6 Replies Latest reply on Jun 13, 2008 8:05 PM by Basil Gello

    airfoil problems

    Nick A
      It's me again,

      Does anyone know if COSMOSFLO has an airfoil example? The lift and drag force is not what I am expecting and does not coincide with real world tests.

      I don't know how much simpler I can make this model. I ripped out everything but one wing. I have the right side of a model airplane (7" x 3" x 0.078") flat plate wing with a +6 degree angle of attack. I know from tests that I am supposed to see 0.127 N of lift at 6 MPH velocity. The frustrating part is that I am not seeing anything near that.

      HELP!!!!!!
        • airfoil problems
          Anthony Botting
          Hi Nick: what values are you seeing in COSMOSFloWorks? I have attempted this and found out that domain distances above and below upper and lower surfaces of the airfoil need to be on the order of 200 chord lengths (for external flow applications - obviously less for wind tunnel type testing). Tony.
            • airfoil problems
              Nick A
              Hello Tony,

              I selected the XY setting for computational domain. I don't have even 1 chord distance all around my airfoil. I will test with your suggested domain value.

              I thought I made a mistake in my actual testing so I retested at 10 MPH velocity at the leading edge of a +6.0 degree angle of attack. My dimensions are 7.5in x 3in x 0.78in, that's right just a plain old plate wing. I measured 13 gram of lift via a load cell under the wing stand.

              COSMOSFLO is giving me 4.0 gram force @ 10 MPH. Oddly when I increase velocity in COSMOSFLO to 20 MPH I get what I measured in the lab.

              Thanks for the reply!

              Nick
                • airfoil problems
                  Basil Gello
                  Nick,
                  have you used 2D real world results to compare them with 3d analysis in Flo? Having quite a big number of runs related to aerodynamics I have realized that the computed values are very dependent on mesh quality and of course on the type of analysis. If you want to get 2d solution, try Zhukovskiy method and you will get clean 2d force representation. If you are using finite-width model, try better using panel methods.
                  Regards, Basil
                    • airfoil problems
                      Nick A
                      Hello All,

                      I want to post my results, so if I am doing something wrong a fresh pair of eyes can spot it.

                      As Tony suggested I increased my domain to a VERY large size ( 100 in x 100 in x 1.5" ). The calculated values are falling within 5% of my measured values at 5 MPH and at 10 MPH.

                      Why it came so close I have no idea, especially since gravity option is not selected. I used the XY option in computational domain and set the domain width to 1/5 of my airfoil length. At a Mesh of 8 and refinement of 4 it took about 1.5hrs to converge. Mind you this is a flat plate wing of which only 30% of its length is being computed. I multiplied my result by 5 (for a wing length of 7.5" ) and I got my total wing lift (right wing). Anyone see a problem with doing it this way on a uniform wing?
                      I realize the above is not doable piecewise if the wing is tapered since its chord is not uniform throughout its wing span. So the above may take a day or more to converge on the full length of a sloped wing.

                      Nick
                        • airfoil problems
                          Bill McEachern
                          gravity has nothing to do with it - it for computing bouyancy forces - natural convection type phenom.

                          You need to dothe whole wing and maybe even the whole bird in youhave nay yaw angles. You need to get the induced drag and lift effects and you need hte whole thing for that . These things take a long compute times to do right.

                          Somebody suggested panel methods. They will get the lift close but the drag tends to be worse than the floworks approach depending on what sort of b'layer approx is being used. Panel methods are essentially potential flow codes so no viscous effects though some of them have the tanspiring wall appox to name one way to account for visous effects in a panel code. Try Katz & Plotkin if you want to know more about panel methods and some aero theory in general. Pretty good book really.
                            • airfoil problems
                              Basil Gello
                              I can agree with Bill considering Katz and Plotkin's books as quite informative ones, but what about mixed viscous and potential flow approaches, probably the best one here should be a kind of combined codes like Boltzmann-lattice or at least multi-parametric Spallart coupled with potential flows. But, it is a matter of time and resources.
                              Regards,Basil