I am running a 1/4 symmetry problem with a bolt preload in the first step and when the model gets a lot of elements (ie refined) it develops convergence problems in the first step. The low density version solves the to two steps in two increments (one increment each). It takes 19 SDI's in the first and then takes 12 SDI's in the second and 1 equilibrium iteration in the second. Small sliding contact is used and NLGEOM is off. The bolt preload happens on two flat faces but the rest of those components are a bit complex so they are meshed with tet 10's improved surface stress formulation. the rest of the bits are 20 noded bricks and some more tet10's. I need the big model to solve like the low density model - one increment in each step as it takes a while. Contact stabilization was used in the second step of the light weight model and in both steps in the heavy weight model. The warning files were almost identical in both, varying only in number of nodes reported for some warnings - the program converted some surface to surface contacts into node to surfaces. The mesh densities are close but they vary over the contact patch. The bolts are meshed with 20 noded bricks. I just tried uping the mesh density on one of component involved in the bot up in the light weight one and it failed to converge and cut the load step in the preload. It is still running. . Anybody have any thoughts? All the bits have constraints applied during the preload step which are deactivated in the second step. General static steps are used for both steps..