Date of Award


Document Type


Degree Name

Doctor of Philosophy (PhD)


Mechanical Engineering

First Advisor

Mehdy Sabbaghian


The objective of this study is to investigate the fatigue behavior of a double-lap, bolted composite joint. The material used for all tests is a quasi-isotropic graphite-epoxy laminate. Prior to fatigue testing, monotonic loading tests were performed in order to establish the important fundamental properties of friction coefficients and the ultimate bearing strength of composite joints. Proper design of friction type bolted joints for composite materials accounts for the force transmitted through friction between the clamped materials. This consideration requires knowledge of the coefficient of friction between the mating surfaces of the bolt, or washer, and the composite material. In this study, the static and kinetic coefficients of friction between standard washers and the graphite-epoxy composite are found as a function of clamping force and outer ply angle. When composite materials are bolted together, the relative diameters of the bolt hole, the constraining washer and the bolt itself have a profound effect on the ultimate bearing strength. In this investigation, the effect of the difference between the bolt diameter and the inside diameter of the washer, which provides the lateral constraint at the edge of the hole, is examined. To explain the significant variation of bearing strength with increasing lateral constraint, a model is proposed to explain the mechanics of load transfer and the lateral deformation at the edge of the hole. The test results support the proposed model and provide data for designers. Fatigue data was generated using constant loading conditions for a double-lap bolted composite joint. The independent variables included six levels of maximum applied stress, four levels of clamping torque, and three different outer ply angles. Analysis of variance, censoring techniques, and goodness-of-fit tests were used to analyze the fatigue data. It was shown that the fatigue life data could adequately be described using the log-normal distribution.