Impact of Data Temporal Resolution on the Energy Yield Modeling of Tandem Photovoltaic Devices: Article No. e70481

Full physics-based equivalent circuit models for tandem photovoltaic (PV) devices provide high accuracy but are computationally intensive. This study investigates reducing simulation runtime by downsampling meteorological input data using three methods-mean, median, and slicing-across two distinct climates (Colorado and Oregon) on the energy yield modeling of tandem devices. These methods represent both post-measurement downsampling (mean and median) and changes in field measurement frequency (slicing). We show that the number of device model evaluations may be reduced by up to 99% while maintaining an annualized energy yield estimate error of less than 0.2% if the resampling method conserves total solar insolation. Conversely, the nonconservative slicing and median methods produce unreliable results with errors exceeding 5% due to inherent energy nonconservation. We identify two primary drivers of modeling error: energy conservation and PV nonlinearity. While these underlying error mechanisms are climate agnostic, the resultant impact on energy yield estimates varies in different meteorological regimes. In high-volatility environments, smoothing effects suppress subcell mismatch and nonlinear losses, leading to overestimation. In low-irradiance conditions, efficiency losses dominate, shifting errors toward underestimation. The results of this work provide a framework for high-speed tandem modeling that balances computational efficiency with site-specific accuracy.