Skip to content

Mass conservation in turbulent pipe flow

Due to recent bot attacks we have chanced the sign-up process. If you want to participate in our forum, first register on this website and then send a message via our contact form.

› Forums › on OpenLB › General Topics › Mass conservation in turbulent pipe flow

Viewing 6 posts - 1 through 6 (of 6 total)
  • Author
    Posts
  • October 27, 2025 at 3:15 am #10893
    nipinl
    Participant

    I’ve slightly modified the geometry of turbulence/nozzle3d example to simulate LES and DNS of turbulent flow through smooth pipe. All boundary conditions are retained the same. While comparing the results of DNS with LES(RLB) I observed some issue with the mass conservation for DNS. The integral of (density * U.n) at upstream of outlet is found to be half of the inlet value (Result) . LES results look good.
    The cpp file : cpp file
    params file:params file

    I had observed excessive diffusion in my earlier attempts for DNS set-up for channel flow. I was wondering if this might be the cause. Details of the simulation: non-periodic, inlet:VortexMethodTurbulentVelocityBoundary, outlet: pressure = 0, dynamics: D3Q27 BGK, walls: bounce back.
    Any suggestion on improving the model is greatly appreciated.

    Best,
    Nipin

    October 27, 2025 at 1:59 pm #10897
    FBukreev
    Keymaster

    Dear Nipin,

    please take for DSN the ThirdOrderRLBdynamics, the standard RLB dynamics doesnt work well.

    And for the vortex method, please use it always inside one cuboid, because it is not parallelized. You can make inlet inside of one cuboid and the rest can be separated as you wish.

    Best,
    Fedor

    October 27, 2025 at 2:26 pm #10898
    nipinl
    Participant

    Hi Fedor,
    Thank you for the prompt reply. For DNS, I use BGK with D3Q27 descriptor. I was using RLB for only LES, which gives good results.Thanks for the suggestion regarding vortex inlet, I’ll use that. However, it is the significant lack of mass conservation in DNS (BGK) that worries me the most. This makes the simulation not useful.

    Best,
    Nipin

    October 27, 2025 at 2:31 pm #10899
    FBukreev
    Keymaster

    You can try to do DNS with ThirdOrderRLBdynamics and D3Q19. That should give correct mass.

    Actually, BGK with D3Q27 should also work, I dont see any reasons why it can loose mass.

    October 27, 2025 at 2:38 pm #10900
    nipinl
    Participant

    Hi Fedor,
    Thank you for the suggestion. I will try DNS with ThirdOrderRLBdynamics and D3Q19. I’ll update how it went.
    Thanks again!!

    Best,
    Nipin

    October 29, 2025 at 10:00 pm #10905
    nipinl
    Participant

    Hi Fedor,

    I ran the pipe flow DNS using ThirdOrderRLBdynamics with D3Q19. It ran successfully for some time but eventually went to “nan” as the average density exceeded around 1.038.

    However, when I used ThirdOrderRLBdynamics for my channel flow case, the re-laminarization issue was resolved, which was my main concern. I had originally attempted the nozzle case to address this issue.

    Thank you for suggesting this!

    Best,
    Nipin

  • Author
    Posts
Viewing 6 posts - 1 through 6 (of 6 total)
  • You must be logged in to reply to this topic.