Professional Documents
Culture Documents
by
Jiyoon Oh
in
(Civil Engineering)
(Vancouver)
April 2016
The experimental study completed in this thesis focused on timber-steel hybrid moment
connections using d=12.7mm diameter mild steel threaded rods glued into Douglas-Fir glulam
with polyurethane based adhesive. Two phases of experiments were conducted: the first to
determine the minimum design parameter values that result in a ductile tensile failure of the
glued-in steel rod, instead of a brittle timber or pull-out failure; and the second, to determine a
relationship between the different design parameters and the moment capacity of the connection.
The work established that the moment connection fails in a ductile manner due to rod yielding
and plasticizing, when the shear force induced into the system was less than 25% of the
maximum axial capacity of the steel rods. Then, ductile failure occurred even when the edge
distances of the steel rods were below the recommendation of a minimum 2.5d to prevent
splitting of the wood. Rod pull-out failure was prevented by having a glued-in embedment length
of the rods equal to or greater than 15d. In addition, ductility and equivalent viscous damping
ratio were found to decrease as the moment capacity of the connection increased.
The theoretical yield moment was calculated based on the assumption that the compression and
tension members are timber and steel, respectively, and by applying the concept that plane-
sections remain plane and the traditional elastic transform theory. The experimentally determined
yield moments were established to be a close match. The results of the research provide guidance
ii
Preface
This thesis is the original, unpublished and independent work by the author, Jiyoon Oh, under the
iii
Table of Contents
Abstract .......................................................................................................................................... ii
Acknowledgements ...................................................................................................................... xi
2.1.2 Recommendations for Preventing Wood Splitting and Group Tear-out .................... 5
2.1.3 Recommendations for Preventing Shearing along Glue Line and Pull-out ................ 6
2.1.4 Recommendations for Shear Concentration at the End of Glued-in Rods ................. 7
2.2.2 Timber Moment Frame Connections with Glued-In Steel Rods .............................. 10
3.2.2 Steel........................................................................................................................... 17
References .....................................................................................................................................71
v
Appendices ....................................................................................................................................76
vi
List of Tables
Table 3.1: Specified Strengths and Modulus of Elasticity for D.Fir 20f-E Glulam (CSA O86-14)
....................................................................................................................................................... 16
Table 3.7: Shear Force Relative to Maximum Pure Shear Capacity for 500mm Long Specimens
....................................................................................................................................................... 34
Table 3.8: Shear Force Relative to Maximum Pure Shear Capacity for 1000mm Long Specimens
....................................................................................................................................................... 37
Table 3.10: Summary of (a)-type Specimen Analysis under Monotonic Loading ....................... 51
Table 3.11: Summary of (b)-type Specimen Analysis under Monotonic Loading ....................... 53
vii
List of Figures
Figure 2.1 Edge Distance and Rod Spacing from Multiple Standards (Tlustochowicz et al., 2011)
......................................................................................................................................................... 5
Figure 2.3 Three Common Glued-In Steel Moment Joints (Fragiacomo & Batchelar, 2012) ..... 10
Figure 2.5 Additional Glued-In Steel Plate for Shear Reinforcement (Andreolli et al., 2011) .... 12
Figure 2.6 Concepts Assumed to Calculate Moment Resistance (Fragiacomo & Batchelar, 2012)
....................................................................................................................................................... 13
Figure 3.1: Layout of Phase 1 (a) U1 & U2, (b) U3, R3, U5 & R5, (c) U4, R4, U6 & R6 .......... 18
Figure 3.2: Reinforced (a) SE-R3&R5 and (b) SE-R4&R6 Specimens ....................................... 20
Figure 3.3: Layout of Phase 2 (a) A1, (b) B2, (c) C3, (d) D2, and (e) D4 Specimens ................. 22
Figure 3.8: Test Specimen Attached: (a) to Base Plate, (b) to Load Cell ..................................... 28
Figure 3.10: (a) LVDT Placement and (b) String Pod Placement ................................................ 29
Figure 3.11: CUREE Deformation Controlled Quasi-Static Cyclic Loading History .................. 30
Figure 3.12: Gap between Load Transfer Plate and Clamping Plate ............................................ 31
Figure 3.13: Shear Failure of 500mm Long Specimens (a), (b), (d) & (e) Wood - Splitting, (c)
Figure 3.19: Bending of Steel Plate (Clamp) while Testing D4(a)-1 (1)...................................... 40
Figure 3.20: Ductile Failure of (a) A1(a), (b) B2(a), (c) C3(a), (d) D2(a), and (e) D4(a)
Figure 3.21: Brittle Splitting Failure of (a) B2(a)-1 (2) and (b) D2(a)-1(2) under Monotonic
Loading ......................................................................................................................................... 44
Figure 3.32: Brittle Pull-out Failures of (a) A1(b), (b) C3(b), (c) D2(b) Specimens ................... 52
ix
Figure 3.35: Moment-Rotation Curve of D2(b) Specimens ......................................................... 55
Figure 3.36: Ductile Failure of (a) A1(a), (b) B2(a), (c) C3(a), (d) D2(a) and (e) D4(a) Specimens
Figure 3.37: Equivalent Viscous Damping Ratio for One Cycle (Piazza et al., 2011) ................. 59
x
Acknowledgements
I owe my endless gratitude to all my teachers and professors throughout my past, who have
inspired and motivated me. I owe special thanks to Dr. Thomas Tannert, my supervisor, for his
support and guidance throughout this study. Without him, this research project would not have
been possible.
I thank Thomas Leung, the industry partner on this project, for not only his extensive expertise
but also for his boundless encouragement and support; as well as, Western Archrib and
Purbond® for their generous contribution of glulam timber beams and CR-421® adhesives,
respectively.
I would also like to acknowledge the hard work and dedication from the technicians at Structures
and Wood Mechanic laboratories who made the fabrication and experimentation of the glued-in
steel rod timber moment connections possible. A special thanks to David Roberts and John
Wong at the Structures laboratory, George Lee at the Wood Mechanic laboratory, and Lawrence
Gunther and Joern Dettmer at the Centre for Advanced Wood Processing.
To my fellow classmates and friends, who have kept me sane through reports, midterms and
Finally, boundless gratitude is owed to my family for their undying support and encouragement
xi
Dedication
To my parents,
xii
Chapter 1: Introduction
1.1 Background
Wood is a renewable resource that is able to sequester greenhouse gases. In addition, timber
structures do not only cut down the carbon footprint of buildings, but are also aesthetically
pleasing and have the ability to create a healthy, nurturing atmosphere (Government of Canada,
2015). However, there are several challenges when using wood as structural material, such as its
combustibility. The British Columbia Building Code (2012) limits construction of combustible
material to 6 storeys, which was a significant development from its previous limit of 4 storeys.
Another challenge for using wood is when it is loaded in tension perpendicular to the grain or
shear parallel to the grain. Under those conditions, wood is a brittle material with very low post-
elastic deformation capacity. Furthermore, due to wood being a natural material, unlike concrete
and steel, its defects, such as knots, and its lack of homogeneity cause the material to exhibit
large variability in its properties (Andreolli et al., 2011). However, even with the challenges of
wood as a structural material, due to the aforementioned advantages and its high strength-to-
weight ratio of wood (Gilbert et al., 2015), the demand and desire for taller wood structures is
increasing around the world, even in high seismic areas, such as British Columbia.
For a high seismic area, ductility, which is the ability for plastic deformation without sudden
brittle failure, is a key design factor. In order to achieve a ductile behavior for timber structures,
the joints of the structure have to be designed to be the ductile and energy dissipative component
while the main wood members remain in their elastic range (Andreolli et al., 2011). One popular
method to achieve the desirable ductile behavior in the joints of timber structures is to create a
hybrid connection using timber and steel such as glued-in steel rods. For a glued-in steel rod
1
timber connection to provide the desired ductility and energy dissipating mechanism, it is crucial
that the steel yields, plasticizes and fails without premature failure elsewhere. Also for non-
seismic design, it is often more favourable to have the failing mechanism as the yielding of the
steel, as steel properties are more predictable and consistent than wood properties.
As the storey limits for wood frame structures increase, more efficient lateral force resisting
systems are required, such as moment frames. It is desirable to have a ductile moment frame to
ensure that there is yielding of the members to dissipate energy and to give sufficient amount of
warning before the system fails. Timber moment frames using connections with glued-in steel
rods have a great potential for becoming a practical, reliable lateral force resisting system.
Considerable research has been done in regards to axially loaded glued-in steel rods within
timber to determine methods to predict the behavior of the system, such as work done by Blass
and Laskewitz (1999) and del Senno et al. (2004). However, there has been little research done
with the glued-in steel rods as moment connections and further research is required to better
1.3 Objectives
The main objectives of this research is: to determine the requirements that will ensure ductile
yielding of the 12.7mm diameter steel rods as the governing failure instead of a sudden brittle
failure of the timber; and to determine the impact of different parameters, such as number of
glued-in rod, edge/rod spacing and embedment length, on the moment capacity of the
connection. Two phase experimental testing is completed in order to fulfill the objectives of this
research.
2
Chapter 2: State of the Art
Experimental research and theoretical studies of connections consisting of steel rods glued into
timber members began in the 1980’s. The studies began in pursue of general criteria and design
guidelines for these connections and the investigation is still ongoing today (Otero Chans et al.,
2010). These connections have become a heavily researched topic due to their many advantages,
such as its high strength and stiffness properties, efficient transfer load mechanism and being a
light weight solution for both existing and new structures (Tlustochowicz et al., 2011). In
addition, having the steel-rods encased into the timber member, the connection is not only
aesthetically pleasing, but the steel rods are protected from fire (Madhoushi & Ansell, 2008) and
corrosion by the surrounding timber (Gattesco et al., 2010). In fact only 20mm of timber layer is
required to covering the steel rods for each 30min of fire resistance (Kangas, 2000).
Individually, the properties, such as stiffness and strength, and the behaviour of timber, steel and
adhesives under loading are well known and can be predicted; however, combining these three
materials to create a hybrid connection, such as a glued-in steel rod timber connection, produces
a complex system for which the behaviour is difficult to predict (Tlustochowicz et al., 2011). As
a consequence, even with all the research put into glued-in rods, no general consensus has been
reached on a common standardization; thus these connections are commonly used in just a few
places around the world, mostly European countries (Tlustochowicz et al., 2011).
3
Out of the materials that can be used for the glued-in rod, such as glass fibre reinforced polymer
(GRFP), carbon fibre reinforced polymer (CFRP), hardwood and steel, the most commonly used
material is steel due to its advantageous property of being a ductile material (Tlustochowicz et
al., 2011). Steel rods that are threaded are more favourable due to the fact that the threads
provide an increased area for adhesion and mechanical interlocking creating a stronger bond
between the rods and the timber member. Additionally, the steel threads allow for easy assembly
if the glued-in timber joint were to be connected to a steel member (Tlustochowicz et al., 2011).
Wood, when subjected to tensile and shear forces, is a brittle material with very low post-elastic
deformation capacity (Andreolli et al., 2011). As a result, timber failures are abrupt, sudden
failures that occur without any given warning, creating an undesirable failure mode. Timber
connections with glued-in rods, however, can be designed in such a way that they exhibit a
ductile failure mode as long as the steel rods provide sufficient yielding and plasticization prior
1. Splitting of wood
3. Rod pull-out
5. Tensile failure of timber at the end of the rod (Parida et al., 2013).
Even though the research community has not come to a consensus on a design standardization of
timber glued-in rod connections, through experimental results, recommendations have been put
4
2.1.2 Recommendations for Preventing Wood Splitting and Group Tear-out
Wood splitting and group tear-out for multiple glued-in rods are common brittle failures modes
that are very susceptible to occur if the necessary precautions are not taken. To avoid these
undesirable failure modes, minimum edge distances and rod spacing are suggested (Parida et al.,
2013). There are many different suggestions between different sources; however, the most
stringent recommendation are given by the German regulation (DIN 1052:2004-08) with an edge
distance of two and a half times the rod diameter (2.5d) and bar spacing of 5d (Tlustochowicz et
al., 2011). Figure 2.1 outlines the different recommendations from different standards on edge
distance and rod spacing for rods glued-in parallel to the grain of the wood. Following these
guidelines should prevent splitting or group tear-out within the wood before all other failures, i.e.
Figure 2.1 Edge Distance and Rod Spacing from Multiple Standards (Tlustochowicz et al., 2011)
5
An alternative measure to prevent splitting of the wood is to reinforce the member by either
using screws perpendicular to the grain or by gluing plates of plywood, or other materials, onto
the ends of the timber member (Jensen & Quenneville, 2009). Self-tapping screws, which are
screws that do not require pre-drilling, are an effective way of strengthening timber in their two
weakest strength properties: tension perpendicular to the grain and longitudinal shear strength
Lam et al. (2008) conducted an experiment using self-tapping screws as reinforcement for the
timber perpendicular to the grain for bolted glulam connections with slotted steel plates. It was
found that while the unreinforced specimens failed in a brittle manner due to splitting of the
wood under monotonic and cyclic loading, the reinforced connections saw no splitting in the
timber and in fact the failure mode shifted from brittle failure to ductile failure. Additionally, the
retrofitted specimens of the original damaged unreinforced specimens achieved some ductility as
well, though not as much as the initial reinforced specimens. The ductility was limited due to the
2.1.3 Recommendations for Preventing Shearing along Glue Line and Pull-out
To prevent rod pull-out failures and shear failure within the glue line, the glued-in rod
connection should be designed in such a way that capacity of the pull-out/shear within the glue
line strength should be greater than the capacity of all other failure modes. For threaded rods, the
threads provide mechanical interlocking between the steel and the adhesive, thus shear failure
6
One way to ensure adequate capacity to prevent rod pull-out due to shear failure along the
adhesive and timber interface is through sufficient embedment lengths (le) or ratio between
embedment length and diameter of the rod (le/d) called the slender ratio. Unfortunately,
experimental studies have found that the relationship between pull-out strength of rods and rod
2011). Otero Chans et al. (2010) concluded that the pull-out capacity of the joint does increase
with increased embedment length; however, at a certain embedment length, the increase in the
capacity of the joint reduces significantly, indicating that the relationship between joint capacity
and embedment length is not a linear one. Recent work by González (2015) showed that glued-in
mild steel rods with diameter of 12.7mm yielded and had a ductile failure mode within the steel
rod when the embedment length was greater than or equal to 10d. However, it was found that for
19mm diameter mild steel rods, an embedment length of 10d led to wood shear failure. Only
when the embedment length was 20d, rod yielding occurred, but even then, it was observed that
some specimens failed due to splitting of the wood. As it can be seen through experimental
studies, estimating the pull-out strength of glued-in rod connections is not a straight forward
linear calculation.
Stress concentrations occur in the presence of sudden changes in section properties and/or when
the direction of the stress path is forced to change (Fragiacomo & Batchelar, 2012). As a result,
the steel rods within the timber member produce a stress concentration at the transition cross-
section where the steel embedment begins, which could lead to tensile rupture of the timber
member creating a premature failure (Fragiacomo & Batchelar, 2012). To prevent this type of
failure, Batchelar (2007) recommended locating the cross-sectional area of timber where the rod
7
forces are transferred, and limiting the tensile stress, at that transitional area, to the allowable
timber tensile stress. Batchelar (2007) suggested the transitional areas (Ac) for different scenarios
1. Single glued-in rod: cross-section width of the timber (b) multiplied by two times the
2. Individual internal rods for multiple layers of glued-in rods: cross-section width of the
timber (b) multiplied by the spacing between the rods (s), Ac = b(s).
3. Outer edge rods for multiple layers of glued-in rods: cross-section width of the timber
member (b) multiplied by the summation of edge distance of the glued-in rod (e) and half
Furthermore, Tlustochowicz et al. (2011) suggested that the two elements being connected, the
steel rod (Er and Ar) and timber member (Ew and Aw), have similar tensile stiffness (EwAw =
ErAr) for a smooth load transfer. Tlustochowicz et al. (2011) further recommended using
multiple smaller rods distributed equally across the timber member rather than utilizing one
single large rod. For multiple tension rods, Batchelar (2007) stated that the tensile stresses within
8
the timber member can be restrained by differing the choice of rod diameters and the tensile
strength of the rod. In addition, multiple tension rods can be designed with different embedment
lengths, i.e. staggering the rod embedment depths, in order to prevent accumulation of stress
In addition to the previous recommendations, using mild steel rods with smaller diameter, as
opposed to high strength steel rods with larger diameter, will favour yielding of the steel rods as
the primary failure mode and necking down the steel rods will provide additional assurance of
plasticization of the steel being the governing failure mode (Parida et al., 2013). Necking of the
steel rod will also ensure that the yielding of the rod will occur at the reduced cross-section; thus
enable the designer to control the location of the yielding/plasticization point as well.
2.2.1 Overview
Moment frames as lateral force resisting systems are favoured over other conventional frame
systems such as concentrically braced frames, crossed braced systems, or chevron brace frames
because of their ability to provide an open, uninterrupted system (Metten, 2012). There are
clauses in the Canadian steel (CSA S16-09 (2010)) and concrete (CSA A23.3-14 (2014))
material code, and the National Building Code of Canada (NBCC 2010) that allow for the design
exist for moment frames constructed from timber. This is mainly due to the fact that ductility is
required for moment frames, which cannot be achieved with timber members alone.
9
2.2.2 Timber Moment Frame Connections with Glued-In Steel Rods
There are many ways that glued-in steel rods can be incorporated into timber joints to create a
timber moment frame, but the three most simple and common configurations are:
A. To have the steel rods through both the column and the beam where the rods are glued
partly along the length, either only within the column or the beam;
B. To have the steel rods through and fully glued within both the column and the beam;
C. To incorporate a steel hub and connecting it to the timber column and beam using glued-
Figure 2.3 Three Common Glued-In Steel Moment Joints (Fragiacomo & Batchelar, 2012)
In configuration A, the steel rods go through both the timber beam and the column, but glued
only within the column or only within the beam. In configuration B, the steel rods go through
and are glued to both the beam and column. Both these configurations create a joint where the
steel rods are fully concealed; however, the joint has an overlapping zone, between the timber
beam and column, where different loading directions in the timber from perpendicular to the
grain to parallel to the grain occur (Fragiacomo & Batchelar, 2012). The overlapping zone
creates a complex region due to the anisotropy nature of timber, which could lead to significant
deformations and stress concentrations. Configuration C, on the other hand, where an isotropic
10
material is introduced at the location where timber beam to timber column overlap, eliminate the
different loading direction of the timber, thus preventing complications that arise from loading
timber in two different directions (Fragiacomo & Batchelar, 2012). The moment connections
shown in Figure 2.3 can alternatively be arranged to have the column extended upwards to have
the beam flush against the column (Fragiacomo & Batchelar, 2012).
A stub of steel sections, such as a T-section welded onto an end plate, can be used for
Configuration C, as shown in Figure 2.4 (Andreolli et al., 2011). However, it should be noted
that depending on the thickness and strength of the steel stub section and the glued-in steel rods,
the failure mode can range from complete yielding and plasticization of only the flange of the
steel stub, to combined yielding and plasticization of the steel stub flange and glued-in steel rods,
to yielding and plasticization of only the glued-in steel rods (Andreolli et al., 2011).
In addition to rods placed at the outer ends of the beam or column for moment resistance, steel
plates or additional glued-in steel rods can be inserted at the center of the beam or column for
shear resistance when the shear force is high (Andreolli et al., 2011; Buchanan et al., 2001).
Figure 2.5 illustrates an example of how a steel plate can be incorporated into a glued-in steel rod
moment connection with Configuration C. However, when the shear force is low and the lateral
11
force can be resisted by the glued-in steel rods utilized for moment resistance alone, then the
Figure 2.5 Additional Glued-In Steel Plate for Shear Reinforcement (Andreolli et al., 2011)
For any structural system, accurately predicting the capacity is a crucial part of the design
process to ensure a safe, efficient force resisting system, whether it be for gravity or lateral
design. The same goes for hybrid timber-steel moment frame systems utilizing glued-in steel
rods in timber. Fragiacomo and Batchelar (2012) proposed a theoretical approach in designing
timber moment connection joints using glued-in steel rods. The assumptions made were the
concept that plane sections remain plane and the traditional elastic transform section theory,
which predicts a linear stress distribution for the timber compression zone. The procedure
outlined by Fragiacomo and Batchelar (2012), illustrated in Figure 2.6, is similar to the
procedure used to calculate the moment resistance of reinforced concrete beams in CSA A23.3-
14 (2014).
12
Figure 2.6 Concepts Assumed to Calculate Moment Resistance (Fragiacomo & Batchelar, 2012)
It is assumed that the tensile force is transmitted purely through the steel rods and the
compressive force is transmitted through the timber-timber interface. Thus similar to reinforced
concrete sections, the tensile forces are resisted by the steel rods, whereas the compression forces
are resisted by only the timber section under compression if the steel rods are not fully glued
along the beam and the column, or by timber and steel rod under compression if the steel rods
are fully glued along their length within the beam and column (Fragiacomo & Batchelar, 2012).
With this in mind, the internal resistive tensile and compression couple forces can be found,
along with the internal lever arm, to calculate the moment resistance of the connection. The
ultimate stress of the steel should be used in calculating the moment resistance of the hybrid
connection instead of yield stress in order to take strain hardening into consideration, which will
The glued-in steel rod connection is recommended to be designed in such a way that the steel
rods yield, plasticize and fail before the timber section reaches its full compressive capacity,
similar to an under-reinforced concrete beam (Fragiacomo & Batchelar, 2012). This way, sudden
brittle system failure will be prevented; and even if the ultimate failure of the “under-reinforced”
13
timber section was due to compression failure, the steel would develop plastic strains and give
warning beforehand (Fragiacomo & Batchelar, 2012). Furthermore, the connection should be
capacity designed to ensure that the timber member and interface stay within the elastic range, to
prevent brittle failures and fully utilize the ductility of the steel (Gilbert et al., 2015). This means
that using the ultimate strength of the steel rods, instead of the yield strength, will give a more
14
Chapter 3: Experimental Investigations
3.1 Overview
The experimental work carried out for this thesis was conducted at the Structures and Wood
Mechanics laboratories of The University of British Columbia between May 2015 and November
2015. The experiments were divided into two phases: Phase 1 (preliminary testing) and Phase 2
(main testing). The objective of Phase 1 was to determine the minimum requirement of edge and
rod spacing, embedment length and length of the glulam beam, which would ensure in ductile
failure. The objective of Phase 2 was to design a ductile connection utilizing the results obtained
from Phase 1 and to determine the relationship between the glued-in steel rod layouts and the
3.2 Materials
Glued-in steel rod timber moment connections are timber-steel hybrid connection that consists of
three materials, timber, steel and adhesive, which could be varied; however, since the main goal
of this research was to determine a method on designing and predicting the capacity of ductile
glued-in steel rod connections by manipulating the location of the glued-in steel rods, the three
component of the materials were kept constant. The timber beams utilized were Douglas-Fir 20f-
E grade glue-laminated (glulam) wood. The steel rods used for this experiment were threaded
mild grade 12.7mm diameter. Finally, the adhesive was CR-421®, a two component
15
3.2.1 Timber
The specimens for the experimental work of this thesis were fabricated using Douglas-Fir glulam
of grade 20f-E of cross-section ranging from 80mm by 266mm to 130mm by 456mm. These are
the typical cross-sections available in Canada for glulam beams and were selected in order to
closely mimic what would be specified in practice. The specified strength for the glulam is
summarized on Table 3.1 (CSA O86-14). The average apparent relative density of the glulam
beams were determined to be 563kg/m3 with the minimum and maximum of 514kg/m3 and
609kg/m3, respectively, The average moisture content (MC) measured of these beams, directly
before testing, for Phase 2, were 10.9% with the minimum and maximum MC of 9.3% and
12.7%, respectively.
Table 3.1: Specified Strengths and Modulus of Elasticity for D.Fir 20f-E Glulam (CSA O86-14)
16
3.2.2 Steel
Threaded mild steel bars of diameter 12.7mm (1/2 inch), with an average yield capacity of
360MPa (σy,mean = 360MPa) were utilized for this thesis. In preceding work, Gonzalez (2015)
experimentally confirmed the yield strength of the rods by testing five random samples following
the ASTM F606M-14 procedure. The 12.7mm diameter steel rods were chosen, as opposed to
the 19mm diameter rods, since it was found that the 12.7mm diameter steel rods were able to
reach ductile tensile failure in the steel rods once adequate embedment length was provided,
whereas the 19mm diameter rods exhibited pull-out failure even with embedment length of 20d.
In addition, the threaded steel rods were selected because, as mentioned previously, the threaded
aspect of the steel rods allow for mechanical interlocking between the steel and the adhesive
3.2.3 Adhesive
A two component polyurethane (PUR) adhesive, CR-421®, provided by Purbon®, was used to
glue the steel threaded rods into the timber. PUR based adhesives are commonly used in the field
(Gonzalez, 2015), thus seemed to be an appropriate choice for this research since one of the
goals was to investigate a feasible design and construction method that could be used in the
engineering world. In addition, PUR has gap filling properties which epoxy glue, another
common adhesive used on-site, does not possess (Gonzalez, 2015). The main properties of the
adhesive utilized were: work time of 10 – 20 minutes, curing time of 7days, viscosity of
9000cps, expected shear strength (parallel to grain) of 7.8MPa, and tensile elongated at failure at
17
3.3 Specimen Descriptions
3.3.1 Phase 1
which are required to ensure a ductile failure within the glued-in steel rods. Two different cross
sections of Douglas-Fir glulam beams were used for phase 1: 80mm by 266mm and 265mm by
266mm in lengths of 500mm and 1000mm. A total of 22 specimens were fabricated and
subsequently tested under quasi-static monotonic and quasi-static revered cyclic loading. Other
parameter variations were the edge distance (e) for the glued-in steel rods, embedment length
(le), and the addition of perpendicular to grain reinforcements, to prevent splitting of the wood.
Figure 3.1 and Table 3.2 summarizes the parameters for each specimen.
Figure 3.1: Layout of Phase 1 (a) U1 & U2, (b) U3, R3, U5 & R5, (c) U4, R4, U6 & R6
18
Table 3.2: Phase 1 Specimen Layout and Parameter Variation
Edge distance, distance from the edge of the glulam beam to the center of the glued-in steel rods,
were either 20mm, less than 2d, or 45mm, greater than 3d, and were given the initials SE (small
edge distance) or LE (large edge distance), respectively. The SE specimens do not meet the
minimum edge distance for glued-in rods recommended by most guidelines (2.5d), thus
theoretically, they should fail due to splitting of the wood beam. The LE specimens with edge
distance greater than 3d exceed the minimum recommended; thus splitting should not occur.
19
Since SE specimens were expected to have brittle wood splitting failure, the SE specimens with
cross section of 265mm by 266mm were fabricated with and without perpendicular to the grain
reinforcement to determine if the additional reinforcements could shift the failure mode from a
brittle to a ductile one. The specimens with cross section of 80mm by 266mm could not be
reinforced since the base dimension of 80mm did not provide adequate width for the required
spacing of the reinforcements. The labels “U” and “R” were placed after “SE-” or “LE-” to
ASSY® VG cylinder head self-tapping screws, with length of 150mm and diameter of 8mm,
were used as reinforcement. The screws were drilled into the glulam as close as possible to either
side of the steel rods. They were placed 50mm from the edge and SE-R3 and SE-R5, with one
steel rod on each side of the glulam, were fitted with an extra row of screws spaced at 50mm on
center from the first row. Figure 3.2 illustrates the placement of the self-tapping screws.
(a) (b)
perpendicular to the grain reinforcement could enhance the capacity of the specimens.
Unfortunately, none of the unreinforced specimens were retrofitted posterior to testing due to the
fact that the failure mode was either shear of the rod or localized splitting of the wood around the
glued-in steel rod that could not be retrofitted with self-tapping screws.
Finally, two different embedment lengths were used for each layout: 15d and 20d. Based on
preceding work by Gonzalez (2015), embeddment length of 15d, for 12.7mm diameter mild steel
rod, should prevent brittle pull-out failure and favour ductile steel tensile failure; furthermore,
increasing the embedment length past 15d should not lead to an increase of the pull-out strength.
Specimens with 20d were designed to double check that this was the case.
3.3.2 Phase 2
Once the minimum edge distance of 2.5d and rod spacing of 5d were determined to ensure a
ductile failure of the connections, for Phase 2, these parameter were set constant and different
layouts were investigated with respect to the connection’s capacity. In addition, the embedment
length of the glued-in steel rod was also varied to verify that an embedment length of less than
15d would produce brittle pull-out failure. Figure 3.3 and Table 3.3 summarizes the parameters
for Phase 2. Four monotonic tests were done on all specimens except for D4(a), where only three
monotonic tests were performed. Four quasi-static cyclic tests were done on specimens A1(a)
and D2(a), and three quasi-static cyclic tests were performed on specimens B2(a), C3(a) and
D4(a).
21
(a) (b) (c)
(d) (e)
Figure 3.3: Layout of Phase 2 (a) A1, (b) B2, (c) C3, (d) D2, and (e) D4 Specimens
22
A different labeling system was used for Phase 2 since the edge distance was kept constant;
specimens were labeled based on their timber cross section, number of rods and the embedment
length of the glued-in steel rods. The first alphabetical letter used indicated the dimension of the
specimen: “A” – 80mm by 266mm; “B” – 130mm by 266mm; “C” – 175mm by 266mm; “D” –
130mm by 456mm. The number following the alphabetical letter specified the number of rods
under tension; thus was half the number of rods utilized in the system. Finally the specimen was
indicated with “(a)” or “(b)” to designate embedment lengths of 16d and 8d, respectively.
Furthermore, in order to maximize the material usage, phase 2 specimens were fabricated with
glued-in steel rods on both sides of the timber beam, as shown in Figure 3.5. This allowed one
timber member to produce four test specimens for monotonic testing since conceptually the
damage would only occur on the glued-in steel rods on the tension side, whereas the glued-in
steel rods on the compression side would obtain no damage; thus the specimen could be flipped
over and be tested again with the compression glued-in steel rods now on the tension side. Self-
tapping screws were utilized to stabilize the specimen in place of the damaged glued-in steel
rods, as show in Figure 3.4, for the second tests performed on each sides.
23
For quasi-static cyclic testing, having glued-in steel rods on both sides of the timber beam
allowed two test specimens per timber member since all glued-in steel rods would at one point be
in tension during the testing. As a consequence, the timber beams were cut to length of 1400mm
in order to provide adequate length of the beam to ensure a ductile failure, while also accounting
for a buffer zone where the other set of glued-in rods are located.
3.4 Manufacturing
The manufacturing of the test species consisted of four stages. First, the glulam was cut to the
desired length and the holes, according to the specified layout with the correct diameter and
depth, were drilled. A CNC heavy timber processor (Hundegger® Robot Drive) was used to cut
and drill all the timber specimens in order to obtain precision and increase efficiency. Precision
of the location of the holes was crucial due to the limited tolerance provide by the base plate used
for testing, which will be further discussed later. In order to account for glue line thickness of
approximately 2mm, a 16mm drill bit was used to drill the holes for the 12.7mm diameter steel
rods. The holes were drilled to the desired embedment length of 191mm (15d), 254mm (20d),
24
Second, the threaded rods, which were available in 3ft lengths, were cut into the desired lengths
using a rebar cutter. The rods were used in the condition provided by the supplier with no
additional treatments, such as cleaning or coating them, to mimic a typical site condition. The
rods were cut while being rotated to prevent the threads being damaged to allow the nuts to be
screwed on; however, even with this precaution, some rods had to be re-threaded manually. In
phase 1, the rods were cut to lengths of the embedment length with an additional 150mm for the
washer and nuts to attach the rods to the end base plate. It was found that this length was
excessive and much time was spend hand tightening the nuts with a wrench. Thus, for phase 2,
the rods were cut with only 75mm additional length in order to use a bolt gun to screw on the
Third, the rods were glued into the holes using a duel cartridge caulking gun to insert the
adhesive. The two part PUR adhesive was filled to approximately three-quarters of the holes,
then the rods were inserted into the hole in a twisting motion to minimize entrapped air.
Overflow of PUR adhesive onto the surface of the wood specimen, as shown in Figure 3.6,
indicated that sufficient adhesive was used to cover the entire length of the rod. Then, toothpicks
were used to ensure the rods were centered and straight, and a level was used to double check the
levelness of the rods. During phase 1, the hardened PUR adhesive due to the excess overflow of
the glue onto the face of the specimen prevented the face of the glulam beam to be flush with the
steel base plate. Theoretically this would shift the compression area, thus in turn the neutral axis
from the initial predictions. As a result, for phase 2, it was ensured that the excess glue was
chipped away to provide a flush timber to steel base plate connection to enable a correct
25
Figure 3.6: Gluing and Centering of Rods
Fourth and finally, after gluing was completed, the specimens were left to air-cure overnight then
the following morning, were placed into constant temperature and humidity room of 20°C and
65%RH, respectively. In order to cure the adhesive up to minimum of 95% of their expected end
bond strength, which occurred after 7 days of curing (Lehringer, 2012) none of the specimens
The general plan view of the test set up is illustrated in Figure 3.7. The base plate to provide a
fixed connection was fabricated out of 25mm thick steel plate with holes drilled through to
accommodate multiple rod layouts. Due to the different size of the beams used and different edge
and spacing between rods, the drilled hole location of phase 1 base plate did not match the rod
layouts of phase 2, in addition, a larger base plate was required; thus a new base plate had to be
fabricated for phase 2. In order to account for imperfection in fabricating, 18mm diameter holes
were drilled through the base plate to give a total of 5mm tolerance for the 12.7mm diameter
steel rods. The 5mm tolerance provided was adequate for set-up without over compensating.
26
Figure 3.7: Overview of Test Setup (Plan View)
The base plate was then bolted onto a steel reaction frame. The rods were slotted into the drilled
through holes within the steel base plate then bolted on with a washer and two nuts, as shown in
Figure 3.8(a), to provide a fixed connection. Two nuts were used because it was found that using
just one was not adequate to withstand the tension forces and resulted in stripping the nuts off the
rods. As one end of the specimen was fixed onto the base plate with nuts, the opposite end was
fixed onto the loading cell by two plates, with depth of 100mm, clamping onto the specimen as
depicted on Figure 3.8(b). The two plates were tightened and held in place by washers and nuts.
Initially, steel plates were used as the clamps for the ease of wrenching the nuts on and off;
however, the plates were found to be lacking in strength and started bending when the larger
specimens were tested. As a result, C-channels, with a depth of 130mm, were used as the
clamping plate instead to provide sufficient strength and stability. This was crucial especially for
the quasi cyclic testing. The result of the experimental set-up was a fixed end connection of a
beam, provided by the rod connection to the base plate, with a load applied perpendicular to the
beam axis at the opposite end as shown in Figure 3.9. The actuator had a displacement capacity
of ± 77.1mm (3inches) and a load capacity of ± 89kN. The specifications for the equipment used
27
Table 3.4: Equipment Specifications
(a) (b)
Figure 3.8: Test Specimen Attached: (a) to Base Plate, (b) to Load Cell
28
Four linear variable differential transformers (LVDT), with specifications outlined in Table 3.4,
were used to measure the displacement of the wood specimens from the steel base plate in the X
and Y direction, on both sides of the beam, as illustrated in Figure 3.10(a). On the other end, a
string pod was hooked onto the clamping plates for phase 1 and placed 1355mm away from the
face of the steel base plate and screwed onto the timber beam for phase 2 to calculate the rotation
of the specimens. The string pod is shown in Figure 3.10(b) and the specification of the string
pod is summarized on Table 3.4. The load was measured with a calibrated load cell.
(a) (b)
Figure 3.10: (a) LVDT Placement and (b) String Pod Placement
The specimens were tested under displacement controlled quasi-static monotonic and reversed
cyclic loading. The actuator was powered by hydraulic pressure and controlled with MTS 458.10
MicroConsole through DASYLab 11.0 software. In addition, DASYLab software was used to
collect the experimental data from the load cell, as well as from the LVDT and displacement
sensor. The maximum displacement for the monotonic loading was 50mm. However, during
phase 2, it was found that the larger specimens reached their capacity at displacements greater
than 50mm, thus failure did not occur even when the maximum displacement was reached. As a
29
result, the peak displacement was increased to 75mm. The loading rate, for phase 1, was set for
For the reversed cyclic tests, the deformation controlled quasi-static cyclic testing protocol from
CUREE publication was used. The protocol is commonly applied for testing woodframe
structures. It consists of initial, primary and trailing cycles, which use a reference deformation as
the peak deformation resulting from the monotonic testing (Krawinkler et al., 2000). The initial
cycles are small cycles that are 5% of the reference deformation. These initial cycles serve as a
check for the load cell, measuring equipment, as well as an indication of the force-deformation
response at low amplitudes (Krawinkler et al., 2000). Primary cycles are peak cycles that
gradually increase in amplitude as the loading history proceeds. The primary cycles are followed
by trailing cycles that are 75% of the amplitude of the preceding primary cycle as illustrated on
30
3.7 Phase 1 Results
Phase 1 of the experiment was set as preliminary testing of phase 2 to determine the parameters
required to induce ductile failure of the glued-in steel rod moment connection. As a result, only
one specimen was fabricated and tested for each configuration. The results of phase 1 are
summarized in Table 3.5 and Table 3.6 for monotonic and cyclic loading, respectively. The
“measured moment arm” was the distance measured from the interface of the steel base plate and
wood beam to the center of the 100mm wide clamping plate; thus the true moment arm could be
±50mm. It should be noted that bending of the steel plate, which was used to transfer the load
from the load cell, was observed during testing due to the opening between the load transferring
steel plate and the clamping steel plate. In consequence, wooden blocks were utilized to help
reinforce the transfer plate, as illustrated in Figure 3.12; however, the wooden blocks were
insufficient. For phase 2 testing, the load transfer plate was replaced with a much stronger and
stiffer I-section.
Figure 3.12: Gap between Load Transfer Plate and Clamping Plate
31
Table 3.5: Summary of Phase 1 Monotonic Testing
None of the specimens with length of 500mm failed in a ductile manner, not even the “LE”
specimens with adequate edge spacing, which was theorized to fail due to yielding and
plasticization of the glued-in steel rods. Instead all specimens of 500mm length failed suddenly
due to shear caused by splitting of the wood or shear failure of the glued-in steel rods. However,
as expected, pull-out failure was avoided for all twenty specimens, except for one (LE-U4), by
32
The 1000mm long specimens, on the other hand, all failed in a ductile manner due to tensile
failure of the steel rods, including the “SE” specimens, which were hypothesized to fail due to
splitting of the wood since they were designed with inadequate edge distance. Slippage between
the clamping steel plates occurred during testing of specimen “SE-U1-1m”, thus the applied load
dropped significantly whenever the clamping plates slipped and no failure of the specimen
occurred. The cyclic testing for the “LE” specimens with length of 1000mm also failed due to
tensile stress of the glued-in steel rods. The failure occurred at a load quite close to the failure
The shear force induced into the 500mm long beam specimens was quite large due to the short
lever arm averaging approximately 458mm. In fact, the shear force introduced into the specimen
was about 53% of the maximum pure shear capacity of the steel rods based on Von Mises Yield
Criteria. As a result, it was not surprising to observe that the shear force had a major effect on the
failure mode of all the 500mm length specimens, whether in the steel rods or in the glulam. The
shear force values relative to the maximum pure shear capacity of the steel rods for each 500mm
long specimens are summarized in Table 3.7 and failure modes outlined on the table are shown
in Figure 3.13.
33
Table 3.7: Shear Force Relative to Maximum Pure Shear Capacity for 500mm Long Specimens
34
(a) (b) (c)
(d) (e)
Figure 3.13: Shear Failure of 500mm Long Specimens (a), (b), (d) & (e) Wood - Splitting, (c) Steel – Shear
However, since shear resistance of the system is not the main objective of this research, failure
due to shear was not investigated in-depth. Instead, the main focus for phase 1, was on
prevention of shear failure. Based on uniaxial tension and shear interaction plot following Von
Mises Yield Criteria, shown in Figure 3.14, in order for the steel rods to achieve 90% of their
axial strength, or greater, the shear stress induced into the rods must be 25%, or less, of the axial
capacity. Thus, since flexure response of the connection is the main focus, the connection should
be designed in a way to ensure that the shear stress introduced into the steel rods are within 25%
of the axial capacity in order to obtain as close to the full axial capacity of the rods.
35
1
0.9
0.8
0.7
τxy/σy 0.6
0.5
0.4
0.3
0.2
0.1
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
σxx/σy
Figure 3.14: Uniaxial and Shear Interaction based on Von Mises Yield Criteria
The 1000mm long specimens had an average of 26% of the pure shear capacity (or 15% of the
ultimate axial stress) applied to the system before failure occurred, as summarized in Table 3.8;
thus it was not surprising to observe the specimens fail in a ductile manner due to plasticization
of the steel rods caused by tensile stresses. What was unexpected and intriguing was that the SE-
U1-1m specimen, which was expected to fail by splitting of the timber since it was designed with
insufficient edge spacing, was also observed to have significant yielding of the steel rod and had
The moment-rotation plot of the 1000mm specimens are shown in Figure 3.15. The SE-U1-1m
specimen had approximately 1.38 times more capacity than the average of the 1m long “LE”
specimens due to the greater moment arm between the tension force within the glued-in steel rod
and compression induced in the timber. Looking at the 1000mm long specimen results, it seems
that ensuring shear stress of 25%, or less, of the ultimate axial strength, within the system, will
36
Table 3.8: Shear Force Relative to Maximum Pure Shear Capacity for 1000mm Long Specimens
18
16
14
Moment [kNm]
12
10
8
6
4
2
0
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04
Rotation [rad]
The quasi-static cyclic testing of 1000mm long “LE” specimens had a ductile failure. The glued-
in steel rod yielded and plasticized through the first six primary cycles and their trailing cycles
before it completely failed during the final primary cycle with the peak amplitude displacement.
The hysteresis of LE-U1-1m and LE-U2-1m are shown in Figure 3.16 and Figure 3.17
respectively. Due to testing just one specimen per layout, no in-depth analysis was done for the
results of the 1000mm specimens. The in-depth analysis of ductile timber moment connection
37
15
10
5
Moment (kNm)
0
-0.02 -0.01 0 0.01 0.02 0.03 0.04
-5
-10
-15
Rotation (Rad)
15
10
5
Moment (kNm)
0
-0.01 -0.005 0 0.005 0.01 0.015 0.02 0.025 0.03
-5
-10
-15
Rotation (Rad)
As mentioned previously, and can be seen in Figure 3.13, the splitting of the timber, for “SE”
unreinforced specimens, occurred in such a way that they could not be retrofitted with screws.
Furthermore, the specimens that were initially fabricated with screws as reinforcement
perpendicular to the grain also failed in shear, including splitting within the timber, which is
38
shown in Figure 3.18. Again these splitting failures in the wood formed in such a way that the
The results (maximum load and displacement) of the monotonic testing are shown in Table 3.9.
Four monotonic tests were performed for most of the series. The “- #” after the specimen name
indicates which specimen beam was used and the “(#)” at the very end indicates the test number
of the specimen. Tests (1) and (3) were done on opposite ends of the timber beam with no
damaged to the glued-in steel rods. Test (2) and (4) were done on the same ends of the beam as
test (1) and (3) respectively, however, with the beam flipped to have the initial compression rods
now in tension, and with self-tapping screws in place to help stabilized the beam instead of the
damaged glued-in steel rods that were under tension from the test prior.
The labeling corresponding to the orientation of the tests was applied for all monotonic tested
specimens except for D4(a). The sudden transfer of loads at failure, from the glued-in steel rods
on the tension side, caused pull-out failure of the glued-in steel rods on the compression side as
39
well as splitting of the timber beam, which is shown in Figure 3.20(e). As a result, one end of the
beam could not be retested flipped over since the glued-in steel rods, which were initially on the
compression side prior, was damaged as well during the first set of testing. In consequence, only
three monotonic tests were performed for D4(a) and each test was done on different ends of the
beam. In addition, when the first test member was tested, D4(a)-1 (1), the capacity of the
specimen was so large that it bent the steel plate that was being used as clamps in the process, as
shown in Figure 3.19, rendering the test result invalid. In consequence, the test result of D4(a)-1
(1) was excluded from analysis. The remaining tests were completed after the steel plate clamp
Figure 3.19: Bending of Steel Plate (Clamp) while Testing D4(a)-1 (1)
It should also be noted that specimen C3(a)-2 (1) did not fail under the maximum displacement
set at 50mm, as a result, no failure load and displacement was found. The remaining of the
monotonic test which included the rest of the C3(a) series, as well as D2(a), D4(a), and D2(b)
40
Table 3.9: Phase 2 Monotonic Testing Results
41
All, but two, of the (a)-type specimens (with adequate glued-in steel rod embedment length of
16d to prevent pull-out failure) failed in a ductile manner due to yielding of the glued-in steel
rods as hypothesized. As mentioned previously, the D4(a) specimens did have pull-out and
splitting of the wood failures in addition to tensile failure of the glued-in rod, but they also
showed significant yielding, thus it is presumed that pull-out and splitting failures occurred after
the tensile failure of the steel rods due to sudden load transfer. As a result, the two exceptions to
ductile tensile failure of the glued-in steel rods for (a)-type specimens were specimens B2(a)-1
(2) and D2(a)-1 (2), which both failed due to splitting of the timber beams. Since these
unexpected failures were one out of four tests for the specific specimen type, or overall, two out
of nineteen tests, they were treated as anomalies and were not taken into consideration in the
analysis. In addition, B2(a)-1 (2) had an initial split in the timber beam prior to testing, which
could have led to the failure due to splitting of the wood. The failures of the (a)-type specimens
under monotonic loading are shown in Figure 3.20 and Figure 3.21.
42
(a) (b)
(c) (d)
(e)
Figure 3.20: Ductile Failure of (a) A1(a), (b) B2(a), (c) C3(a), (d) D2(a), and (e) D4(a) Specimens under
Monotonic Loading
43
(b)
(a)
Figure 3.21: Brittle Splitting Failure of (a) B2(a)-1 (2) and (b) D2(a)-1(2) under Monotonic Loading
The moment-rotation curves for all (a)-type specimens under monotonic loading are shown in
Figure 3.22, Figure 3.23, Figure 3.24, Figure 3.25 and Figure 3.26. The moment was calculated
by multiplying the applied load with the measured moment arm, and the rotation was calculated
by taking the inverse tangent of the displacement measured from the string pot divided by the
length of the beam from the face of the steel base plate to the location of the string pod, which
was measured to be 1355mm. From these moment-rotation curves, the peak moment, ultimate
(or failure) moment and moment at the yield point as well as the corresponding rotations at these
moments were calculated, in addition to the ductility and elastic stiffness. The results of the
44
Figure 3.22: Moment-Rotation Curve of A1(a) Specimens
45
Figure 3.24: Moment-Rotation Curve of C3(a) Specimens
46
Figure 3.26: Moment-Rotation Curve of D4(a) Specimens
The yield point of each specimen was determined following the two methods outlined in EN
12512 (CEN, 2005) and (Piazza et al., 2011). For specimens A1(a), B2(a), C3(a), and D2(a)
where the moment-rotation curve did not have a well-defined linear yield line, the experimental
yielding point was determined following method (b) of EN 12512 (CEN, 2005). It determines the
yield point by finding the interception point of the elastic stiffness slope line and the line tangent
to the curve with a 1/6th slope of the elastic stiffness (Piazza et al., 2011), as shown in Figure
For D4(a) specimens, which had a well-defined linear yield portion, the yield point was found
following method (a) of EN 12512 (CEN, 2005). It takes the yield load, or moment, as the load
at which produces the well-defined, horizontal yield line, and the yield displacement, or rotation,
as the displacement where the elastic stiffness slope line and the horizontal yield line intercept
47
(Piazza et al., 2011), as illustrated in Figure 3.31. The elastic stiffness of the moment-rotation
curve of each test data was calculated by following the equation outlined by Piazza et al. (2011):
0.4𝐹𝑚𝑎𝑥 − 0.1𝐹𝑚𝑎𝑥
𝑘𝑒 = (1)
𝑢0.4 − 𝑢0.1
where 𝑢0.1 and 𝑢0.4 are the displacements at load of 0.1𝐹𝑚𝑎𝑥 and 0.4𝐹𝑚𝑎𝑥 , respectively.
Failure was determined according to EN 12512 as the first point obtained where either failure
occurs, or 80% of the maximum load, or moment, is reached on the descending arm (Piazza et
al., 2011). The ultimate point is also shown in the graph from Figure 3.27 through Figure 3.31. In
addition, these figures also indicate the peak moment and rotation at peak moment. The moment-
rotation curves of the individual tests showing the yield, peak and ultimate point not within the
main body of this thesis are included in Appendix B. The experimental yield, peak and ultimate
(or failure) moment, found for all the (a)-type specimens, were consistent within the test series
49
Figure 3.30: Moment-Rotation Curve of D2(a)-1 (1)
50
Table 3.10: Summary of (a)-type Specimen Analysis under Monotonic Loading
51
The (b)-type specimens, which were designed with insufficient embedment length, of 8d, all
failed in a sudden, brittle manner from pull-out failure along the glue line as anticipated. The
pull-out failures of the (b) type specimens are shown in Figure 3.32.
The results of the (b)-type specimens solidifies Gonzalez’s (2015) finding that for 12.7mm
diameter glued-in steel rods, embedment length less than 10d induce a brittle pull-out failure.
The results of the peak moment and rotation at peak moment, as well as the failure moment and
rotation at failure moment are listed in Table 3.11. These capacities were found by analyzing the
moment-rotation curves of each specimen, which are shown in Figure 3.33, Figure 3.34 and
Figure 3.35.
(a)
(b)
(c)
Figure 3.32: Brittle Pull-out Failures of (a) A1(b), (b) C3(b), (c) D2(b) Specimens
52
Table 3.11: Summary of (b)-type Specimen Analysis under Monotonic Loading
Experimental
Experimental Experimental Experimental
Rotation at
Specimen Peak Moment Rotation at Ultimate
Peak Moment
(kNm) Ultimate (rad) Moment (kNm)
(rad)
A1(b)-1(1) 0.016 11.34 0.017 11.30
A1(b)-1(2) 0.013 10.00 0.014 9.65
A1(b)-1(3) 0.010 9.96 0.010 9.83
A1(b)-1(4) 0.010 9.16 0.010 9.13
Average : 0.013 10.12 0.013 9.98
StDev: 0.003 0.90 0.003 0.93
C3(b)-1 (1) 0.013 22.78 0.013 22.78
C3(b)-1 (2) 0.020 25.25 0.020 25.25
C3(b)-1 (3) 0.020 26.40 0.020 26.40
C3(b)-1 (4) 0.015 21.52 0.015 21.52
Average : 0.017 23.99 0.017 23.99
StDev: 0.003 2.23 0.003 2.23
D2(b)-1(1) 0.009 19.87 0.009 19.68
D2(b)-1(2) 0.013 27.21 0.013 27.21
D2(b)-1(3) 0.013 30.85 0.013 30.85
D2(b)-1(4) 0.012 29.80 0.012 29.80
Average : 0.012 26.93 0.012 26.89
StDev: 0.002 4.95 0.002 5.04
53
Figure 3.33: Moment-Rotation Curve of A1(b) Specimens
54
Figure 3.35: Moment-Rotation Curve of D2(b) Specimens
The summary of cyclic testing of the peak moment and the rotation at the peak moment for both
positive and negative moments, as well as the maximum rotation reached by the specimen and
the corresponding moment at that rotation for both the positive and negative rotation are shown
on Table 3.12. Once again, similar to the monotonic testing, the moment capacity was calculated
by multiplying load applied by the load cell with the measured moment arm, which extends from
the flush face of the steel base plate to the center of the clamping plates, and the rotation was
calculated by taking the inverse tangent of the displacement captured by the string pod divided
by the location of the string pod relative to the steel based plate, placed 1355mm away. In
addition, the same labeling system was used for the cyclic test, except no one edge of the beam
was used twice for testing since quasi-static cyclic loading required all the glued-in steel rods to
55
be under tension at one point or another; thus, each test was done on an untested edge of the
beam.
All specimens tested under the cyclic loading were (a)-type specimens, meaning they were
designed with adequate embedment length of 16d for its glued-in steel rods. Similar to (a)-type
series tested under monotonic loading, these specimens also failed in a ductile manner due to
yielding and plasticization of the glued-in steel rods. In addition, pull-out along the glue line and
splitting of wood was also observed for the D4(a) cyclic series, but analogous to the monotonic
testing of D4(a) series, significant yielding was observed before failure. Thus, it is assumed that
the brittle failure modes were due to the large and sudden transfer of loads from the failed glued-
in steel rods. The failure modes for the quasi-static cyclic testing are shown in Figure 3.36.
56
Table 3.12: Phase 2 Cyclic Testing Results
57
(a)
(b)
(c)
(d)
(e)
Figure 3.36: Ductile Failure of (a) A1(a), (b) B2(a), (c) C3(a), (d) D2(a) and (e) D4(a) Specimens Under Quasi-
58
The energy dissipated at the specific primary peaks of the quasi-cyclic loading protocol is listed
in Table 3.13. EN 12512 was followed to calculate the dissipation of energy of the hysteretic
cuve, or also known as “the equivalent viscous damping ratio by hysteresis” (Piazza et al., 2011).
𝐸𝑑
𝑣𝑒𝑞 = (2)
2𝜋𝐸𝑝
where Ed is the energy dissipated in one half cycle of the hysteresis and Ep is the potential energy
obtainable at the corresponding one half cycle (Piazza et al., 2011). Figure 3.37 illustrates the
energy dissipated and potential energy available in one cycle of a hysteretic curve.
Figure 3.37: Equivalent Viscous Damping Ratio for One Cycle (Piazza et al., 2011)
59
Table 3.13: Summary of Energy Dissipation
60
3.10 Phase 2 Discussions
Ductility, which is the measure of plastic deformation without significant decrease in strength,
thus rotation (or displacement) at ultimate divided by rotation (or displacement) at yield (Piazza
et al., 2011), was also calculated. The A1(a) specimen had the greatest ductility with an average
ductility of 3.0, followed by B2(a) specimens with and average ductility of 2.1, then C3(a) with
an average ductility of 2.0, and finally D2(a) and D4(a) specimens with an average ductility of
1.7. It appears that the ductility of the connection, decreases with the increase in the size of the
timber beam, or in general as the moment capacity of the system increases. However, more tests
Figure 3.38 compares the experimental yield point determined with the theoretical yield point
calculated. The theoretical yield moment was calculated by applying the assumption that plane-
sections remain plane in combination with the traditional elastic transform theory, which
assumes linear stress distribution of the compression stresses within the timber compression zone
(Fragiacomo & Batchelar, 2012). For two rows of tension glued-in steel rods, such as for
specimen D4(a), both layers of the steel were assumed to have yielded and an equivalent moment
arm from the centroid of the steel layers was used to predict the yield capacity of the member.
None, except for three, of the individual specimens achieved yield moments smaller than the
theoretical yield moment hypothesized. The three specimens which had yielding points below
the theoretical yielding capacity were C3(a)-2 (2) with 4.0% less, D2(a)-1(3) with 1.2% less, and
finally D4(a)-2 (3) with 9.2% less. On average, the theoretical yield point calculated for the
61
specimens were close to the experimental yielding point determined, and in fact underestimated
the yield moment for all the different specimen layouts. As a result, the theoretical approach
used, which is similar to the procedure used to calculate moment resistance of reinforced
concrete beams in CSA A23.3-14 (2014) with a replacement of the concrete compression
member with timber, gives a realistic estimation of the yield moment capacity of timber moment
62
3.10.2 Quasi-Static Cyclic Testing
The moment-rotational curves of each specimen, A1(a), B2(a), C3(a), D2(a) and D4(a) under
quasi-static cyclic testing are plotted in Figure 3.39, Figure 3.40, Figure 3.41, Figure 3.42 and
Figure 3.43, respectively. In addition, on the plots, the monotonic backbone curve from the
corresponding monotonic testing is also shown, as well as the hysteretic backbone curve with
markers at the location where the peak of the primary cycle of the CUREE loading protocol
occurs. The detailed moment-rotation curves of specimen members not included within the main
All test series, ranging from A1(a) to D4(a), exhibited pinching during unloading. This was most
likely due to the gap between the steel base plate and the specimen that was introduced through
yielding and plasticization of the steel rods. Upon unloading, the gap produces almost no
stiffness for the connection once the specimens have reached a position where the gap
disengaged the rods from resisting any load. The load would be picked up again by the specimen
The side of the beam where the rods are initially under tension (under positive displacement)
experienced the primary peaks prior to the opposite end of the beam, where the rods are under
compression initially (and in tension under negative displacement). As a result, the positive-
displacement tension rods yield and plasticize before their counter parts introducing a gap
located on just that side of the beam. Looking in-depth into the development of the displacement
between the steel base plate and the face of the timber beam with respect to the loading protocol,
in other words, the gap created with respect to the displacement of the load cell, it appears that
once the loading protocol has reached a displacement that cause the steel rods to yield, it
63
produces a gap large enough that it cannot be completely closed even when the load cell is
unloaded back to zero displacement. Only when the specimen is loaded in the negative direction
does the gap completely close and enable the negative-displacement tension glued-in steel rods
to pick up load. This phenomenon could explain the shift of the hysteresis towards the negative
rotation. In addition, the fact that the negative-displacement tension rods do not get loaded until
the specimen is loaded sufficiently in negative direction contributes to the difference between the
positive and negative moment-rotation hysteretic curves. As a result, it appears, that the more the
hysteretic curve is shifted towards the negative rotation, the less the negative-displacement
tension glued-in rods plasticize, thus creating a more uneven hysteretic curve.
Overall, the positive portion of the hysteretic curve of all the specimens, A1(a) through D4(a),
followed the monotonic moment-rotation backbone curve relatively closely without significant
stiffness deterioration, or cyclic deterioration was present in any of the layouts. Finally, no
Comparing the positive equivalent viscous damping ratio, or energy dissipation, at the primary
cycle with the peak of 0.7 of the maximum displacement, the most energy dissipation is carried
out by specimen A1(a) with an average of 21%, followed by specimen B2(a) with an average of
15%, then specimen C3(a) with an average of 11%, and then specimen D2(a) with an average of
9%, lastly with specimen D4(a) with an average of 7%. Although specimen D4(a) has the highest
moment capacity, similar to the ductility calculated from the monotonic testing, the viscous
damping ratio for specimen D4(a) is the smallest. In contrast, specimen A1(a), with the smallest
moment capacity, had the greatest viscous damping ratio. In addition, no failure occurred for
specimen A1(a)-2 (2) and specimen A1(a)-3 (3) even when the load cell reached the maximum
64
displacement set for the protocol was reached, which was the average failure displacement from
the monotonic testing of A1(a) specimens. Furthermore, the average energy dissipated from
these two specimens on the maximum peak cycle was 27%, which is greater than the energy
dissipated from the preceding peak cycle with 30% less displacement. Overall, it appears, that a
higher moment capacity leads to lower capability to dissipate energy. However, more research is
65
Figure 3.40: Moment Rotation Curve of B2(a)-2 (3)
66
Figure 3.42: Moment-Rotation Curve of D2(a)-2 (1)
67
Chapter 4: Conclusions
Experimental and analytical work was completed on timber moment connections using 12.7mm
diameter mild steel threaded rods glued into Douglas-Fir glulam timber beams with polyurethane
based adhesives to determine the effect of the rod layout on the connection moment capacity.
Through phase 1, it was concluded that ductile failure of the rods is achieved when the shear
force induced into the connection is less than 25% of the axial rod strength. This follows Von
Mises Yield Criteria, which concludes that with shear stress of 25% or less of the ultimate axial
stress, the axial capacity that could be reached within a system is 90% or greater of the ultimate
axial strength. As a result, only the 1000mm long specimens, which experienced shear stress of
less than 20% of axial strength, had a ductile failure. This is true even for the specimens that had
an edge distance of rods of less than the recommended edge distance, 2.5d. The embedment
length of the glued-in rods was found to be a key factor: 8d was found to create brittle pull-out
failures along the glue line, verifying Gonzalez’s (2015) work, while 15d was found to prevent
Phase 2 montonic testing confirmed the computed theoretical yield moment. The theoretical
approach under-predicted the experimental yield moment with a difference of 13% or less; thus
would result in a safe design of the moment connection with a conservative yield moment. The
average ductility was found to be in the range of 1.7 to 3.0. From these values and the moment
capacity, it appears that ductility of the moment connection decreases as the moment capacity of
configurations, which would result in a range of different moment capacity, against its ductility
68
as well as comparing other potential factors that could influence ductility, such as the steel to
Phase 2 reversed cyclic testing exposed some shortcomings of the test set-up, where a gap
between the steel base plate and the timber beam appeared. Non-symmetric hysteresis between
the positive moment-rotation curves and the negative moment-rotation curves were produced
since the negative-displacement tension steel rods were not loaded until the gap closed.
Overall, however, no stiffness deterioration during reloading and unloading was found in the
hysteretic curves. In addition, the positive moment-rotation curve of the hysteresis follow the
The energy dissipation, or equivalent viscous damping ratio, was calculated. The energy
dissipated at the cycle with peak displacement of 70% of the maximum displacement ranged
from 9% to 21%. Similar to ductility, equivalent viscous damping decreased with moment
The experimental work done throughout this research enhanced the knowledge of the behaviour
of timber moment connections using glued-in steel rods. The study focused on connections with
12.7mm diameter mild steel threaded rods glued into Douglas Fir glulam beams with PUR.
between the different configuration layouts of the glued-in rods and the yield moment of the
69
The experimental work was completed with one specific material and properties for wood, steel
and adhesive, thus to provide general design guideline, further experiments are required with
different rod types and diameters, different adhesives, as well as different wood species. The
addition of shear reinforcements and their effect on the moment capacity should also be
The possibility to predict the yield moment of glued-in rod timber connections will, eventually,
allow design of full sized timber frames as the main moment resistance member or in
70
References
Andreolli, M., Piazza, M., Tomasi, R., & Zandonini, R. (2011). Ductile moment-resistant steel-
Batchelar, M. L. (2007). Timber Frame Moment Joints with Glued-In Steel Rods - A Designer's
Blass, H., & Laskewitz, B. (1999). Effect of Spacing and Edge Distance on the Axial Strength of
Glued-In Rods. (pp. CIB-W18/32-7-12). Graz: International Council for Research and
British Columbia, Office of Housing and Construction Standards, National Research Council
Canada. (2012). British Columbia Building Code 2012. British Columbia, Canada: Office
Bruhl, F., & Kuhlmann, U. (2012). Connection Ductility in Timber Structures Considering the
Buchanan, A., Moss, P., & Wong, N. (2001). Ductile moment-resisting connections in glulam
Canadian Commission on Building and Fire Codes. (2010). National Building Code of Canada
Canadian Standards Association. (2010, March). CSA S16-09 Design of steel structures. Ontario,
Canada.
71
Canadian Standards Association. (2014, June). CSA A23.3-14 Design of concrete structures.
Ontario, Canada.
Canadian Standards Assoiation. (2014, May). CSA O86-14 Engineering design in wood.
Ontario, Canada.
CEN European Committee for Standardiztion. (2005). EN 12512: Timber structures - Test
del Senno, M., Piazza, M., & Tomasi, R. (2004). Axial glued-in steel timber joints - experimental
Deutsches Institut fur Normung e.V. (2004). Norm DIN 1052:2004-08 Entwurf, Berechnung und
Fragiacomo, M., & Batchelar, M. (2012). Timber Frame Moment Joints with Glued-In Steel
Fragiacomo, M., & Batchelar, M. (2012). Timber Frame Moment Joints with Glued-In Steel
Engineering, 802-811.
Gattesco, N., Gubana, A., & Buttazzi, M. (2010). Cyclic Behaviour of Glued-In-Joints under
Gilbert, C., Gohlich, R., & Erochko, J. (2015). Nonlinear Dynamic Analysis of Innovative High
Engineering. Victoria.
72
Gonzalez Barillas, E. (2015, September). Performance of Timber Connections with Single and
Multiple Glued-In Threaded Steel Rods. Vancouver, British Columbia, Canada: The
Government of Canada. (2015, January 21). The benefits of wood in building construction.
http://www.nrcan.gc.ca/forests/video/16213
Ibarra, L. F., Medina, R. A., & Krawinkler, H. (2005, June 13). Hysteretic models that
Jensen, J. L., & Gustafsson, P.-J. (2004). Shear strength of beam splice joints with glued-in rods.
Jensen, J. L., & Quenneville, P. (2009). Connections with Glued-In Rods Subjected to Combined
Kangas, J. (2000). Capacity, Fire Resistance and Gluing Pattern of the Rods in V-Connections.
Krawinkler, H., Parisi, F., Ibarra, L., Ayoub, A., & Medina, R. (2000). Development of a Testing
73
Lam, F., Schulte-Wrede, M., Yao, C., & GU, J. J. (2008). Moment resistance of bolted timber
Miyazaki, Japan.
Lehringer, C. (2012). PURBOND 2C-Adhesives. Cost action FP 1004 Report. Wrozlaw, Poland:
Purbond AG.
Madhoushi, M., & Ansell, M. P. (2008). Behaviour of timber connections using glued-in GFRP
rods under fatigue loading. Part II: Moment-resisting connections. Composites: Part B,
39, 249-257.
Metten, A. W. (2012). Structural Steel for Canadian Buildings: A Designer's Guide. Vancouver:
Andy Metten.
Otero Chans, D., Estevez Cimadevila, J., & Martin Gutierrez, E. (2010). Model for predicting the
axial strength of joints made with glued-in rods in sawn timber. Construction and
Parida, G., Johnsson, H., & Fragiacomo, M. (2013). Provisions for Ductile Behavior of Timber-
139(9), 1468-1447.
Piazza, M., Polastri, A., & Tomasi, R. (2011, April). Ductility of timber joints under static and
Tlustochowicz, G., Serrano, E., & Steiger, R. (2011). State-of-the-art review on timber
connections with glued-in steel rods. Materials and Structures, 44, 997-1020.
74
Xu, B., Bouchair, A., & Racher, P. (2012). Analytical study and finite element modelling of
timber connections with glued-in rods in bending. Construction and Building Materials,
34(1), 337-345.
75
Appendices
numrod
b h edge, e
Spec. (in
(mm) (mm) (mm)
tension)
A1(a) 80 266 33 1
B2(a) 130 266 33 2
C3(a) 175 266 33 3
D2(a) 130 456 33 2
D4(a) 130 456 33+64 4
Figure A.1: Theoretical Yield Moment Calculation
With the specimen layout properties summarized in Table A.1, and assuming that plane sections
remain plane and that the traditional elastic transform section theory applies (illustrated on
Figure A.1), the theoretical yield moment was calculated for each specimen layout. The
%Material Properties
%Steel
Es=2*10^5; %MPa
fy=360; %MPa
dim=12.7; %mm
%Wood
Ew=12400; %MPa
fc=30.2; %MPa
76
%calculation
hs=h-edge; %mm
As=pi()*(dim/2)^2*numrod; %mm^2
ew=NA*ey/(hs-NA);
Fw=1/2*b*NA*ew*Ew;
jdw=2/3*NA;
if ew > fc/Ew
warning('Wood has crushed')
end
if ew > fc/Ew
warning('Wood has crushed with ey')
end
Mtheoryield=(Fy*jds+Fw*jdw)*10^-6; %kNm
77
Appendix B Moment-Rotation Curves under Monotonic Loading
The moment-rotation curves and the analysis of each (a)-type test specimen, under monotonic
loading, not included in the main body of the thesis, are shown in Figure B.1 through Figure
B.14.
78
Figure B.2: Moment-Rotation Curve of A1(a)-1 (3)
79
Figure B.4: Moment-Rotation Curve of B2(a)-1 (1)
80
Figure B.6: Moment-Rotation Curve of B2(a)-1 (4)
81
Figure B.8: Moment-Rotation Curve of C3(a)-2 (2)
82
Figure B.10: Moment-Rotation Curve of D2(a)-1 (2)
83
Figure B.12: Moment-Rotation Curve of D2(a)-1 (4)
84
Figure B.14: Moment-Rotation Curve of D4(a)-2 (3)
85
Appendix C Moment-Rotation Curves under Quasi-Static Cyclic Loading
The moment-rotation curves and analysis of each test specimen, under quasi-static cyclic
loading, which were not included within the main body of the thesis, are shown in Figure C.1
86
Figure C.2: Moment-Rotation Curve of A1(a)-3 (3)
87
Figure C.4: Moment-Rotation Curve of B2(a)-3 (1)
88
Figure C.6: Moment-Rotation Curve of C3(a)-3 (2)
91
Figure C.12: Moment-Rotation Curve of D4(a)-3 (2)
92