Application of Genetic Algorithms to Optimum Offshore Plant Design

ABSTRACTThis paper discusses the requirements for reliability/availability studies during the design of oil and gas facilities, both onshore and offshore. It highlights some of the advantages of using Monte Carlo Simulation to make such assessments, and features the BG in-house package SWIFT, which enables modelling of such complex aspects as line pack and seasonal demands. It then discusses the difficulties of optimising Monte Carlo Simulation because of its stochastic nature and the consequent development of a new program, PARAGON, which overcomes these problems by combining Monte Carlo Simulation with a search mechanism based on genetic algorithms. This approach uses the principle of natural selection to automatically search for plant configurations and operational strategies that produce the best outcome, in this case the maximum life cycle return. Finally, one of the successful areas of application within BG E&P projects and Operations is highlighted, along with potential future enhancements.