Benoît G. asked 10 months ago
Hello,
I successfully scripted a full 3D analysis, however the results I get when choosing the VLM2 method are simply “wrong”. I have wind tunnel experimental data for a 3D glider, so prior to scripting the analysis I made a bunch of them using the GUI of flow5 to see which method would give the closest results.
It turned out VLM2 was really close, and fast, which made it my first choice. However after the first scripted analysis I saw that the values for Cl were way lower ( Cl= 0.835 for scripted vs Cl=0.955 for GUI ).
Worst, when I open the scripted made run and then click on “edit” the “wrong” analysis, save without changing anything in the settings and run the exact same analysis I get the previous “correct” Cl values.
I tried to change the flow5 version in the script to match the current one, thinking it could have called a deprecated way to calculate VLM2, but to no avail.
That is a serious hurdle, as all the other methods (trilinear, triuniform, etc ) seem to work perfectly when scripted but give worst values when compared to traditional FVM CFD and wind tunnel experimental methods.
Please let me know if there is a setting I should change, or if that is a bug in the software that needs fixing.
Kind regards,
Benoît
Hi Benoit,
There must be a field which is not set correctly in the script, but I can’t figure out which. The most likely parameters which could change the value of Cl are the reference dimensions. Are they the same in the analysis.xml file and in the scripted.fl5 file?
Another suggestion: could you export the vlm2 analysis from the scripted.fl5 project to a new xml file and compare all the fields to the analysis file used in the script?
I’ll do some more testing on my side since I’ve been unable to replicate the bug so far.
André
The weird part is that if I replace VLM2 with TRILINEAR for instance in the analysis file, without changing anything else, it gives me the same Cl values with the script compared to a GUI-based analysis. Which led me to think there was an error with VLM2 itself.
I can send you my files (plane.xml, analysis.xml, airfoil.dat, airfoil polars ) and a screen recording of the bug in action if you want.
I will try what you asked now.
Best,
Benoît
Just tested the following procedure:
– Defined an analysis on flow5 GUI + ran it + checked that the results were correct
– Exported the analysis-GUI as an xml file
– Added analysis-GUI in the list of analysis to be performed by the script
– Ran the script
But then I checked the results, the error was the same. Furthermore, when I load the script-results, the fuselage visually appears when checking the VLM2 results. However, when setting a VLM2 analysis directly from the GUI only the wing and elevator appear, not the fuselage.
Could it be that it adds the fuselage by mistake to the results of VLM2?
Got it. I did not anticipate this configuration.
The reason is that the script was intended to be used without fuselages since I couldn’t think of a way to build a combined wing+fuse mesh automatically using scripting. In this case flow5 reverts to the incorrect xflr5-type configuration with adjacent wing-fuse panels.
Nonetheless I will try to do something about it and post back.
André
I see, then I will try to remove the fuselage as a temporary solution on my end to see if it solves the issue and update you.
Best,
Benoît
Update
Removing the fuselage from the model yielded different yet similar results from the GUI VLM2 which removed it automatically. I had to try with 1) the full wing 2) a cropped wing mimicking what happens in the VLM2 GUI.
The cropped version is the closest to the GUI one, yet different still. I don’t understand how to upload pictures here but I could send you the results by mail if you want.
Best,
Benoît
The file you sent me worked perfectly, thanks!
Best,
Benoît