Agent-based methods for eliciting customer preferences to guide decision-making in complex energy networks

The key challenge associated with the transition to sustainable energy is dynamically balancing energy supply and demand. Information systems and smart markets play a vital role in this transition. I study electric vehicles as storage and demand response objects, which are a subset of the smart grid solutions to this societal problem. To elicit consumer behavior and deduct inferences on their preferences towards demand response mechanisms and in particular their price elasticity over time I use field experiments. Based on this experimental data, data from driving behavior, and other field experiments in smart grids I device information system artifacts such as machine learning algorithms as solutions to these problems. These artifacts assume the forms of intelligent software agents and decision support mechanisms that are used for smart energy trading and the operation of virtual power plants based on energy market signals. I validate my findings within the large scale smart grid simulation platform Power TAC. First findings underline the advantage of the trading strategy in terms of the triple bottom line: people, plant, profit. Also significant operational efficiencies in the operation of virtual power plants, in particular negative operating reserve capacity could be demonstrated.