Analysis of Weather Research and Forecasting Model for Offshore Wind Turbine Applications in the North Atlantic
In the last two decades, the demand for technological development and planning within renewable energy industries has significantly increased in the United States. Offshore wind energy has shown to provide high power generation while minimizing the consumption of valuable land resources. On March 29th, 2021, the US Department of Interior, Energy, and Commerce announced a collective goal of increasing offshore wind energy production by 30 gigawatts before 2030. Development locations for future wind farms have been selected; however, data for energy loss estimations due to wind turbine blade icing will be necessary for planning purposes at each site. This study focuses on simulating cold weather events at fourteen different wind farm development sites using a Weather Research and Forecasting (WRF) model. First, WRF predictions are compared against measurements from five meteorological towers for four different winter storm events in the last five years to show model reliability. Results show larger errors in air temperature predictions compared to wind speed predictions. Simulated air temperature values were generally underestimated, while wind speed was equally over- and under-estimated. Changes in wind speed along the surface boundary layer were observed but not with the intensity as expected. However, sea surface temperatures were shown to confirm expected behavior, showing temperatures increasing with distance from shore and with decreasing latitude. Second, WRF simulations were evaluated at each wind farm location to estimate the energy loss during each winter storm event. To calculate energy loss, a model developed by Dr. Linyue Gao at the University of Minnesota was utilized. WRF predictions for wind speed, air temperature, relative humidity, precipitation, and frost were input into the model. Results show significant energy loss due to icing, ranging from 112.4 [MW*hr/MW] to 220.9 [MW*hr/MW] across each of the five winter storm events. These values give normalized energy loss as energy loss per unit of wind farm capacity. No correlation was found between icing-event duration and energy loss. For project continuation, six additional winter storm events will be simulated for the fourteen wind turbine planning sites. This will provide a large enough sample size for data significance in energy loss predictions.