Finding optimal combination of distributed renewable energy sources for virtual power plant with minimum output variance

In the current European electricity market, if the actual generation is less than the scheduled contract or excess generation is injected into the grid, a penalty for deviation may be applied to market participants. Virtual power plants (VPPs) are trying to improve their forecasting technologies for managing this risk and increasing their expected profit. However, in this study, the authors would choose another approach to reduce the risk. To minimise the variance of a sum of outputs of all distributed renewable energy resources (DRESs), the covariance between the output of each DRES should be minimised. This covariance depends on the correlation between the outputs of DRESs. If the DRESs in the same area are aggregated as a single VPP unit, the correlation would be high and the variance of the VPP will be large, accordingly. On the other hand, if the DRESs in separated regions are collected as one VPP unit, the correlation between DRESs and the variance of the VPP may be smaller than in the former case. Using this feature, in this study, the authors would find the optimal combinations of VPP compositions with a minimum weighted average of variance of m VPP units among N DRESs based on historical output data.