6 Replies Latest reply on Oct 11, 2012 6:06 PM by Arnold Zeuchner

    Drag coefficients of bends; urgent!!

    Arnold Zeuchner



      I am writing my thesis and do not get ahead. I've got a problem with the drag coefficients of pipe bends. Flow Simulation gives me the values do not fit with my literature. It bends're not overly complicated geometry!


      For example: I have a 90° bend (25mm internal diameter, 50mm radius). In my literature (Hofmann / Wasielewski) which is a drag coefficient of 0.14. SW gives me 0.56. I come to this value even if I calculate the resistance coefficient of the tube length of the arc of 78.5 mm (zeta = 0.24).


      So, can me someone help me, to make sure I take the right way? I Calculate with the formula Zeta=delta p/((roh*velocity^2)/2)


      Before I did't for this simple geometry can create right values, I need not come with anything difficult. So please help me....


      Tanks a lot in forward

        • Re: Drag coefficients of bends; urgent!!
          Stephen Petrock

          Hi Arnold,


          It's unfortunate this isn't working as you would expect it. As always when comparing experimental values to hand calculations there will be some difference.


          I think what I would do here would be to apply a local initial mesh to the region where the bend is. This will allow you to create a lot of cells in that area and hopefully you can get the results you're looking for with relatively minimal extra computer overhead.


          The local initial mesh needs to be applied to a body. I would generate a solid around the bend region in SolidWorks. After that, go to component control to disable this. Now this body will not affect the analysis. Apply a local initial mesh to this body and then you can remesh and rerun the analysis.


          This is just a suggestion to help you out. If it doesn't work at least you learned something new .


          Let me know how it goes.

            • Re: Drag coefficients of bends; urgent!!
              Arnold Zeuchner

              Thanks for the quick response Stephen!


              I have the initial mesh increased to 8, but reached an even worse outcome than the standard mesh.


              That is not for you, Stephen, but maybe that's someone reads from SW.

              Only once as a reminder: The value should be at a ratio of internal diameter and radius (in this case 2) at 0,14. Standard mesh gives me a value of 5 times higher then 0,14 and a mesh with 8 an output of 10 times higher (1.35).

              Of course the simulated value is different from the experimental value. But please, it's a simple elbow, and there is already a difference of I in a flow calculation for this Elbow now take the value of 0.14, or 1.35 to ten times.That means in my spreadsheet that I use ten times the amount of elbows.

              As I said, this is now in my eyes a very simple geometry, which should not cause these deviations.


              Nevertheless, someone has for me yet another solution proposed?


                • Re: Drag coefficients of bends; urgent!!
                  Bill McEachern

                  the above model will never get you what you are seeking. You need to have more upstream and more down stream geometry and you need to sufficient lengths to ensure a well developed flow. You need to put solid in to act as sensors to get the data for just the bend in a well developed flow feild.

                    • Re: Drag coefficients of bends; urgent!!
                      Arnold Zeuchner

                      Hi Bill!


                      I also thank you for your fast answer, but it's not so, if I have a solid as a sensor so I create there turbulences and an additinal hydraulic loss.


                      But your sensor bring me to the piont-goals. I have worked with mesh 8 and point-goals by the inlet und outlet of the elbow and the up and down stream extremely extended. I was good things, but the results showed a zeta value of 0.02.


                      So now comes the thing the other way around. Now is the drag to factor 7 too small.


                      drag with point goals.JPG

                        • Re: Drag coefficients of bends; urgent!!
                          Arnold Zeuchner

                          Sorry, but something is wrong in generally. I have the straight inlet-pipe calculated of pressure loss by Reynolds number and Lamda and compared with the goal values. So if I had (for water, 20 °C, 1 m / s, inside diameter 22.25 mm, length of 2.5 m) the pressure drop calculate, I get a value about 1450Pa. SW tells me in this area 161Pa.


                          Now please not the answer, I need an other up-stream for a well developed flow.

                            • Re: Drag coefficients of bends; urgent!!
                              Bill McEachern

                              did you apply the rght roughness to the pipe? Is the reynolds number inthe right range? You can put in any solid you want and you can disable it so that the program treats it as fluid so the whole turbulence thing is misconstrued. Take a look at the validation file for hydraulic loss and see if you learn anything from that. It is pretty obvious you do not know the ins and outs of the program. You need to construct a simulation that matches the test rig and you need to make sure the assumption inthe hand calc are well accounted for. Good luck.