CYCLIC BEHAVIOUR OF GLUED-IN-JOINTS UNDER

BENDING MOMENTS

Natalino Gattesco1, Alessandra Gubana2, Manuela Buttazzi3

ABSTRACT: A growing interest has arisen towards the possibility of joining wood elements with steel bars glued with
epoxy resin into timber members. This type of joint is interesting, because it allows to obtain an higher joint stiffness, a
more uniform stress distribution and a better aesthetical appearance in comparison with traditional mechanical
connections. A series of experimental tests on glued-in threaded rod joints subjected to bending moments showed that
with a correct design and a proper choice of the bar steel grade it is possible to obtain a very ductile behaviour. In the
present paper the interest is focused on the cyclic behaviour of glued-in-joints under bending moments and some
experimental investigations on joints subjected to cyclic loads are presented. The aim is the understanding of the joint
structural response under cyclic loads simulating seismic action. The results outline that the joint can develop high
ductility, even though it is partially limited by lateral instability due to torsional effects.

KEYWORDS: Glued-in joints, cyclic behaviour, timber structures, experimental tests.

1 INTRODUCTION 123 the failure mechanism. In this way pull-out tests were
primarily carried out on specimens arranged with one bar
In the last years a growing interest has developed and they permitted to define four different failure
towards the possibility of joining glue-laminated beams mechanism: bar failure, ultimate longitudinal wood shear
with steel bars placed into predrilled holes in the timber resistance, splitting of wood and tensile rupture of wood
members and bonded with epoxy resin. over the cross section. Studies about groups of bars and
This type of joint, if properly designed, can offer an high possible ductile behaviour were made by Gehry [9] and
strength, an higher stiffness, a more uniform stress Tomasi [10]. Extensive experimental pull out tests have
distribution in the wood, with no significant peak values been performed also by Gattesco and Gubana [11], [12],
and a better aesthetical appearance in comparison with [13], [14] to investigate the role of bar edge distance in
traditional mechanical connections (e.g. bolts, pins, the joint response and the design criteria to reach a
dowels). Moreover steel bars are surrounded by wood ductile collapse. The behaviour of glued-in rods joints
and so preserved from corrosion. under bending moment was also investigated and
There are several examples of glue-laminated structures experimental results of tests under monotonic load are
where this type of joint was used, from the first ones by here briefly summarized. Some experimental tests under
Buchanan and Fletcher [1], to the most recent ones by cyclic loads have been successively performed. The aim
Piazza [2] and Estévez Cimadevila [3]. was to study the behaviour of the joint under reversible
The first important studies on glued joints were loads, simulating the seismic action and to investigate
performed by Riberholt [4], Piazza et al.[5], Buchanan the level of ductility reachable by the connection.
and Deng [6], Aicher et al [7] and they allowed to state
the first formulations used to predict the pull-out
2 MONOTONIC BENDING TESTS
capacity of the joint.
A great part of the literature about glued-in rods joint The specimens subjected to monotonic bending tests
focused the interest on pull-out test [8], as the first aim were made of two glue laminated beams having a cross
was to investigate the load-slip relationship and to define section 120 mm wide, 295 mm depth and 2850 mm
length, connected together in the mid span with glued-in
1
Natalino Gattesco, Department of Environmental and Civil threaded bars so to obtain a beam 5700 mm length
Engineering, University of Trieste, via Valerio 5, 34127 (Figure 1). The holes drilled in the timber element have a
Trieste, Italy. Email: [email protected] diameter 4 mm larger than that of the bars. The bars
2
Alessandra Gubana, , Department of Civil Engineering and were put inside the holes using a wire guide to centre
Architecture, University of Udine, via delle Scienze 206, them; the epoxy was then injected through 6 mm
33100 Udine, Italy. Email: [email protected] diameter holes, drilled perpendicularly to each
3
Manuela Buttazzi, Department of Civil Engineering and embedment hole, near its end. The embedment length
Architecture, University of Udine, via delle Scienze 206, was assumed equal to 30 times the bar diameter. The
33100 Udine, Italy. Email: [email protected]
Figure 1: Geometric characteristics of the beams and section details

Table 1: Experimental results of monotonic bending tests

Serviceability Plastic Behaviour – Maximum Values

Moisture
Specimen Yielding Yielding Rotation Ductility Global
content Load Deflection
Ident. Load Rotation φu factor rotation
(%) (kN) (mm)
(kN) φy (rad) (rad) φu/ φy (rad)
1st Group
TQ14-40/1 9.5 42.3 0.0044 42.3 44.9 0.0051 1.1 0.0170
TQ14-30/1 9.2 50.5 0.0042 60.5 68.4 0.0073 1.7 0.0208
TD16-32/1 9.9 41.5 0.0055 50.4 69.3 0.0201 3.6 0.0312
TT16-40/1 8.9 44.7 0.0052 44.7 45.1 0.0058 1.1 0.0194
TC16-32/1 9.7 41.9 0.0063 48.3 69.7 0.0187 2.9 0.0311
TC16-40/1 10.4 44.4 0.0069 53.1 75.9 0.0197 2.8 0.0353
TD20-40/1 9.1 47.0 0.0053 57.2 64.7 0.0107 2.0 0.0249
TV20-60/1 10.0 45.3 0.0057 49.1 59.4 0.0114 2.0 0.0228
2nd Group
TT16-32/2 11.1 37.3 0.0059 46.1 91.0 0.0343 6.6 0.0469
TT16-40/2 11.6 38.9 0.0048 44.1 55.8 0.0117 2.4 0.0187
TQ16-32/2 11.9 44.4 0.0053 54.0 85.6 0.0149 2.8 0.0262
TD20-40/2 11.4 34.8 0.0045 42.3 86.9 0.0277 6.2 0.0391
3rd Group
TQ14-30/3 11.7 40.8 0.0037 47.1 296.0 0.1810 48.9 0.1930
TT16-32/3 10.8 37.4 0.0039 44.7 137.0 0.0622 16.0 0.0757
TQ16-32/3 11.2 44.1 0.0068 48.6 258.7 0.1513 22.2 0.1635
TD20-40/3 11.4 39.1 0.0039 45.9 236.7 0.1346 34.5 0.1481

Table 2: Mechanical material characteristics

Materials Mechanical Characteristics 1st Group 2nd Group 3rd Group

Wood fm [MPa] 46.4 46.4 46.4
ft,90 [MPa] 0.81 0.74 0.74
Ec,0 [MPa] 9856 12993 12993
Steel Steel type C40 Fe510 Fe430
fys [MPa]* 590/625/605 412/412/370 328/318/303
fts [MPa]* 690/735/695 558/558/513 458/466/448
(* The values refer to diameter 14 mm, 16 mm and 20 mm respectively)
positions of the bars inside the specimens and the beam materials used for joints. Mechanical material
details are illustrated in Figure 1. Two specimens (TC16- characteristics are summarized in Table 2.
32 and TC16-40) were provided with screwed bolts ( 8 The beams of the first group evidence a brittle failure
mm diameter, 140 mm length) so as to prevent characterised by the tensile rupture in correspondence of
longitudinal splitting of wood along the bars. Sixteen the bar embedment (Figure 2a). The values of ductility
beams were tested. obtained are in fact low. The joints of the specimens of
The specimen are identified by an acronym of two letters the second group were built with bars of lower yielding
followed by two numbers separated by an hyphen; the stress steel: some brittle failure still happened, but a
first two letters are related to bar arrangement: TD for an limited ductility may be noted. In the beams of the third
horizontal layer of two bars, TV for a vertical layer of group mild steel bars were used and before the collapse
two bars, TT for a triangular arrangement, TQ for two the joints presented large rotations with a significant
horizontal layers of two bars each. The first number plastic branch (Figure 2b). The values of ductility
identifies the bar diameter in millimetres and the other obtained are very high in comparison with the other two
indicates the lateral edge distance. groups.
The beams were subjected to two symmetric loads at a The results of monotonic bending tests show that glued
relative distance of 1600 mm; the distance between in rod joints can develop ductile behaviour and ductile
supports was 5500 mm. To avoid the local crushing of failure if they are properly designed: for this purpose it is
wood, in correspondence of point loads and support necessary to choose threaded bars of low yielding stress
reactions, steel plates 120 x 120 x 15 mm were used and, steel. In this way the collapse is governed by steel bars
to allow for free rotation of the beam at the supports, avoiding the brittle mechanism associated to timber
steel cylinders were placed (Figure 1). The beams were collapse.
tested by using an hydraulic servo controlled system. By these results it was possible to identify the steel type
necessary for a ductile behaviour of joints. Being aware
of this conclusion, experimental bending tests under
cyclic load have been performed.

3 CYCLIC BENDING TESTS

3.1 EXPERIMENTAL SET UP AND SPECIMEN
DESIGN
The specimens tested presented exactly the same
geometric features of those subjected to monotonic
bending tests: the only difference was that the steel bars
used to build the glued joint were symmetrically placed
(a) with respect to the longitudinal axes of the beams, as
presented in Figure 3. Four beams were tested and they
were divided in two groups, based on the joint geometry.
Only the joint arrangements which developed the best
results under monotonic loads were reproduced to
perform cyclic tests .
The beams were identified by an acronym, in a similar
way as for monotonic tests. The first two letters were
related to bar arrangement: TD for one horizontal layer
of two bars, TQ for two horizontal layers of two bars; the
third letter, D, indicated that the section was doubly
reinforced. The first number was related to the bar
diameter and the second indicated the number of the
(b)
beam in each of the two groups, as for each group two
Figure 2: Brittle (a) and ductile (b) failure of beams beams of the same characteristics were tested. The bars
were made of Fe430 steel, with the same characteristic
The results presented by Gattesco and Gubana [14] are presented in Table 2, because joints arranged with this
summarized in Table 1, where besides the moisture steel type have given the best performance and the
content of wood at testing, some important information higher values of ductility during monotonic tests.
concerning both serviceability and ultimate limit state The beams were subjected to two symmetric loads at a
are reported. For serviceability state the yielding load distance of 1600 mm applied through a steel beam. As it
and the corresponding yielding rotation are presented. was necessary to invert the load direction during the test,
For collapse condition the ultimate load, the the glue-laminated beam was connected to the steel
corresponding mid-span deflection, the joint rotation and beam by means of four steel plates: both steel and timber
the global rotation, evaluated between two sections at beams presented two holes where end threaded pins are
relative distance of 1200 mm, are indicated. Rotation inserted. The pins were fixed at the ends by nuts that
ductility factor is also calculated. The beams are divided contrast in the steel plates. The same system of plates
into three groups based on the mechanical properties of was used at the edges of the timber beam to connect it to
the steel basement. On the beam front and back surfaces the load was applied following the program described in
two other steel beams were fixed to the steel frame to Table 3.
prevent lateral instability displacement of the tested The beams were tested by using an hydraulic servo
glulam beam. The complete set up is in Figure 4. controlled system. The load speed was equal to 0,2
mm/s. The data acquisition was automatically carried
out. At the beginning of each test wood moisture content
was surveyed by means of an electric hygrometer.

Table 3: Cycle of displacement load

Cycles
Displacement
number
0,25 Vy 1
0,5 Vy 1
0,75 Vy 3
Vy estimated 3
TDD 20-1 TQD 14-1 Vy effective 3
TDD 20-2 TQD 14-2
1,5 Vy effective 3
4Ø20 8Ø14
l = 600 mm l = 420 mm 2 Vy effective 3
30

3 Vy effective 3
40

30

V
215

295

295
175

3 V

3 V

3Vyeff
2Vyeff
1,5Vyeff
0,75Vy Vy Vyeff
40

30 30

0,5Vy
0,25Vy
40 40 40 30 60 30
t
120 120

Figure 3: Section details of beams subjected to cyclic

bending test 3.2 TEST RESULTS
The results of the cyclic test performed on the beams are
summarized in Table 4, where, beside wood moisture
Twelve electrical transducers were arranged to survey content at testing, some information concerning
the main displacement parameters. In particular two serviceability state is reported. Experimental values of
couples of inductive transducers (T1-T2, T3-T4) were the bending moment and central and global rotation for
placed in the mid span section, one on the front side and positive and negative cycle, corresponding to the
the other on the backside of the beam close to the top proportional limit, are presented. In Table 5 information
and the bottom of the beam, respectively. The data were about ultimate limit state is reported: experimental
used to calculate the rotation in the joint section. The values of the maximum bending moment and the central
rotation between two sections at a relative distance of and global rotation for positive and negative cycles
1200 mm (global rotation) was measured by using two corresponding to collapse are given. The ultimate values
inductive transducers (T5-T6). Two other inductive of both type of rotation are considered in correspondence
transducers (T7-T8) were used to survey the mid span of three cycle of complete load and not at the real point
deflection of the beam. Four more gauges (T10-T11, T12- of collapse if this happened during the first or second
T13) were used to measure the rotation at the supports. cycle, because this is the condition for calculation of
These values, as well as those obtained from transducers ductility as indicated for wood structures in EC8.
T5-T6, T7-T8 were used to check the reliability of the In Figure 6 the relationship between bending moment
rotation measured in the joint section. All the transducers and joint rotation are illustrated for the four beams. All
had a sensibility of 0.002 mm, except those used to the curves have an almost linear branch up to a moment
measure the mid span deflection (T9-T10) with a of approximately 30-38 kNm, the same values reached
sensitivity of 0.02 mm. The transducers arrangement is by the beams subjected to monotonic load.
illustrated in detail in Figure 5. After the yielding point, at the progression of the cycle
The load was applied following the indications given by number, beams present different behaviour: all the
UNI EN 12512 (2003); so, once the displacement curves evidence a reduction of stiffness, but while for
corresponding to the start of yielding was estimated (Vy), TDD 20-1 and TDD 20-2 the cycle are full and presented
Figure 4: Set up of cyclic bending tests

F T1 T2 F
T5

T 11 T 12 T 10 T 9
T6 T3 T4
T7 T8

Figure 5: Arrangement of electrical transducers used for bending tests

Table 4: Experimental results for serviceability state

Serviceability State

Moisture Positive Negative

Specimen Positive Positive Negative
content global Negative global
identification moment rotation moment
(%) rotation rotation rotation
(kNm) (rad) (kNm)
(rad) (rad) (rad)

TDD 20-1 32,03 0,0029 0,0049 -38,36 -0,0037 -0,0057

TDD 20-2 8,8 32,60 0,0035 0,0106 -37,38 -0,0034 -0,0117
TQD 14-1 9,4 30,68 0,0020 0,0086 -34,79 -0,0018 -0,0098
TQD 14-2 9,3 29,13 0,0019 0,0079 -35,71 -0,0021 -0,0099

Table 5: Experimental results for ultimate limit state

Ultimate Limit State

Moisture Positive Positive Negative Negative

Specimen Positive
content maximum global maximum Negative global
identification rotation
(%) moment rotation moment rotation rotation
(rad)
(kNm) (rad) (kNm) (rad) (rad)

TDD 20-1 38,29 0,0250 0,0167 -42,40 -0,0213 -0,0146

TDD 20-2 8,8 37,62 0,0296 0,0414 -44,90 -0,0182 -0,0387
TQD 14-1 9,4 37,61 0,0271 0,0348 -43,50 -0,0122 -0,0289
TQD 14-2 9,3 37,85 0,0307 0,0403 -42,84 -0,0126 -0,0277
 TDD 20-1

50
40
30
20
moment (kNm)

10

0
-0,03 -0,02 -0,01 -100,00 0,01 0,02 0,03 0,04 0,05

-20
-30
-40 TDD 20-1
-50
rotation (rad)
(a)
TDD 20-2
50
40
30
20
moment (kNm)

10
0
-0,03 -0,02 -0,01 -100,00 0,01 0,02 0,03 0,04 0,05

-20
-30
-40
-50 TDD 20-2
-60
rotation (rad)

(b)
TQD 14-1

50
40
30
20
moment (kNm)

10
0
-0,06 -0,04 -0,02 -100,00 0,02 0,04 0,06 0,08

-20
-30
-40
TQD 14-1
-50
rotation (rad)

TQD 14-2

50
40
(c)
30
20
moment (kNm)

10
0
-0,02 -100,00 0,02 0,04 0,06
-20
-30
-40
TQD 14-2
-50
rotation (rad)

Figure 6: Moment-Rotation relationship for the 4 tested

beams under cycle load

an almost round form, for TQD 14-1 and TQD 14-2 it is

possible to note a pinching effect, represented by the
reduction of the cycle area. As it is well known this
effect is caused by steel bars. Moreover, at the (d)
progression of the cycle number when the load is
Figure 7: Lateral displacement of steel bars (a), lateral
inverted, the bars that were in tension become subjected failure (b), failure under (c) and over (d) the beam caused
to compression and, at the increase of the load, they tend by lateral instability of bars and torsional effects
to bend laterally and to have instability failure.
The lateral bending produces stresses in the wood around notice that glued joints can develop a good ductile
the bars and so it determines the commencement of behaviour also under cyclic loads, but in a more
cracks as it is visible in Figure 7. restricted way. The ductility development during cyclic
These cracks cause a progressive reduction of the beam test in fact is reduced of almost 5 times with respect to
stiffness and strength as visible in Figure 6, accompanied the one obtained during monotonic tests. This effect is
also by a relative horizontal displacement between the due to the lateral instability of compressed bars at load
two parts of the jointed beam. inversion that causes a torsion in the beam joint.
During the last cycles at the load inversion, the cracks This result imply that this type of joint in seismic zone is
permit free lateral displacement of the bars together with characterized by an appreciable ductility but not so high
a relative torsional rotation between the two beam as seemed to evidence the monotonic tests.
members connected together (Figure 7b). This behaviour
is represented in Figure 6 for TQD 14-1 and TQD 14-2 Table 6: Ductility factors
by the part of the graphic characterised by horizontal Rotation
path corresponding to a constant value of load: by now Δφy Δφu Ductility
Specimen
in fact the bending stiffness is null and there is only (rad) (rad) factor
rotation of the two parts. TDD 20-1 0,0066 0,0464 7,02
TDD 20-2 0,0069 0,0478 6,93
When at the increasing of load the two beam parts enter
TQD 14-1 0,0038 0,0393 10,66
into contact, the torsional effect stops and stiffness and
TQD 14-2 0,0040 0,0433 10,82
strength grow up again until the load inversion or until Global rotation
the beam failure. Δφy Δφu Ductility
It is possible to note that the graphics are not perfectly Specimen
(rad) (rad) factor
symmetric because the beam failure occurred during the TDD 20-1 0,0106 0,0313 2,95
first cycle of a new increment of the load. Generally the TDD 20-2 0,0223 0,0801 3,59
rupture occurs during the negative cycle or the unloading TQD 14-1 0,0184 0,0637 3,46
of positive cycle: this because of compression and lateral TQD 14-2 0,0178 0,0680 3,82
instability of bars that first reached the yielding point in
tension. Only TDD 20-2 failed during positive cycle of
load because of a knot that anticipate the splitting of 4 CONCLUSIONS
timber on the side subjected to tension.
The graphics in Figure 6 represent the behaviour of the Glued-in joint for glue-laminated beams have great
joint in correspondence of the mid-span of the beam: potential to be used in substitution of the traditional
asymmetry of curves is also due to the different point mechanical joints, because they have an high stiffness,
where cracks form, that permit different bending of bars there are no clearances between steel and wood as in
and then a different response of the beam at the bolted joints and steel bars are protected from corrosion
increasing of the load. In Figure 7 it is possible observe by the surrounding timber.
some examples of cracks in tested beams, due to lateral Monotonic bending tests show that it is possible to
instability of bars: laterally (Figure 7a,b), on the bottom obtain a ductile failure of the joint, with a proper choice
of beam (Figure 7c) and on the top of beam (Figure 7d). of the steel bars and of the edge distances. In particular
The plastic branches in Figure 6 illustrate also that under the tests show that with high strength steel bars (fy > 600
cycle loads it is possible to obtain a ductile failure. MPa) the collapse is often brittle, whereas with mild
In Table 6 the values of ductility factor are reported for steel (fy < 320 MPa) the glued-in jointed beams show a
central and global rotation evaluated with equation (1). very ductile behaviour.
Cyclic bending tests were performed on the same type of
∆ u  u    u 
specimens as monotonic tests arranged with mild steel
μ 
∆ y  y    y 
(1) bars with the aim to investigate the behaviour of the joint
under cyclic loads. The results of these tests show that
the glued joint can develop a ductile failure and so they
Where  = ductility factor, u+ = ultimate rotation in
can have dissipative performances under seismic actions,
positive cycle, u- = ultimate rotation in negative cycle,
but at the same time the ductility of the joint is five times
y+ = yielding rotation during positive cycle, y- =
lower in comparison with the values obtained from

yielding rotation during negative cycle, u = difference

monotonic bending tests. This occurs because the load
inversion causes lateral instability of steel bars in

negative cycle and y = difference between the yielding

between the ultimate rotation during positive and compression.

rotation in positive and negative cycle. ACKNOWLEDGEMENT

To understand the behaviour of glued-in threaded bars The authors wish to thank the Stratex Company s.p.a.,
joint under cyclic loads and to evidence its ductility, it is Sutrio, Udine, Italy, that kindly supplied the glue-
useful to make a comparison between the results laminated elements for tests, the Resimix s.r.l., Brendola,
obtained from cyclic and monotonic tests. Vicenza, Italy that offered the epoxy resin, the Friulsider
By the comparison between the values of ductility s.p.a., San Giovanni al Natisone, Udine, Italy that
factors evaluated for monotonic and cyclic tests, reported offered carpentry elements.
in Table 1 and Table 6 respectively, it is possible to
REFERENCES
[1] Buchanan A.H., Fletcher M.R.: Glulam Portal
Frame Swimming Pool Contruction. In 2nd Pacific
Timber Engineering Conference, vol.1, 245-249,
1989.
[2] Piazza M., Tomasi R. and Modena R.: Strutture in
legno. Ulrico Hoepli Editore S.p.a. Milano, 2005.
[3] Estevez Cimadevila J., Vazquez Rodriguez J.A. and
Otero Chans M.D.: Experimental behaviour of
threaded steel rods glued into high-density
hardwood. International Journal of Adhesion &
Adhesives, 27(2):136-144, 2007.
[4] Riberholt H.: Glued Bolts in Glulam, Proposal for
CIB Code International Council for Building
Research Studies and Documentation Working
Commission W18, Proceedings CIB-W18, Meeting
21, September 1988, Parksville, Canada, 21-7-2.
[5] Ballerini M., Piazza M. and Zanon P.: Static and
cyclic behaviour of glued joints in timber. Pacific
Timber Engineering Conference, vol. 2, 811-817,
1994.
[6] Buchanan A.H., Deng X.J.: Strength of epoxied
steel rods in glulam timber. In: International Wood
Engineering Conference 1996, Louisiana, USA, vol.
4, 488-495, 1996.
[7] Aicher S., Gustafsson P.J. and Wolf M.: Load
displacement and bond strength of glued- in rods in
timber influenced by adhesive, wood density, rod
slenderness and diameter. 1st RILEM Symposium on
Timber Engineering, Sweden, 369-378, 1999.
[8] Steiger R., Gehri E. and Widmann R.: Pull-out
strength of axially loaded steel rods bonded in
glulam parallel to the grain. Materials and
Structures, 40(1):69-78, 2007.
[9] Gehri E.: Ductile behaviour and group effect of
glued-in steel rods. In: Joints in Timber Structures.
International RILEM Symposium, Stuttgart,
Germany, 333-342, 2001.
[10] Tomasi R., Zandonini R., Piazza M. and Andeolli
M.: Ductile end connections for glulam beams.
Structural Engineering International, 18(3):290-296,
2008.
[11] Gattesco N. and Gubana A.: Studio sperimentale
sulle unioni incollate di elementi in legno lamellare,
Atti del 13 Congresso C.T.E., Pisa, Italia, 403-411,
2000.
[12] Gattesco N. and Gubana A.: Experimental tests on
glued joints under axial forces and bending
moments. In: Joints in Timber Structures,
International RILEM Symposium, Stuttgart,
Germany, 353-362, 2001.
[13] Gattesco N. and Gubana A.: Giunti di continuità di
elementi in legno lamellare realizzati con barre
resinate, Atti del 14 Congresso C.T.E., Mantova,
Italia, 709-715, 2002.
[14] Gattesco N. and Gubana A.: Performance of glued-
in joints of timber members. In: 9th World
Conference on Timber Engineering, Portland, OR,
USA, 2006 CD-ROM.