Analysing green urban clusters through a multi-node dynamic model and a detailed user-driven approach

This study proposes an integrated multi-node modelling framework for the techno-economic optimisation of energy flows in urban residential clusters. While previous studies have separately advanced urban building energy modelling, Reduced Order Models (ROMs) for dynamic load simulation, and multi-energy system optimisation, their structural integration within a disaggregated, regulation-aware framework remains limited. The proposed approach combines user-level hourly demand simulation based on ROMs and stochastic occupancy profiles with a long-term linear multi-node optimisation model (H2RES) that endogenously represents distributed generation, heat–electricity coupling and regulatory incentive mechanisms. Proposed to support urban energy planning and investment assessment at cluster scale. Unlike conventional single-node or aggregated district models, each dwelling is represented as an autonomous decision-making unit while intra-cluster exchanges are optimised simultaneously. The Italian collective self-consumption energy community framework for shared renewable electricity is directly embedded in the objective function, allowing regulatory design to influence investment decisions endogenously. The model is applied to a Mediterranean multi-family residential block under baseline and Nearly Zero Energy Building-oriented retrofit conditions, with and without energy sharing to analyse intra-cluster exchanges. Results show that passive retrofit drastically reduces thermal demand without improving electrical autonomy, whereas energy sharing increases optimal photovoltaic capacity, promotes homogeneous investment distribution, and generates non-linear collective effects invisible to aggregated approaches. The framework contributes to the literature by structurally bridging dynamic building simulation and cluster-level optimisation under real regulatory conditions, providing a scalable tool for distributed energy planning.