%0 Journal Article %T Modeling snowdrift on roofs using immersed boundary method and wind tunnel test %J Building and environment %D 2019 %A Wang, Jianshuo %A Liu, Hongbo %A Xu, Dong %A Chen, Zhihua %A Ma, Kejian %V 160 %N %P 106208-106208 %X Snow drifts on roofs grow with time under wind blowing during snowfalls. Computational Fluid Dynamics (CFD) is recently developed to predict this process. However, generation of body-fitted CFD grids remains challenging nowadays because of the time-changing snowdrift boundaries. In this paper, a Single-Phase Steady-State (SPSS) CFD model augmented with an Immersed Boundary Method (IBM) was proposed to simulate the snow drifts development on large roofs. Flow field of wind was described under the Eulerian framework and modeled with Reynolds-averaged Navier-Stokes (RANS) equations. The movement of snow particles in the air was modeled using the Eulerian method. The snowfall process was divided into n steady stages, and the snow boundary of each stage was updated by moving the coordinates of Immersed Boundary (IB) points. This method can adapt to complex and changeable snow boundary without re-meshing. The friction velocity over snow surface can be calculated accurately using the modified model of wall shear stress proposed in this paper. Numerical simulation results were verified by field observation and wind tunnel test of snow distribution on stepped flat roof models. The method of conducting snow redistribution test on flat roof models in wind tunnel is difficult to be applied on curved roofs. Therefore, a wind tunnel test of snowfall simulation on spherical roofs was conducted by spraying high-density silica sand particles. Good agreement between numerical simulations and experimental results was achieved and the applicability of this method to curved roofs was verified. The effects of different threshold friction velocities and test durations on the snow distribution on spherical shells were also analyzed.

Language: en

%G en %I Elsevier Publishing %@ 0360-1323 %U http://dx.doi.org/10.1016/j.buildenv.2019.106208