(1) The contact reactions that are reported by WB Simulation appear to be incorrect. I apply a load that is transmitted from one shell-meshed body to another shell-meshed body, through a bonded-contact lap joint. The external reactions equal this load, but the contact reactions are zero. Why are the contact reactions incorrect?
(2) But I have another model for which the specified contact surface also includes a fixed support, yet the contact reactions are correct. Why is there this discrepancy in behavior?
(3) Why doesn't WB simply use PRNLD,,,CONT or FSUM,,CONT to obatin the contact reactions?
(1) By examing the ds.dat file, you can see that the technique that WB uses to calculate contact reactions is to select the nodes and underlying elements at a contact (or target) surface, and then do an FSUM. The various reaction components are then obtained by *GETs of the FSUM data. If the selected node set includes constrained nodes (fixed supports), then the external reaction forces will be included in the FSUM operation. In your case, constrained nodes are included in the selected node set, and the external reaction equals the force being transmitted across the lap joint, causing the FSUM values, and therefore the contact reactions, to be zero.
(2) In some cases, the size of the contact region is limited (i.e., instead of covering the entire face with contact or target elements, WB only creates elements for the contact pair in a region that is close to the overlapping zone). Our Developers call this algorithm "quick rejection". This procedure applies for all bonded contact situations, unless the user has manually defined the contact region (scope mode = manual) or large deformation effects are turned on.
The "quick rejection" procedure applies for your `other` model. As a result, the contact node/element set does not extend over the entire specified contact surface, and does not include the constrained nodes at the fixed support. Because this node set does not include any constrained nodes, the FSUMoperation is unaffected by external reactions, giving the `correct` value for contact reaction for this model. (Your first model used manual scope mode, therefore the contact elements were generated over the entire surface for that model.)
(3) The program obtains contact reactions as it does (using underlying elements rather than the contact elements) because using prnld,,,cont (or fsum,,cont) would not yield `correct` values for symmetric contact, and would also not yield any reactions at all for target surfaces.