Pulsatile Flow over an Hexagon - Residuals Convergence Problem

sergiofrsergiofr Member Posts: 3

Hi,

I am new in simulating pulsatile flows and I am having some troubles with convergence.

My problem constist of an hexagon (2D problem) which is inmersed in a pulsating flow determined at the inlet by the formula -> vx = Vamp·sin(w·t) ; vy = 0.0.


At first, the solver uses the initial time step which is set to 1e-6 seconds. When some time steps has ocurred the solver cannot hold the specified residuals and it reachs the maximum iterations per time step which are set to 50. In that moment the residuals (specially of continuity) starts to rise up to something in the range of 1e-1 - 1. I set the iterations limit to 50 because I see that residuals do not drop down more from the value reached in 40 iterations, it gets stuck there. Below, I attach a example snapshot of what I get.


I have done some researching specially removing the hexagon from the domain, and I have seen a pressure wave propagating over the domain due to the change in the velocity at the inlet (the flow is accelerated in positive X direction when the flow comes in and in the other way around when it is going out), so it logic this pressure wave appears. In this case (without hexagon in the domain) the solver converges to the required residuals accuracy.

Although the pressure wave is logic to appear, I suspect it is doing the velocity-pressure coupling to be more difficult to achieve, so I reach the maximum iterations limit. Obviusly, when I put the hexagon inside it is even more dificult to achive, so I have these poor convergence problems.

Furthermore, I have noticed that lowering the CFL, it gets more stable but at the end it goes to same way.

The general setup of Fluent is described below:

  • Solver type: pressure-based
  • Time solver: implicit-1st-order-unsteady
  • Pressure Coupling: PISO
  • Gradient: Least Squared Based
  • Pressure: Second Order
  • Momentum: Second Order Upwind
  • Viscosity Model: Laminar
  • Fluid: Water (Incompressilbe)
  • Time Step -> Adaptive CFL Based -> CFL <= 1.0 (I test with various values: 1.0 - 0.5 - 0.25)
  • Residuals set to -> Continuity = 1e-3 - Velocities = 1e-6

This is a general view of my mesh

And it is a closer look near the hexagon


I know that the problem should be more rough than a constant uniform flow but, I think I am missing something, because the computational time for this "simple" object is quite high...

Any ideas??

Thanks in advance,

Best Regards,

Sergio

Answers

  • sdebsdeb Posts: 200Forum Coordinator

    Hello,

    Since your inlet boundary condition is periodic with time, you will see some residual fluctuations as well.

    Did you try to make use of Coupled Solver for this case?

    Do you see reversed flow from the outlet? If so, then you might consider extending the outlet further downstream.

    Regards,

    SD

Sign In or Register to comment.