Accurate Computer Codes

Accurate computer codes arc CFD-codcs based on solutions of the Euler equations, sometimes with a coupled boundary layer solution and solutions of the Navier-Stokes equations. They arc

used for:

• configuration optimisation, to check and improve the previous design steps based on sim­pler codes, and to include wind tunnel results.

• interference drag reduction, which is impossible using simpler codes.

• inlet and nozzle design with strong shock/boundary layer interactions.

To allow efficient exploitation of CFD-codcs. these codes must fulfill the following require­ments:

• They must represent the relevant physical properties. For SCT design, these arc: reliable radiation of disturbances (not fulfilled by most CFD-codcs). prediction of shocks and shock reflections.

prediction of separations (this still requires much research on turbulence).

• They must be able to use the exact geometry definitions including suited numerical grids.

• They must provide insight in flow physics by visualisation postprocessing of results.

• They must be able to predict aerodynamic loads.

• They must provide reliable performance predictions (drag prediction is still difficult for most CFD-codcs).

• They must be able to identify the different physical contributions of drag (still a research task, especially for supersonic flow with strong radiation properties).

• They must provide reliable aerodynamic derivatives for flight mechanics calculations.

• They must support the analysis of experiments.

If these codes are only used to check results of previous calculations, the old fashioned procedure of man hour consuming grid adaption and numerical fine tuning may be applied. But as soon as the code is used for configuration optimisation, new requirements must be fulfilled.

• A geometry generator with very few variable parameters must model the variations of interest which the optimizer has to investigate.

• The grid generator must automatically provide a suited high quality grid.

• The code must fast and automatically converge to a useful result. If the code breaks down, a (bad) result must be provided which directs the optimizer to useful variations.

• The results produced by the optimizer’s parameter variations must reflect the variation of physical results