Fluids

Fluids

How to simulate particle release for a certain time using fluent ?

    • Lylad
      Subscriber

      Hi, I am creating a simulation of a cow releasing infectious particles in a barn I just wanted to confirm some things.

      I have been using a transient model with the Eulerian multi- phase approach with DPM turned on for particles to be released, I have set the particles start time =0 and stop time = 0.3 to simulate a cough. But then I want to run the simulation at different times e.g 5 mins, 10 mins, 15 mins to compare the difference in particle dispersion over time.

      I wanted to ask how I can do this ? After doing some googling I think that I can change the time in the running calculating tab by changing the number of time steps and time step size (the number of time steps X time step size = time simulation is ran for) but i'm not too sure if someone could confirm this.

      Also, what's the best option in results to look at the exposure level/ dispersion of the infectious particles ?

    • Rob
      Ansys Employee
      The Eulerian multiphase model is for high concentrations of the "other" phase. DPM is for dispersed systems: I suspect you're wanting the latter.
      You can then use a fully transient simulation (computationally expensive) or steady and see where the particles go.
    • Lylad
      Subscriber
      Hi Rob Thank you for your reply, so just to confirm I'm running a simulation with a cow coughing releasing particles in a cow shed (with inlets and outlets and an airflow) then tracking where the particles go therefore I assumed using a multiphase would be best with two phases running at the same time (airflow + particles), but your suggesting to just use the DPM model not the multiphase eulerian model to track the particles ?
      I just had another question, I want to simulate a cough then track the particles produced over time, but I want to run the simulation at different times e.g 5 mins, 10 mins to look at the difference, is there a way to do this on ANSYS ? (is it number of time steps X time step size= time simulation is running for)
      Thank you for your help and I apologise for all the questions :)
    • DrAmine
      Ansys Employee
      Yes that it the number of time steps * time step size (in general). For particles you are correct: Injection duration and start are steered thr start and stop among other factors which I do not want to explain here :)
    • Rob
      Ansys Employee
      Yes, strictly the DPM model is multiphase, we just track the particles (parcels) and assume they don't hit each other. Have a look in the User's and Theory Guides for the background.
    • Lylad
      Subscriber
      Thanks for answering guys I'll try it out that way, you've been really helpful :)
      I have tried with the transient model and your right the computing time is very long while running the calculation, is there a way to know when the simulation has converged when looking at the scaled residuals graph (because in the transient model the iterations seem to be going on for a long time and the calculation is still not complete).
      I have been using the these conditions for the calculation; number of time steps= 10000, time step size= 0.01 and max iterations/ time step= 20.

    • Rob
      Ansys Employee
      I can't comment on the step size as it's model (mesh & speed) dependent. If you consider a steady model, we'd expect convergence in 2-5k iterations. In the above you could easily use 200,000 (200k) iterations to get to 100s of elapsed time. Given buoyant effects and air change rates in a room you'd expect to need over 20-30minutes to get a response to the particle residence time.
      As we track the DPM parcels and include a particle time (age) with the tracks it's possible to use a steady solution to find the particle (parcel) age as it exits of the domain.
    • DrAmine
      Ansys Employee
      In Transient is not only related to convergence but also to real duration of the process. You can assess the convergence of every time step by looking into residuals over iteration and key monitors: if the monitors do not change towards the end of time step then time step is converged: you can there define extra convergence interruption condition to reduce the number of iterations per time step.
      You friend Fluent User's Guide is helpful there!
Viewing 7 reply threads
  • You must be logged in to reply to this topic.