SIMULIA UK presented its first free online with 77 attendees over the two days. The seminar focused on Contact and Convergence techniques within Abaqus.

• Introduction to General Contact & Contact Pairs
• Overview of Contact Workflow
• Surface Based Contact
• Contact Logic
• General Contact in Abaqus Explicit

If your question was not answered during the webinar or as part of this follow up then please contact your local office with the support question.

Q: When do you use finite sliding or small sliding?

A: the answer is it depends on the relative sliding between the surfaces involved.. If they are going to be sliding quite a bit (greater than an element length) then you need to use finite sliding.. In other cases where there is minimal relative movement such as in a bolted flange example you can use small sliding. If in doubt always use finite sliding. Small sliding is a limiting case of finite sliding.

Q: What are actually the limitations of using surface smoothening? What counts as large deformation and small deformation?

A: The effects of surface-to-surface contact smoothing tend to be most significant for analyses involving small deformation and coarse mesh discretization with first-order elements in the contact region; however, significant improvements to contact stress solutions are common even when the mesh is quite refined or higher-order elements are used. For analyses with large deformation this smoothing technique typically has an insignificant effect on solutions. However, in some cases the smoothing can degrade the solution accuracy after large deformation; therefore, it is not recommended to use surface-to-surface contact smoothing for large-deformation analyses. The effectiveness of surface-to-surface contact smoothing does not degrade upon relative motion between contact surfaces; for example, the smoothing technique works well for cases involving large sliding but small deformation. Further details are provided in Section 38.1.3 Smoothing contact surfaces in Abaqus/Standard of the manuals

Q: Is penalty stiffness scaling not always helpful?

A: It is highly dependent on the type of problem you are solving. If you have very stiff behavior then convergence might suffer. Scaling down the penalty stiffness will allow more penetration but will help with convergence. It is all about a balance between ease of convergence and allowable level of penetrations.

Q: I was also wondering if you knew how the CSHEAR value is calculated. From my model it does not seem to be equal to the product of the friction coefficient and the reaction force. Could you explain how this is calculated?

A: CSLIP and CSHEAR are contact output variables. 

Definition of accumulated incremental relative motion (slip) 

Abaqus/Standard defines the incremental relative motion (also known as slip) as the scalar product of the incremental relative nodal displacement vector and a slip direction. The incremental relative nodal displacement vector measures the motion of a slave node relative to the motion of the master surface. The incremental slip is accumulated only when the slave node is contacting the master surface. The sums of all such incremental slips during the analysis are reported as CSLIP1 and CSLIP2. 

CSLIP has two component CSLIP1 and CSLIP2. CSLIP1/2 provides tangential displacement (slip) of slave surface (more specifically slave node) relative to master surface in first and second slip direction respectively. First and second slip direction for solid element is same as 1 and 2 global direction respectively. For shell and membrane element these are projection of global 1 and 2 direction on element surface. These directions do not always correspond to the global coordinate system, and they rotate with the contact pair in a geometrically nonlinear analysis. Positive value of CSLIP1/2 means slave node is moving in positive 1 direction, relative to master surface. 

CSHEAR (CSHEAR1 and CSHEAR2) provides frictional shear stress at slave surface. Abaqus/Standard calculates tangential results at each constraint point by taking the scalar product of the variable's vector and a slip direction, t1 or t2, associated with the constraint point. The number at the end of a variable's name indicates whether the variable corresponds to the first or second slip direction. For example, CSHEAR1 is the frictional shear stress component in the first slip direction, while CSHEAR2 is the frictional shear stress component in the second slip direction.  The definition of the surface orientation is explained in User’s Manual 1.2.2 Conventions > Subsection: Local tangent directions on surfaces in space.

It may be difficult to manually calculate CSHEAR, as the reaction force/slip will be dependent upon the current state of the frictional response above and beyond tau max.

Q: With limited licenses we do not run the job through CAE. Can we see the diagnostic visualization after job completion - perhaps through Viewer?

A: You can access the job diagnostics tool through viewer, yes - it is part of the visualization module of Abaqus/CAE. Also, if you wish to monitor the convergence information/behavior during a run without utilizing any licenses you can use a text editor to open a number of output files such as the sta file for iteration and increment info and the .msg file for in depth convergence info.

Q: when we use stabilization we need to compare energies? To assure that solution is true? will you mention that as well?

A: You are correct - adding manual or automatic stabilization will introduce additional energy into the system. You will need to ensure that energy balance exists and that this new energy doesn't overload and unbalance the system thereby generating an inaccurate solution. There are history output requests which are generated by default such as viscous damping energy (ALLSD) which you want to compare with total strain energy (ALLIE). A recommended ratio I believe you are aiming at is the SD being lower that 5% of IE, this is a good measure. See Abaqus Analysis Users Manual 7.1.1 for more information.

Q: Why does this noise occur only with Quad elements?  Why is their behavior defined in such a way in the algorithm?

A: Not just quad elements can see noise, it's worth mentioning that if the node-to-surface contact discretization is used it is best to avoid having second-order tetrahedral elements (C3D10, C3D10HS) underlying the slave surface as they are also susceptible to poor convergence and extreme contact pressure noise. In which case use of “modified” versions of these elements (C3D10M) is recommended.