Academic Division: Energy, Fluid Mechanics and Turbomachinery
Research group: Fluid Mechanics
Enhancing engineers' capability to predict flow separation from the surfaces of an aircraft would generate important benefits in aircraft performance. In this study the attention is focused on one component of the aircraft that is usually large and requires a heavy assembly: the vertical tail plane (VTP). For common multi-engine commercial airliners, the size of the VTP is driven by a particular flight condition: loss of an engine during take-off and low speed climb. In this condition, the fin and the rudder have to be sufficient in size to balance the aircraft. The VTP is also crucial during crosswind take-off and landing, so it is import to study the behavior of the flow around its surfaces when a sideslip angle is present. Due to uncertainties in prediction of VTP effectiveness, aircraft designers keep to a conservative approach, which may risk to oversize the VTP. Uncertainties come from difficulties in predicting the separation of the flow from the surfaces of the aircraft using CFD. Improved predictive capabilities would allow a more optimal design approach with resultant weight and drag savings. Currently CFD studies are done using Reynolds Averaged Navier-Stokes (RANS) solvers, with the introduction of a turbulence model for the closure of the equations. At Airbus, for example, RANS equations are solved with two eddy viscosity models (EVM): the Spalart-Allmaras and the Menter-SST. However, the behavior of the flow computed with these models does not always match experimental observations when separation occurs. Moreover, massive separation on the VTP presents a challenge for the RANS approach. In addition, the swept shape of the surfaces leads to increased three-dimensionality and curvature in the separation process, presenting a further challenge to turbulence modelling. For these reasons, it is interesting to evaluate RANS techniques that implement a more advanced approach than eddy-viscosity models, in the form of second moment Reynolds Stress Models (RSM), due to the good abilities they can show in resolving separated turbulent flows on 3D bodies. Results from steady RANS simulations suggest that flow separation cannot be well-predicted with steady simulations. This means a shift towards the use of time dependent techniques, such as URANS and hybrid RANS/LES, is required. In addition, to provide an extensive evaluation of the SSG/LRR-w Reynolds Stress model, LES simulations will be produced on an academic test case (a backward rounded ramp), in order to provide a comparison of the turbulence properties and seek improvements to this turbulence model.