On shear band and homogeneous stress

[MCC1966]

I am modelling triaxial compression test with one element (quadratic).One expects to get homogeneous stress state, but why I am getting nonhomogeneous stresses ( hardening at top nodes and softening at the middle ones ).
Also when plotting the Mises countors I see discontinuity at the middles in the countors (black unidentified spot; as if the element were split). Are these symptoums of shear band ? I do not think so because constitutive laws of ABAQUS cannot deomnstrates a shear band. can it ?
When I use many elements this doesnot appears and I almost get homogenous field.. strange (at least) for me
Please comment about this issue as much as possible
Thanks for support

 

[sendy]

Hallo MCC1966,

I agree with you about only homogeneous deformation occurs when single element triaxial test is simulated. Regarding your problem with non-homogeneous stress at single element test, I ever got that experience with oedometric compression when I didn't use constitutive model provided by Abaqus. It was due to numerical rounding error. At that time I stabilized it by introducing additional Taylor expansion in the constitutive equation (as well as the stiffness equation). Did you also get the non-homogeneous stress at very high displacement ?

As I know, when the discrete elements (many elements) used for simulating plane strain compression the shear band might be observed, but it's due to rounding error.

regards,
sendy.

 

[xerf]

The most accurate stress and strain values are at the integration points, not at the nodes.

The contour plots are accomplished by using the stress values extrapolated to nodes and might be misleading.

Did you probe the values of the stress at the integration points? (visualization module->tools->query->probe values)

 

[MCC1966]

Thank you Sendy and Xerf

Xerf I followed your e-mail then write the file then open as word file but I found nothing please comment with some details about that
THanks

Sendy Actually I am getting high displacement. I just realize that the shear band cannot be observed in ABAQUS with the existing constitutive laws because shear band results from material instability that could be encountered earlier than the failure criterion established by the constitutive law. So what you are talking about is numerical instability (not physical).. Howver, I do not know if ABAQUS can simulate the normal shear zone that coinsides with failure occurence established by constitutive law e.g ; behavior of overconsolidated clay (or dense sand) under drained test (localized softening) what do you think..? in other words under this situation does one get localized failure (as observed in triaxial) or homogenous one

 

[xerf]

what email ..!!???

 

[MCC1966]

soory xerf I meant your post
"Did you probe the values of the stress at the integration points? (visualization module->tools->query->probe values)"

 

[sendy]

Hallo MCC1966,
Yes, in the last post I talked about numerical instability that caused 'fake' shear band, especially when it is element test simulation (homogeneous test, i.e oedometer test: very small shear strain (~10^(-18)) is observed)

For dense sand, the softening behavior after the peak value of friction angle can be simulated by using specific constitutive model applied in Umat (plane shearing and triaxial compression). However, I guess the Cap model in Abaqus might be able to catch this behavior (I'm not sure, never try it).

As I understand, the shear band is not "a must" condition for triaxial or plane shearing. It depends on the soil properties and also the boundary conditions (in my opinion, the boundary condition will give significant influence). The dense properties will triger the shear band and softening behavior.

sendy.