Automatic roughness characterization of simulated ice shapes

Abstract

During in-flight ice accretion, roughness plays an important role since it heavily influences the convective heat transfer and skin friction coefficients. This paper aims to assess the ability of existing ice accretion simulation tools to compute the growing ice’s roughness. To this purpose, a technique based on Self Organizing Maps is applied to numerical simulations of in-flight ice accretion to characterize the roughness. The numerical ice predictions are performed using a standard approach comprising RANS computations, Lagrangian particle tracking, the solution of the unsteady Stefan problem, and a morphogenetic model. Numerical simulations are performed on selected benchmark cases from the 1st AIAA Ice Prediction Workshop. Validation of roughness computation is performed on synthetic test cases, while ice roughness is compared directly to that extracted from ice scans. The results of simulated ice shapes compare reasonably well with experimental data. Computations can replicate the trend of the experimental mean ice shape and roughness distribution for both rime and glaze cases.