Agent-based modeling: Insights into consumer behavior, urban dynamics, grid management, and market interactions

Abstract

A future sustainable energy system is expected to be digital, de-central, de-carbonized, and democratized. As the transition unfolds, new and diverse actors of various sizes will emerge in different segments. Thereby, the future energy system could shift its attention to the actors’ behavior than finding an optimum based on the physical system. Agent based modeling tools can reflect decisions from several actors in a decentralized and digital market setting. Then, such tools can enable a sustainable energy transition.<p>

<p>This work sets out to investigate how agent-based models could tackle various challenges in energy transition. This investigation covers four segments of the energy system — consumer, city, microgrid, and market. It starts with the consumer where consumer behavior is modeled. From there, expands to a city level where the dynamic characteristics of a city are simulated. The next step is distributed microgrids, particularly how to optimally plan the grid expansions. The final step in the investigation is simulating an energy market with national and international stakeholders. The selection of models presents how agent-based models can be applied to decision-making processes in the aforementioned segments. Then a novel framework with metrics for characterization is proposed and validated that addresses the challenge — which are the characteristics that make an agent-based model a better fit to tackle a modeling objective? Additionally, the framework identifies the existing knowledge gaps and the scope for further developments.<p>

<p>In summary, this work outlines how far agent-based models have come to tackle energy system challenges to sustain the energy transition. This work specifically highlights the scope, advantages, challenges, and trends of the agent-based models in energy sector applications. Moreover, this study finds that agent-based models reflect what a solution could be more than the traditional modeling practice that focuses on what a solution should be.