Fluids

Multiphase problem: Divergence detected in AMG solver

• Marco Alborghetti
Subscriber

Good morning everyone,

I'm trying to simulate (transient) a multiphase problem (evaporation) of hydrogen from liquid to gas in a circular pipe with a heated wall. I'm using the Peng-Robinson model in order to have a more realistic simulation, and so I have defined the parameters taken from a file in the Ansys installation folder called "MATERIALS-pengrob.ccl", in particular this file provide the settings for the properties and the coefficients for the cp that provide me the curve you can see in fig. 1 and 2. Over this i set the solution scheme as shown in figure 3 (I've already try every kind of model, included PISO), the problem is that when I go to run the simulation I obtain the errors reported in figure 4 and 5.

Is anyone able to tell me what kind of error is this and why it happens? And mostly important, anyone is able to help me in solving this problem?

Kind Regards,

Marco

• Marco Alborghetti
Subscriber

I forgot the last pic

• Rob
Ansys Employee

Check boundary conditions relative to the various evaporation conditions. What time step are you using?

• Marco Alborghetti
Subscriber

I set a velocity inlet, pressure outlet, no slip wall and heated wall and here you can see the settings of the time step

• Rob
Ansys Employee

OK, and what is the domain temperature relative to the boiling temperature? Getting some of these models going can be a little tricky if you're not careful with the initial conditions.

• Marco Alborghetti
Subscriber

the triple point is around 33 k and the wall is heated at 900 k

• Rob
Ansys Employee

And the fluid temperature at initial condition? There's a good chance you're flashing the liquid so try with a smaller time step. Also check things like cell quality and mesh resolution.

• Marco Alborghetti
Subscriber

the fluid start at 25 k, the mesh quality is very good as the domain is a cylinder, i tried to launch a simulation with a constant cp and all the properties set as above and now it's running, but with a polynomial cp i obtain the problem of before, is there any database for this coefficient?

• Rob
Ansys Employee

Can you replot the properties to include the full range you're modelling over. If phase change is very rapid, or there are dramatic changes in material properties over a small temperature range you may need a very small time step, or a very carefully created initial condition.

• Marco Alborghetti
Subscriber

unfortunately i've changed the properties now, i tried to calculate the cp with coolprop in matlab and than i provided some point with a linearpiecewise

• Rob
Ansys Employee

And if you plot the function you put into Fluent?