Experimental investigation on FRP to steel adhesively-bonded joint under tensile loading

PROCEEDINGS o f

The Sixth International

Composites Conference

COMPOSITES & NANOCOMPOSITES

in C I V I L , O F F S H O R E a n d M I N I N G I N F R A S T R U C T U R E

Edited by:

YuBai

Sri Bandyopadhyay

Xiao Ling Zhao

Raman Singh

Sami Rizkalla

^

M O N A S H U n i v e r s i t y

i

U N S W

NC STATE UNIVERSITY

The topics of ACUN conferences cover all aspects of the science

and technology of novel composite materials, from materials

fabrication, processing, manufacture, structural and property

characterization, sensor technology/structural health monitoring,

theoretical analysis, modeling and simulation, to various applications

in aerospace, automotive, civil infrastructure, packaging,

ship-building, and recreational products.

ACUN conferences have a historical background:

In 1998 at an ICCE Conference in Las Vegas USA, Dr. S r i

Bandyopadhyay (UNSW, Australia), Prof* Sami Rizkalla (then

of University of Manitoba, Canada), Dr. Piyush Dutta (CRREL,

USA), and Prof. Debes Bhattacharyya (U Auckland, NZ)

discussed the possibility of a totally new class of composites

conference.

This led to the birth of ACUN conferences: i in 1999,

ACUN-2 in ACUN-2000, ACUN-3 in ACUN-2001, ACUN-4 in ACUN-200ACUN-2 and ACUN-5 in

2006 all held at University of New South Wales - Australia.

The ACUN-3, ACUN-4 and ACUN-5 conferences held at UNSW

were ranked by majority delegates from 21 countries as amongst the

World's top 5 to 10 conferences they attended in their entire career.

ACUN-5 conference proceedings are available on the website:

PROCEEDINGS of

the 6* Intemational Composites Conference (ACUN-6)

COMPOSITE & NANOCOMPOSITES in C I V I L , OFFSHORE and

M I N I N G INFRASTRUCTURE

Edited by:

Y u B a i

Sri Bandyopadhyay

Xiao Ling Zhao

Raman Singh

Sami Rizkalla

1 MONASH University i UNSW

Disclaimer

The opinions expressed i n this book ( A C U N - 6 proceedings) are solely those o f the individual researchers, workers or groups. A C U N - 6 or Monash University or anyone associated w i t h the organization o f A C U N - 6 does not Accept any responsibility whatsoever f o r any untruthfulness o f the results, or any litigation or compensation claims arising out o f the use o f any information/data included here i n these proceedings by any individual, parties/parties i n Austraha or anywhere else.

The A C U N - 6 proceedings document serves only the purpose o f conferencing/scientific exchange of ideas f o r the delegates present or other readers f o r their o w n individual intellectual exercise and/or f u l f i l m e n t .

The above disclaimer is expUcitly expressed by the editors and publishers o f A C U N - 6 as w e l l as Monash University.

Proceedings of ACUN-6

The 6^*^ Intemational Composites Conference

COMPOSITE & N A N O C O M P O S I T E S i n C I V I L , OFFSHORE and M I N I N G I N F R A S T R U C T U R E

ISBN 978-0-646-58589-5

ISBN

Editors:

Y u B A I , Sri B A N D Y O P A D H Y A Y , Xiao-Ling Z H A O , Raman S I N G H and Sami R I Z K A L L A

ACUN6 -Composites and Nanocomposites in Civil, Offshore and Mining Infrastructure ^ Melbourne 14 -16 November 2012

Experimental Investigation on FRP to Steel Adhesively-bonded Joint

under Tensile Loading

^ Jiang rxu.iiang@tudelft.nl), M.H.Kolstein & F.S.K.Bijlaard

Steel and Composite Structure Group, Section of Structure and Building Engineering, Department of Design

and Construction, Faculty of Civil Engineering and Geosciences, Delft University of Technology, Delft, the

^Netherlands

ABSTRACT: Due to various advantages o f Fibre-Reinforced Polymer (FRP) decks, the FRP to steel compos-ite girder system is being increasingly used i n the construction o f new bridges as w e l l as the rehabilitation pro-jects of old bridges. This paper focus on the mechanical behaviors and failure modes of the adhesively-bonded

oins between FRP sandwich decks and steel girders. A tensile-shear loading device was designed and pre-sented herein. The adhesively-bonded joints were experimentally investigated under tensile loading. The aver-age ultimate failure load of surface pretreated specimens was 17.62kN, which was 9.83% higher than that o f un-pretreated specimens. Further comparison on failure modes confirmed that the surface pretreatment can improve the bonding quality between FRP composites and adhesive layer, and correspondingly increase the strength of the whole adhesive joint under tensile loading.

1 I N T R O D U C T I O N

Fibre-reinforced polymer (FRP) composite materials are being increasingly used i n many c i v i l engineer-ing applications as a competitive alternative f o r con-ventional materials like steel, wood and concrete. One conspicuous utilization o f FRP composites is FRP bridge decks, which offers tremendous potential to meet critical needs f o r rehabilitation and new con-stmction o f pedestrian and highway bridges. This is mainly due to the advantages of FRP decks which al-low the lightweight of bridge superstructures, the ease o f installation, m i n i m u m traffic disturbing, large tolerance f o r environmental corrosion, long service l i f e time, as w e l l as l o w maintenance cost. Current commercially available FRP decks can be classified into two categories according to the types of assembly and construction: sandwich panels and multi-cellular type panels (Zhou and Keller 2005), as shown i n F i g u r e l .

a) Sandwich FRP deck b) Cellular FRP deck (ECOSAFE) (ASSET) Figure 1. FRP bridge decks

To be cost effective, the FRP decks are usually sup-ported by steel gkders, as shown i n Figure 2. Steel girders enhance the ductility o f this composite bridge system after failure loading achieved, which com-pensate f o r the brittle characteristics o f FRP compos-ites. Between the FRP decks and steel girders, adhe-sively bonding technique is usually employed as a preferable connection method. Comparing to bolted or stud connections, adhesively-bonded connections can reduce construction time, save weight by elimi-nating fasteners, introduce more u n i f o r m load trans-fer and provide better long-term performance. Bolted connections usually result i n much higher stress centrations where cracks occur, while adhesive con-nections are a more material-adapted connection

technique since larger surfaces can be glued together, thus reducing stresses. This k i n d o f FRP composite girder system was utilized during recent years (Cassity et al. 2002, Luke et al. 2002, Knippers et al. 2010). I n Knippers's research (Knippers et a/. 2010), it was employed as a flyover across the federal road B3 i n Germany. The high durability of the FRP composites and the fast assembly o f the bridge were decisive factors f o r this apphcation. Through the ex-perimental investigations (Cassity et al. 2002), the degree o f composite action between cellular FRP decks and steel girders was studied and subsequently adopted i n a rehabilitation project o f an old and dete-riorated bridge. Through these projects, valuable ex-perience was gathered conceming in-site constmc-tions, and the good performance o f FRP composite decks was confirmed.

T r a f f i c loads

steel girder

FRP composite deck

Figure 2. FRP composite deck to steel girder bridge sys-tem (SchoUmayer 2009)

Regarding to the mechanical performance o f adhe-sively bonded joints, researches (Keller and Vallee 2005a, b , Vallee and Keller 2006) were conducted, focusing on the adhesively bonded single-lap joints and double-lap joints. These adhesive joints com-posed o f pultmded GFRP composite profiles glued by epoxy adhesives. Parametric studies were con-ducted experimentally and numerically on the over-lap length, the adhesive layer thickness, the adherend thickness and the degree o f chamfering o f the adher-ends. The results indicated that the combination o f local through-thickness tensile (peeling) and shear stresses was the most severe stress-state and usually initiated the failures i n the outer fiber-mat layers o f the adherends below the j o i n t edges. Further re-searches (Vallee et al. 2006b, a) offered a probabilis-tic strength prediction method on the adhesive joints under quasi-static axial tensile loading. Unfortu-nately the technical background and researches on the adhesively bonded joins between FRP decks and steel girders have not been documented adequately in literatures. The researches (Gurtler 2004, SchoUmayer 2009) investigated the mechanical

be-haviors o f FRP-steel composite girder system i n the longitudinal bridge direction and transverse direction respectively, but not particularly f o r the adhesive j o i n t part. The research presented i n this paper was

focusing on the adhesively-bonded joints between FRP sandwich decks and steel gkders. A tensile-shear loading device was designed w i t h the capacity to provide the combination o f tensile and shear loads i n six different ratios. The mechanical behaviors and failure modes o f adhesively-bonded joints under ten-sile loading were investigated experimentally, con-sidering different surface pretreatment methods on FRP sandwich decks and steel girders.

2 T E N S I L E - S H E A R L O A D I N G D E V I C E

Generally, there are three typical stress states for the adhesive j o i n t between FRP decks and steel girders: 1. shear stress x: due to the composite action be-tween FRP deck and steel gkder i n the longitudinal direction o f bridge, the decks and steel girders trend to bend together to carry the traffic load. Thus, the adhesive joint are under the shear stress condition to transfer the loading f r o m FRP decks to steel beams, as shown i n Figure 3 a);

2. tensile stress o: i n the transverse direction of bridge, loading on other traffic lanes causes up-lift forces on adhesive joints, which results i n tensile stress, as shown i n Figure 3b);

3. combination o f above two stress states with dif-ferent ratios o f contributions f r o m tensile stress state and shear stress state.

_^FRP composite deck f FRP composite _^FRP composite deck f steel

7 \ 7 ^ V V A / \ / ^ \ / V V \ / \ / \ / V V A ^ . 0 4 ^

^^^^I-shape steel beam ^

a) Shear stress in the longitudinal direction

M ( ( I M t t M ( t » ( M M I ( * t I t t I I •FRP composite deck

I-shiape s t e e l beam

b) Tensile stress in the transverse direction

Figure 3. Typical stress states of the adhesive joint

Depending on the above three stress states, a smaj^ loading device was needed f o r providing tensi

loading, shear loading and combination of both si-multaneously. The adhesively-boned joint between pjiP deck and steel girder was extracted f o r experi-mental investigation as shown i n Figure 4.

sandwich FRP deck

adhesive layer

steel support

Figure 4. FRP-steel adhesively bonded joint

A 190mmx90mm piece o f sandwich bridge deck was adhesively bonded to the convex shape steel support. I n the middle of sandwich deck was the 38.1mm Balsa SB150, a core material produced from certified kiln-dried balsa wood i n the 'end-grain' configuration. The surface layers were three layers o f 0.94mm EQX1200, which are the glass-fibre reinforced laminated polymer composites (54% glass content by weight). The sandwich profiles were manufactured by resin vacuum infusion. The thick-ness o f 9 0 m m x 9 0 m m adhesive layer was 6 m m . The dimensions o f adhesive joint s were determined de-pending on the actual conditions o f composite bridges as w e l l as lunitations o f loading equipment. The tested adhesive joint was kept as small as possi-ble to make i t easily mobile and independent o f large and complicated experimental facilities. I n order to fix the adhesive joint to the loading system, other accessorial components were designed as shown i n Figure 5. The steel support was grilled w i t h 4 holes to be connected to steel blocks by bolts. For the sandwich decks, no hole was made, since the discon-tinued part i n decks w i l l cause more stress distribu-tion distordistribu-tion, which was not actual i n applicadistribu-tions of composite bridges. A l l the accessorial compo-nents were manufactured by steel. Comparing w i t h the FRP composites and adhesive materials, the de-formation o f the steel components can be neglected during tests, due to the high stiffness o f steel mate-rial. To f i x the sandwich deck part, i t was designed to be fastened by two purple-color L-shape steel plates through four bolts to the top steel block, as shown i n Figure 5a). W h i l e , the steel support was fastened directly through four bolts to the bottom

steel block, as shown i n Figure5b). The two steel blocks were fastened to circular steel plates, as shown i n Figure 5c). The circular steel plates were separated into two pieces. Three bolts were em-ployed to transfer the loading uniformly. B y loading the different angles of circular steel plates, the spe-cific stress-state can be achieved i n the adhesive j o i n t , such as pure tension, pure shear and combina-tion o f both. Correspondingly, six loading condicombina-tions were feasible through this well-designed loading sys-tem.

a) Deck fixed configuration b) Steel support fixed con-figuration

c) Whole loading system

Figure 5. Tensile-shear loading device

3 E X P E R I E M N T A L I N V E S T I G A T I O N

The FRP-steel adhesively-bonded joints were ex-perimentally investigated under tensile loading. The test setup were shown i n Figure 6. A S C H E N C K Hydropuls testing machine w i t h a capacity o f 200 k N i n tension was employed and controlled by the I N S T R O N 8400 controller. The whole tensile-shear device was loaded by jacks through two hinged joints, which could avoid the additional bending moment due to eccentric loading. The quasi-static experiments were performed under LVDT(linear

variable differential transformer)-control at the rate of 0.001 mm/sec and the failure took approximately

13 minutes f o r the strongest specimens. T w o L V D T s were assigned on each side o f the loading system, as shown i n Figure 6, which measured the displacement between the top and bottom loading device. What is more, three displacement sensors were assigned on both sides o f adhesive joints, to track the vertical de-formation between FRP sandwich deck and steel support during the whole test process, as shown i n Figure 7 and Figure 8. Six rephcated specimens were prepared as shown. Before making the gluing, three of specimens were pretreated on the surfaces o f FRP sandwich deck and steel supports by using sand-papers and acetone. For comparison, the other three specimens were glued without any surface pretreatment. The surface pretreated specimens were i n d i -cated as SP-specimens i n the f o l l o w i n g chapters, and un-pretreated specimens were described by UP-specimens.

Figure 6. Experimental set-up

Figure 7. Adhesive joint specimen

Figure 8. Assignment of six displacement sensors

4 RESULTS A N D DISCUSSION

For the UP-specimens, the loads increased ahnost linearly up to failure. As hsted i n Table 1, the aver-age ultimate failure load was 16.04 k N , the deviation f r o m the three specimens was w i t h i n 3%, which means the test resuhs were reliable.

Table 1. Ultimate failure loads of six adhesive joints

- _{UPOl UP02 UP03} Average

Failure (kN) load 15.69 16.43 16.04 16.05 SPOl SP02 SP03 Average Failure (kN) load 19.37 17.93 15.57 17.62

Figure 10 shows the load-displacement curves o f three adhesive joint specimens, which were meas-ured by the L V D T s . I t is manifest that the curves from UP-specimenOl and UP-specimen02 agree w e l l with each other. Bur f o r the UP-specmien03, the

stiffness is a bit different f r o m the other t w o curves, which could be due to the deviation o f material properties and quality o f gluing between FRP sand-wich deck and steel support. I t is obvious that the three curves are almost parallel to each other i n the stable load increasing stage. It can be explained that at the beginning o f loading, the f r i c t i o n between each component o f loading device made the initial stiffness o f specimens different f r o m each other. However, when the loading is large enough beyond the f r i c t i o n , the stiffness o f three adhesive joints was more or less the same.

Figure 9 shows the measurement o f vertical defor-mation between FRP sandwich deck and steel sup-port on the UP-specimenOl f r o m six displacement sensors. Tests on other specimens have the similar mechanical behavior i n the vertical dkection. I t is clearly seen that the measured data f r o m displace-ment sensors DS-01, DS-03, DS-04 and DS-06 are close to each other, while the deformation f r o m DS-02 and DS-05 are relatively smaller. I t implies that the edge part o f bonding area deformed larger than the middle part. However, the total deformation through the adhesive joint was i n a very small scale. Figure 9 also shows that, besides the pure tensile loading, a certain amount o f additional bending mo-ment was also appHed during the whole testing proc-ess, w i t h the deformation f r o m 01, 03, DS-04 and DS-06 deviating f r o m each other. I t means the loading was not applied exactly centrically on the adhesive joint, which cannot be avoided i n such a small scale test.

1 8 - ,

0.000 0.003 0.006 0.009 0.012 0.015 0.018 0.021 0.024

Deformation [mm]

Figure 9. Load-deformation curves measured from six vertical displacement sensors on the UP-specimenOl (DS=displacement sensor)

Displacement [mm]

Figure 10. Load-displacement curves of SP-specimens

For the SP-specimens, the load-displacement curves were plotted i n Figure 10. For comparison, the three curves of UP-specimens were also included i n the same figure. I t is manifest that, besides the similar stiffness w i t h UP-specimens, a certain extent o f duc-tility was achieved after the failure o f SP-specimens, especially f o r the SP-specimenOl. The degree o f ductility was closely related to the failure mode. Fig-ure., 11 shows the failure modes of SP-specimens comparing w i t h UP-specimens. I t can be clearly ob-served that the failure plane f o r un-pretreated adhe-sive joint was through the interface between the FRP sandwich deck and adhesive layer. For the SP-specimens, the failure plane partly moved to the de-lamination of FRP composites, as shown i n Figure 12.

a) Failure modes of SP-specimens

b) Failure modes of UP-specimens

Figure 11. Failure modes of adhesive joints

The average ultimate failure load o f SP-specimens was 17.62kN, which was 9.83% higher than that o f UP-specimens. It means that the surface pretreat-ment can unprove the bonding quality between FRP composites and adhesive layer, and correspondingly increase the strength o f the whole adhesive joint un-der tensile loading. What is more, f r o m Figure 11 a), it can be f o u n d that the areas o f FRP delaminated parts are different among specimens. SP-specimenOl attained the largest FRP delaminated area which covered almost the whole bonding area, while SP-specmien03 attained the smallest. That is why the SPspechnenOl owned the maximum u l t i -mate failure load, better ductile performance and even higher stiffness. Thus, the conclusion can be easily drawn that the strength and ductility o f the surface-pretreated adhesive joints are closely related to the FRP delaminated area.

Figure 12. Delamination failure of SP-specimenOl

5 C O N C U L S I O N

FRP sandwich deck to steel support adhesively bonded joints were experimentally investigated un-der the tensile loading condition. The mechanical behavior o f adhesive joint specimens w i t h surface pretreatment (SP) and un-pretreatment (UP) were compared. For UP-specimens, the joints failed in a brittle mode, which occurred between FRP sandwich deck and adhesive layer. The average ultimate fail-ure load o f UP-specimens was 16.04 k N . For SP-specimens, the failure o f adhesive joint was trig-gered by delamination of FRP composites, which re-sulted i n a relatively ductile failure mode. The aver-age ultimate failure load o f SP-specimens was 9.83% higher than that o f UP-specimens. The further discussion confirmed that strength and ductility of the surface-pretreated adhesive joints were closely related to the FRP delaminated area. Therefore, the sufficient surface pretreatment on FRP sandwich decks and steel girders is necessary to improve the mechanical performance of the adhesively bonded joints under the tensile loading.

REFERENCES

Cassity, P., Richards, D., Gihespie, J., 2002. Compo-sitely acting frp deck and girder system. Structural

Engineering International 2, 71-75.

Gurtler, H.W., 2004. Composite action of frp bridge decks adhesively bonded to steel main girders. Ecole

Polytechnique Federale de Lausanne.

Keller, T., Vallee, T., 2005a. Adhesively bonded lap joints from pultruded gfrp profiles. Part i : Stress-strain analysis and failure modes. Composites Fart D'

Engineering 36 (4), 331-340.

Keller, T., Vallee, T., 2005b. Adhesively bonded lap joints from pultruded gfrp profiles. Part i i : J^^^

strength prediction. Composites Part B-EngineerM 36 (4), 341-350.

Knippers, J., Pelke, E., Gabler, M . , Berger, D., 2010. Bridges with glass fibre-reinforced polymer decks: The road bridge in friedberg, Germany Structural

En-gineering International 20 (4), 400-404.

Luke, S., Canning, L . , Collins, S., Knudsen, E., Brown, P., Taljsten, B., Olofsson, L , 2002. Advanced com-posite bridge decking system—project asset.

Struc-tural Engineering International 12 (2), 76-79.

SchoUmayer, M . , 2009. Through-thickness performance of adhesive connections between frp bridge decks and steel main girders. Ecole Polytechnique Federale de

Lausanne.

Vallee, T., Correia, J.R., Keller, T., 2006a. Probabihstic strength prediction for double lap joints composed of pultruded gfrp profiles - part i i : Strength prediction.

Composites Science and Technology 66 (13), 1915¬

1930. ,^ ,

Vallee, T., Correia, J.R., Keller, T., 2006b. Probabihstic strength prediction for double lap joints composed of pultmded gfrp profiles part i : Experimental and nu-merical investigations. Composites Science and

Tech-nology 66 (13), \903-19 U.

Vallee, T., Keller, T., 2006. Adhesively bonded lap joints from pultruded gfrp profiles. Part hi: Effects of cham-fers. Composites Part B-Engineering 37 (4-5), 328¬ 336.

Zhou, A., Keller, T., 2005. Joining techniques for fiber reinforced polymer composite bridge deck systems.