Вы находитесь на странице: 1из 165

BEHAVIOR OF FASTENED AND ADHESIVELY BONDED COMPOSITES UNDER MECHANICAL AND THERMOMECHANICAL LOADS

by

VINAYSHANKAR LINGAPPA VIRUPAKSHA

A dissertation submitted in partial fulfillment of the requirements for the degree of

DOCTOR OF PHILOSOPHY IN MECHANICAL ENGINEERING

2008

Oakland University Rochester, Michigan

Doctoral Advisory Committee: Sayed A. Nassar, Ph.D., Chair LianXiang Yang, Ph.D. Meir Shillor, Ph.D.
Michael P. Polis, Ph.D.

UMI Number: 3333081

INFORMATION TO USERS

The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleed-through, substandard margins, and improper alignment can adversely affect reproduction. In the unlikely event that the author did not send a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion.

UMI
UMI Microform 3333081 Copyright 2008 by ProQuest LLC. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code. ProQuest LLC 789 E. Eisenhower Parkway PO Box 1346 Ann Arbor, Ml 48106-1346

Copyright by Vinayshankar Lingappa Virupaksha, 2008 All rights reserved

To my dearest mother and father, Sarvamangala and Virupaksha

ACKNOWLEDGMENTS

I would like to express my sincere gratitude and appreciation to my adviser, Professor Sayed Nassar. His wide knowledge and experience have been of great value for me. His understanding, encouraging and personal guidance have been helpful and invaluable. I am grateful to my advisory committee members, Professor Michael Polis, Professor LianXiang Yang and Professor Meir Shillor, for their valuable time and suggestions. Special thanks to Professor Garry Barber and Dr. Forest Wright for providing me the first job in United States of America. I would like to thank all the staff members of Department of Mechanical Engineering for their support through out my stay at Oakland University. I thank all my friends and student colleagues for providing me the required social and academic challenges, and diversions. I devote special thanks to all my relatives for their love, support, and encouragement. Last, but not least, I am very thankful to my family: my mother, Sarvamangala, my father Virupaksha and my brother Dr.Vijayshankar Virupaksha for their unconditional support and encouragement to pursue my interests, even when the interests went beyond boundaries of languages, geography and field. Their love and devotion throughout my life gave me the strength to accomplish my goals.

Vinayshankar Lingappa Virupaksha iv

PREFACE

This document outlines the research conducted to complete the doctoral dissertation entitled "Behavior of fastened and adhesively bonded composites under mechanical and thermomechanical loads". It is submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Mechanical Engineering at Oakland University. The document is organized in the following manner: Chapter 1: Introduction and Literature Review, gives the background of adhesive bonding and mechanical fastening of polymeric composite materials. This chapter outlines the previous research on interfacial stress analysis in adhesive bonded joints, and bolt bearing behavior in composite bolted joints. It briefly describes the limited analytical and experimental work in this field of study. Finally it presents the motivation and the objective of this work. Chapter 2: Effect of Adhesive Thickness and Properties on the Bi-axial Interfacial Shear Stresses in Bonded Joints Using a Continuum Mixture Model, introduces a new analytical model which predicts the interfacial shear stresses due to thermomechanical loading in an adhesively bonded joints. Finite Element Analysis is further carried out to validate the analytical model results. Chapter 3: Effect of Washers and Bolt Tension on the Behavior of Thick Composite Joints, presents an experimental investigation of the effect of washer geometry and initial bolt preload on the strength and stiffness of thick composite bolted joints. Joint clamp load is monitored in real time to correlate the bearing behavior.

Further, failure analysis is carried out to analyze the progression of bearing failure in composite laminates. Chapter 4: Effect of Bolt Tightness on the Behavior of Composite Joints, presents an experimental investigation of the effect of various bolt tightness combinations on the strength and stiffness of double bolted single lap joints. Further a progressive failure analysis is carried out to analyze the bearing failure for different tightening combinations. Finally, a 3-dimensional finite element analysis is carried out to validate the behavior of double bolted joints. Chapter 5: Conclusions and Future Work

VI

ABSTRACT

BEHAVIOR OF FASTENED AND ADHESIVELY BONDED COMPOSITES UNDER MECHANICAL AND THERMOMECHANICAL LOADS by Vinayshankar Lingappa Virupaksha

Adviser: Sayed A Nassar, Ph.D.

The safety and structural integrity of composite structures are determined by their respective joints, which may be either adhesively bonded or mechanically fastened. The strength and reliability of adhesively bonded joints are significantly influenced by interfacial stresses. In the same way the strength and reliability of mechanical fastened joints depend on its laminate bolt bearing strength. In the first part of this dissertation, an analytical model based on continuum mixture theories is developed to study the interfacial shear stresses in adhesively bonded joints. In the second part, experimental and finite element investigations are carried out to study the bolt bearing behavior in composite bolted joints. The analytical model for adhesive bonded joints predicts the effect of adhesive thickness and properties on the bi-axial interfacial shear stresses due to thermomechanical loading. The interfacial shear stresses between the adhesive and each adherend is determined using the constitutive equations. Numerical results show that both the adhesive thickness and the material properties have a significant effect on the thermomechanically induced interfacial shear stresses between the adherends and the

vii

adhesive. The developed model inherently has the capacity for optimizing the selection of the adhesive thickness and material properties that would yield a more reliable bonded joint. For the composite bolted joints, experimental and finite element investigations are carried out to study the affect of bolt-load level, washer geometry and bolt tightness on the bolt bearing behavior. A double lap shear joint is considered to study the effect of bolt-load levels and washer geometry, while a double bolted single lap shear joint is considered to study the effect of bolt tightness and joint material on the bearing behavior of composite bolted joints. Finite element analysis using a commercially available code ABAQUS is used to validate the experimental results. Failure analysis using optical microscope and digital photography is conducted to analyze the progression of bearing failure in composite laminates.

viii

TABLE OF CONTENTS

ACKNOWLEDGMENTS PREFACE ABSTRACT LIST OF TABLES LIST OF FIGURES CHAPTER ONE INTRODUCTION AND LITERATURE REVIEW 1.1 Background on Adhesively Bonded Joints and Interfacial Stresses 1.1.1 Failure Modes in Adhesive Bonded Joints 1.2 Previous Research on Interfacial Shear Stresses in Adhesively Bonded Joints 1.3 Background on Bolt Bearing Behavior in Polymeric Composite Bolted Joints 1.3.1 Failure Modes in Polymeric Composite Bolted Joints 1.3.2 Bolt Bearing Behavior in Polymeric Composite Joints 1.4 Previous Research on Bolt Bearing Behavior in Polymeric Composite Joints 1.4.1 Influence of Coupon Geometry and Laminate Properties on the Bolt Bearing Behavior 1.4.2 Effect of Bolt Tightening Torque and Clamping Pressure on Bearing Behavior

iv v vii xiii xiv

1 3

15

15

16

17

17 23

ix

TABLE OF CONTENTSContinued

1.4.3 Progressive Failure Mechanism in Composite Bolted Joints 1.4.4 Finite Element Analysis of Composite Bolted Joints 1.5 Objective of the Dissertation CHAPTER TWO EFFECT OF ADHESIVE THICKNESS AND PROPERTIES ON THE BI-AXIAL INTERFACIAL SHEAR STRESSES IN BONDED JOINTS USING A CONTINUUM MDCTURE MODEL 2.1 Formulation of the Problem 2.1.1 Equilibrium and Constitute Equations 2.1.2 Continuity Conditions 2.1.3 Continuum Mixture Equations 2.1.4 Evaluation of the Interaction Terms 2.1.5 Relationship between Shear Stresses and Average Displacements 2.1.6 Expression for Shear Stresses in Terms of Displacements 2.1.7 Governing Differential Equations 2.1.8 General Solutions for Governing Partial Differential Equations 2.1.9 Boundary Conditions 2.1.10 Interfacial Shear Stresses 2.2 Numerical Results and Discussions

27

31 34

41 41 42 43 44 46

47 50 50

52 53 54 55

TABLE OF CONTENTSContinued

2.2.1 Finite Element Verification 2.2.2 Effect of Elastic Properties of Adhesive 2.2.3 Effect of Adhesive Thickness 2.3 Summary CHAPTER THREE EFFECT OF WASHERS AND BOLT TENSION ON THE BEHAVIOR OF THICK COMPOSITE JOINTS 3.1 Experimental setup and Procedure 3.1.1 Materials 3.1.2 Test Fixture and Instrumentation 3.2 Results and Discussion 3.2.1 Effect of Bolt Preload 3.2.2 Effect of Washer Size and Thickness on Bearing Behavior 3.2.3 Clamp Load Variation 3.2.4 Failure Analysis 3.3 Summary CHAPTER FOUR EFFECT OF BOLT TIGHTNESS ON THE BEHAVIOR OF COMPOSITE JOINTS 4.1 Experimental Set-up and Procedure 4.1.1 Experiments
XI

56 58 58 59

76 76 77 77 80 81

83 84 86 87

109 109 110

TABLE OF CONTENTSContinued

4.1.2 Progressive Damage Analysis 4.2 Experimental Results and Discussion 4.2.1 Effect of Fastener Tightness Condition 4.2.2 Effect of Joint Materials 4.2.3 Failure Mode Progression 4.3 Finite Element Modeling 4.4 Summary

111 112 112 114 115 116 118

CHAPTER FIVE CONCLUSIONS AND FUTURE STUDY


5.1 Conclusions 5.1.1 Effect of Adhesive Thickness and Properties on the Bi-axial Interfacial Shear Stresses in Bonded Joints Using a Continuum Mixture Model 5.1.2 Effect of Washers and Bolt Tension on the Behavior of Thick Composite Joints 5.1.3 Effect of Bolt Tension on the Behavior of Composite Joints 5.2 Future Work REFERENCES

139
139

139

139

140 141 142

xii

LIST OF TABLES

Table 2.1 Table 3.1

Material properties of Boron and Carbon Phenolic Laminate Initial bolt-load and corresponding clamping pressure for small and large washer j oints

60

88 89 90 91 92 120

Table 3.2 Table 3.3 Table 3.4 Table 3.5 Table 4.1

Bearing properties of single large washer composite joints Bearing properties of single small washer composite joints Bearing properties of double large washer composite joints Bearing properties of double small washer composite joints Material properties of joint components

xm

LIST OF FIGURES

Figure 1.1 Figure 1.2 Figure 1.3

Example of an adhesively bonded joint Shear stresses and peel stresses in an adhesive bonded joint Shear stresses due to difference in coefficient of thermal expansion

36 36

37 37 38 39

Figure 1.4 Figure 1.5 Figure 1.6 Figure 1.7

Failure modes in adhesive bonded joints Joint parameters in a typical composite bolted joint Failure modes in composite bolted joints Single bolted double lap shear joint subjected to in-plane loading

40 63 64 65 66 67 68 69 70 71 72

Figure 2.1 Figure 2.2 Figure 2.3 Figure 2.4 Figure 2.5 Figure 2.6 Figure 2.7 Figure 2.8 Figure 2.9 Figure 2.10

Geometric model Shear stress distribution Theoretical shear stress, (xxy), at the upper interface FEM shear stress, (xxy), at the upper interface Theoretical shear stress, (xzy), at the upper interface FEM shear stress, (xZy), at the upper interface Theoretical shear stress, (xxy), at the lower interface FEM shear stress, (xxy), at the lower interface Theoretical shear stress, (xzy), at the lower interface FEM shear stress, (x^), at the lower interface

xiv

LIST OF FIGURESContinued

Figure 2.11

Effect of adhesive properties on the shear stress at the upper interface Effect of adhesive properties on the shear stress at the lower interface Effect of adhesive thickness on the shear stress at the lower interface

73

Figure 2.12

74

Figure 2.13

75 93 94 95

Figure 3.1 Figure 3.2 Figure 3.3 Figure 3.4 Figure 3.5

Geometry of the test coupon Experimental double lap-shear test fixture Bearing test experimental set-up Schematic representation of bearing stress distribution in a pin loaded joint Bearing stress Vs. bearing strain curve for a small washer finger tightened bolted joint

96 97 97 98 98 99 99

Figure 3.6 Figure 3.7 Figure 3.8 Figure 3.9 Figure 3.10 Figure 3.11 Figure 3.12 Figure 3.13

Bearing stress Vs. strain curve for joints with 50% preload Effect of bolt preload on joint bearing stiffness Effect of bolt preload on offset bearing strength Effect of bolt preload on ultimate joint strength Effect of bolt preload on joint strain Effect of washer size on bearing stiffness of joints: single washer Effect of washer size on bearing stiffness of joints: double washer Effect of washer size on offset bearing strength: single washer

100 100 101

xv

LIST OF FIGURESContinued

Figure 3.14

Effect of washer size on offset bearing strength of j oints: double washer 101 102

Figure 3.15 Figure 3.16

Effect of washer area on bearing stress-strain behavior Effect of washer thickness on bearing stiffness: large washer

102 103

Figure 3.17 Figure 3.18 Figure 3.19 Figure 3.20

Effect of small washer thickness on joint bearing stiffness Effect of washer thickness on bearing strength: large washer Effect of washer thickness on bearing strength: small washer Joint clamp-load variation with joint displacement: zero bolt preload

103 104 104 105 105 106 106

Figure 3.21 Figure 3.22 Figure 3.23 Figure 3.24 Figure 3.25

Joint clamp load Vs. displacement: 50% bolt preload Joint clamp load Vs. applied axial load: small washers Joint clamp load Vs. applied axial load: large washers Bearing damage in finger tightened joint coupons Bearing damage in various joint coupons with large washers

108 121 122

Figure 4.1 Figure 4.2 Figure 4.3 Figure 4.4

Geometry of single lap, double-bolted joint Schematic representation of inspected damage regions Load displacement curves for aluminum-composite joints Initial portion of the aluminum-composite loaddisplacement curve

124 126

xvi

LIST OF FIGURESContinued

Figure 4.5

Load-displacement curves showing the ultimate failure load 127 128 128 129 129

Figure 4.6 Figure 4.7 Figure 4.8 Figure 4.9 Figure 4.10 Figure 4.11

Bearing surface delamination for TT-LB joints Bearing surface delamination for LT-TB joints Bearing surface delamination for LT-LB joints Bearing surface delamination for TT-TB joints Strength comparison of aluminum-composite and composite-composite TT-TB joints Strength comparison of aluminum-composite and composite-composite LT-LB joints Initial portion of the load-displacement data for LT-LB joints Initial portion of the load-displacement data for LT-TB joints

130 130

Figure 4.12

131

Figure 4.13

132 133 133 134

Figure 4.14 Figure 4.15 Figure 4.16 Figure 4.17

Initiation of bearing failure Bearing failure at 70% of ultimate failure load Bearing damage just before the ultimate failure Finite element model of double bolted composite to aluminum joint

135 135 136 138

Figure 4.18 Figure 4.19 Figure 4.20

Contact surface in the finite element model Frictional effect on the load displacement curves Comparison ofFEA and experimental results

xvii

CHAPTER ONE INTRODUCTION AND LITERATURE REVIEW Polymer matrix fiber reinforced laminated composites are widely used in structural and mechanical components across various industries that include automotive, aerospace and defense applications. The safety and structural integrity of composite structures often depend on the integrity and reliability of their respective joints that often are the weak link in the design. The two main commonly followed joining technologies are adhesive bonding and mechanical fastening. The strength and reliability of an adhesively bonded joint depends on their interfacial stresses, while that for mechanical fastened joints depends on their laminate bolt bearing strength [1]. This dissertation deals with both the interfacial stresses in adhesively bonded joints, and as well as the bearing strength in composite bolted joints.

1.1 Background on Adhesively Bonded Joints and Interfacial Stresses Adhesive bonding is a joining technology where low modulus glue is used as a bonding agent to join materials. A typical adhesively bonded joint is shown in Figure 1.1. Adhesively bonded joints provide smoother joint surface, hence they are often preferred in automotive applications. The use of adhesively bonded joints is steadily increasing due to their corrosion resistance and light weight properties. One disadvantage of adhesively bonded joints is that they are permanent and cannot be disassembled. Another drawback of these joints is the uncertainty of long term structural
1

stability due to the lack of standard inspection methodologies for adhesive joint quality. Joint performance largely depends on the bonded surface preparation, the adhesive properties, and the adhesive thickness; adhesively bonded joints are also susceptible to environmental factors such as moisture and high temperature [2]. The main purpose of an adhesively bonded joint is to transfer the loads. These loads produce interfacial stresses and hence understanding the stress distribution is an essential part of the design and analysis. The knowledge of stress - strain state in a bonded joint provides an insight into the joint behavior and potential failure mechanism. Failure due to adhesive shear stresses or adhesive peel stresses are the most common failure modes in bonded joints. Figure 1.2 shows the typical shear stress and peel stress distribution in an adhesive joint. Interfacial shear stresses can be due to mechanical, thermal or thermo-mechanical loading. Figure 1.2a illustrates how the load is transferred by shear in the adhesive layer (as seen by the forces on a section of the substrate). As the load is transferred, the adherend loading decreases and the shear stress is induced on the adhesive. The shear stress is maximum at the overlap edge and decreases along the length of the adhesive [3]; shear stresses are critical in ductile failure of adhesive joints [4]. Figure 1.2b shows the load transfer perpendicular to the adhesive layer. The adherend deforms less as the load is transferred; this induces stress perpendicular to the adhesive layer that is known as peel stress. The peel stress is essentially normal stress that would be critical in brittle failure of adhesive joints. When an adhesively bonded joint is subjected to temperature changes, the adherends and the adhesive expand and contract differently resulting in thermal shear stresses. These thermal shear stresses largely depend on the temperature change and coefficient of thermal expansion (CTE). A 2

material with a higher CTE contracts and compresses the other material when the temperature is decreased and vice-versa with the increase in temperature. This transmits the internal load inducing thermal shear stresses through the adhesive thickness, which decays as the load is transmitted [3], as shown in Figure 1.3. The thermo-mechanically induced shear stresses in an adhesively bonded joint are more complex due to superimposition of mechanical and thermal loads. The adhesive and the adherend thickness, surface topology, taper angle of the adhesive edge, properties of the adhesives and the adherend are some of the critical factors affecting the behavior of adhesively bonded joints. Investigating how these factors affect the stresses would help in improving the reliability of bonded joints.

1.1.1 Failure Modes in Adhesively Bonded Joints Adhesively bonded joints are usually designed for a bulk failure of the adherends and not in the adhesive. Failures in laminated composite material are often at the surface plies of the laminate material (delamination). Care should be taken to ensure that the adhesive layer does not become the weakest link. Some of the failure modes in bonded composite/polymer joints are: cohesive failure in the adhesive (known as the failure of the adhesive layer), adhesive failure at the composite-adhesive interface (known as the interfacial failure), and failure of the adherend (laminate) known as delamination [5]. The interfacial failure and the delamination failure is mainly due to the higher interfacial shear stress. The interfacial shear strength can be obtained from a lap shear test. However, numerical analysis and analytical model would help understand the distribution of shear stresses. Cohesive failure in the adhesive usually occurs when the applied loads
3

exceed the intrinsic strength of the adhesive material. This tends to be a localized effect occurring near stress concentration areas such as the ends of joints. In the laminate materials such as composites, delamination failure generally initiates from the matrix between the layers due to out-of-plane peel stresses or interlaminar shear stresses [6]. Other forms of failures such as through thickness tensile cracking can occur if the composite adherend is not a layered structure. Figure 1.4 shows the failure modes in a typical adhesively bonded joint. Some of the critical issues that need to be investigated in adhesively bonded joints include the prediction of the joint strength, designing the joint parameters, optimizing the joint performance and inspection of bonded joint quality. Stress analysis tools help understanding the adhesive and adherend load distribution in a bonded joint [7]. Interfacial stress prediction is one important aspect in the design of an adhesively bonded structure. Bond breakage/delamination is the most common failure mode in bonded structures with failure initiating at the adherend adhesive interface. The failure may occur due to the progression of existing micro cracks or by delamination developed at the interface. Delamination may be due to thermal, mechanical or thermomechanical loading and interfacial degradation caused by moisture and other chemical species. Delamination also occurs when the interfacial stresses from the loading exceed the strength of the adhesive material.

1.2 Previous Research on Interfacial Shear Stresses in Adhesively Bonded Joints Stress analysis in an adhesively bonded structure can be carried out using two approaches, namely, analytical and/or numerical methods such as Finite Element
4

Analysis (FEA). Analytical approaches have been based on the beam model or continuum model, where a set of differential equations and boundary conditions are formulated. The solution to these differential equations yields analytical expressions which give values of the stresses in the joints [7]. The analytical approach for the solution of interfacial stress distributions in a bonded joint has been progressively refined until recent times. In numerical solution approach, the solutions of the differential equations provide displacements at each point from which corresponding strains and stresses can be computed at each node. Finite element analyses are among the numerical approaches which have been extensively used in many applications [7]. Published work relating to experimental, analytical and numerical approaches to evaluate the stresses in an bonded structures is reviewed in the following section. One of the earliest stress analysis models were derived by Timoshenko (1925) [8] based on the elementary beam theory. Timoshenko analyzed only the normal stresses and assumed it to be unchanged along the length of the bimetal thermostat. The interfacial stresses were not analyzed but he just mentioned that it is higher at the ends of the strips. Various approaches to solve for the interfacial stresses were suggested during the last few decades, mostly in conjunction with the needs of the microelectronics technology [9]. These approaches were mainly extension of Timoshenko theory and were based on strength of materials and structural mechanics. The static analysis were carried out by Nayfeh [10] to estimate the interfacial shear stresses in composites subjected to combined mechanical and thermal loading. Multi-cylindrical periodic fibers, single cylindrical fiber, and single planar fiber reinforcement models were considered in the analysis. Continuum mixture theories for 5

wave propagation, in bilaminated composites by Hegemier et al. [11], yielded a set of partial differential equations which described the displacement behavior of the composite. The interfacial shear stresses were calculated from distribution of the displacements, stresses and the temperatures in each individual constituent. In a subsequent research [12], Nayfeh and Nassar used a continuum theory developed for bilaminated composites to study the bonding material influence on the dynamic behavior of trilaminated composites. Longitudinal wave propagating in the direction parallel to the layers of the linear elastic, homogeneous and isotropic trilaminated composite was considered in the problem formulation. In another study, Nayfeh [13] extended the same methodology in [10] to study the dynamically induced interfacial shear stresses in twocomponent fibrous composites. Nayfeh and Nassar in 1982 [14] studied the influence of bonding agents on the thermo-mechanically induced interfacial shear stresses in laminated composites. The analysis was carried out to determine the influence of bonding material on the statically induced interfacial shear stresses due to various mechanical and thermal loading using the same two dimensional model they introduced in [12]. The laminate representing various materials in the model were assumed to be infinite along the in-plane direction, and were stacked normal to the in-plane direction in such a way that any layer of material 1 and material 2 were sandwiched between the layers of material 0. This arrangement simulated the case where material 0 acted as the bonding agent between material 1 and material 2. Equilibrium equations, constitutive relations along with the continuum mixture relations were used to develop the formulation. A linear variation of shear stresses was 6

assumed about the mid plane of each of the two adherends; the shear stress was also linear through the adhesive thickness. Using these conditions, a system of two coupled differential equations describing the behavior of the two interfacial shear stresses was derived. The solution for the shear stresses was obtained by solving these coupled differential equations. These interfacial shear stresses demonstrated their dependence on the geometry and material properties of the trilaminated model, as well as the combined mechanical and thermal loadings. Various combinations of pressure loadings and uniform temperature changes were utilized to increase, decrease and even neutralize the state of the interfacial shear stresses. The numerical results demonstrated the effect of various thermomechanical loadings, and the influence of the bonding agent properties and thickness, on the interfacial shear stresses [14]. One thing to be noted here is that the interfacial shear stresses were determined based on a two dimensional model. Pao and Eisele [15] developed an analytical model to evaluate the interfacial shear and peel stresses in a multilayered thin stack subjected to a uniform thermal loading. The model was based on Suhir's bimetal thermostat model. The approach provided a system of coupled second order linear differential equations to solve for the interfacial stresses. The interfacial stresses were then used for determining the normal stresses in each layer along with the deflection of the overall stack. A general two-dimensional multi-layered stack with finite length 2L was considered for the analysis. The material behavior was assumed to be linearly elastic, and uniform heating or cooling effect was considered. Two examples, first a five-layered double-shear solder joint and second, a four-layered transistor stack were used to illustrate the application of the approach. The first one showed that the thermally induced bending might increase or decrease the stress level. It 7

was also showed that the maximum interfacial shear stresses were not necessarily at the edges, but were located at the vicinity of the edges. The second one showed that as the thickness of the layer decreased the solution converged to a case where such layer was absent. This approach was considered to be useful to analyze the behavior of multilayer thin stacks in electronic industry. This model considered only the thermal loading and was two dimensional. Suhir in 1998 [9] developed an analytical model for interfacial stresses in bimetal thermostats. This model was based on the elementary beam theory, but in addition he considered the transverse compliance (through thickness). The solution provided the distribution of interfacial shear and normal stresses. Hui-Shen, Teng, and Yang [16] theoretically studied the interfacial stresses in a simply supported beam bonded with a thin fiber reinforced polymeric composite or a steel plate. A simply supported bonded joint was subjected to a uniformly distributed load and a uniform bending moment. A plain stress model was used for the beam and a plain strain model was used for the slabs. The other important feature considered in the analysis was a non-uniform stress distribution in the adhesive layer. The results showed that the maximum normal stress always occurred at the free edges and the maximum shear stress occurred a small distance from the free edge. The interfacial stresses increased as the plate stiffness was increased, or as the plate length was reduced. The location of the maximum interfacial shear stress moved towards the free edge as the plate stiffness was reduced or as the plate length was reduced. Ru [17] developed a non-local modified beam model to evaluate the interfacial thermal stresses in biomaterial elastic beams. The model was based on Suhir (1986) 8

model. The model satisfied the zero shear stress boundry condition and provided the interfacial peel stresses. The predicted interfacial shear stresses were found to be in reasonably good agreement with some of the known numerical results. The model was considered to be best suitable for only multilayered and two-dimensional materials or electronic packaging. Hyonny and Keith [18] worked on. in-plane shear loaded adhesively bonded lap joints. A governing partial differential equation describing the in-plane shear stress in the adherend was obtained. The differential equation was then solved for the shear stress components in the adhesive material. The closed form solution was verified using finite element analysis by predicting the stresses in an in-plane shear loaded bonded joint. The effect of geometric and material parameters on the joint behavior was studied to assist the selection of the design parameters and evaluate the manufacturing tolerance. Yang et al. [19] developed an analytical model using laminated anisotropic plate theory to study the stress and strain distribution in an adhesively bonded composite single-lap joint. The composite adherends were assumed to be linearly elastic material and the adhesive was assumed as an elastic-perfectly plastic material following Von Mises yield criteria. The stresses in the adhesive were considered to be uniform along the thickness direction. The entire coupled system was determined using the kinematics and force equilibrium of the adhesive and the adherends. The system of governing equations was then solved analytically using appropriate boundary conditions. The results from the analytical model were verified with the finite element analysis using ABAQUS . The analytical results showed a good agreement with the finite element analysis. This model along with a failure criterion were used as a tool to evaluate joint strength under the 9

cohesive failure mode of the adhesive. The developed model showed the stress distribution in the adhesive layer and not at the adhesive to adherend interface. Thermal peel, warpage and interfacial shear stresses in adhesive joints were studied by John Rossettos [20]. The author developed a closed form solution for the stresses, in a single lap joint, that were solely due to thermal mismatch and he also indicated the deformation mechanism. The analytical results gave the stress and deformation patterns due to the temperature changes. The thermal mismatch stresses were determined using a bending model. The model predicted the bond line peel stresses, shear stresses, and the axial stresses in the adherend. The analytical solution displayed a sinusoidal deformation consistent with the warpage (bending) of the adherends. Their Modified Shear Lag Model (MSLM), with no adherend bending, showed peek shear stresses at the ends of the joint. The bending model showed the peek stresses not only at the ends of the overlap but also at the interior point of the overlap region. The results for the aluminum adherends with epoxy adhesive showed the distribution of the peel, the warpage and the shear stresses. Seo et al. [21] conducted an experimental and finite element investigation to study the effect of adhesive overlap length and the adhesive thickness on the strength and stress distribution in adhesively bonded joints. Five different over lap lengths with different adhesive thicknesses were considered for the study. Tensile tests with constant cross head speed and a three-dimensional linear finite element analysis were conducted to analyze the strength and the stress distribution for the various adhesive joint configurations. It was found that the stresses were maximum at the ends and minimum at

10

the center of the adhesive area. The joint strength decreased as the adhesive thickness increased. Li, et al. [22] carried out a geometrically nonlinear two-dimensional finite element analysis to study the stress and strain distribution across the adhesive thickness in composite single lap joints. The effects of adhesive thickness and mechanical properties on the stress and strain distributions were investigated. The thin bond line was simulated using 2-element and 6-element mesh schemes, whereas the 10-element mesh was used for thicker bond line. It was found that the maximum peel stresses and shear stresses within the adhesive bond occurred near the adhesive to adherend interface at the corner ends of the overlap. The peak shear and peel stresses increased with the bond thickness and elastic modulus. An elastic three-dimensional finite element analysis was carried out by Sawa et al. [23] to analyze the stress-wave propagation and stress distribution in dissimilar single lap adhesive joints. A commercial finite element software DYNA3D was used for this purpose. The upper end of the single-lap joint was held fixed whereas the other end (lower end) was impacted by a weight. The effect of Young's modulus and adherend thickness on the stress distribution and the stress-wave propagation were investigated. The three main conclusions derived from the analysis were that the maximum principle stress occurred near the edge of the interface of the fixed adherend, the maximum principal stress increased with the Young's modulus of the fixed adherend, and the maximum principal stress increased as the fixed adherend thickness was decreased. Experiments were conducted to validate the analytical results and a good agreement was obtained between the FEM and the experimental results. 11

Yang et al. [24] carried out finite element analysis to study the interfacial stresses in fiber reinforced plastics (FRP) - reinforced concrete (RC) hybrid beams. The effect of FRP thickness, adhesive thickness and the material properties on the interfacial stresses was investigated. Results showed that the interfacial shear stresses and the normal stresses were maximum at the edges and were the main cause for interlaminar delamination. The stiffness of the RC and the FRP greatly influenced the interlaminar shear and the normal stresses. The interfacial stress concentrations and their levels increased with the increase of the FRP thickness. Goncalves, et al. [25] used a specially developed interfacial element in numerical finite element analysis to study the adhesive joint behavior. The element had eighteen nodes distributed in two faces with zero thickness. The main objective of their work was to analyze the stresses at the adherend to adhesive interface. This finite element model was applied to a single lap joint, considering linear elastic and elasto-plastic material properties. The results showed a three-dimensional nature of the stresses suggesting the importance of the three-dimensional analysis. The peek stresses at the interface were much higher than at the middle of the adhesive. This explained the reason for the adhesive joint failures at the interfaces and the importance of the interfacial stresses in bonded joints. Mathias, Grediac and Balandraud [26] derived the solutions for the bi-directional stress distribution in a rectangular composite patch under uniform in-plane loading. An orthotropic composite patch was adhesively bonded on to an isotropic substrate. The stress distribution in the patch, the adhesive, and the substrate showed bi-directional behavior. The solutions were used for comparing uni-directional and bi-directional stress 12

distribution in the adhesively bonded patch. The adhesive was subjected to only transverse shear stresses and these stresses were constant through the thickness. The contribution of the bending moments, tearing, peeling and normal stresses were neglected in deriving the solutions. (These assumptions were standard as they were also used by Adams and Peppiatt, Baker et al. (2002). The bi-dimensional solutions were validated using a finite difference model. A significant difference was noticed when comparing the classical solutions with the bi-directional results, showing the importance of the bidirectional stress formulation. Weijian et al. [4] developed an analytical expression for three-dimensional stress distribution at the bonded interfaces of the dissimilar materials. The mathematical model predicted the stress peeks at the interfaces. The interface was expressed as a general surface in Cartesian coordinates. This was helpful to model the approximate solution for different interface topographies. Finite element analysis was used to compare the mathematical model results. A linear elastic behavior with perfect bonding at the interfaces was assumed for the analysis. Their comparison of the finite element interfacial shear stresses with the three-dimensional mathematical model results showed similar trends in terms of the magnitude and shape, except at the edges. This was attributed to several assumptions in the finite element analysis. The three-dimensional stress solutions were considered more realistic than the two dimensional model [4]. The three-dimensional stress solutions were more helpful to optimize the surface topology or for surface preparation of the bonded surface to produce a reliable joint. This was necessary since ductile adhesives would fail due to shear stresses, while brittle adhesives would fail due to normal stresses. High stresses induce cracks in brittle adhesives, where 13

as a cavity induced failures in deformable adhesives. This bi-material model determined the interfacial normal and shear stresses, but the stresses were due to normal loading on the plane of the bonded joint. The model was more focused on the effect of surface topology on the interfacial stresses. Weijing, Rajesh and Erol [27] used the mathematical model in [4] to study the three dimensional interfacial stress distribution for a scarf interface (y = x / 2) in a bonded joint. A commercial finite element code ALGOR was used for the scarf interface stress analysis. The FEA was not able to replicate the interfacial shear stress distribution obtained from the mathematical model. This was attributed to the difference in their methods in maintaining boundary conditions. The FEA enforced the displacement continuity in the whole system including the interface, but did not maintain the stress free boundaries even when required by the equilibrium conditions. When comparing the normal stresses, it was found that the stresses obtained from FEA were approximately the averages of the corresponding stresses obtained for the bonded materials at their interface by the mathematical model. Based on the results, it was concluded that the mathematical model was able to predict approximately the three dimensional stresses at the bonded interface for various surface topographies As evident from literature survey, most research works [8-20] focus on two dimensional stress analysis of adhesively bonded joints. Literature shows that there are very few three dimensional analytical models which consider the thermomechanical bidirectional loading conditions to determine the effect of adhesive material on the interfacial stress distribution. Mathias, Grediac and Balandraud's [26] study showed the significant difference between the classical solutions and the bi-directional solutions. To 14

understand the need and determine the bi-directional interfacial shear stresses in bonded structures are important. Although finite element analysis is very often used for stress analysis, analytical procedures provide more fundamental insight and helps in analyzing the various critical parameters affecting the stress distribution. 1.3 Background on Bolt Bearing Behavior in Polymeric Composite Bolted Joints Bolted joints for its advantages, such as the ease of assembly and disassembly, are often preferred in many composite joining applications. Bolted joints are considered to be the weakest link in a structure, as drilling bolt holes creates high stress concentration [28]. The design and analysis of fiber reinforced polymeric composite bolted joints involves high degree of complexity and requires a special attention because of the anisotropic, inhomogeneous and viscoelastic properties. Joint geometry, stacking sequence, fiber orientation and bolt pre-load are some of the critical factors to be considered for a reliable joint design [29]. Figure 1.5 shows typical composite bolted joint parameters.

1.3.1 Failure Modes in Polymeric Composite Bolted Joints Predicting the failure load and the failure modes in a composite bolted joint is often a challenge. Previous works have characterized the failure modes and parameters associated with the failure of composite bolted joints [30]. Figure 1.6 shows some of the failure modes. The tensile failure is mainly due to the reduced joint width and is associated with the stress concentration in the fiber and matrix material. The shear out failure is mainly due to the reduced edge distance and results primarily due to the shear and compression failures of the fibers and the matrix materials. This type of failure in 15

most cases can be avoided by proper selection of lay-up and increasing the edge distance. A cleavage failure is due to a combination of reduced edge distance and width of the joint. Fastener pull through failure is due to reduced thickness to bolt diameter ratio. Fastener failure is a secondary type of failure mode and is not common in composite structural applications. Bearing failure is caused by a combination of extensive compressive force exerted on the inner bolt-hole boundary by the shank of the bolt, and the reduced hole diameter to width ratio. The net-tension and the shear-out failures are more catastrophic failure modes, where as the bearing failure is a progressive failure, and may not result in total reduction of load carrying capability of the joints [31]. Most bolted composite structures are designed for the bearing failure [32]; hence methodical understanding of effects of various joint parameters on bearing failure in a joint is of fundamental importance.

1.3.2 Bolt Bearing Behavior in Polymeric Composite Joints FRP composite laminates used in a bolted joint configuration exhibit a complex behavior when subjected to in-plane loading (Figure 1.7). Under the in-plane loading condition the bolt shank compresses the cylindrical surface of the hole (composite); this eventually deforms the composite material and may lead to bearing failure. ASTM D 5961/D 5961M [33] gives the standard bearing test procedure for a composite bolted joint. During a bolt bearing test, load- displacement data is recorded to determine the bearing failure load, bearing strength and the ultimate joint strength. Mechanical fasteners (bolts and rivets) require drilling of holes in composite materials, which ruptures the composite fiber reinforcements. This creates stress 16

concentration and may create micro-cracks and local damage around the drilled holes inducing structural instability [34]. Even with some of these draw backs, mechanical fastening of composites is a proven practical technology. The literature lists [35] the design parameters and the critical factors affecting the structural integrity and reliability of a composite bolted joint. The bearing behavior of FRP composites vary with fiber, matrix and laminate properties (thickness and orientation). Extensive experimental data is required for a reliable joint design, as generalized design formulas is difficult to achieve.

1.4 Previous Research on Bolt Bearing Behavior in Polymeric Composite Joints The study of bearing behavior, bearing strength, and bearing failure in a composite bolted joint becomes essential as bolted joints represent the weakest link in a mechanical system. Bearing strength and bearing failure modes depend on various joint parameters. Some of the critical parameters include joint geometry, hole clearance, type of fasteners, bolt clamping pressure and the operating conditions. The following section gives an overview of previous research carried out on the bolt bearing behavior in composite structures.

1.4.1 Influence of Coupon Geometry and Laminate Properties on the Bolt Bearing Behavior Vangrimde and Boukhili [36] studied the effect of coupon geometry and laminate properties on the bearing stiffness of Glass fiber Reinforced Polyester (GRP) composite. The focus was on the load-displacement response of GRP laminate in a single-bolt double-lap composite joint. Six different laminate materials with different amounts of 0 and 90 roving; Chopped Strand Mat (CSM) were tested to obtain the load-displacement 17

data. Different coupon geometries considered were, a standard coupon with width to diameter ratio of 6 and edge distance to diameter ratio of 3, a long coupon, with width to diameter ratio of 6 and edge distance to diameter ratio of 6. a small coupon, with width to diameter ratio of 2 and edge distance to diameter ratio of 3. Bearing stress was calculated P using o b r = ' where P= applied load, D= bolt hole diameter and h= thickness of the Dh laminate, and bearing strain was calculated using , where 8 represents the

deformation in the bolt hole. These relations were used to obtain the bearing stress verses bearing strain curves. The bearing stress-strain curves had three distinct regions: initial sliding, linear bearing response prior to the damage, and a non-linear post damage stress region. The bearing stiffness was determined from the initial linear part of the curve. On average, joints with reduced width showed 26% more yield than the standard joints. The end distance had an insignificant effect on the bearing stiffness; an average increase of 6% stiffness was noticed in the longer end distance joints compared to the standard joints. The bearing stiffness was higher for the smaller coupon geometry with more axial reinforcement. The experimental observations on the influence of coupon width, the end distance and the laminate properties on the bearing stiffness were approximately verified by 2-D Finite Element Analysis. Li Hou and Dahsin [37] investigated the three-dimensional size effect and thickness constraints on the single-pin double-lap glass-epoxy composite joints. The constraints in the thickness direction, the composite thickness, the bonding strength through the laminate thickness and the clamping force from bolting had a significant 18

effect on the damage and the strength behavior of the joints. The experimental data revealed that the joint strength decreased as the joint size was increased. The failure modes also changed from an initial bearing damage, followed by net section failure to a direct catastrophic net-section failure as the joint size was increased. Composite joints with low thickness constraints showed fiber buckling and delamination, resulting in a bearing failure before net section failure. These joints showed large displacement before final failure and consumed more energy (larger area under the load-displacement curve), hence were more ductile than those which failed by direct catastrophic net-section failure. Cooper and Ibvey [32] experimentally studied the effect of joint geometry and bolt torque on structural performance of the single bolted pultruded GRP (Glass Reinforced Polymer) joints. Tests were conducted to determine the effect of geometric ratios, edge distance to bolt hole diameter ratio and width to hole diameter, and the bolt torque on the failure load, failure modes and stiffness of the single bolted joint. The failure loads of the lightly clamped (3 Nm bolt torque) and the fully clamped (30 Nm bolt torque) joints increased by 45% and 80%, respectively, when compared to pin joints. It was also observed that by increasing the bolt torque the critical e/D and W/D ratios also increased significantly. The initial stiffness of the single-bolt joint was affected mainly by the W/D ratio. The e/D ratio and the bolt clamping torque had only a small effect on the initial joint stiffness. The load vs bolt displacement graphs for each tested joint showed an initial bolt displacement of 0 to 0.3mm despite the tight bolt fitting. After the initial bolt displacement the load displacement response was approximately linear until the joint either failed (small e/D or small W/D values) or the initial stiffness reduced. An

19

irreversible bearing damage was observed at the load where the joint stiffness changed, and this was defined as the damage load. Oh, Kim, and Lee [38] worked on bolted joints for hybrid composites made of glass-epoxy and carbon-epoxy under tensile loading. The design parameters investigated were laminate ply angle, stacking sequence, the ratio of glass-epoxy to carbon epoxy, the outer diameter of the washer and the clamping pressure. Results showed that the peak load occurred before the maximum failure load due to the delamination of the laminate under the washer. The static test results of the hybrid composites with stacking sequences of [0C/ 45G/ 45C/90C]S and [OG/ 45C/90G]S (C: carbon-epoxy, G: glassepoxy) revealed that the bearing strength increased as the 45 plies were distributed evenly along the thickness direction irrespective of the joint material (glass-epoxy or carbon epoxy) and the stacking pattern. The bolted joint of [+45C/-45C/+45C/ (OG) 2, / 45C/+45C/ - 45C/ (90G) 2] s laminate, with 35.5% volume fraction of glass-epoxy yielded the highest bearing strength. The bearing strength increased as the bolt clamping pressure increased to 71.1 MPa, thereafter the bearing strength saturated to a constant value. The failure mode changed from bearing failure to tension failure when a 20mm diameter washer was used. The finite-element analysis predicted the first peak load; however, it could not predict the maximum failure load. For a more accurate prediction of the joint strength it was suggested to consider the effects of material non-linearity, the friction between the joint materials, and the stiffness reduction due to failure. An experimental and numerical study was carried out by Aktas and Dirikolu [39] to investigate the strength of a pinned-joint made of carbon epoxy composite with [0/45/45/90] s and [90/45/-45/0]s stacking configuration. ASTM D953 standard was followed 20

for the experimentation, and a finite element analysis was performed for verifying the experimental results. The ratio of the edge distance to the pin diameter (e/D), and the ratio of the specimen width to the pin diameter (W/D) were systematically considered to analyze the strength and the failure modes in the composite joints. The results from both the analyses showed good agreement. When the e/D > 4 and the W/D > 4, bearing failure was dominant, where as when the ratios were below four, net tension, shear out and mixed mode failure were observed. The [90/45/-45/0]s joint configuration showed 20% higher bearing strength than the [0/45/-45/90]s configuration. The finite element results predicted an average 20% lower bearing strength values when compared to those from the experiments. Yamada-Sun failure criterion was used to determine the failure loads and failure modes in the analysis. Besides its availability in commercial FE codes such as ANSYS , the criterion also gave satisfactory predictions of both failure load and failure modes. Alaattin [40] experimentally investigated both static and dynamic bearing strengths of a pinned-joint carbon epoxy composite plate with [0/45/-45/90]s and [90/45/45/0] s stacking configurations. In order to obtain the optimum geometry the ratio of edge distance to pin diameter (e/D), and the ratio of specimen width to pin diameter (W/D) were varied to obtain the static bearing strength and the S-N fatigue curves. The experiments showed [90/45/-45/0]s sequence was about 12% stronger than the [0/45/45/90]s sequence in terms of bearing strengths. Additionally, the optimum geometry was attained when e/D and W/D ratios were greater than or equal to 4. The fatigue strength reduced up to 63% of the static strength as e/D and W/D ratios increased.

21

McCarthy et al. [41] investigated the effect of bolt-hole clearance on the strength and stiffness of single-lap, single bolted composite joint. The graphite/epoxy with quasiisotropic [45/0/-45/90]5s and zero dominated [(45/02/-45/90)345/02/-45/0]s stacking sequence were considered for the study. Bolt-hole clearanceses of Oum, 80(j.m, 160um and 240um with various torque levels and different bolt types were investigated. The joint stiffness, 2% offset joint bearing strength, the ultimate bearing strength and the ultimate bearing strain were analyzed for various joint configurations. An increase in clearance was found to reduce the joint stiffness and increase the ultimate strain for all test configurations. The finger tightened bolts with negligible clamp load showed a relation between the bolt-hole clearance and the joint strength, where as this was absent for the counter sunk and the torque tightened joints. V.P. Lawlor, et al. [42] continued their experimental investigation on the effect of bolt-hole clearance on the single-bolt, single-shear bolted composite joints. The varying bolt-hole clearance was obtained by using a constant bolt diameter and varying the hole diameter. The same four hole clearances used in [41] (Oum, 80um, 160um and 240um) were used for the analysis with countersunk and protruding head bolts. The initial delay in the load take up was observed in the load-displacement data showing the bolt-hole clearance effect. Two main regions, linear and nearly linear were observed in the loaddisplacement curves. The bolt-hole clearance had an insignificant effect on the maximum load taken by the composite bolted joint. The maximum displacement decreased with the decrease in clearance. All tested joints initially failed by bearing failure and exhibited a drop in stiffness with an increase in bolt-hole clearance. The

22

(& + f\ V2

bearing stress - bearing strain were plotted using o b r = k

and s b r = D h K D

^ ,

where P = load, D = diameter, h - coupon thickness, and k = load per hole factor (1.0 for single fastener or pin tests and 2.0 for double fastener tests ), 81 and 82 are displacements, respectively, in extensometers 1,2, and K=l for double shear tests and K=2 for single shear tests. For the protruding bolt-head joints, the bolt-hole clearance effect on the ultimate strength was essentially negligible. Unlike the ultimate strength, ultimate strain was significantly affected by the bolt-hole clearance. An increase in bolthole clearance increased the ultimate strain. This was due to the extensive laminate damage by the concentrated load on the laminate hole. The bolt hole clearance had a significant effect on the joint stiffness and ultimate strain and less effect on the joint strength. The bolt-hole clearance showed a delay in load take up and this was considered to be a significant factor in multiple bolted joints. 1.4.2 Effect of Bolt Tightening Torque and Clamping Pressure on Bearing Behavior Claire et al. [43] studied the effect of stacking sequence and clamping pressure on the carbon/epoxy bolted composite joints. Three different symmetric lay-ups, cross-ply [(0/90)4] s, angle-ply [(+45/-45)4]s and quasi-isotropic [(0/45/90)2]s were used in the study. The bolt displacement and the local strains around the bolt-hole edge were recorded for the analysis. Bearing stress Vs hole elongation curves and the bearing stress Vs strain curves showed significant effect of clamping pressure on the initial bearing stress and the maximum bearing stress. Tightening the bolt increased the initial bearing stress by 22% and the maximum bearing stress by 105%. The angle ply and the quasi23

isotropic lay-ups showed similar bolt-hole elongation which was larger than the cross-ply bolt-hole elongation. A significant increase in strain was observed for the dowel pin joints when compared to the finger tightened bolted joints. The increase in clamping pressure increased the post-peak stiffness, where as the initial stiffness and the bolt-hole elongation decreased significantly. The bearing stress vs. hole elongation curves showed that the angle ply lay-ups had the lowest initial stiffness and the cross-ply lay-up had the highest initial and post-peak stiffness. Orienting the fibers at an angle of 45 improved the bearing behavior. The results from the rosette strain gage positioned on the bearing zone showed a linear behavior for the angle-ply laminate, where as a nonlinear behavior was observed for the cross-ply and the quasi-isotropic lay-ups. This non-linear behavior was mainly due to the stresses corresponding to the initiation of damage due to local delamination around the hole. This study investigated some of the most significant parameters affecting the bearing behavior in bolted composite joints, and the data from this could be used for the failure mechanism study and for validation of the numerical finite element model. Park [44] investigated the effect of stacking sequence and clamping force on delamination bearing strength and ultimate bearing strength of mechanically fastened carbon/epoxy composite joints using an acoustic emission (AE) and load-displacement technique. Orthotropic and quasi-isotropic laminate lay-up configurations with four different clamping forces were considered for the study. The stacking sequence and the clamping pressure had a significant effect on the delamination and ultimate bearing strength of the mechanically fastened composite joint. The comparison of orthotropic laminate pinned joints, with stacking sequence [9(V06]s and [(V906]s, showed similar 24

ultimate bearing strengths for both lay-ups, but the laminate with [90g/06]s lay-ups had almost twice the delamination bearing strength as compared to laminates with [0^906] s lay-ups. The trends of variation of the delamination and ultimate bearing strengths of bolted joints were similar to that of the pinned joints for different stacking sequences. The laminate with [903/+453/-453/03]s lay-ups had the highest ultimate bearing strength, whereas, the laminate with [903/03/+453/-453]s lay-ups had the highest delamination bearing strength. The 90 layers had an important role in delamination bearing strengths; the laminate with 90 layers on the surface had higher delamination bearing strength than the laminate with 90 layers in the center. An increase in clamping pressure increased the ultimate bearing strength to saturation, whereas the delamination bearing strength increased progressively. The clamping pressure suppressed the delamination and the interlaminar cracks. The failure mode changed from catastrophic fracture to a progressive failure as the clamping pressure was increased. Sun, Chang and Qing [45] experimentally studied the effect of lateral supports on the bearing failure in composite bolted joints. Graphite/epoxy composite coupons were investigated for various washer sizes and clamping forces. The clamping load history was recorded as the tensile load was applied to the joint. The clamp load increased with the applied tensile load; Poisson's ratio contributed for the initial increase in the clamping force until the bearing failure, and the lateral constraints preventing the material damage increased the clamping force after the bearing failure. The pinned joints had the lowest joint strength compared to the bolted joints with lateral clamping force. The experimental

25

results were compared with the predictions from the 3DBOLT code included in ABAQUS . The comparison included failure load, joint response, the failure modes for the joints with various clamping configurations and clamping loads. The predictions from the code agreed well with the experimental results. Yan et al. [46] conducted an experimental study to investigate the effects of clamp-up pressure on the net tension failure of bolt-filled laminated composite plates. The tensile strength and the failure behavior of both open bolt-hole and bolt-filled hole were evaluated for graphite-epoxy composite plates. The effect of washer size on the bolt filled-hole net tension strength, and the net tension failure behavior of composite bolted joints were also investigated. X-rayradiographs were used to analyze the specimens after preloading and at different stress levels for the purpose of characterizing the failure modes and damage progression inside the composite. The bolt filled-hole composite laminate was prone to fiber matrix splitting and delamination due to its sensitivity to the bolt clamp up effect. Higher the clamping pressure lower was the tensile strength of the bolt filled-hole laminate. The tensile strength reduced by about 20% with the bolt clamping pressure. The washer to bolt-hole diameter ratio (Dw/D) (less than 2) had negative effect on the tensile strength of the bolt filled-hole laminate. Unlike bolt filled-hole laminate, the tensile strength of the bolted composite joints increased with clamping pressure. Khashaba et al. [47] investigated the effect of bolt tightening torque and washer size on the bearing behavior of glass fiber reinforced epoxy composite ([0/45/90]s) bolted joints. Damage analysis was carried out to understand the failure mechanism. 26

Tightening torques of T = 0 Nm, 5 Nm, 10 Nm and 15 Nm and the washer sizes of D w o = 14 mm, 18 mm, 22 mm and 27 mm were used in the study. Mechanical properties such as tensile, compressive, and in-plane shear were determined both experimentally and theoretically. The joint stiffness increased with decreasing the washer sizes under constant tightening torque. This was mainly due to the increase in the clamping pressure resulting from reduced washer size. The washer size of 18 mm and the tightening torque of 15 Nm produced the optimum clamping pressure. The composite bolted joints with 14 mm washers had higher clamping pressure but showed reduction in maximum bearing strength. The load displacement curves of the finger tightened bolt joint showed least stiffness with non-linear behavior that indicated the unstable development of internal damage. Most of the tested specimens failed in a sequence, delamination, and net tension failure at 90 laminate, shear out failure at 0 layers and final failure which was nearly catastrophic due to the bearing failure of 45 layers.

1.4.3 Progressive Failure Mechanism in Composite Bolted Joints Lawlor, Stanley and McCarthy studied the effect of bolt-hole clearances on the damage development in carbon-epoxy bolted composite joints [48]. Single-lap singlebolted joints subjected to tensile load were considered in the analysis. The loaddisplacement data was recorded for determining the bearing failure load and the ultimate failure load of the bolted joints. The bolt-hole clearances of 0 urn, 80 urn, 160 |j,m and 240 urn were selected for the 8mm bolt-hole. The joint initially failed due to bearing damage before the final failure. For the larger clearance bolt-boles, the load Vs displacement data showed a delay in load take-up, initial non-linearity after the load take27

up, a lower slope in the linear portion of the force Vs deflection curve, and a significant loss of stiffness. The low clearance joints failed by bolt failure, whiles, the larger clearance joints had no bolt failure but a large joint displacement. Further examination of the failed specimen showed significant damage at the shear plane of the laminate for all clearances Claire et al. [49] studied the failure mechanisms of bolted carbon epoxy composite joints under tensile loading. They studied the effect of stacking sequence and clamping force on the failure mechanism. The evaluation of the external macroscopic damage was done using digital photographs, and the internal damage using the optical microscopy. The damage analysis showed that the fiber orientation around the bolt-hole boundary had an influence on the failure initiation. The matrix cracking, the interlaminar shear delamination and the compressive failure were the prominent failure modes observed in all configurations. The shear cracks in the tight bolted joints (finger tightened and torque tightened) were seen on the surface of the laminates. The micrographs of same specimens showed a severe internal damage over the entire bearing zone. The damage pattern for the angle ply was different than the quasi-isotropic laminate for similar clamping pressure. The quasi-isotropic laminate showed multiple shear cracks spreading over the bearing zone, where as, the damage was more severe and spread over a large area for the angle-ply laminates. For the specimens subjected to higher stresses, the prominent shear cracks appeared on a plane distant from the bearing plane and reached the laminate surface. Yi Xiao [50] investigated the bearing strength and failure process in double-lap single-bolted composite joints. The load displacement data was recorded using the load 28

cell and a non-contact electro-optical extensometer. The bearing strength was based on the bearing load at which the pin relative displacement was deformed by 4% of the pin diameter; the ultimate failure load was determined from the peak point of the loaddisplacement data. The load-displacement data had two prominent regions, an initial linear region and a non-linear region before it reached the load at 4% displacement; this indicated the micro damage beginning in the tested coupon. During the static tests, along with the load displacement data the acoustic emission was recorded for the fracture analysis of the specimen. A sharp change in the AE signal observed at the start of the nonlinear behavior indicated the beginning of the damage. The photographs of the specimen surfaces and the X-ray-radiographs of bearing damage were recorded for the investigation of failure mechanism. The local delamination under the washer, the out-ofplane shear cracks, and the fiber matrix splitting cracks progressed with the increase in the tensile load. These failure modes were responsible for the final failure of the bolted joints. Further the study was extended for double bolted joints to investigate the bearing load proportions (load carried by each individual bolts). The load-displacement and the elongation data were recorded using the load cell and an extensometer placed at each bolt-hole. The bolt-hole elongation for each hole was significantly different with nonlinear behavior. The relative displacement between the bolt-hole was converted into ratio of loading proportion. The damage analysis using SEM and the X-ray radiography showed that the compressive damage state around each hole differed due to the difference between the loading proportions. Hong, Chang and Fu-Kuo [51] experimentally investigated the effect of clamping pressure on the bearing response and bearing failure mechanism of composite bolted 29

joints. The bearing damage was characterized either as a pure bearing failure, which had no lateral support, or as a bolt bearing failure which contained lateral supports with various levels of clamping pressure. Specially designed semi-circular notched specimens were used to characterize the pure bearing damage* while a load cell was used to monitor the clamping pressure in the bolted joint. Three different bolt-hole diameters and two different laminate thicknesses were considered for the study. The failed specimens were inspected using the microscope and Xray-radiography. The shear cracks induced by the compression failure were the main cause for the bearing failure. The bearing failure without lateral constraint was catastrophic. The critical bearing distance (5C) was identified beyond which the catastrophic failure occurred, and this was proportional to the laminate thickness and independent of ply-orientation. Lateral support prevented the catastrophic failure and increased the bearing strength. Liyong [52] studied the effect of two extreme positions of the loose fit fasteners on the bearing failure in double-lap bolted composite joints. When the washers were placed to their extreme positions in the loading direction, an unconstrained gap between the bolt shank and the washers were observed, whereas, when the washers were placed in the extreme positions in the opposite loading direction no unconstrained gap existed. The results showed that the bolted joints with an unconstrained gap had lower initial failure load than those joints having no unconstrained gap. However, there was no difference in the ultimate failure loads. Different strategies for improving the bearing strength were examined by A.Crosky et al. [53]. Fiber steering (directed placement of fibers), matrix stiffening by nano-reinforcement and the through thickness reinforcement using z-pins were the 30

different strategies analyzed in the study. The bearing strength was improved by 36% using two sets of steered fibers in tensile and compressive principle stress direction, as obtained by the FEA analysis. The addition of clay nano-particles to the matrix resin stiffened the matrix but induced a different premature failure mode, which reduced the joint bearing strength. However it was found that the bearing behavior would improve by avoiding the premature failure. Through thickness reinforcement using z-pins increased the ultimate bearing load by 7%, while the bearing strength remained the same

1.4.4 Finite Element Analysis of Composite Bolted Joints Ireman [54] was one of the earliest researchers to develop a three dimensional finite element model of a single lap composite bolted joint to determine the non-uniform stress distribution through the thickness of the composite laminate in the vicinity of a bolt-hole. Number of significant joint parameters including the laminate lay-up, the bolt diameter, the bolt type, the bolt pre-tension and the lateral support conditions were investigated. The commercially available finite element code IDEAS for pre and post processing and ABAQUS were used in the study. The experiments were carried out to verify the strains, the displacements and the bending effects obtained from the finite element analysis. A good agreement for the measured strains was obtained and the primary differences were attributed to the difference between the frictional coefficients in the experimental and the finite element analysis. For the displacements, the agreement between experimental results and analytical results were not as good as the strains. This difference was attributed to the misalignment between the joint coupons and the friction between them. It was clear from this study that the finite element method was able to 31

predict the through thickness stresses, strains at the vicinity of the bolt hole and many such complex design parameters in a composite bolted joint. Finite Element Analysis was carried out by Johan and Joakim [55] to study the effect of secondary bending on the damage behavior and strength of composite bolted joints. The commercial finite element software ABAQUS was used to model the composite joint assembly. An orthotropic linear material property was considered for the composite plates. Both the tensile and the compressive loading were used to study the bolt bearing behavior. For tensile loading, the secondary bending increased the bearing strength and reduced the ultimate joint strength. The bearing strength was much lower for compressive loading, this was due to the reduced bolt to bolt-hole contact. The joint stiffness also reduced due to secondary bending. The secondary bending influenced various macroscopic failure modes and in the process changed the ultimate failure mode. It was recommended to reduce the secondary bending in the bolted joints which resulted in an eccentric load path. Tserpes et al. [56] conducted a finite element analysis to investigate the effect of failure criteria and material property degradation rules on the tensile behavior and strength of graphite/epoxy laminate bolted joints. The analysis was based on three dimensional progressive damage model (PDM) developed earlier by the authors. The PDM comprises the components of the stress analysis, the failure analysis and the material property degradation. The experiments were conducted to compare the finite element results. The effect of various joint geometries and stacking sequence on the loaddisplacement behavior were investigated. The failure load predicted was influenced by

32

the combination of failure criteria and material property degradation rules considered in this study, while the stiffness of the joint was relatively accurate. Dano et al. [57] developed a two-dimensional finite element model to predict the response of the pin-loaded composite plates. The model was developed taking into account the contact at the pin-hole interface, the progressive damage, the large deformation theory, and the non-linear shear stress-strain relationship. To predict the progressive ply failure, the analysis combined Hashin failure criteria and the maximum stress failure criteria. The influence of the failure criteria and nonlinear shear behavior on the strength prediction and the load-pin displacement were investigated. Based on the theoretical and experimental results it was concluded that the maximum stress criteria had more realistic strength predictions for linear shear stress-strain relationship. Both the Hashin failure criteria and maximum stress failure criteria showed same predictions for the non-linear shear stress-strain relationship. The failure strengths predicted from the developed model was within 1-15% of the experimental results McCarthy et al. [58] developed a three dimensional finite element model using MSc. Marc (commercially available software) to study the effect of bolt-hole clearance on the behavior of single-bolted single-lap graphite-epoxy bolted joints. Experiments were conducted to validate the finite element analysis. Issues in modeling the contacts between the joint parts were studied. The surface strains and joint stiffness measured in the experimental study were compared with the finite element study involving the variations in the mesh density, element order, boundary conditions, material model and analysis type. Three dimensional affects such as the through thickness stresses and strains, secondary bending and bolt tightening were represented in this study. 33

It is evident from the literature that the bearing behavior in composite joints is dependent on laminate lay-up, joint geometry, bolt preload and clamping pressure. The initial bolt-load in a joint exerts pressure perpendicular to the plane of the material, and most of the composites haVe their fibers oriented along the plane of the material. The composite bolted joint experiences a biaxial loading condition when subjected to both inplanes loading and bolt preload; hence the bolt preload and joint clamping pressure is a critical component in a composite bolted joint design. Figure 1.7 shows a typical joint subjected to in-plane loading and bolt preload. Very few research works focus on the effect of bolt preload on thick composite bolted joints. Thick composite structures do not necessarily behave in the same manner as thin structures with the same laminate orientation. The strength and failure modes of thick composite joints cannot be scaled, nor predicted based on the results of thin composite joints [33]. The influence of bolt preload level investigated in the literature, according to this author's knowledge, does not simulate the performance of heavily loaded composite structures. The study on behavior of multi-bolted composite joint is also limited. The effect of tight and loose fastener combination on the behavior of multi-bolted composite joints is hardly available in the literature. This has been the main motive for studying the effect of bolt preload on bearing behavior in composite bolted joints.

1.5 Objective of the Dissertation


In the first part of this dissertation, an analytical model based on continuum

mixture theories is developed to study the bi-axial interfacial shear stresses in adhesive bonded joints due to thermo-mechanical loading. The model predicts the effect of
34

adhesive thickness and properties on the bi-axial interfacial shear stresses. The interfacial shear stresses between the adhesive and each adherend is determined using the constitutive equations. Numerical results show that both the adhesive thickness and the material properties have a significant effect on the thermo-mechanically induced interfacial shear stresses between the adherends and the adhesive. The developed model inherently has the capacity for optimizing the selection of the adhesive thickness and material properties that would yield a more reliable bonded joint. In the second part, experimental and numerical investigations are carried out to study the affect of bolt-load level, washer geometry and bolt tightness on the bearing behavior of composite joints. A double lap shear joint is considered to study the effect of bolt-load levels and washer geometry, while a double bolted single lap shear joint is considered to study the effect of bolt tightness and joint material on the bearing behavior of composite bolted joints. Finite element analysis using a commercially available code ABAQUS is carried out to validate the experimental results. Failure analysis using optical microscope and digital photography is conducted to analyze the bearing failure mode in composite laminates.

35

Figure 1.1 Example of an adhesively bonded joint

Adhesive shear stresses

Adhesive Direct (peel) stress

(a)

(b)

Figure 1.2 Shear stresses and peel stresses in an adhesive bonded joint: (a) Shear stresses; (b) Peel stresses. (Modified from [3])

36

Tension

Adherend I Adherent! 2 CTEi> CTE2 Compression

Figure 1.3 Shear stress due to difference in coefficient of thermal expansion (Modified from [3])

Adherend Adhesive m m ^ Failure Location

- Adherend

(a)

(b)

(c)

(d)

Figure 1.4 Failure modes in adhesive bonded joints: (a) Cohesive failure; (b) Interfacial
failure; (c) mixed failure mode; (d) Adherend failure. (Modified from [5])

37

>-" j 'X Thickness =t

N
Width = W

^jw Bolt hole diameter = D

Edge distance =E

Figure 1.5 Joint parameters in a typical composite bolted joint.

38

(a)

-(b)

(c)

(d)

(e)

(f)

Figure 1.6 Failure modes in composite bolted joints: (a) Tension failure; (b) Shear failure; (c) Cleavage failure; (d) Bearing failure; (e) Fastener pull through failure; (f) Bolt failure. (Modified from [48])

earing Test fixture Bolt

1
Composite Laminate

Loading Direction

Figure 1.7 Single bolted double lap shear joint subjected to in-plane loading.

40

CHAPTER TWO EFFECT OF ADHESIVE THICKNESS AND PROPERTIES ON THE BI-AXIAL INTERFACIAL SHEAR STRESSES IN BONDED JOINTS USING A CONTINUUM MIXTURE MODEL

In this chapter, an analytical model is developed to study the interfacial shear stresses which accounts for effects of bidirectional mechanical loading, uniform thermal loading, thickness of the adhesive and the adherends, and the mechanical properties of the adhesive and adherend materials. A continuum mixture theory developed by Nayfeh et al. [10, 14] is extended to analyze the statically induced bi-directional interfacial shear stresses in adhesive bonded joints subjected to various mechanical and thermal loadings. Two sets of governing partial differential equations are solved for the displacement field in each layer of the joint. The interfacial shear stresses between the adhesive and each adherend is determined using the constitutive equations that are developed for the model. Numerical results show, both thickness and the material properties of the adhesive have a significant effect on the thermo-mechanically induced interfacial shear stresses between the adherends and the adhesive. The proposed model inherently has the capacity for optimizing the selection of the adhesive thickness and material properties that would yield a more reliable bonded joint.

2.1 Formulation of the Problem A linearly elastic model is considered for isotropic adherends that are perfectly bonded by a layer of isotropic adhesive as shown in Figure 2.1a. All layers are assumed 41

to have the same length L x and width Lz; layers are stacked normal to the y-axis. Because of symmetry, a one quarter model is used, as shown in Figure 2.1b. The thickness of the adhesive is 2ho and the thicknesses of the adherends are considered to by h a where a represents materials 1 and 2. The displacement vector at any point is described in terms of its components u(x, y, z), v(x, y, z), w(x, y, z) in the x, y, and z directions, respectively. Figure 2.1a shows the model of a mechanically loaded bonded joint that is composed of two plates (adherends) and the sandwiched bonding agent (adhesive); the thermal loading is a uniform temperature change from the ambient temperature. In light of the described thermo-mechanical loading, the y-displacement v is assumed to be independent of x and z
/ d v q _ d v q _ Q x

dx

dz

2.1.1 Equilibrium and Constitutive Equations If body forces are neglected, the equilibrium equations within each material a (a = 1,0,2) are respectively given by [59] 3o- xa dx
|

daXya 3y

3q X za dz

= Q

+ + 3- = 0 (fy dx dz derm , dojun dz dx The constitutive relations are given by 42


|

(2.2)

^gyza _ Q dy

,-

dx

dy

dz

-YaT

(2.4)

Oya = ( 2 M-a + ^a) "T^ + la

3ua

3w_a

8x

3z

Y T
'a

(2.5)

dza = (2 ^ a

^a)-^ + la

dUg

3vg

dz

dx

dy

Y T 'a

(2.6)

tfxya

M^a

dug , dvg

(2.7)

dy

9x

Oyzg - M^a

3vg + 3wg dz dy

(2.8)

0"zxg M* a

dWg

dUg

dx

3z

(2.9)

where A,a and jxa are the Lame' constant and shear modulus, respectively, while a x , c y ,
a

z> xy> yz, CTxz a r e the components of the stress tensor. In equations (2.4-2.6), T

represents the temperature change from the ambient value, y is related to the coefficient of linear thermal expansion p, the Bulk Modulus K, Young's Modulus E, and the Poisson's ratio u as follows
Yg =
3

PgKg

EPg

(l-2ua)

2.1.2 Continuity Conditions At the interface surfaces between the adherends and the adhesive, the displacement field must be continuous. Additionally, the normal stresses o y 43

perpendicular to the laminate thickness, as well as the in-plane shear stresses o"yx, o"yz, and o z x are continuous. With reference to Figure 2.1a, the following continuity conditions are used ui=uo> vi=vo> wi=wo, o-yi= Oyo, cjXyi= c xy o, oZyi = GjyQ at the interface between materials 1 and 0 (y = hi). Similarly with reference to Figure 2.1a, the following continuity conditions are used ui=uo, V2=vo> W2=wo> a y 2= oyo, o xy 2 = oxyo, o zy 2 = Ozyo at the interface between material 2 and 0 (y = h).

2.1.3 Continuum Mixture Equations In the following formulation the aim is to use the above equations to obtain two sets of partial differential equations for x and z displacements that will ultimately yield the solution for the interfacial shear stresses. In arriving at such equations, it is guaranteed that the continuity conditions are satisfied. Hence, the behavior of this adhesively bonded model is described as that of an equivalent higher order interacting continuum. This analysis is carried out in a continuum mixture format by eliminating the ydependence and defining some average values for the displacement and stresses over their respective laminate thickness in y-direction. To this end, equations (2.1) and (2.4) are averaged over the respective laminate thickness (for a =1, 0 and 2) according to
, v l hi

(quantity), = /(quantity \ dy hi o 44

(2.1 Oa)

hl+2ho

(quantity )0 = 2h 0

J (quantity ) 0 dy
hl

(2.10b)

h +h2

(quantity )2 = h2

/ (quantity ) 2 dy
h

(2.10c)

Applying the continuity conditions on a X y ,v a and averaging according to equation (2.10a-2.10c), the following set of equations are obtained 3ojd + 3oxzi 9x 3z da xO j 3o~xz0 3x 3z 3a x2 [ 3o~xz2 3x 3z
/

hi

XxylO

(2.11a)

2ho

TxylO _ T X y02

(2.11b)

h2

T X y02

(2.11c)

hi a -(2u xl 1+

+ YiT ) ^ - U dx v dz J

3wi

zr 'N

= vio

(2.12a)

2h 0

CJ X O""( 2 ^O + ^ O ) T "

3x

^o

dz

+ Y0T

vo2~vio

(2.12b)

h2 Ox2' X2 where,
Txyl0

dx

A.2

3w2

az

y2T
are the

-V02

(2.12c)

= a x y l (hj = axy o (hi), xxy02 x z y i o , a n d Tzy02

equilibrium interaction

terms that represent the interfacial shear stresses on both sides of the adhesive, while vio = vi (hi) = vo (hi) and vo2
are m e

constitutive relation interaction terms.

45

Similarly, if equations (2.3) and (2.6) are averaged (for a =1, 0 and 2) along y direction according to equations (2.10a), (2.10b) and (2.10c), equations similar to (2.11 a, b, c) and (2.12 a, b, c) in terms of oz a r e obtained as follows

hi

dCTzl + dCxzl dz dx

"TyzlO

(2.13a)

2ho

dC z Q ! d a xzO

dz

dx

Tyzl0_tyz02

(2.13b)

h2

dg z 2

dO~xz2

3z

9x z'

Xyz02

(2.13c)

3wi Ozl" foi + A-i) dz

M
v

^ ^ \ 9u_i y

+ YlT

dx

V10

(2.14a)

dwo 2h 0 (2li0 OzO~V^o + ^o) dz Xo h2


Oz2"

Xo

3uo + Y0T dx

vo2 _ vio

(2.14b)

dz

. 3u2 y2T 3x , +

-VQ2

(2.14c)

2.1.4 Evaluation of the Interaction Terms If the constitutive relation (2.5) is averaged according to equations (2.10a, b, c) and the continuity conditions o y i = GyO = a y 2 = a y are imposed, the following equations are obtained dui dwi hi Oy-A,l + YiT dx dz Ul + 2Ri) 46

vio

(2.15a)

2ho
(Xo + 2n 0 )

A.0

8UQ

8x
3u2 3x

3wp ToT 9z +
3w2 3z

V20~vio

(2.15b)

h2

ta + 2n2)

Ov_^2

+ Y2T

V02

(2.15c)

2.1.5 Relationship Between Shear Stresses and Average Displacements The following linear distribution (across each laminate thickness) is assumed for the shear stresses 0" xva (a = 1, 0, 2).
Gxyl- Aiy

0<y

<hi

(2.16a)

OxyO-XxylO"

2h0

lTxylO-TXy02j

hi<y

<h

(2.16b)

axy2 = A 2 ( y - ( h + h 2 ) ) - -

h <y <h

+h2

(2.16c)

Similarly, an assumed linear distribution of the shear stress G v z a in each laminate a (a = 1,0,2) is given by
o-zyl-Biy
r

0<y<hj

(2.17a)

-hi^ 2h0 ,TzylO-Tzy02)


f

CzyO-TzylO"

"

hi ^ y ^ h

(2.17b)

azy2 = B 2 (y - ( h + h 2 ))

--

h < y < h +hi

(2.17c)

where the arbitrary constants A i , A2, B j , B2 are defined as follows

TxylO Ai =

TzylO Bi~

hl

hi

47

A2 =

TXy02 ;

h2

Tzy02 B2 = ~

h2

It can be seen that the assumed shear stress distribution across the laminate thickness, in equations (2.16a-2.17c), is dictated by the continuity conditions on a x y and c z y a s listed in continuity conditions and illustrated in Figure 2.2. Substituting for a x y i from equation (2.16a) into equation (2.7), multiplying the resulting equation by y, and then integrating according to equation (2.10a) by parts gives

TxylO

1L

= ui(hi)- u i

(2.18a)

for material 0, equations (2.16b), (2.7), and (2.10b) are used to obtain 2h 0 TxylO
Ho
XXy02

-uo(n)-u(

(2.18b)

for materials 2, equation (2.16c), (2.7), and (2.10c) are used to obtain
Jl2_ 3u 2

T X y02

U2\n

h)

)~U2

(2.18c)

Following a similar procedure for ozycx (a = 1, 0, 2), equations (2.17a, b, c) and (2.10a, b, c) yield
TzylO" ~ Wl(hl)~wi

(2.19a)

3Hi

2ho tzylO
Ho

TZy02

wo (h)-wo

(2.19b)

Tzy02^

= W2(n)~W2

(2.19c)

3^ 2 48

Finally, two relations similar to equation (2.19b) are derived as follows, for relating the interfacial shear stress values TXylO
an

d TXy02

to

the difference between the

interfacial displacement UQ (hi) and the average displacement u 0 . From Figure 2.2, the shear stress o xy o may be expressed in the alternate form y - ( h i + 2h 0 )
OxyO-XXy02"

2ho

llxylO-Txy02J J

(2.20)

Substituting for a x y o from equation (2.20) into equation (2.7) with cc=0 and then multiplying the resulting equation by [y-(hi+2ho)], and following the same procedure which lead to equation (2.19a, b, c), gives 2h 0 Txy02 Ho
TxylO

-Uo(hl)-uo

(2.21)

Following a similar procedure, the second relationship relates the interfacial shear stresses Tyzio and xyZ02 to the difference between the interfacial displacement wo (hi) and the average displacement WQ
anc

* is derived in a similar fashion by expressing the shear

stress Gyzo in the following alternate form y - ( h i + 2h 0 y


O"zy0-Tzy02"

2h 0
TzylO

(xzylO - Tzy02)

(2.22)

2h 0 Tzy02

-wo(hi)-

W0

(2.23)

49

2.1.6 Expression for Shear Stress in Terms of Displacements Using the displacement continuity conditions, together with the symmetry relations, equations (2.18 a, b, c) and (2.21) are combined and solved for the interfacial shear stresses Txyio and xXy02 as follows TxylO = Di(u2-uo)+D2(uo-ui) TXy02 = D3(u2-uo)+D4(uo-ui) (2.24a) (2.24b)

In a similar way, expressions for the interfacial shear stresses x zy io and T ZV 02 are given by TzylO - Dl \W2 - W0J+ D2 (wo ~ Wl) Tzy02 = D3 (w2 - Woj+ D4\WQ _ Wlj where Di = a 22 -ai2 ,D2 = _ ana22 _ ai2a2i ana22 ai2a2i D4 = an ana22~ai2a2i 2 h o + h2 3u 0 3u 2 2 h o + hi 3^o
3

(2.25a) (2.25b)

D3 =

-a21 ana22_ai2a2i

an-a22:

2hp 6u 0

> a i 2

> a 2 i -

^1.

2.1.7 Governing Differential Equations In this section, the governing partial differential equations are derived for the layered model. Substituting equations (2.15 a, b, c) into equations (2.12 a, b, c), and

50

differentiating with respect to x and substituting the resultant equations in to equations (2.11 a, b, c), the following expressions are obtained.
9 9

Mi +

M2

- T

dx<
3 2 u(i

dxdz
, ^2wp M3 + M4

xylO

(2.26a)

dx'

dxdz

T X ylO _ Xxy02

(2.26b)

d2u2w - ^2w2A, dx' M5 + dxdi M6

Txy02

(2.26c)

Mi through Mg reflect the material properties and layer thicknesses as follows


_ 2 u 1 + Xi MiXi . M2: Xi 1

Xlhi

7-

( 2 ^ + A.iJhi

\ ,

2H! + A.i_ hi
^0

M3 = X,o2ho 2^2 + ^2 ^2h2

7^-\ , v2^i0 + X,oj2ho 7,2


(2^2 + ^2) h 2

M4:

2^0 + ^ 0 . 2 h 0

M5 =

X M6 = 1 2

2 u 2 + ^2 h 2

In light of the model symmetry and the anticipated loading (Figure 2.1a), the shear stress zxa given by equation (2.9) is negligible; the equation may be manipulated to yield
92wa _ 32ua . 9 2 w a _ 32ua

dxdz

az2

and-

dx'

dxdz

. Substituting the expressions for xXyio

an(

Txyo2 from equations (2.24 a, b) into equations (2.26 a, b, c), yields the following set of governing partial differential equations in terms of the average x-displacement u a each layer (a = 1, 0, 2)
m

51

a2m a2 yM2 Mr ax 2 3z

-bl(u2-uo)+D2(uo-ui)J

(2.27a)

a2uo
3x2

M3-

uo M4 = [(^2-^o)(Di-D3)+(uo-m)(D2-D4)] dz'

(2.27b)

a 2 U2 a2u2 M5M 6 = D3(u2-uo)+D4(uo-ui) ax2 az 2

(2.27c)

The second set of governing partial differential equations in terms of the average zdisplacement is obtained by using equations (2.15 a, b, c) and following the same procedure that lead to equations (2.27 a, b, c), which yields

a wi Mr az2

wi M 2 3x'

= -[DI(W^-WO)+D2(W^-W^)]

(2.28a)

a w o . , a wT-M4 p M3 3x' a^
a W2W

= [(w2~wo)(Di-D3)+(wo-wi)(D2-D4)]

(2-28b)

a W2W

az

ax

D3iw2-wo)+D4lwo_wij

(2.28c)

2.1.8 General Solutions for the Governing Partial Differential Equations The general solution for the two sets of governing differential equations (2.27 a, b, c) and (2.28 a, b, c) is given by u>aaePx w>aaePz (2.29) (2.30)

Where a a are constants (a= 1, 0 and 2), and p is the eigen value. Substituting the general solutions given by equations (2.29) and (2.30) into the governing differential equations 52

(2.27a, b, c) and (2.28a, b, c) gives the eigen values for p, by solving a third order 2 polynomial in p . The roots of the polynomial equation are properties of the model that are solely determined by the material properties and thickness of the layers. The non-trivial solution for the average x-displacement u a (a = 1, 0,2) in each layer is given by Ua = ci a sinh(p 1 x)+c2aSinh(p 2 x) (2.31)

where c i a and C2a are arbitrary constants to be determined from the boundary conditions. In a similar fashion, the non-trivial solution for the average z-displacement w a is given by w a = di a sinh(p 1 z)+d2a s m h (p2 z ) where d i a and d2 a are to be determined from the boundary conditions. The normal strains e x a and va (a = 1, 0, 2) are obtained by differentiating equations (2.31) and (2.32) with respect to x and z, respectively. ( 2 - 32 )

2.1.9 Boundary Conditions For the layered model shown in Figure 2.1b, a uniform tensile stress oo is applied only to the faces perpendicular to the x and z directions. Additionally, the model temperature is changed by T from the ambient temperature. These loading conditions are considered to be the generalized case in the study. The model can be used to analyze other types of adhesive joints by modifying the boundary conditions. For example, a single-lap joint may be simulated by limiting the non-zero surface loading (boundary 53

conditions) to the positive and negative x-directions on materials 1 and 2, respectively as shown in Figure 2.1c. Using normal strains and equations (2.16-2.21), (2.12 a, b, c) and (2.15 a, b, c) and assuming that the average stress <yya is negligible yields
2jXa + X,a Xa

o"xa :

Xa
2

^M'a

^a.

[ciap1cosh(p1x)+c2aP2cosh(P2x)]+ (2.33)
2

^a

2 u a + )ia

[diaPiCOsh(p 1 z)+d 2 aP2 c o s h (P2 z )]-YaT

^a

2 u a + A,a Oza
2

^a 2ji a + /,a_

Xa ^a

[dlaPi c o s h (Pl z )+d2aP2 c o s h (P2 z )] +


(2.34)
2

2^ a + ^a

[ciap1cosh(p1x)+c2aP2cosh(P2x)]-YaT

^a

where the constants ci a , C2a, dla> a n ( i ^2a ( a = 1 0 2) in equations (2.33) and (2.34) are > determined from the following boundary conditions xa x=L x /2=0 where OQ is the applied uniform stress. and
c

za|z=L7/2-CO

(2.35)

2.1.10 Interfacial Shear Stresses After the boundary conditions have been implemented, the average displacements u a and w a given by equations (2.31) and (2.32) are substituted into equations (2.24 a, b) and (2.25a, b) to obtain the following expressions for the shear stresses at the interfaces between the adhesive and the two layered materials

54

TXyio = Di[(ci2sinh(p 1 x)+c22sinh(p 2 x))-(ci 0 sinh(p 1 x)+c2osinh(p 2 x))J (2.36) + D2[(ciosinh(pj x)+ C 2 osinh(p 2 x))- (C11 sinh(pjx)+ C 2isinh(p 2 x))] Txy02 = D3Kci2sinh(p1x)+ C22 sinh(p 2 x))- (ciosinh(pjx)+
C2osinh(p2x))J

(2.37) + D4 [(cio sinh (pj x)+C20 sinh (p2 x)) - (C1! sinh (pj x)+c 2 i sinh (p2 x))] TzylO = Di [(di2 sinh(pj z)+ d22 sinh(p2 z))- (dio sinh(pj z)+ d20 sinh(p2 z))J (2.38) + D2 [(dio sinh(pj z) + d2o sinhfpj z))- (di 1 sinh(pj z)+d 2 i sinh(p2 z))] Tzy02 = D3 l(di2 sinh(pj z) + d22 sinh(p2 z)) - (dio sinh (pjz) + d20 sinh(p2 z (2.39)

+ D4 [(dio sinh (pj ) + d20 sinh(p2 z)) - (di 1 sinh(pj z)+ d2l sinh(p2 z))]
2.2 Numerical Results and Discussions The solution for the interfacial shear stresses xXyio, TXy02, TzylO a n ( i Tzy02> given by equations (2.36-2.39), demonstrate their dependence on the geometry and properties of the adherend (material 1 and material 2), the adhesive (material 0) and on the mechanical and thermal loadings, 00 and T, respectively. In the numerical results, two main issues are investigated; namely, the influence of the adhesive thickness and mechanical properties on the interfacial shear stresses. A boron laminate (material 1) and a carbon phenolic laminate (material 2) are used as adherends. Table 2.1 shows the properties of the adherend materials. For the adhesive, various arbitrary sets of elastic properties (Xo, \IQ) and thickness are used. The Lame' constant XQ; and shear modulus uo, for the bonding material were assigned as percentages 55

of the boron laminate properties X\ and \i\. Four cases are considered for the numerical investigations; namely, 5%, 10%, 20% and 30% of the boron laminate properties. 2.2.1 Finite Element Verification The finite element code ABAQUS is used for the comparison of shear stresses at the adherend adhesive interface. Due to symmetry, one quarter model is analyzed using 3-dimensional continuum 8 node reduced integrated elements (C3D8R) [60]. Isotropic material properties for the adherends and the adhesive are considered. Tie constraints are imposed at the adhesive interface with the adherends. A finer mesh, 32000 solid continuum elements, is used for the complete model to analyze the interfacial shear stresses. Uniform tensile stress of 10 MPa was applied to the faces of the joint along with a uniform temperature field of 10C. Figures 2.3-2.10 show the results from both the theoretical and FEM models for interfacial shear stresses, TXY and Tzy on the lower and upper interfaces between the adhesive and the adherends. A tensile stress of 10 MPa is applied to the external boundaries of the joint as shown in Figure 2.1a. The interfacial shear stresses increase in a nonlinear fashion as the distance is increased from the origin towards the edge of the model. The distributions of the 3D interfacial shear stresses from both the theoretical and FEM have similar trends, but there is a difference in the magnitude of the shear stresses. This difference may be attributed to the assumptions in the model and the limitations in the FEM procedure. The through thickness averaging procedure for the displacement in the continuum model, as well as the linear approximation of the shear stresses are one of

56

the reasons. The change in normal stresses c y is considered to be negligible in the theoretical model, which cannot be simulated in the FEM analysis. Further in the FEM model the stresses at the individual nodes are the average between the adjacent nodes and this does not necessarily produce absolute stress free boundary condition. Where as in the theoretical model stress free boundary conditions are maintained when required for equilibrium conditions. All these along with the mesh size of the FEM model of 8-node brick elements contributes to the difference in the shear stresses [9, 60]. Figure 2.3 and Figure 2.4 show the shear stress xXy distribution on the upper interface from the theoretical and FEM analysis respectively. Similar distribution of shear stresses, maximum at the edge of the joint and zero at the center of the joint is observed in both the cases. Figures 2.5 and 2.6 show the shear stresses x zy , distribution on the upper interface and these are rotated version of the shear stresses x xy , about the y axis. Figure 2.7 and 2.8 show the interfacial shear stresses at the lower interface. The theoretical results closely match the FEM results in both the magnitude and the trend of interfacial shear stresses except near the edges. The theoretical and FEM interfacial shear stresses x zy , at the lower interface is shown in Figure 2.9 and 2.10 respectively. For simplification, all the numerical analyses were carried out assuming the geometric model to be symmetric, that is L x and L z were considered equal.

57

2.2.2 Effect of Elastic Properties of Adhesive Due to model symmetry, the numerical results for the effect of adhesive L -L properties are shown for the one quarter model < x < 0, < z < 0. Figures 2.11 and Figure 2.12 demonstrate the effect of adhesive material properties on the interfacial shear stresses at the upper interface (xZy02, Txy02) a n ^ lower interface (xXyio, tzyloX respectively. The thickness of the adhesive is maintained at 0.09 m and the applied stress oo at the boundaries (x = ,z = ) were maintained at 10 MPa. The properties XQ

and (io of the adhesive material are considered to be 5%, 10%, 20% and 30% of the boron laminate and are referred to as case 1, 2, 3 and 4 respectively. Figure 2.11 shows that the shear stresses on the upper interface increased, as the properties no and 1Q of the adhesive are increased from 5% to 30% of those of material 1. However, the corresponding shear stresses on the lower interface (Figure 2.12) decreased. The reason for reduced shear stresses on the lower interface and increased shear stress on the upper interface is due to the increase in the difference in material properties at the upper interface and reduction at the lower interface as the adhesive properties are increased from 5% to 30% of those for the boron laminate (material 1).

2.2.3 Effect of the Adhesive Thickness Figures 2.13 demonstrate the effect of adhesive thickness on the shear stresses xXyi 0, xZyio at the lower interface. The adhesive thickness is described by using a non-

58

dimensional volume fractions n which is the ratio of the adhesive thickness 2ho to the overall thickness h of the model, as shown by equation (2.40) n=^ h In this section, the elastic properties no and XQ of the adhesive are maintained at 10% of those for the boron laminate (material 1). The applied surface traction at the boundaries is 10 MPa. The volume fraction of the adhesive is varied from 0.03 to 0.1 which corresponds to an adhesive thickness between 0.03m to 0.1m. Numerical results in Figures 2.13 show that the magnitude of the maximum shear stress at (x = z = 2 adhesive is increased. ), is increased on both interfaces as the volume fraction of the (2.40)

2.3 Summary New formulas are derived for the bi-axial interfacial shear stresses that develop in an adhesively bonded joint due to static thermo-mechanical loading. The analysis is carried out along the lines of continuum mixture theories of wave propagation. Numerical results that show the effect of both the elastic properties and the thickness of the adhesive on the interfacial shear stresses are investigated. Numerical results show that both the material properties and the thickness of the adhesive have a pronounced effect on the developed interfacial shear stresses due to the thermo-mechanical loading. For the present model, it has been found that increasing the thickness of the adhesive causes a significant increase of the interfacial shear stresses. The larger difference in the 59

elastic and thermal properties between the adhesive and the layered adherends the higher the corresponding interfacial shear stress is. The proposed model inherently has the capacity for optimizing the selection of the adhesive thickness and material properties that would yield a more reliable bonded joint.

Table 2.1 Material properties of Boron and Carbon Phenolic Laminate

Property

Boron laminate (Material 1)


-3

Carbon Phenolic Laminate (Material 2)

Density p (kg/m ) Shear Modulus, u ( MPa) Lam'e constant, X (MPa) y (MPa C)

2370 95100 80600 6.48

1420 6620 11400 9.48

60

Material 2 (Adherend) Material 0 (Adhesive)

^Material 1 (Adherend)

(a) Figure 2.1 Geometric model (a) Complete model of the adhesive bonded joint

61

h2 h+h 2 Material-2

Materta!-0 (Adhesive) Material-1

L x /2

(b) Figure 2.1 (b) One quarter model of adhesive bonded joint

62

Material 2 (Adherend) Material 0 (Adhesive)

Material 1 (Adherend)

Figure 2.1 (c) Model representing an adhesive bonded single lap joint.

63

h+h 0

Figure 2.2 Shear stress distribution

64

Distribution of Shear Stresses (XY) along the Upper interface (Theoretical)

/ J
2 J

,..j

CO

CL
CO CO

2
CO

CO

co a) sz

1.6 -0.05 (Z-axis) Width of the Joint (m) (X-axis) Length of the Joint (m)

Figure 2.3 Theoretical shear stress, (xxy), at the upper interface

65

Distribution of Shea r Stresses (XY) along the Uppeir interface (FEM)


""IN

1 "v

I
l

HH^T 0 ' 3 JHHHo.2 0 1

(0 Q.

HH '
1 Hf^*"" ' . 8
o e l
(Z-axis) Width of the Joint (m) -0.6^

(0

s>

M.1
\ [-0.2 ^^f-0.3
T-

CO to <D CO

V
i

-1.9 -- o

_ - - io

ci

(X-axis) Length of the Joint (m)

Figure 2.4 FEM shear stress, (xXy), at the upper interface

66

Distribution of Shear Stresses (ZY) along the Upper Interface (Theoretical)

0.30
CO CL

!/)

. fi
CO i_ CO (1)

.c CO

(X-axis) Length of the Joint (m)

(Ml

(Z-axis) Width of the Joint (m)

Figure 2.5 Theoretical shear stress, (xZy), at the upper interface

67

Distribution of Shear Stresses (ZY) along the Upper Interface (FEM)

(X-axis) LengtH of the Joint (m)

<^e>

(Z-axis) Width of the Joint (m)

Figure 2.6 FEM shear stress, (izy), at the upper interface

68

Distribution of Shear Stresses (XY) along the Lower Interface (Theoretical) 2 1.5 1 0.5 0 co -0.5
CD w
i_

CD EL

X -

to
CO

* z

C D C O

C O

-1 -1.5 -2

(Z-axis) width of the joint (m)

(X-axis) length of the joint(m)

Figure 2.7 Theoretical shear stress, (xXy), at the lower interface.

69

Distribution of Shear Stresses (XY) along the Lower Interface (FEM)

2.5-r ~^^ririDffiflM9&
1.510.5 0

2 mmmm.
'z

CD Q.

CO CO

8>
CD CD

-0.5

r^^^^H j '"""HI
-S|

.c CO

-1-1 "-1
15

-1.85 -1.35 -0.85 -0.35 / 0.15 / 0.65 /

-zlkl

/
1.15 / m
CO

0.9
-0.55 (Z-axis) Width of the Joint (m)

(X-axis) Length of the Joint (m)

Figure 2.8 FEM shear stress, (Txy), at the lower interface.

70

Distribution of Shear Stresses (ZY) along the Lower Interface (Theoretical)

* z <

(X-axis) length of joint(m)

(Z-axis) width of joint (m)

Figure 2.9 Theoretical shear stress, (xzy), at the lower interface.

71

Distribution of Shear Stresses (ZY) along the Lower Interface (FEM)


, _ . . - " " " i " "* T \

\ >2.5 "2 -ro -1.5 9:

^ !

iX
^ | \

>H^
^

.. ^ ^

-1
0.5 0 -0.5
.1
1

1
$ 2 > +-<
CD

:~"~~r " J ^ ^ ^ ^

^r^"^%
(X-axis) Length of the Joint (m) 0% 1\
2

-1.5

-?

^ f ^ "25 ^ &
T

"

rt

w d ^ 9 d>
'

(Z-axis) Width of the Joint (m)

Figure 2.10 FEM shear stress, (xzy), at the lower interface.

72

Effect Of Adhesive Properties on the Shear Stress at the Upper Interface -3.50 -3.00 -2.50 -2.00 -1.50 -1.00 -0.50 0.00 2000 *case 1 a=10 Mpa AT=10"C

1750

1500

1250

1000

750

500

250

Distance from the center (mm)

2.11 Effect of adhesive properties on the shear stress at the upper interface.

73

Effect of Adhesive Properties on the Shear Stress at the Lower Interface 2 1.8 1.6 1.4
7

a=10 Mpa AT=10"C

1.2 1 0.8 0.6 0.4 0.2

2
Jr

< *

m case 1 --* case 2 case 3 # case 4 '"

2/
z

Si^
1500 1000 Distance from the center (mm) 500

2000

ure 2.12 Effect of adhesive properties on the shear stress at the lower interface.

74

Effect of Adhesi\e Volume Fractions "n" on the Shear Stress at the Lower Interface Volume fraction " n " ^ d h e s i v e Total Thickness

1500

1000 Distance from the center (mm)

Figure 2.13 Effect of adhesive thickness on the shear stresses at the lower interface

75

CHAPTER THREE EFFECT OF WASHERS AND BOLT TENSION ON THE BEHAVIOR OF THICK COMPOSITE JOINTS

In this chapter, experimental characterization of thick composite bolted joints is performed to study the effect of washer size and bolt preload on bearing properties. S2glass fabric-epoxy composite coupons [0/90; +45/-45 @ 10 sets] of 12.5 mm thickness were tested under double shear tensile loading. Two different washer sizes and thicknesses were used in this investigation. A force washer is used to monitor the clamp load variation during the test. It has been found that the initial bolt tension (preload) and washer size have a significant effect on bearing stiffness and bearing strength of thick composite joints. For a low bolt preload, test data shows a significant clamp load increase with the joint displacement. However, the percentage increase in clamp load is reduced as the preload is increased to 50kN. The outward buckling and delamination of the laminate in the composite coupons were found to be the main cause for clamp load increase.

3.1 Experimental Setup and Procedure Extensive experiments were conducted to investigate the effect of bolt preload, washer area and washer thickness on the bearing strength, bearing stiffness and ultimate strength of glass-mat epoxy laminated composite bolted joints. ASTM standard D 5961/D 5961-05 [33] was followed to study the bearing response of single bolted doubleshear tensile loaded joints. 76

3.1.1 Material Multi-directional 20-ply fiber-reinforced polymeric matrix laminated composite coupons were supplied by Sherwood Advanced Composite Technologies. The composite coupons of SO 15 toughened epoxy resin/S-2 fiberglass-mat with [0/90; +45/-45 @ 10 sets] orientation was manufactured by Vacuum Assisted Resin Transfer injection Molding (VARTM) process. The nominal thickness of tested coupons is around 12.5mm; they were designed for bearing failure. Figure 3.1 shows the geometry of the test coupon used in the study. The test coupon was designed for bearing failure. The width to diameter (W/D) ratio of 4 and edge distance to diameter ratio (E/D) of 3.5 was maintained through out the experiments.

3.1.2 Test Fixture and Instrumentation Figure 3.2 shows the double shear test fixture used in this study. The fixture is designed to mount the extensometer for monitoring the coupon displacement relative to the fixture. An MTS hydraulic testing machine is used to apply the tensile loading to the specimen at a rate of 5mm/min. A force washer and a data acquisition system are used to monitor the bolt preload, joint clamp load and as well as the joint displacement during the tests. Figure 3.3 shows the experimental set used in the study. Five levels of bolt preload, two different washer sizes and two washer thicknesses are used in this study. Bolt preload levels are 0, 25, 50, 75 and 100% of the proof strength of 1/2"-20 SAE Grade 5 fasteners. Large washers (USS) with an effective area of 796 mm2 and small washers (SAE) with an effective area of 430 mm2 are used. Single and double washers are used to simulate two different washer thicknesses. Fasteners

77

were tightened to the desired preload level as indicated by the force washer. Table 3.1 presents the actual bolt load and corresponding clamping pressures for different washer configurations. Load control method was used to tighten the fasteners in-order to produce the required reliable initial bolt load. Torque control method for tightening the fasteners does not necessarily produce the required bolt load [61, 62]. T = KdF is the basic equation used to calculate the initial bolt load [61]. Where K is the nut-factor, T is the applied torque, d is the bolt diameter and F is the initial bolt-load. Here, K the nut-factor is generally selected from published tables [62] for various combinations of materials, surface finish, plating, coatings and lubricants. However, the literature [63, 64] has showed that this is highly unreliable and the nut-factor depends on thread friction and washer under head frictional coefficients. Equation 1 [61] gives the relation for the nutfactor, considering various frictional coefficients. T = F + J - L 7r + M,nrn
271 COS p

(3.1)

where T = torque applied to the fasteners (lb-in, N-mm), F= bolt preload (lb, N), P=thread pitch (in, mm), Ut= thread friction coefficient, rt= effective contact radius of the threads (in, mm), p= half-angle of the threads, |i = washer under head frictional coefficient, rn = effective radius of contact between the nut and washer or joint surface (in, mm). In the present study, instead of selecting the nut-factor from published source, a load control method of bolt tightening is followed, in which a force washer is used to monitor the initial bolt load in real time. This procedure ensured that the required (0%,
78

25%, 50%, 75% and 100% of bolt proof load) initial bolt load was achieved in all tested joint configurations.

3.1.2.1 Bearing Properties The effect of various joint parameters, discussed in the previous section, on the bearing stiffness, the bearing strength and the ultimate joint strength and strain is investigated in the study. Figure 3.4 shows the typical bearing stress distribution in a bolt loaded joint [65]. Radial bearing stress p due to a bolt in a hole is generally considered to be distributed around the loaded half of the pin-hole circumference [65] as follows, p = p m cose [-^<0<^] v 2 2) (3.2)

where p m is the maximum radial stress on the bearing region due to bolt load L. The bolt load L can be expressed as L= J pcosOh^dO
-71/2
2

(3.3)

where d is the bolt-hole diameter and h is the laminate thickness. Substituting equation (3.2) into equation (3.3) gives a relation for the bearing stress L=^ p
2

J C O s 2 0de
-nil

(3.4)

L =|pmhd

(3.5)

In this study, the bearing stress a is considered as the average stress acting uniformly over the projected cross-sectional area of the hole given by [33] as follows 79

c =ihd

(3.6)

The corresponding average bearing strain 8 for each displacement 8 is e = , where 8 is d the extension of the composite coupon at the bolt hole region measured using an extensometer. The maximum load prior to failure was used for the ultimate bearing stress. The slope of the initial linear bearing stress- strain curve is used to determine the bearing joint stiffness (Figures 3.5 and 3.6). An effective origin is defined at the intersection of the bearing stress-strain line with the strain axis. The stiffness line is then translated from the effective origin by the offset 2% strain to obtain the 2% offset bearing strength (Figure 3.6). 3.2 Results and Discussion Figure 3.5 and Figure 3.6 show the bearing stress-strain plots for the zero bolt preload and for bolt preload of 50% of its proof load, respectively. The bearing stressstrain data for the zero bolt preload joint shows different behavior when compared to the joints with non-zero bolt preload. The initial portion of the bearing stress - strain curve in Figure 3.6 (strain correction zone) shows a nonlinear response due to combination of joint straightening, overcoming of joint friction, and joint slippage [33]. This initial nonlinear response is observed for the non-zero bolt preloads. Beyond this initial response, the bearing stress-strain curves show a linear increase in stress until the bearing failure in initiated. The slope of the initial bearing stress- strain curve is used to determine the bearing stiffness of the joint (Figures 3.5 and 3.6). A slight stress plateau is observed
80

at point A indicating initial bearing failure for joint with zero bolt preload (Figure 3.5). Beyond this point the joint continues to carry the load until the bearing stress drops significantly, indicating the final rupture at the bolt hole boundary in the composite coupon. A similar stress plateau has not been observed for a joint with a preloaded bolt (Figure 3.6); the change from linear to non-linear behavior was gradual. For this reason, 2% offset bearing strength is considered in this study for comparison purpose. For both preload levels, the joint continues to carry the load after the initial bearing failure until the ultimate bearing strength has been reached. The strain measured at this maximum bearing stress from the effective origin is called as the ultimate joint strain. This joint strain or the joint elongation is due to the combination of increase in the bolt-hole diameter and joint material elongation. In the present study the joint material elongation was negligible.

3.2.1 Effect of Bolt Preload As described earlier, five different bolt preload levels ranging from 0% to 100% of the bolt proof load is considered in the study. Figure 3.7 - Figure 3.10 show the effect of initial bolt load on the bearing joint stiffness, offset bearing strength, ultimate joint strength and joint strain for various joint configurations, respectively. Table 3.2 - Table 3.5 show the average values of the bearing properties for single large washer, single small washer, double large washer and double small washer composite joints, respectively. 3.2.1.1 Effect of Bolt Preload on Joint Bearing Stiffness Figure 3.7 shows the bearing stiffness at various levels of bolt preload for various washer configurations. The bearing stiffness is largest for the untightened joint (zero bolt 81

preload). The stiffness reduces by 15-20% for the joints in which the bolt preload is 25% of the proof load. The axial load transfer in the tightened joint coupons is by the combination of bearing contact on the cylindrical surface of the hole and the frictional contact between the flat coupon surfaces [62]; whereas, in a zero preloaded bolted joints the load transfer is only due the bearing contact. Because the composite coupon used in this study is thick, relatively stronger through the thickness for the bearing load transfer, the joint behavior was stiffer for the untightened joints where the frictional force is negligible. The joints with 25% bolt preload displayed a lower stiffness as their initial behavior was dominated by the frictional force between the washers and the surface of the composite coupon. The frictional effect was reduced as the bolt preload increased to 100% of proof load.

3.2.1.2 Effect of Bolt Preload on Offset Bearing Strength The offset bearing strength increases progressively with increasing the bolt preload. As the bolt preload is increased from 0% to 25%, 50%, 75% and 100% of its proof load, the bearing strength of the clamped composite coupon increased by about 28%, 37%, 37% and 40%, respectively (Figure 3.8). Higher initial bolt load created enough lateral constraint on the composite coupons to delay the initiation of bearing failure on the contact surface between the bolt shank and the hole surface [29, 30, and 32]. 3.2.1.3 Effect of Bolt Preload on Ultimate Joint Strength and Strain Figure 3.9 shows that for the small washer configuration the ultimate joint strength was unaffected by the increase in bolt preload, which is consistent with [31] and 82

[66]. The ultimate joint strain (elongation) was about 15% more for the untightened joint as compared to other bolt preloads (Figure 3.10). The clamping pressure created by the initial bolt load reduced the bolt-hole elongation resulting in lower joint strain. For large washer composite joints, the ultimate failure load exceeded the load cell capacity (MTS 100 kN).

3.2.2 Effect of Washer Size and Thickness on Bearing Behavior Figure 3.11 shows that joints with smaller diameter washers had higher bearing stiffness than those with larger washers for bolt preloads up to 50% of the proof load. The bearing stiffness was unaffected by the washer size for higher bolt preloads. This increase in bearing stiffness for small washer joints was mainly due to the high clamping pressure. For the same bolt preload, smaller washers exerted higher clamping pressure on the composite coupons than the large washers. Figure 3.11 and Figure 3.12 show the effect of washer size on the bearing stiffness for single washer and double washer joints, respectively, with various levels of bolt preload. Table 3.1 presents the lateral clamping pressure exerted by the small and large washers for different initial bolt loads. Figure 3.13 and Figure 3.14 show the results for the effect of washer area on the offset bearing strength. It is observed that joints with smaller washers have a slight higher bearing strength (5%) than those joints with large washers. This suggests that the higher lateral clamping pressure exerted by the smaller washers delays the initiation of bearing failure which translates to an increase in the offset bearing strength. After the initiation of bearing failure, both small washer joints and large washer joints, continue to carry the load as the bearing damage progresses to ultimate failure. 83

The joints with smaller washers fail at about 550 MPa; whereas, the joints with larger washers had its ultimate strength beyond 585 MPa (load cell limit). The larger clamping area of the large washers suppressed the delamination in the bearing zone, resulting in increased ultimate joint strength and strain. The higher lateral clamping pressure by small washers induced additional surface damage, resulting in lower ultimate joint strength Figure 3.16 and Figure 3.17 show the effect of washer thickness on the bearing stiffness for larger and smaller size washers at various preloads levels. It is observed that using single washer causes a slight increase in bearing stiffness (5%), as compared to joints with double washers (thick washer). Figure 3.18 and Figure 3.19 show the corresponding effect on the offset bearing strength. The use of single washer produced a slight increase (5%) in bearing strength. By using a two washer stack in the joint, one introduced an additional frictional surface which induces relative instability in the joint behavior. This unstable joint behavior is likely the reason for the slight decrease in the corresponding bearing stiffness and bearing strength. The washer thickness had no significant effect on the ultimate joint strength and strain. This behavior was expected as friction has a minimal influence on the joint behavior after the bearing failure has been initiated.

3.2.4 Clamp-Load Variation Figure 3.20 and Figure 13.21 show the measured joint clamp-load verses joint displacement/applied load for zero and 50% bolt preload (of proof load) joints, respectively. For the zero bolt preload joints, the clamp load increased linearly with joint

84

displacement. This linear increase continued until the initial bearing failure, beyond which clamp load increased significantly in a nonlinear manner. At the point of ultimate failure the clamp load further increased with a larger slope. For the non-zero bolt preload joints, the clamp load remained almost constant until the joint overcame the friction between the parts (points F on Figure 3.21). At this point the clamp load slightly dropped, and then increased linearly with lower slope until the initial bearing failure initiated (point A). Beyond the initial bearing failure the clamp load variation was similar to that of the zero bolt preload. The clamp load variation for joints with 25, 75 and 100% bolt preload followed the same trend as that of the 50% bolt preload joints (Figure 3.21). This increase in clamp load after the initial bearing failure is mainly due the progressive increase in through thickness of the composite coupon due to the delamination and fiber-matrix outward buckling in composite coupons [66]. Figure 22 shows the variation of joint clamp load with applied load for joints with small washers. The increase in joint clamp load for zero bolt preload, was significantly higher than that for the tightened joint. The lateral clamping pressure induced by the washers increased with the bolt preload, and reduced the delamination failure in the bearing zone. This explains the reduction in clamp load change with increasing the bolt preload. Test data shows the sudden increase in clamp load at ultimate joint failure. This failure load remained almost the same regardless of the level of bolt preload. Figure 23 shows the joint clamp-load variation with the applied axial load when large washers are used. The clamp load variation for large washer joints was similar to the small washer joints, except in the ultimate failure region. The sudden increase in clamp load at the end

85

of each curve was absent, showing that the large washer joints carried load beyond 95kN (load cell capacity).

3.2.5 Failure Analysis Bearing failure in composite coupon was the prominent failure mode observed in all tested joint configurations. The area and the extent of bearing damage varied with washer size and initial bolt load. Figures 3.24 a, b show the bearing damage in small washer and large washer joint coupons with zero bolt preload, respectively. The extent of bearing damage and the bolt-hole elongation in the small diameter washer joints were more sever compared to the large diameter washer joints. For the coupons with large washers the bearing damage was more uniform and was only seen under the washer contact surface. However, the bearing damage for coupons with smaller washers spread over a larger area beyond the washer contact. The delamination was more severe just outside the washer contact region; this was responsible for the significant load drop at the ultimate failure region. The increase in coupon thickness on the loaded side of the bolthole was observed for joints with smaller and larger washers. This was mainly due to the delamination and outward bucking of laminate in the localized bearing regions. As the outward buckling and delamination increased, the laminate were pressed against the washers creating the increase in clamp load observed in Figures 3.22 and 3.23 for untightened joints [32]. Figure 3.25 shows the extent of bearing damage for coupons with large washers. It can be observed that the bearing damage was more significant for joints with zero bolt preload; the damage extent reduced with the increase in bolt preload. The bearing 86

damage was primarily under the washer surface; washer imprint was also observed for all joint coupons. For the joints with higher initial bolt load, the lateral restraint of the washers reduced the delamination and outward bucking of the laminates. This behavior reduced the increase in clamp load as observed in Figure 3.22 and 3.23 for joints with higher bolt preload.

3.3 Summary Experimental data is presented on the affect of initial bolt preload, as well as the washer size and thickness on the behavior of heavily loaded thick composite joints. Friction between the joint parts played a significant role in defining joint stiffness. The joint bearing stiffness was higher for the untightened bolted joint than that with much higher bolt preload (100% of proof load). The bearing stiffness was smallest for the joint with a preload equal to 25% of bolt proof load, and it increased with bolt preload. The offset bearing strength increased progressively with bolt preload. The ultimate joint strength was unaffected by increasing the bolt preload. Joint with small washers had higher bearing stiffness than those with large washers for initial bolt preload of 0%, 25% and 50%. Joints with small washers had higher offset bearing strength than the joints with large washers. The washer thickness had an insignificant effect on the ultimate joint strength and strain. As the axial test load is increased, an untightened bolted joint showed a significant increase in the clamp load (from its zero initial value). This increase in the clamp load was progressively reduced by the increase in initial bolt-load. Bearing damage was the prominent failure mode in all tested joint configurations. The extent of
87

bearing damage was more severe in smaller washer joints, where the damage extended beyond the washer contact. Out-of-plane buckling of laminates exerted lateral pressure on the washers resulting in an increase in clamp load during the experiments. These experimental results help in the selection of an optimum initial bolt preload and washer size and thickness in order to enhance composite joint performance and reliability.

Table 3.1 Initial bolt-load and corresponding clamping pressure for small and large washer joints
Bolt Clamp-Load Clamping Pressure

% of Proof Load

Actual Clamp Load (kN)

Large washer (USS)

Small washer (SAE)

0% 25% 50% 75% 100%

0.2 13.34 26.68 40.03 53.37

0.28 16.74 33.48 50.22 66.97

0.52 31.05 62.55 93.16 124.22

88

Table 3.2 Bearing properties of single large washer composite joints

Single Large (USS) washer composite joint

Clamp Load (% of bolt proof strength)

Bearing Stiffness (MPa)

2% bearing strength (MPa)

Ultimate strength (MPa)

Joint strain (%)

0% 25% 50% 75% 100 %

50.54 41.37 41.50 44.50 49.83

364.00 455.38 495.24 496.45 497.08

>585 >585 >585 >585 >585

>30 >27 >23 >23 >17

89

Table 3.3 Bearing properties of single small washer composite joints

Single Small (SAE) washer composite joint

Clamp Load (% of bolt proof strength)

Bearing Stiffness (MPa)

2% bearing strength (MPa)

Ultimate strength (MPa)

Joint strain (%)

0% 25% 50% 75% 100%

56.46 47.09 46.00 45.40 50.94

372.44 489.64 518.90 512.22 535.46

535.00 551.36 569.63 570.68 572.74

22.10 18.65 18.00 18.50 19.30

90

Table 3.4 Bearing properties of double large washer composite joints

Double Large (USS) washer composite joint

Clamp Load (% of bolt proof strength)

Bearing Stiffness (MPa)

2% bearing strength (MPa)

Ultimate strength (MPa)

Joint strain (%)

0% 25% 50% 75% 100%

48.37 37.89 39.70 41.90 46.90

361.07 424.88 462.01 485.70 491.20

>585 >585 >585 >585 >585

>27 >25 >26 >22 >18

91

Table 3.5 Bearing properties of double small washer composite joints

Double Small (SAE) washer composite joint

Clamp Load (% of bolt proof strength)

Bearing Stiffness (MPa)

2% bearing strength (MPa)

Ultimate strength (MPa)

Joint strain (%)

0% 25% 50% 75% 100%

54.43 46.92 45.39 44.67 49.90

368.00 440.49 470.10 494.10 523.53

524.34 540.56 545.14 548.86 573.22

24.47 19.70 20.33 19.44 20.94

92

25.4 mm
X

*t

D=Bolt Hole diameter 12.5mm

W=50.8mm

Figure 3.1 Geometry of the test coupon

93

I vtensometer mounting plate Force washer Composite coupon

Figure 3.2 Experimental double lap-shear test fixture

94

xtensometer

Figure 3.3 Bearing test experimental set-up.

95

Bolt hole Cross section of Bolt

Figure 3.4 Schematic representation of bearing stress distribution in a pin loaded joint (modified from [65])

96

aing Stress (MI

600 550 500 450 400 350 300 250 200 150 100 50 01 0

~T^
Initial Bearing Failure Bearing Stiffness 5 10 15 20

Ultimate Strength

25

Beaing Strain (%)

Figure 3.5 Bearing stress Vs bearing strain curve for a small washer finger tightened bolted joint

1/3 00 C

a u ffl

10

15

20

25

30

35

Beaing Strain (%)

Figure 3.6 Bearing Stress Vs. strain curve for joints with 50% bolt preload

97

50
OH

40 30
VI 60

-is single large washer - double large washer - single small washer - double small washer

20
ffl
<a

10

25

50

75

100

125

Bolt preload (% of Proof Load)

Figure 3.7 Effect of bolt preload on joint bearing stiffness

600 500
00

400 300 200 100 -A single large washer - double large washer -single small washer - double small washer

s
h
s

'S
o

25

50

75

100

125

Bolt preload (% of Proof Load)

Figure 3.8 Effect of bolt preload on offset bearing strength

98

g a
S/3 U

400 300 Hi Single small washer 200 100 - double small washer

25

50

75

100

125

Bolt Preload (% of Proof Load)

Figure 3.9 Effect of bolt preload on ultimate joint strength

30 25 20
c

'i
1/3

15 -Small single washer - Small double washer

J 10

25

50

75

100

125

Bolt Preload (% of Proof Load)

Figure 3.10 Effect of bolt preload on j oint strain

99

-jk Small single washer - Large single washer 60 1? 50


PH

40
u

I 30.B 20 10 -1
I

0 -

25%

50%

75%

100%

Bolt-preload (% of Proof)

Figure 3.11 Effect of washer size on bearing stiffness of joints: single washer

-ASmall double washers 60


1? 50
pu

- Large double washers

r 40
|
oo
c

30

c 20

o 10

0%

25%

50%

75%

100%

Bolt-preload (% of Proof)

Figure 3.12 Effect of washer size on bearing stiffness: double washers.

100

TA Small single washers 600


500 400
00

- Large single washers

300
00

.S 200 S
m

loo

0%

25%

50%

75%

100%

Bolt-preload (% of Proof)

Figure 3.13 Effect of washer size on offset bearing strength: single washer

-AT Small double washers - Large double washers 600 ^


a

500
400 300

& |

-4- (/}

60

g 200 a pa 100

0%

25%

50%

75%

100%

Bolt-preload (% of Proof)

Figure 3.14 Effect of washer size on offset bearing strength: double washers

101

700

600
TO

single small washer joints with.100%

Ultimate Strength

a
+* OQ 60

PL,

500 400 300 single large washer joints with 100% bok200 100 0 10 Bearing Strain (%) 15
20

e
e

Joint Continue^ to Carry the Lo^id

C O

Figure 3.15 Effect of washer area on bearing stress-strain behavior

B Single large washers H Double large washers 60


50 40 30 a c
PQ
60
03 U

20

10

0%

25%

50% Bolt clamp load

75%

100%

Figure 3.16 Effect of washer thickness on joint bearing stiffness: large washers

102

B Single small washers , B Double small washers 60 50

^-v
OH

e3

s 40
00 Cfi <U

iffii
*-

30 20
10 -t

tzi 00

ffl

0 50% Bolt clamp load 100%

Figure 3.17 Effect of small washer thickness on joint bearing stiffness.

H Single large washers H Double large washers 600


C3
OH

500 400

SO

a
so a

300 200 100 0 0% 25% 50% Bolt clamp load 75% 100%

Figure 3.18 Effect of washer thickness on bearing strength: large washers.

103

M Single small washers II Double small washers

600
1?
Bearing Strength (M

500
400 300 200 100

0
0% 25% 50% Bolt clamp load 75% 100%

Figure 3.19 Effect of washer thickness on bearing strength: small washers.

100 80 Ultimate Strength 60 Initial bearing \ Joint clamp load curve 0 1 2 3 Joint Displacement (mm) 40 20

.2

Figure 3.20 Joint clamp-load Vs. with joint displacement: zero bolt preload

104

Applied Load curve A.

1
'S

1 2 3 Joint Displacement (mm)

Figure 3.21 Joint clamp load Vs. displacement: 50% bolt preload.

20

40

60 Applied Load (kN)

80

100

120

Figure 3.22 Joint clamp load Vs. applied axial load: small washers.

105

60 50 40
O hJ to U 0

Bolt preload =100% of Proof

30 20 10 0 20 40 60 Applied Load (kN) 80 100 120

'S

Figure 3.23 Joint clamp load Vs. applied axial load: large washers.

Single Small washer with 0% Clamp Load

Single Large washer with 0% Clamp Load

(a)

(b)

Figure 3.24 Bearing damage in finger tightened joint coupons: (a) Coupons with small washers; (b) Coupons with large washers.

106

Large washer with 0% boltload

Large washer with 25% boltload

(a)

(b)

P K

^arge washer with 30% boltload Large washer witn / : "< ooit>/ > load
i.L net\/
u_ix

(c)

(d)

Figure 3.25 Bearing damage in various joint coupons with large washers: (a) Coupons with 0% bolt-load; (b) Coupons with 25% bolt-load; (c) Coupons with 50% bolt-load; (d) Coupons with 75% bolt-load.

107

Large washer with 100% bolt-load

(e) Figure 3.25 Bearing damage in various joint coupons with large washers: (e) Coupons with 100% bolt-load.

108

CHAPTER FOUR EFFECT OF BOLT TIGHTNESS ON THE BEHAVIOR OF COMPOSITE JOINTS

In this chapter, experimental and numerical investigations have been carried out to study the affect of bolt tightness and joint material on behavior of double bolted single lap shear composite joints. Various scenarios of bolt tightness were considered for composite-to-composite and composite-to-aluminum bolted joints. Progressive damage analysis of glass mat-epoxy composite coupons was carried out to understand the bearing failure mechanism. Optical microscope was used to study the damage under the bolt head region, and as well as at the region of contact of bolt shank with the hole boundary. Four tightening configurations were used in testing of each double bolted joint. These configurations permit each of the two bolts to be either tight or loose. The numerical part of the study utilizes a 3-D finite element model that simulates the bolt tightness and the multilayered composite coupons. The experimental and finite element results are correlated.

4.1 Experimental Set-up and Procedure Figure 4.1 shows the joint geometry considered in the experimental study. The joint geometry was based on ASTM D5961/D5961M-96 [67] standard. The glass-epoxy woven composite coupons were cut from 6mm thick plaques. The coupons were machined at Oakland University machine shop. The bolt holes were carefully drilled with sacrificial plates on either side of the test coupons to avoid edge delamination at bolt-hole

109

boundary. The aluminum coupons were machined from sheets supplied by McMasterCarr . The non-dimensional geometric variables w/d (joint width to hole diameter ratio), e/d (edge distance to hole diameter ratio), p/d (bolt pitch to hole diameter ratio), and d/t (hole diameter to coupon thickness ratio) were kept constant through out the experiments. The values used for the non-dimensional variables were w/d=6, e/d=3, p/d=4.5, and d/t=2.5. Table 4.1 lists the material properties for the various components of tested joints. Metric M8 x 1.25 Class 8.8 fasteners were used for the experiments.

4.1.1 Experiments A screw driven 50kN capacity MTS tensile machine was used for joint testing. Test Works4 material test software from MTS was used to record the load and displacement data. Four tightening combinations were used for the tightness of the top and bottom bolts. The combinations included Loose Top -Loose Bottom (LT-LB), where both the bolts were loose, Tight Top-Loose Bottom (TT-LB), where the top bolt was tight and the bottom bolt was loose, Loose Top-Tight Bottom (LT-TB), where the top bolt was loose and the bottom bolt was tight, Tight Top-Tight Bottom (TT-TB) bolts, where both the bolts were tight. The tight bolt condition is achieved by using 16 Nm torque to tighten the bolt using a digital torque wrench. The loose bolt condition corresponds to finger tightening that produce negligible bolt load in the joint. Joint material combinations included either composite-to-composite or composite-to-aluminum coupons. 10 Nm torque produced a clamp load of approximately 9000 N, which is about 40% of the proof load of the M8 x 1.25 Class 8.8 fasteners. This level of clamp load was carefully selected so that the composite coupons would not be damaged during bolt 110

tightening. The bearing friction variation was minimized by using a new, cleaned high quality washer under the turning nut in each test [62]. Preliminary testing showed that a 16 Nm torque produced a clamp load that ranged from 8950 N to 9046 N in the composite-to-aluminum joints and from 8850 N to 8900 N in composite-to-composite joints. The load-displacement data was used to evaluate the effect of different bolt tightness condition and joint materials on the strength and stiffness of the double bolted composite joint. The composite joint shown in Figure 4.1 was clamped between the upper and lower grips of an MTS tensile machine. The movement of the upper crosshead applied the tensile load to the specimen, while the lower crosshead remained stationary. The crosshead speed was maintained at 1.5 mm/min throughout the experiments, and the test was continued until the final rupture of the composite coupons. The joint stiffness was determined by the slope of the load-displacement curve.

4.1.2 Progressive Damage Analysis The damage assessment of composite coupons at different intermediate loads before the final rupture was carried out using an optical microscope. The damage inspection at two different regions; namely, the surface under bolt heads and at the contact of the bolt shank with the composite hole boundary was carried out to understand the bearing failure mechanism. The examined damage regions are schematically represented in Figure 4.2.

Ill

4.2 Experimental Results and Discussion In this section, test data is presented and analyzed. The affect of various bolt tightness conditions on the stiffness, strength, and damage is discussed for the compositeto-composite and composite-to-aluminum double bolted joints.

4.2.1 Effect of Fastener Tightness Condition For the double bolted joint shown in Figure 4.1, four scenarios of bolt tightness were investigated; namely, Loose Top-Loose Bottom (LT-LB), Tight Top-Loose Bottom (TT-LB), Loose Top-Tight Bottom (LT-TB), and Tight Top-Tight Bottom (TT-TB) bolts. Figures 4.3a - Figure 4.3d show the load-displacement curves for the composite-toaluminum joints, which correspond to the four scenarios of bolt tightness. Point D represents the ultimate failure load of the composite joint. Figures 4.4a - Figure 4.4d represent an enlarged portion of the load-displacement curves for each of the four tightening scenarios. At point B, the shank of the bolt comes in contact with the curved surface of the bolt hole. Point C represents the beginning of the bearing failure at the contact surface. The slope of the line segment BC represents the joint stiffness. The static friction between the components of the joint dominates the behavior of the joint below point B; which is consistent with [48]. The affect of the four bolt tightness scenarios on the joint stiffness is shown in Figures 4.4. The three scenarios of composite-to-aluminum joints with at least one loose bolt had the same stiffness of 4 kN/mm, as determined from the slope of the loaddisplacement curve BC in Figures 4.4 a, b, and c, for LT-LB, LT-TB and TT-LB, respectively. For the fourth configuration, where both the bolts were significantly tight 112

(i.e. TT-TB), the joint stiffness was increased by 31% to 5.25 kN/mm as determined by the slope of line segment BC in Figure 4.4d. Bearing failure was initiated by fiber - matrix cracking and interlaminar shear delamination of the laminate. This corresponds to point C in Figures 4.4 a, b, c, and d. The failure load at this point was about 3.5 kN for joints that had at least one tightened bolt, This includes the three tightness scenarios of LT-TB, TT-LB, and TT-TB that are shown by Figures 4.4 b, c, and d, respectively. One may observe that an additional line segment AB appeared in the load-displacement curve of the three tightening scenarios, with at least one tight bolt; this includes TT-TB, TT-LB, and LT-TB bolts. For the remaining scenario (LT-LB), where both bolts were loose, the bearing failure load dropped by 29% to 2.5 kN as shown in Figure 4.4a. The ultimate failure in composite coupons for all test configurations was at the net cross section. Figure 4.5 a - Figure 4.5 b show that the ultimate strength of the composite joint, which ranges from 15kN to 16kN. This value remained almost constant for all the tightening configurations, showing no obvious effect on the ultimate strength of both composite-to-aluminum and composite-to-composite joints. Post failure damage inspection was carried out to analyze the micro-failure behavior. Figures 4.6 to Figure 4.9 show the surface damage at the bolt holes for various tightening configurations of the top and bottom bolts in, composite-to-aluminum joints. Figure 4.6a, Figure 4.7b, Figure 4.9a and Figure 4.9b, show that there is no significant delamination near the bolt-hole boundary. This was mainly due to the compressive stress created by the bolt load in the vicinity of the tightened bolt. By contrast, Figure 4.6b, Figure4.7a, and 4.8a and Figure

113

4.8b show significant delamination near the holes where the corresponding bolt was loose. 4.2.2 Effect of Joint Materials Figure 4.10 and Figure 4.11 show the load-displacement data for composite-toaluminum and composite-to-composite bolted joints in which the two bolts are either loose or both tight (i.e. LT-LB or TT-TB), respectively. Test data show that the ultimate strength was almost same for both composite-to-aluminum and composite-to-composite joints. However, the joint material had a significant effect on the joint displacement characteristics. Figure 4.10 shows that the total displacement at failure was 33% more for the composite-to-composite joint when compared to the composite-to-aluminum joint, when both the bolts were tightened. For the scenario where both the bolts were loose, the total displacement at the ultimate failure was 38% higher for the composite-to-composite joint when compared t o composite-to-aluminum joint, as shown in Figure 4.11. The significant increase in the total displacement at failure for the composite-to-composite joints may be attributed to the cumulative damage and delamination in each of the two composite coupons. This also shows that composite-to-composite coupons observed more energy compared to composite-to-aluminum joints. Figures 4.12 and Figure 4.13

show an enlarged portion of the initial load-displacement curve for the composite-tocomposite and composite-to-aluminum joints with loose bolts (LT-LB) and tightened bolts (TT-TB), respectively. It can be observed that the joint material had no significant effect on the joint stiffness or the value of the tensile load that would initiate bearing failure (point C).

114

4.2.3 Failure Mode Progression A bearing failure mode is desired over other modes because of its progressive non-catastrophic nature. The fiber orientation, clearance of the bolt hole and clamping pressure are the factors, which affect the bearing failure mechanism. In this section, damage and failure mode analysis are provided for the composite regions that are near the contact with the shank of the bolt. The affect of bolt tightness on failure modes in double bolted composite joints is investigated. Figure 4.2 illustrates the two regions that are inspected for progressive damage analysis; namely, Region 1 for the damage due to under-head contact surface and Region 2 for the damage due to bolt shank contact. Figure 4.14 - Figure 4.16 show the micrographs of the internal damage through the composite thickness near the bearing contact with the bolt shank (Region 2-Figure 4.2). Figure 4.14a and Figure 4.14b show the damage caused by a tensile load level that is just above point C on Figure 4.4a and Figure 4.4d, respectively, where the bearing failure begins. This corresponds to 3kN for the LT-LB joint and 4kN for the TT-TB tightness conditions. Initiation of matrix cracking and interlaminar delamination was observed in the case of loose bolts (LT-LB), whereas in the case of tightened bolts (TT-TB), the compressive stress from both washers and the bolt shank was responsible for fiber breakage and matrix compression. Figures 4.15a and Figure 4.15b show the micrographs of the damage that corresponds to 12 kN tensile load (70% of the ultimate load), when the bolts are either both loose or both tightened, respectively. Interlaminar delamination was more prominent in the case of LT-LB than TT-TB bolts. In the case of TT-TB bolts, the matrix and fiber cracking was more obvious at the edge of the hole. This is due to the eccentric placing of the washer. Figure 4.16a and Figure 4.16b show the final damage 115

inspection that corresponds to about 15kN tensile load, which is just before the ultimate failure, point D in Figure 4.3a and Figure 4.3d. At this load, the interlaminar delamination and fiber-matrix shear failures increased and resulted in an out-of plane deformation when both bolts were loose (LT-LB), as shown in Figure 4.16a. However, when both bolts were tightened the lateral support from the washers inhibited the out of plane deformation; as a result, fiber compressive failure was observed as shown in Figure 4.16b. The progressive damage was carried out on the flat composite surface in contact with loose bolt heads (LT-LB) for various levels of tensile loads. It was observed that the surface delamination increased by increasing the tensile load. The surface delamination was significantly less for tightened bolt holes. The clamped washers significantly reduced the surface delamination. In some cases the surface delamination was partially caused by unintended eccentric positioning of the washers.

4.3 Finite Element Modeling The commercial finite element code ABAQUS [60] was used for the 3dimensional analysis of the tested composite-to-aluminum double bolted lap joint. Figure 4.17 show the finite element model of the single lap shear composite-to-aluminum joint. A refined mesh was built around the bolt holes and the contact region between the bolt head/shank and the composite coupons. The washer thickness was added to the modeled bolt head/nut in order to create a single entity with the outer diameter of the bolt head being equal to the outer diameter of the washer. A 3-dimensional, 8 nodes, reduced integration brick elements (C3D8R) were used for meshing the joint materials. An

116

orthotopic material model was used for the composite plate with 16 elements throughthickness layers that were stacked at 0 and 90 orientations. Isotropic material properties were used to model the steel fasteners and the aluminum plate. The material properties considered in the analysis are tabulated in Table 4.1 for the composite plates, aluminum plates, steel bolts, washers and nuts. The contact pair option in ABAQUS, which is based on a master-slave approach, was used to model the contact between the two plates, bolt shank and the plates, bolt bearing surfaces and the upper composite plate, and between nut bearing surfaces and the lower aluminum plate. Typical contact pair definitions in the finite element model are shown in the Figure 4.18. The coefficient of friction is different for various contact interfaces, and is generally lower for metal-to-composite contact as compared to that for a metal-to-metal contact [54]. A factional co-efficient value of 0.2 was used for all metal-to-composite interfaces and a value of 0.3 was used for metal-tometal contact. A "small sliding" option in ABAQUS was used in the analyses, which meant that the contact between the master and slave nodes was defined in the initial stage and were not redefined in the later stages of the analysis. The FEA loading was applied in two steps. First, the bolt preload was applied to its middle section; second, a tensile load was applied to the upper composite plate while fixed end boundary conditions were applied to the lower aluminum plate. The tensile load was applied as a concentrated nodal force; the rest of the nodes on the loaded edge are forced to have the same displacement as the loaded node by using the multi-point constraint (MPC) option in ABAQUS. Light springs were used in ABAQUS to compensate for the rigid body motion, as the bolt head sled to cause the shank to contact 117

with the joint, after the clearance in the bolt hole had been consumed. A bolt load of approximately 9000 N for a tight bolt condition and about 800 N for a loose bolt condition were used. The four scenarios of bolt tightness used in the experimental study (TT-TB, TTLB, LT-TB and LT-LB) were considered in a linear finite element model. Figure 4.19 shows the load-displacement results for the TT-TB and LT-LB scenarios using various values of the coefficient of friction at the metal-to-metal interface. It appears that increasing coefficient of friction does not seem to significantly affect the stiffness of the joint (slope of the load-displacement curve). Figure 4.20 shows the load-displacement curves for the four tightness scenarios; namely TT-TB, LT-LB, LT-TB and TT-LB bolts. The coefficients of friction were chosen to be 0.3 and 0.2 for the metal-on-metal and metal-on-composite contact, respectively. The load-displacement data show that the FEA results are in close agreement with the experimental results presented in this study.

4.4 Summary The study provides an experimental procedure and an FEA model for investigating the mechanical behavior and the failure modes of a composite single-lap double bolted joint. It has been demonstrated that sufficiently increasing the bolt tightness (without exceeding the joint strength), would significantly reduce the potential for delamination around the bolt hole when a tensile load is applied to the joint. The opposite has been found to be true as well. The tightening of at least one bolt has increased the tensile load that would just initiate the bearing failure in the composite plate. Joint stiffness is increased only when both bolts are sufficiently tightened. Both

118

the bolt tightening configuration and joint materials are found to have an insignificant effect on the ultimate strength of the single-lap double bolted composite joint. When the bolts were loose, the progressive damage analysis showed interlaminar delamination and fiber-matrix shear failure that is subsequently followed by an out-of-plane deformation at higher tensile loads. However, when both bolts are sufficiently tightened, a fiber compressive failure mode is observed. The linear part of the load-displacement curves obtained from the 3-D FEA show good correlation with the experimental results for all four scenarios of bolt tightness.

119

Table 4.1 Material properties of joint components

E-GlassEpoxy Woven

Alloy 6061 Aluminum

Steel Bolt-Washer-Nut AISI1035

Young's Modulus (E) Young's Modulus (En) Young's Modulus (E22) Young's Modulus
(E33)

68.9 GPa

200 GPa

30.678 GPa 29.782 GPa

7.583 GPa

Poisson's ratio (y) Poisson's ratio


(712)

0.33

0.29

0.114

Poisson's ratio
(Yl3= Y13)

0.27

Shear Modulus
(Gi2=Gi3=G23)

4.756 GPa

120

48 mm 3.18 mm

Upper coupon

, Bolts 0 8 mm 24 mm 191 mm "Top bolt 36 mm 298 mm

24 mm * 32 mm *

Bottom bolt

- Lower coupon

Figure 4.1 Geometry of single lap, double-bolted joint

121

Region 1 (Damage on the under head contact surface)

Region 2 (Damage at the bolt shank contact)

Bolt shank contact line

Loa&is; Difeciion

Figure 4.2 Schematic representation of inspected damage regions.

122

IO

16 14 ^ 12 i . 10 o ~" 6 4

LT-LB

/ ^

2 yftB n 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 Cross Head Displacement (mm)

(a)

16 14 ~ 12
1 10 Load

LT-TB D

| I

j 4 B ^ 2 n /A |

I
Cross Head Displacement (mm)

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6

(b) Figure 4.3 Load displacement curves for aluminum-composite joints: (a) LT-LB; (b) LTTB

123

CO O

0.5

1 1.5 2 2.5

3 3.5 4 4.5 5 5.5

Cross Head Displacement (mm)

(c)

T3

CD O

0.5

1.5

2.5

3.5

4.5

5.5

Cross Head Displacement (mm)

(d) Figure 4.3 Load displacement curves for aluminum-composite joints :(c) TT-LB; (d) TTTB. 124

LT-LB

0.2

0.4

0.6

0.8

1.2

1.4

displacement in mm

(a)

LT-TB

0.2

0.4

0.6

Oi

1.2

1.4

displacement in mm

(b) Figure 4.4 Initial portion of the aluminum-composite load-displacement curve: (a) LTLB; (b) LT-TB.

125

TT-LB

0.2

0.4

0.6

0.8

1.2

1.4

displacement in mm

(c)

TT-TB

0.2

0.4

0.6

0.8

1.2

1.4

displacement in mm

(d) Figure 4.4 Initial portion of the aluminum-composite load-displacement curve: (c) TTLB; (d) TT-TB

126

Composite-Aluminum joints Ultimate

T3

o
J

Cross Head displacement (mm)

(a)

Composite-Compoiste joints Ultimate failure

a S3
o

Cross Head displacement (mm)

(b) Figure 4.5 Load-displacement curves showing the ultimate failure load: (a) Compositealuminum joints; (b) Composite-composite joints

127

Top Hole

(a) Figure 4.6 Bearing surface delamination for TT-LB joints: (a) Top Hole; (b) Bottom Hole.

Top Hole

|:

Bottom Hole

(a)

(b)

Figure 4.7 Bearing surface delamination for LT-TB joints: (a) Top Hole; (b) Bottom Hole

128

HaBgaBBgaHWgmmBBimflflgHgBBm

Too Hole

Bottom Hole

(a)

(b)

Figure 4.8 Bearing surface delamination for LT-LB joints: (a) Top Hole; (b)Bottom Hole.

* & . . - _;
^ v *

,' ''^,''
* ^
%

" l *> I t'i' i


''?;>

^fi^^. *

* *. ^ fc ^ ^
i

\***'
- -.','?..

J^-K' -?*;* , .. A.V#


i, \*

'S '

Rotioni Hole

(a)

(b)

Figure 4.9 Bearing surface delamination for TT-TB joints: (a) Top Hole; (b) Bottom Hole.

129

Aluminum/Composite VS Composite/Composite for TT-TB joints

0.5

1.5

2.5

3.5

4.5

5.5

6.5

Displacement (mm)

Figure 4.10 Strength comparison of aluminum-composite and composite-composite TTTB joints.

Aluminum/Composite VS Composite/Composite for LT-LB joints 17 16 15 14 13 12 11 10 9 8 7H o 6 5 4 3 2 1 0 0

Al/Comp Comp/Comp

0.5

1 1.5

2 2.5

3 3.5

4.5

5 5.5

6 6.5 7

7.5

Displacement (mm)

Figure 4.11 Strength comparison of aluminum-composite and composite-composite LTLB joints.

130

0.5

1 Displacement in mm

1.5

(a)

8 7 6 5 4
(0 O

LTLB

Bearing failure load

3 2 1 0 0.5 1 Displacement in mm 1.5

(b) Figure 4.12 Initial portion of the load-displacement data for LT-LB joints: (a) Composite -composite joint; (b) Aluminum-composite joints.

131

c
T5
CO

0.5

1 Displacement in mm

1.5

(a)

0.5

1 Displacement in mm

1.5

(b) Figure 4.13 Initial portion of the load-displacement data for LT-TB joints: (a) Composite -composite joint; (b) Aluminum-composite joints 132

(a)

(b)

Figure 4.14 Initiation of bearing failure: (a) LT-LB joints; (b) TT-TB joints.

*?*

'sir

^ ) 8 ' -- -' .

(a)

(b)

Figure 4.15 Bearing failure at 70% of ultimate failure load: (a) LT-LB joints; (b) TT-TB joints.

133

3V3Mg3fn|

(a)

(b)

Figure 4.16 Bearing damage just before the ultimate failure: (a) LT-LB joints; (b) TT-TB joints.

134

Composite plate through-thickness section showing the layers Top Bolt Bottom Bolt Composite Plate

Aluminum Plate

Figure4.17 Finite element model of double bolted composite to aluminum joint.

Contact friction between bolt shank and plates

Contact friction between bolt underhead and composite plate

Contact friction between two plates 3

ontact friction between nut and aluminum plate

Figure 4.18 Contact surfaces in the finite element model.

135

Load Displacement Curve (TT-TB) 3500 3000 2500 & 2000


o

Friction Coefficient 0.2 0.2/0.25 0.2/0.3 Experimental 0.2 0.4 0.6

Displacement (mm)

(a)

Load Displacemet Curve (LT-LB)

Friction Coefficient -0.2 -0.2/0.25 0.2/0.3 Experimental 0.2 0.4 0.6

Displacement (mm)

(b) Figure 4.19 Frictional effect on the load displacement curves: (a) TT-TB; (b) LT-LB.

136

TT-TB

0.1

0.2 0.3 Displacement (mm)

0.4

0.5

(a)

LT-LB

0.2

0.4

0.6

0.8

Displacement (mm)

(b) Figure 4.20: Comparison of FEA and experimental results: (a) TT-TB; (b) LT-LB

137

LT-TB 3500 3000 2500 & 2000

15001000 500 0* 0

0.2

0.4

0.6

Displacement (mm)

(C)

TT-LB 3500 3000 2500 2000 1500 1000 500 0


0.2 0.4

0.6

0.8

(d) Figure 4.20: Comparison of FEA and experimental results :( c) LT-TB; (d) TT-LB.

138

CHAPTER FIVE CONCLUSIONS AND FUTURE STUDY


5.1 Conclusions 5.1.1 Effect of Adhesive Thickness and Properties on the Bi-axial Interfacial Shear Stresses in Bonded Joints Using a Continuum Mixture Model New formulas are derived for the bi-axial interfacial shear stresses that develop in an adhesively bonded joint due to static thermo-mechanical loading. The analysis is carried out along the lines of continuum mixture theories of wave propagation. Numerical results that show the effect of both the elastic properties and the thickness of the adhesive on the interfacial shear stresses are investigated. Numerical results show that both the material properties and the thickness of the adhesive have a pronounced effect on the developed interfacial shear stresses due to the thermo-mechanical loading. For the present model, it has been found that increasing the thickness of the adhesive causes a significant increase of the interfacial shear stresses. The larger difference in the elastic and thermal properties between the adhesive and the layered adherends the higher the corresponding interfacial shear stress is. The proposed model inherently has the capacity for optimizing the selection of the adhesive thickness and material properties that would yield a more reliable bonded joint. 5.1.2 Effect of Washers and Bolt Tension on the Behavior of Thick Composite Joints Experimental data is presented on the affect of initial bolt preload, as well as the washer size and thickness on the behavior of heavily loaded thick composite joints.

139

Friction between the joint parts played a significant role in defining joint stiffness. The joint bearing stiffness was higher for the untightened bolted joint than that with much higher bolt preload (100% of proof load). The bearing stiffness was smallest for the joint with a preload equal to 25% of bolt proof load, and it increased with bolt preload. The offset bearing strength increased progressively with bolt preload. The ultimate joint strength was unaffected by increasing the bolt preload. Joint with small washers had higher bearing stiffness than those with large washers for initial bolt preload of 0%, 25% and 50%. Joints with small washers had higher offset bearing strength than the joints with large washers. The washer thickness had an insignificant effect on the ultimate joint strength and strain. As the axial test load is increased, an untightened bolted joint showed a significant increase in the clamp load (from its zero initial value). This increase in the clamp load was progressively reduced by the increase in initial bolt-load. Bearing damage was the prominent failure mode in all tested joint configurations. The extent of bearing damage was more severe in smaller washer joints, where the damage extended beyond the washer contact. Out-of-plane buckling of laminates exerted lateral pressure on the washers resulting in an increase in clamp load during the experiments. These experimental results help in the selection of an optimum initial bolt preload and washer size and thickness in order to enhance composite joint performance and reliability.

5.1.3 Effect of Bolt Tightness on the Behavior of Composite Joints The study provides an experimental procedure and an FEA model for investigating the mechanical behavior and the failure modes of a composite single-lap 140

double bolted joint. It has been demonstrated that sufficiently increasing the bolt tightness (without exceeding the joint strength), would significantly reduce the potential for delamination around the bolt hole when a tensile load is applied to the joint. The opposite has been found to be true as well. The tightening of at least one bolt has increased the tensile load that would just initiate the bearing failure in the composite plate. Joint stiffness is increased only when both bolts are sufficiently tightened. Both the bolt tightening configuration and joint materials are found to have an insignificant effect on the ultimate strength of the single-lap double bolted composite joint. When the bolts were loose, the progressive damage analysis showed interlaminar delamination and fiber-matrix shear failure that is subsequently followed by an out-of-plane deformation at higher tensile loads. However, when both bolts are sufficiently tightened, a fiber compressive failure mode is observed. The linear part of the load-displacement curves obtained from the 3-D FEA show good correlation with the experimental results for all four scenarios of bolt tightness.

5.2 Future Study Based on the developed model, one can further extend the model to predict the interfacial peel stresses which are critical in brittle adhesive applications. An experimental methodology using optical techniques can also be developed to analyze the interfacial stresses under thermomechanical loading.

141

REFERENCES

1. Robert M. Jones, 1999, Mechanics of composite materials, second edition, BrunnerRoutledge, New York and London. 2. A.Baldan, 2004, Adhesively-bonded joints and repairs in metallic alloys, polymers and composite materials: adhesives, adhesion theories and surface pretreatment, Journal of Material Science, 39, 1-49. 3. R.D. Adams, 2005, Adhesive bonding science, technology and applications, Woodhead Publishing Limited, Cambridge England. 4. Weijian Ma, Rajesh Gomatam, Rick D. Fong and Erol Sancaktar, 2001, A Novel Mathematical Procedure to Evaluate the Effect of Surface Topography on the Interfacial State of Stress: Part I: Verification of the Method for Flat Surfaces, Journal of Adhesion Science and Technology, 15,13, 1533-1558. 5. Jin-Hwe Kweon, Jae-Woo Jung, Tae-Hwan Kim, Jin-Ho Choi, Dong-Hyun Kim, 2006, Failure of carbon composite-to-aluminum joints with combined mechanical fastening and adhesive bonding, Composite Structures, 75, 192-198. 6. Christina Helene Ficarra, 2001, Analysis of adhesive bonded fiber-reinforced composite joints, Masters Thesis, North Carolina state university. 7. A Baldan, 2004, Review- Adhesively-bonded joints in metallic alloys, polymers and composite materials: Mechanical and environmental durability performance, Journal of Materials Science, 39, 4729-4797. 8. Timoshenko S. P, 1925, Analysis of bi-metal thermostats, J. Opt. Amer., Vol. 11, pp.233-255. 9. E. Suhir, 1998, Interfacial stresses in bimetal thermostats, Journal of Applied Mechanics, Paper No. 90-WA/APM-l. 10. Adnan H. Nayfeh, 1977, Thermomechanically Induced Interfacial Stresses in Fibrous Composites, Fiber Science and Technology, 10, 195-209. 11. G.A. Hegemier, G.A. Gurtam and A.H. Nayfeeh, 1973, A Continuum Theory for Wave Propagation in Laminated and Fibrous Composites, International Journal of Solids and Structures, 9.

142

12. Adnan H. Nayfeh, Elsayed Abdel-Ati M. Nassar, 1978, Simulation of the Influence of Bonding Materials on the Dynamic Behavior of Laminated Composites, Journal of Applied Mechanics, 78-WA/APM-15. 13. A.H. Nayfeh, 1978, Dynamically Induced Interfacial Stresses in Fibrous Composites, Journal of Applied Mechanics, 45. 14. A.H. Nayfeh and E.A. Nassar, 1982, The Influence of Bonding Agents on the Thermomechanically Induced Interfacial Stresses in laminated Composites, Fiber Science and Technology, 16,157-174. 15. Yi-Hsin Pao, Ellen Eisele, 1991, Interfacial shear and peel stresses in multilayered thin stacks subjected to uniform thermal loading, Journal of Electronic Packaging, 113,164-172. 16. Hui-Shen Shen, J.G. Teng, J.Yang, 2001, Interfacial stresses in beams and slabs bonded with thin plate, Journal of Engineering Mechanics, 127, 4. 17. C.Q.Ru, 2002, Interfacial Thermal stresses in biomaterial Elastic Beams: Modified Beam Models Revisited, Journal of Electronic Packaging, 124. 18. Hyonny Kim, Keith Kedward, 2001, Stress analysis of in-plane, shear-loaded, adhesively bonded composite joints and assemblies, National Technical Information Service (NTIS), Springfield, Virginia 22161, April. 19. Chihdar Yang, Hai Huang, John S. Tomblin and Wenjun Sun, 2004, Elastic-plastic model of adhesive-bonded single-lap composite joints, Journal of Composite Materials, Vol. 38, No. 4. 20. John N. Rossettos, 2003, Thermal peel, warpage and interfacial shear stresses in adhesive joints, Journal of Adhesion science Technology, 17, 115-128. 21. Do Won Seo, Ho Chel Yoon, yang Bae Jeon, Hyo Jin Kim and Jae Kyoo Lim, 2004, Effect of overlap length and adhesive thickness on stress distribution in adhesive bonded single-lap joints, Key Engineering Materials, 270-273, 64-69. 22. Gang Li, Pearl lee-Sullivan, Ronald W. Thring, 1999, Nonlinear finite element analysis of stress and strain distributions across the adhesive thickness in composite single-lap joints, Composite Structures, 46, 395-403. 23. Toshiyuki Sawa, Izumi Higuchi and Hidekazu Suga, 2003, Three-dimensional finite element stress analysis of single-lap adhesive joints of dissimilar adherends subjected to impact tensile loads, Journal of Adhesion Science and Technology, 17, 16, 21572174.

143

24. Q.S. Yang, X.R. Peng, A.K.H.Kwan, 2004, Finite element analysis of interfacial stresses in FRP-RC hybrid beams, Mechanics Research Communications, 31, 331340. 25. J.P.M. Goncalves, M.F.S.F. de moura, P.M.S.T. de castro, 2002, A three-dimensional finite element model for stress analysis of adhesive joints, International Journal of Adhesion and Adhesives, 22, 357-365. 26. J.D.Mathias, M.Grediac, X. Balandraud, 2006, On the bidirectional stress distribution in rectangular bonded composite patches, International Journal of Solids and Structures, 43, 6921-6947. 27. Weijian Ma, Rajesh Gomatam, and Erol Sancaktar, 2003, A Novel Mathematical Procedure to Evaluate the Effect of Surface Topography on the Interfacial State of Stress Part II: Verification of the Method for scarf interfaces, Journal of Adhesion Science and Technology, 17,6,831 -846. 28. V.Kradinov, E.Madenci, D.R. Ambur, 2004, Combined in-plane and through the thickness analysis for failure prediction of bolted composite joints", AIAA 20041703, 45th AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics and Materials Conference, 19-22, April, Palm Springs, California. 29. P. P. Camanho and F. L. Matthews, 1999, A progressive damage model for mechanically fastened joints in composite laminates, Journal of Composite Materials, 33(24), 2248 - 2280. 30. P. P. Camanho, S. Bowron, and F. L. Matthews, 1998, Failure Mechanisms in Bolted CFRP, Journal of Reinforced Plastics and Composites, 17,205-233. 31. Hsien-Tang Sun, Fu-Kuo Chang and Xinlin Qing, 2002, The response of composite joints with bolt-clamping Loads, Part I: Model Development, Journal of Composite Materials, Vol. 36, No. 01. 32. Cooper C and Turvey, G.J., 1995, Effects of joint geometry and bolt torque on the structural performance of single bolt tension joints in pultruded GRP sheet material, Composite Structures, Vol. 32, pp 217-226. 33. ASTM standard D 5961/D 5961M, 2005, Standard test method for bearing response of polymer matrix composite laminates, ASTM international, west Conshohocken, PA 19428-2959, United States. 34. Sandra Polesky Walker, 2001, Thermal effects on the bearing behavior of composite joints, Ph.D Dissertation, University of Virginia, May.

144

35. A.R Hutchinson, Advanced materials engineering and joining-mechanical connections in polymer composite materials, Oxford Brookes University. 36. B. Vangrimde, R. Boukhili, 2002, Bearing Stiffness of Glass fiber-reinforced Polyester: Influence of Coupon Geometry and Laminate Properties, Composite Structures 58,57-73. 37. Li Hou and Dahsin Liu, 2003, Size Effects and Thickness Constraints in Composite Joints, Journal of Composite Materials, Vol. 37, No. 21. 38. Je Hoon Oh, Young Goo Kim & Dai Gil Lee, 1997, Optimum Bolted Joints for Hybrid Composite Materials, Composite Structures, Vol. 38, No. 1-4, pp. 329-341. 39. Alaattin Aktas, M. Husnu Dirikolu, 2004, An experimental and numerical investigation of strength characteristics of carbon-epoxy pinned-joint plates, Composites Science and Technology 64,1605-1611. 40. Alaattin Aktas, 2005, Bearing strength of carbon epoxy laminates under static and dynamic loading, Composite Structures 67,485^489. 41. M.A. McCarthy, V.P. Lawlor, W.F. Stanley, C.T. McCarthy, 2002, Bolt-hole clearance effects and strength criteria in single-bolt, single-lap, composite bolted joints, Composite Science and Technology 62,1415-1431. 42. V.P. lawlor, M.A. McCarthy, W.F.Stanley, 2002, Experimental Study on the Effect of Clearance on Single-Bolt, single-Shear, Composite Bolted Joints, Journal of Plastic, Rubber and Composites, Vol.31, No. 9, pp. 405-411. 43. Claire Girard, Marie-Laure Dano, Andre Picard, Guy Gendron, 2003, Bearing Behavior of Mechanically Fastened Joints in Composite laminates-Part I: Strength and Local Strains, Mechanics of Advanced Materials and Structures, 10:1-21. 44. Heung-Joon Park, 2001, Effect of Stacking Sequence and Clamping Force on the Bearing Strengths of Mechanical Fastened Joints in Composite Laminates, Composite Structures, 53, 213-221. 45. Hsien-Tang Sun, Fu-Kuo Chang, and Xinlin Qing, 2002, The Response of Composite Joints with Bolt-Clamping Loads, Part II: Model Verification, Journal of Composite Materials, Vol. 36, No 01. 46. Y.Yan, W.D.Wen, F.K.Chang, P.Shyprykevich, 1999, Experimental Study on Clamping Effect on the Tensile Strength of Composite Plates with a Bolt-Filled Hole", Composites Part A: Applied Science and Manufacturing, 30, 1215-1229.

145

47. U.A. Khashaba, H.E.M. Sallam, A.E. Al-Shorbagy, M.A.Seif, 2005, Effect of washer size and tightening torque on the performance of bolted joints in composite structures", Composite Structures. 48. V.P.Lawlor, W.F.Stanley, M.A.McCarthy, 2002, Characterization of damage Development in single-shear bolted composite joints, Journal of Plastics, Rubber and Composites, The Institute of Materials, London, UK, Vol. 31, No. 3, pp. 126-133. 49. Claire Girard, Marie-Laure Dano, Andre Picard and Guy Gendron, 2003, Bearing Behavior of Mechanically Fastened Joints in Composite Laminates-Part II: Failure Mechanisms", Mechanics of Advanced Materials and Structures, 10:23-42. 50. Yi Xiao, 2003, Bearing Deformation Behavor of Carbon/Bismaleimide Composites Containing One and Two Bolted Joints, Journal of Reinforced Plastics and Composites, Vol.22, No. 2. 51. Hong-Sheng Wang, Chang-Li Hung and Fu-Kuo Chang, 1996, Bearing Failure of Bolted Composite Joints. Part I: Experimental Characterization, Journal of Composite Materials, Vol. 30, No. 12. 52. Liyong Tong, 2000, Bearing failure of composite bolted joints with non-uniform bolt-to-washer clearance, Composites: Part A 31, 609-615. 53. A. Crosky, D.Kelly, R.Li, X.Legrand, N.Huong, R.Ujjin, 2006, "Improvement of Bearing Strength of Laminated Composites", Composite Structures 76, 260-271. 54. Tomas Ireman, 1998, Three-dimensional stress analysis of bolted single-lap composite joints, Composite Structures, 43,195-216. 55. Johan Ekh, Joakim Schon, 2005, Effect of Secondary Bending on Strength Prediction of Composite, Single shear lap Joints, Composite Science and Technology 65, 953965. 56. K.I. Tserpes, G.Labeas, P.Papanikos, Th. Kermanidis, 2002, Strength Prediction of Bolted Joints in Graphite/epoxy Composite Laminates, Composites: Part B 33, 521529. 57. Marie-Laure Dano, Guy Gendron, Andre Picard, 2000, Stress and Failure Analysis of Mechanically Fastened Joints in Composite Laminates, Composite Structures 50, 287-296. 58. McCarthy, C.T. McCarthy, V.P. Lawlor, W.F. Stanley, 2004, Three-dimensional finite element analysis of single-bolt, single-lap composite bolted joints: part Imodel development and validation, Composite Structures.

146

59. Arthur P. Boresi, Richard J. Schmidt, Omar M. Sidebottom, 1992, Advanced Mechanics of Materials, Fifth edition, John Wiley and Sons, Inc., USA 60. ABAQUS Documentation, 2004, Analysis User's Manual, Version 6.5, ABAQUS Inc., Rhode Island - 02909, USA, www.hks.com. 61. John H. Bickford, 1995, An introduction to the design and behavior of bolted joints, third edition, Marcel Dekker, New York. 62. John H. Bickford, Sayed Nassar, 1998, Handbook of bolts and bolted joints, Hand book of bolts and bolted joints, Marcel Dekker, New York. 63. Nassar, S.A. and Matin, P. H., and G. C. Barber, 2005, Thread friction in bolted joints, journal of pressure vessels technology- ASME Transactions, 127, pp. 387-393 64. Nassar, S.A., Barber, G.C., and Zuo, D., 2005, Bearing friction torque in bolted joints, STLE Tribology Transactions, 48, pp 69-75. 65. T.A. Collings and M.J. Beauchamp, 1984, Bearing deflection behavior of a loaded hole in CFRP, Composites, 15, 1. 66. Nassar, S.A. and Virupaksha, V.L., Ganeshmurthy, S., 2007, Effect of Bolt Tightening and Joint Material on the Strength and Behavior of Composite Joints, ASME Journal of Pressure Vessels Technology, Vol. 129, pp. 43-51. 67. ASTM Standard D5961/D 5961M-96, 1996, Standard Test Method for Bearing Response of Polymer Matrix Composite Laminates, ASTM international, west Conshohocken, PA 19428-2959, United States.

147

Вам также может понравиться