The problem of in-flight icing is widely considered as a severe threat to aircrafts operational safety. In order to minimize the risk associated with icing, aircraft manufacturers must show compliance with safety requirements through flight test campaigns; contrarily, the significant improvements exhibited by the latest generations of CFD-icing simulation tools have made the numerical approach a reference instrument both for supporting aircraft certification and for aircraft design. This work aims at assessing the capabilities of the presented numerical framework in tackling the reference challenges of computational in-flight icing, i.e. the generation of the accreted shape and the evaluation of the ice-contaminated wing aerodynamic performances. Case ED0735 from Broeren et al., 2011, is considered. A glaze-horn ice shape accreted on a NACA23012 straight-wing, a geometry which already showed significant sensitivity to icing. The ice shape is simulated using the 3D stochastic ice accretion model with multi-step capabilities developed by Politecnico di Milano. The stochastic technique advantage is that of generating ice shapes that exhibit the peculiar irregularity and pronounced three-dimensionality observed in experiments that the averaging effect of classic deterministic codes struggles to capture. The aerodynamic performances of the iced geometry are obtained exploiting a RANS-LES method developed by CIRA. The code leverages on an immersed boundary technique that facilitates the domain discretization when complex 3D geometries, such as iced aerodynamic surfaces, are studied. Moreover, the hybrid approach provides more accurate solutions when dealing with the separated flows induced by glaze-horn ice shapes. In this study the computational framework is tested with a full 3D geometry and a 2D-extruded shape obtained via the bi-dimensional version of the stochastic accretion model. Comparisons of the two configurations provide insights about how different levels of fidelity in the shape generation affect performances numerical prediction and how the resulting trends compare to the experimental counterparts.