Recent advances in large language models (LLMs) open new opportunities for simulation‑based energy policy analysis, especially for constructing structured behavioural assumptions and exploring techno‑economic scenarios. Directly replacing adoption models with LLM reasoning raises concerns about interpretability, reproducibility and behavioural validity. This paper proposes a hybrid framework that embeds bounded behavioural rubrics and structured scenario specifications into a calibrated agent‑based model (ABM) of solar photovoltaic (PV) adoption by Irish dairy farms. The approach preserves the original techno‑economic adoption mechanism while augmenting it with interpretable conservative, balanced and optimistic behavioural modulations and with scenario‑driven uncertainty analysis using fixed, rule‑validated specifications. Experiments across multiple policy settings, Monte Carlo worlds and random seeds show stable and economically plausible behaviour; adoption outcomes remain bounded and monotonic across behavioural regimes. Compared with the corresponding logistic baseline, behavioural adoption increases by about 13% without producing unstable or unrealistic saturation dynamics. The results demonstrate that LLM‑assisted specifications can be integrated into calibrated energy ABMs in a controlled, reproducible and policy‑relevant manner.
Review