Impact of Sub-Hourly Resolution on the Design and Reliability of Residential Energy System Models

Energy system optimization has become an indispensable tool for planning the energy transition. However, model accuracy has traditionally been limited to hourly resolution due to data availability and computational complexity. This study quantifies resolution-induced inaccuracies in hourly and sub-hourly energy system optimization models. It focuses on a self-sufficient residential building by converting minutely-resolved renewable supply and demand data from Milan, Italy into data at five-, ten-, 15-, 30-, and 60-minute intervals using both averaging and sampling methods.

The average hourly resolution shows an underestimation of 1.71% in the total annualized cost of the system compared to the minutely resolution. In the electrical sub-system, the photovoltaic inverter is predominantly affected, being twice as large at minutely resolution in order to handle supply and demand peaks on the sub-hourly scale. To test for reliability, the operational performance of the optimal system layouts obtained from different resolutions is tested with minutely-resolved data. Our results show that system designs obtained for lower resolutions are infeasible for minutely data with lost loads of up to 89.37 kWh per year or 1.37% of annual electricity demand. Depending on the value of the lost load cited in the literature, this accounts for up to €893.67 of yearly inconvenience costs. A second method based on regular sampling (i.e., taking every 60^th value of the original time series) shows either an under- or overestimation of the total costs depending on the selected sample (there are 60 in total), with a tendency towards conservative design layouts. The two methods (sampling and averaging) reveal that hourly resolution could be sufficient with respect to total system cost approximations, but is unacceptable for sizing dynamically-operated components and strict reliability requirements.

Future research should seek to provide higher-resolved data on intermittent renewable energy sources and appropriately handle the resulting increased computational complexity of energy system models.