Analyzing maintenance strategies by agent-based simulations: A feasibility study

Authored by M. Kaegi, R. Mock, W. Kroeger

Date Published: 2009-09

DOI: 10.1016/j.ress.2009.02.002

Sponsors: No sponsors listed

Platforms: AnyLogic

Model Documentation: Other Narrative Flow charts

Model Code URLs: Model code not found

Abstract

Thoroughly planned and implemented maintenance strategies save time and cost. However, the integration of maintenance work into reliability analysis is difficult as common modeling techniques are often not applicable due to state explosion which calls for restrictive model assumptions and oversimplification. From authors' point of view, agent-based modeling (ABM) of technical and organizational systems is a promising approach to overcome such problems. But since ABM is not well established in reliability analysis its feasibility in this area still has to be demonstrated. For this purpose ABM is compared with Markov chains, namely by analyzing the reliability of a maintained n-unit system with dependent repair events, applying both modeling approaches. Although ABM and Markov chains lead to the same numerical results, the former points out the potentiality of an improved system state handling. This is demonstrated by extending the ABM with operators as additional “agents” featuring their location (x;y) availability (0:1) and different maintenance strategies. This extension highlights the capability of ABM to analyze complex emergent system behavior and allows a systematic refinement and optimization of the maintenance strategies. (C) 2009 Elsevier Ltd. All rights reserved.

Tags

Agent-based modeling Markov chains Maintenance Monte Carlo simulation Reliability analysis Risk analysis