True degradation estimation of industrial equipment with fuzzy sets: A case study

Abstract

Equipment is prone to degradation during usage, and this is a continuous process. Degradation can be assessed with the help of condition indicators whose levels (near threshold values) indicate the need for maintenance action. A common phenomenon observed in process plants is that, with highly varying process and/or stressor variables, the levels of condition indicators also fluctuate a great deal and may approach threshold values suddenly. Responding to such levels of condition indicators by way of maintenance actions leads to overmaintenance, as these do not represent permanent equipment deterioration, but a combination of temporary and permanent deteriorations caused by changes in the process and/or stressor levels. Thus, there is a need to estimate the permanent deterioration, i.e. true degradation, of the equipment under these situations, excluding the temporary deterioration. The level of true degradation determines the appropriate maintenance actions to be carried out. In this paper, the authors propose a new approach to modelling the degradation process by segregating it into manifested degradation and true degradation. The first component leads to temporary equipment deterioration, and the second to permanent equipment deterioration. Estimation of the true degradation is carried out using fuzzy sets and a fuzzy inference system on the observed condition indicators and process information. Quantitative data and imprecise and vague knowledge are used to fuse into the fuzzy inference process while estimating the true degradation level. A case study on steel rolling mill equipment is considered to demonstrate the applicability of the proposed approach. Additionally, a customized equipment health acceptability criterion is developed for the given equipment using the condition monitoring database.