Seismic Behavior of Beam Lap Splices

The document begins by establishing the context of earthquake-resistant reinforced concrete structures and the challenges posed by large deformations during seismic events. It highlights the inevitable damage to concrete in potential plastic hinge regions. The performance of lap splices is directly linked to the surrounding concrete's condition; therefore, current building codes mandate their placement in areas maintaining concrete integrity. However, older structures might have lap splices located in now-recognized potential plastic hinge regions, leading to potential deficiencies. A particular deficiency is noted in early structures using plain round bars, which require double the anchorage length compared to deformed bars according to current design codes. The introduction sets the stage for investigating the seismic capacity of beam details, specifically focusing on the impact of lap splice placement and column depth on structural behavior in a case study of a building from the early 1960s. This case study involves plain round longitudinal bars lap-spliced within a potential plastic hinge region and a relatively small column depth compared to recent design code requirements. The study will analyze the effects of these design features on the specimen's overall performance.

This section reviews existing structures and pertinent research findings. Research projects conducted at the University of Canterbury Civil Engineering Laboratory tested specimens representing existing members designed according to earlier New Zealand concrete design codes. These specimens, using plain round bars for longitudinal reinforcement, were subjected to simulated seismic loading. The results revealed inadequate bond between the bars and the surrounding concrete, leading to significant softening of the lateral load-displacement relationship. A substantial reduction in strength and low energy dissipation capacity were observed. The measured structural ductility for these specimens ranged from 2 to 2.5. The section also discusses the adverse effect of increasing bar diameter on anchorage capacity, noting larger bond stress concentrations with larger bars. The text highlights studies which demonstrate that longer anchorage lengths can mitigate this issue, and that larger bar diameters lead to larger cracks, potentially causing splitting failure. The research points out that avoiding splitting anchorage failure is nearly impossible for bar sizes exceeding 50 mm. The impact of equal development and splice lengths under cyclic loading conditions is examined, drawing on studies by Gergely et al. which illustrate that splices with limited confinement fail when a bar loses anchorage due to stirrup yielding. In contrast, adequately confined splices redistribute forces before failure, although yielding of the spliced bars might still lead to concrete splitting. The section also presents Orangun et al.'s expression for bond strength, derived from regression analysis of published data. This analysis, however, was limited to specimens tested under monotonic loading and failed before main reinforcement yielded. The limit between pullout and splitting failure is noted as a cover distance to bar diameter ratio (c/db) of approximately 2.5. Finally, the role of concrete cover in confinement is analyzed, concluding that its influence under cyclic loading is less significant compared to monotonic loading, with a minimum clear cover of 1.5db deemed sufficient for load transfer.

The influence of transverse reinforcement on splice strength and failure modes is discussed in this section. Studies by Tepfers under monotonic loading show that stirrups or spiral transverse reinforcement enhance splice strength and modify failure modes. Splices without transverse reinforcement tend to fail in a brittle manner, whereas confining steel leads to a more gradual failure. Concentrating stirrups at the splice ends, where bond stresses are highest, proves more effective than uniform spacing. The research examines the behavior of compression lap splices under inelastic repeated loading in beam and column specimens designed to meet specific design suggestions. Most specimens failed due to concrete crushing outside the splice region after fulfilling initial performance criteria. This led to minor revisions in original design guidelines. The findings suggest these recommendations are suitable for tension and compression lap splices subjected to inelastic cyclic loads. The influence of shear forces is investigated, noting that a moment gradient minimally affects splice strength with moderate shear. However, splices in regions of high varying shear might not perform as well. Cornell University research examining splice behavior during inelastic cyclic loading included the influence of moderate shear levels. The presence of shear confines damage and yield penetration to one splice end, resulting in improved performance. Bond failure was absent in some specimens upon load discontinuation. Sufficient transverse steel to confine the splice was found to be at least four times greater than that required to resist nominal shear stress (up to 1.73 MPa). To counter dowel action, close stirrup spacing is recommended for a distance 'd' (effective beam depth) beyond the high-moment splice end. Earlier New Zealand codes (e.g., NZS 3101:1982) advocated for lap splice placement away from maximum tensile stress points, with increased lap lengths specified for high steel stress or when a large percentage of tensile reinforcement was spliced.

Research at the University of Canterbury's Civil Engineering Laboratory focused on testing specimens representing existing reinforced concrete members designed using earlier New Zealand concrete design codes. The research involved bridge piers and columns reinforced with plain round bars. These specimens were subjected to simulated seismic loading to assess their performance. A key finding was the inadequate bond between the plain longitudinal bars and the surrounding concrete. This deficiency resulted in significant softening of the lateral load-displacement relationship, particularly during the initial cycle, reducing the lateral load capacity to three-quarters of the ideal capacity. The specimens exhibited low energy dissipation capacity and a large strength reduction by the end of testing. The available structural ductility was measured to be between 2 and 2.5, indicating limited capacity to withstand deformation under seismic loads. These findings highlight the vulnerability of structures built to older codes where lap splice placement and the type of reinforcing bar were not as stringently regulated as in current design practices. The study serves as a crucial comparison to modern seismic design standards and emphasizes the need for improved methods to assess and retrofit older buildings.

The influence of bar diameter on anchorage capacity is a key theme within this section. It's noted that increasing bar diameter has a known adverse effect on bar anchorage, leading to higher bond stress concentrations due to the proportional relationship between cross-sectional area and surface area. To counteract this, longer anchorage lengths are needed for larger diameter bars. Larger diameter bars also result in larger cracks along the bar length, acting as points of weakness that can trigger splitting failures. Research indicates that preventing splitting anchorage failures is almost impossible for bars larger than 50mm. This section also delves into the relevance of equal development and splice lengths for cyclic loading conditions, citing research from Gergely et al. This research found that for splices with minimal confinement, failure occurs when one bar loses anchorage because of stirrup yielding. However, sufficient confinement leads to force redistribution prior to failure. Yielding of the spliced bars may lead to splitting of the surrounding concrete, emphasizing the importance of confinement in enhancing the performance of splices under cyclic loading. Orangun et al.'s work is also referenced in developing an expression for the bond strength of embedded and spliced bars using regression analysis; however, their findings are limited to monotonically loaded specimens failing prior to main reinforcement yielding, and they suggest a cover-to-diameter ratio (c/db) of approximately 2.5 as the boundary between pullout and splitting failure modes.

The role of concrete cover in providing confinement for lap splices is analyzed. It is noted that concrete cover plays a less significant role in confinement for cyclically loaded splices than for monotonically loaded ones. Extensive cover cracking before failure led Cornell University researchers to conclude that cover resistance is unreliable. Unless the cover prevents longitudinal splitting, it's not a significant factor in splice strength prediction models. A minimum clear cover of 1.5db is considered sufficient for load transfer. The research highlights that all beams subjected to cyclic loading with splices in constant moment regions experienced longitudinal bond splitting failures. These failures typically involve side and bottom cover cracking, which can lead to corner cover spalling and the loss of force transfer between spliced bars. Further research at Cornell University using reversed cyclic loading also showed an inability to prevent longitudinal bond splitting for splices in constant moment zones. Design recommendations are then made based on the number of cycles a splice can sustain in the inelastic range. This includes a minimum of 15 to 20 reversed load cycles beyond bar yield and a maximum one-off steel strain at least 2.5 times the yield strain. The research also suggests avoiding splices near ground level in first-story columns because of the high demands of plastic hinges in those locations. Earlier New Zealand codes of practice also recognized the need to place lap splices away from points of maximum tensile stress and increased lap lengths for high steel stress or when a high percentage of tensile reinforcement was spliced. NZS 3101:1982 further restricted lap splice placement in seismically designed members, prohibiting their placement within beam/column joint regions or within one effective depth of a potential plastic hinge region.

Specimen One was designed to meet the requirements of NZS 3101:1982, with the exception of lap splice positioning. The lap splices were intentionally placed in the bottom bars within the potential plastic hinge region of a beam, a deviation from the code's recommendations to place lap splices in regions where concrete integrity would be maintained. The construction process involved creating reinforcement cages using templates of the member cross-sections, securing wire ties for cage rigidity during lifting, and oiling plywood formwork to facilitate stripping. Spacers ensured correct concrete cover thickness. Inserts were included for attaching lifting brackets. Concrete was supplied from a ready-mix plant, placed with a skip, and compacted using an electrical vibrator. The specimen was covered with hessian and kept moist for seven days to allow for curing before formwork removal and preparation for testing. This meticulous construction ensured that the specimen’s dimensions and reinforcement were as specified, allowing for a controlled study of the effects of lap splice placement in the plastic hinge region. The method highlights a focus on precise construction to minimize extraneous variables affecting the outcome of the subsequent testing phase.

The methodology for measuring displacements and strains is described. Linear potentiometers measured member distortion and reinforcing bar movement. Internal threaded steel bars (10mm diameter) were placed through the reinforcement cage and embedded in the concrete. These bars protruded, allowing threaded rod to attach linear potentiometers for measurement. Short pieces of timber, polystyrene, or plastic hosing were used around the 10mm bars before concrete casting to prevent concrete bearing and affecting the readings during testing. Electrical resistance strain gauges monitored strains in both longitudinal and transverse reinforcement. Both sets of lapped bottom beam bars in the specimen were strain-gauged to minimize the loss of gauges before or early in the test. One of the four top beam bars, beam stirrups on each side of the column, half of the longitudinal column bars (to provide strain information at the beam faces), and selected joint shear steel and column transverse steel were also gauged. Strain gauge placement on the transverse steel avoided recording bending strains caused by the bowing of stirrups and hoops. Showa 120-ohm foil strain gauges (5mm gauge length) were used after careful preparation of the reinforcing steel (filing, polishing, cleaning) and multiple coats of waterproofing compound and mastic tape for added protection during concrete casting. This comprehensive approach provided detailed data on the response of various components of the specimen under seismic loading. The use of multiple gauges aimed at capturing and minimizing data loss, ensuring reliable strain measurements throughout the experiment.

Specimen One largely conformed to NZS 3101:1982, except for the lap splice location in the bottom bars of the potential plastic hinge region of a beam. The testing started with a load and displacement controlled cycle; however, a voltage discrepancy between the power supply settings impacted data reliability. Calibration equations for the datalogger were generated at a different voltage than the one used for the column top displacement calibration, which controlled the displacement-controlled loading. This led to displacements 24% larger than anticipated during the displacement-controlled loading cycles. Several cycles were completed before this inconsistency was noticed, but loading continued to maintain the ductility factor sequence. This highlights the importance of precise calibration and the potential impact of unexpected errors on the reliability of experimental results. The decision to continue testing despite the error was made to preserve the sequence of ductility factors, which might have been viewed as important for analysis of the specimen's response across a range of conditions, though it should be noted that the actual displacements experienced by the specimen were higher than expected, and this could have affected the test results.

The document details the observed failure mechanism of Specimen One. At a displacement of µ = -4/1, the concrete around the splice split apart, causing splice failure. Prior to this, cracking occurred along the entire splice, increasing beam depth by 7mm. The widening of flexural cracks and the formation of a main crack 150mm from the column face were observed. Concrete powdering was seen between the spliced bars, and at µ = -4/2, the concrete cover along the splice fell, exposing the bottom bars. At higher ductility factors (µ = ±6), large flexural cracks appeared in both beam's plastic hinge regions. Concrete above the splice expanded under compression during positive cycles, while large slip occurred between the spliced bars during negative cycles. Stirrups were pulled out of plane. Buckling indications and further concrete cover loss were also observed at these displacement levels. The loss of the splice capacity resulted in the pinching of the hysteresis loops, decreasing the energy dissipation capability. This section describes a comprehensive failure analysis, emphasizing the various stages of deterioration, from initial cracking to ultimate splice failure. The observed failure behavior correlates with the expected effects of lap splice placement within a potential plastic hinge region and provides evidence of the negative influence of this configuration on the structural behavior under seismic loading.

Specimen Two was constructed to replicate beam reinforcement details from a building designed in the late 1950s. The beam cross-section dimensions and longitudinal reinforcement (plain round bars) were identical to those in the original building. To prevent potential shear failure, additional transverse steel was included in the beams. The column dimensions were kept similar to the original to replicate bond conditions. The beam longitudinal steel areas and cross-section dimensions were identical to those in Specimen One, allowing for a direct comparison between the two specimens. The reinforcement cages were constructed using templates of the member cross-sections, and wire ties were used at frequent intervals to improve the cage's rigidity during lifting. Plywood formwork was oiled before casting to aid in stripping. Spacers at the bottom of the mold ensured correct concrete cover, and inserts were added for attaching lifting brackets. Concrete was supplied by a ready-mix plant, placed using a skip, and compacted using an electrical vibrator. The specimen was kept moist for seven days to cure before formwork stripping and preparation for testing. The replication of the original 1950s building elements within Specimen Two allowed the researchers to analyze the seismic performance characteristics of a typical beam design from that era, making it a valuable case study for evaluating the adequacy of older design practices relative to contemporary standards.

The measurement methods for Specimen Two were similar to those used for Specimen One. Linear potentiometers measured member distortion and reinforcing bar movement using internally threaded steel bars embedded within the concrete. Electrical resistance strain gauges monitored longitudinal and transverse reinforcement strains. Both sets of lapped bottom beam bars were strain-gauged to minimize gauge loss, along with one top beam bar, beam stirrups near the columns, half the longitudinal column bars, and selected joint shear and column transverse steel. Strain gauge placement on the transverse steel was designed to avoid recording bending strains from stirrup and hoop bowing. Showa 120-ohm foil strain gauges (5mm gauge length) were used after preparing and waterproofing the reinforcing steel. This consistent approach to data collection between Specimen One and Specimen Two facilitated direct comparison of the results and offered a comprehensive assessment of the stress distribution within the test specimens under various simulated conditions. The thorough application of multiple, strategically placed strain gauges aimed at collecting a comprehensive dataset for rigorous comparative analysis.

Specimen Two exhibited a lack of stiffness from the first loading cycle. Interstorey drift angles at half the theoretical lateral load (Vi) were 0.83% and 0.61% for positive and negative load directions respectively. The specimen's response was uneven, with the west beam showing no additional load during one negative loading increment, followed by stiffness loss after unloading. The load distribution between beams indicated negative load development in the east beam when the specimen returned from positive ductility factors, attributed to longitudinal bar slip during positive loading. The slipped bars and gaps at the column faces remained when the load was released, resulting in negative load resistance. A similar pattern was observed in the west beam load when returning from negative displacements. The beam carrying the greater load developed an inverted compression arch from the pinned end to the column face. The top longitudinal bars provided the horizontal reaction to balance the compression thrust. The vertical component of the arch opposed the shear force. The opposite beam carried minor loads as the concrete cover above the tensioned top bars compressed. This arching behavior indicated a shift away from typical beam action under load, which suggests bond failure in the longitudinal bars. The test rig restraints seemed to have improved the specimen's performance, highlighting the effect of boundary conditions on the observed structural response. The limited stiffness and uneven load distribution reflect the bond strength issues linked to the use of plain round bars and the inadequate joint depth.

Strain gauge data on Specimen Two revealed gradual tensile strain penetration through the joint into the opposite beam at +0.5Vi. Subsequent cycles showed stiffness loss as longitudinal bars slipped through the joint. At ±1, a clip gauge indicated yield strain exceedance in a top beam bar 300mm from its anchorage, indicating bond failure. The test became dominated by the test rig anchorage after the loss of bar bond, limiting the value of further results. The superior bond conditions for bars in compression due to the absence of cracks and end bearing were noted. Transverse steel was used to reduce development and splice lengths in tension; however, it couldn't maintain sufficient clamping force across the crack between the spliced bars. The results are considered tentative, influenced by the test rig's influence. Strains just below bar yield were recorded at ±0.75Vi. Slip accommodated beam rotations and top bar yielding; the lap splices contributed to the specimen's lack of stiffness. Without the lap splices, bond loss would likely have occurred in both sets of longitudinal beam steel, potentially increasing load resistance and reducing displacements. The rapid bond deterioration, resulting in bar slip, highlighted the issues associated with using plain round bars and insufficient anchorage in the context of seismic loading. The absence of visible signs of anchorage loss in the plain bars is notable, implying the energy required to break adhesion was insufficient to crack the surrounding concrete.

The conclusions summarize the key findings from the experimental testing of beam-column subassemblages. A beam-column subassemblage designed to meet NZS 3101:1982 but containing lap-spliced deformed bar reinforcement in a potential plastic hinge region showed low ductility under simulated seismic loading. Deterioration in the splices began during the initial negative displacement ductility factors, and splice failures resulted in a loss of beam and overall specimen load capacity. There's a potential for substantial lateral load capacity loss if the lapped spliced reinforcement constitutes a significant portion of the total tension beam reinforcement. The testing of Specimen Two, modeled on a 1950s building design with plain round bars, showed a rapid loss of bond strength and stiffness, underlining the vulnerability of older structures under seismic loads. The results from Specimen Two are considered tentative due to potential influence from the testing rig, meaning they are not fully representative of real-world behavior. The study showed how crucial appropriate lap splice design and placement are for achieving adequate seismic capacity in reinforced concrete structures. Specifically, using deformed bars, sufficient concrete cover, and adequate transverse reinforcement are highlighted as necessary components for effective seismic performance. This section effectively summarizes the main observations and outcomes of the experimental work, highlighting the implications of various design features on the seismic response of the test specimens.

The study concludes with recommendations for future research and improvements in testing methodology. Future testing of existing structures with plain longitudinal reinforcing bars needs to mitigate test rig influence on the results. The focus should be on providing suitable anchorage at the member ends to reflect real-world conditions. Using hooks to anchor bars, a method applicable in existing structures where hooks are already present but not in those with lapped bars, is suggested. When welding or bolting longitudinal bars to end plates for testing, researchers need to monitor the bars near the connections to determine the level of restraint. The study clearly points towards a need to improve testing methodologies to avoid results biased by the test apparatus. Future testing must better capture the actual behavior of structures, especially older ones using plain round bars, and account for boundary conditions that may affect results. The observations from Specimen Two, in particular, emphasize the need to account for the effect of the test rig on measured behavior.