Fir and chestnut timber beams reinforced with GFRP pultruded elements

This article was originally published in a journal published byElsevier, and the attached copy is provided by Elsevier for the

author’s benefit and for the benefit of the author’s institution, fornon-commercial research and educational use including without

limitation use in instruction at your institution, sending it to specificcolleagues that you know, and providing a copy to your institution’s

administrator.

All other uses, reproduction and distribution, including withoutlimitation commercial reprints, selling or licensing copies or access,

or posting on open internet sites, your personal or institution’swebsite or repository, are prohibited. For exceptions, permission

may be sought for such use through Elsevier’s permissions site at:

http://www.elsevier.com/locate/permissionusematerial

Autho

r's

pers

onal

co

py

Fir and chestnut timber beams reinforcedwith GFRP pultruded elements

Marco Corradi *, Antonio Borri

Civil and Environmental Engineering Department, School of Engineering, University of Perugia, Via Duranti 93, 06125 Perugia, Italy

Received 27 February 2006; received in revised form 26 June 2006; accepted 16 July 2006Available online 4 October 2006

Abstract

High-performance fibers are being widely researched for repair and rehabilitation in civil engineering structures. The potential ben-efits, liabilities, and architectural considerations regarding the use of high-performance fibers for reinforcing wood beams are discussed.An experimental program based on a four- and three-point bending test configuration is proposed to characterize the stiffness andstrength response of wood beams reinforced with pultruded GFRP (glass fibers reinforced polymers) elements. Improving wood mechan-ical characteristics through the use of fiber reinforced polymers often involves the use of adhesives, generally epoxy resins. For this reasonmechanical, calorimetric and thermo-gravimetric analyses were performed on the resin utilized and bonding effectiveness was studied.Mechanical tests carried out on full-scale wood beams showed that the reinforcement with GFRP beams may produce strong increasesin flexural stiffness and capacity. In addition, an analytical investigation based on a simple linear analysis was conducted to predict ulti-mate load. At the end of this paper results of the experimental program are presented and used for comparison with the analyticalprocedure.� 2006 Elsevier Ltd. All rights reserved.

Keywords: D. Mechanical testing; C. Numerical analysis; B. Elasticity

1. Introduction

Wood is a construction material denoting high tensilestrength. For this reason it was widely used in the pastin many structural situations in which elements weresubjected to flexion or tension loads. Wood is also a veryefficient material. Its notable resistance under both com-pressive and tension loads must be considered as beingnearly unique when compared with its limited weight den-sity. In civil structures existing wood elements such asbeams have usually been subjected either to replacementor reinforcement using classic techniques involving theuse of common building materials, such as concrete orsteel. It should also be noted that wood is characterizedby long-term durability, as long as it undergoes correctmaintenance.

With regard to new wood elements reinforcing wood ele-ments through the use of other materials (steel, reinforcedconcrete, additional wood) is certainly not a new tech-nique. As far as reinforcing with steel or aluminium werefer to works carried out by Sliker [1], Borgin et al. [2],and Hoyle [3] proposed inserting thin plates of steel or alu-minium both vertically and horizontally between the lamel-lae of wood. Other experimental work and studies werecarried out by Bulleit et al. [4] on glulam elements rein-forced by the insertion of steel bars. Significant experimen-tal work on reinforcing wood beams was carried out byPeterson [5] using epoxy adhesives to glue stressed steelplates to the area under tension in order to pre-stress thewood.

Subsequent experiments, carried out to replace metalplates in wood beams, involved the use of high-resistantsteel wire embedded in an epoxy matrix (Krueger [6],Krueger and Eddy [7], Krueger and Sandberg [8], Kobetzand Krueger [9]).

1359-8368/$ - see front matter � 2006 Elsevier Ltd. All rights reserved.

doi:10.1016/j.compositesb.2006.07.003

* Corresponding author. Tel.: +39 075 585 3906; fax: +39 075 585 3897.E-mail address: [email protected] (M. Corradi).

www.elsevier.com/locate/compositesb

Composites: Part B 38 (2007) 172–181

Autho

r's

pers

onal

co

py

All of the above reinforcement techniques refer to inter-ventions on new wood and in general the reinforcementwas to be carried out previous to on-site mounting of thewood elements. Through the years proposals for variousmethods of reinforcing pre-existing wood beams employingtraditional materials, both with and without metal alloys,have been made by Tampone [10].

It is clear that all the proposals involving the use ofmetal plates and bars can be adapted to the world of com-posites, substituting the above-mentioned with sheets andpultruded FRP elements, with some advantages (lighterweight, less invasiveness) as well as some disadvantages(costs, problems of joining and connections). Ad example,the use of aluminium profiles does not permit the achieve-ment of these objectives due to lower in-site machinability,higher costs and weight density compared to compositematerials. The use of FRP materials in reinforcing woodstructures is not as of yet a widespread technique and, tosome degree, adequate experimentation is needed prior toits use on a wide scale. This is particularly true for ‘antique’wood elements, that is of existing structures, since there isalready ample, well-consolidated experience in the field ofnew wood structures.

The use of composite materials in wood reinforcementwas first proposed in the 1980s by Meier [11], Triantafillou[12], Triantafillou and Plevris [13], Kropf and Meierhofer[14], Gentile et al. [15], Borri et al. [16], who applied com-posites based on glass (GFRP) or carbon fiber sheets(CFRP) epoxy-bonded externally on the tension zonesand studied their effect on the mechanical characteristicsof the reinforced wood elements. Corradi et al. [17] usedcomposite materials (GFRP sheets) for in-plane reinforce-ment of existing wood beam floors. These studies havedemonstrated that unidirectional composite sheets are idealin tension reinforcement of wood beams, but are not ableto provide a notable increase in flexural stiffnesscharacteristics.

When a reinforcement work is projected there are manycritical issues including ensuring a durable bond betweenthe FRP and wood given the shrinkage and swelling ofwood due to moisture changes. Glass fibers are economi-cally competitive and are characterized by high mechanicalcharacteristics. However glass transition temperatures forepoxy resins are generally not elevated and equal to 30–80 �C. Recently Tascioglu et al. [18,19] found that E-glassfiber/phenolic resin matrix pultruded composite materialsdesigned for wood reinforcement are susceptible to fungalpenetration by common wood decay fungi highlightingthe risk of strength decrease and moisture increase.

In the interest of increasing stiffness characteristics ofexisting wood beams, in this paper we examine the effectsof the use of pultruded elements placed in the compressionzone on the failure mechanisms, stiffness characteristicsand ductility of the reinforced elements as they are sub-jected to bending loads. An experimental program basedon a four- and three-point bending test configuration isproposed to characterize the stiffness and strength response

of wood beams reinforced with pultruded GFRP elements.These elements do not function as a substitute for the woodbeams but rather effect an increase in their capacitythrough the creation of a mixed wood–composite structure.Mechanical, calorimetric and thermo-gravimetric analyseswere performed on the resin utilized and bonding effective-ness was studied. In addition, a numerical investigationwas conducted to predict ultimate load and to establish amethodology for the optimum design of the hybrid FRP-wood system. Finally the reinforcing scheme was appliedto an historical building (Palazzo Collicola) located in Spo-leto, Italy struck by the Umbrian earthquakes of 1997–98in order to validate its practicality.

2. Experimental method

The experimentation involved an initial mechanical char-acterization of both the wood material and the composite.In particular, a series of experiments were carried out on13 wood beams for the purpose of analysing improvementsin characteristics of stiffness and flexural strength resultingfrom reinforcement by means of pultruded compositeelements. The wood beams, seven of which were of whitefir (identified as A1, A2, A3, A4, A5, A6 and A7) and sixof chestnut wood (C1, C2, C3, C4, C5 and C6), had square,unbevelled cross-sections, 180 · 180 mm and were each3050 mm in length. Moreover, according to the UNI ISO3130 [20], the percentage of humidity measured was11.39%. The density of the fir beams was 452 kg/m3, whilefor those of chestnut it was 800 kg/m3.

Reinforcement of the beams was carried out using nineglass fiber-based pultruded elements (GFRP) 2500 mm inlength: five gray were supplied by Creative Pultrusion(identified as H1, H2, H3, H4 and H5) (type H), while fourwhite were supplied by Atp Pultrusion (identified as I1, I2,I3 and I4) (type I). The I-type elements were obtained bymeans of a coupling of two white C-shaped elements, gluedtogether by means of an epoxy system (Fig. 1). The densityof the pultruded elements was 1810 kg/m3 for those pro-duced by Creative Pultrusion (type H) and 1770 kg/m3

for those of Apt Pultrusion (type I). The pultruded ele-ments were reduced due to execution of notches of theGFRP flange. In fact the common floor joists between

Fig. 1. Cross-section of pultruded elements: (a) type H, (b) type I(dimensions in mm).

M. Corradi, A. Borri / Composites: Part B 38 (2007) 172–181 173

Autho

r's

pers

onal

co

py

wood beams and rafters, for the most part of existingancient floors, necessitate the notching of the GFRP flange.

2.1. Mechanical characterization of the wood beams

In order to find the Young’s modulus of wood EW andflexural stiffness kW a series of elasticity tests were executedon all 13 wood beams: four-point bending tests were carriedout by effecting six load–unload cycles using loads in therange between 10 kN and 30 kN, obtaining for the elasticmodulus the results of 11426 MPa for fir beams and10251 MPa for chestnut beams. The span of each test was3000 mm, while the load span was equal to 1000 mm. Thebeams were loaded with a 245 kN MTS actuator and aspreader beam. The spreader beam, centered about themid-span, created a 1000 mm zone with a constant momentand zero shear. The test was then carried out with a con-stant load rate, so that initially the first load from 0 kN to30 kN was reached in 45 s. The load–unload cycle between10 kN and 30 kN was then repeated another five times,always with the same rate and therefore in cycles of 60 s.

A second series of tests (failure tests) were carried out onfour wood beams. The load was applied up to failure with aconstant rate of 0.2 kN/s and the beam response to theapplied load was measured by means of two inductivetransducers (LVDT). The two LVDTs were positioned toeffect the reading near the neutral axis, midway on the spanon which the test was conducted (span equal to 3 m); thetwo LVDTs were placed on opposite sides of the woodbeam to record eventual rotations of the beam during thetest. The four-point bending configuration was used forthese tests as well. Data acquisition was carried out witha 2 points/s sampling. It was then possible to evaluatethe apparent stiffness (in particular the stiffness value wascalculated for 1/3 of the maximum load and for the maxi-mum load Pmax).

k1=3 ¼P 1=3 � P i

d1=3

ð1Þ

ku ¼P max � P i

dmax

ð2Þ

where Pi is the preload assumed equal to 3 kN, P1/3 is1/3 Pmax and dmax and d1/3 are the midspan deflections atmaximum load and at 1/3 Pmax.

2.2. Mechanical characterization of the pultruded elements

The mechanical characteristics of the pultruded ele-ments were measured by recoding the load–deflection

behavior and by compressing prismatic samples in anInstron testing machine. The references for the mechanicaltests on the pultruded elements were EN 13706-2 (parts Dand G) [21] and ASTM D 695 M [22] standards. In partic-ular the following were carried out:

1. Determination of compression strength: 12 compressiontests on 12 prismatic samples with constant rectangularcross-sections, a height of 40 mm, a length of 28 mm anda thickness equal to that of the flange of the pultrudedelement (12 mm in the case of the I type and 7 mm inthe case of the H type) from which the sample wasextracted;

2. Determination of the Young’s modulus, EP, and shearmodulus, GP: three-point bending tests on three spans(2.5 m, 2.0 m and 1.5 m) were executed. In accordancewith the above-mentioned standards, these tests werecarried out using six load–unload cycles under con-trolled mid-span vertical displacements in 55 s, with dis-placements, respectively, between 1 mm and 12 mm,1 mm and 10 mm, 1 mm and 7.5 mm.

3. Determination of the decrement in flexural stiffness kP

following execution of notches in pultruded elements:four-point bending tests on a span of 2.5 m under sixload and unload cycles were carried out on pultrudedelements before and after execution of 80 · 30 mmnotches at 50 cm intervals on the pultruded elementflange.

Table 1 shows the results for compression strength andthe Young and shear moduli. Both I and H type elementsare characterized by similar Young modulus values (respec-tively, 25407 MPa and 27031 MPa). Compression strengthis higher for the H type and equal to 373.45 MPa.

Before pultruded elements were applied to the woodbeams, notches were made intended, in a simulation of areal case scenario, for the insertion of the wood rafters ofa floor. The notches were executed with particular care toavoid any significant damage to the pultruded element(Figs. 2 and 3). The four-point bending tests highlighteda loss of flexural stiffness in the notched pultruded elements(20.2% for type H pultruded elements and 11.4% for typeI). A synthesis of the results is presented in Table 2. Itcan be noted that type I elements result as much stiffercompared to H type elements. Reinforcement of the woodbeam, by applying a notched pultruded element to theupper part of the beam (compressed area), was carriedout using an epoxy glueing system saturant (produced byMac Spa, Treviso, Italy) and 4.8-type steel screws with adiameter of 10 mm and a length of 80 mm.

Table 1Mechanical characteristics of pultruded elements

Composite element No. Compression strength (MPa) Young modulus EP (MPa) Shear modulus GP (MPa) Density (kg/m3)

H type 373.45 27031 1640 1810I type 206.22 25407 1340 1770

174 M. Corradi, A. Borri / Composites: Part B 38 (2007) 172–181

Autho

r's

pers

onal

co

py

2.3. Characterization of the epoxy system

An epoxy system with room temperature polymerizationwas used to bond the pultruded elements to the woodparts. A series of five tensile strength tests and five com-pression tests were carried out in accordance, respectively,with ASTM D638 [23] and D695 standards to determinethe mechanical characteristics of the epoxy resin. The testson the resin samples highlighted a compression strength of56.54 MPa, a tensile strength of 23.43 MPa and an elasticmodulus of 2112 MPa.

Furthermore, the epoxy system was examined usingcalorimetric, thermo-gravimetric and microscopic analyses.For the former, dynamic analyses were carried out between�30 �C and +200 �C at a heating speed of 10 �C/min todetermine both the polymerization heat, DHD, as well asthe glass transition temperature, Tg. The calorimeter uti-lized was a Setaraam DSC92, capable of working in iso-thermic and dynamic conditions between �100 and+600 �C. Thermo-graphic analyses were carried out usingan Seiko Exstar 6000 TG-DTA, which permits a heatingspeed from 0.5 to 200 �C/min and can function at a maxi-mum temperature of 1200 �C. The samplings were heatedfrom 25 to 6000 �C at a heating speed of 10 �C/min andthe weight-loss curve and its derivative were obtained.

The results of the calorimetric tests showed polymeriza-tion heats DHD and glass transition temperature Tg equalto 350 J/g and 30 �C, respectively. Since the Tg is, respec-tively, close to room temperature and 60 �C, it can behypothesized that the system is stable until 50 �C. Thishypothesis is confirmed by the results of the thermo-gra-phic analyses which show that thermic degradation of thematerial does not initiate until 50 �C. At higher tempera-tures T P Tg the epoxy resin should become rubbery, withnegative effects on the joint performances.

Microscopic analysis (Fig. 4) showed that the penetra-tion of the epoxy resin in the chestnut wood was negligible.This explains the necessity of using a mechanical anchoring

Fig. 2. Pultruded elements after execution of notches.

Fig. 3. Wood beams reinforced with GFRP pultruded elements. Notchespermit the application of reinforcement without interfering with woodrafters.

Table 2Flexural stiffness decrement after execution of notches

Compositeelement No.

Flexural stiffness a

kP (N mm�1)Flexural stiffnessb

kP (N mm�1)Decrement(%)

H type 381 304 20.2I type 1467 1300 11.4

a Before notch execution.b After notch execution. Fig. 4. Microscopic analysis: penetration of the epoxy resin into wood.

M. Corradi, A. Borri / Composites: Part B 38 (2007) 172–181 175

Autho

r's

pers

onal

co

py

system made of screws to connect wood and compositeelements.

2.4. Wood beams strengthened with pultruded elements

A series of tests were carried out, subsequent to rein-forcement with a pultruded GFRP element applied in thecompressed zone, to determine the increase in stiffnessand flexural strength deriving from application of thereinforcement.

Analogously to the bending tests on the un-reinforcedwood beams, the elasticity tests were executed using afour-point bending configuration, carrying out six load–unload cycles between 10 kN and 30 kN. Each pultrudedelement, 2500 m in length, was centered and applied to awood beam 3050 mm long (Fig. 5). The procedure followedin executing the joining was articulated in various phases:(1) careful planing and cleaning of the surfaces to be joined(wood and GFRP elements); (2) positioning and centeringthe pultruded element in GFRP above the wood beam; (3)application of the epoxy resin for the bonding; (4) effectingthe joining by applying screws with washers. A direct com-parison of readings taken on the same wood beam prior toand subsequent to reinforcement highlights significantincreases in flexural stiffness kR following application ofthe pultruded element (Table 3). It can be observed thatthe maximum effect, in terms of a percentage increase inflexural stiffness, was obtained for the beams reinforcedwith I type pultruded elements (+188%) with negligible dif-ference between fir and chestnut beams. Reinforcementwith H type elements caused an average stiffness incre-ments of 83% for fir wood beams and of 58% for chestnut

ones. This is essentially due to the fact that I-type pul-truded element has a greater cross-section area than theH-type pultruded elements used. Fig. 6 shows the load–deflection behavior of wood beam C3 prior to and subse-quent to reinforcement with I1 pultruded element.

The tests up to failure on the reinforced beams were car-ried out according to the same configuration as those onthe un-reinforced wood beams. From the graph in Fig. 7it can be seen how the failure test on all wood beams high-lighted a notable increase, when compared to the un-rein-forced beams, both in terms of the ultimate load Pmax aswell as of stiffness k1/3 and ku and ductility. For all beamsa flexural tensile rupture occurred in the load span near aknot, grain deviation or other defect. Fig. 8 shows a typical

3000

2500

180

1000

No.2 LVDT

Wood beam 180x180x3050

Pultruded element

1000

280(H-type reinforcement)

317(I-type reinforcement)

Metal screws/250 L=80

Fig. 5. Test configuration of reinforced wood beams (not to scale, dimensions in mm).

Table 3Results of elastic tests

Flexural stiffness kW Un-reinforced woodbeams (N mm�1)

Flexural stiffness kP Pultruded elementswith notches (N mm�1)

Flexural stiffness kR Reinforced woodbeams (N mm�1)

kR/kW

A2 1127 H1 305 A2 + H1 2435 2.16A3 2439 H3 291 A3 + H3 3935 1.61A4 2253 H2 303 A4 + H2 3852 1.71A6 2356 I3 1305 A6 + I3 6790 2.88A7 2442 I4 1288 A7 + I4 6956 2.85C2 2153 I2 1310 C2 + I2 6294 2.92C3 2252 I1 1293 C3 + I1 6454 2.87C5 2345 H4 311 C5 + H4 3701 1.58C6 2478 H5 295 C6 + H5 3940 1.59

0

4

8

12

16

20

24

28

32

0 2 4 6 8 10 12 14 16 18Deflexion (midspan) [mm]

Load

[kN

]

C3 C3 + I1

Fig. 6. Elastic tests: behavior of wood beam C3 before and afterreinforcement.

176 M. Corradi, A. Borri / Composites: Part B 38 (2007) 172–181

Autho

r's

pers

onal

co

pytensile rupture evidenced for fir beam A3 reinforced withH3 pultruded element. Failure caused a suddenly drop ofbeam capacity and the subsequent rupture of pultruded ele-ment between the flange and the web. Measurements inthese tests indicated the greatest increase in strength andstiffness for the chestnut beams reinforced with I-type pul-truded elements (Pmax increment 208%, k1/3 increment270%, ku increment 334%) (Table 4). The ultimate capacityof chestnut beams reached 204 kN. Fir beams reinforcedwith I-type elements, except for A2 beam, also demon-strated a very interesting reinforcement action (Pmax incre-ment 143%, k1/3 increment 242%, ku increment 275%).Moreover, the strengthened beams with I-type pultrudedelements demonstrated a behavior almost completely line-arly elastic for all four beams reinforced as a consequenceof the impossibility for the wood to plasticize in the com-pressed area due to the presence of the pultruded element,which has a linearly elastic behavior.

The firwood beam, A2, is characterized by low mechan-ical properties, highlighted both by geometric examination(a great number of knots and splits as well as the presenceof grain deviation) and by the characterization test. For

020406080100120140160180200220240260

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75Deflexion (midspan) [mm]

Load

[kN

]

C3+I1

C1

A7+I4

A3+H3

A1

A4+H2

A5C4

A6+I3 C2+I2

C5+H4

C6+H5

Fig. 7. Tests carried out on un-reinforced wood beams and reinforced.

Fig. 8. Flexural tensile rupture of A3 + H3 beam.

Tab

le4

Res

ult

so

ffl

exio

nte

sts

No

.P

ma

x(k

N)

Incr

emen

tP

ma

x

Rei

nfo

rced

/un

-rei

nfo

rced

k1

/3(N

mm�

1)

Incr

emen

tk

1/3

Rei

nfo

rced

/un

-rei

nfo

rced

ku

(Nm

m�

1)

Incr

emen

tk

u

Rei

nfo

rced

/un

-rei

nfo

rced

C1

65.1

–17

53–

1491

–C

467

.2–

1829

–14

70–

C3

+I1

225.

23.

4075

914.

2470

114.

74C

2+

I218

3.5

2.77

5665

3.16

5837

3.94

C5

+H

413

7.2

2.07

3659

2.04

3102

2.10

C6

+H

510

1.8

1.54

3190

1.78

1816

1.23

A1

85.8

–19

75–

1971

–A

578

.0–

1988

–15

47–

A3

+H

317

3.2

2.11

3699

1.87

3187

1.81

A2

+H

180

.80.

9923

931.

2123

071.

31A

4+

H2

148.

31.

8139

441.

9922

841.

30A

6+

I319

3.0

2.36

6721

3.39

6535

3.72

A7

+I4

205.

82.

5168

273.

4566

473.

78

M. Corradi, A. Borri / Composites: Part B 38 (2007) 172–181 177

Autho

r's

pers

onal

co

py

this reason it is not possible to effect a comparison with thefailure loads of the other wood beams. The measured fail-ure load of 80.8 kN is much lower than those of the otherreinforced beams. However, a direct comparison betweenthe behavior of the beam A2 before and after reinforce-ment is shown in Fig. 9, where it is possible to note the con-siderable increase in stiffness following application of thepultruded element.

The timber beams, A4 and C6, were reinforced byapplying an H-type pultruded element. The two elements,in wood and composite, were joined using only screwswithout applying the epoxy resin on the contact surface.The result of the flexure test highlights that the absenceof the epoxy resin did not compromise the effectivenessof the reinforcement. The beam A4 + H2 failed at a loadof 148.3 kN compared to the average capacity load of81.9 kN for the un-reinforced beam A1 and A5.

The effectiveness of the reinforcement is mainly deter-mined by the presence of the metal screws which operatedup to the failure of the reinforced beams. However, thediffering flexural stiffnesses of the wood beam and thepultruded element, respectively, determined phenomenaof withdrawal of the screws from the wood. The with-drawal of screws was particularly evident for fir beamsreinforced H-type elements. This means that failure initi-ated at the screws in the GFRP-timber beams. In orderto avoid these phenomena metal screws should be insertedinto the wood at a slight angle (15–20%) with respect to theperpendicular and toward the supports. In fact, all fivebeams A2 + H1, A3 + H3, A4 + H2, C5 + H4, C6 + H5(expect for C5 + A4 beam) clearly exhibit non-linearresponse as a consequence of partial withdrawal of thescrews though local tearing and fracture of the wood. Asa consequence the adopted screw spacing equal to 25 cmis to long for fir beams characterized by lower mechanicalproperties. Another possible solution is to increase screwsize. It is also probable that a limited compression yieldingof the wood occurred for timber beams reinforced with H-type elements.

3. Analysis

On the basis of the experimental results for the compres-sion strength of the pultruded elements in GFRP and themodality of collapse observed, it is possible to estimatethe failure load as well as the flexural stiffness for each ofthe reinforced wood beams, hypothesizing a conservationof plane sections.

The behavior of the pultruded element in GFRP is com-parable to linear-elastic, while the behavior of wood isnotorious for being practically linear-elastic under tension,while presenting linear-plastic behavior under compression(Fig. 10).

In particular, the Bazan–Buchanan law for the wood[24,25] can be expressed by:

rWc ¼ EW � eWc if eWc < eWc0

rWc ¼ rWc0 � m � ðeWc � eWc0Þ if eWc > eWc0

rWt ¼ EW � eWt

ð3Þ

Where m represents the slope of the plastic branch of theBazan–Buchanan law:

m ¼ rWc0 � rWcu

eWcu � eWc0

ð4Þ

and rWc and rWt are, respectively, the wood compressionand tensile stress, EW is the wood Young modulus, eWc

and eWt are the wood strains in compression and in tension.eWc0 is the strain value at yield stress rWc0.

In the case of I-type reinforcement the wood materialmay be assumed to be completely in tension or only par-tially lightly compressed in a limited zone. As a conse-quence it is possible to assume that:

eWc max < eWc0 ð5Þ

0

10

20

30

40

50

60

70

80

90

100

0 5 10 15 20 25 30 35 40 45Deflexion (midspan) [mm]

Load

P [k

N]

A2+H1

A2

Fig. 9. Wood beam A2 before and after reinforcement.

εW

σW

εWc0

σWc0

11

EW

EW

εWcu

σWcu

Fig. 10. Bazan–Buchanan law for wood.

178 M. Corradi, A. Borri / Composites: Part B 38 (2007) 172–181

Autho

r's

pers

onal

co

py

i.e., that the wood does not yield under compression due tothe presence of the pultruded element in the compressedzone.

As far as the FRP material is concerned, the generic lawwill be given by:

rP ¼ EP � eP ð6Þdescribing a linear-elastic behavior, having assumed thesame elastic modulus EP for the wings and web of thepultruded element.

Under these hypotheses, the position of the neutral axisis easily found:

y ¼AWðh1 þ h2

2Þ þ AP

EPEW

h1

2

AW þ EPEW

APð7Þ

where h1 is the height of the pultruded element and h2 is theheight of the wood beam.

Based on this simple linear-elastic model, we may useexperimental results of capacity P in order to estimate boththe compression and tensile stresses for wood and the com-pression stress for the GFRP pultruded element:

rWt ¼MJ x� ðh1 þ h2 � yÞ

rWc ¼MJ x� ðy � h1Þ

rP ¼MJ x� y � EP

EW

ð8Þ

Table 5 shows that the experimental and numerical valuesfor chestnut and fir beams reinforced with I-type elementsare similar and the error is always in the range 10–20%.The data obtained makes it clear that the linear-elasticmodel utilized is capable of providing acceptable resultsfor the reinforced beams C3 + I1 and C2 + I2 for whichthe maximum calculated compression stresses of the woodreached (5.96 and 4.85 MPa, respectively) turn out to besufficiently low so as to exclude the wood yielding. Anacceptable error was also calculated for fir beams rein-forced with I-type elements.

With regard to H-type reinforcement, a non-linearbehavior was observed during the experimental work. Thisnon-linearity may be due to compression yielding of thewood and/or imperfect composite action between the woodand FRP as discussed above, which the approximatedlinear-elastic model cannot account for.

The numerical non-linear model takes into account thepresence of upper reinforcement on the section under thehypothesis of preservation of plain sections and perfectadhesion between wood and pultruded element. Assumingthe failure is not reached in the composite materials, it ispossible to reduce the study of the problem to the followingfailure case: attainment of limit strain eWtu in compressionregion without exceeding limit strain in the compressionregion for wood or pultruded element.

From the condition of equilibrium it follows that(Fig. 11):

F I þ F II þ F IV ¼ F III ð9Þ

where the forces in the compression region are given by:

F I ¼2rWc0 � m eWtu

y�h1

h�y

� �� eWco

h i2

8<:

9=; � b � ½y � h1 � aðh� yÞ�;

ð10ÞF II ¼

rWc0

2� b � aðh� yÞ; ð11Þ

F IV ¼ eP maxEP ad þeP maxEP � eWtuEP

y�h1

h�y

� �2

24

35ðh1 � 2dÞc

þ eWtuEPy � h1

h� y

� �ad ð12Þ

and the force FIII in the tension zone is given by:

F III ¼eWtu

2� EWðh� yÞ � b; ð13Þ

where:

a ¼ rWc0

EWeWtu

ð14Þ

eP max ¼ yeWtu

ðh� yÞ ð15Þ

From Eq. (9) and from the stress–strain laws of materials itis possible to find the position of the neutral axis and thevalue of the ultimate bending moment of the section.

Once the neutral axis position is found, it is possible toproceed to the calculation of the ultimate bending momentof the section, which can be expressed as follows:

Mu ¼ F III � s ð16Þ

Table 5Experimental vs. numerical: capacity values

No. Reinforcement Experimental Pmax

(kN)Numerical Pmax

(kN)

C (chestnutwood)

I-type 204.3 159.4

A (fir wood) I-type 199.4 190.6C (chestnut

wood)H-type 119.5 96.1

A (fir wood) H-type 160.7 111.5

h = h1+h2y

εWtu

εWc

σWtu

εP

σPmax

σWtεWt

σWc0εWcmax

a

b

dc

εPmax

h1

h2

FIII

FII FI

FIV

STRAIN STRESS

Fig. 11. Cross-section strain and stress analysis.

M. Corradi, A. Borri / Composites: Part B 38 (2007) 172–181 179

Autho

r's

pers

onal

co

py

where s is the lever arm of internal forces FI, FII, FIII andFIV.

For application of Eq. (16), it is necessary to know thelimit and yield strain for wood in compression and the limitstrain in tension. Considering that a major problemencountered when developing flexural strength models forwood is that tensile stress at failure is greater in bendingthan in axial tension, the tensile stress at failure rWtu (par-allel to grain) was calculated from tests carried out on un-reinforced beams (30.7 MPa for fir wood and 24.8 MPa forchestnut wood). The remaining data were assumed [26,27]:

rWc0 ¼ 0:7 � rWtu

eWcu ¼ 2 � eWc0

rWcu ¼ 0:8 � rWc0

ð17Þ

4. An example of application

The design criteria described above were applied to thetimber floor of the XVII century palace (Palazzo Collicola)located in Spoleto, Italy. The reinforcement work of thefloors was entirely focused on strengthening and stiffeningof existing ancient beams, which poses construction chal-lenges, including the field application of adhesive. Toachieve the aim the supplementation of the beams shownin Fig. 12, where the common floor joists necessitate thenotching of the GFRP flange, an attempt was made toavoid any work which would be particularly traumaticfor the important historical structure. As a consequencean important issue of a reinforcement work was lightnessand minimum increase in dead loads. The reinforcementwith pultruded H-type elements is characterized by thisproperty and by an high machinability of pultruded ele-ments to be carry out in-site. Due to particular specifica-tions of the local authority it is was not possible tooperate at beam intrados (zone under tension) for the pres-ence of decorations. The reinforcement of the compressedzone has the positive characteristic to be un-visible and

the use of pultruded elements induced very interestingincrease in flexural stiffness and strength. Reinforcing oper-ations, under the direction of Dr. Andrea Giannantoni,described in detail in [28] with full description of the caseof study, turned out to be extremely fast and effective.

5. Conclusions

The application of new, advanced materials for the rein-forcement and seismic upgrading of civil engineering con-structions can represent a valid alternative to traditionaltechniques. With regard to existing constructions, the useof composite materials presents positive characteristics interms of reversibility, effectiveness, non-invasiveness, chem-ical stability and compatibility in a large number of rein-forcement typologies.

The flexural reinforcement of wood elements using pul-truded elements in composite material is capable of deter-mining significant increases in terms of strength, stiffness,and ductility when compared to the un-reinforced woodbeams tested. In particular, the increase in flexural stiffnessobtained via this method is capable of overcoming a lacktypically present in existing wood beam floors. Moreover,this intervention is easily carried out on site, without hav-ing to dismantle the overlying part of the structure, andwithin a short time, depending exclusively on the curingtimes of the epoxy resins used for glueing.

The series of tests performed on composite-reinforcedand un-reinforced wood included calorimetric, thermo-gravimetric and microscopic analyses of the epoxy resinused to bond wood and composite elements. The resultsdemonstrated that the bonding is effective at least up to50 �C, but the penetration of the epoxy resin into the woodis negligible.

The joining of the elements was also effected through theuse of screws in order to avoid problems connected to lossof adherence between the epoxy resin and wood surface.Moreover, even the use of these screws alone seemed capa-ble of guaranteeing an effective joining of the wood beamand the composite element, but more tests will be necessaryto validate this. The adopted screw spacing of 25 cm turnedout to be adequate only for chestnut beams. The presenceof soft wood like fir beams needs shorter screw spacingsor higher screw sizes. However, it is advisable to placethe screws at an inclination of 10–20� with respect to theperpendicular, and not directly perpendicularly, in orderto avoid phenomena of screw withdrawal. Finally a linear(based on the hypothesis that compression stresses forwood are sufficiently low so as to exclude the wood yield-ing) and non-linear analyses furnished acceptable resultsin terms of estimated capacity of reinforced beams.

Acknowledgements

Special thanks to engineers Dr. Andrea Giannantoniand Dr. Antonio Triboli for the cooperation and assistancegiven. ECT s.r.l. located in Jesi (Italy), ATP Pultrusion,Fig. 12. The in-site application of reinforcement: Palazzo Collicola.

180 M. Corradi, A. Borri / Composites: Part B 38 (2007) 172–181

Autho

r's

pers

onal

co

py

Creative Pultrusion, Ediltecnica s.r.l. are also thankedfor their technical assistance during the strengtheningoperations.

References

[1] Sliker A. Reinforced wood laminated beams. Forest ProdJ 1962;12(1):91–6.

[2] Borgin KB, Loedolff GF, Saunders GR. Laminated wood beamsreinforced with steel strips. J Struct Engng ASCE 1968;94(7):1681–705.

[3] Hoyle RJ. Steel-reinforced wood beam design. Forest ProdJ 1975;25(4):17–23.

[4] Bulleit WM, Sandberg LB, Woods GJ. Steel-reinforced gluedlaminated timber. J Struct Engng, ASCE 1989;115(2):433–44.

[5] Peterson J. Wood beams prestressed with bonded tension elements.J Struct Engng, ASCE 1965;91(1):103–19.

[6] Krueger GP. Ultimate strength design of reinforced timber: state ofthe art. Wood Sci 1973;6(2):175–86.

[7] Krueger GP, Eddy FM. Ultimate strength design of reinforcedtimber: moment-rotation characteristics. Wood Sci 1974;6(4):330–44.

[8] Krueger GP, Sandberg LB. Ultimate strength design of reinforcedtimber: evaluation and design parameters. Wood Sci 1974;6(4):316–30.

[9] Kobetz RW, Krueger GP. Ultimate strength design of reinforcedtimber: biaxial stress failure criteria. Wood Sci 1976;8(4):252–62.

[10] Tampone G. The slow progress of the conservation of woodenstructures and wooden architecture. In: Proceedings of the interna-tional conference in conservation of historic wooden structures,February 22–27, 2005.

[11] Meier U. Bridge repair with high performance composite materials.Mater Tech 1987;4:125–8.

[12] Triantafillou TC. Shear reinforcement of wood using FRP materials.J Mater Civil Engng, ASCE 1997;9(2):65–9.

[13] Triantafillou TC, Plevris N. Post-strengthening of r/c beams withepoxy-bonded fiber-composite materials. Proc ASCE specialty confon Advanced Composites for Civil Engng Struct. ASCE; 1991, p.245–6.

[14] Kropf FW, Meierhofer U. Strengthening, retrofitting and upgradingof timber structures with high-strength fibres. SEI 3/2000, 2000.

[15] Gentile C, Svecova D, Saltzberg W, Rizkalla SH. Flexural strengh-tening of timber beams using GFRP. In: 3rd International conferencein advanced composites materials in bridge and structures, proceed-ings, Ottawa-Canada, August 15–18, 2000.

[16] Borri A, Corradi M, Grazini A. A method for flexural reinforcementof old wood beams with CFRP materials. J Compos, Part B 2005;36/2:143–53.

[17] Corradi M, Speranzini E, Borri A, Vignoli A. In-plane shearreinforcement of wood beam floors with FRP. J Compos, Part B2006;37/4–5:310–9.

[18] Tascioglu C, Goodell B, Lopez-Anido R. Bond durability character-ization of preservative treated wood and E-glass/phenolic compositeinterfaces. Sci Technol 2003;63:979–91.

[19] Tascioglu C, Goodell B, Lopez-Anido R, Peterson M, Halteman W,Jellison J. Monitoring fungal degradation of E-glass/phenolic fiberreinforced polymer (FRP) composites used in wood reinforcement.Int Biodeter Biodegrad 2003;51:157–65.

[20] UNI ISO 3130, Wood. Humidity determination.[21] EN 13706-2: Reinforced plastic composites – specifications for

pultruded profiles – Part. 2: methods of test and general requirements,CEN (European Committee for Standardization), October 2002,2002.

[22] ASTM D 695M-91. Standard test method for properties of rigidplastics.

[23] ASTM D 638-95 Standard test method for tensile properties ofplastics.

[24] Buchanan AH. Combined bending and axial loading in lumber. JStruct Engng, ASCE 1986;112(12):2592–609.

[25] Buchanan AH. Bending strength of lumber. J Struct Engng, ASCE1990;116(5):1213–29.

[26] Boding J, Jayne BA. Mechanics of wood and wood composites. NewYork: Van Nostrand Reinhold; 1982.

[27] Forest Products Laboratory. Wood handbook – Wood as anengineering material. Gen. Tech. Rep. FPL-GTR-113. Madison,WI: US Department of Agriculture, Forest Service, 1999.

[28] Borri A, Giannantoni A. Esempi di utilizzo dei materiali compositiper il miglioramento degli edifici in muratura, Examples of possibleuses of composite materials for seismic upgrading of masonrystructures. In: XI Conference in ANIDIS, January 25th–29th, CDrom, Genova, Italy, 2004.

M. Corradi, A. Borri / Composites: Part B 38 (2007) 172–181 181