Computational Fluid Dynamics (CFD) simulation can be considered undoubtedly a valid means for investigating the fluid dynamic properties of complex configurations for different engineering applications. The study of the aerodynamic degradation due to ice-accretion represents a typical example of research effort to prevent in-flight disasters. In particular, aero-icing tools have been recognized in the last decade by the aeronautical regulatory agencies as future and desirable key instruments for building-up proper certification roadmaps [1]. Different European projects have been launched in the last years [2,3] linking aeronautical industries with academic and research entities. The development of accurate and robust numerical methods is crucial for mitigating the high costs linked with wind-tunnel experiments and flight tests on icing conditions [4], while increasing safety. A number of numerical issues arise due to the multi-physics nature of the problem which involves multi-phase flow, thermodynamics as well as geometry handling. Moreover, the mesh generation process itself represents one of the major bottlenecks during icing analyses especially for body-conforming methods. Indeed, the study of flows involving complex geometries requires extensive manpower for generating body conforming meshes. An alternative is the use of immersed boundary (IB) techniques based on Cartesian meshes. Their non-body conforming nature allows the coding of automatic and very-fast algorithms for grid generation as well as wall-surface tagging. Adaptive mesh refinement (AMR) procedures help clustering cells in proximity of the wall and in zones of high flow gradients. In the past, questions were raised about near-wall accuracy due to the use of wall-models especially at high Reynolds numbers. Besides, the unstructured nature of the AMR meshes makes the Cartesian methods less efficient than classic body-fitted structured codes in terms of data management and computational effort. Moreover, the application of IB methods to the icing topic is today uncommon. All the above points, and others such as High-Performance Computing, geometrical pre/post-processing and Level-Set boundary interface, represent open research areas that deserves to be addressed to allow the IB methods raising themselves as assessed tools of analysis and design towards a new generation of icing-free surfaces and/or ice-protection systems (IPS).