Shansep MC in plaxis 2d for stability analysis

Dear Jian,

For some reason, I know that Martin Hawkes posted the answer but I cannot see it now. I will repost it here:

Hi Jian,

I think the use of the word "switch" leads to some confusion. One often switches from one material model to another, for example when soil is replaced with concrete. However in the context of the SHANSEP MC model the word "switch" is used to switch the MC strength from a C-Phi strength to an undrained strength (Su), or to update the Su strength based on changing effective stress history (i.e. strength increase from consolidation).

To update (switch) the strength you do not need to create two groups of properties. You can start with the SHANSEP MC in the initial conditions. The initial conditions will calculate the starting σ'1 and σ'1max. For a later phase, you would enter a 1 as the special option parameter in the General section of the phase calculation parameters. What this does is instruct the material model to update (or "switch") the soil strength in the model to the SHANSEP strength using the values of  σ'1 and σ'1max to calculate the OCR. The model keeps track of σ'1max for all subsequent calculation phases. If you want to model the increase in strength due to increased  σ'1 or OCR then you would enter a 1 as the special option parameter to update or "switch" the SHANSEP strength.

Without this updating strength method, the SHANSEP MC model is just like the MC model.

Be sure to check the undrained strength in the model from the SHANSEP update (switch) by plotting Stresses menu in the output module to be sure the strength values are reasonable and as expected.

Hope this helps, I like plotting the SHANSEP MC undrained strength model a lot, and use it frequently as a reality check.

Martin

Dear Stefanos,

Thank you for your reply. I have figured out it by using 1 in the special option. I have another question regarding the use of Shansep for tailings. 

I am using Shansep MC mode for stability analysis. As I choose “user defined” in the material model, the drainage type can only be undrained A, drained, non-porous, there is no option undrained B. So I used undrained A and it was required to input the c and phi for the Mohr Coulomb parameters and also alpha, power, G/Su, etc. for Shanspe model parameters, as shown in the below figure.

If I understand correctly, the MC parameters will be used in the initial phase to calculate the initial stress, and then it will switch to Shansep MC to update the Su based on the calculated initial stress. At the same time, the model will behave as undrained B, with the friction and dilation angles being switched to zero, and C=Su.

I was wondering what is the value of C and phi I should choose for tailings in the MC model parameters. I chose c=0 and phi=0, the model failed as there was no strength. I also tried to use C=Su,min and phi=0, the model also failed. The model can work when I use c=0, and phi=effective friction angel at the critical state, but this seems not reasonable. Because for tailings, I suppose it should have no cohesion and low value of friction angle if MC model parameters should be set, my case is different from the case shown in the manual for shansep, which is usually for clay.

Thank you for your help in advance.

Dear Jian,

What does your initial geometry looks like? Since you're using K0 procedure, I presume your initial geometry is only horizontal layers, no dams or anything alike in place.

If you're trying to calculate a fully or partially installed dam as initial conditions you can't use K0 procedure and you should use Gravity loading instead....or alternatiively after the K0-procedure do a plastic nil-step in order to get correct stresses. Just a K0-procedure would leave you with unbalanced stresses, and that can indeed very well lead to a very low safety factor.

With kind regards,

Dennis Waterman

Dear Dennis,

Thank you for your reply. The initial geometry is not only horizontal layers but with interbedded non-horizontal tailings layers and berms.

Thank you,

Jian 

Dear Jian,

The K0 procedure ONLY gives correct stress results if layers and phreatic levels are horizontal. So to answer your question under 2): YES, it is necessary to do a plastic nil step after K0 procedure, because for your project the K0 procedure gives a stress state that in reality cannot exist and this must be corrected with the nil step. The fact that you like those unrealistic stresses and plastic points more than the real stresses and plastic points (after nil step or after gravity loading) is irrelevant. 

What I find more suprising is that you're so focused on the FEM results and not on the LE results. With a phi=0 and c=10 I would expect the factor of safety being close to or even below 1 ... it's liquefied tailings as you say, it should be at the brink of being unsafe or just unsafe. You mention that the LE analysis gave you a safety factor of 1.7....that is way too high for an dam with phi=0 and c=10.

With kind regards,

Dennis Waterman

Dear Dennis,

Thank you very much for your reply. 

I did not prefer any stress state, what I need is a real stress state for the safety calculation. In my case, there is no plastic points in the K0 procedure and many red plastic points in the plastic nil-step. And I checked the manual that the plastic nil-step should be adopted when the k0 procedure generates an initial stress field that is not in equilibrium or where plastic points occur. After the Plastic nil-step, the stress field should be in equilibrium and all stresses will obey the failure condition. This is not the case in my calculation and that is why I asked the question. 

So I suppose it is more suitable to use the gravity loading as there are non-horizontal layers in my case. But the problem is that the model will fail in the initial phase as the liquefied tailings do not have enough shear strength.

I also checked the LE method by using the zero cohesion and a small value of phi corresponding to the undrained shear strength ratio for the safety calculation, it also gives a similar value of safety factor of 1.7.  It is not a very steep slope so this value of safety factor can be possible. 

thank you,

Jian 

Dear Jian,

After the plastic nil-step the stresses ARE obeying the failure criterium, that is why you have all those plastic points. Apparently K0-procedure gave a lot of stresses outside the failure criterium (which cannot be) and the nil-step reduced/corrected them so that they fall on the failure criterium, due to which they are now marked as failure points. So the stress state after the nil step is fine.

Another comment....you're calculating with an undrained shear strength, but as the name already says an undrained shear strength is for the undrained situation in which effective stresses do not change. As soon as effective stresses change, the undrained shear strength changes (that is why you want to use SHANSEP).
So what is the effective stress state for which you have the undrained shear strength? Only if you obtain that effective stress state you can calculate with the undrained shear strength you have. 
If you do a gravity loading the effective stresses are built up from zero to full effective stresses as if the material is drained. That is most likely not the stress state for which you have the undrained shear strength - probably it's a higher stress state for which the undrained shear strength should be higher then the value you use.
That is the problem of calculating with undrained shear strength, the value of undrained shear strength is valid for for 1 specific effective stress state. If one knows that stress state it can be possible to achieve it, but if one doesn't know to what effective stress state that undrained shear strength belongs one can trial-and-error for weeks and still not find it. 


With kind regards,

Dennis Waterman

Dear Jian,

After the plastic nil-step the stresses ARE obeying the failure criterium, that is why you have all those plastic points. Apparently K0-procedure gave a lot of stresses outside the failure criterium (which cannot be) and the nil-step reduced/corrected them so that they fall on the failure criterium, due to which they are now marked as failure points. So the stress state after the nil step is fine.

Another comment....you're calculating with an undrained shear strength, but as the name already says an undrained shear strength is for the undrained situation in which effective stresses do not change. As soon as effective stresses change, the undrained shear strength changes (that is why you want to use SHANSEP).
So what is the effective stress state for which you have the undrained shear strength? Only if you obtain that effective stress state you can calculate with the undrained shear strength you have. 
If you do a gravity loading the effective stresses are built up from zero to full effective stresses as if the material is drained. That is most likely not the stress state for which you have the undrained shear strength - probably it's a higher stress state for which the undrained shear strength should be higher then the value you use.
That is the problem of calculating with undrained shear strength, the value of undrained shear strength is valid for for 1 specific effective stress state. If one knows that stress state it can be possible to achieve it, but if one doesn't know to what effective stress state that undrained shear strength belongs one can trial-and-error for weeks and still not find it. 


With kind regards,

Dennis Waterman