Predictive Maintenance and Deterministic Maintenance Scheduling for Continuous Process Industry Rotary Equipments

Abstract

Abstract This paper explores and analyzes the methods for predictive maintenance and deterministic maintenance scheduling for continuous process industry rotary equipments. This paper consists of a brief introduction of various types of predictive maintenance techniques, maintenance schedules and the methodology to develop the optimization models. The paper also considers a turbine as a case example for an equipment type, analysis of results using the optimization model and conclusion from the analysis and future scope related to the optimization model. In the overall end-end flow, this paper outlines the use of analytical modeling techniques to first predict for example, the likelihood of a component of an equipment failing, the recommendations that can be generated based on the predictive model scores, and then use an optimization model to generate the maintenance schedule. The predictive model and the optimization model are modular and can also be used together or in isolation to provide different set of capabilities and advantages to the end customer. Binary integer programming model is used to generate the maintenance schedule considering the deterministic part life, pulling back of events, non maintenance period of equipment and minimum number of outage. There are mainly two decisions in the optimization model which are occurrence of event for each part and outage for the equipment. Few of the business constraints are as part life for repair and replace, non maintenance period due to business requirement. The objective of the optimization model is to minimize total maintenance cost over time horizon. The generated maintenance schedule using the optimization model reduces the wastage of part life and number of outage for equipment. This also increases the utilization of the equipment. This also reduces the downtime of equipment due to unscheduled and unnecessary maintenance. Over and above this minimize the total maintenance cost of the equipment. This model can be extended to stochastic model considering stochastic failure time of parts life. Net present value can also be included to calculate the maintenance cost for long term. Production reduction is a major issue due to unscheduled maintenance. This optimization model gives an unique idea to generate the maintenance schedule for very complex machine such as turbine which has many sections, each section has many part-kit, each part-kit consist of many parts with various failure time for each type of event in a very efficient and effective way. This idea can be used for any industry which has very large and complex machine to get the maintenance schedule. Dragline in opencast mine, turbine, manufacturing industry, production industry are the suitable place to use this type of maintenance schedule.