Business jet aircraft configuration
Discretised surface of the jet
Tetrahedral mesh of the domain
Contours of properties on the jet surface
The FLITE system for aerodynamic analysis
Over the last two decades, Professor Ken Morgan, Professor Nigel Weatherill and Professor Oubay Hassan at Swansea University have lead the development of state of the art techniques for the simulation of complex aerodynamic flows. The procedure has been fully automated and highly tuned for the use in the aerospace design cycle. These procedures form the basis of the original FLITE system, extensions of which are now in production use in AIRBUS, BAE SYSTEMS and Rolls Royce.
The input to this system is the definition of the vehicle geometry, in the form of an assembly of mathematically defined surfaces and their intersection curves. The computational domain was defined to be the region surrounding the vehicle, and extending a prescribed distance from it in all directions.
The FLITE system requires that the computational domain be divided into an unstructured mesh which is an assembly of hybrid cells of hexahedra, prism, pyramid and tetrahedral elements meeting at nodes positioned at the element vertices. To accomplish this, the boundary of the domain is first discretised into an assembly of quadrilateral or triangular planar elements and the discretisisation of the domain volume then follows. These discretisation processes are fully automatic and generate points, at cell vertices, according to a user-specified point spacing.
The equations governing the fluid flow are the compressible Navier-Stokes equations. These equations, expressing the conservation of mass, momentum and energy in the fluid. In 3 dimensions these are a set of 5 coupled partial deferential equations, involving time and the fluid velocity, pressure, density, total specific energy and temperature. For the air flow simulations that are of interest here, the equation set is completed by the addition of the perfect gas equation of state. Finally, a model to account for turbulence in the flow is included. Before the solution of these equations can be attempted, appropriate boundary conditions must be prescribed on each surface bounding the computational domain.
The solution algorithm of FLITE is based upon an integral Galerkin approximate variational formulation of this classical problem statement. To produce a practical algorithm for the simulation of high speed flows, consistent stabilisation and discontinuity capturing terms have to be added to this basic formulation. In the algorithm the solution vector is assumed to vary linearly over each cell. Finite difference procedures are employed to discretise the time dimension and the solution is advanced by using a standard multi-stage explicit time stepping procedure. The convergence of the pseudo-time stepping procedure is enhanced by the addition of a multigrid procedure. In the approach adopted, the coarse meshes needed for the multigrid procedure are automatically generated from the fine mesh by agglomeration. Full parallel implementation of the FLITE system has enabled the solution of practical industrial cases to be achieved in an acceptable time scale.
The results of any computational simulation may be presented in both qualitative and quantitative form. An overall impression of the flow is obtained by using black and white, or colour-shaded, contours of selected flow variables. From such plots, flow features such as shock waves are readily detected. A more detailed analysis of the predicted aerodynamic performance can be determined from quantitative data, such as the contribution made to the lift and pitching moment by the individual geometrical components.
The FLITE system had been extensively validated on aerospace geometries for a wide range of vehicle Mach numbers.