NUMERICAL MODELING OF THE DISTRIBUTION OF SNOW LOAD ON A HYPERBOLIC PARABOLOID. THEORETICAL BASIS

Abstract

Abstract. The paper discusses the choice of a method for studying the distribution of snow loads on a biconcave roof of a hyperbolic paraboloid and its theoretical justification. It is noted that the numerical modeling of the aerodynamic characteristics of buildings and structures is a difficult and resource-intensive task due to the design features of building objects, which, as a rule, have a complex geometric shape, as well as due to a complex unsteady flow resulting from their flow around them. In addition, the task becomes more complicated due to the interference of vortex structures between different objects. Overcoming these objective difficulties became possible with the advent of modern specialized software systems, primarily ANSYS Fluent. Opportunities have appeared for accurate modeling with verification of the results obtained, which implies the use of an effective, well-tested mathematical apparatus. To implement the theory of two-phase flow, two methods based on numerical modeling are mainly used: the Euler-Lagrange method and the Euler-Euler method. The second method is used in the work. Comparative analysis, which investigates two-phase flow around different structures using different turbulence models (including RSM model, SST k-ω model, k-ε model and k-kl-ω model), shows that the k-kl-ω model is the best fit with experiment. ANSYS Fluent supports four multiphase models, i.e. VOF model, Mixture model, Wet Steam and Euler model. Compared to the other three models, the Mixture model provides better stability and lower computational costs, while the Euler model provides better accuracy, but at a higher computational cost . With a rather complex geometry and flow conditions, the use of the RANS approach does not lead to reliable simulation results. Moreover, unsteady turbulent flows cannot be reproduced. In real situations, landslides, saltations, and the suspended state of snow particles are closely related to the real effects of microbursts and bursts present at the surface of the boundary layer. Therefore, in further research, it is advisable to apply alternative approaches to RANS, which include Direct Numerical Simulation (DNS), Large Eddy Simulation (LES), and the hybrid RANS-LES approach to turbulence modeling, which combine efficiency LES techniques in tear-off free zones and the cost-effectiveness of RANS in near-wall areas.