Modeling Electricity Wholesale Markets With Model Predictive and Profit Maximizing Agents

In this paper, a new agent-based electricity market model is presented in which participants correspond to generation plants as well as storage power plants. In contrast to agent-based models where agents use learning heuristics and trial-and-error approaches to maximize their profits, the proposed model predictive bidding uses multi-step optimization to find bidding curves which maximize the expected discounted profit over a time horizon in the future. The profit is calculated based on an hourly price forward curve (HPFC), whereby the HPFC is constructed taking several factors into account. In addition, a price adjuster is used in these calculations which allows the participant to take into account his market power. The resulting optimization problem for each agent is solved using dynamic programming. A case study is carried out in which the proposed agent-based market model is applied to the four countries Switzerland, Germany, Italy, and France to study the effects of constrained cross-border capacities. The simulations show that the transmission system operators as well as the power generating units have no incentive to build additional cross-border capacity, since it lowers their profits.