Combining real-time processing streams to enable demand response in smart grids

Modern smart grids are trending towards direct involvement of end customers in reaching the optimum in balancing energy consumption and generation. In the traditional model, supply follows demand while the modern grids are shifting to the opposite where the load follows supply. Since smart grid systems process a lot of data in real-time, we have researched how we could use the data in order to save energy and reduce peaks. Reducing peaks in energy consumption can save investments on utility side by supplying the same number of customers with less power generation units. The paper presents a possible solution that gives insight into the amount of energy that could potentially be saved at any time by turning off particular devices in the Demand Response (DR) program. Moreover, the proposed solution allows utility to easily and effectively manage network. The solution relies on real-time big data processing and is implemented as Apache Storm topology. Storm processes the gathered data in two data streams - location of customers and devices measurements. By combining two data streams, we check whether a household is empty and how much energy could be saved in every moment. We measured throughput for three distinct loads which were used to simulate three different city sizes. By increasing parallelism and the number of nodes we have noticed that those two factors have a significant influence on the obtained results. What is more, these results provide us with a valuable insight into the overall and complete state of the network in real time.