Load Shifting in Energy Communities by Providing User-Centered Recommendations – Forecast, Optimization and Potential

A novel approach for the optimisation of the operation of energy communities (EC) by computing individual human centred day-ahead load shift recommendations for the participants is presented. It relies solely on already available data from the smart meters / grid operator and photovoltaic inverters installed. The load shifting actions can then be carried out by the user or an existing energy management system (EMS). Therefore, no additional hardware or investment is required. In order to provide load shift recommendations prior to the load shift event, forecasts have to be implemented. The developed forecasting methods and their accuracy as well as the optimization algorithm are described. Where MOS turned out to deliver rMAE below 30 % for day ahead PV-power forecasting, while load forecasting is most efficient and accurate with a PAR model. The approach has been tested in the field and a potential analysis based mostly on calculated data has been carried out, revealing significant increases in the own consumption rate of about 11 %pt at a conversion rate of the load shift recommendations of 27 %. The potential analysis also showed that an increase of around 30 %pt could be achieved in theory if all recommendations had been executed. In order to determine the real effect of the described approach on the load profile of an energy community, further long-term field studies would be necessary.