Search
Close this search box.
Notifications
Clear all

Using an Abaqus UMAT in 2-D versus 3-D

4 Posts
2 Users
0 Reactions
1,388 Views
Posts: 5
Topic starter
(@lipaul)
Active Member
Joined: 13 years ago

Hello,

I am in the process of learning how to implement a UMAT for my abaqus models.

I am starting small - using a single element and implementing isotropic linear-elasiticity theory so I can easily confirm my answers. I have implemented this behavior successfully on a single C3D8 brick element (verified by using Abaqus built-in *ELASTIC keyword).

I am trying the same thing on a single CPS4 element (plane strain). I am attempting to use the same UMAT file as in the 3-D case, with only the following exceptions:

1) Changing the definition of the Jacobian for a plane stress element. That is, using Hookes law for [URL= http://www.efunda.com/formulae/solid_mechanics/mat_mechanics/hooke_plane_stress.cfm?search_string=plane%20stress%20stiffness ]plane stress[/URL] as opposed to [URL= http://www.efunda.com/formulae/solid_mechanics/mat_mechanics/hooke_isotropic.cfm?search_string=hookes%20law ]3D isotropic[/URL] Hookes Law.

2) Using the strain indices appropriate for the CPS4 element (i.e., in plane stress, I have stress(3) referring to S12 as opposed to S33, as it would in the 3D case).

Something is still not working for my formulation. I submit my job, and in the Abaqus Model Tree, the job status goes from Submitted -> Running and hangs there indefinitely. Whats more, the .msg file only goes so far as writing INCREMENT 1 STARTS. ATTEMPT NUMBER 1, TIME INCREMENT 1.00 and then just ends - no warnings, no errors, no feedback. I have uploaded my plane stress UMAT as ELASTIC_PL_STRESS.txt.

I was hoping someone could help me with some feedback if they can identify what it is Im doing wrong. Im happy to answer any questions that would facilitate a response.

Thanks!

-MN

PS - Id be happy for some input regarding another quirk Ive noticed (that may be inconsequential). When testing/troubleshooting my code, Ill often write output text messages and the like to the .dat file. This works fine, but I find the output is printed 2X for each integration point (e.g., if I specify hello world to be output to the .dat file only at int. pt. 5 and increment 5 in a single step simulation, hello world is printed twice).

3 Replies
Posts: 5
Topic starter
(@lipaul)
Active Member
Joined: 13 years ago

I found an [URL= http://polymerfem.com/archive/index.php/t-282.html ]archived forum post[/URL] that might help any potential reply-ers. Much of this persons problem reflects my own (e.g. .023 .CID .LCK. STT files present in folders). He says his .dat file will hang halfway into computing a 3X3 tensor, as I said, my .dat file dosent even make it to the first increment.

I must admit my ignorance regarding some of the content (e.g. monitoring instabilities caused by shear strain). Still, Im a little baffled as to what might be happening as Im only trying to implement linear-elastic behavior.

In addition, Im pretty sure it doesnt have to do with the input file because the model will run perfectly using Abaqus built in linear elastic material model.

One final question: the archived post suggests outputting tracer write statements to the default write output as opposed to the .dat file (output 6). Were I to do this, where could I find the write statements? Do they just show up on the command line screen?

Thanks again!

-MN

3 Replies
Posts: 93
(@kiranckst)
Trusted Member
Joined: 14 years ago

Without going in to too much detail, as I dont really have time to go through your code right now, if the job is hanging then the problem is almost certainly your UMAT. Have a look for the book computational plasticity on the website oup.co.uk. Its gives a number of simple linear elastic UMAT formulations that you could compare yours to.

Debugging will allow you to find out what the problem is. Just use a simple print statement which writes to the log file or directly to the command window if youre running the job interactively.

Reply
Posts: 5
Topic starter
(@lipaul)
Active Member
Joined: 13 years ago

Thanks Lumpwood.

Turns out Im still not used to Fortrans indexing and referred to what should have been the first item in PROPS (e.g. PROPS(1)) as PROPS(0). Doh.

Reply
Share: