Load Forecasting and Demand Response Management in Distribution Grid with High Renewable Energy Penetration

Abstract

With increasing distributed renewable energy penetration, new tools are necessary to cope with variability and intermittency of these resources. Demand response and battery energy storage system (BESS) have emerged as potential solutions which provide more flexibility to the power system operator. In this paper, several active and reactive power experiments are performed on the BESS to find out the impact of BESS power flow on the voltage level. A day ahead load forecasting approach is developed which is used for the BESS power optimization. Moreover, twenty minute ahead load forecasting method is also proposed which utilizes the day ahead load forecast. Using the day ahead load forecasting method, a load following formulation is developed by which utility can shape the power curve with the help of BESS. In order to remove the power curve fluctuations, a smoothing approach is utilized in which a power line is defined and power fluctuations around the line are removed by the BESS. SOC of battery and ramp rating of generators can be used to update the slope of the power line. Operation of storage devices such as residential battery storage and water heater are optimized using dynamic programing. A method for scheduling of storage devices is proposed where a given risk function is decomposed to a given number of devices. Operation of each device is then optimized based on the obtained cost function. The demand response resources in the distribution grid can be managed either in a vertical environment or in a market environment. In the vertical structure, utility is directly controlling the available devices to come up with an optimized load curve. Customers are paid compensation for providing this flexibility to the utility. In the market environment, aggregators who manage the distributed demand response resources compete with each other to sell demand response quantity to the utility. Utility provides an inverse demand function by performing optimal power flow at each node so that the function between the price and quantity of power at each distribution node is determined. Two models for clearing price in the market is introduced and the contribution of each aggregator as well as market price is calculated for each model.