[lb13g11]

Hi all. This is my first post so bare with me.

I'm fairly new to CFD, and am currently working on my first project. It basically boils down to a 3D study of flow over a wing looking at stalling angles. I have set the case up with the wing in a box, with 4 velocity inlets (all with the appropriate velocity components), a symmetry plane from which the wing is extrude into the box (cut), and a pressure outlet. The wing has a chord of 0.1m and a span of 0.2m. The domain is meshed with tetrahedral cells, but with an inflation region at the wing surface 4mm thick.

I have done a mesh refinement study to check that the inflation region and cell sizes are ok. I have been running the solver with the k-w SST solver in double precision. All controls are second order upwind on the SIMPLE method. The solution converges very well.

So far everything sounds ok, and I thought it was until I validated my results. Looking at my values of Cd and Cl, I noticed that although my Cd values were accurate, my Cl values were far too small. For the Eppler profile I had used the Cl should have been around 1.275. I was actually getting about 0.25. I was stunned because it looks as thought my viscous forces are correct (suggesting my inflation region has worked!), but my pressure forces are far too small.

I have tried running on a different solver and I had the same problem. It is also not confined to this airfoil profile, as I had the same problem with 4 alternative profiles used. I am completely confused by it, and I have only 10 days to fix the problem. If anyone has any idea what could be causing this I would be very grateful.

Thanks
Lee

[ghost82]

Hi,
did you check reference values in fluent?

[lb13g11]

That was my first thought as well, but I checked the reference values and they are all correct for the flow conditions.

[zhaomu67]

dear lb13g11, I'm a new hand at fluent, I encountered the same problem with you. I'm running flow over cylinder, but both the lift coefficient and the drag cofficient are small. The drag cofficient should be around 1, but the actual drag cofficient is 0.15, please help me,thanks!
d=0.127mm
v=0.2m/s
Reynolds numbers:25234

[rolloblues]

Originally Posted by zhaomu67

dear lb13g11, I'm a new hand at fluent, I encountered the same problem with you. I'm running flow over cylinder, but both the lift coefficient and the drag cofficient are small. The drag cofficient should be around 1, but the actual drag cofficient is 0.15, please help me,thanks!
d=0.127mm
v=0.2m/s
Reynolds numbers:25234

Hi,
at your Re value about the 90% of drag should be pressure drag so if you are underestimating it, maybe you are not resolving correctly the wake behind the cylinder.

Can you be more precise about your mesh (posting some pics would be nice) and turbulence modelling?

[zhaomu67]

Originally Posted by rolloblues

Hi,
at your Re value about the 90% of drag should be pressure drag so if you are underestimating it, maybe you are not resolving correctly the wake behind the cylinder.

Can you be more precise about your mesh (posting some pics would be nice) and turbulence modelling?

Hi,rolloblues,
Thanks for your reply. there are some pics:



viscous modelk-epsilon RNG enhanced wall treatment)
I used Y+ wall distance estimation to calculate wall distance which is 0.075mm(here y+=1), but 0.075mm seems not right. I used 0.095mm to calculate lift coefficient and drag cofficient, in the result lift coefficient is about 0.43 and drag coefficient is about 0.8, the result seems better but still small. It seems a little change of the wall distance has a big difference on the result.
dear rolloblues,
how to choose the right wall distance?
is there any problem with y+=1?
how to resolve the wake behind the cylinder?
Looking forward to your reply！Thanks!

[rolloblues]

how to choose the right wall distance?

Getting the right wall distance it is an iterative process:
- you make a first guess mesh
- you run a few iterations
- you calculate the y+ with Fluent post-processor
- you adjust the mesh accordingly
- you re-run the case

is there any problem with y+=1?

y+ is good but isn't evertything:
also important is how many layers you have inside the boundary layer and what is the expansion ratio. For near wall calculations (y+=1) 10-15 layers with a 1.2 ratio could be a good choice.
Looking at your near wall mesh it seems to me that you have few layers and a rough transition in mesh size outside the last layer.
In my opinion this isn't any good, try to make smoother transitions.

how to resolve the wake behind the cylinder?

Again first of all try to avoid the rough transition between the O-Grid and the downstream block.