Hourly Modeling Correction in Solar Energy Assessment
DNV’s Solar Resource Compass now also includes the hourly modeling corrections (HMC) automatically in its indicative energy assessments.
The vast majority of solar projects built today are developed and financed using hourly resolution data. This is a natural consequence of both the data sets and the turn-key modeling tools that are readily available. The use of hourly data in modeling leads to biased P50s that over-predict generation. This is increasingly relevant as solar photovoltaic projects are now installed in new regions with fewer clear days and more resource variability. The use of higher DC to AC ratio of projects has also increased the importance of understanding intra-hourly variability and its impact on energy modeling. Finally, the Inflation Reduction Act (IRA) implementation allows the use of Production Tax Credits (PTCs) for solar projects along with the traditional Investment Tax Credits (ITCs), which will amplify the importance of the accuracy of energy estimates.
The hourly modeling error is primarily due to inverter clipping with smaller contributions from transposition and temperature models. The Figure below adapted from Anderson et al (2022) (IEEE, NREL), illustrates how 1 minute DC power (blue) contain minutes that are both above and below the inverter clipping level (dashed black), yet an hourly average (orange) is strictly below the clipping level. In these hours, inverter clipping causes the true hourly AC power to be significantly lower than the power that would be computed using hourly DC data. A variety of techniques have been proposed to quantify and correct for these hourly modeling errors, including temperature-corrected insolation ratios, use of NSRDB PSM 5-minute irradiance data, and machine learning or other statistical techniques. We refer readers to Anderson et al (2022) (IEEE, NREL) for a review of the hourly modelling error literature.
DNV models the HMC using the Anderson and Perry (IEEE, NREL) model with two extensions. The first extension is a temporal adjustment factor that accounts for the sampling frequency underlying the hourly data. The temporal adjustment factor is detailed in Mikofski et al (2022) (JPV). The second extension is a spatial adjustment factor, detailed in Hayes et al (2022) (IEEE), in which the wavelet variability model is used to simulate the irradiance over projects of various areas.
The HMC model is a standard component of every DNV solar energy assessment. DNV’s Solar Resource Compass (SRC) now also includes the HMC automatically in its indicative energy assessments. SRC is intended for feasibility analysis, but it shares the same analytical tools that DNV analysts use and allows users to align with IE practices earlier in the project lifecycle. SRC also provides the estimated HMC as an annual loss factor for those users that would prefer to add this loss factor into their existing solar energy modeling workflow (see Figure below). For existing SRC users, this additional functionality is available automatically for all new and historical projects.
Learn more about hourly modeling corrections for accurate solar energy assessments in DNV's on-demand webinar with Southern Company which discusses errors in PV simulations caused by hourly resolution weather data.