Fluids

Fluids

2 type of particles, coal particle + biomass particle, diverged in cold flow

    • ll00023
      Subscriber

      I have set up a combustion case with two types of particle, coal and biomass.

      the mesh for the case is proved to be good for coal particles in LES simulation.

      now I add biomass particles. 

      I run cold flow without particles; it converged quite well.

      after injecting particles, it just diverged.

      kw-sst model is used. steady RANS case, 3D model. 

    • Rob
      Ansys Employee

       

      So, it’s OK for coal combustion but not biomass?  If you look at the flow field just before it fails what’s happening? Also is "converged quite well" saying it converged or it nearly converged? Just checking my understanding. 

       

    • ll00023
      Subscriber

      If I injected both particles, the velocity field will be wrong.

      now I injected coal particle first, it didn't converge well, but not diverge too.

      I just added biomass particles and increased the DPM iteration interval from 300 to 500, to make it converge better.

      Thank you for your reply.

    • ll00023
      Subscriber

      kw-sst model is used. steady RANS case, 3D model, mesh has been validated.

      First cold flow without particles is converged. 

      Then two type of particles are injected;  DPM iteration interval 1000 still does not work.

      if just injecting coal particles, cold flow with particle will converge.

      biomass particle's density is half of coal particle's density.

      I will try to only inject biomass particles to test whether it converges.

      converged cold flow without particle, as expected

      diverged cold flow with two types of particles, 

       

    • Rob
      Ansys Employee

      There's nothing obvious from the above. The poor convergence with particles on suggests a very high particle loading somewhere, or some weird DPM source terms. Turn on hi-res tracking, and in the DPM Numerics tab turn on the Linearise Source terms and Node based averaging. The Theory Guide will explain what they do. 

    • Rob
      Ansys Employee

      As an aside, how many particle streams are there, and are you using stochastic tracking?

    • ll00023
      Subscriber

      2 particle streams. coal and biomass.

      I use discrete random walk model.

      I will test the high-resolution tracking.

      I can only use Node Based Averaging or Linearize Source term.

      Thank you for the suggestion Rob.

    • Rob
      Ansys Employee

      Increase the number of tries, I suspect you've got too much DPM source in one place. Read up on the model & parcel theory. 

    • ll00023
      Subscriber

      after I switch second order upwind of pressure to PRESTO!.  the simulation is stable

    • Rob
      Ansys Employee

      OK, good to hear. 

Viewing 9 reply threads
  • You must be logged in to reply to this topic.