RESILIENT INFRASTRUCTURE June 1 4, 2016

RESILIENT INFRASTRUCTURE June 1 4, 2016 EXPERIMENTAL STUDY ON THE CAPACITY OF BARRIER DECK ANCHORAGE IN MTQ PL-3 BARRIER REINFORCED WITH HM-GFRP BARS WITH HEADED ENDS Michael Rotami Department of Civil Engineering, Ryeron Univerity, Toronto, Ontario, Canada Khaled Sennah Department of Civil Engineering, Ryeron Univerity, Toronto, Ontario, Canada S. Milad Dehnadi Department of Civil Engineering, Ryeron Univerity, Toronto, Ontario, Canada ABSTRACT A recent deign work conducted at Ryeron Univerity on PL-3 bridge barrier ha led to an economical gla fibre reinforced polymer (GFRP) bar detailing for utainable contruction. A PL-3 barrier wall of 27.6 m length wa contructed uing the propoed GFRP bar configuration, incorporating the ue of V-Rod headed-end bar. The propoed barrier configuration wa recently crah teted to qualify it ue in Canada highway bridge. Then, wall egment of thi barrier were teted under tatic loading to-collape to determine their tructural behavior, crack pattern and ultimate load carrying capacity under imulated vehicle impact load. Tet reult led to etablihing two Standard drawing by Ontario Minitry of Tranportation (MTO) for ue by conulting engineer and contractor. The crah-teted barrier dimenion were identical to thoe pecified by Minitry of Tranportation of Quebec (MTQ) for PL-3 barrier except that the bae of the barrier wa 40 mm hort and the deck lab i of 200 mm thickne, leading to reduction in the GFRP embedment depth into the deck lab. A uch, Ryeron Univerity reearch team propoed an experimental program to enure that the reitance of barrier-deck junction, with the reduced width of barrier bae and thickne of the deck lab, i greater of equal to the pecified factored deign load applied to the barrier wall imulating vehicle impact. Thi paper ummarize the experimental program to jutify the modified barrier deign to fit with MTQ barrier and deck lab dimenion and experimental finding when compared to the available factored applied moment pecified in CHBDC of 2006 for the deign of barrier-deck junction. Correlation between the experimental finding and the factored applied moment from CHBDC equivalent vehicle impact force reulting from the finite-element modelling of the barrier-deck ytem wa conducted followed by recommendation for ue of the propoed deign in highway bridge in the Province of Quebec. Keyword: Bridge barrier deign, cantilever deck lab, truck impact, bridge code, experimental teting, FRP. 1. BACKGROUND OF RESEARCH Until recently, the intallation of GFRP bar wa often hampered by the fact that bent bar have to be produced in the factory ince GFRP bar cannot be bent at the ite. Alo, bent GFRP bar are much weaker than traight bar, due to the redirection and aociated rearrangement of the fibre in the bend. A a reult, number of bent GFRP bar i increaed and even doubled at uch location where bar bent are required. The ue of headed-end GFRP bar, hown in Fig. 1, i intended to eliminate the unneceary and expenive ue of cutom made bar bend. The headed end i cat over the and-coated bar a hown in Fig. 1. Recently, Ryeron Univerity reearch team developed a cot-effective GFRP bar detailing for PL-3 barrier with MTO barrier dimenion. The GFRP bar ued to develop reinforcement detail, a hown in Fig. 2, were of and-coated urface profile to enure optimal bond between concrete and the bar. The 13M (#4) high-modulu (HM) GFRP bar of pecified tenile trength of 1312 MPa, modulu of elaticity of 65.6 ± 2.5 GPa, and train at rupture of 2%, a lited in the manufacturer catalogue (Pultrall, 2013), were ued a vertical bar at the back face of the barrier wall. Alo, the 15M (#5) high-modulu MAT-732-1

(HM) GFRP bar of pecified tenile trength of 1184 MPa, modulu of elaticity of 62.5 ± 2.5 GPa, and train at rupture of 1.89% were ued in all other bar hape hown in Fig. 2. The ue of headed-end GFRP bar wa propoed in thi reearch to allow for anchorage in concrete at lower cot than the bend bar. Fig. 1 View of HM-GFRP bar with end anchorage head Fig. 2. Barrier cro-ection howing the difference in geometry between the crah-teted MTO barrier and the propoed MTQ PL-3 barrier (the black-haded area of 40 mm width i removed from the MTO crah-teted barrier dimenion to reach MTQ barrier bae width of 435 mm) Recent reearch work conducted at Ryeron Univerity on PL-3 bridge barrier propoed a cot-effective barrier configuration incorporating high-modulu GFRP bar with headed end. To qualify the developed barrier configuration for ue in bridge contruction, a full-cale PL-3 barrier wall of 27.6-m length wa contructed to perform vehicle crah teting in December 2011 (Sennah and Khederzadeh, 2014). The crah tet wa performed in accordance with MASH Tet Level 5, TL-5, (MASH, 2009). Evaluation criteria for full-cale vehicle crah teting were baed on three appraial area namely: (i) tructural adequacy; (ii) occupant rik; and (iii) vehicle trajectory after colliion. Reult from the tet qualified uch innovative barrier ytem to reit vehicle impact per MASH crah tet requirement. Crah tet reult howed that the developed barrier contained and redirected the vehicle. The vehicle did not penetrate, underride or override the parapet. No detached element, fragment, or other debri from the barrier were preent to penetrate or how potential for penetrating the occupant compartment, or to preent undue hazard to other in the area. No occupant compartment deformation occurred. The tet vehicle remained upright during and after the colliion. After conducting the vehicle crah tet on the developed GFRP-reinforced barrier hown in Fig. 2, full-cale tatic tet to-collape were performed in other part of thi barrier at interior and exterior location (Khederzadeh and Sennah, 2014) to invetigate it ultimate load carrying capacity per Canadian Highway Bridge Deign Code, CHBDC, (CSA, 2006a). The experimental ultimate load carrying capacity of the barrier wa oberved to be far greater than CHBDC factored deign tranvere load. The failure pattern wa initiated by a trapezoidal crack pattern at the front face of the barrier, followed by punching hear failure at the tranvere load location. Baed on the punching hear failure developed in the barrier wall and comparion with available punching hear equation in the literature, an empirical punching hear equation wa propoed to determine the tranvere load carrying capacity of PL-3 bridge barrier wall reinforced with GFRP bar. To perform the above-mentioned tatic tet on the contructed barrier, Ryeron reearch team conidered CHBDC proviion that pecifie tranvere, longitudinal and vertical load of 210, 70 and 90 kn, repectively, that can be applied imultaneouly over a certain barrier length. CHBDC pecifie that tranvere load hall be applied over a barrier length of 2400 mm for PL-3 barrier. Since tranvere loading create the critical load carrying capacity, both the longitudinal and vertical load were not conidered in the deign of barrier wall reinforcement and anchorage between the deck lab and the barrier wall. It hould be noted that CHDBC pecifie a live load factor of 1.7. Thu, the deign equivalent impact load on PL-3 barrier wall over 2.4 m length i 357 kn. MAT-732-2

Fig. 3. Current dimenion and GFRP bar reinforcement for MTQ PL-3 tapered barrier (MTQ, 2013) Fig. 4. View of the propoed MTQ PL-3 barrier with 200-mm thick deck lab,165 mm bar embedment in the deck and 435 mm bae width 2. THE PROBLEM Minitry of Tranportation of Quebec (MTQ) pecifie PL-3 barrier dimenion and GFRP bar detailing, hown in Fig. 3. By comparing MTQ barrier dimenion with thoe in the crah-teted barrier in Fig. 2, one may oberve the difference in the width of barrier bae at the barrier-deck junction (i.e. 475 mm in the crah-teted barrier and 435 mm in MTQ barrier). By drawing the two barrier in one graph a hown in Fig. 2, one may oberve the blackhaded area, on the left ide of the barrier-deck junction, of 40 mm width and about 117 mm deep, taken out from the crah-teted barrier to reach MTQ barrier dimenion of 435 mm. Although the barrier bae width decreaed by about 8%, the diagonal headed bar at the lower tapered portion of the front face of the barrier become teeper by about 13% (i.e. with 68 lope in the MTQ barrier compared to 60 lope in the crah-teted barrier). It hould be noted that increaing the lope of the diagonal GFRP bar from 60 to 68 increae the available vertical tenile force in the bar by 7% (i.e. in 68 / in60 = 1.07). On the other hand, MTQ requeted that the concrete cover of the headed-end bar at the barrier-deck junction to be 50 mm a depicted in Fig. 4 in contrat to the 35 ± 10 mm concrete cover ued the MTO crah-teted barrier detail (Sennah and Khederzadeh, 2014). Since thi change i at the barrier deck junction, and the amount and hape of reinforcement a well a other barrier dimenion are identical to thoe in the crah teted barrier, Ryeron reearch team believe that a tatic tet in a contructed MTQ barrier egment in the laboratory would jutify the change in the barrier width at the barrier deck junction per CHBDC Claue 12.4.3.5. In addition, reult on the tatic load tet on the crah teted barrier (Khederzadeh and Sennah, 2014) reulted in a factor of afety in deign in the order of 1.9 to prevent punching hear failure at the top of the barrier wall. Since the crah-teted barrier and MTQ barrier are of the ame width, height and lope for the top tapered portion a well a the amount, type and arrangement of GFRP bar, one may oberve imilar behavior for punching hear capacity in the MTQ barrier. A for the thickne of the deck lab cantilever upporting the PL-3 barrier, the deign of the crah-teted PL-3 barrier conidered a minimum deck lab thickne of 225 mm, leading to a 195 mm vertical embedment length of the headed-end GFRP bar into the deck and a minimum concrete cover of 25 mm to the bottom urface of the deck lab cantilever. On the other hand, MTQ conider 200 mm thick lab with top and bottom reinforcement. To accommodate uch change compared to the crah-teted barrier, Ryeron reearch team changed the embedment length of headed-end GFRP bar at the lower tapered portion of the front face of the crah-teted barrier hown in Fig. 2 to be 165 mm in the vertical direction for the propoed MTQ PL-3 barrier hown in Fig. 4. Thi allow for a concrete cover to the headed bar of 35 mm from the bottom urface of the 200-mm thick deck lab cantilever a hown in Fig. 4. A a reult, the headed-end bar embedment length i propoed to change from 225 mm in the crah-teted barrier to 182 mm in the propoed MTQ barrier hown in Fig. 4. Alo, the embedment length of the traight-end vertical bar at the back face of the barrier a well a the diagonal, traightend, bar extending from the top tapered portion of the front face of the barrier changed from 185 mm in the crahteted barrier to 165 mm in the propoed MTQ barrier. A mentioned earlier, the change in bar anchorage in the MAT-732-3

deck lab can be qualified by conducting tatic load teting in the laboratory for actual-ize barrier of 900 mm length only. So, there i no need to repeat vehicle crah teting per CHBDC Claue 12.4.3.5. The objective of thi reearch i to provide experimental data and numerical evaluation, uing the finite-element modelling, a a proof of concept of, or to validate, the propoed change to the GFRP bar anchorage at the barrier-deck junction to meet MTQ barrier dimenion compared to the crah-teted barrier-deck junction dimenion. The reult from the experimental program on elected barrier egment are correlated to CHBDC factored applied moment at the barrier-deck junction pecified in Table C5.4 of CHBDC Commentarie (CSA, 2006b) a well a the reult from the finiteelement computer modelling conducted by Ryeron reearch team. 3. EXPERIMENTAL PROGRAM Three full-cale PL-3 barrier pecimen of 900 mm length were erected and teted to-collape to determine their ultimate load-carrying capacitie and failure mode at deck barrier joint. Figure 5 how chematic diagram of pecimen S-1 that repreent the interior egment of the propoed MTQ barrier. The barrier i reinforced with M15 GFRP vertical bar at the front face at 300 mm pacing and 13M GFRP vertical bar at the back face of the barrier wall at 300 mm pacing. All vertical GFRP bar are embedded into a 200-mm thick deck lab cantilever with vertical embedment length of 165 mm. The cantilever deck lab wa reinforced in the main direction with M20 teel bar at 100 mm pacing. Specimen S-2 repreent the exterior egment of the propoed MTQ barrier on which the vertical reinforcement at the front face of the barrier i doubled compared to thoe hown in Fig. 5 for pecimen S-1. In thi cae, M15 GFRP vertical bar were ued at the front face of the barrier at 150 mm pacing while the vertical reinforcement at the back face of the barrier i kept a M13 at 300 mm pacing. It hould be noted that pecimen S- 1 and S-2 repreent barrier contruction in cae of lab-on-girder bridge ytem. Fig. 5. Cro-ection of pecimen S-1 for interior portion of the barrier reting over deck lab cantilever Fig. 6. Cro-ection of pecimen S-3 for interior portion of the barrier reting over thick deck lab In cae, barrier wall intalled on top of olid lab bridge, voided lab bridge and adjacent box beam ytem, the barrier wall i conidered connected to non-deformable thick lab compared to the cae of barrier wall intalled over 200-mm thick deck lab cantilever in lab-on-girder bridge ytem. Specimen S-3 hown in Fig. 6 repreent thi cenario where the barrier wall i fixed to a 500 mm thick lab reting on the laboratory floor to preent it flexural deformation. The dimenion and GFRP arrangement are the ame a thoe for pecimen S-1. The embedment length of the barrier vertical GFRP bar into the thick olid lab bae i maintained 165 mm. Thee three pecimen repreent the cae of new contruction. All pecimen were cat uing ready mix concrete with target compreive trength of 35 MPa. The tet pecimen were cat in the ame day. In order to determine the trength of the concrete, MAT-732-4

three 100 200 mm concrete cylinder collected from the concrete ued to cat each barrier and then they batch were teted jut after teted on the ame day the barrier pecimen were loaded to collape. To take into account the effect of the number of teted cylinder and the deviation of the trength value of each cylinder with the average value, concrete characteritic trength wa calculated uing the available equation pecified in CHBDC Chapter 14 for bridge evaluation (CSA, 2006a). Analyi of data reulted in the characteritic compreive trength of concrete a 39.72, 38.84 and 37.74 MPa for barrier pecimen S-1, S-2 and S-3, repectively. Figure 7 how the arrangement of GFRP and teel bar in the teted pecimen. More information can be found elewhere (Rotami, 2016). Figure 8 how a chematic diagram of the tet etup, while Fig. 9 how view of pecimen S-1 before teting. Each barrier pecimen wa upported over the tructure laboratory floor, then, tied down to the floor uing 50 mm diameter threaded rod. Each rod wa placed @ 600 mm c-c and tightened by applying a pecified torque to control the lab uplift during teting. A 900- kn hydraulic jack wa ued to apply horizontal load to the barrier wall. A univeral flat load cell of 900 kn capacity wa ued to meaure the applied load on barrier model. SYSTEM 6000 data acquiition unit wa ued to record reading from all enor. Each pecimen wa teted under increaing monotonic load up-to-collape. During the tet, jacking load wa applied in increment 10 kn. At each load increment, the load wa maintained for a few minute to oberve crack initiation and propagation a well a change in barrier geometry a depicted from potentiometer (POT) reading. Failure of the model wa attained when the reading from enor were increaing while the model did not take any further increae in load. Specimen S-1 b) Specimen S-2 c) Specimen S-3 Fig. 7. View of internal reinforcement in the teted barrier pecimen Fig. 8. Tet etup howing the tie-down ytem for the deck Fig. 9. View of the tet etup for pecimen S-1 4. EXPERIMENTAL RESULTS 4.1 Specimen S-1 Figure 9 how the tet etup of pecimen S-1 before teting. While Fig. 10 how different view of the crack pattern of the teted barrier pecimen. In thi pecimen, the firt viible crack wa oberved in the front ide of the barrier wall at barrier-deck junction at 30 kn jacking load. Flexural crack were oberved in the deck lab at it fixed end and at the interection of the tapered portion of the front ide of the barrier wall at a jacking load of 60 kn. MAT-732-5

Thee flexural crack penetrated further at a higher load, along with other flexural crack in the deck lab. Although flexural crack appeared in the barrier wall and the deck lab portion, a udden diagonal tenion crack appeared in the deck lab at a jacking load of 80 kn on the right ide of the barrier and at a jacking load of 86 kn at the left ide of the barrier. Thee crack propagated further till the barrier could not take a jacking load beyond 100.42 kn. Given the width of the barrier of 900 mm and the height of the applied load over the top urface of the deck lab of 990 mm, the experimental reiting moment i calculated a 110.46 kn.m/m. Thi experimental reiting moment at the barrier deck junction i greater than the CHBDC deign value of 83 kn.m/m at the barrier-deck lab junction per Table 1 in thi report. Thi lead to a factor of afety of 1.33 in the deign of the propoed MTQ barrier wall at interior location a lited in Table 1 Table 1. Comparion of experimental failure load and CHBDC factored deign moment at barrier-deck junction S-1 S-2 S-3 Interior portion with cantilever lab and 300 mm bar pacing Exterior portion with cantilever lab and 150 mm bar pacing Interior with nondeformable olid lab and 300 mm bar pacing Experimental failure load, kn 100.42 131.83 117.02 Experimental failure load, kn/m 111.58 146.48 130.02 Experimental reiting moment, kn.m/m 110.46 145.02 128.72 CHBDC-2006 deign moment, kn.m/m 83.00 102.00 83.00 Factor of afety in deign (experimental failure moment/ CHBDC-2006 deign moment) 1.33 1.42 1.55 a) Crack pattern at the right ide of the pecimen b) Crack pattern at the left ide of the pecimen Fig. 10.View of crack pattern for pecimen S-1 after failure 4.2 Specimen S-2 Specimen S-2 i identical to S-1 except that the vertical bar at the front face of the barrier wall are doubled to repreent the exterior location of the barrier wall. Figure 11 how different view of the crack pattern of the teted barrier pecimen. In thi pecimen, the firt viible crack wa oberved in the front ide of the barrier wall at barrierdeck junction at 30 kn jacking load. Flexural crack were oberved in the deck lab at jacking load of 30 kn with other flexural crack appeared at higher load. Alo, flexural crack appeared at the interection of the tapered portion of the front ide of the barrier wall at a jacking load of 80 kn. Thee flexural crack penetrated further into the barrier thickne at a higher load. Flexural crack propagated into the barrier thickne at the barrier-deck junction with increae in applied load. However, a udden diagonal tenion crack appeared in the deck lab under the barrier wall at a load of 90 kn on the left ide of the barrier wall and at a load of 100 kn at the right ide of the barrier wall a depicted in Fig. 11. The recorded data in the data acquiition ytem howed that the barrier could not carry a jacking load beyond 131.83 kn. Given the width of the barrier of 900 mm and the height of the applied load over the top urface of the deck lab of 990 mm, the experimental reiting moment i calculated a 145.02 kn.m/m. Thi experimental reiting moment at the barrier deck junction i greater than the CHBDC deign value of 102 kn.m/m MAT-732-6

at the barrier-deck lab junction per Table 1 in thi report. Thi lead to a factor of afety of 1.42 in the deign of the propoed MTQ barrier wall at exterior location a lited in Table 1. a) Crack pattern at the left ide of the pecimen b) Crack pattern at the right ide of the pecimen Fig. 11. View of crack pattern for pecimen S-2 after failure 4.3 Specimen S-3 Specimen S-3 repreent a barrier connected to non-deformable concrete lab. The amount of vertical reinforcement at the front face repreent the cae of interior egment of the barrier wall. Figure 12 how different view of the crack pattern of the teted barrier pecimen. In thi pecimen, the firt viible crack wa oberved in the front ide of the barrier wall at barrier-deck junction at 60 kn jacking load. Thi obervation coincide with the jacking load of about 60 kn at the end of the lower traight line portion of the load-deflection relationhip for S-3. Flexural crack appeared at the interection of the tapered portion of the front ide of the barrier wall at a jacking load of 50 kn. Although flexural crack at the barrier-deck junction penetrated further into the barrier thickne at a higher load, udden concrete breakout appeared approximately at the embedded GFRP bar headed end at a load of 102 kn in each ide of the barrier wall. Thee concrete breakout crack extended toward the top urface of the olid lab and toward the back face of the barrier wall at higher load till the barrier could not aborb jacking load beyond 117.02 kn. Given the width of the barrier of 900 mm and the height of the applied load over the top urface of the deck lab of 990 mm, the experimental reiting moment i calculated a 128.72 kn.m/m. Thi experimental reiting moment at the barrier deck junction i greater than the CHBDC deign value of 83 kn.m/m at the barrier-deck lab junction per Table 1 in thi report. Thi lead to a factor of afety of 1.55 in the deign of the propoed MTQ barrier wall at interior location when it i rigidly connected to non-deformable concrete deck lab a lited in Table 1. a) View crack pattern at left ide of pecimen b) Cloe-up view of crack pattern at the right ide of pecimen Fig. 12. View of crack pattern for pecimen S-3 after failure MAT-732-7

5. LIMITATION OF USE OF CHBDC DESIGN TABLE FOR FACTORED APPLIED MOMENTS AT THE BARRIER-DECK JUNCTION When an errant vehicle collide with the concrete bridge barrier, the effect of the lateral impact force i ditributed in the barrier and the deck with diperal angle hown in Fig. 13. Thi led to the deign force pecified in CHBDC commentarie (CSA, 2006b). Thee force were obtained by finite-element analyi a tated in the literature, auming bridge cantilever of 1 m length. However, the bridge deck lab cantilever may be of different length which i the cae of lab-on-girder bridge or the barrier wall can be rigidly fixed to a non-deformable bae which i the cae of thick olid lab bridge, voided lab bridge and adjacent box beam bridge ytem. A uch, the change in the upport ytem of the barrier wall hould be invetigated to enure that it i properly addreed in the value of the deign factored applied moment at the barrier-deck junction due to equivalent vehicle impact force. The deign force pecified in CHBDC commentarie (CSA, 2006b) comprie moment and tenion force for the inner and end portion of the barrier. However, the Table doe provide any further information for deigner to obtain appropriate deign moment and tenile force for different barrier length and lab thicknee. The length of a barrier i conidered between two free end of the barrier or between expanion joint. The length of the barrier will affect the diperion angle of applied force hown in Fig. 13. A uch, finite-element computer modelling hould be conducted to examine the effect of barrier length and the change in deck lab thickne on the factored applied moment at the barrier-deck junction. Fig. 13. Tranvere load diperion at (a) interior portion and (b) end portion of barrier wall 6. PARAMETRIC STUDY, USING FINITE-ELEMENT COMPUTER MODELLING, ON BARRIER- DECK ANCHORAGE MOMENTS A parametric tudy wa conducted to invetigate the effect of key parameter on the barrier-deck anchorage force. The deign force conidered in thi tudy were bending moment and tenion force at the inner and end portion of the barrier hown in Fig. 13. The tudied parameter include the deck lab thickne, t, cantilever overhang length, L c, and the barrier length, L b. The aociated value for each parameter were taken a 175, 225, 275 and 350 mm for t ; 0, 0.5, 1.0, 1.5, and 2.0 m for L c; and 3, 4, 6, 8, 10, and 12 m for L b. Finite-element (FEA) model were created for the PL-3 barrier with tapered face. The general purpoe SAP2000 oftware wa ued to contruct threedimenional (3D) FE model to conduct linear elatic analyi of concrete barrier. The barrier wall and the deck lab were modeled uing hell element with ix degree of freedom at each node. The cantilever length, L c, of 0 repreent a fixed bae of the barrier wall, imulating non-deformable deck lab. Thi i the cae when the rigidity of the deck lab i ignificantly higher than that of the barrier wall. In addition, the minimum barrier length, L b, i uually conidered a 3m in practice. The maximum barrier length of 12 m wa taken into conideration in thi tudy becaue analyi howed that greater length would have inignificant effect on the ditribution of force in the inner portion of the barrier and no effect at the end portion of the barrier. On each barrier-deck FEA model, the lateral load repreenting vehicle impact loading were placed at the middle of barrier for inner portion and at the end of barrier for the end portion. Thee force were ditributed over pecified length at the given height per CHBDC. Then, the reultant force (i.e., moment and tenion force) were obtained at the bottom of the barrier for a unit length parallel to lateral load. For inner portion, the unit length wa conidered at the centerline of the applied load, wherea for the end portion, the unit length wa conidered at the end of barrier. The reult of thi FEA analyi were preented elewhere (Azimi et al., 2014). Repreentative reult of the developed equation for applied factored moment at the barrier-deck junction obtained from FEA are hown in Table 2 for PL-3 barrier conidered in thi paper. By correlating data obtained from FEA modeling with thoe pecified in CHBDC of 2006, it can be oberved that CHBDC factored deign moment are generally conervative for many cae of cantilever length and lab thicknee, wherea CHBDC underetimate thee value in other cae. Thi can be attributed to the fact that MAT-732-8

CHBDC equation were conducted uing a FEA barrier model with deck lab cantilever of 1 m length and a certain lab thickne that i not mentioned in the literature. However, the current tudy howed that both lab thickne and cantilever length have ignificant effect on applied factored moment at the barrier-deck junction, epecially for horter barrier. Table 2. Developed formula for factored deign moment at the barrier wall-deck interface (Azimi et al., 2014) M inner (kn.m) M end (kn.m) Fixed bae 132 1 Cantilever deck lab 100 Lb 2. 3t 0. 2 2. 83t L 1 143t Fixed bae 148 b 23 0. 7 0. 8 Lc Cantilever deck lab 1 1 14t Lb 2. 3t 2 0. 2 0. 7 2. 83t L 1 L 240t 25 Note: Formula are bet applicable for: 175mm t 350mm ; 0 L c 2. 0m ; t = overhang thickne (m); L c = cantilever length (m) ; L b = barrier length (m) M inner = moment in the inner portion of barrier; M end = moment in the end portion of Barrier. b 0. 7 c 3. 0m When applying the developed equation in Table 2 to the tudied pecimen S-3, it can be oberved that the factor of afety in deign (FOS) for the propoed barrier over non-deformable bae i 0.98 compared to 1.55 obtained baed on CHBDC-pecified factored applied moment hown in Table 1. A uch, it i recommended to increae the embedment length of the GFRP bar of the barrier wall reting over non-deformable deck lab from 165 mm to 195 mm. Reult from FEA modelling and reulting equation in Table 2 howed that the FOS for the propoed barrier wall intalled over deck lab cantilever increae with increae in deck lab cantilever length, for a given barrier length. Alo, it can be oberved that the FOS increae with increae of barrier length up to about 6 m, beyond which the increae in FOS i inignificant (Azimi et al., 2014). It hould be noted that for barrier length equal of greater than 6 m, the FOS range from 1.41 to 1.63 with increae in barrier length up to 12 m, compared to a FOS of 1.33, hown in Table 1 baed on the available CHBDC deign factored moment for pecimen S-1. Given the fact that the barrier length hould be long enough to promote two-way action of truck impact load ditribution, it i recommended to ue the propoed GFRP-reinforced barrier reting over deck lab cantilever, hown in Fig. 4 for barrier length greater than or equal 6 m. Thi concluion i alo applicable to the barrier wall exterior location given the fact that for barrier length equal of greater than 6 m, the FOS for exterior location range from 1.41 to 1.64 with increae in barrier length up to 12 m, compared to a FOS of 1.42, hown in Table 1 baed on the available CHBDC deign factored moment. L b 7. CONCLUSIONS Baed on the data generated from the experimental program and the finite-element modelling, it i recommended to ue the propoed MTQ GFRP-reinforced barrier ytem reting over 200-mm thick deck lab cantilever, hown in Fig. 4, for barrier length greater than or equal 6 m. Thi deign i acceptable for lab-on-girder bridge a well a multiple-pine bridge with deck lab thickne of 200 mm and cantilever length ranging from 0.5 to 2 m. It i alo recommended to ue the propoed MTQ GFRP-reinforced barrier ytem hown in Fig. 4 in bridge with nondeformable deck lab provided that the embedment length of the GFRP bar into the non-deformable deck lab increae from 165 mm to 195 mm. Given the dimenion of the non-deformable concrete bae upporting the barrier wall during tet, the propoed deign i acceptable for application in olid-lab and voided-lab bridge cro-ection with minimum deck thickne of 500 mm. MAT-732-9

ACKNOWLEDGEMENTS The author acknowledge the upport to thi project by Pultrall Inc. of Quebec, Canada. The continuou upport, commitment and dedication of Mr. Nidal Jaalouk, the enior technical officer of Ryeron Univerity, were an integral part of the experimental work reported in thi tudy. REFERENCES Azimi, H., Sennah, K., Tropynina, E., Goremykin, S., Lucic, S., and Lam, M. 2014. Anchorage Capacity of Concrete Bridge Barrier Reinforced with GFRP Bar with Headed End. ASCE Journal of Bridge Engineering, 19(9): 04014030 (1-15). CSA. 2006a. Canadian Highway Bridge Deign Code. CAN/CSA-S6-06. Canadian Standard Aociation, Toronto, Ontario, Canada. CSA. 2006b. Commentary on CAN/CSA-S6-06, Canadian Highway Bridge Deign Code. Canadian Standard Aociation, Toronto, Ontario, Canada. Khederzadeh, H., and Sennah, K. 2014. Development of Cot-Effective PL-3 Concrete Bridge Barrier Reinforced with Sand-Coated GFRP Bar: Static Load Tet. Canadian Journal for Civil Engineering, 41(4): 368-379. MASH. 2009. Manual for Aeing Safety Hardware, MASH. American Aociation of State Highway and Tranportation Official, Wahington. D.C. 276 page. MTQ. 2009. Manuel de conception de tructure, Chapter 16. Minitry of Tranportation of Quebec. Pultrall. 2013. V-Rod HM Data Sheet. Pultrall Inc. Quebec. Rotami, M. Development of GFRP-reinforced bridge barrier deck-lab ytem for utainable contruction. Ph.D. Diertation (expected December 2016), Civil Engineering Department, Ryeron Univerity, Toronto, Ontario. Sennah, K., and Khederzadeh, H. 2014. Development of Cot-Effective PL-3 Concrete Bridge Barrier Reinforced with Sand-Coated GFRP Bar: Vehicle Crah Tet. Canadian Journal for Civil Engineering, 2014, 41(4): 357-367. MAT-732-10