SPLTRAK Abstract Submission
Impact of Snow Depth on Single-Axis Tracked Bifacial Photovoltaic System Performance
Annie C. J. Russell1, Christopher E. Valdivia1, Joan E. Haysom1,2, Karin Hinzer1
1SUNLAB, Centre for Research in Photonics, Ottawa, ON, Canada
/2J. L. Richards & Associates Limited, Ottawa, ON, Canada

Mid-to-high latitude photovoltaic (PV) capacity is rapidly expanding due to competitive system costs and global decarbonization efforts. However, uncertainty around PV performance under latitude-specific conditions, such as ground-accumulated snow, contribute to investment risk. In this work, we employed our custom bifacial PV modelling tool DUET, previously validated against a system at 55˚N latitude, to study variable ground clearance resulting from snow accumulation. We compare modelled rear irradiance and energy yield resulting from this snow-dependent ground clearance method against the traditional fixed ground clearance method at 69˚N and 45˚N during the snowy season for four generic 2-in-portrait single-axis tracked (SAT) systems with baseline ground clearances from 1.6-2.8 m. Hourly snow depth is mapped from a 30-year average of daily snow accumulation data. Albedo in both methods is set based on snow depth. To inform applications from forecasting to financing, we provide a discussion of hourly, daily, and seasonal impacts. Over the snowy season in Cambridge Bay, Nunavut (69˚N) and Ottawa, Ontario (45˚N), rear insolation decreases by 3.4-9.6%, resulting in 0.36-0.69% energy yield loss. These losses are comparable to annual structure shading and electrical mismatch losses which suggests that snowy season energy yield predictions may benefit from a snow depth derate. The average daily energy yield losses are 0.57-0.79%, amounting to 0.034% loss per centimeter of accumulated snow in both locations. Hourly power loss exceeds 5-9% in some hours, depending on location. When power loss is binned by hour of day, the per-bin averages peak at 1.2-1.4% loss, suggesting implications for real-time and short-term forecasting.