[Bdew8556]

Hi guys,

Got a few questions confusing me.

So I see that the time step for solids is much larger than for gases, leading one to think that's because it takes longer for the solid to heat up and cool down than the gas.
However air has a lower thermal conductivity than solids like metal.
Could anyone set me straight on the theory?

B

[FliegenderZirkus]

The key difference between air and the solids is that air can flow, which allows for very quick mixing of the hot and cold areas. If you can somehow make sure that the air is perfectly still, then you're right that in that setup it is a great insulator. This is the idea behind double-pane windows. But when talking about heat-up times, you should also be concerned about specific heat, perhaps even more so than the conductivity. In any case, typically the choice of the time step needs to take other factors into account, such as the flow field itself. Depending on what you want to capture in terms of unsteady flow field, this may require much shorter time steps than the energy equation.

[Gerry Kan]

Dear Bdew:

From my experience CHT is usually a steady-state affair, but your application might be different. If it is a transient run, the time steps must align between the solid and fluid domains.

With this in mind, the term "time scale" is not really used to indicate different transient behaviors (i.e., different time step sizes) between the solid and fluid domains, but a way to reconcile the non-uniform rates of convergence. This is because thermal diffusion in solids reacts much more slowly than in fluids, mainly due to the additional convection. So one would need to solve heat diffusion equation for the solid phase much more frequently to reach convergence than, say, the fluid phase. Hence the "time scale" for the solid domain is "higher" than the "time scale" for the fluid domain.

From the CHT work I have done before (my memory is a little spotty these days), the solid "time scale" usually about 20 to 50 times higher, meaning that for each iteration of the fluid domain solver, the solid domain solver will be called corresponding many times. However, the physics of solid heat conduction is much simpler than the fluid convection, and you don't have to deal with additional effects like cavitation and two-phase flows. So in the end the additional solid solver calls still wouldn't impact the overall solver performance too much.

Hope that helps, Gerry.

[Bdew8556]

Thanks guys,

I think you've answered my question. Essentially the part I was missing is that in a fluid there is convection and that is what allows the fluid to heat up much faster than solids, hence the need for a smaller time step. Then you of course need to consider things like how the fluid is moving.

[Gerry Kan]

You got it backwards. Within the same time step, the solid domain solver needs to iterate more often than the fluid domain solver to reach convergence for that time step. The "time scale" has nothing to do with the time stepping itself. Gerry.

[y896li]

Dear Gerry,

I'm working on a CHT problem and the residual for H-energy converges slowly, same with H-energy imbalance. Other equations for mass, momentum, and T-energy are all well converged. I'm wondering if there is a way to control the timescale for H-energy in fluid domain only without changing the timescale for fluid mass and momentum? I know there are two ways to control the timescales, one way is to enable Beta feature but this way cannot control the timescale for different equations seperately. Another way is to specify the equation class in solver control, however this would also change the timescale for solid domain. But both method cannot control timescale for fluid energy individually.

[johnkhalil91]

Originally Posted by Gerry Kan

Dear Bdew:

From my experience CHT is usually a steady-state affair, but your application might be different. If it is a transient run, the time steps must align between the solid and fluid domains.

With this in mind, the term "time scale" is not really used to indicate different transient behaviors (i.e., different time step sizes) between the solid and fluid domains, but a way to reconcile the non-uniform rates of convergence. This is because thermal diffusion in solids reacts much more slowly than in fluids, mainly due to the additional convection. So one would need to solve heat diffusion equation for the solid phase much more frequently to reach convergence than, say, the fluid phase. Hence the "time scale" for the solid domain is "higher" than the "time scale" for the fluid domain.

From the CHT work I have done before (my memory is a little spotty these days), the solid "time scale" usually about 20 to 50 times higher, meaning that for each iteration of the fluid domain solver, the solid domain solver will be called corresponding many times. However, the physics of solid heat conduction is much simpler than the fluid convection, and you don't have to deal with additional effects like cavitation and two-phase flows. So in the end the additional solid solver calls still wouldn't impact the overall solver performance too much.

Hope that helps, Gerry.

Hello Mr. Gerry,

I have a question please, and it is urgent.
I have to make a transient CFD simulation using Ansys Fluent to render the temperature profile inside solid spheres that are set in a tank and heated with a hot steam. As the time-scales are very far for conduction and convection, it is not practical to use small time step for the two zones. However, is it physically correct to use a solid-time step that is 20-50 higher than that for fluid?

Thank you in advance,
John

[Gerry Kan]

Dear John (and y896li):

As I mentioned previously, the concept of "time scale" is to address the numerical issue that the solid (conduction) domain converges much more slowly than the fluid (convection) domain. It has nothing to do with the physical time required for the two domains to reach thermal equilibrium.

I did not come up with the term "time scale" but it is used in many commercial codes. That having said, the "time step" for the solid domain should, as reciprocal of the time scale, much smaller than that of the fluid domain.

Gerry.