Just as computer-aided design software has pervaded today's aircraft structures design offices, computational fluid dynamics (CFD) is gaining the respect of aerodynamicists for its ability to speed up aerofoil selection.

While useful for trade-off studies, for years the tool was regarded suspiciously by traditional aerodynamicists due to its often untrustworthy absolute results, earning it the nickname "colour for directors" because of the elaborate flow visualisations it could generate.

Now, as the computer forms as much a part of final aircraft shape definition as the windtunnel, engineers are tackling one of the most challenging modelling requirements - complex high-lift systems on transport aircraft.

One of the objectives of the European Union's three-year European High Lift programme (Eurolift), which concluded at the end of last year under the EU's Fifth-Framework research programme, was to improve the ability of CFD methods to model high-lift flow. The work extended the 2D research of the earlier Garteur programme to a 3D Airbus A310-representative wing shape and focused on flow conditions up to flight Reynolds numbers.

Traditionally, CFD programmes have difficulty in converging on high-lift flow solutions, mainly because of the widely varying flow speeds involved. The modelling of turbulence at high Reynolds numbers must accurately simulate large areas of flow separation, re-circulation zones and wake effects.

CFD tends to under predict low Reynolds number lift due to grid-resolution issues and leading-edge suction-peak underprediction. However, at maximum lift conditions at high angles of attack, numerical methods tend to overpredict lift due to the difficulty of recreating stall behaviour. High-lift flow contains both low-speed and transonic regions, so finding a steady-state solution, especially in the transition regions, can be difficult.

For this reason windtunnel data is still the prime method of evaluating a wing's high-lift and maximum-lift performance.

The Eurolift programme compared results for the A310 wing obtained from several European manufacturers' CFD methods to windtunnel results, including flight Reynolds number results obtained from the European Transonic Windtunnel in Cologne, Germany. The programme looked at grid-generation algorithms, turbulent flow transition prediction and solver accuracy and efficiency. Results published by Germany's DLR, Sweden's FOI defence research agency and France's ONERA show that the inaccuracies can be overcome with intelligent grid generation and by applying appropriate solver preconditions - essentially "forcing" certain transition conditions.

One conclusion was that efficient grid generation, structured or unstructured, can be invaluable in increasing the accuracy and efficiency of the solving process. More interestingly, aerodynamicists may be able to get away with imperfect flow representations and coefficient of pressure distributions, providing the CFD solution matches the total lift at high angles of attack. By using the knowledge that the magnitude, if not the detailed solution, is correct, designers may be able to cut down windtunnel work, beginning testing further into the preliminary design process, saving costs.

Source: Flight International