Abstract

How do you measure the performance of an estimated variable without knowing the counter factual to compare it with? This is one of the challenges facing the Operational Forecasting team at the Australian Energy Market Operator (AEMO) with respect to behind the meter (BTM) rooftop photovoltaic (PV) electricity generation.

Accurate information on BTM PV generation is essential for AEMO to manage and operate the power system in a safe, secure and reliable manner. However, BTM PV generation is not directly measured like utility scale generators. Instead, AEMO’s Operational Forecasting team estimates BTM PV generation. Therefore, it is important to understand the performance of this generation to ensure the estimates are fit for purpose.

This presentation proposes a methodology to determine the performance of the BTM PV estimates using a proxy variable that is directly measured. This proxy variable is the first derivative of demand with respect to time, known as demand ramps. The demand ramps are then compared to the first derivative with respect to time of BTM PV estimated generation, known as PV ramps, where typical statistical performance metrics can be applied. These metrics are calculated for daytime periods only.

The proposed methodology is grounded in the inverse relationship that exists between demand and BTM PV generation, where a ramp up in BTM PV generation corresponds to a ramp down in demand, and vice versa. This relationship exists because for every unit of BTM PV energy generated, either for self-consumption or exporting to the grid, is one unit less utility generators need to supply from the grid. The inverse relationship is especially the case in the state of South Australia, where the uptake in BTM PV is high compared to demand and therefore the PV generation noticeably impacts the demand during daytime. Because there can be other impacts on demand, such as temperature and time of day/year affects, the statistical performance metrics are divided into categories to account for these impacts.