E ective GDP optimization models for modular process synthesis

Abstract

In this paper, we propose an optimization-based strategy to systematically evaluate tradeo s associated with modular alternatives for the multi-period design of a chemical processing network. We give a general formulation as a Generalized Disjunctive Program (GDP) and discuss a linearizing reformulation that exploits structure common to modular design problems. By modeling the GDP in the Pyomo algebraic modeling language, we gain access to a exible set of automatic reformulations and solution algorithms, from which the best tool may be selected to optimize a given model. We apply the design strategy to a set of illustrative case studies, including capacity expansion, bioethanol processing, and heat exchange network design. The results show that the proposed design strategy is able to solve modular design problems and provide general insights on tradeo s between investment and transportation costs in which incorporation of modular facility constructions may or may not prove to be advantageous. Introduction The past few years bear witness to growing interest in modular chemical plants, which o er improved exibility, quality, and schedule e ciency characteristics over traditional plant 1 constructions. In a modular plant, major pieces of processing equipment are assembled as standardized modules rather than custom-designed and constructed on-site. A module includes the processing equipment and associated control instruments, piping, valves, and interconnection points mounted in a structural steel framework. Each module forms a selfcontained processing unit, which can be built and tested in a controlled environment at the manufacturer s workshop. In this way, modular design can be seen as a move towards greater standardization in the chemical process industry. Modular design is not a new idea in manufacturing or chemical processing, but its combination with the ideas of distributed manufacturing and process intensi cation comprises a compelling solution to modern chemical process industry challenges. However, modular plant design may not be appropriate in all places and situations. Therefore, our main challenge is the assessment of modular design as a partial or complete replacement for existing processes and the exploration of new applications that are enabled by modular design. Mathematical programming can be a powerful tool for this assessment. Through optimizationbased process synthesis, we can systematically evaluate trade-o s between the advantages and disadvantages of modular process design. The rst step of an e ective analysis is development of the appropriate physical and cost models. Several authors have investigated modular design as a central element of Europe s Industry 4.0 initiative, with exibility and time to market identi ed as key bene ts of modular designs. 11 Lier and Grünewald provide a net present value (NPV) analysis of modular versus conventional constructions, showing that for rapid market growth scenarios, modular plants can outperform conventional designs due to their exibility and responsiveness. The European literature also explores the value of exibility in providing extra value in the presence of uncertainty. Other authors have examined modular design in conjunction with speci c applications. Modular chemical facilities appear well-suited to address the recent issue of stranded natural gas processing. The distributed nature of gas sources in some regions makes constructing pipelines to a traditional centralized processing facility eco-