Optimization of Availability and Cost for the Heating System in an Oil Refinery through Semi-Markov Process and Simulated Annealing

摘要

In an oil refinery, the heating system availability of distillation columns plays an important role to determine the system's performance. This article aims to develop an optimal preventive maintenance policy that optimizes the heating system availability and maintenance costs in an oil refinery based on continuous time non-homogeneous semi-Markov processes (CTNHSMP). CTNHSMPs are used since a heating system has components that upon a failure condition might either be completely renewed or have intensity functions dependent on the process (global) time. Moreover, some of these components exhibit non-exponential times to failure. Simulated Annealing are adopted for addressing the optimization problem. The proposed approach is exemplified by means of an application in an oil refinery.