Cart3D Sharp Wedge Study
This example highlights the use of PySAGAS on a CFD solution. Here, a diamond wedge geometry has been simulated in Cart3D, though the flow solution from any solver could be used instead of Cart3D.
Problem Defintion
The geometry for this example was generated using the parameteric geometry generation tool HyperVehicle. This tool provides the capability of generating cell vertex sensitivities to geometric parameters.
Free Stream Conditions
The wedge is simulated at Mach 6, with a 3-degree angle of attack. Since Cart3D is an inviscid flow solver, this is all that is required to define the non-dimensional flow state.
Parameterisation
To simplify this case study, a single parameter of wedge thickness is used to alter the wedge geometry.
Cart3D Solution
Parameter Sensitivities via Finite Differencing
Running a series of simulations in Cart3D for geometric perturbations of the wedge thickness about the nominal value, the following sensitiviites can be generated using finite differencing. Note that the values are reported for sensitivities in the non-dimensional coefficients \(C_x, C_y\) and \(C_z\).
Parameter |
\(\partial C_x/ \partial \theta\) |
\(\partial C_y/ \partial \theta\) |
\(\partial C_z/ \partial \theta\) |
---|---|---|---|
Thickness |
0.14517 |
0.126153 |
0.0000 |
PySAGAS Solution
Using the nominal geometry’s Cart3D solution, the following sensitivities can be generated using PySAGAS. Note, the error of each sensitivitiy, as calculated using the Cart3D solution for reference, is shown in brackets.
Parameter |
\(\partial C_x/ \partial \theta\) |
\(\partial C_y/ \partial \theta\) |
\(\partial C_z/ \partial \theta\) |
---|---|---|---|
Thickness |
0.15496 (5.0%) |
0.11912 (-5.6%) |
0.00312 (-) |
Special Considerations
Having a coarse geometry mesh will impact the accuracy of the Cart3D solution. In terms of computational expense, there is little reason to use a coarse geometry mesh, since this is loaded by Cart3D just once. A coarse geometry mesh may also not accurately capture the features of the geometry.
The computational expense of PySAGAS scales with the number of cells which must be transcribed from the Cart3D solution, so the resolution of the geometry mesh should not be excessive.