Inventor Nastran Thermal Conductance

[daniel_luescher]

Hallo everyone,

I want to make a little example about thermal conductance, see image, with two bodies connected by a a bonded connection, and temperature boundary condition at the two ends of the assembly (400°C and 300°C).

I got a result like the following:

BUT: doing the calculation by hand I got a heat flux = 1'430 W/m2, while Inventor Nastran gave me a result q = 1'064 W/m2. This a big discrepance.

Also. the heat flux seems very "fuzzy" at the contact.

Are there some settings I missed?

See attachment for the calculation setup and the material data.

Thanks, Daniel

[Anonymous]

Hello Daniel,

I was just passing through the forum in my spare time and saw your post.

I decided to run a quick test of my own. I couldn't get the FEA setting from your attached models. So, I have created my own.

The setup is shown in Attachment 1. The FEA result is shown in Attachment 2. My analytical hand calculation is shown in Attachment 3, which agrees with the FEA result.

I have used two different materials to make it interesting. The FEA result gets patchy if I reduce the contour range. I think this is just showing small variances in the heat flux averaging.

There is also something called contact resistance between surfaces. I have not accounted for it. But, regardless, the sample probe results are close to the mark.

Regards,
R Rai

[daniel_luescher]

Hi  Roshan,

thanks for you reply and sorry for my late answer (usually I get an email when someone reacts to a post, but not this time...).

I am explicitely interested in a simulation with thermal contact conductance resistance. Since the results I got are not understandable I use the same thermal conductance for both bodies, because of semplicity reasons.

I attached my hand calculation of a thermal flow through two bodies separated by a contact, the same example I calculated with Inventor Nastran.

I attached two images of the Inventor Nastran result:

The first is the SOLID TOTAL-HEAT FLUX (Since in this case the heat flux is 1-dimensional I displayed the total heat-flux, wich is almost equal to SOLID Z-HEAT FLUX).

The second is the temperature distribution in z-direction.

Summary:

                | hand calulation | Inventor Nastran

---------------------------------------------------

heat flux | 1'430 W/m2       | 1'060 W/m2

Tc1           | 385.7 °K            | 389 °K

Tc2          | 314.3 °K             | 311 °K

[daniel_luescher]

Hi Roshan,

thanks for you reply and sorry for my late answer (usually I get an email when someone reacts to a post, but not this time...).

I am explicitely interested in a simulation with thermal contact conductance resistance. Since the results I got are not understandable I use the same thermal conductance for both bodies, because of semplicity reasons.

I attached my hand calculation of a thermal flow through two bodies separated by a contact, the same example I calculated with Inventor Nastran.

I attached two images of the Inventor Nastran result:

The first is the SOLID TOTAL-HEAT FLUX (Since in this case the heat flux is 1-dimensional I displayed the total heat-flux, wich is almost equal to SOLID Z-HEAT FLUX).

The second is the temperature distribution in z-direction.

Summary:

                 | hand calulation | Inventor Nastran

---------------------------------------------------

heat flux | 1'430 W/m2         | 1'060 W/m2

Tc1           | 385.7 °K              | 389 °K

Tc2           | 314.3 °K              | 311 °K

[John_Holtz]

Hi @daniel_luescher 

I have not done too much work with the contact resistance (or conductance), but it looks like there are a number of effects occurring in your analysis:

• The variation at the contact face is due to two random meshes being in contact.
• The total heat flow (or the temperature gradient across the contact, both related) is dependent on the contact stiffness. I think you would not see this effect in most models, but it becomes evident in your model because the contact resistance is large compared to the resistance of the material itself.  (just a guess)

In other words, if the contact resistance is 0, the temperature difference should be 0. Mathematically, the result probably shows some non-zero temperature difference. (It is as if the contact between the unmatch mesh creates a small resistance.) When a contact resistance is used, the effect of the unmatched mesh is magnified.

Now the question is what does all of this mean in a model such as yours. I will have to think about that.

[daniel_luescher]

Hi John,

thank you for your reply.

I noticed that decreasing the contact stiffness some kind of decreasing of the conductivity (that is, the temperature gap increases) occours.

Since the contact stiffness plays a role in the spatial domain (like penetration), the stiffness should not play a role in heat conductance?