Failure of interfaces not working how I expect

Question

Hi, 
 
 I am doing my undergraduate thesis on soil nail walls and I have a question about the failure of interfaces. I have derived the strength of the interfaces based on the soil using an Rinter of 0.67. The soil properties are based on the Mohr-Coulomb soil model with a cohesion of 10, unit weight of 18 and a friction angle of 30. Therefore, using the Mohr-Coulomb failure criterion I would expect interface failure at a shear stress of around 11 kN/m^2 for the top row of nails since they a have normal stress of 18 kN/m^3 based on 1m of soil above them. I would expect this failure value to increase as you go down the rows since there is a higher normal stress on them but I haven't observed this. The failure of the interfaces stayed at around 11 kN/m^2 for each row of nails when looking at the shear stresses at the interface and comparing these to the failure points of the interface. What would be the explanation of this? Also there are no water pressures in this analysis.

Thanks.

Faseel Khan · Answer

For any calculation, first a part of the out-of-balance is applied (Msf), followed by the elastic reaction of the model. If then a stress point's stress state falls outside the failure envelope, the stresses will need to be corrected by projecting the stresses back to the failure envelope. If this is the case for that stress point, it is marked as a red failure point. Then the stresses are recalculated and it might need some redistribution after that to ensure local and global equilibrium, taking into account the tolerated error. 
 The parameter Rinter relates the strength of the soil to the strength in the interfaces, according to the equations: tan(&phi;interface ) = Rinter tan (&phi;soil ) and c inter = Rinter c soil where: c soil = c ref . Hence, using the entered Rinter-value gives a reduced interface friction (wall frictions) and interface cohesion (adhesion) compared to the friction angle and the cohesion in the adjacent soil. 
 The frictional sliding at the interface is modeled by the Coulomb friction contact law. This law connects the normal and tangential components of the contact constraints, by an intermediary coefficient of friction, supposed constant. The Coulomb laws can be expressed as: 
 - Adherent contact: &tau; < &sigma;n tanϕintr + Cintr 
 - Slipping contact: &tau; = &sigma;n tanϕintr + Cintr 
 where qs = &sigma;n tan&phi;intr + Cintr is unit side friction limited with &tau; the tangential stress at contact; &phi;intr the angle of friction to the interface; Cintr the cohesion of interface; &mu; = tan&phi;intr the coefficient of friction of the interface and &sigma;n indicates the positive normal constraint in compression. 
 Please also note that you can directly assign the elastic interface normal stiffness, K N , and the elastic interface shear stiffness, K S when defining the interface strength of your materials by changing the stiffness mode from Standard to Kn/Ks in the Interfaces tab of your material. 
 The interface is basically a line element and as such it has no real thickness in the model. We have however introduced the concept of a virtual interface thickness to be able to assign a stiffness to the interface. The virtual thickness refers to the interface rather than the structure itself and it is a purely numerical value, which can be used to optimize the numerical performance of the interface. The virtual thickness is not supposed to be equal to the actual thickness of the wall. 
 When selecting an interface in Structures mode you can see and adjust the virtual thickness factor in the Selection explorer. The actual virtual thickness is calculated as the virtual thickness factor * global element size (t_virtual = 0.1 * global element size). Virtual thickness under interface tab is an additional control on interface stiffness, you may want to stick to the default value unless you have good reasons. Real interface thickness is relevant only when advanced soil model is selected and dilatancy cut-off option is activated. 
 Interface behaves stiffer with smaller thickness and vice-versa. From numerical point of view, we want the elastic deformation of interface to be negligible, yet not too high an interface stiffness that would cause numerical problems. The default value is good for most of the problems, in order to investigate the effects you should set up different test models. 
 The interface elements are supposed to generate very little elastic deformation. Interface stiffness is also influenced by the poisons ratio. So, for the interface elements, a Poisson ratio of 0.45 will lead to a stiff interface and therefore negligible elastic deformations of the interface in the model. We want a stiff interface such that elastic deformations are in general negligible in relation to other deformations in the model. 
 Very low stiffness values for the interface on the other hand, results in high(er) displacements of the interface. So for such cases, when your primary interest is to control the stiffness of the interface you may also directly assign a material set to the interface and directly set the desired stiffness. It can be more useful to define a separate material data set. This should give results less dependent on the interface stiffness values varied by changing the Rinter factor. You have to set the interface strength of material to Rigid in order to avoid additional account for stiffness/ strength reduction. In general too stiff an interface is not preferred from the perspective of numerical convergence. Excessive reduction in virtual thickness would result in too stiff of an interface behaviour, and causes numerical issues.

Failure of interfaces not working how I expect

Top Replies