Conference Agenda

I provide a dynamic, stochastic central planner model of the optimal management of electricity storage at the grid level where wind electricity supply follows a realistic stochastic process in continuous time and is the main (or only) source of electricity. By choosing an infinite horizon discounted objective and a Markov process for wind electricity production, the optimal amount of electricity to be stored or unstored at any given point of time only depends on the current levels of storage and wind electricity production. The model parameters are estimated using German data on the onshore wind electricity production at the national level. Fixing an amount of installed capacities, the optimal policy for the management of storage is numerically determined. Based on the value function, the optimal electricity mix can be obtained.

The numerical results highlight the importance of demand elasticity for optimal storage capacity. While the optimal storage capacity is orders of magnitude higher of what is currently installed, the levelized cost of electricity of a wind- and storage only system is in the same order of magnitude as the levelized cost of electricity produced with fossil fuels or nuclear power. When storage costs outweigh production capacity costs, it is optimal to opt for lossy storage technologies if these are significantly cheaper. It is shown theoretically that market power can decrease the installed storage capacity drastically with respect to the social optimum.

By introducing settings with a (preexisting) fossil fuel backup, it results that any fossil fuel backup heavily decreases the optimal storage capacity. Finally, an inflexible fossil fuel backup increases the optimal storage capacity with respect to a flexible fossil fuel backup.