Fluids

Fluids

The Discrete Phase Domain Drop-down List of the Injection Window

    • Sardar
      Subscriber

      Hi

      Lets assume you have defined a granular phase called "powder".

      In a DDPM model, for an injection of a granular phase, there's a "discrete phase domain" drop-down list in the injections window having this choice of "none" as well as the name of the discrete phase you have ( the "powder" phase in our case).


      I have noticed via monitoring of particles volume that when choosing none, injected particles have zero volume, although with their mass being respected. This is opposed to when "powder" is chosen as the drop-down choice.


      There are a number of occasions where the user's guide talks about how choosing none activates parcel definitions panel for manual defining of parcel specs. However, there seems to be no mention of occasions where using "none" is preferred over choosing the discrete phase already defined in the models window.


      So, which one is used when?

      Any ideas?

    • Rob
      Ansys Employee
      Copied from other thread.

      In the injections window there is a little (probably mostly passed-by one) drop-down titled "discrete phase domain" which includes a "none" choice as well as your already defined granular phase.
      When choosing none as the dp domain, the particles injected are of zero volume, though having non-zero mass. My issue is that I am using ddpm and this does not fit my case I think - primarily because in ddpm, volume fraction is not negligible.
      1. So my first question is about when this none can be used in addpmcase.
      2. Choosing the discrete phase instead of none leaves me with limited options on parcels, rendering their size dependent on the smallest cell volume in the mesh. This results in a high number of particles and super slow iterations.
      What should I do? And what are occasions where either one is preferred over the other?
      And Ansys staffs, if this is mentioned in a course, please let me know which one it is.
    • Rob
      Ansys Employee
      The system has been a little slow and the spam filters have been fairly keen of late.
      You need to think about what the model is of. If you're looking at a dispersed system then DPM is used. For a very much non-dispersed system we favour Eulerian. DDPM sort of sits in the middle and in theory bridges the gap.
      The lecture you want is the Intro lecture in the Fluent Multiphase Course. The slides were updated by for 2021, and I did the recordings for R19. We'll update for 2022 as and when.
    • DrAmine
      Ansys Employee
      Are you talking about parcel Method? What is your question?
    • Sardar
      Subscriber
      Thank you and for reaching out.
      DrAmine, I am talking about Models > Discrete Phase > Injections. Once you define an injection, inside the injection window there is this short drop-down list titled "Discrete Phase Model".

      In the Image below, none is always there; powder is the name of the discrete phase I have already defined as an Eulerian model.


      Rob, I have no doubt that I should be using DDPM as my volume fraction is 20%; So I wouldn't double check my general setup for the simulation.
      My problem is:
      Setting the "Discrete Phase Domain" to "powder" (i.e. the name of my discrete phase) instead of "none", results in Fluent crashing sometimes or otherwise, very slow iterations Why? Because, as User/theory Guide says, choosing discrete phase domain other than none (i.e. powder in my case), fits five parcels into the smallest cell in the mesh. creating very high overall number of parcels, (2e+9 in my case).
      So why do I not choosenone? Because it does not inject particles with volume!! It only keeps their mass in calculations. This is not suitable either, because I am using DDPM (20% volume fraction), thus neglecting the volume of my discrete phase is underestimating their effect.
      If my English is not too rusty yet, I guess my problem is clear. I can not inject effectively! :|
    • Sardar
      Subscriber

    • Rob
      Ansys Employee
      The issue DPM is the particles have no volume. DDPM gets around this by switching to the Eulerian model when volume fraction hits a set value. However, DPM still has no volume so it's not a perfect solution. If the particles are all moving and not packing 20% isn't going to be too bad if you're using 2-way coupling. You need to understand the flow & physics to decide just how far you can bend the guidelines.
    • Sardar
      Subscriber
      I am already using two-way coupling. But now with your explanations, my problem is changed into one of entire domain volume. Since particles have no volume in dpm, should I add their volume (20%) to water volume (80%) or should I only include water volume?
    • Rob
      Ansys Employee
      With DPM you model the domain volume (whatever it is) and assume it's full of the carrier fluid.
    • DrAmine
      Ansys Employee
      If you relating the volume injection to a DDPM phase the parcel method choice should become inactive and does not support any user change. The method is set to constant. The diameter will be calculated by Fluent. Fluent will inject certain number of parcels trying to fulfil the requested mass or volume fraction without going over the packing limit in each cell. It will pick up the minimum cell volume in the region you select and based on that volume it will allow for 5 parcels.
    • Sardar
      Subscriber
      Thanks DrAmine.
      Just a tiny correction: the injection will set diameters based on the smallest cell in the whole domain, and not in the (sub-)domains you have selected for volume injection.

    • DrAmine
      Ansys Employee
      Yes.
Viewing 11 reply threads
  • You must be logged in to reply to this topic.