Abstract

Weather-data representation can materially influence the simulation-based assessment of residential photovoltaic (PV) systems with battery storage, particularly when conclusions depend on the timing of generation, household demand, and grid exchange. While previous studies have shown that weather-data choice can affect building and PV simulation outputs, fewer have examined how different representations of the same reanalysis source influence residential PV-battery assessment. This question is particularly relevant for ERA5, which is increasingly used to generate both typical meteorological year (TMY) files and long-term simulation inputs. This study develops a framework for modelling in the whole-building performance EnergyPlus simulation engine using weather data from multiple consecutive years and year-specific load files, processed using Python scripts. This framework is applied to a residential grid-connected PV-battery system, modelled and validated against monitored 2020 data, and compares an ERA5-derived TMY with a 30-year chronological ERA5 sequence (1996-2025) within an EnergyPlus workflow. In comparison to the chronologically ordered multi-year ERA5 weather data, the ERA5-derived TMY overpredicted the annual energy delivered at the inverter output by 3.5% and underpredicted the net grid import by 6.3%, indicating that it provides a less reliable basis for assessing grid dependence.