Experimental behaviour and failure of beam-column joints with plain bars, low-strength concrete and different anchorage details

Engineering Failure Analysis xxx (xxxx) xxxx Contents lists available at ScienceDirect Engineering Failure Analysis journal homepage: www.elsevier.com/locate/engfailanal Experimental behaviour and failure of beam-column joints with plain bars, low-strength concrete and different anchorage details Cumhur Cosguna, , Ahmet Murat Turkb, Atakan Mangirc, Turgay Cosgund, Guven Kiymaze ⁎ a College of Information Technology and Engineering, Marshall University, WV 25755, USA Department of Civil Engineering, Istanbul Kultur University, Istanbul 34158, Turkey c School of Engineering and Natural Sciences, Istanbul Medipol University, Istanbul 34810, Turkey d Department of Civil Engineering, Istanbul University-Cerrahpasa, Istanbul 34320, Turkey e Department of Civil Engineering, Istinye University, Istanbul 34010, Turkey b A R TICL E INFO A BSTR A CT Keywords: Reinforced concrete Beam-column joint Anchorage detail FRP retrofit Low-strength concrete In framed structures, both steel and reinforced concrete, beam-column joints play a very crucial role in terms of seismic resistance. Under the effects of high lateral seismic loads, beam-column joints are subjected to high forces and moments and their behaviour have a significant influence on the response of the structure. Poor seismic performance of inadequately detailed joints can lead to the total or partial collapse of reinforced concrete frame structures. The use of low strength concrete, plain reinforcing bars, problematic anchorage details and inadequate transverse reinforcement in beam-column joints are the factors increasing the failure risk of the structures during severe earthquakes. In this paper, an experimental study on the cyclic behaviour of reinforced concrete exterior beam-column joints is presented. The study aims at investigating the effects of the longitudinal beam reinforcement anchorage detail on the joint performance and quantifying the level of contribution of retrofitting the joints by fiber reinforced polymer sheets (FRP). Three different details were considered in the test program including the longitudinal reinforcement of the beam being anchored within the joint with 90-degree hooks, 180-degree hooks and straight bar (no hook). All of the test specimens were produced using low strength concrete and plain bars to represent the conditions of joints of existing deficient reinforced concrete building structures. In the first series of tests, four 2/3 scale reinforced concrete beam-column joint specimens were tested by adopting a displacement controlled and quasi-static load application method to assess the performance of joints with the above-mentioned anchorage details. The load was applied in a reversed cyclic fashion. The second series of tests were carried out on two additional specimens with the same details as described above but strengthened using FRP sheets. The response of the specimens were evaluated and compared in terms of load-drift, displacement hysteretic behaviour. It was found out that the problematic anchorage details have a very significant adverse effect on the seismic performance of the joints. On the other hand, FRP retrofitting has resulted in a significant increase in peak loads and sustained ductility particularly for the specimens for which reinforcement slippage was not a governing mode of failure. ⁎ Corresponding author. E-mail address: cosgun@marshall.edu (C. Cosgun). https://doi.org/10.1016/j.engfailanal.2019.104247 Received 19 February 2019; Received in revised form 30 October 2019; Accepted 3 November 2019 1350-6307/ Published by Elsevier Ltd. Please cite this article as: Cumhur Cosgun, et al., Engineering Failure Analysis, https://doi.org/10.1016/j.engfailanal.2019.104247 Engineering Failure Analysis xxx (xxxx) xxxx C. Cosgun, et al. 1. Introduction Beam-column joints are one of the most critical regions of building structures subjected to seismic effects. Insufficient transverse reinforcement details, use of the plain bars and low strength concrete and problematic anchorage details in beam-column joints are quite common in relatively old existing buildings. These deficiencies have caused severe damages or partial/total collapse of structures during earthquakes [9,7,5]. Recent earthquakes reconnaissance has identified the failure in beam-column joints with inadequate transverse reinforcement. In these observations, it was easily noticeable that the joints contributed to partial or total building collapse [8,23]. Examples of building collapse due to the deficiencies related to joint regions are presented with photographs of collapsed buildings (Fig. 1) in Tabanli and Edremit earthquakes in Turkey, 2011. In developing countries, structures with these features are quite common. Therefore, a number of important studies have been conducted to better understand the behaviour of such joints and to improve their performance [1–4,6,13,14,16,19,24,25]. One of the first experimental studies on RC beam-to-column joints reported in the literature was by Hanson and Conner [15]. In this study, full size reinforced concrete beam-column joints were tested under simulated earthquake loading with the main purpose of investigating the suitability of recommended reinforcing details for use in earthquake resistant design. Test parameters that were adopted were column dimensions, column axial load, and amount of joint lateral reinforcement. Anchorage behavior of the longitudinal beam reinforcement within the beam-to-column joint region was studied by Marques and Jirsa [18], Soroushian et al. [22], and much later by Bedirhanoglu [5]. In Marques and Jirsa [18], the effects of column axial load, vertical column reinforcement, side concrete cover, and lateral reinforcement through the joint on the performance of standard hooked bars were studied whereas Soroushian et al. [22] focused on the effects of the anchored bar diameter, confinement of the joint, and compressive strength of concrete on the hook behavior. Bedirhanoglu et al. [5] tested full-scale exterior corner beam-to-column joints with plain reinforcing bars and low-strength concrete. The parameters investigated in this study included column axial load, displacement history, conditions of anchorage within the joint and amount of joint reinforcement. The test results showed that an increase in column axial load led to an increase in the dissipated energy and less pinching behavior of the hysteresis curves. With regards to the effect of displacement history it was found out that the effect was negligible. On the other hand, the use of transverse reinforcement in the joint resulted in greater energy dissipation capacity. The structural response of full-scale beam-to-column joints designed according to the pre-1970s codes was studied experimentally by Liu and Park [17] tested. The main test variables included the type of application in which the longitudinal beam bars were hooked in the joint core and the level of the column axial load. Specimens with plain bars displayed significantly lower stiffness and strength, less joint shear distortion but high opening of beam bar hooks in tension, and column bar buckling. Beam-to-column joints with insufficient amount of transverse reinforcement and poor anchorage details was tested by Pampanin et al. [20]. The experimental work was carried out on four exterior and two interior beam-column joints designed for gravity loads only. Two different types of anchorage solutions were considered; continuous reinforcement or lapped splices with hooked-end anchorage outside the joint region. A better global joint behavior was obtained for the specimens with lapped splices and hooked end anchorage. The effects of engineered cementitious composite (ECC) on the behaviour of RC exterior beam-column joints was experimentally investigated by Said and Razak [21]. The specimens were subjected to reverse cyclic loading under controlled deformation at the tip of the beam until failure. It was found out that at post cracking stages, the ECC joint showed significant improvement in the ultimate shear and moment capacities of the beam-column joint. Cyclic behavior of beam-to-column connections have also been investigated using numerical simulation. In Feng et al. [11] a 3D finite element analysis procedure for precast concrete beam-to-column connections subject to reversal cyclic loading was presented. Focus was given on two important modelling aspects including the use of a softened damage-plasticity model for concrete and the modification of the Menegotto-Pinto model with the aim of reproducing the behavior of reinforcing steel Fig. 1. Beam-column joint failure at Van Tabanli and Edremit earthquakes. 2 Engineering Failure Analysis xxx (xxxx) xxxx C. Cosgun, et al. bars particularly within the critical bond and anchorage regions. The numerical model was utilized to simulate a number of precast concrete beam-to-column connections exhibiting different failure modes. The authors suggested that their developed numerical modelling scheme provides an effective and efficient way to modelling the cyclic behavior of precast beam-to-column connections with good accuracy. In Feng and Xu [10], a new fiber beam-column element was proposed for predicting the cyclic responses of RC structures. The proposed element which can account for flexure–shear interaction and bond-slip effect was validated by physical cyclic tests on RC members. In the model, shear deformation is assumed to be uniformly distributed along the section and only resisted by concrete fibers. The bond-slip effect at the critical regions e.g. beam-to-column connection, is represented by using a modified stress–strain model for steel fibers. The proposed element is validated through numerical simulations of a number of RC structures under cyclic loading. Considering the overall results of the study, the authors suggest that the new element can be used as a reliable tool for analysis of RC structures. In a separate study by Feng et al. [12] which focused on the progressive collapse performance analysis of precast reinforced concrete structures, a numerical simulation framework for precast RC structures was developed. In the simulation models, which were produced by using the OpenSEES software, beam and column members were modeled by fiber frame elements and Joint2D element was used for modelling of the precast concrete beam-to-column connections. Based on these modelling assumptions, progressive collapse analysis of a 10-storey prototype precast RC structure was numerically carried out by static pushdown analysis and dynamic column-removal analysis. The analysis was focused on verifying the developed numerical framework as well as investigating the progressive collapse resisting mechanism of the structure. 2. Experimental study 2.1. Details of test specimens Six 2/3 scale reinforced concrete beam-column joint specimens were tested under quasi-static cyclic loading. Two of the specimens were retrofitted by using fiber reinforced polymer (FRP) sheets, and four of them were tested without the application of any kind of retrofit. The specimens represented different exterior beam-to-column joints with various reinforcement anchorage detailing found typically in existing reinforced concrete structures (Figs. 2–5). In all of the tested specimens, beam cross-section dimension is 150 mm × 200 mm; column cross-section is 150 mm × 150 mm and the length of the beam is 900 mm from the column face, and column height is 1200 mm. All dimensions and reinforcement details of the specimens are presented in Figs. 2–5. The specimens were supported at both ends of the column and load reversals were applied to the end of the beam (Fig. 6). Specimen reference names, concrete strength values, joint reinforcement size, anchorage type and column axial load ratio values are given in Table 1. The reference specimen (J1-REF) was designed according to Turkish Seismic Code (TSC-2007) and Building Code Regulation (TS500) to avoid failure under shear forces. Hence the expected failure mode for this reference specimen is the failure of the beam cross-section by pure flexural yielding. Other specimens, namely, J2, J3 and J4, represent joint conditions with different anchorage details typically found in existing reinforced concrete structures. J2 specimen includes anchorage with 90-degree hooks, J3 includes 180-degree hooks whereas J4 includes no hooks, i.e. straight bar in the joint core. On the other hand, J2-R and J3-R specimens are the specimens that were retrofitted by externally wrapping FRP sheets around Fig. 2. Reference specimen (J1-REF). 3 Engineering Failure Analysis xxx (xxxx) xxxx C. Cosgun, et al. Fig. 3. Anchorage type of J2. Fig. 4. Anchorage type of J3. their joint regions. These two specimens incorporate all the geometry, material and reinforcement detail features that were adopted for the un-retrofitted J2 and J3 specimens. Hence the only difference is that they were retrofitted by FRP sheets. Retrofitting was carried out by wrapping three plies of carbon FRP sheets which were bonded diagonally over the external face of the joints. Before the FRP application, corners of the joint region were rounded to a radius of 25 mm and the concrete surfaces were prepared by cleaning and covering with an epoxy-based primer layer, to obtain ideal surface conditions for FRP-concrete bonding. Then, the 80 mm wide and 1550 mm long continuous carbon FRP sheets were bonded to the joint regions by using a double component epoxy adhesive. The level of column axial load, which was kept constant during the tests as 20 kN, corresponded to an axial compressive stress level of 0.1f'cbh, where f'c is the cylinder compressive strength, b is the width and h is the height of the column cross-section. Axial load ratios ν are calculated using Eq. (1), where b and h are the width and depth of the cross-section of the column and P is the axial load. = P bhfc (1) 4 Engineering Failure Analysis xxx (xxxx) xxxx C. Cosgun, et al. Fig. 5. Anchorage type of J4. Fig. 6. General view of the test setup. Table 1 Specimen details. Specimens fc , MPa Joint reinforcement Anchorage detail υ, Axial load ratio, % J1-REF J2 J3 J4 J2-R J3-R 9 9 9 9 9 9 3 8 No No No No No 90-degree hooks 90-degree hooks 180-degree hooks No hook 90-degree hooks 180-degree hooks 10% 10% 10% 10% 10% 10% 3. Material properties In order to represent the actual conditions of relatively older buildings, the concrete poured during the construction of the specimens was designed to have low strength. The concrete mix proportions are given in Table 2. The relatively high water/cement 5 Engineering Failure Analysis xxx (xxxx) xxxx C. Cosgun, et al. Table 2 Concrete mixture proportions (kg/m3). Concrete Aggregate size (mm) Cement Water Sand Crushed stone Coarse aggregate High-range water-reducing admixture Normal-Weight 8 165 231 563 851 271 1.8 ratio of the concrete mixture (1.17), resulted in a 180-day cylinder (150 mm diameter and 300 mm high concrete cores) compressive strength of 9 MPa. The average modulus of elasticity of the concrete was obtained as 13000 MPa. These values are means of results from six standard cylinder test results. Plain bars with 14 mm diameter were used as longitudinal reinforcement both in column and beam. For the transverse reinforcement that was only used in the joint region of J1-REF specimen, 8 mm diameter ribbed bars were used. The ratios of the longitudinal reinforcements of the columns and beams are 2.7% and 2.0%, respectively. The spacing between the transverse reinforcements of columns and beams are 100 mm. Towards the support regions of the members, the spacing was decreased to 50 mm (Figs. 2–5). The mechanical characteristics of the longitudinal and transverse reinforcements are given in Table 3. In this Table, fy is the yield stress, fs,max is the tensile strength of steel, fsu is the rupture stress of steel, Es is the elastic modulus of steel, y is the yield strain, s,max is the strain corresponding to maximum stress, and su is the rupture strain for steel. The values given in Table 3 are average readings of the test values for five tensile tests. Obtained yield stresses for 14 mm and 8 mm diameter bars varied between 258 and 284 MPa and 382 and 552 MPa, respectively. As for the FRP material used for retrofitting, the unidirectional carbon FRP sheets used for joint retrofitting had a modulus of elasticity of 240 GPa, a tensile strength of 3800 MPa, single ply thickness of 0.176 mm and a rupture strain of 1.55%. The twocomponent epoxy used for bonding the FRP sheets to the concrete surface had a compressive strength of 80 MPa, a direct tensile strength of 12 MPa and an ultimate tensile strain of 1.6%. These values were all indicated by the FRP sheet manufacturer. 4. Test setup and loading pattern The experimental study included the cyclic testing of six 2/3 scale reinforced concrete exterior beam-column joint specimens as described above. The tests were carried out to assess the performances of different anchorage details and to investigate and quantify the effects of FRP retrofitting on the joint behaviour. The constant axial load (20 kN) on the column was applied using hydraulic jack as shown in Fig. 6. These specimens were designed to study the effect of anchorage detailing on the overall exterior beam-column joint behaviour in terms of load-drift/displacement relationship and failure mechanisms. A schematic of the test setup used in the study is shown in Fig. 6. As shown in this figure, the beam member was arranged to be in the horizontal plane, and the column member was in the vertical plane. All tests were conducted under displacement control. The specimens were supported at both ends of the column, and a 100 kN capacity servo-controlled hydraulic actuator was used to apply quasi-static vertical displacement at the beam end (Fig. 6). Constant axial load corresponding to 10% of the column axial load capacity was applied on the column by using a hydraulic jack. The displacements and deformations were measured using linear variable displacement transducer (LVDT) and strain gauges. In each specimen, eleven linear variable displacement transducers (LVDT) and eight strain gauges were used. The load cell in the actuator arm measured the applied load, and the LVDT was connected to the specimen to monitor the displacements. The details of the measurement system are presented in Fig. 6. The loading pattern consisted of a simple history of the reversed symmetric cycles of increasing displacement amplitude in specified steps (Fig. 7). This loading pattern permits the evaluation of parameters such as cyclic strength deterioration, stiffness degradation, ductility, energy dissipation, failure mode and crack propagation, which are essential in understanding the seismic performance of the structures. 5. Test results and discussions 5.1. Behavior of un-retrofitted specimens The applied total load versus drift-displacement response results of the test specimens without retrofit application (J1-REF, J2, J3, J4) are presented in Fig. 8 and test results for these specimens are compared in Fig. 9. As can be seen, the hysteretic load-driftdisplacement relationships of the specimens are mainly characterised by significant pinching in unloading and reloading branches. The pinching behaviour observed on the hysteretic curves of the specimen is mainly due to slip of beam longitudinal bars within the joints caused by crashing of weak concrete. This type of behaviour was verified with large vertical cracks over the beam-column Table 3 Mechanical properties of reinforcing bars. Reinforcement 14 8 Diameter (mm) fy, MPa 14 8 276 445 fs,max , MPa y= fy/Es 0.0013 0.0021 420 625 6 s,max 0.12 0.18 fsu, MPa 275 540 su 0.35 0.26 Es, MPa 212,000 209,000 Engineering Failure Analysis xxx (xxxx) xxxx C. Cosgun, et al. Fig. 7. Displacement pattern applied to test the specimens. Fig. 8. Load-drift and displacement response of un-retrofitted specimens. Fig. 9. Comparison of the load-displacement relationships for J1-REF, J2, J3 and J4. 7 Engineering Failure Analysis xxx (xxxx) xxxx C. Cosgun, et al. Fig. 10. Damage pattern of the J1-REF specimen (2%, 4%, 7%). interface regions. Figs. 12–14 present the observed damage mechanisms for the vertical cracks for all the specimens tested. The pinched hysteretic load-displacement relationship of the specimen J1-REF is mainly due to shear damage within the joint region rather than the slip of beam longitudinal bars. To compare the responses of four specimens; cyclic load-drift, displacement responses are plotted together in Fig. 9. In Figs. 10–15, damage photographs and damage maps of the un-retrofitted specimens for the 2–7% drift range in pushing and pulling directions are also presented. As shown in Fig. 9, compared with J1-REF curve, significantly decreased load capacity values were obtained for other specimens with different hook configurations and lacking transverse reinforcement in the joint region. As mentioned earlier, the load capacity of the specimens was mostly controlled and limited by slipping of longitudinal reinforcements. None of the specimens could reach its nominal beam flexural capacity except the reference J1-REF specimen. All specimens sustained their load capacities during cycles up to maximum drift ratios of approximately 4%. After 4% drift ratio, the joint shear deformations increased, and the load capacity of the specimens started to drop for J2, J3 and J4 specimens whereas the J1-REF specimen sustained the load capacity up to 7% drift. Initial cracks that took place on the J1-REF specimen started forming at 0.5% drift ratio and were in the form of bending cracks at the beam support region and inclined shear cracks at the beam-column joint core. At 1% drift ratio, these cracks were more apparent (Fig. 10). When 2% drift ratio was reached, the width of the diagonal shear cracks was around 2.3 mm. At 4% drift ratio, the width of the same cracks increased to 2.8 mm in the pushing and 4 mm in the pulling directions. At 6% drift-ratio, crushing of concrete at the beam-column joint core was observed and spalling of concrete cover initiated at 7% drift ratio (Fig. 10). Load-drift and displacement response of the J1-REF specimen is given in Fig. 8(a). The peak load capacities of the specimen were found to be 12.97 kN at 5% drift in the pushing direction and 12.84 kN at 6% drift in the pulling direction. Compared with the theoretical, applied force corresponding to the bending capacity value of the specimen J1-REF (13 kN-indicated with the red horizontal line in Fig. 8a) very close values were obtained in the test. The first cracks that were observed on the J2 specimen started with the formation of the bending cracks at the beam support section and inclined shear cracks at the beam-column joint core at 0.1% drift ratio. At 1% drift ratio, the width of the beam bending cracks was 1.2 mm in the pulling, and 0.6 mm in the pushing direction. At 2% drift ratio, the diagonal shear cracks increased at the joint core (Fig. 11). The width of the largest shear crack was 1.9 mm in pushing and 1.8 mm in pulling directions. At the following 3% drift ratio, crushing of concrete at the joint core was apparent. At 4% drift ratio, the crack width at the beam support section was 5 mm in pulling, and 4 mm in pushing directions (Fig. 11). At 7% drift ratio, the width of the same crack increased to 12 mm in the pushing and 11 mm in the pulling directions (Fig. 11). 8 Engineering Failure Analysis xxx (xxxx) xxxx C. Cosgun, et al. Fig. 11. Damage pattern of the J2 specimen (2%, 4%, 7%). Additionally, shear cracks were observed to increase at the joint core, and the largest crack width was 3.1 mm in pulling direction. Load-drift and displacement response of the J2 specimen is given in Fig. 8(b). The peak load capacities of the specimen were obtained as 8.84 kN at 3% drift in the pushing, and 8.63 kN at 4% drift in the pulling directions. The load capacity of the specimen could not reach the 13 kN theoretical load level (indicated with red the horizontal line in Fig. 8b) corresponded to the bending capacity of the beam support section. The significant slip of the beam longitudinal bars controlled the behaviour of the specimen. The load capacity was reduced by 32% of the load capacity corresponding to the calculated beam bending capacity. However, due to the pseudo ductility introduced to the system with the slipping of the beam bars, this load level could be maintained even up to 7% drift. As for the J3 specimen, at 0.5% drift ratio, first cracks started forming at the beam support section both in pulling and pushing directions. At 1% drift ratio, the width of these cracks was 0.9 mm in pulling, and 1.3 mm in pushing directions. Additionally, the shear cracks of the joint core increased, and the largest crack width was measured as 0.7 mm. At 2% drift ratio, the crack width of at the beam support section was 2.4 mm in pulling, and 0.9 mm in pushing directions (Fig. 12). Additionally, the largest shear crack width on the joint core reached around 1 mm. At 4% drift ratio, this shear crack width was more than 4 mm (Fig. 12). After 5% drift ratio, the crack width could not be measured (Fig. 12). Load-drift and displacement response of the J3 specimen is shown in Fig. 8(c). The maximum load capacity of the specimen was obtained as 7.51 kN with 3% drift in the pushing, and 7.78 kN at 2% drift in the pulling directions. The load capacity of the specimen could not reach the 13 kN theoretical load level (as is indicated with red the horizontal line in Fig. 8c) that corresponded to the bending capacity of the beam support section. The significant slip of the beam longitudinal bars controlled the behaviour of the specimen. The load capacity was reduced by 40% of the load capacity corresponding to the calculated beam bending capacity. For the J4 specimen, first crack widths were measured as 0.5 mm in pushing direction at the beam support section around 0.5% drift ratio. Additionally, at 1% drift ratio, the width of this crack reached 1.7 mm. At 1.5%, 2%, and 3% drift ratios, width values for the same cracks were measured as 2.7 mm, 3.1 mm, and 6 mm in pushing, and 1.5 mm, 2 mm, and 3.3 mm in pulling directions, respectively (Fig. 13). The shear crack width was 0.1 mm in pushing, and 0.6 mm in pulling directions and the cracks formed in the joint core at 4% drift ratio. After 4% drift ratio, the crack widths were not measured (Fig. 13). Load-drift and displacement response of the J4 specimen is given in Fig. 8(d). The peak load capacities of the specimen were obtained as 5.86 kN at 3% drift in the pushing, and 6.08 kN at 4% drift in the pulling directions. The load capacity of the specimen could not reach the 13 kN theoretical load level (indicated with red the horizontal line in Fig. 8d) that corresponded to the bending capacity of the beam support section. The significant slip of the beam longitudinal bars controlled the behaviour of the specimen. The load capacity was reduced by 53% of the 9 Engineering Failure Analysis xxx (xxxx) xxxx C. Cosgun, et al. Fig. 12. Damage pattern of the J3 specimen (2%, 4%, 7%). load capacity corresponding to the calculated beam bending capacity. Table 4 presents a summary of the findings as mentioned earlier for the four un-retrofitted beam-to-column joint specimens. In order to compare the response of the tested specimens, load-drift and displacement hysteresis are plotted together for all the specimens in Fig. 9. As can be seen in these figures, the load capacities obtained for the specimens with different anchor details (90degree hooks, 180-degree hooks and straight bar-no hook) are significantly lower than the reference specimen. The behaviour of J2 and J3 was dominated by the slip of the beam longitudinal reinforcement and beam-column joint shear deformations. For the J4 specimen, on the other hand, it was only the slip of the beam longitudinal reinforcement that dominated the behaviour. Consequently, none of the specimens except the reference specimen could reach the section bending capacity. The reductions from the theoretical section bending capacity for J2, J3 and J4 specimens were calculated as 31.32%, 41.65% and 54.48% in the pushing and 32.27%, 38.94% and 52.28% in the pulling directions, respectively. 5.2. Behavior of FRP retrofitted specimens As previously mentioned, the second series of tests were carried out on two additional specimens with the same details as described above but retrofitted using fiber reinforced polymer (FRP) sheets. Specimens designated as J2-R and J3-R were retrofitted by externally wrapping FRP sheets around their joint regions. These two specimens incorporate all the geometry, material and reinforcement detail features that were adopted for the un-retrofitted J2 and J3 specimens. Hence the only difference is that they were retrofitted by FRP sheets. No retrofit was applied for J4 type joint. Based on the observed behaviour of the un-retrofitted J4 specimen with a premature bar slippage mode of failure, the contribution of an FRP retrofit was deemed unnecessary. The hysteretic load-drift-displacement relationships of the FRP retrofitted specimens are presented in Fig. 14(a-b). For the J2-R specimen, it was observed that the specimen failure was concentrated over the beam cross-section manifesting itself as a flexural yield type of behaviour. In line with this observed behaviour, nominal flexural yield capacity was achieved for the J2-R specimen. As for the J3-R specimen, failure was prevented within the joint core region. Compared with the un-retrofitted J3 specimen, a 22% increase was achieved in the peak load sustained; however beam nominal flexural yield capacity could not be achieved since the failure was ultimately governed by the slip of reinforcements. In Fig. 15, the comparison of the force-drift and displacement relationships of reference (J1-REF) and retrofitted (J2-R and J3-R) specimens is presented. Concerning the peak load achieved, an approximately 16% increase in noted for the J2-R specimen in comparison to the J1-REF specimen. On the other hand, the J3-R specimen could not achieve the peak load achieved by the J1-REF. The reinforcement slipping that occurred in the test for the J3-R specimen resulted in a 10 Engineering Failure Analysis xxx (xxxx) xxxx C. Cosgun, et al. Fig. 13. Damage pattern of the J4 specimen (2%, 4%, 7%). Fig. 14. Force-drift and displacement response of retrofitted specimens. drop, in the load capacity of around 43% of the peak load obtained for the J1-REF specimen. Damage patterns observed in the tests for the FRP retrofitted specimens are presented in Fig. 16 for different drift ratios. 5.3. Energy dissipation behavior As discussed earlier, the hysteretic load-displacement relationships obtained for the un-retrofitted specimens are mainly characterised by significant pinching in unloading and reloading branches. On the other hand, for the FRP retrofitted specimens (J2-R and 11 Engineering Failure Analysis xxx (xxxx) xxxx C. Cosgun, et al. Fig. 15. The comparison of the force-drift and displacement relationships of reference (J1-REF) and retrofitted (J2-R and J3-R) specimens. Table 4 Summary of key results for the un-retrofitted specimens. Obtained result Loading direction Peak load obtained (kN) Pulling Pushing Pushing Pulling Drift corresponding to peak load (%) Average percentage difference between beam section capacity and test peak load (%) Specimens J1-REF J2 J3 J4 12.97 12.87 5 7 0.1 8.84 8.63 3 4 31.8 7.51 7.78 3 2 40.3 5.86 6.08 3 4 53.4 J3-R), the hysteretic curves exhibited relatively less pinching behaviour. These findings with regards to the characteristics of the hysteretic curves for un-retrofitted and FRP retrofitted specimens are in line with the energy dissipation behaviour under the complete cyclic loading. Fig. 17 presents the cumulative dissipated energy values for all the test specimens for 13 successive cyclic loading steps among which the last loading step is the step where the highest energy dissipation is achieved. Comparing the energy dissipation levels calculated at this last loading step, it is noted that the highest level is achieved for the FRP retrofitted J3-R specimen (with 180° hook anchorage and no joint reinforcement). The difference in total energy dissipation for the J3 and J3-R specimens is around %43 and for the J2 and J2-R specimen is around %49. These results are indicative of the significant contribution of applying FRP sheet retrofit over deficient beam-to-column joints. 6. Conclusions Exterior beam-column joints of substandard buildings generally suffer from the use of insufficient or no transverse reinforcement, low quality of concrete and problematic beam bar anchorage details. Observations made after major destructive earthquakes indicate that, in some cases, the weakness of the joint may significantly contribute to the partial or total collapse of the buildings. In this study, an experimental study on the cyclic behaviour of reinforced concrete exterior beam-column joints was carried out to investigate the effect of the longitudinal beam reinforcement anchorage detail on the joint performance under cyclic loading conditions. Three different details were considered in the test program including the longitudinal reinforcement of the beam being anchored within the joint with 90-degree hooks, 180-degree hooks and straight bar (no hook). Apart from the above-described specimens with various hook anchorage conditions, a reference specimen was designed and prepared following the Turkish Seismic Code (TSC-2007) and tested for benchmarking purposes. All of the test specimens were produced using low strength concrete and plain bars to represent the conditions of joints of existing deficient reinforced concrete building structures. Reinforced concrete beam-column joint specimens (2/3 scale) were tested adopting a displacement controlled and quasi-static load application method to assess the performance of joints with the above-mentioned anchorage details. The load was applied in a reversed cyclic fashion. The response of the specimens were evaluated and compared in terms of load-drift, displacement hysteretic behavior. On the other hand, the effectiveness of retrofitting such beam-column joints with Fiber Reinforced Polymer (FRP) composite sheets is demonstrated through testing of two additional specimens with similar substandard characteristics. Based on the aforementioned experimental investigation, the following conclusions can be drawn: • It was demonstrated that the anchorage details have a very significant effect on the seismic performance of the joints both in terms of joint capacity and mode of failure. • Combined with other adverse effects of low strength concrete and plain reinforcement bars, load-carrying capacities of the tested specimens decreased to levels much lower than the code calculated nominal capacity values. The • behavior of the specimens was mostly dominated by the slip of longitudinal reinforcements within the beams which has resulted in premature failure of the joint core region, and hence it was not possible for the specimens to reach the flexural capacity 12 Engineering Failure Analysis xxx (xxxx) xxxx C. Cosgun, et al. Fig. 16. Damage pattern photographs for the FRP retrofitted specimens at different drift ratios. Fig. 17. Cumulative dissipated energy values for all specimens. of the beams. • Beam flexural yield capacity was only achieved for the J1-REF specimen for which the reinforcement was designed according to valid ultimate limit state design rules. • Under cyclic loading, maximum drift ratio that the specimens could sustain was approximately 4% except for the J1-REF specimen that could sustain a drift level of around 7%. • The application of FRP sheet retrofitting for the substandard specimens with various joint core anchorage details resulted in significant improvements in the cyclic behavior in terms of strength and ductility as well as energy dissipation capacity. It was 13 Engineering Failure Analysis xxx (xxxx) xxxx C. Cosgun, et al. observed that in the FRP strengthened specimens the mode of failure was transformed from a dominant joint core region failure to a more beam flexural yielding mode concentrated away from the core region. Appendix A. Supplementary material Supplementary data to this article can be found online at https://doi.org/10.1016/j.engfailanal.2019.104247. References [1] U. Akguzel, S. Pampanin, Effects of variation of axial load and bidirectional loading on seismic performance of GFRP retrofitted reinforced concrete exterior beam-column joints, ASCE J. Compos. Constr. 14 (1) (2010) 94–104. [2] F. Alemdar, H. Sezen, Shear behaviour of exterior reinforced concrete beam-column joints, Struct. Eng. Mech. 35 (1) (2010) 123–126. [3] T. Amoury, A. Ghobarah, Seismic rehabilitation of beam-column joints using GFRP sheets, Eng. Struct. Eng., Elsevier Science Ltd. (2002) 1397–1407. [4] C.P. Antonopoulos, T.C. Triantafillou, Experimental investigation of FRP-strengthened RC beam-column joints, ASCE, J. Compos. Constr. 7 (1) (2003) 408–416. [5] I. Bedirhanoglu, A. Ilki, S. Pujol, N. Kumbasar, Behavior of deficient joints with plain bars and low-strength concrete, ACI Struct. J. 107 (3) (2010) 300. [6] G.M. Calvi, G. Magenes, S. Pampanin, Relevance of beam-column joint damage and collapse in RC frame assessment, J. Earthquake Eng. 6 (1) (2002) 75–100. [7] C. Cosgun, Seismic retrofit of joints of 3D reinforced concrete frames with FRP (In Turkish)”, PhD Dissertation, Istanbul Technical University, Turkey, 2014. [8] C. Cosgun, A.A. Dindar, E. Seckin, Y.H. Onen, Analysis of building damage caused by earthquake in eastern Turkey, GRADEVINAR 8 (2013) 743–752. [9] C. Cosgun, C. Mustafa, C. Demir, A. İlki, Seismic retrofit of joints of a full-scale 3D reinforced concrete frame with FRP composites, J. Compos. Constr., ASCE 23 (2) (2019). [10] D.C. Feng, J. Xu, An efficient fiber beam-column element considering flexure–shear interaction and anchorage bond-slip effect for cyclic analysis of RC structures, Bull. Earthq. Eng. 16 (11) (2018) 5425–5452. [11] De-Cheng Feng, Wu. Gang, Lu. Yong, Finite element modelling approach for precast reinforced concrete beam-to-column connections under cyclic loading, Eng. Struct. 174 (2018) 49–66. [12] D.C. Feng, Z. Wang, G. Wu, Progressive collapse performance analysis of precast reinforced concrete structures, Struct. Des. Tall Special Build. 28 (5) (2019) e1588. [13] G. Genesio, R. Eligehausen, S. Pampanin, Application of post-installed anchors for seismic retrofit of RC beam-column joints: design method, 9th Pacific Conference on Earthquake Engineering, (2011). [14] A. Ghobarah, A. Said, Seismic rehabilitation of beam-column joints using FRP laminates, J. Earthquake Eng. 5 (1) (2001) 113–129. [15] N.W. Hanson, H.W. Conner, Seismic resistance of reinforced concrete beam-column joints, J. Struct. Divis. ASCE ST5 (1967) 533–560. [16] A. Ilki, I. Bedirhanoglu, N. Kumbasar, Behaviour of FRP-Retrofitted joints built with plain bars and low-strength concrete, ASCE J. Compos. Constr. 15 (3) (2011) 312–326. [17] A. Liu, R. Park, Seismic behaviour and retrofit of pre-1970's as-built exterior beam-column joints reinforced by plain round bars, Bull. New Zealand Natl. Soc. Earthquake Eng. 34 (1) (2001) 68–81. [18] J.L.G. Marques, J.O. Jirsa, A study of hooked bar anchorages in beam column joints, ACI J. 72 (18) (1975) 198–209. [19] D.F. Meinheit, O. Jirsa, Shear strength of reinforced concrete beam-column connections, ASCE J. Struct. Eng. 107 (ST11) (1981) 2227–2244. [20] S. Pampanin, Alternative performance-based retrofit strategies and solutions for existing RC buildings, Seismic Risk Assessment and Retrofitting, Springer, Dordrecht, 2009, pp. 267–295. [21] S.H. Said, H.A. Razak, Structural behaviour of RC engineered cementitious composite (ECC) exterior beam-column joints under reversed cyclic loading, Constr. Build. Mater. 107 (2016) 226–234. [22] P. Soroushian, K. Obaseki, M. Nagi, M.C. Rojas, Pullout behaviour of hooked bars in exterior beam-column connections, ACI Struct. J. 85 (S28) (1988) 269–276. [23] M. Tapan, M. Comert, C. Demir, Y. Sayan, K. Orakcal, A. Ilki, Failures of structures during the October 23, 2011 Tabanli (Van) and November 9, 2011 Edremit (Van) earthquakes in Turkey, Eng. Fail. Anal. 34 (2013) (2013) 606–628. [24] A.G. Tsonos, I.A. Tegos, G.G.R. Penelis, Seismic resistance of type 2 exterior beam-column joints reinforced with inclined bars, ACI Struct. J. 89 (S1) (1992) 3–12. [25] A.G. Tsonos, Effectiveness of CFRP-jackets and RC-jackets in post-earthquake and pre-earthquake retrofitting of beam–column sub assemblages, Eng. Struct. 30 (3) (2008) 777–793. 14