In linear implicit FE, the fundamental quantity calculated is displacement, which is derived from the relationship F=K.x (F=nodal force, K=stiffness, x=displacement). The stiffness (matrix) is obtained from the structure, whilst the nodal forces can be obtained separately from the loading and model boundary conditions. The stiffness matrix K can then be inverted to find displacement at the nodes. Once this is known, the strains are obtained across the element and are calculated at the element integration points from which the stresses are then derived (also at the IPs).
In FE post-processing packages, the stresses can be listed/contoured at the integration points or can be extrapolated to the nodes. Most packages show stresses as contour plots, and by default these are either a pure extrapolation at the nodes, or an extrapolation PLUS averaging across elements.
------------
See faq569-1083 for details on how to make best use of Eng-Tips.com