Data-Driven Modeling of Wind Farm Power and Revenue Generation

Accurate power and revenue prediction in wind farms is essential for optimizing performance and economic planning, particularly under variable and turbulent wind conditions. This study presents a data-driven model that refines traditional power curves by incorporating statistical corrections for turbulence and wake effects, enabling robust predictions of wind power generation and revenue. Using simulations in OpenFAST and turbulent inflow generated by TurbSim, the model is validated through a case study of the Pyron Wind Farm in Texas. The analysis leverages a comprehensive fouryear dataset, including hourly wind speed data from ERA atmospheric data and electricity price data from the ERCOT West zone, capturing real-time generation and market conditions. Model accuracy comparisons show that the smoothed turbulence-corrected curve reduces mean absolute error (MAE) by 24 percent compared to the steady flow curve, with total predicted power production within 4 percent and predicted revenue within 1 percent of actual observations. Additionally, the modeled correlation between hourly wind speed and electricity price reveals that revenue peaks often occur at lower wind speeds, countering traditional approaches that rely on mean prices and tend to overestimate revenue.