Data Driven Infrastructure Planning for Offshore Windfarms

Offshore wind is anticipated to play a central role in achieving the net-zero targets established by many countries, particularly within future energy systems. In the context of political and economic uncertainty, accurately estimating costs during the planning stage is critical for ensuring project bankability and attractiveness. This thesis presents a Bayesian analysis framework for predicting key performance indicators (KPIs) for offshore wind. Operational and event data from onshore and offshore wind farms are translated into turbine states and failure or repair events, and are modelled using time-to-failure and repair distributions, Poisson processes, and wind-dependent Markov chains with a tiered prior structure. Posterior distributions, derived via Markov chain Monte Carlo, are integrated into a prototype decision-support tool to generate predictive KPI distributions. The results demonstrate that explicit treatment of uncertainty can substantially influence perceived costs and investment decisions.