Development of composite beam theory and its application in composite and prestressed concrete structures

Abstract

For some complex structural engineering problems like behavior of sandwich structures and prestressed concrete structures, extensive experimental works and finite element analyses might not be able to help understand the mechanics completely rather than providing the numerical results of specific cases. On the other hand, conventional structural analysis approach like Euler-Bernoulli beam theory is inherently incapable of dealing with such composite behaviors due to the plane section assumption. For those complex problems in structural engineering, an appropriate and rigorous theory has been needed for a while but unfortunately absent.

This dissertation provides a new perspective, composite beam theory that defines the structures as interacting components. The structural mechanics and mathematical manipulation are the primary tools and the balance between accuracy and applicability is carefully maintained so that the conclusions arrived in this research can be readily applied to engineering practice.

This dissertation firstly derives the composite theory in a general form, and then applies the theory to various concrete structure analysis applications, including sandwich structures and prestressed concrete. Specifically, it provides a systematic analysis methodology for sandwich structures with symmetrical and unsymmetrical wythes (sandwich structures that have identical wythes and properties will be referred as symmetrical wythes sandwich structures since the wythes are symmetrical about the neutral axis). Both longitudinal and transverse governing equations and closed form solutions are derived and studied. For prestressed concrete structures, the immediate prestress loss formulas in current design provisions and specifications are evaluated and improved, and a new type of loss due to slip is presented. The transfer length of prestressed concrete structures is redefined by investigating the mechanics and new methodologies to improve current transfer length prediction are presented. The validation against comprehensive test data demonstrates the success of the new approach, and comparisons with current formula suggest significant improvement.