Search
Close this search box.
Notifications
Clear all

zero deformation passed in (DFGRD=1)!!!

7 Posts
2 Users
0 Reactions
1,389 Views
Posts: 55
Topic starter
(@james.lockley)
Trusted Member
Joined: 8 years ago

Hi there, I had posted a similar question before but I stoped working on this problem and now Im back to it and stuck with the same issue!! I hope some one can help with this.

I have written a UMAT for a time/stress adaptive material. In my UMAT I derive the elasticity tensor, stress, etc. from a strain energy function that I have picked, along with its material constants, also using the deformation gradient that must be passed in by ABAQUS. In order to make sure first that apart from the adaptive part, DDSDDE and STRESS are formulated correctly, I used a strain energy function of an isotropic linear elastic material, giving it the Youngs modulus and Poissons ratio in the input file as the constants. Well, the problem is that neither DFGRD0 nor DFGRD1 update! They remain equal to the identity matrix. As the test element, I used a CPE4 with its two lower nodes pinned, and the upper two constrained in x direction. I then tried to compress the square once using a moving BC:

*NODE,NSET=NALL

1,0.,0.,0.

2,2.,0.,0.

3,2.,4.,0.

4,0.,4.,0.

*ELEMENT,TYPE=CPE4,ELSET=umat

1,1,2,3,4

*SOLID SECTION,ELSET=umat,MATERIAL=umat

**----------------------------------------------------

*MATERIAL,NAME=umat

*USER MATERIAL, CONSTANTS=2

.15e3, 0.3

**----------------------------------------------------

*BOUNDARY

1,PINNED

2,PINNED

3,1

4,1

**-----------------------------------------------------

*Step, name=LOAD, AMPLITUDE=ramp

*STATIC

0.01, 0.10

*BOUNDARY

3,2,,-.3

4,2,,-.3

I calculated STRESS in the umat in two different ways:

1) Using DSTRAN provided by ABAQUS:

DO K1=1,NTENS

DO K2=1,NTENS

STRESS(K2)=STRESS(K2)+DDSDDE(K2,K1)*DSTRAN(K1)

END DO

END DO

2) Using a slid mechanics equation giving the stress using stress invariants and the left Cauchy-Green deformation tensor.

In both cases, DFGRD0 and DFGRD1 do remain 1, while strains are correct. In case 1, stress values are also correct since they were obtained using STRAN, but in the second case stresses are zero, since they were obtained using DFGRD. The second formulation is what I eventually need, since I will need to calculate my stresses using some modified variable obtained from DFGRD. In either case, DFGRD doesnt update and I dont understand why. Can some body provide some input? Ill appreciate it! Thanks 🙂

Here is a sample of the output to the log file:

1) First case:

1.00000000000000 0.000000000000000E+000 0.000000000000000E+000

0.000000000000000E+000 1.00000000000000 0.000000000000000E+000

0.000000000000000E+000 0.000000000000000E+000 1.00000000000000

END DEF. GRADIENT (DFGRD1)

1.00000000000000 0.000000000000000E+000 0.000000000000000E+000

0.000000000000000E+000 1.00000000000000 0.000000000000000E+000

0.000000000000000E+000 0.000000000000000E+000 1.00000000000000

CURRENT STRESSES

-6.45000000000000 -15.0000000000000 -6.45000000000000

0.000000000000000E+000

CURRENT STRAINS (STRAN)

0.000000000000000E+000 -6.843749999999998E-002 0.000000000000000E+000

0.000000000000000E+000

5/12/2009 4:52:13 PM

End ABAQUS/Standard Analysis

Begin Extrapolator

5/12/2009 4:52:13 PM

Run Extrapolator.exe

5/12/2009 4:52:13 PM

End Extrapolator

ABAQUS JOB DISP COMPLETED

2) Second case:

INI. DEF. GRADIENT (DFGRD0)

1.00000000000000 0.000000000000000E+000 0.000000000000000E+000

0.000000000000000E+000 1.00000000000000 0.000000000000000E+000

0.000000000000000E+000 0.000000000000000E+000 1.00000000000000

END DEF. GRADIENT (DFGRD1)

1.00000000000000 0.000000000000000E+000 0.000000000000000E+000

0.000000000000000E+000 1.00000000000000 0.000000000000000E+000

0.000000000000000E+000 0.000000000000000E+000 1.00000000000000

CURRENT STRESSES

0.000000000000000E+000 0.000000000000000E+000 0.000000000000000E+000

0.000000000000000E+000

CURRENT STRAINS (STRAN)

0.000000000000000E+000 -6.843749999999998E-002 0.000000000000000E+000

0.000000000000000E+000

5/12/2009 5:01:22 PM

End ABAQUS/Standard Analysis

Begin Extrapolator

5/12/2009 5:01:22 PM

Run Extrapolator.exe

5/12/2009 5:01:22 PM

End Extrapolator

ABAQUS JOB DISP COMPLETED

Topic Tags
6 Replies
Posts: 55
Topic starter
(@james.lockley)
Trusted Member
Joined: 8 years ago

I have an update on this. After examining different things, I found out using the NLGEOM option passes in the deformation gradients into the umat just fine. But why is that?? Its very strange. There is no notion of necessity of using nlgeom in the documentation and it doesnt make sense either. Does any one know??

There is another problem now. The analysis will exit with an error saying the elements are distorting excessively even though there is not much deformation. Any thoughts on how to fix that?

Thanks.

Topic Tags
6 Replies
Posts: 3998
(@jorgen)
Member
Joined: 5 years ago

I think I read somewhere that you need NLGEOM to get the deformation gradient in Abaqus - but I dont remember where I saw it. I agree that it is a silly requirement.

My guess is that one or more elements become unstable very rapidly, causing the excessive distortion.

-Jorgen

Reply
Posts: 55
Topic starter
(@james.lockley)
Trusted Member
Joined: 8 years ago

Dear Jorgen, Thanks for your response. Can you elaborate a little on what you mean by elements becoming unstable quickly?

I have these foot bone geometries that are stl files which I have meshed in Patran. Since the stl file contains triangle information, the only mesh type that I can have is tetrahedra. According to ABAQUS documentation, these are the worst types of elements and should be avoided as much as possible. At this point I dont have a way of having my geometry in a format I can mesh with brick elements. The medical image processing software that we have (Analyze) can also export iges files but they only contain curve information and not surface information in the form of NURBS patches. So Patran can not mesh the model with solid elements.

So my question is: how essential is it to be able to use brick elements? Could this excessive distortion be caused by the fact that Im using tetrahedral elements? I have tested my umat on a simple geometry and it worked fine. I also used my complex bone geometry with ABAQUS built-in linear elastic model, and that was O.K. too. But now the combination of my umat and the complex geometry meshed with tetrahedra leads to this error. I wonder if I should invest in finding a way to get a different mesh type, or focus on some thing else. I will appreciate your insight. I dont know how important it is to avoid tetrahedron elements. Thanks.

Reply
Posts: 3998
(@jorgen)
Member
Joined: 5 years ago

1) When I said becoming unstable quickly, I meant that it is possible that the elements became unstable in one increment (or a very few increments). That might be why you did not see any signficant element distortions in Abaqus/Viewer.

2) It is better to use brick elements, but the main reason for that is that you typically get more accurate results using bricks instead of tetrahedral elements. Tetrahedral elements usually work just fine though. And a lot of people use then for the same reason you are: it is much easier to use them for meshing. My guess is that your problem is not caused by the element type.

-Jorgen

Reply
Posts: 55
Topic starter
(@james.lockley)
Trusted Member
Joined: 8 years ago

Dear Jorgen, thanks. Are the STABILIZATION option and RIKS normally useful in this case?

Reply
Posts: 3998
(@jorgen)
Member
Joined: 5 years ago

My guess is that they are not 🙁
Feel free to give it a try though...

-Jorgen

Reply
Share: