455577 A Hybrid Framework for Process Synthesis-Design: Application to Biorefineries

Tuesday, November 15, 2016: 3:15 PM
Monterey II (Hotel Nikko San Francisco)
Maria-Ona Bertran1, Ana-Sofia Sanchez-Arcilla1, John M. Woodley2 and Rafiqul Gani2, (1)Department of Chemical and Biochemical Engineering, Technical University of Denmark (DTU), Kongens Lyngby, Denmark, (2)Department of Chemical and Biochemical Engineering, Technical University of Denmark, DK-2800 Lyngby, Denmark

As energy demand and emissions increase steadily, the need for more sustainable processes is becoming evident. To this end, innovation in synthesis and design of processing routes needs to take into account raw materials availability, existence of appropriate processing technologies and products that have demand and use. When more than one alternative (technology or process interval) exists for one or more processing steps, a processing network results from which the optimal processing route needs to be determined.

The synthesis and design of processing networks for any chemical process in general and biorefineries in particular is an industry-relevant problem. Mathematical programming techniques, where the model equations representing all alternatives present in the network are included as constraints within an optimization problem, are commonly applied to solve this type of synthesis-design problems. The determination of the optimal processing route from a network (superstructure) of alternatives therefore involves decision-making through continuous variables (such as mass flow-rates and separation factors) and discrete variables (such as selection of an alternative in a processing step). The size of the problem depends on the number of alternatives included within the network, while, the complexity of the problem depends on the type of the models and decision variables used to represent the process intervals. Within the context of a biorefinery, considering a network involving a maximum of NS processing steps, NI process intervals (alternative) organized in processing steps, NF different biomass (raw materials), NP different products, and, allowing for processing step bypass and material flow recycle, would result in a mixed integer nonlinear programming problem. The size of the problem depends on the values of NS, NI, NF, NP and NC (number of chemicals). Note also that the models for each process interval also include model parameters that would need to be specified before any problem can be solved.

The objective of a hybrid framework for biorefinery synthesis-design is to provide the means to study the feasibility of establishing biorefineries as a function of biomass availability, location and characteristics, product types (fuels and/or chemicals), available technologies (use of different types of conversion technologies) and many more. The requirements for such a framework are: availability of appropriate numerical solvers for MILP and MINLP problems; a tool to generate the network model equations and to formulate the problem specific mathematical programming problems; and a database of information-knowledge-data, such as, different types and sources of biomass, different types of products, different types of known conversion technologies, processing routes and prices. This hybrid framework is based on a previous work (Quaglia et al. 2012) and has been extended in order to accommodate a wider range of process synthesis-design problems that can be formulated and solved. Key additions to the framework are a data management system that allows the generation of the superstructure (network) based on the biorefinery case study under investigation and retrieval of location-dependent data. A data structure has been designed and implemented in a knowledge base, which contains all necessary information for a range of problem formulations and their solution. The knowledge base currently contains 11 types of biomass, from 10 different locations, 102 intervals representing processing technologies and 9 products. It is a dynamic knowledge base that is ever-growing as more data are being collected and more problems are being solved. Moreover, once data is stored in it, superstructures are generated systematically including only the relevant alternatives for a specific problem definition with respect to a specified location. In addition, a user-friendly interface has been developed covering several of the steps (Bertran et al. 2016).

The main features and application of the hybrid framework are illustrated through the formulation and solution of a multi-feedstock multi-product problem. Six biomass-based raw materials are considered, yielding various products, in eight geographical locations; this leads to a superstructure with 9 processing steps (NS) and 35 processing intervals (NI). The solution is a (set of) raw material(s), a (set of) products, a location and the process configuration. Furthermore, several scenarios, such as minimum waste, maximum profit and different product demands, derived from the main problem have been investigated and the location dependency highlighted. A second case study involving different biomass to produce liquid fuel products is also briefly highlighted, which involves 2 feedstocks (NF), 2 products (NP), 13 processing steps (NS) and 34 processing intervals (NI), which leads to 79118 equations. Note that although the hybrid framework was originally designed for chemical flowsheet synthesis-design and extended for biorefineries, it is generic and has also proven useful for other processes such as carbon dioxide utilization and waste-water treatment networks.


Bertran M., Frauzem R, Zhang L, Gani R. A generic methodology for superstructure optimization of different processing networks. In: Kravanja Z, editor. Proc 26th Eur Symp Comput Aided Process Eng – ESCAPE 26. Portorož, Slovenia: Elsevier B.V.; 2016.

Quaglia A, Sarup B, Sin G, Gani R. Integrated business and engineering framework for synthesis and design of enterprise-wide processing networks. Computers & Chemical Engineering, 2012, 38:213–23.

Extended Abstract: File Not Uploaded
See more of this Session: Process Design II
See more of this Group/Topical: Computing and Systems Technology Division