A prediction model for roof snow distribution during long-period snowfall using Euler-Lagrangian method

Abstract This paper proposed a roof snow prediction model based on the Euler-Lagrangian method. The key objective is to improve the performance and robustness of the snow surface's evolution on roof during long-period snowfall. To this end, this paper implemented a COMSALT model expressed as a pseudo-transient form to simulate the erosion and deposition process of snow particles on a roof, and added a curvature-based smoothing algorithm to the update of the snow surface on the roof. As verification, a snowfall process around a stepped roof is reproduced using a physical time of about one day. The results are compared with data obtained from field measurements and previous CFD simulations, which demonstrate that the model proposed in this paper can use a large time step of several seconds or even hundreds of seconds to estimate the snow distribution on the building roof, and the snow surface can ensure a second-order continuity to maintain the quality of the CFD calculation.