Stochastic petri net modeling for availability and maintainability analysis

• Verfasst von: Fabricius, Stefan [VerfasserIn]
• Verfasst von: Badreddin, Essameddin [VerfasserIn]
• Titel: Stochastic petri net modeling for availability and maintainability analysis
• Verf.angabe: S.M.O. Fabricius, E. Badreddin
• Jahr: 2001
• Umfang: 8 S.
• Fussnoten: Gesehen am 15.01.2020
• Titel Quelle: Enthalten in: Condition monitoring and diagnostic engineering management
• Ausgabe Quelle: 1. ed
• Ort Quelle: Amsterdam [u.a.] : Elsevier Science Ltd, 2001
• Jahr Quelle: 2001
• Band/Heft Quelle: (2001), Seite 833-840
• ISBN Quelle: 0-08-044036-3
• ISBN Quelle: 978-0-08-044036-1
• Sprache: eng
• K10plus-PPN: 1687411433

Abstract

Discrete-event simulation methodology for the evaluation of Petri nets can deal with general distributions and allow for further integration of dynamic effects at the same time. In this document, a commercially available Petri net simulator called 'PACE' is used. Examples of PACE models for individual repairable components and for system structures are introduced and their suitability for both transient and steady-state availability analysis is illustrated. In particular, maintenance strategies can be modelled, simulated and the results can be used to optimise the strategies applied to real plants. The potential availability improvements are largely dependent on the type of failures happening in a system. Yet even when the failure models are nor known with precision, theoretical considerations can be useful and can suggest advantages of a certain strategy over another. Two main critiques must be formulated with respect to PACE when being used for modelling as outlined in this text. First, it does not support object-orientation as a modelling paradigm which makes model generation and evolution somewhat tedious. Second, the model execution is considered rather slow since the PACE nets - dues to the Smalltalk foundation - are interpreted. It is assumed that native machine code would allow for significantly faster execution. The demonstration models presented here are rather simple. The investigations are of course more interesting for larger models with a variety of failure characteristics and interdependencies between individual components. In future work, costs (investment, depreciation, labour, parts, down-time), resource scarcity as well as additional dynamic effects could be included in extended, possibly hybrid models (combined discrete-event and continuous behaviour). Further the models could be altered to allow for simulations aimed at suggesting dynamic change to maintenance procedures during real plant operation, depending on the plants current states.