Planning of on/off devices with minimum run-times

To be able to reach objectives such as peak shaving or self-consumption within smart grids, it becomes increasingly important to control smart grid ready devices within households. This paper explores the planning of devices with minimal run-time constraints and constraints on total cumulative production (e.g., buffer level), such as heat pumps and electric vehicles. For such devices, a novel dynamic programming formulation is presented that makes it possible to find the optimal planning in $\mathcal{O}(N^2)$ time and space, where $N$ is the number of time intervals. To evaluate the algorithm, we integrate it into an existing decentralized demand side management approach. This evaluation shows using simulations that a load curve within a neighborhood can be flattened this way.