## Fluids

#### How to model convective heat transfer on a two-sided wall

• dodonn4
Subscriber

Hi everyone.

I am working on a simulation which involves a cold volume of liquid in a plastic container being placed in an ambient air environment, and the goal is to determine when the center of the liquid heats up to a certain degree. The plastic container is very thin, and I modeled it as a shell and applied shell conduction to it. Otherwise, the only method of heat transfer is by natural convection.

My set-up involves a larger enclosure around the bag of liquid, which represents the ambient air. The volume of liquid is completely enclosed by the plastic container, which is an elliptical cylinder.

I am having trouble figuring out how to properly apply convection to the volume of liquid. I have tried applying a convection boundary condition to the outer walls of the ambient environment, but the results remain the same no matter what I set the heat transfer coefficient as (I tried anywhere from 0.5 to 500 w/m^2 K). Since the plastic container is a two-sided wall I cannot apply a convective boundary condition to it, so I have it set as a coupled wall.

Another method I tried was by applying a temperature boundary condition to the walls of the ambient air enclosure, then using the "Patch" feature (in the initialization step) to set the enclosure at a constant temperature. In the material properties for the liquid that is inside the plastic container, I set the density as "boussinesq" since the overall temperature change I am dealing with is small (ambient environment is 23C, volume of liquid starts at 5C). I am unsure if this method actually simulates convection or not. Or, is there a better way to go about doing this?

It is important to note that a physical experiment was performed, and my simulation is about 25-30% off from the physical experiment.

• Rob
Ansys Employee
For a coupled wall Fluent will calculate the heat transfer on both sides, so there is no HTC etc to set. All you set is the material & thickness on one of the wall & wall:shadow pair.
Depending on the set up you'll set outer boundaries (for flow etc) and set the fluid zone material. You may then need to patch in a temperature etc. Assuming the two fluids don't contact each other you don't need a multiphase model.
• dodonn4
Subscriber
There are two fluids in contact inside the plastic container. The plastic container is not 100% filled with a liquid, it also has some air at the top of it. The entire plastic container and its contents are just warming up with natural convection, and the plastic container remains stationary. The overall difference in temperature is only 18C, so there won't be significant changes in densities or convective flow currents in the volume of liquid.
Do I need to use the multiphase model in this case? Or is there another way to properly simulate those two fluids in contact, such as imposing an "imaginary" wall with zero thickness between the liquid and the air inside the plastic container?
• Rob
Ansys Employee
Yes, you'll need a multiphase model, VOF in this case. Also as the volume is sealed are you looking at buoyant effects in the air too?
• dodonn4
Subscriber
Yes, I would like to consider buoyant effects as well, even though they may be small. Currently I have the material properties of the air density set as ÔÇ£ideal incompressible gasÔÇØ, is this correct?
• Rob
Ansys Employee
As it's a sealed volume you'll need ideal gas as the density change means the gas may expand/contract and that won't work well with a fixed bag.
• dodonn4
Subscriber
I see, so I will make sure to change that property of air. How would I go about making sure the bag is not a fixed size and allows for the expansion of the air inside?
I have not run into this issue before, so if you could give me some tips or some documentation it would be greatly appreciated.
thank you.
• Rob
Ansys Employee
If you want the bag to flex you'll need FSI. The issue only occurs if the density of the fluid varies with temperature only in a fixed volume as the extra "volume" due to expansion has to go somewhere.