CFD Simulation of Aircraft Icing Effects using Roughness Modeling

In-flight ice accretion and other contamination, e.g. by insect impact, can severely degrade airfoil characteristics and have serious consequences for aircraft handling and performance. Notwithstanding the use of Ice... [ view full abstract ]

In-flight ice accretion and other contamination, e.g. by insect impact, can severely degrade airfoil characteristics and have serious consequences for aircraft handling and performance. Notwithstanding the use of Ice Protection Systems (IPS) to avoid ice accretion, contamination effects form a critical concern in the aerodynamic design. Besides IPS failure cases and icing on unprotected surfaces, certification regulations specifically require demonstration of safe operation with up to two minutes of ice accretion on all leading edges to allow for a time lag between detection of icing conditions and IPS activation, the so called ‘delayed turn on’ (DTO) condition. Even when only a very thin layer, the roughness of DTO ice at the leading edge can severely reduce the maximum lift of a wing [1][2].
Over the past two decades Bombardier Aerospace has undertaken experimental and numerical research efforts to understand and predict the aerodynamics effects of leading edge contamination [3]-[6]. Wind tunnel and flight tests with simulated leading edge roughness allowed establishing a correlation between the loss in maximum lift and the height and density of roughness elements using a roughness parameter Rp [3]. The loss in maximum lift was shown to relate well to results from 2D panel/boundary-layer method with the Cebeci-Chang roughness model using Rp as equivalent sand grain roughness height. This also led to develop a method for estimating the contamination effect on CLmax of wings using the pressure difference rule [4].
The present paper presents the continuation of Bombardier’s research efforts in predicting the aerodynamic impact of leading edge contamination using 3D Navier-Stokes methods. The application of roughness extensions to the S-A and k-ω turbulence models is described and results for leading edge roughness on cases ranging from airfoils to aircraft are discussed and compared to experiments.