Price-Induced Load-Balancing at Consumer Households for Smart Devices

Abstract

The rising share of renewable energies in today’s power grids poses challenges to electricity providers and distributors. Renewable energies, like, e.g., solar power and wind, are not as reliable as conventional energy sources. The literature introduces several concepts of how renewable energy sources can be load-balanced on the producer side. However, the consumer side also offers load-balancing potential. Smart devices are able to react to changing price signals. A rational behavior for a smart device is to run when electricity rates are low. Possible devices include washing machines, dryers, refrigerators, warm water boilers, and heat pumps. Prototypes of these devices are just starting to appear. For a field experiment with 500 households we simulate adequate device behavior. The simulation leads to a mapping from price signal to load change. We then train a neural network to output an appropriate price signal for a desired load change. Our results show that even with strong consumer-friendly constraints on acceptable price changes the resulting load change is significant.We currently implement the results with a leading energy services provider.