The 6 x 12 in timber beam has been strengthened

Composite Materials Term Paper

Flexural Stiffness and Strength of GFRP-Reinforced Timber Beams

Hanan Alhayek1 and Dagmar Svecova2

Abstract: An experimental program was conducted at the University of Manitoba to test salvaged timber stringers strengthened with glass fiber-reinforced polymer (GFRP) laminates. A total of 20 creosote-treated Douglas Fir beams with dimensions of 130 × 330 × 4;500 mm were tested in three-point bending. Two groups were studied to examine the effect of GFRP reinforcement on the stiffness increase of the beams. The first group of 10 samples were reinforced with GFRP laminates on the tension side only (Group T), whereas the other group of 10 samples was reinforced with GFRP laminates alongside both tension and compression zones (Group TC). This study found that the strengthening with GFRP laminates, on average, increased the strength and the stiffness of the beams, respectively, by 36 and 3% for Group T and by 31 and 3.5% for Group TC. An analysis of a database of fiber-reinforced polymer (FRP)–strengthened timber beams tested by other researchers facilitated further study on the effect of FRP on the behavior of timber. The comprehensive analysis revealed a minimal stiffness increase in timber beams strengthened with FRP. Some evidence exists, however, that the beam span-to-depth ratio is an important factor to consider when strengthening timber beams. Beams with smaller span-to-depth ratios showed some increase in stiffness with in- creasing the reinforcement ratio; however, beams with larger span-to-depth ratios showed no real enhancement of beam stiffness, unless the reinforcement ratio was around 7 times the minimum code-recommended value. This result supports the current CSA provisions that do not advocate for stiffness increase when this strengthening method is used. The analysis shows that stiffness increase in GFRP-strengthened timber beams, based on results of small-scale samples, is minimal. DOI: 10.1061/(ASCE)CC.1943-5614.0000261. © 2012 American Society of Civil Engineers.

CE Database subject headings: Wood beams; Fiber-reinforced polymer; Stiffness; Flexural strength.

Author keywords: Timber; Fiber-reinforced polymer; Strengthening; Strength; Stiffness.

Introduction

North America has a large number of timber bridges that will need rehabilitation or replacement in the near future. The owners of tim- ber bridges report a frequent requirement to resurface these struc- tures. The low stiffness of these bridges is suspected to contribute to this problem. For this reason, and to reduce maintenance costs, Manitoba Infrastructure and Transportation supported the research to investigate stiffening options for timber bridges. The estimated cost of infrastructure replacement in the Province of Manitoba alone, for example, is projected at more than $40 billion dollars (CAN). Considering the high cost of bridge replacement, an essen- tial need exists for novel economical bridge rehabilitation schemes. In recent years, increased attention has focused on the rehabilitation of timber bridges as one solution to deteriorating infrastructure. To a great extent, this heightened interest is the result of new tech- nologies and advances in engineering materials.

In previous decades, many studies were performed on strength- ening timber beams with fiber-reinforced polymers (FRP) (Kuilen 1991; Svecova and Eden 2004; Amy and Svecova 2004; Buell and

Saadatmanesh 2005; Gomez and Svecova 2008). FRPs are noncor- roding, have a low modulus of elasticity and high strength, and a strength-to-weight ratio; therefore, they are an excellent choice for rehabilitation projects. GFRP strengthening is known to increase the strength of timber; however, this research will focus on the use of passive GFRP laminates to increase the stiffness of existing timber stringers.

Previous Research

Fueled by the growing demand for aesthetically pleasing wood products and depletion of forest resources, a considerable amount of research has been undertaken on GFRP-reinforced tim- ber beams in the last few decades. The use of GFRP-reinforced timber beams has increased the structural strength and efficiency of such beams, allowing them to compete against the other main- stream materials such as steel and concrete.

Kuilen (1991) tested 43 beams in four-point bending, in which the height of the beams varied between 94 and 104 mm with widths of 100 mm and span lengths of 4,000 mm. Ten beams were tested as control specimens without reinforcement. The rest were rein- forced with GFRP sheets of 4-mm and 8-mm thickness on either the tension or compression side, or on both sides. The findings showed that the strength and the stiffness increased by 24 and 17%, respectively, in beams that were reinforced with 4-mm sheets, by 49 and 27% when reinforced with 8-mm sheets, by 58 and 32% with double 4-mm sheets, and by 84 and 54% with double 8-mm sheets. No significant difference was recorded in the stiffness increase between single 8-mm or double 4-mm reinforcement.

1Ph.D. Candidate, Dept. of Civil Engineering, Univ. of Manitoba, Winnipeg, Manitoba, Canada R3T 5V6.

2Professor, Dept. of Civil Engineering, Univ. of Manitoba, Winnipeg, Manitoba, Canada R3T 5V6 (corresponding author). E-mail: svecovad@cc .umanitoba.ca

Note. This manuscript was submitted on February 3, 2011; approved on October 14, 2011; published online on May 15, 2012. Discussion period open until November 1, 2012; separate discussions must be submitted for individual papers. This paper is part of the Journal of Composites for Con- struction, Vol. 16, No. 3, June 1, 2012. ©ASCE, ISSN 1090-0268/2012/3- 245–252/$25.00.

JOURNAL OF COMPOSITES FOR CONSTRUCTION © ASCE / MAY/JUNE 2012 / 245

J. Compos. Constr., 2012, 16(3): 245-252

D ow

nl oa

de d

fr om

a sc

el ib

ra ry

.o rg

b y

U ni

ve rs

it y

of H

ou st

on o

n 04

/2 6/

19 . C

op yr

ig ht

A S

C E

. F or

p er

so na

l us

e on

ly ;

al l

ri gh

ts r

es er

ve d.

http://dx.doi.org/10.1061/(ASCE)CC.1943-5614.0000261
http://dx.doi.org/10.1061/(ASCE)CC.1943-5614.0000261
http://dx.doi.org/10.1061/(ASCE)CC.1943-5614.0000261
http://dx.doi.org/10.1061/(ASCE)CC.1943-5614.0000261
Gentile et al. (2002) investigated a total of 22 half-scale beam specimens that had been cut from creosote-treated timber bridge stringers. The specimen dimensions were 100- mm wide × 300-mm deep and 4;300-mm long. Seven out of 22 were control beams with no reinforcement, and the other 15 were reinforced at a ratio between 0.27 and 0.82% using GFRP bars. The beams were tested in four-point bending until failure. A total of four full-scale timber stringers 200- mm wide × 600- mm deep × 10;000- mm long were also reinforced with GFRP bars and tested in four-point bending. The experimental results showed a strength increase of 30% and a stiffness increase for the full-scale beams of 5 to 7%.

Buell and Saadatmanesh (2005) applied carbon FRP (CFRP) in the form of fabric wrap and laminate strips to 10 creosote- treated cut Douglas Fir beams. The beams were 203-mm wide× 483-mm deep × 9;100-mm long. Seven beams were reinforced using carbon fiber and the other three were tested as control spec- imens. The first beam was reinforced by wrapping it with a piece of carbon fabric that covered the tension face of the beam, the two wide faces, and approximately 102 mm of the compression face. The second beam was wrapped perpendicular to the longitudinal axis of the beam using seven strips of carbon fabric that wrapped around the beam starting at mid span. Each strip of fabric was over- lapped approximately 254 mm onto the adjacent strip. The third beam was reinforced using two strips of carbon laminate that bonded to the bottom face of the beam. The fourth beam was re- inforced with two strips of carbon laminate on the bottom face of the beam and then wrapped with a piece of carbon fabric. Finally, the last beam was reinforced using four pieces of wood that were attached to the bottom face of the beam and two strips of carbon laminate that were bonded to the bottom of the pieces of the wood. Bending strength, shear strength, and stiffness were calculated; the results showed that applying carbon fabric to the timber beams increased the stiffness by 7 to 27%, the bending strength by 40 to 53%, and the shear strength by 36 to 68%.

Akbiyik et al. (2007) repaired six salvaged sawn timber beams with horizontal splits. The beams were 200-mm wide × 400- mm deep ×4:6- m long. Lag screw bolts, FRP plates, and plywood were used to increase the shear strength of the beams. One beam was repaired by installing hex bolts vertically through holes from the top to the bottom at the center of the beam. Another two beams were repaired by installing lag screw bolts in the same way as the hex bolts. The other three beams were repaired using plywood and FRP plates on the sides of the beams. All six beams were tested in four-point bending to failure. The results showed that the residual load capacity increased an average of 62%. They also showed that the stiffness of the unsplit beams was larger than the stiffness of the repaired beams.

Yang et al. (2008) tested 27 beams with dimensions of 100 × 50 × 1;800 mm in three-point bending. A total of 6 out of 27 were control beams with no reinforcement, and 21 were reinforced using CFRP and GFRP sheets. Their findings showed that by using FRP-reinforcement ratios from 0.37 to 1.13%, the ultimate load of FRP-strengthened timber beams increased from 17.7 to 77.3% compared with the nonstrengthened control beams.

Silva-Henriquez et al. (2010) tested 45 glulam beams in four-point binding to failure. The beams were 130-mm wide× 305-mm deep × 6:7-m long. A total of 15 beams were prestressed with GFRP laminate (121-mm wide × 3-mm thick) on the tension side. Another 15 beams were reinforced with GFRP laminate on the tension side. The remaining 15 beams were control beams without any reinforcement. The results showed that the strength of the prestressed glulam beams increased by 38% compared with the reinforced GFRP glulam beam and by 95% compared with the

control beams. Also, the stiffness for both the prestressed and reinforced GFRP beams increased by 8% compared with the control beams.

Experimental Program

Materials

A total of 20 creosote-treated Douglas Fir (Pseudotsuga menziesii) beams (130 × 330 × 4;500 mm) were graded using specifications for stringers rough or surfaced 127 mm (5 in.) or thicker with width more than 55 mm (2 in.) greater than thickness from the National Lumber Grades Authority (NLGA 2003). On the basis of the visual classification conducted on the beams, one was graded as “Stan- dard,” three were graded as “No. 1,” 11 were graded as “No. 2,” and the rest were graded as “Utility.” All of the beams were strengthened using rectangular 5 × 50 mm GFRP laminates manu- factured by Rotafix (Rotafix 2011) in the United Kingdom. The tensile modulus and the ultimate strength of the laminates were recorded as 27 GPa and 650 MPa, respectively, on the basis of a typical data sheet of mechanical properties provided by the manufacturer.

Specimen Preparation

All of timber beams for experimental testing were prepared at the W. R. McQuade Structures Laboratory at the University of Manitoba by a local contractor. Two grooves were routed along the bottom tension side of the beam for all 20 beams and cleaned with a brush and vacuum to ensure a satisfactory bond between the grooved timber surfaces (wood substrate) and installed GFRP laminate. An additional groove was routed in the compression zone for 10 out of the 20 beams. Half of the beams were rein- forced with two GFRP laminates 5 mm × 50 mm in the tension zone only (Group T), and the other half was reinforced with GFRP laminate on both the tension and compression sides (Group TC). To avoid weakening the cross-section, the grooves at the top of the beam were not located at the same distance from the top of the beam. Cross-sections of the two specimen groups are shown in Fig. 1.

GFRP laminate with cross-section (5 mm × 25 mm) was used in the compression zone. An epoxy Fibreglass Evercoat (FIB 622) was used to bond the laminates to the timber. The epoxy resin was squeezed into the grooves and the GFRP laminates were then in- serted into the epoxy. Following this, the epoxy was leveled out with a trowel until it was flush with the beam surface.

Fig. 1. Beam cross section

246 / JOURNAL OF COMPOSITES FOR CONSTRUCTION © ASCE / MAY/JUNE 2012

J. Compos. Constr., 2012, 16(3): 245-252

D ow

nl oa

de d

fr om

a sc

el ib

ra ry

.o rg

b y

U ni

ve rs

it y

of H

ou st

on o

n 04

/2 6/

19 . C

op yr

ig ht

A S

C E

. F or

p er

so na

l us

e on

ly ;

al l

ri gh

ts r

es er

ve d.

Test Setup

All of the beams were tested in accordance with ASTM Standard D198 [ASTM (1999)]. The beams were simply supported on rollers with a span of 4,500 mm and tested in three-point bending. A static load was applied by a servohydraulic testing machine with a dis- placement rate of 3 mm∕ min to achieve failure of the beams within 10 to 20 min. A 500-mm long bearing plate was used to distribute the applied load. Lateral support was provided at either ends of the beam. A photograph of the test setup is shown in Fig. 2.

Beam Instrumentation

The beams were instrumented with four linear variable differential transformers (LVDTs) to measure the vertical displacement at the mid span and quarter spans. Four 200-mm pi-gauges were used to measure the strains in the beam at mid span at different heights. The four gauges were placed symmetrically, 65 and 115 mm, from the mid height of the beam. All readings from the LVDTs, pi-gauges, machine load, and the stroke were recorded using a data acquisition system. A schematic drawing of the beam instrumentation is shown in Fig. 3.

Experimental Results

This paper will present the experimental results in terms of the ul- timate strength of the beams and their stiffness. No control beams were tested; however, initial apparent stiffness of each beam was obtained before strengthening. As a result, strength increase will be on the basis of strength predicted using the apparent stiffness of

individual beams before strengthening. The model suggested by Buchanan (1990) was used to find the strength of the beams on the basis of the experimental apparent stiffness measured before strengthening was applied.

Moisture Content

The moisture content of each beam was measured before and after testing to ensure an accurate reading. Measurements were taken using the J-2000 moisture meter (manufactured by Delmhorst Instruments), which uses the principle of electrical resistance. Two pins capable of conducting electricity were driven into the beam parallel to the grain. The degree of conductivity depends on the moisture content of the wood. The meter allows the user to set the species of wood and the temperature of the room and to accurately read values between 6 and 40%. Three moisture- content readings were taken for each beam at different locations. The values were then averaged to obtain an accurate reading of the moisture content. The moisture content in the tested beams ranged from 12 to 16%.

The average measured moisture content before strengthening for Group T and for Group TC was 16 and 14%, respectively, whereas the readings were found to be 15 and 14% after strengthening for Group T and Group TC, respectively. These are acceptable limits for treated timber. The moisture content decreased but did not change dramatically after cutting the slots in the beams because they were filled with epoxy resin and FRP shortly after the slots were cut. Prolonged exposure to the environment may cause more pronounced changes.

Beam Stiffness

All of the beams were tested in bending up to a load of 45 kN be- fore strengthening so that the effect of strengthening on bending stiffness could be evaluated for each beam. The apparent stiffness was calculated from the initial linear-elastic portion of the load- deflection curve of each beam using Eq. (1)

EI ¼ ΔPL 3

48Δδ ð1Þ

where △P = given range of applied load; L = span length; △δ = given range of deflection for load range △P; E = modulus of elasticity; and I = moment of inertia.

Tables 1 and 2 present the stiffness of all beams before and after strengthening for Groups T and TC, respectively. The experimental results obtained for beams TC1, TC4, and TC7 show that the stiff- ness increased by 9.45, 9.41, and 9.11%, respectively. These values were relatively high compared with the values of the other beams in

Fig. 2. Timber beam test setup

Fig. 3. Typical schematic drawing of beam instrumentation

JOURNAL OF COMPOSITES FOR CONSTRUCTION © ASCE / MAY/JUNE 2012 / 247

J. Compos. Constr., 2012, 16(3): 245-252

D ow

nl oa

de d

fr om

a sc

el ib

ra ry

.o rg

b y

U ni

ve rs

it y

of H

ou st

on o

n 04

/2 6/

19 . C

op yr

ig ht

A S

C E

. F or

p er

so na

l us

e on

ly ;

al l

ri gh

ts r

es er

ve d.

Group TC, which ranged from �0:36 to 3.33%. The justification for that divergence could be the high variability in the individual timber beams. The higher increase in stiffness in beams TC1, TC4, and TC7 was primarily the result of their visible defects such as splits and checks in the original beams. Beams of higher quality will experience negligible changes in stiffness.

Strengthening using GFRP laminates has increased the stiffness of the beams by 3% when used only on the tension side (Group T) and by 3.5% when used on both the tension and compression sides (Group TC). Hence, the compression reinforcement had minimal effect on stiffness increase.

Modulus of Rupture

The modulus of rupture (MOR) was calculated from the recorded ultimate loads for the reinforced beams to compare and verify the results obtained from the experiment and from the analysis as shown in Tables 3 and 4. Table 3 shows the MOR for the beams in Group T, whereas Table 4 shows MOR for beams in Group TC. Only beams that failed in flexure were considered when calculating the MOR using Eq. (2)

MOR ¼ MF Sg

ð2Þ

Table 1. Experimental Results for Group T

Beam NLGA grade Stiffness of unstrengthened

beam (N:mm2 × 1012) Stiffness of strengthened beam (N:mm2 × 1012) % Change

Failure load (kN)

Test duration (min)

Failure mode

T1 No. 1 4.75 4.89 3.08 143.86 21 Flexure

T2 Utility 6.02 6.15 2.15 147.46 17 Shear

T3 Utility 4.53 4.58 1.10 117.42 23 Shear

T4 No. 2 4.18 4.37 4.54 101.58 16 Flexure

T5 No. 2 5.30 5.54 4.56 100.59 12 Flexure

T6 No. 2 4.47 4.58 2.30 100.48 18 Flexure

T7 No. 2 4.08 4.23 3.61 125.70 21 Flexure and shear

T8 Standard 4.06 4.29 5.33 100.99 18 Flexure

T9 No. 2 5.47 5.62 2.88 139.34 18 Shear

T10 No. 2 4.54 4.55 0.20 114.70 19 Flexure

Table 2. Experimental Results for Group TC

Beam NLGA grade Stiffness of unstrengthened

beam (N:mm2 × 1012) Stiffness of strengthened beam (N:mm2 × 1012) % Change

Failure load (kN)

Test duration (min.)

Failure mode

TC1 No. 1 5.61 6.15 9.45 159.60 18 Shear

TC2 No. 2 3.94 3.93 �0:36 62.56 11 Flexure and shear TC3 No. 2 5.17 5.26 1.66 158.38 23 Shear

TC4 No. 2 4.36 4.77 9.41 78.70 11 Shear

TC5 No. 2 4.45 4.47 0.41 145.44 15 Shear

TC6 No. 2 5.35 5.37 0.19 148.30 18 Flexure and shear

TC7 Utility 4.21 4.60 9.11 83.58 14 Flexure and Shear

TC8 Utility 3.94 4.07 3.33 107.96 19 Flexure

TC9 Utility 4.33 4.33 0.01 117.40 15 Shear

TC10 No. 1 5.25 5.35 1.96 122.30 20 Shear

Table 3. Experimental and Analytical Results of MOR for Group T

Beam NLGA grade MORtheocont: (MPa) MOR exp reinf (MPa) Change in MOR (%) MOR

theo reinf (MPa) MOR

exp reinf ∕MOR

theo reinf

1 2 3 4 ð4 � 3Þ∕4 5 4∕5 T1 No. 1 22.79 58.91 48.34 33.07 1.78

T4 No. 2 24.48 43.64 43.91 37.98 1.15

T5 No. 2 47.73 45.08 � 5:54 67.36 0.67 T6 No. 2 27.99 42.91 34.78 40.82 1.05

T7 No. 2 22.51 53.62 58.02 34.43 1.56

T8 Standard 31.02 45.81 32.29 44.93 1.02

T10 No. 2 29.46 49.34 40.30 40.84 1.21

Average 35.97 1.21

Standard deviation 18.85% 0.28

248 / JOURNAL OF COMPOSITES FOR CONSTRUCTION © ASCE / MAY/JUNE 2012

J. Compos. Constr., 2012, 16(3): 245-252

D ow

nl oa

de d

fr om

a sc

el ib

ra ry

.o rg

b y

U ni

ve rs

it y

of H

ou st

on o

n 04

/2 6/

19 . C

op yr

ig ht

A S

C E

. F or

p er

so na

l us

e on

ly ;

al l

ri gh

ts r

es er

ve d.

where MF = bending moment at the failure location; and Sg = section modulus based on gross section properties.

Load-Deflection Behavior

The average failure load for Group T and Group TC was found to be 119.2 kN and 118.4 kN, respectively. Figs. 4 and 5 show the load-deflection behavior of the beams in the two groups. Inspec- tion clearly showed that the beams in Group T and Group TC behaved in a linear-elastic fashion until failure. Beams in Group TC were also, on average, of poorer quality compared with those in Group T. Therefore, the strength increase in the two groups is very similar. The increased nonlinearity in behavior of beams TC6 and TC3 is beneficial as it serves as a warning before im- pending failure.

Most of the beams failed in the 10-to 20-min range with the exception of four that failed after a maximum of 21 to 23 min. Estimating the test duration for strengthened beams is difficult, and the rate of loading in this test has been kept the same as in all of the authors’ previous research. The experimental results show that six timber beams failed in shear and three beams failed in a combination of flexure and shear from Group TC compared with three shear failures and one flexure-shear failure in Group T.

Strength and Ultimate Load

The strength of the unreinforced beams was calculated using the following formula developed by Buchanan (1990):

f m ¼ � k3 þ 1

c

� 1∕k3

f tu ð3Þ

where f m = bending strength; k3 = stress distribution parameter; c = ratio of the depth of the neutral axis to the overall depth; and f tu = axial tensile strength.

The strength of the reinforced beams was predicted using the following formula, which was developed for beams reinforced for flexure only (Gentile et al. 2002):

f m ¼ α � k3 þ 1

c

� 1∕k3

f tu ð4Þ

where α = adjustment factor to account for FRP reinforcement. A value for α equal to 1.3 is suggested for beams with flexural

reinforcement only. The values predicted using Eq. (3) compared with the experimental values are shown in Tables 5 and 6. For beams in Group T, the observed average strength compared with the predicted value from Eq. (3) increased on the basis of beam grade by 32%, 48%, and 34% for Standard, No. 1, and No. 2, respectively. For beams in Group TC, the average strength increase was 28% for grade No. 2 and 34% for Utility grade. On average, the strength in both groups increased by 36 and 31% for Groups T and TC, respectively. Beams T1 and T7 had an increase in strength of 48.3 and 58.0%, respectively. Because of the large variability in timber strength, it is common to get higher strength increases in beams of lower grade, which was the case for these two beams.

Fig. 6 shows the relationship between the modulus of rupture (MOR), the modulus of elasticity (MOE) for unstrengthened,

0 10 20 30 40 50 60 70 0

20

40

60

80

100

120

140

160

Deflection mm

L o

a d

k N

T1 T2 T3 T4 T5 T6 T7 T8 T9 T10

T1T2

T3

T4

T5

T9

T10

T7

T8 T6

Fig. 4. Load deflection for beams in Group T

0 10 20 30 40 50 60 70 0

20

40

60

80

100

120

140

160

Deflection [mm]

L oa

d [

k N

]

TC1

TC2

TC3

TC4

TC5

TC6

TC7

TC8

TC9

TC10

TC2

TC1

TC6

TC3

TC5

TC8

TC9

TC10

TC7

TC4

Fig. 5. Load deflection for beams in Group TC

Table 4. Experimental and Analytical Results of MOR for Group TC

Beam NLGA grade MORtheocont: (MPa) MOR exp reinf (MPa) Change in MOR (%) MOR

theo reinf (MPa) MOR

exp reinf ∕MOR

theo reinf

1 2 3 4 ð4 � 3Þ∕4 5 4∕5 TC2 No. 2 20.44 26.37 22.48 29.40 0.90

TC6 No. 2 41.95 62.51 32.89 56.50 1.11

TC7 Utility 27.35 35.15 22.19 42.86 0.82

TC8 Utility 24.77 46.44 46.67 37.13 1.25

Average 31.06 1.02

Standard deviation 9.99 0.17

JOURNAL OF COMPOSITES FOR CONSTRUCTION © ASCE / MAY/JUNE 2012 / 249

J. Compos. Constr., 2012, 16(3): 245-252

D ow

nl oa

de d

fr om

a sc

el ib

ra ry

.o rg

b y

U ni

ve rs

it y

of H

ou st

on o

n 04

/2 6/

19 . C

op yr

ig ht

A S

C E

. F or

p er

so na

l us

e on

ly ;

al l

ri gh

ts r

es er

ve d.

and the modulus of elasticity for reinforced timber beams. The MOEs were obtained from the experimental load-deflection curves. The dashed line represents the trend for unreinforced beams and shows that MOR is greatly affected by the MOE. Beams with a smaller value for the MOE have a smaller MOR as well. Therefore, it is apparent that strengthening makes the behavior of timber beams more uniform. This conclusion is reiterated in Fig. 7. The results obtained from Fig. 7 indicate that the stronger beams do not benefit from strengthening to the same degree shown in the weaker beams. For example, beams with a strength of 20 MPa can be strengthened up to 40%, whereas beams with a strength of 40 MPa may increase their strength by only 15%.

Database Analysis

A database of FRP-strengthened timber beams was assembled to provide additional information about factors that affect the increase

in stiffness. The beams tested in this research are combined with results obtained from earlier published research by Gentile et al. (2002), Svecova and Eden (2004), Amy and Svecova (2004), Buell and Saadatmanesh (2005), Walker (2006), and Gomez and Svecova (2008). The total number of beams in the database is 167, of which 35 are control beams and 132 are reinforced for flexure or shear. The beams were tested either in three-point bending or four-point bending with a beam length ranging from 1.8 to 10 m as shown in Table 7. A total of 132 beams from the current database were analyzed for stiffness (Alhayek 2009).

Analysis of the database revealed that, on average, the stiff- ness of strengthened timber increased by 7.8%, with a standard deviation of 9.6% and a coefficient of variance of 1.23. The largest increase in stiffness (31.8%) was found in the work by Gomez and Svecova (2008), in which nine beams were strengthened for both flexure and shear and the original stringers were mostly Utility grade, virtually separated into two portions before strengthening was applied. Once again this shows that strengthening is very beneficial for low-grade timber; therefore, even utility stringers

Fig. 6. Relationship between MOE and MOR of unstrengthened and strengthened samples

Fig. 7. Efficacy of strengthening

Table 5. Experimental and Analytical Results of Ultimate Loads for Group T

Beam NLGA grade Ptheoult (control) (kN) P exp ult (reinf.) (kN) P

theo ult (reinf.) (kN) Strength increase (%) P

theo ult ðreinf:Þ∕Pexpult ðreinf:Þ

1 2 3 4 5 ð4 � 3Þ∕4 5∕4 T1 No. 1 74.31 143.86 80.74 48.34 0.56

T4 No. 2 56.97 101.58 88.39 43.91 0.87

T5 No. 2 106.49 100.59 150.29 � 5:87 1.49 T6 No. 2 65.54 100.48 95.59 34.78 0.95

T7 No. 2 52.76 125.70 80.71 58.02 0.64

T8 Standard 68.38 100.99 99.05 32.29 0.98

T10 No. 2 68.48 114.70 94.95 40.30 0.83

Average 35.97 0.90

Standard deviation 18.85 0.28

Table 6. Experimental and Analytical Results of Ultimate Loads for Group TC

Beam NLGA grade Ptheoult (cont.) (kN) P exp ult (reinf.) (kN) P

theo ult (reinf.) (kN) Strength increase (%) P

theo ult ðreinf:Þ∕Pexpult ðreinf:Þ

1 2 3 4 5 ð4 � 3Þ∕4 5∕4 TC2 No. 2 48.50 62.56 69.75 22.48 1.11

TC6 No. 2 99.52 148.3 134.03 32.89 0.90

TC7 Utility 65.03 83.58 101.90 22.19 1.22

TC8 Utility 57.58 107.96 86.33 46.67 0.80

Average 31.06 1.02

Standard deviation 9.99 0.17

250 / JOURNAL OF COMPOSITES FOR CONSTRUCTION © ASCE / MAY/JUNE 2012

J. Compos. Constr., 2012, 16(3): 245-252

D ow

nl oa

de d

fr om

a sc

el ib

ra ry

.o rg

b y

U ni

ve rs

it y

of H

ou st

on o

n 04

/2 6/

19 . C

op yr

ig ht

A S

C E

. F or

p er

so na

l us

e on

ly ;

al l

ri gh

ts r

es er

ve d.

removed from service can be strengthened and used for emergency repairs on existing timber bridges.

The next highest increase in stiffness (17.5%) was in the work of Buell and Saadatmanesh (2005), which used CFRP. Using reinforcement with a higher modulus of elasticity proved to be beneficial for increasing stiffness. However, the reinforcement ratios used by Buell and Saadatmanesh (2005) were, in some cases, 8 to 9 times larger than the minimum values required by the Canadian Standards Association (CSA 2006), and those used in other research.

Fig. 8 shows that the stiffness increased only for beams with a smaller span-to-depth ratio (6 to 8.5). Beams with a large span- to-depth ratio usually did not achieve stiffness increase, with the exception of the beams from Buell and Saadatmanesh (2005) dis- cussed previously. Placing the minimum amount of reinforcement as suggested in CSA (2006) will have virtually no effect on the overall stiffness of the system.

The analysis of the database showed that the modulus of elas- ticity (MOE) increases as long as there is an increase in the span length of timber beams reinforced with FRP, as shown in Fig. 9. The trend of increasing stiffness with beam span is evident. This effect was recognized in the earlier work of Madsen (1992) and will be further studied.

The Canadian Highway Bridge Design Code CSA (2006) presents numerical values for the MOR of timber beams reinforced with GFRP. For timber beams graded Standard, No. 1, and No. 2, the MOR values are 20.5, 19.0, and 13.5 MPa, respectively. The experimental MOR for both Groups T and TC graded Standard, No. 1, No. 2, and Utility were 32.9, 58.9, 46.2, and 40.8 MPa,

respectively. Fig. 10 shows the normal distribution of the MOR for strengthened beams with GFRP and unstrengthened beams from the database. The mean and the 5th-percentile values for the MOR for strengthened beams obtained in the database were 42.82 and 23.52 MPa, whereas for unstrengthened beams were 26.77 and 13.97 MPa, respectively. There is a 15, 24, and 74% increase of the 5th percentile for the reinforced beams compared with the CSA (2006) design value for timber beams graded Standard, No. 1, and No. 2, respectively.