(171e) Pattern Recognition in Chemical Process Flowsheets

Process design is a nontrivial and complex task that has motivated the development of many systematic search and optimization techniques over the past four decades. As a result, a practical designed flowsheet will follow certain design patterns. In this work, we are interested in the development of a methodology for discovering patterns in existing chemical process flowsheets. There are two primary motivations for this work. First, a systematic comparison of a large number of successful chemical processes may help us identify design rules and operational characteristics that contribute to the success of these processes. Second, comparing the flowsheet of a potentially new process against flowsheets of existing processes may help assess the costs associated with the detailed design of the new process. In particular, potential new designs with a high degree of similarity with existing processes are likely to be easier to analyze as well as less expensive to build. A tool capable of performing such comparisons can be used to screen potential new projects and optimize the selection of product portfolios.

There is no prior literature on tools and methodologies for mining process flowsheets. To close this gap, we have developed a systematic methodology for discovering process patterns in existing chemical process flowsheets. The proposed methodology proceeds in three major steps to: (1) generate graph representations to denote general process flowsheets; (2) convert process flowsheets to string notations [1]; and (3) identify process patterns with sequence alignment algorithms [2, 3]. This three-step methodology has polynomial complexity at each step and can be applied automatically to compare general process flowsheets. We demonstrate the effectiveness of the proposed method through several case studies. By comparing flowsheets that produce the same chemical product, we assess the similarity of process flowsheets and discover structure features that contribute to successful processes. While comparing flowsheets that produce different chemical products, we identified many general structural patterns that reflect engineering practice or heuristic rules in literature [4].

References cited

[1] Tula, A. K., M. R. Eden, and R. Gani. Process synthesis, design and analysis using a process-group contribution method. Computers & Chemical Engineering 81 (2015): 245-259.

[2] Smith, T. F., and M. S. Waterman. Identification of common molecular subsequences. Journal of molecular biology 147.1 (1981): 195-197.

[3] Needleman, S. B., and C. D. Wunsch. A general method applicable to the search for similarities in the amino acid sequence of two proteins. Journal of molecular biology 48.3 (1970): 443-453.

[4] Biegler, L. T., I. E. Grossmann, and A. W. Westerberg. Systematic methods for chemical process design. Prentice Hall (1997).