Solar Energy Assessments: When Is a Typical Meteorological Year Good Enough?
Abstract
This study compares probability-of-exceedance values (P-values) for photovoltaic systems derived using multiple years of Vaisala’s 3TIER Services weather data to those derived using a Typical Meteorological Year (TMY) based on the same resource data. Both approaches were used to estimate the year-1 yield of eighteen Megawatt-scale photovoltaic projects located in different parts of North America, South America and Asia. Pvalues in the TMY case were derived by using the standard deviation of annual global horizontal insolation as a proxy for inter-annual variability. All other uncertainties were treated identically in both approaches. Since our analysis included only eighteen case studies, we supplemented it by examining extreme cases where the differences between the two approaches should be maximum, namely cases where inter-annual variability dominates all other uncertainties. P50 values derived from a 3TIER Services TMY are usually within 0.5% of those derived using a full time series. Meanwhile, other P-values derived using a TMY exhibited a positive bias, indicating that this approach systematically underestimates uncertainty. A simple method for removing this bias was developed using ten projects as a training data set, and tested on the remaining eight projects. Overall, differences in P90 and P99 values are typically less than 1%, but can reach up to 2-5% in extreme cases. These results can serve as benchmarks for deciding whether and when TMY analysis is good enough.
Keywords
Photovoltaic system, Energy modeling, Typical meteorological year, Tmy, Energy assessment