Quantifying Impact of Model Uncertainty and Weather Variability on Solar Generation

The solar industry continues to lead capacity additions in the United States, emerging as the largest contributor to new generation deployment. However, multiple reports of portfolio underperformance, including a Wells Fargo analysis citing an average 6.5% energy shortfall, highlight that approximately $100 million per gigawatt of project value may be at risk over the operational lifetime. The financial success of solar photovoltaic (PV) projects is influenced significantly not only by total lifetime energy generation but also by how consistently that generation occurs and the credibility of the underlying performance models. These factors are driven respectively by inherent weather variability (aleatoric uncertainty) and the limitations of performance modeling approaches, including data accuracy (epistemic uncertainty). Aleatoric uncertainty reflects the inherent randomness in environmental and system factors, such as annual variability in solar insolation driven by phenomena like El Niño. Conversely, epistemic uncertainty stems from incomplete knowledge of truth, inaccuracies in modeling assumptions, and measurement limitations, which can be systematically reduced through enhanced data collection and improved modeling techniques. Conventionally, uncertainty in PV project performance is represented by P50 (typical) and P90 (conservative or downside) generation scenarios, calculated by arithmetically combining variability and model uncertainty. This oversimplification leads to a bias in expected energy generation, as evidenced by reports of underperforming solar portfolios. Further, a combined "all-in" approach impedes traceability of contributing factors, limiting what project stakeholders can realistically do about it. A structured uncertainty quantification (UQ) approach systematically models both types of uncertainty - weather-driven variability and model inaccuracies - and their interactions, leading to clearer distinctions between the two. This method produces more accurate and traceable predictions for typical and downside energy scenarios. In this study, we will introduce PowerUQ, a structured uncertainty quantification framework designed to support risk-weighted decision-making across the project lifecycle.