AN EVALUATION OF 3 MODELS DESIGNED FOR SITING WIND TURBINES IN AREAS OF COMPLEX TERRAIN

The performance of three micrositing models to simulate the winds in complex terrain was evaluated. Two of these models, MS3DJH/3R and BZ, are descendants of the Jackson-Hunt (J-H) theory, while the third, NOABL, is a mass-consistent model. The NOABL model was operated in both optimizing and normal modes. The models were evaluated by comparing model-derived simulations of the wind to actual wind data collected from two areas of complex terrain: Askervein Hill in Scotland and the Altamont Pass area of California. The overall performance of the models was about the same. When the models were applied to Askervein Hill, the errors in the prediction of averaged winds were in a range of about 8% to 10% for all models. Wind speed prediction errors for Altamont Pass were sensitive to the number of wind observations available for either model input, in the case of mass-consistent models, or post-processing of the model simulations, in the case of J-H models (or the NOABL model operated without optimization). With only one wind input wind station employed for model input, all models performed poorly and exhibited wind speed prediction errors of about 25%. Using six wind input stations reduced the errors to about 9% to 10%. Errors of this magnitude are essentially identical to those found for the Askervein Hill model simulations. The greatly improved model simulations that follow from the use of the six wind observations underscore the necessity for wind data at more than one point for some modeling situations.