Use this URL to cite or link to this record in EThOS: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.787881
Title: Integrated workload allocation and condition-based maintenance threshold optimisation
Author: Li, Hao
Awarding Body: University of Cambridge
Current Institution: University of Cambridge
Date of Award: 2019
Availability of Full Text:
Access from EThOS:
Full text unavailable from EThOS. Thesis embargoed until 18 Oct 2020
Access from Institution:
Abstract:
Effective asset management is considered key to reducing total costs of asset ownership while enhancing machine availability, guaranteeing security, and increasing productivity. Amongst all the activities involved in asset management, maintenance has been one of the major focus areas of academic research due to its potential in helping manufacturers to generate the most value from their assets. The emergence of condition-based maintenance (CBM) in which decisions are made based on the real-time condition of assets, has opened up new possibilities in developing more comprehensive approaches to improve the performance of production systems. For instance, a trend has been observed where attempts are made to couple CBM decisions with those on other production-related factors such as inventory control, spare parts management, and labour routing. The intrinsic link between the degradation behaviour of and the workload allocated to an asset, however, has not been sufficiently studied. Consequently, the potential benefits of intervening in machine degradation, either in the context of a single asset or a fleet of assets, are rarely explored. It is therefore essential that a systematic approach is at hand to improve system performance by exploiting the inter-relationship between production and maintenance. This thesis is dedicated to developing a dynamic integrated decision-making model to improve the system-level performance of a fleet of parallel assets. The aim of the model is to realise the potential benefits, mainly in the form of lower maintenance costs and reduced penalty costs incurred due to loss of production, by simultaneously optimising workload allocation and the CBM threshold. The decision-making model is implemented using an agent-based system involving two types of agents - 1) machine agents that reside within each individual machine; and 2) a coordinator agent that oversees the entire system. The integrated decision-making model is constituted of two components - 1) a workload-dependent condition-based maintenance optimisation model based on Gamma Process at the asset level through a machine agent; and 2) a workload allocation strategy at the system level implemented by a coordinator agent. Numerical analysis is performed to demonstrate the rationale behind the decision-making process, which is to reach the most desirable balance between maintenance costs and penalty costs incurred by loss of production. The capability of the model to reduce total costs is demonstrated via comparison with traditional strategies such as uniform and random workload allocation. Additionally, the sensitivity analysis conducted has helped to reveal the respective factors that impact the potential reduction in maintenance costs and that in penalty costs, which include the sensitivity of asset degradation to workloads, heterogeneity of assets, penalty cost for a unit of production loss, redundancy of the system, etc. The model presented in this study not only assists operation and maintenance managers to make decisions on the optimal combination of workload allocation and maintenance plans for assets in a production system, but also provides guidance on whether they should invest in workload control capabilities. Furthermore, the proposed approach allows practitioners to evaluate the long-term impacts of sudden events such as an increase in demand, a decrease in the number of redundant machines, and a change in the cost of maintenance actions.
Supervisor: Parlikad, Ajith Kumar Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.787881  DOI:
Keywords: Workload-dependent degradation ; Condition-based maintenance ; Workload optimisation ; Distributed decision-making ; Multi-agent systems
Share: