Online Demand Response Strategies for Non-Deferrable Loads With Renewable Energy

We propose two online demand response strategies to minimize the operational expenditure incurred by a non deferrable load facility over a finite planning horizon. The facility is permanently connected to the grid, and is equipped with a rechargeable battery and a renewable energy (RE) harvester. The rechargeable battery can be operated in a linear or a nonlinear regime, the latter being modeled by using Peukert's Law. The first proposed demand response strategy (DRS) is based on forecasting techniques. We specifically use a time-inhomogeneous Markov chain model to estimate future RE arrivals. The second proposed DRS is based on continuous-time optimal control theory, and does not require explicit estimates of future RE arrivals when the battery operates in its non-linear regime. Existing works in the literature require extensive computations, e.g., solving a linear program on a rolling basis, and ignore the non-linear relationship between the discharging rate and the remaining charge. In contrast, the two proposed DRSs take into account the non-linear behavior of the discharging operation, and the second proposed DRS requires fewer computations than existing solutions.