Fluids

Fluids

DPM problem with system coupling

    • Luigi0
      Subscriber
      Hi,I am doing fluid structure analysis with system coupling to study acoustophoretic phenomena. A piezoelectric layer vibrates to generate acoustic waves that interact with dpm phase in a fluid. nMy analysis is divide in 2 parts:at first the piezo is off and the particles + fluid enter in the channel and the analysis is stopped. Then it is restarted (from the last timestep of the prevoius analysis) switching on the piezo and changing the timestep.nAfter the first part, I checked if the particles enter in the channel, so I export the file from fluent to CFD post, and it was ok. At the end of the second part, I made the same thing (export file from fluent to cfd post) but the trajectories are completely different with respect to the first part of the analysis and I don't know why. Any help?n(Interaction with continuum phase is not activated and the tracking is steady)n
    • Rob
      Ansys Employee
      Please can you post some images? n
    • Luigi0
      Subscriber
      nnIn the first image the trajectories referred to the first part of analysis are shown. Particles enter from the inlet (on the left) and go to the right. (The flow is laminar)nIn the second image the trajectories of the particles referred to the second part of the analysis. My problem is related to the fact that also the initial trajectories change, but the analysis is restarted from last time-step of the previous analysis part, and not from the beginning.nnn
    • Rob
      Ansys Employee
      Can you plot velocity vectors and contours of y velocity on the flow field. n
    • Luigi0
      Subscriber
      Yes, the first picture shows the contour plot of Y velocity at end of the first part of the analysis.nThe second image shows the the contour plot of Y velocity at end of the second part of the analysis.nThe third imageshows the the vector plot of Y velocity at end of the first part of the analysis.nThe fourth image shows the the vector plot of Y velocity at end of the second part of the analysis.nSorry for the different post processing method : the first two are obtained with CFD post, the last two with fluent post processing. (In CFD post I am not able to plot the component Y of velocity as a vector plot).n
    • Rob
      Ansys Employee
      Looking at the y velocity it's up in the lower half and down in the upper half of the domain in the second set of results. With that information what would that do to the particle positions? In the first result y component is zero or near enough zero. n
    • Luigi0
      Subscriber
      The second set of results refer to the second part of the analysis, when the piezo is on, so there is generation of acoustic waves in the fluid, and the trajectories of the particles should be deviated. I expect that particles move to the centre of the channel.nanalysis from 0 to 0.03s --> piezo is off, particles enter in the domain and the trajectories should be straight more or less.nanalysis from 0.03 to t-final --> piezo is on, generation of pressure waves in the fluid, particles trajectories should be deviated.nI should see trajectories deviations only when the piezo is on. Please let me know if I was clear. Thanksn
    • Rob
      Ansys Employee
      The second DPM image is before the Fluent solver has done anything? If not isn't the image what you're expecting to see? n
    • Luigi0
      Subscriber
      The second DPM image is at the end of the analysis (t = t-final). I expect a situation like the sketch below:nAfter 0.03 s particles should concentrate in the center, but the trajectories before 0.03 s should remain straight as in the first DPM image that I sent. Maybe I'm confused concerning dpm tracking or representation: I inject 10 particles (parcels), one per cell (I have 10 cells on the inlet), so the remaining that are present in the DPM's images that I post are not physical particles but they only represent the trajectory, am I alright?n
    • Luigi0
      Subscriber
      Sorry do you have any further advices?n
    • Rob
      Ansys Employee
      The particles you see represent the parcel location, so are where the particle would be. If the whole flow field is affected by the new velocity field the particles will behave as per the velocity field: it'll take a bit of time hence the particles taking some time to contract towards the middle. The particle tracks look about right for the flow field, so I think you're getting an explainable result. n
    • Luigi0
      Subscriber
      Ok, but i don't understand why if I plot particles trajectories from 0 to 0.03 s (when the piezo is off) , they are deviated. In steady tracking are the particles continuosly injected from 0 s to the end time of the analysis or not?n
    • Rob
      Ansys Employee
      If you're turning the piezo on and off with time isn't the solution transient? n
    • Luigi0
      Subscriber
      Yes , the solution is transient.n
    • Rob
      Ansys Employee
      So particles flow into the system with a forward speed, and it takes a certain distance for them to cluster in the middle due to the y velocity component. If you take the y velocity how long will it take for a particle to reach the centre line; using the x component how far into the domain will that be? n
    • Luigi0
      Subscriber
      The height of the channel is 150 micrometers, the length is 6 mm. The particles are injected at time= 0 s with velocity (X component) = 0.11m/s.nFrom 0 to 0.03s piezo is off and particles position at t=0.03s is = velocityX * 0.03 s = 3.3 mm (along X).nThen the piezo is turned on, pressure waves are generated in the fluid and the particles should be deviated. Considering, for example, a particle that is injected at position (X=0 m; Y= 30 micrometer), in order to reach the centre line it has to travel 45 micrometers along Y. nSo, considering the velocity Y plot that I have post above, particle's Y velocity is 0.01 m/s (green/yellow color). The time for the particle to reach the centre is t = 0.0045 s. The final position along X of that particle should be X = 3.795 mmn
    • Luigi0
      Subscriber
      I forgot to say that when I turn on the piezo, I change the time step to 1e-8 s, because the actuation frequency is 5 MHz. If I want to simulate 0.0045 s , I need 450000 timesteps but this is a too high number.n
    • Luigi0
      Subscriber
      maybe should I switch to unsteady tracking, instead of steady?n
    • Rob
      Ansys Employee
      If the solver is unsteady the particles are already unsteady. The option in the DPM panel just allows you to use a different time step for the injection. nWithout the files I can't tell if there's something wrong, and that's not permitted (US Export rules). Go through the whole set up with your supervisor and/or an RA who's good with the software & physics. They may well spot something or ask a question that picks up if there is something missing. n
Viewing 18 reply threads
  • You must be logged in to reply to this topic.