7 Replies Latest reply on Jul 24, 2011 5:19 PM by Eamon Ronan

    Flow simulation accuracy

    Adam Fain

      Has anyone ever had the opportunity to compare flow simulation results to actual test data?  How did it compare?  I'm running internal flow on a swivel head for a hydraulic torque wrench to find the pressure drop across it.  When compared to actual lab data, flow simulation predicted consistently lower values.  I defined an inlet volume flow, and an outlet static pressure (ambient), then surface goals on the inlet and outlet to find the static pressure difference.  I defined hydraulic oil as a custom liquid, so perhaps the viscosity or density is slightly different, but it wouldn't be as significant as what was seen.  I'll attach a screenshot of the graphs for reference.

        • Re: Flow simulation accuracy
          Igor Isayev

          My typical application is different then yours, but I'm always trying to correlate CFD results with Lab. Looking at the both results you posted -empirical & CFD  , it does not look too bad . Depends on how CAD model used in CFD study is close to the actual one, material properties, BC - the result correlation may vary.

          However , if you still concern in regards to poor correlation, I would recommend to look at the mesh size you have .

          Pressure is a main driver in your flow domain ( or in flow channels). Make sure that you have an adequate amount of mesh cells across the channel ( at least 3-4 ). I would recommend to run your apps using different mesh sizes . It's always a good habit to check out the mesh sensitivity results. Also, you may want to specify roughness ( avg.) that can contribute something to total pressure loss.

            • Re: Flow simulation accuracy
              Adam Fain

              It turned out to be the mesh size.  I increase the resolution from 3 to 5 in the wizard and the values were a lot closer.

                • Re: Flow simulation accuracy
                  Joe Galliera

                  Mesh has a huge influence on the results calculated; I always interrogate the mesh before running it.  I also like using the adaptive mesh to capture gradients in the flow field and where there is no geometry (that the mesh will automatically focus on initially).

                   

                  Other typical issues with matching lab results are things that we don't have much control over such as recreating the physical enviroment and some things we have more control over, like material properties, although not exactly.

              • Re: Flow simulation accuracy
                Rich Bayless

                Hello,

                 

                Looking at the swivel.jpg, I don't see the connecting hoses or the fittings for the pressure gages.  You may want to add something to better match your actual lab setup, with hoses, fittings, etc.  They will add a bit more pressure drop.

                 

                Slight changes in viscosity or temperature can have a big impact.  You need some way of finding out your actual fluid properties.  Another approach might be to work with a well documented fluid, such as water or purchase your hydraulic fluid with a material certificate.  Perhaps a silicone oil with a certificate might be a better fluid for you in the lab.

                 

                The viscosity index of your hydraulic fluid is another issue, where changes in temperature cause changes in viscosity.  Do you have thermocouples at the inlet and outlet of your test rig?

                 

                Just a few thoughts,

                 

                Rich.

                  • Re: Flow simulation accuracy
                    Adam Fain

                    On an unrelated note, when I try to use fluid pressure as a load for a static analysis, I get the error "no loads applied to the model."  Anyone know why this occurs and how to fix it?  The .fld file gets saved and I import it into simulation saying using the "flow effects" option, but for some reason the software won't recognize any loads.

                      • Re: Flow simulation accuracy
                        Joe Galliera

                        In Flow, you can do the Export results to Simulation at any point, it actually just simply marks the surfaces that will be transferred; the important point is that you save the model after doing this export command.  Try this out to see if it gets rid of your problem with recognizing loads.

                      • Re: Flow simulation accuracy
                        Eamon Ronan

                        Adam,

                        From a new user  - trying to correlate with actual test results is exactly what I've had trouble with,  What threw me was the the results for some test-comparisons were close - but done the wrong way.......

                        Some progress by following advice which concurrs with above, including Rich's comments on making the CAD model replicate as close as possible to a physical test rig, and the especially the comments by Igor and Joe on adaptive meshing and refinement. 

                        If of any use (experts - put me right if this is wrong), I've also put in 'dummy, dis-abled lids' away from the inlet/outlet lids for greater accuracy, and changing the criterion from the defaults.  All else I can suggest, because I'm still rying to figure it out, is to be sure of which, and why, you pick from the selections; whether avg or max, static or total pressure, etc