Discovery Import

Discovery Import

Internal CFD: Volume flow rate and meshing

    • daniel.trauner
      Subscriber

      Hi,

      I am testing Discovery Live for internal CFD and stumbled upon some questions:

      • I want to measure the in- and outflow in a system. I therefore set multiple charts for volume flow: 1.) using “Inlet or Outlet Faces” as a location type, 2.) using planar surface and selecting the in- and outlet, 3.) using planar surface and choosing a surface at some distance from the in/outlet. All of those charts measure different values for the in- and outlets (all in m³/s): 1.) 0.312 and 0.333, 2.) 0.356 and 0.378, 3.) 0.374 and 3.365. Which of those measurements can I trust? I would guess the ones some distance away from the in/outlet?
      • I recognized some parts of the flow being represented worse with higher fidelity (compared to fluent results). I think the problem is at the inlet, where some edges of the domain get rounded during meshing. Is there a way to force sharp edges?
      • As mentioned above, some edges at inlet faces get rounded. That also means that depending on the fidelity, the inlet area changes. Does that also mean, that when I apply an inlet velocity boundary condition, that the volume flow rate changes with the fidelity or is the volume flow rate somehow preserved?

      Thanks in advance and best regards,

      Daniel

    • Naresh Patre
      Ansys Employee

      Hello dtrauner

      Please find below the answers to your questions:

      I want to measure the in- and outflow in a system. I therefore set multiple charts for volume flow: 1.) using “Inlet or Outlet Faces” as a location type, 2.) using planar surface and selecting the in- and outlet, 3.) using planar surface and choosing a surface at some distance from the in/outlet. All of those charts measure different values for the in- and outlets (all in m³/s): 1.) 0.312 and 0.333, 2.) 0.356 and 0.378, 3.) 0.374 and 3.365. Which of those measurements can I trust? I would guess the ones some distance away from the in/outlet?

      The planar surface location type is typically intended to measure parameters inside the flow domain and should be avoided on inlets and outlets. 

      The main intention of Discovery Live is to rapidly perform simulations on your design changes and understand the trend of the results, rather than the accuracy of the results. For example, Discovery Live will help to understand how the mass flow will change if you change the geometry, or change the setup properties like velocity, etc. The trend of results is what will help to determine which design or setup change would give you better results. Once identified, you can then use high fidelity solvers like Discovery AIM to get more accurate results from your simulation.

      Having said this, I did a quick test on a simple straight pipe and monitored the volumetric flow rate at inlet, outlet and a few regions within the pipe using planar surface option. I could notice small difference between all the values. See below snapshot.

      I would be happy to look at your model if you are able to share it. Can you also tell how much graphics card memory does your machine have?

      I recognized some parts of the flow being represented worse with higher fidelity (compared to fluent results). I think the problem is at the inlet, where some edges of the domain get rounded during meshing. Is there a way to force sharp edges?

      Other than increasing the fidelity, using higher GPU, and reducing the flow domain if possible, there is currently no other option to improve the results.

      As mentioned above, some edges at inlet faces get rounded. That also means that depending on the fidelity, the inlet area changes. Does that also mean, that when I apply an inlet velocity boundary condition, that the volume flow rate changes with the fidelity or is the volume flow rate somehow preserved?

      Increasing the fidelity increases the accuracy of the results. As you increase the fidelity, the geometry is better captured which results in increased accuracy. However, there are some limitations in capturing flow regions within thin sections of the model. I am not sure if your model has thin regions or not. In case it does, then you can check the minimum feature size in your model which the software can capture. The option to display feature size is located in the options panel. Refer below image.

       

      Please let me know if you have any questions.

    • daniel.trauner
      Subscriber

      Hi,

      Thanks for your response. I am aware that this tool is intended to quickly analyse the relative change of certain outputs in response to changes in boundary conditions and that to validate the final design an analysis in AIM/Fluent/CFX is necessary. I am still wondering about the discrepancies in volume flow measurements and the algorithm behind their calculation, especially because the volume flow also changes with fidelity due to changes in mesh/geometry.

      I am currently working with an NVIDIA Quadro M4000 graphics card with 8GB Memory.

      I am unfortunately not sure if I am allowed to exchange files in this forum, but I would be happy to discuss my problem.

    • Naresh Patre
      Ansys Employee

      Hello dtrauner

      I can send you a separate secure transfer link for uploading the file through email. Please let me know if that is ok with you.

    • daniel.trauner
      Subscriber

      Hi,

      thanks, that would be great!

      • Naresh Patre
        Ansys Employee

        dtrauner I have sent you a secure transfer through email.

    • Naresh Patre
      Ansys Employee

      Hello dtrauner

      Thanks for sharing the file. I will check it out and share my findings.

    • Naresh Patre
      Ansys Employee

      Although there is difference between the volumetric flow rate recorded at inlet and outlet (A charts) as well as locations near to them (C charts), the difference seems to be very small (less than 5%) when compared with the V_A_in chart. I think such small difference is acceptable considering the speed at which you get the results. Again to reiterate, Discovery Live is not intended to measure the accuracy of the values, but the trends in the results.

Viewing 6 reply threads
  • You must be logged in to reply to this topic.