Multiphase Flows

Multiphase Flows

EWF model fails with floating point overflow I am running an EWF model simulation that is very simple. There is no DPM splashing and only one-way coupling between the film flow and the fluid flow. The run fails almost immediately with floating point overflow even though I am using very reasonable timesteps for the fluid.

    • FAQFAQ
      Participant

      The immediate failure of an EWF simulation is almost always due to non-convergence of the EWF solver. The EWF model has an option to post its solution residuals to the Fluent message window If you note these, they will usually be very high (1.0e+9 or higher) right before the failure. You can switch to the variable time-stepping mode for the EWF model to let a better sense of what the timesteps should be, but often this also allows the timestep to increase to overly large values. If this is the case, set a constant small timestep for the EWF solution (1.0e-06 s, ie:) and reduce the fluid timestep by one order of magnitude or more. The fluid and the film flow may not be coupled, but the film model has to adjust every time the fluid solution is updated, If the change in the fluid solution is too large, the EWF model will not undergo enough residual reduction to properly recover, and significant solution error will start to accumulate because of the transient nature of the film solution.