a. .Q l "nemva "no” W"“V‘§‘C"\ l u _ ‘1': ugh-19“ 9.... : n31 . » .3 Sq. 7 . .z. A o. 73 .1. .I ‘ .: S» ‘ .5 :4. s .25. .: THESIS This is to certify that the dissertation entitled INTEGRATION OF ENVIRONMENT INTO PRODUCT DESIGN AND MANUFACTURING: THEORY AND IMPLEMENTATION presented by Youngsun Chun has been accepted towards fulfillment of the requirements for Ph.D. degreein EIECtricaI Eng [£2 ;[1 §JRAJVKa¢uaA,IZ——. Major professor 3 - 7‘- 02000 Date MSU i: an Affirmative Action/Equal Opportunity Institution 0- 12771 ————A E* LIBRARY Miehlgan State Univeretty PLACE IN RETURN BOX to remove this checkout from your record. TO AVOID FINES return on or before date due. MAY BE RECALLED with earlier due date if requested. DATE DUE DATE DUE DATE DUE 6/01 eJCIF-IC/DateDuepBS-pJS ‘ l INTEGRATION OF ENVIRONMENT INTO PRODUCT DESIGN AND MANUFACTURING: THEORY AND IMPLEMENTATION By Youngsun Chun A DISSERTATION Submitted to Michigan State University in partial fulfilment of the requirments for the degree of DOCTOR OF PHILOSOPHY Department of Electrical and Computer Engineering 2000 ABSTRACT Integration of Environment into Product Design and Manufacturing: Theory and Implementation by Youngsun Chun Environmentally Conscious Manufacturing(ECM) has two major goals such as developing green technology and analytical modeling tool which can assess the consequences of different strategies. Overall, modeling interactions between manufacturing plants and environment is very huge and complex task. Most quantitative tools are concerned with environmental accounting sys- tem without concerning of feasibility and impact of process network struc- ture. This disseratation is an attempt to develop a framework to study the impact of alternative technologies, strateges, and designs based on process network theory. As a result, a computer modeling and simulation tool, Mass- Energy Based Simulation(MEBS), instantiating an ECM tool is presented. Copyright © 2000 By Youngsun Chun Dedicated to My Father and Mother 580]: Hyun Chan and Hyung Ae K im iv ACKNOWLEDGEMENTS Looking upon the pathway that I have chosen, this is the moment to acknowledge those who shaped me and helped me for being where I am. My first thank goes to my beloved mother and father, and my family. They made me proud of what I am. Also my gratitude and appreciation goes to my academic advisor Dr. Lal Thmmala for his guidance and financial support during my research work. This dissertation could not be completed without his support. And this work is partially supported by NSF grant DMI—9528759. I am also grateful to my guidance committee members: Dr. Hassan Khalil, Dr. Steven Melnyk, and Dr. Charles MacCluer for their interests, guidance, advice, and encouraging. lam indebted to Ms. Mary and Mr. Norman Robison, Mr. Charles McNease, Mr. H. Kay for their encouragement and friendship. Ms. and Mr. Robison, taught me how to communicate in written English. Mr. McNease, beginning with one of my volunteer examiners for my amateur radio license, shared lots of his experiences and friendship. Finally, I am asking forgiveness to my wife Kyounghwa, and to my dear children Sungah and Sungwoo for not sharing time together much. Contents I INTRODUCTION 1 1 Introduction , l 1.1 Why ECM? ............................ 2 2 Previous Work 8 2.1 Qualitative methods ....................... 16 2.2 Quantitative methods ....................... 18 3 Problem Statement 20 II Methodology 23 4 System and Modeling 23 4.1 System Definition ......................... 23 4.2 Model ............................... 24 5 Methodology 26 6 What is NIEB model? 29 6.1 Overview .............................. 30 6.1.1 Production Process .................... 31 6.1.2 Recycling Process ..................... 33 6.1.3 Storage Process ...................... 34 vi 6.1.4 Junction Process ..................... 6.1.5 Goal Process ....................... 6.1.6 Wire Object ........................ 6.1.7 Library Process ...................... 7 NIEB NETWORK 7.1 MEB Graph Representation ................... 7.2 Inputs and Outputs ........................ 8 NIEB Language(NIEBL) 8.1 System Structure ......................... 8.2 Netlist ............................... 8.3 Variables .............................. 8.4 Constants ............................. 8.5 Control Flow ........................... 8.6 Database .............................. 8.7 Expression ............................. 8.7.1 Matrix ........................... 8.7.2 Vector ........................... 8.8 Operators ............................. 8.9 Output Functions ......................... 8.10 Math Functions .......................... 9 Process Network Execution Model vii 36 37 46 48 50 53 53 54 62 . 63 63 66 68 68 69 69 69 69 73 9.1 Execution of a Process ...................... 76 10 Graphic User Interface 87 10.1 Screen Preperty Composition .................. 90 10.2 Graphic Database(GDB) ..................... 92 11 Case Studies 94 11.1 Swine/ Crop System ........................ 94 11.2 Paper Cup LCA .......................... 94 11.3 Ford F150 'Ii‘uck Tail Light Assembly .............. 113 11.4 Water Plant Modelling ...................... 113 III CONCLUSION 125 12 Conclusion 125 12.1 Contributions ........................... 130 12.2 Future Direction ......................... 131 IV APPENDICES 134 A INTRODUCTION 139 B GETTING STARTED 140 B.l Convention ............................ 140 8.2 Overview .............................. 140 viii 3.3 Example .............................. 143 R31 Drawing an Example Plant ............... 143 8.3.2 Simulation of an Example Plant ............. 146 8.3.3 Defining Library ..................... 150 C MEBL SYNTAX 152 (3.1 Plant Structure .......................... 152 C2 Netlist ............................... 155 0.3 Node Structure .......................... 155 0.4 Wire Variables .......................... 156 C5 Read Only Variables ....................... 156 C6 Constants ............................. 157 D CONTROL FLOW 157 DJ If Else ............................... 158 D.2 For ................................. 158 D.3 While ............................... 158 DA Loop Control ........................... 159 D5 Comments ............................. 159 E DATABASE 159 El Database Structure ........................ 159 E2 Database Statement ....................... 160 F EXPRESSION 161 F.1 Matrix ............................... F.2 Vector ............................... F.3 Operators ............................. F.4 Output Functions ......................... F.5 Math Emctions .......................... USER’S REFERENCE MANUAL MEBL GRAMMAR AN NIEBL PROGRAM EXAMPLE About the CD Rom J .1 System Requirements ....................... J.2 Installation ............................ 162 162 162 166 168 172 178 178 179 List of Tables NTQCHA Benefits of DfE implementation and ECM strategies used by a sample of domestic manufacturers ............... UnitszELU / kg. Source: B.Steen and S.Ryding, The EPS Enviro- Accounting Method: An Application of Environmental Ac- counting Principles for Evaluation and Valuation of Environ- mental Impact in Product Design, Stockholm:Swedish Envi- ronmental Research Institute(IVL),1992. ............ The product use is based on 1 year of use. Calculation of ELU for automobile front ends. Source: S.Ryding, B.Steen, A.Wenblad, and R.Karlson, The EPS system - An LCA con- cept for cleaner Technology and Product Development Strate- gies, and Design for the Environment, Paper presented at EPA Workshop on Identifying a Hamework for Human Health and Environmental Risk Ranking, Washington, DC, June 30-July Inbound wire;N,o,,, x Noam", constraints in tabular form Outbound wire; NM, x New,” constraints in tabular form . . Block node subclass number used in MEBL .......... Precedences of operators ..................... 14 52 52 57 7O 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 Conversion table to transform MEB graph into MEB Petri net in top-down approach; row and column are current or next MEB block class respectively and the first and second element of a two-tuple is for inbound and outbound wire respectively . 82 Conversion table to transform MEB graph into MEB Petri net in bottom-up approach; row and column are current or next MEB block class respectively and the first and second element of a two-tuple is for inbound and outbound wire respectively . 82 Inbound wire;Nm,, x Nwzumn constraints in tabular form 90 Inbound wire;Nm,, x Neda,“ constraints in tabular form 91 Unit cost of material flux ..................... 95 The technical coefficients of swine/crop agroecosystem 1 96 The technical coefficients of swine/crop agroecosystem 2 97 The technical coefficients of swine/crop agroecosystem 3 98 The technical coefficients of swine/crap agroecosystem 3 99 Unit input costs .......................... 100 Swine/Crop agroecosystem 1 ................... 102 Swine/ Crop agroecosystem 2 ................... 103 Swine/CrOp agroecosystem 3 ................... 104 The technical coefficients of paper manufacturing ....... 105 The technical coefficients of cup use and disposal ....... 106 The technical coefficients of cup use and diSposal ....... 107 The unit costs of system resources for paper cup use and disposallll 25 26 27 28 F150 tail light assembly modeling 1 ............... 114 F 150 tail light assembly modeling 2 ............... 115 Block node subclass number used in MEBL .......... 154 Precedences of operators ..................... 163 xiii List of Figures 00 consume. 10 11 12. 13 14 15 16 17 Sources of solid waste ....................... 4 Activities in the five life-cycle stages of a product ....... 9 The environmentally responsible product assessment matrix and the target plot ........................ 10 Facilities LCA matrix ....................... 11 Matrix: Supply line to environmental design practices ..... 12 Application of the Eco-Indicator as a tool ........... 13 Life cycle modeling of manufacturing .............. 17 Integrated approach to manufacturing system analysis and de- sign ................................ 22 Abstraction of material transformation ............. 32 Multi layered tree structured description of a system ..... 38 System as a component at the next level ............ 4O Illustrative process network diagram for a manufacturing en- terprise ............................... 42 Extended process network diagram for a manufacturing enter- prise ................................ 44 MEB simulation architecture ................... 49 A system with junction between processes ........... 55 Automatically generated netlist in MEBL ............ 56 Example of a database file “elu. db” to show database structure 67 xiv 18 19 20 21 22 23. 24 25 26 27 28 29 3O 31 32 33 34 35 Translation of the MEB graph figure 8.1 into tOp-down Petri net ................................. 78 Algorithm for execution of a Petri net .............. 81 A system model having all classes ................ 83 Translation of the MEB graph figure 9.1 into top-down Petri net ................................. 86 Moore machines for GUI interactions .............. 89 Simulation environment set—up for the system in Table 8.1. . . 92 MEB GDB table of the system in Figure 8.1 .......... 93 MEB network swine/crop agroecosystem ............ 101 MEB model of paper manufacturing plant ........... 108 MEB model of paper use and disposal system ......... 109 MEB model for paper cup life cycle analysis .......... 110 Ford F150 tail light assembly plant ............... 116 Recycle option I .......................... 117 Integrated approach to manufacturing system analysis and de- sign ................................ 128 An example plant ......................... 144 A procedure to draw a plant ................... 147 XV 36 37 38 39 40 Pop—up capacity query window ................. 148 Pop—up cost query window .................... 148 A procedure to simulate the plant ................ 149 Capacity vs. cost 2—D plot .................... 150 Procedure to define a library part ................ 153 xvi Part I INTRODUCTION 1 Introduction Traditionally, products are designed for their appearance, technical (electri— cal, mechanical, and etc.) performance, and functionality. The environmen- tal impact of this design on the manufacturing processes, product use and disposal are seldom considered. With the increasing demands on conser- vation of natural resources and environment, modern firms have begun to incorporate environmental concerns into product design and manufacturing. This process is variously called environmentally conscious product design, en- vironmentallly responsible product design, design for environment(DfE) or green design[24, 10]. As an extension of existing DD((Design for X) strate- gies, the DfE focus begins at the product development stage and runs all the way through the distribution[29]. It is important to recognize that decisions made during the design phase have a profound impact on the entire life cycle which involves the manufacture, product use, and product reuse or disposal [40]. Manufacturing systems that incorporate environmental considerations similarly are called Environmentally Conscious Manufacturing system(ECM) or green manufacturing systems. Billatos and Basaly [33] define the goals of Eivironmentally Conscious Manufacturing(ECM), also coined as green engi— neering or green technology, as follows: Waste reduction is justified based on financial analysis without concern for the added environmental benefits. Total Quality Management(TQM) and Just In-Time(JIT) manufacturing are example strategies for achiev- ing this goal. Materials management aims for economical recovery of materials or fin- ished products for reuse. The three categories of strategies to achieve this goal are Design for recycling(DfR), Design for disassembly(DfD), and toxic management. Pollution prevention has the goal of eliminating the use of manufacturing processes that generate pollution. This differs from pollution control, also known as end—o -pipe(EOP) solution, which refers to the treatment of harmful by-products after they have been produced. Product enhancement is a design activity to reduce resource require- ments, waste, and pollution during product’s use through its Operable life, usually motivated by regulations to control harmful by-products. 1.1 Why ECM? Pepulation grows in a geometric ratio in an environment which supplies un- limited resources and tolerates unlimited waste, unless the environment is managed. In reality this exponential growth is not true, because every pop- ulation depends on others in one way or another, the earth has limited re- sources, and humans have limited tolerance of waste in the environment[12]. There are many signs of environmental stress indicating that the health of the environment today is worse than that of yesterday. If management or regulation toward sustainable products and services which can be produced indefinitely without adding any environmental stress are not done, the only one earth we share with others will be worse day by day until disaster may strike all of us. The importance of supporting the environment is increas- ing as both the products and services demanded by the human population grow at the cost of environment resources and continuous increase in the human population. While better technologies and more focused effort by individuals have increased the productivity and services, the pressure under- standing interactions among different entities such as plants, economies, and environmental loads also have increased. Considering industry is the major producer of the solid waste as in Fig- ure 1, environmentally friendly design in manufacturing plants would greatly affect the rest of the life cycle of a product. Due to increasing environmental awareness, the companies have recog- nized the economic and social advantages of designing and manufacturing environmentally responsible products, so called “green products” and placed greater emphasis on incorporating environmental concerns into product de- sign and manufacturing. AT&T, Xerox, Intel, Hewlett Packard, Tektronix, 3M, and Texas Instruments corporations have integrated DfE concepts into their product development and design and the benefits of ECM strategies are shown in Table 1 from [29]. For example, Intel Corporation has retooled their Other 17.2% Municipal solid waste 9% Agricultural Manufacturing 60% Sources of solid waste(U.S. Congress, Office of Technology Assessment) Figure 1: Sources of solid waste Corporation Estimated Savings DfE Strategies Intel $1 Million Recycling/ Reuse Xerox ' $200 Million Remanufacturing Hewlett-Packard $17 Million Recycling 3M $1 Billion Recycling, Remanufacturing Table 1: Benefits of DfE implementation and ECM strategies used by a sample of domestic manufacturers product design so they do not generate waste in the first place. This lead to savings on chemical purchases, as well as on disposal costs. Similarly, Xerox and 3M Corporation have incorporated remanufacturing and recycling into their design. Once expired products are returned, the parts are segregated into reusable and unusable parts which will be made available as spare parts for newly manufactured products. To minimize the impact on the environment, designers should took into consideration the materials used, energy efficiency of the processes used, wastes generated during manufacture, product use and disposal. In order to achieve these objectives, some guidelines are provided and are as follows [17]: 1. Choose abundant, nontoxic, nonregulated materials if possible. If toxic materials are required for a manufacturing process, try to generate them on site rather than by having them made elsewhere and shipped. 2. If possible, choose natural materials rather than synthetic materials. 3. Design for minimum use of materials in products, in processes, and in service. 4. Try to get most of the needed materials through recycling streams rather than through raw materials extraction. Darnell et al[29] state that two essential future needs for successful ECM are 1) green techn010gy development to minimize waste in processes and 2) the development of analytical modeling tools that can assess the environ- mental consequences of different design and managing strategies. Sweatman and Simon [36] view green products as different from sustain- able products which depend on what kinds of products are made in what quantity. In other words, the degree of sustainability - also known as eco- efficiency — is measured in terms of biodegradability, DfE emphasizing re- newability, and consumption patterns. They made three categories of prod- ucts by the degree of sustainability as follows: 100% eco—eficiency : sustainable products, those that can be produced in large quantities indefinitely. high ecu—efficiency : products having environmentally-conscious features but which can be produced eithter in limited quantity or for a limited time low ecu—efficiency : products which deplete non-renewable resources, dam- ageshuman health, or pollute the environment. This thesis presents a new methodology and computer aided tool which can assist in decision making and green technology assessment to realize those goals of ECM. The thesis is organized as follows: 6 10. ll. 12. 13. . Literature survey of the previous work . Problem definition Problem statement System and modeling . Methodology . Mass—Energy Based(MEB) model description MEB network . MEB Language(MEBL) MEB network execution model Graphic user interface Case studies Survey Conclusion 2 Previous Work Computer tools are available to aid designers in analyzing the impacts of designs on the environment or providing guidelines of design strategies. They are either analysis tools based on Life Cycle Analysis(LCA) [32, 35, 4] or strategy and planning tools which are often linked with other Computer Aided Design(CAD) softwares[15] or handbooks. Generalized description of LCA is described in [30, 1, 9]. AT&T proposed a quick way to assess environmental impacts using an evaluation questionaire to be answered by people who are involved in the life cycle of a product and its alternative [18]. They are asked to specify the degree of environmental assessment using numbers between zero and four and they are asked to fill the product assessment of the 5x5 abridged matrix with rows describing five stages of a product life cycle stages as in Figure 2 and columns representing five categories of envirnonmental concern. Guided by checklists, DfE assessor assigns a number from 0(highest impact) to 4 (lowest impact) to each element of the matrix. Then the final 25 scoring elements are plotted on a target plot which is a polar form of a transformed bar graph to display environmental impact. The circumference of target plot is divided into 25 sections. The outermost circle represents the value 0 and innermost circle represents the value 4. Then the bull’s-eye represents a product of the lowest environmental impact. Although this method is easy to apply, and may become a step toward DfE, this does not provide objective scoring nor , Material 5 Material . Virgin I manufacture 3 processing I . Product material . i . assembly gxflcfign I . Component l Module I manufacture : assembly __________________________________ 12 Refurbish ; cu???“ :L Ship Package Figure 2: Activities in the five life-cycle stages of a product guidance regarding the relative importance of different issues. The abridged matrix proposed by AT&T[18] and ecoindicator[15] are shown in Figure 3 and Figure 6 as examples of qualitative and quantita- tive abridged life cycle assessment tools respectively. This method was also applied to facilities design and planning and supply line analysis[28]. Examples are shown in Figure 4 and 5 The example of Figure 6 is rather a nice example of data visualization than a system modeling tool. Sheldon[34] made an attempt to assess the environmental responsibility of an manufacturing process and propsoed Environmental Quotient(EQ) by EQ=AUXU where AU (atom utilization) is calculated by dividing the molecular weight of the desired product by that of the sum total of all substances produced and . Material Ene Solid Li uid Gaseous 1““ Stage choice uslegy residues resiilues residues Resource (l: 1) (19 2) extraction 4 l 3 3 2 Product I manufacturing \ Product I \ delivery Product I \ use Refurbishment [ \ Disposal \ III] a “1) ________ r1, 2) 0 ' """" ..__(l,3) .- ......... . ........ -:. 1' E 5:. 3(19 4) 2 """"" """ ' :" 3:... ‘- -------- -_ . ..... .......... o '''''' ..... ............. o o . ..... ..... TTTTTTT ------------- Figure 3: The environmentally responsible product assessment matrix and the target plot ' 10 $82 gassesam as e 888.85 8 888% as nausea é eases 883 38888 85:85 a so 508223 2F. .8 88m 8.8883 858a 885885 8383.685 <3 82:5 283 2.0m nous—6m use? cou==om 5‘ 8883680 ambam 8508M 3532 8:9:me 803.3on 0.585 558$ 8:338 8803 .8 038.5 823.8 ”88cm 8:82:82 95 8880 Sea 3089:? 880E so 893ng 888055 conscrocomxcouogmcoo 0535ch one seamen .2582 86 Figure 4: Facilities LCA matrix 11 $32 Joofieoacm 05 on moEoboo—m no 828983 53 ...n88&< 302 a ”0:3 3.55 05 E .3220 388:8?qu .momwgoonooh €83 ._owez.m.2v8_ou§m summon— §§§>om 8 on: manna “x332 88532 88889 spasm seam was? a 855 mass as. saves: a. 8.86 a. 038 Baogaou cinema 298580 Ram 888980 -maom basemoqsa e8 5609 .0055 Rheum .8.“ ammmon bnmnfiaqom no.“ summon 55323 new :33 £88.85 .8.“ summon excesses new omfion— 83:3 8: amusem— cogsoom 08:8 Seesaw a. 383 8888 SD 08835 8:33.35 BESS Figure 5: Matrix: Supply line to environmental design practices 12 . \\\\\\\ 3 a was \\ ~15 .11 WW '&§§§x\ \alxxm " u -{.\.\.\ \ \ \\ ‘ ,_ \' \RV\\ \ \\ wxx' s. . m \"\*-\\€§1f\\\v'\ \\\§ § . \ fix x \ “\sm \\\\x§\\1\\ - \ \ 0% §\_\ 0‘33 \ * ‘ ' \\\ ‘9‘ ~.\‘\\\' \_."\ .1“; -. MW \ \\ K 0‘1 ‘ QwR- 1 ~ x \\\<‘\9\:\\- ‘1 New“ \ \ \ \\ . \ . . t.\\ \X . 1. \. §"\'§~‘§$= \\ \ Rx. ‘§§‘K 1x \\\ 3: \1 \ \ \ \\\\\ \Qx \\ v ~\: >\\\ . \\1.\.\ V\?§\ \\ “i“ k \\ ‘ s . . . \‘23‘\ 3\ . . ~ . E» \‘m \\ \\\\w~. - “Wu ‘~ A \ \ \ .\ \ ~ \ . \\\‘\\ §§§k§$&%\i \\ “\w‘ \ '\: {<\ ‘ ‘ i ‘ A ‘ Impact ffect Damage Valuation Result CFC Ozone layer depl. Pb \ Heavy metals Fatilities Cd Cardiogenics PAH Du“ S smog Subjective Eco Mi ‘ Health damage -1 cator VOC Winter smog im . ent mem value DDT Pesticides ——-———-—‘ CO Greenhouse effect SO Ecosystem No m "H mm ,————— P Eutrophication Figure 6: Application of the Eco-Indicator as a tool 13 R aw Materials E missions-Air E missions-Water Co 76 0'02 0.09 Nitrogen 0.1 Cr 8.8 CO 0.27 Phosphorus 0.3 Fe 0.09 N0, 0.22 Mn 0.97 N20 7.0 Mo 1.5e3 SO; 0.10 Ni 24.3 CFC — 11 300 Pb 180 CH, 1.0 Pt ' 3.5e5 Rh 1.8e6 Sn 1.2e3 V 12 Table 2: UnitszELU/kg. Source: B.Steen and S.Ryding, The EPS Enviro- Accounting Method: An Application of Environmental Accounting Princi- ples for Evaluation and Valuation of Environmental Impact in Product De— sign, Stockholm:Swedish Environmental Research Institute(IVL),1992. U is an environmental index, a measure of toxicity[34]. Although Sheldon did not suggest how to assign the index, the Swedish Environmental Insti- tute(IVL) and Volvo Car Corporation have developed an analytic tool, the Environmental Priority Strategies(EPS) system. The index is represented in ‘Environmental Load Units’(ELUs) per kilogram(ELU / kg), per square meter(ELU/m2), per spots(ELU/spot), and etc. Those indices are calcu- lated by environmental scientists, ecologists, and materials specialists for every raw material[6]. Table 2 show for some examples of environmental indices. An example of the use of EPS system to compare the front end made of GMT composite and galvanized steel is shown in Table 3. The proposed ELU concept, an agreed set of environmental indices, has 14 .82 .H 378 2.3. rod 630:2me .05:ch SEE _S=mE:o.:>:m 0:.» 5.8m 535m .50 foBQESm 0 053553 no 0058083 «mm 3 @3583 SEE £58:o.:>:m 2: .80 :wfiofl 0:.» £203.85 05802960 3200.5 0:0 $30288 .5520 .80 38:8 <04 :4 - 839nm mam “EH. 68—ng 0.8 .0m_a:o3.< .zwmumd .mfivzmw ”moasom .mvcw E80 Efioaofis 8.0 Dam 0o cosine—so .35 .«o 39» H no women fl mm: 8:003 BE. ”0 wish. 00.3 00.0- 00.00 5.0 82.. 730.0 «0.0- and- 0.0 00.0- BEBE: 63m "memuaoufl 00.00 00.00 0.00V «0.0 ~8qu ”2.0 835$ 85 3o 3 So Asoaemécam 3.0 3.0 we $00.0 $335523, uoam “Show 05.0. 05.0- 0.0 00.0- 552605 vmmswx 3.0 3.0 0.0 8.0 means: 38m gm «ma od 80 Ease 38m "eomguaufim ~83 vofigiw 83me 05.00 050- 360 50.0 8% 020.0 05.0- 05.0- 50 00.0- BESS: .020 0.500908% 3+0 36m 0.00 00.0 Ebon— "25 83.8% 2.0 $0 0.v 00.0 05208 563588 92% :0- :0- 0.0 00.0- 830353 08st 00.0 mm.m 06 00.0 3.838 .020 ”cemuoamfim 36an8 .020 3a m: 5.3m 0.5 a: $3.5 3m we 33m 8:98:65 83803 DAN Each. $0.03 83 uosvem :oEoszunm um. £023.02 15 great advantage of objective scoring system, flexibility of modeling a system, and capability to compare different DfE strategies. Still, it provides neither how processes interact together nor feasibility of implementing a DfE strat- egy. Matthews and Lave[21] proposed another method which accounts for costs in a manufacturing setting and shows the optimal price for all cases as in Figure 7. In general the ECM tools range from simple to complex[32, 35, 4]. Fiskel [l4] classified these tools into qualitative and quantitative methods and discussed the advantages and disadvantages of these mothods. [21] proposed a gener- alized system model which accounts for costs in a manufacturing setting and shows the optimal price for all cases as in Figure 7. An expert system tool was discussed to determine improvements for easy assembly, disassembly, and material suggestions based on CAD software or input of a product specification[8]. 2.1 Qualitative methods Qualitative methods are further divided into two typas of methods: checklists and matrices. 1. Advantages 0 Easy to apply 0 Minimal data is required 0 Not as expensive as quantitave method. 16 Raw Reuse \Q Transfer Recycling Materials K Material F bn' t A bl sing a ca e ssem y Repair Remanufacture The Model C1 = M1, M1 is the initial cost of manufacturing Ci = (1-k)M1 + T, where k is the reusable product of a product and T is turnaround cost Disposal Use Man cost Turnaround Use disp End disp End toxic Cost(i) Life cost Price 1 40 2 12 3 12 4 12 5 12 10 12 20 12 End 0 0 0 2 3.6 2 3.6 2 3.6 2 3.6 2 3.6 2 3.6 2 0 MOOOCOO¢ 40 17.6 17.6 17.6 17.6 17.6 17.6 14 N000¢000 40 57.6 75.2 92.8 1 10.4 198.4 374.4 388.4 40 28.8 25.7 23.2 22.8 19.8 18.7 19.4 Assuming k = 0.7, M]: 40, T = 2, n = 20 Life Cycle Modeling of Manufacturing by H.S.Mathews and L.B.Lave(Camegie Mellon University) Figure 7: Life cycle modeling of manufacturing 17 Because of above characteristics, this method may be the first step in implementing DfE especially in identifying probable improvements. 2. Disadvantages o This method can show existence of performance improvement, but not how much improvement, even though using numerical scores. 0 This method provides no guidance regarding the relative impor- tance of different issues. For example, is it more important to reduce source volume or to assure recyclability? 0 People may fail to become sufficiently involved in DfE issues and may overlook important opportunities or problems that are not covered on the list. 2.2 Quantitative methods 1. Advantages 0 Can deve10p an inventory of the environmental burdens associated with a product and process by identifying and quantifying energy ' and materials used and wastes released to the environment. 0 Assess the impact of those energy and material uses and releases on the environment. 0 Evaulate and implement opportunities to effect environmental im- provements. 18 2. Disadvantages 0 Defining system boundaries for LCA is controversial. o LCA is data—intensive and expensive to conduct. 0 Inventory assessment alone is inadequate for meaningful compar- ison, yet impact assessment is fraught with scientific difficulties. 0 LCA does not account for other nonenvironmental aspects of prod- uct quality and cost. 0 LCA cannot capture the dynamics of changing markets and tech- nologies. e LCA results may be inappropriate for use in eco-labeling. In order to overcome these shortcomings, an ECM tool was developed based on PNT theory also developed at Michigan State University. 19 3 Problem Statement Due to the complexity of LCA, the subjective, vague, inconsistent guidelines inherent in these semi-quantitative approaches are likely to lead to ad hoc evaulation. Its primary weakness is that results are often subject to individ- ual interpretation[11]. And the major disadvantage of the above methods is that they provide very limited guidance for the improvement of the process. My thesis will provide a systematic way to deal with this problem without sacrificing the detail necessary for environmental and economic impact of the various strategies used in product design and manufacturing. Rirthermore, we have developed a new paradigm for the improvement and management of the process using the technology and the tools developed in this thesis as shown in Figure 8. More specifically, we will discuss a new quantitative tool which provides a systematic way to deal with this problem without sacrificing the detail necessary for evaluating environmental and economic impact of the various strategies used in product manufacture, use and disposal as follows: 1. It is comprehensive and thus can be used for the entire life cycle of the product. This is important because of conflicting requiremnets between different life stages. 2. It is isomorphic to the physical activities of the life cycle(material flows) that are responsible for pollution. Materials cause polution and the 20 tools should be able to provide this information as a function of man- agement strategies used. . The models are based on fundamental principle of material energy and balance. . It allows the user the capability to perform sensitivity analysis. This will help to evaluate the impact of less accurate data on the outcome. . It allows “what-if” simulation capability. . Helps to evaluate the impact of changes in processes and/ or technolo— gies(for example, the impact of automation or recycling). 21 ASSESSMENT l Economic OF ECONOMIC PERFORMANCE vael ALTERNATIVES (__ I CAL Economic mapping PERFORMANCE ‘ (Environmental response) — PHYSICAL PERFORMANCE Alternative structures I Ecological mapping Physical mapping {—J ['—'* Product Resources Enterpise ENVIRONMENT PROCESS NETWORK Level Env. Load Alternative technologies 7 Process 'lHIHNOLOGlES & PROCESSES Level Figure 8: Integrated approach to manufacturing system analysis and design 22 Part II Methodology 4 System and Modeling 4.1 System Definition A system is defined as an aggregation or assemblage of objects joined in some regular interaction or interdependence, simply put, a set of interacting objects called subsystems[5]. And the system is often affected by changes occurring outside the system[7]. Some system activities may also produce changes that do not react on the system. Such changes occurring outside the system are said to occur in the system environment [16]. The definitions of system in Webster Dictionary are: e structural design 0 a usually miniature representation of something 0 a pattern of something to be made 0 an example for imitation or emulation a description or analogy used to help visualize something that cannot be directly observed 0 a system of postulates, data, and inferences presented as a mathemat- ical description of an entity or state of affairs. 23 Often, physical systems under studies are likely to be too large, too broad, or too complex to characterize as a whole. Many theoretical suggestions about how to partition such a system have been suggested in [20]. In order to circumvent such problems to get a satisfactory solution which might not be the best or exact solution, the large system needs to be broken down to a number of subsystem small enough to be tractable problems and then reduce the number of objective measurement parameters to a smaller number of parameters relevant to the study objectives. Then the system can be represented as interconnection between each block which is an aggregation of entities. A model, whether it is physical or mathematical, is used to study a system as a substitute and in most case as a simplification of the system. Based on assumptions on the physical system, this process of selecting pa- rameters through system data gathering and data analysis chooses the system boundary and identifies its entities along with their relationships together. 4.2 Model Simulation refers to a broad collection of methods and applications to mimic the behavior of real systems[25]. Simulation Meriam Webster dictionary defines simulation as: o The initiative representation of the functioning of one system or process by means of the functioning another 24 0 Examination of a problem often not subject to direct experimentation by means of simulating. With the exception of design by creation without any help of previous knowledge of similar problems before, most designs can be done from previous designs by modification or selection [3]. Measurability study in the modeling phase is one of the key issues in studying a system, whether the primary purpose of the study is looking for a new creation of a system, or enhancement of an exisiting system, or even controlling an existing real system. Even though most small store managers do not use computer simulation, they are making every effort to maximize their profits by constructing their own store models and by making “what-if” analysis in their minds. Depending on the size of a system and the goal of extent of fine detail, an appropriate scale of modeling is required to meet the goal of a system study. While simulation can be a replica of a real physical system, or mathematical model, computer simulation refers to methods for studying a wide variety of models of real world systems by numerical evaluation using software designed to initiate the sytem’s operations or characteristics, often over time. It quantifies assumptions represented by conceptual maps and explores their impact on various “what-if” situations This is very valuable to managers who want to test new assumptions in new ways. One of the advantages of simulation is that it enables us to gain valuable insight into how their assumptions interact with each other. This insight offers an unprecedented 25 competitive advantage by improving decision-making and problem-solving skills. It is especially effective when used as a scenario planning tool. The ben- efits of simulation are: e Risk—free strategy experimentation e Enables managers to explain their ideas more easily and insights to other pe0ple inside and outside the organization. 0 “what-if” type questions in comparison to other alternatives Modeling is a formal representation of a system followed by simulation which assigns semantic meanings for its formal representation. 5 Methodology Top—down design methodology traditionally has been used to cope with de- sign complexity[22]. Here, both top—down design which is obtained by goal oriented approach and bottom-up design which is used to estimate the costs of products are used together. In order to represent a system, MEB DfE tool have both textual descrip- tion and graphical description. The analytical description of each process in MEB DfE tool is based on both PNT and MEB economic model [38, 39, 37]. The textual description of a sytem, MEBL in section 8, has been designed for brief mathematical modeling of a process especially in dealing with matrix computation. 26 The MEB graph model is also used for better description of interprocess communication and interactions among processes. This modeling method hides the cryptic nature of textual description and provides a global concep- tual map of a very large system. Modeling a system begins with identifying every process and its output products, input materials, and byproducts(waste) of each process. Then with all the measurements available after construction of an MEB modeling and simulation, the next question is what to do with all those evidences. Any reasoning, validating, scientific judgement is based on those evidences which may lead to modification of a model, or different judgements. To make a judgement, possible decision categories need to be defined first. Then the decision problem would be assigning measurements to each of the categories. The next question is how each category is judged compared to other cate- gories to quantify a global environmental burden. It seems to be next to impossible to find an unified formula to lead to an unique decision agreed upon by all the communities. Still, it would be nice having such a formula pleasing all the communities. Here, the environmental impact is computed for each byproduct first and then summed up as an eco-indicator value of a whole system. The envi- ronmental impact metrics used in this tool came from the EPS Enviro- Accounting method by Swedish Environmental Researach Institute(IVL). The MEBL description of this metrics is in Section 8.6. To access such an interoperable database provided by environmental communities, the textual 27 description — MEBL - also understand SQL—like syntax. 28 6 What is MEB model ? Nowadays, the importance of impacts on the environment by manufactur- ing processes motivates to evaluate environmental burdens associated with a product. The Mass-Energy Based Modeling System(MEBMS) is attempting to realize the evaluation of environmental burdens. This paper describes a tool based on the Mass-Energy based economic model[38, 39, 37]. The manufacturing environment consists of many processes and proce- dures applying to materials, and disposals associated with its energy, and cost. Historically, the effectiveness of manufacturing has been evaluated by monetary accounting system. Trends in manufacturing towards decentral- ization and outsourcing of business requirements need an effective modeling tool to coordinate the business activities. While the majority of financial accounting systems is powerful, this ap- proach alone does not show environmental factors, technical factors, energy cost, and monetary factors easily due to the complexity of the interconnection between processes or between processes and environment. As more information is flourishing from various disciplines and processes, the difficulties of system modeling increase in terms of creating a model, evaluation of the model, maintaining the model, and proficiencies in pro- gramming language skill. This section prOposes Mass-Energy Based Modeling System Tool(MEBMST) 29 based on the the Mass—Energy Based Economic Models[38, 39, 37]. MEBMST was developed to evaluate environmentally conscious product designs, man- agement of manufacturing facilities to evaluate the strategies for reducing waste flows into the environment, and life cycle assesment. Along with motivation of modeling environmental problems, the obser- vation of similarities between physical laws of preserving material and en- ergy and economic characteristics of physical production process, and the classic economic input-ouput analysis lead to proposing the Mass-Energy Based(MEB) economic model in [38, 39, 37]. The difference between the classic input-ouput analysis is how labor is formulated. In the MEB model, labor is formulated as an energy cost rather than as a flow of services as in classical input—output analysis. The MEB model views a production process as a sequence of transformations on the state of materials by energy. This enables one to break a large system into tractable smaller systems. Further, MEB model divides output as useful product and by-product(waste). 6.1 Overview The primary goals of MEBMST are as follows: 0 To build a mathemtical model for processes which are used as building blocks of a plant. 0 To implement a Graphical User Interface(GUI) which hides all the de- tails of programming languages and visualizes the presentation of anal- 30 ysis results such as monetary factors, environmental factors given quan- tities of final products, and the unit cost of final products in addition to a report of the results. A manufacturing plant is modularized with building blocks of several classes according to process flow until the desired detail description is reached. In order to contain a whole plant in a limited property of screen resource, certain blocks are described as a library which has a full description at some- where else. The MEBMST provides six basic kinds of process building blocks as fol- lows: 0 Production process 0 Recycling process 0 Storage process 0 Junction process 0 Goal process 0 Wire object 0 Library Process 6.1.1 Production Process For example, in the model in Figure 9, the y,- represents the flow rates of materials, and :r; represents energy cost per each unit of material flow rate 31 O—w—i 5 3 C} > Figure 9: Abstraction of material transformation where i = 1, 2, - - ~ , 5. And assumptions are made that 115 is the useful product and y, is the by-product. Then the product 315$; becomes the energy flow rate. ill 161 312 k2 def = = K . 1 313 ’63 315 3’5 ( ) 314 ’94 where the column vector K = [k1k2k3k4]T is called “technological coefficients of productions” following Leontief. The law of conservation of mass requires that y1+y2+y3-y4—y5=0. And applying the law of conservation of energy, output energy + input energy + processing energy = 0. 32 01' 4 275315 = - Zxfllj - f(y5)y5. (2) i=1 where f (315) is the processing energy per unit of output 315. Substituting Equation 1 for Equation 2, we have, for y5 7t 0, the cost equation 4 $5 = - Z kjicj — f (115) (3) j=l or 1‘5 = "KTX " f (315) (4) in vector form. The Mass—Energy Based(MEB) Simulation tool is developed to evaluate environmentally conscious product designs, management of manufacturing facilities to evaluate the strategies for reducing waste flows into the environ- ment, and life cycle analysis. 6.1.2 Recycling Process Given the two choices of whether to produce the exact required input mate- rials by recycling part of byproducts with possible leftovers and whether to recycle all amounts of byproducts and to postpone compensation of required input materials after recycling process, the latter is the philosophy behind the recycle class. Depending on the lack or excess of recycled products, the 33 difference of amount between the required input materials and recycled prod- ucts may be brought from outside of the system or took out of the system with associated cost. Considering that the goal of recycling process is to recycle all byproducts, the execution sequence of computing flow rates becomes Opposite of the pro- duction class. Given the flow rates and the unit costs of the byproduct or a production class, the flow rates of the recycled products and its associated unit costs are described by [:21th and the cost of by-products and intermediate recycled products are de- scribed by [ jg; ] = K - X. — F.(y.) (6) 6.1.3 Storage Process The Storage and Recycle classes are somewhat different from other classes, while still being closely related to each other. If the flow rate of a reprocessed end product to be recycled into a pro- duction line matches the exact flow rate requirement of a production process, then the system would form a perfect closed cycle system. But in reality, what if the flow rate of reprocessed materials does not match the flow rate required by a Production instance ? Or is it possible to design a plant which 34 exactly matches the amounts of needed materials ? Depending on the suffi- ciency or insufficiency of the recycled end product, the same kind of material - possibly with different unit prices - may need to be imported from outside a system. To deal with such inconsistency, the Storage class comes to the rescue between a Recycling instance and a Production instance and behaves as a buffer between demand and service. Given the recycled material flow Y,, the required input material flow by a production class Yo, the quantity of out-sourcing or surplus recycled material Yb is determined by n=Yo—)/i (7) Similarly, given the recycled material unit cost Xi, and out-sourcing ma- terial unit cost X5, the unit cost of input materials required by a production class is determined as follows: XO=C¥°X§+B'X5. (8) where a=Yo/Y.- [3:1—01 35 6.1.4 Junction Process This process is used to deliver an intermediate useful product to the next several production processes. Let YJD be the flow rate of intermediate products, Y}, the flow rate of the next m production processes where o E [1, - . - , m], with X, and Xp, the unit energy cost for Y, and Y,, respectively. Two constraints which are met by the junction process are the continuity constraint and the compatibility constraint X0 = X,. Figure 8.1 illustrates how junction class is used. 6.1.5 Goal Process Before performing MEB simulation, goals need to be defined. For example, “what-if” analysis of different flow rates of final products are specified as follows: Y0 = constant vector. 6.1.6 Wire Object This object is used to interconnect processes within a system carrying mea- sures such as flow rate, unit engergy cost, name, and etc. 36 6.1.7 Library Process Aggregation of processes is necessary in order to overcome the following prob- lems: e The screen size may be too small to describe a complex system. 0 As processes clog together, readibility of processes deteriorates. 0 Repeated modeling of frequent use of a system can be tedious, time consuming, and prone to errors. A system is defined in terms of above processes. By defining the Library Process as a system recursively, system can be described hierarchically and structurally. The library class process is rather a simulation directive which manages what processes are to be simulated next. The larger a system be comes so is the degree of cluttering in a limited screen space. The library class is introduced as a building block to model a system which is too large to accomodate in the limited space of a drawing screen. By its capability of encapsulation of complex processes with a simple rep- resentative object and its capability of expansion of a Library class instance into the full blown description of the system, not only does this class enhance readability but also it allows a user to choose the degree of detail description of the system. As a library instance can have other library instances, a complicated sys- tem can be organized in a tree structure allowing a user the freedom of 37 \ \ \ \ \ Figure 10: Multi layered tree structured description of a system viewing any detailed level of a system as in Figure 10. And the reusability of a proven library class helps modeling with confidence and saves time. A library class is created by adding its representative encapsulation shape to the existing system model. In normal classes other than the library class, the property of incident wire is determined by the neighboring process con- text. However, that is not the case for the library class. The communication of one level of a system with the next level of a sys- tem occurs through the specific wiring of the next level system. Thus the encapsulation procedure involves defining the next level system boundary 38 and defining the location of points incident upon a library class. This cor- reSpondence in the interconnection of the library class and another class is determined by the incidence points alongside the enscapsulation shape. A li- brary class process implicitly has a fixed number of incident points associated with library a labeled internal wire, besides the representative encapsulation shape. By doing so, it is possible to to map the external wire into a certain internal wire in the next level system. The graphical user inteface(GUI) for this process is described in Appendix A. As an illustration, the system in Figure 8.1 can be simplified by creating a new library class j j in Figure 11. The Production Process block, which describes a transformation process is based on the Mass Energy Based Economic Models[37]. The first part of the model, transformation process, has been explained in Section 6.1.1. The production material flow equation Eq. 1 and the energy equation Eq. 3 in Sectionseczwmeb can be partitioned and generalized as in Eq. 9 and Eq. 10 to view a system as a component at the next level as in Figure 11. [::}=[1’§::§::][::ng[::] <9) where y; refers to the required input material flow supplied to the processes within a system boundary, ya to intermediate products produced inside a sys- tem boundary, and y, to the material flow response variables outside a system boundary which include both input supplies and production of byproduct or waste. 39 :32 axe: 2: as aswcoafioo s we 839$ H: wSmE 39.59..an 30.6293 ~330an 39.83 9:839... mcuuooocaok ucozcongcm 4v em Aw mm Av wm ucgcoham « .oz "1--.? no: on .l-.. A, a a B > «S 91:3] We: 33 nip 3953 _.|AIV IAIX all lb _ m: I on v.3 umaz N .02 "luml co: 0... .i-.. v a w: 30 can: “wwm 303 can» AW J €0.53 nguco flags” acumen?“ mm 33: an 38% nouns magnum confinEwm 83 .33 no.5“: #333 #885500 «90?... aaoo new}: 2.: 23.3: .6: 00%;: no: soaks don saunas as: oucovqucH queuing—H axe.— odoauu read-am con-36.. 2.31.3... couvocsfi 3.52.30 UHF—cum to...— Hocd... 3.0300: sedan-afloat 0.3.3.3.. done.— ucossou anon—lama anon 2.00 .3000 con.— zoq> 034.6 0365 yea 03.2, you H25 _._._m 1: econ—ox 30.50: p.75 :ul—z 90.: cat—Ix: audio: “End..— 0.3.“— 40 The K, characterizing the transformation, is called the technological co- efficients of production. This goal-driven architecture allows one to answer the question such as “given a final product flow, what the material flow of intermediate products and byproducts would be?” And the second part of the model, generalized energy equation, is given by: Xb _ Kl: K3; X; Fb(yb) [Ll-[Kg ,., [Ll-[W <10) where X,- is the cost related with y,- for i E {1, r, b, a}. With this equation, every material flow rate is computed to meet a final goal by back-prOpagation. Once every flow rate is known, the unit cost of the final product is computed in reverse order based on the unit cost of out-sourcing material unit costs. As an example, consider a manufacturing enterprise in Figure 12. Then those material flow equation and the generalized energy cost equa- tion would be 3131 31r1 3141 11r2 311: :3: , yr = Z: 3153 yr5 (11) . 3154 . _ yrs .1 F . 3103 ' 31b = .1104 , yo = lg: _ 3,05 _ . l 41 t . 4 2 . : r6 . 05 n O—*:— Prtproccssin; *0 A : 5 l“ I 5 44 : 5 3 : Ol Produd 3 —.—-()—>—-l 1 : 6‘; r5 la 1 r1 Envrronmcm Wbypmm Mgbyprodnct Anembly byproduct Figure 12: Illustrative process network diagram for a manufacturing enter- prise By the continuity constraints imposed by the junction process, 3105 - 3154 — 3153 = 0 3103‘3131—3132=0 31o4-3l4z-3141=0 1 0 1 0 0 0 (12) =>yb= 0 l 0 l 0 O y; 0 0 1 1 With the substitution of (12) for yb in (9), the requirement of outside material and its associated cost is described in terms of the final product: 315 (I - AKlb)—1AKlo llfi Kbyo yr = (Krb(I "' AKrb)_lAKlo + Kro)yo 42 (12f (Krbe+Kro)ya dg Kayo X5 = —(I — AK$)-1K,.,X, — (I — AKg)-1Fb(yb) The system model given in Eq. 13 quantitatively establishes the relation- ship between the flow rates of the raw materials entering the system boundary and the products leaving the boundary along with the wastes released into the environment of the system. Furthermore, the model represents these flow rates as an explicit function of the process technologies incorporated within the system boundary. This provides us with the powerful method of handling complex system without losing logical or physical consistency [37]. Consider the extended system model as in Figure 13. _ 3167 J [ 31r7 J 31;: = 3178 , 31h = 3m F 3182 31r9 yIG (13) 3m. on = 3117., , 21;.» = [ 3’03 ] 3118 _ 3108 l The technical coefficients of the storage class connecting i-th production class and the j-th recycling class, is determined by subtracting j-th row from i-th row of K. [214%] 04> 43 Reprocessing byproduct Processing byproduct Assembly byproduct Systemboundary r5 r4 ll r2 7' ---..-..-_--___....__.-_--__-_-______._..-__---._.._-_-----___-_-.._ ____--__-__' I 04 4‘2 I I ‘ I I 5-4 T I 02 PIOdUCI I = Processing Assembly —’I'—O No 2 I 3-2 I : 4 2 g I I . > 05 v I I Pre ' A I I 8—5 processmg —'—(3 ll I I T ' 5 I E M E I 53 : 01 Product : > Processing 03 34 Assembly F—‘t—O No 1 g 3 -—-O->— 1 g I H n : Praccssingbyptoduct 3.7 1'6 : : ‘3 Assemblybyprodm ‘1 : I I I r7_. 16 I I I I : """""" : """"""""""""""""""""""""" : I 03 ' 08' I - 67 I ' Product I 18 7-8 Recycling ”J, Recyclmg : I l ——-I WW i—‘O—e .-——D0—-( ) : . S SS . ProoessZ Process! 5 : N03 . . ___________ E: -.........7. ........................ £5 .2 . I _______________________________________________________________________ I R I byproducts mm: Environment Figure 13: Extended process network diagram for a manufacturing enterprise and the cost of byproducts, intermediate, or final products from the re- cycling processes are described by l§iil=KCI§$I:]-[%IZIZI - 05> Then overall extended system model could be constructed by augmenting the original system model with the extended system model as follows: Y1 K n Klo * Yo K = Krb Kro * Yo ( 1 6) be * Kfu Kiel Yfl on * K ’0; K,“ Yfr and the cost of byproducts, intermediate, or final products from the re- cycling processes are described by X. K3,; K3, * XI F5018) X0 _ K1: K3; * Xr _ Fo(yo) (17) XI» _ * K018: K011» XII Ffl(31n) Xflo * KCfd KCfor Xfr Ff,(yf,) 45 7 MEB NETWORK The proposed Mass-Energy Based(MEB) System Modeling tool is composed of several main components: Process and Network Representation, Environ- mental Load Unit Database, Goal Definition and System Environment Setup, MEB Language(MEBL) Execution Unit, and Data Visualization Unit. There are two extreme cases of how simulation can be accomplished. One approach is to design a simulation with a textual simulation language which can handle what general purpose language does. After all, it is the machine codified behavioral descriptions which make simulation possible by computer. And the capability of a simulation in fine detail is only limited by the capability of simulation language. However, the disadvantage becomes obvious when a system grows larger. The larger the system grows, so does the size of the codified simulation pro- gram. Even the author of a simulation design is likely to become confused about what are the boundaries of subprocesses and how they interact to- gether, as time goes by. Adding to that, the learning of a new simulation language may take long time. And the cost of initial learning, retaining that learning of the simulation language can be high. At the other side of the first apporach stands the graphical representation . of a system. Describing a system by only graphical objects greatly enhances the readibility of a system, helps the reader to grasp overview of the whole system, and to understand interactions between subprocesses more easily. 46 In reality, both methods should be mixed apprOpriately and manageably in performing simulation, as does the design process. Despite the ease and other advantages of graphic modeling of a system, the graphic objects are not as flexible as simulation language itself. And there is some system behavior which can be described only by simulation language itself. By embedding the conversion from visual semanties to simulation lan- guage into the MEB modeling, the learning time of using a new simulation tool is greatly reduced. I would like to recommend that the first approach of textual description to be confined in a process and the second approach of graphical description be used in defining system boundaries and interactions to accomplish a simulation. Rom this discussion, I would like to assert that graphical modeling cannot replace simulation language itself, even though the reverse is true. The proposed Mass-Energy Based System Modeling tool, combining both approaches, begins with dividing a large system into managegeable subpro- cesses with MEB building blocks and wires as shown in Figure 14. This tool allows one to input the description of the main structure of a plant using a drawing pallet available to the user. This pallet contains built-in drawing buttons in a graphic user interface(GUI) implemented on the X-Window environment. The GUI relieves the user from knowing all the mathematical details of the models which describe each process within a plant and the interconnection constraints associated with the structure of 47 the plant or process. Besides having features which represent network information succinctly, MEBS also introduces the Mass-Energy Based Simulation Language(MEBL) which borrows many aspects from C language, MATLAB‘, and SQL database language. A source program is automatically created by the user with the GUI. I will describe some of the details of the program with an example. 7 .1 MEB Graph Representation The process network is a mirrored acyclic data flow graph even though pro- cesses contains feedback 100ps. A network is described with several types of building blocks such as production, junction, library, goal blocks, and wires which connect the blocks together. The forward connections are done by all the types of blocks except the recycle type block while backward feedback connection uses only the recycle type block as a subprocess. The special storage type block is used when the backward connection feeds to a forward connected process block to form a feedback connection. The various values of intermediate products or byproducts are propa- gated through wires. The entities of wire, which might be a final product, byproduct, or intermediate product, propagate through wire communicating bidirectionally. And these wires determine preset nodes which current node is dependent on and a postset nodes which are to be exected next. lMATLAB is a trademark of Math Works Inc. 48 r w - Production class "li— - Recycling class - Storage class - Junction class - Library class \ - Wire class Plant network descri tion _ with Graphic User nterface System modelmg (Graphic object databasej Netlist generation I l I I I @rocess dependency neg Process descriptioj Modifying in MEB language Process attributes —+ [Graphical load unit I fin definition j 9 database I Land environment set u Semantic analysis Simlation results ] S‘mulam“ Data visualization L Figure 14: MEB simulation architecture 49 Considering that wire attributes might be changed only through execution of process modeling, this process execution becomes a transition to change a state of a process into the next state and to move on to other dependent processes to do the same. Definition 1 A process network, G, is a seven-tuple graph G = (V, T, W', A, 6, 1', fl); V = {v1, v2, . . . , uk} where k = IVI, is afinite set ofprocesses shaped as rect- angulars, and the attributes of processes are extended by five difl'erent classes of {PRODUCTION, JUNCTION, RECYCLE, STORAGE, and LIBRARY}; T = {t1,t2, . . . ,tm} where m = |T|, is a finite set of terminal nodes shaped as small circles, and the attribute of T are extended by two different types of{GOAL, SIGNAL}; W = {w1,w2,. ..,wa} where a = IWI, is afinite set of ordered pair of different nodes such that w.- = (vuvd), s 915 d where w.- is a wire from a node v, to a node 1).; represented by an arrow; A and 6 is a top-down and bottom-up hook-up function mapping from W to W; T C V and 13 C V are initial set of nodes at which top-down or bottom-up execution sequences are to be originated. 7 .2 Inputs and Outputs Wire,W, which connect processes together into a system is extended by its attributes as follows: W = (Label, Capa, Cost) Label Entity name; set of alphabets 50 Capa Flow rate of entities Cost Unit cost of entities Since A and 6 are function of W, the distinction of inputs and outputs are necessary. Given a process with neighbor wires on it, the wires are fall into one of a catetory, input or output. The decision of input and output of a wire, w,-, is based on PNT mathematical model mentioned in Section 6. In PNT theory, wire has a context sensitive meaning in terms of their direction, types of its associated processes, and the direction of execution sequence i.e. top-down or bottom-up execution environment. Those resulting constraints to form a sensitive meaning of wire is described in Table 4 and Table 5 along with the constraint such that the number of inbound wires of a JUNCTION, STORAGE, and GOODS should be one. This context sensitive meaning of a wire is used in several ways as follows: 1. To determine whether connection of a certain nature should be allowed. 2. To construct a set of functions through wire examination whether it is a stimulus or a response variable to form a function. 3. To determine the next execution sequence. The pair of values in Tables 4 and 5 are used to construct the tOp-down function and the bottom-up function respectively. The meaning of the num- bers are as follows: 51 SIGNAL JUNC. PROD. RECYCLE GOODS STORAGE LIBRARY SIGNAL (-1.-1) (3.3) (3.3) (3.3) (3.3) (33) (-l.-l) JUNCFION ('lr'l) (1:0) (110) ('lr'l) (1:0) (1:0) ('11'1» PRODUCTION (0.2) (1,0) (1.0) (0.2) (1.0) (0,2) (-1.-1) RECYCLE (0:0) ('lr'l) (’lr'l) (090) ('lr'l) (0:0) ('lr‘l) GOODS ('lr'll ('lr'l) (-l,0) (‘lr’l) ('lr’l) ('lr‘l) ('lr'l) STORAGE ('lr'l) (120) (1:0) (’lr'l) (190) ('lr'l) ('lr'l) LIBRARY (0,3) (0,3) (0.3) (0.3) (0.3) (0.3) (1.0) Table 4: Inbound wire;N,.o.,, x chumn constraints in tabular form SIGNAL JUNC. PROD. RECYCLE GOODS STORAGE LIBRARY SIGNAL {-1,-1) (0.1) (0.1) (1.1) (-1.-l) (-l.-l) (0.1) JUNCPION (0:1) (071) (0)1) (’lr'l) (323) (0’1) (091)) PRODUCTION (0.1) (0.1) (0.1) (1.1) (3.3) (0.1) (1.0)) RECYCLE (0:1) (‘lr'l) (1’1) (1’1) (’lr’l) ('lr'l) (lrll) GOODS (393) (3:3) (3:3) (‘lr’1) ('lr‘l) (313) ('19'1» STORAGE (0:1) ('lr'l) (0:1) (1:1) ('lr'l) (’lr'l) (0,1)) LIBRARY (0:1) ('lr'l) ('lr'l) (1:1) (’lr'l) (0,1) (0,1)) Table 5: Outbound wire; wa x Nwzumn constraints in tabular form 0 The wire is a dependent variable. 1 The wire is an independent variable. 2 The wire is an independent variable, but the cost of the wire needs to be initialized first in a process as part of the initialization function. 3 Connections are allowed, but the process have an empty function. 52 8 MEB Language(MEBL) The graph model is a very good start to model a big system without delving into a mire of language quirks, and is quite a useful conceptual global map to show the extent of detail interactions between different processes and the data flow between processes. Still, the mathematical modeling of each process is often based on a tex- tual description for its brevity, simplicity, and flexibility. After all, the com- puter only understands numbers and alphabets at its core. Thus, every aspect of system modeling should be translated into a textual description describing how each process should evaluate and how each process should interact with one another. However, a computer code is not a good tool for designing a program, nor is it a good communication language for people[l3]. In this context, our language MEBL is to be described in two parts, i.e. evaluation of processes and its execution environment. The MEBL expression for evaulation uses a similar model of C language, MATLAB, and SQL. In the next section how a system can be described in textual MEBL will be presented. The exact syntax and grammar of MEBL are presented in Appendix H. 8.1 System Structure The system structure described in MEBL consists of two types of system descriptions; they are declarations and the series of process description blocks 53 some of which might be directly related to wire variables and some of which are related to execution environment for initialization or set-up for simulation or post—processing of simulation results. The keywords to differentiate such types are as follows: wire Define a set of entities conveyed by wire netlist Build MEB graph and translate MEB graph into MEBL init Assign name entity of wire instances with label post Calculate overall environmental impact junction Specify junction process based on MEB theory signal Specify terminal node process block Specify process description The first declaration is about defining entities of wire which become per- fomance measures based on MEB theory described as wire { capa; cost; name; } A wire class is a composite object consisting of flow rate capa, energy cost cost, and a label name of wire itself. A wire instance name followed by a period and one of three entity names is the way of referring to the individual wire data object. 8.2 N etlist Again considering the above example, the netlist body consists of the list of the wires with its source node to the left and its destination node to the right. For example, 54 $3803 5953 516:3 8:3 889$ < "ma Sana AwMH sovv 3.5.5.1. dean; 34w _ acescou _ oncolaaa a: unonou ll :38: sec vUOHJG mace oDOflao :6: xouLo> >0: 000530 and xoauo> Hon 85:2: 18:2: £8» 39.3 Elem cadnucafi cqaaaflu OUHLOum uOLm AOCHu Oduauou :0dfl0316ht OLd: 38 use“. 28“. :qu so; 2:8 31> com 31> one 8% 18 losvou Bucu culqz OOHL Addlxt undauoz afloat Oddu 55 netlist { H1: 81 -> 01; H2: 01 -> J1; H3: J1 -> U3; H4: U2 -> J2; H5: U3 -> J3; H6: J3 -> U4; H7: J2 -> US; H8: U4 -> 06; H9: US —> UY; H10: 01 -> 82; H11: U3 -> 33; H12: 05 -> S4; H13: 02 -> 85; H14: U4 -> 86; H15: J1 -> U2; H16: 32 -> U4; H17: 33 -> 05; Figure 16: Automatically generated netlist in MEBL H1: 81 -> J1; implies that the wire H1 goes from the signal node 81 to junction node J1. Note that -> has a different meaning from the meaning of the C language. The second declaration consists of lists defining all the MEB graph in- formation of how processes are connected together. This netlist will partly determine the sequence of process execution in simulation. For example, in a system of Figure 8.1, the netlist delclaration will look like Block type Subclass No. Production 0 Recycle l Goods 2 Storage 3 Libray 4 Table 6: Block node subclass number used in MEBL The remaining part of MEBL describes either the process itself or initial- ization before simulation begins and post-processing after the completion of simulation. A block node represents a subprocess and contains apprOpriate state— ments in its body. Accordingly, there are two block nodes corresponding to two subprocesses in this example. init node is a special kind of block node which is done first before execution of functions associated with each block. The subclass number of the block used in MEBL statement as in “block subclassmumber blockmame { - - ~ }” is shown in Table 6. Table 6 shows five kinds of process types used to build a system. The automatically generated example of an MEBL system description corresponding to Figure 8.1 would be translated as: junction J1 { shape J1; out H15 ; in H2 ; out H3 ; k=[1.1;]; if (backward) { [H2.capa; ] = k * [H15.capa; H3.capa; ]; 57 } else { [ H15.cost; H3.cost; ] = k’ * [ H2.cost; ]; } } junction J2 { shape J2; out H16; in H4; out H7; 1: = [1, 1; 1; if (backward) { [ H4.capa; ] = k * [ H16.capa; H7.capa; J; } else { [ H16.cost; H7.cost; ] = k’ * [ H4.cost; ]; } } junction J3 { shape J3; out H17; in H5; out H6; 1: = [1. 1: 1; if (backward) { [H5.capa; ] = k * [H17.capa; H6.capa; J; } else { [ H17.cost; H6.cost; J = k’ * [ H5.cost; J; } } signal 81 { shape none; out H1; } signal 82 { shape none; in H10; } 58 signal S3 { shape none; in H11; } signal 84 { shape none; in H12; } signal 85 { shape none; in H13; } signal 86 { shape none; in H14; } block 0 UI { shape U1; in H1; out H10; out H2; H10.cost = zeros(size(H2.capa)); k = [ 1; 1; 1; if (backward) { [ H1.capa; H10.capa; ] = k t [ H2.capa; ]; } else { [ H2.cost; ] = k’ t [ H1.cost; H10.cost; J; } block 0 U2 { shape U2; out H13; in H15; out H4; 59 H13.cost = zeros(size(H4.capa)); k = [ 1; 1; 1; if (backward) { [H13.capa; H15.capa; ] = k * [H4.capa; 1; } else { [ H4.cost; J = k’ * [ H13.cost; H15.cost; J; } } block 0 US { shape U3; out H11; in H3; out H5; H11.cost = zeros(size(H5.capa)); k = [ 1; 1; 1; if (backward) { [ H11.capa; H3.capa; ] = k * [ H5.capa; J; } else { [ H5.cost; ] = k’ t [ H11.cost; H3.cost; J; } } block 0 U4 { shape U4; out H14; in H16; in H6; out H8; H14.cost = zeros(size(H8.capa)); k = [ 1; 1; 1; 1; if (backward) { [H14.capa; H16.capa; H6.capa; J = k * [H8.capa; J; } else { [ H8.cost; ] = k’ * [ H14.cost; H16.cost; H6.cost; ]; } } 60 block 0 US { shape US ; out H12; in H17 ; in H7 ; out H9 ; H12.cost = zeros(size(H9.capa)); k=[1; 1:1;1; if (backward) { [H12.capa; H17.capa; H7.capa; ] = k * [H9.capa; J; } else { [ H9.cost; ] = k’ t [ H12.cost; H17.cost; H7.cost; ]; } } block 2 06 { shape U6; in H8 ; } block 2 U7 { shape U7; in H9; } The junction node abides by special constraints. In the framework of the MEB theory, the sum of output flow rates are the same as the input flow rate and the unit costs are the same for all wires. A signal node is created automatically either at the beginning or at the end of a wire which is not connected to any other process. Normally, there is nothing to compute in the signal body except the initial cost provided in the phase of environmental setup to initiate simulation. 61 Finally, the special post body which may or may not be executed is available to evaluate environmental impact along with environmental index database as in Section 8.6 after all other processes have completed their simulation. General building block classes as in Table 6 are specified by different shape of geometrics in drawing. For the sake of readability, the declaration part associates a specific block class with a graphic object in a drawing and indicates which 'wires are coming in and going out of a process. Then, a specific MEB evaulation statement of a block class follows the declaration part. It is worth mentioning that any wire variables can be accessed in any nodes. However, considering that the messages coming in and going out of a process are highly correlated with the network of processes, it seems to be a good practice to access only the wire variables which are in contact with the node. With this limitation of choice in the practice, it helps the program to be modular and structured. 8.3 Variables There are two execution environmental reserved read-only global variables such as backward and forward. Simulation goes through the phases of back- ward and forward computation; in other words, top—down or bottom—up computation. Those special variables shows the direction of traverse during simulation 62 so that each process can define its own segment of a program in MEBL inside the same block depending on the state of traversal. Users can read those values but users are not allowed to set the values of those special variables. 8.4 Constants To accommodate the frequently used symbol 1r in trigonometry function, a specific symbol PI is reserved for a constant. And the character constant is a character between single quoatation marks while the string constant like "this is string" is a list of characters within double quotation marks. Both the character and the string constants are accepted as in C language. Some invisible characters are represented by escaping as follows: \n newline \t tab \f form feed \\ back slash For the numerical representation, both decimal and hexa-representation are accepted for an integer value. For example, the decimal number 10 is equal to the hexa-number Oxa. For the floating number representation, only decimal numbers are allowed, as in the following examples: .1234 1.234 12.34E5 123.48—5 8.5 Control Flow The control flow is similar to that of the C language except that the expres- sion body between control flow keywords to be selected or iterated should 63 be enclosed by { and } even though the expression body contains only one statement. if else These non—iterative conditional selective control keywords associate exclusive statements to be executed with dynamic expressions while simulation is going on. As a result, this control flow selects a specific part of the body separated by if and else keywords for computation. Example) if (expr1) { statements; } else if (expr2) { statements; } else (expr3) { statements; } Every expression body between if and else should be within { and }, even though the body has only one line statement. for This iterative conditional control key word is a very concise iterative statement especially if initial statement before the iterative body or/ and the post statement after the iterative body is/ are necessary. Example) for (exprl; expr2; expr3) { statements; } 64 arm] is the initial condition before any other iterative statements of for is considered for computation. If ezpr2 is true, the main body within { and } is computed followed by computation of expr3 and erpr2 to make a full cycle again for next iteration. Otherwise, the for statement is completed without computation of the main iterative statements. while The statements in the while body are computed repeatedly as long as the ezprI is true before the execution of the while body. This is another iterative selective control flow simpler than the for key word. Example) while (expr1) { statements ; } As a note the control flow do . . . while in C language is not avail- able. break This control flow key word terminates the innermost enclosing loop by for and while. continue returns the next computation program pointer immediately to the innermost enclosing while or for control statement. goto goto identifier; renders the next computation to be the statement of the label identifier. An identifier followed by : is considered as an address label within the scope of a process. 65 Example) goto labell; 8.6 Database MEBL has a statement similar to Structured Query Language(SQL) for sim- ple database manipulation. And for the compatibility with other databases, the database table can be managed by a normal text editor because it con- tains only the plain ASCII text. Four keywords used in database file are: version requires only one argument telling a version number. delim describes the delimiter between fields. field requires three arguments. The first argument is the field name and the second argument describes the type of field. There are only two types; c implies the field is character type and f implies the type of floating number. The third argument is the maximum field size. Record delimiter is set to the new line character. end implies the end of head information. To understand the database structure an example of Environmental Load- ing Unit(ELU) database file is shown as follows: The syntax for the database statement is of the form: 66 version 1 delim field material c 20 field ec c 4 field elu f 10 end Co, RM, 76 Cr, RM, 8.8 Fe, RM, 0.09 Mn, RM, 0.97 Mo, RM, 1.5E3 CD2, EA, 0.09 CD, EA, 0.27 501, EA, 0.10 CFC—11, EA, 300 CH4, EA, 1.0 Nitrogen, EH, 0.1 Phosphorus, EH, 0.3 Figure 17: Example of a database file “eludb” to show database structure 67 select field.name from database where query The database uses the name omitting the suffix .db from the database file name. To specify a field material in the database elu, elu#material is allowed in the query statement. 8.7 Expression 8.7.1 Matrix The elements of the matrix are either separated by a comma or a semicolon. A semicolon is for the change to the next row of a matrix while a comma is for delimiting elements column-wise. For example, a = [ 1 , 2, 3; 4, 5, 6] ; implies two by three matrix. Larger matrices can be generated by using variables as shown in the following example: a= [1, 2; 3, 4]; b= [5; 6]; C = [7, 89 9]; A 3x3 matrix d is generated by using a, b, c as follows: d = [a, b; c] ; will construct d matrix to be three by three matrix resulting in l 2 5 3 4 6 7 8 9 68 8.7.2 Vector Two or three elements are needed to represent a range of values as follows: [expr1: expr2] or [exprlz expr2: expr3] . The first element exprl is the value to start from and the second value expr2 is the final value of a vector. The third element determines the step size for the next element to generate. If the third element is missing, one is used for the default step size. For example [0: 10:2] will generate a vector [0, 2, 4, 6, 8, 10]. 8.8 Operators Operators and their precedences are shown in Table 7. 8.9 Output Functions prval(a) Displays the value of a in the message window. plot(x, y) Draws x-y plot on a pop-up window. title(s) Sets the title message of a pop-up drawing window to s. 8.10 Math Functions det(a) Determinant of the square matrix a. inv(a) Inverse of the square matrix a. 69 Positive of a Negative of a Negation of a Transpose of a matrix a \ O 0‘ Multiplication of a and b Division of a by b Element-wise multiplication Element-wise division A\"IVI+' F? A ~ “0‘ O‘O‘O‘ O'O'O‘O'O’ 0’ Sum of a and b Subtraction of a by b Greater than Greater than or equal to Less than Less than or equal to Equal to Not equal to a and b a or b NNWWUWWWWWNNN 8b, Assignment Table 7: Precedences of operators 7O diag(a) If a is a matrix, diag(a) is the main diagonal matrix. _Or if a is a vector diag(a), creates a square matrix with the diagonal elements the same as a and off-diagonal elements zeros. size(a) Returns the number of rows and the number of columns. eye(a) Returns an identity matrix with the same size of a. zeros(a) Returns a matrix of the same size of a with its elements zeros. ones(a) Returns a matrix of the same size of a with its elements ones. exp(a) Returns a matrix of the same size of a with its elements exponential of the elements of a. ln(a) Returns a matrix of the same size of a with its elements natural logarithms of the elements of a. log(a) Returns a matrix of the same size of a with its elements base ten logarithms of the elements of a. cos(a) Returns a matrix of the same size of a with its elements cosine of the elements of a. sin(a) Returns a matrix of the same size of a with its elements sine of the elements of a. tan(a) Returns a matrix of the same size of a with its elements tangent of the elements of a. 71 acos(a) Returns a matrix of the same size of a with its elements inverse cosine of the elements of a. asin(a) Returns a matrix of the same size of a with its elements inverse V sine of the elements of a. atan(a) Returns a matrix of the same size of a with its elements inverse tangent of the elements of a. cosh(a) Returns a matrix of the same size of a with its elements hyperbolic cosine of the elements of a. sinh(a) Returns a matrix of the same size of a with its elements hyperbolic sine of the elements of a. tanh(a) Returns a matrix of the same size of a with its elements hyberbolic tangent of the elements of a. 72 9 Process Network Execution Model A process network graph also can be represented by Petri Net, a bipartite graph, to aid a visual communication among processes similar to fiowgraphs with concurrent processes capability. Recalling the nature of a system, it has separate processes or components some of which interact independently and some of which have to wait until the required entites are available to begin its own processing. Petri net is a tool for the study of systems [31]. And a definition for the Petri net is as follows: Definition 2 A marked Petri Net Structure, C, is a five-tuple, C = (P, T, I , 0, p); P = {p1,p2,.. . ,p,,} is a finite set of places; T = {t1,t2,. . .,t,,,} is a finite set of transitions where m,n > 0 such that P n T = d); I Q {P x T} is 0 input function of t,,i = 0,1,...,n; O Q {T X P} is 0 output function of pj,j = 0, 1, . . . ,m; and the vector p = (uh/.02, . . . ,pn) is the marking where p.- E N is the number of tokens in place p,-. The state of a Petri net is represented by marking u, an assignment of some number of tokens, represented by large black dots, to places. A transition is able to fire in marking y if each input place to that transi- tion has at least one token in it. A transition is fired by removing one token from each input place and adding one token to each output places, resulting in a new marking. 73 The Petri net can be extended by transition labeling with the addition of tOp-down function or bottom-up function as in Equation 1-8 to each transi- tion depending on the direction of traversing. By the way of constructing an MEB graph, every place of the equivalent Petri net is reachable either in top-down or bottom-up approach. Although the exact execution order of Petri net is not predetermined, due to its concurrency and asynchronous nature, some sequences by which nodes are invoked depend on the goal setup and environmental cost setting. The desired goal of the final product flow rate initiates backward pr0pagation to figure out how much material is needed or how much by-product is produced. The simulation by forward pr0pagation computes every unit cost in a system with the unit costs of materials which are provided from outside of a system. Having defined both the process network and the PN, it is helpful to define the process network in terms of PN for analyzing characteristics of the process network. Another definition of process network graph can be achieved by translation of process network graph into two kinds of extended Petri nets, one for the t0p-down simulation PN, Pt and the other for the bottom-up simulation PN, Pb, as follows: For each instance of building block process, a pair of place and transition is created as a way to deal with the process synchronization problem. For each independent variable of a building block determined by Tables 4 and 5, Petri edge E is created so that each independent wire variable becomes 74 incoming Petri edges of P x T, incident upon a transition node, and each dependent wire variable becomes outgoing Petri edges from a transition t E T to a place p E P, incident upon a place. Again whether a wire variable is dependent or not is determined by Tables 4 and 5. The initial marking of R and P, is determined by series of goals to achieve and those unit costs of materials provided from outside of a modeled system respectively. Given a process network G = (V, T, W, A, 6, T, {3), the algorithm to trans- late the process network into marked t0p-down PN, Pt, using the Table 8 is as follows: P, T, I, 0 +— 43; #0 +- T ; for each n,- E V,T P (— PU {11,-} unless the class of n,.type is SIGNAL type T (— T U {11,-}; for each w = {(n,,nd)} E W if (n, = n,-) then inbound = 0; else inbound = 1; if (g(ns.type, nd.type, inbound) == 1) I <— I U {m} 0 (— 0 U {714} Similar algorithm can be constructed to translate the process network into marked bottom-up PN, Pb, using the Table 9. As an example, two translations from the process network graph of Fig- ure 8.1 consisting of Junction, Production, and Signal class instances into a 75 pair of extended Petri net are shown in Figures 18 and 19. Note that two ordering of nodes in the resulting two PNs are Opposite against each other. In either simulation, whether it is top-down or bottom-up, the solvability of a process creates a token necessary to fire a transition for each input Petri edge in P x T. By the proposed way of constructing an MEB graph, every place is guaranteed to be a reachable place, in other words, every process is guaranteed to have its own solution. 9.1 Execution of a Process The execution environment is responsible for execution of a process and deter- mines the Petri edges coming into a transition. Depending on the solvability of a place which is the source of incoming Petri edges, a token may or may not be assigned for each incoming Petri edge to the transition. If every incoming edge to a transition has a token, then the tOp-down or bottom-up associated with the transition is ready to be fired; otherwise the current process will wait in the queue until all the required tokens are available. If the library class process is encountered, after creation of the related process context and allocation of the required resources, process is switched to the new libray class process. The number of process contexts in a system will be the same as the number of library class instances. The execution environment converts the MEB graph into the MEB Petri net and then, if a transition is ready to be fired, the associated function will 76 U6 U 86(Co) S4(V) 12 J3 0 00“ 0 0 0’0. O x 0 85(Cr) 330’") 82(Ni) ,. o I“ O O .1 0 Figure 18: 'Ii‘anslation of the MEB graph figure 8.1 into tOp—down Petri net 77 82(Ni) 55(Cr) S3(Pb) J2 J3 . U4 . .US . 86(Co) 54(V) Figure 19: Translation of the MEB graph figure 8.1 into bottom-up Petri net “0 “1.. ,. O 78 be executed. The conversion of the MEB graph into the MEB Petri net is made possible by Table 8 which is used in tOp-down approach and Table 9 which is used in bottom-up approach. And if a transition corresponding to a MEB building block is ready to be fired, meaning all the required tokens are available, then the associated top- down or bottom-up function will be executed. Depending on the solvability of the associated function, a new token may or may not be placed on the next place. The corresponding algorithm is shown in Figure 20. Theorem 1 The the process network PN is safe and bounded. Proof: An enabled transition leads to execution of either top-down and bottom-up function associated with the place. Since the number of output function |0| = 1, there is only one place from an enabled transition. As a result only one token is deposited into the place at most. Also there is no loop in the process network PN. Therefore, The PN translated from a process network is always safe and bounded. Cl Theorem 2 There exists a firing sequence to reach every place of the process network PN. Proof: The process network PN is acyclic directed graph with root places given by r or 5. The root places have level of O and its children have level of 1. 79 Base step: The tokens in the root places are initialized by 'r or 13. For the first level transition, |I| = l and |0| = 1. These conditions satisify the firing condition of the first level transition leading to emptying tokens from root places and passing those tokens to its direct descendant places. Hypothesis step: Assume that for all n—th level of places, there exists a firing sequence to have tokens in n-th level places. Induction step: Since the process network PN is an acyclic graph, the level of places needed by (n + 1)-th places are less than (n + 1) which were already deposited from previous steps. Having all the required tokens needed to fire (n + 1)-th transitions are available, the (n + 1)-th transitions are fired transfering tokens to (n + 1)-th places or descendant transitions. Thus there exists a firing sequence to reach every place in the process network PN. Cl. Noting that tokens represent solvability of processes, the above theorem guarantees every process in the process network have a solution, if each subprocess can be solved. The translation of the two-tuple elements in Tables 8 and 9 is as follows: -1 Connection itself is not allowed 0 Connection is allowed but excluded from being the next place 1 Connection is allowed and should be the next place to be exeucted Consider Figure 9.1 to show how to model a system having comprehensive use of all kinds of classes. The library class used in this example encapsulates the system in Figure 8.1. 80 Execute() { IF the next process is a library class, THEN { IF the current process context is not the same as the next one, THEN { Set up enviroments for parameter passing; Switch to a new process context; Execute(); Switch back to the parent process; Post-evaluate the wire variables of a child; process to synchronize with those of a parent process; } IF the relevant tOp-down or bottomrup function associated with this transition is not solvable, THEN { Do not generate a token for this Petri edge going out of this transition; } ELSE { Generate a token for this Petri edge; } FOR each outgoing Petri edge of the current process { Find the next place to execute; If the transition is not ready to be fired, then wait until all the tokens are ready; Execute(); Figure 20: Algorithm for execution of a Petri net 81 PROD. RECY. GOODS STOR. LIB. JUN C . SIGNAL PRODUCTION (1,0) (-1, 1) (-1, 0) (1,0) (1,0) (1,0) (1,0) RECYCLE (oi'l) (0, 1) ('12'1) (“121) (oi'l) (-l,-l) ('1,0)) GOODS (1"1) ('li‘l) (-l,-l) (19'1) (1"!) (1"1) (“110) STORAGE (1:0) (oi‘l) ('1 :0) (1 ,0) (110) (120) “f” LIBRARY (1:0) ("110) ('19‘1) (-l,-l) (li'l) (1,-1L (lv'l) JUNCTION (1,0) (-1.-l) (4.0) (-1.0) (1,0) (1.0) (1,-1) SIGNAL (0:0) (010) (' 1 )0) (" l ,0) (”liol (' 1:0) (’1 fl) Table 8: Conversion table to transform MEB graph into MEB Petri net in top-down approach; row and column are current or next MEB block class respectively and the first and second element of a two-tuple is for inbound and outbound wire respectively _ PROD. RECY. GOODS STOR. LIB. JUN C . SIGNAL PRODUCTION (o, 1) (-1, 0) (.1, 0) (o, 1) (o, 1) (o, 1) (0,0) RECYCLE (oi'l) (0, 0) ('li‘l) ('lio) (0"1) (~l,-l) (0:0)) . GOODS (0,-1) (-1.-1) (-1.-1) (0.:1) (0.4) (0,-1) (0,-1) STORAGE (1.1) (0,-1) (4.0) (0,1) (0.1) (0.1) (0,-1) LIBRARY (1’0) ('120) (“la'll ('la'l) (li'l) (1"1) (1"1) JUNCTION (0:1) ('li'l) (-1,0) (0, l) (0, 1) (0i 1) (Did-T SIGNAL (0.1) (0.0) (4.0) (4.1) (0. 1) (4.1) (-l.-1) Table 9: Conversion table to transform MEB graph into MEB Petri net in bottom—up approach; row and column are current or next MEB block class respectively and the first and second element of a two-tuple is for inbound and outbound wire respectively 82 momma—o :5 mike: 3308 Seaman < "Hm 8&3 mm was 3895 9.20303. 3m 3000...“. ma: 80.3.5 1m m: g 2.. 3. i, was. >183) {>33 \ "213": m3: .. g“ a 3.5.53 _ \ “it 3 .31 a 808k . ......... naoooca \ \ as 1 1‘ m a Ln " 3: N m:.. _ .5 .2. 83032.; ma: Rafi rIIIAIII aIIIIIIIIIIIAIIIIIIIIIIAv w--.c.o.c.fi.m Sanguok {8.3: 8.3.. mm 3 a S a: m: 25 n! or mm c a ma1 6mm .3: Gunfiao aloe aoomao >0: xDvLob >0: confine do: xoaL°> Hon x00L9> at: scantdo:H duoulucH asap edocdu oeddnm Cormack aLaLada :Oquucafi Odawaou onlLoam uOLL Hanan onoaoox couaoavosm 0L4: oeddadom I'll — Hana; _ ucoctou _ anOUIamn aweu nnmu 55000 such 30M> o>Hom cadob com 03Ho> vow anew :4w 3: «Lemon acute: vucu calm: anal Janis: uuddaoz used; .3 . .. . . . . . ... OHHL 83 The translation of the MEB graph Figure 9.1 into MEBL is shown in Appendix section IV. And the correspnding translations of the MEB graph into MEB Petri net are shown in Figures 22 and 23. 84 0 U3 I .‘n . us - . n m. . m Figure 22: Translation of the MEB graph figure 9.1 into tOp—down Petri net 85 .* J1 U1 U3 c: as Figure 23: Translation of the MEB graph figure 9.1 into bottom-up Petri net 86 10 Graphic User Interface Because MEB modeling partitions an overall system into a tractable amount of processes and MEB standardized modules, MEB theory is capable of suc- cinct and crisp modeling of a complicated, very large system. As a system becomes larger and complex, the more importance of designing user inter- face should be emphasized for handling large system comfortably, safely, and efficiently [27] . In the drawing course to partition 3 system, graphical interactions be- tween the user and the MEB simulation system play an important role more crucial than any other phase of simulation. With the goal and sc0pe definition of simulation in mind, a satisfactory modeling comes out of numerous repetitive corrections of models based on its resulting interpretation and its validity check. As is one of MEB simulation characteristics, the mix of top-down and bottom-up approaches makes modeling look a lot more like a real world system because that is the basic nature involved in many design, analysis, and synthesis processes, though such characteristics might add one more complexity to a system. The ease of drawing graphical objects embedding MEB theory determines smooth riding over the important phase of modeling with less pain. With the MEB GUI, from partitioning a model to seeing the results are but a few clicks of a mouse button away. The easiness and the simplicity of interactions make 87 it possible for a user to focus only on defining a system or process boundaries in this phase of partitioning a large system. And this easiness of MEB GUI simulation helps to make a system more understandable. The GUI buttons are used to accomplish and control every aspect of system simlation activities. The interactions to accomplish most of the GDB related activities are designed with the Moore machine[23]. The literal B followed by numerals are the set of mouse buttons and M is the event of mouse movement. Some Moore machines are shown in Figure 24 and Most drawing com- mand buttons have their own automaton. Creation of rectangle shape by (Production, Recycle, Final Prod, Storage, Capsule> command but- tons have Moore machines as in Figure 24 (a). The automata (b) and (c) in Figure 24 are for the , activities associated with its state diagram (a) in Figure 24 are shown as follows: 0 Clear GDB temporary buffer. 1 Store a vertex which forms a corner of rectangle. 2 Display rubber band rectangle. 8 Add a new instance to GDB. In the next section, MEB GUI will be introduced in terms of screen property composition, and interactions to accomplish simulation activities will be described. 88 (C) Figure 24: Moore machines for GUI interactions 89 File Plant Netlist MkJib Plot Nl_fn Grid Redraw Report Up ELU Goal Get Value Solve View Tech Coefl' Capa Cost Byp_Cost Comment Label Save/ load GDB to/ from file, Save/ load simulation envrionment to/from file Save/ load system description in MEBL to/ from file Translate MEB graph into MEBL Use current drawing objects for Library shape definition Plot 2—D graph for a visualization of simulation result Define different non—linear function in MEB theory Define grid size for easy selection of GDB Redraw GDB on canvas Dump all the simulation results to a file Process context switch to parent Evaulate total Environmental Load Unit(ELU) Simulate after Setting up simulation environment Measure system performance Simulate for an individual process Observe resources for a process Modify Technical Coefficient in a process Modify tOp—down approach function Modify bottom—up approach function Change a byproduct cost Assign a comment to a wire variable Change label of building blocks or wire Table 10: Inbound wire;N,,,,,, x chumn constraints in tabular form 10.1 Screen PrOperty Composition The screen prOperty is divided horizontally into three regions as in Figure 11. The first region contains command buttons specifying one of the simula- tion activities. The functionality of the activities is summarized in Tables 10 and 11. The second region is used as the window to show the simulation status Lastly, the third one is the canvans window having the coordinate system with the origin in the upper left-hand corner. By simulation environment I mean the list of goals to be achieved in 90 Polyline Wire Production Recycle Final Prod Storage Capsule Junction Library Polygon Spline Circle Text Inc_Goal Incidence Place polyline Place staircase polyline Place Production class object Place polyline class object Place Goods class object Place Storage class object Define Library class object boundary Place Junction class object Place Library class object Place polygons Place splines Place circle Place Text Define incident points having independent wire variables in creating a new Library class Define incident points having dependent wire variables in creating a new Library class Add Vertex Del Vertex Del Object Mov Vertex Mov Object copy Object Table 1 1: Add vertex Delete vertex Delete object Move vertex Move object Copy object Inbound wire;N,,,,,, x chumn constraints in tabular form 91 goal 06 H8.capa goal U7 H9.capa cost $1 H1.cost [1:10:1]; [1:10:13; 1*ones(size(H1.capa)) + exp(-0.5*abs(H1.capa)); Figure 25: Simulation environment set-up for the system in Table 8.1. terms of flow rate of final products and the unit cost of materials supplied from outside of a system. Sometimes if a sytem gets larger, setting up those simulation environments by hand becomes very cumbersome. If those sim- ulation environmental setups just were saved for later use or for changing part of them, it would save a lot of time or effort doing tedious interactions in setting up the environment every time a simulation is about to be per- formed. An example of such a simulation environment setup of the system in Table 8.1 is shown in Figure 25. The expression of Figure 25 is automatically generated as default tem- plate statements. By changing part of the templates, it is not necessary to remember the whole exact syntax or to enter all of the expressions. 10.2 Graphic Database(GDB) As noted in Figure 14, MEB simulation begins with drawing a system - MEB graph - using MEB building blocks to make a graphic database. MEB GDB is the resultant collection of such graphical objects. An example of a GDB table of the system Figure 8.1 will look as follows: 92 U1 rect 2 0 164 168 236 208 U2 rect 2 0 304 88 376 128 U3 rect 2 0 304 248 376 288 U4 rect 2 0 468 88 540 128 U5 rect 2 0 472 244 544 284 U6 rect 2 2 612 100 672 112 U7 rect 2 2 612 256 672 268 88 188 164 188 236 188 264 188 264 188 264 268 304 268 376 100 428 100 376 280 404 280 404 280 404 124 468 124 428 100 428 256 472 256 540 108 612 108 544 264 612 264 196 208 196 292 340 288 340 336 508 284 508 336 336 88 336 44 504 88 504 44 H1 polyline H2 polyline H3 polyline H4 polyline H5 polyline H6 polyline H7 polyline H8 polyline H9 polyline H10 polyline H11 polyline H12 polyline H13 polyline H14 polyline J1 junction 264 188 J2 junction 428 100 J3 junction 404 280 264 188 264 108 304 108 H16 polyline 428 100 468 100 H17 polyline 404 280 472 280 TEl text 1 348 60 Cr H15 polyline NNNHHHNMNMMMNQWMMWMN TE2 text 1 516 64 Co TE3 text 1 172 248 Ni TE4 text 1 316 312 Pb T35 text 1 488 312 V Figure 26: MEB GDB table of the system in Figure 8.1 93 11 Case Studies 11.1 Swine/CrOp System MEB theory is applied to a pasture—based farrow-to—finish swine production system. Three crops(wheat, soybeans, and corn) are grown on the farm to provide the bulk of the feed ration for the swine. The swine/crOp agroecosys- tem is partitioned into physical and biological production processes[2]. The process flow diagram of a agroecosystem are depicted in Figure 11.1. Based on the data used in [2], the technical coefficients of the complex net- work of paper manufacturing are described in Tables 13,14, 14, and 15. Given the unit costs of input material flows in 12, the unit costs of inter- mediate product and the final product are shown in Tables 18 and 19 along with the amount of each material flow. 11.2 Paper Cup LCA The process flow diagram of a paper manufacturing plant and that of paper use and disposal are depicted in Figures 28 and 29 respectively. To enhance the view of overall LCA of paper cup, the whole LCA is described in Fig- ure 30. This hierarchical modeling not only gives a bird’s eye view of LCA but also gives detail system description as the system is further explored. Based on the data used in [26], the technical coefficients of the complex network of paper manufacturing are described in Table 21 where response variablm on the left column are expressed in terms of stimulus variables. Also the technical coefficients of the use and disposal phase of LCA is described 94 Variable Description Unit cost W1 wheat seed 0.12500 W2 soybean seed 0.18750 W3 corn seed 1.18750 W4 grass seed 1.57500 W30 NPK fertilizer 0.35000 W31 pesticides 2.27000 W28 NPK fertilizer 0.35000 W29 pesticides 4.76000 W44 NPK fertilizer 0.35000 W45 pesticides 1.57 500 W70 manure 0.00000 W20 water 0.00000 W21 medication 0.05000 W59 boars 233.33333 W60 sows 0.00000 W49 water 0.00000 W50 medication 0.05000 W67 gilts 155.000 W46 water 0.00000 W47 medication 0.05000 W26 purchased feed 0.04464 W27 supplement 0.15500 Table 12: Unit cost of material flux 95 Res. Description Tech. Unit var. coeff. U1(SOIL/ PROD Wheat) W1 seed 2.5 lb/bu W30 NPK 1.14286 lb/bu W31 pesticides 0.12114 lb/bu W32 leaching 0.00000 lb/bu W5 biomass stimulus bu U2(HARVESTING Wheat) W33 losses 0.05263 bu / bu W5 biomass 1.05263 bu/ bu W6 grain stimulus bu U3(STORAGE Wheat) W34 losses 0.01010 bu/bu W6 biomass 1.01010 bu/bu W7 grain stimulus bu U4(Transport Wheat) W35 losses 0.00017 bu/lb W7 gainss 0.01684 bu/lb W14 wheat stimulus lb U5(SOIL/ PROD Soybeans) W2 seed 1.77778 lb/bu W28 NPK 1.03333 lb/bu W29 pesticides 0.07189 lb/bu W39 leaching 0.00000 lb/bu W8 biomass stimulus bu U6(MAR.KETING Soybeans) W38 losses 0.05263 bu / bu W8 biomass 1.05263 bu / bu W10 soybeans stimulus bu Table 13: The technical coefficients of swine/crOp agroecosystem 1 96 Res. Description Tech. Unit var. coeff. U7(STOR.AGE Soybeans) W10 soybeans 1.01010 bu/bu W37 lossess 0.01010 bu/bu W12 soybeans stimulus bu U8(TR.ANSPORT soybeans) W12 soybeans 0.01684 bu / lb W36 losses 0.00017 bu / lb W15 soybeans stimulus lb U9(SOIL/ PROD Corn) W3 seed 0.09836 lb/bu W40 leaching 0.00000 lb/ bu W44 NPK 1.17612 lb/bu W45 pesticides 0.07297 lb/ bu W10 biomass stimulus bu U 10(HARVESTIN G Corn) W41 losses 0.05263 bu/ bu W9 biomass 1.05263 bu / bu W11 grain stimulus bu U 11 (STORAGE Corn) W11 grain 1.01010 bu/ bu W42 losses 0.01010 bu/bu W13 grain stimulus bu U 12(TRANSPORT Corn) W13 losses 0.01804 bu / lb W43 biomass 0.00018 bu/ lb W16 soybeans stimulus lb U13(SOIL / PROD Pasture) W100 leaching 0.00000 lb/ lb W4 seed 0.01400 lb/lb W69 manure 61.11110 lb/lb W17 biomass stimulus lb Table 14: The technical coefficients of swine/crop agroecosystem 2 97 Res. Description Tech. Unit var. coeff. U14(GRAZING Pasture) W17 biomass 8.69565 lb/lb W18 grass feed stimulus lb U15(Manure 'Ikansport) W68 manure 1.00000 lb/lb W51 manure stimulus lb U16(Breeding and Farrowing) W19 grass feed 0.44190 lb/piglet W20 water 147.780 gal/piglet W21 medication 25.0000 mg / piglet W24 feed 134.270 lb/piglet W52 manure 31 1.116 lb / piglet W54 gilts 0.03000 hd/piglet W57 cull 0.03400 hd/piglet W59 boars 0.00600 hd/piglet W60 sows 0.04200 hd / piglet W61 piglets stimulus piglet U17 (Nursery) W49 water 21.0000 gal/piglet W50 medication 2.00000 mg/piglet W53 manure 48.29974 lb/piglet W56 dead animals 0.29070 hd/piglet W61 piglets 1.29199 hd / piglet W64 feed 15.0000 lb/piglet W62 piglets stimulus piglet Table 15: The technical coefficients of swine/crOp agroecosystem 3 98 Res. Description Tech. Unit var. coeff. U18(GROWING/FINISHING) W46 water 452.08333 gal/SH. W47 medication 1.00000 mg / S.H. W48 gilts 0.04167 hd/S.H. W58 dead animals 0.03225 hd/S.H. W62 piglets 1.07500 hd/S.H. W63 pasture 2.58056 lb/S.H. W65 feed 784.19444 lb/S.H. W66 manure 1043.75 lb/S.H. W55 slaughter hogs stimulus S.H.(Slaughter Hog) U19(FEED/MILL) W14 wheat 0.36293 lb/lb W15 soybeans 0.14282 lb/lb W16 corn 0.47274 lb/lb W25 losses 0.02041 lb/lb W26 purchased feed 0.00000 lb/lb W27 supplement 0.04192 lb/lb W22 feed stimulus lb Table 16: The technical coefficients of swine/ crop agroecosystem 3 99 Process Description Fixed Variable Unit cost name cost cost U1 SOIL/ PROD Wheat 716.05 0.32143 $/ bu U2 HARVESTING Wheat 741.81 0.30827 $/ bu U3 STORAGE Wheat 356.59 0.08069 $/bu U4 Transport Wheat 0.00 0.00008 $/1b U5 SOIL/PROD Soybeans 453.33 0.97778 $/bu U6 MARKETING Soybeans 306.51 0.48655 $/bu U7 STORAGE Soybeans 192.26 0.08664 $/bu U8 TRANSPORT soybeans 0.00 0.00008 $/lb U9 SOIL/PROD Corn 503.87 0.32836 $/bu U10 HARVESTING Corn 387.59 0.22938 $/bu U11 STORAGE Corn 854.53 0.23011 $/bu U12 TRANSPORT Corn 0.00 0.00008 $/lb U13 SOIL/PROD Pasture 0.00 0.06000 $/lb U14 GRAZING Pasture 0.00 0.02609 $/lb U15 Manure Transport 183.75 0.00005 $/lb U16 Breeding and Farrowing 1,001.84 3.54662 $/piglet U17 Nursery 1,732.79 3.71150 $/piglet U18 GROWING/FINISHING 3,170.35 11.08271 $/S.H. U19 FEED / MILL 2,805.67 0.00200 $ / lb Table 17: Unit input costs 100 88630883 aoSFEBm x833: mmE Km PSmE 9:3 9.888% ~90. xo 9.: 3.25.» kg: an... 85 ,llle 8.3.8.. 2.32.... 2.38 t»... . soon 3.. . 9.58 SIILMO 3.. _ m g— .“ mum 3.: N80552: :80 m3“ [Mtllo 8.55.5. 8.8.“! SED§ .888 3.3% 8...! 0.5833 no... gggfiool 1 N31 980309 91 , w: 3 4:2 coonmom gm 80038 9.3.9.5! SIMS agmow voommm 8: 8 .5 8 8.. mm: a. mum... 33033... x9. 0003. 383 393 o .5; mm... a8 3: Sm m8 3. .8 8.68. 3.. 33...: E1 so; 9.1 our... 91 03.... n. a... m9 3...... EszcE SM... 335 some 2:33.... Sousa :55 88.3” acosoESmm 3 m... a... on... too... macaroni 3.03983 N.“ x9. mold: 38 31.8 3.. :3...» 3.. 80?: a... 3...... 1.. 23...: I... _ Ogden—H 3.10.; 080.— 0.—.o...—.u 0.3de 8.1.0.. _ FISH. _ 8365.. _ 61.2.8 _ 8.85 3.... ~25. 53.5.... 8363...... 0.5. 833.: 1...... _ acoiou _ 381...... _ anon. _ 2.8 .28”. so.» :8: Beam 81.. uom 31.. a... How in .5 €83. its... 3.6 5...... our. 3.3.: and»... 2.3.. .3“. Variable Description Material outputs Cost per unit of output W55 HOGS(goal) 360 370 101.74 100.7 W1 wheat seed 5771.4 5931.7 0.125 0.125 W2 soybean seed 1615 1659.9 0.1875 0.1875 W3 corn seed 316.85 325.65 1.1875 1.1875 W4 grass seed 139.99 143.88 1.575 1.575 W5 biomass 2308.5 2372.7 1.6191 1.6107 W6 wheat grain 2193.1 2254 2.3508 2.3329 W7 wheat grain 2171.2 2231.5 2.6195 2.5969 W8 biomass 908.46 933.69 2.514 2.5005 W9 biomass 3221.3 3310.8 1.1281 1.1239 W10 soybeans 863.04 887.01 3.488 3.4642 W11 grain 3060.3 3145.3 1.5436 1.5357 W12 soybeans 854.41 878.14 3.8349 3.8048 W13 grain 3029.7 3113.8 2.0548 2.0397 W14 wheat 1.2893e+05 1.3251e+05 0.044192 0.043812 W15 soybeans 50737 52146 0.06466 0.064152 W16 corn 1.6794e+05 1.7261e+05 0.037149 0.036876 W17 biomass 9999.6 10277 0.10318 0.10268 W18 pasture 1150 1181.9 0.92327 0.91895 W19 pasture 220.95 227.09 0.92327 0.91895 W20 water 73890 75943 0 0 W21 medication 12500 12847 0.05 0.05 W22 feed 3.5525e+05 3.6512e+05 0.05723 0.056677 W23 feed 72940 74966 0.05723 0.056677 W24 feed 67135 69000 0.05723 0.056677 W25 losses 7250.7 7452.1 0 0 W26 purchased feed 0 0 0.04464 0.04464 W27 supplement 14892 15306 0.155 0.155 W28 NPK 938.74 964.81 0.35 0.35 W29 pesticides 65.309 67.123 4.76 4.76 W30 NPK 2638.3 2711.6 0.35 0.35 Table 18: Swine/ Crop agroecosystem 1 102 Variable Description Material outputs Cost per unit of output W55 HOGS(goal) 360 370 101.74 100.7 W31 pesticides 279.66 287.43 2.27 2.27 W32 leaching 0 0 0 0 W33 losses 115.42 118.63 0 0 W34 losses 21.929 22.538 0 0 W35 losses 21.918 22.527 0 0 W36 losses 8.6253 8.8648 0 0 W37 losses 8.6295 8.8692 0 0 W38 losses 45.422 46.683 0 0 W39 leaching 0 0 0 0 W40 leaching 0 0 0 0 W41 losses 161.06 165.54 0 0 W42 losses 30.6 31.449 0 0 W43 losses 30.229 31.069 0 0 W44 NPK 3788.7 3893.9 0.35 0.35 W45 pesticides 235.06 241.59 1.575 1.575 W46 water 1.6275e+05 1.6727e+05 0 0 W47 medication 360 370 0.05 0.05 W48 gilts 15.001 15.418 0 0 W49 water 8127 8352.8 0 0 W50 medication 774 795.5 0.05 0.05 W51 manure 5.5e+05 5.6528e+05 0.00038 0.00038 W52 manure 1.5556e+05 1.5988e+05 0.00038 0.00038 W53 manure 18692 19211 0.00038 0.00038 W54 gilts 15 15.417 -0.01236 -0.01236 W56 dead animals 112.5 115.63 0 0 W57 culls 17 17.472 0 0 W58 dead animals 11.61 11.933 0 0 W59 boars 3 3.0833 233.33 233.33 Table 19: Swine/CrOp agroecosystem 2 103 Variable Description Material outputs Cost per unit of output W55 HOGS(goal) 360 370 101.74 100.7 W60 sows 21 21.583 0 0 W61 piglets 500 513.89 16.41 16.28 W62 piglets 387 397.75 30.368 30.07 W63 pasture 929 954.81 0.92327 0.91895 W64 feed 5805 5966.2 0.05723 0.056677 W65 feed 2.8231e+05 2.9015e+05 0.05723 0.056677 , W66 manure 3.7575e+05 3.8619e+05 0.00038 0.00038 W67 extra gilts -0.0011961 -0.0012293 155 155 W68 manure 5.5e+05 5.6528e+05 0.00038409 0.00037506 W69 manure 6.1109e+05 6.2806e+05 0.0003457 0.00033757 W70 excess manure 61085 62782 0 0 Table 20: Swine/CrOp agroecosystem 3 in Table 22 and 23. With the unit cost of system resourses in Table 24, some of the simulation results, for brievity, are shown as follows: PROCESS(0): plca H1. H1. H2. H2 . H3. H3 H4. H4. H5 H5. H6 . capa: cost: capa : cost: capa: .cost: capa: cost: .capa: cost: capa : [0.001 0.002 [0.25 0.25 0 [0.06 0.12 0 [0.25 0.25 0 [0.02 0.04 0 [0.25 0.25 0 [0.03 0.06 0 [ 0.25 0.25 0 [0.01 0.02 0 [0.25 0.25 0 [0.01 0.02 0 0.003 0. .25 .18 .25 .06 .25 .09 .25 .03 .25 .03 0. O CO CO CO CO 25 .24 .25 .08 .25 .12 .25 .04 .25 .04 O CO CO CO CO C 104 Variable Description Tech. Coeff. U3(Pulp Manufacturing) W12 r3(chlorine) 0.06 W49 r4(sodium hydroxide) 0.02 W54 r5 (sodium chlorate) 0.03 W55 r6(sulfuric acid) 0.01 W56 r7(sulfur dioxide) 0.01 W57 r8(calcium oxide) 0.01 W58 r2(water) 0.10 W39 wood chip 2.2 W1 sodium sulfate(s.s.) 0.009 W11 recycled as 0.01 W16 r22(H20) 0.07 W17 r23(suspended solids) 0.01 W18 r24(BOD) 0.005 W19 r25(organochlorides) 0.003 W20 r26(cellulosic fiber) 0.001 W21 r27(inorganic salts) 0.06 W22 r19(chlorine) 0.0002 W23 r20(chlorine dioxide) 0.0002 W24 r2l(reduced sulfides) 0.0015 W46 black liquor 1.2 W24 pulp stimulus U6(Waste Wood & Black Liquor Combustion) W41 C02 0.0 W42 CC 0.028 W43 N Ox 0.046 W44 S02 0.100 W45 particulates 0.015 W2 bark and waste stimulus W59 smelt 0.17 W46 black liquor stimulus 105 Table 21: The technical coefficients of paper manufacturing Variable Description Tech. Coeff. U4(Waste Paper Re-pulping) W29 r3(chlorine) 0.06 W28 r4(sodium hydroxide) 0.02 W27 r5(sodium chlorate) 0.03 W26 r6(sulfuric acid) 0.01 W25 r7(sulfur dioxide) 0.01 W13 r8(calcium oxide) 0.01 W14 r2(water) 0.10 W53 waste pulp 1.0 W30 r22(H2) 0.04 W31 r23(suspended solids) 0.005 W32 r24(BOD) 0.003 W33 r25(organochlorides) 0.002 W34 r26(cellulosic fiber) 0.001 W35 r27 (inorganic salts) 0.03 W36 r19(chlorine) 0.001 W37 r20(chlorine dioxide) 0.001 W38 r21(reduced sulfides) 0.0008 W48 recycled pulp stimulus U5(Wood Processing) W2 bark and waste 0.06 W40 wood logs 1.06 W39 wood chips stimulus U7(Paper Manufacturing) W47 pulp 1.0 W48 recycled pulp 0.0 W51 H20 0.01 W50 paper stimulus U2(Cup Manufacturing) W2 paper 1.0 W6 adhesive 0.0 W8 waste paper 0.03 W3 cups stimulus Table 22: The technical coefficients of cup use and disposal 106 Variable Description Tech. coeff. U2(Cup Use) W4 cups 1.0 W7 beverages 1.0 W9 used cups 1.0 W5 beverages stimulus U6(Waste Paper Transport) W10 incin. wasted paper 0.0 W12 landfill wasted paper 1.0 W8 waste paper stimulus U7 (Used Cup Transport) W11 landfill used cups 1.0 W13 incin. used cups 0.0 W9 used cups stimulus U8(Used Cup Ti‘ansport) W14 r14(C02) 0.0 W15 r15(CO) 0.028 W16 r16(NOx) 0.046 W17 r17 (302) 0.100 W18 r18(particulates) 0.015 W19 fuel needed 0.0 W20 ash 0.03 W13 used cups stimulus U9(Landfill) W21 r14(COZ) 0.0 W22 r32(methane) 0.0 W23 r33(leachate) 0.0 W24 r34(cellulosic fiber) 1.0 W25 r24(BOD) 0.0 W11 used cups stimulus W12 wasted paper stimulus Table 23: The technical coefficients of cup use and disposal 107 Shaman mkmcgofincefi .8an mo ESE mimbz “mm PSmE s... w - .1. - a... _| .1 -1. 00033058... 1 l 2.58 1 Now CENSUS“. 303:3» tognozvac [All _ Loans. 32.: [Al Ill .9. [LII .83 33.25 8.8288,. S 05.5 1. [All 8 [All 8:...333 Lose: 5.8; A . 28m.“ 333 was {on N8 [All 333 3:080:35? [All 9.5an ouon3-oquNL [All academic». .. Anon . L2 £008.95 5N... [I n QaJQ B~UNOOM AQBVVNL n Loam Dvwo3 21.000)... .rL. Ann—Mow newcoaosmvmmc [All 2 “.003 m l 838.. lull, . e. o. . > .0003 ON: 9.79.: cocoa A . > ”3:0 too: m.---.hmm:._ll, as. s 1 ............. .5833” 833.51 [Ail . 1 3336:... 8.82809. [Al 1 1. (L oflnwonpum: £30? 358...“.va l, < . , u "1 1 03.8%. (523.0. 323 3:095:35? [Al . A1 1 Boo 2.523 Con; 3333988.. [All 9.5.3: 33 1 Av 1 32020 .5236 [ll Anette—505938.. s. 1 03x05»... .533 38.3.. [l < 3320” novcoanzmvmwc [I has 1 0:782”. SNIVNNL Ill. J .Illl. 1 m ".1 lull r [1.— 30.2.8 530m now.“ on: don—1.3 aaoo coo-.8 .5: sou-50> .5: via: do: .393: do: .3052, at: coco—$.95 ducalup-H 0x0... own-.56 0.54am gem 3.3.5.: Ewart-.5... in. ouauoam tor...— ddcdu 0.3300“ ceduosok. and: 8.351.; Henna disco unsure—5 £39 :3 .3009 £00... 30.; cad-x." 81> van 0:1) com— :66 =._w n5 neon-ox 30.503 3.59 eel—z 00.: bur—Ix: 001—502 055. 0.3.... 108 889mm Emoammv 93 mm: .593 m0 .258 mmE "an 8&3 mu: OIIITI AQBVQNL mHm QNI at 22:... 0303:03me 0 “an “320035"... 0 Mum. :5:qu €8th Aotncuorvwmg nummm «um. man no»: AmoovvaL «am 8 5 m5 «3 m use! is“. Olldgnl AfiSHL 0 was. Aoouo~:o_»gnmvmag anHMIIMHanIIIIII coda» mou 3095.. O was. agement 26:57. Okhlflsfl b 3... 38m: OIMmImAJOII couuogosntu 1 519m. 33: A893... . 9.. a g m g _ o no»: Leia 0225 .............. m: v: m: «an ,IllmmnIMWAu gonna “-1--------mn-h on: nasu uLonncoLh a: m:«.mmz Lemon uuoaocngh m: .3 a3 a u as Locum 3 m: 8 m. S 1 11111Au o>.oo;u¢ 3 am 1 11AU «doaLoaoa mm Aooa .owv £0052: amoo HOOHAD :0: soauo> be: aUOfiaa no: xouLo> Ho: xovLo> at: oocoumocH HmouIOCH axoh odosfiu oeddam abandon aLoLadJ codaocafi oqaancu ouauoaw tOLL HQCdu Oduaoom :oqvoavcgm OLda ochadom fi1 done; _ acetsou _ vuoulaam anon _ omuu sheen such aud> cadom 03Ho> dam oaqu> new H009 =4“.— .5 9.3%.: 30.503 1.75 :5142 00,—... 43's: unmade: arid; addu 109 mam—128 £98 £2 96 Sada 15“ E58 mmE ”om 8:me jll'l. _ . 1 _ . J «a _ _ 731* 1 +2.11. 09 m nbmflktia IIEAI .IIAT. m . 11111111111111 L ”inflkxaa _n5«93%& _ TITLTII _ _ 1|:I:II.2X8:9£5_ _ _ ~099kzn th~o5:.kXIm _ m F:IIII:ILMMHHMMW a: HIIIIIIIIIL .rlll..| 33:3 .368 _ 8 r c 8.5983 83.. _ 9amaass§§§ I fil I | 51 El 5| 3 a H 3:1 8 a 4 Lu A. MHMAV a awmm ~ovv is 2.8 goo?! 2.: x3...» 8: coo—18 «on 33.8: 1.. a3...» 2.: oocovuocn Hooalocfl asap Odosdu ocddtm abandon 5395.. _ 33.3 owls; 3.... 15.1. 3N3”. 8383.... 95. 833.... 1 _ 2.8 too“. so: an; 038 3:5 sum 815 van anew veto _ aLILmHu ~33 _ 1.3.3.8 _ vacuum... 3w 1.: 1“ 3 :bldz and; Addlxz uuddvoz Head; eddu v.1... \\<«\\t¢\\\\t1\\1\\x xxxx A11\\\\\..11 .11 xx .111 1,.1..H11.1H1.“u W. ,111 111.1 111 1.11 1111€,\1.w\w Mmdwxwxfiw.U..\N11\1XYVNHM¢\1.11111VXuxfliflxycxxc .1x1,1.&\.H..1»111\11.\1\1\$11\1,f\1.-H\1:s “.1 1x11..1...1-”\.11.11.11\.’.\1;..1\1>H111\1,..111\H3111111111., 1911\11911311111611\11i11\1;.11.11,1..,,\..11 1 1 . . 1. 1 11.1.1 . . ,1. 11(11... (\1. . (\1. . \1. .1 1 11.1. ‘1. 11.1. 11. 1.1 .11 .11 .1 1. ..\.11..1\111 1.1 13.. \1 . 117.11.1111 . 3. .99 16.51 1. . 1.0.31. .1\.\\.,1.u\1 11.1.\1 13 11 .11.. 1x ...1. .116... 11.1(u. 1. 11.. 1.. 1.\.(11.(... 11.)\1. 3!... Qt.\ 1.1/1.143511” ma. 11.1.31 .1 ‘11; .1511 .1111...\1.1 1) .1 .1 1., 1. \x. 11 1 , 1.1\121\1 1 1. \111211 1. 1111 11:1. 1111.. 1.11.1. . ,1 x, \19 a x, 1 11 1 a...“ \xnx\1>\ue\\31\\3\®,\\>\\vw\>\\\11,1”. ”1111111111111” 11.111111111111 11111, 1111 MR?“ 1\>1\1,>\\1111Y\Cu> 11x” 3&fikfikwm? .muixfi 1111.11,.1m11111131ux1flb11m1111.110 1.11111\.11\11311_.\111u1\1m1 1.1«1fivx11111v111w111u.n111111m1flfl1fi11.«111111,11,11111114111111HA1JM1AJ111M.11, $111“ 1 11 111.11...» 11,111.»,111111, 1181». 110 Res. Description Energetic var. cost W1 Sodium sulfate $0.25 W2 Chlorine $0.25 W3 Sodium hydroxide $0.25 W4 Sodium chlorate $0.25 W5 Sulfuric acid $0.25 W6 Sulfur dioxide $0.25 W7 Calcium hydroxide $0.25 W8 Water $0.00 W9 Wood logs $0.10 W10 Waste paper $0.00 W11 smelt $0.00 W14 Beverage $0.00 W15 Adhesive $0.25 Table 24: The unit costs of system resources for paper cup use and disposal W6.c08t: [ 0. 25 0. 25 0. 25 0. 25 0.25 ; J H7.capa; [ O. 01 0. 02 Q 03 Q 04 0.05 ; J W7.c08t: [ Q 25 Q 25 Q 25 Q 25 0.25 ; ] 98.capa: [ 0.1 0.2 0.3 Q 4 Q 5 ° ] W8.c08t: [ 4. 546 -05 2. 06126-09 9.35766-14 4.24846-18 1.92876-22 ; J U9.capa: [ 2. 332 4. 664 6. 996 9. 328 11.66 ; ] W9.c08t: [ Q 08 Q 08 Q 08 Q 08 0.08 ; J 910.capa: [ 0 0 0 0 0 ; ] H10.cost: [ 1 1 1 1 1 ; ] H11.capa: O ; H11.cost: 1 ; W12.capa: [ 1 2 3 4 5 ; J 812.c08t: [ 0.22181 Q 22181 Q 22181 Q 22181 Q 22181 ; ] 913.capa: [ 1 2 3 4 5 ; J U13.C08t: [ 0.22181 Q 22181 Q 22181 Q 22181 Q 22181 ; ] 111 W14.capa: [ 1 2 3 4 5 ; ] W14.cost: [ 3.72016-44 .38398-87 5.14826-131 1.91526-174 7.12469-218 , ] W15.capa: [ O O 0 O O ; J W15.cost: [ 0.25 0.25 0.25 0.25 0.25 , ] PROCESS(1): puse W1.capa: [ 1 2 3 4 5 ; J W1.C08t: [ 0.22181 0.22181 0.22181 0.22181 0.22181 ; J W2.capa: [ 1 2 3 4 5 ; ] W2.c08t: [ 0.22181 0.22181 0.22181 0.22181 0.22181 ; J W3.capa: [ 1 2 3 4 5 ; ] W3.C08t: [ 0.22181 0.22181 0.22181 0.22181 0.22181 ; J W4.capa: [ 1 2 3 4 5 ; J W4.C08t: [ 0.22181 0.22181 0.22181 0.22181 0.22181 ; J W5.capa: [ 1 2 3 4 5 ; ] W5.C08t: [ 0.22181 0.22181 0.22181 0.22181 0.22181 ; ] W6.capa: [ O 0 O 0 0 ; J W6.C08t: [ 0.25 0.25 0.25 0.25 0.25 ; ] PROCESS(2): pmfg W1 W1 W2. W2. W3. W3. W4. W4. .capa: .cost: capa: cost: capa: cost: capa: cost: HP! HP“! HH FM“! 00 CO CO 00 .027 0.036 0.045 ; J 18 o 0 0 ] 264 0.396 0.528 0.66 ; J 0;] 2 5 112 W5.capa: [0.03 0.06 0.09 0.12 0.15 ; ] W5.cost: [0.25 0.25 0.25 0.25 0.25 ; ] W6.ca.pa: [0.01 0.02 0.03 0.04 0.05 ; ] W6.c08t: [0.25 0.25 0.25 0.25 0.25 ; ] 11.3 Ford F150 Truck Tail Light Assembly The tail light assembly structure plant is shown in Figure 31. The technical coefficients of the Ford F150 truck tail light assembly plant are give in Table 25 and Table 26. 11.4 Water Plant Modelling The water plant is modelled in Figure 32. And all the byproduct costs are assumed to be zero dollar. The electricity cost of pump with efficiency of 0.85 is computed as follows: 12¢,“ = $0.05 - 0.73kwh/1KGD/100PSI- kgd . psi/0.85 = 0.042941 - kgd ° psi Given the plant water demand, the problem is to find all the costs of products in the water plant. The result is summarized in Section 11.4. The plant water demands are as follows: 113 Res. Description Tech. unit var. coeff. U1(Body molding) W1 plastic 0.551 oz. W2 gates 0.05 unit W3 scrap 0.05 unit W4 body stimulus unit U2(Body metallized) W15 metal waste 0.0 mg W19 metal 75 mg W4 body 1 unit W5 metalized body stimulus unit U3(Glued body) W16 purge 0.0088 oz. W17 plastic 0.0088 oz. W20 glue 0.013 oz. W5 body 1 unit W6 glued body stimulus unit U4(Lens mating) W30 lens 1 unit W6 glued body 1 unit W7 body stimulus unit U5(Dry on rack) W7 body 1 unit W14 body stimulus unit Table 25: F150 tail light assembly modeling 1 114 Res. Description Tech. unit var. coeff. U6(Drive studs) W14 dried body 1 unit W31 studs 2 unit W6 body stimulus U7(Leak test) W18 scrap 0.0045 unit W8 body 0.9955 unit W9 body passed stimulus unit U8(Put bulbs in) W21 body 1 unit . W9 socket 1 unit W10 body stimulus unit U9(Inspection) W10 body 0.95 unit W12 scrap 0.05 unit W11 body stimulus unit U13(Socket assembly) W32 bulb 2 unit W33 socket 2 unit W21 socket stimulus unit 115 Table 26: F150 tail light assembly modeling 2 23a fess 2%: :3 SE Bow ”a 2:5 5 «£3 3.. a: 3: NH o 83» 83. 83 (Ml—82:3... 38 3 032:“. aszaM B 32% 3: 8 n: S «a o «:3 n3 mm mm mm m ~32. no.6“ ~33 03m 6 a 3!: 8...: 3m 16!. A8.“ ~th 11...: anon non—.8 .6: 33.5., 2.: 12.8 1: :3..o> 1: :35; ave 00:03.0:H doom—I05” 11.30.— ononuu 04¢.de cog—6.. 1.3.5.3 sedan-Ia. odd-anon. ouanouw to...“— 3... Oduaoou flown-03.0.... 0.5.: 0.5.1.50; _|1._3 _ 05:5... _ 18:5: 13 2.3 :98 so: 33> 2.1m 31> 1m 31> «on 18 :._m .5 0,-3.0». 20.50“ Eda fluid: as... fiddly! 9.1—Jo: 09...: 0.3.... 116 H 2230 298mm ”mm Eswfi “ovum: 0.52301 0 .. ............ L -mbfiFukm.-- m: _ CLsuom H . ............ N: ma. 9.. m2 “-mwmm-flfl---m :3 28:8 mm: 1% .. "as: u 9.55 " wmmm>mm ll m .5 . ............ EL 3: a h 33 m: cm: .32 a 3 nm m o mm “60 acoocou 3.3: L303 35 Ass: 05:80». E .35 2... Egg 2:3... EE .5 . fl 117 Product Amount of Flow Cost per Unit($) Unit Soft Water(W26) 500 ? $/ yr KGD DI(W26) 300 ? $/yr KGD UPW(W17) 800 ? $/yr KGD Total Cost '? $/yr . REVERSE OSMOSIS(U5) 1. Pump dP = 300 psi, Eff. = 85% 2. Rejection 95% of TDS ppm, 100% of R0 Feed TSS 3. Concentrate Flow(W21) = 5% of R0 Feed(W25) 4. Bleed Flow(W20) = 5% of R0 Product(W_f) 5. Return Flow(W5) = 5% of R0 Product(W_f) Born (3), (4), (5), and by the conservation of mass respectively, 3121 = (1051125 3120 = 0-05yf y5 = 0.0531, y25=y5+ygo+y21 +31] 31} = 3117+y2o + 315 de => y5 = y20 = 0.05yf =f a => 3125 = (a) + (a) + 3121 + (20(1) ¢ 203/21 '2 220 + yzl 118 => y21 = 212-99- = 1.15790 => 3,25 = (a) + (a) + (1.1579a) + (205) =.- 23.1579a => 3'! = 3117 + (a) + (a) => 20a = 3117 + 20. => 0 = 11181 gm 1 0.05555 3121 __ 1.1579 _ 0.06433 ‘12! => 3,25 ‘ 23.1579 “ ‘ 1.28655 3’" " kg” 3,5 1 0.05555 1320 _ I 5321 x” ‘ k $25 + 0.042941 :1 3 $5 . ULTRA FILTER(U4) 1. Pump dP = 100 psi, Eff. = 85% 2. Concentrate Flow(W19) = 10% of UP Feed(W4). 3. Cleaning Flow(Wll) = 1% of UF Feed(W4). 4. Rejection = 0% of TDS ppm, 95% of TSS ppm From (2), (3), and by the conservation of mass respectively, 3119 = 01.714 3111 = 0.01114 119 314 = 315 + .1111 +3119 114 = 115 + (0-01y4) + (01114) (1 — 0.11)y4 = y5 y4 = 1.1236315 {LUUULL yll 0.011236 a. 3119 = 0.11236 y5=!ky5 314 1.1236 ll 11511 $24 = ’6’ 1319 $4 + 0.042941 o DEGAS(U3) 1. Cooling Water (City) Usage(W18,W23) = 0.5 gpm/KGD = 0.0005 kgpm/KGD 2. Degas Water Feed(W7) = Degas Water Product(W4) From (1), (2), and by the conservation of mass respectively, .7118 = 3123 = 0-53/4 3/7 = 314 => 3):, = 0.05y17 + 0.052117 + 3117 120 => 115 = 1-13117 3118 0.0005 dc => 1,23 = 0.0005 y4 =’ km 317 1 . ION EXCH.(U1) All the by-product costs are assumed to be zero dollars. 1. Pump dP = 100 psi, Efl'. = 85% 2. Regen. Flow(WIO) = 35% of IX Product(W7) (50% Feed(W16), 50% Soft(W2), + Acid(W14) & Base(W13)) 3. Effectiveness = 100% removal of IX Feed TDS Ion 4. Acid Usage 2 l gal/kgal/ 100ppm ion Acid Cost = $0.30/gal = $300/kgal 5. Base Usage = 0.5 gal/kgal/ 100ppm ion Base Cost = $1.50/gal = $1500/kgal 6. TDS in City Feed 300 ppm From (2), (4), (5), (6), and by the conservation of mass respectively, ym = 0.35y7 1116 = 312 «1;; a 3114 = (0.001)(300ppm/100ppm)(y16 + 212) = 0.0060 3,13 = (0.0005)(3001mm/100ppm)(y15 + 312) = 0.0036 121 317 + 3110 = 3116 + 312+3113 + 3114 => 317 + 3110 = (a) + (a) + (0.006a) + (0.003a) = 2.00961 => (0—65 + l)3110 = 2.009a => ylo = 0.520852a => (0.5208526) = 0.353,7 f 3110 ' F 0-35y7 ‘ ' 0.35 3113 0.003a 0.002016 .1. => y” = 0.0060 = 0.004032 y7 =’ km 3116 0 0.671976 _ y2 . _ a 1 _ 0.671976 .1 - $10 1 P 0 $13 + 0.042941 1500 + 0.042941 $7 = ’13, $14 + 0.042941 = k, 300 + 0.042941 316 + 0.042941 2 + 0.042941 _ $2 + 0.042941 _ _ $2 + 0.04294]. . SOFTENER(U2) All the by-product costs are assumed to be zero dollars. 1. Pump dP = 100 psi, Eff. = 85% 2. Regen. Flow(Wl) = 1% of Soft Product(W6) 09 . Salt Usage(W12) = 2lb/kgal/ 100 ppm hard, Salt Cost = $.03/lb 4. Hard Ion in City Feed(W9) = 200 ppm 0" . City Feed Cost(W9) = $2.00/kgal 122 From (2), (3), (4), and by the conservation of mass respectively, 311 = 0-01316 3112 = (2)(200PPm/100Ppm)319 = 4319 316 + 311 = U9 => 319 = 1.01y6 7" 3112 = 4(1.01y5) = 4.04y6 311 0.01 => y12 .= 4.04 316 dg ky6 319 1.01 $1 0 $6 = 1" $12 = = k' 003 $9 + 0.042941 2 + 0.042941 0 RESULT All the amounts of material flows and the related costs are in the luc.1pt report file. The Cost($/ yr) is computed as follows: Cost($/yr) = Flow Rate - Unit Cost - 365 Important measures are tabulated as follows: 123 Product Flow Rate(KGD) Unit Cost($/ Day / KGD) Cost($/ yr) Soft Water(W6) 500 2.1846 398,690 DI(W26) 300 8.0309 879,384 UPW(W17) 800 10.052 2,935,184 Total Cost 4,213,258 124 Part III CONCLUSION 12 Conclusion Good environmental performance measures are crucial to good environmental decision making. The measurability makes it possible to make the public be aware of environmental consequences, to set up the environmental goals to achieve, and to enforce such commitments. The environmental performance measure is the cornerstone of all environmental management systems(EMS). Because the environmental systems are likely to be very large, complex, multi-disciplined, and often conflicting multi-objective, there is no unique method on how LCA should be done. Those difficulties are mentioned in [19]. Especially in the phase of interpretation, the judgement may be political. Any quantitive environmental impact assessment can be used to measure the eco-efliciency in deciding whether a product, service, or process is greener than other alternatives. Greenness is a subjective term, however, identifying greenness may help to produce environmentally compatible products. The goal of ECM is waste minimization through pollution prevention. Two components of Environmentally Conscious Manufacturing(ECM) are design and analysis and design of manufacturing strategies. Design of ECM systems requiras quantitative tools to study the impact of alternative tech- nologies, schedulers, materials and designs used. Furthermore, to assist the environmental decision making, analysis, and 125 understanding of EMS, we need an effective modeling method to deal with large scale EMS while preserving the overall structure. This dissertation is an attempt to develop an ECM tool based on MEB theory[38, 39, 37], to design a simulation language, to present a computer pro- gram instantiating a DfE tool called Mass-Energy Based Simulation Tool(MEBST). Most quantitative DfE tools are concerned with environmental account- ing system without concerning of feasibility and impact of process network structure. This dissertation is an attempt to answer such questions and to present a computer modeling and simulation tool, Mass-Energy Based Sim- ulation(MEBS), instantiating an ECM tool(MEBST). The unique features of this aproach are: 1. The MEBST is logical, mathematical, and has an expandable struc- ture to model a system of various size and scale. Above all, processes are modeled based on physical parameters which does not change in terms of geographic location or different time, such as materials, en- ergy cost(i.e. land, labor, and energy). Based on the sound physical and mathematical modeling, the MEBST can objectively assess envi- ronmental, economical, technological, network performances. 2. It is comprehensive and thus can be used for the entire life cycle of the product. This is important because of conflicting requirements between different life stages. 3. The models are based on fundamental principle of material energ and 126 balance. 4. It is computer based and is easy to use. 5. It allows the user the capability to perform sensitivity analysis. This will help to evaluate the impact of less accurate data on the outcome. 6. It allows “what—if” simulation capability 7. It helps to evaluate the impact of changes in processes and/or tech- nolog'es(for example, the impact of automation or recycling). All of these measures can be used for process improvement and manage- ment as shown in Figure 33. Even if a small store managers do not use a simulation program, they are always drawing pictures in their minds, to maximize their profits using their best knowledge. This tool provides a graphical interface to evaluate these Options rather easily. One of the conclusions is that modeling is a formal representation of a system followed by simulation which assigns semantic meanings to its formal representation. Because the framework of MEB modeling partitions an overall system into a tractable amount of processes and MEB standardized modules, MEB theory is capable of succinct, crisp, and structural modeling of complicated, very large system. By allowing the library class, a system can be constructed 127 ASSESSMENT | Economic OF ECONOMIC PERFORMANCE [ml ALTERNATIVES I ICAL Economic mapping PERFORMANCE I (Environmental response) L- PHYSICAL PERFORMANCE Alternative structures 500106031 mapping Physical mapping __1 __. 1...... Resources We ENVIRONME‘JT PROCESS NETWORK Level Env. Load Alternative technologies Process TECHNOLOGIES & PROCESSES level Figure 33: Integrated approach to manufacturing system analysis and design 128 in multi—layered structures along with a GUI capable of zoom-in and zoom- out presentation. In the drawing course to partition a system, graphical interactions be- tween the user and the MEB simulation system play an important role, more crucial than any other phase of simulation for a system to be understandable. As is the one of a MEB simulation characteristics, the mix of tOp-down and bottom-up approaches makes modeling look a lot more like real world system because that is the basic nature involved in many design, analysis, and synthesis processes, though such characteristics might add more complexity to a system. The ease of drawing graphical objects embedding MEB theory determines smooth riding over the important phase of modeling with less pain. With the MEB GUI, from partitioning a model and to seeing the results are but a few clicks of mouse button away. The ease and the simplicity of interactions make it possible for a user to focus only on defining a system or process boundaries in this phase of partitioning a large system. Afier successful construction of the MEB graph, the semantics of the MEB graph are done through translating into MEBL and MEB Petri net. Each process solves its own problem using MEBL which can handle not only vector and matrix object expression but also interOperable database object succinctly with control flow statements. As a results, this MEB tool can be used as highly complex information management system. 129 12.1 Contributions For the realization of ECM, the proposed framework of MEB research and its methodology provides the following contributions: 1. Most importantly, the framework of MEB modeling and simulation tool which identifies related measures and processes to accomplish goals or specification of a system has been designed and implemented. With the pr0posed framework, measurement can range from a physi- cally detailed description of raw material flow to an empirical view of environmental impacts of each life cycle stage. The framework is also useful for on-line evaluation of process improve- ment and management(Figure 33). 2. In order to express MEB theory, a powerful language - MEBL - which can deal with concurrent processes with composite data models such as vectors, matrices, interoperable database, and control flows has been designed and implemented. 3. A formal representation of MEB execution environment which inher— ently contains concurrency has been defined by using MEB Petri net. 4. MEB graph grammar providing a hierarchically structured multi-layered system modeling tool has been designed and implemented. 5. Grammar which translates MEB graphs into MEBL to express formulas 130 and to compute wire variables and MEB Petri net has been defined and implemented. 6. In order to determine the sequence of process execution due to paral- lelism embedded in a system, an algorithm has been developed to select a process to execute in the MEB Petri net execution environment. 7. Assuming user’s minimal knowledge of MEBL, a GUI which painlessly guides the user through complex system modeling processes from draw- ing MEB graphs to viewing simulation results has been meticulously designed and implemented. 8. Reporting of MEB simulation results and 2-D data visualization have been implemented. 9. A variant of SQL which can query interOperable databases has been deve10ped. Those databases may contain not only the environmental burden by each byproduct but also environmental impact categories and associated weight to aid in computing coo-indicator value consid- ering the lack of unique measure of “how clean is green '3”. 12.2 Future Direction This research can be further improved in the following issues: l 1. The framework provides the necessary information regarding the waste flows as a function of technology. The environmental performance of 131 these waste flow is heavily dependent upon the measures of impact. More research needs to be done to incorporate this incomplete and sometimes conflicting information to determine the environmental im- pact of a given technology. . With all the measurements available after construction of an MEB modeling and simulation, the next question is what to do with all those evidences. Any reasoning, validating, scientific judgement is based on those evi- dences which may lead to modification of a mode], or different judge— ments. To make a judgement, possible decision categories need to be defined first. Then the decision problem would be assigning measurements to each of the categories. The next question is how each category is judged compared to other categories to quantify a global environmental burden. Combined with different weights for each categories, it seems to be next to impossible to find a unified formula to lead to a unique decision agreed upon by all the communities. Still, it would be nice having such a formula pleasing all the communities. . The database containing environmental load units needs to be filled with meaningful values agreed upon by environmental communities and scientists. 132 4. The framework does not provide any means of automatically selecting the Optimal strategy from many of the strategies available. A feedback mechanism needs to be incorporated for this purpose. It may take a shape Of an expert system. 5. The framework is implemented using Linux Operating system. Trans- forming this tO other Operating system platform would be helpful. 133 Part IV APPENDICES 134 LISTING The translation of the MEB graph Figure 9.1 into MEBL is as follows: wire { capa; cost; name; } netlist { W1: U2 -> J1; WW2: 01(W8) -> 03; WW3: Ul(W9) -> 03; W4: 01(W10) -> 05; W5: U1(W11) -> US; W6: 01(Wl2) -> 05; W7: 01(W13) -> 81; W8: 01(W14) -> 82; W9: J1 -> 03; W10: 03 -> U4; W11: US -> 06; W12: 06 -> 02; W13: 83 -> 02; W15: S4 -> 06; W16: 05 -> 85; WW1: J1 -> 01(W1); } init comment { } post elu { total_elu = O; b = select elu from elu where (elu#material tota1_elu = tota1_e1u + W7.capa*b; b = select elu from elu where (elutmaterial tota1_elu = tota1-e1u + W8.capa*b; b = select elu from elu where (elu#material total-elu = tota1_elu + W16.capa*b; title("x: W10 vs. y: tota1_elu"); plot(W10.capa, tota1_e1u); W7.name); W8.name); W16.name); } junction 31 { 135 shape J1; 1n W1; out W9; out WW1; k = [ 1. 1; 1; if (backward) { [ W1.capa; ] = k * [ W9.capa; WW1.capa; J; } else { [ W9.cost; WW1.cost; ] = k’ * [ W1.cost; 1; } } signal 81 { shape none; in W7; } signal 82 { shape none; in W8; } signal 83 { shape none; out W13; } signal S4 { shape none; out W15; } signal 85 { shape none; in W16; } class jj Ul { shape 01; out W4; out W5; out W6; out W7; 136 out W8; in WW1; out WW2; out WW3; } block 0 02 { shape 02; out W1; in W12; in W13; k = [1; 1; 1; if (backward) { [ W12.capa; W13.capa; ] = k * [ W1.capa; 1; } else { [ W1.cost; ] = k’ * [ W12.cost; W13.cost; 1; } } block 0 U3 { shape 03; out W10; in W9; in WW2; in WW3; k = [1; 1; 1; 1; if (backward) { [W9.capa; WW2.capa; WW3.capa; 1 = k * [W10.capa; ]; } else { [ W10.cost; ] = k’ * [ W9.cost; WW2.cost; WW3.cost; 1; } block 2 U4 { shape 04; in W10; } block 1 US { shape 05; 137 out W11; out W16; in W4; in W5; in W6; kcost = [ 1, 1, 1; 1, 1. 1; 1; k=[1,1,l;1,1,1;]; if (backward) { [ W11.capa; W16.capa; 1 k * [ W4.capa; W5.capa; W6.capa; J; [ W11.cost; W16.cost; J = kcost * [ W4.cost; W5.cost; W6.cost; J; } } block 3 06 { shape 06; in W11; out W12; in W15; if (backward) { W15.capa = W12.capa - W11.capa; alphal = W11.capa./W12.capa; alpha2 = ones(size(a1pha1)) - alphal; } else { W12.cost = a1pha1.*W11.cost + alpha2.*W15.cost; } } 138 Mass-Energy Based Simulation User’s Guide A INTRODUCTION The Mass—Energy Based Simulation(MEBS) tool is deve10ped to evaluate environmentally conscious product designs, management of manufacturing facilities to evaluate the strategies for reducing waste flows into the environ- ment, and life cycle analysis on the Linux platform. This tool allows to input the description of the main structure Of a plant using a drawing pallet. This pallet contains built-in drawing buttons in a graphic user interface(GUI) implemented on the X—Window environment with X11. The GUI relieves the user from having to know all the mathe- matical details of the models which describe each process within a plant and the interconnection constraints associated with the structure of the plant or a process. Besides having features to represent network information succinctly, MEBS also introduces the Mass—Energy Based Simulation Language(MEBL) which borrows many aspects from C language, MATLABz, and SQL database lan- guage. A source program is automatically created by the user with the GUI. Followings are case-by-case examples which will illustrate the details Of the program. 2MATLAB is a trademark of Math Works Inc. 139 B GETTING STARTED B.1 Convention denotes the command button with the name embedded between an- gled brackets and executed by clicking the left mouse button once. Similarly, , , . . . denotes a sequence Of command buttons. Usually the left mouse button is interpreted as selection operation and the right button as ESC key. The double click Of the left mouse button is interpreted as the RETURN key or equivalent to a click Of the middle button on a three button mouse. 3.2 Overview A network is described with several types of building blocks such as produc- tion, junction, library, and goal blocks, and wires which connect the blocks together. The forward connections are done by all the types of blocks except the recycle type block while backward feedback connection uses only the re- cycle type block as a subprocess. The special storage type block is used when the backward connection feeds to a forward connected process block to form a feedback connection. MEBS has a command name of meb. The detail command line Options are described in Section G. There are buttons in the GUI of meb for drawing a plant, modification of the subprocess description, simulation, and the report of simulation results. With the specific command meb, a large empty canvas shows up for draw- 140 ing subprocesses connected by directed lines or wires. The basic building blocks to describe a plant are rectangles of type (Production), (Recycle), (Storage), (Junction), (Library), and (Goods) for a subprocess which are interconnected by (Polyline) and (Wire). It should be noted that at least a block of type Goods be included in a network to specify the desired flow rate of final goods. (Text), (Comment), and (Label) buttons together are used to label the input and output of the subprocesses and later used for the report of simulation result. All other drawing primitives such as (Polygon), (Spline), and (Circle) are available. There are four kinds of files used to model a plant. The first one is the figure file which describes the network of the plant. The figure file is later used to provide the relationships of subprocesses and it contains all geometric information of the subprocesses on the canvas. The second file is the plant file which has all the information to describe a plant except the geometric information to draw on the canvas and it is created by the (Netlist) button. The suffix Of the file explicitly explains what kinds of files they are; i.e. a file name with the suffix “.fig” implies that the file contains figures, and the file name with the suffix “.pl” implies that the file is a type of plant file. The file name with the suffix “.rpt” may also be created to store the simulation results with the (Report) button. To address environmental impact assessment, an environmental index is assigned to each type of material used in a plant. And all of the indices used in a plant are stored into the default database named “elu.db”. A quantitive 141 measure of greeness of a plant is computed with the (ELU) button. (File), (Save) will save a figure file after drawing a graph while (Plant) , (Save) will save a plant file generated by (Netlist) button. The plant file contains MEBL to simulate a plant. After a network description is completed, we begin the simulation with the (Netlist) command button. This will generate a sample source program in MEBL in the framework of the MEB model. For the modification of the program in any subprocess, click the left mouse button on the subprocess which needs to be modified. There are four parameters which can be modified through GUI. They are Technical Coefl‘lcients, Capo, Cost, and By—product cost. Once a modeling by MEB graph is done, the next step is to simulate by specifying (Goa1>. The goal is Specified by selecting the output nodes associated with Goods and providing input costs. If the beginning of the terminal node is selected, meb will ask you to input the unit cost. When the goals and the material unit costs have been established, simulation is initiated with a press of the mouse middle button. The simulation results may be visualized with the (Get Value) button. This will plot a 2-D graph with the flow rate on the X-axis and the unit cost on the Y-axis. There is also a (Plot) button to obtain an x—y plot Of any two arbitrary data. Finally all the simulation results can be stored in filerpt by pressing the (Report) button. 142 B.3 Example As an illustration, consider a plant with three Production blocks and two Goods block as shown in Figure 34. 3.3.1 Drawing an Example Plant 0 Invoke simulation program meb. 1. meb 0 Use (Production> to create rectangular type production blocks . [ Production ] H 2. Select two points to determine the size and position Of rectangular production block with mouse. 3. Com Object 4. The above button is used to duplicate the rectangular boxes that represent production processes. 5. Place them at different positions on the canvas to create three more production blocks. 0 Create (Goods) type rectangular blocks 1[mePmd] 2. Similarly create two Goods blocks with mouse. 0 Connect the blocks with the (Wire) button. 143 ESQ 29:98 :< #m EsmE 144 mm mm «aha .whv 01...: 2.8 ace—.8 so: safe: .6: 363: «a: :38: do: .39.; we: oocovach #6810an 9:0... 040...: “Edam Sande...— aLQLo—dd cedvopll. Odd-wann- ouoLoum 39.... H5:.“- Odoaolu Evan-033.... OCqHadeL r Moan; _ 0:08:00 _ anon-1H5 snow can“. #230 such 30; gdom 2.5.2» new 25.3, How doom. :sIJz and“. 3.415.: 00.350: .254... 9.1: 3m .5 fee... 3.63. 3.6 1. r Wire lor Polyline 2. Interconnect blocks with mouse. To complete a line segment, click the middle mouse button or double click the left mouse button. 3. Complete the rest of interconnections between blocks following the same method. 0 Place junctions on lines if necessary. 1.[ Junction ] 2. Double click the left mouse button or click the middle mouse but- ton at the junction of lines. Note that there should be only one input line into a junction. This junction represents the points Of interaction between the processes. 0 Place comments on lines and blocks. 1. [ Text ] 2. Write description Of lines or blocks if necessary. 3. [ Comment ] 4. First, select a line to be associated with comments followed by a description with double click of the left mouse button or a click of the middle mouse button. a When the drawing is done, save the drawing usingl File N Save as ] and name the file as ex]. fig. 145 Figure 35 describes a procedure to draw the plant where the numbers be- side the boxes of the process represent the sequence of Operations in drawing the plant. The corresponding mouse Operation for each user input number is also shown at the bottom of Figure 35. 3.3.2 Simulation of an Example Plant 0 Generate a simulation program. 1. L Netlist ] 2. The result would look like Figure 34. e Provide information for the good. 1L Goal ] 2. Select blocks of Goal type to specify the desired flow rate of final goods. POp—up window will appear for input like Figure 36 3. Select beginning of lines to provide unit cost of materials which are fed into the production blocks. Similarly, a pop—up window will appear for input like Figure 37 4. Move the mouse on empty space and double click the left mouse button or click the middle mouse button. 5. If everything goes right, The “All Solved’ message will appear on the status window above the canvas. Otherwise, the blocks which are not solved will be displayed on the status window. 146 Eda a 32v 3 83805 < ”mm 95me .350 E _. mm 3. as. an ocean: +2. mé 9 .13.3.3.{Rainmaflfimflmqomf E 6mm a tee... a." _ t 1 mm mm in. l _ ............... m um mm mm QM 3. m o : m W3 1 an Bo: .movooch $11 :2. 68.. a e “9.5m: .2. R. 3 too: 2. 1.. ms" H“. «a , m 5a m a _fi _mm 8 “on 2 mm. 1 RN 3 _WNils “ ................ 9 mm a a ma 3 3 a me 8 0 $5 noon 88o: N... am t «at _e S _8 a at: _ fin .33.. _ l. a woe—.8. ER: 583.. :33: nun-.2 «on can...» a... 1356... S... ta- . 30cm. 051m _ 52.15 _ 536...... 1 oflfium _ 33a Jam]... 33036: 0.5. 1.236.. 5.8.3.3 one... _ 2.8 7:08 scorp— ao§ 633 2.15 new 6:15 new «new focus can?!) .lsS 3.6.3 3.5 :41... our. can»... nine and... 147 Enter caps for ”8 | [HB.capa - [1:10:1131 l lflone ltancel I Figure 36: POp—up capacity query window Enter unit cost 1‘ or III [H1.cost = 1*onos(size(H1.capa)) + eup(-0.2§3abs(fl1.capa))s] Done I Came]. Figure 37: POp—up cost query window Figure 38 describes a procedure to simulate the plant. e To change values of technical coefficients push [ Tech Coeff J and select a subprocess. Then a query will appear to facilitate the modification of technical coefficients. If we want to save modifications for later reference, save the results with the drag submenu button I Plant J I Save 1 The file will be stored as e31. pl 0 Examination of simulation results 1. met Value I 2. Select variables to visualize the simulation result. An example 2D plot is shown in Figure 39. 3. Or use [ Plot 1 button. 148 pas—Q 2: REESE 3 23680.3 < "mm. 55mg Ila. m . _I_ E e E . a a nuanced. a .5 E m _.:om8cd=vflitmoolvai .9 «Agogvflflovofli‘d I 608253 a a .5. Lou uueu vac: 509:“ . . 5.53»... Smear... new e.” . a FEES-.. m... 5:0 >91. mm . . :ocen ll Mn. . Q ..:........~....._: 3.. _. . m: a boo: 33.; 9 SE coo: LIA—v 5% A... V00: 3 n .3 _. . - . . no... a; ~||1| :on LI mu: 3.. H u E n . IIAIOIAII .o-olelooeI-eloo. NH. m... 3 n: ma 9 on 733:: I 2.3.93 .380 BI.- gun! a... .3.— Otao Log-fl weenie-561669359: mm mm . 1 . . . .. .vx...mo.z.wNM a 3?me e N w . Anna 6!: _ m 7.01139 3.30 _ coo—.8 36! 800.35 :62 «91.8 3 33:-0:. do: 208...? 3.1 n. . vi... CAP-du- ‘Ififinw 603.1..- — Saran-:1. _ DUFF-00m 0108 a? afloat—wot... 0.5.3.. guides action. _ .53 — 2.8 7:25 gua- 3M3 038 In.) sum .51: be... 13 815.303 .3 do .5 > 500 .3 301.101 “.de :bl! a 0.7- nomfluoz aunt—m 0.3...— . ................................................................................................................................................................................................................................................................................................................................... cost 2.74 ------------------------------------- P r'""'"""'""'"" ---L__-J_--J__- ---:- n 4L, 1 8 2 3 4 5 6 7 (D H ‘_-------------- go Figure 39: Capacity vs. cost 2—D plot 4. Report button will save all the simulation results in eerpt file. _ B.3.3 Defining Library A library convention is used to represent a big plant in the limited size of a GUI screen. Assuming that a plant description is done and saved as part2. fig and part2. pl, a library part is saved as lib.part2.fig and takes the following steps: 0 Change the GUI’s current mode into the library creation mode by se- lecting the[ Make Lib. Jsubmenu button in the] Mk_Lib Jmenu. 0 Define the shape of the library part and associate wire variablas with its names. 150 1. Capsule allows the user to draw the boundary of a library part for interaction with the rest of the plant. 2. Select two points to determine the size and position of the Capsule type rectangle block with mouse. 3. Draw smaller Production type rectangular inside the Capsule type block. 4. Draw wires between the Capsule and Production type blocks which are used to interact with the rest of a plant. 5. Associate the newly created wire with the same name used in par2. p1 with the I Label J button. 0 Define the incident points where interaction occurs between the library part and the rest of a plant. 1. The incident points are defined by the wire direction which is drawn above and the incident markers of I Inc_Goal Iand I Incidence I. 2. If the incident point is either an input to the library part or a byproduct of a library part, choose the I Incidence I button, and click the left mouse button at the incident point. Otherwise, choose the I Inc_Goal Ibutton and and click the left mouse but- ton at the incident point. 0 Save the resultant drawing with the name em lib.part2.fig by selecting the I Save As Isubmenu button in the r Mk_Lib Imenu. 151 0 Finish the defining library part by selecting the I Use Lib. Isubmenu button in theI Mk_Lib Imenu. 0 Using a library block is similar to using the normal building block. Select the Library button and select a position where the library is to be located. When asked to enter a library name, just enter part2 omitting the prefix lib- and the suffix. fig. Figure 40 describes a procedure to define the library part lib.part2. fig where the numbers beside the boxes of the process represent the sequence of operations in drawing the plant. C MEBL SYNTAX 0.1 Plant Structure ' The plant structure consists of two types of declarations and a series of the node blocks. There are five kinds of visible node blocks: junction node, signal node, block node, recycle node, storage node, and two kinds of special blocks: init node and post node. I If the plant structure is to be saved, it will be saved in the form of a plant file with the suffix of the file name ending with .pl. An example MEBL pro- gram corresponding to Figure 34 generated by (Netlist>, , is given in Section I. The first part of a plant description begins with the declaration of mem- bers of wire structure followed by the netlist keyword. The netlist dec- 152 tam ESQ: .m Big 3 9:689an Nov 2:th 353...... E:v~u.x¢mxfltz:fi~: .5038 Ann 9.me 6m $51»? 25! .5 Baas “N513 AA: 833?. N unfigigea Rum—3K»! u: 5.55 ea 3: 53 «has: 333 wit. a?» R: «m . Q: n 2,? «N mummy—.8 D —I J _ 1 _ _ ............ R _ a _ a: Ran—LIE Elia}: .TILI 53 a: _ _ 5.. u " ITIWO an _ fl «3 Il... v .935 an mm m _£. T— _H Tm _0 _o 3v ‘3 329 2% 3qu an 13.. > > 82.8 «on. 23...“; H :38: 2:. ugtdoeun Aloud-EH _ h _ 3.9.30 _ 2.;an — tennfiem . Tainan—d4 _ 601.05... _ Ogntlu 00.7.0.5 not; 3“. o.— 003 gusset—a. 0...:— _ 2:340: 421.. _ 23.200 _ unsung: _ v.39 _ 2.3 ocean. so: 50; imam nan-5 vem can; you 13 . 3m a: $.13. 3.62. 3.5 El! 32 , 3.3! £3.33. .23.. 3: 153 Block type Subclass No. Production 0 Recycle 1 Goods 2 3 4 Storage Libray Table 27: Block node subclass number used in MEBL laration part contains all the network information about how subprocesses are connected together. A junction node is also considered as a special subprocess which abides by special constraints. In the framework of MEBS, the sum of output flow rates are the same as the input flow rate and their unit costs are the same. A junction node can take only one input with arbitrary number of output. A signal node is created where either the beginning or the end of the wire does not go to any subprocess. There is nothing to compute in the signal node body except the initial condition provided in . A block node represents a subprocess and contains appr0priate state ments in its body. Accordingly, there are two block nodes corresponding to two subprocesses in this example. The init node is a special kind of block node which is done first once before execution of functions associated with each block. The subclass number of the block used in an MEBL statement as in “block subclass_number blockJmme { - - - }” is shown in Table 27. The order of which node is invoked first depends on the goal set-up in (Goal), solvability of the subprocess, and the network of a graph. First, 154 the simulation solves for flow rates noted by capa for every block by back prepagation from the goal. Then it solves for cost for each block by forward propagation from the input material costs which were given in the button. Finally, after all other solutions of processes are finished, the special post block - which might be executed as an option by (ELU) - is available to evaluate environmental impact assessment along with environmental index database as in Section 153.1. 0.2 N etlist Again considering the above example, the netlist body consists of the list of the wires with its source node to the left and its destination node to the right. For example, W1: S1 -> J1; implies that the wire H1 goes from the signal node 81 to junction node J 1. Note that -> has a different meaning from the meaning in the C language. C.3 Node Structure Considering that every subprocess has an associated building block of rect- angular shape in the graph, the shape declaration begins first. For the sake of readability, the rest of the declarations indicate which wires are coming in and which ones are going out of a subprocess. 155 After the declaration part, the statements of a node which determine the behavior or the function of a subprocess begin. It is worth mentioning that any wire variables can be accessed in any nodes. However, considering that the messages coming in and going out of a subprocess are highly correlated to the network of subprocesses, it seems to be a good practice to access only the wire variables which are in contact with the node. With this limitation of choice in the practice, it helps the program to be modular and structured. C .4 Wire Variables A wire variable is a composite object consisting of flow rate capa, energy cost cost, and a label name of wire itself. The wire variable name followed by period and one of capo, cost, name refers to simple data types of either real value or string. 0.5 Read Only Variables There are two reserved read-only global variables such as backward and forward. Simulation goes through the phases of backward and forward com- putation. Those Special variables show the direction of traverse during sim- ulation so that each subprocess can define its segment of a program to be computed. Users can read those values, but users are not allowed to set the values of those special variables. 156 0.6 Constants To accommodate the frequently used symbol 7r in trigonometry functions, a specific symbol PI is reserved for a constant. And the character constant is a character between single quotation marks while the string constant like "this is string" is a list of characters within double quotation marks. Both the character and the string constants are accepted as in C language. Some invisible characters are represented by escaping as follows: \n newline \t tab \f form feed \\ back slash For the numerical representation, both decimal and hexa-representation are accepted for an integer value. For example, the decimal number 10 is equal to the hexa-number Oxa. For the floating number representation, only decimal numbers are allowed as in the following examples: .1234 1.234 12.34E'5 123.48—5 There are two reserved variables, backward and forward. Those variables show the direction of traverse during simulation so that each subprocess can define a specific process in MEBL, depending on the state of traversal. D CONTROL FLOW The control flow is similar to that of the C language except that the ex- pression body between the control flow keywords to be selected or iterated 157 should be enclosed by { and }, even though the expression body contains only a statement. D.1 If Else This control flow keywords select a specific part of the body separated by if and else keywords for computation. Every expression body between if and else should be within { and } even when the body has only a line statement. D.2 For for (exprl; expr2; expr3) { statements; } The above for expression has all the control conditions which are Optional in one line followed by for. exprI is the initial condition before any other expression related to the for is considered for computation. If erpr2 is true, the main body within { and } is computed followed by computation of expr3 and erpr2 to make a full cycle again for the next iteration. Otherwise, the main body is skipped from computation. D.3 While The control flow do . . . while in C language is not available yet. while (exprl) { statements; I 158 The statements in the while body are computed repeatedly, as long as the exprI is true when tested before the execution of the while body. D.4 L00p Control break terminates the smallest enclosing 100p by for and while. continue returns the next computation immediately to the smallest enclosing while or for control statement. goto “identifier;” renders the next computation to be the statement after the label identifier. An identifier followed by : is considered to be a label or address in a process program. D.5 Comments A comment is a list of characters between /* and *l in a line and ignored from computation. E DATABASE MEBL has a statement similar to SQL for simple database manipulation. A database table can be managed by a normal text editor because it contains only the plain ASCII text. E.1 Database Structure Four keywords which explain database itself used in the database are: version requires only one argument telling a version number. delim describes the delimiter between fields. 159 field requires three arguments. The first argument is the field name and the second argument describes the type of field. There are only two types; c implies character type and f implies floating number type. The third argument is the maximum field size. Record delimiter is set to the new line character. end implies the end of the head information. An example of a database file is as follows: version 1 delim , field material c 20 field ec c 4 field elu f 10 and Co, RH, 76 Cr, RH, 8.8 Fe, RM, 0.09 Mn, RM, 0.97 No, RM, 1.583 002, EA, 0.09 CD, EA, 0.27 80x, EA, 0.10 CFC-11, EA, 300 CH4, EA, 1.0 Nitrogen, EH, 0.1 Phosphorus, EU, 0.3 E.2 Database Statement The syntax for the database statement is of the form: 160 select fieldJiame from database where query The database uses the name omitting the suffix .db from the database file name. To specify a field material in the database elu, elu#material is allowed in the query statement. F EXPRESSION F.1 Matrix The elements of the matrix are separated by either a comma or a semicolon. A semicolon is for the change to the next row of a matrix while a comma is for delimiting elements column-wise. For example, a = [ 1 , 2, 3; 4, 5 , 6] ; implies 2x3 matrix. Larger matrices can be generated by using variables as shown in the following example: a = [1, 2; 3, 4]; b = [5; 6]; c = [7, 8, 9]; A 3x3 matrix d is generated by using a, b, c as follows: d = [a, b; c] ; will construct d matrix to be three by three matrix resulting in 1 2 5 3 4 6 . 7 8 9 161 F.2 Vector Two or three elements are needed to represent a range of values as follows: [expr1: expr2] or [expr1: expr2: expr3]. The first element exprl is the value to start from and the second element expr2 is the final value of a vector. The third element determines the step size for the next element to generate. If the third element is missing, one is used for the default step size. For example [0:10:2] will generate a vector [0,2, 4, 6, 8, 10]. E3 Operators Operators and their precedences are shown in Table 28. FA Output Functions prval(a) Displays the value of a in the message window. plot(x, y) Draws x-y plot on a pop-up window. title(s) Sets the title message of a pop-up drawing window to 3. E5 Math Functions det (a) Determinant of the square matrix a. inv(a) Inverse of the square matrix a. 162 l+ 9’” Positive of a Negative of a Negation of a Transpose of a matrix a \ Multiplication of a and b Division of a by b Element-wise multiplication Element-wise division GG;.\*U O‘O‘ Sum of a and b Subtraction of a by b Greater than ”WNWWNNNNWNWNNWQLE; VI >= b Greater than or equal to < b Less than <= b Less than or equal to == b Equal to != b Not equal to u b a and b I l b a or b = b Assignment Table 28: Precedences of Operators 163 diag(a) If a is a matrix, diag(a) is the main diagonal vector. Or if a is a vector, diag(a) creates a square matrix with the diagonal elements the same as a and off-diagonal elements zeros. size(a) Returns the number of rows and the number of columns. eye(a) Returns an identity matrix with the same size of a. zeros(a) Returns a matrix of the same size of a with its elements zeros. ones(a) Returns a matrix of the same size of a with its elements ones. exp (a) Returns a matrix of the same size of a with its elements exponential of the elements of a. ln(a) Returns a matrix of the same size of a with its elements natural logarithms of the elements of a. log(a) Returns a matrix of the same size of a with its elements base ten logarithms of the elements of a. cos(a) Returns a matrix of the same size of a with its elements cosine of the elements of a. sin(a) Returns a matrix of the same size of a with its elements sine of the elements of a. tan(a) Returns a matrix of the same size of a with its elements tangent of the elements of a. 164 acos(a) Returns a matrix of the same size of a with its elements inverse cosine of the elements of a. asin(a) Returns a matrix of the same size of a with its elements inverse sine of the elements of a. atan(a) Returns a matrix of the same size of a with its elements inverse tangent of the elements of a. cosh(a) Returns a matrix of the same size of a with its elements hyperbolic cosine of the elements of a. sinh(a) Returns a matrix of the same size of a with its elements hyperbolic sine of the elements of a. tanh(a) Returns a matrix of the same size of a with its elements hyberbolic tangent of the elements of a. 165 G USER’S REFERENCE MANUAL NAME meb - Mass-Energy Based simulation tool SYNOPSIS meb [filefig I filepl] DESCRIPTION There are two kinds of files to model a plant. One of them is the figure file which describes the network of subprocesses in a plant. The figure file is later used to describe the relationship of subprocesses and only contains the geometric information of the subprocesses on the canvas. The other kind of file is the plant file which has all the information to describe a plant except the geometric information to draw on the canvas. The suffix of the file explicitly explains what kinds of files they are i.e. the suffix “.fig" implies a figure file and the suffix “.pl” implies the plant file. The file name with the suffix “.rpt” contains all the simulation results. If meb is invoked without a file name, a new canvas shows up with the default figure file name “untitled.fig” and with the default plant file name “untitled.pl”. For the graphic user interface, the left mouse button is usually inter- preted as selection operation and the right button as ESC key. The 166 double click of the left mouse button is interpreted as RETURN key or as equivalent to a click of the middle button. When the left but- ton is pressed, 3 rubber band appears on the canvas to show what a consequence would be. If that is a right choice, then the choice can be confirmed by the second press of the left button. AUTHOR Youngsun Chun email: chun©pilot.msu.edu 167 H MEBL GRAMMAR program wire netlist nodes wire wire { wirebody } netlist netlist { netbody } nodes nodes node node wirebody wirebody identifier ; identifier ; netbody netbody netelem netelem netelem identifier : identifier -> identifier ; node classhead { body } classhead init identifier post identifier block block.type identifier junction identifier signal identifier body declarations statements declarations declshape declportvar declshape shape identifier ; shape none ; 168 declportvar declportvar portvar portvar portvar in identifier ; out identifier ; statements statements astatopt astat expression if expression statements if expression statements else { statements } while statements for ( forexpr ) { statements } identifier : statements goto identifier ; continue break return forexpr expressionlopt ; erpression2opt ; erpression3opt expression mathfunction expression ) title ( string plot ( expression, expression ) prval ( expression ) term + expression - expression ! expression expression’ expression binOpl expression expression binop2 expression matrix = expression lval = expression lval = dbstmt dbstmt select identifier from identifier where expressionl mathfunction det inv exp diag size eye zeros ones In log 169 cos sin tan acos asin atan cosh sinh tanh term ( exar ) matrix range dbval lval identifier PI constant constant integer real string backward forward matrix [ matrest ] [ rows ] matrest matrest rows matrest rows ; rows ; rows rows comma expr4 expression binopl expression range [ expr ; expr ] I expression ; expression ; expression ] dbval identifier # identifier lval identifier . identifier identifier 170 ~1-_ r “'3’ ll—lelAAVV —Pll II II 171 I AN MEBL PROGRAM EXAMPLE wire { capa; cost; name; } netlist { H1: H2: H3: H4: H5: H6: H7: H8: H9' H10: H11: H12: H13: H14: H15: H16: H17: } 81 UI J1 02 03 J3 J2 U4 US U1 US U5 U2 U4 J1 J2 J3 init comment { H11.name H13.name } post elu { -> -> -> -> -> -> -> 01; J1; 03; J2; J3; U4; U5; U6; 07; > $2; > 83; > $4; > SS; -> 02; > U4; > 05; "Hasted paint"; "Hasted wood chip"; total_elu = O; b: select elu from elu where (elu#material = total_elu b = select elu from elu where (elu#materia1 = = tota1_elu + H10.capa*b; total_elu = total_elu + H11.capa*b; b: select elu from elu where (elu#material total_elu = total-elu + H12.capa*b; b = select elu from elu where (elu#material = total-e1u b a select elu from elu where (elu#material = total_elu + H13.capa*b; 172 H10.name); H11.name); - H12.name); H13.name); H14.name); total_elu = total_elu + H14.capa*b; title("x: H14 vs. y: total_elu"); plot(H14.capa, total_elu); junction J1 { shape J1; out H15; in H2; out H3; k = [ 1, 1: 1; if (backward) { [ H2.capa; ] = k t [ H15.capa; H3.capa; ]; } else { [ H15.cost; H3.cost; 1 = k’ * [ H2.cost; ]; } } junction J2 { shape J2; out H16; in H4; out H7; k = [ 1. 1; 1; if (backward) { [H4.capa; ] = k * [H16.capa; H7.capa; ]; } else { [ H16.cost; H7.cost; ] = k’ * [ H4.cost; ]; } } junction J3 { shape J3; out H17; in H5; out H6; k = [ 1, 1; 1; 173 if (backward) { ' [H5.capa; 1 = k * [H17.capa; H6.capa; l; } else { [ H17.cost; H6.cost; ] = k’ * [ H5.cost; J; l } signal 81 { shape none; out H1; } signal 82 { shape none; in H10; } signal 83 { shape none; in H11; } signal S4 { shape none; in H12; } signal 85 { shape none; in H13; } signal 86 { shape none; in H14; } block 0 01 { 174 shape 01; in H1; out H10; out H2; H10.cost = zeros(size(H2.capa)); k=[1;1;]; if (backward) { [ H1.capa; H10.capa; 1 = k * [ H2.capa; ]; } else { [ H2.cost; ] = k’ * [ H1.cost; H10.cost; ]; } } block 0 U2 { shape U2; out H13; in H15; out H4; H13.cost = zeros(size(H4.capa)); k = [1: 1; 1; if (backward) { [H13.capa; H15.capa; ] = k * [H4.capa; I; } else { [ H4.cost; ] = k’ * [ H13.cost; H15.cost; 1; } } block 0 U3 { shape 03; out H11; in H3; out H5; H11.cost = zeros(size(H5.capa)); k = [ 1; 1; 1; 175 if (backward) { [ H11.capa; H3.capa; ] = k * [ H5.capa; ]; } else { I H5.cost; ] = k’ * [ H11.cost; H3.cost; ]; } } block 0 U4 { shape U4; out H14; in H16; in H6; out H8; H14.cost = zeros(size(H8.capa)); k = [1; 1: 1; 1; if (backward) { [H14.capa; H16.capa; H6.capa; ] = k * [H8.capa; l; } else { [ H8.cost; ] = k’ * [ H14.cost; H16.cost; H6.cost; ]; l } block 0 U5 { shape 05; out H12; in H17; in H7; out H9; H12.cost = zeros(size(H9.capa)); k = [1; 1: 1; 1; if (backward) { [H12.capa; H17.capa; H7.capa; ] = k * [H9.capa; ]; } else { [ H9.cost; ] = k’ * [ H12.cost; H17.cost; H7.cost; ]; } } 176 block 2 U6 { shape I in } block 2 U7 { shape in } 06; H8; U7; H9; 177 J About the CD Rom Besides the DfE tool, meb, the enclosed CD Rom contains the system mod- elings described in the Section 11 and a hypothetical model to show how to measure the environmental impact of a manufacturing plant. A system model is composed of four parts as follows: Graphical description describes plant structure and shows how processes interact together in a plant. File name ends with the suffix .fig. Textual description which is automatically translated from above plant structure describes how to solve each process and integrate those in- dividual solution into various measures to assess a plant in various perspective view. File name ends with the suffix . pl. Environment description describes measures which are provided from out- side a system boundary. File name ends with the suffix .inp. Environmental Load Unit(ELU) database stores measures which are agreed upon by environmental scientists. These values define environ- mental impacts of various kinds of materials in various forms. Default file name of ELU database is elu. db. J .1 System Requirements At least physical or virtual display size of 800 by 600 pixels is required in X window configuration. 178 The first generation Of this DfE tool meb version 1.18 has been tested under following environments: 0 Linux Version 2.0.32 0 X Window System Version 11 J .2 Installation The DfE tool, meb, can be anywhere as long as the full path name of meb is reachable by $PATH shell environment variable. However, since meb needs to write on current working directory, system modeling files which are on CD Rom should be c0pied to hard disk which has writing permission on it. Assuming that the enclosed CD Rom is mounted under the directory /cdrom, following procedures will install the meb package under new directory meb18. $ / cdrom/ setup For those who are not ready for running Linux operating system and X window system, all the MEB modeling files and the simulation results are stored under /cdrom/meb18 directory ready to be viewed by any Operating system, although simulation cannot be done on other Operating system than Linux. Following is an illustration of the setup procedures. 179 Thank you for trying Mass Energy Based(MEB) DfE tool. Please let me know where to find your CD Rom(/cdrom)? Please let me know where to install this package(/u1/chun)? /tmp I am about to install the MEB package under the ltmp/meb18 from the /cdrom. Are you ready ([yles, [n]o)? y Installation completed. For an example of a HEB modeling try followings. 3 cd Itmp/meb18 $ meb farm.pl For more explanation of included MEB modeling examples, please read /tmp/meb18/README.p1 I hOpe you to enjoy the DfE tool meb. 180 References [1] A.A.Jensen, J .Elkington, and five others. Life cycle assessment(lca) / a guide to approaches, experiences, and information sources. Techni- cal report, dk—TEKNIK Energy & Environment, 15 Gladsaxe Mollevej, Soborg, Denmark, 1997. [2] A.R.Tilma, C.B.Tilma, and E.C.Aloci1ja. Process network theory(pnt) and analysis Of a swine/crop system. American Society of Agricultural Engineers Meeting Presentation, 90(7577), 1990. [3] RA. Baldwin. A Discipline Independent Framework for Engineering Design. PhD thesis, Michigan State University, 1994. [4] B.Kassahan, M.Saminathan, and J .Sekutowski. Green design tool. IEEE Symposium on Electronics 85 Environment, pages 118—125, 1995. [5] B.S.Blanchard and W.J.Fabrycky. Systems Engineering and Analysis. Prentice—Hall, 1981. [6] B.Steen and S.Ryding. The epa envirO-accounting method: An applica- tion Of environmental accounting principle for evaluation and valuation of environmental impact in product design. Technical report, Stockholm: Swedish Environmental Research Institute(IVL), 1992. [7] C.Hall and J.Day. Ecosystem Modeling in Theory and Practice. Wiley Interscience, 1977. 181 [8] Chen. An approach for material selection that integrates mechanical design and life cycle environmental burdens. IEEE Symposium on Elec- tronics 8 Environment, pages 68-74, 1995. [9] D.F.Ciambrone. Environmental Life Cycle Analysis. Lewis, 1997 . [10] D.J.Richards. The Industrial Green Game. National Academy Press, 1997. [11] D.P.SmOlik. Material Requirements of Manufacturing. Van Nostrand Reinhold CO., 1983. [12] E.C.Pielou. Population and Community Ecology: Principles and Meth- ods. Gordon and Breach, 1974. [13] E.Henley and H.Kumamoto. Designing for Reliability and Safety Con- trol Prentice-Hall, 1985. [14] J. Fiksel. Design for Environment. McGraw-Hill, 1996. [15] M. GoedkOOp. The Eco-indicator95, Final report, 1995. (URL:http://www.pre.nl/eco—ind.html). [16] G. Gordon. System Simulation. Prentice-Hall, 1978. [17] T.E. Graedel and RR Allenby. Industrial Ecology. Prentice-Hall, 1995. [18] T.E. Graedel and ER. Allenby. Design for Environment. Prentice-Hall, 1996. 182 [19] Y.Y. Haimes. Modelin of large scale systems in a hierarchical- multiobjective framework. North-Holland, 1961. [20] H.Khalil. Nonlinear System. Prentice-Hall, 1995. [21] H.S.Matthews and L.B.Lave. Price setting for green design, 1998. [22] J.Berge, O.Levia, and J.Reuillard. Models in System Design. Kluwer Academic Publishers, 1997. [23] J.E.Hopcroft and J.D.Ullman. Introduction to Automata Theory, Lan- guages, and Computation. Addison Wesley, 1979. [24] J.E.M.Klostermann and A.'Ihkker. Product Innovation and Eco- efi‘iciency. Kluwer Academic Publishers, 1998. [25] W.D. Kelton, R.P.Sadowski, and DA. Sadowski. Simulation Language with Arena. McGraw—Hill, 1998. [26] BE. Koenig. Enterprise Models for the Design and Management of Manufacturing Systems. PhD thesis, Michigan State University, 1992. [27] L.Barfield. The User Interface Concepts and Design. Addison Wesley, 1993. [28] L.Weinberg. The development of a streamlined, environmental life cy- cle analysis matrix for facilities. IEEE Symposium on Electronics 8! Environment, pages 65-70, 1998. 183 [29] [30] [311 [32] [33] [34] ~ [35] N.M.Darnall, G.I.Nehman, J.W.Priest, and J.Sarkis. A review Of envi- ronmentally conscious manufacturing theory and practices. Int’l Jour- nal of Environmentally Conscious Design 6'! Manufacturing, 3(2):49—57, 1994. Society Of Environmental Toxicology and Chemistry. A technical frame- work for life-cycle assessments, January 1991. WorkshOp Report, Wash- ington, D.C.: Society Of Environmental Toxicology and Chemistry, and SETAC Foundation for Environmental Education, Inc.,. J .L. Peterson. Petri Net Theory and the Modeling of Systems. Prentice- Hall, 1981. JR. Poyner and M. Simon. Integration Of dfe tools with product de- velopment. Technical report, Dept. Of Mechanical Eng., Manchester MetrOpolitan University, 1995. DFE/TR25. S.B.Billatos and N.A.Basaly. Green Technology and Design for the En- vironment. Taylors&Francis, 1997. RA. Sheldon. Organic synthesis-past, present, and future. Chemistry and Industry, (23), 1992. A. Sweatman and M. Simon. Integrating dfe tools into the design pro- cess. Technical report, Dept. of Mechanical Eng., Manchester MetrOpoli- tan University, 1996. DFE/TR30. 184 [35] [37] [38] [39] [40] A. Sweatman and M. Simon. Dfe/tr34—products Of sus- tainable future. Technical report, Dept. Of Mechan- ical Eng, Manchester MetrOpolitan University, 1997. (URLzhttp: / / sunl .mpce.stu.mmu.ac.uk / pages / projects / dfe/ dfe/ .html). R.L. Tummala and L.J. Connor. Mass-energy based economic models. IEEE Trans. Sysst., Man, Cybem., 3(6):548—555, 1973. R.L. Tummala and BE. Koenig. Models for life cycle assessment Of manufactured products. IEEE Int’l Symposium on Electronics and the Environment, pages 548—555, 1994. R.L. Tummala, B.E. Koenig, and HE. Koenig. Process network the- ory and implementation for technology assessment in manufacturing. Proceedings of a Joint German/ US Conference, Hagen, Germany, 1992. SH. Weissman and J.C.Sekutowski. Environmentally conscious manu- facturing: A technology for the nineties. Technical report, AT&T Tech- nical Journal, 1997. 185