DEVELOPMENT OF A NOVEL INSERT TO INCREASE SERVICE LIFE AND PERFORMANCE OF
BOLTED CONNECTIONS
By
Bryon J. Hall

A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
MASTER OF SCIENCE
Mechanical Engineering
2011

ABSTRACT

DEVELOPMENT OF A NOVEL INSERT TO INCREASE SERVICE LIFE AND PERFORMANCE OF
BOLTED CONNECTIONS
By
Bryon J. Hall

As composite materials take on a more prominent role in ground vehicles, there is an
increasing need to find efficient methods of joining composites to dissimilar materials; one
of the most common methods is to use a mechanical fastener within a thin metallic sleeve.
Studies have shown this method can decrease the strains in the vicinity of the bolted
connection. This study explores the use of new, novel insert in joining composite to
dissimilar materials. Instead of using a metallic insert, a standard 12.70 mm hex bolt is
modified to allow a liquid to be injected into the joint through the bolt. Once the material
has hardened, it creates a sleeve around the bolt, similar to the metallic insert. In this
study, single-lap joints were fabricated using the novel insert. These materials were tested
both quasi-statically and using ballistic-strain rates. The results were recorded using Digital
Image Correlation as well as several qualitative methods. The results show the novel insert
is preferable over a non-reinforced joint and does provide some benefits compared to a
metallic insert.

Copyright by
BRYON J. HALL
2011

TABLE OF CONTENTS

LIST OF FIGURES ............................................................................................................. VI
INTRODUCTION ................................................................................................................1
Purpose ....................................................................................................................................... 3
CHAPTER 1 – NOVEL INSERT BOLT DESIGN .......................................................................5
Previous Research....................................................................................................................... 5
Mold Design ................................................................................................................................ 9
Initial Injection Bolt Design ....................................................................................................... 12
Injection Method ...................................................................................................................... 13
Injection Bolt Design Qualitative Testing ................................................................................. 15
Injection Bolt Design Qualitative Testing Results ..................................................................... 17
Initial Injection Bolt Design Test and Results ........................................................................... 18
Injection Bolt Design Modifications ......................................................................................... 19
Chapter 1 Conclusions .............................................................................................................. 21
CHAPTER 2 – LOAD-DISPLACEMENT RESULTS OF INSERTS................................................ 23
Previous Research..................................................................................................................... 23
General Testing Procedure ....................................................................................................... 24
Baseline Results ........................................................................................................................ 28
Aluminum Insert Results .......................................................................................................... 31
Steel Insert Results ................................................................................................................... 33
Chapter 2 Conclusions .............................................................................................................. 35
CHAPTER 3 – DIGITAL IMAGE CORRELATION RESULTS OF INSERTS................................... 37
Digital Image Correlation Introduction .................................................................................... 37
Digital Image Correlation Configuration................................................................................... 38
Baseline Results ........................................................................................................................ 39
Machined Insert Results ........................................................................................................... 40
Injected Inserts ......................................................................................................................... 41
Strain Profiles............................................................................................................................ 43
Chapter 3 Conclusions .............................................................................................................. 48
CHAPTER 4 – PERFORMANCE OF INSERT MATERIALS UNDER BALLISTIC LOADS................ 50
Previous Research..................................................................................................................... 50
Panel Design ............................................................................................................................. 51
Determining Equivalent Thicknesses ........................................................................................ 53
Testing Procedure ..................................................................................................................... 55
iv

Testing Results .......................................................................................................................... 56
Chapter 4 Conclusions .............................................................................................................. 59
SUMMARY ..................................................................................................................... 61
Conclusions ............................................................................................................................... 61
Future Work.............................................................................................................................. 62
APPENDICES ................................................................................................................... 64
APPENDIX A .............................................................................................................................. 65
REFERENCES ................................................................................................................... 73

v

LIST OF FIGURES

Figure 0.1 – Schematic view of a single-lap bolted joint ................................................................ 1
Figure 0.2 – Exploded view of single-lap bolted joint with machined insert ................................. 2
Figure 0.3 – Schematic view of a single-lap bolted joint with an injected insert ........................... 2
Figure 1.1 – Strain parallel to load axis for some machined inserts. [2] ........................................ 6
Figure 1.2 – Strain perpendicular to load axis for some machined inserts. [2] .............................. 7
Figure 1.3 – Insert geometry for reinforcing thin composite plates. [3] ........................................ 8
Figure 1.4: Sample Square used for bolt design testing. .............................................................. 11
Figure 1.5 – Sample mold assembly with injected insert. ............................................................ 12
Figure 1.6 – Selected injection bolt designs ................................................................................. 13
Figure 1.7 – Injection mold after removing nut............................................................................ 16
Figure 1.8 – Bolt assembly with lower nut and washer removed. ............................................... 16
Figure 1.9 – Bolt assembly and injected insert pre-loaded to 2.82 N-m. ..................................... 18
Figure 1.10: Injection bolt failure with 3.18 mm wall thickness insert. ....................................... 19
Figure 1.11: Modified injection bolt schematic (left) compared to original design (right). ......... 20
Figure 2.1 – Plate dimensions for single-lap shear joint testing................................................... 26
Figure 2.2 – Side view of single-lap shear joint. ........................................................................... 27
Figure 2.3 – Mounting configuration for single-lap bolted configuration. ................................... 27
Figure 2.4 – Typical net tension failure of composite plate. ........................................................ 28
Figure 2.5 – Raw load-displacement data for baseline tests. ....................................................... 29
Figure 2.6 – Normalized load-displacement data for baseline tests. ........................................... 30
vi

Figure 2.7 – Failure region for normalized data for baseline tests............................................... 30
Figure 2.8– Normalized load-displacement data for aluminum inserts. ...................................... 32
Figure 2.9– Failure region of normalized data for aluminum inserts. .......................................... 32
Figure 2.10– Normalized load-displacement data for steel inserts. ............................................ 34
Figure 2.11– Failure region of normalized data for steel insert. .................................................. 34
Figure 3.3 – Legend for DIC derived line profiles.......................................................................... 44
Figure 3.4 – X-X Profile for single-lap bolted connection derived from DIC................................. 44
Figure 3.5 – X-Y profile for single-lap bolted connection derived from DIC. ................................ 45
Figure 3.6 – Y-X Profile for single-lap bolted connection derived from DIC. ................................ 46
Figure 3.7 – Y-Y profile for single-lap bolted connection derived from DIC. ................................ 47
Figure 4.1 – Completed test panel for ballistic loading. ............................................................... 52
Figure 4.1 – Sample panel for determining equivalent thicknesses............................................. 54
Figure 4.2 – Initial and final thicknesses for film adhesive samples. ............................................ 55
Figure 4.3 – Representative tile adhesion of polyurethane sample from Puente. [11] ............... 58
Figure 4.4 - Representative tile adhesion of film adhesive samples from Puente. [11] .............. 58
Figure A.1.1 - X-axis strains for non-reinforced at 30kN. ............................................................. 65
Figure A1.2 – Y-axis strains for non-reinforced at 30kN. .............................................................. 65
Figure A1.3 – X-axis strains for 3.18 mm 7075 aluminum at 30kN. ............................................. 66
Figure A1.4 – Y-axis Strains for 3.18 mm 7075 aluminum at 30kN. ............................................. 66
Figure A1.5 – X-axis strains for 3.18 mm 1144 steel at 30kN. ...................................................... 67
Figure A1.6 – Y-axis strains for 3.18 mm 1144 steel at 30kN. ...................................................... 67
Figure A1.7 – X-axis strains for 1.59 mm 7075 aluminum at 30kN. ............................................. 68
vii

Figure A1.8 – Y-axis strains for 1.59 mm 7075 aluminum at 30kN. ............................................. 68
Figure A1.9 – X-axis strains for 1.59 mm 1144 steel at 30kN. ...................................................... 69
Figure A1.10 – Y-axis strains for 1.59 mm 1144 steel at 30kN. .................................................... 69
Figure A1.11 – X-axis strains for 1.59 mm epoxy resin at 30kN. .................................................. 70
Figure A1.12 – Y-axis strains for 1.59 mm epoxy resin at 30kN. .................................................. 70
Figure A1.13 – X-axis strains for 3.18 mm polyurethane at 30kN. ............................................... 71
Figure A1.14 – Y-axis strains for 1.59 mm polyurethane at 30kN. ............................................... 71
Figure A1.15 – X-axis strains for 3.18 mm film adhesive at 30kN. ............................................... 72
Figure A1.16 – Y-axis strains for 1.59 mm film adhesive at 30kN. ............................................... 72

viii

INTRODUCTION
The purpose of this research is to investigate one possible method of increasing the failure load
of a single-lap bolted joint, specifically by adding an insert to the joint. For the purposes of this
paper, a single-lap bolted joint is a joint in which two materials are overlapped and are joined
together using a mechanical fastener. The mechanical fastener consists of a bolt, two washers,
and a nut consists of a bolt, two washers, and a nut; the washers are placed between the head
of the bolt and the composite panel and between the nut and the metallic panel. Figure 0.1
shows a sectional view of the single-lap shear joint.

Composite Panel
Aluminum

Mechanical Fastener

Figure 0.1 – Schematic view of a single-lap bolted joint
An insert is any material which is inserted into the joint and acts as a boundary between the
bolt and the outer diameter of the bolt. Prior researched has focused on machined inserts;
machined inserts are thin bushing, usually metallic in nature, inserted into the joint prior to
securing the panels by means of a mechanical fastener. A machined insert is fabricated so as to
be flush with the outer diameter of the bolt hole. The inside diameter is designed for minimal
clearance between the bolt and the machines insert. Machined inserts are typically designed
using hard materials such as aluminum or steel, although softer materials such as epoxies,
1

nylons, or other plastics can also be used. Figure 0.2 shows an exploded view of a single-lap
lastics
2
single
shear joint with a machined insert.

Figure 0.2 – Exploded view of single
single-lap bolted joint with machined insert
The intent of this research is to focus on the development of the novel, or injected
injected,
insert. The insert is formed by injecting a liquid into a fastened single lap bolted joint using a
single-lap
standard grade 8 hex bolt that has been modified by machining channels into the bolt. After
the liquid has cured it will form a soft insert around the bolt. Figure 0.3 shows a schematic view
of a single-lap bolted joint with an injected insert.
lap
Injection Channels
Washer
Composite Panel
Aluminum Panel

Joint Offset
Injected Insert

Washer
Nut
Injection Bolt

Figure 0.3 – Schematic view of a single lap bolted joint with an injected insert
single-lap

2

Purpose
The injected insert has several possible over a machined insert. A machined insert must be
machined to specifically match both the size and the geometry of the hole. Even machined
inserts intended for use on similar diameter or shape holes must be carefully machined to
ensure a tight fit between the insert and the inside diameter of the hole. Furthermore, a
machined insert requires the bolt hole to be aligned in every panel. An injected insert is
dependent only on the size of the bolt used in joining the objects. Once the size of the fastener
has been determined, an injected insert would be able to conform to a variety of hole sizes and
geometries. Additionally, as shown in Figure 0.3, an injected insert is able to accommodate an
offset within the bolt hole.
Metallic inserts can be susceptible to environmental factors. In addition to the clearance
between the bolt and the inside diameter of the machined insert, if any defects exist between
the outside diameter of the insert and the hole, a liquid or fine particle such as sand could
become entrapped within the joint, causing long-term damage. An injected insert would seal
the joint, preventing water or other liquids from becoming entrapped in the fastening system.
A machined insert can be susceptible to long term fatigue or vibration caused by poor
clearances between the bolt and the insert or between the insert and the outer diameter of the
hole. Additionally, if the joint is subjected to vibration, the nut could loosen, resulting in failure
of the joint. An injected insert conforms to the hole geometry and results in a tight fit
preventing damage caused by clearance within the fastening system. Additionally, an injected
insert creates a seal within the fastening system and can be designed to create a locking

3

mechanism between the nut and the bolt, reducing the possibility of prolonged vibration
loosening the bolt and causing failure.
Three possible disadvantages of using the injected insert have been identified and will
be investigated. The removal of material from the bolt to create the injection channels could
weaken the bolt and change the failure mode of the composite. The locking mechanism noted
earlier could make removal of the bolt from the joint more difficult, decreasing the reparability
of the joint. The material properties of a self-curing liquid, specifically the elastic modulus and
the shear modulus, will typically be lower than for a machined insert. This could result in a less
efficient distribution of strain.

4

Chapter 1 – Novel Insert Bolt Design
Chapter 1 will focus on the development of the bolt design for the injected insert. This chapter
will investigate different channel designs and how each channel design affects the quality of the
cured injected insert. This chapter will investigate different methods which can be used to
inject the liquid into the joint; it will consider the advantages and disadvantages several
different means of injection. The chapter will also introduce the standard testing procedure
which will be used throughout the research.
Previous Research
Investigation into the addition of an adhesively bonded metallic insert to a composite bolted
joint was conducted by Nillsson [1]. In his study, a FE analysis was performed using a 2-D FE
analysis with a width-diameter ratio of 4.0 and edge-diameter ratio of 2.5. A hole diameter of
10 mm was chosen for the composite only model and a diameter of 14 mm, with a insert wall
thickness of 2 mm for the bonded insert models. It was determined high compressive and
tensile forces existed respectively at 0 and 90 degrees from the direction of loading. The
addition of an aluminum and steel insert resulted in a stress concentration reduction of 40%
and 55%, respectively; experimental verification was performed using a (90/0/90/0/452/+452/90/0/-45/45)s 3.6mm laminate of similar dimensions loaded in dual shear using a
finger-tight bolt. These results showed an increase in failure of load of 30% and 55% while
using bonded steel and aluminum inserts, respectively. Unbonded inserts reduced the failure
load 10% to 20%, respectively, from the bonded inserts.
Herrera-Franco and Cloud [2] performed similar studies regarding strain relief inserts. In their
study, they used R1500/1581 3.55 mm thick laminate with an edge-diameter ratio >3, width5

diameter ratio >8, and hole diameter of 6.35 mm. They loaded the panel in double-shear
without inserts and using soft (photoelastic cement) and hard inserts (aluminum) with a 3.18
mm wall thickness. Using moiré fringe patterns, the results showed a significant reduction in
stress at an angle 0 degrees from the axis of loading. Figure 1.1 shows strains parallel to the
load axis for various types of machined inserts and Figure 1.2 shows the strain perpendicular to

800.0

the load axis for various types of machined inserts.

0.0

NO INSERT
PLASTIC INSERT
ALUMINUM INSERT AFTER GLUE FAILURE
ALUMINUM INSERT BEFORE GLUE FAILURE

y

-800.0

STRAIN ey

P

Dd
x
-1600.0

xa
EDGE OF PIN HOLE

-2400.0

INSERT LAMINATE INTERFACE

-15.2

-7.6

0.0

7.6

POSITION Y (mm)
Figure 1.1 – Strain parallel to load axis for some machined inserts. [2]

6

16.0

NO INSERT
PLASTIC INSERT
ALUMINUM INSERT AFTER GLUE FAILURE
ALUMINUM INSERT BEFORE GLUE FAILURE

14.0

y

12.0
TENSILE STRESS RATIO (σy/σo)

P
10.0
8.0
Dd

x

6.0
4.0

Locus of tensile
failure

2.0
0.0
Insert-Laminate Interface
0.0

0.5

1.0

1.5

2.0

y

( /d)
Figure 1.2 – Strain perpendicular to load axis for some machined inserts. [2]
Camanho et.al [3] conducted research into the addition of bonded metallic inserts to both
single-shear bolted joints as well as double-sheer bolted joins. Their research concluded that
although bonded metallic inserts reduce the stress concentrations in both types of joints,
double-shear bolted joints did not experience an increase in failure load. Furthermore, they
showed aluminum inserts are preferable to steel inserts because of lower developed tensile
stresses. Based on this research, using thin composite panels they investigated the effects of

7

using a tapered insert that protrudes from the joint. They tested three types of inserts: type
“A”, type “B”, and type “C”. A type “A” is shown in Figure 1.3. A type “B” is similar to type “A”
except the edges of the insert are not tapered. A type “C” insert is a traditional insert which
does not protrude from the joint. They found that, due to increased load transfer surfaces, a
Type “A” insert resulted in a 27% and 282% increase in failure load opposed to a Type “B” and
Type “C” insert, respectively.
A

45°
R
h
t
d

t

Figure 1.3 – Insert geometry for reinforcing thin composite plates. [3]
Herrington and Sabbaghian [4] investigated the effect of radial clearance on the failure strength
of thin composites. Using thin composites loaded in double-shear, they examined the failure
loads using washer to bolt-diameter ratios between 1.124 and 1.005. Their results showed a
significant correlation between high ratios and low ratios. A low ratio significantly increased
the failure load whereas, and despite high clamping loads, a high ratio resulted in a failure load
approaching that of a pin-loaded joint. There results concluded the increase in failure load due
of a pre-clamping force is due to both the increased frictional load on the plate as well as the
ability of a washer to prevent out-of-plane “brooming”.
Riccio and Marciano [5] investigated the effects of material and geometrical features on the
failure load of bolted composite joints. Using a composite plate 4.16mm thick loaded in single
shear, they compared the effects of hole diameter, type of bolt (countersunk vs. protruding),
8

and joint type (composite-composite vs. composite-aluminum) against the failure load of each
combination. Based on their experiments, they concluded that although bolt type does not
affect failure load, protruding bolts withstand higher loads prior to damage whereas
countersunk bolts experience greater displacement prior to failure. They also found that
whereas aluminum-composite interfaces have lower failure loads than composite-composite
interfaces, composite-aluminum interfaces result in a more uniform distribution of damage
within the composite plate.
Erikkson [6] investigated the effects of lateral constraints, ply orientation, hole diameter and
thickness and loading direction on the bearing strength of laminated composited. They used
brittle graphite-epoxy T300/914C and toughened graphite-epoxy HTA7/6376 available from
Ciba-Geigy. The results were consistent for each composite type and were as follows. Lateral
constraints increase the strength because of load transfer through friction forces. Without a
constraint, the damage will occur at the bolt-composite interface; with a constraint, the
damage will occur at the outer diameter of the constraint. Ploy orientation has little effect on
the strength if the majority of the plies are oriented at 0-degree or + 45-degrees. Increasing the
hole diameter or increasing the thickness of the composite were found to slightly increase the
bearing strength. The loading direction (tension or compression) was found to have little effect
on the bearing strength of the composite joint.
Mold Design
Before testing any injection bolt design in an actual single-lap bolted joint it was
necessary to determine the effectiveness of the injection bolt design. An effective injection
bolt design should result in an injected insert that conforms fully to the geometry of the hold
9

and is free of any significant defects such as air entrapped within the insert. The optimal
injection bolt design should also require the removal of minimal material from the bolt.
This test required the design of a mold that allow for qualitative observation of the
injected insert without affecting the insert itself. If a standard single-lap bolted joint were used,
the process needed to remove the fastening assembly, which includes the bolt, washers, nut,
and injected insert, would require the removal of the nut and washer and would likely result in
damage to the injected insert. A break-away mold that would replicate the conditions of the
single-lap bolted joint and allow for removal of the bolt assembly without causing damage was
designed for this experiment. Note: typically all dimensions were performed using the English
system of measurement. However, for consistency only the SI values are shown
Each mold consisted of a square 76.20 mm in length; this length allowed for easy
handling and reduced the amount of excess material. A hole 19.05 mm in diameter was drilled
through the center of each square. This diameter lessened the tolerances necessary when
centering the injection bolt within the hole and created an injected insert equivalent to a
machined insert with a 3.18 mm wall thickness.
Each square was cut down through the center of one edge. This cut allowed for the
removal of the bolt assembly without causing damage to the injection insert. This split also
eliminated the need to removal any part of the bolt assembly. Transparent tape was placed on
the exposed edges resulting from the split. The tape acted as a compressible gasket that
prevented the liquid from entering the boundary between the adjoining edges of the mold.
The tape was also placed on the outer diameter of the bolt hole. This tape resulted in an
injected insert that was more transparent than those created without using tape to smooth the
10

edges of the bolt hole. Figure 1.4 shows one square of a mold with some of the features
highlighted.

(1)

(2)

(3)

sed
Figure 1.4: Sample Square used for bolt design testing.
The mold consists of two squares, a top half and a bottom half. Each square was 12.7
mm thick and fabricated using the above specifications; this thickness was chosen to mimic the
thickness of the composite and aluminum plates used in the actual single lap bolted joints. The
ckness
single-lap
two squares were stacked on each other, with the cut edges aligned parallel to each other. The
mold was fabricated using an optically clear high
high-strength polycarbonate; this provided a cheap
bonate;
lightweight alternative to steel and aluminum and allowed for limited visual observation of the
boundary layer between the top and bottom half of the mold. The four pieces of the mold
were secured using a commonly available q
quick-clamp. A completed mold with injection bolt is
clamp.
shown in Figure 1.5.

11

(1)
(2)

(3)

Figure 1.5 – Sample mold assembly with injected insert.
ssembly
Three important features of the completed mold assembly are shown in Figure 1.5. (1) The
assembly
material is injected into the bolted through an opening located on the head of the bolt. (2)
Centerlines on the mold and the washer allowed the bolt to be nominally centered within the
hole providing a uniform insert thickness for the tests. (3) The quick clamp provided a quick
(3)
method for securing the mold during the curing phase. The 76.2 mm length of the mold was
ideal for use with the quick clamp.
Initial Injection Bolt Design
Having developed a mold capable of allowing visual observation of the injected insert without
f
damaging the injected insert or requiring the removal of any portion of the bolt assembly, it
was necessary to test the effectiveness of different bolt channel designs. Two different
parameters were chosen as variable for the design of the injection bolt; channel diameter and
variables
e
channel quantity. Using the parameters, three different designs were initially developed. The
developed
first design uses two thru-channels; thru channels span the entire diameter of the bolt. The
channels;

12

channels are perpendicular to each other and are equally spaced through the 25.4 mm length
corresponding to the length of the bolt contained within the bolt hole. The second design
contains one thru-channel; centrally located within the 25.4 mm length. The third design
contains two half-channels; half channels extend radially from the center of the bolt to the
outside of the bolt. The channels are parallel to each other and are equally spaced through the
25.4 mm length. Figure 1.6 shows a schematic view of these three channel designs. The
channels of all three designs are 3.18 mm in diameter. The uppermost 6.35 mm of the vertical
injection channel was been enlarged to 4.76 mm to accommodate the injection device. Once
the liquid been injected into the joint, the injector can be removed from the bolt and a steel rod
inserted into the vertical injection channel. This rod serves to replace some of the material
removed during fabrication of the injection bolt as well as prevent backflow of the liquid during
the curing process.

Figure 1.6 – Selected injection bolt designs
Injection Method
Three different methods were considered for the injection method: a human-powered injector,
an air-powered injector, and an electrically powered injector. The advantages and

13

disadvantages of each system were analyzed to determine the most effective injection method
for these experiments as well as the best overall method.
The hand powered injector uses disposable syringes with tapered nozzles. The shape of the
nozzle allows the syringe to be directly inserted into the injection channel as noted in the
injection bolt design dimensions, eliminating the requirement of an intermediary tube. These
syringes are available in sizes suitable for the experiments and are readily available at retail
stores. However, these syringes have a limited injection pressure and do not have a repeatable
method of controlling the injection rate.
The air-powered method uses disposable syringes compatible with shop air, typically 690 kPa.
The shape of the nozzle allows the syringe to be inserted directly into the injection channel,
eliminating the need of an intermediary connecting tube. These syringes are also available in
sizes suitable for these experiments and provide a constant pressure of injection. However, this
method requires standard 12VAC power and a supply 690 KPa compressed air. Furthermore,
the system requires a significant, onetime expense for initial setup and does not provide an
adequate method to prevent excessive spillage of the liquid.
The electrically powered method uses an electrically powered motor to drive the injection of
the liquid. As the system relies on a variable-speed battery driven motor, the rate of injection is
controllable and the system is portable. However, currently available systems are designed for
reservoir volumes of 300 ml or larger. A single injection bolt requires less than 6 ml of liquid.
This system also requires a significant onetime expense prior for initial setup.

14

Based on the advantages and disadvantages of each system above, a disposable 30 ml syringe
with a tapered nozzle was chosen as the best option for the experiments conducted for this
research. The low cost, ease of use and availability of the components for this system
outweighed the lack of control of either the injection rate or the injection pressure. However, if
a higher rate of production was required, the electrically powered device would be the best
method of the three listed. The reservoir size is suitable for quickly filling multiple injection
bolts and the electrically driven motor provides a controlled and repeatable injection rate and
pressure. Furthermore, the one-time cost required is diminished as the volume of the
production increases.
Injection Bolt Design Qualitative Testing
Tests were performed to validate the mold design and concept of injecting a liquid into a singlelap bolted joint through an injection bolt. Standard grade 8 bolts, 63.5 mm in length with a
diameter of 12.7 mm, with varying channel designs and channel diameters of 4.76 mm were
fastened in the molds using washers with a 34.93 mm outer diameter and 2.54 mm thickness
placed between the mold and nut and between the mold and the bolt head. Epoxy resin was
injected into each sample using a disposable 30 ml syringe with tapered nozzle. The injection
insert was allowed to cure for 24 hours. After 24 hours, the mold was removed from the quick
clamp and visual observations were used to qualitatively estimate each configuration.
Figure1.7 and Figure 1.8 show some observations

15

(1)

(2)
Figure 1.7 – Injection mold after removing n
nut.
Two important features of the injection mold after removing the nut are highlighted in Figure
are
1.7. (1) Epoxy resin infiltrated the space between the lower washer and the bol between the
bolt,
lower washer and nut and (already removed) and between the nut and the bolt. (2) Epoxy
infiltrated the boundary between the upper and lower halves of the mold. This leakage could
be used to create a hybrid adhesive
adhesive-mechanically fastened joint.

(1)

(2)

Figure 1.8 – Bolt assembly with lower nut and washer removed.
ith
Two important features of the bolt
bolt-insert assembly are highlighted in Figure 1.8. (1) Significant
1.8
air voids were observed near the interface between the insert and the lower washer (
(already
removed). (2) Epoxy infiltrated the boundary between the top washer and top half of the mold.

16

Injection Bolt Design Qualitative Testing Results
The purpose of the qualitative testing was to determine whether an injection bolt could
be used to effectively inject a liquid into a single-lap bolted joint and to determine what factors
would significantly affect the quality and consistency of the injected insert. Based on these
tests, which used a thin epoxy resin with a medium rate hardener, neither the channel design
nor the channel diameter had any noticeable effect on the quality and consistency of the
injected insert. Every test performed using these two parameters as variables ended with
similar results. Every injected insert was of similar quality and consistency. Furthermore, every
injected insert had air voids, similar in size and quantity to those shown in Figure 1.8 above, at
the location in the injected insert which had the highest elevation as the injected insert was
curing.
During the testing a third parameter was observed that did affect the quality and
consistency of the insert. During the initial tests, the bolt assembly was hand-tightened,
producing a minimal pre-load on the assembly. These samples experiences a significant
number of air voids. Additional tests were conducted using bolts that were tightened to 2.82
N-m by means of a torque wrench. These samples had a significantly reduced number of air
voids. Figure 1.9 shows a sample tightened to 2.82 N-m.

17

Figure 1.9 – Bolt assembly and injected insert p
pre-loaded to 2.82 N-m.
Initial Injection Bolt Design Test and Results
After identifying the parameters that most significantly affected the quality and consistency of
the injected insert, it was necessary to determine whether the material removal of the bolt
would affect the failure mode of the single
single-lap bolted joint. Using the results from the
validation testing above, a single thru channel bolt, see Figure 1.6 (center figure) with 3.18 mm
diameter channels was chosen for the initial testing. The center of the injection channel was in
the center of the 25.4 mm length corresponding to the length of the bolt contained within the
ngth
bolt hole. Two samples were prepared using this injection bolt. One sample used a single
.
single-lap
bolted joint with an hole diameter of 15.88 mm, corresponding to an insert wall thickness of
,
1.59 mm. The second sample used a single
.
single-lap bolted joint with a hole diameter of 19.05 mm,
corresponding to an insert wall thickness of 3.18 mm. Both samples were tested to failure
using a controlled displacement rate of 1mm/min to determine the effect of removing material
from the injection bolt.

18

Previous research states that thick composites will fail under net tension and as such these
samples were expected to fail by net tension. However neither sample failed through net
However,
tension of the composite. The sample with the3.18 mm wall thickness insert failed by
e
catastrophic shear within the bolt at the plane corresponding to the center of the injection
channel. Figure 1.10 shows the failed injection bolt. The sample with the 1.59 mm wall
.
thickness insert was stopped after a brittle failure, presumably in the injected insert, was
failure,
audibly observed during the test. Upon inspection of the bolt assembly, cracks were observed
within the injected insert.

.
Figure 1.10: Injection bolt failure with 3.18 mm wall thickness insert.
ailure
Injection Bolt Design Modifications
Based on the unexpected failure of the sample by shear failure of the bolt, it was necessary to
reinvestigate the design of the bolt to improve the strength of the bolt. Upon investigation,
three factors were identified which could contribute to increased strain concentrations in the
actors
bolt which contributed to the shear failure observed within the injection bolt. (1) A decreased
cross-sectional area of the bolt resulting from material removal from the bolt to accommodate
sectional
the
19

the injection channels. (2) The position of the injection channel was at a location corresponding
to the boundary plane between the plates; this plane is also the plane of greatest shear strain
within the bolt. (3) The injection channel was positioned at a location in the threaded portion
of the bolt. With these observations, a modified design was proposed.
The purpose of the modified injection bolt design was to minimize the effect of the three
conditions listed above. The modified injection bolt design accomplishes this goal by the
following changes. (1) The modified design uses a single-half channel opposed to a single thruchannel to minimize the material removal from the bolt. (2) The design moves the plane of
minimal cross section (the horizontal injection channel) away from the plane corresponding to
boundary between the aluminum plate and the composite plate. (3) The design moves the
injection channel to an unthreaded portion of the bolt, minimizing the effect of thread strain
concentrations on the failure of the bolt. (4) The design reduced the injection channel diameter
from 3.18 mm to 2.38 mm. Figure 1.11 shows a schematic view of the modified bolt design
compared to the original design. Using the modified design, two additional tests were run
using the testing procedure above. Both tests ended by net tension failure within the
composite, validating the modified design of the injection bolt.

Figure 1.11: Modified injection bolt schematic (left) compared to original design (right).
20

Chapter 1 Conclusions
Chapter 1 focused on the initial and design of the injected insert. The novel insert was
proposed as a novel method of reinforcing the bolted connection of a composite plate.
Opposed to the more traditional method of using a pre-machined metallic insert to reinforce
the bolted connection, the novel insert uses a liquid which is injected into the bolted
connection by means of a specially designed bolt and is then allowed to cure.
However, before the performance of the injected inserts could be tested under different
loading conditions, it was necessary to verify the feasibility of such a design. A bolt had to be
created which could be used to inject the liquid into the bolted connection, a method had to be
designed which could test the effectiveness of different bolt designs, a method to inject the
liquid into the bolt had to be selected, and the bolt had to be tested to ensure the alterations to
the bolt would not change the failure mode from composite failure to bolt failure.
The bolt design used a standard 12.70 mm grade 8 bolt; a hole was drilled from the
head downward, and channels radiating from this central hole were used to inject the liquid
into the bolted connection. Similarly, a break-away mold, which could be removed after the
insert had cured, was used to allow evaluation of the quality of the insert. Through testing with
multiple channel configurations and diameters, the research showed the design was insensitive
to either the channel configuration or the channel diameters. The only factor which noticeably
affected the quality of the insert was the preload applied to the bolt before injecting the liquid;
higher preloads resulted in cured inserts with fewer air voids than inserts created with lower
initial preloads.

21

Three possible injection methods were identified in this research: human powered, air
powered, and electrically powered. The human powered methods proved the most cost
effective, being low-cost, disposable, and available in a large variety of reservoir sizes. The
most noticeable disadvantage is the lack of pressure which can be applied using a human
powered applicator. The air powered method was similar to the hand-powered methods as
many of the applicators can be adapted for pneumatic injection; however, this method was not
cost-effective as it required a significant one-time cost to purchase the necessary equipment.
In regards to a high-volume production, the electrically powered method was the most
efficient, having larger capacity reservoirs as well as well as precise control over volume and
rate of injection. However, for these experiments this method was also not cost-effective.
Testing of the bolt design in a single-lap bolted connection showed a significant design
flaw. The material removal of the bolt, coupled with several strain concentrations factors,
caused the bolt to catastrophically fail in shear prior to failure of the composite. A redesign of
the bolt, which decreased the channel diameter, moved the injection channel away from the
plane corresponding to the boundary between the aluminum plate and the composite plate,
and moved the injection channel to the unthreaded portion of the bolt, sufficiently reduced the
stress concentrations within the bolt such that the single-lap joint once again failed by net
tension of the composite plate.

22

Chapter 2 – Load-Displacement Results of Inserts
Chapter 2 will focus on the use of load and displacement data to examine the performance of
metallic-composite joints mechanically fastened using a single-lap shear joint. This chapter will
compare two different types of single-lap shear joints: non-reinforced joints and joints
reinforced by means of a machined insert. Due to the limitations of force-displacement
measurement, injected inserts were not analyzed using this method. The test will examine the
results of joints reinforced with steel and aluminum inserts to those of non-reinforced singlelap joints. The chapter will consider the performance of the inserts at two different wall
thicknesses.
Previous Research
Ekh [7] examines the secondary bending caused by attaching carbon fiber and epoxy resin
composites to aluminum via a multi-fastener single-lap bolted joint. Ekh seeks to develop a
method to measure the bending using a contact-free method. They use a [+45/0/90] 4S
composite with a ply thickness of 0.13mm The aluminum is 8mm thick and is fabricated from
AA7475-T76, and the bolts from titanium. Tensile force is applied to the sample by means of a
hydraulic tester. Measurements are obtained using a two-camera system utilizing digital
speckle photography. Both cameras were aligned in a plane parallel to the loading direction.
When all fasteners were equally tightened, the results showed the primary deflection occurred
after the fourth and last fastener. The results also showed the overlap as well as the pre-torque
applied to the fasteners will affect both the location and magnitude of the maximum curvature.
McCarthy [8] investigated the load distribution around the fasteners in a multi fastener
single-lap shear joint undergoing a tensile load. McCarthy applied the spring model to examine
23

the load distribution in a multi fastener joint. The spring model is used to account for the
clearance of a particular fastener as well as an unequal load distribution. The research uses this
method to analyze a [45/0/-45/90]5S using 8 mm titanium bolts as well as steel nuts and
washers. The results show that for both single-lap shear joints as well as double-lap shear
joints, the results of the spring model reasonably agree with the results of the spring model.
Using this agree, a parametric study was conducted which showed that compared to plate
thickness and bolt pitch, the clearance of the bolts was the most significant aspect in affect the
load distribution of the sample joint.
Khashaba [9] investigated the stress concentrations in the vicinity of a bolted composite
joint using both varying washer sized and pre-torques. Khashaba used a [0/+45/90]S glass fiber
composite fabricated by hand lay-up. The specimen was 135mm by 36mm with a 6mm hole
drilled 18mm from one end. The experiment considered both the outer diameter of the washer
used as well as the pre-torque applied to the bolt before the specimen was tested. The results
showed with a constant pre-torque, even though each sample failed at an approximately equal
load, the samples with larger washers failed at a greater displacement. When the washer size
was held constant and the pre-torque was varied, the results showed the samples with lower
pre-torques not only failed at lower loads but also experienced several knees, partial internal
failures prior to ultimate failure.
General Testing Procedure
Every test using the application of uniaxial tension was carried out in a similar fashion
with minor deviations depending on the type of test performed. Every test was performed on a

24

Material Testing System 810 (MTS810) with a maximum applicable load of 100kN. The load was
applied by a controlled cross-head rate of 1mm/min.
Every test was performed using a single-lap shear joint. Each joint was fabricated using
two similarly sized plates joined together using a single 50.80 mm x 12.70 mm bolt with
associated nut and washers. Every bolt was a standard grade 8 zinc-plated steel hex bolt with
minimum tensile strength of 150 ksi, threaded length of approximately 31.75 mm. Similarly,
the nuts were standard grade 8 zinc-plated steel hex nuts with a height of 11.11 mm. The
washers were grade 8 zinc-plated high strength steel with inside diameter of 14.30 mm, outside
diameter of 34.93 mm, and height between 2.29 mm and 4.57 mm.
Every plate was fabricated using a similar process. Every plate was 203.20 mm in length,
101.60 mm in width, and 12.70 mm in thickness and was oriented vertically. Each plate had
two holes drilled 25.40 mm from the bottom of the plate and offset 25.40mm in either
direction from the center line of the plate; these holes were used to attach the plates to the
testing rig. Each plate had a third hole drilled 50.80 mm from the top edge of the plate located
on the center-line of the specimen. This hole diameter, 12.70 mm, 15.88 mm, or 19.05 mm,
was determined by the insert wall thickness for the specific test. Figure 2.1 shows the
dimensions for an individual plate.

25

101.60

50.80

Ø 12.70
Ø 15.88

203.20
Ø 12.70
25.40

25.40

25.40

Figure 2.1 – Plate dimensions for single-lap shear joint testing.
Every single-lap shear joint consisted of two plates joined together by means of a mechanical
fastener. The front plate, corresponding to the plate closest to the head of the bolt, was
fabricated from a 12.70 mm Grade 10 Garolite (glass-epoxy composite) manufactured to
military specification MIL-I-24768 with an experimental elastic modulus of 25.6GPa. The back
plate was fabricated from a 12.70 mm plate of 7075 aluminum with an elastic modulus of
71.7GPa. The central holes in each plate were aligned with each other and fastened using the
above mentioned bolts and accessories. The bolt was secured using a torque of 2.82 N-m and
mounted in the uniaxial testing machine using a specially designed fixture which aligned the
loading grips with the boundary plane between the two plates. The single-lap shear joint was
secured in the mounting fixture by friction gripping and by the two decentralized holes in each
plate mentioned above. Figure 2.2 shows a side view of the completed single-lap shear joint

26

Composite Panel
Aluminum Panel

Mechanical Fastener

Figure 2.2 – Side view of single
ingle-lap shear joint.
The sample was mounted in the uniaxial testing machine using a mount designed to
accommodate the single-lap bolted configuration. Figure 2.3 shows the mounting configuration
lap
.3
of the single-lap joint within the uniaxial testing machine. Each sample was loaded at a
lap
constant cross-head rate of 1 mm/min The loading was started at 0 N and was continued until
head
mm/min.
the sample failed catastrophically either by failure within the fastening mechanism or by net
catastrophically,
tension failure of the composite panel. The load and displacement data was recorded at a rate
ailure
of 10 points per second.

Hydraulic Grips

Aluminum Plate
Composite Plate
Mounting Device with
decentralized mounting holes

onfiguration
Figure 2.3 – Mounting configuration for single-lap bolted configuration.
27

Baseline Results
The first set of tests was run using non
non-reinforced single-lap configuration joints. Four tests
lap
were ran using the General Testing Procedure described above. Every sample failed by net
tension of the composite plate originating on the boundary between the composite plate and
the mechanical fastener at a location approximately 45 degrees from horizontal; Figure 2.4
shows a typical composite plate failure.

Figure 2.4 – Typical net tension failure of composite plate.
ension
The results were compared against each other to provide a baseline comparison from which the
he
results of the machined inserts could be compared. Figu
s
Figure2.5 shows the load-displacemen
displacement
data for the baseline tests. Examining Figure 2.5 it is apparent the loading mechanism allowed
slack to develop within the mounting configuration prior to testing; this resulted in an initial
area of non-linear loading prior to the region of elastic loading after the experiment had
linear
removed the slack.

28

70000
60000

Force (N)

50000
40000
30000
20000
10000
0
0

1

2

3

4
5
6
7
Displacement (mm)

8

9

10

11

Figure 2.5 – Raw load-displacement data for baseline tests.
To allow for a direct comparison between the data, each data set was normalized to remove
the non-linear portion of the data from the results. The normalization process consisted of two
steps. During the first step, all data points below 8000 N were removed from the data set. The
displacement at this point was then subtracted from the remainder of the data points. This
normalization removed the majority of the initial non-linear portion and aligned the resulting
elastic loading regions of each data set with each other. Figure 2.6 shows the normalized data
for the baseline load-displacement tests and Figure 2.7 shows the failure region of the
normalized load-displacement tests.

29

70000
60000
50000
Force (N)

40000
30000
20000
10000
0
0

1

2

3
4
5
6
7
Normalized Displacement (mm)

8

9

Figure 2.6 – Normalized load-displacement data for baseline tests.

70000
65000

Force (N)

60000
55000
50000
45000
3

4

5
6
7
Normalized Displacement (mm)

Figure 2.7 – Failure region for normalized data for baseline tests.

30

8

9

Using Figures 2.6 and 2.7, several results can be observed. Using Figure 2.6 it is seen the
samples first deviate from the linear loading relationship at approximately. 45 kN. Additionally,
each sample has a similar slope, indicating a consistency within the samples. The results also
show each sample has little to no bearing failure, indicated by knees in the data, prior to net
tension failure. Only one sample, shown in Figure 2.7 at a normalized displacement of 4.3 mm
and a load of 53 kN, shows indications of bearing failure. Figure 2.7 also shows, with the outlier
exception, that failure occurs at a normalized displacement of 7 to 7.5 mm and a load of
approximately 62 kN.
Aluminum Insert Results
The second set of tests was conducted using single-lap configuration joints reinforced using
aluminum inserts. Six tests were conducted using the General Testing Procedure described
above; three tests used inserts with a 3.18 mm wall thickness insert and three tests used inserts
with a 1.59 mm wall thickness insert. Every sample failed by net tension of the composite plate
similar to Figure 2.4. The raw data for each test was normalized using the method described in
the baseline results. Figure 2.8 shows the normalized load-displacement data for the aluminum
inserts and Figure 2.9 shows the failure region for the aluminum insert data. Tests using 3.18
mm inserts are shown by the darker curves and tests using 1.59 mm inserts are shown by the
lighter curves.

31

70000
60000

Force (N)

50000
40000
30000
20000
10000
0
0

1

2

3
4
5
6
Normalized Displacement (mm)

7

8

9

Figure 2.8– Normalized load-displacement data for aluminum inserts.

65000

Force (N)

60000

55000

50000

45000
3

4
5
Normalized Displacement (mm)

6

Figure 2.9– Failure region of normalized data for aluminum inserts.

32

7

The results shown in Figure 2.8 and 2.9 are similar to the baseline results shown above. Figure
2.8 shows the loading changes from linear to non-linear at approximately 45 kN. Additionally,
all six tests exhibited elastic slopes similar to baseline results, indicating an insert does not
affect the stiffness of the joint. The greatest difference is shown in Figure 2.9. The average
failure load decreased from 62 kN to approximately 55 kN, a difference not entirely by the
decrease I cross-sectional area due to the addition of the machined insert. The normalized
displacement until failure also decreased from 7 mm to an average of 5.5 mm. Furthermore,
several samples experienced showed signs of significant bearing failure to complete net tension
failure. However, compared to the baseline results, the aluminum insert results showed
considerable variation compared to similar tests. Furthermore, no significant difference was
shown between the inserts with a 3.18 mm inserts and the 1.59 mm inserts.
Steel Insert Results
The third set of tests was conducted using single-lap joints reinforced with steel inserts.
Six tests were conducted using the General Testing Procedure described above; three tests
used 3.18 mm inserts and three tests used 1.59 mm inserts. Every sample failed by net tension
of the composite plate similar to Figure 2.4. Figure 2.10 shows the normalized loaddisplacement data for the steel inserts and Figure 2.11 shows the failure region for the steel
inserts data. Tests using 3.18 mm inserts are shown by the darker curves and tests using 1.59
mm inserts are shown by the lighter curves.

33

70000
60000

Force (N)

50000
40000
30000
20000
10000
0
0

1

2

3
4
5
6
Normalized Displacement (mm)

7

8

9

Figure 2.10– Normalized load-displacement data for steel inserts.

65000

Force (N)

60000

55000

50000

45000
3

4

5
6
Normalized Displacement (mm)

Figure 2.11– Failure region of normalized data for steel insert.

34

7

The results shown in Figure 2.10 and 2.11 are similar to the baseline results shown above with a
few exceptions. Figure 2.10 shows, similar to the baseline results, the loading changes from
linear to non-linear at approximately 45 kN. Additionally, similar to the aluminum results, the
stiffness is similar to the baseline results again indicating an insert does not affect the stiffness
of the joint. Several differences were also observed. Five of six samples showed signs of
bearing failures prior to complete net tension failure. Additionally, the failure load decreased
from 62 kN for the baseline to approximately 58 kN, a difference not entirely explained by the
decrease in cross-sectional area due to the addition of the machined insert. Differing from the
aluminum inserts, the thickness of the insert did appear to affect the failure of the joint.
Although the insert thickness did not significantly affect the failure load samples using 1.59 mm
inserts failed at a normalized displacement of 6.25 mm whereas samples using with 3.18 mm
inserts failed at a normalized displacement of 5 mm.
Chapter 2 Conclusions
The results of Chapter 1 showed that the concept of a novel insert was indeed feasible;
it was possible to inject a liquid into a single-lap bolted joint through a bolt without changing
the overall failure mode of the joint. The next step was to examine the behavior of both nonreinforced single-lap joints and single-lap joints reinforced different types of inserts. Forcedisplacement was chosen for the initial analysis because of the simplicity of the test in addition
to the range of the test; zero load to failure. However, because of the device used to mount
the samples, the results an initial non-linear region resulting from slack within the mounting
device. As such, it was necessary to normalize the results to allow for comparison between
each test.
35

The results of the test yielded several results. The baseline tests recorded the highest
load to failure as well as the greatest normalized displacement to failure of the three sets of
tests. The baseline tests also showed the fewest signs of bearing failure, or ‘kneeing’ of the
three sets of tests conducted. The results also showed the addition of an insert had minimal
impact upon the stiffness of the composite plate; however, the addition of an insert tended to
cause partial bearing failure prior to complete net tension failure.
The results of the machined inserts were equally useful. The failure load for aluminum
inserts decreased from 62 kN for the baseline results to approximately 55 kN. The normalized
strain to failure also decreased from 7 mm to 5.5 mm. There wall thickness did not appear
significant. The steel inserts showed similar results. The load to failure decreased to 58 kN and
the normalized strain to failure decreased to 6.25 mm for the 1.59 mm inserts and 5 mm for the
3.18 mm inserts. The wall thickness did appear to affect the normalized displacement to failure
but not the load to failure.
However, the above results are otherwise of limited use. The variation within the each
set of tests and the use of load and displacement make it difficult to provide a baseline to which
to compare injected inserts. Furthermore, the nature of the net tension failure, as shown in
Figure 2.4, indicate the strain fields within the samples are more complex than can be
adequately descreibed by load-displacement measurements. It was determined a full-field
measurement technique, such as digital image correlation, would be necessary to adequately
analyze the performance of a single-lap bolted joint undergoing uniaxial tension loading
condition.

36

Chapter 3 – Digital Image Correlation Results of Inserts
Chapter3 will focus on the use of Digital Image Correlation (DIC) to examine the strain field
resulting from applying a tensile loading to the single-lap bolted joints with mechanical
fasteners that have not been reinforced with any type of insert (baseline tests). This chapter
will compare the results of the baseline tests to tests performed on single-lap bolted joints with
mechanical fasteners reinforced with machined inserts as well as joints reinforced with the
injected insert. The chapter considers both the effect of material properties as well as the
inserts wall thickness.
Digital Image Correlation Introduction
Some common methods of strain measurement systems such as resistance strain gages
or fiber-optic strain gages, only measure the average strain that is experienced over the length
of the strain gage. With specimens of simple geometry or simple loading conditions, such as a
plate in uniaxial tension or pure bending, the use of strain gages is adequate because the strain
field is relatively simple and well understood. However, when using a single-lap bolted joint
fastening a cross-ply composite laminate to an aluminum plate, the strain field is much more
complicated and is less understood. In this situation, the use of a full-field measurement
technique such as DIC has many advantages.
DIC is a non-contact optical measurement method used to measure the full-field strain
distribution of an object. In DIC, the desired surface is speckled with a random dot pattern,
often by using spray-paint. As the object is loaded, images are captured using a digital camera
and transferred to a computer. An algorithm is then used to correlate the displacement of each
speckle subset relative to the position indicated by a reference image; once the displacements
37

are known, the strains can be calculated through the use of elasticity equations. Additionally, if
necessary a DIC system can be modified by adding a second camera; this configuration is then
capable of measuring out-of-plane displacements.
Digital Image Correlation Configuration
For these experiments, DIC was used to better understand the full-field strains
developed in the single-lap bolted joint as a result of the uniaxial tension loading configuration.
The DIC setup utilized a single Allied Vision Technologies Stingray Progressive CCD camera with
a 1624 x 1234 pixel resolution utilizing a bayer mask; a bayer mask is a color filter which
increases the sensitivity to green light by using a pattern with 50% green, 25% blue, and 25%
red filters. The specimen was illuminated by a single directed light source containing 16 green
light emitting diodes (LED) as well as ambient lighting consisting of hi-bay fluorescent fixtures.
The samples were mounted in the MTS810 testing apparatus and the load applied according to
the specifications established in the general testing procedure. Deviating from the general
testing procedure, each sample was only loaded from 0 kN to 30 kN. The loading was paused at
every 5 kN (i.e. 5 kN, 10 kN, etc) to acquire the correlation images. Additionally, load and
displacement data was not recorded during these tests. The images were transmitted to a
personal computer and processed using the Istra 4d software, version 4.2.2.15, to determine
the full-field normal strains for both the horizontal axis (X Axis) and the vertical axis (Y Axis).
The area of each image showing the bolt assembly was masked (excluded) to remove the
reporting of false strains resulting from a non-continuous strain field.

38

Baseline Results
DIC was used to better understand the in-plane strain fields developed as a result of the
uniaxial loading on the single-lap shear joint. DIC was also used to understand what effect, if
any, the addition of an insert, either machined or injected, has on the full field strain
distribution of the single-lap shear joint. Three sets of experiments were performed to
investigate this effect. The first set of experiments examined a non-reinforced joint. The
second set of experiments examined the results of samples reinforced with machined inserts.
The third set of experiments examined the results of samples reinforced with different types of
injected inserts.
The first set of experiments was performed on non-reinforced single-lap shear joints.
These results serve two purposes. The first is to provide a comparison between empirical
results and DIC. The second is to provide a baseline for the composite joint. The normal strains
for the X axis and the Y axis are shown in Figure A1.1 and Figure A1.2, respectively. As the bolt
is pushed into the composite because of the loading, it would begin to cleave the composite,
(push the material aside), at this point of contact creating a tensile strain in the normal X- axis.
This strain is observed in Figure A1.1.
The load is applied to the sample in the negative Y direction (downward) and is transferred to
the aluminum plate through the bolted connection; this condition is expected to create a
compressive strain in the Y-direction immediately above the bolt hole. From previous testing,
ultimate failure of the composite is expected to initiate at the boundary between the
composite and the bolted connection at an angle approximately 45 degrees from horizontal.
The test configuration utilizes a single-lap shear joint; this configuration creates a bending
39

moment on the sample, resulting in a compressive strain in the lower portion (the portion
below the centerline of the bolt) of the sample. All three of these phenomena are observable
Figure A1.2.
Machined Insert Results
A second series baseline tests were performed on preformed inserts. The purpose of this set of
experiments was to determine whether similar research performed on thin composite plates
could be applied to thick composite plates; this set of experiments also provided a secondary
baseline from which to compare the performance of the injected inserts. These tests used
1144 steel and 7075 aluminum inserts with an inner diameter of 12.70 mm and wall thicknesses
of 1.59 mm and 3.18 mm.
Figure A1.3 and Figure A1.4 show the x-axis and y-axis normal strains for samples using 3.18
mm aluminum 7075 insert. Based on the results of Figure A1.3 and Figure A1.4 samples using
3.18 mm wall thickness inserts show significant improvement over the baseline results; the
insert reduces the y-axis compressive normal strain located above the bolt hole.. Similarly, the
insert reduces the cleaving effect seen above the bolt in the x-axis normal strains. In general,
the 3.18 mm inserts most effectively reduced the strain gradients.
Figure A1.5 and Figure A1.6 show the x-axis and y-axis normal strains for samples using
a 3.18 mm 1144 steel inserts. Comparing these results of the 3.18 mm 7075 aluminum insert,
the 1144 steel inserts behaved in a similar manner to the of 3.18 mm aluminum inserts. Both
inserts have similar strain profiles for both normal strains and both inserts reduced the strains
seen in the baseline results.

40

The second set of tests was performed using machined considered inserts with a 1.59
mm wall thickness. Figure A1.7 and Figure A1.8 show the x-axis and y-axis normal strains for
samples using 1.59 mm 7075 aluminum insert. These results showed that even though these
inserts had higher strain magnitudes than the 3.18 mm inserts, the strain profiles are similar to
the 3.18 mm 7075 aluminum inserts, with the exception of a more pronounced compressive
strain directly above the bolted joint seen in Figure A1.8. The results from the 1.59 mm 1144
steel inserts are comparable to the aluminum inserts of the same wall thickness. These results
are shown in Figure A1.9 and Figure A.1.10.
Injected Inserts
The third set of tests was performed using injected inserts. The injection bolts were
machined using the modified bolt design described in Chapter 1. The injection materials
consisted of a thin epoxy resin with a medium rate hardener, a two-part polyurethane adhesive
and a pre-impregnated film adhesive. After application, each material was allowed to cure for
at least 24 hours before testing. Inserts with wall thicknesses of 1.59 mm and wall thicknesses
of 3.18 mm were tested.
Figure A1.11 and Figure A1.12 show the x-axis and y-axis normal strains for tests using epoxy
resin with 1.59 mm epoxy inserts. As the results for the 3.18 mm epoxy inserts are similar to
the 1.59 mm epoxy inserts, only the results of the 1.59 mm epoxy inserts are shown.
Comparing the results to the previous results, the results show that away from the bolted
connection, the strain fields are similar to the strain fields for the 3.18 mm 7075 aluminum
inserts. However, as the strain field approached the bolted connection, the behavior of the
strain field was similar to the non-reinforced samples. The compressive strains above the bolt
41

hole, shown in Figure A1.12 are almost identical to the strains produced in the non-reinforced
samples.
The second injected insert material tested was a two-part polyurethane. This material is both
ductile and has a low elastic modulus. Figure A1.13 and Figure A1.14 show the x-axis and y-axis
normal strains for samples using 3.18 mm polyurethane inserts. Using these results, two
distinct observations are noted. The compressive strain above the bolted connection is greater
in magnitude than the compressive strain for the non-reinforced samples. However, the
cleaving effect appears to be significantly decreased when compared to the non-reinforced
samples.
The third injected insert material tested was a film adhesive, similar to a pre-impregnated
lamina. This material is considered brittle with an elastic modulus similar to the epoxy resin
tested; however, unlike the epoxy resin, it has a reinforcing mesh contained within the matrix
material. Figure A1.15 and Figure A1.16 show the x-axis and y-axis normal strains for samples
using 3.18 mm film adhesive inserts. Using these results, two observations were noted. Unlike
the previous injected insert materials, the film adhesive did decrease the compressive strain
above the bolted connection. Additionally, similar to the other injected insert materials tested,
the film adhesive did decrease the cleaving effect above the bolted connection. Of the three
injected insert materials tested, the results of the film adhesive were most suitable when
compared to the results of the machined inserts, specifically the 1.59 mm steel or aluminum
inserts.

42

Strain Profiles
As the three sets of experiments described above were taken at different times with the
camera positioned at different distances, comparing the full field strain distributions for each
image against the other sets of experiments is relatively difficult. To further the analysis, line
profiles were recorded for each set of images. The line profiles were created as to capture
certain behavioral aspects of each sample that can then be easily compared with the same
behavioral aspects from a different sample.
The first line profile was constructed using the x-axis normal strain at a y-value corresponding
to the top of the washer (x-x profile); this profile was recorded to analyze the cleaving effect
above the bolted connection. Figure 3.3 shows the legend for all four line profiles. Figure 3.4
shows the x-x profile. These results show the non-reinforced joint has the greatest cleaving
effect. The strain increases from -0.25 mm/m away from the bolted connection to a magnitude
of 2.4 mm/m directly above the bolted connection for a change in magnitude of 2.25mm/m.
The 3.18 steel and aluminum inserts had the smallest change, increasing from -1.2 mm/m to 0.3 mm/m. Similarly, the 1.59 mm steel and aluminum inserts also have similar behaviors. The
injected inserts show a similar behavior. The polyurethane inserts appears to have the greatest
magnitude change, 2.5 mm/m, followed by the epoxy inserts, 2.0 mm/m with the film adhesive
having the least change of 1.75 mm/m. This data supports the concept that stiffer inserts,
those with a higher elastic modulus (steel and aluminum), perform better than inserts with a
low elastic modulus (resin, two-part polyurethane, film adhesive).

43

Non-Reinforced

Aluminum (3.18 mm)

Aluminum (1.59 mm)

Steel (3.18 mm)

Steel (1.59 mm)

Epoxy Resin (1.59 mm)

Polyurethane (3.18 mm)

Film Adhesive (3.18 mm)

“For interpretation of the references to color in this and all other
figures, the reader is referred to the electronic version of this thesis.”
Figure 3.3 – Legend for DIC derived line profiles

3
2.5
2

Strain (mm/m)

1.5
1
0.5
0
-0.5
-1
-1.5
-2
-40

-30

-20

-10
0
10
Location (mm)

20

30

40

Figure 3.4 – X-X Profile for single-lap bolted connection derived from DIC.
The second profile was constructed using the x-axis normal strain at an x-value corresponding
to center of the bolted connection (x-y profile); this profile was recorded to analyze the
magnitude between the tensile and compressive strains located above and below the bolted
connection, respectively. Figure 3.5 shows the x-y profile. The results of Figure 3.5 show that,
in general, the type of insert does not seem to significantly affect the change in magnitude
between the tensile strain located above the bolted connection and the compressive strain
44

located below the bolted connection. Inserts with low tensile strains, the 3.18 mm steel and
aluminum inserts, have the highest compressive strain. Similarly, inserts with high tensile
strains above the bolted connection, the 1.59 mm steel and aluminum inserts, have a low
compressive strain below the bolted connection.

40
30
20
Location (mm)

10
0
-10
-20
-30
-40
-4

-3

-2

-1

0
1
Strain (mm/m)

2

3

4

Figure 3.5 – X-Y profile for single-lap bolted connection derived from DIC.
The third profile was constructed using the y-axis normal strain at a y-value corresponding to
the top of the washer (y-x profile); this profile was recorded to analyze the compressive strain
above the bolted connection. Figure 3.6 shows the y-y profile. The noted results from this
profile are similar to those noted for the x-x profiles shown in Figure 3.4. The 3.18 mm steel
and aluminum results have similar magnitude changes in strain. Similarly, the material with the
lowest elastic modulus has the great change in strain. The most noticeable change is the
difference in behavior between the 1.59 mm steel and aluminum inserts. Whereas both inserts
show almost identical behavior in the x-x profile (Figure 3.4), in the y-x profile (Figure 3.6), the
45

steel insert has a significantly increased change in strain compared to the aluminum insert; this
data shows that although inserts with higher elastic moduli generally perform better, this
performance is also affected by the insert thickness. When using thinner inserts, a more
compliant material, such as aluminum, may be more desirable than a less compliant material
such as steel.

1
0.5
0
-0.5
Strain (mm/m)

-1
-1.5
-2
-2.5
-3
-3.5
-4
-40

-30

-20

-10
0
10
Location (mm)

20

30

40

Figure 3.6 – Y-X Profile for single-lap bolted connection derived from DIC.
The forth profile was constructed using the y-axis normal strain at a x-value corresponding to
the left of the washer (y-y profile). This profile was recorded to analyze the effect created by
the single-lap bolted connection. Figure 3.7 shows the y-y profile. If the plate where under
simple tension in the y-axis, it is expected that every point in the plate would experience a
tensile strain in the y-axis. However, because of the single-lap configuration, a moment is
created about the y-axis that acts to bend the composite plate and cause out-of-plane (z-axis)

46

displacement. Unfortunately, because the experimental used a one-camera DIC setup, it was
not possible to directly measure the out-of-plane displacement. However, as the plate bends,
it is expected to most significantly affect the y-axis strains. Therefore, by comparing the y-axis
normal strain along a vertical profile, it might be possible to indirectly examine the out-of-plane
displacement. The results of the Figure 3.7 show that although the strain for each sample
varies, the slope of the strain for every sample in the area of highest interest is approximately
equal. Therefore, it might be reasonably concluded that although out-of-plane displacement
affects the y-axis normal strain during the tests, as is noted by the unexpected compressive
strains shown in the DIC results, the effects of the out-of-plane displacement are similar during
each test, indicating it is still valid to compare the results of these experiments to each other,
even if it is not possible to extend these results to experiments carried out under other
conditions.

40
30
20

Location (mm)

10
0
-10
-20
-30
-40
-5

-4

-3

-2
-1
Strain (mm/m)

0

1

Figure 3.7 – Y-Y profile for single-lap bolted connection derived from DIC.
47

2

Chapter 3 Conclusions
As chapter 2 examined, although useful, the use of load-displacement data was
insufficient for analyzing the performance of the injected inserts. Scatter within the results and
the proximity of the results to each other required the use of a more detailed method to
analyze the performance of the inserts. Digital image correlation (DIC) was chosen to analyze
the performance of the inserts within the elastic, non-damaging, region of the single-lap bolted
configuration.
Three sets of data were acquired using DIC. The first set of data analyzed the results of
non-reinforced joints. These results agreed with the visual observations shown in chapter 2.
The results of chapter 2 showed failure originated at the boundary of the bolt hole at a location
roughly 45 degrees from horizontal. The results of chapter 3 showed this location corresponds
with the area of highest normal strain in the vertical direction. The baseline DIC results also
showed several other areas of interest: a compressive “contact” stress shown in the vertical
normal strain field located above the bolt hole, a “cleaving” tensile strain shown in the
horizontal normal strain field located above the bolt hole, and a general compressive strain
shown in the vertical normal strain field starting below the mid-point of the bolted connection.
Tests performed on machined inserts as well as some injected inserts reinforced trends
seen throughout the previous chapter as well as trends seen in previous research. Stiffer
inserts, those with a higher elastic modulus, tended to result in lower “contact” stresses above
the bolt hole as well as lower cleaving stress above the bolt hole.
Two other trends were also noticed. Inserts which results in lower cleaving strains tended to
have increased compressive strains immediately below the bolt hole whereas inserts with
48

higher cleaving strains tended to have lower compressive strains immediately the bolt hole.
The second trend identified was the un-predicted compressive strain below the midpoint of the
bolt was not affected by any insert. The gradient of this strain remained constant for all
experiments and the magnitude of the strain appeared independent of the insert material.
The primary disadvantage of this digital image correlation setup was the inability to track outof-plane displacements. The single-lap bolted configuration by nature produced an out-ofplane displacement within the testing configuration. This bending created an additional stress
which was not measureable by the DIC setup utilized. However, as these results were used
comparatively and as the out-of-plane bending was predicted to affect each sample equally, it
was determined that although the actual numbers reported would be changed by the addition
of the bending induced strain , the comparison of the digital image correlation results to each
other within these experiments was valid.

49

Chapter 4 – Performance of Insert Materials under Ballistic Loads
The loading conditions to which these mechanically fastened joints would be exposed to is not
best modeled by a low-strain rate environment, as was seen in both Chapter 2 and Chapter 3, it
becomes useful to look at the performance of the joints under more severe loading conditions.
As such, Chapter 4 seeks to examine the performance of two of the injected inserts, the
polyurethane and the film adhesive, under ballistic loading under ballistic loading conditions.
As the single-lap bolted joint setup is not practical for this loading condition, an alternative
means was developed to test the performance of each of these materials. The delamination,
adhesion, and deformation of each sample was considered in the analysis of each sample.
Previous Research
Zaera [10] investigates the effect the adhesive thickness and adhesive properties play on the
performance of a ceramic-composite panel undergoing ballistic loading. Zaera researches
ceramic-composite armors, specifically, the adhesive layer, as an alternative to monolithic plate
armors. These experiments use a 6mm alumina tiles bonded to a 6mm alumina plate with two
different types of adhesives: a polyurethane and rubber-toughened epoxy with different
thicknesses. The samples were impacted by a 7.62 tungsten carbide AP projectile at 940 m/s.
The results showed that both the adhesive type and thickness affected the damage of the
sample; thicker layers resulted in less damage to the surrounding tiles but resulted in larger
deformations to the aluminum plate as well as increased fragmentation of the ceramic tile. The
stiffer adhesive, epoxy, resulted in a lower supporting effect of the aluminum plate.
Lopez-Puente [11] investigated the effect of the thickness of the adhesive layer on the
performance of a composite-armor. The experiments used 100x100mm and 51x51mm alumina
50

tiles arranged in a square enclosed in a 4130 steel frame The samples were bonded to a 2017
aluminum plate using Hysol EA-9361 toughened epoxy resin of carrying thicknesses. The
samples were impacted by a 7.62 tungsten carbide round at 940 m/s. After each test, the
aluminum plate was removed and cut transversely with a diamond thread. The results showed
that the impact caused shear strain in the adhesive and that samples with a thicker adhesive
layer better attenuated the shear strain and also better retained portions of the fractured tile.
However, samples with thinner adhesive layers experienced less deformation. The optimum
thickness was determined to be 0.3mm for all configurations tested.
Panel Design
The tests conducted for both load-displacement and DIC testing used a single-lap bolted
configuration; this setup was advantageous because each sample had a simple geometry and
simple fabrication method. However, for ballistic loading conditions, this setup is not
satisfactory. Under ballistic loading conditions, there is a high probability that upon impact by
the projectile, the insert itself as well as the composite plate would be destroyed. The
aluminum plate would also be damaged, resulting in an exponential increase of testing costs.
As such, a panel design using the desired insert material was used.
The panel design consisted of four basic layers. The back layer used a 0.5” thick
aluminum plate; this layer is used to absorb most of the energy from the projectile as well as
provide a foundation for the rest of the panel. The second layer consists of the adhesive; the
adhesive is used to transmit the shockwave from the front layer to the back layer as well as
adhere the front layer to the back layer. The third layer consists of ceramic tiles. The ceramic
tiles are used to dull the projectile, if necessary, and absorb some of the impact energy of the
51

projectile. A small offset is included between each tile and is filled with a rubbery material
designed to attenuate the transmission of the shockwave from tile to tile. A forth layer
consisting of glass-reinforced composite is also used; however, the purpose of this layer is to
contain ceramic spalling as well as prevent bowing of the ceramic tiles. This layer does not
serve a significant role in the performance of the panel. Figure 4.1 shows a diagram of a
finished panel without the glass reinforced composite overwrap layer.

Figure 4.1 – Completed test panel for ballistic loading.
Although this design is not the same as the single-lap bolted configuration used for the
earlier experiments as well as several advantages. 1) The purpose of the inserts in the singlelap configuration is to distribute the forces created by the bolt and decrease the localized
strains as well as decrease strain gradients. The purpose of the adhesive in the panel design is
to distribute the shockwave to the aluminum panel. 2) Similar to the wall thickness of the
single-lap configuration, the adhesive thickness of the panel can easily be adjusted to examine
the effect of thickness on the performance. 3) The size of the panel can easily be scaled,

52

allowing for multiple tests to be performed on a single panel with a minimal increase in
fabrication time.
Determining Equivalent Thicknesses
The primary advantage of using the above panel design is the ease with which the
adhesive thickness can be altered. This is significant because the method used to apply the
polyurethane to the panel is fundamentally different than the method used to apply the film
adhesive to the panel.
The polyurethane remains a liquid shortly after mixing and retains a constant thickness
during the cooling process. As such, the thickness of the adhesive layer in the panel can be
controlled by using spacers between the aluminum panel and the layer of ceramic tiles.
However, the film adhesive is initially applied to the panel in discrete layers as a solid object.
Furthermore, as the film adhesive is cured, a portion of the adhesive will liquefy, causing the
thickness to decrease during the cooling process. As previous research has indicated the
thickness of the adhesive can significantly affect the performance of the panel, it became
necessary to determine a correlation between the initial thickness of the film adhesive before
curing and the final thickness of the film adhesive after curing.
In order to establish a correlation between the initial, uncured thickness of the film
adhesive to the final, cured thickness of the film adhesive, an experiment was conducted. Four
small samples representative of the panel design was fabricated. The back layer of each sample
was constructed from an aluminum plate. The adhesive layer of each sample consisted of
preselected number of layers of the film adhesive; for this experiment samples were fabricated
using three layers, four layers, six layers, and seven layer. The top layer consisted of ceramic
53

tiles positioned similar to the ceramic tiles on the actual test panels. Figure 4.1 shows a test
sample used in this experiment.

Figure 4.1 – Sample panel for d
determining equivalent thicknesses.
After each panel had been fabricat
fabricated, the thickness of the panel was measured and
recorded at several points; the average value were normalized and plotted. These values are
values
shown in Figure 4.2 by the diamond markers. The panels were then cured according to the
manufacturing recommendations; a weight was placed on each sample to minimize movement
ions;
of the ceramic tiles during the curing process. After the curing process had completed, the
thickness of each sample was measured at several location and the average values were
normalized and plotted. These values are shown in Figure 4.2 by the square markers. Using
.
this data, it was determined five layers of the film adhesive corresponds most accurately to the
thickness of the adhesive when using the polyurethane.

54

2

Normalized Thickness

1.5

Initial Thicknes
Final Thickness

1

0.5

0

-0.5
0

1

2

3
Layers

4

5

6

7

Figure 4.2 – Initial and final thicknesses for film adhesive samples.
Extrapolating the initial thickness values to a theoretical sample with no layers of film adhesive
the data correlated precisely to the thickness of the aluminum panel and ceramic tiles. A
similar method, when applied to the final thickness data, yielded a slight offset as is shown in
Figure 4.2. However, due to movement of the ceramic tiles during the curing process this is
considered acceptable error.
Testing Procedure
Based on the results from Figure 4.2 four panels were fabricated for experimental
purposes. Two panels were fabricated using the polyurethane using the standard thickness of
the adhesive. One panel was created using four (4) layers of the film adhesive. One panel was
created using five (5) layers of the film adhesive. The purpose of the experiment was to

55

qualitatively measure the deformation of the aluminum panel (shape and location), the level of
delamination between the aluminum plate and the ceramic tiles, and the adhesion of the
fragmented ceramic tiles to the adhesive. However, as no previous research existed to identify
how the panels would perform under the ballistic conditions, one panel fabricated using the
polyurethane was used as a gauging panel. This panel was tested using several projectiles at
different projectiles to determine which combination of projectile/velocity was best suited for
the objectives.
Using the results of the gauging panel, it was determined two different tests could be
performed on each panel without the results of one experiment affecting the results of the
other experiment. The first experiment used a low mass, pointed-edge projectile at a high
velocity (Projectile A). The purpose of the experiment was to determine the effectiveness of
the adhesive by analyzing the deformation of the aluminum panel at the impact site as well as
the retention of the ceramic fragments by the adhesive at the impact site; this experiment was
designed not to fully penetrate the sample. The second experiment used a higher mass, bluntedged projectile at a lower velocity (Projectile D). The purpose of this experiment was to
examine the adhesion properties of each material when subjected to a high level of damage.
This experiment was designed to cause complete penetration of the sample. The experiment
using Projectile A was completed a total of six times per panel; the experiment consisting of
Projectile D was tested a total of 3 times per panel
Testing Results
After the panels had been tested, the glass-reinforced composite layer is carefully
removed and the panels examined; four aspects of each panel were examined. 1) The
56

deformation, shape and location, of the aluminum panel caused by the Projectile A tests. 2) The
adhesion of the fragments of the ceramic tiles impacted by the projectile during the Projectile A
tests. 3) The cumulative level of delamination between the aluminum plate and the ceramic
tiles caused by the Projectile A and Projectile D tests. 4) The localized delamination between
the aluminum plate and the ceramic tiles caused by the Projectile D impacts. In general, the
results showed the panel using the polyurethane adhesive outperformed the panels using the
film adhesive as the adhesive layer.
The shape and location of the aluminum deformation for the panel fabricated using
polyurethane tended to be more spherical in nature; this shape is indicative of a better
distribution of the shockwave as it propagated through the adhesive. The shape and location of
the deformation resulting from the panels fabricated using the film adhesive tended to be more
parabolic in nature with a greater maximum deformation, as defined by the distance from the
plane of the undeformed aluminum to the farthest point of the deformation. Additionally,
indications of plugging, or material failure within the aluminum, were also observed in these
samples. The shape and location of the deformation indicate a lesser degree of distribution of
the shockwave as it propagated through the adhesive. Furthermore, the indications of plugging
indicate the material was closer to complete failure.
The sample fabricated using the polyurethane retained more of the ceramic shards
resulting from the Projectile A impact. Other than the area immeadiately adjacent to the
impact site of the projectile, the majority of the ceramic shards remained adhered to the
adhesive. As actual images of the impact area were not available, Figure 4.3 shows a image
representative of the fragmented ceramic tile adhesion obtained from previous research.
57

Figure 4.3 – Representative tile adhesion of polyurethane sample from Puente. [11]
ile
ample
Puente
The samples fabricated using the film adhesive performed poorly in retaining ceramic
poorly
shards resulting from the Projectile A impacts. Once the glass reinforced composite layer had
glass-reinforced
been removed, it was found little to none of the ceramic shards remained adhered to the
adhesive. Furthermore, when viewed from the front, a greater level of damage to the
front,
aluminum plate was observed when compared to the sample fabricated using the polyurethane
adhesive. As an actual image of the impact area was not available, Figure 4.4 shows a
representative image of the fragmented tile adhesion of the film adhesive from previous
tile
research.

Figure 4.4 - Representative tile adhesion of film adhesive samples from Puente. [11]
ile
amples
Puente
58

The sample fabricated using the polyurethane retained good adhesion during the tests.
After the glass-reinforced composite layer was removed, it was observed the ceramic tiles not
directly affected by the projectile impacts remained adhered to the aluminum panel. However,
for both samples created using the film adhesive, it was observed that complete delamination
had occurred between the aluminum panel and the ceramic tiles. However, due to the
manufacturing process of the samples, it was not possible to determine whether the
delamination was a result of the projectile impacts or whether the delamination occurred
during the curing process.
The level of localized delamination resulting from the Projectile B impacts was similar
for both the panel fabricated using polyurethane and the samples fabricated using film
adhesive. For all tests using Projectile B, the tiles within the area of the aluminum affected by
the impact delaminated from the aluminum panel but remained adhered to the adjacent
ceramic tiles. Additionally, for all tests, regardless of the adhesive used, the primary damage
was limited solely to the tile impacted by the adhesive.
Chapter 4 Conclusions
Although the panel design represents a departure from the single-lap bolted
configuration in the previous chapters, the design has the distinct advantage that it provides
the opportunity to examine the behavior of the adhesive under more severe the loading
conditions of ballistic impacts; a feature which the single-lap bolted configuration does not
allow.
One of the primary difficulties encountered with this design is that previous research
has shown the overall performance of the panel is highly sensitive to the thickness of the panel;
59

as the film adhesive is applied in discrete layers and will change in thickness as the adhesive is
cured, it was necessary to determine a method to correlate the initial thickness of the film
adhesive to the final thickness of the film adhesive. The high degree of correlation and
accuracy within the results of the experiment showed a simple linear correlation could be used
to determine an equivalent thickness.
Using the results of the previous experiment, three samples were fabricated, one using
polyurethane as an adhesive and two using film adhesive, and tested. The results from these
tests showed the polyurethane adhesive, which is more ductile and has a lower modulus of
elasticity than the film adhesive, performed better in every area of interest. The sample using
the polyurethane retained a majority of the ceramic fragments created by the projectile
impacts and did not show any signs of delamination during the impact. Inversely, the panels
fabricated using the film adhesive retained few if any of the ceramic fragments created by the
project impacts and showed signs of complete delamination during the experiment.
Additionally, the samples created using the film adhesive proved more difficult to fabricate and,
because of the initially solid, discrete nature of the film adhesive, likely contained more flaws
within the original construction. Contrary to the previous chapters, which indicate a harder
insert with a higher modulus of elasticity is preferable, the results of this chapter indicate a
softer material with a lower modulus of elasticity is preferred as those materials have a greater
ability to distribute and attenuate shock waves resulting from ballistic impacts.

60

Summary
Three objectives were considered when conducting this research. These objectives were: to
explore the process of mechanically joining dissimilar materials using the concept of a injected
insert; asses some of the advantages and disadvantages of using an injected insert; compare
the results obtained from samples using injected inserts against similar samples containing
more traditional machined inserts. In general, the research was successful in meeting all of
these objectives.
Conclusions
Chapter 1 explored the feasibility of mechanically joining dissimilar materials using an injected
insert. Preliminary experiments concluded a phase-changing material, such as an epoxy resin,
could be injected into the bolted connection through a modified bolt to form an injected insert
with sufficient quality. Additional experiments were performed to further optimize the bolt;
through these experiments It was determined that neither the channel design or the channel
diameter had any significant effect on the consistency of the injected insert. Only the pretorque applied to the bolt before injecting the material affected the consistency of the bolt.
Samples with higher initial pre-torques resulted in samples with a greater consistency in the
injected insert. Samples with lower pre-torques exhibited significant air voids within the
injected insert.
Chapter 2 and Chapter 3 explored methods of comparing traditional machined inserts against
injected inserts. Load-displacement measurements obtained from tension testing experiments
provides some results, but were of limited value. The load-displacement results showed that
samples without any type of insert generally failed at higher loads and greater normalized
61

displacements. Additionally, samples with machined inserts typically had more numerous and
more pronounced examples of “kneeing” prior to net tension failure of the composite plate.
However, the limited nature and variations within the results prevented further analysis.
Digital Image correlation provided a method of comparing the types of inserts. In general, and
consistent with previous research conducted on thick composites, stiffer insert, such as
aluminum or steel performed best. These inserts showed smaller strain gradients across each
cross section. Injected inserts, with the injection of the film adhesive, performed worst,
sometimes resulting in strain gradients greater than the baseline tests.
Chapter 1, chapter 3, and chapter 4 assessed some of the advantages and disadvantages of
using the injected inserts. The sealing aspect of injected inserts helped protect the connection
from liquids or other particles. Injected inserts conformed to the geometry of the connection,
lowering the tolerances when compared to a machine insert. Additionally, injected insert
materials performed better in attenuating and distributing impulse forces. However, injected
inserts consistently performed worse than machined inserts when subjected to static or simple
tensile loads and the material removal from the bolt can substantially affect the failure mode of
the bolted connection. In general, in applications where ultimate failure load is not the primary
concern, injected inserts showed several advantages over a traditional machined insert.
Future Work
Several areas in which the injected insert research can be be furthered have been noted. The
majority of research into inserts has been conducted in the static and quasi-static range; in this
range it has been established stiffer inserts, typically aluminum or steel, will perform better
than softer inserts, as used in injected inserts. However, in many instances, materials will
62

perform significantly differently under dynamic loading. A continuation of the research would
be to perform similar experiments on the injected inserts in a single-lap shear joint using
dynamic loading conditions.
The current research was conducted using a single-lap shear joint. Although a single-lap shear
joint is simple to build and test, the configuration complicates the analysis of the samples. In
addition to the tensile stress, the single-lap shear joint creates a bending moment on the
sample which results in an uneven, and significantly more complicated, strain field across the
sample. A continuation of the research would be to conduct experiments, such as simple
tension of double-lap shear joints that eliminate the bending moment.
The current research was conducted using a limited number of commonly available injected
insert materials. A continuation of the research would be to conduct experiments using a
greater variety of commonly available materials as well as materials altered specifically for use
in an injected insert, i.e. by using nano-particles or carbon fibers.

63

APPENDICES

64

APPENDIX A

Figure A.1.1 - X-axis strains for non-reinforced at 30kN.

Figure A1.2 – Y-axis strains for non-reinforced at 30kN.

65

Figure A1.3 – X-axis strains for 3.18 mm 7075 aluminum at 30kN.

Figure A1.4 – Y-axis Strains for 3.18 mm 7075 aluminum at 30kN.

66

Figure A1.5 – X-axis strains for 3.18 mm 1144 steel at 30kN.

Figure A1.6 – Y-axis strains for 3.18 mm 1144 steel at 30kN.

67

Figure A1.7 – X-axis strains for 1.59 mm 7075 aluminum at 30kN.

Figure A1.8 – Y-axis strains for 1.59 mm 7075 aluminum at 30kN.

68

Figure A1.9 – X-axis strains for 1.59 mm 1144 steel at 30kN.

Figure A1.10 – Y-axis strains for 1.59 mm 1144 steel at 30kN.

69

Figure A1.11 – X-axis strains for 1.59 mm epoxy resin at 30kN.

Figure A1.12 – Y-axis strains for 1.59 mm epoxy resin at 30kN.

70

Figure A1.13 – X-axis strains for 3.18 mm polyurethane at 30kN.

Figure A1.14 – Y-axis strains for 1.59 mm polyurethane at 30kN.

71

Figure A1.15 – X-axis strains for 3.18 mm film adhesive at 30kN.

Figure A1.16 – Y-axis strains for 1.59 mm film adhesive at 30kN.

72

REFERENCES

73

REFERENCES
[1] Nillsson, Sören. (1989) “Increasing Strength of Graphite/Epoxy Bolted Joints by Introducing
an Adhesively Bonded Metallic Insert.” Journal of Composite Materials 23: 642-650
[2] Herrera-Franco, Pedro J., Cloud, Gary L. (1992) “Strain-Relief Inserts for Composite Fasteners
–An Experimental Study.” Journal of Composite Materials 26: 751-768
[3] Camanho, P.P., Tavares, C.M.L., et al. (2005) “Increasing the efficiency of composite singleshear lap joints using bonded inserts.” Composites Part B: Engineering 36: 372-383
[4] Herrington, P.D., Sabbaghian, M. (1992) “Effect of Radial Clearance between Bolt and
Washer on the Bearing Strength of Composite Bolted Joints.” Journal of Composite Materials
26: 1826-1843
[5] Riccio, A., Marciano, L. (2005) “Effects of Geometrical and Material Features on Damage
Onset and Propagation in Single-Lap Bolted Composite Joints under Tensile Load: Part 1 –
Experimental Studies.” Journal of Composite Materials 39: 2071-2090
[6] Eriksson, Ingvar. (1990) “On the Bearing Strength of Bolted Graphite/Epoxy Laminates.”
Journal of Composite Materials 24: 1246-1269
[7] Ekh, Johan, Schon, Joakim, et al. (2005). “Secondary bending in multi fastener, compositeto-aluminum single shear lap joints.” Composites Part B 36:195-208
[8] McCarthy, M.A., McCarthy, C.T. et al. (2006). “A simple method for determining the effects
of bolt-hole clearance on load distribution in single-column multi-bolt composite joints.”
Composite Structures. 73:78-87
[9] Khashabam U.A., Sallam, H.E.M., et al. (2006) “Effect of washer size and tightening torque
on the performance of bolted joints in composite structures.” Composite Structures. 73:310317.
[10] Zaera, R., Sanchez, S., et al. (2000). “Modeling of the adhesive layer in mixed
ceramic/metal armours subjected to impact. Composites Part A:. 31:823-833

74

[11] Lopez-Puente, J., Arias, A., et al. (2005) “The effect of the thickness of the adhesive layer
on the ballistic limit of ceramic/metal amours. An experimental and numerical study.”
International Journal of Impact Engineering. 32:321-336.

75