Divergence issues with higher Lee coefficient for cryogenic chill-down problems

    • Jayachandran Narayanan


      We are trying to simulate the chilldown process of liquid nitrogen through a pipe. The simulation includes both fluid and solid regions (pipe dimensions). Mixture model along with dispersed option has been selected for modeling the multiphase flow. We are trying to simulate the film boiling phase of the cryogenic chilldown process by adjusting the Lee coefficients. It has been observed that Lee coefficient value of around 1000 with first order accurate time and space schemes provides the closest estimate of the solid temperature distribution. But, due to the diffusive nature of the first order schemes, we need second order time and space stepping, but the simulations run smoothly upto Lee coefficent of around 100 and diverges above values of 100 with the second order schemes. The second order simulations are running around 1.5-2s (10^-4s time step size) and then diverging. Hence, the use of first order schemes for stabilising the simulation may not work for our case. Kindly help to resolve this issue. 

    • Rob
      Ansys Employee

      The coefficients are a frequency unit so try with a smaller time step: you may be moving mass too quickly for the current time step. 

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