Modelling and predicting wind velocity patterns for windbreak fence design

Abstract The use of windbreak fences has become a common practise to reduce wind erosion and dispersion of soil particle or spray-drift in potentially risky areas. For effective and economical use of windbreak fences, this paper introduces regression equations to predict wind speed reduction by windbreak fences based on their screen porosity, fence height and wind speed. The prediction equations provided straightforward procedures to predict the effects of a single fence and multiple-fence arrays. The equations were developed by non-linear regression analysis based on data obtained from computational fluid dynamics simulations, which were validated in advance by wind-tunnel tests and experimental data of literatures. The prediction equations, derived by a genetic algorithm, for a single fence showed good agreement with the simulation results, with R 2 =0.99. The equations for multiple fences were derived from the product of the equations for a single fence and also showed good agreement with the simulation results.