Efficient availability assessment of reconfigurable multi-state systems with interdependencies. (September 2017)

• Record Type: Journal Article
• Authors: George-Williams, Hindolo; Patelli, Edoardo
• Is Part Of: Reliability engineering & system safety. Volume 165(2017)
• Page Start: 431
• Page End: 444
• Publication Date: 2017-09
• Subjects: Multi-State system -- Simulation -- Interdependencies -- Availability -- Limited maintenance -- Multi-Commodity; Reliability (Engineering) -- Periodicals; System safety -- Periodicals; Industrial safety -- Periodicals; Fiabilité -- Périodiques; Sécurité des systèmes -- Périodiques; Sécurité du travail -- Périodiques; 620.00452
• DOI: 10.1016/j.ress.2017.05.010
• Languages: English
• ISSNs: 0951-8320

Abstract

Highlights: Load-flow simulation for multi-state systems with interdependencies, multi-commodity flows, and limited maintenance teams. A straight-forward procedure for uncoupling interdependencies with an intuitive mathematical model for their representation. An intuitive and general approach that requires less human effort and, therefore, robust to human-induced errors. Real-time component ranking to help decide the sequence of maintenance response that maximises system performance. Easily implementable algorithms for the efficient extraction of multi-state system availability and performance indices. Abstract: Realistic engineering systems often possess attributes that complicate their availability assessment. Notable examples being complex topology, multi-state behaviour, component interdependencies, and interactions with external phenomena. For such systems, analytical techniques have limited applicability, and efficient simulation techniques are, therefore, required. In this paper, a novel load-flow simulation approach is proposed to simplify the availability assessment of realistic engineering systems. The approach is simple and generally applicable to systems, including those with limited maintenance teams, reconfiguration requirements, and multiple commodity flows. A novel metric for assessing maintenance inadequacy and a real-time component ranking procedure are also introduced. In real-time ranking, failed components are assigned maintenance priorities during simulation in accordance with how much their availability improves system performance and how many idle maintenance teams there are. This eliminates the need for component importance ranking algorithms prior to simulation, which for some systems may be unnecessary. The applicability of the approach is demonstrated by analysing an offshore plant producing oil, gas, and water. The solution obtained is compared against another Monte Carlo simulation-based solution that requires the enumeration of the plant's cut-sets. The proposed approach is shown to be more intuitive, robust to human-induced errors, and require less human effort.