Creep calculation and results

22

Calc_em:

You can try this one, it can be solved by Abaqus, but even run a few step in Calculix, the results are not correct at all. Thanks.

*HEADING
BASIC VERIFICATION OF CREEP (CREEP TEST)
*NODE
1,1.,10.,1.
2,2.,10.,1.
3,2.,10.,0.
4,1.,10.,0.
11,1.,5.,1.
12,2.,5.,1.
13,2.,5.,0.
14,1.,5.,0.
21,1.,0.,1.
22,2.,0.,1.
23,2.,0.,0.
24,1.,0.,0.
*NSET,NSET=BOT
21,22,23,24
*NSET,NSET=TOP
1,2,3,4
*NSET,NSET=LEFT
21,24
*ELEMENT,TYPE=C3D4
1,21,22,23,12
2,21,12,23,14
3,21,14,23,24
4,21,12,14,11
5,14,12,23,13
6,11,12,14,1
7,1,12,14,3
8,14,3,12,13
9,12,3,1,2
10,1,3,14,4
*ELSET,ELSET=DATTIM
1,2,3,4,5,6,7,8,9,10
*ELSET,ELSET=CREEPERS
DATTIM
*SOLID SECTION,ELSET=DATTIM,MATERIAL=A1
*MATERIAL,NAME= A1
*ELASTIC
20E6,.3
*PLASTIC
0.,0.
0.,10.
*CREEP,LAW=NORTON
2.5E-27,5.,-0.2
**2.5E-27,5.,0.0
*BOUNDARY
BOT,1,1
BOT,2,2
BOT,3,3
**LEFT,1

*STEP
** STEP 1 ----- ELASTIC LOADING
*STATIC
1.0E-7,1.0E-7
*DLOAD
1,P1,20000.
2,P1,20000.
*EL FILE,FREQUENCY=20
SDV,
S,
CEEQ
*NODE FILE, FREQUENCY=20
U
*END STEP
*STEP,INC=2500000
** STEP 2 ----- CREEP TEST
*VISCO, CETOL=5.E-3
1.e-8,100.,1.E-18
**PRINT,FREQUENCY=5,RESIDUAL=NO
*EL PRINT,ELSET=CREEPERS,FREQUENCY=20
SDV,S
CEEQ,
*EL FILE,ELSET=CREEPERS,FREQUENCY=20
SDV,S,CEEQ,
*NODE FILE,FREQUENCY=20
RF,U
*END STEP

23

This definition won’t pass in Abaqus. And I would avoid such coarse meshes consisting of C3D4 elements for these tests.

To actually confirm the incorrectness of results obtained with CalculiX’s in this case, it is necessary to achieve convergence in both solvers (CalculiX and the reference one - Abaqus) and test different element types with a sufficiently dense mesh. I would expect C3D8R elements to be the most accurate but it’s also worth checking out other types. Here you can find my results of plastic bending test with different elements: Elasto-plastic bending - elements, accuracy

24

Calc_em:
Thanks.
I have checked part of CCX source code for using the Norton law’s contribution to the creep analysis.
I only can find one location to use this, that is, umat_aniso_creep.f uses this formulation. But this function is for user material, ANISO_CREEP material. I wonder how the CCX counts the Norton’s Law in the code for the default creep material.
I have tried many ways to get the creep results. The results from CCX are all not reasonable.

25

Is there any other solver than Pardiso that working gut with Calculix? Or is it possible to get that plot with gnuplot from Calculix?

26

Hi,

I have start to search about Creep and I have recently found a very interesting explanation of why convergence could be difficult with Calculix in some cases. The origin could be the Creep LAW=NORTON formulation itself. (Explained in the first two minutes of the vid).

According to the manual the stress is not normalized in Calculix, and it could lead to a tiny A parameter which further leads to rounding/convergence issues. Reviewing the example shown in the previous posts of ZFDong , I can see he is using an A parameter of A=2.5E-27. That seems a bad starting point for any numerical analisys. Maybe other unit system for the model would help to reduce that A value.
ÂżWhat do you think? ÂżOr maybe directly normalizing the NORTON LAW and changing A by the initial Strain rate parameter in Calculix?

screenshot.38

Regards,