thermal expansion

General discussion about Elmer
Roland
Posts: 226
Joined: 12 Apr 2018, 11:29
Antispam: Yes

Re: thermal expansion

Post by Roland »

Hi Kevin,
Thanks a lot for this help!
I will look deeper into this and keep you informed.
Can you just explain why you have added following BC's at the end of the le.sif and what this means:
"Boundary Condition 3
Target Nodes(2) = 1 65
Name = "Disp_x_zero"
Displacement 1 = 0
End

Boundary Condition 4
Target Nodes(1) = 15
Name = "Disp_y_zero"
Displacement 2 = 0
End"
Thanks in advance
kevinarden
Posts: 2237
Joined: 25 Jan 2019, 01:28
Antispam: Yes

Re: thermal expansion

Post by kevinarden »

The linear solver does not like for there to be any rigid body motion. Makes it hard to converge. So in addition to the vertical restraint, I target a couple of nodes to keep the body restrained in the X, Y and rotation about the axis.
Roland
Posts: 226
Joined: 12 Apr 2018, 11:29
Antispam: Yes

Re: thermal expansion

Post by Roland »

Hi Kevin (and Peter?),
Concerning this thermal expansion matter , what I want to do is to setup an electromagnetic heating process (see the here attached jpg drawing) with a spiral inductor which should heat and deform a bi-metal (aluminium material for the lower part and isolating material for the upper part of the plate which should bend due to the 2 different thermal expansion coefficients). This whole model couples the electromagnetic physic (say EM), the Heat transfer physic (say HT) and the linear elasticity physic (say LE), in transient state from 0s to 20s. As this whole model (case.sif_EM_HT_LE here attached, in which the EM is disabled for the moment) does not work, I first made a simplified model coupling only HT with LE (case.sif_HT_LE here attached) in which I replaced the Joule heating(coming from EM) heat source in HT by a simple constant heat source (7400 W/kg).
1/After some advices from Peter (like using the quadratic mesh elements for LE (line "Element = p:2" in the LE solver), the simplified model HT_LE works nicely and gives(after 20s giving a temperature of about T = 365 K) a maximum displacement_Z of about 10 mm (which is the same as in Comsol) corresponding to a realistic plate bending.
2/ Then I came back to the whole EM_HT_LE model, but in which I disabled in a first state the EM physic and, like in the simplified model HT_LE, replaced the Joule heat heat source in HT by a simple constant heat source (7400 W/kg), meaning that this EM_HT_LE model is in this state (EM disabled) the same as the simplified HT_LE model.
3/ But unfortunately this EM_HT_LE model (with disabled EM) gives a Z displacement, after 20s (with also a T=365 K reached temperature) , of only around 40 microns, which is completely different from the 10mm of the HT_LE model, when these 2 models should give the same result!
After many investigations and comparisons of the 2 models, I am now out of ideas and don't see why the 2 models, which should be the same, give 2 completely different results for the plate vertical Z displacement !
It would be very kind if somebody could take a look at the 2 attached case.sif files and give me some help elements so that the 2 models give the same results...
Many thanks in advance!
Attachments
case_HT_LE.zip
(1.32 KiB) Downloaded 42 times
case_EM_HT_LE.zip
(2.08 KiB) Downloaded 40 times
spiral_inductor_heating.JPG
spiral_inductor_heating.JPG (100.89 KiB) Viewed 570 times
kevinarden
Posts: 2237
Joined: 25 Jan 2019, 01:28
Antispam: Yes

Re: thermal expansion

Post by kevinarden »

If you can post the mesh files than I can double check the set up.
Roland
Posts: 226
Joined: 12 Apr 2018, 11:29
Antispam: Yes

Re: thermal expansion

Post by Roland »

Hi Kevin,
Here attached are the mesh files (they come from Comsol and are imported in Elmer).
- plate.mphtxt corresponds to the HT_LE simplified model
- spiral.mphtxt corresponds to the whole problem EM_HT_LE
It would be very nice if you can take a look and keep me informed!
Many thanks in advance!
Attachments
spiral.zip
(358.97 KiB) Downloaded 38 times
plate.zip
(7.04 KiB) Downloaded 40 times
kevinarden
Posts: 2237
Joined: 25 Jan 2019, 01:28
Antispam: Yes

Re: thermal expansion

Post by kevinarden »

I interrogated the model you provided and could not find any error in the modeling or sif file. I verified that different results were obtained. I performed tests that indicated that if the additional body is present, even though not included in the equation, impacts the answer of the linear solver.

I then did an independent study of the deflection of a cantilever beam, with just the linear solver. And then repeated the simulation of the cantilever beam in air, where the air is modeled. Even though the air is not included as a body, nor is it included in the equation, its presence changes the answer.

These tests were just with the linear elastic solver.
Roland
Posts: 226
Joined: 12 Apr 2018, 11:29
Antispam: Yes

Re: thermal expansion

Post by Roland »

Hi Kevin,
Thank you for your efforts and your conclusion about the influence of the surrounding media even if it is not activated.
But this puzzles me very much as you can understand ! what kind of additionnal tests could be made to fix this problem?
Can perhaps Peter take a look at it and give his opinion?
What do you think about that?
We keep in touch
Thanks for your efforts!
kevinarden
Posts: 2237
Joined: 25 Jan 2019, 01:28
Antispam: Yes

Re: thermal expansion

Post by kevinarden »

I was wrong about the test case I put together, the linear solver does get the same answer with or without the air present. I accidentally used two different meshes for the beam. With the same mesh, with or without air the answer was the same. I need to test it with heat.
Roland
Posts: 226
Joined: 12 Apr 2018, 11:29
Antispam: Yes

Re: thermal expansion

Post by Roland »

Hi Kevin,
Ok that's good news!
Keep me informed when you have again tested my 2 models with heat ok?
I will also investigate further on my side, since I am convinced that the 2 models should give same result and that there must just be something in the settings which is different and which leads to different results!
Please keep me informed as soon as you have fresh results about that ok? Say that it is very important for me to get the right result in the whole EM_HT_LE model since it will then validate the Elmer coupling ability of EM with HT and LE, which is a coupling we need for our electromagnetic processes modeling...
Thanks again and see you soon!
Rich_B
Posts: 421
Joined: 24 Aug 2009, 20:18

Re: thermal expansion

Post by Rich_B »

Hello,

This may or may not be relevant, but loading each model in ElmerGUI and opening Model -> Summary, shows that the plates are meshed differently, 100 elements versus 81 elements per plate.

HT_LE:
VOLUME BODIES
Body 1: 100 volume elements
Body 2: 100 volume elements
Undetermined: 0
Total: 2 volume bodies

EM_HT_LE:
VOLUME BODIES
Body 1: 28414 volume elements
Body 2: 81 volume elements
Body 3: 81 volume elements
Undetermined: 0
Total: 3 volume bodies

Rich.
Post Reply