Posters

Abstract

Accurate representation of wind power time series is essential for techno-economic assessment. Standard wind industry  relies on long-term hourly simulations, which capture interannual variability and P50 outcomes. While long-term hourly simulations best represent wind behaviour, many hybridisation, storage, PV and financial tools are limited to single-year (8760-hour) inputs, creating a methodological gap. The growth of hybrid plants, and the rising importance of hourly electricity price variability have created a clear need to bridge this gap. This work presents a practical methodology to adapt long-term hourly wind production data into a representative 8760-hour time series. The proposed approach is structured in three steps. First, a statistically representative 8760-hour wind year is derived from long-term hourly wind data using a Reference Wind Year methodology. For each calendar month, candidate months from all available years are evaluated against long-term behaviour using three complementary criteria: the wind speed frequency distribution, the wind direction frequency distribution, and the deviation of the monthly average wind speed from the long-term mean. Wind speed and wind direction distributions are compared using cumulative distribution functions and the Finkelstein–Schafer statistic, while the average wind speed term ensures consistency in monthly energy content. These metrics are combined into a weighted score, assigning 60% weight to wind speed distribution, 20% to wind direction distribution, and 20% to average wind speed deviation. For each month, the candidate with the minimum weighted score is selected, and the twelve selected months are concatenated to form a synthetic 8760-hour time series that reproduces long-term energy yield, directional characteristics, and realistic intra-annual and intraday variability. Second, the representative year is adjusted to match long-term P50 energy using an energy-consistent scaling approach. Instead of applying a uniform multiplicative factor to power output, which would artificially compress high-production periods, the adjustment is performed by shifting operating points backwards along the power curve through an optimisation process. The scaling factor is determined iteratively so that the resulting annual production matches the P50 target while preserving the temporal structure and relative magnitude of high-wind events. This ensures that wind power maxima and high-production periods critical for hybrid plant interactions, storage charging dynamics, curtailment assessment while remaining representative of long-term behaviour. Finally, operational availability is incorporated using an energy-based downtime framework covering both turbine and grid-related outages. Total annual availability losses are defined as energy targets and applied through discrete events rather than uniform derating. Grid (non-WTG) downtime is modelled as full-plant unavailability over randomly selected full-day blocks until the energy-loss target is met. WTG downtime is split into minor and major events, accounting for 20% and 80% of total WTG downtime energy, respectively. Minor events are modelled as the unavailability of one turbine over fixed 6-hour blocks, while major events are modelled over 15-day blocks. Events are randomly positioned along the 8760-hour time series and applied sequentially until their respective energy-loss targets are reached. This approach preserves temporal clustering of outages and avoids artificial smoothing, resulting in an hourly production profile consistent with assumed availability levels and realistic operational behaviour.