Do you have contact definitions in your model?

With higher values of cohesion, your interfaces are probably still intact during the entire simulation. Low values of cohesion will cause failure of the interface and therefore different parts might be in contact.

Please check if this is the case.

Dr. Reiner makes a very good point.

Also, may be your model is under tensile stress causing the failure, a higher cohesion will allow for some tensile strength. On the other hand, Mohr-Coulomb is a material model with "sharp edges", condition that makes convergence difficult some times. For soil, Ducker-Prager/Cap plasticity model is my recommendation.

Hello Reiner

yes i do have a contact property assigned

i am modelling underground tunnels so i have defined normal and tangential behavior between soil and concrete tunnel lining

Yes.. The cohesionless properties (using MC model) can cause convergence problems.

1. You can check the message file and identify the node which causes the problem. You can adjust that and check if it solves. But that should not affect the solution considerably.

2. Or else, the convergence criteria can be loosened with caution. there are some other ways too.. like using automatic stabilization/or specifying the damping factor (adaptive stabilization method part should not be ignored depending on the type of the problem, pl. check the manual for that).

Several trials would be necessary to arrive at the approximate solution.

It is suggested to define a small increment in time step in ABAQUS when modelling for a cohesionless soil.

In Abaqus software, may researchers are successfully used to simulate the behaviour of cohesionless soil. The cohesion value can be used as 0.01 kN/m2 and the interface bond strength needs to be define in the model. The convergence criteria can be satisfied by increasing the load steps, mean loads are to be applied in minimal increments as possible. All the best..

Hi

The problem arises when you set the cohesion of the soil as zero. It can be solved by utilizing a small number in this field (even 0.1 kPa can be used). However, another issue emerges that is the slow convergence. To resolve this, I would suggest using a thin-layer interface between the soil and structural elements with the properties that are explained in the paper below. It worked for me very well.

Article Numerical analysis of settlement and bearing behavior of pil...

Thanks.

Hi

The zero value of cohesion will create problem in the convergence. So, it suggested to use small value of cohesion and also the small element size near the interface may help you in convergence issue.

Thanks Dhananjay