Reducing Grid Distortions Utilizing Energy Demand Prediction and Local Storages

Abstract

Energy storage systems will play a key role in the establishment of future smart grids. Specifically, the integration of storages into the grid architecture serves several purposes, including the handling of the statistical variation of energy supply through increasing usage of renewable energy sources as well as the optimization of the daily energy usage through load scheduling. This article is focusing on the reduction of the grid distortions using non-linear convex optimization. In detail an analytic storage model is used in combination with a load forecasting technique based on socio-economic information of a community of households. It is shown that the proposed load forecasting technique leads to significantly reduced forecasting errors (relative reductions up-to 14.2%), while the proposed storage optimization based on non-linear convex optimizations leads to 12.9% reductions in terms of peak to average values for ideal storages and 9.9% for storages with consideration of losses respectively. Furthermore, it was shown that the largest improvements can be made when storages are utilized for a community of households with a storage size of 4.6-8.2 kWh per household showing the effectiveness of shared storages as well as load forecasting for a community of households.