Assumption: The mesh is sufficient to capture the behavior of interest.
Evaluating, after solution, to decide if mesh refinement is needed = convergence
Accuracy is controlled by the mesh.
Page 316: "... an "ideal" model is one that enables you to solve your problem in the time allotted."
Page 317: "... convergence is directly related to the load and constraints applied to a specific run."
... "uncertainty is preferable to the term "error" ....... NOT !
"Error estimates in FEA refer to the relative difference between results across an element edge or at a node."
Page 319: Simple error estimate equations. ANSYS uses strain energy.
Figure 10.6: NOTICE, displacement converges before stress
Figure 10.8 and 10.9: are "converged" when <10% change in stress occurs
P-elements go through automatic convergence during a solution
H-elements are usually manually converged (ANSYS: adaptive meshing)
Page 323, referring to P-elements: "... if they indeed converge" ???
Element Quality: effects accuracy (and therefore, effects convergence)
Refers to page 246: aspect ratio, taper, corner angles, warp
Summary: "... only through correlation to test data and reconciliation of ALL detected discrepancies, that confidence in analysis results can be achieved."