A stochastic approach to predict flow properties over irregular rough walls

The influence of complex terrain orography on the aerodynamic flow field is important in a wide range of applications, from meteorology to wind energy. At the present state of art, it is mainly modeled through some aerodynamic quantities, such as the overall drag coefficient, the roughness length or the roughness function. The aim of this study is to compute with an efficient method the aerodynamic quantities characterizing a irregular rough surface by using a stochastic surrogate model that relies on high fidelity data obtained from DNS. The geometrical parameters of the roughness are treated as uncertain to build a response model in the parameter space of the quantities of interest through the polynomial chaos expansion. Then, a irregular roughness can be described in terms of probability density function of its geometrical parameters and the aerodynamic properties can be obtained by using the stochastic surrogate model together with an efficient sampling of the parameter space based on the Latin Hypercube Sampling method.  The need of carrying out computationally expensive wall resolved numerical simulations is avoided in this way. The predictions obtained with the proposed approach are compared against DNS of different irregular rough surfaces showing good agreement.