Automatic generation of interlock designs using genetic algorithms. (9th June 2017)

• Record Type: Journal Article
• Authors: Lepar, Yeremia Yehuda; Wang, Yu-Chih; Chang, Chuei-Tin
• Is Part Of: Computers & chemical engineering. Volume 101(2017)
• Page Start: 167
• Page End: 192
• Publication Date: 2017-06-09
• Subjects: Corrective maintenance -- Preventive maintenance -- Interlock -- Cold-standby -- Spare -- Superstructure -- Genetic algorithm; Chemical engineering -- Data processing -- Periodicals; 660.0285
• DOI: 10.1016/j.compchemeng.2017.02.042
• Languages: English
• ISSNs: 0098-1354

Abstract

Highlights: A generic GA-based MATLAB code is developed for generating structural and maintenance specifications. An interlock superstructure can be constructed to facilitate systematic search for the optimal configurations. Novel and comprehensive models of the corrective and preventive maintenance programs can be incorporated. This optimization code is adopted to achieve simplicity, portability and maintainability in practical applications. Four examples are provided to demonstrate the aforementioned benefits. Abstract: The hazardous units in a modern chemical plant are often equipped with safety interlocks to mitigate the detrimental effects of potential accidents. To achieve a desired level of reliability (or availability), not only the interlock configuration but also the maintenance policies of its components must be properly stipulated before actual installation. Although a number of deterministic programming models have already been developed for this purpose, their laborious formulation and solution steps are usually carried out on a case-by-case basis. To attain the essential qualities of conciseness, portability and maintainability for easy implementation, there is a definite need to develop a generic and modularized code according to an evolutionary algorithm. The structural and maintenance specifications of an interlock can be represented with: (1) the number of measurement channels and the corresponding alarm logic, (2) the numbers of online and spare sensors in each channel and the corresponding voting gate, (3) the number of shutdown channels and the corresponding tripping configuration, (4) the numbers of online and standby actuators in each channel, and the corresponding activation mechanism, and (5) the inspection intervals of shutdown channels. All of them are encoded in this study with binary numbers for use as inputs to implement the genetic algorithm (GA). An interlock superstructure is also developed to facilitate the search for the best configuration and maintenance plan in any application. By minimizing the overall life-cycle expenditure, a generic MATLAB code has been developed for generating all aforementioned specifications in any application. Four examples are provided to demonstrate the benefits of the proposed optimization approach.