Location-dependent Synthesis of Biorefinery Networks Mariona Bertran, John M. Woodley and Rafiqul Gani Department of Chemical and Biochemical Engineering Technical University of Denmark (DTU) DK-2800 Lyngby Denmark
The Current Global Situation 6-7 x Global GDP growth over next ~50 years (in constant dollars) 5-6 x Production capacity for most commodities (steel, chemicals, lumber, etc.) 3.5 x Energy demand 7 x Electricity demand Increase Water demand Increase GHG emissions Siirola (2012) Proc 11th Symp PSE (Ed Karimi and Srinivasan) 1
The Design Challenge PROCESS Biomass CO2 ... New process synthesis-design problems arise from: (i) Switch to renewable raw materials (biomass, CO2) (ii) Discovery of new technologies (catalysts, solvents, bioprocesses) (iii) New design objectives and constraints (sustainability)
Design Problem Formulation The decision-making nature of the process design problem makes it an optimization problem Problems: LP, NLP, MILP, MINLP, Simulation… Solution strategies: simultaneous, decomposition-based
Process Synthesis Needs Identify alternatives Mathematical model Solution strategy Database Need: Superstructure representation Problem Need: Generic process model Need: Integration with an optimization environment Solution
1. Superstructure Representation The Processing Step-Interval Network (PSIN) representation is suitable for a wide range of problems Generic processing interval mixing reaction separations Quaglia et al. Comput Chem Eng, 2014 Bertran et al. Comput Chem Eng, 2017
2. Generic Process Model A generic process model can represent multiple process options at various scales Bertran et al (2016) Computer Aided Chemical Engineering
3. Optimization Problem Objective function Process interval model A generic process model can represent multiple process options at various scales Objective function Process interval model Composition, availability and demand constraints Superstructure connections Location dependent / Location independent
Superstructure of alternatives Data Management Databases are used to collect existing process data to make it readily available Superstructure of alternatives Model Solution strategy Database Problem Need: Knowledge management The bottleneck is data management! Solution Bertran et al. Computers and Chemical Engineering (submitted).
Data Management with Databases Process steps Technologies Mixing data Reaction data Waste data Separation data Added Reference Ratio Reaction Key reactant Conversion Compound Fraction Recovery Inlet material stream Outlet material stream Utilities data Utility … Step Interval Components Properties Reaction sets Reactions MW BP ... Utilities Cp Hvap Reaction data Stoichiometry Catalyst … Locations Countries Name Code Feedstock data Availability Composition Price Products Location Product data Demand Specs Feedstocks Utilities Properties Cp Hvap ... Data Biorefinery Database Components 71 Utilities 4 Processing steps 21 Processing intervals 102 Feedstocks 11 Products 9 Reactions 63 Locations 10 Bertran et al. Computers Chemical Eng (Submitted)
Super-O Bertran et al. Computers and Chemical Engineering (submitted).
Superstructure of alternatives Super-O Superstructure of alternatives Model Solution strategy Database Problem Super-O Solution
Conceptual Examples
Application Problems Which biomass-derived feedstocks can be used? Where are they available? What are the different routes to convert the feedstocks to the product? What are the processing technologies available? Is the solution location-dependent? Which set of feedstock-topology-location is optimal?
Biomass to Chemicals
Synthesis Constrained to a Single Location* *synthesis problem solved for different locations
New (more flexible) Model Output information: Optimal processing route (steps & technologies); Flowrates; Capacities of technologies; Environmental impacts, LCA indicators; Location of each section; Economics (revenue, capital costs, operating costs, waste handling costs, transport costs, …) Input information: Network data (steps, intervals, connections); Processing data (performance of alternatives); Supply/demand data (availability, demand, market price); Location data (distances, transport prices) Allows to investigate many more scenarios
Example: Biomass to Ethanol Bertran et al. Computers and Chemical Engineering (submitted).
Adding Transportation
Distributed Production Transportation
Revisit: Biomass to Ethanol
1. No Transport Cost Profit 91.19 M$/y 152 kt/y ethanol Raw material 700 kt/h cassava rhizome Pretreatment Process Product Profit 91.19 M$/y Bertran et al., AIChE Annual Meeting, 2017
2. Transport Product: Process based in Asia 602 kt/y cassava rhizome 98 kt/y sugarcane bagasse 63 kt/y ethanol Raw material 100 kt/y ethanol Pretreatment Process Product Profit 30.90 M$/y Bertran et al., AIChE Annual Meeting, 2017
3. Transport Product: Process Based in N. America 700 kt/y wheat straw 152 kt/y ethanol Raw material Pretreatment Process Product Profit 84.90 M$/y Bertran et al., AIChE Annual Meeting, 2017
4. Transport Intermediate: Process based in N. America 700 kt/y wheat straw 135 kt/y ethanol Raw material Pretreatment Process Product Profit 30.90 M$/y Bertran et al., AIChE Annual Meeting, 2017
Overview of Problems / Applications Synthesis of a new process Selection of potential products Supply-chain management Distributed production Process retrofitting Plant allocation … Biorefinery CO2 utilization Chemical processes Water management Pharma processes
Concluding Remarks A framework for biorefinery process synthesis using superstructure optimization has been developed. The associated methods and tools are: superstructure representation, generic process model, data management system. A software implementation of the framework is available (Super-O). The framework has been exemplified in a series of applications. Options for transportation between locations to be developed further. We are interested in collaboration, to build the database and refine information.