The Role of Individuals in the City-Scale Energy Transition: An Agent-Based Assessment

Abstract

An energy transition in cities is underway, bringing new technical and financial chal- lenges for multi-utility companies. The implementation of decarbonization policies has a long-term impact on the design, operation, and profitability of electrical, heat, and mobility infrastructure, resulting in significant investments and various types of risks. These risks are amplified by the individual choices of thousands of customers whose present and future behavior is uncertain. Therefore, it is necessary to analyze the inter- dependencies among the various actors in the energy transition and quantitatively assess its implications. Multi-utility companies generally address these unknowns by relying on top-down models based on historical data, statistical models, and architectural standards. These models often do not provide the granularity needed for city-scale assessments and are not well suited to model discontinuities, such as the introduction of new technologies. Thus, this study’s main contribution is the development of three predictive, integrated, bottom-up models to simulate electricity, heat, and electric mobility demand at the urban level. The most significant methodological advances of this study include the integration of the demand models into a single, high-resolution “digital twin” of an actual city, as well as the inclusion of behavioral models of individuals, obtained through the extensive and large-scale use of agent-based models. Three case studies are used to evaluate the accuracy and performance of the developed models and their suitability in supporting a sustainable energy transition. In the mobility case study, it was found that providers face considerable financial exposure at today’s low electric vehicle penetration, and the initial investment in public charging stations would only break even from a 4% electric vehicle penetration. Moreover, the study revealed that revenues based on parking duration break even faster than tariffs based on power supplied. However, the revenues from parking fees are also found to be more sensitive to user charging behavior. Furthermore, at 20% penetration, charging at public and charging at work can cause a local increase in grid load of up to 78%. In the heat case study, a model considering building occupants’ behavior was found to quantify the time- resolved heat demand better, achieving an average annual error of -4.8%. To support the city’s plans to expand the district heating network, the model was used in conjunction with a predictive model capable of quantifying the probability of a building connecting to the future network. By considering both the spatial distribution of heat demand and the probability that a building would connect to the future network, the internal rate of return of the district heat infrastructure could be increased by 25%, compared to a network extension in which the probability of connection was not modeled. Finally, the electricity study demonstrated how the implementation of behavioral models linked to location-specific habits improves electricity demand forecasts. The impact on the grid of future population dynamics and increased electricity demand due to the electrification of mobility and decarbonization of the building stock through heat pumps was quantified by identifying the areas of the city where the increase in load on the grid would be the greatest. In conclusion, this study demonstrates the superiority of bottom-up agent-based energy demand forecasting models compared to other approaches, as well as the importance of incorporating individual behavioral patterns into the models. The results of this work have already been used in the test-case city to concretely support decision-making processes, improve business models, and ensure that future investments are socially accepted.