Non-linear material

Discussion about building and installing Elmer
raback
Site Admin
Posts: 4060
Joined: 22 Aug 2009, 11:57
Antispam: Yes
Location: Espoo, Finland
Contact:

Re: Non-linear material

Post by raback »

Anil,

Great pointer to WSL! And nice to have you back Anil after some time!

We have been using VirtualBox for the same purpose and this virtual machine:
https://www.nic.funet.fi/pub/sci/physic ... lMachines/
I guess WSL meets the same need.

Have you been using WSL yourself? It would seem that the Ubuntu 20.04 version + launchpad would be a straight-forward way to go if one wants to avoid compiling the whole thing.

-Peter
Sayan
Posts: 92
Joined: 06 Dec 2020, 12:44
Antispam: Yes

Re: Non-linear material

Post by Sayan »

Hi,,
Can anyone tell me if I use bi-linear material and define 8 constants "nu, strain 1, stress 1, strain 2, stress 2, strain 3, stress 3, final strain" where all stresses and strains are positive, then in case of negative stress for a time dependent forcing function whether will it take symmetrical negative value of stress and strain or separately I have to define negative stress strain value...Also if I have to define negative stress and strain with positive part, what will be the last input which is generally final strain as I have both positive and negative extreme strain value...
kevinarden
Posts: 1021
Joined: 25 Jan 2019, 01:28
Antispam: Yes

Re: Non-linear material

Post by kevinarden »

It uses mises to determine where it is on the sress-strain curve and mises is always a positive value, even if the normal stresses are negative, so there is no need to define a negative stress strain curve.
Sayan
Posts: 92
Joined: 06 Dec 2020, 12:44
Antispam: Yes

Re: Non-linear material

Post by Sayan »

Okkk...Thank you..
Another thing is, whether the sequence "nu, strain 1,stress 1, strain 2, stress 2, final strain" for bi-linear material is constant or it doesn't depend on the sequence...I have followed some test case, where sometime it's written that final input is "final stress" in place of "final strain". Also somewhere density is also used in that section...which one is right? After defining the ultimate strain how material will behave? Will it continue as plastic section or treat it as breaking point?
kevinarden
Posts: 1021
Joined: 25 Jan 2019, 01:28
Antispam: Yes

Re: Non-linear material

Post by kevinarden »

It has been a while since I looked at these subroutines, so it would help if you post the F90 source code you are actually using. There is no breaking, the material just goes perfectly plastic, or stays on on the second slope depending on which subroutine is being used.
Sayan
Posts: 92
Joined: 06 Dec 2020, 12:44
Antispam: Yes

Re: Non-linear material

Post by Sayan »

I have attached both F90 and sif file. Here I am trying to plot a load-displacement curve to check whether it's actually behaving like bi-linear or not and also to check how it's tracing back when unloading is done. But result is not matching with what I expected. A part of result I have attached,where you can see at some point the curve is going out of track. I thought may be it's because of high load intensity but still it should behave as plastic deformation which is not happening here. I have used steel stress strain data with yield strength 400Mpa and ultimate strength 500MPa at 0.025 strain.If you kindly throw some light on this issue it will be helpful to overcome the problem. (I didn't plot full loading and unloading part as very early to that point time when unloading starts curve went wrong. Also several LU decompose matrix returning warning that singular matrix formed ). During unloading, it should either follow the same path or parallel to first linear part,which I am not sure how it's programmed here in this F90 file.
Attachments
beam.grd
(587 Bytes) Downloaded 12 times
1case.sif
(2.75 KiB) Downloaded 9 times
Untitled.png
(25.67 KiB) Not downloaded yet
Sayan
Posts: 92
Joined: 06 Dec 2020, 12:44
Antispam: Yes

Re: Non-linear material

Post by Sayan »

Sorry I forgot to attach F90 file..Here it is..
Attachments
UMATbi_linear.F90
(11.42 KiB) Downloaded 11 times
kevinarden
Posts: 1021
Joined: 25 Jan 2019, 01:28
Antispam: Yes

Re: Non-linear material

Post by kevinarden »

Bi-linear only uses 5 constants so it only took your first two points on the stress strain curve. Bi-linear should look like this

! UMAT test
Number of Material Constants = Integer 5
Number of State Variables = Integer 7
! List material constants as {}:
Material Constants(5) = Real 0.3 .002 4E8 .025 5.5E8
UMAT Subroutine = File "UMATbi_linear" "bi_linear"
Name = "bi_linear" ! This specifies the CMNAME argument of UMAT

several LU decompose matrix returning warning that singular matrix formed happens when the mesh becomes so deformed due to the displacements that the elements no longer form a good matrix. At this point the solution is done and any results after that is bogus.

I also noticed you are using linear triangle elements, these are not very good for solid mechanics. I changed the mesh to squares and increased the order of the elements to 2. This will produce much improved results.
beam.grd
(588 Bytes) Downloaded 11 times
Sayan
Posts: 92
Joined: 06 Dec 2020, 12:44
Antispam: Yes

Re: Non-linear material

Post by Sayan »

Thank you....I have understood the LU decomposition part...and also that irregularity...But what about unloading part? I have tried many times by reducing the loading limit upto which the result is showing a good bilinear curve but when it's unloading from that point to zero the result is not good...it's showing random displacement instead following the same path....
Also with 5 material constant it's showing error, but with 6 material constants where last term I mentioned as final strain it's running,,,, I tried your grd file also which is showing segmentation error.... Don't know why...But till now with so many trials by reducing slope of triangular loading, I have successfully generated the bilinear curve for loading zone but not the unloading one...That's where I am stucked.
kevinarden
Posts: 1021
Joined: 25 Jan 2019, 01:28
Antispam: Yes

Re: Non-linear material

Post by kevinarden »

I just noticed in your sif there are two Material 1 definitions, in Elmer the last overwrites all of the previous ones if it has the same ID number.

Material 1

! UMAT test
Number of Material Constants = Integer 8
Number of State Variables = Integer 7
! List material constants as {}:
Material Constants(8) = Real 0.3 0.001 1.0E8 .002 4E8 .025 5.5E8 0.025
UMAT Subroutine = File "UMATbi_linear" "bi_linear"
Name = "bi_linear" ! This specifies the CMNAME argument of UMAT
!
End

Material 1
Name = "Material 1"
Density = 7850
Youngs modulus = 200.0e9
Poisson ratio = 0.3
Rayleigh Damping=Logical True
Rayleigh Beta=Real .0001
End
Post Reply