## General Mechanical

#### Convection question!

• jvardaman
Subscriber

Hello all,

In ANSYS Mechanical, I am modeling a simple plastic pipe, as shown below. The pipe is set to a uniform temperature of 100C and cooled down. The temperature vs. time cool-down curve for the model does not match very well with my controlled, experimental results. In the model, the pipe is modeled as standing vertical. I then apply convection to all surfaces of the pipe (the experimental measurement is from a suspended pipe). I also provide the program with a look-up table for the heat transfer coefficients, as determined from a heat transfer equation. The big question is: What am I missing? Does Mechanical model the flow of air to simulate the chimney effect? Any help is greatly appreciated!!
-JP

• peteroznewman
Subscriber

Mechanical Transient Thermal does not model the flow of air up the center of the vertical pipe. When you apply a Convective BC in Mechanical, it simply applies the convective heat transfer equation using the convective heat transfer coefficient and the bulk fluid temperature. Here is the ANSYS Help entry.

In Mechanical, the inside surface of the pipe gets a single value of tf but in reality, as the air rises in the pipe, it gets warmer but that effect is not included in the model unless special effort was made to try to include it.

ANSYS has Fluent could build a combined Heat Transfer and CFD model of the pipe and the air around the pipe. In the Fluent model, the simulation will compute the rising air inside the pipe warming up.  If you assume axisymmetric conditions, this could be a very small model.

I want to build that model to gain some experience with Fluent as I haven't done a Heat Transfer / Natural Convection model before. Are you willing to share the experimental results?  What are the properties of the experiment?

• Pipe length

• Pipe ID

• Pipe OD (or wall thickness)

• Pipe lip geometry (squared cut, small radius, full radius, or provide details)

• Pipe conductivity

• Pipe specific heat capacity

• Pipe Initial Temperature

• Air bulk temperature

• jvardaman
Subscriber

Thank you for your reply. Mechanical works as I suspected. Modeling in Fluent sounds like it could prove promising. The properties of the pipe are as follows:

• Pipe length = 0.17 m

• Pipe ID = 0.048 m

• Pipe OD (or wall thickness) = 0.06 m

• Pipe lip geometry (squared cut, small radius, full radius, or provide details) = square cut

• Pipe conductivity = 0.48 W m^-1 C^-1

• Pipe specific heat capacity = 1900 J kg^-1 C^-1

• Pipe Initial Temperature = 94 C

• Air bulk temperature = 28 C

The pipe material itself is fairly standard high density polyethylene.

• peteroznewman
Subscriber

I forgot to ask for density.  I found in Granta a HDPE that has a Density = 958 kg/m^3. Okay?

• jvardaman
Subscriber

I used 950 kg/m^3, so I would think 958 would not have much of a difference

• peteroznewman
Subscriber

Please tell me where you measured the temperature vs. time and I will plot data from the 2D axisymmetric model.

Also, what time duration have you recorded?  At some point the graph goes flat as the pipe comes to equilibrium.

• peteroznewman
Subscriber

Here is a 2D axisymmetric model (axis is on the left edge) of the temperature after 15 seconds.

• jvardaman
Subscriber

We measured the temperature halfway up the pipe on the outside surface. We have the ambient air at 27C and the pipe at 100C. I think we were measuring about 20 minutes for the pipe to reach about 47C. I am in contact with someone else about getting using Fluent under a research license. I don't have the program yet but am hoping to use it! I have a question: How did you determine the film coefficients (heat transfer coefficients) for the model? In Mechanical the user has to find those value, but I am not sure about them. Thank you for your help!

• peteroznewman
Subscriber

Find a second computer and download the free Student license using the big buttons at the top of this webpage. Don't mix Research and Student licenses on the same computer. After you install ANSYS Student 2019 R2, you can have a copy of the Fluent model I am running.

You don't enter a film coefficient for heat transfer, you just let the fluid motion carry away the heat using conduction from solid to air and natural convection where air has a velocity induced by gravity and the buoyancy of warmer air.

• jvardaman
Subscriber

I have Student 2019 R2 running and am able to open Fluent, this looks like a really cool program. Do you have to set a boundary condition between the object and the fluid flow or does Fluent take that into account already?

• peteroznewman
Subscriber

In this simulation, the midpoint temperature has fallen to 49 C at 20 minutes for an initial temperature of 100 C and an air temperature of 27 C.

Attached is the archive file from 2019 R2.  Use File Open to look at the simulation, which was saved with results. You will see all the boundary conditions. You might need to watch a few Fluent introductory tutorials.  You can see the distribution of temperature along the length of the pipe after 20 minutes. I don't expect that would be created in the Mechanical model with convection BC and a constant film coefficient.

• jvardaman
Subscriber

This seems very accurate to my experimental results. Fascinating, thank you!

• peteroznewman
Subscriber

I'm thinking of turning this into a tutorial video. It would be interesting to overlay your experimental data on the simulation data, especially if you took data out to a longer end time.

You can attach zip files to these posts after you Add Post, so if your data is in Excel, you have to put that in a zip file first.  You can attach .txt files directly.

• peteroznewman
Subscriber
Subscriber

Hi,

I have watched your videos, they are very informative, thank you for putting them up.

I am trying to model a heatsink with natural convection and radiation.

I manged to make the natural convection work based on your videos and posts, but I can't figure out how to add the radiation of the heatsinks surface into the picture.

Do you know, how to do that?

The problem is, that the surface of the heatsink is coupled with the air (as it needs to be) and so I can't add a radiation boundary condition.

• peteroznewman
Subscriber