The prediction of the flow around helicopters by means of CFD methods (CFD: Computational Fluid Dynamics) is one of the most challenging problems in aerodynamics. This is due to unsteadiness of the flow and the large number of flow phenomena which must be accurately resolved, e.g. the transonic flow regions on the advancing rotor blade, dynamic stall and reverse flow areas on the retreating blade, tip vortices of the main and tail rotors and their interaction with other rotor blades and the fuselage. The blade vortex interactions (BVI) generate high load peaks and represent one of the main noise sources of a helicopter. In contrast to the rotors the flow around the fuselage is basically incompressible and many helicopters have a blunt body with large flow separations behind the fuselage. Depending on the flight conditions there may be strong interactions between main and tail rotors, rotor head, fuselage and the empennage, e.g. the tail shake phenomenon which is mainly caused by separations behind the rotor head.
In order to resolve accurrately the flow phenomena described above the solution of the unsteady Reynolds-averaged Navier-Stokes equations (URANS) is required. During one rotor revolution the unsteady airloads and centrifugal forces cause blade flapping and lead/lag motions as well as large aeroelastic blade deformations. A rotor trim procedure and a method to calculate the elastic deformations (CSD: Computational Structure Dynamics) are therefore needed to get the correct blade shape and to adjust the rotor controls for the desired flight state.
In Europe first coupled rotor simulations have been presented in the year 2000 by ONERA and in 2001 by DLR. Both approaches use a weak coupling strategy between trim, CSD and CFD-method. The weak approach is based on the exchange of periodical data and is therefore limited to steady flight conditions. In the following years the weak coupling has been continually developped and extended in a strong cooperation of DLR, ONERA, University of Stuttgart and the German-French helicopter manufacturers, see e.g. www.dlr.de/as/desktopdefault.aspx/tabid-3384/5247_read-7664/
Nowadays, the weak coupling method is regularly used in Germany and France for aerodynamic simulations of rotors and complete helicopter configurations. At DLR the CFD codes FLOWer and TAU are used. Blade deformation and trim are computed by the rotor simulation codes HOST (developed by Airbus Helicopters) or S4. The coupled aerodynamic simulation of a complete helicopter configuration is always a demanding and time consuming task. The computational grid for a helicopter without rotor head contains at least 30-40 million cells. For the unsteady coupled simulation the computing time on a supercomputer requires several days simulating 10 - 20 rotor revolutions for a steady flight condition. Improved solutions using finer grids, methods with enhanced physical modelling or higher order methods require much longer computing times, especially if the aerodynamics of maneuvering helicopters will be investigated in future.
Current research activities consider the application of unstructured CFD methods like the DLR TAU code in order to simplify the grid generation of complex helicopter fuselages including rotor heads and the development of a strong coupling procedure using the multibody simulation code SIMPACK.
Another focal point of the research and development activities is the improvement of the accuracy of the CFD methods. First topic concerns the consideration of the laminar-turbulent transition. Second topic is the exact representation of the tip vortices and the wake behind the rotor head. The interactions of the tip vortices with the other rotor blades (BVI) have a large impact on the rotor performance and especially on the external noise. The exact prediction of BVI using standard 2nd order CFD method is not possible, even on very fine grids the vortex representation is insufficient. Therefore, advanced CFD methods for simulations of rotors and complete helicopter configurations require an higher order spatial discretization scheme combined with local grid refinement.