[kiwishall]

I was computing the case with a multiple-phase flow, which consists of liquid air and gas air. The temperature of working condition is very low. So the properties of liquid air and gas are defined myself. But only after 5 steps the solver will suggest the error of overflow. And in the converge figure of heat transfer, the value is up to about 1.0e08. I have tried the common ways to solve the overflow problem, but in this case I think the heat transfer is the basic problem. How can I solve it ? Help me please!

[ghorrocks]

The problem is almost certainly with your material properties. If you run this with a simple material property (maybe ideal gas or some other simple model) does it run successfully?

[kiwishall]

With a simple ideal gas or liquid, it works successfully. And only with the liquid I defined, it can also work well. But combining them, the problem came.

[ghorrocks]

First thing to check is that your material properties are correct.

If you still get the problem then you need to improve the numerical stability of your simulation. This means better mesh quality, smaller time steps, double precision numerics and/or a better initial condition.

[kiwishall]

Thanks ghorrocks. But could you tell me what is the best way to define a compressible gas? How to describe the thermal and dynamic properties correctly?

[ghorrocks]

It depends on what you want the gas to do. You will have to describe what you are modelling and the fluids you intend to use.

[kiwishall]

I was trying to simulate a liquid expander, taking the cavitation into consideration. And that is why I need to define the gas. Of course I have read the tutorial in the help. I define the gas with the expressions and functions of specific heat, viscosity, and the thermal conductivity. So is the liquid. But during the solving, the heat transfer can't converge or I should call it "diverge".

[ghorrocks]

This is a very tricky model to achieve and I would expect big problems with convergence. The massive change in material properties are always hard for numerics to resolve and converge.

To get this working you are going to have to look closely at the details of your simulation. Any simplifications be made? For instance assuming the liquid phase is incompressible, or that the gas phase only undergoes adiabatic temperature changes. These simplifications can help a lot.