Warning: Use of undefined constant the_title - assumed 'the_title' (this will throw an Error in a future version of PHP) in /var/www/html/wp-content/themes/uofmsustainable/single-research.php on line 3

A method to estimate residential PV generation from net-metered load data and system install date

  • May 2020
  • Peer-Reviewed Articles
  • Multiple

Stainsby, W., D. Zimmerle, and G.P. Duggan. A method to estimate residential PV generation from net-metered load data and system install date. Applied Energy (267).

ABSTRACT: In the USA, residential photovoltaic (PV) systems are often configured for net metering “behind-the-meter”, where PV energy generation and building energy demand are reported as a combined net load to advanced metering infrastructure (AMI) meters, impeding estimates of PV generation. This work presents a methodology for modeling individual array and system-wide PV generation using only weather data, premise AMI data, and the approximate date of PV installation – information available to most distribution utilities. The study uses 36 months of data spanning nearly 850 homes with installed PV systems in Fort Collins, Colorado, USA. The algorithm estimates building energy consumption by comparing time periods before PV installation with similar periods after PV installation that have common weather and activity characteristics. Estimated building energy consumption is then compared with AMI meter data to estimate otherwise unobservable solar generation. To assess accuracy, modeled outputs are compared with directly metered PV generation and white-box physical models of PV production. Considering aggregate, utility-wide, generation estimates for the three year study period, the proposed method estimates over 75% of all days to within 20% of established physical models. The method estimates more effectively in summer months when PV generation peaks and is of most interest to utilities. The model often outperforms physical models for days with snow cover and for arrays with shading or complex multi-roof implementations. The model also supports day-ahead PV prediction using forecasted weather data.

Read the full article.