This paper deals with the question of the validity of the concept and the applications of "updating" analytically derived mathematical models of structures by the use of measured data taken from the actual structure. A major point of the discussion is the common error of not distinguishing between the finite element model and the actual structure itself. This leads to such misleading concepts as: "complete set of data," banded matrices, and unique solutions. Illustrations of some of the problems with these concepts are presented. The often attempted extension of these procedures to identify structural damage is also discussed. An illustration of multiple distinct models that equally well represent the same structure is presented. It is suggested that it may be reasonable to expect that different models of the same structure may be appropriate for different applications. Examples to illustrate this concept are presented. The difficult question as to how to determine which model should be used for a particular application is raised. Significant new research is required in this area and suggestions are presented.
展开▼