Not at a computer with flow this weekend.
A few questions:
1. When you did your cl and cd calcs. What flow parameters did you use? I prefer to work in forces to eliminate that as a variable at first.
2. Your comp domain sounds ok but can you post a picture of it and your mesh.
3. Are nd3020 and sd3020 the same? What page is your reference in that document?
Thanks for your prompt reply.
1. For Cl Cd, I first insert Force (Y) and Force (X) as goals under Input Data, then, based on them, create equation goals to calculate Cl Cd.
2. Kindly see attached pictures (overall and zoomed-in).
3. Sorry for the typo, the airfoil is SD2030 with 1 ft chord, which is used by the paper I attached in my 1st post. Its raw airfoil data can be accessed at: http://aerospace.illinois.edu/m-selig/ads/coord_database.html
I have modified it slightly (delete the 0,0 in the middle and add all-zero z-axis data) and it is also attached here.
Your further help is appreciated.
1. i'm talking more about the flow parameters, are you sure they are the same that are used to calculate Cd and Cl in your documentation? If you back calculate the force how do they compare (taking the flow characteristics out of the equation) I would also recommend surface goals in addition to your global goals just as a double check on the force being calculated. (slightly different methods but should give the same numbers)
2. you gave us the basic mesh, what does the final mesh look like?
3. as bill mentioned, have you combed over that document to make sure your flow experiment is exactly the same? geometry is definitely important but the size of the comp domain should be similar to the wind tunnel like bill said. or at least my guess is that it shoudl be a bit longer. if you post a velocity/pressure plot that might be helpful.
but in the long run, if everything is exactly the same. EXACTLY. it may still be difficult to get the accuracy that you're looking for at such slow speeds and 0 angle of attack. you may have to be ok with trending. another suggestion here would be to look at more than 1 data point to see if it trends the same as the experiment. as bill mentioned, the turbulence model may not be appropriate for this problem.
To me it sounds like your domain is way to large. I would try something in the order of 10 chords down streamm and maybe 5 above and below & 5 upstream. Run at mesh setting of 8 and see where that gets you. Goals on lift and drag. The lift should be quite close..... Belay all that. I just took a look at the paper you posted. The prediction of laminar separation bubbles is likely beyond what flow is capable of but feel free to give it a whirl. You may try it with adaptive meshing......good luck as there are not a lot of codes (none come to mind other than maybe epplers code - give Dan Somers a call if you want to buy it). The other thing on domain sizing is consider how large the the wind tunnel was. Try matching the test set up exactly - i.e put the wind tunnel in the analysis - you will need to be careful on whether the results of the experiment were "corrected" to account for the wind tunnel effects. You might want to do some research on numerical methods (CFD) and sub critical R'numbers and the effects you are trying to capture. Start with the higher R'numbers you willhave better success with those - highest to lowest and figure what is close and what is not.
Hi Bill and Jared,
Many thanks for your kind suggestions. Really helpful. I will conduct the tests according to your advices and update my result here. Truly sorry for my late reply...