14 Replies Latest reply on Apr 13, 2011 9:56 AM by Joe Galliera

    Wing lift off by a lot

    Brian McVeigh

      Hi!

      This is my first time using Flow Simulation - I'm trying to find the lift generated by a wing.  Using a lift calculation, it should theoretically be around 62N, however using flow simulation its found to be just 8.4N. Any idea what could be causing this?  I have set the speed to the same used in the calculation, and attached an image of my set up (also set to external and exlude internal cavities).

       

      Also, it took over an hour to compute on quard core i5 with 4gb ram...does that sound right?  I'm currently running another and it's saying it'll take 5 hours, tutorials seem to suggest it should just a few minutes.

       

      Thanks for any help, pulling my hair out here.

       

      Cheers,

      Brian

        • Re: Wing lift off by a lot
          Bill McEachern

          If you aren't going near sonic I would say your domain is much to small for one thing. Use the default and then bring the symmetry plane in - leave te rest alone.

           

          You typically don't need all that fine mesh to get a good lift number - drag is another story.

           

          How sure are you on the "known" number? Was it calculated, if so how?

            • Re: Wing lift off by a lot
              Brian McVeigh

              Hi Bill!

              Thanks for the help, I really appreciate it!  I will re-run it with the default domain just now.  The number was calculated ising a pretty simple lifting force calculation: L = \tfrac12\rho v^2 A C_L

               

              I also used NASA's FoilSim which combines theory and collected experimental data and got in the same ballpark as the theory, but still miles off from the CFD.

              Thanks

                • Re: Wing lift off by a lot
                  Bill McEachern

                  I should have been more clear on the calculation question. Was it a calc based on just 2D or was it corrected for 3D -ie spanwise effects? Regardless, it should be in the same ball park.......sp look to the basics...is the densites used the same? How is your goal set up to compute the lift? Check all that stuff. They should line up. Angle of attack?

                    • Re: Wing lift off by a lot
                      Brian McVeigh

                      It was a 3D Calculation yes, across the entire length of the wing.  The goal was set up to compute as 'Global goal - Y component of Force'.  The angle of attack was 1 degree, and was included in the geommetry and accounted for in the theoretical calculation.  Will double check the air density just now, but I the calculation was not for high altitude flight so I imagine any difference will be negligible if there is one.

                       

                      Cheers for the help, will hopefully be able to get to the bottom of this soon!

                       

                      EDIT:  Thanks Bill!  It was as simple as having a slightly different pressure  (least I think that was it!)  Got withing 2N of my theoretical value.  Cheers for your time!

                      EDIT: 2 Turns out I had the speed set incorrectly when I re-ran it, it's not solved afterall, still getting 1/4 of what I should.

                  • Re: Wing lift off by a lot
                    Brian McVeigh

                    I made the computational domain much bigger which got the force up a small amount, but it's still only a quarter of what it should be.  I have no idea what to try now....has anyone experienced this before?

                      • Re: Wing lift off by a lot
                        Rich Bayless

                        Hello,

                         

                        Take a look at this thread:

                         

                        https://forum.solidworks.com/message/62136#62136

                         

                        He had the same problem, of the predicted force being too low with his first attempt at modeling.

                         

                        Rich.

                          • Re: Wing lift off by a lot
                            Brian McVeigh

                            Thanks Rich, I have already had a look and followed the guidance, it improved my result somewhat but its still 1/4 of what it should be.

                              • Re: Wing lift off by a lot
                                Rich Bayless

                                Hello Brian,

                                 

                                First off, I haven't worked with airfoils before, so you've been warned.....

                                 

                                That said, I did a little searching and at this site, http://www.mh-aerotools.de/airfoils/javafoil.htm they show examples of validation of their code, JavaFoil.

                                 

                                You may want to try one of these reference examples?  Or any validation example would help, there are plenty on the web it seems.

                                 

                                They also mention how their code works better with more nodes.  Would you mind posting some pictures of your mesh, as a cross section of the wing?

                                 

                                Feel free to post your model, using Pack and Go.

                                 

                                Rich.

                                  • Re: Wing lift off by a lot
                                    Brian McVeigh

                                    Hi Rich,

                                    Thanks for the help! I'm new to airfoils myself!

                                    I

                                    have attached my file, and a cross section image of the wing.  I will try a validation example and see how that goes.

                                     

                                    Cheers,

                                    Brian

                                     

                                    (Also, does anyone know how to check the numerical procedure / model being used by Flow Sim?)

                                      • Re: Wing lift off by a lot
                                        Joe Galliera

                                        Hi Brian,


                                        I downloaded your wing model to have a look at it.  I was able to very quickly get a much better answer than you reported getting (it's still calculating but after 5 mins it's at about 52 N).


                                        You have to remember that a theoretical result is most likely based off a 2D simplification and then multiplied to the length of the span, in your case 1 meter.  So I went about doing the calc as a 2D problem (in X-Y plane) using only 1mm in the Z-direction, then I used an equation goal to multiply the calculated global Y-force by 1000 to get the full 1m length.  I made the computational domain in X: -0.7m, 1.5m; Y: -1m, 1m; and Z: -0.0005m, 0.0005m.  I used a manual meshing definition for the initial mesh (to refine the partial cells that cut into the geometry) and added automatic adaptive mesh refinements at 1 and 2 travels.  You could alternatively use a mesh level of 8 with minimum wall thickness of some small value to capture the trailing edge geometry thickness.


                                        Also as noted earlier, double check your inputs as related to theoretical inputs, for example: ambient air properties (affects density), geometry, airspeed (is velocity given by Re instead and if so are you using appropriate fluid values and characteristic chord length).

                                         


                                        Updates--> with settings used above and after finished calc, resultant force stayed at about 52N.


                                        Now using auto-level 8 with expanded domain of X: -1m, 2.5m; Y: -1.5m, 1.5m.  This alone did not do much of a job because the thin trailing edge was not captured well enough; I got about 39N with this calc (although didn't finish all the way).  I created a second solid body around the trailing edge (didn't merge it) and refined the mesh there but it didn't yield the difference I was looking for.  I reduced the computational domain (CD) back to the above and that didn't change much either.


                                        Sticking with an auto-level approach, I instead created a new body around the entire wing profile by offsetting a spline (by 2.75mm) of the original profile and then extruding that mid-plane for the entire span length.  Thus I can control the meshing around the whole profile with a local mesh control.  With an auto-level of 4, I was able to get back to the better results produced by the first manual mesh above.

                                         

                                        Lessons learned: CD size does not have as much as an effect as the mesh around the entire wing profile.

                                         

                                        The files that I've uploaded are done with SW & Flow Sim 2011 SP3.0.

                                          • Re: Wing lift off by a lot
                                            Brian McVeigh

                                            Thanks so much Joe, that is a brilliant help.  I am continually amazed by this community.  When you say:

                                             

                                            Sticking with an auto-level approach, I instead created a new body around the entire wing profile by offsetting a spline (by 2.75mm) of the original profile and then extruding that mid-plane for the entire span length.  Thus I can control the meshing around the whole profile with a local mesh control.  With an auto-level of 4, I was able to get back to the better results produced by the first manual mesh above.

                                             

                                            Why did you need to create a new body and how did you use this to control the mesh?

                                            Thanks for your help, I really appreciate it.  Will hopefully be able to have a look at the file too if the department has any workstations on SP3.

                                             

                                            Cheers,

                                            Brian

                                            • Re: Wing lift off by a lot
                                              Brian McVeigh

                                              Finally got round to trying this!  So I created the study, and when it asked for initial mesh, I set this to automatic and level 4.  I then created a local mesh using a solid body that had been offset from the wing as a component, with 'disable solid body' checked, and then selected automatic again and level 4 again. Is this the process you adopted Joe?  Running as and will hopefully get the result I'm looking for!

                                               

                                              Thanks,

                                              Brian

                                                • Re: Wing lift off by a lot
                                                  Joe Galliera

                                                  Brian, turn off the "Automatic settings" option in the lower left corner of the Local Mesh Control duialog box.

                                                   

                                                  local_mesh_auto_off.jpg

                                                  Then go to the "Refining Cells" tab that now appears in the changed dialog, and turn on the "Refine fluid cells" and "Refine partial cells" options, and set to the number of cuts to the parent cell that you want performed.  In the picture below it is set for 2.

                                                   

                                                  local_options.jpg

                                              • Re: Wing lift off by a lot
                                                Joe Galliera

                                                Attached here is another good example of a airfoil profile, NACA 4412, which is a well-documented profile.  The model attached is from an earlier version of SW so more people should be able to open it.