1. SUSTOIL: Work Package 5 Modelling, Simulation, Multi-Objective Optimisation and Life Cycle Analysis of Integrated Biorefining Schemes Michael Binns and Constantinos Theodoropoulos School of Chemical Engineering and Analytical Science University of Manchester, Manchester M60 1QD

2. SUSTOIL WPs and Interconnections

3. Tasks and deliverables Tasks –To construct computationally efficient biorefinery simulators including both reaction and separation units in order to investigate the viability of schemes, which will be proposed by WPs1-3. –To perform optimisation studies on the biorefinery schemes proposed in order to better compare them and to assess their economic and environmental viability. –To combine the modelling and optimisation studies with Life Cycle Analysis information based on appropriate databases, by constructing multi-objective optimisation schemes –To assess upstream and operation options based on directions obtained through multiobjective optimisation Deliverables –D5.1: Construction of biorefinery process simulators with explicit parametric dependence –D5.2: Model validation for biorefinery schemes proposed in WPs 1-4 (month 12) –D5.3: Optimisation of biorefining options with respect to economic and environmental objectives –D5.4: Construction of multi-objective optimisation schemes taking in account LCA information –D5.5: Delivery of holistic process options assessments (month 20) –D5.6: Report to be fed to WP4 for preparation of discussion document for workshop on WP5- 6 D5.7: Final report on social, Environmental and Economic cost-benefits analysis of developed biorefinery schemes

4. Contents 1.Modelling in Aspen Plus 2.Optimisation Methodology 3.Life Cycle Analysis Data 4.Biodiesel Scheme 5.Biogas Scheme 6.Super Critical CO 2 Extraction Scheme 7.Oil Extraction Schemes 8.Protein Extraction Scheme 9.Thermomoulding scheme 10. Levulinic Acid Scheme

5. Modelling and Simulation Tools Aspen Plus Simulation Software Linked to libraries of chemical properties Good for modelling many common units (reactors, separators etc.) Computes mass balances, energy balances and duties for each unit Can be used to construct and simulate large refineries/plants Aspen Custom Modeller Can be used to model uncommon or Novel units Links simulation software with programmable code (anything is possible) Units involving Solids Units using complex kinetics Uncommon Units

6. Graphical User interface Data Entry Forms Graphical Display

7. Simulated Annealing based Process Optimisation Evaluate objective (Including Aspen Simulations SA move accepted? No Base Case Conditions SA parameter moves Evaluate objective (Including Aspen Simulations) Revert to previous state Converged? Yes No Yes Stop Flexible Treats Aspen models as black boxes Can converge close to global optimum Moves are random (based on probability)

8. Objective Function This objective function works in the following way Run Aspen using System call Read input parameters Enter Parameters into input file using scripting Read simulation results from report file Use results to evaluate Objective function Objective Function (minimisation) Objective = -VP + CR + EC + CP + WC + EF VP = Value of Products CR = Cost of Raw Materials EC = Energy Costs CP = Capital costs (scaled to €/hr based on 10 years lifespan) WC = Cost of liquid and solid waste disposal EF = Error function (=1 x 10 9 if an error in the simulation occurs otherwise = 0)

9. Multi-Objective Optimisation We have based our multi-objective optimisation on the methods of using “non-inferior” or Pareto curves a a. Azapagic A., Clift R., (1999), Computers and Chemical Engineering, 23, 1509-1526. Profits Environmental Impact In this project the competing priorities are -Profits -Environmental impact

10. Life Cycle Analysis Data Electricity data 0.0772 € / kWh a 0.537 b a.FERA, 2007 b.DEFRA, Market transformation programme, BNXS01 c.GEMIS database, http://www.oeko.de/service/gemis/en/http://www.oeko.de/service/gemis/en/ d.Turton et al. (2009), Analysis, synthesis and design of chemical processes, Pearson Education Inc Heating Data 0.0179 € / kWh c 0.201 c Cost (a typical value) Emissions Factor ( kgCO 2 / kWh) Waste 0.0248 € / kg d

11. Scheme 1. Biodiesel Production Inputs Rapeseed3,125 kg/hr Methanol216.29 kg/hr Potassium Hydroxide15 kg/hr Outputs Biodiesel1,237 kg/hr Rapeseed meal1,875 kg/hr Crude Glycerol93.56 kg/hr In this simulation the crude glycerol is 80% glycerol (mass fraction) 99% conversion 70% methanol recovery

12. Scheme 1. Biodiesel Optimisation Varying the methanol recovery With fixed methanol feed at 7 kmol/hr ProfitsEmissions

13. Scheme 1.1 Bioconversion of Glycerol to Succinic Acid Batch Fermenter Inputs Crude Glycerol93.56 kg/hr Water3,971 kg/hr Biomass0.492 kg/hr Outputs Succinic Acid41.1 kg/hr

14. Scheme 1.1 Bioconversion of Glycerol to Succinic Acid Dilution of the mixture seems to increase the production rate Possibly due to inhibition effects ProfitsEmissions

15. Scheme 1. Comparison of different Biodiesel Schemes ParameterWith co-production of Succinic acid With co-production of crude glycerol With co-production of Purified glycerol Methanol Recovery (%) 84.07 86.94 Methanol Feed (ton/ton of feed) 0.0531 0.0519 Profit ( € /ton of feed) 138.3385.7791.73 Emissions (tons CO 2 /ton of feed) 0.1520.1510.155 Waste (tons/ton of feed) 0.0421 0.0452 Optimal Oil : Methanol ratio ~ 1 : 13

16. Scheme 1. Multi-objective optimisation With Crude GlycerolWith Succinc acid

17. Scheme 2. Biogas Production Inputs Residue1,350 kg/hr Water1,450 kg/hr Air kg/hr Outputs Digestate1,424 kg/hr Digestate(dry matter)198 kg/hr CO 2 2,324 kg/hr CH 4 trace This simulation includes a recycle rate of 50% for the digestate Simulation based on information provided by FORTH 99% conversion 90% conversion Energy Outputs 3,199 kW 3,656 kW 2,285 kW

18. Scheme 2. Biogas Optimisation Optimisation performed using the nonlinear SQP method in Matlab (Have also tested Simulated annealing for this method) Recycle bound to the range 0 to 99% Economic Optimum Recycle is 97.49% ( 379.78 € per hour ) Profits Emissions

19. Scheme 2. Biogas Multi-Objective Optimisation Performing optimisation using different constraints gives a range of solutions (a pareto curve) This graph show how the optimum economics change as the limits on CO2 emissions are reduced

20. Scheme 3. Super Critical CO 2 Extraction Inputs Residue (Straw) 100 kg/hr CO 2 20 kg/hr Outputs Spent straws97 kg/hr Waxes3 kg/hr CO 2 20 kg/hr Process constructed based on information from UoY Example capacity 100 kg/hr (876 tons/year)

21. Scheme 3. CO 2 Extraction Capacity based on information from UoY 1. Brunner G., Journal of Food Engineering, 67, 21-33 (2005). Log (cost per ton) = -0.76894 x Log (capacity) + 13.807 A linear fit can be found as demonstrated in the literature 1

22. Scheme 3. CO 2 Extraction optimisation based on information from UoY Optimal capacity = 36,227 tons per year (profit = 488.40 € per ton) Profits Emissions

23. Scheme 3. CO 2 Multi-objective optimisation based on information from UoY Limiting total emissionsLimiting emissions per ton of feed

24. Scheme 3. CO 2 Multi-objective optimisation based on information from UoY If multi-objective optimisation is based on tons of emissions per ton of feed Minimum emissions are achieved at the highest capacity

25. Scheme 4. Protein Extraction from Rapeseed Meal Process information from WUR Protein Extraction Rapeseed Meal Fibres (41%) 3,003.7 tons LMW (24%) 1,758.2 tons Protein S (15%) 1,098.9 tons Protein P (20%) 1,465.2 tons Electricity required: 475 kWh / ton Heating required: 3,431 kWh / ton Costs: 547.7 €/ton Profits: 176.7 €/ton 7,326 tons

26. Scheme 5. Oil Extraction Process information from CREOL, CETIOM 3,125 kg/hr EfficiencyOil extractedSeed meal 83.3%1,042 kg/hr1,823 kg/hr 95%1,188 kg/hr1,641 kg/hr Method Cold Pressing Hexane-based Input: Rapeseed Assuming 40% oil content

27. Process information from CREOL, CETIOM and Desmet Ballestra 28.92 €/ton Comparison of Oil Extraction methods 33.19 €/ton Cold Pressing Hexane-based Extraction So the larger Hexane based plant has cheaper costs per ton Costs Energy Emissions Electricity: 68 kWh / tonElectricity required: 38.2 kWh / ton Natural gas:240 kWh / ton 36.5 kg/ton of seeds68.8 kg/ton of seeds However the hexane based method also emits more CO 2

28. Scheme 6. Thermomoulding Process information from INPT Injection molding Sunflower meal ExtrusionWater Sodium Sulfite Plant pots Extrusion produces the intermediate product TEGS Total cost 0.104 €/potCapacity: 69.7 tons/year Energy: 8,363 kWh/ton Giving 120 pots/hr Assumed value of pots = 0.12 €/pot (Biodegradable)

29. Scheme 7. Levulinic Acid Information provided by Biorefinery.de via WUR Formic acid Lignin and Humins Levulinic acid 0.45 kg cellulose  0.23 kg Levulinic Acid 1 mol Glucose  1 mol Levulinic Acid + 1 mol formic acid Cellulose content of straws = 45%

30. Scheme 7. Levulinic Acid Economics 1 ton Rapeseed Straw ~ 1.73 tons HCL(20%) 230 kilograms Levulinic Acid (25 €/ton) (46 €/ton) Information provided by Biorefinery.de via WUR 91 kilograms Formic Acid (12,000 €/ton) (1,150 €/ton) Costs = 104.58 €/ton Revenue = 2864.65 €/ton Based on these figures, a very promising biorefinery option

31. Economic and Environmental Comparison Process Optimal profits ( € per ton of feed) Emissons (kg CO2 per ton of feed) Scale of plant (tons per year) Cold pressing oil extraction 33.030.036511,000 Hexane based oil extraction 52.280.06881,000,000 Biodiesel with crude glycerol 85.770.15125,000 Biodiesel with purified glycerol 91.730.15525,000 Biodiesel with succinic acid 138.330.15225,000 Protein extraction 201.450.9457,326 Biogas production 281.321.82110,854 Supercritical CO 2 extraction 488.400.98336,227 Thermomoulding 1,413.204.76269.70 Levulinic acid production 2,758.95*

32. Conclusions 1) Optimisation has been performed linking aspen software with MatLab For Biodiesel an optimum oil:methanol ratio was found to be ~ 1:13 Very high recycling of digestate gave the optimum profits from Biogas A capacity of 36,227 tons per year was shown to be the most profitable 2) Multi-objective optimisation shows how profits and operating conditions would be affected by emissions limits. 3) Comparison of biorefinery schemes were performed Bioconversion of glycerol to succinc acid is a promising option for biodiesel producers Thermomoulding and levulinic acid show potential Hexane-based oil extraction is more profitable but emits twice as much CO2 Biogas, supercritical CO 2 extraction and protein extraction are profitable but emit around one ton of CO 2 per ton of feed

33. Thank you!