Optimal Management of community Demand Response

More than one-third of the electricity produced globally is consumed by the residential sectors [1], with nearly 17% of CO2 emissions, are coming from residential buildings according to reports from 2018 [2] [3]. In order to cope with increase in electricity demand and consumption, while considering the environmental impacts, electricity providers are seeking to implement solutions to help them balance the supply with the electricity demand while mitigating emissions. Thus, increasing the number of conventional generation units and using unreliable renewable source of energy is not a viable investment. That’s why, in recent years research attention has shifted to demand side solutions [4]. This research investigates the optimal management for an urban residential community, that can help in reducing energy consumption and peak and CO2 emissions. This will help to put an agreement with the grid operator for an agreed load shape, for efficient demand response (DR) program implementation. This work uses a framework known as CityLearn [2]. It is based on a Machine Learning branch known as Reinforcement Learning (RL), and it is used to test a variety of intelligent agents for optimizing building load consumption and load shape. The RL agent is used for controlling hot water and chilled water storages, as well as the battery system. When compared to the regular building usage, the results demonstrate that utilizing an RL agent for storage system control can be helpful, as the electricity consumption is greatly reduced when it’s compared to the normal building consumption