diverge with 2nd order upwind scheme for momentum

    • zhengxuanwu

      Dear Ansys Community,

      Hi everyone, i am going on with the simulation of a melting process now, during which a phase change material(liquid phase at the beginning) is heated up by fluid flowing in a u-pipe heat exchanger and begin to melting. The geomrtry is like in the figure

      It is a transient simulation and what i want to study is the heat transfer phenomenon between the fluid and phase change material. for simulation, i used the solidification and melting model and k-omega for turbulent.

      it was all good when i used first order upwind schema for discretization of momentum equation and i managed to simulate the system. However, when i switched it to second order schema for higher accuency. It starts to diverge after several times steps like the following pic

      Then i tried to solve it with the method which i can find online including :

      1. improve the mesh quality. The min. orthogonal quality is now 0.15 and i am not able to improving it any more. 

      2. start the simulation with 1st order schema till a stable converge and then switch to 2nd order  

      3. decreasing the time step size

      4. with coupled solver (used SIMPLE at first).


      But sadly all the methods don´t really help.

      So i am back here and hope anyone can help or give some suggestion.


      Best  Regards


    • DrAmine
      Ansys Employee

      1/Try using actual version

      2/Mesh quality is not only orthogonal quality. Check values of skewness, aspect ratio and cell volume change. (increase mesh check verbosity in TUI to 2)

      3/Enable HOTR

      4/Save residuals for post-processing (/solve expert..). Check residuals of the equations to know where they accumulate.

Viewing 1 reply thread
  • You must be logged in to reply to this topic.