Pursuing the goal of carbon neutrality has led to an unprecedented transformation in the energy system. Generation out of thermal power plants is increasingly replaced by renewable sources. Massive installment of photovoltaic generation led to an increasingly complex regulatory and economic environment. Therein, incentives to stimulate renewable generation while preserving grid stability also entailed the use of storage systems in residential  homes. Forecast based operation strategies of these systems are deployed to integrate more renewables into the electricity system. However, solar generation as well as electrical load on site are subject to forecast uncertainties decreasing the control system’s performance.

 

Targeting this issue, this  thesis presents a control scheme using a model of forecast uncertainties to control storage systems combined with photovoltaic generation and electric  load. This control scheme is applied to a battery storage system and an electric vehicle charger in several economic scenarios. Depending on the specific  economic framework, stochastic control outperforms comparable heuristic and deterministic optimization-based control schemes. Simulations show that  it is particularly successful when competing incentives are in place. An example is weighing self-consumption of solar generation against avoiding high  power feed-in to the distribution grid. Besides the simulation studies, both application cases are demonstrated in real households as a proof of concept.

 

The studied systems were set in a residential context where the financial margins are slim. However, using stochastic model predictive control showed a  promising prospect for smart grid systems and the methods used in this work will be applied to additional systems in the future.