GREEN DESIGN, GREEN ENERGY, AND SUSTAINABILITY: A SYSTEMS ANALYSIS PERSPECTIVE

1. GREEN DESIGN, GREEN ENERGY, AND SUSTAINABILITY: A SYSTEMS ANALYSIS PERSPECTIVE Urmila Diwekar Center for Uncertain Systems: Tools for Optimization and Management (CUSTOM) Vishwamitra Research Institute Clarendon Hills, IL 60514 urmila@vri-custom.org

2. Economic Objectives Environmental Control Cost Reduction Process Simulation Products Raw Materials By-Products Chemical Synthesis in Lab. Process Integration Options Process Synthesis Traditional Process Design Steps

3. Operability Controllability Performance Indices Management and Planning Environmental Impacts Health Impacts Cost Reduction Profitability Simulation Discovery Improved Performance Material/ Chemical Selection Plant/ Network Synthesis Greener by Design Risk, Reliability Safety Products Integration and Control

4. 1 5 2 2 3 6 4 4 Vitrify Waste 5 1 Glass Formed Add Frit 6 3 7 7 8 8 Vitrify Waste Synthesizing Optimal Waste Blend Glass Formed

5. Conversion of Waste to Glass Waste 1 Vitrified Blend Blend-1 Waste 2 Glass Logs Frit Waste 3 Vitrified Blend Blend-2 Waste 4 Objectives: To select the combination of blends so that the total amount frit used is minimized For 21 tank wastes and 3 blends: 66,512,160 possible combinations

6. Recursive Loops Decision Variables Optimal Design Optimizer Output CDFs Uncertain Variables Stochastic Modeler Optimization Loop Stochastic Modeler Sampling Loop Sampling Loop • Model Model Stochastic Modeling Stochastic Optimization

7. Important Properties of Sampling Techniques • Independence / Randomness • Uniformity In most applications, the actual relationship between successive points in a sample has no physical significance, hence, randomness of the sample for approximating a uniform distribution is not critical (Knuth, 1973). Once it is apparent that the uniformity properties are critical to the design of sampling techniques, constrained or stratified sampling becomes appealing (Morgan and Henrion, 1990).

8. A: Monte Carlo B: Latin Hypercube • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 0.0 0.4 0.8 0.0 0.4 0.8 x x C: Median Latin Hypercube D: Hammersley Sequence • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 0.0 0.4 0.8 0.0 0.4 0.8 x x Wozniakowski-Hammersley

9. HSS LHHS HSS2 New Sampling Technique Hammersley Sequence Sampling (HSS) based on a Quasi-random number generator HSS sampling is at least 3 to 100 times faster than LHS or MCS. HSS is preferred sampling for stochastic modeling and/or stochastic optimization.

10. Operability Controllability Performance Indices Management and Planning Environmental Impacts Health Impacts Cost Reduction Profitability Simulation Discovery Improved Performance Material/ Chemical Selection Plant/ Network Synthesis Greener by Design Risk, Reliability Safety Products Integration and Control

11. Single and Multi-objective Optimization Problems Multi-Objective e.g. Min Cost, Max Safety, Min Emissions, Max Flexibility Min/Max Z=(Z1,Z2,..,Zp) Zi = fi(x) Subject to: h(x) = 0 g(x) < 0 x - Decision variables Single Objective - e.g. Min Cost Min/Max Z Z = f(x) Subject to: h(x) = 0 g(x) < 0 x - Decision variables

12. A Simple Multi-objective Linear Example Max Z1 = 6x1 + x2 Z2 = - x1 + 3x2 subject to: 3x1 + 2x2  12 3x1 + 6x2  24 x1  3 x1, x2  0

13. 2 + 1 + Weighting Method: Max.Y=1*Z1+2*Z2 Objective Space 14 E Max. Z1= 6x1+x2 12 Max. Z2=-x1+3x2 10 D 8 6 Z2 4 2 Pareto set C 0 A 0 5 10 15 20 25 -2 -4 B Z1

14. 100000 10000 Conventional Method New Method 1000 Number of Optimization Problems solved 100 10 1 2 3 4 5 Number of objective functions Figure 18: Compare the new algorithm and the existing constraint method with different numbers of objective functions Comparison of the New MINSOOP Algorithm with the Conventional Method

15. Pareto Optimal Solutions MOP • Defining Optimization Problems • Optimal Solutions Discrete Optimizer • Discrete decisions • Feasible Solutions Continuous Optimizer • Probabilistic objective & constraints • Continuous decisions Sampling Model Multi-objective Optimization under Uncertainty

16. Natural Gas Stack Gases 5.61 kg- CO2/kWh Steam Solid Oxide Fuel Cell PEM Fuel Cell Heat Recovery Steam Generator Fuel Pre Reformer Selective Catalytic Oxidizer Low Temp Shifter Water Air O2 Air Water SOFC-PEM Hybrid Power Plant Rating: 1472 kW Efficiency: 72.6% Capital Cost: 1773 $/kW Cost of Electricity: 6.35 c/kWh 176°F, 25 psi 190 mA/cm2 1750°F, 20 psi 75 mA/cm2 70% fuel utilization

17. Uncertainties and Trade-off Surface Stochastic trade-off surface Uncertainties in PEM and SOFC models Deterministic trade-off surface

18. Eco -friendly Management Industrial Symbiosis ` Industrial Ecology Socio-Economic Policies Ecology Sustainability Clean Products Green Energy Clean Processes

19. Conceptual Framework for Industrial Ecology Spatial Global National Sector Region Firm Community Division Industrial Plant Unit Operation Sustainability Quality of Life Environmental Quality Environmental Impact Energy Reduction Material Reduction Profitability CostProduction Throughput Eco-Efficiency Thermal Efficiency Local & National Economic Data Qualitative Firm, Plant Production Data Scale of Application Mass and Energy Balances Process Simulation and Optimization Quantitative Uncertainty Analysis Thermodynamic Constraints Physical Constraints Criteria for Evaluation Sources of Information Tools for Analysis

20. Eco -friendly Management Industrial Symbiosis ` Industrial Ecology Socio-Economic Policies Ecology Sustainability Clean Products Green Energy Clean Processes

21. Sustainability “Development that meets the needs of the present without compromising the ability of the future generations to meet their own need” -- The world commission on environment and development

22. Emissions to air Deposition in water bodies Bioaccumulation in aquatic biota The Mercury Cycle Mercury Research Strategy (USEPA) Consumption of mercury is highly dangerous to humans • Long term exposure effects: • Permanent damage to- • Brain • Kidney • Developing fetus • Short term exposure high levels exposure effects: • Lung damage • Diarrhea • Blood pressure increase Mercury management at various points important for successful control of mercury pollution • Consumption of contaminated fish is the biggest source of human exposure to mercury • Fish consumption advisories at many lakes and rivers in United States

23. Savannah River Basin Water Shed: Mercury Problems Olin Corporation Augusta PCS Nitrogen Fertilizer PCS Nitrogen Fertilizer Peridot Chemicals Pooler/Bloomingdale Richmond Co Spirit Cr. Rincon Sardis WPCP Savannah Elec Effingham Savannah Elec Riverside Savannah Elec Wentworth Savannah Electric & Power Savannah President St Savannah Sugar Refinery Savannah Travis Field Savannah Wilshire/Windsor Savannah Yacht Club Solutia Inc South Carolina Electric Southern Aggregates Columbia Southern States Phosphorous & Fert Springfield Stone Container Corp Sylvania Yarns Systems Inc WQ-IP-047 Thermal Ceramics Inc Thiel Kaolin Hobbs Tybee Island Union Camp Corporation USA Ft. Gordon USA Hunter AFB STP Waynesboro WPCP Wrens WPCP 2004 Mercury-Related Fish Advisories206 Georgia issued 178 fish consumption advisories – relating to 40 different rivers and 34 lakes and ponds. Near the Olin plant, in the Savannah River Basin, there were 24 advisories, affecting five rivers and seven lakes and ponds.

24. Trading Optimization with Two Objectives Additional constraint calculates the actual pollutant discharge reduction

25. Stochastic Programming Formulation Uncertainty affects the constraint: • Possible sources of uncertainty: • Discharge quality/quantity of the point sources • Reduction capabilities of the technologies • Mercury fate, transportation and concentrations • For the trading problem, uncertain discharge results in uncertain target reduction redi • Different levels of constraint satisfaction reflects different levels of protection against uncertain information

26. Industrial Symbiosis via Trading Cumulative technology selections for complete TMDL range • Stochastic model: • Change in the distribution of technologies • Higher implementations of most efficient technology

27. Multi-objective Optimization

28. Eco -friendly Management Industrial Symbiosis ` Industrial Ecology Socio-Economic Policies Ecology Sustainability Clean Products Green Energy Clean Processes

29. Controlling Ecological Impact • PH Control by Liming • Reduce methyl mercury formation • Food-web, Predator pray models • Reduce bioaccumulation

30. External lime input to water Water compartment Variable: Lime in water Active sediment compartment Variable: Lime in active sediment Lake pH: Using algebraic and logical relation Passive sediment compartment Variable: Lime in passive sediment Lake Liming Model • Based on the model developed by Ottoson and Håkanson (1997) • Model inputs: Lake physical characteristics, lake chemistry and lime input to water (e.g. sedimentation rate, distribution coefficient etc.) Lake pH: Using sigmoidal curve and a linear approximation

31. Stochastic Lake Liming Model • Various sources of uncertainty: • Time independent: Lake physical attributes (mean depth, area) • Time dependent: Natural seasonal variation in lake pH • Natural seasonal pH variation: • Approximate yearly variation given in Ottoson and Håkanson (1997) • Possible to use stochastic processes for modeling • Stochastic processes used to model: • Stock prices, interest rates • Relative volatility in distillation • Human mortality rate • Use of Ito mean reverting process to model natural pH variation

32. Stochastic Model Comparison

33. Algorithmic Framework

34. Lake Liming Results

35. Mercury Bioaccumulation • Factors affecting • bioaccumulation: • Mercury chemistry and abundance • Species-specific effects • Geochemical influences (water salinity, temperature) • Food uptake • Manipulation of food-chain regimes to affect the food intake thereby affecting mercury bioaccumulation • Use of control theory and information theory to derive regime manipulation strategies

36. Sustainable Systems Hypothesis • Sustainability is a multi-disciplinary concept • Any type of data or model can be converted to some kind of information irrespective of their disciplinary origin • Necessary condition: for the system dynamic regime to be sustainable,the fisher information must be constant Reference: Fath and Cabezas, 2000

37. Time (days) Optimal Control Problem: Results Control variable: Nutrient flow rate D Controlled system shows lower mercury bioaccumulation

38. Summary • Decision making in process design and industrial ecology is a multi-objective problem with uncertainties • A Multi-objective Framework • provides trade-offs among objectives • provides environmentally friendly and economical designs • Uncertainties can change decisions and designs significantly • Sustainability is a multi-disciplinary concept • Ecological systems involve time dependent uncertainties and time dependent decision making • Financial literature and optimal control theory provides a systematic decision making

39. Contributors Yan Fu, Ford Motor Company Venkatesh Narayan, Avery Dennison Mark Hoza, Pacific Northwest Laboratories Yogendra Shastri, UIUC Jayant Kalagnanam, IBM T. J. Watson Lab. • Financial Support: • National Science Foundation • Pacific Northwest Laboratory • National Energy Technology Laboratory