(186b) Development of a Multi-Objective Optimization-Tool for Simulation-Based Chemical Process Synthesis and Design Tasks

For the synthesis and design of chemical processes numerous methods have been developed. Generally speaking, they can be divided into two main approaches. The first and traditional approach follows the so-called “onion logic”. In this approach first the reactor followed by the separation system etc. are designed. Due to this procedure, every decision is based on the information available at the particular stage of the process design and therefore on an incomplete picture [1]. As a consequence, a series of best local decisions are made [1]. These processes satisfy the chemical and technical requirements very well, but no guarantee on global optimality with respect to for instance its cost can be given. Moreover, the ability to look ahead to the completed design might have led to different decisions [1]. The second approach overcomes this disadvantage by creating a so-called superstructure. Superstructures contain every feasible variant of the process structure. In a simulation-based design approach, the process is modeled in one of the various flowsheet-simulators available on the market. Subsequently, the design problem is formulated as a mathematical optimization problem. A set of objective functions, e.g. operating and investment costs, as well as constraints, e.g. the chemical purity of the product streams, are chosen. In addition, decision variables are defined. In most cases, they are operative or constructive parameters of the process. The completed optimization problem is then solved by a solution algorithm. Yet the suitability of different solution algorithms depends highly on the optimization problem. In order to provide a universally applicable method to process optimization, an innovative tool, the Advanced Process Optimizer (AdvPO), was developed and intensively tested at the Institute of Process and Plant Engineering.

In this presentation the AdvPO as well as selected optimization methods will be introduced. The AdvPO is capable of performing single- or multi-objective structural and parameter (SOSP-/MOSP-) optimizations automatically. As a result, constructive and operative parameters are optimized as well as the process structure itself. This was realized by a bidirectional interface between the AdvPO and various flowsheet simulators, for example Aspen Plus. Due to the modular concept of the AdvPO basically every process can be optimized. The solution algorithm is a specially tailored evolutionary algorithm. The basic structure and functionality of the AdvPO will is part of the presentation. In case of a MOSP-optimization the obtained Pareto-optimal solutions, visualized as the Pareto-front, represent the best trade-offs between the multiple and often opposed objectives. As the optimization problem is highly complex, different innovative concepts and methods had to be developed and extensively tested. One example for such a method is the Two-Step-Optimization which will be part of the presentation as well. The Two-Step-Optimization combines SOSP and MOSP approaches to enlarge the obtained Pareto-Front. As a result, the decision maker can choose his personal optimal design on the basis of a complete picture of the best trade-offs. The exemplary optimization problem deals with a dimethyl ether process. The objective functions are the investment and operation costs of the process. The superstructure is composed of three different reactor types and two different separation systems. The optimization results will be compared with reference solutions achieved by conventional approaches [2]. The results show a significant reduction of both the investment and operating costs compared to the reference solutions.

References

[1] R. Smith, Chemical process design and integration, Wiley, Chichester, West Sussex, England, Hoboken, NJ 2005.

[2] W.L. Luyben, Principles and case studies of simultaneous design, Wiley, Hoboken, N.J. 2011.