TURBOdesign CFD - Theory & Flow Model

TURBOdesign CFD is a finite volume based structured mesh solver which solves the Reynold's averaged Navier-Stokes equations by using a time marching method. The method employs 4 stage Runge-Kutta to march the solution in time to steady state. Eigen value scaling is used for artificial dissipation terms.

In order to reduce the computational costs TURBOdesign CFD employs three techniques to speed up the convergence to the steady state-solution. These techniques are each described briefly below;

Local Time Stepping

In the present version of TURBOdesign CFD the local time step limit is computed accounting for both the convection and diffusive contributions.

Residual Smoothing

An implicit smoothing of residuals is used to extend the stability limit and the robustness of TURBOdesign CFD. The variable coefficient formulations are used to obtain effective viscous calculations on highly stretched meshes. The time step is then computed on the basis of a fixed courant number 5.0.

Multigrid

The multigrid technique is based on the full approximation storage (FAS) schemes. A V-type cycle with sub iterations is used where the process is advanced from the fine grid to the courser one without any intermediate interpolation and when the coarser grid is reached corrections are passed back. One Runge-Kutta step is performed on the fine grid, two on the first coarse grid and three on all other coarser grids. For viscous flows with very low Reynolds number or strong separation, it is important to compute the viscous terms on the course grids. The turbulent viscosity is evaluated only the finest grid level and interpolated on coarse grids. On each grid the boundary conditions are treated in the same way and updated at every Runge-Kutta stage.

The method is currently only available in incompressible flow. Further details on the solver can be found in paper by Arnone and Pacciani (1995).