Citation
Connection capacity of pultruded GRFP channels in multidirectional loading

Material Information

Title:
Connection capacity of pultruded GRFP channels in multidirectional loading
Creator:
Wang, Michael C. ( author )
Place of Publication:
Denver, CO
Publisher:
University of Colorado Denver
Publication Date:
Language:
English
Physical Description:
1 electronic file. : ;

Subjects

Subjects / Keywords:
Glass-reinforced plastics ( lcsh )
Loads (Mechanics) ( lcsh )
Glass-reinforced plastics ( fast )
Loads (Mechanics) ( fast )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Abstract:
Fiber Reinforced Polymer (FRP) composites is a relatively new material that offers unique advantages for many different engineering applications. While there has been plenty of research performed to understand composite fiber failure mechanisms numerically and experimentally, the exact nature of failure based on fiber orientation is unknown. Due to the complex nature of FRP composites, fiber failure mechanisms are not clearly understood and the goal of this study is to gain further knowledge regarding FRP failure, particularly with bolted connections. The primary objective of this research is to investigate the behavior of bolted Glass Fiber Reinforced Polymer (GFRP) structural channel members subject to cyclic loading. The investigation consisted of testing various channel specimens with different flange geometric configurations. Monotonic loading is performed to determine ultimate strength of the connection in the axial and transverse directions for the first phase. The second phase comprises of how unidirectional cyclic loading effects stiffness and the overall ultimate strength of the connection in the axial and transverse direction. The last phase studies the behavior and the residual connection capacity of the GFRP specimens subject to multidirectional cyclic loading. Various load levels for cyclic loading were established as a fraction of the ultimate strengths obtained from testing. The testing showed that the absence of the flange did not have a significant impact with respect to the ultimate strengths in the axial and transverse directions. Failure modes observed consisted of initial bearing failure followed by shear-out failure for axial specimens and net tension splitting for the transverse specimens. The GFRP channel specimens undergo a period of nonlinear progressive damage after initial failure. Furthermore, it was found that the residual behavior and strength was affected by unidirectional and multidirectional load cycles. A probability-based design was adopted to provide design recommendations and guidelines. Resistance factors are determined using the Monte-Carlo simulation for the Load and Resistance Factor Design (LRFD) design method.
Thesis:
Thesis (M.S.)--University of Colorado Denver. Civil engineering
Bibliography:
Includes bibliographic references.
System Details:
system requirements: Adobe Reader.
General Note:
Department of Civil Engineering
Statement of Responsibility:
by Michael C. Wang.

Record Information

Source Institution:
|University of Colorado Denver
Holding Location:
|Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
880351177 ( OCLC )
ocn880351177

Downloads

This item has the following downloads:


Full Text
CONNECTION CAPACITY OF PULTRUDED GFRP CHANNELS
IN MULTIDIRECTIONAL LOADING
by
MICHAEL C. WANG
B.S., University of Colorado Denver, 2008
A thesis submitted to the
Faculty of the Graduate School of the
University of Colorado in partial fulfillment
of the requirements for the degree of
Master of Science
Civil Engineering
2013


This thesis for the Master of Science degree by
Michael C. Wang
has been approved
for the Civil Engineering Program
by
Yail Jimmy Kim, Advisor
Kevin Rens, Chair
Frederick Rutz
July 12, 2013
n


Wang, Michael, C. (M.S., Civil Engineering)
Connection Capacity of Pultruded GFRP Channels in Multidirectional Loading
Thesis directed by Associate Professor Dr. Yail Jimmy Kim
ABSTRACT
Fiber Reinforced Polymer (FRP) composites is a relatively new material that offers
unique advantages for many different engineering applications. While there has been
plenty of research performed to understand composite fiber failure mechanisms
numerically and experimentally, the exact nature of failure based on fiber orientation is
unknown. Due to the complex nature of FRP composites, fiber failure mechanisms are
not clearly understood and the goal of this study is to gain further knowledge regarding
FRP failure, particularly with bolted connections.
The primary objective of this research is to investigate the behavior of bolted Glass Fiber
Reinforced Polymer (GFRP) structural channel members subject to cyclic loading. The
investigation consisted of testing various channel specimens with different flange
geometric configurations. Monotonic loading is performed to determine ultimate strength
of the connection in the axial and transverse directions for the first phase. The second
phase comprises of how unidirectional cyclic loading effects stiffness and the overall
ultimate strength of the connection in the axial and transverse direction. The last phase
studies the behavior and the residual connection capacity of the GFRP specimens subject
to multidirectional cyclic loading. Various load levels for cyclic loading were established
as a fraction of the ultimate strengths obtained from testing.


The testing showed that the absence of the flange did not have a significant impact with
respect to the ultimate strengths in the axial and transverse directions. Failure modes
observed consisted of initial bearing failure followed by shear-out failure for axial
specimens and net tension splitting for the transverse specimens. The GFRP channel
specimens undergo a period of nonlinear progressive damage after initial failure.
Furthermore, it was found that the residual behavior and strength was affected by
unidirectional and multidirectional load cycles.
A probability-based design was adopted to provide design recommendations and
guidelines. Resistance factors are determined using the Monte-Carlo simulation for the
Load and Resistance Factor Design (LRFD) design method.
The form and content of this abstract are approved. I recommend its publication.
Approved: Jimmy Kim
IV


ACKNOWLEDGMENTS
My sincere gratitude goes to my advisor Dr. Yail Jimmy Kim and co-advisor Dr. Rui Liu.
It has been a pleasure to work with both and I thank them for their guidance, support, and
time during the preparation of this thesis. They provided me with invaluable knowledge
and expertise necessary for completion of the research project. Also Id like to thank Dr.
Kevin Rens and Dr. Frederick Rutz for participating on my thesis defense committee.
The material testing conducted in this research study would not have been possible
without the support provided by the following staff and students: Tom Thuis, Jac Corless,
Eric Losty, and Dennis Dunn. A special thanks to Shahlaa Alwakeel, for her guidance
and assistance in the lab. Your hard work is greatly appreciated. Finally, the author
gratefully acknowledges support from International Cooling Towers.
I would also like to express my appreciation to my fellow colleagues and mentors David
Blanchette and Jerry Isler, both have been instrumental in the development of my
professional and academic career.
Lastly, I would like to thank my friends and family for their support and encouragement
along the way.
v


TABLE OF CONTENTS
Tables ...........................................................................x
Figures ..........................................................................xi
Chapter
1. Introduction................................................................1
1.1 General.....................................................................1
1.2 Objectives..................................................................2
1.3 Scope.......................................................................3
1.4 Outline of Thesis...........................................................3
2. Literature Review...........................................................5
2.1 Introduction................................................................5
2.2 Overview....................................................................5
2.3 Material Properties of FRP..................................................6
2.4 Fabrication of FRP Members.................................................7
2.5 Bolted FRP Members..........................................................8
2.6 Fasteners...................................................................9
2.7 Failure Modes...............................................................9
2.7.1 Tension Failure............................................................10
2.7.2 Shear Failure..............................................................11
2.7.3 Bearing Failure............................................................11
2.7.4 Cleavage and Pull-out Failure..............................................12
2.8 Factors Affecting Joint Strength...........................................12
2.8.1 Fiber Orientation..........................................................12
vi


2.8.2 Lateral Contraint........................................................13
2.8.3 Stacking Sequence........................................................13
2.8.4 Joint Geometry...........................................................14
2.9 Static Behavior of Pultruded GFRP Beams..................................14
2.9.1 Experimental Setup and Test Procedure....................................15
2.9.2 Results and Conclusions..................................................15
2.10 Multi-Bolted Joints For GFRP Structural Members..........................16
2.10.1 Test Parameters..........................................................17
2.10.2 Test Setup and Instrumentation...........................................17
2.10.3 Experimental Results.....................................................18
2.10.4 Conclusions.............................................................19
2.11 Finite Element Modeling of Damage Accumulation...........................19
2.11.1 Progressive Damage Model.................................................20
2.11.2 Finite Element Modeling.................................................23
2.11.3 Results and Discussions.................................................24
3. Experimental Program.....................................................39
3.1 Introduction.............................................................39
3.2 Materials used for Test Specimens........................................39
3.2.1 GFRP Channels............................................................39
3.2.2 GFRP Flat Plates.........................................................40
3.3 Experimental Setup and Loading...........................................41
3.3.1 Phase I and Phase II Setup for Axial Tests...............................41
3.3.2 Phase I and Phase II Setup for Transverse Tests..........................42
3.3.3 Phase III Testing Procedure..............................................42
vii


48
48
48
49
50
54
54
55
56
57
57
58
59
60
62
63
63
63
64
64
64
65
Finite Element Modeling
Introduction.............
Modeling and Elements.................................
Failure Criteria......................................
Model Results.........................................
Test Results and Discussion...........................
Introduction..........................................
Phase I Testing Results of GFRP Channels..............
Phase I Failure Modes of GFRP Channels................
Phase I Testing Results of GFRP Thickened Plates......
Phase I Failure Modes of GFRP Thickened Plates........
Phase II Testing Results of GFRP Channels.............
Phase II Failure Modes of GFRP Channels...............
Phase III Testing Results of GFRP Channels............
Phase III Failure Modes of GFRP Channels..............
Comparisons of Phases.................................
Comparison Among Different Flange Geometries in Phase I.
Comparison of Post-Cyclic Residual Behavior...........
Effect of Cyclic Loading on Residual Capacity.........
Design Recommendations................................
Monte-Carlo Simulation................................
Calibration of Resistance Factor......................
vm


6. Summary and Conclusions................................................82
6.1 Summary................................................................82
6.2 Conclusions............................................................82
6.3 Recommendations for Future Work........................................85
Appendix
A Load Displacement Curves for Phase I Axial and Transverse............86
B Load Displacement Curves for Phase II Residual Capacity..............98
C Load Displacement Curves for Phase III Residual Capacity............101
D Phase I Testing Failure Photos Axial and Transverse.....................103
E Phase II Testing Photos Damage From Cyclic Loading..................113
F Phase III Testing Photos Damage From Multi Directional Cyclic Loading ....116
G Phase III Testing Failure Photos......................................118
References..................................................................119
IX


LIST OF TABLES
Table
2.1 Comparison of Flexural Rigidity.............................................26
2.2 Comparison of Shear Rigidity................................................26
2.3 Geometrical Data for the Investigated Geometry..............................27
2.4 Elastic Properties of HTA/6376 Material.....................................27
4.1 Material Properties for GFRP Channel Specimens (Extren).....................51
4.2 Failure Criteria Strength Values............................................51
5.1 Phase I Ultimate Failure Load of GFRP Channel Specimens.....................67
5.2 Phase I Ultimate Failure Load of GFRP Thickened Members.....................67
5.3 Phase II Residual Connection Capacity of Members Subject to Cyclic Loading.... 68
5.4 Phase III Residual Connection Capacity of Members Subject to Cyclic Loading... 68
5.5 Resistance Factor Calibration...............................................69
x


LIST OF FIGURES
Figure
2.1 FRP Pultrusion Process.....................................................28
2.2 Modes of Failure for Bolted Joints in FRP Composities......................28
2.3 Stress Concentration Relief in Fibrous Composites by Delamination..........29
2.4 Relation Between Stress Concentration Factors..............................29
2.5 Terminology.................................................................30
2.6 Influence of CFRP Fiber Proportion..........................................30
2.7 Effect of Bolt Torque on Bearing Strength..................................31
2.8 Comparison between Bearing Strength with and without Clamping Pressure......31
2.9 Effect of Grouped 0 Plies on Bearing Strength..............................32
2.10 Effects of Stacking Sequence on Bearing Strength for GFRP Laminates.......32
2.11 Variation of Net Tensile Strength with Width of 0/45...................33
2.12 Cross Sections of Test Specimens..........................................33
2.13 Connection Configuration...................................................34
2.14 Flowchart of Progressive Damage Model......................................35
2.15 Geometry of Bolted Single-Lap Joint........................................36
2.16 Finite Element Model of Bolted Joint......................................36
2.17 Mesh Around Hole and Finite Element Model of Bolt.........................37
2.18 Comparison of Strains......................................................37
2.19 Illustration of Damage Propagation in Tipper Surface of Laminate...........38
2.20 Illustration of Damage Propagation in Lower Surface of Laminate............38
xi


3.1 Condition #1.............................................................43
3.2 Condition #2.............................................................43
3.3 Condition #3.............................................................44
3.4 Tension Test.............................................................44
3.5 Shear Test...............................................................45
3.6 Geometry of Thickened Flat Plate.........................................45
3.7 Axial Test Setup.........................................................46
3.8 Transverse Test Setup....................................................46
3.9 Thickened Flat Plate Test Setup..........................................47
3.10 Thickened Flate Plate Test Setup.......................................47
4.1 Channel Geometry of ANSYS Model..........................................52
4.2 Applied Loading of Model.................................................52
4.3 Displacement from 25% of Ultimate Load...................................53
4.4 Tsai-Wu Strength Index 25% of Ultimate Load...........................53
5.1 Load-Displacement Curve for Phase I Axial................................70
5.2 Load-Displacement Curve for Phase I Transverse...........................70
5.3 Load-Displacement Curve for Phase I Thickened Member Axial...............71
5.4 Typical Failure Mode: Monotonic Axial Load...............................71
5.5 Typical Failure Mode: Monotonic Transverse Load..........................72
5.6 Failure Mode: (a) Monotonic Axial Load; (b) Monotonic Transverse Load....72
5.7 Typical Failure Mode of GFRP Thickened Plate from Axial Loading..........73
5.8 Debonding of GFRP Thickened Plate........................................73
5.9 Cyclic Unidirectional Load for Phase II Specimens........................74
5.10 Load-Displacement Curve for Phase II Monotonic Axial Loading...........74
5.11 Load-Displacement Curve for Phase II Monotonic Transverse Loading......75
xii


5.12 Damage Propagation with Load Cycles.....................................75
5.13 Geometry of GFRP Channel Specimens for Phase III........................76
5.14 Cyclic Multidirectional Load for Phase II Specimens.....................76
5.15 Load-Displacement Curve for Phase III Monotonic Axial Loading...........77
5.16 Load-Displacement Curve for Phase III Monotonic Transverse Loading......77
5.17 Damage Propagation with Load Cycles.....................................78
5.18 Typical Failure Mode: Monotonic Axial Load for Phase III................78
5.19 Typical Failure Mode: Monotonic Transverse Load for Phase III...........79
5.20 Failure Mode: Monotonic Transverse Load for Phase III...................79
5.21 Load-displacement of Phase I-Comparison of Various Flange Geometry......80
5.22 Comparison of Post-Cyclic Residual Behavior.............................80
5.23 Effect of Cyclic Loading on Residual Capacity...........................81
5.24 Comparison of Strength Reduction Factor (0) Based on Phase 1............81
xiii


1.
Introduction
1.1 General
High cost from repair and maintenance of steel structures damaged from corrosion lead
engineers and researchers to search for alternative materials. Fiber Reinforced Polymer
(FRP) composites is a relatively new material that offers unique advantages for many
different engineering applications. A composite material is formed by the combination of
two or more distinct materials to produce a new material with enhanced properties. FRP
utilizes a polymer resin matrix that is typically reinforced with glass, carbon, basalt or
aramid fibers. The advantages of using FRP are the excellent strength-to-weight ratio
and stiffness-to-weight ratio which makes them highly desirable as a building material for
structural systems. The composite material has proven to be economical and efficient for
new construction and repair and rehabilitation of damaged or deteriorating structures in
civil engineering. In addition, they provide favorable corrosion and weathering
resistance.
The application of FRP composites in structural engineering involves strengthening of
beams, columns, and slabs in existing structures. They can be used for repairing structural
members that have been damaged from loading conditions. For example, FRP can be
applied to provide flexural and shear strengthening of a damaged beam or column. The
strengthening will improve the stiffness and deflection capacity of the member. If
flexural strengthening is desired FRP sheets or plates are applied to the bottom or tension
face of the beam. When FRP is applied to the web or sides of the beam the shear strength
is improved.
1


FRP composites can also be molded into different structural shapes and is used as an
alternative to steel and aluminum due to its benefits. Today manufacturers can produces a
variety of structural members including channels, angles, wide flange beams, tubes, and
flat-sheets similar to structural steel shapes.
While there has been plenty of research performed to understand composite fiber failure
mechanisms numerically and experimentally, the exact nature of failure based on fiber
orientation is unknown. Due to the complex nature of FRP composites, fiber failure
mechanisms are not clearly understood and the goal of this study is to gain further
knowledge regarding FRP failure, particularly with bolted connections. Glass fibers are
commonly used for reinforced composites. Glass fiber reinforced polymers (GFRP)
channel members with bolted connections will be tested and researched in this study.
1.2 Objectives
The objective of this research is to evaluate GFRP channel members subject to cyclic
axial and transverse loading. The components involved in the investigation include the
following:
1. Review of previous research and testing conducted on GFRP structural members.
2. Examine the ultimate connection load capacity of the GFRP specimens in the
axial and transverse directions.
3. Observe various failure modes based on different loading and geometry
conditions.
2


4. Evaluate the impact of unidirectional and multidirectional loading and unloading
cycles and determine the effect it has on the residual strength capacity of the
connection. Observe the change in stiffness at various load levels.
5. Develop a finite element model to predict the load-deflection behavior and local
stress characteristics of the specimens.
6. Summarize results to provide design recommendations.
1.3 Scope
The scope of the research consists of an experimental program followed by a finite
element model to study the behavior of GFRP channel members with bolted connections
subject to unidirectional and multidirectional cyclic axial and transverse loading. Testing
was performed to determine ultimate strengths and to apply load cycles at different
levels. The experimental investigation studies the local deterioration around the bolt holes
influenced by high stress concentrations and the reduction of ultimate strength and
stiffness caused by repeated loads. A finite element model was developed to validate and
predict the deflection response and stress concentrations. The model accounts for change
in material properties due to cyclic loading.
1.4 Outline of Thesis
The contents of this thesis include the following:
Chapter 2: presents a review of literature related to previous research conducted on
GFRP structural members.
3


Chapter 3: provides a detailed description of the experimental program, fabrication of
test specimens, instrumentation, test setup and procedures. In addition, all ancillary tests
required for necessary material properties are discussed.
Chapter 4: provides a detailed description of the finite element model developed to
validate and predict the displacement and stress response to cyclic loading of GFRP test
specimens.
Chapter 5: discusses the experimental program results, which includes the strength and
stiffness degradation due to cyclic loading, various failure modes from different
geometric and loading conditions, and results of all ancillary material tests performed.
Design recommendations are provided using the results obtained.
Chapter 6: presents the summary and conclusion of the research along with
recommendations for further research on GFRP members.
4


2.
Literature Review
2.1 Introduction
The literature presented in this chapter will investigate past research and tests performed
on FRP members. The focus of this section will be failure behavior and mechanism from
bolted connections and how they are affected by fiber orientation.
This section will discuss the literature available on the subject of GFRP structural
members. It is evident based on the study of current literature available that there is a lack
of information about how to thoroughly characterize pultruded GFRP members.
Currently there is no design standard for FRP.
2.2 Overview
In recent years there has been significant growth of FRP, there usage has spread from
aerospace and automotive industry to civil structures. When compared with traditional
materials the significant advantages are the resistance to corrosion, lightweight, high
strength, and ease of installation. However, negative aspects of FRP would include the
high initial cost, durability problems caused by freeze-thaw cycles, moisture, or sustained
loading, and the lack of design standards and experience. Their strength and stiffness
properties are dependent on the type, quantity, and orientation of the fibers within the
member.
5


2.3 Material Properties of FRP
Material properties of FRP are complex due to the orthotropic nature of the composite
material and a clear explanation can be quite difficult. Unlike the American Concrete
Institute (ACI) for reinforced concrete design and the American Institute of Steel
Construction (AISC) for structural steel design, FRP pultruded structural shapes currently
do not have a design code, although two design guides are available for engineers:
Structural Plastics Design Manual and Eurocomp Design Code and Handbook (ASCE
1984 and Eurocomp 1996). To complicate matters further, there are many FRP
manufacturers in the industry who provide different material properties for their products.
Depending on the fiber orientation within the polymer matrix FRP can exhibit different
material properties. Other important factors that influence the behavior of FRP materials
include the type of fiber and polymer used in the composite (Tibbetts 2008). Common
reinforcement fibers used include glass, carbon, aramid, and boron fibers. The unique
anisotropic behavior requires special considerations in the manufacturing process and
design of FRP materials.
FRP composites pose a high resistance to corrosion regardless of the materials used for
the fiber and polymers. However, the durability can be reduced when exposed to harsh
environmental conditions which should be avoided for adequate performance, ranging
from high temperatures, ultraviolet exposure, and water with high alkalinity levels. The
resistance to environmental effects is greatly dependent on the fabrication process along
with the material properties of the fibers (Benmokrane et al 2002).
6


In comparison to steel, FRP structural members have anisotropic properties, low stiffness,
and high elastic to shear modulus ratios. From previous experience and tests composite
materials must be carefully analyzed due brittle failure modes. FRP differ from metals as
typically there is no yielding, strain hardening or elongation prior to failure (Deng 2004).
Generally GFPR members have a linear stress-strain relationship to failure. Certain types
of GFRP can exceed the strength of conventional steel. A study preformed reported the
ultimate strength of GFRP bars at 150 ksi (1,035 MPa) (Pleimann 1987). However, the
modulus of elasticity when compared to steel is significantly smaller at approximately
25% of steel which places limitation for the use of GFRP as the main load-carrying
element in many types of structures. Graphite reinforcing fibers can be used to increase
the modulus of elasticity by up to three times (Saadatmanesh and Ehsani 1991).
According to ACI (2006), the tensile strength of GFRP bars can up to twice as large as
the tensile strength of steel bars and is even higher for carbon and aramid FRP.
2.4 Fabrication of FRP Members
Fabrication of FRP composites can be completed by the method of pultrusion or by hand
layup (Tibbetts 2008). This research uses FRP products that were manufactured through
the process of pultrusion. For fabrication by pultrusion, FRP members are formed into
different standard structural shapes by an automated process. The manufacturing process
produces continuous lengths of various shapes and materials by combining fibers with a
polymer resin matrix under heat and pressure and pulled through using a heated steel
forming die. Fibers typically used are glass and carbon, although other fibers such as
7


aramid or polyethylene can be used for structural applications (Bank 2006). The basic
manufacturing process concept is described in Figure 2.1.
2.5 Bolted FRP Members
Plenty of research has been performed studying the behavior of FRP in the aerospace and
automotive engineering industries; however there has been a lack of research conducted
for structural engineering applications for FRP materials particularly in the field of bolted
connections (Hassan et al. 1994). Similar to steel, FRP structural members utilize bolted
connections and although mechanically fastened, FRP joints share the same failure modes
as metals. It has been determined that the damage mechanism and propagation is not
quite the same. Therefore, the design criteria for steel are not applicable to FRP members
(Duthinh 2000). It is important to understand the behavior or FRP members with bolted
connections because the strength of the structural member is limited by the strength of the
connection.
FRP structural members can be significantly weakened by the presence of bolt holes or
cut outs. The strength reduction is attributed to high stress concentrations around
discontinuities, damage to the reinforcing fibers and because FRP composites do not
yield. The stress concentration value around a hole for isotropic materials is 3 while a
uni-directionally FRP sheet can have a value as large as 8 (Codings 1987). Generally
isotropic materials exhibit plasticity that relieve high stress concentrations resulting in a
small effect on the net failure stress. However, for uni-directional FRP due to the lack of
8


plasticity the material is elastic to failure where the stress concentrations reduce the net
failure stress.
2.6 Fasteners
Mechanical fasteners have proven to be an effective method for connecting FRP
members. Different types of steel fasteners can be used for connection joints such as nuts,
bolts, threaded rods, rivets and self-tapping screws. Self-tapping screws are favorable
when it is not possible to access the reverse side of the joint. The use of rivets are
appropriate for joining laminates up to 0.1 in (3 mm) thick. However, the installation
operation of rivets can potentially damage the laminates due to uncontrollable clamping
pressure. In addition to steel fasteners FRP mechanical fasteners are also available which
include nut, bolts, threaded rods, and screws. These fasteners present disadvantages when
compared to steel due to the high cost and are susceptible to shear failure at low loads.
Although there are various options for fasteners Collins (1977) determined that bolted
joints were the most efficient and effective method for FRP materials.
2.7 Failure Modes
FRP have similar failure modes to steel member connections by failing in shear, tension,
or bearing. The connector can also fracture and pull through the laminates. In addition,
FRP can also fail by delamination between layers, debonding, and crack propagation. The
different types of failure modes for bolted joints in FRP composites are presented in
Figure 2.2. The best performance for shear, tension, and bearing failure is obtained
utilizing 0/45 fiber orientations.
9


2.7.1 Tension Failure
When members are subject to a tensile load the average net stress on across a section is:
P
n (w nd)t
(2-1)
Where P is the tensile load at the joint having a width w and thickness t. The number of
bolts within the section is n having a diameter d. Generally, stress concentration at the
hole will initiate failure because FRP does not yield to alleviate high stress
concentrations. However, a study by Hart-Smith (1980) showed that slippage between the
resin and fibers near bolt holes and cut-outs provide some relief of high stress
concentrations (Figure 2.3 and Figure 2.4). When the fibers experience high local stress
they pull out of the resin resulting in delamination or debonding of the fibers. Hart-Smith
(1980) concluded that around holes improving adhesion between the fibers and resin
results in a reduction of joint strength. However, reinforcing fibers near the hole in
various directions diminishes the degree of anisotropy allowing minor plastic behavior
and softening.
Fiber orientation of the FRP dictates the tensile strength of the composite. The majority
of the load is carried by the fibers parallel to the load. Potter (1978) determined that the
failure near holes in CFRP occurred at the edge of the hole perpendicular to the loading
axis. For GFRP the failure is more complicated and is influenced by both shear and
tension. Failure around the hole is initiated by in-plane shearing across the width of the
10


laminate. The shear results in only the 45 plies taking the axial load leading to tension
failure immediately after (Godwin et al. 1982).
2.7.2 Shear Failure
The shear stress t in a FRP joint is calculated by the following equation:
P
T ~2et
(2-2)
Where P is the applied load with thickness t and a distance from the bolt hole center to
the ends of the connected plates is e. See Figure 2.5 for a diagram of the terminology
used. It is noted that the shear stress is determined using the same procedure as for
isotropic materials. Similar to tensile loading the shear strength around the hole is
dependent on the fiber orientation of the member. The in-plane shear strength is
significantly lower with 0 fibers when compared with joints that have 45 fibers.
2.7.3 Bearing Failure
The average bearing stress at the cross section of the hole can be calculated by:
P
a = r
nat
(2-3)
Where P is the load applied, n is the number of bolts with diameter d and with a material
thickness of t. Bearing stress is caused by compression of half the bolt hole transferred by
the bolt. The compressive strength within the fibers and the clamping pressure are the
main parameters that will dictate the bearing strength around the hole (Collins et al.
11


1977). In a study by Collins (1987) he showed that the shear failure initiated at the hole
edge through the fibers and matrix. FRP laminates with a combination of 45 and 90
plies performed well under bolt bearing loads. This discovery contradicted the
compressive performance of laminates with 45 and 90 fiber orientations which
indicates the failure mechanisms for bearing and compressive loadings are different. The
difference can be attributed to the clamping pressure generated by the bolt. The fibers
perpendicular to the load and laminates are constrained and fail in constrained transverse
compression.
2.7.4 Cleavage and Pull-out Failure
In addition to shear, tension, and bearing failure FRP joints can exhibit cleavage and pull-
out failure. Cleavage failure develops only during 0 fiber orientation in a single shear
mode and is followed by net section failure occurring on one side of the laminate.
Reinforcement should be placed around the bolt holes to prevent cleavage failure. Pull-
out failure is initiated by out of plane bending at the joint caused from in plane axial
loads which results in peeling at the joint. Typically, they occur with the use of rivets in
single shear. Bending or shear failure of the bolt is also possible under extreme loading
conditions (Duthinh 2000).
2.8 Factors Affecting Joint Strength
2.8.1 Fiber Orientation
The biggest factor affecting the joint strength of FRP is the fiber orientation. In a study
by Collins (1977), concluded that optimum joint performance for CFRP is achieved using
12


a combination of 0 / 45 plies. It can be seen from Figure 2.6 that shear failure occurs
with low levels of 45 fibers. The shear strength of the joint increases as the 45 fibers
increase in the section, eventually bearing becomes the critical mode of failure. As a
result of 45 fibers being weak in tension, high amounts of 45 will shift the failure
mode to tension.
2.8.2 Lateral Constraint
Lateral constraint caused by clamping pressure from the bolts can significantly affect the
joint strength. However, excessive pressure from over-tightening of the bolts can damage
the surface of the laminate. The recommended optimum bolt clamping pressure was
determined to be 3,190 psi (22 MPa) for CFRP joints (Garbo and Ogonowski 1981).
Overtime clamping pressure is decreased by resin creep but does not necessarily reduce
the joint strength (Shivakumar and Crews 1982). See Figure 2.7 Figure 2.8 for the
effect that clamping pressure has on the bearing strength of FRP laminates.
2.8.3 Stacking Sequence
In a study performed by Collins (1977) it was shown that there was no change in shear
strength for bolted CFRP joints that consisted of 2/3 0 and 1/3 45 plies, however there
was a difference of 6% in tensile strength when compared with two different stacking
sequences. The bearing strength had a substantial drop with laminates grouped together.
The highest bearing strength for pin-loaded holes was achieved by using 90/45/0
laminates whereas using 0/90/45 reduced the strength by 30% (Quinn and Mathews
13


1977). See Figure 2.9 and Figure 2.10 for the effect that the stacking sequence has on the
bearing strength of GFRP laminates.
2.8.4 Joint Geometry
The tensile failure stress of FRP members with a hole heavily depend on the width due to
no stress relief from lack of plasticity. The width has a significant impact on the strength
when the laminates consists of mostly 0 fibers and an opposite effect on 45 fibers. This
can be seen in Figure 2.11. Collins (1977) determined that the hole size didnt have a
significant impact on the net tensile and shear strength with fiber orientation consisting of
0 / 45 for CFRP. The same could be said for GFRP laminates (Kretsis and Matthews
1985). The bearing strength of CFRP is not affected by hole size as long as there is
sufficient clamping pressure provided by the bolt. On the other hand, because GFRP has
low elastic modulus of glass, out-of-plane cracking can occur regardless of clamping
pressure for d/t > 3. Values of d/t < 3 is not recommended due to the possibility of bolt
shear failure. It is desirable that the joint geometry selected experiences tension and shear
failure simultaneously near the bearing failure stress.
2.9 Static Behavior of Pultruded GFRP Beams (Nagaraj and GangaRao 1997)
A study was performed by Nagaraj and GangaRao (1997) to investigate the experimental
and theoretical characterizations of mechanical properties of pultruded GFRP structural
members. A total of 187 test were conducted using wide flange and box beams to
examine the effects of shear influence, shear lag, warping, and manufacturing quality.
14


2.9.1 Experimental Setup and Test Procedure
The testing consisted of two different structural shapes with different sizes. The sizes
were 102 x 102 x 6 mm (4 x 4 x Vi in.) for the box section and the wide flange section
was 102 x 102 x 6 mm (4 x 4 x Vi in.) and 152 x 152 x 6 mm (6 x 6 x Vi in.). See Figure
2.12 for the cross section dimensions of the test specimens. The members were made up
with a vinylester matrix reinforced with E-glass fibers. The specimens were testing using
a span length of 1,828 mm (72 in.) with simply supported boundary conditions under
three-point and four-point bending. Linear variable differential transducers (LVDTs)
were used to measure the deflection under the load points at midspan. Electrical
resistance strain gauges were placed at a distance of 203 305 mm (8 12 in.) away
from the load point to overcome stress concentration effects. The strain gauges were
installed on both the compression and tension face at equal distance from the midspan.
2.9.2 Results and Conclusions
The experimental results were compared using theoretical computations of simplified
equations based on classical lamination theory (CLT) (Jones 1975) along with three-
dimensional finite-element analysis using ANSYS 1994. Both the flexural and shear
rigidities were determined based on the approximate CLT. The finite-element model
utilized SHELL 91 elements which had the ability to model the composite material layer
by layer and specify the material properties and fiber orientation. The finite-element
results compared well with the theoretical calculations and experimental results and were
within 0-4%. The comparisons are displayed in Table 2.1 and 2.2.
15


It was determined that the shear influence on deflection measurements for both three-
point and four-point bending was significant. The testing resulted in a shear influence of
36% for three-point bending test and 25% for the four-point bending test. Using 45
layers reduced the shear influence by 9% compared to the specimen with unidirectional
fibers in the web. The variation in strain readings was 10-20% and was attributed to
interfacial slip between the top layer and the layer beneath it. In addition, variation in
strain reading also existed due to asymmetric fiber distribution located in the top and
bottom flange laminates. Nagaraj and GangaRao (1997) concluded that the approximate
theoretical expressions provided by Jones (1975) could be used to determine the flexural
and shear rigidities in a pultruded GFRP beam.
2.10 Multi-Bolted Joints for GFRP Structural Members (Hassan et al. 1997)
An experimental investigation performed by Hassan et al. (1997) at the University of
Manitoba studied the behavior of multi-bolted connections using glass fiber-reinforced
plastic materials. The study consisted of testing a total of 105 multi-bolted double lap
shear connections with various parameters that included the width of the structural
member, edge distance, number of bolts, bolt pattern, pitch, thickness of the member, and
direction of the fibers with respect to the applied load direction.
The orthotropic composite material used for the tests were EXTREN Flat Sheet Series
500, pultruded glass fiber sheet with alternating stacked layers of E-glass rovings and E-
glass continuous strand mat.
16


2.10.1 Test Parameters
The influence of the placement and number of bolts was investigated by using 5 members
with different connection configurations and were designated as Joints A, B, C, D, and E,
see Figure 2.13. Various ratios of edge distance to hole diameter (e/d) and side distance
to the pitch (s/p) were selected to obtain different failure modes with each bolt patterns.
The 105 total specimens tested were 12.7 mm (1/2 in.) thick and had principal
unidirectional fibers layers orientated at 0, 45, and 90 with respect to the applied load.
High-strength structural bolts with a diameter of 19 mm (3/4 in.) and a bolt hole diameter
of 20.6 mm (13/16 in.) providing adequate clearance was used for all connections. The
bolts were tightened to a constant torque of 32.5 Nm (24 ft-lb).
2.10.2 Test Setup and Instrumentation
The specimens were tested using a 1,000 kN (220,000 lb) MTS closed-loop servo-
controlled loading system. The members were loaded axially in tension at a constant
stroke rate of 0.001 mm/s (0.0254 in./s). Concentric loading was achieved by a using
double-shear configuration thus removing any bending effects. The relative
displacements were measured using a linearly variable differential transducer (LVDT)
and strain gauges were used to determine the stress distribution in the member along with
load distribution of each bolt.
2.10.3 Experimental Results
The results indicated that the ultimate load was within 10% of the identical specimens
tested and the main factors effecting strength were the width, edge distance, and fiber
17


orientation with respect to the applied load. Similarly, these factors have a significant
effect on the failure mode of the specimens. It was determined that connection type B and
D which had smaller edge distances of 38.1 mm (1.5 in.) resulted in cleavage failure and
was characterized by crack propagation parallel to the applied load. For connection type
E with the same edge distance the section experienced net tension failure located at the
inner bolt row without any crack propagation between bolt rows. For the specimens with
larger edge distances ranging from 63.5 mm to 152.4 mm (2.5 in. to 6 in.) the resulting
failure mode was net tension failure located at the inner row of bolts, which had the
highest stress concentrations measured. All connections and tests showed bearing damage
after failure around the area adjacent to the loaded section of the hole.
When studying the load distribution among the bolts it was determined the specimens
with only one row of bolts (type B and type D) generally experienced equal distribution
of the load. However, this was not the case for connection type A, C, and E which had
unequal forces at each bolt.
Typically, for mechanically fastened joints with in-plane loading, bearing failure is likely
to occur as the width to diameter ratio (w/d) increases. This was not the case for this
experimental investigation using composite fiber members as bearing failure did not
occur; only localized bearing damage was apparent. Large widths reduced the ultimate
net tensile capacity and were relatively low when compared to specimens with smaller
widths; this can be attributed to the high stress concentrations for specimens with larger
widths. It is concluded that the load of a bolted joint is resisted by the material around the
18


vicinity of the bolt hole and increasing the width of the specimen has no benefits in terms
of strength efficiency.
2.10.4 Conclusions
After the experimental investigation performed by Hassan et al. (1997) the following
concluding remarks can be made. All connections and specimens tested experienced
linear behavior up to failure. The connections with only one row of bolts had equal load
distribution whereas, connections with more than one row experienced unequal load
distribution. The edge distance to diameter ratio, e/d, has a significant impact on the
failure mode for connections with either one row or one column of bolts. For small edge
distances with e/d ratios less than 3 cleavage failure occurred. Large edge distances with
e/d ratios larger than 3 experienced tension failure. All connections tested with 0, 45,
and 90 fiber orientation had higher bearing strengths as the e/d ratio increased up to a
ratio of 5. The ultimate capacities and bearing strengths for all connection types increased
as the side distance to pitch s/p ratio increased up to a value of 1.2. Having a higher
number of bolts is not directly proportional to the increase in ultimate strength capacity in
the member.
2.11 Finite Element Modeling of Damage Accumulation (Kermanidis et al. 2000)
A study was performed by Kermanidis et al. (2000) to model the effects of damage
accumulation using ANSYS, a finite element analysis program. A three-dimensional
model was developed to simulate the progressive damage and predict residual strength
and stiffness in single-lap bolted composite joints under in-plane tensile loading.
19


2.11.1 Progressive Damage Model
The progressive damage model involves an iterative procedure to determine stress
analysis, failure analysis, and material degradation. ANSYS has the ability to calculate
the stresses between each ply to be used as the model progresses. Due to the complex
nature of failure mechanisms in bolted composite laminates, many previous empirical
methods are used as failure criteria. The following seven expressions (Shokrieh et al.
1996) below represent stress-based criteria to predict different failure modes. The modes
of failure consist of matrix tensile and compressive failure, fiber tensile and compressive
failure, fiber-matrix shear-out and delamination of fibers in tension and compression.
Matrix tensile failure for (o y>0):
(2-4)
Matrix compressive failure for (o y<0):
(2-5)
Fiber tensile failure for (ox > 0):
(2-6)
Fiber compressive failure (ox < 0):
(2-7)
Fiber-matrix shear-out for (ox < 0):
20


2
(2-8)
Delamination in tension for (oz > 0):
2
> 1
(2-9)
Delamination in compression for (o z < 0):
2
> 1
(2-10)
Where Oy are the layer-stress components in the ij direction and the denominators are the
strengths in the corresponding directions.
Once ply failure occurs, the material properties are disabled from carrying a specific load,
this is known as the degradation rule. Therefore, each failure mode has its own
corresponding degradation rule. The ply-by-ply degradation rules are based on
assumptions and empirical methods from constraints in the composite material properties.
The material property degradation rules for failure analysis in this study were taken from
Shokrieh et al. (1996) and were believed by Kermanidis et al. (2000) to be best suited for
the failure criteria used.
When matrix tensile and compressive failure is observed in a ply, the assumption is that
the matrix cannot sustain any load. Therefore, the material properties of the failed ply are
reduced to:
Ey = Vxy = 0
21


(2-11)
When fiber tensile and compressive failure is observed in a ply, the assumption is that the
material cannot sustain any load in the vicinity where the failure has occurred. Therefore,
the material properties of the failed play are reduced to:
Ex = Ey = Ez = GXy GyZ GXz xy Vyz Fvr 0
(2-12)
When fiber-matrix shear-out failure is observed in a ply, the assumption is that the
material can only sustain load in the fiber and transverse to fiber directions. The material
does not have the ability to carry shear load, therefore, the material properties of the
failed ply are reduced to:
G
xy

(2-13)
Delamination failure in tension and compression affect the properties in the z-direction in
the delaminated region. This results in the material losing the ability to sustain any load
in the z-direction in addition to shear loads. Therefore, the material properties of the
failed ply are reduced to:
Ez Gyz Gxz Vyz ^XZ 0
(2-14)
A progressive model is developed in ANSYS using a programmed macro-routine
iterative process. The macro-routine is described in the flowchart Figure 2.14 and
involves the following steps.
1. Developing a FEM model of the composite joint with initial material properties,
specimen geometry, boundary conditions, initial load and load step.
22


2. Determine the stresses for each ply by performing a non-linear analysis.
3. Perform a failure analysis by implementing the failure criteria.
4. Determine if ply failure exist. If no failure is calculated the applied load is
increased. If any failure mode is recognized the program continues to the next
step.
5. Apply appropriate material property degradation rules on the failed ply.
6. Determine if final failure is reached. If so, the program is completed. If final
failure has not been reached the program returns to step 2 and the analysis is
performed again with the same load to compute the redistributed stresses. Final
failure is reached when the program converges. Convergence is assumed when no
additional failures are detected.
7. Repeat procedure until final failure occurs.
2.11.2 Finite Element Modeling
The study consisted of developing a finite element model of the bolted single-lap joint
using ANSYS. The geometry of the member is depicted in Figure 2.15. The model
consists of a composite laminate upper plate with fibrous unidirectional layers
(0/90/45) and an aluminum lower plate connected with a bolt. The model utilizes 8-
noded SOLID46 3-D ANSYS layered elements with three displacement DOFs per node.
The solid element used for the composite material is defined by the orthotropic material
properties, the fiber orientation, and layer thickness. For the interaction between two
surfaces, 3-D CONTAC49 ANSYS elements were used to model the contact between
two plates, between the bolt and the surface of the hole, and between the bolt head and
23


the plate. See Figure 2.16 and Figure 2.17 for the finite element model. The loading
conditions for the models consisted of in-plane tensile loading at the ends as well as a
pre-tension load at the fastener by applying thermal expansion properties in the axial
direction of the bolt. To prevent secondary bending the model was supported laterally in
the z-direction.
2.11.3 Results and Discussion
The strains were measured experimentally with an 8 kN (1,798 lbs) tensile load applied
to the lap joint. The strains were determined for the global x-direction at the angles of 0,
22.4 and 45 and are displayed in Figure 2.18. Comparisons were made to experimental
and numerical values from a study performed by Ireman (1998). The experimental
program consisted of installing strain gauges around the hole at angles of 0, 22.4 and 45
to measure the response. The comparison of the numerical to experimental model
resulted in sufficient accuracy. The material properties and strengths used for the
composite material are shown in Table 2.3 and Table 2.4.
The progressive damage model was subjected to incrementally increasing tensile loading
in which the first failure (matrix compressive failure) occurring at a load of 3 kN (674
lbs) near the stress concentrations around the hole. Before the first ply failure the
composite material behaved elastically. The damage occurs at an angle of 45 with
respect to loading and continues to prorogate due to the pressure from the bolt. Figure
2.19 and Figure 2.20 illustrates the progressive damage at different load steps predicted
by the model. The shaded elements in the model represent fiber breakage, which
24


represents the most critical failure mode in the model. Ultimate failure is reached when
the composite laminate is unable to carry anymore load and is represented in Figure
2.19(c). The ultimate failure load in the model was determined to be 20.2 kN (4,541 lbs)
25


Table 2.1. Comparison of flexural rigidity (Nagaraj and GangaRao 1997)
Ei (1 5 < 1010) (N *mma)
Section 102 x 102 x 6 mm Box 6.89 7.8 ll 7.03 2.0
102 X 102 X 6 mm Wide flange 7.29 8.37 13 7.57 3.8
152 X 152 X 6 mm Wide flange (unidi- rectional) 27.0 26.7 1.1 27.0 0
152 X 152 x 6 mm Wide flange (bidi- rectional) 25.7 25.5 0 25.8 0
Table 2.2. Comparison of shear rigidity (Nagaraj and GangaRao 1997)
Shear Rigidity 1 x 10* N
Section (1) From theory (2) From exper- iment (3) Percent dif- ference (4)
102 X 102 X 6 min Box 102 X 102 X 6 mm Wide 2.9 3.06 5.3
flange 152 x 152 X 6 mm Wide 1.63 1.68 2.9
flange (unidirectional) 152 x 152 X 6 mm Wide 2.87 2.77 3.6
flange (bidirectional) 3,15 3.12 1.0
26


Table 2.3. Geometrical Data for the Investigated Geometry (Kermanidis et al. 2000)
L W D d e h t
(mm) (nan) (ram) (nun) (nan) (mm) (mm)
240 60 16 10 30 30 4.16
Table 2.4. Elastic Properties of HTA/6376 Material (Kermanidis et al. 2000)
£x (GPa) Ey (GPa) Ez (GPa) Git (GPa) Gyz (GPa) G1Z (GPa) !' AT V.XZ (GPa) (GPa) VYZ (GPa)
145 10.3 12.1 5.3 5.275 3.95 0.301 0.5 0.495
27


MAT CREELS
SURFACING
MATERIAL
Figure 2.1. FRP pultrusion process (Extren 2003)
Figure 2.2. Modes of failure for bolted joints in
FRP Composites (Hart-Smith 1987)
28


t t t 01 t
Figure 2.3. Stress concentration relief in fibrous composites by delamination
(Hart-Smith 1977)
STRESS CONCENTRATION
FACTOR ktc
ON NET SECTION
OF FIBROUS COMPOSITE
LAMINATE AT FAILURE
UNLOADED
HOLE TESTS

'S TENSION FAILURES^!#
------------* I
'BEARING FAILURES
^'(APPARENTLY PREMATURELY
WITH RESPECT TO I
TENSILE gTRENCTHS)

ZV-i&i-'-b- COMPLETE
STRESS CONCENTRATION
-TYPICAL TEST RESULTS*; RELIEF ----
i
APPROXIMATE
w/d VALUES
OPEN HOLES 2 4 610 THEORETICAL ELASTIC STRESS CONCENTRATION FACTOR. A,e
LOADED HOLES 2 3 4 5 6 7 8
Figure 2.4. Relation between stress concentration factors observed at failure of fibrous
composite laminates predicted for perfectly elastic isotropic materials (Hart-Smith 1977)
29


Figure 2.5. Terminology (Duthinh 2000)
d;8-35jnm (torque 3*4 Nm)
4d
Figure 2.6. Influence of CFRP fiber proportion (0 / 45) on failure mode (Garbo and
Ogonowski 1981)
30


200
BEARING STRENGTH
IKSI)
100
PIN-LOADED HOLE 0 W(-H DIAMETER BOLTS
fINGER TIGHT
0
NORMAi TORQUES NORMAL METAL
FOR COMPOSITES BOLT TORQUES
BOLT TORQUE
Figure 2.7. Effect of bolt torque on bearing strength of fibrous composite laminates
(Hart-Smith 1987)
140
120
100
e
s
8
8
00
bearing strength
40
$
8
20
0
BIN CONNECTION
T3D0/NS2CS CARBON EPOXY LAMINATES
quasi isotropic laminate pattern
BOLT DIAMETER 0.15 INCH
TESTS CONDUCTED
AT ROOM TEMPERATURE
protruding-head bolt
Figure 2.8. Comparison between bearing strengths of connections with and without
clamping pressure (Hart-Smith 1987)
31


d* 6-35 mm
Figure 2.9. Effect of grouped 0 plies on bearing strength (Garbo and Ogonowski 1981)
Figure 2.10. Effects of stacking sequence on bearing strength for GFRP laminates
(Quinn and Mathews 1977)
32


100Q-)
X AS/914 vf0-6
d 6- 35 mm
Torque 3-4 Nm
1 45
0
o
T
W fmm >
n
40
Figure 2.11. Variation of net tensile strength with width of 0/45 CFRP composites
with 6.35 mm (0.25 in) hole (Collins 1987)
152mm
LEGEND
Strain Gauge
£)t>i
Dial Gaige
n
(c) WF 152il52x6mm (uni & bidirectional)
Figure 2.12. Cross sections of test specimens (Nagaraj and GangaRao 1997)
33


i +pt i_ i > i 1 1 r4 1 4 li4
.... P ii-i-ih j*"*dr*** * "7tV" f
1 h i i 1 1 t
Joint A Joint B Joint C
I > 1
-T--I----

I I i
I t t
n i*rTf
t
i i
i i i
Joint D
r4
i *
i i
i-ih
LU*
i i
i
Joint E
Figure 2.13. Connection configuration (Hassan et al. 1997)
34


Marl
Figure 2.14. Flowchart of the Progressive Damage Model (Kermanidis et al. 2000)
35



L

T
Composite
Aluminium
]_______z______
-r

v ,D i 1
h
Vl (Zj) i

t
II
Figure 2.15. Geometry of the Bolted Single-Lap Joint (Kermanidis et al. 2000)
Figure 2.16. Finite Element Model of Bolted Joint (Kermanidis et al. 2000)
36


Figure 2.17. (a) Mesh of Area Around Hole (b) Finite Element Model of Protruding
Head Bolt (Kermanidis et al. 2000)
Figure 2.18. Comparison of Calculated Strains with Experimental and Numerical
Results (Kermanidis et al. 2000)
37


(c) 20.2kN
Figure 2.19. Illustration of Damage Propagation Predicted by the Present Model at
Different Load Steps. Upper Surface of the Laminate (Kermanidis et al. 2000)
\ \ N
(b) 12tN
(c) 20.2kN
Figure 2.20. Illustration of Damage Propagation Predicted by the Present Model at
Different Load Steps. Lower Surface of the Laminate (Kermanidis et al. 2000)
38


3. Experimental Program
3.1 Introduction
The experimental program examined the axial and transverse strength of GFRP structural
channel members with bolted connections using various geometric configurations. The
parameters studied included the effects of load cycles at various load levels on the
ultimate strength and stiffness. In addition, the effect different flange geometries have on
the ultimate strength when loaded in the axial and transverse direction. Phase I consisted
of testing a total of 60 specimens with monotonic loading applied in the axial and
transverse directions to determine the ultimate strengths of each member (30 axial and 30
transverse). In addition, Phase I tested 10 thickened GFRP members 1 in. (25 mm) thick.
Phase II consisted of testing specimens to failure after each specimen underwent
unidirectional cyclic loading at various load levels. Phase III tested each member to
failure after the specimen underwent multidirectional load cycles in both the axial and
transverse directions at different load levels. This chapter will discuss the fabrication
process and materials used for the test specimens, instrumentation, test setup, and
experimental procedure. All material testing was performed in the Structural Engineering
Laboratory at the University of Colorado Denver.
3.2 Materials used for Test Specimens
3.2.1 GFRP Channels
Extren GFRP structural shapes produced by Strongwell (Bristol, Virginia, USA) were
used for the experimental testing. Extren is a product line manufactured by Strongwell
which consists of more than 100 fiberglass structural shapes. The shapes are made up of
39


long glass fibers intertwined and bound with resin to form a mat. In addition, the glass
reinforcements consist of continuous strand rovings containing 800 4,000 fiber
filaments in each strand. The resins used in Extren are isophthalic polyester and vinyl
ester, which provide corrosion resistance and high strength properties. The mechanical
properties of the GFRP members were determined through tensile testing of GFRP
coupons and using the data provided by the manufacturer.
Channel structural shapes were selected for testing with dimensions of 3 V2 x 1 V2 x
3/16 (90 mm x 38 mm x 5 mm) and l-2 (356 mm) in length. Each channel member
had a bolt hole with a diameter of 9/16 (14 mm) for loading purposes. The channel
members consisted of three different geometric configurations with a portion of the
flange cut near the bolt hole. Condition #1 involved the whole section with no flange
modifications. For Condition #2 and #3 both the top and bottom flange were cut at a
length of 1 5/8 (41 mm) and 3 1/8 (79 mm), respectively. See Figure 3.1 Figure 3.3
for member dimensions with various flange configurations.
3.2.2 GFRP Flat Plates
The GFRP material testing also included thickened flat plate members that were loaded
axially to failure. The GFRP flat plates were V2 in. (13 mm) thick with an additional V2 in.
thick plate epoxied at the end where loading occurs. The flat plates had 1 in. chamfers at
the corners of both plates. Similar to the channels, a hole with a diameter of 1 9/16 (40
mm) was drilled thru both plates to install a bolt to be loaded. The plates are 3 in. (76
40


mm) wide and the flat plate that is epoxied has a length of 4 in. (102 mm) See Figure 3.6
for dimensions of the GFRP flat plate.
3.3 Experimental Setup and Loading
3.3.1 Phase I and Phase II Setup for Axial Tests
The tensile testing for each GFRP members consisted of applying a point load through
the use of a steel bolt. An MTS closed-loop electro-hydraulic universal testing machine
was used for the monotonic axial load application. The loading was performed at a rate of
1 mm/min (0.0394 in./min) and was controlled by the displacement of the actuator. The
restraints of the member were provided by fixtures that were connected to the specimen
with three V2 (13 mm) diameter bolts. Load was applied at a distance of 1 V2 (38.1 mm)
from the end of the specimen using a V2 diameter steel bolt attached to a fixture. The
tensile testing was divided into three phases. Phase I consisted of monotonic axial loading
to ultimate failure of each channel specimen. The results of Phase I were used to establish
various load levels by taking 25%, 50%, and 75% of the ultimate failure to be used for
Phase II testing. Phase II testing consisted of axial and transverse loading to failure after
the specimens underwent 10 load cycles applied to the specimen using different load
levels mentioned above in the axial and transverse directions. Phase III consisted of
loading to failure after the specimens were subject to multidirectional cyclic loading. See
Figure 3.6 for the axial load experimental testing setup.
41


3.3.2 Phase I and Phase II Setup for Transverse Tests
The shear tests for each GFRP members were tested in a similar fashion by applying a
point load through use of a steel bolt in the transverse direction. The same testing
equipment and loading rate of 1 mm/min from the tensile testing was used. Instead of
utilizing fixtures with bolted connections, a fixed support was provided by gripping a
portion of the channel web with clamps. The restraint required cutting the bottom channel
flange to allow for installation of the clamp. The same fixture from the tensile testing was
used which applied a transverse concentrated load through the V2 (13 mm) steel bolt.
Similar to the axial test the transverse testing was divided into Phase I, Phase II, and
Phase III, using the same load levels based on the ultimate failure to be used for the
cyclic loading. See Figure 3.7 for the transverse load experimental testing setup.
3.3.3 Phase III Testing Procedure
The channel specimens in Phase III testing are subject to multidirectional loading, which
requires the specimen to be loaded in the axial direction at a certain load level followed
by loading in the transverse direction at another load level. This procedure constitutes one
load cycle and is performed for a total 10 cycles. Phase III requires the fixtures to be
alternated for each load cycle to accommodate for loading in the axial and transverse
direction. This procedure is different from Phase II, which allows for consecutive
unidirectional cyclic loading in one test for all 10 cycles.
42


BOLT
Figure 3.1. Condition #1
BOLT
Figure 3.2. Condition #2
43


1 3/4
Figure 3.3. Condition #3
2-FRP CHANNELS
(3" CLEAR SPACING)
TENSION TEST
Figure 3.4. Tension Test
44


2-FRP CHANNELS
(3" CLEAR SPACING)
1/2 BOLT
CONNECTION
SHEAR TEST
Figure 3.5. Shear Test
CM C'J
Figure 3.6. Geometry of Thickened Flat Plate
45


Figure 3.7. Axial Test Setup
Figure 3.8. Transverse Test Setup
46


Figure 3.9. Thickened Flat Plate Test Setup
Figure 3.10. Thickened Flat Plate Test Setup
47


4.
Finite Element Modeling
4.1 Introduction
A three-dimensional finite element model was developed to predict the behavior of the
GFRP channel specimens using the structural analysis software ANSYS. The specimens
were modeled using structural solid elements with orthotropic material properties. The
model is a non-linear stress analysis incorporating an established failure criterion to
determine stress, deflection, and predicted failure. ANSYS uses two general modes for
modeling, interactive and batch. The interactive mode allows the user to use the GUI and
other various tools to develop the model in the graphics window and allows for any
modifications related to the analysis whereas, the batch mode uses command files that
have been created or previously generated for the analysis. The model created utilizes the
interactive mode. This chapter will discuss the development of the model along with the
approach and methods used.
4.2 Modeling and Elements Preprocessor
Building a model for the channel composite specimen consists of several steps. An eight-
node brick solid elements (SOLID 185) was selected to represent the GFRP channel
member. This solid element has three translational degrees of freedom at each node.
Orthotropic material properties are required and are defined using values from the
manufacturer. The material properties include the modulus of elasticity E, in the x, y, and
z directions, Poissons ratio v, in the xy, yz, xz directions, and the shear modulus G, in
the xy, yz, xz directions, they are listed in Table 4.1. The channel geometry modeled has
a depth of 3 U in. (89 mm) a flange width of 1 U in. (38 mm) and a thickness of 3/16 in
48


(5 mm), the dimensions are shown in Figure 3.1. A V2 in. (13 mm) diameter hole is added
at a distance of 1 V2 in. (38 mm) from the end of the member.
Several types of constraints were applied to the model. The edge of the channel member
was constrained in all six degrees of freedom. To prevent lateral movement and
secondary bending effects of the member, a lateral support was added in the z-direction
located at the bolt. The Mesh Tool was used for automatic mesh generation of the model
which is beneficial for the ease of mesh refinement. Both the channel and bolt were
meshed using tetrahedral solid elements. The load was applied in the longitudinal x-
direction using 25% and 75% of the ultimate tensile failure load determined from the
experimental testing. The corresponding load values are 584 lbs (2.6 kN) and 1,752 lbs
(7.8 kN) respectively. Concentrated loads were applied over multiple nodes to replicate
uniform pressure caused by loading of the bolt, see Figure 4.2.
4.3 Failure Criteria
The Tsai-Wu failure criterion used to predict the composite failure of the GFRP channel
members. The Tsai-Wu failure criterion is a quadratic, interactive stress-based criterion
that identifies failure, but does not distinguish between different failure modes (Tsai and
Wu 1971). Failure occurs when the following condition is satisfied:
FjO-j + FijOiOj < 1.0
(4-1)
Where Ft and Fy are experimentally determined material strength parameters and o, and oj
are the laminate stress in the fiber direction and the laminate stress transverse to the fiber
49


direction, respectively. The failure criterion requires input values for material strengths
which include the compressive, tensile, and shear strength of the composite in the x, y,
and z directions. See Table 4.2 for the material strength used in the analysis.
4.4 Model Results
After the analysis was performed the results indicate a maximum deflection of 0.004
inches (0.1 mm) at the bolt hole. This value was based on an axial load of 584 lbs (2.6
kN) which was determined to be 25% of the ultimate load in the axial direction. When
75% of the ultimate axial load was applied of 1,752 lbs (7.8 kN) resulted in a maximum
deflection of 0.012 inches (0.3 mm) at the same location adjacent to the bolt hole. These
deflection values were significantly different from the displacements measured through
testing. The finite element model yielded values considerably smaller. Based on load
testing of the channel specimens the average deflection from an applied axial load of 584
lbs and 1,752 lbs were approximately 0.025 inches (0.6 mm) and 0.0725 inches (1.8 mm)
respectively. The difference can be attributed by the restraint conditions of the two. The
ANSYS model uses a fixed restraint at the end of the channel whereas; the testing of the
channel specimens used three bolts for the support conditions. This resulted in high stress
concentrations at the bolt locations leading to a larger total displacement from each bolt
hole. In addition, there may have been a very small gap between the bolt and bolt hole
surface in the three bolted connections used as a restraint which caused additional
displacement. See Figure 4.2 and 4.4 for the deflection results from ANSYS.
50


Table 4.1. Material Properties for GFRP Channel Specimens (Extren 2013)
Ex Ey Ez Gv ^ xy Vxz v*
(psi) (psi) (psi) (psi) (psi) (psi)
2,600,000 800,000 800,000 425,000 425,000 n/a 0.33 0.33 n/a
Table 4.2. Failure Criteria Strength Values
X y z
Stress in Tension (psi) Stress in Compression (psi) 30.000 30.000 7.000 15.000 7.000 15.000
xy yz xz
Stress in Shear (psi) 4,500 4,500 4,500
51


Figure 4.1. Channel Geometry of ANSYS Model
Figure 4.2. Applied Loading of Model
52


Figure 4.3. Displacement from 25% of Ultimate Load

ANSYS
R14.5
21 2013
13;. 02:40
Figure 4.4. Tsai-Wu Strength Index 25% of Ultimate Load
53


5.
Test Results and Discussion
5.1 Introduction
This chapter presents the results obtained from material testing of the GFRP channel
specimens. The testing consisted of loading the specimens in the axial and transverse
direction. The behavior of the loaded specimens along with the various failure modes are
discussed in detail. The mechanical properties of the GFRP specimens are established
from the material testing and are evaluated in terms of the load-displacement response.
The effect of different geometric configurations where the flange was cut near the applied
load was studied. Stiffness degradation and ultimate strength from load cycles at different
load levels are evaluated. The experimental program is divided into three phases. The
primary objective of Phase I is to determine the ultimate strength of the GFRP specimens
through monotonic loading in the axial and transverse directions. For Phase II the
residual capacity of the connection is investigated with the specimens subjected to
unidirectional cyclic loading at various load levels. Phase III involves investigating the
residual capacity of the connection subjected to multidirectional cyclic loading at various
load levels. The test results from all three phases will be discussed in the following
sections.
5.2 Phase I Testing Results of GFRP Channels
A total of 30 GFRP specimens were used for axial testing to determine ultimate load
capacity and to generate load-displacement curves. For transverse testing, a total of 30
specimens were used as well. The GFRP specimens have various flange geometry as
discussed in Chapter 3. Table 5.1 displays the load capacity results for flange geometries
54


01, 02, and 03 which represent Condition #1, Condition #2, and Condition #3
respectively, in the axial and transverse directions. Ten specimens were tested for each
flange geometry with monotonic loading to failure in both directions. The load-
displacement curves for each test in Phase I is provided in Appendix A of this thesis. The
load-displacement response indicates linear behavior until initial failure, followed by
non-linear behavior until ultimate failure. Initial failure is represented by the first peak on
the load-displacement curve and the ultimate failure is represented by the maximum load
of the response. Figure 5.1 and Figure 5.2 displays a typical load-displacement curve for
the GFRP specimens in the axial and transverse directions.
For AXIAL 01 the average of the 10 specimens tested had an initial failure load capacity
and standard deviation of 2,214 lbs (9.8 kN) and 162 lbs (0.72 kN) respectively. TRANS
01 the average initial failure load capacity and standard deviation was 2,601 lbs (11.6 kN)
and 345 lbs (1.5 kN), respectively. For the subsequent testing, the load levels of 25%,
50%, and 75% for Phase II and Phase III were obtained using the initial failure load from
Phase I. AXIAL 01 and TRANS 01 was the full channel section with no flange cut
modifications made. AXIAL 02, AXIAL 03, TRANS 02, and TRANS 03 had different
flange geometries as noted in Chapter 3. Based on the final results there was no distinct
trend and it can be concluded that removing a portion of the flange in the channel section
does not impact the overall performance of the channel member, when loaded in the axial
and transverse directions. The total average of all the specimens tested in the axial and
transverse direction was 2,337 lbs (10.4 kN) and 2,588 lbs (11.5 kN) respectively.
Therefore, the load levels used for Phase II and Phase III testing were 584 lbs, 1,168 lbs,
55


and 1,752 lbs (2.6 kN, 5.2 kN, and 7.8 kN) for axial load levels of 25%, 50%, and 75%
and 647 lbs, 1,294 lbs, and 1941 lbs (2.9 kN, 5.8 kN, and 8.6 kN) for transverse loading.
5.2.1 Phase I Failure Modes of GFRP Channels
Investigation of the failure mode for the axial and transverse monotonic load test showed
different failure mechanisms for the axial and transverse tests. For the test with the
applied load in the axial direction initial bearing failure occurred at the contact zone
between the bolt and composite specimen followed by shear-out failure. As noted in
Chapter 2 bearing failure occurs when excessive compressive stresses develop at the hole
boundary surface. Shear-out failure is typically consequence of bearing failure with a
short edge distance, e and can be characterized as a combination of in-plane and
interlaminar shear failures. Initial bearing damage is followed by a series of peaks on the
load-displacement curve that represents damage accumulation with the eventual sudden
drop in load carrying capacity. The test is completed when the specimen reaches this
point.
In the transverse direction the specimen experienced initial bearing failure similar to the
axial test followed by net tension splitting failure. This was evident by the crack
propagation in the transverse direction of the applied load. The behavior of the specimens
loaded in the transverse direction differs from the axial tests; this is evident by the load-
deflection curve which indicates less progressive damage before a sudden drop in load
carrying capacity. The transverse load-displacement curves show less activity of damage
accumulation with a smaller region of varying peaks after initial failure. Net tension
56


failure is a function of both the joint geometry and the strength of the GFRP specimen.
See Figure 5.4 through Figure 5.6 for the axial and transverse typical failure mode of the
GFRP channel specimens.
5.2.2 Phase I Testing Results of GFRP Thickened Plates
A total of 10 GFRP plate thickened members were tested with monotonic loading to
failure in the axial direction. The average failure load was 14,824 lbs (66 kN) with a
standard deviation of 1,308 lbs (5.8 kN) As expected the failure load was significantly
higher than the channel members due to the increased thickness of 1 inch (25 mm) which
allows for more area at the contact zone between the bolt and the specimen. Table 5.2
displays the failure load for each thickened specimen with loading in the axial direction.
The load-displacement curves indicate a linear response until ultimate failure. In
comparison to the channel specimen load-displacement curves in the axial direction, the
thickened plates do not experience progressive failure which is evident by one peak on
the curve as opposed to numerous peaks for the channel specimens. Figure 5.3 shows a
typical load-displacement curve of the thickened GFRP plates.
5.2.3 Phase I Failure Mode of GFRP Thickened Plates
Similar to the GFRP channel members the thickened plates experienced shear-out failure
adjacent to the bolt hole. Unlike the channel members which experienced initial bearing
failure around the bolt hole, the thickened specimens had no visible signs of bearing
failure. This can be attributed to the fact the bearing strength capacity is larger than the
shear-out capacity for the thickened specimens. On three of the thickened specimens
57


tested debonding of the plates occurred followed immediately by shear-out failure due to
the reduced cross section caused by the plates debonding. See Figure 5.7 for typical
failure mode for the GFRP thickened members with monotonic loading in the axial
direction and Figure 5.8 for debonding of the plates.
5.3 Phase II Testing Results of GFRP Channels
A total of eighteen specimens were tested for Phase II, nine specimens in the axial
direction and nine specimens in the transverse direction. Each load level test (25%, 50%,
and 75%) consisted of the same flange geometries used in Phase I, represented by 01, 02,
and 03 in the axial and transverse direction. Each specimen underwent ten unidirectional
load cycles in the axial and transverse direction of the appropriate load level before being
loaded to failure in the respective direction to determine the residual capacity of the
connection. For the cyclic tests, the specimens were loaded to a specific predefined load
than unloaded when that load was reached. This procedure was performed consecutively
ten times.
Figure 5.9 displays the load-displacement curves for the unidirectional cyclic load
applied for Phase II. Based on the cyclic loading curves it appears that there is no
significant stiffness degradation as the slope of the curves is relatively constant, any
decrease in stiffness is minor. The load-displacement response was similar to Phase I
with apparent progressive damage and linear behavior up until initial failure after the load
cycles were completed and the specimens were loaded to failure. The load-displacement
curves for the transverse loading to failure closely resembled Phase I testing as well. See
58


Figure 5.10 and Figure 5.11 for typical load-displacement curves with monotonic loading
in the axial and transverse direction for Phase II.
The average residual load capacity of the three specimens that experienced axial cyclic
loading of 25%, 50%, and 75% of ultimate load were 3,000 lbs, 2,836 lbs, and 2,537 lbs,
(13.3 kN, 12.6 kN, and 11.3 kN) respectively. Based on these values we can see a distinct
trend where the ultimate failure load is reduced as a higher load is applied for cyclic
loading. This can be attributed to minor stiffness and strength degradation in the
composite material due to repeated load effects. It is noted that the values above, the
maximum load capacity was obtained from the load-displacement curves not the initial
failure represented by the first peak on the curve.
Similarly, the average residual load capacity of the three specimens that experienced
transverse cyclic loading of 25%, 50%, and 75% of the ultimate load were 3,120 lbs,
2,347 lbs, and 2,521 lbs, (13.9 kN, 10.4 kN, and 11.2 kN) respectively. The effect of
cyclic loading in this case does not have a clear trend of decreasing strength with
increasing load during the ten cycles. Although, it is clear that the failure load of the 75%
specimens decreased in comparison to the 25% specimens. Table 5.3 summarizes the
testing results for the channel specimens subject to cyclic loading.
5.3.1 Phase II Failure Modes of GFRP Channels
The failure modes for the GFRP channel specimens were similar to the failure modes
experienced in Phase I. The only difference came from minor damage propagation and
59


displacements caused by the load cycles which are shown in Figure 5.12. When the
specimens were tested in the axial direction to determine the residual capacity of the
connection, they experienced initial bearing damage at the contact zone followed by
shear-out failure in the region adjacent to the bolt, similar to Phase I failure. Likewise, the
transverse testing resulted in initial bearing damage around the bolt hole followed by
tension splitting of the section.
5.4 Phase III Testing Results of GFRP Channels
Phase III consisted of multidirectional cyclic load testing of nine specimens. Three
specimens were used for each load level of 25%, 50%, and 75% of the ultimate load
determined in Phase I. All nine specimens had the same geometric configuration, see
Figure 5.13. On one side of the flange no modifications were made, however, the
opposite side required cutting the flange to fit the clamp fixture necessary for the
transverse loading test. As described in Section 3.3.3 the multidirectional cyclic loading
required alternating fixtures to load in both the axial and transverse direction as one load
cycle. After ten cycles of multidirectional loading the specimens were loaded to failure.
Two of the three sets (01-25, 01-50, 01-75, 02-25, 02-50, and 02-75) were loaded axially
to failure to determine the residual capacity in the connection. The last set (03-25, 03-50,
and 03-75) was loaded to failure in the transverse direction to determine the residual
connection capacity.
The load-displacement curves for the multidirectional cyclic loading are displayed in
Figure 5.14. Cyclic loading in the axial or transverse direction is not performed
60


consecutively as the curves show; the axial cyclic loading and transverse cyclic loading
are alternated. Thus, each specimen is subject to a total of twenty cycles of applied
loading at the appropriate level rather than a total of ten cycles in Phase II. As a result
there is more damage accumulation and stiffness degradation as seen in Figure 5.14;
however, it is very minor. The load-displacement curves in the axial and transverse
directions to failure after multidirectional load cycles resemble the curves generated in
from Phase I and Phase II. Recall from Phase I and Phase II, linear behavior up until
initial failure from the first peak followed by damage accumulation resulting in ultimate
failure.
The average residual load capacity in the axial direction of the two specimens that
experienced multidirectional cyclic loading of 25%, 50%, and 75% of the ultimate load
were 2,826 lbs, 2,893 lbs, and 2,386 lbs (12.57 kN, 12.87 kN, and 10.61 kN) respectively.
Based on these values it is apparent that there was a significant reduction in strength from
the specimens which were subject to 75% multidirectional loading. The values above
were determined using the maximum load from the load-displacement curves which is
not necessarily the initial failure represented by the first peak. However, the average
residual connection capacity of the two specimens based on initial damage that
experienced multidirectional cyclic loading of 25%, 50%, and 75% of the ultimate load
were 2,626 lbs, 2,321 lbs, and 1,894 lbs (11.68 kN, 10.32, and 8.42 kN) respectively.
This shows significant strength degradation with each increase in load level.
61


For the residual connection capacity in the transverse direction after multidirectional
cyclic loading only one set of specimens were tested. The connection capacity with cyclic
loading of 25%, 50%, and 75% of the ultimate load was 2,168 lbs, 2,595 lbs, and 2,231
lbs (9.64 kN, 11.54 kN, and 9.92 kN) respectively. The effect of cyclic loading in this
case does not have a clear trend of decreasing strength with increasing load during the
multidirectional load cycles. The same is observed using the residual capacity based on
initial failure. Table 5.4 summarizes the testing results for the channel specimens subject
to multidirectional cyclic loading.
5.4.1 Phase III Failure Modes of GFRP Channels
For Phase III, the failure modes from monotonic axial and transverse testing after
multidirectional load cycles were similar to Phase I and Phase II. The only difference
came from displacements and damage caused from the repeated loads. For the 25% and
50% multidirectional load cycle tests the damage and displacements was minimal,
however, for the 75% multidirectional load cycle tests the damage was more as shown in
Figure 5.17. For failure in the axial direction the specimen experienced initial bearing
failure followed by shear-out failure adjacent to the bolt hole. In the transverse direction,
two of the three specimens experienced initial bearing failure followed by tension
splitting at the bolt hole as expected based on the previous test, displayed in Figure 5.18
and Figure 5.19. However, for specimen 50-3 the tension splitting was located just above
the clamp and below the hole seen in Figure 5.20 which was unusual based on the
previous failure modes from testing in the transverse direction. Recall, once the test
shows a sudden drop in load carrying capacity the test is completed.
62


5.5 Comparison of Phases
The following section will compare the behavior and results between each phase of the
testing. Strength reduction, stiffness degradation, failure mechanisms, etc. will be
discussed.
5.5.1 Comparison Among Different Flange Geometries in Phase I
Figure 5.21 show the comparison of load-displacement curves for specimens tested in
Phase I. The comparison shows the effect different flange geometries have on the
connection strength. Based on the load values and observation of the curve it can be
concluded that variation in flange geometries has no significant effect on the behavior
and ultimate strength of the connection in the axial and transverse directions. However, it
is noted that for the axial specimens the initial failure load capacity increased with larger
sections cut from the flange of the channel.
5.5.2 Comparison of Post-Cyclic Residual Behavior
The comparison of post-cyclic residual behavior is displayed in Figure 5.22. It is clear
from the data and curves that multidirectional cyclic loading has a substantially higher
impact on the residual behavior and strength of the connection compared to the residual
behavior subject to unidirectional cyclic loading. In some instances the load capacity for
specimens that underwent unidirectional cyclic loading was higher than the load capacity
determined in Phase I which did not experience any cyclic loading. The curves show that
the post-cyclic residual behavior from multidirectional cyclic loading generally had the
smallest slope. In addition, the transverse specimens from Phase III had a longer region
63


of damage accumulation with multiple peaks before failure compared to the transverse
specimens from Phase I and Phase II.
5.5.3 Effect of Cyclic Loading on Residual Capacity
The effect of cyclic loading on the residual capacity for Phase II and Phase III is shown
in Figure 5.23. The figure is divided into the capacity at the first peak and the capacity at
the maximum peak. Generally, the residual capacities for Phase III experienced reduced
connection capacity as the load level percent of cyclic loading increased. The same
cannot be said for Phase II as there was no clear trend apparent that the cyclic loading
had on the residual capacity. It is noted, that for the transverse specimens in Phase II the
trend was similar when comparing the capacity at the first peak to the capacity at the
maximum peak. In addition, there was very small variation in the capacity between the
three specimens tested for each load level resulting in similar chart lines for each of the
three specimens.
5.6 Design Recommendations
This section describes discusses design recommendations for pultruded GFRP structural
channel members with bolted connections subject to axial and transverse loading.
5.6.1 Monte-Carlo Simulation
A Monte-Carlo simulation was performed to assist with developing design guidelines and
recommendations for GFRP structural channel members. The simulation involves
generating many random values to overcome limited experimental data. The algorithm
64


used for the random sampling was the inverse transformation method (Devroye 1986).
For the simulation 10,000 samples were used for the calibration of the resistance factor.
5.6.2 Calibration of Resistance Factor
The resistance factor for bolted GFRP channel members was determined using the
following calibration method. The safety index fi is used to adjust the level of safety
performance in structural members and is calculated by the following equation:
Where R and E are the mean capacity and mean instantaneous load applied, respectively,
and errand 07. are the corresponding standard deviations. The target safety indexfi was set
to 2.5, 3.0, and 3.5. The denominator of Equation 5-1 can be approximated as the
following equation:
Where a is the directional cosine or empirical constant to be determined. Combining
equation 5-1 and 5-2 the following Load and Resistance Factor Design (LRFD)
relationship is determined (Barker and Puckett 1997).
R E
(5-1)
(5-2)
(pRn = yEn
(5-3)
= 1R(1 a(3VR)
(5-4)
65


Y = AjjC 1 a(3VE)
(5-5)
Where /(, and En are the nominal resistance and applied load, respectively, XR and /./, are
their bias factors, Vr and Ve are the coefficients of variations, and cp and y are the
resistance factor and load factor, respectively.
Table 5.5 summarizes the resistance factor calibration for specimens in monotonic
unidirectional loading. As expected as the safety index fi decreases the resistance factor (p
increases. The resistance factor calibration indicates the transverse specimens yielded
lower resistance factor value than the axial specimens, which is caused by higher
coefficients of variation in the transverse specimens. The resistance factor calibration is
divided into the first peak and maximum peak of the load-displacement curves. For
design purposes resistance values determined from the first peak should be used as they
are typically more conservative.
66


Table 5.1. Phase I Ultimate Failure Load of GFRP Channel Specimens with Monotonic
Loading in the Axial and Transverse Directions.___________________________________________
Speciman Capacity (lbs) Max Capacity (lbs) Speciman Capacity (lbs) Max Capacity (lbs) Speciman Capacity (lbs) Max Capacity (lbs)
AXIAL-01 -1 2090 2493 AXIAL-02-1 2118 2493 AXIAL-03-1 2103 2711
AXIAL-01 -2 2206 2789 AXIAL-02-2 2167 2228 AXIAL-03-2 2357 2839
AXIAL-01-3 2080 2520 AXIAL-02-3 2473 3063 AXIAL-03-3 2296 2920
AXIAL-01-4 2175 2650 AXIAL-02-4 2458 2946 AXIAL-03-4 2419 2575
AXIAL-01-5 1996 2556 AXIAL-02-5 2302 3273 AXIAL-03-5 2057 2650
AXIAL-01-6 2465 2571 AXIAL-02-6 2286 2524 AXIAL-03-6 3136 3140
AXIAL-01-7 2357 2465 AXIAL-02-7 2048 2535 AXIAL-03-7 2680 2960
AXIAL-01-8 2109 2508 AXIAL-02-8 2501 2763 AXIAL-03-8 2200 2403
AXIAL-01-9 2214 2381 AXIAL-02-9 2476 2823 AXIAL-03-9 2669 2714
AXIAL-01-10 2455 2823 AXIAL-02-10 2637 3121 AXIAL-03-10 2569 2602
Average 2215 2576 2347 2777 2449 2751
Std Dev 162 140 192 330 326 216

Speciman Capacity (lbs) Max Capacity (lbs) Speciman Capacity (lbs) Max Capacity (lbs) Speciman Capacity (lbs) Max Capacity (lbs)
TRANS-01 -1 2071 2161 TRANS-02-1 2091 2100 TRANS-03-1 2763 2772
TRANS-01 -2 2394 2406 TRANS-02-2 1813 2315 TRANS-03-2 2754 2763
TRANS-01-3 2634 2667 TRANS-02-3 2529 2564 TRANS-03-3 2708 2738
TRANS-01-4 2852 2865 TRANS-02-4 2829 2879 TRANS-03-4 2586 2627
TRANS-01-5 2768 2776 TRANS-02-5 2585 2617 TRANS-03-5 2283 2305
TRANS-01-6 2476 2500 TRANS-02-6 2874 2905 TRANS-03-6 2546 2567
TRANS-01-7 2317 2338 TRANS-02-7 2498 2517 TRANS-03-7 2607 2645
TRANS-01-8 2538 2564 TRANS-02-8 2587 2622 TRANS-03-8 3152 3158
TRANS-01-9 3347 3362 TRANS-02-9 2840 2844 TRANS-03-9 2593 2616
TRANS-01-10 2617 2649 TRANS-02-10 2541 2562 TRANS-03-10 2441 2461
Average 2601 2629 2519 2593 2643 2665
Std Devation 345 332 336 252 230 225
Table 5.2. Phase I Ultimate Failure Load of GFRP Thickened Plate Members with
Monotonic Loading in the Axial Direction.
Specimen Capacity (lbs)
AXIAL-04-1 15655
AXIAL-04-2 14195
AXIAL-04-3 16229
AXIAL-04-4 16607
AXIAL-04-5 15021
AXIAL-04-6 13878
AXIAL-04-7 14711
AXIAL-04-8 13430
AXIAL-04-9 15931
AXIAL-04-10 12579
Average 14824
Std Dev 1308
67


Table 5.3. Phase II Residual Connection Capacity of Members Subject to Unidirectional
Cyclic Loading__________________________________________________________________________
Specimen Capacity (lbs) Max Capacity (lbs) Specimen Capacity (lbs) Max Capacity (lbs)
AXIAL-01-25 2835 2838 TRANS-01-25 2931 2934
AXIAL-02-25 2462 3022 TRANS-02-25 3185 3191
AXIAL-03-25 2161 3141 TRANS-03-25 3220 3236
AXIAL-01-50 2934 2934 TRANS-01-50 2057 2362
AXIAL-02-50 3016 3016 TRANS-02-50 2001 2285
AXIAL-03-50 2558 2558 TRANS-03-50 2083 2394
AXIAL-01-75 2295 2295 TRANS-01-75 2275 2523
AXIAL-02-75 2740 2740 TRANS-02-75 2573 2574
AXIAL-03-75 2575 2575 TRANS-03-75 2163 2465
Average-25 2486 3000 Average-25 3112 3120
Std Dev-25 338 153 Std Dev-25 158 163
Average-50 2836 2836 Average-50 2047 2347
Std Dev-50 225 244 Std Dev-50 42 56
Average-75 2537 2537 Average-75 2337 2521
Std Dev-75 225 225 Std Dev-75 212 55
Table 5.4. Phase III Residual Connection Capacity of Members Subject to
Multidirectional Cyclic Loading
Specimen Capacity (lbs) Max Capacity (lbs) Specimen Capacity (lbs) Max Capacity (lbs)
MULTI-AXIAL 01-25 2742 3103 MULTI-TRANS 03-25 1954 2168
MULTI-AXIAL 02-25 2549 2549 MULTI-TRANS 03-50 2184 2595
MULTI-AXIAL 01-50 2594 3129 MULTI-TRANS 03-75 2084 2231
MULTI-AXIAL 02-50 2047 2656
MULTI-AXIAL 01-75 1942 2774
MULTI-AXIAL 02-75 1846 1997
Average-25 2646 2826
Std Dev-25 136 392
Average-50 2321 2893
Std Dev-50 387 334
Average-75 1894 2386
Std Dev-75 68 549
68


Table 5.5. Resistance Factor Calibration for Specimens in Monotonic Unidirectional Load (Phase I)
Where p = mean (kN), o = standard deviation, COY = coefficient of variance, = resistance factor
Specimem Test Monte-Carlo simulation
First peak Maximum peak First peak Maximum peak
ft a cov d> ft a COV d> ft a COV P =3.5 AXIAL01 9.85 0.72 0.07 0.89 11.46 0.62 0.05 0.94 9.86 1.15 0.12 0.92 11.44 1.10 0.1 0.96
AXIAL02 10.44 0.86 0.08 0.86 12.35 1.47 0.12 0.74 10.33 1.18 0.11 0.89 11.71 1.25 0.11 0.83
AXIAL03 10.89 1.45 0.13 0.69 12.24 0.96 0.08 0.87 10.72 1.27 0.12 0.80 12.39 1.15 0.09 0.93
TRANS01 11.57 1.54 0.13 0.69 11.69 1.48 0.13 0.71 11.16 1.30 0.12 0.78 11.72 1.25 0.11 0.84
TRANS02 11.20 1.50 0.13 0.69 11.53 1.12 0.10 0.82 10.67 1.28 0.12 0.79 11.38 1.16 0.10 0.92
TRANS03 11.76 1.02 0.09 0.85 11.86 1.00 0.08 0.86 11.35 1.16 0.10 0.92 11.38 1.16 0.10 0.92
P =3.0 AXIAL01 9.85 0.72 0.07 0.90 11.46 0.62 0.05 0.94 9.88 1.15 0.12 0.93 11.44 1.10 0.10 0.97
AXIAL02 10.44 0.86 0.08 0.88 12.35 1.47 0.12 0.78 10.35 1.18 0.11 0.91 11.74 1.25 0.11 0.86
AXIAL03 10.89 1.45 0.13 0.73 12.24 0.96 0.08 0.89 10.73 1.27 0.12 0.83 12.34 1.15 0.09 0.94
TRANS01 11.57 1.54 0.13 0.73 11.69 1.48 0.13 0.76 11.17 1.30 0.12 0.81 11.69 1.25 0.11 0.86
TRANS02 11.20 1.50 0.13 0.73 11.53 1.12 0.10 0.84 10.63 1.28 0.12 0.82 11.37 1.16 0.10 0.93
TRANS03 11.76 1.02 0.09 0.87 11.86 1.00 0.08 0.88 11.37 1.16 0.10 0.93 11.39 1.16 0.10 0.93
P- 2.5 AXIAL01 9.85 0.72 0.07 0.92 11.46 0.62 0.05 0.95 9.89 1.15 0.12 0.94 11.44 1.10 0.10 0.97
AXIAL02 10.44 0.86 0.08 0.90 12.35 1.47 0.12 0.82 10.35 1.18 0.11 0.92 11.72 1.25 0.11 0.88
AXIAL03 10.89 1.45 0.13 0.78 12.24 0.96 0.08 0.91 10.71 1.27 0.12 0.85 12.40 1.15 0.09 0.95
TRANS01 11.57 1.54 0.13 0.78 11.69 1.48 0.13 0.80 11.15 1.30 0.12 0.84 11.73 1.25 0.11 0.88
TRANS02 11.20 1.50 0.13 0.78 11.53 1.12 0.10 0.87 10.69 1.27 0.12 0.86 11.39 1.16 0.10 0.94
TRANS03 11.76 1.02 0.09 0.89 11.86 1.00 0.08 0.90 11.39 1.16 0.10 0.94 11.37 1.16 0.10 0.94
69


Figure 5.1. Load-Displacement Curve for Phase I Axial
Figure 5.2. Load-Displacement Curve for Phase I Transverse
70


Displacement {in)
Figure 5.3. Load-Displacement Curve for Phase I Thickened Member Axial
Figure 5.4. Typical Failure Mode: Monotonic Axial Load
71


Tension Splitting Failure
Figure 5.5. Typical Failure Mode: Monotonic Transverse Load
Loading Direction
A
\
Shear-Out
Failure
Loading Direction
Tension
Splitting
Failure with
Bearing
Damage
Figure 5.6. Failure Mode: (a) Monotonic Axial Load; (b) Monotonic Transverse Load
72


Figure 5.7. Typical Failure Mode of GFRP Thickened Plate from Axial Loading
Figure 5.8. Debonding of GFRP Thickened Plate
73


(d) (e) (f)
Figure. 5.9. Cyclic Unidirectional Load for Phase II Specimens: (a) axial load at 25%PU;
(b) axial load at 50%/f,; (c) axial load at 7 5 (d) transverse load at 25%Pu\ (b)
transverse load at 50%PU; (c) transverse load at 75%PU
Figure 5.10. Load-Displacement Curve for Phase II Monotonic Axial Loading
74


Figure 5.11. Load-Displacement Curve for Phase II Monotonic Transverse Loading
(a) (b) (c)
Figure 5.12. Damage Propagation with Load Cycles: (a) Axial Load at 25%PU; (b) Axial
Load at 50%/f,; (c) Axial Load at 75%/f,
75


Loud (lb) m
Figure 5.13. Geometry of GFRP Channel Specimens for Phase HI
(d)
(e)
(f)
Figure. 5.14. Cyclic Multidirectional Load for Phase III Specimens: (a) axial load at
25%PU; (b) axial load at 50%PU', (c) axial load at 75%/f,; (d) transverse load at 25%/f,; (e)
transverse load at 50%Pu; (f) transverse load at 75%PU
76


Figure 5.15. Load-Displacement Curve for Phase III Monotonic Axial Loading
Figure 5.16. Load-Displacement Curve for Phase III Monotonic Transverse Loading
77


Figure 5.18. Typical Failure Mode: Monotonic Axial Load for Phase III after 75%PU
Multidirectional Cyclic Loading
78


Figure 5.19. Typical Failure Mode: Monotonic Transverse Load for Phase III after
Figure 5.20. Failure Mode: Monotonic Transverse Load for Phase III after 75%PU
Multidirectional Cyclic Loading
79


Figure 5.21. Load-displacement of Phase I-Comparison of Various Flange Geometry: (a)
Axial load; (b) Transverse Load
s
1
(a)
(b)
(c)
£
1
(d)
(e)
(g)
Figure 5.22. Comparison of Post-Cyclic Residual Behavior: (a) Axial Load at 25%PU; (b)
Axial Load at 50%/f,; (c) Axial Load at 7 5 (d) Transverse Load at 25%PU; (e)
Transverse Load at 50%PU; (f) Transverse Load at 75%PU
80


(a) (b)
3,500 -i 3,500 -i
="3,000 . 13 *3,000 -
12,500 1 ^ H -2,500 1
s 2,000 12,000
1*1,500 -e-Trars01 11,500 - -e- Trans 01
2.1,000 - -B-Trans 02 a1i000. -a-Trans 02
-a-Trans 03 S -e-Trans 03
500 500 -s Multi Trans 03 3. -<- Multi Trans 03
0 , , , 1 3 o -J 1 1 1
20
40
Load level (%)
60
80
20
40
Load level (%)
60
80
(c) (d)
Figure 5.23. Effect of Cyclic Loading on Residual Capacity: (a) Axial Load at 1st Peak;
(b) Axial Load at Maximum Peak; (c) Transverse Load at 1st Peak; (d) Transverse Load
at Maximum Peak
&
o
CO
O
=5

OB
c
m
Safety index Figure 5.24. Comparison of Strength Reduction Factor (0) Based on Phase I
81


6.
Summary and Conclusion
6.1 Summary
The connection capacity of Glass Fiber Reinforced Polymer (GFRP) structural channel
members has been studied through experimental and analytical investigations. A total of
97 GFRP specimens were tested in the axial and transverse directions to determine
strength capacity and was divided into three phases. Post-cyclic residual behavior was
studied after unidirectional (Phase II) and multidirectional (Phase III) cyclic loading was
performed on the test specimens. The effects of various parameters including various
flange configurations, loading direction, and level of cyclic loading have been evaluated.
The change in stiffness from the tests is also investigated due to unloading and reloading
at various load levels.
A GFRP channel structural channel member with a single bolted connection near the
edge is vulnerable to premature failure caused by high stress concentrations leading to
initial bearing failure. Bearing failure is an undesirable failure mode because it fails to
utilize the entire material strength of the member. The proposed solution to this problem
is to increase the number of bolts used for the joint as well as adjusting the edge distance
for optimal performance.
6.2 Conclusions
The following conclusions are drawn from the experimental study of GFRP structural
channel members:
82


1. The load-displacement curves indicate linear behavior up to initial failure when
testing the GFRP channel members in the axial and transverse directions.
2. The initial failure mechanism for the bolted connection results in initial bearing
failure and is represented by the first peak on the load-displacement curve.
3. Progressive damage occurs and the specimens exhibit inelastic deformation
around the bolt hole and does not yield similar to steel.
4. Testing show that using different flange configuration which consists of cutting
the flange at various degrees, do not affect the overall connection capacity in the
GFRP specimens in both the axial and transverse directions.
5. Monotonic axial loading results in a two phase failure mechanism. Initial bearing
failure is observed followed by shear-out failure which results in significant loss
of load carrying capacity.
6. Monotonic transverse loading also results in a two phase failure mechanism.
Initial bearing failure first occurs followed by net tension failure. The net tension
failure is characterized by splitting in the transverse direction of the load. The
majority of the testing show the splitting occurring at the bolt hole with a couple
of instance where the splitting occurred below the bolt hole and directly above the
fixture clamp.
7. The thickened GFRP members with an additional bonded plate exhibit
significantly increase in strength capacity which is expected due to the increased
area. The failure mode observed is debonding of the bonded plate followed
immediately by shear-out failure. This is expected due to a sudden reduction in
83


area. Initial bearing failure is not witnessed similar to all the other specimens
tested.
8. Initial failure represented by the first peak on the load-displacement curve does
not necessarily constitute the maximum capacity. After initial failure the
specimen can still achieve a higher maximum capacity which is represented by
the maximum peak on the load-displacement curve. This occurs in both the axial
and transverse directions.
9. After initial failure the specimens experience a period of damage accumulation
which is represented by a region of multiple peaks in the load-displacement curve
until final failure when the specimens loses a significant amount of load-carrying
capacity.
10. The load testing results do not indicate a reduction in strength capacity when
comparing Phase II to Phase I. In some instances specimens from Phase II after
unidirectional cyclic loading lead to higher connection capacities.
11. For Phase II axial loading the results indicate a reduction in residual strength
capacity as the load level percent for cyclic unidirectional loading is increased.
12. For Phase III axial load to failure, the results indicate a reduction in residual
strength capacity as the load level percent for multidirectional loading is
increased.
13. The residual connection capacity of Phase III transverse specimens loaded to
failure is significantly reduced after multidirectional cyclic loading when
compared to Phase II and Phase I.
84


14. The residual connection capacity of Phase III axial specimens loaded to failure
has a small reduction after multidirectional cyclic loading when compared to only
unidirectional loading in Phase II.
15. The post-cyclic residual failure mode is the same as specimens from Phase I,
which is shear-out and net tension failure. The only difference is deflections and
damage accumulation that occur from the cyclic loading.
16. Minor stiffness degradation occurs from multidirectional cyclic loading.
6.3 Recommendations for Future Work
Additional research is required to fully understand the behavior of GFRP structural
members with bolted connections. The following aspects should be addressed:
1. Developing a finite element model to predict damage accumulation around the
bolt hole.
2. Investigating GFRP connections with multiple bolts to fully optimize the material
strength in the member.
3. Investigation of edge distance on GFRP connections to fully optimize the material
strength capacity.
4. Evaluation of the stress concentrations experimentally and numerically around the
bolt hole.
5. Investigation of simultaneous multidirectional monotonic and cyclic loading on
the connection capacity.
6. Generate and establish a complete design standard used for practicing structural
engineers for bolted GFRP connection
85


Lead 0b) Load (lb)
Appendix A. Load Displacement Curves for Phase I Axial and Transverse
toptaesmentCh}
Displacement (ki)
Dtaptaaamant(h)
86


Load (lb) Load (lb) Load (lb)
Displacement (it)
87


Full Text

PAGE 1

CONNECTION CAPACITY OF PU LTRUDED GFRP CHANNELS IN MULTIDIRECTIONAL LOADING by MICHAEL C. WANG B.S., University of Colorado Denver, 2008 A thesis submitted to the Faculty of the Graduate School of the University of Colorado in partial fulfillment of the requirements for the degree of Master of Science Civil Engineering 2013

PAGE 2

ii This thesis for the Mast er of Science degree by Michael C. Wang has been approved for the Civil Engineering Program by Yail Jimmy Kim, Advisor Kevin Rens, Chair Frederick Rutz July 12, 2013

PAGE 3

iii Wang, Michael, C. (M.S., Civil Engineering) Connection Capacity of Pultruded GFRP Channels in Multidirectional Loading Thesis directed by Associate Professor Dr. Yail Jimmy Kim ABSTRACT Fiber Reinforced Polymer (FRP) composites is a relatively new ma terial that offers unique advantages for many different engine ering applications. While there has been plenty of research performed to unders tand composite fiber failure mechanisms numerically and experimentally, the exact nature of failure based on fiber orientation is unknown. Due to the complex nature of FRP co mposites, fiber failure mechanisms are not clearly understood and the goal of this st udy is to gain furt her knowledge regarding FRP failure, particularly with bolted connections. The primary objective of this research is to investigate the behavior of bolted Glass Fiber Reinforced Polymer (GFRP) structural cha nnel members subject to cyclic loading. The investigation consisted of te sting various channel specim ens with different flange geometric configurations. Monotonic loading is performed to determine ultimate strength of the connection in the axia l and transverse directions for the first phase. The second phase comprises of how unidirectional cyclic loading effects stiffness and the overall ultimate strength of the connec tion in the axial and transverse direction. The last phase studies the behavior and the residual connection capacity of the GFRP specimens subject to multidirectional cyclic loading. Various load levels for cyclic loading were established as a fraction of the ultimate strengths obtained from testing.

PAGE 4

iv The testing showed that the absence of the fl ange did not have a significant impact with respect to the ultimate strengths in the ax ial and transverse directions. Failure modes observed consisted of initial bearing failure followed by shear-out failure for axial specimens and net tension splitting for the transverse specimens. The GFRP channel specimens undergo a period of nonlinear prog ressive damage after initial failure. Furthermore, it was found that the residual behavior and strength was affected by unidirectional and multidire ctional load cycles. A probability-based design was adopted to provide design re commendations and guidelines. Resistance factors are determined using the Monte-Carlo simulation for the Load and Resistance Factor Design (LRFD) design method. The form and content of this abstract are approved. I reco mmend its publication. Approved: Jimmy Kim

PAGE 5

v ACKNOWLEDGMENTS My sincere gratitude goes to my advisor Dr Yail Jimmy Kim and co -advisor Dr. Rui Liu. It has been a pleasure to work with both and I thank them for their guidance, support, and time during the preparation of this thesis. They provided me with invaluable knowledge and expertise necessary for completion of the re search project. Also IÂ’d like to thank Dr. Kevin Rens and Dr. Frederick Rutz for part icipating on my thesis defense committee. The material testing conducted in this re search study would not have been possible without the support provided by the following st aff and students: Tom Thuis, Jac Corless, Eric Losty, and Dennis Dunn. A special thanks to Shahlaa Alwakeel, for her guidance and assistance in the lab. Y our hard work is greatly ap preciated. Finally, the author gratefully acknowledges support from International Cooling Towers. I would also like to express my appreciation to my fellow colleagues and mentors David Blanchette and Jerry Isler, both have been instrumental in the development of my professional and academic career. Lastly, I would like to thank my friends a nd family for their support and encouragement along the way.

PAGE 6

vi TABLE OF CONTENTS Tables ...................................................................................................................... .......... x Figures ...................................................................................................................... ........ xi Chapter 1. Introduction ............................................................................................................. 1 1.1 General ................................................................................................................. ... 1 1.2 Objectives ............................................................................................................. .. 2 1.3 Scope .................................................................................................................... ... 3 1.4 Outline of Thesis ..................................................................................................... 3 2. Literature Review.................................................................................................... 5 2.1 Introduction. ............................................................................................................ 5 2.2 Overview. ................................................................................................................ 5 2.3 Material Properties of FRP. .................................................................................... 6 2.4 Fabrication of FRP Members .................................................................................. 7 2.5 Bolted FRP Members .............................................................................................. 8 2.6 Fasteners ................................................................................................................ 9 2.7 Failure Modes ......................................................................................................... 9 2.7.1 Tension Failure ..................................................................................................... 10 2.7.2 Shear Failure. ........................................................................................................ 11 2.7.3 Bearing Failure. ..................................................................................................... 1 1 2.7.4 Cleavage and Pull-out Failure ............................................................................... 12 2.8 Factors Affecting Joint Strength ........................................................................... 12 2.8.1 Fiber Orientation ................................................................................................... 12

PAGE 7

vii 2.8.2 Lateral Contraint ................................................................................................... 13 2.8.3 Stacking Sequence ................................................................................................ 13 2.8.4 Joint Geometry ...................................................................................................... 14 2.9 Static Behavior of Pultruded GFRP Beams. ......................................................... 14 2.9.1 Experimental Setup and Test Procedure. .............................................................. 15 2.9.2 Results and Conclusions. ...................................................................................... 15 2.10 Multi-Bolted Joints For GFRP Structural Members ........................................... 16 2.10.1 Test Parameters ..................................................................................................... 17 2.10.2 Test Setup and Instrumentation ............................................................................ 17 2.10.3 Experimental Results ............................................................................................ 18 2.10.4 Conclusions ........................................................................................................... 19 2.11 Finite Element Modeling of Damage Accumulation ............................................ 19 2.11.1 Progressive Damage Model .................................................................................. 20 2.11.2 Finite Element Modeling ...................................................................................... 23 2.11.3 Results and Discussions ........................................................................................ 24 3. Experimental Program .......................................................................................... 39 3.1 Introduction ........................................................................................................... 3 9 3.2 Materials used for Test Specimens ....................................................................... 39 3.2.1 GFRP Channels ..................................................................................................... 39 3.2.2 GFRP Flat Plates ................................................................................................... 40 3.3 Experimental Setup and Loading .......................................................................... 41 3.3.1 Phase I and Phase II Setup for Axial Tests ........................................................... 41 3.3.2 Phase I and Phase II Setup for Transverse Tests .................................................. 42 3.3.3 Phase III Testing Procedure .................................................................................. 42

PAGE 8

viii 4. Finite Element Modeling ...................................................................................... 48 4.1 Introduction ........................................................................................................... 4 8 4.2 Modeling and Elements ........................................................................................ 48 4.3 Failure Criteria ...................................................................................................... 49 4.4 Model Results ....................................................................................................... 50 5. Test Results and Discussion.................................................................................. 54 5.1 Introduction ........................................................................................................... 5 4 5.2 Phase I Testing Results of GFRP Channels .......................................................... 55 5.2.1 Phase I Failure Modes of GFRP Channels. .......................................................... 56 5.2.2 Phase I Testing Results of GFRP Thickened Plates. ............................................ 57 5.2.3 Phase I Failure Modes of GFRP Thickened Plates. .............................................. 57 5.3 Phase II Testing Results of GFRP Channels ........................................................ 58 5.3.1 Phase II Failure Modes of GFRP Channels .......................................................... 59 5.4 Phase III Testing Result s of GFRP Channels ....................................................... 60 5.4.1 Phase III Failure Mode s of GFRP Channels ......................................................... 62 5.5 Comparisons of Phases. ........................................................................................ 63 5.5.1 Comparison Among Different Flange Geometries in Phase I. ............................. 63 5.5.2 Comparison of Post-Cyc lic Residual Behavior. ................................................... 63 5.5.3 Effect of Cyclic Loading on Residual Capacity ................................................. 64 5.6 Design Recommendations. ................................................................................... 64 5.6.1 Monte-Carlo Simulation. ...................................................................................... 64 5.6.2 Calibration of Resistance Factor. .......................................................................... 65

PAGE 9

ix 6. Summary and Conclusions. .................................................................................. 82 6.1 Summary. .............................................................................................................. 8 2 6.2 Conclusions. .......................................................................................................... 8 2 6.3 Recommendations for Future Work .................................................................... 85 Appendix A Load Displacement Curves for Phase I – Axial and Transverse ...........................86 B Load Displacement Curves for Phase II – Residual Capacity ..............................98 C Load Displacement Curves for Phase III – Residual Capacity ...........................101 D Phase I Testing Failure Photos – Axial and Transverse ......................................103 E Phase II Testing Photos – Damage From Cyclic Loading ..................................113 F Phase III Testing Photos – Da mage From Multi Directional Cyclic Loading ....116 G Phase III Testing Failure Photos ..........................................................................118 References ........................................................................................................................119

PAGE 10

x LIST OF TABLES Table 2.1 Comparison of Fle xural Rigidity. ............................................................................ 26 2.2 Comparison of Shear Rigidity. ................................................................................. 26 2.3 Geometrical Data for th e Investigated Geometry. ................................................... 27 2.4 Elastic Properties of HTA/6376 Material. ............................................................... 27 4.1 Material Properties for GFRP Channel Specimens (Extren). .................................. 51 4.2 Failure Criteria Strength Values. .............................................................................. 51 5.1 Phase I Ultimate Failure Load of GFRP Channel Specimens. ................................ 67 5.2 Phase I Ultimate Failure Load of GFRP Thickened Members. ............................... 67 5.3 Phase II Residual Connection Capacity of Members Subject to Cyclic Loading. ... 68 5.4 Phase III Residual Connection Capacity of Members Subject to Cyclic Loading. .. 68 5.5 Resistance Factor Calibration. ................................................................................. 69

PAGE 11

xi LIST OF FIGURES Figure 2.1 FRP Pultrusion Process. ............................................................................................. 28 2.2 Modes of Failure for Bolted Joints in FRP Composities. .......................................... 28 2.3 Stress Concentration Relief in Fibrous Composites by Delamination. ...................... 29 2.4 Relation Between Stress C oncentration Factors. ....................................................... 29 2.5 Terminology. ............................................................................................................. 30 2.6 Influence of CFRP Fiber Proportion. ......................................................................... 30 2.7 Effect of Bolt Torque on Bearing Strength. ............................................................... 31 2.8 Comparison between Bearing Strength wi th and without Clamping Pressure. ......... 31 2.9 Effect of Grouped 0 Plies on Bearing Strength. ....................................................... 32 2.10 Effects of Stacking Sequence on Bear ing Strength for GFRP Laminates. .............. 32 2.11 Variation of Net Tensile Strength with Width of 0/45. ...................................... 33 2.12 Cross Sections of Test Specimens. .......................................................................... 33 2.13 Connection Configuration. ....................................................................................... 34 2.14 Flowchart of Progressive Damage Model. ............................................................... 35 2.15 Geometry of Bolted Single-Lap Joint. ..................................................................... 36 2.16 Finite Element Model of Bolted Joint. ..................................................................... 36 2.17 Mesh Around Hole and Finite Element Model of Bolt. ........................................... 37 2.18 Comparison of Strains.............................................................................................. 37 2.19 Illustration of Damage Propagati on in Upper Surface of Laminate. ....................... 38 2.20 Illustration of Damage Propagation in Lower Surface of Laminate. ...................... 38

PAGE 12

xii 3.1 Condition #1.............................................................................................................. 43 3.2 Condition #2.............................................................................................................. 43 3.3 Condition #3.............................................................................................................. 44 3.4 Tension Test. ............................................................................................................ .. 44 3.5 Shear Test................................................................................................................ ... 45 3.6 Geometry of Thickened Flat Plate. ............................................................................ 45 3.7 Axial Test Setup. ........................................................................................................ 46 3.8 Transverse Test Setup. ............................................................................................... 46 3.9 Thickened Flat Plate Test Setup. ............................................................................... 47 3.10 Thickened Flate Plate Test Setup. ............................................................................ 47 4.1 Channel Geometry of ANSYS Model. ...................................................................... 52 4.2 Applied Loading of Model. ........................................................................................ 52 4.3 Displacement from 25% of Ultimate Load. ............................................................... 53 4.4 Tsai-Wu Strength Index – 25% of Ultimate Load. .................................................... 53 5.1 Load-Displacement Curve for Phase I Axial. ............................................................ 70 5.2 Load-Displacement Curve for Phase I Transverse. ................................................... 70 5.3 Load-Displacement Curve for Phase I Thickened Member Axial. ............................ 71 5.4 Typical Failure Mode: Monotonic Axial Load. ......................................................... 71 5.5 Typical Failure Mode: Monot onic Transverse Load. ................................................ 72 5.6 Failure Mode: (a) Monotonic Axial Lo ad; (b) Monotonic Transverse Load. ............ 72 5.7 Typical Failure Mode of GFRP Thic kened Plate from Axial Loading. .................... 73 5.8 Debonding of GFRP Thickened Plate. ....................................................................... 73 5.9 Cyclic Unidirectional Load for Phase II Specimens. ................................................. 74 5.10 Load-Displacement Curve for Phase II Monotonic Axial Loading. ........................ 74 5.11 Load-Displacement Curve for Phase II Monotonic Transverse Loading. ............... 75

PAGE 13

xiii 5.12 Damage Propagation with Load Cycles. .................................................................. 75 5.13 Geometry of GFRP Channel Specimens for Phase III............................................. 76 5.14 Cyclic Multidirectional Load for Phase II Specimens. ............................................ 76 5.15 Load-Displacement Curve for Phase III Monotonic Ax ial Loading. ...................... 77 5.16 Load-Displacement Curve for Phase III Monotonic Transverse Loading. .............. 77 5.17 Damage Propagation with Load Cycles. .................................................................. 78 5.18 Typical Failure Mode: Monotonic Axial Load for Phase III. .................................. 78 5.19 Typical Failure Mode: Monotonic Tr ansverse Load for Phase III. ......................... 79 5.20 Failure Mode: Monotonic Transverse Load for Phase III. ...................................... 79 5.21 Load-displacement of Phase I-Compar ison of Various Flange Geometry. ............. 80 5.22 Comparison of Post-Cyc lic Residual Behavior. ...................................................... 80 5.23 Effect of Cyclic Loading on Residual Capacity. ..................................................... 81 5.24 Comparison of Strength Reduction Factor ( ) Based on Phase I. .......................... 81

PAGE 14

1 1. Introduction 1.1 General High cost from repair and maintenance of st eel structures damage d from corrosion lead engineers and researchers to search for alte rnative materials. Fiber Reinforced Polymer (FRP) composites is a relatively new materi al that offers unique advantages for many different engineering applications. A composite material is formed by the combination of two or more distinct materials to produce a new material with enhanced properties. FRP utilizes a polymer resin matrix that is typi cally reinforced with glass, carbon, basalt or aramid fibers. The advantages of using FR P are the excellent strength-to-weight ratio and stiffness-to-weight ratio which makes them highly desirable as a building material for structural systems. The composite material has proven to be economi cal and efficient for new construction and repair and rehabilitation of damaged or deteriorating structures in civil engineering. In addition, they pr ovide favorable corrosion and weathering resistance. The application of FRP composites in struct ural engineering involves strengthening of beams, columns, and slabs in existing structures They can be used for repairing structural members that have been damaged from lo ading conditions. For example, FRP can be applied to provide flexural and shear streng thening of a damaged beam or column. The strengthening will improve the stiffness and deflection capacity of the member. If flexural strengthening is desire d FRP sheets or plates are app lied to the bottom or tension face of the beam. When FRP is applied to the web or sides of the beam the shear strength is improved.

PAGE 15

2 FRP composites can also be molded into differe nt structural shapes and is used as an alternative to steel and aluminum due to it s benefits. Today manuf acturers can produces a variety of structural members including channe ls, angles, wide flange beams, tubes, and flat-sheets similar to structural steel shapes. While there has been plenty of research pe rformed to understand composite fiber failure mechanisms numerically and experimentally, the exact nature of failure based on fiber orientation is unknown. Due to the complex na ture of FRP compos ites, fiber failure mechanisms are not clearly understood and the goal of this study is to gain further knowledge regarding FRP failure, particularly with bolted connections. Glass fibers are commonly used for reinforced composites. Glass fiber reinforced polymers (GFRP) channel members with bolted connections will be tested and researched in this study. 1.2 Objectives The objective of this research is to evalua te GFRP channel members subject to cyclic axial and transverse loading. The components involved in th e investigation include the following: 1. Review of previous resear ch and testing conducted on GF RP structural members. 2. Examine the ultimate connection load cap acity of the GFRP specimens in the axial and transverse directions. 3. Observe various failure modes based on different loading and geometry conditions.

PAGE 16

3 4. Evaluate the impact of unidirectional and multidirectional loading and unloading cycles and determine the effect it has on the residual strength capacity of the connection. Observe the change in st iffness at various load levels. 5. Develop a finite element model to predic t the load-deflection behavior and local stress characteristics of the specimens. 6. Summarize results to provide design recommendations. 1.3 Scope The scope of the research consists of an experimental program followed by a finite element model to study the behavior of GFRP channel members with bolted connections subject to unidirectional and multidirectional cyclic axial a nd transverse loading. Testing was performed to determine ultimate strength s and to apply load cycles at different levels. The experimental investigation studies the local deteriorati on around the bolt holes influenced by high stress concentrations and the reduction of ultimate strength and stiffness caused by repeated loads. A finite element model was developed to validate and predict the deflection response and stress conc entrations. The model accounts for change in material properties du e to cyclic loading. 1.4 Outline of Thesis The contents of this thesis include the following: Chapter 2: presents a review of literature related to previous research conducted on GFRP structural members.

PAGE 17

4 Chapter 3: provides a detailed description of th e experimental progr am, fabrication of test specimens, instrumentation, test setup a nd procedures. In additi on, all ancillary tests required for necessary material properties are discussed. Chapter 4: provides a detailed description of th e finite element model developed to validate and predict the displa cement and stress response to cy clic loading of GFRP test specimens. Chapter 5: discusses the experimental program resu lts, which includes the strength and stiffness degradation due to cyclic load ing, various failure modes from different geometric and loading conditions, and results of all ancillary mate rial tests performed. Design recommendations are provid ed using the results obtained. Chapter 6: presents the summary and conclusion of the research along with recommendations for further research on GFRP members.

PAGE 18

5 2. Literature Review 2.1 Introduction The literature presented in this chapter will i nvestigate past research and tests performed on FRP members. The focus of this section w ill be failure behavior and mechanism from bolted connections and how they ar e affected by fiber orientation. This section will discuss the literature avai lable on the subject of GFRP structural members. It is evident based on the study of current literature av ailable that there is a lack of information about how to thoroughly characterize pultruded GFRP members. Currently there is no design standard for FRP. 2.2 Overview In recent years there has been significant gr owth of FRP, there usage has spread from aerospace and automotive industry to civil stru ctures. When compared with traditional materials the significant advantages are th e resistance to corros ion, lightweight, high strength, and ease of installation. However, negative aspects of FRP would include the high initial cost, durability problems caused by freeze-thaw cycles, moisture, or sustained loading, and the lack of design standards a nd experience. Their strength and stiffness properties are dependent on the type, quantity, and orientation of the fibers within the member.

PAGE 19

6 2.3 Material Properties of FRP Material properties of FRP are complex due to the orthotropic nature of the composite material and a clear explanation can be qu ite difficult. Unlike the American Concrete Institute (ACI) for reinforced concrete de sign and the American Institute of Steel Construction (AISC) for structural steel de sign, FRP pultruded structural shapes currently do not have a design code, although two desi gn guides are available for engineers: Structural Plastics Design Manual an d Eurocomp Design Code and Handbook (ASCE 1984 and Eurocomp 1996). To complicate matters further, there are many FRP manufacturers in the industry who provide diffe rent material propertie s for their products. Depending on the fiber orientation within the polymer matrix FRP can exhibit different material properties. Other important factors th at influence the behavi or of FRP materials include the type of fiber and polymer used in the composite (T ibbetts 2008). Common reinforcement fibers used include glass, carbon, aramid, and boron fibers. The unique anisotropic behavior requires special cons iderations in the manufacturing process and design of FRP materials. FRP composites pose a high resistance to corro sion regardless of the materials used for the fiber and polymers. However, the durabil ity can be reduced when exposed to harsh environmental conditions which should be avoided for adequate performance, ranging from high temperatures, ultraviolet exposure, and water with high alkalinity levels. The resistance to environmental effects is great ly dependent on the fabr ication process along with the material properties of the fibers (Benmokrane et al 2002).

PAGE 20

7 In comparison to steel, FRP structural member s have anisotropic properties, low stiffness, and high elastic to shear modulus ratios. Fr om previous experience and tests composite materials must be carefully analyzed due bri ttle failure modes. FRP differ from metals as typically there is no yielding, strain hardening or elongation prior to failure (Deng 2004). Generally GFPR members have a linear stress-strain relations hip to failure. Certain types of GFRP can exceed the strength of conventional steel. A study preformed reported the ultimate strength of GFRP bars at 150 ks i (1,035 MPa) (Pleimann 1987). However, the modulus of elasticity wh en compared to steel is signifi cantly smaller at approximately 25% of steel which places limitation for the use of GFRP as the main load-carrying element in many types of structures. Graphite reinforcing fibers can be used to increase the modulus of elasticity by up to thr ee times (Saadatmanesh and Ehsani 1991). According to ACI (2006), the tensile strength of GFRP bars can up to twice as large as the tensile strength of steel bars and is even higher for carbon and aramid FRP. 2.4 Fabrication of FRP Members Fabrication of FRP composite s can be completed by the method of pultrusion or by hand layup (Tibbetts 2008). This research uses FRP products that were manufactured through the process of pultrusion. Fo r fabrication by pultrusion, FRP members are formed into different standard structural shapes by an automated process. The manufacturing process produces continuous lengths of various shapes and materials by combining fibers with a polymer resin matrix under heat and pressu re and pulled through using a heated steel forming die. Fibers typically used are gl ass and carbon, although other fibers such as

PAGE 21

8 aramid or polyethylene can be used for st ructural applications (Bank 2006). The basic manufacturing process concept is described in Figure 2.1. 2.5 Bolted FRP Members Plenty of research has been performed studyi ng the behavior of FRP in the aerospace and automotive engineering industries; however th ere has been a lack of research conducted for structural engineering applic ations for FRP materials partic ularly in the field of bolted connections (Hassan et al. 1994). Similar to steel, FRP structural members utilize bolted connections and although mechan ically fastened, FRP joints share the same failure modes as metals. It has been determined that the damage mechanism and propagation is not quite the same. Therefore, the design criteria for steel are not applicable to FRP members (Duthinh 2000). It is important to understa nd the behavior or FRP members with bolted connections because the strength of the structur al member is limited by the strength of the connection. FRP structural members can be significantly weakened by the presen ce of bolt holes or cut outs. The strength reduc tion is attributed to high stress concentrations around discontinuities, damage to the reinforci ng fibers and because FRP composites do not yield. The stress concentration value around a hole for isotropic materials is 3 while a uni-directionally FRP sheet can have a value as large as 8 (Colli ngs 1987). Generally isotropic materials exhibit plasticity that re lieve high stress concentrations resulting in a small effect on the net failure stress. However, for uni-directional FRP due to the lack of

PAGE 22

9 plasticity the material is elastic to failure where the stress concentrations reduce the net failure stress. 2.6 Fasteners Mechanical fasteners have proven to be an effective method for connecting FRP members. Different types of steel fasteners can be used for connection joints such as nuts, bolts, threaded rods, rivets and self-tapping screws. Self-t apping screws are favorable when it is not possible to access the reverse si de of the joint. Th e use of rivets are appropriate for joining laminates up to 0.1 in (3 mm) thick. However, the installation operation of rivets can potentially damage the laminates due to uncontrollable clamping pressure. In addition to steel fasteners FRP mechanical fasten ers are also available which include nut, bolts, threaded r ods, and screws. These fastener s present disadvantages when compared to steel due to the high cost and are susceptible to shear failure at low loads. Although there are various options for fasteners Collins (1977) determined that bolted joints were the most efficient and effective method for FRP materials. 2.7 Failure Modes FRP have similar failure modes to steel me mber connections by failing in shear, tension, or bearing. The connector can also fracture and pull through the la minates. In addition, FRP can also fail by delamination between layers, debonding, and crack propagation. The different types of failure modes for bolted joints in FRP composites are presented in Figure 2.2. The best performance for shear, te nsion, and bearing failure is obtained utilizing 0/45 fiber orientations.

PAGE 23

10 2.7.1 Tension Failure When members are subject to a te nsile load the average net stress n across a section is: (2-1) Where P is the tensile load at the joint having a width w and thickness t The number of bolts within the section is n having a diameter d Generally, stress concentration at the hole will initiate failure because FRP doe s not yield to alleviate high stress concentrations. However, a study by Hart-Smith (1980) showed that slippage between the resin and fibers near bolt holes and cutouts provide some relief of high stress concentrations (Figure 2.3 and Figure 2.4). When the fibers experien ce high local stress they pull out of the resin resu lting in delamination or debondi ng of the fibers. Hart-Smith (1980) concluded that around holes improving adhesion between the fibers and resin results in a reduction of joint strength. Howe ver, reinforcing fibers near the hole in various directions diminishes the degree of anisotropy allowing minor plastic behavior and softening. Fiber orientation of the FRP di ctates the tensile strength of the composite. The majority of the load is carried by the fibers parallel to the load. Potter (1978) determined that the failure near holes in CFRP occurred at the e dge of the hole perpendicular to the loading axis. For GFRP the failure is more compli cated and is influenced by both shear and tension. Failure around the hole is initiated by in-plane shea ring across the width of the

PAGE 24

11 laminate. The shear results in only the 45 p lies taking the axial load leading to tension failure immediately afte r (Godwin et al. 1982). 2.7.2 Shear Failure The shear stress in a FRP joint is calculated by the following equation: (2-2) Where P is the applied load with thickness t and a distance from the bolt hole center to the ends of the connected plates is e See Figure 2.5 for a diagram of the terminology used. It is noted that the sh ear stress is determined using the same procedure as for isotropic materials. Similar to tensile lo ading the shear strength around the hole is dependent on the fiber orientation of th e member. The in-plane shear strength is significantly lower with 0 fibers when comp ared with joints that have 45 fibers. 2.7.3 Bearing Failure The average bearing stress at the cross s ection of the hole can be calculated by: (2-3) Where P is the load applied, n is the number of bolts with diameter d and with a material thickness of t Bearing stress is caused by compression of half the bolt hole transferred by the bolt. The compressive strength within th e fibers and the clamping pressure are the main parameters that will dictate the b earing strength around the hole (Collins et al.

PAGE 25

12 1977). In a study by Collins (1987) he showed that the shear failure in itiated at the hole edge through the fibers and matrix. FRP la minates with a combination of 45 and 90 plies performed well under bolt bearing load s. This discovery contradicted the compressive performance of laminates with 45 and 90 fiber orientations which indicates the failure mechanisms for bearing and compressive loadi ngs are different. The difference can be attributed to the clamping pressure generated by the bolt. The fibers perpendicular to the load and laminates are c onstrained and fail in constrained transverse compression. 2.7.4 Cleavage and Pull-out Failure In addition to shear, tension, and bearing failu re FRP joints can exhibit cleavage and pullout failure. Cleavage failure develops only dur ing 0 fiber orienta tion in a single shear mode and is followed by net section failure occurring on one side of the laminate. Reinforcement should be placed around the bol t holes to prevent cleavage failure. Pullout failure is initiated by out of plane bendi ng at the joint caused from in plane axial loads which results in peeling at the joint. T ypically, they occur with the use of rivets in single shear. Bending or shear fa ilure of the bolt is also possible under extreme loading conditions (Duthinh 2000). 2.8 Factors Affecting Joint Strength 2.8.1 Fiber Orientation The biggest factor affecting th e joint strength of FRP is th e fiber orientation. In a study by Collins (1977), concluded that optimum join t performance for CFRP is achieved using

PAGE 26

13 a combination of 0 / 45 plies. It can be seen from Figure 2.6 that shear failure occurs with low levels of 45 fibers. The shear streng th of the joint increases as the 45 fibers increase in the section, eventually bearing becomes the critical mode of failure. As a result of 45 fibers being weak in tens ion, high amounts of 45 will shift the failure mode to tension. 2.8.2 Lateral Constraint Lateral constraint caused by clamping pressure from the bolts can significantly affect the joint strength. However, excessive pressure fr om over-tightening of the bolts can damage the surface of the laminate. The recommended optimum bolt clamping pressure was determined to be 3,190 psi (22 MPa) fo r CFRP joints (Garbo and Ogonowski 1981). Overtime clamping pressure is decreased by resin creep but does not necessarily reduce the joint strength (Shivakumar and Crew s 1982). See Figure 2.7 – Figure 2.8 for the effect that clamping pressure has on th e bearing strength of FRP laminates. 2.8.3 Stacking Sequence In a study performed by Collins (1977) it was shown that there was no change in shear strength for bolted CFRP joints that consisted of 2/3 0 and 1/3 45 plies, however there was a difference of 6% in tensile strength when compared with two different stacking sequences. The bearing strength had a substa ntial drop with laminates grouped together. The highest bearing strength for pin-loaded holes was achieved by using 90/45/0 laminates whereas using 0/90/45 reduced the strength by 30% (Quinn and Mathews

PAGE 27

14 1977). See Figure 2.9 and Figure 2.10 for the eff ect that the stacking sequence has on the bearing strength of GFRP laminates. 2.8.4 Joint Geometry The tensile failure stress of FRP members w ith a hole heavily depe nd on the width due to no stress relief from lack of plasticity. The width has a sign ificant impact on the strength when the laminates consists of mostly 0 fi bers and an opposite eff ect on 45 fibers. This can be seen in Figure 2.11. Collins (1977) de termined that the hole size didnÂ’t have a significant impact on the net tensile and shear st rength with fiber orientation consisting of 0 / 45 for CFRP. The same could be said for GFRP laminates (K retsis and Matthews 1985). The bearing strength of CFRP is not affected by hole size as long as there is sufficient clamping pressure provided by the bolt. On the other hand, because GFRP has low elastic modulus of glass, out-of-plan e cracking can occur re gardless of clamping pressure for d/t > 3. Values of d/t < 3 is not recommended due to the possibility of bolt shear failure. It is desirable that the joint geometry selected experiences tension and shear failure simultaneously near the bearing failure stress. 2.9 Static Behavior of Pultruded GF RP Beams (Nagaraj and GangaRao 1997) A study was performed by Nagaraj and GangaRao (1997) to investigate the experimental and theoretical characterizations of mechani cal properties of pultruded GFRP structural members. A total of 187 test were conduc ted using wide flange and box beams to examine the effects of shear influence, shear lag, warping, and manufacturing quality.

PAGE 28

15 2.9.1 Experimental Setup and Test Procedure The testing consisted of two different struct ural shapes with different sizes. The sizes were 102 x 102 x 6 mm (4 x 4 x in.) for th e box section and the wide flange section was 102 x 102 x 6 mm (4 x 4 x in.) and 152 x 152 x 6 mm (6 x 6 x in.). See Figure 2.12 for the cross section dimensions of the test specimens. The members were made up with a vinylester matrix reinforced with Eglass fibers. The specimens were testing using a span length of 1,828 mm (72 in.) with simply supported boundary conditions under three-point and four-point bending. Linear variable differential transducers (LVDTs) were used to measure the deflection unde r the load points at midspan. Electrical resistance strain gauges were placed at a distance of 203 – 305 mm (8 – 12 in.) away from the load point to overcome stress con centration effects. The strain gauges were installed on both the compression and tension face at equal distance from the midspan. 2.9.2 Results and Conclusions The experimental results were compared us ing theoretical comput ations of simplified equations based on classical lamination theo ry (CLT) (Jones 1975) along with threedimensional finite-element analysis usi ng ANSYS 1994. Both the flexural and shear rigidities were determined based on the a pproximate CLT. The finite-element model utilized SHELL 91 elements which had the ability to model the composite material layer by layer and specify the material properties and fiber orientation. The finite-element results compared well with the theoretical ca lculations and experimental results and were within 0-4%. The comparisons are displayed in Table 2.1 and 2.2.

PAGE 29

16 It was determined that the shear influen ce on deflection measurements for both threepoint and four-point bending was significant. The testing resulted in a shear influence of 36% for three-point bending test and 25% for the fourpoint bending test. Using 45 layers reduced the shear influence by 9% co mpared to the specimen with unidirectional fibers in the web. The varia tion in strain readings was 10-20% and was attributed to interfacial slip between the t op layer and the layer beneath it. In addition, variation in strain reading also existed due to asymmetr ic fiber distribution located in the top and bottom flange laminates. Nagaraj and Ganga Rao (1997) concluded that the approximate theoretical expressions provide d by Jones (1975) could be used to determine the flexural and shear rigidities in a pultruded GFRP beam. 2.10 Multi-Bolted Joints for GFRP Stru ctural Members (Hassan et al. 1997) An experimental investigation performed by Hassan et al. (1 997) at the University of Manitoba studied the behavior of multi-bol ted connections using glass fiber-reinforced plastic materials. The study consisted of testing a total of 105 multi-bolted double lap shear connections with various parameters that included the widt h of the structural member, edge distance, number of bolts, bol t pattern, pitch, thickness of the member, and direction of the fibers with respec t to the applied load direction. The orthotropic composite material used for the tests were EXTREN Flat Sheet Series 500, pultruded glass fiber sheet with alternating stacked laye rs of E-glass rovings and Eglass continuous strand mat.

PAGE 30

17 2.10.1 Test Parameters The influence of the placement and number of bolts was investigated by using 5 members with different connection configur ations and were designated as Joints A, B, C, D, and E, see Figure 2.13. Various ratios of edge distan ce to hole diameter (e/d ) and side distance to the pitch (s/p) were selected to obtain di fferent failure modes with each bolt patterns. The 105 total specimens tested were 12.7 mm (1/2 in.) thick and had principal unidirectional fibers layers orie ntated at 0, 45, and 90 with respect to the applied load. High-strength structural bolts w ith a diameter of 19 mm (3/4 in.) and a bolt hole diameter of 20.6 mm (13/16 in.) providing adequate clearance was used for all connections. The bolts were tightened to a consta nt torque of 32.5 Nm (24 ftlb). 2.10.2 Test Setup and Instrumentation The specimens were tested using a 1,000 kN (220,000 lb) MTS closed-loop servocontrolled loading system. The members were loaded axially in tension at a constant stroke rate of 0.001 mm/s ( 0.0254 in./s). Concentric load ing was achieved by a using double-shear configuration thus removi ng any bending effects. The relative displacements were measured using a linearl y variable differential transducer (LVDT) and strain gauges were used to determine th e stress distribution in the member along with load distribution of each bolt. 2.10.3 Experimental Results The results indicated that the ultimate load was within 10% of the identical specimens tested and the main factors e ffecting strength were the wi dth, edge distance, and fiber

PAGE 31

18 orientation with respect to the applied loa d. Similarly, these factor s have a significant effect on the failure mode of the specimens. It was determined that connection type B and D which had smaller edge distances of 38.1 mm (1.5 in.) resulted in cleavage failure and was characterized by crack propagation parallel to the applied loa d. For connection type E with the same edge distance the section ex perienced net tension fa ilure located at the inner bolt row without any crack propagation between bolt rows. For the specimens with larger edge distances rangi ng from 63.5 mm to 152.4 mm (2.5 in. to 6 in.) the resulting failure mode was net tension failure located at the inner row of bolts, which had the highest stress concentrations measured. All connections and tests s howed bearing damage after failure around the area adjacent to the loaded section of the hole. When studying the load distribution among the bolts it was determined the specimens with only one row of bolts (t ype B and type D) generally experienced equa l distribution of the load. However, this was not the case for connection type A, C, and E which had unequal forces at each bolt. Typically, for mechanically fastened joints with in-plane loading, bear ing failure is likely to occur as the width to diameter ratio (w /d) increases. This was not the case for this experimental investigation using composite fiber members as bearing failure did not occur; only localized bearing damage was apparent. Large widths reduced the ultimate net tensile capacity and were relatively lo w when compared to specimens with smaller widths; this can be attributed to the high st ress concentrations for specimens with larger widths. It is concluded that th e load of a bolted joint is resi sted by the material around the

PAGE 32

19 vicinity of the bolt hole and increasing the width of the speci men has no benefits in terms of strength efficiency. 2.10.4 Conclusions After the experimental investigation perf ormed by Hassan et al. (1997) the following concluding remarks can be made. All connections and specimens tested experienced linear behavior up to failure. The connections with only one row of bolts had equal load distribution whereas, connections with more than one row experienced unequal load distribution. The edge dist ance to diameter ratio, e/d has a significant impact on the failure mode for connections with either one row or one column of bolts. For small edge distances with e/d ratios less than 3 cleavage failure occurred. Large edge distances with e/d ratios larger than 3 experi enced tension failure. All conne ctions tested with 0, 45, and 90 fiber orientation had hi gher bearing strengths as the e/d ratio increased up to a ratio of 5. The ultimate capacities and bearing strengths for all connection types increased as the side distance to pitch s/p ratio increased up to a value of 1.2. Having a higher number of bolts is not directly proportional to the increase in ultimate strength capacity in the member. 2.11 Finite Element Modeling of Damage A ccumulation (Kermanidis et al. 2000) A study was performed by Kermanidis et al. (2000) to model the effects of damage accumulation using ANSYS, a finite element analysis program. A three-dimensional model was developed to simulate the progres sive damage and predict residual strength and stiffness in single-lap bolted composite joints under in-plane tensile loading.

PAGE 33

20 2.11.1 Progressive Damage Model The progressive damage model involves an iterative procedure to determine stress analysis, failure analysis, and material degradation. ANSYS has the ability to calculate the stresses between each ply to be used as the model progresses. Due to the complex nature of failure mechanisms in bolted composite laminates, many previous empirical methods are used as failure criteria. The fo llowing seven expressions (Shokrieh et al. 1996) below represent stress-based criteria to predict different failure modes. The modes of failure consist of matrix tensile and comp ressive failure, fiber tensile and compressive failure, fiber-matrix shear-out and delamina tion of fibers in te nsion and compression. Matrix tensile failure for ( y > 0): (2-4) Matrix compressive failure for ( y < 0): (2-5) Fiber tensile failure for ( x > 0): (2-6) Fiber compressive failure ( x < 0): (2-7) Fiber-matrix shear-out for ( x < 0):

PAGE 34

21 (2-8) Delamination in tension for ( z > 0): (2-9) Delamination in compression for ( z < 0): (2-10) Where ij are the layer-stress components in the ij direction and the denominators are the strengths in the corr esponding directions. Once ply failure occurs, the material propertie s are disabled from ca rrying a specific load, this is known as the degradation rule. Th erefore, each failure mode has its own corresponding degradation rule. The plyby-ply degradation rules are based on assumptions and empirical methods from constrai nts in the composite material properties. The material property degradation rules for failu re analysis in this study were taken from Shokrieh et al. (1996) an d were believed by Kermanidis et al. (2000) to be best suited for the failure criteria used. When matrix tensile and compressive failure is observed in a ply, th e assumption is that the matrix cannot sustain any load. Therefore, the material properties of the failed ply are reduced to: Ey = xy = 0

PAGE 35

22 (2-11) When fiber tensile and compressi ve failure is observed in a ply, the assumption is that the material cannot sustain any load in the vicin ity where the failure ha s occurred. Therefore, the material properties of th e failed play are reduced to: Ex = Ey = Ez = Gxy = Gyz = Gxz = xy = yz = xz = 0 (2-12) When fiber-matrix shear-out failure is obser ved in a ply, the assumption is that the material can only sustain load in the fiber and transverse to fiber directions. The material does not have the ability to car ry shear load, therefore, th e material properties of the failed ply are reduced to: Gxy = xy = 0 (2-13) Delamination failure in tension and compression affect the properties in the z-direction in the delaminated region. This results in the mate rial losing the ability to sustain any load in the z-direction in addition to shear loads. Therefore, the material properties of the failed ply are reduced to: Ez = Gyz = Gxz = yz = xz = 0 (2-14) A progressive model is developed in ANSYS using a programmed macro-routine iterative process. The macro-routine is described in the flowchart Figure 2.14 and involves the following steps. 1. Developing a FEM model of the composite jo int with initial material properties, specimen geometry, boundary conditions, initial load and load step.

PAGE 36

23 2. Determine the stresses for each ply by performing a non-linear analysis. 3. Perform a failure analysis by implementing the failure criteria. 4. Determine if ply failure exist. If no failure is calculated the applied load is increased. If any failure mode is recogni zed the program continues to the next step. 5. Apply appropriate material property degradation rules on the failed ply. 6. Determine if final failure is reached. If so, the program is completed. If final failure has not been reached the program returns to step 2 and the analysis is performed again with the same load to compute the redistributed stresses. Final failure is reached when the program c onverges. Convergence is assumed when no additional failures are detected. 7. Repeat procedure until final failure occurs. 2.11.2 Finite Element Modeling The study consisted of developi ng a finite element model of the bolted si ngle-lap joint using ANSYS. The geometry of the member is depicted in Figure 2.15. The model consists of a composite laminate upper pl ate with fibrous unidirectional layers (0/90/45) and an aluminum lower plate connected with a bolt. The model utilizes 8noded SOLID46 3-D ANSYS layered elements with three displacement DOFÂ’s per node. The solid element used for the composite material is defined by the orthotropic material properties, the fiber orientat ion, and layer thickness. For the interaction between two surfaces, 3-D CONTAC49 ANSYS elements were used to model the contact between two plates, between the bolt and the surface of the hole, and between the bolt head and

PAGE 37

24 the plate. See Figure 2.16 and Figure 2.17 for the finite element model. The loading conditions for the models consisted of in-plane tensile loading at th e ends as well as a pre-tension load at the fast ener by applying thermal expans ion properties in the axial direction of the bolt. To pr event secondary bending the mode l was supported laterally in the z-direction. 2.11.3 Results and Discussion The strains were measured experimentally w ith an 8 kN (1,798 lbs) tensile load applied to the lap joint. The strains were determined for the global x-direction at the angles of 0, 22.4 and 45 and are displayed in Figure 2.18. Comparisons were made to experimental and numerical values from a study perfor med by Ireman (1998). The experimental program consisted of installing strain gauge s around the hole at angles of 0, 22.4 and 45 to measure the response. The comparison of the numerical to experimental model resulted in sufficient accuracy. The material properties and strengths used for the composite material are show n in Table 2.3 and Table 2.4. The progressive damage model was subjected to incrementally increasing tensile loading in which the first failure (matrix compressive failur e) occurring at a load of 3 kN (674 lbs) near the stress concen trations around the hole. Befo re the first ply failure the composite material behaved elastically. The damage occurs at an angle of 45 with respect to loading and continue s to prorogate due to the pre ssure from the bolt. Figure 2.19 and Figure 2.20 illustrates the progressive damage at diff erent load steps predicted by the model. The shaded elements in the model represent fiber breakage, which

PAGE 38

25 represents the most critical failure mode in the model. Ultimate failure is reached when the composite laminate is unable to carry anymore load and is represented in Figure 2.19(c). The ultimate failure load in the mode l was determined to be 20.2 kN (4,541 lbs)

PAGE 39

26 Table 2.1. Comparison of flexural rigidi ty (Nagaraj and GangaRao 1997) Table 2.2. Comparison of shear rigidity (Nagaraj and GangaRao 1997)

PAGE 40

27 Table 2.3. Geometrical Data for the Investigated Geometry (Kermanidis et al. 2000) Table 2.4. Elastic Properties of HTA/6376 Ma terial (Kermanidis et al. 2000)

PAGE 41

28 Figure 2.1. FRP pultrusion process (Extren 2003) Figure 2.2. Modes of failure for bolted joints in FRP Composites (Hart-Smith 1987)

PAGE 42

29 Figure 2.3. Stress concentration relief in fibrous composites by delamination (Hart-Smith 1977) Figure 2.4. Relation between stress concentration f actors observed at failure of fibrous composite laminates predicted for perfectly elastic isotropic materials (Hart-Smith 1977)

PAGE 43

30 Figure 2.5. Terminology (Duthinh 2000) Figure 2.6. Influence of CFRP fiber proportion (0 / 45) on failure mode (Garbo and Ogonowski 1981)

PAGE 44

31 Figure 2.7. Effect of bolt torque on bearing strength of fibrous composite laminates (Hart-Smith 1987) Figure 2.8. Comparison between bearing strengths of connections with and without clamping pressure (Hart-Smith 1987)

PAGE 45

32 Figure 2.9. Effect of grouped 0 plies on bear ing strength (Garbo and Ogonowski 1981) Figure 2.10. Effects of stacking sequence on bear ing strength for GFRP laminates (Quinn and Mathews 1977)

PAGE 46

33 Figure 2.11. Variation of net tensile strength wi th width of 0/45 CFRP composites with 6.35 mm (0.25 in) hole (Collins 1987) Figure 2.12. Cross sections of test specimens (Nagaraj and GangaRao 1997)

PAGE 47

34 Figure 2.13. Connection configurati on (Hassan et al. 1997)

PAGE 48

35 Figure 2.14. Flowchart of the Progressive Damage Model (Kermanidis et al. 2000)

PAGE 49

36 Figure 2.15. Geometry of the Bolted Single-La p Joint (Kermanidis et al. 2000) Figure 2.16. Finite Element Model of Bolted Joint (Kermanidis et al. 2000)

PAGE 50

37 Figure 2.17. (a) Mesh of Area Around Hole (b) Fi nite Element Model of Protruding Head Bolt (Kermanidis et al. 2000) Figure 2.18. Comparison of Calculated Strains with Experimental and Numerical Results (Kermanidis et al. 2000)

PAGE 51

38 Figure 2.19. Illustration of Damage Propagation Pr edicted by the Present Model at Different Load Steps. Upper Surface of the Laminate (Kermanidis et al. 2000) Figure 2.20. Illustration of Damage Propagation Pr edicted by the Present Model at Different Load Steps. Lower Surface of the Laminate (Kermanidis et al. 2000)

PAGE 52

39 3. Experimental Program 3.1 Introduction The experimental program examined the axial an d transverse strength of GFRP structural channel members with bolted connections us ing various geometric configurations. The parameters studied included the effects of load cycles at various load levels on the ultimate strength and stiffness. In addition, the effect different flange geometries have on the ultimate strength when load ed in the axial and transverse direction. Phase I consisted of testing a total of 60 sp ecimens with monotonic loading applied in the axial and transverse directions to determine the ultimate strengths of each member (30 axial and 30 transverse). In addition, Phase I tested 10 thickened GFRP members 1 in. (25 mm) thick. Phase II consisted of testing specimens to failure after each specimen underwent unidirectional cyclic loading at various load levels. Phase III tested each member to failure after the specimen unde rwent multidirectional load cy cles in both the axial and transverse directions at different load leve ls. This chapter will discuss the fabrication process and materials used for the test sp ecimens, instrumentation, test setup, and experimental procedure. All ma terial testing was performed in the Structural Engineering Laboratory at the University of Colorado Denver. 3.2 Materials used for Test Specimens 3.2.1 GFRP Channels Extren GFRP structural shapes produced by Strongwell (Bristol, Virginia, USA) were used for the experimental testing. Extren is a product line manufactured by Strongwell which consists of more than 100 fiberglass st ructural shapes. The sh apes are made up of

PAGE 53

40 long glass fibers intertwined and bound with re sin to form a mat. In addition, the glass reinforcements consist of continuous strand rovings containing 800 4,000 fiber filaments in each strand. The resins used in Extren are isophthalic polyester and vinyl ester, which provide corrosion resistance and high strength properties. The mechanical properties of the GFRP members were dete rmined through tensile testing of GFRP coupons and using the data prov ided by the manufacturer. Channel structural shapes were selected for testing with dimensions of 3 ” x 1 ” x 3/16” (90 mm x 38 mm x 5 mm) and 1’-2” ( 356 mm) in length. Each channel member had a bolt hole with a diameter of 9/16” (14 mm) for loading purposes. The channel members consisted of three different geomet ric configurations with a portion of the flange cut near the bolt hole. Condition #1 involved the whole section with no flange modifications. For Condition #2 and #3 both th e top and bottom flange were cut at a length of 1 5/8” (41 mm) and 3 1/8” (79 mm ), respectively. See Figure 3.1 – Figure 3.3 for member dimensions with va rious flange configurations. 3.2.2 GFRP Flat Plates The GFRP material testing also included thic kened flat plate member s that were loaded axially to failure. The GFRP flat plates were in. (13 mm) thick w ith an additional in. thick plate epoxied at the end where loading occurs. The flat plates had 1 in. chamfers at the corners of both plates. Similar to the cha nnels, a hole with a diam eter of 1 9/16” (40 mm) was drilled thru both plates to install a bolt to be loaded. The plates are 3 in. (76

PAGE 54

41 mm) wide and the flat plate that is epoxied has a length of 4 in. (102 mm) See Figure 3.6 for dimensions of the GFRP flat plate. 3.3 Experimental Setup and Loading 3.3.1 Phase I and Phase II Setup for Axial Tests The tensile testing for each GFRP members c onsisted of applying a point load through the use of a steel bolt. An MTS closed-loop electro-hydraulic universal testing machine was used for the monotonic axial load applica tion. The loading was performed at a rate of 1 mm/min (0.0394 in./min) and wa s controlled by the displace ment of the actuator. The restraints of the member were provided by fixt ures that were connected to the specimen with three ” (13 mm) diameter bolts. Load was applied at a distance of 1 ” (38.1 mm) from the end of the specimen using a ” diam eter steel bolt attached to a fixture. The tensile testing was divided into three phases. Phase I consiste d of monotonic axial loading to ultimate failure of each channel specimen. Th e results of Phase I were used to establish various load levels by taking 25%, 50%, and 75 % of the ultimate failure to be used for Phase II testing. Phase II testing consisted of ax ial and transverse loading to failure after the specimens underwent 10 load cycles app lied to the specimen using different load levels mentioned above in the axial and tran sverse directions. Phase III consisted of loading to failure after the specimens were subject to multidirectional cyclic loading. See Figure 3.6 for the axial load e xperimental testing setup.

PAGE 55

42 3.3.2 Phase I and Phase II Setup for Transverse Tests The shear tests for each GFRP members were tested in a similar fashion by applying a point load through use of a st eel bolt in the transverse direction. The same testing equipment and loading rate of 1 mm/min from the tensile testing was used. Instead of utilizing fixtures with bo lted connections, a fixed suppor t was provided by gripping a portion of the channel web with clamps. The restraint required cut ting the bottom channel flange to allow for installation of the clamp. The same fixture from the tensile testing was used which applied a transverse concentrat ed load through the ” (13 mm) steel bolt. Similar to the axial test the transverse testing was divided into Phase I, Phase II, and Phase III, using the same load levels based on the ultimate failure to be used for the cyclic loading. See Figure 3.7 for the transv erse load experimental testing setup. 3.3.3 Phase III Testing Procedure The channel specimens in Phase III testing are subject to multidirectional loading, which requires the specimen to be loaded in the axia l direction at a certain load level followed by loading in the transverse direction at anothe r load level. This procedure constitutes one load cycle and is performed for a total 10 cy cles. Phase III requires the fixtures to be alternated for each load cycl e to accommodate for loading in the axial and transverse direction. This procedure is different from Phase II, which allows for consecutive unidirectional cyclic loading in one test for all 10 cycles.

PAGE 56

43 Figure 3.1. Condition #1 Figure 3.2. Condition #2

PAGE 57

44 Figure 3.3. Condition #3 Figure 3.4. Tension Test

PAGE 58

45 Figure 3.5. Shear Test Figure 3.6. Geometry of Thickened Flat Plate

PAGE 59

46 Figure 3.7. Axial Test Setup Figure 3.8. Transverse Test Setup Clamp Fixture for Transverse Loading

PAGE 60

47 Figure 3.9. Thickened Flat Plate Test Setup Figure 3.10. Thickened Flat Plate Test Setup

PAGE 61

48 4. Finite Element Modeling 4.1 Introduction A three-dimensional finite element model was developed to predict the behavior of the GFRP channel specimens using the structural analysis software ANSYS. The specimens were modeled using structural solid elements with orthotropic material properties. The model is a non-linear stress analysis incorpor ating an established failure criterion to determine stress, deflection, and predicted fa ilure. ANSYS uses two general modes for modeling, interactiv e and batch. The interactive mode a llows the user to use the GUI and other various tools to develop the model in the graphics window and allows for any modifications related to the analysis wher eas, the batch mode uses command files that have been created or previously generated fo r the analysis. The model created utilizes the interactive mode. This chapte r will discuss the development of the model along with the approach and methods used. 4.2 Modeling and Elements Preprocessor Building a model for the channel composite spec imen consists of several steps. An eightnode brick solid elements (SOLID185) was se lected to represent the GFRP channel member. This solid element has three tran slational degrees of freedom at each node. Orthotropic material properties are require d and are defined using values from the manufacturer. The material propertie s include the modul us of elasticity E in the x, y, and z directions, PoissonÂ’s ratio in the xy, yz, xz directi ons, and the shear modulus G in the xy, yz, xz directions, they are listed in Table 4.1. The channel geometry modeled has a depth of 3 in. (89 mm) a flange width of 1 in. (38 mm) and a thickness of 3/16 in

PAGE 62

49 (5 mm), the dimensions are shown in Figur e 3.1. A in. (13 mm) di ameter hole is added at a distance of 1 in. (38 mm) from the end of the member. Several types of constraints we re applied to the model. The edge of the channel member was constrained in all six degrees of freedom. To prevent lateral movement and secondary bending effects of the member, a la teral support was added in the z-direction located at the bolt. The Mesh Tool was used for automatic mesh generation of the model which is beneficial for the ease of mesh refinement. Both the channel and bolt were meshed using tetrahedral solid elements. Th e load was applied in the longitudinal xdirection using 25% and 75% of the ultimate tensile failure load determined from the experimental testing. The corresponding lo ad values are 584 lb s (2.6 kN) and 1,752 lbs (7.8 kN) respectively. Concentrated loads were applied over multiple nodes to replicate uniform pressure caused by loadi ng of the bolt, see Figure 4.2. 4.3 Failure Criteria The Tsai-Wu failure criterion used to predict the composite failure of the GFRP channel members. The Tsai-Wu failure criterion is a quadratic, interactive stress-based criterion that identifies failure, but doe s not distinguish between diffe rent failure modes (Tsai and Wu 1971). Failure occurs when the following condition is satisfied: (4-1) Where Fi and Fij are experimentally determined material strength parameters and i and j are the laminate stress in the fiber direction a nd the laminate stress transverse to the fiber

PAGE 63

50 direction, respectively. The failure criterion requires input va lues for material strengths which include the compressive, tensile, and sh ear strength of the composite in the x, y, and z directions. See Table 4.2 for the mate rial strength used in the analysis. 4.4 Model Results After the analysis was performed the re sults indicate a maxi mum deflection of 0.004 inches (0.1 mm) at the bolt hole. This value was based on an axial load of 584 lbs (2.6 kN) which was determined to be 25% of the ultimate load in the axial direction. When 75% of the ultimate axial load was applied of 1,752 lbs (7.8 kN) resulted in a maximum deflection of 0.012 inches (0.3 mm) at the same location adjacent to the bolt hole. These deflection values were signifi cantly different from the disp lacements measured through testing. The finite element model yielded va lues considerably smaller. Based on load testing of the channel specimens the average deflection fr om an applied axial load of 584 lbs and 1,752 lbs were approximately 0.025 inches (0.6 mm) and 0.0725 inches (1.8 mm) respectively. The difference can be attributed by the restraint conditions of the two. The ANSYS model uses a fixed restraint at the end of the channel whereas; the testing of the channel specimens used three bolts for the support conditions. This resulted in high stress concentrations at the bolt loca tions leading to a larger to tal displacement from each bolt hole. In addition, there may have been a ve ry small gap between the bolt and bolt hole surface in the three bolted c onnections used as a restraint which caused additional displacement. See Figure 4.2 and 4.4 for the deflection results from ANSYS.

PAGE 64

51 Table 4.1. Material Properties for GFRP Channel Specimens (Extren 2013) Table 4.2. Failure Criteria Strength Values E x E y EzGx y GxzG y zvx y vxzv y z(psi)(psi)(psi)(psi)(psi)(psi) 2,600,000800,000800,000425,000425,000n/a0.330.33n/a xyz Stress in Tension (psi) 30,0007,0007,000 Stress in Compression (psi) 30,00015,00015,000 xyyzxz Stress in Shear (psi) 4,5004,5004,500

PAGE 65

52 Figure 4.1. Channel Geometry of ANSYS Model Figure 4.2. Applied Loading of Model

PAGE 66

53 Figure 4.3. Displacement from 25% of Ultimate Load Figure 4.4. Tsai-Wu Strength Index – 25% of Ultimate Load

PAGE 67

54 5. Test Results and Discussion 5.1 Introduction This chapter presents the results obtained fr om material testing of the GFRP channel specimens. The testing consisted of loading the specimens in the axial and transverse direction. The behavior of the loaded specime ns along with the various failure modes are discussed in detail. The mechanical propert ies of the GFRP specimens are established from the material testing and are evaluated in terms of the load-displacement response. The effect of different geometric configuratio ns where the flange was cut near the applied load was studied. Stiffness degradation and ultima te strength from load cycles at different load levels are evaluated. The experimental program is divided into three phases. The primary objective of Phase I is to determine the ultimate strength of the GFRP specimens through monotonic loading in th e axial and transverse di rections. For Phase II the residual capacity of the connect ion is investigated with the specimens subjected to unidirectional cyclic loading at various load levels. Phase III involves investigating the residual capacity of the connection subjected to multidirectional cyclic loading at various load levels. The test results from all th ree phases will be discussed in the following sections. 5.2 Phase I Testing Results of GFRP Channels A total of 30 GFRP specimens were used fo r axial testing to determine ultimate load capacity and to generate load-displacement cu rves. For transverse testing, a total of 30 specimens were used as well. The GFRP sp ecimens have various flange geometry as discussed in Chapter 3. Table 5.1 displays the load capacity results for flange geometries

PAGE 68

55 01, 02, and 03 which represent Condition #1, Condition #2, and Condition #3 respectively, in the axial and transverse dire ctions. Ten specimens were tested for each flange geometry with monotonic loading to failure in both directions. The loaddisplacement curves for each test in Phase I is provided in Appendix A of this thesis. The load-displacement response indicates linear behavior until initial failure, followed by non-linear behavior until ultimate failure. Initia l failure is represented by the first peak on the load-displacement curve and the ultimate fa ilure is represented by the maximum load of the response. Figure 5.1 and Figure 5.2 displa ys a typical load-displacement curve for the GFRP specimens in the axial and transverse directions. For AXIAL 01 the average of the 10 specimens te sted had an initial failure load capacity and standard deviation of 2,214 lbs (9.8 kN ) and 162 lbs (0.72 kN) respectively. TRANS 01 the average initial failure load capacity and standard deviation was 2,601 lbs (11.6 kN) and 345 lbs (1.5 kN), respectively. For the subs equent testing, the lo ad levels of 25%, 50%, and 75% for Phase II and Phase III were obtained using the initial failure load from Phase I. AXIAL 01 and TRANS 01 was the fu ll channel section with no flange cut modifications made. AXIAL 02, AXIAL 03, TRANS 02, and TRANS 03 had different flange geometries as noted in Chapter 3. Base d on the final results there was no distinct trend and it can be concluded th at removing a portion of the fl ange in the channel section does not impact the overall performance of th e channel member, when loaded in the axial and transverse directions. The total average of all the specimens tested in the axial and transverse direction was 2,337 lbs (10.4 kN) and 2,588 lbs (11.5 kN) respectively. Therefore, the load levels used for Phase II and Phase III testing we re 584 lbs, 1,168 lbs,

PAGE 69

56 and 1,752 lbs (2.6 kN, 5.2 kN, and 7.8 kN) for ax ial load levels of 25%, 50%, and 75% and 647 lbs, 1,294 lbs, and 1941 lbs (2.9 kN, 5.8 kN, and 8.6 kN) for transverse loading. 5.2.1 Phase I Failure Modes of GFRP Channels Investigation of the failure m ode for the axial and transverse monotonic load test showed different failure mechanisms for the axial a nd transverse tests. For the test with the applied load in the axial direction initial b earing failure occurred at the contact zone between the bolt and composite specimen fo llowed by shear-out failure. As noted in Chapter 2 bearing failure occurs when excessive compressive stresses develop at the hole boundary surface. Shear-out failure is typically consequence of bearing failure with a short edge distance, e and can be characterized as a combination of in-plane and interlaminar shear failures. In itial bearing damage is followe d by a series of peaks on the load-displacement curve that represents damage accumulation with the eventual sudden drop in load carrying capacity. The test is completed when the specimen reaches this point. In the transverse direction th e specimen experienced initial bearing failure similar to the axial test followed by net tension splitting failure. This was evident by the crack propagation in the transverse di rection of the applied load. Th e behavior of the specimens loaded in the transverse direction differs from the axial tests; this is evident by the loaddeflection curve which indicates less progres sive damage before a sudden drop in load carrying capacity. The transverse load-displacem ent curves show less activity of damage accumulation with a smaller region of varying peaks after initial failure. Net tension

PAGE 70

57 failure is a function of both the joint geom etry and the strength of the GFRP specimen. See Figure 5.4 through Figure 5.6 for the axial an d transverse typical failure mode of the GFRP channel specimens. 5.2.2 Phase I Testing Results of GFRP Thickened Plates A total of 10 GFRP plate thickened member s were tested with monotonic loading to failure in the axial direction. The averag e failure load was 14,824 lbs (66 kN) with a standard deviation of 1,308 lbs (5.8 kN) As e xpected the failure load was significantly higher than the channel members due to the in creased thickness of 1 inch (25 mm) which allows for more area at the contact zone between the bolt and the specimen. Table 5.2 displays the failure load for each thickened sp ecimen with loading in the axial direction. The load-displacement curves indicate a li near response until ultimate failure. In comparison to the channel spec imen load-displacement curves in the axial direction, the thickened plates do not experience progressive failure which is evident by one peak on the curve as opposed to numerous peaks for the channel specimens. Figure 5.3 shows a typical load-displacement curve of the thickened GFRP plates. 5.2.3 Phase I Failure Mode of GFRP Thickened Plates Similar to the GFRP channel members the thic kened plates experienced shear-out failure adjacent to the bolt hole. Unlike the channel members which experienced initial bearing failure around the bolt hole, the thickened sp ecimens had no visible signs of bearing failure. This can be attributed to the fact the bearing strengt h capacity is larger than the shear-out capacity for the thickened specime ns. On three of the thickened specimens

PAGE 71

58 tested debonding of the plates occurred fo llowed immediately by shear-out failure due to the reduced cross section caused by the plates debonding. See Figure 5.7 for typical failure mode for the GFRP thickened memb ers with monotonic loading in the axial direction and Figure 5.8 for debonding of the plates. 5.3 Phase II Testing Results of GFRP Channels A total of eighteen specimens were tested for Phase II, nine specimens in the axial direction and nine specimens in the transverse direction. Each load level test (25%, 50%, and 75%) consisted of the same flange geomet ries used in Phase I, represented by 01, 02, and 03 in the axial and transv erse direction. Each specimen underwent ten unidirectional load cycles in the axial and transverse direc tion of the appropriate load level before being loaded to failure in the resp ective direction to determine the residual capacity of the connection. For the cyclic tests, the specimens were loaded to a specific predefined load than unloaded when that load was reached. This procedure was performed consecutively ten times. Figure 5.9 displays the load-displacement cu rves for the unidirect ional cyclic load applied for Phase II. Based on the cyclic lo ading curves it appears that there is no significant stiffness degradation as the slope of the curves is relatively constant, any decrease in stiffness is minor. The load-d isplacement response was similar to Phase I with apparent progressive damage and linear be havior up until initial fa ilure after the load cycles were completed and the specimens were loaded to failure. The load-displacement curves for the transverse loading to failure closely resembled Phase I testing as well. See

PAGE 72

59 Figure 5.10 and Figure 5.11 for typical load-d isplacement curves with monotonic loading in the axial and transverse direction for Phase II. The average residual load capacity of the thr ee specimens that experienced axial cyclic loading of 25%, 50%, and 75% of ultimate load were 3,000 lbs, 2,836 lbs, and 2,537 lbs, (13.3 kN, 12.6 kN, and 11.3 kN) resp ectively. Based on these values we can see a distinct trend where the ultimate failure load is redu ced as a higher load is applied for cyclic loading. This can be attributed to minor stiffness and strength degradation in the composite material due to repeated load eff ects. It is noted that the values above, the maximum load capacity was obtained from the load-displacement curves not the initial failure represented by the fi rst peak on the curve. Similarly, the average residual load capacity of the three specimens that experienced transverse cyclic loading of 25%, 50%, a nd 75% of the ultimate load were 3,120 lbs, 2,347 lbs, and 2,521 lbs, (13.9 kN, 10.4 kN, a nd 11.2 kN) respectivel y. The effect of cyclic loading in this case does not have a clear trend of decreasing strength with increasing load during the ten cy cles. Although, it is clear that the failure load of the 75% specimens decreased in comparison to the 25% specimens. Table 5.3 summarizes the testing results for the channel specimens subject to cyclic loading. 5.3.1 Phase II Failure Modes of GFRP Channels The failure modes for the GFRP channel spec imens were similar to the failure modes experienced in Phase I. The only differen ce came from minor damage propagation and

PAGE 73

60 displacements caused by the load cycles which are shown in Figure 5.12. When the specimens were tested in the axial direction to determine the residual capacity of the connection, they experienced initial bearing damage at the contact zone followed by shear-out failure in the region ad jacent to the bolt, similar to Phase I failure. Likewise, the transverse testing resulted in initial b earing damage around the bolt hole followed by tension splitting of the section. 5.4 Phase III Testing Results of GFRP Channels Phase III consisted of multidirectional cyclic load testing of nine specimens. Three specimens were used for each load level of 25%, 50%, and 75% of the ultimate load determined in Phase I. All nine specimens had the same geometric configuration, see Figure 5.13. On one side of the flange no modifications were made, however, the opposite side required cutting the flange to fit the clamp fixture necessary for the transverse loading test. As described in S ection 3.3.3 the multidirectional cyclic loading required alternating fixtures to load in both th e axial and transverse direction as one load cycle. After ten cycles of multidirectional loading the specimens were loaded to failure. Two of the three sets (01-25, 01-50, 01-75, 0225, 02-50, and 02-75) we re loaded axially to failure to determine the residual capacity in the connection. The last set (03-25, 03-50, and 03-75) was loaded to failure in the tran sverse direction to determine the residual connection capacity. The load-displacement curves for the multidir ectional cyclic loading are displayed in Figure 5.14. Cyclic loading in the axial or transverse direction is not performed

PAGE 74

61 consecutively as the curves show; the axial cy clic loading and transverse cyclic loading are alternated. Thus, each speci men is subject to a total of twenty cycles of applied loading at the appropriate level rather than a total of ten cycles in Phase II. As a result there is more damage accumulation and stiffness degradation as seen in Figure 5.14; however, it is very minor. The load-displacem ent curves in the axial and transverse directions to failure after multidirectional load cycles resemble the curves generated in from Phase I and Phase II. Recall from Ph ase I and Phase II, linear behavior up until initial failure from the first peak followed by damage accumulation resulting in ultimate failure. The average residual load capacity in the ax ial direction of the two specimens that experienced multidirectional cyclic loading of 25%, 50%, and 75% of the ultimate load were 2,826 lbs, 2,893 lbs, and 2,386 lbs (12.57 kN, 12.87 kN, and 10.61 kN) respectively. Based on these values it is appa rent that there was a significant reduction in strength from the specimens which were subject to 75% multidirectional loading. The values above were determined using the maximum load fr om the load-displacement curves which is not necessarily the initial failure represen ted by the first peak. However, the average residual connection capacity of the two sp ecimens based on initial damage that experienced multidirectional cyclic loading of 25%, 50%, and 75% of the ultimate load were 2,626 lbs, 2,321 lbs, and 1,894 lbs (11.68 kN, 10.32, and 8.42 kN) respectively. This shows significant strength degradati on with each increase in load level.

PAGE 75

62 For the residual connection capacity in the transverse direction after multidirectional cyclic loading only one set of specimens were tested. The connection capacity with cyclic loading of 25%, 50%, and 75% of the ulti mate load was 2,168 lbs, 2,595 lbs, and 2,231 lbs (9.64 kN, 11.54 kN, and 9.92 kN) respectively. The effect of cyclic loading in this case does not have a clear tre nd of decreasing strength with increasing load during the multidirectional load cycles. The same is observed using the residual capacity based on initial failure. Table 5.4 summarizes the testi ng results for the cha nnel specimens subject to multidirectional cyclic loading. 5.4.1 Phase III Failure Modes of GFRP Channels For Phase III, the failure modes from m onotonic axial and transverse testing after multidirectional load cycles were similar to Phase I and Phase II. The only difference came from displacements and damage caused fr om the repeated loads. For the 25% and 50% multidirectional load cycle tests the damage and displacements was minimal, however, for the 75% multidirectional load cycle tests the damage was more as shown in Figure 5.17. For failure in the axial directi on the specimen experien ced initial bearing failure followed by shear-out failure adjacent to the bolt hole. In the transverse direction, two of the three specimens experienced in itial bearing failure followed by tension splitting at the bolt hole as expected based on the previous test, di splayed in Figure 5.18 and Figure 5.19. However, for specimen 50-3 th e tension splitting was located just above the clamp and below the hole seen in Figure 5.20 which was unusual based on the previous failure modes from testing in the transverse direction. Recall, once the test shows a sudden drop in load carrying capacity the test is completed.

PAGE 76

63 5.5 Comparison of Phases The following section will compare the behavi or and results between each phase of the testing. Strength reduction, stiffness degrada tion, failure mechanisms, etc. will be discussed. 5.5.1 Comparison Among Different Fl ange Geometries in Phase I Figure 5.21 show the comparison of load-displacement curves for specimens tested in Phase I. The comparison shows the effect different flange geometries have on the connection strength. Based on the load values and observation of the curve it can be concluded that variation in flange geometri es has no significant e ffect on the behavior and ultimate strength of the connection in th e axial and transverse directions. However, it is noted that for the axial specimens the initia l failure load capacity increased with larger sections cut from the flange of the channel. 5.5.2 Comparison of Post-Cyclic Residual Behavior The comparison of post-cyclic residual behavi or is displayed in Figure 5.22. It is clear from the data and curves that multidirectiona l cyclic loading has a substantially higher impact on the residual behavior and strength of the connection compared to the residual behavior subject to unidirecti onal cyclic loading. In some instances the load capacity for specimens that underwent unidirectional cyclic loading was higher than the load capacity determined in Phase I which did not experien ce any cyclic loading. The curves show that the post-cyclic residual behavior from multid irectional cyclic loading generally had the smallest slope. In addition, the transverse specimens from Phase III had a longer region

PAGE 77

64 of damage accumulation with multiple peaks before failure compared to the transverse specimens from Phase I and Phase II. 5.5.3 Effect of Cyclic Loading on Residual Capacity The effect of cyclic loading on the residua l capacity for Phase II and Phase III is shown in Figure 5.23. The figure is divided into the cap acity at the first peak and the capacity at the maximum peak. Generally, the residual cap acities for Phase III experienced reduced connection capacity as the load level per cent of cyclic loading increased. The same cannot be said for Phase II as there was no cl ear trend apparent that the cyclic loading had on the residual capacity. It is noted, that fo r the transverse specimens in Phase II the trend was similar when comparing the capacity at the first peak to the capacity at the maximum peak. In addition, there was very sm all variation in the capacity between the three specimens tested for each load level resu lting in similar chart lines for each of the three specimens. 5.6 Design Recommendations This section describes discusses design reco mmendations for pultruded GFRP structural channel members with bolted connections s ubject to axial and transverse loading. 5.6.1 Monte-Carlo Simulation A Monte-Carlo simulation was performed to a ssist with developing design guidelines and recommendations for GFRP structural channel members. The simulation involves generating many random values to overcome limited experimental data. The algorithm

PAGE 78

65 used for the random sampling was the inve rse transformation method (Devroye 1986). For the simulation 10,000 samples were used for the calibration of the resistance factor. 5.6.2 Calibration of Resistance Factor The resistance factor for bolted GFRP ch annel members was determined using the following calibration method. The safety index is used to adjust the level of safety performance in structural members and is calculated by the following equation: (5-1) Where R and E are the mean capacity and mean instantaneous load applied, respectively, and R and E are the corresponding standard devia tions. The target safety index was set to 2.5, 3.0, and 3.5. The denominator of Equation 5-1 can be approximated as the following equation: (5-2) Where is the directional cosine or empirical constant to be determined. Combining equation 5-1 and 5-2 the following Load and Resistance Factor Design (LRFD) relationship is determined (Barker and Puckett 1997). (5-3) (5-4)

PAGE 79

66 (5-5) Where Rn and En are the nominal resistance and applied load, respectively, R and E are their bias factors, VR and VE are the coefficients of variations, and and are the resistance factor and lo ad factor, respectively. Table 5.5 summarizes the resi stance factor calibration fo r specimens in monotonic unidirectional loading. As exp ected as the safety index decreases the resistance factor increases. The resistance factor calibrati on indicates the transver se specimens yielded lower resistance factor value than the ax ial specimens, which is caused by higher coefficients of variation in the transverse specimens. The resistance factor calibration is divided into the first peak and maximum peak of the load-displacement curves. For design purposes resistance values determined fr om the first peak should be used as they are typically more conservative.

PAGE 80

67 Table 5.1. Phase I Ultimate Failure Load of GFRP Channel Specimens with Monotonic Loading in the Axial and Transverse Directions. Table 5.2. Phase I Ultimate Failure Load of GF RP Thickened Plate Members with Monotonic Loading in the Axial Direction. SpecimanCapacity (lbs)Max Capacity (lbs)SpecimanCapacity (lbs)Max Capacity (lbs)SpecimanCapacity (lbs)Max Capacity (lbs) AXIAL-01-120902493AXIAL-02-121182493AXIAL-03-121032711 AXIAL-01-222062789AXIAL-02-221672228AXIAL-03-223572839 AXIAL-01-320802520AXIAL-02-324733063AXIAL-03-322962920 AXIAL-01-421752650AXIAL-02-424582946AXIAL-03-424192575 AXIAL-01-519962556AXIAL-02-523023273AXIAL-03-520572650 AXIAL-01-624652571AXIAL-02-622862524AXIAL-03-631363140 AXIAL-01-723572465AXIAL-02-720482535AXIAL-03-726802960 AXIAL-01-821092508AXIAL-02-825012763AXIAL-03-822002403 AXIAL-01-922142381AXIAL-02-924762823AXIAL-03-926692714 AXIAL-01-1024552823AXIAL-02-1026373121AXIAL-03-1025692602 Average221525762347277724492751 Std Dev162140192330326216 SpecimanCapacity (lbs)Max Capacity (lbs)SpecimanCapacity (lbs)Max Capacity (lbs)SpecimanCapacity (lbs)Max Capacity (lbs) TRANS-01-120712161TRANS-02-120912100TRANS-03-127632772 TRANS-01-223942406TRANS-02-218132315TRANS-03-227542763 TRANS-01-326342667TRANS-02-325292564TRANS-03-327082738 TRANS-01-428522865TRANS-02-428292879TRANS-03-425862627 TRANS-01-527682776TRANS-02-525852617TRANS-03-522832305 TRANS-01-624762500TRANS-02-628742905TRANS-03-625462567 TRANS-01-723172338TRANS-02-724982517TRANS-03-726072645 TRANS-01-825382564TRANS-02-825872622TRANS-03-831523158 TRANS-01-933473362TRANS-02-928402844TRANS-03-925932616 TRANS-01-1026172649TRANS-02-1025412562TRANS-03-1024412461 Average260126292519259326432665 Std Devation345332336252230225 SpecimenCapacity (lbs) AXIAL-04-115655 AXIAL-04-214195 AXIAL-04-316229 AXIAL-04-416607 AXIAL-04-515021 AXIAL-04-613878 AXIAL-04-714711 AXIAL-04-813430 AXIAL-04-915931 AXIAL-04-1012579 Average14824 Std Dev1308

PAGE 81

68 Table 5.3. Phase II Residual Connection Capacity of Members Subject to Unidirectional Cyclic Loading Table 5.4. Phase III Residual Connection Capa city of Members Subject to Multidirectional Cyclic Loading SpecimenCapacity (lbs)Max Capacity (lbs)SpecimenCapacity (lbs)Max Capacity (lbs) AXIAL-01-2528352838TRANS-01-2529312934 AXIAL-02-2524623022TRANS-02-2531853191 AXIAL-03-2521613141TRANS-03-2532203236 AXIAL-01-5029342934TRANS-01-5020572362 AXIAL-02-5030163016TRANS-02-5020012285 AXIAL-03-5025582558TRANS-03-5020832394 AXIAL-01-7522952295TRANS-01-7522752523 AXIAL-02-7527402740TRANS-02-7525732574 AXIAL-03-7525752575TRANS-03-7521632465 Average-25 24863000 Average-25 31123120 Std Dev-25338153Std Dev-25158163 Average-50 28362836 Average-50 20472347 Std Dev-50225244Std Dev-504256 Average-75 25372537 Average-75 23372521 Std Dev-75225225Std Dev-7521255 SpecimenCapacity (lbs)Max Capacity (lbs)SpecimenCapacity (lbs)Max Capacity (lbs) MULTI-AXIAL 01-2527423103MULTI-TRANS 03-2519542168 MULTI-AXIAL 02-2525492549MULTI-TRANS 03-5021842595 MULTI-AXIAL 01-5025943129MULTI-TRANS 03-7520842231 MULTI-AXIAL 02-5020472656 MULTI-AXIAL 01-7519422774 MULTI-AXIAL 02-7518461997 Average-25 26462826 Std Dev-25136392 Average-50 23212893 Std Dev-50387334 Average-75 18942386 Std Dev-7568549

PAGE 82

69 Table 5.5. Resistance Factor Calibration for Specimens in Monotonic Unidirectio nal Load (Phase I) Where = mean (kN), = standard deviation, COV = coefficient of variance, = resistance factor Specimem Test Monte-Carlo simulation First peak Maximum peak First peak Maximum peak COV COV COV COV =3.5 AXIAL01 9.85 0.72 0.07 0.89 11.46 0.62 0.05 0.94 9.86 1.15 0.12 0.92 11.44 1.10 0.1 0.96 AXIAL02 10.44 0.86 0.08 0.86 12.35 1.47 0.12 0.74 10.33 1.18 0.11 0.89 11.71 1.25 0.11 0.83 AXIAL03 10.89 1.45 0.13 0.69 12.24 0.96 0.08 0.87 10.72 1.27 0.12 0.80 12.39 1.15 0.09 0.93 TRANS01 11.57 1.54 0.13 0.69 11.69 1.48 0.13 0.71 11.16 1.30 0.12 0.78 11.72 1.25 0.11 0.84 TRANS02 11.20 1.50 0.13 0.69 11.53 1.12 0.10 0.82 10.67 1.28 0.12 0.79 11.38 1.16 0.10 0.92 TRANS03 11.76 1.02 0.09 0.85 11.86 1.00 0.08 0.86 11.35 1.16 0.10 0.92 11.38 1.16 0.10 0.92 =3.0 AXIAL01 9.85 0.72 0.07 0.90 11.46 0.62 0.05 0.94 9.88 1.15 0.12 0.93 11.44 1.10 0.10 0.97 AXIAL02 10.44 0.86 0.08 0.88 12.35 1.47 0.12 0.78 10.35 1.18 0.11 0.91 11.74 1.25 0.11 0.86 AXIAL03 10.89 1.45 0.13 0.73 12.24 0.96 0.08 0.89 10.73 1.27 0.12 0.83 12.34 1.15 0.09 0.94 TRANS01 11.57 1.54 0.13 0.73 11.69 1.48 0.13 0.76 11.17 1.30 0.12 0.81 11.69 1.25 0.11 0.86 TRANS02 11.20 1.50 0.13 0.73 11.53 1.12 0.10 0.84 10.63 1.28 0.12 0.82 11.37 1.16 0.10 0.93 TRANS03 11.76 1.02 0.09 0.87 11.86 1.00 0.08 0.88 11.37 1.16 0.10 0.93 11.39 1.16 0.10 0.93 =2.5 AXIAL01 9.85 0.72 0.07 0.92 11.46 0.62 0.05 0.95 9.89 1.15 0.12 0.94 11.44 1.10 0.10 0.97 AXIAL02 10.44 0.86 0.08 0.90 12.35 1.47 0.12 0.82 10.35 1.18 0.11 0.92 11.72 1.25 0.11 0.88 AXIAL03 10.89 1.45 0.13 0.78 12.24 0.96 0.08 0.91 10.71 1.27 0.12 0.85 12.40 1.15 0.09 0.95 TRANS01 11.57 1.54 0.13 0.78 11.69 1.48 0.13 0.80 11.15 1.30 0.12 0.84 11.73 1.25 0.11 0.88 TRANS02 11.20 1.50 0.13 0.78 11.53 1.12 0.10 0.87 10.69 1.27 0.12 0.86 11.39 1.16 0.10 0.94 TRANS03 11.76 1.02 0.09 0.89 11.86 1.00 0.08 0.90 11.39 1.16 0.10 0.94 11.37 1.16 0.10 0.94

PAGE 83

70 Figure 5.1. Load-Displacement Curve for Phase I Axial Figure 5.2. Load-Displacement Curve for Phase I Transverse Loss of Load Carrying Capacity Loss of Load Carrying Capacity Linear Behavior

PAGE 84

71 Figure 5.3. Load-Displacement Curve for Phase I Thickened Member Axial Figure 5.4. Typical Failure Mode: Monotonic Axial Load Shear-Out Failure Loss of Load Carrying Capacity

PAGE 85

72 Figure 5.5. Typical Failure Mode: M onotonic Transverse Load (a) (b) Figure 5.6. Failure Mode: (a) Monotonic Axial Lo ad; (b) Monotonic Transverse Load Loading Direction Shear-Out Failure Loading Direction Tension Splitting Failure with Bearing Damage Tension Splitting Failure

PAGE 86

73 Figure 5.7. Typical Failure Mode of GFRP Th ickened Plate from Axial Loading Figure 5.8. Debonding of GFRP Thickened Plate Shear-Out Failure Debonding of Plate

PAGE 87

74 (a) (b) (c) (d) (e) (f) Figure. 5 9. Cyclic Unidirectional Load for Phase II Specimens: (a) axial load at 25% Pu; (b) axial load at 50% Pu; (c) axial load at 75% Pu; (d) transverse load at 25% Pu; (b) transverse load at 50% Pu; (c) transverse load at 75% Pu Figure 5.10. Load-Displacement Curve for Phase II Monotonic Axial Loading Initial Bearing Damage Maximum Capacity

PAGE 88

75 Figure 5.11. Load-Displacement Curve for Phase II Monotonic Transverse Loading (a) (b) (c) Figure 5.12. Damage Propagation with Load Cycles: (a) Axial Load at 25% Pu; (b) Axial Load at 50% Pu; (c) Axial Load at 75% Pu Initial Damage Failure Loadin g Direction Loadin g Direction Loadin g Direction

PAGE 89

76 Figure 5.13. Geometry of GFRP Channe l Specimens for Phase III (a) (b) (c) (d) (e) (f) Figure. 5 14. Cyclic Multidirectional Load for Phas e III Specimens: (a) axial load at 25% Pu; (b) axial load at 50% Pu; (c) axial load at 75% Pu; (d) transverse load at 25% Pu; (e) transverse load at 50% Pu; (f) transverse load at 75% Pu Flange Cut for Transverse Loading

PAGE 90

77 Figure 5.15. Load-Displacement Curve for Phase III Monotonic Axial Loading Figure 5.16. Load-Displacement Curve for Phase III Monotonic Transverse Loading Dama g e Accumulation Maximum Load Capacity

PAGE 91

78 (a) (b) (c) Figure 5.17. Damage Propagation with Load Cycl es: (a) multidirectional load 25% Pu; (b) multidirectional load at 50% Pu; (c) multidirectional load at 75% Pu Figure 5.18. Typical Failure Mode: Monotonic Ax ial Load for Phase III after 75% Pu Multidirectional Cyclic Loading Failure Loading Direction Shear-Out Failure

PAGE 92

79 Figure 5.19. Typical Failure Mode: Monotonic Tran sverse Load for Phase III after 75% Pu Multidirectional Cyclic Loading Figure 5.20. Failure Mode: Monotonic Transver se Load for Phase III after 75% Pu Multidirectional Cyclic Loading Failure Loading Direction Tension Splitting Failure Bearing Damage Tension Splitting Failure

PAGE 93

80 Figure 5.21. Load-displacement of Phase I-Compari son of Various Flange Geometry: (a) Axial load; (b) Transverse Load (a) (b) (c) (d) (e) (g) Figure 5.22. Comparison of Post-Cyclic Residual Behavior: (a) Axial Load at 25% Pu; (b) Axial Load at 50% Pu; (c) Axial Load at 75% Pu; (d) Transverse Load at 25% Pu; (e) Transverse Load at 50% Pu; (f) Transverse Load at 75% Pu

PAGE 94

81 (a) (b) (c) (d) Figure 5.23. Effect of Cyclic Loading on Residual Capacity: (a) Axial Load at 1st Peak; (b) Axial Load at Maximum Peak; (c) Transverse Load at 1st Peak; (d) Transverse Load at Maximum Peak Figure 5.24. Comparison of Strength Reduction Factor ( ) Based on Phase I

PAGE 95

82 6. Summary and Conclusion 6.1 Summary The connection capacity of Glass Fiber Reinfo rced Polymer (GFRP) structural channel members has been studied through experimental and analytical inves tigations. A total of 97 GFRP specimens were tested in the axia l and transverse directions to determine strength capacity and was divided into thr ee phases. Post-cyclic residual behavior was studied after unidirectional (Phase II) and mu ltidirectional (Phase III) cyclic loading was performed on the test specimens. The effect s of various parameters including various flange configurations, loading direction, and le vel of cyclic loading have been evaluated. The change in stiffness from the tests is also investigated due to unloading and reloading at various load levels. A GFRP channel structural channel member with a single bolted connection near the edge is vulnerable to premature failure cau sed by high stress concen trations leading to initial bearing failure. Bearing failure is an undesirable failure mode because it fails to utilize the entire material strength of the me mber. The proposed solution to this problem is to increase the number of bolts used for th e joint as well as adjusting the edge distance for optimal performance. 6.2 Conclusions The following conclusions are drawn from th e experimental study of GFRP structural channel members:

PAGE 96

83 1. The load-displacement curves indicate lin ear behavior up to initial failure when testing the GFRP channel members in the axial and transverse directions. 2. The initial failure mechanism for the bolte d connection results in initial bearing failure and is represented by the first peak on the load-displacement curve. 3. Progressive damage occurs and the sp ecimens exhibit inelastic deformation around the bolt hole and does not yield similar to steel. 4. Testing show that using different flange configuration which consists of cutting the flange at various degr ees, do not affect the overal l connection capacity in the GFRP specimens in both the axial and transverse directions. 5. Monotonic axial loading results in a two phase failure mechanism. Initial bearing failure is observed followed by shear-out failure which results in significant loss of load carrying capacity. 6. Monotonic transverse loadi ng also results in a two phase failure mechanism. Initial bearing failure first occurs followe d by net tension failure. The net tension failure is characterized by splitting in th e transverse direction of the load. The majority of the testing show the splitting occurring at the bolt hole with a couple of instance where the splitti ng occurred below the bolt ho le and directly above the fixture clamp. 7. The thickened GFRP members with an additional bonded plate exhibit significantly increase in strength capacity which is expected due to the increased area. The failure mode observed is debonding of the bonded plate followed immediately by shear-out failure. This is expected due to a sudden reduction in

PAGE 97

84 area. Initial bearing failure is not witn essed similar to all the other specimens tested. 8. Initial failure represented by the first peak on the load-displacement curve does not necessarily constitute the maximu m capacity. After initial failure the specimen can still achieve a higher maxi mum capacity which is represented by the maximum peak on the load-displacement curve. This occurs in both the axial and transverse directions. 9. After initial failure the specimens e xperience a period of damage accumulation which is represented by a region of multip le peaks in the load-displacement curve until final failure when the specimens lose s a significant amount of load-carrying capacity. 10. The load testing results do not indicate a reduction in strength capacity when comparing Phase II to Phase I. In some instances specimens from Phase II after unidirectional cyclic loading lead to higher connection capacities. 11. For Phase II axial loading the results i ndicate a reduction in residual strength capacity as the load level percent for cycl ic unidirectional loading is increased. 12. For Phase III axial load to failure, the results indicate a reduction in residual strength capacity as the load level percent for multidirectional loading is increased. 13. The residual connection capacity of Phas e III transverse specimens loaded to failure is significantly reduced afte r multidirectional cyclic loading when compared to Phase II and Phase I.

PAGE 98

85 14. The residual connection capacity of Phase III axial specimens loaded to failure has a small reduction after multidirectional cyclic loading when compared to only unidirectional load ing in Phase II. 15. The post-cyclic residual failure mode is the same as specimens from Phase I, which is shear-out and net tension failure The only difference is deflections and damage accumulation that occur from the cyclic loading. 16. Minor stiffness degradation occurs from multidirectional cyclic loading. 6.3 Recommendations for Future Work Additional research is require d to fully understand the beha vior of GFRP structural members with bolted connections. The follo wing aspects should be addressed: 1. Developing a finite element model to predict damage accumulation around the bolt hole. 2. Investigating GFRP connections with multip le bolts to fully optimize the material strength in the member. 3. Investigation of edge distance on GFRP c onnections to fully optimize the material strength capacity. 4. Evaluation of the stress concentrations experimentally and numerically around the bolt hole. 5. Investigation of simultaneous multidirec tional monotonic and cyclic loading on the connection capacity. 6. Generate and establish a complete design standard used for practicing structural engineers for bolted GFRP connection

PAGE 99

86 Appendix A. Load Displacement Curves for Phase I – Axial and Transverse

PAGE 100

87

PAGE 101

88

PAGE 102

89

PAGE 103

90

PAGE 104

91

PAGE 105

92

PAGE 106

93

PAGE 107

94

PAGE 108

95

PAGE 109

96

PAGE 110

97

PAGE 111

98 Appendix B. Load Displacement Curves for Phase II – Residual Capacity

PAGE 112

99

PAGE 113

100

PAGE 114

101 Appendix C. Load Displacement Curves for Phase III – Residual Capacity

PAGE 115

102

PAGE 116

103 Appendix D. Phase I Testing Failure Photos – Axial and Transverse Axial 01-01 Axial 01-02 Axial 01-03 Axial 01-04 Axial 01-05 Axial 01-06

PAGE 117

104 Axial 01-07 Axial 01-08 Axial 01-09 Axial 01-10 Axial 02-01 Axial 02-02 Axial 02-03 Axial 02-04

PAGE 118

105 Axial 02-05 Axial 02-06 Axial 02-07 Axial 02-08 Axial 02-09 Axial 02-10 Axial 03-01 Axial 03-02

PAGE 119

106 Axial 03-03 Axial 03-04 Axial 03-05 Axial 03-06 Axial 03-07 Axial 03-08 Axial 03-09 Axial 03-10

PAGE 120

107 Transverse 01-01 Transverse 01-02 Transverse 01-03 Transverse 01-04 Transverse 01-05 Transverse 01-06 Transverse 01-07 Transverse 01-08

PAGE 121

108 Transverse 01-09 Transverse 01-10 Transverse 02-01 Transverse 02-02 Transverse 02-03 Transverse 02-04 Transverse 02-05 Transverse 02-06

PAGE 122

109 Transverse 02-07 Transverse 02-08 Transverse 02-09 Transverse 02-10 Transverse 03-01 Transverse 03-02 Transverse 03-03 Transverse 03-04

PAGE 123

110 Transverse 03-05 Transverse 03-06 Transverse 03-07 Transverse 03-08 Transverse 03-09 Transverse 03-10 Axial 04-01 Axial 04-02

PAGE 124

111 Axial 04-03 Axial 04-04 Axial 04-05 Axial 04-05 Axial 04-06 Axial 04-06

PAGE 125

112 Axial 04-07 Axial 04-07 Axial 04-08 Axial 04-09 Axial 04-10

PAGE 126

113 Appendix E. Phase II Testing Photos – Damage From Cyclic Loading 75% Load Level Damage 75% Load Level Damage 75% Load Level Damage 75% Load Level Damage 75% Load Level Damage 50% Load Level Damage

PAGE 127

114 50% Load Level Damage 50% Load Level Damage 50% Load Level Damage 50% Load Level Damage 25% Load Level Damage 25% Load Level Damage

PAGE 128

115 25% Load Level Damage 25% Load Level Damage

PAGE 129

116 Appendix F. Phase III Testing Photos – Da mage From Multi Directional Cyclic Loading 25% Load Level Damage 25% Load Level Damage 25% Load Level Damage 25% Load Level Damage 50% Load Level Damage 50% Load Level Damage

PAGE 130

117 50% Load Level Damage 50% Load Level Damage 75% Load Level Damage 75% Load Level Damage 75% Load Level Damage 75% Load Level Damage

PAGE 131

118 Appendix G. Phase III Testing Failure Photos 25% Load Level Failure 25% Load Level Failure 50% Load Level Failure 50% Load Level Failure 75% Load Level Failure 75% Load Level Failure

PAGE 132

119 REFERENCES Bank, L. (2006). Composites for Construction John Wiley and Sons, Inc., Hoboken, New Jersey. Barker, R.M., and Puckett, J.A. (1997). Design of Highway Bridges based on AASHTO LRFD Bridge Design Specifications New York, NY, John Wiley & Sons, Inc. Collins, T.A. (1977). “The Strength of Bo lted Joints in Multi-Directional CFRP Laminates.” Composites, 8(1), 43-55. Collins, T.A. (1982). “On the Bearing Strength of Carbon Fibre-Reinforced Plastic Laminates.” Composites, 13(3), 241-252. Collins, T.A. (1987). “Experimentally Determined Strength of Mechanically Fastened Joints.” Joining of Fibre-re inforced Plastics, ed. Matthews, F.L., Elsevier. Deng, Q. (2004). Three-Point Bending Fatigue Life for Unidirectional GFRP Beams. Phd Dissertation, Southern Illi nois University, Carbondale. Devroye, L. (1986). Non-Uniform Random Variate Generation. New York, NY, Springer-Verlag. Duthinh, D. (2000). Connections of Fiber-Reinfor ced Polymer (FRP) Structural Members: A Review of the State of the Art, Gaithersburg, MD, National Institute of Science and Technology. GangaRao, V.S., and Nagaraj, V. (1997). “Sta tic Behavior of Pultruded GFRP Beams.” J. Compos. Constr., 1, 120-129. Garbo, S.P. and Ogonowski, J.M. (1981). AFWAL-TR-81-3041 Vol.1. Godwin, E.W., Matthews, F.L. and Kilty, P.F. (1982). “Strength of Multi-Bolt Joints in GFP.” Composites, 13(3), 268-272. Hart-Smith, L.J. (1980). “Mechanically-Fas tened Joints for Advanced Composites – Phenomenological Considerations and Simple Analyses.” Fibrous Composites in Structural Design, ed. E. M. Lenoe, D.W. Oplinger and J.J. Burke, Plenum Press, New York. Hassan, N., Mohamedien, M., and Rizkalla, S. (1997). “Multibolte d Joints for GFRP Structural Members.” ASCE Journal of Composites for Construction, 1, 3-9. Ireman, T. (1998). “Three-Dimensional St ress Analysis of a Bolted Single-Lap Composite Joint.”, Composite Structures, 43, 195-216.

PAGE 133

120 Jones, R.M. (1975). Mechanics of Composite Materials. Hemisphere Publishing Corp., Bristol, Pa. Kermanidis, T., Labeas, G., Tserpes, K.I., and Pantelakis, S. (2000). “Finite Element Modeling of Damage Accumulation in Bolted Composite Joints Under Incremental Tensile Loading.” European Congress on Computational Methods in Applied Sciences and Engineering, Barcelona, 11-14. Larsen, E.B. (2007). Pressure Bag Molding: Manufa cturing, Mechanical Testing, NonDestructive Evaluation, and Analysis. Master’s Thesis, Montana State University, Bozeman. Peimann, L.G. (1991). “Strength, Modulus of Elasticity, and Bond of Deformed FRP Rods.” Adv. Composites Mat. In Civ. Engrg. Struct., Proc., of th e Specialty Conf., Mat. Engrg. Div., ASCE, 99-110. Potter, R.T. (1978). “On the Mechanism of Tensile Fracture in Notched FRP.” Proc. R. Soc. London, A361, 325-341. Quinn, W.J. and Matthews, F.L. (1977). “Effect of Stacking Sequence on the Pin-Bearing Strength in Glass FibreReinforced Plastics.” J. Composite Materials, 11, 139145. Saadatmanesh, H., and Ehsani, M.R. (1991). “R/C Beams Strengthened with GFRP Plates: Experimental Study.” J. of Str. Engrg., ASCE, 117(10), 3417-3433. Shivakumar, K.N. and Crews, J.H. (1 982). Tech. Memo 83268, Jan., NASA, Langley, Va., USA. Shokrieh, M.M., Lessard, L.B. and Poon, C. (1996). “Three-Dimensional Progressive Failure Analysis of Pin/Bolt Loaded Composite Laminates.” Paper presented at the 83rd Meeting of the AGARD SMP on “Bolted Joints in Polymeric Composites”, Florence, Italy. Tibbets, A. (2008). Durable Fiber Reinforced Polymer Connections for Precast Concrete Structures. Master’s Thesis, University of Wisconsin, Madison. Tsai, S., and Wu, E. (1971). “A General Theo ry of Strength of An isotropic Materials.” Journal of Composite Materials, 5, 58-80. Kim, Y.J, Yoshitake, I., and Yang, M. (2013). “A Predictive Investigation Associated with Design Recommendations for CFRP -Confined Concrete in Aggressive Service Environments.” Construction and Building Materials, 43, 69-79.