Modular modeling and optimal sizing of electrified heating systems in renewable-powered communities considering system complexity and heterogeneity

Electrified heating is a key decarbonization pathway for renewable-powered communities. However, existing studies commonly adopt constant coefficient of performance (COP) heat pumps, single-node storage tanks, or aggregated single-building loads, and rarely quantify how these simplifications bias the joint sizing of the heat pump and thermal energy storage (TES) system. This paper presents a modular ordinary-differential-equation (ODE) coupling model comprising a variable-COP sewage-source heat pump (SSHP), a multi-node stratified TES tank, a controlled substation with district heating network, and 2R2C thermal networks of 15 heterogeneous residential, office, and teaching buildings, with inter-unit heat conduction aggregated into building-level representative nodes. A bi-objective non-dominated sorting genetic algorithm II (NSGA-II) framework embeds this simulation as the inner-level evaluator to jointly size the heat pump and the TES tank under time-of-use pricing and distributed wind power. A 28,800 m² community in a severe-cold region of China, with 1.5 MW of co-located wind power over a 170-day heating season, was analyzed through six progressive scenarios. The NSGA-II-optimized design reduced heat pump capacity by 23.5% and annual total cost by 23.2%, raising the seasonal COP (SCOP) from 3.29 to 3.49. The simplified model biased SCOP by −14.0%, overestimated electricity consumption by 13.1%, and underestimated comfort by 7.4%; embedded in the sizing loop, these biases would drive heat pump oversizing and TES undersizing. Compared with a coal-fired baseline, carbon intensity dropped from 89.0 to 9.7 kgCO₂/m² · yr, with wind accommodation driving 56% of this reduction. These results indicate that sizing models for such systems need to resolve off-design component behavior.