check convergence in polyflow

    • Jasmina92


      I modeled with Polyflow a high viscous flow extrusion through a die - after the complete set up I updated the solver-button (in the workbench project) and the calculation started to run. After aprx. one hour the solver ended with the message "The problem has not converged".

      I checked the numerical parameter - settings:

      max. number of iterations: 1200

      convergence test: 0,005

      divergence test: 0,001

      conv. test for species: 0,005

      solver aggressivity: 2

      Unfortunatly polyflow does not write an result file - the instance.clp file doesn't say anything (to me).

      Has anyone an idea what to look for to bring the calculation to convergence?

      thank you very much

    • José Mantovani

      Hello Jasmina, I've not experience with this type of flow simulation. 

      But, in around my knowledge, I would recommend iterating more often and tracking the waste to see if over the iterations they are approaching the specified convergence factor. If this does not happen you may need to change to a transient solution with small time steps.

      If that is not enough, wait for ANSYS employees  to respond and give you more ideas soon.

      I hope this helps.

      Best regards,



    • Jasmina92

      Thank you for your suggestion!

      yes, I already adapted the max. number of iterations, (and so does polyflow by default when the calculation does not converge - by about 20% I think). But with this number of iterations the calculation needs more than an our with a quite powerful pc - so i thought there must be a deeper problem (the same material model with CFX needs just a few minutes to converge). It is a komplex high viscous shear-thinning fluid.

      I think it is not possible in polyflow to track anything while calculating...? you just need to wait and see if it works.

      But i try to change to a transient solution - I hope this works better

      Again, thank you! Best regards


    • Jasmina92

      Maybe someone who may be as well a beguinner with Polyflow might be interested in an important point that I was missing here:

      I figuered out, that I missed to choose an Evolution-problem. By activating the EVOL-Button you can set an iteration procedure for the "critical parameter" that makes the problem hightly nonlionear and a convergence problematic.

      Still, I started with a higher value for the critical parameter in my calculation (the power law index n in the material data: Herschel-Bulkley fluid) than the actual n. This helps for reaching convergence, that setup did work.

      Now I want to use the previous sulution output for any next calculation, but I am recieving an Error-Message. "Cannot load the output file" or something like that. I checked the path, its alright. (output file-> .res)

      Has anyone an idea how to start the calculation from a previos solution? (I read the manual and couldn't find the answer)

      I enabled the "polyflow output file", but there are several in the output file after calculation (same data size) - which one to choose here for the next calculation?

      thank you for some hint.

    • Rob
      Ansys Employee

      Polyflow uses a fundamentally different solver to Fluent and CFX, and is designed for extrusion modelling and fluids with visco-elastic properties and it's the numerical differences that make it take longer. If you're "just" needing to model a fluid with HB viscosity I'd suggest sticking with one of the CFD codes: my personal preference is Fluent, but CFX is equally capable. 

    • Jasmina92

      Yes, thank you, I modelled the same problem in CFX, but the results (massflow out) didn't match the experimental messurements at all. The paste is very complex (viscoelastic, high viscous jelly-like, shear thinning), so I was recommanded to use polyflow.

      Now I try to use the "real" material parameters and start the calculation again (with no previous solution). See if it will converge..


    • Rob
      Ansys Employee

      Given that I'd agree that Polyflow is probably the correct tool. We're expanding staff usage on here, but we also need to wait for the Polyflow group to join in, it's not a tool I know how to use. 

Viewing 6 reply threads
  • You must be logged in to reply to this topic.