of each chemical element entering the system is also leaving it. If they are satisfied, then there is no need for reconciling the mass flows, and the procedure advances to the next phase, which involves exergy analysis and verifying the second law constraints. If the material balances are not satisfied, then the process data is reconciled.
A missing flow is added every time that a chemical element is present in the inputs but not the outputs or vise versa. The next step is to test the reconciled data to identify gross errors. Depending on how incomplete the inventory is, the data reconciliation gives
138 meaningless results or large deviations. If the results are not satisfactory, the algorithm suggests formulating an alternative for the inventory as it will be illustrated later on this chapter. This can be done by obtaining more information about the process, changing some of the existing assumptions or incorporating new ones. Once the mass flows are reconciled with satisfactory results, the next phase is to perform an exergy analysis. If the thermodynamic state of the streams cannot be determined, then one may assumed that they are at standard state, i.e. temperature and pressure of the surroundings. The second law is verified by assuring that the exergy losses in the process exceed the inevitable exergy loss. Alternatively, one can check that the exergetic efficiency of the process is lower than the maximum achievable efficiency. If the second law is satisfied, one last assessment is performed to make sure that further improvement of the inventory is unlikely. In this case, the resulting inventory is selected; otherwise, more alternatives are produced until a satisfactory rectified inventory is obtained.
Given the wide variety of LCI process inventories, the effectiveness of the algorithm can vary drastically. For instance, an inventory missing too many streams may require making unsure assumptions about the process, adding uncertainty about whether the data is representative of the real process. Nevertheless, since a process that violates the laws of physics is unfeasible, it is guaranteed that any rectified inventory is more likely and therefore preferable. The following sections describe every step of the algorithm in detail.
139 5.1.1 Process Inventory
LCA Process data can be site specific for a company or an area, or can be more general. The kind of data chosen depends on the goal of the LCA study. For example, comparing two companies that produce the same item requires site specific data, while comparing two general activities allows data from public sources.
Given the wide variety of process inventories, there is no generic way to describe them. A typical kind of inventory is shown in Figure 5.3. The input data contains separate listings for the process feedstock—raw material—and the energy inputs. Energy inputs are used for providing the energy requirements of the process, for instance heating, cooling, pumping, compressing, transporting, etc. Figure 5.3 shows two different sub- modules: the main process and the energy system. In general, material flows from the energy system do not mix with those from the process itself. In such cases, only energy is transferred between these sub-systems, allowing their material balance to be performed separately. Separating the process inventory in these two sub-modules simplifies the problem and assures that species common to both subsystems will not mix up. However, as represented in Figure 5.3, reported data of the outputs is typically present together for both subsystems. Consequently, outputs require careful allocation among the subsystems.
Products are produced in the main process, while emissions and other possible by- products are generated by both sub-modules. Emissions can be classified as air emissions, water effluents or solid waste. There may be missing flows anywhere in the process inventory.
140
Start
Process External data Inventory and assumptions
Calculation Inventory of Material Alternative Balances Formulation
No Data Reconciliation Is mass No Is inventory and Gross Error conserved? satisfactory? Detection Test Yes Yes Exergy Analysis
nd Is 2 Law No satisfied?
Yes
Is inventory No satisfactory?
Yes
Stop
Figure 5.2: Algorithm for the LCI Process Data Rectification.
141
Raw Material Main Products Process Air Emissions Energy Flow Water Effluents
Solid Waste Energy Inputs Energy System
Figure 5.3: Material Flows in Process Inventories.
5.1.2 Calculation of Material Balances
In this phase of the approach, the material balances are determined to check if there is a violation of the law of mass conservation. The material balances are calculated with the process data. First of all, the streams are separated in inputs and outputs and categorized by their functionality, e.g. raw materials, energy inputs, air emissions, etc.
The most common mass unit in the inventory data is selected as the common unit.
Streams that are expressed in other units may be converted to the common unit.
Nevertheless, if this conversion factor is an approximation, then it is possible that this
142 factor will have to be adjusted as the other measured variables. The overall mass balance can be formulated for the inputs as follows
m& n = ∑ m& j , ∀j ∈{}Set of input streams (5.1) j
where m& n is the total mass flow of the input streams and for this equation, m& j is the mass flow of the j-th input stream. Similarly, the overall mass balance can be formulated for the outputs as
m& p = ∑ m& j , ∀j ∈{}Set of output streams (5.2) j
where m& p is the total mass flow of the output streams and for this equation, m& j is the mass flow of the j-th output stream. The absolute values are used in the equations above to avoid confusion about the sign of the streams.
Ideally, a material balance is also calculated for each component. When the chemical reactions involved in the process are unknown, the mass balance for every chemical element is formulated instead. The mass flow of a stream can be divided into its component mass flows, which are either the species of chemical elements. The chemical formula of all the species present in the streams must be determined first. For complex mixtures such as fuels, the mass fraction of the chemical elements can be used to formulate the material balances. The total mass input flow of the i-th component, m& n,i , can be formulated as
143 m& n,i = ∑ wij m& j , ∀j ∈{}Set of input streams ∧ i = 1,2,...,m (5.3) j
where wij is the mass fraction of the i-th component or chemical element present in the j- th stream. Similarly, the total mass output flow of the i-th component, m& p,i , can be formulated as
m& p,i = ∑ wij m& j , ∀j ∈{}Set of output streams ∧ i = 1,2,...,m (5.4) j
The mass fraction of the i-th chemical element in the j-th species can be calculated as
υij MWi wij = (5.5) MW j
where υij is the number of atoms of the i-th chemical element in the j-th species, while
MWi and MW j are the molecular weights of the chemical element and the species, respectively.
5.1.2.1 External Data and Assumptions
The ideal case is to obtain all the information from the inventory itself and any accompanying instructions or reports. Useful information includes main chemical reactions involved, the origin and fate of pollutant emissions and description of the process plant structure. When required information is not provided by the process inventory, making some assumptions and using data from external sources are necessary. 144 Since different assumptions can result in different configurations, the proposed approach encourages the formulation of various alternatives, so that the assumptions can be tested and the most reasonable alternative can be selected.
5.1.2.2 Verifying the Law of Mass Conservation
Satisfaction of the overall material balance can be checked by the following equation
m& n − m& p ≤ ε (5.6) 1 ()m& n + m& p 2 where ε is a threshold that allow small variations of the relative difference between the input and output streams. If this difference is lower than the threshold, then the overall material balance is considered to be satisfied.
Satisfaction of the components material balance can be checked with
m& n,i − m& p,i ≤ ε, i = 1,2,...,m (5.7) 1 ()m& n,i + m& p,i 2
If these differences are lower than the threshold, then there is no need for data rectification. Satisfying the material balances assures that the mass of each chemical element entering the system is also leaving it. If they are satisfied, the procedure advances to the next phase, which involves exergy analysis and verifying the second law
145 constraints. If any of the material balances is not satisfied, then the data rectification is required.
5.1.3 Data Rectification
Data rectification can be applied by following the procedure explained in Section
2.2.1. All the values reported in the inventory are the non-reconciled measured variables, denoted yi. A missing flow stream must be added every time that a chemical element is present in the inputs but not in the outputs or vice versa. Every missing flow becomes an unmeasured variable, denoted ui. Each chemical reaction used to solve the data rectification problem, also adds an unmeasured variable, namely the extent of reaction.
In order to construct the incidence matrix, A, a process flow diagram of the process is needed. In the worst case scenario, where there is absolutely no information about the process structure, the model shown in Figure 5.4 can be used for reconciling the process inventory. The model is composed of three units: a feedstock, a reactor and a separator. The reactor is the main unit, where the products and by-products are created.
The feedstock unit converts the input flows into the reactants for the reactor’s chemical reactions. The separator unit allocates the outputs into products, air emissions, water effluents and solid waste. Streams readily available as reactants for the reactor’s chemical reactions and inert streams enter the reactor as clean inputs. If they need to be converted into the reactants, then they enter the feedstock unit as raw inputs. Missing outputs can be assigned to one of the streams leaving the separator.
146 5 Products 6 Air Emissions Raw 1 2 4 Inputs Feedstock Reactor Separator 7 Water Effluents 3 8 Clean Inputs Solid Waste
Figure 5.4: Model for setting up the LCI data reconciliation problem.
The incidence matrix, A, for the model shown in Figure 5.4 can be determined with the procedure described in Section 2.2.1.1. This gives
⎡1 −1 0 0 0 0 0 0 ⎤ ⎢ ⎥ A = ⎢0 1 1 −1 0 0 0 0 ⎥ (5.8) ⎣⎢0 0 0 1 −1 −1 −1 −1⎦⎥
The balance matrix B and the master stoichiometric matrix S can be formulated similarly, but its final form depends on the species and reactions involved. Solution to situations that are different to the ones described in Section 2.2.1 may be found in the data reconciliation literature (Crowe, 1986; Romagnoli and Sánchez, 2000; Narasimhan and Jordache, 2000). Measured variables can be reconciled with equation 2.13 and unmeasured variables can be estimated with equation 2.14.
147 5.1.3.1 Validating the Reconciled Inventory
The next step is to test the reconciled data in order to identify gross errors. The test statistics for gross errors detection is described by equation 2.17. The reconciled values exceeding the confidence limits contain gross errors. Gross errors can be compensated using techniques described elsewhere (Romagnoli and Sánchez, 2000;
Narasimhan and Jordache, 2000). Alternatively, gross errors can be reduced by obtaining more information about the process, changing some of the existing assumptions or incorporating new ones. Depending on how incomplete the inventory is, the data reconciliation can give meaningless results or large deviations. If the results are not satisfactory, the algorithm suggests formulating an alternative for the inventory. Once the mass flows are reconciled with satisfactory results, the next phase is to perform an exergy analysis.
5.1.4 Exergy Analysis
Ensuring satisfaction of the second law is generally impossible through data rectification because there is not an equality constraint. This methodology performs exergy analysis to test for second law violations. This is applied over the reconciled data and primarily verifies that the energy inputs are sufficient to drive the process.
Satisfaction of the second law can be achieved by adjusting the energy requirements.
Exergy of material and heat streams can be calculated with equations 2.22 and 2.41, respectively. If the thermodynamic state of the streams cannot be determined, then one
148 may assumed that they are at standard state, i.e. temperature and pressure of the surroundings.
5.1.4.1 Verifying the Second Law of Thermodynamics
The second law is verified by assuring that the exergy losses in the process exceed the inevitable exergy loss. This is
W&lost > W&lost,INE (5.9)
Alternatively, one can check that the exergetic efficiency of the process is lower than the maximum achievable efficiency. This is
η < η max (5.10)
Satisfaction of the second law can be enforced by modifying the assumptions and external data incorporated, or simply by scaling up the energy input flows. If the second law is satisfied, one last assessment is performed to make sure that further improvement of the inventory is unlikely. In this case, the resulting inventory is selected; otherwise, more alternatives are produced until a satisfactory rectified inventory is obtained.
5.1.5 Inventory Alternative Formulation
A list of considerations that leads to creation of alternatives is discussed below.
These considerations are precisely connected to the availability or lack of information in the process inventory.
149 5.1.5.1 Chemical Composition of the Streams
Determining the chemical formula of all the streams is absolutely necessary. This means knowing the number of atoms of each chemical element present in every species.
If a substance is a mixture, then it is necessary to know what species are present in the mixture. The chemical composition of inert substances is not critical information. Once the chemical formula of all species is known, the next step is to identify which elements are missing in the inputs and in the outputs. It is possible that the chemical formula of a reacting species is not known. In this case, various chemical formulas are specified resulting in a group of alternatives. Another possibility is that species present in a mixture or their composition are not known. In such case, assumptions must be made about the composition of the mixture, which may leads to more alternatives. For instance, one of the inputs may be coal. The composition of coal can vary drastically. A new alternative follows from each coal composition specified.
5.1.5.2 Chemical Reactions
When there are no chemical reactions occurring in the process, the mass balance is quite straightforward because the same substance that enters the system must also leave it. In the case of processes with chemical reactions, it is necessary to have a formulation linking the reactive species of the outputs with those of the inputs. This formulation is given by the chemical reactions involved in the process. An advantage is that it is not required to know the reaction conditions or the chemistry of the reaction. The only requirement is that the material balance must be satisfied for each chemical element.
150 Many possible outcomes can be tested using simple chemical combinations alone. For instance, dummy chemical reactions can be created to decompose the reactive species into their forming elements. This is
M µ Nν → µM +νN for input reactive species (5.11) and
µM +νN → M µ Nν for output reactive species (5.12) where M and N are the chemical elements that constitute the substance and µ and ν are the stoichiometric coefficients that balance the reaction. The condition is that the dummy reaction satisfies the mass balance. Another possible situation is that the data is reported for a mixture of substances, like a solution or a blend. If the components of this mixture participate in the chemical reactions, then it is necessary to create another dummy formulation that separates this mixture into its components, or at least the reacting ones.
For instance, the decomposition of a mixture composed of three substances P, Q and R can be formulated as follows
Mixture(wP of P,wQ of Q, wR of R) → wP P + wQQ + wR R (5.13)
where wP, wQ and wR are the mass fractions of the substances P, Q and R in the mixture, respectively.
5.1.5.3 Measurement Units Conversion Factors
151 There must be consistency in the units system used. In the case of the mass flows, they should all be expressed in the same units. For those variables expressed in moles, molecular weights can be used to create the corresponding connection. For those variables expressed in volumetric units, densities can be used to connect these variables with the ones in mass units. Some data may be reported in volumetric or energetic units, e.g. natural gas. In such case, it is necessary to convert these units in units of mass. Some assumptions about the density or caloric value may be necessary.
5.2 Case Study: The Chlorine/Caustic Soda Production Inventory
The National Renewable Energy Laboratory (NREL) is in the process of developing a publicly available U.S. life-cycle inventory (LCI) database (NREL, 2005).
The case study done here is based on the process inventory provided by NREL under the module name: Chlorine/Caustic Soda Production (NREL, 2005). This example does not intend by any means to discredit the quality standards of NREL, especially because this database has not been completed yet. It rather aims at illustrating the methodology described here.
The process inventory is for the simultaneous production of caustic soda (sodium hydroxide) and chlorine. Hydrogen is also a co-product. The process is based on the electrolysis of a solution of sodium chloride (salt brine). The inventory shows unallocated data on the basis of 1,000 lbs of caustic soda and 893 lbs of chlorine production. The system boundaries of this module are limited to the process itself, starting with the input of salt brine and ending with the output of chlorine, caustic soda, and hydrogen.
152 Transportation of incoming salt brine is included, but it is not used in this example. The data is a combination of data from 4 confidential industry sources and data available in publicly available documents. There are three: the diaphragm cell process, the mercury cathode cell process, and the membrane cell process. Data represent weighted average of two of the commercially existing processes for the electrolysis of sodium chloride: the diaphragm electrolysis cell (85% of production) and the mercury cathode electrolysis cell
(15% of production). The chemistry of this process is described by the following reaction
2 NaCl + 2 H 2O → 2 NaOH + Cl2 + H 2 (5.14)
Because of the stoichiometry of the reaction, it is not possible to control the cell to increase or decrease the amount of chlorine or caustic soda resulting from a given input of salt. Mercury is lost in the mercury cathode cell process, thus becoming an emission.
5.2.1 Calculation of Material Balances
Table 5.1 shows the material flows of the streams for the process inventory.
Except for electricity, natural gas and residual oil, all the streams are reported in pounds
(lb). The first step is to verify the total material balance. Determining the chemical formula of most streams is straightforward. The first assumptions result from approximating natural gas to methane (CH4), sulfur oxides to sulfur dioxide (SO2) and sulfides to hydrogen sulfide (H2S). The units of natural gas and residual oil are converted into pounds (lb) by assuming a specific volume of 22.68 ft3/lb for pure methane (EPPO,
2005) and a density of 7.83 lb/gal for residual oil (EPPO, 2005). All the measured
153 variables in the same mass units are shown in the last column of Table 5.1. Total mass of inputs and outputs, shown at the bottom of the last column of Table 5.1, were calculated with equations 5.1 and 5.2, respectively. The difference is -2.5 lb and the relative difference, calculated with the left hand side of equation 5.6, is 0.1%. A common threshold, ε, is 5%. Therefore, according to equation 5.6, the overall mass balance is satisfied.
The next assumptions are the composition of coal and residual oil. Table 5.2 shows the weight fractions of the species that form both fossil fuels. These compositions were taken from examples in Szargut et al. (1988). The last column of Table 5.2 shows the component mass flows for the given compositions. Total mass of inputs and outputs for each chemical element present, shown in the fourth and fifth columns of Table 5.2, were calculated with equations 5.3 and 5.4, respectively. The material balance is shown in the sixth column of Table 5.2. The relative differences, calculated with the left hand side of equation 5.7, are shown in the last column of Table 5.2. For a threshold, ε, of 5%, every relative difference violates equation 5.7 and thus the law of mass conservation is not satisfied for any chemical element. This is the first remarkable result from this example: satisfying the overall material balance is not sufficient to ensure satisfaction of the law of mass conservation.
5.2.2 Data Rectification
For illustrative purposes, it is assumed that the chemical reaction described by equation 5.14 is unknown at first. As a matter of fact, four different alternatives are
154 developed in this example. Each successive alternative results from a modification of its previous one and aims at improving the reconciled inventory. The following sections describe the external data used and the assumptions made, and discuss the reasoning behind the formulation of each alternative.
5.2.2.1 Alternative 1 – Single Black Box
The first alternative illustrates the worst case scenario, which is when there is no information about the chemical reactions involved or about the process flow diagram.
The model used as the process flow diagram is that shown in Figure 5.4. An inspection of the component material balances shown in Table 5.3 reveals that lead (Pb), mercury (Hg), nickel (Ni) and zinc (Zn) are missing in the inputs and carbon (C) and Nitrogen (N) are missing in the outputs. Their corresponding streams are added as unmeasured variables.
The missing inputs enter directly to the reactor as clean inputs (stream 3 in Figure 5.4).
Carbon is assigned to the solid waste stream (stream 8 in Figure 5.4) and molecular nitrogen (N2) is assigned to the air emissions stream (stream 6 in Figure 5.4). It is known that mercury is lost in the mercury cell process, which means that there is no mercury in the inputs. However, if this missing input is not included, then the output would have to be removed in order to apply data rectification.
155 Measured Original Chemical Molecular Measured Stream Variable Units Formula Weight Variable (lb) Raw Materials Salt 1,670 lb NaCl 58.4 1,670.0 Energy Inputs Electricity 813 kWh ------a 3 Natural gas 4,036 ft CH4 16.0 178.0 Coal 68.1 lb -- -- 68.1 Residual oil b 0.48 gal -- -- 3.8 Products Caustic soda 1,000 lb NaOH 40.0 1,000.0 Chlorine 893 lb Cl2 70.9 893.0 Hydrogen 16.9 lb H2 2.0 16.9 Air Emissions (AE) Mercury 3.2E-04 lb Hg 200.6 3.2E-04 Chlorine 5.4E-04 lb Cl2 70.9 5.4E-04 Particulates 4.6E-04 lb Ash -- 4.6E-04 Sulfur oxides 1.0E-03 lb SO2 64.1 1.0E-03 Water Effluents Mercury 1.5E-06 lb Hg 200.6 1.5E-06 BOD 1.9E-05 lb O2 32.0 1.9E-05 Dissolved Solids 4.33 lb Ash -- 4.33 Lead 9.4E-07 lb Pb 207.2 9.4E-07 Nickel 9.4E-07 lb Ni 58.7 9.4E-07 Sulfides 1.5E-04 lb H2S 34.1 1.5E-04 Zinc 9.4E-07 lb Zn 65.8 9.4E-07 Solid Waste (SW) Solid Waste 3.08 lb Ash -- 3.08 Total Inputs 1919.8 Outputs 1917.3 Balance -2.5 a Assumed to be pure methane with a specific volume of 22.68 ft3/lb (EPPO, 2005); b Assumed to be fuel oil with a density of 7.83 lb/gal (EPPO, 2005).
Table 5.1: Process Inventory of the Chlorine/Caustic Soda Production.
156 Weight Chemical Molecular Measured Stream Fraction (%) Formula Weight Variable (lb) Coal a -- -- 68.1 Sulfur 1.3 S 32.1 0.9 Hydrogen 4.1 H2 2.0 2.8 Carbon 57.7 C 12.0 39.3 Oxygen 11.2 O2 32.0 7.6 Nitrogen 0.7 N2 28.0 0.5 Water 10.0 H2O 18.0 6.8 Ash 15.0 Ash -- 10.2 Residual Oil b -- -- 3.8 Sulfur 1.0 S 32.1 0.0 Hydrogen 15.0 H2 2.0 0.6 Carbon 84.0 C 12.0 3.2 a Assumed to be bituminous coal from Szargut et al. (1988, example 3.7, pp. 105); b Assumed to be fuel oil from Szargut et al. (1988, example 3.11, pp. 109).
Table 5.2: Composition of the Coal and Residual Oil streams.
The incidence matrix, A, is given by equation 5.8. The balance matrix, B, and the master stoichiometric matrix, S, are shown in Appendix C. They were determined by following the procedure described in Section 2.2.1.1. For the construction of the master stoichiometric matrix, formulation of dummy chemical reactions was required. These chemical reactions—reactions R1 thru R11—are shown in Table 5.4. The chemical reactions are formulated in mass units instead of moles. Equation 5.5 was used for the mass coefficients of most chemical reactions. In Table 5.4, R1 represents the decomposition of sodium chloride into elemental sodium and chloride. Similarly, R2, R3 and R4 represent the mass breakup of the fuels into their element components, in which
157 the mass fractions were used directly. Chemical reactions R1 thru R4 occur in the feedstock unit, while R5 thru R11 occur in the reactor unit. R5, R6 and R7 represent the formation of the products caustic soda, chlorine and hydrogen, respectively. R8, R9, R10 and R11 represent the formation of the by-products sulfur oxides, BOD, sulfides and nitrogen.
Table 5.5 shows the results for the data reconciliation problem. At a first glance, one can notice that the reconciled mass flows for residual oil and the BOD output are negative. Secondly, the mass flows of sodium chloride, caustic soda and chlorine are much smaller than the measured values. This is one of the cases, when results from data reconciliation are not meaningful. The cause can be analyzed better by looking at Table
5.6 to identify which variables exhibit gross errors. Recall that the variables with gross errors are those that do not satisfy equation 2.17, i.e. the test statistics is larger than 1.96
(which represents a 5% level of confidence). The largest gross error is detected in the
BOD stream, followed by sodium chloride, caustic soda, chlorine, residual oil, the sulfides and the sulfur oxides. This suggests that the problem occurs from trying to balance oxygen and sulfur simultaneously. Oxygen is provided in the inputs by coal only and is required for the production of caustic soda, BOD and sulfur oxides. In order to satisfy the demand for oxygen, more coal has to enter the system. However, more coal entering the system results in more sulfur being produced. This sulfur production causes deviations from the original values of the sulfur containing output streams. This trade-off holds back the production of oxygen, which translates in less production of caustic soda, propagating to less sodium chloride used and subsequently in less chlorine produced.
158 Chemical Chemical Molecular Inputs Outputs Balance Relative Element Formula Weight (lb) (lb) (lb) Error (%) Ash Ash -- 10.2 7.4 -2.8 -31.8 Carbon C 12.0 175.7 0.0 -175.7 -200.0 Chlorine Cl 35.5 1013.1 893.0 -120.1 -12.6 Hydrogen H 1.0 48.8 42.1 -6.7 -14.8 Lead Pb 207.2 0.0E+00 9.4E-07 9.4E-07 200.0 Mercury Hg 200.6 0.0E+00 3.2E-04 3.2E-04 200.0 Nickel Ni 58.7 0.0E+00 9.4E-07 9.4E-07 200.0 Nitrogen N 14.0 4.8E-01 0.0E+00 -4.8E-01 -200.0 Oxygen O 16.0 13.7 400.0 386.3 186.8 Sodium Na 23.0 656.9 574.8 -82.1 -13.3 Sulfur S 32.1 9.2E-01 6.4E-04 -9.2E-01 -199.7 Zinc Zn 65.4 0.0E+00 9.4E-07 9.4E-07 200.0 Total 1919.8 1917.3 -2.5 -0.1
Table 5.3: Component Material Balance (for each chemical element present).
5.2.2.2 Alternative 2 – Separate Energy Sub-module
Alternative 1 is now modified by assuming that this process inventory can be represented by Figure 5.3, in which the raw materials and products can be treated in a different system from that of the energy inputs. Each sub-module is modeled by the process flow diagram shown in Figure 5.4. The main process sub-module contains the sodium chloride input and the caustic soda, chlorine, hydrogen and mercury outputs. An inspection of the chemical elements present in the streams reveals that mercury (Hg), oxygen (O2) and hydrogen (H2) are missing in the inputs. Their corresponding streams
159 are added as unmeasured variables. Oxygen and hydrogen enter to the feedstock unit as raw inputs (stream 1 in Figure 5.4). The energy sub-module contains the energy inputs and the by-products not assigned to the main process sub-module. Missing inputs are lead
(Pb), nickel (Ni) and zinc (Zn) and missing outputs are carbon (C) and Nitrogen (N).
Their corresponding streams are added as unmeasured variables. As for alternative 1, the missing inputs enter directly to the reactor as clean inputs (stream 3 in Figure 5.4), carbon is assigned to the solid waste stream (stream 8 in Figure 5.4) and molecular nitrogen (N2) is assigned to the air emissions stream (stream 6 in Figure 5.4).
The incidence matrix, A, for both sub-modules does not change from that of alternative 1, which is given by equation 5.8. The balance matrix, B, and the master stoichiometric matrix, S, are shown in Appendix C. For the construction of the master stoichiometric matrix, the same chemical reactions—reactions R1 thru R11—were used.
Additionally, the reverse reactions of R7 and R9 were used in the feedstock unit of the main process sub-module to breakup the incoming hydrogen and oxygen into their atomic forms.
160 Name Chemical Reaction (in mass units) R1 1.00 NaCl → 0.39 Na + 0.61 Cl R2 1.00 Natural Gas → 0.75 C + 0.25 H R3 1.00 Coal → 0.01 S + 0.05 H + 0.58 C + 0.20 O + 0.01 N + 0.15 Ash R4 1.00 Residual Oil → 0.01 S + 0.15 H + 0.84 C R5 0.57 Na + 0.40 O + 0.03 H → 1.00 NaOH
R6 1.00 Cl → 1.00 Cl2
R7 1.00 H → 1.00 H 2
R8 0.50 S + 0.50 O → 1.00 SO2
R9 1.00 O → 1.00 O2
R10 0.06 H + 0.94 S → 1.00 H 2 S
R11 1.00 N → 1.00 N 2
R12 1.00 NaCl + 0.31 H 2O → 0.68 NaOH + 0.61 Cl2 + 0.02 H 2
R13 1.00 Natural Gas + 3.99 O2 → 2.74 CO2 + 2.25 H 2O
R14 1.00 Coal +1.76 O2 → 2.11 CO2 + 0.47 H 2O + 0.03 SO2 + 0.01 N 2 + 0.15 Ash
R15 1.00 Residual Oil + 3.44 O2 → 3.08 CO2 +1.34 H 2O + 0.02 SO2
R16 1.00 SO2 + 0.28 H 2O → 0.53 H 2 S + 0.75 O2
Table 5.4: Chemical Reactions used in this case study.
Table 5.5 shows the results for the data reconciliation problem. The mass flows of the energy inputs are practically zero, with natural gas showing a negative value. By looking at Table 5.6 it is clear that the main process sub-module shows better results, with all gross errors greatly reduced. From these streams, only sodium chloride and caustic soda contain gross errors. These errors appear because the measurements are not reflecting the fact that caustic soda and chlorine must be produced in stoichiometric proportions because they are both originated from sodium hydroxide, this is 1.12 pounds 161 of caustic soda per pound of chlorine. As for the energy sub-module, the fuel and ash streams exhibit gross errors. Except for natural gas and coal, all gross errors were greatly reduced as compared to alternative 1. The separation of the streams into two sub- modules, forces all the hydrogen content in the energy inputs to be converted into sulfides and all the incoming oxygen to end up as BOD. This creates a tradeoff between the adjustment of values of the BOD, sulfides and ash streams. For this same reason, the resulting values of the fuel streams are almost zero. A second law constraint or a modification of the assumed compositions of the fuels would solve the problem of the fuel streams degenerating to zero. Before any of these modifications are applied, the following alternative illustrates how knowing the chemical reactions affects the results of the data reconciliation problem.
5.2.2.3 Alternative 3 – Known Reactions
In this case, it is assumed that the chemical reactions involved in the process are known. This is not really an assumption for the case of the main process, because the chemical reaction was provided in the inventory. The chemical reaction for the production of caustic soda and chlorine is described in molar units by equation 5.14, and is shown in mass units in Table 5.4 by the name R12. For the reactions in the energy sub- module, it is assumed that the fuels undergo combustion reactions. This is not an arbitrary assumption, since the energy from fuels is usually harnessed through combustion processes and these streams are energy inputs. These chemical reactions are formulated by assuming complete combustion, which means that carbon monoxide is not an output.
162 They are shown in Table 5.4 by the names R13, R14 and R15 for the combustion of natural gas, coal and residual oil, respectively. The fact that there are sulfides present in the outputs is acknowledged with the chemical reaction shown in Table 5.4 by the name
R16. By incorporating these chemical reactions, the chemical reactions from the previous alternatives are unnecessary. Consequently, reactions R1 thru R11, shown in Table 5.4, are not used for the construction of the master stoichiometric matrix. Since all chemical reactions are known, there is no need of converting the species into their chemical elements. For the main process sub-module, there are only two missing inputs: water
(H2O) and mercury (Hg). Their corresponding streams are added as unmeasured variables. For the energy sub-module, the missing inputs are mercury (Hg), oxygen (O2), nickel (Ni) and zinc (Zn) and hydrogen (H2) and the missing outputs are carbon dioxide
(CO2), water (H2O) and nitrogen (N2). All the outputs are assigned to the air emissions stream (stream 6 in Figure 5.4). All the inputs of both sub-modules are clean inputs, thus the feedstock unit, and streams 1 and 2 in Figure 5.4 are not needed.
The incidence matrix, A, for both sub-modules are now given by the following equation
⎡1 −1 0 0 0 0 ⎤ A = ⎢ ⎥ (5.15) ⎣0 1 −1 −1 −1 −1⎦
The balance matrix, B, and the master stoichiometric matrix, S, are shown in
Appendix C. For the construction of the master stoichiometric matrix, chemical reactions
R12 thru R16 were used.
163 Chemical Reconciled Values (lb), Alternatives a Stream Formula 1 2 3 4 Raw Materials Salt NaCl 6.61,548.7 1,542.1 1,542.1 Energy Inputs Electricity ------Natural gas CH4 134.2 -0.5 178.0 178.0 Coal -- 9.2 0.0 2.2 68.1 Residual oil -- -4.4 0.8 0.5 3.8 Missing Inputs Lead Pb 9.4E-079.4E-07 9.4E-07 9.4E-07 Mercury Hg 3.2E-04 3.2E-04 3.2E-04 3.2E-04 Nickel Ni 9.4E-079.4E-07 9.4E-07 9.4E-07 Zinc Zn 9.4E-07 9.4E-07 9.4E-07 9.4E-07 Hydrogen H2 -- 43.6 -- -- Oxygen O2 -- 424.0 712.3 762.6 Water H2O -- -- 475.3 475.3 Products Caustic soda NaOH 4.5 1,059.9 1,055.4 1,055.4 Chlorine Cl2 4.0 939.5 935.5 935.5 Hydrogen H2 33.4 16.9 26.6 26.6 Air Emissions (AE) Mercury Hg 3.2E-043.2E-04 3.2E-04 3.2E-04 Chlorine Cl2 2.4E-06 5.7E-04 5.7E-04 5.7E-04 Particulates Ash 8.6E-05 2.8E-07 2.0E-05 4.6E-04 Sulfur oxides SO2 0.0977 0.0119 0.0438 0.0010 Water Effluents (WE) Mercury Hg 1.5E-061.5E-06 1.5E-06 1.5E-06 BOD O2 -1.4E-04 8.5E-05 1.9E-05 1.9E-05 Dissolved Solids Ash 0.81 0.00 0.19 4.33 Lead Pb 9.4E-079.4E-07 9.4E-07 9.4E-07 Nickel Ni 9.4E-079.4E-07 9.4E-07 9.4E-07 Sulfides H2S 2.9E-02 2.8E-03 1.2E-02 1.5E-04 Zinc Zn 9.4E-079.4E-07 9.4E-07 9.4E-07
Continued
Table 5.5: Reconciled and estimated values of the process variables.
164 Table 5.5 continued
Solid Waste (SW) Solid Waste Ash 0.57 0.00 0.13 3.08 Missing Outputs Carbon (SW) C 102.1 0.3 -- -- Nitrogen (in AE) N2 0.06 0.00 1.5E-02 5.1E-01 Carbon Dioxide (AE) CO2 -- -- 491.1 566.0 Water (AE) H2O -- -- 401.4 438.5 Total b 145.6 2,016.7 2,910.3 3029.8 a Measured variables reconciled with equation 2.13 and unmeasured variables or missing flows estimated with equation 2.14; b Total represents the sum of the mass of all inputs or outputs (all Material balances are satisfied).
Table 5.5 shows the results for the data reconciliation problem. The mass flows of coal and residual oil are still close to zero. By looking at Table 5.6, it is noticed that gross errors in the main process sub-module have gotten slightly worse as compared to alternative 2, with hydrogen now showing gross errors. This is expected because the incorporation of the chemical reaction imposes a fixed ratio of hydrogen and oxygen entering the system, i.e. 2 moles of atomic hydrogen (H) per mole of atomic oxygen (O).
As for the energy sub-module, coal, residual oil, ash, sulfur oxides and sulfides exhibit gross errors. There is an overall improvement that results from the addition of the water and carbon dioxide streams, because they relax the constraint of oxygen and hydrogen production. Smaller adjustments in the values of coal and residual oil are still constrained 165 by the production of sulfur and ash. Therefore, a last improvement to the inventory is attempted in alternative 4 by modifying the assumptions about the compositions of the fuels.
5.2.2.4 Alternative 4
This alternative is based on a modification of alternative 3. In this case, it is assumed that the compositions of coal and residual oil are such that their total amount of sulfur and ash completely matches the total amount reported by the inventory. For this purpose, two main modifications were made. First of all, it is assumed that residual oil does not contain sulfur, resulting in a composition of 15% hydrogen and 85% carbon, both values given on a mass basis. The second assumption is that the weight fractions of hydrogen, carbon, oxygen, nitrogen and water present in the coal stream are proportionally adjusted to match the changes made in the amounts of sulfur and ash.
Consequently, the chemical reactions R14 and R15, associated with the combustion of coal and residual oil, must be adjusted to match the changes in composition of these fuels. Other than that, there are no further changes with respect to alternative 3. The incidence matrix, A, for both sub-modules is also given by equation 5.15. The balance matrix, B, and the master stoichiometric matrix, S, are shown in Appendix C. The resulting forms of chemical reactions R14 and R15 are respectively
−5 1.00 Coal + 0.67 O2 → 1.06 CO2 + 0.50 H 2O + 2×10 SO2 + 0.01 N 2 + 0.11 Ash
(5.16)
166 Chemical Power Measurement Test Statistics, Alternatives a Stream Formula 1 2 3 4 Raw Materials, Salt NaCl 40.73 3.85 4.04 4.04 Energy Inputs Electricity ------Natural gas CH4 5.20 13.40 0.00 0.00 Coal -- 7.378.25 8.25 0.00 Residual oil -- 21.02 2.11 8.68 0.00 Products Caustic soda NaOH 31.50 2.29 2.11 2.11 Chlorine Cl2 29.76 1.83 1.67 1.67 Hydrogen H2 5.20 0.00 2.37 2.37 Air Emissions (AE) Mercury Hg 0.00 0.00 0.00 0.00 Chlorine Cl2 29.76 1.83 1.67 1.67 Particulates Ash 2.23 2.72 2.62 0.00 Sulfur oxides SO2 19.23 0.63 8.68 0.00 Water Effluents (WE) Mercury Hg 0.00 0.00 0.00 0.00 BOD O2 59.64 1.40 0.00 0.00 Dissolved Solids Ash 2.23 2.72 2.62 0.00 Lead Pb 0.000.00 0.00 0.00 Nickel Ni 0.000.00 0.00 0.00 Sulfides H2S 20.02 1.40 8.68 0.00 Zinc Zn 0.000.50 0.00 0.00 Solid Waste (SW) Ash 2.23 2.72 2.62 0.00 Objective Function b 3648.1 271.7 96.6 20.4 a Equation 2.17 (values exceeding 1.96 indicate that measurement contains gross errors); b Optimized values of the objective function given by equation 2.3 (it represents the minimum of the sum of the normalized squared errors), where wi is the reciprocal of the measured value yi.
Table 5.6: Values of the gross error detection test and the optimized objective function.
167 and
1.00 Residual Oil +1.81 O2 → 1.47 CO2 +1.34 H 2O (5.17)
Table 5.5 shows the results for the data reconciliation problem. Results for the main process sub-module remain the same as for alternative 3. On the other hand, the reconciled measured variables in the energy sub-module do not change at all from their measured values. The reason is that when the fuel composition constraints are relaxed, the system is no longer redundant. The problem is no longer reconciliation, but estimation instead. By looking at Table 5.6, it can be verified that there are no errors in the energy sub-module variables.
5.2.3 Exergy Analysis
According to the algorithm described by Figure 5.4, the approach proposed here is an iterative process. When the reconciled inventory is satisfactory, the procedure advances to the next phase, which is to verify satisfaction of the second law of thermodynamics by performing exergy analysis on the reconciled data. Nevertheless, for illustrative purposes, exergy analysis is performed on all the alternatives. Table 5.7 shows the resulting values for exergy loss and exergetic efficiency of the process inventory alternatives. Exergy losses include the exergy of the emissions, which are normally regarded as external exergy losses. Exergetic efficiency was calculated with equation
2.48. The second column of Table 5.7 shows a reported value for this process (Szargut et al., 1988). If the reported exergetic efficiency represented the maximum, then all the
168 alternatives would have to be adjusted to satisfy the second law of thermodynamics. If this value represented an average, then all the alternatives would represent processes more efficient than the average.
From this example, it is clear that the inventory that seems most reasonable is that given by alternative 4. Adjustments in the inventory may be made to match the exergetic efficiency with the value reported by Szargut et al. (1988), but it is not necessary if there is assurance that the inventory does not violate the second law. A remarkable additional benefit of reconciling the inventory is that it accounts for missing streams. As shown in
Table 5.5 for alternative 4, the inventory was not showing possible emissions of carbon dioxide. According to these results, emissions of carbon dioxide amounts to approximately half the weight of the caustic soda produced. The missing inputs are also important because their life cycle also contributes to environmental impact and should be taken into account.
Reported Alternatives
Value a 1 2 3 4 b Exergy Loss (MJ/kg NaOH) 39.5 2065.5 1.2 9.1 10.8 Exergetic Efficiency (%) c 14.1 29.7 81.4 41.3 37.1 a Value obtained from Szargut et al. (1988, example 5.1, pp. 183); b this value is the exergy of the inputs subtracted the exergy of the products and normalized for 1 kg of caustic soda; c equation 2.48.
Table 5.7: Values of exergy loss and exergetic efficiency of the process inventory alternatives.
169 5.3 Summary
The performance of the LCA will strongly depend on the quality of its process inventory data. Obtaining good quality data can be quite expensive and even inaccessible.
Consequently, companies solve this shortcoming by using general publicly available databases. Since these databases collect information from different sources, disagreement is very common. Process inventories may be outdated, unreliable, incomplete and/or unverifiable (Ayres, 1995). LCA has no utility if physical data is wrong with respect to critical pollutants (Ayres, 1995). Schaltegger (1997) argues that from an economic point of view current LCAs provide a small potential benefit given the high probability of potentially wrong decisions and high costs.
This Chapter introduced an approach for data rectification of process inventories to satisfy material balance and the laws of thermodynamics, thus enhancing the quality of
LCI databases. The proposed approach improves the quality of process inventories by adjusting the values of the reported variables to satisfy the mass balance. This is carried on using techniques for data rectification. These techniques also provide the means for estimating missing data. No deep knowledge of the theory behind these approaches is required, making this approach more accessible to the general community of Life Cycle
Assessment analysts. The example of the data rectification of a process inventory for the production of caustic soda and chlorine illustrates the strength and challenges of implementing such techniques. This methodology also performs exergy analysis to test for second law violations. This is applied over the reconciled data and primarily verifies that the exergy inputs are sufficient to drive the process. Data rectification with the
170 second law constraint can be achieved by increasing the input flows of energy. A higher benefit could be obtained from this approach if the companies start reporting values for their processes exergy losses and exergetic efficiency so they can be used as guideline. In the meantime, attempts are being made to have a database of reported exergy data for industrial processes (Szargut et al., 1988; U.S. DoE, 2004).
171
CHAPTER 6
EVALUATING ECOLOGICAL AND ECONOMIC TRADE-OFF OF PROCESSES
Give no decision till both sides thou’st heard.
–Phocylides
This Chapter provides unique insight into the trade-off between ecological and
economic aspects of manufacturing processes when they are analyzed at multiple spatial
scales. Economic aspects are accounted for via traditional cost analysis and do not change
at different scales. Ecological factors are considered via exergy analysis of the inputs at
each scale, and depend on the selected processes. The scales considered in this analysis
correspond to existing methods at the manufacturing, value chain, economy and
ecosystem scales. The finest equipment or manufacturing scale corresponds to a
traditional exergy analysis (Szargut et al., 1988). The trade-off between ecological and
economic aspects is usually quite large at this scale because the narrow manufacturing
boundary ignores the environmental impact of other processes in the life cycle that incur
an economic cost such as capital equipment. This next coarser scale is at the value chain
scale and corresponds to a cumulative exergy consumption or exergetic life cycle
analysis. The boundary is further expanded to include activity in the entire economy by 172 combining exergy analysis with economic input-output analysis (Ukidwe and Bakshi,
2003,2004). Finally, contributions from ecosystem goods and services are included at the coarsest ecosystem scale (Hau and Bakshi, 2004a; Ukidwe and Bakshi, 2003, 2004). The multiscale approach is closely related to existing hybrid LCA methods (Suh et al., 2004), but represents inputs and outputs in terms of cumulative exergy consumption (CEC)
(Szargut et al., 1988). The proposed methodology considers two objective functions: economic cost and exergetic efficiency of the process life cycle at multiple scales. Trade- off between these objectives is represented via a series of Pareto optimal surfaces at various scales, thus avoiding arbitrary combinations until the final stages of decision making. Case studies of a heat exchanger and the CGAM cogeneration system (Valero et al., 1994) compare the proposed approach with existing methods, and highlight the benefits of adopting a multiscale and multiobjective view.
Such a hierarchical and multiobjective analysis of ecological and economic aspects shows that expanding the analysis boundary shrinks the disparity between ecological and economic aspects if consideration of more processes will introduce aspects in ecological analysis that are already internalized in economic analysis. Thus, expanding the boundary to include environmental impact of equipment decreases the disparity between ecological and economic aspects since equipment costs are already included in economic aspects. However, if expanding the boundary results in consideration of environmental impacts that are economic externalities, then the disparity between the two objectives will increase. This would imply that market prices do not capture ecological aspects that increase the disparity. Addressing such disparity should be
173 of interest to those decision makers interested in quantifying natural capital needed for assessing weak and strong sustainability. It should also be relevant to corporations interested in balancing short- and long-term costs since it is very likely that future market prices will reflect ecological aspects more accurately via instruments such as restrictions on Carbon emissions and new regulations. Such analysis is also expected to be essential for developing ecologically and economically conscious methods and heuristics for engineering design.
6.1 System Boundaries and Scales of Analysis
The old ways of pollution prevention in Process Design concentrated on recycling, treatment and disposal (Turton et al., 1998). Although valuable, these practices are not cost-effective and likely to lead to ecologically suboptimal designs because their limited scope cannot prevent shifts of environmental problems to out-of-reach domains.
For instance, the wet limestone scrubbing process for removal of sulfur dioxide is an end- of-pipe solution for the problem of acidification, but it creates a new problem of global warming associated with the life cycle of limestone (Azapagic, 1999). These practices, which are focused on remediation, not only require additional spending, but they may also indirectly generate additional pollution. Preventive strategies are more promising because they reduce pollution by minimizing the use of pollution sources, thus eliminating the need of remediation. Since it is of interest to reduce environmental impact globally, the boundaries of analysis in process design must be greatly expanded to capture and optimize these effects. Expanding the analysis boundaries, however, adds a
174 great challenge to traditional process design because it greatly increases the amount of data required.
The issue of expanding the analysis boundaries has been widely discussed in LCA
(ISO 14040, 1997; Suh et al., 2004). For instance, a LCA of the industrial process shown on the left side of Figure 6.1 would require, among other things, expanding the analysis boundaries to include all the processes in the supply chain. The small circles in Figure 6.1 represent individual processes added to the supply chain. Noticeably, there is a limit to the number of individual processes that can be included in the life cycle of the process under consideration. Each process incorporated in the life cycle means a whole new cluster of data that must be collected. The deviation from the true value caused by not considering all the processes that actually form part of the value chain is regarded as the truncation error. Treloar (1997) and Lenzen (2001) have shown that truncation errors in
LCAs can be larger than 50%, which clearly indicate that they cannot be ignored.
In practice, data and models are available at multiple spatial scales ranging from individual equipment and processes, to the supply and demand chains, to the economy and ecosystem. An industrial sector, for instance, is a conglomerate of similar processes, as depicted in Figure 6.1. In the same figure, the shaded areas with dashed-lined borders enclose the individual processes to their corresponding sectors. Data handled in macroeconomics and other large-scope fields are typically aggregated by sectors or other kind of conglomerates. Incorporating this information, to complete the life cycle of the process under consideration, definitely tackles the problem of truncation errors, but it increases the uncertainty of the results because of the use of lower quality aggregated
175 data. As a matter of fact, the quality of the data collected for the individual processes in the supply chain is also expected to be lower than that of the process studied. Evidently, there is a hierarchy of information quality determined by the spatial scales at which the data is collected. The bottom line is that although these scales provide useful sources of data input, their sharp quality and uncertainty differences can drag the results down to a point were they become meaningless and highly unreliable. Suh et al. (2004) review the most relevant techniques for combining various scales in LCA.
Financial Chemicals Sector Sector
Transportation Sector
Petroleum Refining Sector
Mining Sector Machinery Sector
Figure 6.1: Life Cycle of an industrial process.
176
Holistic Ecosystem Coarser Economy Value Chain Manufacturing
Ecological Residual Primary Primary Main Residual Ecological Goods & Supply Supply Demand Process Demand Services Services Chain Chain Chain Chain & Impacts
Uncertain Aggregated
Figure 6.2: Characteristics of data and models at different scales of analysis.
Figure 6.2 illustrates how data and models may vary at different scales of analyses. The Equipment scale encloses only the plant under design, which is the level at which decisions are required. It is characterized as having the finest resolution, with the most disaggregated and precise information. Details at this scale are greatly transparent.
Models at this scale include those of traditional process design (Douglas, 1988; Turton et al., 1998): cost-benefit analysis, processes simulation software, etc. The Life Cycle scale includes the most relevant processes in the production chain to which the plant is part of.
Models at this scale are less detailed than at the equipment scale, often considering just a
177 few critical characteristics of the units. Data sources at this scale can be obtained from handbooks of industrial processes, life cycle inventories and public databases such as
Riegel’s Handbook of Industrial Chemistry (Riegel, 1992), LCAccess (EPA, 2004),
SimaPro (PRé, 1990) or Franklin Associates’ U.S. LCI Database (Franklin Associates,
1974). The sort of information provided by these sources is typically limited to material and energy streams of the plant and other relevant processes in the supply chain. The
Economy scale encompasses the entire U.S. economy. Data is typically disaggregated by industrial sectors and/or states. Models at this scale include economic input-output tables and other data from the Bureau of Economics, Toxic Release Inventory from EPA,
Economic Input-Output LCA, Thermodynamic Input-Output tables from Ukidwe and
Bakshi (2004), etc. The Ecosystem scale incorporates all the natural systems that sustain the whole economic activity. These natural systems provide goods such as water, coal and air; services such as wind, photosynthesis and evaporation; and impact absorption such as loss of biodiversity and disappearance of wetlands. Models at this scale include case studies in emergy analysis, simulations of atmospheric processes such as global climate change, and Thermodynamic Input-Output tables (Ukidwe and Bakshi, 2004).
Most thermodynamic methods only use the equipment and life cycle scales.
Emerging hybrid LCA approaches combine information at the equipment, life cycle and economy scales (Suh et al., 2004). Based on these hybrid techniques, this Chapter presents a multiscale approach for systematically utilizing available information at all these scales and providing the most comprehensive scenario on which process alternatives can be tested and better decisions at the plant scale be made. Hence,
178 structuring the problem by scales allows classifying sets of data with similar characteristics.
6.2 Methodology
The proposed approach evaluates the process hierarchically, starting from the process itself and gradually expanding the system boundaries as successive scales are added. As discussed by Douglas (1988), a hierarchical approach facilitates the evaluation procedure by starting with simple systems and increasing complexity gradually as successive layers of information are added. It may lead to the recognition of a general pattern that could be used as a basis for evaluating new processes. It also allows screening out bad alternatives at earlier stages thus saving time and computational power.
Then, for instance, alternatives with negative economic potential can be ruled out before evaluating its life cycle. Other benefits of a hierarchical approach include tracking down truncation errors, facilitating uncertainty analysis and providing the whole range of alternatives to the decision makers.
This Chapter describes the methodology by assuming that there are four well- defined scales: Equipment, Life Cycle, Economy, and Ecosystem, denoted S0, S1, S2 and S3 respectively. Two objective functions are evaluated: monetary cost and exergetic efficiency of the system. In order to maintain the general structure of existing thermodynamic methods, the methodology focuses on the supply chain and excludes the demand chain. Nevertheless, this approach can be applied with more than four scales if
179 the sources of information are more diverse. Moreover, it can be extended to include the demand chain as well as other objective functions.
6.2.1 Level 1 – Manufacturing Scale: Traditional Exergy Analysis
This first level of analysis considers only the main process, which is the process of interest. Figure 6.3 shows a systematic way of representing the system as an exergy flow network. The units classified as inputs represent the raw materials and utilities required for the process, e.g. water and electricity. Similarly, the units classified as final products represent the products of the process, including waste and emissions. The process units represent the main equipment or subsystems. Transactions between the process units are also indicated. Data at the equipment scale is for most cases obtained through traditional process design (Turton et al., 1998; Douglas, 1988). This network representation is not unique but it is convenient because it facilitates the interpretation of results in matrix algebra. An alternative network representation is described in Chapter 4 or in Hau and Bakshi (2004a).
Let i and j be any process unit of the system S0 shown in Figure 6.3,
(0) (0) corresponding to the process or equipment level. Then, Bij and Cij are respectively the exergy and cumulative exergy transferred from the i-th to the j-th unit at the equipment level S0. Notice that the superscript indicates the level at which the streams are defined.
(0) (0) Similarly, Bn,i and Cn,i are the external exergy and cumulative exergy inputs to the
(0) (0) equipment level S0. B p,i and C p,i are the external exergy and cumulative exergy outputs
180 from the i-th unit at the equipment level S0. Based on this network representation,
(0) cumulative exergy of the process at the equipment level, denoted C p , is
(0) (0) (0) C p = ∑ Bn,i = ∑C p,i (6.1) i∈S0 i∈S0
(0) Exergetic efficiency of the process at the equipment level, denoted η p , is
S0 Process Final Inputs Units Products (0) (0) (0) (0) C (0) (0) Bn,1 B12 , C12 B23 , 23 B p,3 (0) 1 2 3 C (0) Cn,1 p,3
(0) (0) Bn,4 B p,6 (0) 4 5 6 (0) Cn,4 C p,6
(0) (0) Bn,7 B p, j (0) (0) 7 (0) (0) i (0) (0) j Cn,7 C p, j B7i , C7i Bij , Cij M M M N (0)-2 N (0)-1 N(0)
Figure 6.3: Network representation of a system at the Manufacturing Scale (B is exergy and C is cumulative exergy).
181 (0) ∑ B p,i (0) i∈S0,U η p = (0) , S0,U = {}i ∈ S0 | p,i is useful (6.2) ∑ Bn,i i∈S0
It is also possible to determine how cumulative exergy of the process is allocated among its products. Applying the network algebra based approach described in Chapter 4
(0) (0) and by Hau and Bakshi (2004a), the transaction matrix γ with coefficients γ ij can be calculated as
(0) (0) Bij γ ij = (0) (0) , ∀i, j ∈ S0 (6.3) ∑ Bij + B p,i j∈S0
(0) The diagonal matrix γ p can also be determined with its diagonal coefficients given by
(0) (0) B p,i γ p,i = (0) (0) , ∀i ∈ S0 (6.4) ∑ Bij + B p,i j∈S0
(0) (0) The vector of cumulative exergy of the outputs Cp , with coefficients C p,i , is given by
(0) (0) (0) T −1 (0) Cp = γ p ⋅ ()I − (γ ) ⋅ Bn (6.5)
(0) (0) where Bn is the vector of the exergy of the inputs with the coefficients Bn,i . The
(0) cumulative exergy of the process C p can also be calculated as
182 (0) (0) C p = ∑C p,i (6.6) i∈S0
6.2.2 Level 2 – Value Chain Scale: Cumulative Exergy Consumption Analysis
The second coarser level of analysis considers the main process and relevant processes in the supply chain. Figure 6.4 shows a network representation of the system at
(0) the life cycle level. N is the total number of process units in S0. The units from i =
N(0)+1 to N(1) represent the added units. These units do not typically have the same degree of details as the units at the manufacturing scale. For instance, a unit may correspond to a whole process. In addition, although not indicated in Figure 6.4, the added process units may have final products leaving the system without entering the main process. For a fair assessment, it is necessary to consider all the processes upstream of the main process, this is, from the processes that extract the natural resources. However, this can be quite challenging since many materials involve innumerable production steps.
(1) (1) The transaction matrix γ with coefficients γ ij can be calculated with equation
6.3 by replacing the super index (0) by (1) and S0 by S1. Similarly, the coefficients of the
(1) (1) diagonal matrix γ p and the vector of cumulative exergy of the outputs Cp can be determined with equations 6.4 and 6.5 by replacing the super index (0) by (1) and S0 by
S1.
(1) The cumulative exergy of the process C p can be calculated with equation 6.6 by replacing the super index (0) by (1). Exergetic efficiency of the process at the value chain
(1) scale, denoted η p , is
183
S1 S0 Process Process Final Inputs Units Inputs Units Products (1) (1) (1) (1) C (1) (1) B (0) B12 , C12 B23 , 23 B p,3 n,N +1 N(0)+1 N (0)+2 (1) 1 2 3 (1) C (0) C p,3 n,N +1 (1) (1) B (0) B n,N +3 (0) N (0)+4 p,6 (1) N +3 4 (1) C 5 6 n,N (0) +3 C p,6
(1) (1) B (0) B n,N +5 N(0)+5 N (0)+6 p, j (1) (1) 7 (1) (1) i (1) (1) j C (0) C p, j n,N +5 B7i , C7i Bij , Cij M M M M M N(1)-1 N (1) N (0)-2 N (0)-1 N(0)
Figure 6.4: Network representation of a system at the Value Chain Scale.
(1) ∑ B p,i (1) i∈S0,U η p = (1) , S0,U = {}i ∈ S0 | p,i is useful (6.7) ∑C p,i i∈S0
6.2.2.1 Interface between Manufacturing and Value Chain Scales
As for the relation between the streams at the manufacturing and value chain scales, they essentially remain the same with the exception of the inputs and outputs crossing the interface. This is 184 (1) (0) Bij = Bij , ∀i, j ∈ S0 (6.8)
The input streams to the main process at the equipment level may get disaggregated at the Life Cycle level, as illustrated in Figure 6.5. For instance, natural gas used in the process may be a mixture from natural gas directly from the well and natural gas from a refinery. Similarly, part of the final products may be used by processes in the supply chain, thus splitting the stream.
(a) Equipment Scale B (0) n,1 1
(0) (1) (1) (1) Bn,1 = Bn,1 + B31 + B41 (b) Life Cycle (1) Scale 3 B31
(1) 1 B41 4 B (1) n,1
Figure 6.5: Disaggregation of exergy streams resulting from the expansion of system boundaries.
185 The relation of these streams between both levels of analysis can be expressed as
(0) (1) (1) Bn,i = ∑ B ji + Bn,i , ∀i ∈ S0 (6.9) j∈S1\S0
(0) (1) (1) B p,i = ∑ Bij + B p,i , ∀i ∈ S0 (6.10) j∈S1 \S0
where S1\S0 is the relative complement of S0 in S1, which is the subset of all the process units that belongs to the value chain S1 with the exception of those that are part of the main process S0.
6.2.2.2 Cumulative Exergy of Capital
Although exergy of capital investment is not defined, its cumulative exergy may be estimated by determining the exergy invested in manufacturing the equipment as done in ELCA (Cornelissen and Hirs, 2002). Including capital investment this way does not affect the approach described above, since the value chain of equipment manufacturing for the main process can be treated like any other process unit of the supply chain. Not including capital investment at this level is analogous to applying ICEC analysis (Szargut et al., 1988).
6.2.3 Level 3 – Economy Scale: Thermodynamic Input-Output Analysis
The Economy level of analysis attempts to estimate the cumulative exergy of the process by expanding the system boundaries to include the whole supply chain. Since it is intractable to include the whole supply chain by including each process as done at the life
186 cycle level of analysis, a different approach is used. Closely related to existing hybrid
(tiered) LCA methods (Suh et al., 2004), the rest of the economy is included by using cumulative-exergy-to-money ratios based on the ICEC analysis of the 1992 U.S
Economy carried out by Ukidwe and Bakshi (2004).
Let ξ1 ,ξ 2 ,K,ξ E , form a partition of the whole Economy represented by the set S-
2, i.e. ξ k is the k-th sector of the economy. Figure 6.6 shows the interaction of an economy sector with the system defined at the life cycle level. ξk\S1 symbolizes the set of all the processes that belongs to the economic sector ξk with the exception of those that are already taken into account at the life cycle level of analysis. Strictly speaking the sectors should exclude the units that comprise the set S1. However, if the processes included at the value chain scale does not contribute significantly to the cumulative exergy of the sectors where they belong, then their exclusion will not make a significant difference. This is
C (2) << C (2) for ξ ∈ S ⇒ C (2) = C (2) − C (2) ≅ C (2) (6.11) ∑ i ξk k 2 ξk \S1 ξk ∑ i ξk i∈ξk ∩S1 i∈ξk ∩S1
There are two kind of inputs to the system represented by the set S1: purchased and environmental. A purchased input comes from the industry sectors and can be
(1) represented in monetary units, for example by its cost Z n,i . An environmental input comes directly from ecological processes and therefore has a monetary cost of zero. The
(1) cost of a purchased input, Z n,i , is composed of money transactions from various industry sectors. This is
187 C ξki
S0 C iξ ξk\S1 k S1
Figure 6.6: Exergy exchange between an economic sector, ξk and the system at the value chain scale, S1.
Z (1) = Z (2) , ∀i ∈ S = {}i ∈ S | n,i is purchased (6.12) n,i ∑ ξ k i 1,P 1 ξ k ∈S2 where Z (2) is the monetary transaction from the sector ξ to the i-th process unit. It is ξki k possible to assume that the ratio of cumulative exergy to money transacted from the sector to the unit is approximately equal to the ratio of the total throughput of cumulative exergy to the total monetary activity of that sector, this is
C (2) C (2) ξki ≅ ξk = ρ (2) (6.13) Z (2) Z (2) ξk ξk i ξk
188 where Z (2) is the monetary activity of the sector and ρ (2) is its ICEC-to-money ratio. ξk ξk
The accuracy of this approximation depends on the level of disaggregation of the economy and on the variety of products in each sector. Based on this assumption,
(2) Cumulative exergy of purchased inputs to S1 at the economy level Cn,i can be estimated as
C (2) = (ρ (2) ⋅ Z (2) ), ∀i ∈ S = {i ∈ S | n,i is purchased} (6.14) n,i ∑ ξ k ξ k i 1,P 1 ξ k ∈S2
(2) Cumulative exergy of environmental inputs to S1 at the economy level Cn,i equals their exergy value. This is
(2) (2) Cn,i = Bn,i , ∀i ∈ S1,E = {}i ∈ S1 | n,i is environmental (6.15)
(2) (2) The vector of cumulative exergy of the outputs Cp , with coefficients C p,i is given by
(2) (1) (1) T −1 (2) Cp = γ p ⋅ ()I − (γ ) ⋅Cn (6.16)
(2) (2) where Cn is the vector of the exergy of the inputs with the coefficients Cn,i .
(2) The cumulative exergy of the process C p can be calculated with equation 6.6 by replacing the super index (0) by (2). The exergetic efficiency of the process at the
(2) economy scale η p can be calculated as
189 (1) ∑ B p,i (2) i∈S0,U η p = (2) , S0,U = {}i ∈ S0 | p,i is useful (6.17) ∑C p,i i∈S0
6.2.3.1 Cumulative Exergy of Capital
If capital investment had been included at the life cycle level, then the approach explained above would include capital investment at the economy level. An alternative way to include capital investment and labor inputs is by using the cost analysis performed on the main process, in which case the supply chain for the manufacturing of equipment cannot be considered. The cumulative exergy of every expense associated with capital investment can also be estimated with equation 6.14.
6.2.4 Level 4 – Ecosystem Scale: Thermodynamic Input-Output Analysis
Accounting for nature’s contribution is critical for determining the impact and environmental performance of industrial activities. Ecological goods and services constitute the productive base that is essential for all industrial and economic activity.
Examples of ecological goods include water, fertile soil, wood, and coal, while examples of ecological services include, rain, pollination, carbon sequestration and wind. The
Ecosystem level of analysis attempts to estimate this contribution. As shown in Chapter 4 or by Hau and Bakshi (2004a), this contribution can drastically change the outcome of a decision, especially in comparative analyses, because many natural resources that are easier to harness have already undergone transformations in which nature has taken a low quality raw material into a high quality product. 190 The approach taken at the ecosystem level of analysis is similar to that at the economy level, in which the rest of the economy and ecological processes are included by using cumulative-exergy-to-money ratios based on the ECEC analysis of the 1992 U.S
Economy carried out by Ukidwe and Bakshi (2004). All natural resources are created from the exergy contained in the energy sources supplying the Earth. For reasons explained in Chapter 4 or by Hau and Bakshi (2004a) and Ukidwe and Bakshi (2003,
2004), ECEC of natural resources follow an allocation scheme in which co-products are assigned the total throughput of the process, as opposed to a fraction proportional to their relevance. As a consequence, environmental inputs in an ECEC analysis cannot be added arbitrarily because there is a risk of double counting. Instead, ECEC of all co-products are compared and the maximum value is taken. Renewable resources are co-products because there are produced from the current inflow of exergy to the Earth. Non- renewable resources, on the other hand, are not considered to be co-products because they were created in the distant past and at different time scales. As a result, while ECEC of non-renewable resources can be added, ECEC of renewable resources cannot.
(3) Cumulative exergy of all process inputs of S1 at the ecosystem level Cn,i can be estimated as
(3) (3) (3) Cn,i = C NR,i + CRR,i , ∀i ∈ S1 (6.18)
(3) (3) where CNR,i and CRR,i are the ECEC of the process inputs entering the i-th process unit and coming respectively from non-renewable and renewable resources. For purchased inputs, these are determined as
191 C (3) = (ρ (3) ⋅ Z (2) ), ∀i ∈ S = {i ∈ S | NR,i is purchased} (6.19) NR,i ∑ NR,ξk ξki 1,P 1 ξk ∈S2
C (3) = max{ρ (3) ⋅ Z (2) }, ∀i ∈ S = {i ∈ S | NR,i is purchased} (6.20) RR,i RR,ξk ξki 1,P 1 ξk ∈S2 where ρ (3) and ρ (3) are the ECEC-to-money ratios coming respectively from non- NR,ξk RR,ξk renewable and renewable resources for the ξk–th economy sector (Ukidwe and Bakshi,
(3) (3) 2004). For environmental inputs, C NR,i and CRR,i can be estimated as explained in
Chapter 4 or by Hau and Bakshi (2004a)
(2) (2) (3) ⎧τ n,i ⋅ Bn,i NR,i is non renewable C NR,i = ⎨ , ⎩ 0 NR,i is renewable (6.21)
∀i ∈ S1,E = {}i ∈ S1 | NR,i is environmental
(3) ⎧ 0 NR,i is non renewable CRR,i = ⎨ (2) (2) , ⎩τ n,i ⋅ Bn,i NR,i is renewable (6.22)
∀i ∈ S1,E = {}i ∈ S1 | NR,i is environmental
(2) where τ n,i is the solar transformity, as defined by Odum (1996), of the process inputs
(3) entering the i-th process unit. The vector of cumulative exergy of the outputs Cp with
(3) coefficients C p,i can be calculated as
(3) (3) (3) Cp = Cp,NR + Cp,RR (6.23)
(3) with Cp,NR given by
192 (3) (1) (1) T −1 (3) Cp,NR = γ p ⋅ ()I − (γ ) ⋅CNR (6.24)
(3) and the coefficients of the vector Cp,RR given by
−1 C(3) = max γ (1) ⋅ I − (γ (1) )T ⋅C(3) , ∀i ∈ S (6.25) ()p,RR i {( p ( )RR )ij } 1 j∈S1
(3) (3) (3) where CNR is a vector with the coefficients CNR,i and CRR is a diagonal matrix with the
(3) (3) coefficients CRR,i . The cumulative exergy of the process C p is then
C (3) = C(3) + max C(3) (6.26) p ∑( p,NR )i {( p,RR )i } i∈S0 i∈S0
(3) Similarly, the exergetic efficiency of the process η p can be calculated as
(1) ∑ B p,i (3) i∈S0,U η p = (3) , S0,U = {}i ∈ S0 | p,i is useful (6.27) C p
The same approach as in the economy level is used for including capital investment. The ECEC of every expense associated with capital investment can be estimated with equation 6.18.
6.2.5 Evaluating the Ecosystem-Economy Trade Off
The proposed methodology considers two objective functions: monetary cost and exergetic efficiency. The optimization problem is solved at each level of analysis. Trade- off between these objectives is represented via a series of Pareto optimal surfaces at 193 various levels, thus avoiding arbitrary combinations until the final stages of decision making.
Since the cost of the plant is invariant to the scale of analysis, the economic objective is the same for every scale, this is
(s) (0) (0) Z p = Z p = ∑ Z n,i (6.28) i∈S0
Given the m ×1 vector of decision variables X, the design problem is to minimize economic cost and cumulative exergy of the process at every scale of analysis, this is
Find X to
(0) Minimize Z p (6.29)
(s) Maximize η p (6.30)
S.t. X = {X ∈ ℜm | X is feasible} (6.31)
The technique used to solve the multiobjective optimization problem in this approach is known as Data Envelopment Analysis (DEA). DEA was initially created to compare the relative efficiency of different production units in economics, such as banks, mail centers and manufacturing plants, referred as decision making units (DMU)
(Charnes et al., 1993). These efficiencies range from zero to unity. A DMU is Pareto- efficient if its efficiency is unity. The set of Pareto-efficient DMUs forms what is called the efficient frontier, i.e. a surface where the improvement of an objective is only possible
194 in detriment of the others. It is important to notice that the efficient frontier is not necessarily the Pareto optimal surface. Before DEA can be applied to the multiobjective problem encountered here, the DMUs have to be produced. This can be done by creating a set of discrete alternatives from the feasible region. Whether the efficient frontier and the Pareto optimal surface coincide depends on the extent of the set and the resolution of the grid on the feasible region. Among the advantages of using DEA to solve multiobjective optimization problems are that the decision maker does not have to establish neither a preference structure nor prior explicit targets on the multiple objectives
(Cabrera-Rios, 2002).
There are various models in DEA (Charnes et al., 1993). The model used here is known as the input oriented Banker-Charnes-Cooper (BCC) Model, designed to deal with variable returns-to-scale problems. The design problem can then be formulated as
+ − Find µ,ν,µ0 , µ0 to
+ − Maximize µY0 + µ0 − µ0 (6.32)
S.t. νX 0 = 1 (6.33)
+ − µY j − νX j + µ0 − µ0 ≤ 0 j = 1,K,n (6.34)
µ,ν ≥ ε (6.35)
+ − µ0 , µ0 ≥ 0 (6.36)
195 where µ and ν are multipliers for the economic and the exergetic functions, respectively.
+ − µ0 and µ0 are scalar variables whose value represents an intercept, n is the number of
DMUs or discrete elements of the feasible set, and ε is a non-Archimedean infinitesimal required to prevent certain suboptimal solutions from being part of the efficient frontier.
The economic and the exergetic functions, respectively Yj and Xj, are defined as
(s) Y j = η p,j (6.37)
(0) X j = Z p,j (6.38)
The optimization problem is solved for each DMU at each level of analysis, this is
4× n times. More details on DEA can be found in Charnes et al. (1993).
6.3 Case Studies
The following section illustrates the proposed approach via two case studies. The first case study is concerned with the design of a heat exchanger intended to cool down a stream. The second case study is the well known CGAM cogeneration system (Valero et al., 1994). These examples compare the proposed approach with existing methods, and highlight the benefits of adopting a multiscale and multiobjective view.
In both case studies, total cost and exergetic efficiency are optimized simultaneously for each tier of analysis. The ecological objective optimum at the equipment level is equivalent to exergy analysis. The ecological objective optimum at the life cycle level is equivalent to a traditional cumulative exergy consumption analysis
196 (Szargut et al., 1988). As in exergy and cumulative exergy consumption analyses, these first two tiers do not consider capital investment. There are no restrictions about including capital investment in these levels, however, more assumptions would be required. The ecological objective optimum at the economic tier of analysis is similar to hybrid LCA (Suh et al., 2004). The ecological objective optimum at the ecosystem tier of analysis is not comparable to any existing methods in thermodynamics. At the economy and ecosystem levels, capital investment is considered.
Cumulative exergy analysis at the economy and ecosystem levels uses the ICEC- to-money and ECEC-to-money ratios of the economic sectors supplying the inputs. These ratios are obtained from Ukidwe and Bakshi (2004). These values have been calculated with the economic input-output tables provided by the Bureau of Economics and are representative of the U.S. economy in 1992 (Ukidwe and Bakshi, 2004).
6.3.1 The Heat Exchanger
This case study consists of the design of a heat exchanger required to cool a stream of benzene at a mass flow rate of 20 kg/s from a 70ºC to 55ºC. A shell-and-tube heat exchanger with benzene flowing in the shell side is suggested. Figure 6.7 illustrates the design problem. The coolant to be used is water entering the tube side at 17ºC.
To simplify the design problem, it is specified that the heat exchanger type is a carbon steel floating head with one shell pass and two tube passes, a baffle cut of 25%, a square pitch-tube layout with a pitch size 25% larger than the tube outside diameter, and a thermal conductivity of the tube material of 60 W/m2·K. The maximum tube length
197 allowed is 5m. Moreover, the water enters a cast iron centrifugal pump at 1.013 bars and leaves the exchanger at 1.013 bars. The decision variables selected are the coolant mass flow rate m& t , the shell inside diameter Ds, and the tube outside diameter do. In order to comply with commercial standards and to restrict the feasible region, the shell inside diameter cannot exceed 31 inches and the tube outside diameter must be 1, 1¼ or 1½ inches.
Correlations have also been specified to determine the rest of the geometrical parameters of the heat exchanger as a function of the decision variables. The number of tubes Nt can be approximated to the nearest integer of
Figure 6.7: Flow Diagram of the Heat Exchanger system.
198 −2 2 −1 ⎧ 0.5096in Ds − 3.1678in Ds +12.8510 for d 0 = 1in ⎪ −2 2 −1 1 N t ≈ ⎨0.2863in Ds −1.8218in Ds + 6.9046 for d 0 = 1 4 in (6.39) ⎪ −2 2 −1 1 ⎩0.2037in Ds −1.6130in Ds + 6.2321 for d 0 = 1 2 in
The baffle spacing B is taken as the arithmetic mean of the recommended limits, this is
3.189in + 74in 0..25d 0.75 B = o (6.40) 2
The Kern Method (Kakaç and Liu, 2002, pp. 307) was used to determine the heat transfer coefficient and pressure drop of the shell side. Gnielinski’s correlation (Kakaç and Liu, 2002, pp. 313) was used for the analysis on the tube side. Cost functions are obtained from Turton et al. (1998). More details are available in Appendix D.
6.3.1.1 Hierarchical Analysis
Figure 6.8 shows the network representation of the heat exchanger system and its supply chain. The system boundaries for each level of analysis are indicated in the figure.
These are the manufacturing (S0), economy (S2) and ecosystem (S3) scales. Since there is no process data of any process unit in the value chain, the value chain scale, S1, is identical to S0. The supply chain for benzene is not included because it does not affect the design of the system under study, which is the cooling system for the benzene stream.
The system at the equipment level consists of the heat exchanger and the pump required for overcome the pressure drop on the tube-side, which is carrying the water stream. At the Economy level, the system boundaries are expanded to include capital investment and 199 the supply chain providing electricity to the pump. Figure 6.8 shows the system for this tier of analysis, where the dashed arrows represent capital investment inputs, which are provided by the sector of general industry machinery and equipment. Electricity is delivered from the sector of electric services. ICEC-to-money ratios for the required sectors are shown in Table 6.1. The second column of Table 6.1 shows the Standard
Industrial Classification (SIC) code of the sectors. At the ecosystem level, an exergetic efficiency of 1.18×10-3% has been used for water, which corresponds to a value of
40,000 sej/g as reported by Odum et al. (2000). ECEC-to-money ratios for the required sectors are shown in Table 6.1.
ICEC/Money ECEC/Money Economic Sector SIC (MJ/$) (MJ/$) Crude petroleum and natural gas 8 1452.16 3.69×107 Engines and turbines 43 835.26 4.18×106 General industry machinery and 49 1096.84 3.09×106 equipment Electric Services (Utilities) 68A 1010.51 5.20×106
Table 6.1: CEC-to-money ratios for the required economic sectors (Ukidwe and Bakshi, 2004).
200
S3 S2 S0 Process Final Inputs Units Products
Water (17ºC) Pump Hot Electric Electricity Water Services
Benzene Heat Benzene (70ºC) Exchanger (55ºC)
General industry machinery and equipment
Figure 6.8: Network representation of the Heat Exchanger system at multiple scales.
6.3.1.2 Results
Figure 6.9 shows the Pareto optimal surface for the first level of analysis: the manufacturing scale. The economic optimal solution, located at the lower left extreme of the Pareto surface, results in an exergetic efficiency of 99.9926% and a total cost of 4,137
$/yr. The ecological optimum suggests an exergetic efficiency of 99.9956% and a total cost of 20,565 $/yr. There is a small variation in efficiency because the exergy loss is small when compared to the exergy of the benzene stream. In thermoeconomics,
201 exergetic efficiency for a heat exchanger is defined in such a way that only the exergy change of the streams is considered. Applying such definition to this example would give a sharper difference of efficiency. However, the definition of efficiency used in this
Chapter is standard and more suitable for the multiscale approach. It can be observed in
Figure 6.9 that the ecological optimum at this tier of analysis results in the largest heat exchanger allowed by the constraints. Such result is due to ignoring exergy of capital inputs.
99.9960
99.9955 Ecological Optimum 99.9950 m& water = 3.0kg / s 99.9945 Ds = 31in
99.9940 Economic Optimum d 0 = 1.5in
CDP (%) m = 8.3kg / s 99.9935 & water
Ds = 13.25in 99.9930 d 0 = 1in 99.9925
99.9920 4 6 8 10 12 14 16 18 20 22 Cost (1,000 $/y)
Figure 6.9: Heat Exchanger – Optimal Pareto Curve at the manufacturing scale. 202 $99.9765 Ecological Optimum
$99.9760 m& water = 6.0kg / s
Ds = 12in
$99.9755 d 0 = 1in
Economic Optimum
CDP (%) $99.9750 m& water = 8.3kg / s
Ds = 13.25in $99.9745 d 0 = 1in
$99.9740 4.0 4.1 4.2 4.3 4.4 4.5 Cost (1000 $/y)
Figure 6.10: Heat Exchanger – Optimal Pareto Curve at the Economy scale.
Figure 6.10 shows the Pareto optimal surface for the third level of analysis: the economy scale. The economic optimal solution has an exergetic efficiency of 99.9755%.
The ecological optimum has an exergetic efficiency of 99.9760% at a total cost of 4,200
$/yr. The economic objective has not change because it is on the company’s interest to minimize the cost of the plant and not of the entire value chain. Consequently, all feasible solutions only suffer changes in their ecological objective, which explains why the minimum cost does not change its value from that at the equipment level of analysis. The
203 ecological objective at this level includes cumulative exergy of capital. Notice that the optimal curve at this level has shrunk sharply when compared to that of the equipment level. This is because cumulative exergy of capital investment has been included, which sets limits over designs that operate at maximum efficiency because they require more sophisticated equipment. Such equipment demand greater exergy consumption for their manufacture.
7.00 Ecological Optimum 6.00 m& water = 3.0kg / s
5.00 Ds = 8in d = 1in 4.00 0
3.00
CDP (%) Economic Optimum
2.00 m& water = 8.3kg / s D = 13.25in 1.00 s d 0 = 1in 0.00 0 5,000 10,000 15,000 20,000 25,000 30,000 Cost ($/y)
Figure 6.11: Heat Exchanger – Optimal Pareto Curve at the Ecosystem scale.
204 Figure 6.11 shows the Pareto optimal surface for the fourth level of analysis: the ecosystem scale. The economic optimal solution has an exergetic efficiency of ~2%. The ecological optimum has an exergetic efficiency of ~6% at a total cost of 4,137.5 $/yr. In addition to including capital inputs, this level also considers the exergy consumed by ecological processes to produce the natural goods and services that are necessary for the main process and its supply chain.
6.3.2 The CGAM Cogeneration System
The CGAM problem is a predefined optimization problem created by Valero et al.
(1994) to be used for comparison of diverse thermodynamic and economic methodologies. The system is a cogeneration plant operating at steady state that delivers
30 MW of electricity and 14 kg/s of saturated steam at 20 bars. Figure 6.12 shows a flow diagram of the CGAM problem. Air is passed through a compressor (AC) to achieve high pressure and then through a preheater (PH) to increase its temperature. The compressed air reacts with Natural gas (taken as pure methane) in a combustion chamber (CC). The combustion gases drive a turbine (GT) that produces energy to generate the 30MW of electricity and additional work to operate the air compressor. The combustion gases leaving the turbine are used as heating fluid for the air preheater and the heat-recovery steam generator (HRSG). The heat-recovery steam generator produces 14 kg/s of saturated steam at 20 bars from feedwater at 25ºC. It is composed of a preheater and an evaporator.
205
Figure 6.12: Flow Diagram of the CGAM cogeneration system (Valero et al., 1994).
The CGAM problem is too simplistic to be implemented as a design for a real cogeneration system, yet it is ideal to make an instructive optimization problem. The assumptions made by Valero et al. (1994) are:
206 • The air and the combustion gases behave as ideal gases with constant specific
heats.
• As mentioned before, the fuel (natural gas) is taken as methane. It is provided to
the combustion chamber at high pressure by throttling from a high-pressure
source. The price of natural gas is 2.78 $/GJ.
• All components, except the combustion chamber, are adiabatic. Heat transfer
from the combustion chamber is 2% of the fuel lower heating value.
• Constant values are chosen for the pressure loss of the air and gas flows in the
combustion chamber, air preheater and heat-recovery steam generator.
• Combustion in the combustion chamber is complete and Nitrogen is inert.
The decision variables selected for the optimization are the compressor pressure ratio (P1/P2), the compressor (ηAC) and turbine (ηGT) isentropic efficiencies, and the temperatures of the air at the preheater exit (T3) and the combustion gases at the exit of the combustion chamber (T4). Thermodynamic and economic data and models are provided by Valero et al. (1994) and they are all expressed as function of these decision variables. Constraints to these variables are P1/P2 cannot be higher than 16, ηAC must be lower than 90%, ηGT must be lower than 92%, and T4 cannot be higher than 1550 K.
The objective function presented in Valero et al. (1994) is the total monetary cost of the plant, which is the sum of the cost of fuel and the levelized cost of equipment. The optimization problem consists of minimizing this purely economic objective. Their results are presented in Table 6.2.
207 Variable P2/P1 ηAC (%) ηGT (%) T3 (K) T4 (K) Value 8.5234 84.68 87.86 914.28 1492.63
Table 6.2: Economic optimum values for the CGAM Problem (Valero et al. 1994).
6.3.2.1 Multiscale Analysis
Figure 6.13 shows the network representation of the CGAM cogeneration system and its supply chain. The system boundaries for each tier of analysis are indicated in the figure. These are the equipment (S0), life cycle (S1), economy (S2) and ecosystem (S3) levels. The Life Cycle level includes the process of extraction of the natural gas.
According to Szargut et al. (1988), this process unit has an exergetic efficiency of 87.5%.
Specifically, it requires 14.3 kJ in exergy of fuel per every 100 kJ in exergy of natural gas extracted. At a price of 4.25 $/GJ, Fuel oil No. 2 has been assumed to be the fuel used for natural gas extraction.
208 S3 Process S1 S0 Units
CC Inputs Final Products Extraction GT
Methane Electricity Fuel Oil AC
Air Steam PH
Water HRSG Emissions
Petroleum Engines General industry refining and and machinery and related products turbines equipment S2
Figure 6.13: Network representation of the CGAM cogeneration system at multiple scales.
As shown in Figure 6.13, the economic sectors involved are assumed to be petroleum refining and related products, engines and turbines, and general industry machinery and equipment. Petroleum refining and related products sector supplies the fuel oil to be used for the natural gas extraction. The dashed arrows represent capital costs being included. The engines and turbines sector supplies the gas turbine and the general industry machinery and equipment sector provides the rest of the equipment of 209 the cogeneration system. This is clearly a simplification because there are surely more sectors supplying inputs to the system. Nevertheless, this case still serves the purpose of illustrating how the connections are made between the life cycle and the economy scales.
Table 6.1 shows the values of ICEC-to-money and ECEC-to-money ratios for the required sectors. At the ecosystem level, exergetic efficiency air, water and natural gas has been taken as the reciprocal of their transformity. Such values correspond to 2,450 sej/J for air, 40,000 sej/g for water and 48,000 sej/J for natural gas (Odum et al., 2000).
6.3.2.2 Results
Figure 6.14 shows the Pareto optimal surface for the first tier of analysis: the equipment level. The economic optimal solution, located at the lower left extreme of the
Pareto surface, results in an exergetic efficiency of 50.31% and a total cost rate of 0.3620
$/s (10,425,234 $/yr). The ecological optimum converges asymptotically to an exergetic efficiency of 54.21%. This result suggests minimum consumption of natural gas
(methane) and acquisition of the most thermodynamically efficient equipment. The cost of both the air compressor and the gas turbine approaches infinity as they achieve their efficiency limit value, corresponding to 90 and 92% respectively. It is clear then that achieving the ecological optimum is not feasible in practice. Optimal solutions are typically the result of trade-offs between capital and operational inputs. The asymptotic behavior of the ecological optimum is a result of ignoring exergy of capital inputs. It is important to emphasize that the ecological objective at this tier of analysis corresponds to a traditional exergy analysis.
210 54.5
54.0 Ecological Optimum 53.5 P2 / P1 = 16
53.0 η AC = 90%
η GT = 92% 52.5 Economic Optimum T3 = 792 .00 K
52.0 T4 = 1550 K P2 / P1 = 8.54
51.5 η AC = 84.68% η = 87.86% 51.0 GT Exergetic Efficiency (%)Exergetic Efficiency T3 = 914.28K 50.5 T4 = 1492.63K 50.0 0 5 10 15 20 25 30 Total Cost ($/s)
Figure 6.14: CGAM Problem – Optimal Pareto Curve at the Equipment level.
Figure 6.15 shows the Pareto optimal surface for the second tier of analysis: the life cycle level. The economic optimal solution has not change its total cost rate, which is
0.3620 $/s, but it has lowered its exergetic efficiency to 44.02%. The ecological objective at this tier of analysis corresponds to a traditional CEC analysis (Szargut et al., 1988).
The ecological optimum converges asymptotically to an exergetic efficiency of 47.43%.
Although all feasible solutions are shifted to lower exergetic efficiencies due to the boundaries expansion, the set of optimal solutions for this case is identical to that at the 211 equipment level. This is explained by the fact that the expansion only incorporates units upstream of the natural gas inlet, whose exergy contributes almost entirely to the equipment-level cumulative exergy of the cogeneration plant. Consequently, the relation between the feasible solutions at the first and second tiers can be approximately established through one proportionality constant, i.e. the exergetic efficiency of the natural gas extraction unit.
47.5
47.0 Ecological Optimum 46.5 P2 / P1 = 16
46.0 η AC = 90%
η GT = 92% 45.5 Economic Optimum T3 = 792 .00 K
45.0 T4 = 1550 K P2 / P1 = 8.54
44.5 η AC = 84.68% η = 87.86% 44.0 GT Exergetic Efficiency (%) T3 = 914.28K 43.5 T4 = 1492.63K 43.0 0 5 10 15 20 25 30 Total Cost ($/s)
Figure 6.15: CGAM Problem – Optimal Pareto Curve at the Life Cycle level.
212 Given the idealized definition of the CGAM problem, the only serious emission is that of carbon dioxide which is directly proportional to the mass flow of methane. Thus it is fair to state that the ecological optimal also corresponds to the minimal environmental impact. As shown in Figure 6.15, this optimum suggests selecting the most efficient equipment, even if that means paying an infinitely high price for it and regardless of the resources consumed and emissions incurred in the manufacture of such equipment. The bottom line is that ignoring capital inputs can lead to perverse analysis or nonsensical results, thus imposing serious shortcomings on methods such as Exergy analysis, traditional CEC and process life cycle assessment.
Figure 6.16 shows the Pareto optimal surface for the third tier of analysis: the economy level. The economic optimal solution has an exergetic efficiency of 8.40%. The ecological optimum has an exergetic efficiency of 8.42% at a total cost rate of 0.3628 $/s
(10,449,460 $/yr). As opposed to previous tiers, the ecological objective at this level includes the exergy consumed in manufacturing the equipment. The ecological optimum shows a value that is much closer to the economic optimum. Although both extreme optima differ by only 24,226 $/yr, the ecological optimum still suggests investing in better equipment and reducing natural gas (methane) consumption.
Figure 6.17 shows the Pareto optimal surface for the fourth tier of analysis: the ecosystem level. The economic optimal solution has an exergetic efficiency of 3.52×10-
4%. The ecological optimum has an exergetic efficiency of 3.63×10-4% at a total cost rate of 0.3951 $/s (11,380,155 $/yr). Similar to the economy level, this tier also includes capital inputs. In addition, this level considers the exergy consumed by ecological
213 processes in order to produce the natural goods and services that are necessary for the main process and its supply chain. An interesting result is that the distance between the two extreme optima is wider at this level than at the economy level. The ecological optimum indeed suggests a design quite different to the most economical, with an additional cost of nearly one million dollars a year.
8.44 Ecological Optimum 8.42 P2 / P1 = 9.13 η = 84.99% 8.40 AC ηGT = 88.24% T = 904.93K 8.38 3 Economic Optimum T4 = 1501.86K 8.36 P2 / P1 = 8.54
η AC = 84.68% 8.34 ηGT = 87.86%
Exergetic Efficiency (%) T = 914.28K 8.32 3
T4 = 1492.63K 8.30 0.3615 0.3620 0.3625 0.3630 0.3635 0.3640 0.3645 Total Cost ($/s)
Figure 6.16: CGAM Problem – Optimal Pareto Curve at the Economy level.
214 3.64E-04
3.62E-04 Ecological Optimum 3.60E-04 P2 / P1 = 12.75
η AC = 87.49% 3.58E-04 η = 89.61% Economic Optimum GT T = 853.75K 3.56E-04 3 P2 / P1 = 8.54 T4 = 1526.25K η AC = 84.68% 3.54E-04 ηGT = 87.86%
Exergetic Efficiency (%) Exergetic T = 914.28K 3.52E-04 3 T4 = 1492.63K 3.50E-04 0.360 0.370 0.380 0.390 0.400 0.410 Total Cost ($/s)
Figure 6.17: CGAM Problem – Optimal Pareto Curve at the Ecosystem level.
6.4 Discussion
The equipment tier of analysis is the most reductionist and the most commonly applied in engineering thermodynamics. The ecological optima at this tier for both examples, shown in Figure 6.9 and Figure 6.14, clearly show the shortcomings of making decisions based solely on exergy analysis. It leads to solutions that minimize operational inefficiencies at the expense of highly expensive capital investment, which may not mean
215 more environmentally conscious designs. A clear benefit of the multiobjective approach over the classical thermoeconomics or exergy analyses is that it allows observing other optimal values that are in between the two extreme optima. For instance, it is possible to identify solutions that improve sharply upon exergetic efficiency without increasing the total cost significantly. Nevertheless, exergy analysis at this tier is shortsighted and should only be used to get a rough set of potential optimal designs.
The ecosystem level of analysis is the most comprehensive because it captures the overall effect of the process evaluated. As opposed to narrowly focused analyses, that are prone to shift environmental problems beyond the reach of the system, this approach identifies the designs that improve the efficiency of the entire production chain. Even though this aspect is critical for assessing sustainability, most thermodynamic methods for process evaluation lack a proper handling of the system boundaries. This became clear for exergy analysis and traditional CEC in this case study. By ignoring capital inputs, their outcome not only are misleading by leaning towards infinitely expensive equipment, but are also perverse because the alternatives suggested sharply increase resource utilization and pollutants’ emissions from the production chain of the capital inputs.
Rather than identifying the ecological optimum at the ecosystem level, the hierarchical approach can offer a far more powerful feature, which is a basis for designing environmentally conscious processes. In principle, such approach can be used to efficiently identify the best alternatives and discard poor designs at early stages of analysis. Nevertheless, it is necessary to understand and recognize a general pattern.
216 Figure 6.18 shows a qualitative graph of exergetic efficiency vs. total cost at all tiers of analysis. The more obvious trends can be identified in the figure. These are:
• Exergetic efficiency decreases with increasing levels of analysis. The reason is
because there is more exergy consumed as the boundaries of analysis includes
more processes. A similar behavior occurs with the hierarchical approach for
process synthesis described by Douglas (1988), in which the economic potential
inevitably decreases as increasing tiers add costs to the analysis.
Equipment Level
Life Cycle Level
Economy Level
Ecological Ecosystem Level Optimum Exergetic Efficiency Economic Optimum Total Cost
Figure 6.18: Optimal Pareto Surfaces at different levels of analysis.
217 • The economic optimum does not change with increasing levels of analysis. As
discussed earlier, it is in the firm’s interest to optimize the costs of the plant, and
not of the entire production chain. There may be exceptions if the firm is
conducting a particular analysis.
• The optimal curve at the economy level is expected to shrink as compared to the
optimal curves at the life cycle and equipment levels. As discussed earlier, the
inclusion of capital inputs tend to brusquely drop exergetic efficiency in designs
with high capital investment.
• A process alternative that is lying on any of the Pareto optimal surfaces is an
optimal solution. The extreme optima are not necessarily the best decisions. As a
matter of fact, points in between are likely to be better for the firm. Selection of
alternatives depend on the priorities and goals set in the decision making stage. In
general, moving toward the ecosystem level increases environmental performance
of the system, but it also increases uncertainty. Moving toward the economic
optimum reduces total cost and increases economic potential, but moving toward
the ecological optimum has the potential of increasing the intangibles’ market
value of the company, especially with an effective marketing strategy. If the
market value of intangible assets could be related to the optimal solutions
illustrated in Figure 6.18, then it would be possible to identify the best process
alternative.
A more interesting behavior is the length of the optimal curves at the economy and ecosystem levels. What determines whether the optimal curve at the ecosystem level 218 will be longer or shorter than the optimal curve at the economy level is not obvious.
Similarly, it is not clear under which conditions may the economic and the ecological optimal solutions coincide, thus degenerating the optimal curve to a single point.
The value of the process inputs in the economic objective are reflected in their market price. Natural resources are not priced directly, but their value is captured as these resources are refined in the economic system. In any case, the economic objective reflects how natural resources are valued in the economy. For instance, the economic optimum would adjust accordingly if the price of crude oil increased. Cumulative exergy at the ecosystem level also reflects the value of the process inputs, but this value is based on how much exergy has been invested in producing the input. Then, cumulative exergy of crude oil would remain invariant regardless of fluctuations in its price and it would reflect how much exergy has been required by ecological processes to make it available to human beings. Consequently, it is fair to imply that the length of the optimal curve at the ecosystem level signifies a bias in economic valuation of natural capital. Nevertheless, much research is required before a statement is issued.
6.5 Future Work
As part of future work, there is to expand the multiobjective problem to include impact of emissions and to develop the general framework for Multiscale analysis.
Increasing understanding that may lead to recognizing patterns and developing heuristics for selection and quick screening of environmentally conscious alternatives or designs is part of ongoing research.
219 Finally, given the complexity and size of the systems treated by the approach presented here, uncertainty plays a vital role in the evaluation process. Since the hierarchical approach treats each scale of analysis separately, uncertainty analysis can be easily and efficiently incorporated in the methodology. Similarly, it is possible to integrate this approach with the hierarchical approach for process synthesis described by
Douglas (1988) in order to create a general framework for environmentally conscious process design. These tasks are material of future work.
220
CHAPTER 7
HIERARCHICAL THERMODYNAMIC METRICS
The expectations of life depend upon diligence; the mechanic that would perfect his work
must first sharpen his tools.
–Confucius
Performance metrics for industrial activities aim to quantify the ecological, economic and social aspects of processing systems and their life cycles to facilitate sound decision-making. The challenges in developing such metrics for industrial processes and the variety of existing approaches are described in recent papers (Azapagic and Perdan,
2000; Marteel et al., 2003; Sikdar, 2003). Popular approaches relevant to chemical processes include those developed by the American Institute of Chemical Engineers in the U.S. (AIChE, 2004; Schwarz et al., 2002) and by the Institution of Chemical
Engineers in the U.K. (IChemE, 2004). Similar efforts are also being made by industry groups such as, the Global Reporting Initiative (www.globalreporting.org) and the World
Business Council for Sustainable Development (www.wbcsd.org). Because of these efforts, a variety of practical and industrially relevant metrics have already been developed and applied to quantify the environmental performance of economic and
221 industrial activities (Wirdak, 2003). These metrics typically include measures of pollutant output, process performance and direct and indirect effects of an activity on the environment and society.
In general, these approaches categorize the environmental effects of industrial processes into input side and output side metrics as shown in Figure 4.2. Basic environmental performance metrics include material intensity, energy intensity, water consumption, toxic emission, pollutant emission and carbon dioxide emission. These input and output variables are normalized by measures such as mass of product, dollars of value added, or dollars of revenue. Additional metrics for specific types of impact of pollutants, land use, and social aspects may also be developed. The desirable characteristics for performance metrics of industrial activities include: simple to calculate, useful for decision making, understandable to different audiences, cost- effective, robust and non-perverse in indicating progress towards environmental protection, stackable to permit combination with metrics for other processes, and protective of proprietary information (Schwarz et al., 2002). The calculations for these metrics are relatively straightforward, and have been carried out for a large number of chemical processes by Schwarz et al. (2002).
However, some of the shortcomings faced by such approaches for practical metrics include the following.
• Curse of Dimensionality. A large number of, often conflicting, metrics and
variables make the decision-making task quite challenging.
222 • Perverse Results due to Lack of Theoretical Rigor. Adding the mass or energy of
different streams to compute the material or energy intensity focuses only on the
first law of thermodynamics and ignores the second law. This can lead to perverse
results such as improvement in metrics by switching to a higher quality but scarce
energy source. Furthermore, mass and energy usually cannot be separated for any
stream, and the separate consideration of both streams may introduce redundancy
and double counting.
• Challenge of Multiple Scales. A rigorous and comprehensive method for
considering inputs and impacts of the selected process at multiple spatial scales is
crucial. Otherwise, use of a narrow spatial boundary may improve environmental
performance by simply shifting the impacts outside the boundary.
•Material intensity •Green house gases •Energy intensity •Human health metrics •Water consumption Processes •Ecotoxicity metrics •Land use •Acidification •etc. •etc.
Figure 7.1: Environmental performance metrics of AIChE-CWRT (Schwarz et al., 2002).
223 This Chapter proposes the use of thermodynamic methods at multiple spatial scales to overcome these shortcomings while retaining the attractive characteristics of practical environmental performance metrics. The use of thermodynamic methods for evaluating environmental performance of industrial products and processes is motivated by the fact that all activities on earth rely on the availability of energy and its conversion to various goods and services. Ultimately, all planetary activities depend on exergy or available energy (Ayres, 1994; Odum, 1996; Szargut et al., 1988), making it the ultimate limiting resource. Exergy provides a scientifically rigorous way to compare and combine streams of material and energy, and represents environmental impact and information content. It has been most popular for analyzing chemical and thermal processes to improve their efficiency (Szargut et al., 1988). An influential report by Ford et al. (1975) demonstrated the benefits of exergy analysis for improving energy efficiency. Unlike energy or mass, exergy is able to jointly represent material and energy streams. Ayres et al. (1998) also suggested that the exergy of emissions could provide a proxy for the potential impact of the emissions. Such an approach is useful for quick and approximate evaluation of environmental impact without a detailed impact analysis. The limited validity of this connection between exergy of emissions and their impact is demonstrated by Dincer (2000) and Seager and Theis (2002).
These ideas have also been incorporated into environmental performance metrics.
For example, Berthiaume et al. (2001) have proposed a renewability indicator of a biofuel based on the net exergy produced in a process and the exergy required for reducing the impact of emissions by converting them into a benign state. The difference between these
224 two exergy values is the useful work, and the renewability indicator is calculated as a ratio of the useful work to the produced work. Such an approach is difficult to use without detailed knowledge about the process life cycle. Dewulf et al. (2000) have quantified environmental performance using renewability and efficiency parameters. The renewability parameter is the ratio of the exergy consumption of renewable resources to the total exergy consumption. The efficiency parameter is the ratio of the exergy value of the useful products to the sum of exergy consumed in the process and that required for abatement of harmful emissions. Lems et al. (2002) have used the depletion time of a resource, exergy efficiency of process, and exergy consumption for abatement of emission. The depletion time of a resource is used as an environmental performance index since it can represent both renewable and nonrenewable resource in a consistent manner. The depletion time is converted into an abundance factor to have a scale of zero to one. This approach faces challenges in determining the relevant parameters for each resource, and ignores the effect of market forces on the depletion time.
The approach proposed in this Chapter enhances the metrics described in Schwarz et al. (2002). Exergy is used for reducing dimensionality on the input side of Figure 4.2 by combining material and energy streams in a theoretically rigorous manner. This combination is possible because the utility of any material or energy stream is in its ability to do work, and exergy represents this useful part of any material or energy stream. It is quantified as the thermodynamic distance or distinguishability from the reference environment. On the output side, the dimensionality may be reduced in two ways. The exergy of all the output streams can be calculated, as for the inputs, and may
225 roughly correspond to the impact of the outputs on the environment (Ayres et al., 1998;
Seager and Theis, 2002). Unfortunately, the relationship between the exergy of emissions and their impact is often tenuous. Consequently, if end-point impact assessment methods are available (Bare et al., 2000), then the impact of emissions may be represented in terms of exergy loss of the impacted system, or its ability to do useful work. Finally, the input and output side results may be combined to yield a single aggregate metric. This aggregation may be done in a few different ways depending on the normalization. This approach results in a hierarchy of metrics at different levels of aggregation, and uses a scientifically sound approach for addressing the first two shortcomings listed above.
Existing data on environmental performance metrics used in Schwarz et al. (2002) and related work for individual processes may be readily used for developing this hierarchy at the process scale. The challenge of multiple spatial scales is addressed by including information about the life cycle, other economic sectors and contribution of ecological goods and services. Expansion to the life cycle scale is accomplished by using information about selected processes in the life cycle, as suggested by ISO 14000, (ISO
14040, 1997), while information about economic sectors and ecosystems is incorporated via input-output analysis (Hendrickson et al., 1998; Ukidwe and Bakshi, 2003). The result is a doubly nested hierarchy, which consists of multiple spatial scales and different levels of aggregation for the nodes at each scale. These scales are selected to capture the global nature of environmental performance, along with the need of connecting it to local decisions at smaller scales. The selected scales of equipment, life cycle, economy and ecosystem correspond with type of data and models that are currently available.
226 This Chapter demonstrates the features and benefits of exergy analysis and multi- scale hierarchical metrics. It does not focus on decision-making or evaluating specific metrics, but other exergy metrics such as those described earlier may be readily calculated from the results of the proposed approach. The hierarchical structure permits gradual expansion of the scale of analysis from the equipment to the economy scales.
This approach is illustrated via application to an ammonia process and its coarser spatial scales.
The rest of the Chapter is organized as follows. The use of exergy analysis is explained to evaluate the emission impacts in Section 7.1. Benefits of using exergy analysis are illustrated through various examples in Section 7.2. Hierarchical metrics for process scale and life cycle scale are discussed using industrial and ecological cumulative exergies in Section 7.3. Case studies are implemented for an ammonia process in Section
7.4 followed by discussion and conclusions in Section 7.5.
7.1 Exergy of Output Streams and their Impact
As mentioned earlier, this Chapter considers two approaches for combining the exergy of process outputs – direct addition of the exergy of the output streams, or assessing the impact of emissions and representing it as a loss of exergy of the impacted system. Both approaches are described in brief in this section.
Exergy of useful products is commonly used for calculating thermodynamic efficiency. In addition, exergy of waste products has been suggested as a way of quantifying the impact of emissions. Several researchers have hypothesized that since
227 exergy of emissions represents their ability to do work on the environment, it may be related to their impact. A recent study by Seager and Theis (2002) used the exergy of mixing to quantify environmental impact of air pollutants.
⎛ y ⎞ B = nRT ln⎜ ⎟ (7.1) mix 0 ⎜ 0 ⎟ ⎝ y ⎠
0 where Bmix is the exergy change of a substance when its activity changes from y to y , n is the total number of moles, y the activity in the thermodynamic system under consideration, and y 0 the reference activity in the environmental sink. However, the computation results for exergy of mixing are very sensitive to the choice of reference state, and chemical species that do not exist at all in nature would have an infinite exergy of mixing. Seager and Theis (2002) have found only a limited relationship between the exergy of emissions and their environmental impacts. However, this measure may provide an approximate estimate and may be useful if detailed impact analysis is not available.
Impact assessment by “end-point” methods considers the likely effect of emissions on human and ecological systems (Bare et al., 2000). If such analysis is available, then the results from end-point methods can be converted into exergetic terms
(Hau, 2002; Ukidwe and Bakshi, 2004). For example, an end-point impact assessment method such as, Eco-indicator 99 measures human health impact in terms of disability- adjusted life years (DALY). DALY represents the total years of healthy life lost due to premature mortality or some degree of disability caused by the emission. The results of
228 Eco-indicator 99 may be converted into thermodynamic terms since impact may be viewed as a loss of human or ecosystem ability to do work. For converting DALY of substance i , DALYi, to exergy loss due to human health impact, Bimpact,i, it must be
multiplied with its mass flow, mi, and a conversion factor, ξi , representing the exergy consumption of the population concerned.
Bimpact,i = mi ⋅ DALYi ⋅ξi (7.2)
The conversion factor, ξi may be calculated as a product of the exergy consumed by an average person and the ratio of the average payroll of the affected population to the mean average payroll. This ratio, assumes that human exergy is proportional to their skill level, as quantified by their payroll.
7.2 Illustrations of the Benefits of Thermodynamic Metrics
The following simple illustrations show how environmental performance metrics based on material or energy intensity can lead to perverse results and how exergy based metrics may be able to overcome this shortcoming.
7.2.1 Energy versus Exergy Metrics for Hot Water
Consider two sources of hot water, Stream 1 at 75 oC and flowrate of 20 kg/hr, and Stream 2 at 45 oC and flowrate of 50 kg/hr. The enthalpy content of both streams is equal implying that energy metrics for both streams would also be equal. The fact that the hotter stream is much more useful due to its ability to heat a wider range of temperatures,
229 is not captured by energy metrics. However, the exergy content of both streams is very different, and is able to capture the difference in the quality of both streams. Stream 1 has an exergy of 75.6 kcal/hr versus 32.1 kcal/hr for Stream 2. Thus, in this case, exergy metrics provide much more meaningful information. In fact, the ratio of exergy to enthalpy has been suggested as a measure of quality (de Swaan Arons et al., 2004). If material and energy metrics are used together, then the smaller material intensity of
Stream 1 indicates the higher quality of this stream. However, this approach requires consideration of two dimensions (material and energy) and ignores the second law, while exergy is more rigorous due to consideration of the first and second laws with only one dimension.
7.2.2 Maleic Anhydride Processes
Input side metrics are calculated for two alternative processes for making maleic anhydride (Schwarz et al., 2000). Both processes consume n-butane and oxygen as raw materials, but in different amounts. Table 4.1 shows material and exergy intensities of the two processes per pound of maleic anhydride produced. Material intensity of Process A is larger than that of Process B, which implies that Process B may be preferable from the view of resource consumption. Exergy analysis of both processes leads to the opposite conclusion since this approach accounts for the fact that oxygen is a lower quality material source than butane. That is, butane is farther away from the ambient environment than oxygen, or oxygen is more plentiful than butane. Using less butane and more oxygen is better for the conservation of resources. Thus, exergy-based metrics
230 guide towards the conservation of limited resources because chemical exergy calculated with an environmental reference state can roughly correspond to the scarcity of natural resources (Munoz and Michaelides, 2000).
7.2.3 Thermodynamic Metrics for Processes
The first example above illustrates how exergy per unit enthalpy is a metric of energy quality of a material. The corresponding metric for evaluating processes is suggested to be the ratio of exergy to the ICEC for the process supply chain (Szargut et al., 1988). This ratio, called industrial cumulative degree of perfection (ICDP) ignores the contribution of ecosystem goods and services or natural capital, and may result in misleading metrics. As illustrated by Hau and Bakshi (2004a), the ICDP of coal based electricity is higher than that of solar thermal electricity. This is a potentially misleading result from the viewpoint of environmental performance and conservation of natural capital since coal is a finite resource, unlike sunlight. This approach ignores the fact that coal is a much higher quality and scarce energy source than sunlight. In contrast, the ecological CDP (ECDP), calculated by including the ecosystem goods and services required for making natural resources, provides the opposite result. This is because emergy or ECEC is a better indicator of the scarcity of natural resources since greater reliance on ecological goods and services usually implies less availability of the resource
(Odum, 1996). More research is required to evaluate the benefits of ECEC for weak and strong sustainability assessment and metrics.
231 Intensity Process A Process B Material (lb / lb product) Butane 0.073 0.316 Oxygen 5.745 2.176 Total 5.818 2.492 Exergy (Btu / lb product) Butane 2.236×104 2.742×104 Oxygen 2.079×10-1 9.798×10-2 Total 2.236×104 2.742×104
Table 7.1: Comparison of material intensity and exergy for two maleic anhydride processes.
7.3 Hierarchical Metrics
Due to the complex and multi-dimensional nature of sustainability, a hierarchy of metrics is convenient for combining results from different studies (Schwarz et al., 2002), capturing the trade-offs between natural, economic and social capital (Graedel and
Allenby, 2002), and allowing easier evaluation among alternatives (Sikdar, 2003).
Analysis of individual industrial processes is not enough for evaluating their environmental performance since such a narrow view may simply shift the impacts to other parts of the life cycle. Consequently, environmental performance metrics must be capable of linking small scales such as, an individual equipment or process with life cycle scales, and extend further to scales of the economy and ecosystems (Bakshi and Fiksel,
2003). Methods for including the life cycle are widely used and standardized via ISO 232 14000. Combining this Process LCA with Economic Input-Output LCA (EIO-LCA) is the focus of Hybrid LCA methods (Suh et al., 2004). Another desirable property of environmental performance metrics is that they must be stackable. Metrics for multiple systems or coarser scales should be obtainable by direct combination of individual metrics of constituent systems. This section describes a doubly nested hierarchy of metrics at multiple spatial scales, with another hierarchy based on the degree of aggregation at each scale. The spatial hierarchy represents the equipment, life cycle, economy and ecosystem scales, while the aggregation hierarchy uses exergy to reduce dimensionality by combining material and energy flows.
7.3.1 Multiscale System
Figure 7.2 shows the conceptual diagram of flows for industrial and ecological processes (Odum, 1996; Bakshi, 2002). Industrial processes consume non-renewable
resources, N , renewable ecosystem services and products, R1 , and input from economy,
F . Economic inputs represent the things that are valued by the economy, and involve a monetary transaction. The outputs of industrial processes include the main products that are sold in the market, Y , and emissions that are returned to the environment, W . The
ecosystem output, R2 , represents nature’s services needed to dissipate the emissions, and absorb their impact, respectively. This Chapter only considers the human impact of emissions.
233
Ecosystem scale Economy scale N Nonrenewable resources F
Y Economic Ecological R Sun 1 Industrial processes resources processes
R2
W
(a)
Process 5
Process scale Process 1 Process 2 Process 3 Process 4
Process 7 Process 6
Process-based life cycle scale
(b)
Figure 7.2: Energy flow diagram at multiple spatial scales. (a) Flow diagram for economy and ecosystem scales, (b) Industrial processes considered for process and life cycle scales. 234 The box representing industrial processes in Figure 7.2a may include a single process or a network of processes depending on the scale of analysis. At the process scale, only individual processes are considered, as depicted in Figure 7.2b. Information from selected multiple processes is combined to form the life cycle scale. The economy scale also includes economic sectors represented by the box of “economic resources” in
Figure 7.2a. Finally, the ecosystem scale includes ecological processes that lie outside the market.
7.3.2 Aggregation Hierarchy
The hierarchy representing different levels of aggregation or dimensionality reduction is depicted in Figure 7.3 for a selected system. Such a hierarchy may be readily developed at any spatial scale. At the base level (Level 1a), the left and right halves contain data about the inputs and outputs, respectively. These data may be in a variety of units such as mass, energy, or money. At Level 1b, all the data are converted into consistent thermodynamic units. This requires the use of methods described in Section 2.
It includes chemical and physical exergies because raw materials usually react to form products. At this level, details about all input and output streams are available without any aggregation resulting in a high dimensional space. Properties of the products such as, cost or exergy may be incorporated in the metrics via normalization.
235 Single Level 3 metric
Exergy Exergy of Level 2 loss due to inputs impact
Exergy loss due to Exergy for impact by Level 1b individual input individual output
Mass, energy or money for Mass, energy or money for Level 1a individual input individual output
Figure 7.3: Hierarchical structure of environmental performance metrics for a selected system.
At the next level of the hierarchy (Level 2), the multitude of inputs and outputs are combined to yield more aggregate but separate metrics for the inputs and the impact of emissions. The impacts may be approximated via the exergy of the waste streams, or if end-point impact assessment is available, by the exergy loss of human and ecological systems. The examples in this Chapter use the latter approach but only consider human impact. This aggregation can be done in a few different ways. Exergy of outputs may be classified as products (Y) and waste (W), and of inputs may be classified as direct renewable (R1), direct nonrenewable (N), and direct economic (F). The impact of
236 emissions represents the exergy required for dissipation of the emissions and the loss of exergy due to the ecological and human impact caused by the emissions.
At Level 3, a single metric may be obtained to represent the environmental performance of the selected system. Many different types of metrics may be defined by combining the variables at Level 2. The simplest and potentially most useful among these is the thermodynamic efficiency (or its reciprocal), which may be defined with or without including the impact of emissions. Alternatively, the economic value added or exergy of useful products may be used for normalization. In this work, the exergy values of input streams and the exergy loss due to the impacts of emissions are aggregated to obtain the single metric at Level 3. This represents the exergy change of the environment due to the consumption of raw material and the release of pollutants from the system at the selected scale. The proposed hierarchical structure eases the curse of dimensionality by aggregating multiple variables in a scientifically sound manner, while ensuring that details are still available. This approach allows the user to select the level of aggregation that is most useful for making decisions.
7.3.3 Spatial Hierarchy
Assessment of sustainability principles cannot be accomplished by considering only a single spatial scale, and techniques and data are required for expanding the system boundary of a process or equipment to the entire life cycle, economy or ecosystem
(Bakshi and Fiksel, 2003). Conversely, if data and metrics are available at a coarse scale, they need to be translated to finer scales to permit detailed engineering decision-making.
237 The aggregation hierarchy presented in Section 3.2 focuses on a system at a single scale.
Similar hierarchies of metrics may be developed at other scales, and may be connected with each other to result in a spatial hierarchy, as described in this section.
Information for developing the aggregation hierarchy at the process scale may be readily obtained from mass and energy balances and cost information about the process from simulation or literature sources. Gate-to-gate inventory modules developed for specific processes may also provide useful information (Jiménez-González et al., 2000).
Such information also forms the basis of the metrics developed by AIChE-CWRT and others. Developing the aggregation hierarchy for the process scale would permit a conventional thermodynamic analysis, and inclusion of the impact of emissions allows consideration of some broader life cycle aspects.
Expanding the analysis to the life cycle scale involves selection of the most important processes in the life cycle. This approach is analogous to that used for process
LCA, and may utilize the extensive life cycle inventory databases included in various software packages. Converting the results of a process LCA into thermodynamic terms is quite straightforward if information about the physical and chemical properties of various streams is available. Data about cumulative exergy consumption (CEC) and cumulative degree for perfection (CDP) have been calculated for many common industrial processes, and may be used to calculate the life cycle exergy consumption of selected products
(Szargut et al., 1988; Sussman, 1980). Analysis at this scale ignores a large number of processes in the life cycle network, which together may introduce a significant error in the results.
238 The coarser economy scale considers activities in the entire economy to satisfy the requirements of the processes selected in the life cycle scale. This analysis relies on combining economic input-output LCA (EIO-LCA) with process LCA, resulting in an approach analogous to a tiered hybrid LCA (Suh et al., 2004). EIO-LCA is applied to the streams that enter the process LCA based on their economic value, resulting in the material, energy and emissions information due to the economic inputs considered at the life cycle scale. These may be converted into exergy values for Level 1b and Level 2 by the approaches described in Section 2. The same result may also be obtained more conveniently by using the ratios of ICEC to money for economic sectors derived by the thermodynamic input-output analysis approach of Ukidwe and Bakshi (2004). This ratio indicates how much cumulative exergy is consumed in the entire economy per unit of monetary throughput in each industrial sector. Thus, the ICEC of each economic input may be calculated as,
ICECi = mi ⋅Ci ⋅ RICEC,i (7.3)
where, ICECi is industrial cumulative exergy for product i , mi mass flow of product i ,
Ci price of product per unit of mass and RICEC,i the ratio of ICEC to money in the economic sector corresponding to product i . This analysis ignores the contribution of ecological goods and services for making various natural resources, which is equivalent to assuming that all natural resources require similar ecological effort. This assumption is clearly incorrect since coal and oil require much more input from nature than sunlight.
The shortcomings of this assumption are discussed and illustrated in the last example of
239 Section 2.5 and by Odum (1996) and Hau and Bakshi (2004a). The ecosystem scale overcomes this shortcoming.
The ecosystem scale expands the analysis boundary to also account for the contribution of ecological goods and services. These inputs of natural capital form the basis of all economic activity, and ignoring them may result in transferring impacts to erosion of natural capital (Ekins et al., 2003). Determining the ecological cumulative exergy consumption throughout the life cycle for resources and impact of emissions is facilitated via the ratios of ECEC to money (Ukidwe and Bakshi, 2004). ECEC of each economic input is calculated by equation 7.4.
ECECi = mi ⋅Ci ⋅ RECEC,i (7.4)
where, ECECi is ecologically cumulative exergy for product i , mi mass flow of product
i , Ci price of product per unit of mass and RECEC,i the ratio of ECEC to money in the economic sector corresponding to product i . This ratio provides information about how much exergy is consumed from the generation of natural resources to the generation of products per unit of monetary throughput in each industrial sector. It is represented in solar equivalent joules (sej) following the approach of emergy analysis (Odum, 1996).
The appropriate level of detail for decision-making can be selected by the decision-maker according to the type of decision-making task. For example, upper level management may prefer using the most aggregated numbers, while process engineers may prefer the details provided by the bottom levels of the hierarchy. More aggregated metrics may be used for quick screening between alternatives.
240 7.4 Case Study – Ammonia Process
This case study demonstrates use of the proposed hierarchical environmental performance metrics via application to an ammonia process. It provides the detailed procedure for the calculation of metric values for process scale, life cycle scale, economy scale, and ecosystems scale. Detailed calculations are available in supplementary material
(Yi and Bakshi, 2004).
7.4.1 Process Scale
Figure 7.4 and Figure 7.5 show the detailed information of input and output streams for an ammonia process, based on data from Shreve and Brink (1977). From Figure 7.4 and Figure 7.5, the hierarchical metrics at the process scale are prepared very easily, as shown in Figure 7.6 and in Yi and Bakshi (2004). For a metric to be meaningful and permit comparison across alternatives it must be normalized by an appropriate output measure such as monetary value or exergy content. Normalized metrics may be readily obtained by dividing the numbers in Figure 7.4 and Figure 7.5 by mass of ammonia (38 t/h) or its exergy (212 MW). The normalized metric value by the production rate of ammonia for material consumption at Level 1a for the ammonia process is 1.6 t/h because it only includes natural gas and air. Fuel is not included as material consumption but as energy metric in the Level 1a. The metric value for energy consumption at Level
1a is 4.3 MW, which is the sum of net calorific value for fuel and electric power. The value for water consumption at Level 1a is 3.1 t/h based on assuming 10% loss of process water in cooling tower. Exergy values for material, water and fuel are calculated by
241 equation 2.22, but exergy of electricity is assumed equal to its energy content. The values of emission of toxics, pollutants and carbon dioxide are based on the work of US
Department of Energy Office of Industrial Technology (Pellegrino, 2000). These emissions are converted into exergy via equation 7.2. Although the mass of emissions is much smaller than that of resource consumption, the converted exergy values of impact are three times the exergy value of input streams. The impact considers the affected systems to the end-point, while the input exergy does not consider all the inputs to the starting point. With increasing spatial scale, this disparity decreases. The single metric in
Level 3 is the sum of input exergies and exergy losses due to environmental impact and dissipation in Level 2.
7.4.2 Life Cycle Scale
The system boundary for environmental performance metrics is expanded using process-based life cycle analysis. The expanded system includes a refinery process for the generation of natural gas from crude oil and a power plant for the generation of electric power from coal. The flow information for mass and exergy is based on data from Maple (2000) and Taftan Data (www.taftan.com). Such information could also be obtained from commercial life cycle inventory databases. Natural gas is produced in the refinery from crude oil, and is fed to the ammonia plant as raw material. Electric power for the ammonia plant is supplied from the power plant where coal is consumed to generate electricity. Figure 7.7 shows input and output flows of material and exergy for the process-based life cycle scale. From Figure 7.7, environmental performance metrics
242 for individual processes of the power plant, refinery and ammonia plant are shown at the process scale in Figure 7.8. The calculation procedures for the refinery and power plant are similar to those for the ammonia plant (Yi and Bakshi, 2004).
18 LNG
41 Air
CO2 46
12 Fuel Ammonia process 0 Electricity Mass flow (t/h)
TOC 1.8x10-1
117 Water NH3 7.9x10-2 SO2 1.1x10-3 CO 3.0x10-1
Figure 7.4: Mass flow diagram for an ammonia process. The energy value for 12 t/h of fuel is 160 MW.
243 Ecological processes
LNG CO 797 259 2 5.8 0
Air TOC 582 0 1.7 Ammonia process 5 Fuel 167 Exergy (MW) NH3 19 0.44
Electricity 1 SO2 1.48x10-3 Water 1.6 CO 0.82
Figure 7.5: Exergy flow diagram for an ammonia process.
Figure 7.8 shows the normalized hierarchical metrics for life cycle scales of an ammonia process. The hierarchical metrics of a power plant and a refinery plant at process scale are normalized by the production rate of electricity and natural gas, respectively. However, the hierarchical metrics in process-based life cycle scale, economic scale and ecological scale are normalized by the production rate of ammonia.
Therefore, the hierarchical metrics of a power plant and a refinery plant are multiplied by production rates of electricity and natural gas to prepare the hierarchical metrics at coarser scales. Material metric value in Level 1a for the life cycle scale is 32.2 t/h, which 244 is calculated from the flow rates of crude oil, i-butane, MTBE and air. The corresponding energy value at Level 1a for the life cycle scale is 0.7 MW, which corresponds to net calorific value of 0.12 t/h of coal. Water metric value at Level 1a for the life cycle scale is about 11.7 t/h with the 10% loss of cooling water. Emission flow rates at this scale are obtained by addition of this information at the finer scale, and exergy values of input streams and exergy loss due to impact are calculated as for the ammonia plant. Not surprisingly, the exergy consumption at the life cycle scale of an ammonia process is larger than that for the single ammonia process. The input side metric for the entire life cycle scale of an ammonia process is about thirty times larger than that for the ammonia process alone, while the output side metric of this scale is twenty times larger than that of the ammonia process only. However, the analysis is still incomplete since the contribution of the rest of the life cycle network is ignored at the process-based life cycle scale.
7.4.3 Economy Scale
The incompleteness of process-based LCA may be addressed via a thermodynamic hybrid life cycle analysis based on the ICEC to money ratios discussed in Section 3. The economy scale in Figure 7.8 shows the results by hybrid life cycle analysis of an ammonia process. The metric values of Level 1a at the economy scale at Figure 7.8 are the sum of the process-based life cycle analysis and of applying EIO-LCA to cut-off streams at the life cycle scale. These include crude oil, i-butane, MTBE and coal. For each cut-off stream, the corresponding economic sector is selected for applying EIO-
245 LCA. This approach results in information about consumption of energy, water, and ores and production of pollutants, green house gases, toxics, etc. in the entire economy, which are added to those from process-based LCA to give metric values in Level 1a of the economy scale. Exergy at Level 2 is the sum of exergy consumption and exergy losses due to impacts by process-based analysis and by EIO-LCA. The exergy consumption by process-based analysis is shown in Figure 7.8, which is calculated by equation 7.1. The exergy consumption by EIOLCA is calculated by equation 7.3. For example, the ratio of
ICEC to money for the sector of coal mining is 2.03 × 109 J/$. This is used to estimate the total ICEC consumption in the economy due to the coal consumption of 0.12 t/h at the cost of 5 $/t. The ratios of ICEC to money are available for each US economic sector in
Ukidwe and Bakshi (2004). Exergy loss due to impact is calculated by equation 7.2 and
DALY values of emission materials. For example, the emission of carbon dioxide is 5.8 t/h and its DALY value is 2.1× 10-7 DALY/kg. The converting coefficient, ξ , of carbon dioxide is 3.1× 1014 J/DALY, which is available from Ukidwe and Bakshi (2004).
Therefore, exergy loss due to emission of carbon dioxide is about 104.9 MW. The exergy losses for other streams can be calculated in the same way.
246 Single metric 48.5 MWh/t Exergy of Exergy loss inputs due to impact 11.3 MWh/t 37.1 MWh/t Material Energy Water Toxics Pollutants CO2 6.8 4.4 0.04 15.4 0.6 21.1 MWh/t MWh/t MWh/t MWh/t MWh/t MWh/t Material Energy Water Toxics Pollutants CO2 2.1x10-3 1.3x10-3 1.6 4.3 MWh/t 3.1 1.2
Figure 7.6: Hierarchy of metrics for an ammonia process. The values of metrics are normalized by the production rate of ammonia.
The exergy of input streams at the economy scale is smaller than exergy loss due to impact, which means that cumulative exergy consumption by all processes to make ammonia is smaller than the potential exergy changes in the environment by waste emission. In addition, energy consumption, water consumption and carbon dioxide emission are much increased compared with the metric values of process life cycle scale.
Including the entire economy for life cycle analysis is not complete because ecological products and services are indispensable for industrial and economic activities and need to be included in the analysis.
247 M: 977 Crude oil E: 11633 M: 38 M: 3 E: 508 E: 12561 i-Butane SO2 M: 34 M: 3 Refinery E: 16 MTBE E: 360 Acid gas M: 321 plant M: 1014 E: 4 E: 12714 Many kinds of products M: - M: 30 Natural gas and E: 1.85 E: 396 fuel gas M:38 E: 212 NH3 M: 168 M: 41 M:46 Air E: 0 E: 0 E: 797 CO Ammonia 2 M: 443 M: 117 M:367 E: 6.1 E: 1.6 plant H2O E: 5 H2O M: - M:0.4 E: 0.6 E: 606 Others
M: - M: 127 E: 2.45 Electricity E: 0 M: 96.1 N2 E: 0.69 M: 4.5 E: 0.1 M: 17.6 O 2 E: 6.1 Power Exhaust gas M: 15.1 CO M: 4.5 plant 2 E: 1.89 E: 30 Coal M: 0.16 Ar E: 0.01 M: 47.6 H O 2 E: 0.95 M: mass flow rate (t/h) M: 0.09 SO E: Exergy flow rate (MW) 2 E: 0.12
Figure 7.7: Mass and exergy flow for an ammonia plant and selected processes in its supply chain.
248 7.4.4 Ecosystem Scale
ECEC analysis expands the system boundary of ICEC to include all ecological processes required to make natural resources available to industries (Hau and Bakshi,
2004). The ECEC values for inputs and impact of emissions may be readily obtained via an approach similar to that used for the economy scale in Section 4.3. The ECEC to money ratio for sectors of the US economy (Ukidwe and Bakshi, 2004) are used instead of the ICEC to money ratios. ECEC is represented by a consistent thermodynamic unit of solar equivalent joule (sej) and is calculated for the streams in Level 2 of the ecosystem scale by the ratio of ECEC to money for each economic sector via equation 7.4.
Alternatively, the ECEC to ICEC ratio of each sector may be used. For example, ICEC consumption of 0.12 t/h of coal is 0.18 MW, and is converted to ECEC by multiplication with the ECEC/ICEC ratio of 19 001 sej/J resulting in the ECEC of about 3 420 Msej/s.
The exergy loss due to impact by emissions in Level 2 for the ecosystem scale is calculated by multiplying the output side metrics of Level 2 in the economy scale by the ratio of ECEC/ICEC for US economic sectors, which is also available in Ukidwe and
Bakshi (2004). The cumulative exergy values at the ecosystem scale are much larger compared with those of the economy scale because energy is required to make natural product and/or services by ecological processes. When raw materials are made from ecological processes, exergy is consumed or lost. Therefore, the exergy consumption during ecological processes is not negligible when quantifying the rate of change of natural capital or assessing weak and strong sustainability of industrial processes. The
249 ratio of ECEC to money is based on the consumption of the ecological exergy when the environmental performance of a process is evaluated.
7.5 Discussion
Quantification of environmental performance is important for sound engineering and strategic decision-making. A variety of environmental performance metrics developed by various academic and industrial groups rely on quantifying the material and energy flow and impact of emissions. Such metrics suffer from the curse of dimensionality, may lack adequate scientific rigor due to ignorance of the second law, and often fail to address the multiscale nature of environmental challenges. This Chapter proposes the use of thermodynamic methods combined with input-output and hybrid life cycle assessment to overcome these shortcomings. Exergy analysis is used for scientifically rigorous combination of material and energy streams, and the quantification of emission impacts. These results are represented via an aggregation hierarchy for the selected system. Evaluation of environmental performance requires such aggregation hierarchies at multiple spatial scales including, process, life cycle, economy and ecosystem, resulting in a doubly nested tree of metrics. Expansion to coarser spatial scales is achieved via the thermodynamic input-output analysis approach of Ukidwe and
Bakshi (2004). The appropriate level of analysis may be selected depending on user preference and the type of analysis.
The focus of this Chapter is on the use of thermodynamic and multiscale methods for improving existing practical metrics such as those developed by AIChE CWRT
250 (Schwarz et al., 2002). The goal of any metric is to compare alternatives and guide decision making. Detailed empirical study is essential for evaluating various kinds of thermodynamic metrics, and is the subject of on-going work. The suitability of normalization by the monetary, exergetic or some other value of the outputs also requires more study. The proposed thermodynamic methods may also be improved, particularly for evaluating the impact of emissions. This work relies on eco-indicator 99 and only considers the human impact of emissions. Similar methods are needed for evaluating the impact on ecosystems more completely.
The focus of this Chapter has been mainly on metrics for environmental performance and quantification of natural capital. However, sustainability principles require consideration of natural, economic and social capital. The proposed metrics based on exergy seem to be capable of considering all the three dimensions since exergy is the ultimate limiting resource for all economic and ecological activities. Such metrics are also referred to as 3-D metrics (Sikdar, 2003). The aggregated exergy-based metrics at
Level 3 may be useful as 3-D metrics, while metrics at lower levels could be 2-D and 1-D metrics. For example, exergy consumption via inputs is an effective 2-D indicator of economic and ecological efficiency. Thus, the proposed framework may permit decision making over a hierarchy that considers 3-D metrics for quick screening followed by smaller dimension metrics. Techniques for multiobjective decision-making and handling uncertainty will also need to play an essential role in the practical use of the proposed metrics.
251 Single metric 2.1×1010 Msej/t Exergy of Exergy loss input due to impact 1.9×1010 Msej/t 1.8×109 Msej/t Energy Water Toxics Pollutants Material 4.2 67.3 8.5×10-2 1.5×10-1 CO2 32.5 MWh/t 5.8 Ecosystem scale by hybrid analysis and ECEC/money
Single metric 5811.1 MWh/t Exergy Exergy loss of input due to impact 539.7 MWh/t 5271.4 MWh/t Energy Water Toxics Pollutants Material 4.2 67.3 8.5×10-2 1.5×10-1 CO2 32.5 MWh/t 5.8 Economy scale by hybrid analysis and ICEC/money Single metric 722.4 MWh/t Exergy Exergy loss of input due to impact 331.6 MWh/t 390.8 MWh/t Energy Water Toxics Pollutants Material 0.7 11.7 7.7×10-2 8.1×10-2 CO2 32.2 MWh/t 1.8 Process-based life cycle scale
Single Single Single metric metric metric 289.8 MWh/t 48.5 MWh/t 853.4 MWh/t Exergy Exergy loss Exergy Exergy loss Exergy Exergy loss of input due to impact of input due to impact of input due to impact 12.0 MWh/t 277.8 MWh/t 11.3 MWh/t 37.1 MWh/t 423.6 MWh/t 429.8 MWh/t Energy Energy Energy Water Toxics Pollutants Water Toxics Pollutants Water Toxics Pollutants Material 11.0 Material 4.3 Material 0.1 1.9 0 3.6×10-2 CO2 3.1 2.1×10-3 1.3×10-3 CO2 10.9 9.6×10-2 9.0×10-2 CO2 51.8 MWh/t 6.2 1.6 MWh/t 1.2 35.5 MWh/t 0.2
Power plant Ammonia plant Refinery plant Process scale
Figure 7.8: Multiscale hierarchy of environmental performance metrics for ammonia process. The values are normalized by the production rate of ammonia.
252
CHAPTER 8
CONCLUSIONS
What I dream of is an art of balance.
–Henri Matisse
In general, this dissertation developed systematic approaches for integrating economic and ecological considerations in process engineering decision making. It tackled the challenges of evaluating environmental performance and implementing environmentally conscious decision making. These challenges were evaluation of ecosystem goods and services, expansion of the analysis boundaries beyond the reaches of the plant, combination of data of different quality, and diversity of perspectives regarding the right balance between economic prosperity and environmental protection.
The following sections summarize the main objectives achieved in this dissertation and provide recommendations for implementing and enhancing the approaches presented here.
8.1 Concluding Remarks
253 Conclusions were classified by the problems tackled in this dissertation. They are described in the following sub-sections.
8.1.1 Evaluation of ecosystem goods and services
Ecological goods and services constitute the productive base that is essential for all industrial and economic activity. The principle of strong sustainability states that the natural capital available for current economic activity should also be available to future generations for their needs. Unfortunately, engineering and economic approaches for industrial decision making tend to ignore or take for granted most ecological inputs, since their contribution is not reflected in market prices. This dissertation:
• Introduced Ecological CEC (ECEC) analysis, a novel approach that accounts for
the contribution of ecosystem goods and services. This approach is an extension
of traditional or Industrial CEC (ICEC) analysis, in which the system boundaries
were broadened to include the exergy consumed by ecological processes.
Practical challenges in computing ECEC for industrial processes were identified
and a formal algorithm based on network algebra was proposed.
• Showed that a main obstacle to the implementation of ECEC analysis is that there
is no proper model for describing the exergy flows of the ecological network.
• Recognized that Emergy analysis is a tool that accounts for the contribution of
ecosystem goods and services, and has devised ways to quantify the exergy flows
of the ecological network. Unfortunately, emergy has encountered a lot of
254 resistance and criticism that has hindered its use, particularly by economists,
physicists and engineers.
• Analyzed thoroughly the main features and criticisms of emergy analysis needed
to clarify many of the common misconceptions about emergy, inform the
community of emergy practitioners about the aspects that need to be
communicated better or improved, and suggest solutions.
• Concluded that criticisms pertaining to uncertainty, sensitivity, and quantification
apply not just to emergy analysis but to all methods that focus on a holistic view
of industrial activity, including life cycle assessment, material flow analysis, and
exergy analysis.
• Established a rigorous relation between ECEC and emergy analysis that proved
their equivalence under certain conditions, enabled combination of their best
features and addressed one of the criticisms regarding the disconnection to other
thermodynamic quantities. It is also expected to help “bridge the gap between
Ayres’ industrial ecology and Odum’s systems ecology” (Anonymous Reviewer,
2003) and lead to new methods and insight for evaluating and improving the
environmental performance of industrial activity.
• Illustrated the implementation of ECEC analysis with the examples of the chlor-
alkali process and solar and coal-based processes for generating electricity,
demonstrating the benefits of accounting for nature's inputs.
255 8.1.2 Combination of data of different quality
Evaluating environmental performance requires analysis boundaries that surpass the reaches of the plant. As a result, analysts must make the additional effort of collecting all data and information of each process included in the expanded system. Because process data is typically collected from diverse sources, they tend to be inconsistent, even violating the material and energy balances. This dissertation:
• Introduced an approach for data rectification of process inventories to enforce
satisfaction of material balance and the laws of thermodynamics. The proposed
approach uses two existing techniques in process systems engineering: data
rectification and exergy analysis. The approach is user-friendly and requires no
deep knowledge of the theory behind these techniques, making it more accessible
to the general community of Life Cycle Assessment (LCA).
• Illustrated the implementation of this approach with the example of a process
inventory for the production of caustic soda and chlorine, demonstrating the
benefits of using data rectification techniques to validate LCA and process data
and guaranteeing more reliable results.
• Developed a hierarchical thermodynamic approach for utilizing available
information at various spatial scales of processes’ life cycle to evaluate ecological
objectives.
8.1.3 Evaluation of Ecological and Economic Trade-off
256 Reports show that it is common to encounter disagreement between economic and ecological objectives, i.e. there is no definitive solution but a trade-off between monetary profit and environmental impact. This dissertation:
• Provided unique insight into the trade-off between ecological and economic
aspects of manufacturing processes when they are analyzed at multiple spatial
scales. Economic aspects are accounted for via traditional cost analysis and do not
change at different scales. Ecological factors are considered via exergy analysis of
the inputs at each scale, and depend on the selected processes. The scales
considered in this analysis correspond to existing methods at the manufacturing,
value chain, economy and ecosystem scales.
• Proposed hierarchical approach that considers two objective functions: economic
cost and exergetic efficiency of the process life cycle at multiple scales. Trade-off
between these objectives is represented via a series of Pareto optimal surfaces at
various scales, thus avoiding arbitrary combinations until the final stages of
decision making.
• Illustrated the implementation of this approach with the case studies of a heat
exchanger and the CGAM cogeneration system (Valero et al., 1994), comparing it
with existing methods and highlighting the benefits of adopting a multiscale and
multiobjective view.
8.1.4 Evaluation of Environmental Performance
257 Evaluating environmental performance requires meaningful, practical and scientifically sound metrics. Most existing metrics rely on information about material and energy inputs and emissions from the main process and selected processes in its life cycle. Such metrics often result in multiple conflicting variables, making it difficult to use them for decision-making. Furthermore, environmental performance metrics need to be scientifically rigorous and capable of evaluating the broader economy and ecosystem scale impacts of selected processes and products. This dissertation
• Proposed the use of thermodynamic methods combined with input-output and
hybrid life cycle assessment to overcome these shortcomings. Exergy analysis is
used for scientifically rigorous combination of material and energy streams, and
the quantification of emission impacts. These results are represented via an
aggregation hierarchy for the selected system that results in a doubly nested tree
of metrics.
• Focused on the use of thermodynamic and multiscale methods for improving
existing practical metrics such as those developed by AIChE CWRT (Schwarz et
al., 2002).
• Proposed improvements for thermodynamic methods, particularly for evaluating
the impact of emissions. This work relies on eco-indicator 99 and only considers
the human impact of emissions.
• Illustrated the implementation of this approach with the example of the ammonia
process. 258 8.2 Future Work
Recommendations for implementing and enhancing the approaches are presented in this section.
• The challenge of combining resources over multiple temporal and spatial scales
plagues many holistic techniques, including the ones discussed in this dissertation.
The transparency and utility of existing methods could be improved by
developing a tiered system which distinguishes between resources according to
their replenishment time. Instead of categorizing resources as renewable or non-
renewable, this system could separate resources according to their renewability
over daily, short-cycle, long-cycle, or cosmological time scales.
• A similar spatial hierarchy could also be defined. The hierarchical approach
proposed here is a step toward this goal, but much is needed. Ideally, it should be
a systematic multiscale statistical framework that considers differences in the
quality (uncertainty) of data at multiple temporal and spatial scales, and combines
these data in an “optimal” and transparent manner. Concepts of “opportunity” and
“sunk” costs from economics could also be useful for considering opportunities
and alternatives that may be lost due to a decision. Research in these and other
related areas is necessary for recognizing the full potential of thermodynamic
methods for environmentally conscious decision making.
259 • The hierarchical approach should also expand the multiobjective problem to
include impact of emissions and to develop the general framework for Multiscale
analysis.
• The hierarchical approach can offer a far more powerful feature, which is a basis
for designing sustainable processes. In principle, such approach can be used to
efficiently identify the best alternatives and discard poor designs at early stages of
analysis. Nevertheless, it is necessary to understand and recognize a general
pattern. Increasing understanding that may lead to recognizing patterns and
developing heuristics for selection and quick screening of sustainable alternatives
or designs is part of future research.
• Sustainability principles require consideration of natural, economic and social
capital. The proposed metrics based on exergy seem to be capable of considering
all the three dimensions since exergy is the ultimate limiting resource for all
economic and ecological activities. Such metrics are also referred to as 3-D
metrics (Sikdar, 2003). The aggregated exergy-based metrics at Level 3 may be
useful as 3-D metrics, while metrics at lower levels could be 2-D and 1-D metrics.
For example, exergy consumption via inputs is an effective 2-D indicator of
economic and ecological efficiency. Thus, the proposed framework may permit
decision making over a hierarchy that considers 3-D metrics for quick screening
followed by smaller dimension metrics. Techniques for multiobjective decision-
making and handling uncertainty will also need to play an essential role in the
practical use of the proposed metrics. 260
APPENDIX A
DETAILS FOR ECEC ANALYSIS OF EXAMPLE IN SECTION 4.3.4
261 The steps of the algorithm introduced in Section 4.3.3 are illustrated here. Table
A.1 summarizes the data obtained from Figure 4.5, which is required for the algorithm.
From line 2 to 9 of the algorithm, as indicated in Table 4.3, the data in Table A.1 is introduced. From line 10 to 15, the transaction coefficients are calculated. From line 15 to
20, the vectors and matrices are constructed, the results are:
⎡0 0.5⎤ ⎡30⎤ ⎡10⎤ ⎡0.03 0 ⎤ ⎡0.5 0⎤ γ = ⎢ ⎥ , Bn = ⎢ ⎥ , Bp = ⎢ ⎥ , ηn = ⎢ ⎥ and γp = ⎢ ⎥ ⎣0 0 ⎦ ⎣20⎦ ⎣10⎦ ⎣ 0 0.05⎦ ⎣ 0 1⎦
Line 21 calculates the Allocation matrix, i.e.
−1 ⎡0.5 0⎤ ⎛ ⎡1 0⎤ ⎡ 0 0⎤⎞ ⎡0.5 0⎤ ⎜ ⎟ Γi = ⎢ ⎥ ⋅⎜ ⎢ ⎥ − ⎢ ⎥⎟ = ⎢ ⎥ ⎣ 0 1⎦ ⎝ ⎣0 1⎦ ⎣0.5 0⎦⎠ ⎣0.5 1⎦
Line 22 calculates ECEC of the natural resources, i.e.
−1 ⎡0.03 0 ⎤ ⎡30⎤ ⎡1000⎤ Cn = ⎢ ⎥ ⋅ ⎢ ⎥ = ⎢ ⎥ ⎣ 0 0.05⎦ ⎣20⎦ ⎣ 400 ⎦
Unit i Bi1 Bi2 Bn,k Bp,k ηn,j 1 0 10 30 10 0.03 2 0 0 20 10 0.05
Table A.1: Data required for starting ECEC analysis algorithm. 262
From line 23 to 27, the allocation scheme is chosen. If the natural resource streams can be added, as in ICEC analysis, the only step left is to calculate ECEC of products by using the equation shown by line 28, i.e.
⎡0.5 0⎤ ⎡1000⎤ ⎡500⎤ Cp = ⎢ ⎥ ⋅ ⎢ ⎥ = ⎢ ⎥ ⎣0.5 1⎦ ⎣ 400 ⎦ ⎣900⎦
If the natural resource streams cannot be added, then the MaxSelect sub-algorithm shown in Table 4.4 is applied. The algorithm compares the ECEC of the products for each natural resource input, i.e.
⎡500 0 ⎤ ()aij = ⎢ ⎥ ⎣500 400⎦
The resulting values are shown in parenthesis below the streamlines in Figure 4.5.
The largest value for each stream is indicated by a box in Figure 4.5. For products 1 and
2, 500 and 500 are the largest values, respectively. These values correspond to a11 and a22. The Allocation matrix is then modified to give
⎡max{}a11 ,a12 ⎤ ⎡a11 = 500⎤ ⎡0.5 0⎤ ⎢ ⎥ = ⎢ ⎥ ⇒ Γ = ⎢ ⎥ ⎣max{}a21 ,a22 ⎦ ⎣a21 = 500⎦ ⎣0.5 0⎦
In this case of non-additive inputs, ECEC of products is
⎡0.5 0⎤ ⎡1000⎤ ⎡500⎤ Cp = ⎢ ⎥ ⋅ ⎢ ⎥ = ⎢ ⎥ ⎣0.5 0⎦ ⎣ 400 ⎦ ⎣500⎦
263
APPENDIX B
DETAILS FOR ECEC ANALYSIS OF CHLOR-ALKALI CASE STUDY IN
SECTION 4.4.1
264 The allocation matrix of the Mercury cell process obtained from an ICEC analysis is
⎡ 0 0 0 0 0 0 0 0 0 0 0 0 0⎤ ⎢ ⎥ ⎢ 0 0 0 0 0 0 0 0 0 0 0 0 0⎥ ⎢ 0 0 0 0 0 0 0 0 0 0 0 0 0⎥ ⎢ ⎥ ⎢ 0 0 0 0 0 0 0 0 0 0 0 0 0⎥ ⎢ 0 0 0 0 0 0 0 0 0 0 0 0 0⎥ ⎢ ⎥ ⎢ 0 0 0 0 0 0 0 0 0 0 0 0 0⎥ Γ = ⎢ 0 0 0 0 0 0 0 0 0 0 0 0 0⎥ i,mercury ⎢ ⎥ ⎢ 0 0 0 0 0 0 0 0 0 0 0 0 0⎥ ⎢0.011 0.042 0.039 0.194 0.042 0.042 0.042 0.008 0.194 0 0 0 0⎥ ⎢ ⎥ ⎢ 0 0 0 0 0 0 0 0 0 0 0 0 0⎥ ⎢ ⎥ ⎢0.296 0.245 0.240 0 0.245 0.245 0.245 0.301 0 0 1 0 0⎥ ⎢0.648 0.537 0.525 0 0.537 0.537 0.537 0.658 0 0 0 1 0⎥ ⎢ ⎥ ⎣0.045 0.176 0.197 0.806 0.176 0.176 0.176 0.033 0.806 1 0 0 1⎦
This was obtained from algorithm in Table 4.3. Morris’ results differ from these because losses are attributed neither to the loops nor to the waste streams. In those cases,
Morris does not split according to equation 4.24. The allocation matrix for the process obtained from the ECEC analysis is
⎡0 0 0 0 0 0 0 0 0 0 0 0 0⎤ ⎢ ⎥ ⎢0 0 0 0 0 0 0 0 0 0 0 0 0⎥ ⎢0 0 0 0 0 0 0 0 0 0 0 0 0⎥ ⎢ ⎥ ⎢0 0 0 0 0 0 0 0 0 0 0 0 0⎥ ⎢0 0 0 0 0 0 0 0 0 0 0 0 0⎥ ⎢ ⎥ ⎢0 0 0 0 0 0 0 0 0 0 0 0 0⎥ Γ = ⎢0 0 0 0 0 0 0 0 0 0 0 0 0⎥ i,mercury ⎢ ⎥ ⎢0 0 0 0 0 0 0 0 0 0 0 0 0⎥ ⎢0 0 0.039 0.194 0 0 0 0 0 0 0 0 0⎥ ⎢ ⎥ ⎢0 0 0 0 0 0 0 0 0 0 0 0 0⎥ ⎢ ⎥ ⎢0 0.245 0.240 0 0 0 0 0 0 0 0 0 0⎥ ⎢0 0.537 0.525 0 0 0 0 0 0 0 0 0 0⎥ ⎢ ⎥ ⎣0 0 0.197 0.806 0 0 0 0 0 0 0 0 0⎦
265
APPENDIX C
DETAILS FOR THE CASE STUDY OF THE CAUSTIC/CHLORINE PROCESS
INVENTORY IN SECTION 5.2
266
NaCl (1) CH4 (1) Coal (1) R. Oil (1) Ash (2) C (2) Cl (2) H (2) N (2) O (2) Na (2) S (2) Pb (3) Hg (3) NaCl10000000000000 CH401000000000000 Coal00100000000000 R. Oil00010000000000 Ash0000-1000000000 C00000-100000000 Feedstock Cl000000-10000000 H0000000-1000000 N00000000-100000 O000000000-10000 Na0000000000-1000 S00000000000-100 Ash00001000000000 C00000100000000 Cl00000010000000 H00000001000000 N00000000100000 O00000000010000 Na00000000001000 S00000000000100 Pb00000000000010 Reactor Hg00000000000001 Ni00000000000000 Zn00000000000000 NaOH00000000000000 Cl200000000000000 H200000000000000 SO200000000000000 O200000000000000 H2S00000000000000 N200000000000000 NaOH00000000000000 Cl200000000000000 H200000000000000 Hg00000000000000 Ash00000000000000 SO200000000000000 Separator O200000000000000 Pb00000000000000 Ni00000000000000 H2S00000000000000 Zn00000000000000 C00000000000000 N200000000000000 Measured Variables = 1,670.0178.068.13.8------
Figure C.1: 1st to 14th Columns of the Balance Matrix, B, for Alternative 1.
267
Ni (3) Zn (3) NaOH (4) Cl2 (4) H2 (4) Hg (4) Ash (4) SO2 (4) O2 (4) Pb (4) Ni (4) H2S (4) Zn (4) C (4) N2 (4) NaOH (5) 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 000000-1000000000 0000000000000-100 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 000000000-1000000 00000-10000000000 1000000000-100000 010000000000-1000 00-10000000000000 000-1000000000000 0000-100000000000 0000000-100000000 00000000-10000000 00000000000-10000 00000000000000-10 001000000000000-1 0001000000000000 0000100000000000 0000010000000000 0000001000000000 0000000100000000 0000000010000000 0000000001000000 0000000000100000 0000000000010000 0000000000001000 0000000000000100 0000000000000010 ------1,000.0
Figure C.2: 15th to 30th Columns of the Balance Matrix, B, for Alternative 1.
268 Cl2 (5) H2 (5) Hg (6) Cl2 (6) Ash (6) SO2 (6) Hg (7) O2 (7) Ash (7) Pb (7) Ni (7) H2S (7) Zn (7) Ash (8) C (8) N2 (8) 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 -100-1000000000000 0-100000000000000 00-1000-1000000000 0000-1000-10000-100 00000-10000000000 0000000-100000000 000000000-1000000 0000000000-100000 00000000000-10000 000000000000-1000 00000000000000-10 000000000000000-1 893.0 16.9 3.2E-04 5.4E-04 4.6E-04 0.0010 1.5E-06 1.9E-05 4.33 9.4E-07 9.4E-07 1.5E-04 9.4E-07 3.08 -- --
Figure C.3: 31st to 46th Columns of the Balance Matrix, B, for Alternative 1.
269 R1 R2 R3 R4 R5 R6 R7 R8 R9 R10 R11 -10000000000 0-1000000000 00-100000000 000-10000000 000.1500000000 00.750.580.840000000 0.610000000000 00.250.050.150000000 000.0100000000 000.2000000000 0.390000000000 0 0 0.01 0.01 0 0 0 0 0 0 0 00000000000 00000000000 00000-100000 0 0 0 0 -0.03 0 -1 0 0 -0.06 0 0000000000-1 0000-0.4000-0.50-100 0000-0.57000000 0000000-0.500-0.940 00000000000 00000000000 00000000000 00000000000 00001000000 00000100000 00000010000 00000001000 00000000100 00000000010 00000000001 00000000000 00000000000 00000000000 00000000000 00000000000 00000000000 00000000000 00000000000 00000000000 00000000000 00000000000 00000000000 00000000000
------
Figure C.4: Transpose of Master Stoichiometric Matrix, ST, for Alternative 1.
270 B 1,670.0 ------1,000.0 893.0 16.9 3.2E-04 5.4E-04 1.5E-06 NaCl (1)NaCl H2(1) O2(1) (2) Cl (2) Na (2) H (2) O (3) Hg (4) NaOH (4) Cl2 (4) H2 (4) Hg (5) NaOH (5) Cl2 (5) H2 (6) Hg (6) Cl2 (7) Hg Block Matrix, H000001000000000000 O000000100000000000 N000000000000000000 H00000-1000000000000 O000000-100000000000 Cl000100000000000000 Cl000-100000000000000 Na000010000000000000 Na0000-10000000000000 Hg00000001000-1000000 Hg000000000001000-10-1 H20000000000-10000000 H200000000001000-1000 H2010000000000000000 O2001000000000000000 Cl2000000000-100000000 Cl20000000001000-100-10 NaCl100000000000000000 NaOH00000000-1000000000 NaOH000000001000-100000 Reactor Separator easured Variableseasured = Feedstock
M
Figure C.5: Balance Matrix, B, for the Main Process Module of Alternative 2.
271
Transpose of Master Stoichiometric Matrix, Transpose( S) R1 R7' R9' R5 R6 R7 -100000 0-10 0 0 0 00-1000 0.6100000 0.3900000 010000 001000 0000-10 0 0 0 -0.57 0 0 0 0 0 -0.03 0 -1 0 0 0 -0.40 0 0 000000 000000 000100 000010 000001 000000 000000 000000 000000
------
Figure C.6: Transpose of Master Stoichiometric Matrix, ST, for Main Process Sub- Module of Alternative 2.
272 ) Pb (3) (3) Ni (3) Zn (4) Ash (4) C (4) Pb (4) Ni (4) Zn (4) SO2 178.0 68.1 3.8 ------CH4 (1)CH4 (1) Coal (1) Oil R. (2) Ash (2) C (2) N (2) H (2) O (2 S S000000001000000000 S00000000-1000000000 C000000000000010000 C0000100000000-10000 C0000-10000000000000 N000001000000000000 H000000100000000000 O000000010000000000 N00000-1000000000000 H000000-100000000000 O0000000-10000000000 Ni000000000010000-100 Ni000000000000000100 Pb00000000010000-1000 Pb000000000000001000 Zn0000000000010000-10 Zn000000000000000010 O2000000000000000000 N2000000000000000000 O2000000000000000000 N2000000000000000000 Ash000000000000100000 Ash000100000000-100000 Ash000-100000000000000 SO200000000000000000-1 H2S000000000000000000 SO2000000000000000001 H2S000000000000000000 CH4100000000000000000 Coal010000000000000000 R. Oil001000000000000000 R. Reactor Separator easured Variables= Feedstock
M
Figure C.7: 1st to 18th columns of the Balance Matrix, B, for the energy sub-module of Alternative 2.
273
00000000000000 00000000000000 00000000000000 00000000000000 00000000000000 00000000000000 00000000000000 00000000000000 00000000000000 00000000000000 00000000000000 00000000000000 00000000000000 00000000000000 00000000000000 00000000000000 00000000000000 00000000000000 00000000000000 0-1000000000000 00-100000000000 000-10-100000-100 000000000000-10 000000-10000000 0000000-1000000 00000000-100000 0000-1000000000 100000000-10000 0100000000-1000 0010000000000-1 ------4.6E-04 0.0010 4.33 9.4E-07 9.4E-07 9.4E-07 1.9E-05 1.5E-04 3.08 -- -- -10000000000000 O2 O2 (4) (4) H2S (4) N2 (6) Ash (6) SO2 (7) Ash Pb (7) (7) Ni (7) Zn (7) O2 (7) H2S (8) Ash (8) C (8) N2
Figure C.8: 19st to 32nd columns of the Balance Matrix, B, for the energy sub-module of Alternative 2.
274
R2 R3 R4 R8 R9 R10 R11 -1000000 0-100000 00-10000 0 0.1500 0 0 0 0 0 0.7487 0.5770 0.8400 0 0 0 0 0 0.0070 0 0 0 0 0 0.2513 0.0522 0.1500 0 0 0 0 0 0.2008 0 0 0 0 0 0 0.0130 0.0100 0 0 0 0 0000000 0000000 000000-1 0 0 0 0 0 -0.0592 0 0 0 0 -0.4995 -1 0 0 0 0 0 -0.5005 0 -0.9408 0 0000000 0000000 0000000 0001000 0000100 0000010 0000001 0000000 0000000 0000000 0000000 0000000 0000000 0000000 0000000 0000000
------
Figure C.9: Transpose of Master Stoichiometric Matrix, ST, for energy sub-module of Alternative 2.
275 NaCl (3) H2O(3) Hg (3) NaOH (4) Cl2 (4) H2 (4) NaCl100000 H2O010000 Reactor Hg001000 NaOH000-100 Cl20000-10 H200000-1 NaOH000100 Cl2000010 Separator H2000001 Hg000000
Measured Variables = 1,670.0 ------
Figure C.10: 1st to 6th Columns of Balance Matrix, B, for the Main Process Module of Alternative 3.
Hg (4) NaOH (5) Cl2 (5) H2 (5) Hg (6) Cl2 (6) Hg (7) R12 0000000-1.00 0000000-0.31 -10000000 00000000.68 00000000.61 00000000.02 0-1000000 00-100-100 000-10000 1000-10-10
-- 1,000.0 893.0 16.9 3.2E-04 5.4E-04 1.5E-06 --
Figure C.11: 7th to 13th Columns of Balance Matrix, B, and Transpose of Master Stoichiometric Matrix, ST, for the main process module of Alternative 3. 276 178.0 68.1 3.8 ------CH4 (3)CH4 (3) Coal (3)Oil R. (3) O2 Pb (3)(3) Ni (3) Zn (4) O2 Pb (4)(4) Ni (4) Zn (4) CO2 (4) H2O Ni0000000001000 Ni000001000-1000 Pb0000000010000 Pb00001000-10000 Zn0000000000100 Zn0000001000-100 N20000000000000 N20000000000000 O20000000100000 O20001000-100000 Ash0000000000000 Ash0000000000000 SO20000000000000 H2S0000000000000 SO20000000000000 H2S0000000000000 CH41000000000000 Coal0100000000000 CO20000000000010 CO200000000000-10 H2O0000000000001 H2O000000000000-1 R. Oil0010000000000 R. Reactor easured Variables = Variables easured Separator
M
Figure C.12: 1st to 13th Columns of the Balance Matrix, B, for the Energy Sub-Module of Alternative 3. 277 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0-100000000000000 00-10000000000000 000-1000000000000 000000000-1000000 0000000000-100000 00000000000-10000 000000000000-1000 0000-100000000000 00000-10000000000 100000-1000000000 0100000-100000000 00100000-10000-10-1 00010000000000-10 ------0.0010 -- 4.6E-04 1.9E-05 9.4E-07 9.4E-07 9.4E-07 4.33 1.5E-04 3.08 -1000000000000000
SO2 (4)SO2 (4) N2 (4) Ash (4) H2S (6) CO2 (6) H2O (6) SO2 (6) N2 (6) Ash (7) O2 (7) Pb (7) Ni (7) Zn (7) Ash (7) H2S (8) Ash
Figure C.13: 14th to 30th Columns of the Balance Matrix, B, for the Energy Sub-Module of Alternative 3.
278
R13 R14 R15 R16 -1.00 0 0 0 0-1.000 0 00-1.000 -3.99 -1 -1.81 0.75 0000 0000 0000 2.74 1.00 1.45 0 2.25 0.47 1.34 -0.28 0 0.03 0.02 -1.00 00.010 0 00.150 0 0000.53 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000
------
Figure C.14: Transpose of Master Stoichiometric Matrix, ST, for Energy Sub-Module of Alternative 3.
279 NaCl (3) H2O(3) Hg (3) NaOH (4) Cl2 (4) H2 (4) NaCl100000 H2O010000 Reactor Hg001000 NaOH 0 0 0 -1 0 0 Cl20000-10 H200000-1 NaOH000100 Cl2000010 Separator H2000001 Hg000000
Measured Variables = 1,670.0 ------
Figure C.15: 1st to 6th Columns of the Balance Matrix, B, for the Main Process Sub- Module of Alternative 4.
Hg (4) NaOH (5) Cl2 (5) H2 (5) Hg (6) Cl2 (6) Hg (7) R12 0000000-1.00 0000000-0.31 -10000000 00000000.68 00000000.61 00000000.02 0-1000000 00-100-100 000-10000 1000-10-10
-- 1,000.0 893.0 16.9 3.2E-04 5.4E-04 1.5E-06 --
Figure C.16: 7st to 13th Columns of the Balance Matrix, B, and Transpose of Master Stoichiometric Matrix, ST, for Main Process Sub-Module of Alternative 4. 280
178.0 68.1 3.8 ------CH4 (3)CH4 (3) Coal Oil (3) R. (3) O2 Pb (3) (3) Ni (3) Zn (4) O2 Pb (4) (4) Ni (4) Zn CO2 (4) H2O (4) Ni000001000-1000 Ni0000000001000 Pb00001000-10000 Pb0000000010000 Zn0000001000-100 Zn0000000000100 O20001000-100000 N20000000000000 O20000000100000 N20000000000000 Ash0000000000000 Ash0000000000000 SO20000000000000 H2S0000000000000 SO20000000000000 H2S0000000000000 CH41000000000000 Coal0100000000000 CO200000000000-10 CO20000000000010 H2O000000000000-1 H2O0000000000001 R. Oil0010000000000 R. Reactor easured Variables = Separator
M
Figure C.17: 1st to 13th Columns of the Balance Matrix, B, for the Energy Sub-Module of Alternative 4.
281
0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0-100000000000000 00-10000000000000 000-1000000000000 000000000-1000000 0000000000-100000 00000000000-10000 000000000000-1000 0000-100000000000 00000-10000000000 100000-1000000000 0100000-100000000 00100000-10000-10-1 00010000000000-10 ------0.0010 -- 4.6E-04 1.9E-05 9.4E-07 9.4E-07 9.4E-07 4.33 1.5E-04 3.08 -1000000000000000
SO2 (4)SO2 (4) N2 (4) Ash (4) H2S (6) CO2 (6) H2O (6) SO2 N2 (6) (6) Ash (7) O2 Pb (7) (7) Ni (7) Zn (7) Ash (7) H2S (8) Ash
Figure C.18: 14th to 30th Columns of the Balance Matrix, B, for the Energy Sub-Module of Alternative 4. 282
R13 R14 R15 R16 -1.00 0 0 0 0-1.000 0 00-1.000 -3.99 -1 -1.81 0.75 0000 0000 0000 2.74 1.06 1.47 0 2.25 0.50 1.34 -0.28 0 0.00 0.00 -1.00 00.010 0 00.110 0 0000.53 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000
------
Figure C.19: Transpose of Master Stoichiometric Matrix, ST, for Energy Sub-Module of Alternative 4.
283
APPENDIX D
DETAILS FOR THE CASE STUDY OF THE HEAT EXCHANGER PROBLEM
IN SECTION 6.3.1
284 The heat duty Q is
Q = m& s c p,s ()Th1 − Th2 (D.1)
where m& s is the shell-side mass flow rate, cp,s is the specific heat of the shell-side fluid and Th1 and Th2 are the temperature of the inlet and outlet of the hot side, respectively.
The temperature at the outlet of the cold side Tc2 is
Q Tc2 = Tc1 + (D.2) m& t c p,t
where cp,t is the specific heat of the tube-side fluid and Tc1 is the temperature at the inlet of the cold side, respectively. The log-mean temperature difference ∆Tlmcf is
()()Th1 − Tc2 − Th2 − Tc1 ∆Tlmcf = (D.3) ln()()()Th1 − Tc2 Th2 − Tc1
The true mean temperature difference ∆Tm is
∆Tm = F∆Tlmcf (D.4) where the correction factor F is determined as follows
1/ 2 ()R 2 +1 ⎛ 1− P' ⎞ ln⎜ ⎟ R −1 1− P' R F = ⎝ ⎠ (D.5) 1/ 2 ⎛ 2 P' −1− R + ()R 2 +1 ⎞ ln⎜ ⎟ ⎜ 2 1/ 2 ⎟ ⎝ 2 P'−1− R − ()R +1 ⎠ with 285 1/ N ⎛1− PR ⎞ 1− ⎜ ⎟ ⎝ 1− P ⎠ P'= 1/ N (D.6) ⎛1− PR ⎞ R − ⎜ ⎟ ⎝ 1− P ⎠
T − T P = c2 c1 (D.7) Th1 − Tc1
T − T R = h1 h2 (D.8) Tc2 − Tc1
The Kern Method (Kakaç and Liu, 2002, pp. 307) is used to determine the heat transfer coefficient and pressure drop of the shell side. The shell-side heat transfer coefficient hs can be determined with McAdams correlation (Kakaç and Liu, 2002, pp.
308) as
0.55 1/ 3 0.14 hs De ⎛ DeGs ⎞ ⎛ c p,s µ s ⎞ ⎛ µ s ⎞ = 0.36⎜ ⎟ ⎜ ⎟ ⎜ ⎟ (D.9) ks ⎝ µ s ⎠ ⎝ ks ⎠ ⎝ µ w ⎠
3 Gs De 6 for 2×10 < Re s = < 1×10 (D.10) µ s
where De is the equivalent diameter of the shell side, Gs is the shell-side mass velocity and µw is the dynamic viscosity of the shell-side fluid at the average wall temperature. ks,
µs and cp,s, are respectively thermal conductivity, dynamic viscosity and isobaric specific heat of the shell-side fluid at average temperature.
286 The equivalent diameter of the shell, for a square pitch-tube layout, can be determined as:
2 2 4(PT − πd o 4) De = (D.11) πd o
where PT is the pitch size. The shell-side mass velocity is
m& s PT Gs = (D.12) Ds ()PT − d o B
where m& s is the shell-side mass flow rate, Ds is the shell inside diameter and B is the
baffle spacing. f s is the shell-side friction factor, calculated as
Gs De 6 f s = exp()0.576 − 0.19ln Re s for 400 < Re s = < 1×10 (D.13) µ s
The tube-side heat transfer coefficient ht can be determined by using Gnielinski’s correlation for turbulent flow, this is
h d ()(f 2 Re −1000 )Pr Nu = t i = t t t (D.14) t 1/ 2 2 / 3 kt 1+12.7()ft 2 ()Prt −1
ρt ut di 4 for Ret = > 10 (D.15) µt
where ut is the average velocity inside tubes and ft is the tube-side friction factor. These can be determined as
287 8m& t ut = 2 and (D.16) ρt N tπdi
−2 f t = ()1.58ln Ret − 3.28 (D.17)
The clean surface overall heat transfer coefficient Uc can be calculated as
1 U = (D.18) c d d ln()d d 1 o + o o i + di ht 2k hs where k is the thermal conductivity of the tube material. The fouled surface overall heat transfer coefficient Uf can be calculated as
1 U f = (D.19) 1 d o + R f ,t + R f ,s U c di
where Rf,t and Rf,s are the fouling factors of the tube-side and shell-side fluids, respectively. The transfer area Ao can then be calculated as
Q Ao = (D.20) U f ∆Tm
The length is
A L = o (D.21) πd o N t
The pressure drop on the shell side ∆ps is calculated as
288 2 fGs ()L B De ∆ps = 0.14 (D.22) 2ρ s De ()µ s µ w
where L is the effective tube length of heat exchanger between tube sheets and ρs is the shell-side fluid density at average temperature. The pressure drop on the tube side ∆pt is calculated as
⎛ LN ⎞ u 2 ⎜ p ⎟ t ∆pt = ⎜4 f t + 4N p ⎟ρt (D.23) ⎝ di ⎠ 2
where Np is the number of tube passes, di is the tube inside diameter.
Exergy calculations are obtained using Szargut et al. (1986) procedure. On the other hand, costing of the process is obtained from Turton et al. (1998).
289
BIBLIOGRAPHY
Adriaanse, A., Bringezu, S., Hammond, A., Moriguchi, Y., Rodenburg, E., Rogich, D., Schütz, H., 1997. Resource Flows: The Material Basis of Industrial Economies (1st edition). World Resources Institute, Washington D.C., 66 pp.
AIChE, 2000. Total Cost Assessment Methodology: Internal Managerial Decision Making Tool. Center for Waste Reduction Technologies, American Institute of Chemical Engineers, http://www.aiche.org/cwrt/pdf/TCAM.pdf (Accessed on November 30th, 2004).
AIChE, 2004. Focus Area: Sustainability Metrics. Center for Waste Reduction Technologies, American Institute of Chemical Engineers, http://www.aiche.org/cwrt/projects/sustain.htm, (Accessed on April 10, 2004).
Allen, D.T., Shonnard, D.R., 2002. Green Engineering: Environmentally Conscious Design of Chemical Processes. Prentice Hall, Upper Saddle River, NJ, 552 pp.
Anonymous Reviewer, 2003.
Arrow, K., Bolin, B., Costanza, R., Dasgupta, P., Folke, C., Holling, C.S., Jansson, B.-O., Levin, S., Mäler, K.-G., Perrings, C., Pimentel, D., 1995. Economic Growth, Carrying Capacity, and the Environment. Science 268(5210), 520-521.
Ayres, R.U., 1994. Information, Entropy and Progress: A New Evolutionary Paradigm (1st Edition). AIP Press, Woodbury, NY, 301 pp.
Ayres, R.U., 1995. Life Cycle Analysis – A Critique. Resources, Conservation and 290 Recycling 14(3-4), 199-223.
Ayres, R.U., 2000. Ecology vs. Economics: Confusing Production and Consumption. In: Perloff, H. (Ed.), Cybernetics, Ecology and Bioeconomics. Cybernetics Academy.
Ayres, R.U., Ayres, L.W., Martinas, K., 1998. Exergy, Waste Accounting and Life-Cycle Analysis. Energy 23(5), 355-363.
Azapagic, A., 1999. Life Cycle Assessment and Its Application to Process Selection, Design and Optimization. Chemical Engineering Journal 73(1), 1-21.
Azapagic, A., Clift, R., 1999. The Application of Life Cycle Assessment to Process Optimization. Computers and Chemical Engineering 23(10), 1509-1526.
Azapagic, A., Perdan, S., 2000. Indicators of Sustainable Development for Industry: A General Framework. Process Safety and Environmental Protection 78(B4), 243-261.
Bakshi, B.R., 2002. A Thermodynamic Framework for Ecologically Conscious Process Systems Engineering. Computers and Chemical Engineering 26(2), 269-282.
Bakshi, B.R., Fiksel, J., 2003. The Quest for Sustainability: Challenges for Process Systems Engineering. AIChE Journal 49(6), 1350-1358.
Bare, J.C., Hofstetter, P., Pennington, D.W., Udo de Haes, H.A., 2000. Life Cycle Impact Assessment Workshop Summary – Midpoints versus Endpoints: The Sacrifices and Benefits. The International Journal of Life Cycle Assessment 5(6), 319-326.
Bejan, A., 1997. Advanced Engineering Thermodynamics (2nd edition). John Wiley & Sons, New York, 850 pp.
Bejan, A., Tsatsaronis, G., Moran, M., 1996. Thermal Design and Optimization. John Wiley & Sons, New York, 543 pp.
Berthiaume, R., Bouchard, C., Rosen, M.A., 2001. Exergetic Evaluation of the Renewability of a Biofuel. Exergy 1, 256-268.
291 Bockstael, N.E., Freeman III, A.M., Kopp, R.J., Portney, P.R., Smith, V.K., 2000. On Measuring Economic Values for Nature. Environmental Science and Technology 34(8), 1384-1389.
Boltzmann, L., 1886. The Second Law of Thermodynamics. In: McGuiness, B.F. (Ed.), Theoretical Physics and Philosophical Problems: Selected Writings by Ludwig Boltzmann (1st edition). Springer, New York, 1974, 304 pp.
Brandt-Williams, S.L., 2001. Folio #4, Emergy of Florida Agriculture. Handbook of Emergy Evaluation, Center for Environmental Policy, Environmental Engineering Sciences, University of Florida, Gainesville, 40 pp.
Brown, M.T., Bardi, E., 2001. Folio #3, Emergy of Ecosystems. Handbook of Emergy Evaluation, Center for Environmental Policy, Environmental Engineering Sciences, University of Florida, Gainesville, 94 pp.
Brown, M.T., Herendeen, R.A., 1996. Embodied Energy Analysis and Emergy Analysis: a Comparative View. Ecological Economics, 19(3), 219-235.
Brown, M.T., Odum, H.T., Murphy, R.C., Christianson, R.A., Doherty, S.J., McClanahan, T.R., Tennenbaum, S.E., 1995. Rediscovery of the World: Developing an Interface of Ecology and Economics. In: Hall, C.A.S. (Ed.), Maximum Power: The Ideas and Applications of H.T. Odum, University Press of Colorado, Niwot, 393 pp.
Brown, M.T., Ulgiati S., 1997. Emergy-Based Indices and Ratios to Evaluate Sustainability: Monitoring Economies and Technology Toward Environmentally Sound Innovation. Ecological Engineering 9(1-2), 51-69.
Brown, M.T., Ulgiati S., 2002. Emergy Evaluations and Environmental Loading of Electricity Production Systems. Journal of Cleaner Production 10(4), 321-334.
Burgess, A.A., Brennan, D.J., 2001. Application of Life Cycle Assessment to Chemical Processes. Chemical Engineering Science 56(8), 2589-2604.
Cabrera-Rios, M., 2002. Multiple Criteria Optimization Studies in Reactive In-Mold Coating. Ph.D. Dissertation, Department of Industrial, Welding, and Systems Engineering, The Ohio State University, Columbus, 191 pp.
292 Campbell, D.E., 2001. Proposal for Including What is Valuable to Ecosystems in Environmental Assessments. Environmental Science & Technology 35(14), 2867- 2873.
Cano-Ruiz, J.A., McRae, G.J., 1998. Environmentally Conscious Chemical Process Design. Annual Review of Energy & the Environment 23(1), 499-536.
Chang, R., 2001. Chemistry (7th edition). McGraw-Hill, Boston, 1088 pp.
Chapman, D., 2000. Environmental Economics: Theory, Application, and Policy. Addison Wesley Longman, Reading, MA, 415 pp.
Charnes, A., Cooper, W.W., Lewin, A.Y., Seiford, L.M., 1993. Data Envelopment Analysis: Theory, Methodology, and Applications. Kluwer Academic Publishers, Boston, 513 pp.
Chen, H., Shonnard, D.R., 2004. Systematic Framework for Environmentally Conscious Chemical Process Design: Early and Detailed Design Stages. Industrial & Engineering Chemistry Research 43(2), 535-552.
Cleveland, C.J., Kaufmann, R.K., Stern, D.I., 2000. Aggregation and the Role of Energy in the Economy. Ecological Economics 32(2), 301-317.
Connelly, L.; Koshland, C.P., 2000. Two aspects of Consumption: Using an Exergy Based Measure of Degradation to Advance the Theory and Implementation of Industrial Ecology. Resources Conservation and Recycling 19(3), 199-217.
Cornelissen, R.L., Hirs, G.G., 2002. The Value of the Exergetic Life Cycle Assessment Besides the LCA. Energy Conversion and Management 43(9-12), 1417-1424.
Costanza, R., d’Agre, R., de Groot, R., Farber, S., Grasso, M., Hannon, B., Limburg, K., Ñame, S., O’Neil, R.V., Raruelo, J., Raskin, R.G., Sutton, P., van den Belt, M., 1997. The Value of the World’s Ecosystem Services and Natural Capital. Nature 387(6630), 253-260.
Crowe, C.M., 1986. Reconciliation of Process Flow Rates by Matrix Projection, Part II: The Nonlinear Case. AIChE Journal 32(4), 616-623. 293 Crowe, C.M., Garcia Campos, Y.A., Hrymak, A., 1983. Reconciliation of Process Flow Rates by Matrix Projection, Part I: Linear Case. AIChE Journal 29(6), 881-888.
Daily, G.C., 1997. Introduction: What Are Ecosystem Services? In: Daily, G.C. (Ed.) Nature’s Services: Societal Dependence on Natural Ecosystems. Island Press, Washington D.C., 392 pp.
Dasgupta, P., 2002. Is Contemporary Economic Development Sustainable? Ambio 31(4), 269-271. de Swaan Arons, J., Van Der Kooi, H.J., Sankaranarayanan, K., 2004. Efficiency and Sustainability in the Energy and Chemical Industries (1st edition). Marcel Dekker, New York, 299 pp.
Dewar, R.L., 2003. Information Theory Explanation of the Fluctuation Theorem, Maximum Entropy Production and Self-Organized Criticality in Non-Equilibrium Stationary States. Journal of Physics A: Mathematical and General 36(3), 631-641.
Dewulf, J., van Langenhove, H., 2002. Assessment of the Sustainability of Technology by Means of a Thermodynamically Based Life Cycle Analysis. Environmental Science and Pollution Research 9(4), 267-273.
Dewulf, J., Van Langenhove, H., Mulder, J., Van Den Berg, M.M.D., Van Der Kooi, H.J., de Swaan Arons, J., 2000. Illustrations Towards Quantifying the Sustainability of Technology. Green Chemistry 2(3), 108-114.
Diamond, J., 1995. Easter’s End – Easter Island. Discover August 1995, 63-69.
Dincer, I., 2000. Thermodynamics, Exergy and Environmental Impact. Energy Source 22, 723-732.
Diwekar, U.M., 2003. Greener by Design. Environmental Science and Technology 37(23), 5432-5444.
DoE, 2004. Chemical Bandwidth Study – Exergy Analysis: A Powerful Tool for Identifying Process Inefficiencies in the U.S. Chemical Industry. Industrial Technologies Program, Energy Efficiency and Renewable Energy, U.S. Department 294 of Energy. http://www.eere.energy.gov/industry/pdfs/chemical_bandwidth_report.pdf (Accessed on May 12th, 2004).
Douglas, J.M., 1988. Conceptual Design of Chemical Processes. McGraw-Hill, Boston, 601 pp.
Ekins, P., Simon, S., Deutch, L., Folke, C., de Groot, R., 2003. A Framework for the Practical Application of the Concepts of Critical Natural Capital and Strong Sustainability. Ecological Economics 44(2-3), 165-185.
Elkington, J., 1999. Triple Bottom Line: Implications for the Oil Industry. Oil and Gas Journal 97(50), 139-141.
El-Sayed, Y.M., Gaggioli, R.A., 1989. A Critical Review of 2nd Law Costing Methods .1. Background and Algebraic Procedures. Journal of Energy Resources Technology – Transactions of the ASME 111(1), 1-7.
Emblemsvag, J., Bras, B., 2001. Activity-Based Cost and Environmental Management, A Different Approach to ISO 14000 Compliance. Kluwer Academic Publishers, Boston, 317 pp.
EPA, 2004. LCAccess. http://www.epa.gov/ORD/NRMRL/lcaccess (Accessed on May 12th, 2005).
EPPO, 2005. Oil Industry Conversion. Energy Policy and Planning Office, Ministry of Energy, Royal Thai Government, Thailand. http://www.eppo.go.th/ref/UNIT- OIL.html (Accessed on May 12th, 2005).
European Environment Agency, 1997. Life Cycle Assessment – A guide to approaches, experiences and information sources. Environmental Issues Series 6, 119 pp.
Farber, S.C., Costanza, R., Wilson, M.A., 2002. Economic and ecological concepts for valuing ecosystem services. Ecological Economics 41, 375-392.
Fath, B.D., Patten, B.C., Choi, J.S., 2001. Complementarity of Ecological Goal Functions. Theoretical Biology 208(4), 493-506.
295 Feng, X., Zhu, X.X., 1997. Combining Pinch and Exergy Analysis for Process Modifications. Applied Thermal Engineering 17(3), 249-261.
Feng, X., Zhu, X.X., Zheng, J.P., 1996. A Practical Exergy Method for System Analysis. Presented at the 31st Inter-Society Energy Conversion Engineering Conference, Washington D.C., 2068-2071.
Fischer U., Hungerbuhler K., 2000. Application of indicators for assessing environmental aspects of chemical processes to case studies from pharmaceutical production. Chimia 54(9), 494-500.
Ford, K.W., Rochlin, G.I., Socolow, R.H., 1975. Efficient use of energy. American Institute of Physics, NY.
Franklin Associates, 1974. Franklin Associates’ U.S. LCI Database. http://www.fal.com (Accessed on April 7th, 2005).
Giannantoni, C., 2003. The Problem of the Initial Conditions and their Physical Meeting in Linear Differential Equations of Fractional Order. Applied Mathematics and Computation 141(1), 87-102.
Goedkoop and Spriensma, 2000. The Eco-Indicator 99, a Damage Oriented Method for Life Cycle Assessment: Methodology Report ( 2nd edition). PRé, Product Ecology Consultants, The Netherlands, 132 pp.
Goldstein, M., Goldstein, I.F., 1993. The Refrigerator and the Universe: Understanding the Laws of Energy. Harvard University Press, Cambridge, MA, 433 pp.
Graedel, T.E., 1998. Streamlined life-cycle assessment. Prentice Hall, Upper Saddle River, NJ, 310 pp.
Graedel, T.E., Allenby, B.R., 1995. Industrial ecology. Prentice Hall, Englewood Cliffs, N.J., 412 pp.
Graedel, T.E., Allenby, B.R., 2002. Hierarchical Metrics for Sustainability. Environmental Quality Management 12(2), 21-30.
296 GREET, 2005. Argonne National Laboratory, U.S. Department of Energy. http://greet.anl.gov/download.html (Accessed on May 31st, 2005).
Guinée, J.B., Heijungs, R., Udo de Haes, H.A, Huppes, G., 1993b. Quantitative Life Cycle Assessment of Products 2: Classification, valuation and improvement analysis. Journal of Cleaner Production 1(2), 81-91.
Guinée, J.B., Udo de Haes, H.A., Huppes, G., 1993a. Quantitative Life Cycle Assessment of Products 1: Goal Definition and Inventory. Journal of Cleaner Production 1(1), 3- 13.
Hall, C.A.S., 1995a. Introduction: What is Maximum Power. In: Hall, C.A.S. (Ed.), Maximum Power: The Ideas and Applications of H.T. Odum, University Press of Colorado, Niwot, 393 pp.
Hall, C.A.S., 1995b. Emergy Analysis. In: Hall, C.A.S. (Ed.), Maximum Power: The Ideas and Applications of H.T. Odum, University Press of Colorado, Niwot, 393 pp.
Hau, J.L., 2002. M.S. Thesis, Department of Chemical and Biomolecular Engineering, The Ohio State University, Columbus, 143 pp.
Hau, J.L., Bakshi, B.R., 2004a. Expanding Exergy Analysis to Account for Ecosystem Products and Services. Environmental Science and Technology 38(13), 3768-3777.
Hau, J.L.; Bakshi, B.R, 2004b. Promise and Problems of Emergy Analysis. Ecological Modelling 178(1-2), 215-225.
Hediger, W., 1999. Reconciling “Weak” and “Strong” Sustainability. International Journal of Social Economics 26(7-9), 1120-1143.
Heinzle E, Weirich D, Brogli F, Hoffmann VH, Koller G, Verduyn MA, Hungerbuhler K, 1998. Ecological and economic objective functions for screening in integrated development of fine chemical processes. 1. Flexible and expandable framework using indices. Industrial & Engineering Chemistry Research 37(8), 3395-3407.
Heiskanen, E., 2000. Managers’ Interpretation of LCA: Enlightenment and Responsibility or Confusion and Denial? Business Strategy and the Environment 9, 239-254. 297 Hendrickson, C., Horvath, A., Joshi, S., Lave, L., 1998. Economic Input-output Models for Environmental Life-Cycle Assessment. Environmental Science and Technology 32, 184A-191A.
Hoffmann, V.H., Hungerbuhler, K., McRae, G.J., 2001. Multiobjective screening and evaluation of chemical process technologies. Industrial & Engineering Chemistry Research 40(21), 4513-4524.
Hoffmann, V.H., McRae, G.J., Hungerbuhler, K., 2004. Methodology for early-stage technology assessment and decision making under uncertainty: Application to the selection of chemical processes. Industrial & Engineering Chemistry Research 43(15), 4337-4349.
Hofstetter, P., Baumgartner, T., Scholz, R.W., 2000. Modeling the Valuesphere and the Ecosphere: Integrating the Decision Makers’ Perspectives into LCA. The International Journal of Life Cycle Assessment 5(3), 161-175.
Holliday, C., 1999. DuPont CEO Sees Sustainability as Main Industrial Challenge of the New Century. http://www.dupont.com/corp/news/speeches/holliday_11_29_99.html (Accessed on November 13th, 2001).
Holliday, Jr, C.O., Schmidheiny, S., Watts, P., 2002. Walking the Talk - The Business Case for Sustainable Development. Greenleaf Publishing, Sheffield, U.K, 288 pp.
Horlock, J.H., 1992. Combined Power Plants: Including combined cycle gas turbine (CCGT) plants (1st edition). Pergamon Press Ltd, Oxford, 184 pp.
Huppes, G., 1994. A general method for allocation. Proceedings of the European workshop on allocation in LCA, Leiden, SETAC-Europe, Brussels, February 1994.
IChemE, Institution of Chemical Engineers. http://www.icheme.org/sustainability (Accessed on April 10th, 2004).
ISO 14040, 1997. Environmental Management – Life Cycle Assessment – Principles and Framework. International Organization for Standardization. http://www.iso.org (Accessed on November 30th, 2004).
298 Jensen, A.A., Hoffman, L., Møller, B.T., Schmidt, A., Christiansen, K., Berendsen, S., Elkington, J., Van Dijk, F., 1997. Life cycle assessment (LCA): a guide to approaches, experiences and information sources. Copenhagen, Denmark: European Environment Agency; Luxembourg: Office for Official Publications of the European Communities, 116 pp.
Jiménez-González, C., Kim, S., Overcash, M.R., 2000. Methodology for Developing Gate-to-Gate Life Cycle Inventory Information. The International Journal of Life Cycle Assessment 5(3), 153-159.
Jørgensen, S.E., 2002. Integration of Ecosystem Theories: A Pattern (3rd edition). Kluwer Academic Publishers, Boston, 420 pp.
Jorgensen, S.E., Patten, B.C., Straskraba, M., 1992. Ecosystems emerging: toward an ecology of complex system in a complex future. Ecological Modelling 62, 1-27.
Kakaç, S., Liu, H., 2002. Heat Exchangers: Selection, Rating and Thermal Design. CRC Press, Boca Raton, FL, 501 pp.
Lagerberg, C., Brown, M.T., 1999. Improving agricultural sustainability: The case of Swedish greenhouse tomatoes. Journal of Cleaner Production 7(6): 421-434.
Lave L.B., Cobas-Flores, E., Hendrickson, C.T., McMichael, F.C., 1995. Using input- output analysis to estimate economy-wide discharges. Environmental Science & Technology 29(9), 420A-426A.
Lazzaretto, A., Toffolo, A., 2004. Energy, Economy and Environment as Objectives in Multi-criterion Optimization of Thermal Systems Design. Energy 29, 1139-1157.
Lee, J., O’Callaghan, P., Allen, D., 1995. Critical review of life cycle analysis and assessment techniques and their application to commercial activities. Resources, Conservation & Recycling 13, 37-56.
Lems, S., van der Kooi, H.J., de Swaan Arons, J., 2002. The Sustainability of Resource Utilization. Green Chemistry 4, 308-313.
Lenzen, M., 2001. Errors in conventional and input-output-based life-cycle inventories. 299 Journal of Industrial Ecology 4(4), 127-148.
Leontief, W.W., 1986. Input-Output Economics (2nd edition). Oxford University Press, New York, 436 pp.
Linnhoff, B., 1997. Introduction to Pinch Analysis. In: Boehm, R.F. (Ed.), Developments in the Design of Thermal Systems. Cambridge University Press, Cambridge, UK, 1997, 288 pp.
Lombardi, L., 2001. Life cycle assessment (LCA) and exergetic life cycle assessment (ELCA) of a semi-closed gas turbine cycle with CO2 chemical absorption. Energy Conversion & Management 42, 101-114.
Lorenz, R., 2003. Full Steam Ahead—Probably. Science 299, 837-838.
Lotka, A., 1922. Contributions to the Energetics of Evolution: Natural Selection as a Physical Principle. Proceedings of the National Academy of Science 8, 147-154.
Mallick, S.K., Cabezas, H., Bare, J.C., Sikdar, S.K., 1996. A pollution reduction methodology for chemical process simulators. Industrial & Engineering Chemistry Research 35(11), 4128-4138.
Mansson, B.A., McGlade, J.M., 1993. Ecology, thermodynamics and H.T. Odum’s conjectures. Oecologia 93, 582-596.
Maple, R.E., 2000. Petroleum Refinery Process Economics (2nd edition). PennWell Corporation, Tulsa, 474 pp.
Marteel, A.E., Davies, J.A., Olson, W.W., Abraham, M.A., 2003. Green Chemistry and Engineering: Drivers, Metrics, and Reduction to Practice. Annual Review of Environment and Resources 28(1), 401-428.
Matthews, E., Amann, C., Bringezu, S., Fischer-Kowalski, M., Hüttler, W., Kleijn, R., Moriguchi, Y., Ottke, C., Rodenburg, E., Rogich, D., Schandl, H., Schütz, H., van der Voet, E.; Weisz, H., 2000. The Weight of Nations: Material Outflows from Industrial Economies (1st edition). World Resources Institute, Washington D.C., 125 pp.
300 Mellor, W., Wright, E., Clift, R., Azapagic, A., Stevens, G., 2002. A mathematical model and decision-support framework for material recovery, recycling and cascade use. Chemical Engineering Science 57, 4697-4713.
Miller, R.E., Blair, P.D., 1985. Input-Output Analysis: Foundations and Extensions. Prentice Hall, Englewood Cliffs, NJ.
Morris, D.R., 1991. Exergy Analysis and Cumulative Exergy Consumption of Complex Chemical Processes: The Industrial Chlor-Alkali Processes. Chemical Engineering Science 46(2), 459-465.
Munoz, J.R., Michaelides, E.E., 2000. The Impact of the Model of the Environment in Exergy Analyses. Journal of Energy Resources Technology 121, 268-276.
Narasimhan, S., Jordache, C., 2000. Data Reconciliation & Gross Error Detection: An Intelligent Use of Process Data (1st Edition). Gulf Publishing Company, Houston, 406 pp.
Nordhaus, W.D., Kokkelenberg, E.C., 1999. Nature’s Numbers, Expanding the National Economic Accounts to Include the Environment. National Academy Press, Washington D.C., 250 pp.
NREL, 2005. The National Renewable Energy Laboratory. http://www.nrel.gov (Accessed on May 31st, 2005).
Odum, E.P., 1953. Fundamentals of Ecology. Saunders, Philadelphia, 384 pp.
Odum, E.P., 1971. Fundamentals of Ecology (3rd edition). Saunders, Philadelphia, 574 pp.
Odum, H.T., 1984. Embodied energy, foreign trade and welfare of nations. In: Jansson, A-M. (Ed.), Integration of Economy and Ecology—An Outlook for the Eighties. Proceedings of the Wallenberg Symposium, Stockholm, Asko Laboratory, University of Stockholm, 185-199.
Odum, H.T., 1988. Self-Organization, Transformity, and Information. Science 242, 1132- 1139.
301 Odum, H.T., 1995a. Energy-Systems Concepts and Self-Organization – A Rebuttal. Oecologia 104(4), 518-522.
Odum, H.T., 1995b. Self-Organization and Maximum Empower. In: Hall, C.A.S. (Ed.), Maximum Power: The Ideas and Applications of H.T. Odum, University Press of Colorado, Niwot, 393 pp.
Odum, H.T., 1996. Environmental Accounting: Emergy and Environmental Decision Making (1st edition). John Wiley & Sons, New York, 370 pp.
Odum, H.T., 2000. Folio #2: Emergy of Global Processes. Handbook of Emergy Evaluation, Center for Environmental Policy, Environmental Engineering Sciences, University of Florida, Gainesville, 28 pp.
Odum, H.T., Brown, M.T., Brandt-Williams, S.L., 2000. Folio #1, Introduction and Global Budget. Handbook of Emergy Evaluation, Center for Environmental Policy, Environmental Engineering Sciences, University of Florida, Gainesville, 17 pp.
Odum, H.T., Odum, E.C., 1981. Energy Basis for Man and Nature. McGraw-Hill, New York, 337 pp.
Patten, B.C., 1993. Toward a More Holistic Ecology, and Science—the Contribution of Odum, H.T. Oecologia 93(4), 597-602.
Patterson, M., 1993. Approaches to energy quality in energy analysis. International Journal of global Energy Issues. Special Issue on Energy Analysis, 19-28.
Pellegrino, J.L., 2000. Energy and environmental profile of the U.S. chemical industry, ENERGETICS, Columbia. http://www.eere.energy.gov/industry/chemicals (Accessed on May 12th, 2005).
PRé, 1990. SimaPro. http://www.pre.nl/simapro (Accessed on April 7th, 2005).
Riegel, E.R., 1992. Riegel’s Handbook of Industrial Chemistry (9th Edition). Kent, J.M. (Ed.), van Nostrand Reinhold Company, New York, 1288 pp.
302 Rifkin, J. 2003. The Hydrogen Economy: The Creation of the Worldwide Energy Web and the Redistribution of Power on Earth. Jeremy P. Tarcher/Penguin, New York, 294 pp.
Romagnoli, J.A., Sánchez, M.C., 2000. Data Processing and Reconciliation for Chemical Processes Operations (1st Edition). Academic Press, San Diego, 270 pp.
Sandler, S.I., 1989. Chemical and Engineering Thermodynamics. John Wiley & Sons, New York, 622 pp.
Schaltegger, S. (Ed.), 1996. Life cycle assessment (LCA) – Quo Vadis? Birkhäuser Verlag, Boston, 187 pp.
Schaltegger, S., 1997. Economics of Life Cycle Assessment: Inefficiency of the Present Approach. Business Strategy and the Environment 6, 1-8.
Schwarz, J., Beaver, E., Beloff, B., 2000. Sustainability Metrics: Making Decisions for Major Chemical Products and Facilities. BRIDGES to Sustainability, Houston. www.bridgestos.org (Accessed on May 12th, 2005).
Schwarz, J., Beloff, B., Beaver, E., 2002. Use Sustainability Metrics to Guide Decision- Making. Chemical Engineering Progress 8, 58-63.
Sciubba, E., 2001. Beyond Thermoeconomics? The Concept of Extended Exergy Accounting and its Application to the Analysis and Design of Thermal Systems. Exergy an International Journal 1(2), 68-84.
Seager, T.P., Theis, T.L, 2002. A Uniform Definition and Quantitative Basis for Industrial Ecology. Journal of Cleaner Production 10(3), 225-235.
Seider, W.D., Seader, J.D., Lewin, D.R., 2003. Process Design Principles: Synthesis, Analysis, and Evaluation (2nd edition). John Wiley & Sons, New York, 820 pp.
Shonnard, D.R., Hiew, D.S., 2000. Comparative Environmental Assessments of VOC Recovery and Recycle Design Alternatives for Gaseous Waste Streams. Environmental Science & Technology 34(24), 5222-5228.
303 Shreve, R.N., Brink, J.A., 1977. Chemical Process Industries, McGraw-Hill, New York, 813 pp.
Sikdar, S.K., 2003. Sustainable Development and Sustainability Metrics. AIChE Journal 49, 1928-1932.
Singh, N., Kaushik, S.C., Misra, R.D., 2000. Exergetic Analysis of a Solar Thermal Power System. Renewable Energy 19(1-2), 135-143.
Spreng, D.T., 1988. Net-Energy Analysis and the Energy Requirements of Energy Systems. Praeger Publishers, New York, 289 pp.
Stefanis, S.K., Livingston, A.G., Pistikopoulos, E.N., 1995. Minimizing the Environmental Impact of Process Plants: A Process Systems Methodology. Computers & Chemical Engineering 19 (Suppl.), S39-S44.
Stromberg, L. et al., 1997. Guidelines on life cycle inventory analysis of pulp and paper. Nordpap DP 2/30, February 1997.
Suh, S., Lenzen, M., Treloar, G.J., Hondo, H., Horvath, A., Huppes, G., Jolliet, O., Klann, U., Krewitt, W., Moriguchi, Y., Munksgaard, J., Norris, G., 2004. System Boundary Selection in Life-Cycle Inventories Using Hybrid Approaches. Environmental Science and Technology 38, 657-664.
Sussman, M.V., 1980. Availability (exergy) Analysis: A Self Instruction Manual. Mulliken House, Lexington, MA, 100 pp.
Szargut, J., Morris, D.R., Steward, F.R., 1988. Exergy Analysis of Thermal, Chemical and Metallurgical Processes, (1st Edition). Hemisphere Pubs., New York, 332 pp.
Tilman, D., Cassman, K.G., Matson, P.A., Naylor, R., Polasky, S., 2002. Agricultural sustainability and intensive production practices. Nature 418(6898), 671-677.
Treloar, G., 1997. Extracting Embodied Energy Paths from Input-Output Tables: Towards An Input-Output-Based Hybrid Energy Analysis Method. Economic Systems Research 9(4), 375-391.
304 Turton, R., Bailie, R.C., Whiting, W.B., Shaeiwitz, J.A., 1998. Analysis, Synthesis, and Design of Chemical Processes (1st Edition). Prentice Hall, New Jersey, 814 pp.
Ukidwe, N.U., and Bakshi, B.R., 2003. Accounting for ecosystem contribution to economic sectors by thermodynamic input-output analysis, part I. Approach, Technical Report, Department of Chemical and Biomolecular Engineering, The Ohio State University, Columbus, 18 pp.
Ukidwe, N.U., Bakshi, B.R., 2004. Thermodynamic Accounting of Ecosystem Contribution to Economic Sectors with Application to 1992 U.S. Economy. Environmental Science and Technology 38, 4810-4827.
Ulgiati, S., 1999. Energy, Emergy and Embodied Exergy: Diverging or Converging Approaches, Proceedings of the First Biennial Emergy Analysis Conference, Gainesville, Florida.
UNEP, 1999. Towards the global use of life cycle assessment. United Nations Environment Programme, Division of Technology, Industry and Economics, Production and Consumption Unit; United States Environmental Protection Agency; Centre of Environmental Science, Leiden University; France.
Valero, A., Lozano, M.A., 1997. An Introduction of Thermoeconomics. In: Boehm, R.F. (Ed.), Developments in the Design of Thermal Systems. Cambridge University Press, Cambridge, UK, 288 pp.
Valero, A., Lozano, M.A., Serra, L., Tsatsaronis, G, Pisa, J., Frangopoulos, C., von Spakovsky, M.R., 1994. CGAM Problem: Definition and Conventional Solution. Energy 19(3), 279-286.
WCED, 1987. Our Common Future. The World Commission on Environment and Development, Oxford University Press, Oxford and New York, NY.
Weidema, B., 2001. Avoiding Co-Product Allocation in Life-Cycle Assessment. Journal of Industrial Ecology 4(3), 11-33.
Wirdak, C., 2003. GEMI Survey: EHS Metrics and Processes. Global Environmental Management Initiative (GEMI). http://www.gemi.org/docs/bench/bench.htm (Accessed on May 12th, 2005). 305 WRI, 2000. World Resources 2000-2001: People and Ecosystems: The Fraying Web of Life. World Resources Institute, Washington D.C. http://www.wri.org/wr2000 (Accessed on May 12th, 2005).
WWF, 2000. Living Planet Report 2000. World Wildlife Fund. http://panda.org/livingplanet/lpr00 (Accessed on May 12th, 2005).
Yi, H.-S., Bakshi, B.R., 2004. Hierarchical Thermodynamic Sustainability Metrics for Ammonia Process – Detailed Calculations, Technical Report, The Ohio State University, www.che.eng.ohio-state.edu/~bakshi/AmmoniaMetrics.html (Accessed on May 12th, 2005).
306