OpenFOAM's Arbitrary Mesh Interface (AMI) was used to communicate between a spinning mesh region containing the shaft and fan blades and a stationary region containing the main wing structure.
The simulation was initialized with a potential flow solution and then run unsteady incompressible Navier-Stokes solver. A high Courant number on the order of 10 was used in the beginning of the simulation until the mean force results stabilized. Then the Courant number was lowered to the order of 1 to get more accurate force results. As the Courant number was lowered, the changes in mean force values appeared to be proportional to the time step size!
For stability and speed, first-order upwind spatial discretization was used. It appeared to have little effect on accuracy to use second-order spatial discretization schemes. First order Euler time-stepping was used. Interestingly, second-order backward or Crank-Nicholson time-stepping resulted in much more inaccurate and unstable simulations.
The following shows the mesh.
The simulation was initialized with a potential flow solution and then run unsteady incompressible Navier-Stokes solver. A high Courant number on the order of 10 was used in the beginning of the simulation until the mean force results stabilized. Then the Courant number was lowered to the order of 1 to get more accurate force results. As the Courant number was lowered, the changes in mean force values appeared to be proportional to the time step size!
For stability and speed, first-order upwind spatial discretization was used. It appeared to have little effect on accuracy to use second-order spatial discretization schemes. First order Euler time-stepping was used. Interestingly, second-order backward or Crank-Nicholson time-stepping resulted in much more inaccurate and unstable simulations.
The following shows the mesh.