Are you using the adaptive meshing? You might be able to decrease your overall cell count but increase your resolution around the wing.
Or could you accept a 2D plane simulation at the center plane of the wing?
thanks for answering.
Yes I use adaptative mesh.
I agree I could reduce the number of cells.
But at this stage the big problem is a problem of accuracy (30% error).
2D will not fit as I consider different angles, including pitch.
Chris meant to reduce the overall cell count so you can gain more accurate results near the wing surface when you increase your mesh density there. This may help your error, much like stress concentrations in a static simulation.
You should be able to a 2D with no issues, unless you want to see tip vorticies or induced drag tendencies, or how it may act in yaw/crosswind. In 2D you can still change your pitch/angle of attack, but use a much smaller model/mesh.
Just a thought!
if it was me I would make sure the domain is the right size - same as the fluent sim.
You need sufficient mesh resolution on the wing surface to give a good aprox of hte surface geometry.
What is hte angle of attack? What is hte expected lift coefficient? If it they are high then you need to have adequate mesh inthe separated flow region or adaptive meshing with the right resolution and tightness on the gloals to ensure the flow develops sufficiently to get you to where you need to go. These are a bit tricky but inthe R'number range you are in yoyu should be able to get closer than 30%. I would suggest you are an order of magnuitude low on cell count. You need to be in the multiple million cell range or higher it my expereince for this class of problem..
thanks for helping.
Attack angle is 20° for present simulation. I will simulate angles between 0 and 50° (potential used angle, it's a marine application with high attack angle, not an air foil).
Expected lift coef at 20° is about 1.7 (drag about 0.5).
Re of the presente simulation is rather 300 000 (scaled model in tank), This should be before transition (Schlichting) thus it is simulated "only laminar". "Turbulent+laminar" is about the same.
Final application will have Re about 2 000 000.
Concerning fluent sims, cell number is about 400 000 for the scaled model and give good results.
I m trying 2D simulation, I will come back soon !
Hi Benoit, I would definitely suggest the meshing techniques suggested above. lots of cells close to the area of interest, solution adaptive, using 2d and making sure that the computational domain is big enough. There are also several other suggestions for aero flow in the solidworks KB.
The other thing that I'd make sure is that you're comparing apples to apples. I've had a couple situations where we were comparing results from flow with a "known" solution but they weren't equivalent setups. For example the testing conditions were different or the geomerty of our model was oversimplified.
Also, you might want to double check the faces that you have selected for calculating the drag/lift from Flow Simulation and the method that you're using. If you select individual faces and sum them, you might run into issues. I'd suggest using a goal and then comparing forces in Simulation to the back calculated forces from the Cl/Dd that you're comparing.
If you want to get accurate results on a 2-D airfoil I have just successfully worked through one. I'm getting close to accurate results on a 3-D wing but only after very long calculation times. I'll keep you posted.
Here's the link to my thread.