Construction and Building Materials 29 (2012) 24–32

Contents lists available at SciVerse ScienceDirect

Construction and Building Materials

journal homepage: www.elsevier .com/locate /conbui ldmat

Structural behaviour of glued laminated timber beams pre-stressedby compressed wood

B. Anshari a,c, Z.W. Guan a,⇑, A. Kitamori b, K. Jung d, K. Komatsu b

a Department of Engineering, University of Liverpool, Liverpool, UKb Research Institute for Sustainable Humanosphere, Kyoto University, Uji, Kyoto, Japanc Department of Civil Engineering, University of Mataram, NTB, Indonesiad Faculty of Education, Shizuoka University, Japan

a r t i c l e i n f o a b s t r a c t

Article history:Received 22 March 2011Received in revised form 31 August 2011Accepted 2 October 2011Available online 24 November 2011

Keywords:Compressed woodGlulamMoisture-dependent swellingPre-camberPre-stressingReinforcement

0950-0618/$ - see front matter � 2011 Elsevier Ltd. Adoi:10.1016/j.conbuildmat.2011.10.002

⇑ Corresponding author. Tel.: +44 1517945210; faxE-mail address: zguan@liverpool.ac.uk (Z.W. Guan

In this study, glulam beams were strengthened by inserting compressed wood (CW) blocks into the pre-cut rectangular holes with one-thirds of the beam depth from the top of the beams. This practice was tomake use of moisture-dependent swelling nature of the compressed wood which was conditioned withthe moisture content significantly lower than the ambient one. The test results showed that a pre-camberwas produced in the mid-span of the beam reinforced due to expansion of the compressed wood blockson the top part of the beam. As a result, significant initial tensile and compressive stresses were generatedon both the top and the bottom extreme fibres of the beam, respectively. Subsequent bending testsrevealed that the initial stiffness and load carrying capacity of the pre-stressed beams were increased sig-nificantly in comparison to the beam without pre-stressing.

� 2011 Elsevier Ltd. All rights reserved.

1. Introduction

Glued laminated timber or glulam have been used in Europesince the middle of the 19th century [1]. Glulam is made of woodlaminations glued together to form a specific piece of wood for aspecific load. The interest to use this technology is to decreaseproduct variability and make it less affected by natural growthcharacteristics like knots. Besides, the glulam technology offersalmost unlimited possibilities of shape and design for construction,and is widely used for load bearing structures in houses, ware-houses, pedestrian bridges, etc.

Reinforcement of structural wood products using bonded rein-forcing materials and pre- or post-stressing techniques has beenstudied for many decades. In the earlier stages of the research,the focus was mainly on using metallic reinforcement, includingsteel bars, pre-stressed stranded cables and bonded steel and alu-minium plates [2–4]. Recently, research on glulam beams rein-forced with fibre and fibre-reinforced polymers (FRPs), such ascarbon, aramid and glass fibres, has been increased significantly,due to the high specific strength and stiffness of the FRP materials.Plevris and Triantafillou [5] studied the effect of reinforcing firwood with carbon/epoxy fibre-reinforced plastics (CFRPs). Plevris

ll rights reserved.

: +44 1517945218.).

and Triantafillou [6] investigated the creep behaviour of FRP-rein-forced wood and developed an analytical approach to predict time-dependent deflections of timber beams reinforced with CFRPlaminates with different thickness. Triantafillou and Descovic [7]also studied the effect of pre-stressed CFRP reinforcement bondedto European beech lumber. A lot of research work has been under-taken to investigate structural behaviour of glulam beams andsolid timber beams reinforced by FRP sheets or bars [8–13].

Issa and Kmeid [14] undertook research on glulam beams rein-forced with two types of reinforcement: steel plate and carbonfibre reinforced polymer. The reinforcement has changed the modeof failure from brittle to ductile and has increased the load-carry-ing capacity of the beams. Borri et al. [15] discussed the use ofFRP materials to strengthen the existing wood elements underbending loads.

Guan et al. [16] studied glulam beams pre-stressed by pultrud-ed GRP tendons. Finite element models were developed and vali-dated, which is capable of simulating the pre-camber introducedinto the beam due to transfer of the pre-stressing force. Corradiand Borri [17] also studied reinforcement of timber beams rein-forced with pultruded GFRP elements. The results indicated signif-icant improvement in flexural stiffness and capacity comparedwith unreinforced timber beams. Fiorelli and Dias [18] developeda theoretical model of fibre glass reinforced glulam beams withnecessary validation. Johnsson et al. [19] studied reinforced glulamusing pultruded rectangular carbon fibre rods and established the

Nomenclature

a distance between the support and the point load (mm)b beam width (mm)h beam depth (mm)f the actual pre-camber (mm)t time (day)t0 the thickness of the wood plate in the radial direction

before the densification processes (mm)t1 the thickness of the wood plate in the radial direction

after the densification processes (mm)w2–w1 incremental deflection corresponding to F2 � F1 (mm)y the measured deflection (mm)ySB pre-camber deflection for the short beam (mm)A1, B1 constants for the best fit line equation of pre-camber

deflection at mid span of the short beamA2, B2 constants for the best fit line equation of the initial ten-

sile strain on the top extreme fibre of the short beamA3, B3, C3 constants for the best fit line equation of the initial com-

pressive strain on the bottom extreme fibre of the shortbeam

A4, B4, C4 constants for the best fit line equation of the initial ten-sile strain on the bottom extreme fibre of the long beam

CR compression ratio (%)CW compressed woodEL modulus of elasticity in the longitudinal direction (MPa)ER modulus of elasticity in the radial direction (MPa)F2–F1 incremental load in elastic range (N)L, R, T longitudinal, radial and tangential direction, respec-

tivelyMC moisture content (%)MoEg global modulus of elasticity (MPa)ðetÞSB the measured extreme fibre tensile strain on the top of

the short beamðeCÞSB the measured extreme fibre tensile strain on the bottom

of the short beamðetÞLB the measured extreme fibre tensile strain on the top of

the long beam

B. Anshari et al. / Construction and Building Materials 29 (2012) 24–32 25

anchorage length for this system. The proposed reinforcementmethod increased the short-term flexural load-carrying capacityby 49–63% on average.

However, reinforcements currently used to strengthen glulambeams cover almost whole bottom surface of the beam. Althoughthey are effective, however due to various limitations, applicationsof strengthening the beam using CFRP, GFRP and other metallicmaterials are limited up to date. Possible problems of the existingstrengthening techniques facing are likely compatibility, de-bond-ing, stress relaxation and complex procedures.

The above problems may be overcome by the newly developedstrengthening techniques using compressed wood, which is tomake use of moisture-dependent swelling nature of compressedwood. In the reinforcing practice, glulam beams were strength-ened by inserting compressed wood blocks with the lower mois-ture content than the ambient one into the pre-cut rectangularholes on the top part of the glulam beams. Once the CW blockswere inserted, they would be gradually swelling due to absorbingmoisture from air until they reached the equilibrium state, i.e. thebalance between the moisture-dependent swelling and the con-strained expansion by surrounding glulam. The expansion onthe CW blocks on the top part of the beam would generate bend-ing moments with respect to the neutral axis that would create apre-camber of the beam. As a result, the up-lift deflection wouldalso produce initial tensile and compressive stresses at the topand bottom extreme fibres of the beam before applying a serviceloading. When such beam is used in construction, the pre-camberwill cancel out some downward working deflection and the initialextreme fibre stresses will cancel out some working stresses.Therefore, the beam size can be reduced (so material saved) butstill having the desired load carrying capacity, or more load canbe carried by the same beam, also with the bending stiffness in-creased. There are some major advantages of the above reinforc-ing technique, i.e. (1) it is very cost effective, (2) it uses timberto strengthen timber beams so that it is purely green, (3) verysmall amount of compressed wood is needed, (4) it is an easyand simple practice, (5) it will have significant impact on timberconstruction (domestic houses and large span commercial struc-tures), material savings, sustainability and environmental protec-tion by reducing CO2 and volatile organic compounds (VOC)emissions.

This paper aims to investigate how the moisture dependentswelling of the compressed wood, which was inserted in a pre-cut hole on the top part of a glulam beam, could generate pre-cam-ber and further enhance the structural performance of the beamreinforced. In addition, the ultimate failure modes of both thepre-stressed short and long glulam beams were presented.

2. Sample preparation

In this research, Japanese cedar (Crytomeria japonica D Don) wasused to manufacture compressed wood blocks and glulam beamswith the initial density of 300–420 kg/m3 and moisture content(MC) of 12% in a dry air condition.

2.1. Compressed wood

Compressed wood is made of lower grade timber through den-sification processes, which requires wood with free of knot and noother defects to ensure that it can be compressed uniformly in theradial direction. The manufacturing processes consist of preheat-ing, pressing and cooling. The preheating temperature was 180 �Cto ensure dimension stability of compressed wood produced. Thelevel of densification, which is also known as compression ratio(CR), can be represented as follows:

CR ¼ t0 � t1

t0� 100% ð1Þ

where t0 and t1 are the thickness of the wood plate in the radialdirection before and after the densification processes, respectively.In this research programme, the compression ratio of CW blockswas set to 70%. In order to obtain the possible maximum swellingof compressed wood, low moisture content needs to be conditioned.Here the initial moisture content of 6% was chosen.

The density of the compressed wood block is 1163 kg/m3 inaverage which was increased by over 200% from that of the soft-wood. Material properties of CW blocks with CR = 70% were greatlyenhanced, e.g. the Young’s modulus in the L (longitudinal) and R(radial) directions increase significantly to 32,858 MPa and3111 MPa respectively in comparison to the normal Japanese cedarwith EL = 8017 MPa and ER = 753 MPa [20]. Compressed wood

Table 1Summary of short glulam beams with and without CW reinforcement.

Specimenname

CW size(in mm)

No. of CWblock

No. ofbeam

CW/beamvolume (%)

L R T

SB-0CW 0 0 0 0 3 0SB-1CW30 35 30 63 1 3 0.40SB-3CW15 35 15 63 3 3 0.60SB-3CW30 35 30 63 3 3 1.20SB-3CW45 35 45 63 3 3 1.80

Fig. 2. Locations of strain gauges.

26 B. Anshari et al. / Construction and Building Materials 29 (2012) 24–32

lamina was made of Japanese cedar with the initial dimensions of1200 � 60 � 150 (in mm) to produce CW lamina size of1200 � 30 � 120 (in mm) coincident with the L, R, T (tangential)directions. The dimension reduction from 150 to 120 mm in thetangential direction was just a cut to fit the host beam. However,the compression ratio of CW lamina was 50%.

2.2. Glulam beams

The raw material for making short glulam beams was Japanesecedar log which was available in Laboratory of Structural Function,Kyoto University, Japan. The dimensions of the short glulam beamto be reinforced by CW blocks were 105 � 105 � 1500 in mm. Thesize of lumber used to make the glulam beam was 40 � 120 � 1600in mm.

The long glulam beams were also produced in a glulam factory(Meiken Co. Ltd., Yamaguchi) in Japan, which were made of Japa-nese cedar with specifications as follows. The dimensions of thelong glulam were 3800 mm long, 200 mm deep and 120 mm wide.The grade of glulam beam was based on the Japanese IndustrialStandard (JIS), JASE65f225. According to the data from the factorythe modulus of elasticity of glulam beams without CW lamina wasvaried between 7205 and 7832 MPa.

Table 1 shows 15 short glulam beams produced, including threecontrol beams without insertion of any CW blocks. The table alsoshows 30 CW blocks produced, consisting of nine CW blocks witht = 15 mm, 12 CW blocks with t = 30 mm, and nine CW blocks witht = 45 mm.

There are total seven long beams in which two of them werereinforced at their bottom surfaces with a lamina made of com-pressed Japanese cedar. Among them, there were three beamswithout any reinforcement for control purpose. CW laminas asreinforcement for two glulam beams with 1200 mm length wereprepared in Laboratory of Structural Function, Kyoto University.The top figure in Fig. 1 shows the beam reinforced by three CW

100

45

120

200

120

65 65

Top view

Side view

900

100

45

600600

120

200

120

65 65

Top view

Side view

FJ

Fig. 1. Full-size glulam beams pre-stressed by thre

blocks without CW lamina where the arrangement of CW blockswas symmetrical. The bottom figure in Fig. 1 shows the beam rein-forced by five CW blocks and a CW lamina at the bottom surfacewith three finger joints located at 900, 1800 and 2700 mm fromthe left support. The dimensions of the CW block are65 � 45 � 65 (in mm) coincided with the L, R, T directions of com-pressed wood.

A series of parametric studies were carried out using the finiteelement analysis to identify reasonable size and number of thecompressed wood block, which would not generate the critical lo-cal stresses on the beam due to expansion of the compressed woodblocks.

3. Experimental work

3.1. Pre-stressing processes and measurements

Pre-stressing technology developed is to make use of moisture-dependentswelling nature of compressed wood. The main reason to use compressed wood, in-stead of pre-dried wood, is that the former shows much less stress relaxation [21],which would maintain a reasonable level of the pre-stress. Since there is the largestswelling in the radial direction of compressed wood, the CW block was placed in away in which its radial direction is coincident with the longitudinal direction of thebeam to be strengthened. As indicated previously, the CW blocks were conditionedto have MC of 6% prior to the insertion. The beams pre-stressed would be placed inan ambient condition with MC of 12%.

3.1.1. Short glulam beamsAfter completed insertion of the CW blocks, beams were subjected to measure-

ments of the pre-camber deflection and strains at selected positions on the beam.The displacement transducer CDP-25 and strain gauge PFL-10-11 (Tokyo SokkyKenkyujo Co. Ltd.) were used. Seven strain gauges were attached to the beam alongits depth at the mid span (four strain gauges) and 130 mm away from the mid-span(three strain gauges), as shown in Fig. 2. One transducer was placed underneath themid span at the selected beams, as shown in Fig. 3.

All beams were placed in a conditioning chamber where the Relative Humiditywas set to be fluctuated between 40% and 80% (typical RH range in Japan) in a 3-daycycling but the temperature was kept at 20 �C. Readings of transducers and strain

L=3600

100

900

L=3600100

600 600

FJFJ

e CW block and five CW blocks with a lamina.

Fig. 3. Deflection measurement set-up.

B. Anshari et al. / Construction and Building Materials 29 (2012) 24–32 27

gauges were recorded by a Personal Computer through a data logger. The measure-ments would be terminated when readings from displacement transducers andstrain gauges were well into a stable phase, i.e. the compressed wood blocks hadreached the maximum swelling after absorbing moisture and further swellingwas constrained by glulam surrounded. The time interval set to record measure-ments was 30 min for about 8 weeks.

3.1.2. Long glulam beamsThe long glulam beams were placed in a conditioning room with the same

ambient conditions set to the short glulam beams. The beams were subjected tostrain measurements at the selected positions. Total 46 strain gauges were attachedto five beams, in which 10 strain gauges were for each of four beams strengthened,as shown in Fig. 4, and six gauges for the beam without reinforcement. The timeinterval set to record measurements was 30 min for a period of six and a half weeks.

Measurements of the pre-camber deflections were taken manually by using asteel I-beam, long clamp, and also automatically by a transducer CDP-25 set atthe mid-span underneath the beam. Fig. 5 shows the setting up of pre-camberdeflection measurement. There are several thin Teflon sheets attached the glulambeam to minimize friction between steel and timber during the measurement.The actual pre-camber (f) is assumed to be equal to the measured deflection (y),as shown in Fig. 5. To validate the result, it was repeated the same procedures bychanging the clamped end.

3.2. Destructive flexural tests

3.2.1. Short glulam beamsFour-point bending tests were undertaken for all beams with test arrange-

ments, as shown in Fig. 6. All strain gauges used to measure strains related to thepre-camber deformations were also used during the destructive tests. One CDP-50 transducer and five CDP-25 transducers were set-up to measure the ultimatedeflection and curvature of the beam, respectively. The Universal Testing System(UTS) Instron 2250 was used to carry out the tests, with the maximum capacityof 100 kN and the crosshead speed set to 2 mm/min.

From the measured parameters of the above tests, an apparent elastic bendingmodulus and strength can be calculated based on simple bending theory [21],which will be compared for beams with and without the local pre-stressing inthe results.

3.2.2. Long glulam beamFour-point bending tests were also undertaken for 7 long glulam beams. The

Universal Testing System (UTS) Instron was used to carry out the tests, with themaximum capacity of 1000 kN and the crosshead speed set to 2 mm/min. Thearrangements of the destructive bending test are shown in Fig. 7. Two transducers(DTP-05) were placed at the mid span and two CDP-50 at each support to measurethe relative displacements caused by the rotation of the beam. There were also twotransducers (CDP-10) and one CDP-25 at two side of the mid span and the top sur-face of mid span to measure curvature of the beam deformed.

Based on the measured parameters from the above tests, the global modulus ofelasticity in bending can be obtained by using the following equation [22]:

Fig. 4. Strain gauge location t

MoEg ¼L3ðF2 � F1Þ

bh3ðw2 �w1Þ3a4L

� �� a

L

� �3� �

ð2Þ

where MoEg is the global modulus of elasticity (MPa).

4. Results and discussion

4.1. Pre-camber deflection and pre-stressing state

4.1.1. Short glulam beamsFig. 8 shows the pre-camber deflections of pre-stressed glulam

beams built-up for 60 days after inserting the CW blocks. For glu-lam beams reinforced by one 30 mm thick CW block or by three15 mm CW blocks, although there were significant increases (espe-cially for the former beam) of the pre-camber deflection in the first20 days the increasing rates were slow down greatly after that. Infact, the deflection–time curves were then almost horizontal, i.e.with very small slopes. The ultimate upward deflections measuredfor both beams were about 1.5 mm. This indicated that such rein-forcements might not be effective due to only the small momentgenerated by the moisture-dependent swelling from a relativelysmall volume of the CW blocks. However, for beams reinforcedby three 30 or 45 mm thick CW blocks, there were sharp increaseson the pre-camber deflections in the first 20 days followed by stea-dy increases of the pre-camber deflection at the mid-span until theend of measuring period of 60 days. The measured upward deflec-tions were 2.8 mm and 3 mm for the former and the latter beams,respectively.

Fig. 8 also shows the best fit curves of the measured pre-camberdeflections of beams reinforced, which represent the averaged val-ues of the up and down deflection measurements caused by thecontrolled fluctuated relative humidity in the chamber. The best-fit equations for the pre-camber deflections of the short beamsshown in Fig. 8 are expressed as:

ySB ¼ A1ð1� Bt1Þ ð3Þ

where t is the time (day), A1 and B1 are constants. The aboveequations represent the basic features of the experimentalmeasurements.

Fig. 9a shows the measurements of the top extreme fibre strainsof the pre-stressed beams reinforced by three CW blocks with

hrough long beam depth.

Clamping

Steel I-beam

Transducers CDP-25

Pre-camber

f

y

Fig. 5. Set-up for measuring the pre-camber deflection of the long beam.

h

l=1200

CDP-25

CW

150150l1=300

350 CDP-25

Loading point

CDP-25

a=425

b

supportsupport

Fig. 6. Set-up of the four-point bending test for the short beam.

3600100

100

1200

200 Side view

1200 1200900

Top view

CDP-50 CDP-10

DTP-05

1200 1200

120

CDP-50

CDP-25

CDP-25 CDP-10

Loading point

Fig. 7. Set-up of the four-point bending test for the long beam.

Time (day)

0 10 20 30 40 50 60

Def

lect

ion

(mm

)

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

1CW_t=30mm3CW_t=15mm3CW_t=30mm3CW_t=45mm

Fig. 8. Pre-camber deflections at the mid span of the short beams reinforced byvarious CW blocks.

28 B. Anshari et al. / Construction and Building Materials 29 (2012) 24–32

thickness of 45, 30, 15 mm and one 30 mm thick CW block, built-up for 60 days. Tensile strains also show significant increases in thefirst 20 days of measurement and subsequent mild increases after-wards. The tensile strains on beams reinforced by three CW blockswith thickness of 30 and 45 mm after the measuring period of60 days were about twice of those on beams reinforced by three15 mm thick CW blocks and one 30 mm thick CW block. The max-imum strains measured were 1800 and 1600 micro-strains forbeams strengthened by three CW blocks with thickness of 45

and 30 mm respectively. Within the elastic range, this indicatedthat the initial tensile stresses of 15.3 and 13.6 MPa were gener-ated at the top extreme fibres of the beams pre-stressed, respec-tively. The best fit equations of the measured extreme fibretensile strains on the top of the beam take the similar form as thatshown in Eq. (3).

ðetÞSB ¼ A2ð1� Bt2Þ ð4Þ

where A2 and B2 are constants.The initial compressive strains were also generated at the bot-

tom extreme fibres of the beams reinforced due to pre-camberdeflection, as shown in Fig. 9b. Similarly, there were significant in-creases on the extreme fibre strains in the first 20 days, althoughthe increasing rates in comparison to those on the top extreme fi-bres were more than halved. This is due to the locations of com-pressed wood blocks, as they are placed in the top one-thirds ofthe beam depth. Hence, the expansion of CW blocks induces theinitial strain at the top extreme fibre much greater than that onthe bottom extreme fibre. The results showed the maximum com-pressive strain of 700 micro-strains was obtained on the beamreinforced by three 45 mm thick CW blocks, which gives an elasticstress of 6 MPa. The related regression equations of the best fitlines of the measured bottom extreme fibre strains are listedbelow.

ðeCÞSB ¼ A3ð1� Bt3Þ � C3 ð5Þ

where A3, B3 and C3 are constants.

Time (day)

Stra

in (

mic

ro-s

trai

n)

0

500

1000

1500

2000

25001CW_t=30mm3CW_t=15mm 3CW_t=30mm 3CW_t=45mm

Time (day)0 10 20 30 40 50 60 0 10 20 30 40 50 60

Stra

in (

mic

ro-s

trai

n)

-800

-600

-400

-200

01CW_t=30mm 3CW_t=15mm3CW_t=30mm3CW_t=45mm

(a) (b)Fig. 9. The measured extreme fibre strains at the mid-span of the short glulam beams pre-stressed: (a) at the top and (b) at the bottom.

Table 2The measured pre-camber deflection.

Beam specimen Pre-camber (mm) Deflection/span

LB-0CWB-0CWL 0 0LB-3CWB-0CWL 4.48 1:804LB-5CWB-0CWL 8.13 1:443LB-5CWB-1CWL 7.46 1:483LB-7CWB-1CWL 8.63 1:417

B. Anshari et al. / Construction and Building Materials 29 (2012) 24–32 29

4.1.2. Long glulam beamTable 2 shows the measured pre-camber deflections of pre-

stressed long glulam beams in 45 days after inserting the CWblocks. Clearly, glulam beams reinforced by three, five and seven45 mm thick CW blocks produced pre-camber significantly atmid span. The highest pre-camber deflection occurred at the glu-lam beam reinforced by seven CW blocks which reaches the ulti-mate deflection of 8.6 mm. The pre-camber deflection of theglulam beam reinforced by five CW blocks without CW laminawas higher than that of the similar glulam beam but with CW lam-ina. This was due to the high stiffness of the compressed woodlamina at the bottom layer which reduces the pre-camber deflec-tion at mid span generated by the expansion of CW blocks on thetop part of the beam.

Fig. 10a shows the measurements of the top extreme fibrestrains of the pre-stressed beams building up for 45 days. Clearly,the measured strains were gradually increased until the end oftesting. The maximum extreme fibre strain measured was 1150micro-strain on the top of the glulam beam reinforced by sevenCW blocks and a CW lamina, which is related to an extreme fibrestress of 8.6 MPa based on modulus of elasticity of 7478 MPa. Forbeams reinforced by three 45 mm CW blocks and five 45 mm CWblocks without CW lamina, the increases in the tensile strain weremild throughout the whole stage of testing. The measured strainson the top extreme fibre were 481 and 556 micro-strain for the for-mer and the latter beams, respectively. Within elastic range, thesestrains indicated that the initial tensile stresses of 3.8 and 4.3 MPawere generated on the top extreme fibres of the beams pre-stressed, respectively. Comparing with the maximum initial tensilestress (15 MPa) generated on the short beams pre-stressed, that(8.6 MPa) on the long beam reinforced by seven CW blocks is over40% lower. This hints that amount of CW used to pre-stress thebeams needs to be increased significantly.

Due to the limited time, all measurements were terminatedafter 45 days. However, the trends of all curves were well estab-lished (Fig. 10a). Therefore, the reliable extrapolation may be pro-duced based on the best fit lines up to a certain period not too faraway from 45 days, see 60 days. Eq. (6) below represents theregression lines of the top extreme fibre tensile strains measured.

ðetÞFB ¼ A4ð1� Bt4Þ ð6Þ

where A4 and B4 are constants.The initial compressive strains were also generated at the bot-

tom extreme fibres of the beams due to the pre-camber deflections,as shown in Fig. 10b. It also indicated that the glulam beam whichwas not reinforced by CW lamina shows relatively constant strainreadings in comparison to those reinforced by CW lamina. This is

due to the effects of swelling and shrinkage of CW lamina duringmeasurements in a chamber with cyclic RH. The best fit curve isnot meaningful because of the tiny increase in relation to relativelylarge fluctuations of the strain measurements.

4.2. Destructive flexural tests

4.2.1. Short glulam beamFig. 11a shows the averaged load–deflection relationships for all

beams tested. Generally, the curves demonstrate the overall linearfeatures until the load reaches 30 kN. However, there were reduc-tions of the slope in the later stage before the rupture, which werelikely caused by local compressive deformations in the CW–glulaminterface areas. The chart clearly shows distinguished initial stiff-ness of beams with different pre-stressing manners. The beamsreinforced by three 30 and 45 mm thick CW blocks have the initialstiffness improved by 19.2%, and 21.6% respectively compared withbeams without any reinforcement. The enhancement of the initialstiffness is mainly contributed by the pre-camber deflection, whichreduces the ultimate deflection.

In all the tests carried out, the tensile rupture failure occurred atthe bottom mid-span. In general, the initial crack was initiated atthe bottom extreme fibre around the mid span and followed by afast crack growth due to brittle splitting. Noted that the ratio be-tween the span and the depth was much less than 18 as requiredin Eurocode 5 [22], the failure mode tended to be shear–tensioncombined.

Results of the destructive bending tests are summarised in Ta-ble 3. They demonstrated that in comparison to the control beamthe load carrying capacities were increased by 14% and 19% for glu-lam beams reinforced by three 30 mm and three 45 mm thick CWblocks, respectively.

4.2.2. Long glulam beamFig. 11b shows the load–deflection relationships of the long

beams tested. Generally, the curves demonstrate overall linear fea-tures initially until a certain loading level, which was varied from

Time (day)Time (day)0 10 20 30 40 50 60

Stra

in (

mic

ro-s

trai

n)

Stra

in (

mic

rost

rain

s)

-300

-200

-100

0

100

7CWB-1CWL5CWB-0CWL5CWB-1CWL3CWB-0CWL

7CWB-1CWL5CWB-0CWL

5CWB-1CWL3CWB-0CWL

1400

1200

1000

800

600

400

200

00 10 20 30 40 50 60

1,2,3,4=best fit

1

2

3

4

(a) (b)Fig. 10. The measured extreme fibre strains at the mid span of the long glulam beams pre-stressed: (a) at the top and (b) at the bottom.

Fig. 11. Load–deflection relationships obtained from destructive tests after the pre-stressing: (a) the short glulam beams and (b) the Long glulam beams.

Table 3Summary of bending test results of the short beams.

Specimen MoE[avg]

(MPa)MoR[avg]

(MPa)Improvement(%)

Pmax

[avg]

(kN)

Improvement(%)

SB-0CW 6691 39 0 35.32 0SB- 1CW30 6301 4310.03 38.85 9.99SB- 3CW15 6327 4515.25 40.70 15.23SB- 3CW30 5912 4413.97 40.41 14.41SB- 3CW45 6741 4619.02 42.03 18.99

Table 4Summary of bending test results of the long beams.

Specimen Pmax

(kN)Stiffness(N/mm)

MoEL

(MPa)MoEg

(MPa)MoR(MPa)

LB-0CWB-0CWL 57.76 707.7 7789 7186 43LB-3CWB-0CWL 58.59 768.6 7645 7311 44LB-5CWB-0CWL 58.67 826.6 6881 7978 45LB-5CWB-1CWL 64.30 970.2 10,833 10,187 48LB-7CWB-1CWL* 32.67 1032.2 10,356 9185 25

* Pre-mature failure.

30 B. Anshari et al. / Construction and Building Materials 29 (2012) 24–32

curve to curve. However, there were slope reductions in the laterstage, which were again likely caused by local deformations inthe CW–glulam interface regions. The chart clearly shows the dis-tinguished initial stiffness of beams with different pre-stressingarrangements comparing with beams without any reinforcement.The beam reinforced by seven 45 mm thick CW blocks and a CWlamina at the bottom layer has the highest stiffness of 1032 kN/m, which is a 45.8.1% increase in comparison to the beam withoutany CW reinforcement. The initial bending stiffness for beams rein-forced by three 45 mm and five 45 mm CW blocks but without CWlamina also increased by 8.6% and 16.8% respectively. It clearlyindicated that the reinforcement with CW lamina contributed toincrease on bending stiffness, as expected.

Results of the destructive bending tests are summarised inTable 4. It demonstrated that in comparison to the control beam

the load carrying capacity were increased by 11.2% for glulambeam reinforced by five 45 mm thick CW blocks and a CW lamina.However, the load carrying capacities for glulam beams reinforcedby three and five 45 mm CW block without CW lamina were in-creased slightly by 1.4% and 3.8% respectively. Bending test of glu-lam beams reinforced by seven 45 mm thick block and a CWlamina was terminated at load of 32 kN due to premature failure,which was caused by knots at the second laminate from bottomof the beam and deterioration of the finger joints located at themid-span of the extreme tensile fibre. The finger joints on the lam-ina were not strong due to the thin nature of the connections asso-ciated with the lamina and the extreme loading condition there.However, this specimen did show the highest initial bending stiff-ness. In terms of MoR, the bending strengths of full size glulambeams were not significantly increased in comparison to those ofthe short glulam beams, except for the beam reinforced by five45 mm thick CW blocks and a CW lamina with the maximumMoR of 48 MPa obtained. This may be due to a high tensile strengthand sound finger joints of the CW lamina at mid span.

Crack Delamination

(a) (b)Fig. 12. The typical failure modes obtained from the destructive tests after the pre-stressing: (a) the short glulam beam and (b) the long glulam beam.

Fig. 13. Comparion of the pre-camber and the peak load for all beams reinforced with different CW/beam volume ratios: (a) the short glulam beam and (b) the long glulambeam.

B. Anshari et al. / Construction and Building Materials 29 (2012) 24–32 31

Global modulus of elasticity in bending for the pre-stressedbeam was obtained from the relationship between the load andthe total deflection at the whole beam region. This modulus wascalculated based on the load–deflection curve between 10% and40% of the ultimate load using Eq. (2). The results indicated thatthe global bending modulus increased significantly by increasingthe amount of CW blocks with or without CW lamina. The maxi-mum bending modulus was 10187 MPa for the beam reinforcedby five 45 mm thick CW block and a CW lamina. However, theMoR for a beam reinforced by seven CW blocks and a CW laminawas quite low, which indicated a premature failure as mentionedbefore. The bending strength or modulus of rupture (MoR) forthe beam reinforced with five 45 mm CW blocks and a CW laminaincreased by 11.6%, i.e. from 43 MPa to 48 MPa.

Fig. 12 shows the failure modes obtained from the destructivetests after the pre-stressing for both the short and the long glulambeams. For the short glulam beam (Fig. 12a), shear failure was ini-tiated from the bottom mid-span and propagated towards one ofthe supports, as expected. For the long glulam beam (Fig. 12b),delamination was appeared at the interface related to the bottomlaminate at the mid-span. There was no damage on the top halfof both the short and the long beams, as the compressed woodblocks made the top half of the beam stronger in compression.

Fig. 13 gives comparison of the pre-camber and the peak loadfor beams reinforced with different CW/beam volume ratios. Forthe short glulam beams (Fig. 13a), both the pre-camber and thepeak load are on a general up-trend with increasing the CW/beamvolume ratio, as expected. For the long glulam beams (Fig. 13b), thepeak load is on a up-trend (excluding the one for the pre-maturefailure beam). However, the pre-camber corresponding to thebeam with only 1.04 of CW/beam volume ratio is higher than that

of the beam with the much higher ratio of 15.48, and only slightlylower than that of the beam with the ratio of 15.95. The latter twobeams are all reinforced with a large volume of CW lamina. Thisindicates that the attachment of lamina on the tension face (bot-tom layer) of the beam reinforced is not effective to enhance thepre-camber. However, there is a significant increase on the peakload for the beam with CW lamina, as expected.

5. Conclusions

A new approach to strengthen glulam beams has been estab-lished in this research. The use of compressed wood made of a low-er grade wood through densification as a reinforcing material hasbeen approved to be effective. As only a small amount of com-pressed wood is required and no bonding between the CW andthe beam is necessary, the techniques developed are economicaland environmental friendly. This research shows that the con-trolled moisture-dependent swelling of compressed wood is crucialto pre-stress glulam beams and to produce pre-camber as well asthe initial stresses to cancel out working counterparts partially.

Destructive bending tests for all beams pre-stressed by thecompress wood have also indicated that there are significantenhancements on the bending capacity and the initial stiffness.For the short glulam beams reinforced by three 30 and 45 mmthick CW blocks corresponding to the reinforcement volume ofonly 1.2% and 1.8% of the total beam volume, there are increasesof 19% and 22% in bending stiffness and 14% and 19% in load carry-ing capacity respectively. For the long beams the enhancements ofthe bending stiffness were 37.1% and 45.8% for the beam reinforcedby five and seven CW blocks and a CW lamina respectively. Interms of load carrying capacity, a beam reinforced by five CW

32 B. Anshari et al. / Construction and Building Materials 29 (2012) 24–32

blocks and a CW lamina carries the maximum load of 64.3 kN,which is an 11% increase in comparison to the control beam. Thebeam reinforced by seven 45 mm thick CW blocks should show ahigher improvement on load carrying capacity following its initialstiffness if there was no premature failure. Amount of CW used tostrengthen the long beam can be increased to enhance the pre-camber and the corresponding initial extreme fibre stresses.

Cyclic and visco-elastic creep behaviour of the pre-stressedbeam systems are currently under investigation in the numericalmodelling. In addition, stress relaxation is being investigated. Thiswork will enhance further understanding of structural behaviour ofglued laminated beams reinforced by compressed wood.

Acknowledgments

Authors are grateful to the Higher Education Directorate Gen-eral, Ministry of National Education, Republic of Indonesia for thefinancial support of this Project. We specially would like to expressour sincere thanks to the Laboratory of Structural Function, RISH,Kyoto University for supporting all experimental work.

References

[1] Andre A. Fibres for strengthening of timber structures, research report, forestand wood product research and development corporation, Australia, vol. 3;2006. p. 1–91.

[2] William MB, Sandberg LB, Greg JW. Steel-reinforced glued laminated timber. JStruct Eng, ASCE 1989;115:433–44.

[3] Bohannan J. Prestressed wood members. Forest Prod J 1962;12:596–602.[4] Peterson J. Wood beams prestressed with bonded tension elements. J Struct

Eng, ASCE 1965;91:103–19.[5] Nikolaos P, Thanasis CT. FRP-reinforced wood as structural material. J Mater

Civil Eng 1992;4:300–17.[6] Plevris N, Triantafillou TC. Creep behavior of FRP-reinforced wood members. J

Struct Eng, ASCE 1995;121:174–86.

[7] Triantafillou TC, Deskovic N. Prestressed FRP Sheets as external reinforcementof wood members. J Struct Eng, ASCE 1992;118:1270–84.

[8] Dagher H, Kimball TE, Shaler SM, Abdel-Magid B. Effect of FRP Reinforcementon Low Grade Eastern Hemlock Glulams. National conference on woodtransportation structures. Department of Agriculture, Forest Service, ForestProducts Laboratory; 1996.

[9] Hernandez R, Davalos JF, Sonti SS, Kim YJ, Moody R. Strength and stiffness ofreinforced yellow-poplar glued-laminated beams. Forest Products Laboratory,Research Paper; 1997. FPL–RP–554:1-28.

[10] Blab HJ, Romani M. Reinforcement of glulam beams with FRP reinforcement.Research Report; 1998–2000.

[11] Romani M, Blab HJ. Design model for FRP reinforced glulam beams. VeniceItaly: International Council for Research and innovation in Building andConstruction; 2001.

[12] Gentile C, Svecova D, Rizkalla SH, Plevris N, Triantafillou TC. Timber beamsstrengthened with GFRP bars: development and applications creep behavior ofFRP-reinforced wood members. J Compos Constr 2002;6:11–20.

[13] Fiorelli J, Dias Antonio A. Analysis of the strength and stiffness of timber beamsreinforced with carbon fiber and glass fiber. Mater Res 2003;6:193–202.

[14] Issa CA, Kmeid Z. Advanced wood engineering: glulam beams. Constr BuildMater 2005;19:99–106.

[15] Borri A, Corradi M, Grazini A. A method for flexural reinforcement of old woodbeams with CFRP materials. Compos Part B: Eng 2005;36:143–53.

[16] Guan ZW, Rodd PD, Pope DJ. Study of glulam beams pre-stressed withpultruded GRP. Comput Struct 2005;83:2476–87.

[17] Corradi M, Borri A. Fir and chestnut timber beams reinforced with GFRPpultruded elements. Compos Part B: Eng 2007;38:172–81.

[18] Fiorelli J, Dias AA. Fiberglass-reinforced glulam beams: mechanical propertiesand theoretical model. Mater Res 2006;9:263–9.

[19] Johnsson H, Blanksvärd T, Carolin A. Glulam members strengthened by carbonfibre reinforcement. Mater Struct 2007;40:47–56.

[20] Anshari B, Guan ZW, Kitamori A, Jung K, Hassel I, Komatsu K. Mechanical andmoisture-dependent swelling properties of compressed Japanese cedar. ConstrBuild Mater 2011;25(4):1718–25.

[21] Kitamori A, Jung K, Komatsu K. Utilization of compressed wood as mechanicalfasteners of friction joints in timber buildings. In: Proceedings of the 11thinternational conference on non-conventional materials and technologies(NOCMAT 2009), 6–9 September 2009, Bath, UK.

[22] Anonymous. Eurocode 5, Timber structures-structural timber and gluelaminated timber-determination of some physical and mechanicalproperties. BS EN 408:2003.1-15.