9 Replies Latest reply on Jul 9, 2013 2:59 AM by Elvira Thonhofer

    Flat Plate External Wind Flow - Verification of Results with Theory

    Teodor Penchev

      Dear SolidWorks Forum Community,


      I have to ask you once again for your opinion. I will have to perform an external flow analysis and I thought that the best thing is to start with a very simple and basic study. If the results are satisfying and correlate with theory results then later on I can move on to more complex study situations. So, I want to perform an external flow analysis with the wind blowing perpendicular to a flat plat (e.g. advertisement panel). For this problem I also have a problem statement from a CFD book which is as follows:

      Problem Statement from Book.pngProblem Statement.png

      The solution to the problem is theoretically very easy. One has to consider the drag coefficient of the body/shape (here we have a flat plate) and use the formula for the drag force to perform the calculation. The solution is as follows (just for the force acting on the front face of the panel):

      Solution to the problem.png

      You can see that we just have to use the formula and put in the values, e.g. 2 for the drag coefficient of a flat plate, 1.3 for the density of air and so on. Basically, the drag coefficient of a flat plate does not change with the Reynolds number and is approximately 2.0 for most problems as you can see from the following screenshot taken from a CFD book:

      Value for Flat Plate.png

      The resultant drag force calculated according to the above formula is 18kN! The next thing to do is to set up a Flow Simulation study and check the results with those from theory. So, I would expect to end up with about 18kN of force on the front face. If you want you can download the complete model (without the study files) and have a look, but here I want to share some screenshots with you. The problem is that the result from the study shows a normal force (x-direction) of 7.5kN only which is 2.5 times smaller value than the theoretical one! The following screenshots show the settings of the study:

      General Settings - Analysis type.pngGeneral Settings - Fluid.pngGeneral Settings - Initial conditions.pngGeneral Settings - Wall conditions.pngCalculation Control Options - Finish.pngCalculation Control Options - Refinement.pngComputational Domain 1.pngComputational Domain 2.pngSo, the initial mesh refinement level is set to 7 with a minimum wall thickness of 25 mm (because the plate is 50 mm thick). The minimum refinement number is 2, the refinement level is set to 3 and the refinement strategy is set to Tabular Refinement. With these setting I want to make sure that the mesh is going to be good enough, but the model is really simple, so it should work. I set a surface goal for the Normal Force in X-Direction. Then I ran the study and got the result of 7.5kN on the front face. The following screenshots show the results:

      Solver.pngGoal Plot - Results.pngCut Plot - Pressure.png

      Cut Plot - Velocity.pngFlow Trajectories.pngNow, the very first question that comes to my mind is: why do I get a force of 7.5kN with Flow Simulation and 18kN with the theoretical formula? Am I doing something wrong? Are the settings that I am using suitable for this very simple part? It took really long for the solver to finish, why is it doing so many iterations? Could you quickly perform a study with your own settings and share the results with me? Do I have to change some settings? Or if not, which result should I trust? Later on it would be important because if the force is bigger than I would have to use more material (e.g. thicker plate).


      I hope you can help me with this simple problem so that I have the feeling that I am well equiped to start a more complex problem.


      Thank you in advance for your hints!


      Message was edited by: Teodor Penchev (link to the solution added)

        • Re: Flat Plate External Wind Flow - Verification of Results with Theory
          Jared Conway

          This is going to be an interesting one. You're correct that getting confidence with the software is a good first step.


          I haven't looked at your model yet, but from the screenshots, the setup looks reasonable. I haven't checked your calcs or conversions but assume that has already bend done. (I've caught myself here before)


          The first thing that jumps out at me is that your measurement is on the front face. That would be considered an open surface which may result in erroneous results. You should use a surface goal with all the faces selected for calculating the force. There are a couple of good kb articles on this.


          Next, the comp domain looks too small. This won't change your results by double, but will be needed for the long term solution.


          A couple other things that would be good to note are a screenshot of your mesh at the end and a note on what caused the solution to complete. Max iterations...etc

            • Re: Flat Plate External Wind Flow - Verification of Results with Theory
              Teodor Penchev

              Mr. Conway,


              I have checked the conversions and calculations many times and moreover, the calculations from the solution are the same as mine. So, this thing is checked. The only thing that comes to my mind is that the drag coefficient must have a lower value for the force at the end to be smaller. But everywhere in literature you can find that the drag coefficient for a flat plate is about 2.0 for most cases. That is why I expect to get more or less the same result from the Flow Simulation study - after all, this is a really simple study but the result is still not the same or close to the theoretical one.


              Here are a few screenshots of the mesh:

              Mesh 1.pngMesh 2.pngMesh 3.pngMesh 4.png


              Another very important thing that you are touching is the solver - I have to admit that it did not finish, I stoped it! Since this is a very simple flow simulation study I expected the solver to finish very quickly, perhaps within a couple of minutes. But it kept on simulating for hours and then I saw that it is settling at the value of 7.5kN with each iteration and I decided to stop it. Why did it take so long? It did not even finish. This is something that bothers me!


              Mr. Conway, would it be too much if I asked you to perform a quick simulation of the problem - the way you would do it, with your own settings and your own understanding of the problem? I am asking you for this because, for sure, you have a lot more experience then I do, I actually have none! I want that we end up having a good example the will be a starting point for some unexperienced users like me!


              Thank you in advance for any further help.