Design and Analysis of Maintenance Scheduling Algorithms for Offshore Aeroderivative Engines: An Operations Research Problem

Efficient maintenance scheduling is critical for the optimal operation and longevity of offshore plant engines, where any downtime can have significant economic and operational repercussions. This thesis presents the design and analysis of maintenance scheduling algorithms to address this complex operations research problem. The study focuses on a simulated offshore plant environment comprised of two platforms, each containing three engine packages, and with a total of eight engines. The challenge lies in ensuring that six engines are operational at all times while the remaining two are either in maintenance or on standby, under a structured maintenance regime. The first Genetic Algorithm approach is utilizing a combination of heuristic approaches and stochastic simulations so that the proposed algorithms generate and optimize maintenance schedules over a five-year period. The second Recurrent Neural Networks approach utilizes the dataset generated from our Genetic Algorithm Simulator and tries to generate optimized maintenance schedules with Machine Learning as an experimental approach. The analysis demonstrates the robustness and efficiency of the proposed algorithms in handling the stochastic nature of engine operations and maintenance requirements. This research contributes to the field of operations research by providing a scalable and adaptable solution to maintenance scheduling, with potential applications in various industrial settings beyond offshore plants. The final Maintenance Plans are presented in a Dashboard for better visual representation, making this complete pipeline from initial generation to visual representation for the client and their direct use.