Stochastic Dynamic Programming For Energy Management of An Overplanted Offshore Wind Farm With Dynamic Thermal Rating and Storage

We study the optimal energy management of an offshore wind farm which combines ``overplanting'' (more production than transmission capacity), ``dynamic rating'' (DR, transiently exporting more than the steady-state transmission capacity thanks to the large thermal inertia of the soil surrounding the export cable) and an energy storage (to mitigate both curtailements and forecast errors). This forward-looking setting, which aims at further reducing the Levelized Cost of Energy of offshore wind power, creates an optimization problem with both temporal couplings and uncertain inputs. The difficulty of this energy management problem comes from having time constants separated by several orders of magnitude due the thermal inertia of the cable surroundings. We propose an approximate solution based on large GPU implementation of Stochastic Dynamic Programming (SDP). In our performance comparisons, SDP outperforms simpler rule-based energy management schemes while we also explore the benefit of DR in the context of overplanting.