Multi-objective decision making for maintenance planning of deteriorating offshore floating systems

George, Biju (2022) Multi-objective decision making for maintenance planning of deteriorating offshore floating systems. Doctoral thesis, University of West London.

[img]
Preview
PDF
George - PhD Final Thesis (March 2022).pdf - Accepted Version

Download (10MB) | Preview

Abstract

Maintenance planning program of offshore assets is a complex activity due to its impact on the operational and safety risks and consequences, dependence on personnel resource availabilities, site constraints due to operational requirements and environmental factors, and uncertainties related to various vulnerabilities on asset. This thesis elaborates the challenges on offshore maintenance frameworks and have carried out a review of recent state-of-the-art literature from which have observed that the current state-of-the-art does not incorporate site constraints of the asset related to offshore personnel resource availability and impact of time required to carry out activities, into the maintenance plan and its impact on other activities due to the maintenance. Also, it has been identified that dynamic and autonomous resource allocations for maintenance activities are not employed in the offshore maintenance planning program that allows each maintenance item to independently adjust its resource allocation based on the time required to complete the activity, to improve the resource utilisation.
In this work, a novel approach has been utilised to formulate a maintenance plan optimisation problem for a Floating Production Storage and Offloading Facility (FPSO) that maximises the maintenance personnel resource utilisation and enable FPSO condition enhancement, considering the priorities with respect to design features, operating conditions, deteriorations, and the consequences of not doing the maintenance, taking into consideration the personnel resource time required for activity completion. To find the Pareto-optimal solution, an overall objective function has been developed corresponding to maintenance priorities with respect to Stress Unity Check, Fatigue Damage Ratio, Bending Moment Ratio, Shear Force Ratio, Degree of Corrosion Scale, Degree of Metal Loss, Safety Risks in the event of not doing maintenance and Financial Risks in the event of not doing maintenance, taking into consideration the personnel resource time required for activity completion using the weighted sum approach. This formulation provides flexibility to direct the focus of the overall objective function towards any one or more of the objective functions by adjusting their respective weight according to the maintenance strategy followed, which would supplement the Regulatory oversight requirements of the FPSO.
Also, in this work, a novel work management framework has been proposed that comprises of Deep Q-Reinforcement Learning (DQN) problem formulation as a solution to multi-objective optimisation problem for maintenance activities of FPSOs. The framework enables carrying out activities that have minimal site constraints, considering the design features, operating conditions, deteriorations, consequences of not doing the activities and time required to complete the activities, to get higher weighted sum of the completion times at short time as possible, whereby achieving higher resource utilisations. A greedy algorithm benchmarks the performances of DQN model and a hybrid model comprising of greedy and DQN parameters. This formulation enables achieving the optimal path for carrying out activities that liquidates the risks to the asset’s performance, which would in turn supplement the Regulatory oversight requirements of the FPSO.

Item Type: Thesis (Doctoral)
Subjects: Computing
Depositing User: Biju George
Date Deposited: 25 Apr 2022 10:30
Last Modified: 25 Apr 2022 10:30
URI: http://repository.uwl.ac.uk/id/eprint/8992

Downloads

Downloads per month over past year

Actions (login required)

View Item View Item

Menu