54 Replies Latest reply on Apr 7, 2015 7:58 AM by Josiah Lund

    Lift Coefficient on 2D Airfoil

    Josiah Lund

      I've been working with a model trying to calculate the lift coefficient on a 2D section of a HM60 airfoil. Data for the airfoil can be found here. The airfoil is 1 ft in length in flow at 150 ft/s giving a Reynolds number of ~950,000. I have done my own XFLR5 (based on Xfoil) simulation on the airfoil to confirm the data found on the website and my results are in agreement. I zoomed in much further and Cl is about 0.0788

      XFLR5.PNG

      Cl v Alpha.PNG

       

      On the SolidWorks side, I have set up a 2D flow simulation with air at a velocity 150 ft/s and density of 1.225 kg/m^3 (0.002377 slug/ft^3). Air properties are standard properties of air referenced from table in some of my classroom textbooks. These are the same values that are used in XFLR5.

      Initial Conditions.PNG

      The domain I left the automatic size as it seemed sufficiently large.

      Domain.PNG

      The lift coefficient was calculated as follows using the global Y force as the lift.

      Cl Equation.PNG

      The mesh was created using the highest level automatic mesh with a local mesh with a level 1 overall refine. In the control conditions, the refine level was set to 4 with an approximate max number of cells set to 10,000,000. To my understanding, it only refined the mesh twice. The final mesh is shown below.

      Total Domain.PNGLocal Initial Mesh.PNGAirfoil Refinement.PNGSurface Precision.PNG

       

      And now the results. The jump at the beginning was due to me having the Cl equation off. There were approximately 312,000 fluid cells with 16,000 partial cells. The coefficient sat right around 0.07.

      Solver.PNG

       

      I realize that my calculation is not fully converged, but it did not seem to be going anywhere, so I stopped it.

       

      Can anybody advise me on how to get a little more agreement between my calculated value of ~0.07 and what I believe is the more accurate and true ~0.08, or is this as close as I can hope to get?

        • Re: Lift Coefficient on 2D Airfoil
          Bill McEachern

          I would just restart the calc with the mesh now existing - it should get close it was on the way up by inspection of your plot. Flow sim does not have the appropriate turbulence model for aerodynamic work at moderate angles of attack. I would call it more of an industrial aerodynamics code. Pretty decent estimates for separated and quite messy flows at least in my experience. That said though it should produce a decent number for lift (normal force). You really need the spalart-allmaras model for well behaved flows to get a decent drag number..

          Drela's Xfoil code will do a better job for 2D section flows I suspect.

            • Re: Lift Coefficient on 2D Airfoil
              Josiah Lund

              Thanks for the response. Are you suggesting that this is about as good as I can get with the software? When you say that flow sim does not have the appropriate turbulence model for aerodynamic work at moderate angles of attack, what is considered a "moderate AoA"? I would expect to run into issues at high angles of attack and have to run a transient simulation, but should I expect any issues at a 0 deg AoA?

               

              I did continue my calculation to convergence and the Cl stuck right around 0.07.Solver Complete.PNG

              I am aware of software such as Xfoil and XFLR5 and their usefulness, but the purpose of this 2D sim was to do a proof of concept to move towards a 3D sim of an RC plane I am designing with a team as a capstone project. That also raises some issues seeing that it has taken 7 hours for convergence for a 2D simulation, so I am concerned about the amount of time to run, but I would first like to see what level of accuracy can be achieved.

                • Re: Lift Coefficient on 2D Airfoil
                  Bill McEachern

                  in my experience, which is not extensive in using flow sim for aerodynamics, but considerable otherwise suggests that in well behaved flows (small regions of separation) where the tangential drag is dominated by the turbulence model flow sim will provide very conservative estimates of drag. With large separated regions it performs better than a lot of other codes but it still isn't particularily close (near and beyond stall). There is a paper done by who I can't remember on assessing flow sim (probably Floworks at the time) compared with Star CCM+ (Or CD can't recall which) and the trends just mentioned were shown.  Experimental data was also presented. The paper assessed the codes ability to predict the flow over a whale flipper type geometry with turbulces, which seems like the same R'number regime that an RC aircraft might fly in (ie subcritical).  The lift, since it is dominated by normal forces was usefully  close. If I can dig it up I will post it for you. I would compare flow sim on 3d flows, 2d might not be sensible use given the alternatives. And really it is the ability to do a 3d flow that is likely of primary interest and the performance of a 2d assessment for a 3d tool relative to a 2d tool might not actually be as meaningful as you might think.

                  by the way, how does the 2d code work on the prediction of laminar separation bubbles? In my experience this is particularily challenging and I think will be a phenomena of concern in the small scale model aircraft flight regime. The turbulces are suspected of avoiding this and this would be an inherently 3d (span wise) effect. Good luck.

                  I would call flow sim a good industrial aerodynamics code and a useful aerodynamics code though it would not be a code where you would rely on where largish sums of money are dependent on it being very close. Then again not many codes fit that bill without extensive experience and validation for a particular class of problem.

                    • Re: Lift Coefficient on 2D Airfoil
                      Josiah Lund

                      Regrettably, it took me too long to figure out this simulation, so my team already has a plane designed that we are well into the process of manufacturing. The simulation results will only be an addition to the report that we turn in rather than a driving factor in the design. That being said, I will take what I can get. I am aware that drag can be particularly difficult due to viscous effects and turbulence. In this case our simulation results will be compared with results from Xflr5 and Xfoil as well as in an attempt to verify hand calcs that are already complete.

                       

                      I would be interested in reading that article if you are able to place your hands on it.

                       

                      Thanks for your insight.

                      Josiah

                • Re: Lift Coefficient on 2D Airfoil
                  Josiah Lund

                  I believe I have figured out my problem. I stumbled across another thread that suggested using a computational domain on the order of 10 cord lengths behind and 5 on top and bottom. My original domain was only about 4 behind and 2 top and bottom. I now converge to a Cl of 0.0764. I am trying another simulation where I extend the domain even further behind, but I am very happy with the current value being off from the Xflr5 value by just over 3%.

                   

                  Solver Complete 3%.PNG

                    • Re: Lift Coefficient on 2D Airfoil
                      Jared Conway

                      great example.

                        • Re: Lift Coefficient on 2D Airfoil
                          Peter Thannhauser

                          Josiah, you are light years ahead of me, but I'm very interested in your post. This is the first time I've seen SW actually compared like this. I've (hackishly) run XFOIL in the past through Profili front-end interface, but never evolved to XFLR & don't have SW-simulation. But one of common Q&A issues I recall hearing about when people were experiencing similar Cl-Cd accuracy issues sometimes stemmed back to airfoil input geometry. I'm paraphrasing & might be off base, but if the x,y data points were:

                          - not dense enough, particularly around critical nose area

                          - were not sufficiently accurate (# sig figs if imported from other sources or blended between dissimilar airfoils)

                          - generated from spline curve intercepts, or even dependant on different spline algorithms yielding different curves between honored x,y intercepts

                          - roughness settings (here my memory is getting foggy, but I thought there was an adjustment type factor within Xfoil?..might have this mixed up with airfoil blending subroutines)

                           

                          Anyway, what I'm wondering out loud is: is SW-sim seeing the same 'curve/resolution' in between the imported airfoil X,Y data points? If its being viewed as 'rougher' or a smidge different geometry, that might explain the descrepancy?

                           

                          Hope you get it resolved & good luck. If you have any project links, be happy to view. Speaking of tubercles, maybe your next assignment is propeller simulation (these are all the rage in F5B/F5D electric competition)

                          2-20-2015 0000.jpg

                      • Re: Lift Coefficient on 2D Airfoil
                        Osayi Richmond

                        Hey guys, I followed all your comment. so far, I have learned a lot. However my result is a Nightmare. I am simulating a NACA 4412 at an AoA of 6deg using this paper (Computational Fluid Dynamics Study Of Fluid Flow And Aerodynamic Forces On An Airfoil ). The Author uses ANSYS FUENT to get Lift Coefficient of  0.65 but In my study on SolidWorks I am getting 0.25. I am using 10 chord length, 5 up and down, 2.5 forward. Temp 288.17K and Density of 1.225Kg/m^3, chord lenght of 0.1m. I followed the instruction above, level 8 mesh level (Auto) and level 1 refine mesh in Local Mesh (I had 2 location 1 on a part around the airfoil and on the Surface of the airfoil). In my Control setting, I used level 4 refinement in the Domain and Local meshes, No maxi travel. I will be very great for any reply now. ThanksNACA 1.PNG

                          • Re: Lift Coefficient on 2D Airfoil
                            Jared Conway

                            can't see the mesh for yours

                            it looks like it is no where near converging

                            can you back calculate the force so that all the fluid parameters aren't part of the equation

                            and can you describe how you're measuring the force? global? on surfaces?

                              • Re: Lift Coefficient on 2D Airfoil
                                Osayi Richmond

                                Okay. My initial mesh is Automatic (Level 8) and I have 2 local meshes (1. on the surface (2. on a  part that is invisible and they were all in level 1 in Solid/Fluid interface and Refinement. On the calculation control, the refinement level was all in Level 4. Here is a picture showing the mesh before and after calculation (close view). I didn't get what you mean but I am getting my Lift Force using the Y-axis Force and Drag using the X-axis. I didn't get what you meant by "can you back calculate the force so that all the fluid parameters aren't part of the equation". Working backward using the Result Pressure Coefficient (Integration)? or using SolidWorks integral result?

                                BEFORE!

                                Final Mesh.PNG

                                AFTER CALCULATION!

                                Final Mesh 2.PNG

                            • Re: Lift Coefficient on 2D Airfoil
                              Osayi Richmond

                              Hey Guys, I will be presenting my analysis on Wednesday but the Force I am getting from SolidWorks is very small compare to the Force on ANSYS. This analysis was carried out in with Surface goal, however I have tried Global goal too but still small. I think the Force at Y in SolidWorks for a certain point?. Please Guys how do I work out the FORCE Lift or Drag, 0.001N doesn't look reasonable. here is a comparison of my result to the experiment paper I am using (this was carried out in ANSYS).

                              AoA.PNG

                              AoA 6.PNG

                                • Re: Lift Coefficient on 2D Airfoil
                                  Jared Conway

                                  your results are so far off, i'd guess that the setups aren't the same or the geometries aren't the same

                                   

                                  maybe post your model

                                   

                                  and have you tried xfoil? i think it is free and might give you another number to work with and compare against

                                    • Re: Lift Coefficient on 2D Airfoil
                                      Osayi Richmond

                                      Thank you very much. I think I was comparing a 2D model with a 3D model. Please is this a 3D model? (see table 2) please. http://www.ijer.in/ijer/publication/v3s3/IJER_2014_305.pdf 

                                        • Re: Lift Coefficient on 2D Airfoil
                                          Amit Katz

                                          I don't know how much I would trust that paper. Besides the many grammatical and spelling errors (which I attribute to writing in a language other than one's native tongue), it doesn't seem like an in-depth analysis carried out by an expert. There's no discussion of the fundamentals of CFD or the math behind the solution, it's just a report about a CFD project. Looks like the work of a single student.

                                          • Re: Lift Coefficient on 2D Airfoil
                                            Josiah Lund

                                            It says right in the abstract of both papers that the analysis is for a 2d airfoil. I finished one simulation on the NACA 4412 and here is what I came up with.

                                             

                                            NACA 4412.PNG

                                            I am not sure why I was able to get such good agreeance between the HM60 xflr5 analysis and the SW flow sim, but for the NACA 4412 at 2 deg AoA I got a xflr5 Cl of about 0.7 and for the SW flow sim I got about 0.54. Again, a little confused about this myself but this is what I have so far.

                                              • Re: Lift Coefficient on 2D Airfoil
                                                Josiah Lund

                                                The computational domain is much larger than displayed. I just did a detail view of the mesh. The domain is 4 cord lengths in front, 5 top and bottom, and 10 behind.

                                                  • Re: Lift Coefficient on 2D Airfoil
                                                    Josiah Lund

                                                    I'm thinking that the airfoil used may be incorrect. Compare below

                                                     

                                                    Airfoild Side by Side.PNG

                                                    • Re: Lift Coefficient on 2D Airfoil
                                                      Osayi Richmond

                                                      Thanks for your work, I really appreciate the time. I am doing another simulation in NACA 0012 I CL of 50 (Using a domian of 10C, 4C,4C) and I got CL of 56 (Using a domain of 13C,13C,13C) my Target was 0.57 from the paper I post early on http://www.ijer.in/ijer/publication/v3s3/IJER_2014_305.pdf. this is the force I am getting but PLEASE why is my force very low?... and I have notice that SW forces are very LOW is XFOIL or ANSYS force thesame as SW? Thanks

                                                          • Re: Lift Coefficient on 2D Airfoil
                                                            Josiah Lund

                                                            Dont try to match the force because that is dependent of how the domain was set up. A "thicker" 2d domain will give a larger force. Cl is what you want to match as it gets rid of any size effects. The other persons results do seem to be for a 3D wing. I dont see how you could get that much force by analyzing a thin slice.

                                                              • Re: Lift Coefficient on 2D Airfoil
                                                                Osayi Richmond

                                                                Thanks very much brother, all your replies are too useful ... Anyway, I was having a doubt about this the 3D wing too. what does this mean? I highlighted this on the image below. Thanks

                                                                3D Mesh.PNG

                                                                • Re: Lift Coefficient on 2D Airfoil
                                                                  Osayi Richmond

                                                                  Please guys. I want to do a CFD analysis on NACA 0020, so I am doing a simulation on NACA 4412 and NACA 0012 but my current setting (Autmatic mesh level 8) my current result is very low with my NACA 4412 (Non symmetry), but on the NACA 0012 I am getting CL 50 and the target is 56. this is not too bad I guess but Using Automatic Mesh is not reliable for Validation because this will change anytime I change the angle of attack, I mean I can't be able to control this which mean I can't validate the CFD setting I am using. So, do anyone know what Mesh Cells I should be looking at? and Should this study be on Non-refinement or Should I control the refinement using Iterations?. I AM CONFUSE or you I just ask how quick can I validate my CFD result? THANKS

                                                                  • Re: Lift Coefficient on 2D Airfoil
                                                                    Jared Conway

                                                                    remember that in 2d analysis the cell is still 3d, so it should be as thin as possible and to make things comparable you should know the thickness of your cell so that you can come up with force/l. multiply by length to get the total force (minus end effects). using force means that you remove the other elements like flow parameters.

                                                                     

                                                                    and you cannot compare 2d to 3d which I think you are saying. because of end effects.

                                                                     

                                                                    I suspect that not every airfoil will be successful in CFD. this is well documented and applies to many codes, it really comes down to the strength of the turbulence model. as noted by the kb, tech reference, help...etc when separation occurs, that is generally hard for flow simulation to get very good results for lift and drag.

                                                                      • Re: Lift Coefficient on 2D Airfoil
                                                                        Osayi Richmond

                                                                        Jared Conway Thank you for your reply. Anyway, when you said CELL thickness, do you mean the Z-axis length of the computation domain? and when you said force/l. is the l length? and if we divide the force by l length while multiplying the force again? I really appreciate for the reply. I think I get what you mean now force/l means force per unit length? thank you.

                                                                          • Re: Lift Coefficient on 2D Airfoil
                                                                            Jared Conway

                                                                            z axis > yes, if your airfoil is in XY

                                                                             

                                                                            l is length

                                                                             

                                                                            i'm not sure i'm following your last question

                                                                             

                                                                            2d analysis is a slice

                                                                            your output is Force

                                                                            but since the slice has a finite thickness you're really getting force per length (or width...however you want to call it)

                                                                            once you have that, you can scale it based on teh actual length or width of your actual wing. but this does not account for end effects. that would require a full 3d analysis

                                                                              • Re: Lift Coefficient on 2D Airfoil
                                                                                Osayi Richmond

                                                                                Jared Conway Thank you again. I have a presentation on Wednesday. I have no result yet because I am getting CL of 0.8 and my target is 1.0 for NACA 0012 @ AoA of 10 degree. I am trying to validate the NACA 0012 with past experiment before running a simulation on NACA 0020 for my presentation. So, this is really though for me now. This result was obtained using 10C behind, 4C at front and 5C top and bottom. The contour very close to the result from CFX Simulation. However, using 13C behind, front, top and bottom. I am getting 0.91 CL. So, what do you think? Is 10C, 5C top and bottom reliable? or should I got on with this? PLEASE

                                                                                  • Re: Lift Coefficient on 2D Airfoil
                                                                                    Jared Conway

                                                                                    I don't really understand what you are saying. It might be a good idea to talk to your prof or TA about your questions.

                                                                                     

                                                                                    My opinion is if you are getting 0.8 and experimentally you want 1.0, you're in the right zone. Especially if you are getting 0.91 in another CFD software.

                                                                                     

                                                                                    But again, your paragraph isn't very clear to me on what your question is. I can interpret it as you got 0.91 with a bigger comp domain or with CFX. If the former, and your goal is 1.0, I don't understand why you wouldn't go with that value. If the latter, I don't know why you wouldn't try extending the comp domain in Flow. And overall, I don't know what your presentation or goal with this particular analysis is. There is a very good discussion about accuracy for aerodynamics in both the KB and technical documentation. I think you're in the ballpark and that is what is important but you'll have to be confident in your results and be able to explain why there is a delta above "some guy on the internet said it was ok". This is where I think your prof and TA can help.

                                                            • Re: Lift Coefficient on 2D Airfoil
                                                              Osayi Richmond

                                                              Hello Jared Conway. What Value are you talking about? the Turbulence parameter or the Domain? by the way. this is an update of my simulation. Do anyone have this problem, I think this is due to the turbulence setting and boundary conditions. I am validating my result with NACA 0012 'Abbott & von Doenhoff Experiment'. Thanks guys.

                                                              Current result.PNG