Estimating the Total Installed Capacity of Rooftop Solar Photovoltaics Using Two Approaches: A Dalsig, Stellenbosch Case Study

There has been a large-scale adoption of behind-the-meter rooftop solar PV generation in South Africa. In most cases, these systems are not visible to distribution utilities, making it difficult to estimate the financial impact of these installations on utility kWh sales. These variable renewable energy systems also complicate power system operations. Therefore, a prudent operator's first step is quantifying this local resource. Several different estimation approaches exist to this end. This paper demonstrates two especially promising methods, using a case study that focuses on the residential suburb of Dalsig, located in the South African town of Stellenbosch. Method A is based on analyzing satellite imagery in the public domain. Method B is deducted from the changes in the electricity profile supplied by the utility; this data is typically confidential. Method A uses satellite imagery and the Mask Region-based Convolutional Neural Network to detect and assess the surface area of rooftop solar panels, ultimately calculating the installed capacity of these solar PV systems. Method B assesses fluctuations in electricity demand on different days with significant changes in cloud cover and minimal temperature variations to infer behind-the-meter solar PV output. Using 2024 satellite imagery, Method A estimated a capacity of approximately 500 kWp, compared to the manually identified capacity of around 570 kWp. Electricity supply data for Dalsig was only available for 2020-2021; thus, Method B could only be deployed over this period. Method B underestimated the embedded capacity at 110 kWp compared to the manually identified 165 kWp. The paper finds that Methods A and B provided reliable yet conservative insights. The paper aims to raise awareness amongst distribution utility operators and system operators that methods exist to regularly estimate the installed capacity of behind-the-meter rooftop solar photovoltaics in their local supply areas.