Int J Performability Eng ›› 2023, Vol. 19 ›› Issue (4): 283-290.doi: 10.23940/ijpe.23.04.p7.283290

Previous Articles    

Assessment of the Effectiveness of Maintenance Actions and the Influence of Covariates on the Reliability of Gas Turbines using the Extended Generalized Proportional Intensity Model

Sidali Bacha* and Ahmed Bellaouar   

  1. Transport Engineering and Environment Laboratory, Frères Mentouri Constantine 1 University, Constantine, Algeria
  • Contact: *E-mail address: sidali.bacha@umc.edu.dz

Abstract: Imperfect repair (IR) models are strongly recommended in reliability modeling because of the limitations presented by homogeneous (HPP) and non-homogeneous Poisson processes (NHPP). Among the imperfect repair (IR) models, the generalized proportional intensity model (GPIM) highlights the effect of preventive (PM) and corrective (CM) actions of maintenance and certain covariates on the future performance of complex repairable systems (CRS). In this work, the GPIM with log-linear and power laws was adapted for the reliability and maintenance of two gas turbines that had operated for nearly 6 years on one of the production divisions of SONATRACH (Algeria, Hassi Messaud). The objective was to evaluate the effect of preventive (PM) and corrective (CM) maintenance actions on the reliability of the system and to show the influence of the covariates representing the number of shutdowns, the time elapsed since the last maintenance action (TSLMA), gas leakage, and vibration failures on the future gas turbine performance. A comparison study between the basic generalized proportional intensity model (basic GPIM), extended GPIM, intensity reduction model (IRM), and non-homogeneous Poisson process (NHHP) made it possible to propose GPIM as the best-fitting model. This judgment was based on the maximum likelihood approach using the MATLAB programming language to estimate the parameters of the models and perform a likelihood ratio test (LR).

Key words: imperfect repair (IR) models, generalized proportional intensity model (GPIM), effect of preventive (PM) and corrective (CM) actions of maintenance, covariates, complex repairable systems (CRS), maximum likelihood approach