"... In the simulation model, between the contact flanks nonlinear springs (invisible for the user) are connected to transfer the loads in case of compression Note: These (penalty) springs are often called “gap elements” in other FEM codes. The stiffness of these springs is adjusted automatically by the software: Simulate tries to iteratively set the penetration depth by adjusting this stiffness to a small value, so that both local stress and the global load balance are accurately captured. A penetration depth of zero is mathematically impossible, because then the stiffness of these spring elements would become infinite. Before convergence of the underlying nonlinear matrix equation Simulate calculates the residual error corresponding to the latest solution of the displacement vector u: r=f-Ku. Here, the residual vector r has the dimensions of force (this force must be zero for system convergence). The Newton-Raphson solution then solves for Kdu=r to determine the change in u in the next iteration. The residual norm is the dot product rdu. It can be thought of physically as a residual energy, which should be zero when the system has converged. Simulate normalizes the residual norm with the dot product of the total displacement and the total force vector, so the residual norm is: (rdu)/(uf). This residual norm must be smaller than the default value of 1.0E-12 to achieve convergence for the "Residual Norm Tolerance" listed in the engine .pas-file ..."
1.0E-12 ist high accuracy
1.0E-10 ist medium accuracy
1.0E-08 ist low accuracy
