Engineering Structures 33 (2011) 3018–3026

Contents lists available at SciVerse ScienceDirect

Engineering Structures

journal homepage: www.elsevier.com/locate/engstruct

Ductility aspects of reinforced and non-reinforced timber jointsHans Joachim Blaß, Patrick Schädle ∗

Timber Structures and Building Construction, Karlsruhe Institute of Technology, Germany

a r t i c l e i n f o

Article history:Available online 11 March 2011

Keywords:DuctilityTimber jointsReinforced connectionsCross-laminated timber

a b s t r a c t

Even though brittle failuremodes in timber joints may be avoided by the proper design of the connection,the use of minimum timber dimensions and minimum spacing and distances of fasteners often leadsto timber splitting in the connection area. Due to the highly nonlinear behaviour of timber loaded incompression aswell as the steel used formechanical fasteners, timber joints can behave in a rather ductilemanner. Ductile behaviour is preferable in timber structures.

Technical innovations regarding engineered wood products as well as fastener and steel technologyled to the development of high-performance timber connections. In these high-performance connections,brittle failure modes are prevented by reinforcing the timber in the connection area perpendicular to thegrain or using cross-laminated timber members. The improvement of the ductility levels is shown basedon several experimental studies comparing non-reinforced to reinforced connections.

© 2011 Elsevier Ltd. All rights reserved.

1. Introduction

The failure mechanism of a timber joint with mechanicalfasteners among other things depends on the geometry of the jointand the type of fastener. Under compressive stresses, timber asa material may be loaded far over its elastic limit. Also the steelused for mechanical fasteners is able to deform in a distinctlyplastic manner. Even though, not in all cases timber connectionsdeform plastically before failure resulting in brittle failure modes.Avoiding the causes for brittle failure modes, specifically hightensile perpendicular to the grain and shear stresses, leads todistinctly plastic failure modes of connections with mechanicalfasteners. This case is preferable in timber structures.

Based on technical innovations regarding engineeredwood products as well as fastener and steel technology, high-performance timber connections were developed in recent years.High-performance connections often are carried out as reinforcedconnections. The reinforcement relates to the timber in the con-nection area, where especially tensile stresses perpendicular tothe grain are transferred by the reinforcement. The cross layersof cross-laminated timber members act as reinforcement; hencejoints in CLT also show high potential for ductile joints.

This paper deals withmodern timber joints that are designed insuch a way that brittle failure is avoided and ductile behaviour canbe achieved by using cross-laminated or subsequently reinforcedmembers. The improvement of the ductility levels is shown based

∗ Corresponding author. Tel.: +49 721 60848127; fax: +49 721 60844081.E-mail addresses: Hans.Blass@kit.edu (H.J. Blaß), Patrick.Schaedle@kit.edu

(P. Schädle).

0141-0296/$ – see front matter© 2011 Elsevier Ltd. All rights reserved.doi:10.1016/j.engstruct.2011.02.001

on several experimental studies comparing non-reinforced toreinforced connections.

The failure mechanisms as well as the common reinforcementtechniques are also shown. Due to the highly nonlinear behaviourof timber joints, the definition of ductility is also discussed.

2. Fundamentals about timber joints

When considering timber connections, the joints can beclassified into three types:• Carpenter joints.• Glued joints.• Joints with mechanical fasteners.

Carpenter joints are mainly used to transfer compression forces.If ductility is observed within these connections, it is becauseof compression perpendicular to the grain failure in at least onemember. Glued joints generally do not show ductile behaviour,their failure is brittle. Joints with dowel-type fasteners arepotentially ductile ‘‘by nature’’ due to the interaction between thehighly nonlinear behaviour of the wood under embedding stressesand the bending behaviour of the steel fasteners. The challenge isto avoid the brittle failure mechanisms described in Section 2.3 byeither avoiding perpendicular to the grain tensile stresses or byusing reinforcement techniques described in Section 2.4.

2.1. Embedding behaviour of timber

Mechanical fasteners in timber members loaded perpendicularto the fastener axis cause embedment stresses and subsequentdeformations in the surrounding wood. Due to the mostly circularcross-section of the fasteners, the wood is compressed both,

H.J. Blaß, P. Schädle / Engineering Structures 33 (2011) 3018–3026 3019

0

F

assumptiontrue deformation

u

Fmax

c)

Fig. 1. (a) Close-up view of the embedding deformation in hardwood species parallel to the grain and (b) perpendicular to the grain, (c) assumption for the load–embedmentdeformation behaviour according to Johansen’s theory [4].

Fig. 2. (a) Stress–displacement relationship loading parallel to the grain, (b) stress–displacement relationship loading perpendicular to the grain [2].

φ

0

Mmax

F

assumptiontrue deformation

b)

Fig. 3. (a) Deformation of dowel after a bending test, (b) assumption for the bending behaviour according to Johansen’s theory.

Fig. 4. (a) Splitting of specimen, (b) plug shear failure.

parallel and perpendicular to the grain even if only loaded parallelto the grain (see Fig. 1(a)).

Depending on the characteristic density ρk of the timber andthe diameter d of the fastener, the embedding strength fh,0,k of adowel-type fastener parallel to the grain is calculated according toEurocode 5 [1] as follows:

fh,0,k = 0, 082 · ρk · d−0,3

in N/mm2 without predrilled holes (1)fh,0,k = 0, 082 · (1 − 0, 01 · d) · ρk

in N/mm2 with predrilled holes. (2)Embedding behaviour strongly depends on the grain directionof the wood: dowel-type fasteners loaded parallel to the grain(Fig. 1(a)) show a high initial stiffness and a plastic plateau whilefasteners loaded perpendicular to the grain are less stiff at lowerloads but show a continuous load increase before failure (see Fig. 2,taken from [2]).

The embedding strength is the first important parameter usedin Johansen’s yield theory [3] which is widely used for calculatingthe load-carrying capacity of dowel-type joints. The embeddingbehaviour of timber or wood-based materials is assumed to beinfinitely stiff/perfectly plastic (see Fig. 1(c)). This assumptionleads to the exclusively ductile failure modes described byJohansen.

2.2. Yielding of fasteners

Bending deformation of the steel fasteners is observed espe-cially for slender fasteners and the yield moment is the second im-portant parameter used in Johansen’s yield theory. Depending on

3020 H.J. Blaß, P. Schädle / Engineering Structures 33 (2011) 3018–3026

Fig. 5. (a) Splitting of a moment-resisting connection, (b) reinforcement using glued-on boards while embedment strength increases also.

Fig. 6. (a) Reinforcement by punched metal plate fasteners and (b) nail plates, (c) reinforced connection, screws in contact with dowels.

u

F

umax uFailure

F

F

max

Failure

Fy

uy

Kα

Kβ

tanβ = 1/6 * tanα

u

F

uy umax uFailure

F

F

F

y

max

Failure

0.5 Fmax

u

F

umax uFailure

F

F

max

Failure

Fy

uy

K10-40

K40-90

K//40-90

u

F

uFailure

0.8 Fmax

0.4 Fmax

K

Fy

uy

a

dc

b

α

α

β

Fig. 7. Different methods to determine ductility: (a) CEN-1/6-Method, (b) EEEP-Method (c) 0,5 Fmax-Method, (d) 10-40-90-Method.

the type and diameter d of the fastener as well as the steel qual-ity (given by the ultimate tensile strength) fu,k, the yield momentMy,k of bolts, dowels, nails or staples according to Eurocode 5 [1] iscalculated using the following equation:

My,k = 0, 3 · fu,k · d2,6 for dowel-type fasteners. (3)

The deformation of a dowel is shown in Fig. 3(a), as well as thecorresponding assumption for themoment–bending angle relation(Fig. 3(b)).

Modern timber screws are hardened; the tensile strengthof the steel often exceeds 1000 N/mm2. Due to the influence

H.J. Blaß, P. Schädle / Engineering Structures 33 (2011) 3018–3026 3021

Fig. 8. Typical load–displacement curves of non-reinforced and reinforced connections.

1,va 1,va a1 a1

(M1 to M4 in Fig. 10)

(M5 to M6 in Fig. 10)

(M7 to M10 in Fig. 10)

Fig. 9. Geometry of specimen for tests M1–M10 with different numbers ofreinforcing screws.

of the thread, the yield moment of these fasteners has to bedetermined by tests. The yield moment is then stated in thetechnical specification of the screw, e.g. in an ETA.

2.3. Failure modes in timber joints

Contrary to the assumptions in Johansen’s yield theory, thefailure of timber in connections is not always characterised by a

ductile embedding failure. Depending on timber thickness, load-grain angle, the spacing as well as the end and edge distances ofdowel-type fasteners and their slenderness ratio, timber showsa tendency to split in the connection area before the embeddingstrength is reached. Increasing fastener spacing decreases thesplitting tendency; however, large connection areas are necessary,which usually are not favoured. Decreasing fastener spacingparallel to the grain results in increasing splitting tendency andis considered in the design by using nef , an effective number offasteners.

nef =

[min

n; n0,9

·4

a1

10 · d

]·90 − α

90+ n ·

α

90

According to DIN 1052 [4]. (4)

Here, n is the number of fasteners parallel to the grain direction, a1is the fastener spacing, d is the fastener diameter and α is the anglebetween force and grain direction.

Fig. 4 shows two different types of splitting failure forloads parallel to the grain: after the critical tensile stressperpendicular to the grain is reached, uncontrolled crack growthdevelops and the split follows the dowel line (Fig. 4(a)). Thefailure type in Fig. 4(b) is caused by a combination of tensileperpendicular to the grain and shear stresses: here, timberplugs are sheared off between two dowels or between the lastdowel and the end grain. Both failure types are brittle andgenerally occur before the embedding strength of the timber isreached.

Fig. 10. Test results of non-reinforced (M1–M4) vs. reinforced (M5–M10) connections.

3022 H.J. Blaß, P. Schädle / Engineering Structures 33 (2011) 3018–3026

Fig. 11. Test configurations for (a) 1-24-2S 1.1, (b) 1-20-22_1.2, (c) 1-12-42 1.2.

If the tensile stresses perpendicular to the grain are transferredby reinforcement, e.g. by self-tapping screws with a full threadalong the shank and arranged perpendicular to the grain and to thedowel axis, the splitting type in Fig. 4(a) is avoided. In order to alsoprevent the failure type in Fig. 4(b), glued-on wood-based panelsor punched metal plate fasteners are needed as reinforcement.

If timber splitting is prevented by a reinforcement perpendic-ular to the grain, the connection failure modes coincide with theductile failure modes described by Johansen and nef reaches theactual number of fasteners n. If reinforcement is used provid-ing higher embedment strength than the timber in the connec-tion area, the reinforcement increases the load-carrying capacityof the connection beyond the values predicted by Johansen’s yieldtheory.

2.4. Reinforcing timber joints

Splitting of timber in the connection area is not only causedby the wedge effect of the dowel-type fasteners; restrainedshrinking may increase the tensile stresses perpendicular tothe grain and hence facilitate splitting (Fig. 5(a)). Both effects

can be counteracted by reinforcing the connection area. Thereinforcement in general has two effects: first, tensile stressesperpendicular to the grain and possibly shear stresses aretransferred and, depending on the reinforcement type and itsarrangement, the embedding capacity of the reinforced timberarea increases. Common methods for the reinforcement of timbermembers in the connection area are:

• Glued-on wood-based panels on both sides of the shear planein timber-to-timber connections and on the timber memberin steel-to-timber connections. The embedding strength of thereinforcing panels generally exceeds the embedding strength ofthe timber (Fig. 5(b)).

• Punched metal plate fasteners or nail plates as an alternativeto wood-based panels. The embedding strength significantlyexceeds the embedding strength of the timber (Fig. 6(a) and(b)).

• Self-tapping screws with a continuous thread over the shankeither without or with contact to the dowel-type fasteners. Inthe latter case the screws increase the embedment capacity(Fig. 6(c)).

H.J. Blaß, P. Schädle / Engineering Structures 33 (2011) 3018–3026 3023

v1v2Connections Top

v3v4Connections Bottom

v1v2Connections Top

Displacement in mm

Loa

d in

kN

0

20

40

60

80

100

120

140

160

180

200

10 11 12 13 14 15 160 1 2 3 4 5 6 7 8 9

Displacement in mm

Loa

d in

kN

0

5

10

15

20

25

30

35

40

Loa

d in

kN

Displacement in mm10 11 12 13 14 1815 17160 1 2 3 4 5 6 7 8 9

0

20

40

60

80

100

120

140

160

180

200

10 11 12 13 14 15 160 1 2 3 4 5 6 7 8 9

a

b

c

Fig. 12. Typical load–displacement curves for (a) 1-24-2S xx, (b) 1-20-22 xx, (c) 1-12-42 xx.

The load-carrying capacity of reinforced connections with glued-on wood-based panels or with punched metal plate fastenersor nail plates may be calculated e.g. according to [5]. The loadintroduced in the reinforcing plate is transferred through theadhesive bond or the nailed connection, respectively, into thetimber member.

Self-tapping screwswith continuous threads represent a simpleand economic reinforcement method. The screws are placed infront of the dowel-type fasteners perpendicular to the dowel axisand to the grain direction. If the reinforcing screws are placedin contact with the dowels, an increase of load-carrying capacitysimilar to the reinforced connections with wood-based panelsor nail plates is achieved. The load-carrying capacity may be

Table 1Classification of ductility according to Eurocode 8 [7].

Classification Static ductility ratio

Low ductility µ ≤ 4Medium ductility 4 ≤ µ ≤ 6High ductility µ ≥ 6

calculated using a mechanical model developed by Bejtka [6](chapter 3).

If no contact exists between reinforcing screw and the dowel,the lateral load-carrying capacity of the dowel can be calculatedaccording to Johansen’s yield theory.

3024 H.J. Blaß, P. Schädle / Engineering Structures 33 (2011) 3018–3026

Table 2Test matrix for non-reinforced vs. reinforced connections.

Test no. Density ρ (kg/m3) Number of reinforcing screwsn × m

Distance of reinforcement todowel a1,v (mm)

Maximum load R (kN) Ductility

M1 421 – – 32,3 2,1M2 422 – – 28,2 1,3M3 428 – – 33,7 1,6M4 423 – – 29,5 1,3

Average value non-reinforced 1,6M5 411 5 × 1 60 34,1 3,4M6 430 5 × 1 60 34,6 7,3M7 428 5 × 2 60 36,5 4,9M8 408 5 × 2 60 36,1 4,7M9 420 5 × 2 60 36,1 5,7M10 410 5 × 2 60 39,2 5,8

Average value reinforced 5,3

Reinforced tests: n = 5 dowels with d = 24 mm. a1/d = 5, fu = 360 N/mm2 , b = 100 mm.

Table 3Test matrix for joints with cross-laminated timber.

Test no. Number of tests Type of fastener Connection geometryt1 mm t2 mm a1,t mm a1 mm a2,c mm s m n

1-24-2S_1.1 4 (1)a Dowel d = 24 mm 60 (3)b – 7d 5d 3d 2 1 31-24-2S_2.1 3 (2) Dowel d = 24 mm 60 (3) – 5d 4d 5, 6d 2 1 51-24-2S_3.1 3 (2) Dowel d = 24 mm 128 (5) – 5d 4d 3d 2 1 31-20-22_1.1 6 (3) Dowel d = 20 mm 60 (3) 128 (5) 5d 5d 3d 2 1 31-20-22_1.2 6 (3) Dowel d = 20 mm 60 (3) 128 (5) 5d 4d 3d 2 1 31-20-22_1.3 3 (2) Dowel d = 20 mm 60 (3) 128 (5) 4d 4d 3d 2 1 31-12-42_1.1 3 (1) Screw 12 × 200/100 27 (3) 146 (5) 10d 4d 2, 5d 2 1 21-12-42_1.2 3 (1) Screw 12 × 200/100 27 (3) 146 (5) 12d 5d 3d 2 1 21-12-42_1.3 3 (1) Screw 12 × 200/100 27 (3) 146 (5) 6d 3d 3d 2 1 2

s: number of shear planes,m: number of columns of fasteners, n: number of fasteners in a row.a In brackets: number of tests with fastener in gap between two adjacent boards.b In brackets: number of layers in CLT member.

2.5. Determination of ductility

Ductility in general describes the ability of a structure toundergo large deformations in the plastic range before its collapse.Ductility often is defined as the ratio between ultimate and yielddisplacement:

µ =umax

uy. (5)

Ductile behaviour is especially important e.g. for structures inseismic regions. Classifications of ductility hence can be found inseismic codes like in Eurocode 8 [7] (Table 1).

To estimate ductility, the determination of a yield point (givenby yield force and yield displacement) is necessary. Since the yieldcriterion for timber connections is not well-defined as it is e.g. forsteel structures, different methods for the determination of theyield point for timber connections or structures exist (Fig. 7). Somewell-known procedures are (1) the 1/6 method according to EN12512 (CEN Method) [8], (2) the equivalent energy elastic–plasticmethod (EEEP), which is widely used in northern America, (3) the0, 5 ∗ Fmax-Method from Karacabeyli and Ceccotti and (4) the10–40–90-Method according to Yasumura and Kawei. A detaileddescription of the methods can be found in [9].

Since the CENmethod is specified in EC8, it is used to determinethe static ductility ratio in this paper. In some of the following testsit can be seen that none of the procedures in Fig. 7 seems to beadequate to evaluate the ductility of highly deformable joints orstructures.

3. Reinforced connections

A research project at KIT (formerly: Universität Karlsruhe)concluded in 2005 [6] dealswith the reinforcement of timber joints

using self-tapping screws. The experiments proved that crackgrowth can effectively be prevented by placing screws in front ofthe dowels of a dowel-type connection. Consequently, by avoidingsplitting of the specimen, the load-carrying capacity is increased(Fig. 8). Furthermore, the screws can be placed in contact with thedowel-type fasteners. In this case, the load-carrying capacity ofthe connections can be further increased while the joint stiffnessincreases aswell. A calculationmodel as an extension of Johansen’syield theory and based on theoretical and experimental studies ispresented in [6].

Preventing brittle failure modes by the use of screws asreinforcement, not only strength and stiffness values are increased.The ductile characteristics of timber joints also change in afavourable manner. Fig. 10 shows load–displacement curves offour non-reinforced vs. six reinforced test specimens. Whilebrittle failure characterises the behaviour of the non-reinforcedspecimensM1–M4, ductile behaviour is observed in all other tests.The tests M5 and M6 were carried out using one screw per doweland shear planewithout contactwith the dowel (according to Fig. 8(middle)). Starting with small cracks, the failure finally occurreddue to plug shear failure. TestM5 did not show pronounced ductilebehaviour. Tests M7–M10 were carried out using two screws perdowel and shear plane. Ductile behaviour can be observed in thesetests. The test matrix can be found in Table 2. The geometry of thespecimen is shown in Fig. 9.

The advancement in ductility can be seen in Table 2.While tests1–4 show static ductility ratios between 1 and 2— tests 5–10 showan average static ductility ratio of 5,3.

4. Connections in cross-laminated timber (CLT)

Cross-laminated timber is increasingly used as a structuralproduct in timber engineering. Connections between CLTmembers

H.J. Blaß, P. Schädle / Engineering Structures 33 (2011) 3018–3026 3025

Fig. 13. Static ductility ratios for 1-24-2S 1.1 (diamonds), 1-20-22_1.2.2 (squares), 1-12-42 1.2 (circles). The low values for test nos. 5 and 6 are the result of preliminarytensile failure of the side members.

Fig. 14. Failure modes of dowelled connections with cross-laminated timber: (a) dowels in gaps between adjacent boards, (b) embedment deformation, (c) brittle failurewith reduced spacing.

or between a CLT member and a solid or glued laminatedtimber member are often carried out using self-drilling screwsas fasteners. The fasteners may be arranged either perpendicularto the plane of the panels or in their edges. A research projectregarding connections in cross-laminated timber was carried outat Universität Karlsruhe in 2007 [10]. The aim of this project wasto determine the load-carrying capacity of joints with dowel-typefasteners in cross-laminated timber members.

Cross-laminated timber contains cross layers acting as rein-forcement: cross layers eliminate splitting to a large extent, cracksand plug shear failure, if any, are only observed in the outer layers.

Here, three test configurations are analysed:

• 10 testswith dowels in double shearwith outermembers of CLTand a steel plate as middle member.

• 15 tests with dowels in double shear and cross-laminatedtimber as outer and middle member.

• 9 tests with screws in double shear and cross-laminated timberas outer and middle member.

The test configurations can be seen in Fig. 11, the test matrix isgiven in Table 3. The different configurations exhibit rather typicalload-displacement curves which can be seen in Fig. 12.

Fig. 13 shows the static ductility ratios for the three testconfigurations. Comparing static ductility ratios of CLT connections

to the non-reinforced and screw reinforced timber connections,it can be seen that CLT connections show higher static ductilityratios. It was observed that the failure in CLT connections wasmostly caused by large embedment deformationswhile cracks andplug shear failure could only be observed in the outer CLT layers(Fig. 14(a) and (b)). Reduced spacing parallel to the grain from 5dto 4d more often leads to brittle failure (Fig. 14(c)).

5. Conclusions

Several methods for the reinforcement of timber joints arepresented in this paper. The primary reason for reinforcing timberjoints is to prevent brittle failure modes of timber due to splittingor shear and hence improve ductility. Depending on the type ofconnection to be reinforced, different methods may be applied.

For members loaded mainly in tension, the plastic fastenerdeformation in bending requires large embedding deformationwithout preliminary timber failure due to splitting or shear failure.An easyway for the reinforcement perpendicular to the grain of theconnections in tensile members is the use of self-drilling screws.The application is very fast and the screws are comparativelycheap. This leads to a cost-effective way for the reinforcement ofthe member’s joint areas. The screws may as well be placed incontact to the fastener which increases the load-carrying capacityand the stiffness of the joint.

3026 H.J. Blaß, P. Schädle / Engineering Structures 33 (2011) 3018–3026

The reinforcementmay also be carried out using punchedmetalplate fasteners or glued-on boards to increase the embeddingstrength of the timber. This method in general leads to a higherincrease of the joint’s load-carrying capacity, the application of thereinforcement, however, is more laborious and expensive.

To date, the main field of application for CLT has been its useas load-carrying panels loaded either in or perpendicular to itsplane. Connections in CLT show very favourable behaviour dueto the crosswise lamination of the layers, which can be seen as anatural reinforcement. Ongoing research projects are dealing withthe application of CLT as linear load-carryingmembers for exampleas tensile or compressive members in trusses or as beams.

References

[1] Eurocode 5: design of timber structures – part 1–1: general – common rulesand rules for buildings. German version EN 1995-1-1:2004+A1; 2008.

[2] YasumuraM. Determination of yield strength and ultimate strength of dowel-type timber joints. In: Proceedings of CIB-W18. Paper 33-7-1. Delft (TheNetherlands); 2000.

[3] Johansen KW. Theory of timber connections. In: International associationof bridge and structural engineering. Publication No. 9:249–262. Bern(Switzerland); 1949.

[4] DIN 1052 2008-12: design of timber structures — general rules and rules forbuildings. German version.

[5] Blaß HJ, Werner H. Stabdübelverbindungen mit verstärkten Anschluß-bereichen. In: Bauen mit Holz 90: S. 1988. p. 601–7.

[6] Bejtka I. Verstärkung von Bauteilen aus Holz mit Vollgewindeschrauben.Band 2 der Reihe Karlsruher Berichte zum Ingenieurholzbau. Herausgeber:Universität Karlsruhe (TH), Lehrstuhl für Ingenieurholzbau und Baukonstruk-tionen, Univ. -Prof. Dr. -Ing. H.J. Blaß; 2005.

[7] Eurocode 8: design of structures for earthquake resistance — part 1: generalrules, seismic actions and rules for buildings. German version EN 1998-1;2004.

[8] EN 12512. Timber structures – test methods – cyclic testing of joints madewith mechanical fasteners; 2001.

[9] Munoz W, Mohammad M, Salenikovich A, Quenneville P. Need for aharmonised approach for calculations of ductility of timber assemblies. In:Proceedings of CIB-W18. Paper 41-15-5. St. Andrews (Canada); 2008.

[10] Blaß HJ, Uibel T. Tragfähigkeit von stiftförmigen Verbindungen in Brett-sperrholz. Band 8 der Reihe Karlsruher Berichte zum Ingenieurholzbau.Herausgeber: Universität Karlsruhe (TH), Lehrstuhl für Ingenieurholzbau undBaukonstruktionen. Univ.-Prof. Dr. -Ing. H.J. Blaß; 2007.