Note: This discussion is about an older version of the COMSOL Multiphysics^{®} software. The information provided may be out of date.

**Discussion Closed** This discussion was created more than 6 months ago and has been closed. To start a new discussion with a link back to this one, click here.

## "Failed to compute elastoplastic strain variables" Help Please

Posted Feb 4, 2013, 6:55 AM EST Geomechanics, Structural Mechanics & Thermal Stresses Version 4.3a, Version 4.3b 12 Replies

I am an absolute beginner at COMSOL Multiphysics and I am trying to model the effect of the movement of soil on a steel pipeline buried in it. I have attached the model file with this post. I am trying to use the Soil Plasticity Criterion but as the tutorial describes, It needs to be given Initial values. So following the built in tutorial, I have included two studies, one for calculating the initial stresses and the other one for calculating the actual stresses on the pipeline.

I would be really grateful if someone could take a look and point out the mistake in my method.

Thanks in advance.

- Rohit

Knowledgebase entry 1150 discusses this error message.

www.comsol.com/support/knowledgebase/1150/

Regards,

Henrik

I had come across it before posting the question. I tried the solution suggested but I am still stuck at the same issue.

May be you have solved your problem by now, if not, what is important is when you calculate initial stresses for a given domain as your first step that you check if the calculated stresses (if possible, visually from your plots) everywhere in your elastop-plastic domain are below the initial yield stress (this you can obtain from your yield function). If not, you will continue to have the problem. For example, I had certain corners of my domain where von mises stresses were way too high because of a boundary condition and relaxing the geometry a bit helped reduce stresses in such zones. In some cases, introducing an adjacent linear elastic material (which I had ignored to reduce model complexity) solved the problem.

Suresh

Thanks for the prompt Reply.

I had come across it before posting the question. I tried the solution suggested but I am still stuck at the same issue.

Hi

Have you solved this problem? Can you talk a little bit about it?

May be you have solved your problem by now, if not, what is important is when you calculate initial stresses for a given domain as your first step that you check if the calculated stresses (if possible, visually from your plots) everywhere in your elastop-plastic domain are below the initial yield stress (this you can obtain from your yield function). If not, you will continue to have the problem.

Dear Suresh,

I got this problem and I found your reply very interesting. But what I don't understand is what you mean for initial stresses from the first step, that you mentioned. For example, if apply a load on a domain not previously stresses (so initial stresses are 0 at t=0, in a time dependent study e.g.), and the load will lead to plastic deformation.

Since the initial yield stress is the stress level where plastic deformation starts, why the stresses (von Mises) should be below this level? In that case, plasticity will never occur...and it's not possible...

Can you give me an explanation?

Thank you in advance!

\Fabio

Instead of applying the load at once, can you try to ramp it up gradually using a parametric continuation study (or time dependent) and see if it solves.

Suresh

I would say, check the results without using plasticity and compare them with the values of the initial stress...

If you are just using random values of initial stresses (if you are doing some test etc.) you can insert in the matrix of initial stress a reduced value of the yield stress (in Comsol this value is called "Initial yield stress", because of hardening), for instance if your yield stress is 60MPa, try using 6MPa

I hope this helps

Regards

Regards,

Xuefeng

What value did you use for the cohesion? For a fine sand it is probably low.

With a very low cohesion, the analysis is likely to fail with this error. On the free surface, for example, the stresses are zero. Due to numerical errors, there will a some places be small tensile stresses, and then there must be some cohesion to stay inside the yield surface.

Such a problem could be cured by adding a small extra compressive initial stress close to the free surface, since the results there are not important anyway. This approach will however only be applicable if the tensile stresses are fictitious. In the excavation model, there will be tensile stresses for physical reasons when material is removed. The maximum principal stress around the tunnel is of the order of 16 kPa, so unless the cohesion is larger enough, the problem cannot be solved. This reflects that if is difficult to make a tunnel in sand.

Regards,

Henrik

Thank you for your recommendation, and I have being working about this problem recently. You are right. The cohesion has a significant impact on the calculation.

The model that I have being working with is just like the Tunnel Excavation. The soil properties are:

Poisson’s ratio: 0.43,

Young’s modulus: 15.03e6 Pa,

Density: 1570kg/m^3,

Cohesion: 20e3 Pa,

Internal friction angle: 15 [deg]

It is a kind of heavy loam, the cohesion is 20KPa. In the Tunnel Excavation, the soil cohesion is 130KPa, which is more like a rock. Well, what I am doing is after excavation, the drilling fluid is full fill the hole to support the borehole wall and balance the formation pressure. In practice, the pressure of drilling fluid is 1800KPa, but it will lead the calculation to fail if the pressure is so high. Therefore, I start the calculation with 300KPa, and increase the pressure 100KPa per calculation. The calculation is successful before 1400KPa, but failed at 1400KPa (Failed to compute elastoplastic strain variables.). The effective plastic strain is very large when the pressure is 1300KPa.

I think maybe the pressure is too high (1400KPa~1800KPa), hence, the mashes may distort or undefined failure criterion (Open-Hole Multilateral Well defined a failure criterion), which may lead the calculation to fail.

Change the soil properties are not a proper way to get through the failure. Once the soil properties changed, the model is no longer close to the real situation. As we know, the soil properties that we test in the laboratory are a little different with the in-situ soil properties. For example, the cohesion we get form the laboratory is smaller than the in-situ soil cohesion. However, the difference between these two cohesions is not much, and we cannot just enlarge it 10 times.

Above all, I still cannot find an effective approach except enlarge the soil cohesion to get through this error (Failed to compute elastoplastic strain variables.) when I put a high pressure (1800KPa) on the borehole wall.

Regards,

Xuefeng

Note that while COMSOL employees may participate in the discussion forum, COMSOL^{®} software users who are on-subscription should submit their questions via the Support Center for a more comprehensive response from the Technical Support team.