Thermal simulation of a long narrow conductor

General Questions
1

Hello everyone.
Usually I can thermally simulate any kind of object, very successfully.
But now I am simulating the temperature of a cylinder with base area of 1 mm^2 and height of 1 meter.
It is very thin and long.

The problem is probably in creating the Mesh, which also takes a lot of time and then the simulation does not give the desired results.

What advice can you give me on the creation of the Mesh?

Thanks to all
Giovanni Di Maria

2

1D elements (beam/truss) are the way to go in such cases. They have temperature DOF too. But they are not supported in PrePoMax yet so you would have to prepare such an analysis in another way. But you can open the results in PrePoMax if the output is set to 3D.

3

Okay.
Thank you very much for you help.
Giovanni

4

Anyway, I also solved it by the following ways:

  1. I exported the 3D model to .STEP and not to .STL.
  2. I used a cubic shape instead of a cylindrical shape.
  3. I used a Mesh that was not too dense.

GIovanni

5

Exactly, I do wish 1D elements were included. Large structure can’t be modeled in 3d, and surfaces can work but there are times when beams would be better (and much faster).

When I got to Boeing many years ago I was surprised to see that the global loads were generated with a Patran model using only 1D and 2D elements but it really makes sense once you understand the scale of that sort of project. They run over 1000 load cases and even if there were only a few there is no practical way to create a valid model of a full airframe with 3D elements and expect it to run this century.

6

¿Are you running steady state or transient?

7

Hello ANYS
I don’t know who you are referring to.
I am running “steady state” simulations
Thank you
Giovanni

8

I was referring to you, sorry. I forgot the @calimero22
I read your request for advice about the mesh creation.

I recall the mesh in transient thermal should not exceed certain size that can be determined from the time step and material propertie. If you are doing Steady State no problem.

9

Yes, there’s that formula recommended for Abaqus analyses, used to choose the initial increment based on mesh size and material properties:

Δt >= ((ρc)/(6k))*(Δl^2)

where: ρ - density, c - specific heat, k - thermal conductivity, Δl - the distance between the nodes of the element close to the surface with the largest temperature gradient.

10

this may applicable to linear elastic analysis only, i’m not sure for nonlinear analysis using 1D beam and 2D shell is faster than solid since it depend on number of integration point in fiber beam or layered shell element.

i found many reporting (Code_Aster and OpenSees) even 1D/2D element used, still the computational resource are huge, convergence problems and hard to finish.