Title: Modelling composite performance variable of deteriorating systems using empirical evidence and artificial neural network

Authors: P.A. Ozor; S.O. Onyegegbu; J.C. Agunwamba

Addresses: Department of Mechanical Engineering, Faculty of Engineering, University of Nigeria Nsukka, Enugu State, Nigeria; Department of Quality and Operations Management, Faculty of Engineering and the Built Environment, University of Johannesburg, Johannesburg, South Africa ' Department of Mechanical Engineering, Faculty of Engineering, University of Nigeria Nsukka, Enugu State, Nigeria ' Department of Civil Engineering, Faculty of Engineering, University of Nigeria Nsukka, Enugu State, Nigeria

Abstract: The use of operational and environmental conditions combined with Artificial Neural Networks (ANNs) to model the composite performance of deteriorating repairable systems is presented. The proposed variable is obtained by combination of reliability, availability, maintainability and profitability (RAMP). Probability distributions and empirical evidence observed on an example system, namely centrifugal pumps at the gas plant of an energy company, were relied upon to model the operation process. The results show that the input variables, preventive maintenance, spare parts availability, efficiency of operating personnel and efficiency of maintenance personnel, with cumulative performance enhancement of 56.1%, 39.97%, 30.8% and 30.6%, respectively, improve RAMP appreciably. The results also show that proper assessment and control of the input variables, administrative delays, repair period, service crew strength and mostly environmental factors with cumulative performance enhancement of 23.6%, 19.4%, 17.3% and −14.62%, respectively, had significant potential for improving RAMP further.

Keywords: deteriorating repairable system; empirical evidence; probability distribution; composite performance variable; artificial neural network; maintenance policies.

DOI: 10.1504/IJRS.2017.088546

International Journal of Reliability and Safety, 2017 Vol.11 No.1/2, pp.23 - 49

Received: 13 Oct 2016
Accepted: 08 Sep 2017

Published online: 11 Dec 2017 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article