Methods and Algorithms for Process Synthesis and Intensification using Building Blocks

Abstract

Process intensification is a design concept with transformative prospects for the chemical industries. It refers to any design activities that substantially improves one (or more) of several performance metrics, including size, energy efficiency, environmental footprint, and safety. Intensification can be achieved by combination of physicochemical phenomena or extreme integration in a single equipment. Examples of process intensification include dividing-wall columns, reactive distillation columns, membrane reactors, and more. However, process intensification with phenomena combination may not be as good as separate phenomena interactions. Examples include nonideal reactive mixtures with minimum driving force such as methyl tert-butyl ether (MTBE) production process. On the other hand, the traditional unit-operation-based paradigm for process synthesis and integration does not always have the mechanisms required for systematic identification of novel designs. The objective of this Ph.D. work is to address three critical questions: (1) when process intensification is desirable, (2) how to represent chemical processes for automatic generation and screening of intensification pathways without knowing their existence beforehand and (3) how to effectively solve the optimization problems for the representation involving intensification opportunities.

In this dissertation, optimality conditions are derived for both nonintensified and intensified systems. It is found that the superiority of intensification is subject to system parameter interactions. Furthermore, a new building block-based representation approach for processes intensification is presented to foster creativity at the conceptual design stage. The building block superstructure is modeled using a mixed-integer nonlinear (MINLP) optimization formulation to design flowsheets that reduce total annualized cost or increase total annual profit. The building block representation can be used as a general platform for systematic process synthesis, integration and intensification. We apply the proposed approach on large-scale process synthesis, simultaneous process synthesis and intensification, and general process integration. While the building block approach facilitates process design, this representation involves symmetry leading to multiple representations for a single chemical process. Therefore, a novel branching rule based on the formulation group of optimization problem is proposed which is general to symmetric optimization problems. The performance of this branching rule is illustrated through computational experiments which demonstrates its ability to break symmetry and improve convergence.