Study on the numerical predictive accuracy of wind pressure distribution and air flow characteristics -Part 1 optimization of turbulence models for practical use; Part 2 prediction accuracy of wind pressure distribution of various shaped buildings

To evaluate wind pressure distribution on a building by using CFD (computational fluid dynamics), it has been general practice to use k-s models. However, it is known that the use of the standard k-s model has disadvantages such as overestimation of wind pressure coefficient and turbulent kinetic energy on the windward surface where wind impinges on the building. To overcome these problems, various modifications of the k-ε model have been proposed. In the present study, a number of modified k-ε models and a k-ω model were applied for the estimation of wind pressure distribution on a parallelepiped shaped building. The characteristics of each of these turbulence models were confirmed using a wind tunnel model. The results suggest that a modified k-ε model incorporating Durbin's limiter (model parameter α=0.5) showed satisfactory results for the estimation of wind pressure distribution. In the overall evaluation, the modified k-ε models (incorporating Durbin's limiter (α=0.65), RNG model (renormalization group theory) and Quadratic model provided good results. Part one of the study was performed on an object of extremely simple shape, and questions may arise if this is applied on an actual building. In this respect, Part 2 of the paper covers a similar evaluation on a complicated shaped object. For this case, it was found that a RNG model provides high reproduction accuracy just as in the case of the object with simple shape. Also, a problem with the model incorporating Durbin's limiter (α=0.65) was found when considering the object with a complicated shape. Consequently, a modified model incorporating Durbin's limiter with a higher value for a shows better results when compared to the RNG model.