Messy Residuals for 2D Flow over an Airfoil

  • Last Post 21 December 2017
oyallou posted this 06 December 2017

Good evening, 


I am having trouble getting my residuals to decrease nicely with no oscillation. I have refined my mesh numerous times, as well as increased the flow domain, especially moving the outlet further far field from the airfoil but to no avail. I have attached a picture of the residuals plot that I get.. And this is after 5000 iterations. I have also included a photo of my mesh which I believe to be in good condition. The orthogonal quality is good and the skewness is below 0.95. 

Properties of the 2 Element Airfoil: 

Mainplane - 0 AoA, Flap - 25 AoA, Overall chord length - ~1.2 m

Boundary Conditions: 

Velocity Inlet - 14 m/s in the x-direction

Top and Bottom - Stationary Wall with Specified Shear components set at 0

Outlet - 0 Gauge pressure outlet

The results make physical sense but the residuals are leading me to believe that I shouldn't take this values as accurate. This is for my degree project so accuracy is key. 

Any help is greatly appreciated. Thank you

Order By: Standard | Newest | Votes
Raef.Kobeissi posted this 07 December 2017

Hello, I recommend to have an inflation layer around the airfoils, use the K Omega SST model or the Spallart Almaras 1 turbulence equation. I would also use farfield boundary condition and ideal gas for air.


Raef Kobeissi

Raef.Kobeissi posted this 07 December 2017

Your mesh could also be refined in a better way by reducing the refinement far away from the airfoil and by creating a semi-cricle on the left hand side (front side of the airfoil)

Raef Kobeissi

oyallou posted this 07 December 2017

Hi Mr. Kobeissi, 

I do have an inflation layer sizing option. I have 6 layers on my airfoils. The reason why there are so little are so they do not encounter each other, otherwise I find that the inflation layers get canceled out and the quad mesher fails. I will also try to reduce the refinement as you have directed. 

Research I have done showed that the realizable k-epsilon model is best for capturing this kind of flow domain but I will try to use the k-omega sst. 


While I have you, is the k-omega SST still applicable for 3D flow? I am currently doing a 3D simulation and my lift coefficient does not make sense. 

 Also what is the difference between a semi circle and the box domain I have currently? 

Thank you

Raef.Kobeissi posted this 07 December 2017

Hello, I disagree with the choice of model. The SST model or the Spalart Almaras are favourable. As for your question regarding a 3D model, yes the SST like all other turbulence model are used in 3D simulations. Another thing Insuggest you to do is using thr Farfield boundary conditions. Best Regards

Raef Kobeissi

José Mantovani posted this 18 December 2017

In addition, from the ideas already suggested, try to adopt a numerical solution strategy. Of course, choosing a turbulence model appropriate to the problem is the first thing to do, but try to solve the solution with the first-order interpolations, then recalculate in the second order and finally change the absolute criterion. It may be that only one of them does not respect the criterion (which is already better than everyone does not respect ...). Taking into account that a good mesh should be close to the wall so that it can capture the effects of the boundary layer and mainly give it the expected numerical values (consistent with the experimental). Use the Y + calculator to calculate the wall distance of elements near to the wall according to the conditions of the free stream and airfoil length. Below have a link to Y+ calculator.



Raef.Kobeissi posted this 21 December 2017

Here is another online app to calculate the thickness of the first layer for an appropriate Y+ value which I recommend it to be with an order of 1.

Raef Kobeissi