A Systemic Agent-Based Model for Analysing Energy Policies

This thesis presents TerraSim, an agent-based model designed to simulate the interactions between human activities (both individual and collective), economic dynamics at the level of households and firms, and their environmental consequences, including but not limited to land use change and greenhouse gas emissions. A central objective is to assess the systemic impact of public policies (regulations, incentives, legislative measures...) on this coupled socio-economic and environmental system.

A key strength of the agent-based approach lies in its ability to produce emergent collective phenomena that cannot be straightforwardly derived from individual behavioral rules. TerraSim is developed incrementally: starting from a macroscopic framework that captures the main aggregate dynamics of the system, the model is progressively enriched by refining specific sectors at a finer granularity. Each enrichment cycle introduces new agent behaviors and interactions while preserving consistency with the existing structure. In this thesis, this approach is applied to two domains: the energy sector, whose granularity is refined to better capture the societal and environmental effects of diverse policy interventions, and the household sector, where aggregate representative agents are disaggregated into heterogeneous individual households to represent the diversity of behaviors, consumption patterns, and responses to policy measures.

Given that TerraSim aims for a high degree of realism, particular attention is devoted to model validation. We follow the MOSIMA methodology, which structures validation in three stages: (1) the collection of empirical data and domain knowledge to guide model design; (2) the gathering of calibration data and the implementation of automatic parameter calibration through optimization algorithms; (3) generalization testing, whereby the model is evaluated on its ability to reproduce established stylized facts. This validation process is applied at each enrichment step, ensuring that the model remains grounded as its complexity grows.

Publications 2024

• 2024
  • T. Bersoux, M. Franceschetti, C. Herpson, J.‑D. Kant : “Terrasim : An Agent-Based Simulation for systemic policy evaluation. Application to electricity mix transition”, 19^th annual Social Simulation Conference, Kraków, Poland (2024)