Frequency control with aggregated heat pumps in buildings: A comparison of complex and simple control strategies

Abstract

As the share of electricity generation from non-dispatchable sources like wind and photovoltaics grows, so does the need for active grid stabilization. For this purpose, we use aggregated heat pumps in buildings as a virtual power plant. Unlike other technologies used to stabilize the grid, such as pumped hydro or natural gas power plants, the aggregation of buildings is not hindered by geography, nor does it incur any additional strain on the environment. However, a large number of buildings is required to reach a substantial level of deployable power. The right trade-off between cost and performance in the aggregation method is essential to be competitive with the more established technologies in this field. To this end, we evaluate simple and complex frameworks for frequency control with a simulated cluster of 1000 buildings, based on real data, and estimate their respective cost-effectiveness. The most complex methods, using full data access and bi-level optimization, yield the highest total flexibility provided and the lowest tracking error. However, the lower system costs of the less complex methods, relying on limited data and simplified formulations of the optimization problem, outweigh their inferior performance in terms of expected profitability. With the assumed tariffs, we only expect a net profit when using the more cost-effective methods with large buildings. To achieve profitability with small buildings within the specifics of our simulations, the remunerations would have to increase by a factor of two to seven.