Learning Forum

[puh69]

Hello,

I notice a massive spike in my pressure derivatives close to the wall in my airfoil simulation and am looking for some direction to remove them. I believe this is unphysical and for the work I want to do from here, I need this to be accurate. My goal is to write a UDS where the pressure gradient is a source term and it'll have many issues near the wall if I employ the UDS with the current state of the flow solution.

Thanks!

[Rob]

Can you show contours in that region, node values off. Is the above spike the cell or facet value?

[puh69]

Hi Rob, 

This is not a single value problem but occurs at many cells close to the wall and I've observed these artifiical gradients on various simulations. Attached is the contour in the region along with a closer look at the dp/dx behavior in those final cells. I've attached plora od both cell and nodal values.  Please note values are slightly different as I continue to work on this issue to find a solution but the problem remains regardless of the numerical approach.

 

 

[Rob]

Are you running single or double precision? How thick are those cells?

[puh69]

DP always.

Using TMR 0012 1793x513 grid which states minimum spacing at the wall of y=4 x 10^-7

[Rob]

Look very carefully at the near wall values. At that scale you may find the cell height (dx) to be so small that the gradient is  d(big number)/d(near enough zero) to give some noise. The solver gradients will be fine, it's the reporting I'd be checking. 

[puh69]

Rob, I want to use the dp/dx and dp/dy terms as source terms in a UDS and when I do so these errors are significantly magnified. I understand that the margin of error increases with the reduction in the dx factor, but its also present in the dp/dy component too.

 

Cheers,

 

Pierce

[Rob]

You need to zoom in so we can see what's happening. 

[puh69]

[Rob]

And now the pressure & contours? 

[puh69]

Hi Rob, 

These look fine, but the devil is in the detail. If I put the dp/dx term as a source term of a UDS the value in the BL will blow up!

 

[Rob]

As I can't see the contours it's very hard to comment. Please plot with node values off and post here. 

[puh69]

 

 

 

[Rob]

Odd, how many cells are there in the unstable region? You may need to alter the colour map scale to improve clarity.