Computational Aerodynamics Questions & Answers  




As I explained today, ximax probably won't go down to zero for your problems if you use the TVD Van Leer limiter or WENO. The residual will thus “hang”. So, you can't rely on ximax to determine convergence. You need to proceed differently as explained in class.




I need more information. Please show a plot of the pressure contours. Also, show your code in the Post() module. Further, how are you invoking warp with opm? I hope you read in your data file... Otherwise, this will simply postprocess the initial conditions..




Everything seems fine except your matlab values. Are you doing a simple averaging using matlab? This will of course be wrong. You need to do a mass flux averaging (do $\frac{1}{\dot{m}}\int P d\dot{m}$ in matlab not $\frac{1}{H}\int P dy$).




The problem may be that you are imposing free stream boundary conditions on all boundary nodes..? You should check your boundary conditions are well imposed using the on flag.




This looks fine: your boundary conditions are well imposed. The contours you showed previously seem to have not been iterated at all. After the iteration process, are you reading in the data file (obtained after 1000 iterations or so) before outputting the post file?




Very good question. We made this assumption only when deriving the GCI. So, within the $\rm GCI_{f}$ equation, you should change the term to $\left\vert\left(\frac{\triangle x_c}{\triangle x_f }\right)^p1\right\vert$. That is, the GCI should always be positive. However, when determining order of accuracy $p$, we did not use the GCI and we did not assume that $p$ should be greater than 1. So you should not change any of the equations used to determine $p$.




The userdefined constant $\epsilon$ is included to prevent a division by zero. Set it to a very small value. As for $\gamma_0$ and $\gamma_1$, they are fixed constants: don't change them.




For the inflow BC, the stagnation pressure can be assumed equal to the one in the freestream because the flow along a streamline is isentropic. But such is not the case for the outflow BC. What if there is a shock somewhere within the domain? Then, the entropy rises and the stagnation pressure will go down and not be equal to the one in the freestream. However, for external flows around a body, the pressure will eventually become equal to the freestream pressure even if shocks are present (as long as the BC is far away from the body). Hence why it's better at the outflow BC to choose to fix pressure rather than stagnation pressure.




OK, will look into this now.




You don't necessarily need to express $\phi$ as a function of $U$ to determine $\partial \phi/\partial U$ in the same way as you don't need to express $F$ as a function of $U$ to obtain $\partial F/\partial U$.




Use a 1D extrapolation polynomial. 2D is too time consuming to compute.




No, $A$ is a $2 \times 2$ matrix determined from an average state function of $Z_L$ and $Z_R$.



$\pi$ 