The effect of concrete shear key on the performance of segmental columns subjected to impact loading

Abstract

Conventional precast segmental columns with seismic resistance design consist of only flat concrete segments with prestress tendon. This is because friction between adjacent segments is sufficient to resist the lateral forces from earthquake-induced actions. However, the friction between segments is not necessarily sufficient to resist lateral impact loads such as vehicle impact the column might experience during its service life. This article investigates the effectiveness of using concrete shear key in segments of precast segmental column in resisting the lateral impact loading. The precast reinforced concrete segments were designed with concrete shear keys to improve the column shear resistance capacity and minimize the relative displacement between adjacent segments. Two groups of segmental columns with and without shear key were designed and tested using a pendulum impact system. The effectiveness of shear key in resisting lateral impact loads was analysed by comparing the performance of the two groups of segmental columns. The testing results revealed that by introducing concrete shear key to segmental column, the relative displacement between adjacent segments could be effectively reduced. However, the large concrete shear key increased stress concentration in the concrete segments. Relatively, more severe damages to concrete segments were found on the columns with shear key. Further improvements on shear key designs should be made for better performance of segmental columns against impact loading.

Keywords

impact loading prefabrication reinforced concrete segmental column shear key

Introduction

Precast segmental columns have been more and more popularly used in construction in recent years. This is primarily because construction with prefabricated structural members could significantly improve on-site efficiency, reduce construction cost, minimize environmental impact and enhance construction quality. Moreover, application of new materials, for example, fibre-reinforced concrete and ultra-high strength concrete, which often requires careful mixing and/or curing under elevated temperatures, could become more feasible for precast segmental columns as they are prefabricated in factories. Segmental columns have been used in constructions of buildings and bridges since ancient times. Many iconic structures around the world today were built hundreds or even thousands of years ago with segmented stone pillars. Numerous modern constructions have also successfully implemented precast segmental concrete columns. Examples of building and bridge structures using segmental columns or piers can be found in Sprinkel (1985) and Ou (2007). Most of those applications are in low-seismicity regions. Recently, intensive studies of the performances of segmental columns under cyclic loadings have been reported for their possible applications in high-seismicity regions (Chou et al., 2013; Guerrini et al., 2014; Ichikawa et al., 2015; Kawashima et al., 2012; Motaref et al., 2013; Nikbakht et al., 2015; Ou et al., 2013; Wang et al., 2014). However, very few studies on the behaviour of segmental column under impact loading can be found in the literature.

Current understandings on segmental column response under dynamic lateral loading rely primarily on its seismic performance. Many researches have been carried out to investigate the behaviour of segmental column under seismic loading. Cyclic tests and shake table tests on segmental column with different designs and column specifications have been carried out in the laboratory or simulated using numerical methods. Under seismic loading, segmental column was found to be capable of undergoing larger lateral displacement than conventional monolithic column (Ou et al., 2007). Under earthquake-induced lateral loading, plastic hinge was normally formed near the base of the column (Shim et al., 2008). Failure of segmental column under seismic loading was, therefore, always flexural response dominated, whereas shear failure such as shear slip between adjacent segments was not observed (Kim et al., 2010). Some recent studies by Sideris et al. (2014a, 2014b) introduced segmented bridge pier featured with hybrid sliding-rocking joints by reducing the friction between adjacent segments. More energy was reported to be consumed through frictions between different segments. However, amplified moment would be experienced by the column because of P-delta effect.

Comparing with the many studies on seismic response of segmental columns, available investigation on the impact resistance capability of segmental column is limited. Considering the fact that during the design life, a bridge pier or structural column might face the risk of experiencing impact loads, such as vehicle collision and falling rock strike. It is, therefore, necessary and important to properly understand the response of segmental column under impact loading. Recently, Chung et al. (2014) reported their numerical study on bridge pier made of segmental concrete elements with post-tensioning tendon under vehicle impact. A relatively large segmental pier and a small truck striking at low velocity were modelled and studied. No severe damage or failure was resulted on the segmental column in the considered example. Zhang et al. (2016) performed laboratory tests on scaled segmental column models using pendulum impact system. The column comprised plain reinforced concrete segments without shear resistance component as commonly researched and used in the current practice. The segments were post-tensioned with unbonded prestress tendons. The response of segmental column under lateral impact loading was studied experimentally. Comparing with monolithic column, the segmental columns showed outstanding impact resistance capacity. Upon the application of impact loading, the segmental joints would open instead of resulting in concrete tensile cracks and failure. Because of the relatively low stiffness, lower peak impact loadings were measured on segmental columns owing to the interaction of the impactor and the column. Also, segmental columns were found to exhibit much better self-centring capability as compared with monolithic column. The effect of segment/joint number has also been studied. Experimental study by Zhang et al. (2016) revealed that segmental columns of the same height, but with different numbers of segments, performed differently under impact loading. The column with more segments is more flexible, and therefore, the impact load acting on the column under the same impacting condition is lower but the loading duration is longer. Better self-centring capability and energy dissipation capability could be achieved using segmental column consisting of more segments.

In the design of segmental columns, a shear key is often introduced to improve the link between different segments and to increase the shear resistance capacity of the entire column. Shear key can be made of steel (or alloy) and/or concrete. The former normally comprises steel bars which are grouted between adjacent segments or through part or the entire column (Kim et al., 2010). The steel bars will increase shear resistance capacity of the column, and more importantly under seismic loading, deformation of these bars could help to dissipate imposed energy (known as energy dissipation bar). Concrete shear key is also used in practice. Shear keys of different shapes and geometries have been developed to interlock adjacent segments (Dyskin et al., 2001, 2012). Commonly used geometries for segmental column include prism, circular truncated cone and hemispherical block (Mashal et al., 2013; Wang et al., 2008). These shear keys could provide in-plane interlocking mechanism. Interlocking blocks with more complex geometries such as plate-like assembles of tetrahedrons or osteomorphic elements have also been developed in recent years (Ali et al., 2013; Molotnikov et al., 2007). Ramamurthy and Nambiar (2004) provided the history of development of shear key.

In seismic analysis, the effect of shear key is normally ignored. As mentioned above, under earthquake excitation, the magnitude of lateral load is relatively small. The primary response mode of segmental column is flexural bending. Friction force between contacting segments is generally sufficient to transfer shear force. In engineering practice, small concrete shear keys are used only for easy alignment of segments during erection (Wang et al., 2008). However, when the segmental column is subjected to impact or blast loading, the lateral load could be very large. Friction between segments may not necessarily always be sufficient to resist such large lateral dynamic loads. In the authors’ recent experiment on segmental columns with flat segments subjected to lateral impact loads, a large lateral drift was observed between the central segment which was impacted and its adjacent segment (Zhang et al., 2016). Similarly, in a full-scale field blast test on segmental tower (Figure 1(a)), small sliding was observed on the base segment when it was subjected to the blast loading from small explosion (Hulton et al., 2013). For much slender segmental columns with larger slenderness ratio, the resulted lateral drift could amplify the moment on the segmental column due to P-delta effect, which as a result increases the overturn potential of the column. It is, therefore, necessary to include shear key to the concrete elements of segmental column to improve their shear resistance capacity against lateral impact and blast loading. Moreover, Numerical studies by Li et al. (2015) also found that introducing a large concrete shear key to segmental column (as shown in Figure 1(b)) could help to relieve spalling damage when it is under close-in detonation. This is because more interfaces are created within concrete segments due to the existence of shear key, and less intensive blast wave reflects at the back face of concrete segments. Therefore, introducing concrete shear keys to segmental column has the potential to improve the segmental column blast resistance capacity. Studies of the effect of concrete shear key on the impact resistance capacity of segmental column cannot be found in the literature yet.

Figure 1.

Previous studies of segmental columns subjected to blast loads: (a) segmental tower (Hulton et al., 2013) and (b) segmental column with shear key (Li et al., 2015).

In this study, we continued and extended our experimental study on segmental column under lateral impact loading. Two segmental columns comprised, respectively, five segments, and seven segments with unreinforced concrete shear key were tested using a pendulum impact system. The responses of the columns were recorded by a high-speed (HS) camera. The impact loading time history and the column lateral displacement time histories were monitored and used to analyse column performance. Comparisons between segmental columns with and without shear key were made to evaluate the effectiveness of concrete shear key in segments on the lateral impact loading resistance capacity of segmental columns.

Experiment setup

Two pairs of segmental columns are tested in this study. Table 1 summarizes the details of the four columns. Columns S5N and S5K comprise five plain segments and five segments with concrete shear key, respectively. In the second group, columns S7N and S7K both have seven segments with and without concrete shear key. A pendulum impact test system was used to generate the impact loading onto the column. Detailed information including specimen specification, system setup and measurement system is described in the following sections.

Table 1.

Summary of tested column specimens.

Column	Height (mm)	Segment number	Segment height (mm)	Cross section (mm × mm)	φ_{long, rebar}^a (mm)	φ_tie^a (mm)
T02-S5N^b	800	5	160	100 × 100	6	4
T06-S5K	800	5	160	100 × 100	6	4
T04-S7N^b	800	7	115	100 × 100	6	4
T05-S7K	800	7	115	100 × 100	6	4

φ stands for the diameter of reinforcement.

Zhang et al. (2016).

Column design

Figure 2 depicts the design of the column specimens. The columns had a total height of 800 mm² and 100 mm² × 100 mm² cross-sectional area, which were ¼ scale column models. The dimension of the tested columns was restricted by the dimension of the pendulum impact testing system for the current experiment. Previous studies by Woodson and Baylot (1999, 2000) have demonstrated the additivity of ¼ scale model for reinforced concrete column with similar dimensions subjected to blast loading. Column damage and response including load and deflection were properly documented in their tests. It is, therefore, believed the current adopted scaled model could be able to properly represent the behaviour of segmental column under impact loading. As shown, the column segments were reinforced with four 6-mm-diameter longitudinal bars, and the longitudinal rebar were trimmed and did not extend between adjacent segments. Four 4-mm-diameter ties were used for each 160-mm segment, and three ties were used for each 115-mm segment. A 10-mm-thick concrete cover was designed for all the columns. Two 6-mm-diameter starter bars were used to connect the segmental columns to their footings. No grout or epoxy was applied to glue the base segment to the footing. All the 6-mm-diameter longitudinal rebar and the starter bars were made of ribbed deformed bars; the tie were made of plain steel wires. A 15-mm-diameter hole at the centre of each segment was made for applying the post-tensioning tendon. A total of seven wire strands with a combined diameter of 9.3 mm were used as prestress tendon. The footing was 400 mm wide by 400 mm long by 140 mm deep with heavy reinforcements. A set of barrel and wedge was used in the footing to anchor the prestress tendon in the segmental columns. Grade 25 self-compacting concrete with super plaster was used to cast the column segments and their footings. Considering the scale of the column models, aggregates with maximum dimension of 10 mm were used for the concrete. The material properties of the reinforcement, tendon and concrete are listed in Table 2.

Figure 2.

Design of the test specimens: (a) T02-S5N, (b) T06-S5K, (c) T04-S7N and (d) T05-S7K.

Table 2.

Material properties.

Material	ρ (kg/m³)	fc′ (MPa)	f_t, f_y (or f_proof) (MPa)	E (GPa)
Concrete	2400	34	5	30
Longitudinal rebar	7800	–	500	200
Tie	7800	–	300	200
Prestress tendon	7850	–	1860	195

The scaled column models were casted with care. Much attention was paid to installing the segmental columns and to ensure alignment. A 400 mm × 400 mm × 450 mm (length × width × depth) concrete block together with a series of steel plates was stacked on top of the columns as the added weight (approximately 288 kg total weight). The ratio of supported weight over the column self-weight (about 18 kg) was 16. The top segment of each segmental column was in direct contact with the concrete block without application of grout or reinforcement. After installing the added weight on top of each column, a total prestress load of 30 kN was jacked to the prestress tendon. The setup of the column is a free-standing column with supported weight on its top. Such design reflects free-standing columns or piers with minimum later constrain at its top. Upon later impact, the top of the column sways sideway introducing significant P-delta effect, which amplifies the moment onto the column.

Test setup

Figure 3 depicts the pendulum impact testing system for this study. As shown, it comprised a steel rig fixed on solid floor to support the entire test system. A 300-kg steel impactor was connected through a 2.8-m-long arm to the frame. An inclinometer installed on top of the steel rig was used to measure the release angle of the impactor. In each impact test, the impactor was lifted to a designated height and then released to impact the centre (mid-height) of the column. The impactor hits the column and rebounds, which was pulled back to avoid the column being impacted for a second time.

Figure 3.

Schematic view of pendulum impact system.

Measurement system

Figure 3 also illustrates the measurement system. To measure the applied impact load, a load cell was fixed in front of the impactor. Three laser linear voltage displacement transducers (LVDTs) were placed at column mid-height, column top and near the column base behind the specimens. The LVDT has a measuring range of −150 to 150 mm. The sensors were wired to a NI USB-9237 acquisition system, and the data were captured at a sampling frequency of 50 kHz. A HS camera (Photron^® SA-Z) was installed to monitor the deformation-to-failure process of the columns. The filming rate of the HS camera was set to 8000 fps. The exposure time was set to balance with the aperture. Four halogen lights were used to provide intensive light for HS filming. Five tracking dots were glued to the middle of each segment for the segmental columns S5N and S5K on its side elevation. Similarly, seven tracking dots were used for columns S7N and S7K. The HS camera images were post-processed using digital image correlation software to derive the column displacement time histories at these tracking dots. For easy interpreting, the test results, the segments and joints are numbered as Segments 01–07 and Joints 01–08, respectively, as shown in Figure 2.

Experimental results

The experimental results from the pendulum impact tests on segmental columns with shear keys are presented in this section. HS camera images on column deformation-to-failure processes together with the column local damage and global failure modes are shown to give a clear account of the behaviour of segmental column with shear key under lateral impact loading. Quantitative results including impact load time histories and column lateral displacement time histories are then provided to better describe the impact resistance capacity of the tested segmental columns.

In this study, each column was tested under multiple impacts with increased impact speed in each time. As shown in Table 3, in Impact 01, the impactor was released at 2.5° which corresponded to an impact speed of 0.23 m/s. The release angles of the impactor were then increased to 7° and 15° in Impact 02 and 03. The corresponding impact speeds were about 0.64 and 1.37 m/s, respectively. In Impact 04, a final strike was conducted with the impactor released at 30° and an estimated impact velocity of 2.71 m/s. The designed impact velocity was to assess the performance of the specimens at different loading levels, for example, elastic response, minor damage, medium damage and ultimate failure response of the columns under impact loading. Detailed test results are provided in the following sections.

Table 3.

Summary of impact parameters.

Test no.	Impactor weight (kg)	Release angle (°)	Est. impact velocity (m/s)
Impact 01	300	2.5	0.23
Impact 02	300	7	0.64
Impact 03	300	15	1.37
Impact 04	300	30	2.71

HS camera images

The deformation-to-failure process of the segmental columns in each impact test was monitored by a HS camera. Figures 4 and 5 show the snapshots of the HS camera images on columns T06-S5K and T05-S7K, respectively.

Figure 4.

Snapshots of HS camera images of T06-S5K: (a) Impact 01, (b) Impact 02, (c) Impact 03 and (d) Impact 04.

Figure 5.

Snapshots of HS camera images of T05-S7K: (a) Impact 01, (b) Impact 02, (c) Impact 03 and (d) Impact 04.

As shown in Figure 4(a), when column T06-S5K (five segments with shear key) was subjected to a very small impact load (impact speed 0.23 m/s), no significant deformation or apparent damage could be found. Close observation found that at t = 36.25 ms, the column began to deform under the action of the impactor. The mid-height region of the column (Segments 03 and 04) showed the largest flexural deformation. Because of inertia resistance from the added mass at the top of the column, the upper part of the column responded relatively slower than that at the centre of the column. When repeated the impact on the column with a higher speed (about 0.64 m/s) in Impact 02, the column responded similarly as in Impact 01, but larger flexural deformation was developed near the centre initially as shown in the image at t = 35.75 ms in Figure 4(b) as expected. An insignificant joint opening was formed between Segments 04 and 05. As the vibration continued in the free-vibration phase, the inertia effect from the added mass diminished and the upper segments swayed sideway. The column primary response mode transferred from the local flexural mode to the global flexural mode with the largest deformation occurring at the top of the column as shown in Figure 4(b) at t = 222.625 ms. The column vibrated back and forth freely around its original position (t = 222.625 ms and 590.75 ms). When the impactor was released at 15° in Impact 03 (Figure 4(c)), the column bent more significantly at its centre (t = 30 ms). As the compressive stress in the left corner of Segment 03 exceeded concrete strength, concrete crack was formed. Similar as in Impacts 01 and 02, the inertia restraint on top of the column diminished quickly and the local mode gradually turned into global mode with the largest deformation at the centre of the column shifted upwards. Joint opening between Segments 02 and 03 closed, but the joint between Segments 04 and 05 opened due to the combined action of the successive compressive stress from flexural bending, the shear stress due to the impact and also the axial compressive stress from the added mass and prestress tendon. Concrete crack was found in Segment 05 at t = 60 ms. Despite the damages to segments, all opened joints closed as the column rebounded. The entire column vibrated freely and eventually restored its original position. In the ultimate Impact 04, a large flexural bending was formed at t = 32.5 ms. More damages were observed in the central segment due to large direct impact force and also the flexural-induced compression. As the response of the column transferred from local mode to global mode, joint opening between Segments 02 and 03 gradually closed, but a large opening was formed at the joint between Segments 04 and 05. In the meanwhile, the base segment (Segment 01) also rotated as the entire column swayed sideway. A significant joint opening could be observed between the base segment and the footing (t = 122.5 ms). The column rebounded a little bit after the impact. Because of the severe damage on Segment 05, the column lost its stability with the supported mass fell over at the end of the test.

The responses of column T05-S7K (seven segments with shear key) were slightly different from the five-segment column. With more segments, column S7K was more flexible than column S5K. As shown in Figure 5(a), in Impact 01, the central part of the column was forced to bend. At t = 24.125 ms, a small crack was formed at the left corner of Segment 04 due to flexural-induced compression. A longitudinal crack was also developed in Segment 02 near the base. This was because of the combined actions of the flexural-induced compression on the right-hand side of the lower segments and also the axial compressive stress due to the added mass and the prestress force. The existence of the shear key resulted in some stress concentration. When the column was subjected to a larger impact load in Impact 02 (Figure 5(b)), the column responded similarly to that in Impact 01. At t = 32.625 ms, more apparent flexural bending was observed at column centre due to larger impact load. In Impact 03 (Figure 5(c)), the release angle of the impactor was increased to 15°. Under the elevated impact loading, a larger central deflection was formed on the column at t = 26.875 ms. Because of excessive flexural bending, the up-left corner of Segment 02 was also damaged. Small joint openings occurred between Segments 03 and 04 and also Segments 04 and 05. In the meanwhile, a small joint opening was also formed between the upper two segments (Segments 06 and 07) due to inertia restrain. The longitudinal crack near the bottom of the column was observed (in Segment 02), which extended into the base segment (Segment 01). As the impactor forced the column to further deform (t = 53.75 ms), larger central deflection can be observed together with wider joint openings. The cracks in the two base segments (Segments 1 and 2) grew wider. The maximum displacement moved upwards from the column centre because the response of the column gradually transferred from the local mode to the global flexural mode after the forced-vibration phase (t = 99.75 ms), and the column vibrated freely until coming to a full stop owing to damping effect. When subjected to the ultimate impact (Impact 04), catastrophic damage of the column was observed. As the impactor struck at the centre of the column (t = 23 ms), the column was forced to bend with a significant joint opening between Segments 03 and 04 near the column centre. More cracks were formed in Segment 04 as a result of excessive flexural compression and impact force directly on the segment. Severe damage was resulted on this segment (t = 116.25 ms). The column deformed substantially. Cracks in the two bottom segments further grew as the column bent. A large opening was also formed between these two segments. Under the effect of inertia restrain, the top segment (Segment 07) also bent as the column deflected. Concrete on the left-hand side of Segments 06 and 07 failed and were spalled under compression. Joint opening could have also been observed between these two segments. The localized flexural response mode gradually changed to global flexural response mode as the inertia effect on the top of the column diminished. The top of the column swayed sideway at t = 214.625 ms. Due to excessive lateral displacement and severe damage in concrete segments, the column collapsed under the action of the added mass (t = 1012.875 ms). It is to be noted that the added mass (concrete block and the steel plates) were hanged with shackles and slings during installation. Before the impact, the slings were loosen, which would hang and lift the mass only when the column completely lost its supporting capability.

Column and segment damages

Figures 6 and 7 depict the damage status of columns S5K and S7K after each impact. As shown in Figure 6(a) for column S5K, close observation after the Impact 02 spotted an unnotable longitudinal crack in the top segment (Segment 05). In Impact 03, the crack extended through Segment 05 leading to the entire corner of the segment chopped. Minor corner damage was also found on the central segment (Segment 03). Figure 6(c) shows the impact elevation of the entire column after the impact, which clearly depicts the damages of concrete due to flexural compression. After Impact 04, the column was carefully disassembled for detailed examining the segment damage. Snapshots of typical segment damages are given in Figure 6(d). As shown, the concrete shear keys of Segments 02 and 04 both suffered certain level of damage. The concrete mortises of Segments 03 and 05 also experienced severe damage. This was mainly because of the combined action of flexural compressive stress and shear stress. In addition, stress concentration tended to form in such locations with non-smooth geometry.

Figure 6.

Damages to column T06-S5K: (a) Impact 02, (b) Impact 03, (c) Impact 04 and (d) damaged segments.

Figure 7.

Damages to column T05-S7K: (a) Impact 01, (b) Impact 02, (c) Impact 03, (d) Impact 04 and (e) damaged segments.

More severe damages to concrete segments were found on the column with more segments (column S7K). As shown in Figure 7(a), concrete cracks at the corner of central segment of column S7K were formed as early as in Impact 01, which, in comparison, was only observed on column S5K after Impact 02. Moreover, longitudinal crack was also found in the segment near the base (Segment 02). These cracks developed and grew wider in the subsequent impacts (Impacts 02 and 03) from large flexural deformation in Impact 03. The crack in Segment 02 extended through Segment 01, which eventually grew into major cracks leading to the collapse of the column in Impact 04 (Figure 7(d)). Much more severe damages to segment were resulted on this column with seven segments. As evidenced in Figure 7(e), most segments suffered serious concrete damages. This was probably because the segmental column with more segments was more flexible. Under similar level of impact loading, larger deflection and curvature were developed on the column. As a result, larger flexural compressive stress was resulted on relatively smaller areas of the segments. Since large shear force was also transferred through the segments because of the introduced shear key, more severe damages were resulted on column S7K under the combined action of the above stresses together with stress concentration due to shear key.

Displacement time history

The lateral displacements of the column top, mid-height and near the column base (at the centre of the base segment) were recorded by LVDTs. Figure 8 depicts the displacement time histories of column S5K. As shown, the displacements at the base segment of the column were always quite small compared to those at the column centre and top. This was mainly because of the restrain from the footing through the starter bars. Lateral displacement of the base segment was caused because of the rotation of the base segment when the column deformed due to the impact load hitting at the column centre. During the forced-vibration phase, the column lateral displacements at the three measurement locations quickly pick up and reach a peak value (the first peak in the displacement time history). They then decrease slightly. As can be noted, the first peak displacement values in the first three tests at the column top and centre before column collapsed in Impact 04 are more or less the same, and in Impact 03, with higher impact velocity, the column central displacement is even slightly larger than that at the column top. During the forced-vibration phase, the column responded following a mode equivalent to a column with a partial restraint at its top. After the action of the impact load (the impact loading duration is indicated in Figure 10 of the recorded impact loading time histories), the column responded freely. The column vibrated according to its fundamental mode with the displacement value at the column top significantly larger than that at the mid-height of the column (second peak in the displacement time history). The maximum displacement response of the column top occurred in the second peak, that is, the first peak in free-vibration phase, while the maximum displacement at the column centre occurred in the first peak during the forced-vibration phase. As can be seen in Figure 8(a) to (c), as the impactor releasing angle increased gradually, the responses of the column also increased as expected. In Impact 04 (Figure 8(d)), the damaged column collapsed under the significant impact loading. The displacements at both the centre and top of the column quickly increased to peak deflections of about 131 and 188 mm, respectively. Because of inertia restrain at column top, the deflection at the centre of the column increased faster than that at the top of the column at the beginning. It should be noted that in the tests, after reaching peak deflections, the column was still pulled back by the prestress tendon. However, as some segments and joints suffered severe damage, the column lost stability and was considered totally damaged.

Figure 8.

Displacement time histories of column T06-S5K: (a) Impact 01, (b) Impact 02, (c) Impact 03 and (d) Impact 04.

Figure 9 shows the displacement time histories of column T05-S7K. Similar observations made above can be drawn here again. As shown in Figure 9(a), in Impact 01, a peak deflection of 4.4 mm was resulted at the central segment, while a maximum deflection of about 9.2 mm was captured at the top of the column during the free-vibration phase. Because of the damage to some segments on S7K, a small residual displacement of less than 1 mm was found at both the top and the centre of the column. The peak deflections and residual displacements on the column increased as the impact forces increased in Impacts 02 and 03. The column collapsed in Impact 04. As shown in Figure 9(d), the lateral displacements at column centre and top increased quickly under the impact loading. Again, due to inertia resistance, which provided certain level of restraint at the column top during the action of impact loading, the column response followed the local mode with the displacement response at the column centre and top very similar. The maximum column response occurred in the free-vibration phase in the second response peak with the displacement response at the column top substantially larger than that at the column centre. Because of the large lateral displacement and the seriously damaged column in Impact 04, the column lost its stability.

Figure 9.

Displacement time histories of column T05-S7K: (a) Impact 01, (b) Impact 02, (c) Impact 03 and (d) Impact 04.

Impact load

The impact load time histories on the two columns (S5K and S7K) in each test are shown in Figures 10 and 11, respectively. The load time history for column S5K in Impact 01 was missing due to load cell malfunction, which was immediately repaired after this impact and calibrated again before being used in the subsequent tests. As shown in Figure 10(a), the impact load in Impact 02 increased quickly to about 14 kN after the impactor struck on the column, which then reduced to about 4 kN. The impact load fluctuated then reached a second peak of about 6.5 kN before it diminished to 0. This second peak on load time history was because of the interaction between the impactor and the column. As depicted above in Figure 8(a), the central segment quickly rebounded after reaching its first peak displacement. The rebounded central segment compressed the load cell and resulted in the second peak in the load time history. As the release angle of the impactor further increased to 15° in Impact 03, the recorded peak impact force also increased to about 19 kN. Despite the impactor was dropped at a larger angle in Impact 04 (30°), the peak impact load did not further increase because the stiffness of the column degraded as a result of accumulated damage in previous impacts, and the column collapsed quickly.

Figure 10.

Load time histories of column T06-S5K: (a) Impact 02, (b) Impact 03 and (c) Impact 04.

Figure 11.

Load time histories of column T05-S7K: (a) Impact 01, (b) Impact 02, (c) Impact 03 and (d) Impact 04.

Figure 11 shows the recorded load time histories on column S7K. Similar pattern of impact loadings were recorded on this column as those on S5K. A clear second peak can be observed in Figure 11(a) for Impact 01, in which the impact did not cause any damage to the column. The column rebounded fast and resulted in a large and significant second peak load as it compressed back on the load cell. In Impact 02 a peak load of 11 kN was measured on column S7K. The peak load was 22% smaller than that on column S5K because with more segments and joints, column S7K was more flexible than column S5K. In addition, certain segments of column S7K experienced minor damage which could also contribute to the degraded stiffness of the column. The peak load (F_peak) and Impulse, together with the maximum central displacement, residual centre displacement, maximum top displacement and residual top displacement, are summarized and listed in Tables 4 and 5.

Table 4.

Summary of impact loads.

	T02-S5N		T06-S5K		T04-S7N		T05-S7K
	F_max (kN)	Impulse (kN ms)	F_max (kN)	Impulse (kN ms)	F_max (kN)	Impulse (kN ms)	F_max (kN)	Impulse (kN ms)
Impact 01	7.3	141	–	–	8.2	203	7.9	183
Impact 02	13.4	301	14.1	338.6	11.8	323	11.0	364
Impact 03	20.9	560	19.0	649.3	16.6	557	13.1	631
Impact 04	21.8	918	17.2	1002.4	23.9	926	11.8	1007
Impact 05	–	–	–	–	19.6	1179	–	–

Table 5.

Summary of maximum and residual deflections.

	T06-S5K		T02-S5N		T04-S7N		T05-S7K
	Max. (mm)	Resi. (mm)	Max. (mm)	Resi. (mm)	Max. (mm)	Resi. (mm)	Max. (mm)	Resi. (mm)
(a) Maximum (Max.) and residual (Resi.) deflections at column centre
Impact 01	2.9	0.3	3.0	0.5	5.1	0.7	4.4	0.6
Impact 02	9.4	0.8	7.7	0.5	11.5	0.9	12.6	1.3
Impact 03	31.1	3.1	32.8	2.6	24.3	1.1	37.1	7.9
Impact 04	131.1	7.4	108.4	24.4	104.2	4.7	–	–
(b) Maximum (Max.) and residual (Resi.) deflections at column top
Impact 01	6.1	0.9	6.1	0.7	10.6	0.7	9.2	0.9
Impact 02	22.4	1.6	17.6	1.8	21	2.4	28.5	2.0
Impact 03	68.0	4.5	52.9	3.8	37.6	4.4	88.5	13.7
Impact 04	187.9	−4.7	116.3	41.8	111.9	6.9	–	–

Analysis and discussion

The effectiveness and influence of shear key on the impact resistance performance of segmental column is analysed and discussed in this section by comparing the response of segmental columns with and without shear key. Segmental columns – T02-S5N and T04-S7N – from the same group of laboratory tests but reported in a previous paper (Zhang et al., 2016) are also included here for comparison. Analysis is carried out with respect to the column damage, peak impact load and impulse, column lateral displacement and energy dissipation.

Column damage

Figure 12 compares the damages of the two groups of columns after Impact 03. As shown, damage to segmental columns was primarily in the form of concrete compressive failure. None of the columns suffered tensile concrete cracks. The flexural-induced tension was undertaken by the opening of the segmental joints and extension of prestress tendon. After the impact, the joints gradually closed and the columns restored almost to their original positions with small residual displacement despite some concrete damage. No global shear failure occurred on the segmental columns. However, due to large impact loading, relative displacement between adjacent segments could occur. It is evidenced in Figure 12(a) that an apparent relative displacement was formed between Segments 03 and 04 in column S5N. In contrast, no such relative displacement was found on column S5K (Figure 12(b)) indicating the effectiveness of the shear key in resisting shear force and preventing relative displacement between segments. Although the shear key is effective in providing shear resistance because of stress concentration around the concrete shear key, more severe concrete damage occurred around the unreinforced concrete mortise on Segments 03 and 05 in column S5K. Similarly, the segments of column S7K suffered more severe concrete damages (Figure 12(d)) than those on column S7N (Figure 12(c)). More longitudinal cracks were found in column S7K than in S7N. Therefore, through comparing the column damage, it can be easily found that introducing concrete shear key could help to reduce the lateral relative displacement between segments when it was subjected to lateral impact loading. However, with the current design, the large unreinforced shear key also resulted in stress concentration which led to more concrete damage in the segments around segmental joints. Optimized design should be made in further research and practice.

Figure 12.

Comparison of column damages after Impact 03: (a) T02-S5N, (b) T06-S5K, (c) T04-S7N and (d) T05-S7K.

Impact load

The recorded impact loads were integrated along time to derive the impulse for each impact. Together with the peak impact loads for the tested columns in each impact, the derived impulses are summarized and plotted in Table 4 and Figure 13. As shown for the segmental columns with five segments (S5K), shear key increased the stiffness of the undamaged column. As evidenced in Impact 02, larger peak load (14 kN) and impulse (339 kN ms) were recorded, while those on column S5N without shear key were 13.5 kN and 301 kN ms. In all, 5% higher peak impact load and 12% larger impulse were resulted on the segmental column with shear key. As discussed above, the existence of shear key caused more concrete damage which degraded column stiffness, in Impact 03 and 04, lower peak impact forces were found on column S5K than those on column S5N. But as the impact duration increased, higher impulses were still measured on column S5K than those on S5N. For instance, in Impact 03 about 10% smaller peak impact load was measured on column S5K than that measured on S5N, but the impulse on S5K was 16% higher. With more segments and joints, the peak impact loads measured on column S7K and S7N were in general lower than the corresponding peak impact loads measured on segmental columns with five segments because of relatively smaller column stiffness. For example, in Impact 02, a peak impact load of 11 kN was measured on S7K, which was 20% less than that on S5K. Despite the existence of shear key in S7K which was supposed to increase the stiffness of the column, even in Impact 01, this column suffered damages to some segments, which led to the degraded column stiffness. A peak impact load of 7.9 kN resulted on S7K which was 3% lower than that on S7N. Smaller peak impact forces always resulted on column S7K as more damages accumulated in the subsequent tests. In Impacts 02 and 03, 7% and 21% lower peak loads were found on column S7K than those on S7N. Nevertheless because the impactor acted longer on column S7K, larger impulses were resulted. In Impact 02, 364 kN ms impulse was measured on column S7K indicating a 13% larger impulse than that on S7N.

Figure 13.

Recorded impulses and peak impact loads.

Column response

The central deflection time histories for the segmental columns are compared in Figure 14. The deflections of the columns with seven segments (S7K and S7N) were in general larger than those with five segments (S5K and S5N), because the formers had smaller stiffness due to more joints. As shown in Figure 14(a), under small impact loading in Impact 01, the columns with shear key (S5K and S7K) both exhibited smaller central deflections comparing with those without shear key. This is because of the increased stiffness of column due to shear key. A maximum central deflection of over 5 mm was measured on column S7N, while that on column S7K was 4.4 mm (−12%); the maximum deflection on column S5K was only 2.9 mm, which was slightly smaller than that on column S5N (3.0 mm). It is worth noting that S5N was the first column tested and Impact 01 at 2.5° was the very first impact test. After the initial impact, the impactor rebounded and the probe of the impactor touched the column again before being manually pulled back, which resulted in a sharp pulse on the column deflection time history during free vibration. Much more care was paid to the following tests to avoid such second impact. Smaller residual deflections were also found in segmental columns with shear key. For instance, after Impact 01, a residual central deflection of only 0.3 mm was found on column S5K, while that on column S5N was 0.5 mm. It can also be seen clearly that after the forced vibration, the column S5K vibrated faster with smaller amplitude. Again, this was because of the higher stiffness of S5K due to the concrete shear key. Despite smaller peak deflection resulted on the column S7K in Impact 01, concrete cracks resulted due to stress concentration near shear keys. As a result, larger central deflections were always found on this column in the following impact tests comparing with the one without shear key (S7N). The column also responded relatively slower as its stiffness degraded. Similarly, in Impact 02, concrete cracks in certain segments of column S5K were also generated, which resulted in degraded column stiffness and larger central deflection consequentially.

Figure 14.

Comparison of column central deflections: (a) Impact 01, (b) Impact 02, (c) Impact 03 and (d) Impact 04.

Table 5 summarizes the maximum and residual deflections at column centres and column tops. The deflections are plotted against the corresponding impulses in Figure 15. As shown in Figure 15(a), the maximum central deflections in general correlated with the applied impulses. When subjected to higher impulse, larger maximum central deflection resulted. Despite the shear key improved column stiffness, more concrete damages also resulted due to stress concentration. As a result, slightly larger residual deflections were found on the segmental columns with shear key. For example, a residual central deflection of about 0.8 and 3.1 mm were recorded on column S5K, while those on column S5N were 0.5 and 2.6 mm in Impact 02 and 03, respectively. It is more apparent that because of more severe damage to segments of column S7K, a residual central deflection of 7.9 mm was measured. In comparison, that of column S7N was only 1.1 mm. Similar trend can also be observed on column top deflections (as shown in Figure 15(b)).

Figure 15.

Summaries of the maximum and residual deflections at column centre and top: (a) column centre and (b) column top.

Relative displacement

Despite shear key did not contribute significantly to reducing the column maximum and residual deflections, the existence of the shear key could help to minimize relative displacement between adjacent segments. Preliminary assessment as shown in Figure 6 found unnotable relative displacement between segments of segmental column with shear key (S5K). In comparison, significant relative displacement was observed on the column without shear key (S5N). To better evaluate the effect of shear key, relative displacements between adjacent segments on the four tested segmental columns were quantified after each impact test. Quantification was made by measuring the misalignment of vertical grid lines plotted on the surface of each column. The measured relative displacements were then summarized and plotted against the corresponding impulse in Figure 16. As shown, for the two columns with five segments, no segmental drifts resulted under small impact loads in Impact 01 and 02. When column S5N was subjected to about 560 kN ms impulse, a relative displacement of about 4 mm resulted between the two central segments. In contrast, only 1.2 mm relative displacement (70% less) was found on column S5K with shear key (while the corresponding impulse was even larger than that on column S5N). In the subsequent test, the relative displacement on column S5N further enlarged to about 8.7 mm, while that on the segmental column with shear key was only 2.2 mm (75% less).

Figure 16.

Relative displacements versus impulses.

For segmental columns comprising more segments, the effect of shear key in controlling the relative displacement was more significant. As shown in Figure 16, on column S7N without shear key, a relative displacement of about 2 mm was found between the two central segments when the column was subjected to about 560 kN ms impulse. And the value increased to about 2.3 mm in the subsequent test. No segmental drift was observed between the central segments on the column with shear key (S7K). Nevertheless, because of stress concentration on concrete near the shear key, cracks were formed and developed on the segments near column base. As a result, a small relative displacement (about 0.7 mm) occurred between Segments 02 and 03 when subjected to 184 kN ms impulse in Impact 01. The segmental drift grew to about 1.3 and 1.5 mm under the subsequent impacts. As crack extended into the base segment (Segment 01) when the column was under 624 kN ms impulse, very small relative displacement (about 0.5 mm) between Segments 01 and 02 was also found. Overall, for the group of segmental columns with seven segments, less relative displacements were found on the column with shear key. It can, therefore, be easily concluded that the shear key was effective in reducing relative movement between segments especially when the column was subjected to large impact loading. However, the drawback of the current shear key design is also obvious because it results in stress concentration in the segment. Further improvement in shear key design to reduce stress concentration is needed.

Energy dissipation

Energy dissipation capability is an important factor when considering impact resistance capacity. The energy absorption depends on the impact loading and deformation time history. The load–central displacement curves of the segmental columns in each test are plotted and compared in Figure 17. The enclosed area of the curves represents the dissipated energy by the column. It can be seen that as the impact level increased, the corresponding energy dissipated by the segmental columns also increased. In general, higher peak impact loads were recorded on segmental columns with five segments, while larger central displacements were found on columns with seven segments. Since shear key led to more concrete damages in column S7K, the stiffness of this column degraded significantly. As shown in Figure 17(a), in Impact 01, similar stiffness (tangent of load–deflection curve) can be observed between columns S7K and S7N because no joint opening or relative displacement was formed under this low-level impact. As impact load increased to over 4 kN, the stiffness of column S7K reduced as some segments cracked within this column. Similar comparison can be observed between these two columns in Impact 02. As column S7K suffered more damages, the central deflection was larger than that on column S7N. The influence of shear key on segmental columns comprising five segments was not apparent in terms of energy dissipation capability. The dissipated energies by each column is summarized and plotted against the corresponding impulses. As shown in Figure 18, it can be found clearly that the energy dissipated by the segmental columns is approximately proportional to the impulse. More energy was dissipated when the column was subjected to larger impulse. In general, the columns with shear key consumed more impact energy. For example, comparing the pair of segmental columns with five segments, it could be found that about 15% more energy was dissipated by column S5K in Impact 02. The dissipated energy was also 9% and 22% more on column S5K in Impacts 03 and 04. As column S7K experienced more concrete damage, much more energy was consumed than column S7N especially when subjected to high-level impact. For instance, in Impact 03, about 1300 J energy was consumed by column S7K which was 25% higher than that on column S7N. Even more energy (about 5900 J) was dissipated in the following test because of the substantial damage to column S7K, and the dissipated energy was nearly 50% more than that by column S7N.

Figure 17.

Comparison of load–deflection curves: (a) Impact 01, (b) Impact 02, (c) Impact 03 and (d) Impact 04.

Figure 18.

Energy dissipation with respect to impulses.

Conclusion

In this study, pendulum impact tests were carried out on precast concrete segmental columns. The effect of shear key on the performance of segmental columns under lateral impact load was investigated experimentally. Comparison was made between segmental columns with and without shear key. Segmental columns comprising five segments and seven segments were tested. The test results found that the proposed concrete shear key could help to improve the lateral shear resistance and reduce segmental drift. Damage to concrete shear key was observed on the segments, indicating the effectiveness of shear key. Nevertheless, it was also found that the large shear key led to stress concentration within segments which resulted in more severe damage to concrete segments. The phenomenon was more apparent on the segmental column with more segments. Because of more damage to concrete segments, the stiffness of the segmental columns with shear key degraded. Lower peak impact loads were measured on the segmental columns with shear keys. More energy was found to have been dissipated by these segmental columns with shear keys. Modification and optimization of design on the geometry and dimension of the shear key could be performed to reduce stress concentration and hence damage to concrete segments.

Footnotes

Acknowledgements

The first author (X.Z.) would like to thank Mr James Water in the structure lab of the University of Western Australia for his assistance in conducting the test. The authors would also like to address Mr Matt Brockman in actively involving in this test for his final year project.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The authors would like to acknowledge ARC for financially supporting this project.

References

Ali

Briet

Chouw

(2013) Dynamic response of mortar-free interlocking structures. Construction and Building Materials 42: 168–189.

Chou

C-C

Chang

H-J

Hewes

(2013) Two-plastic-hinge and two dimensional finite element models for post-tensioned precast concrete segmental bridge columns. Engineering Structures 46: 205–217.

Chung

Lee

Gil

(2014) Structural performance evaluation of a precast prefabricated bridge column under vehicle impact loading. Structure and Infrastructure Engineering 10: 777–791.

Dyskin

Estrin

Kanel-Belov

. (2001) Toughening by fragmentation – how topology helps. Advanced Engineering Materials 3: 885–888.

Dyskin

Pasternak

Estrin

(2012) Mortarless structures based on topological interlocking. Frontiers of Structural and Civil Engineering 6: 188–197.

Guerrini

Restrepo

Massari

. (2014) Seismic behavior of posttensioned self-centering precast concrete dual-shell steel columns. Journal of Structural Engineering. Available at: http://ascelibrary.org/doi/abs/10.1061/(ASCE)ST.1943-541X.0001054

Hulton

Walker

Eytan

. (2013) The response of a tower to blast loading: analysis, design, testing. In: 15th international symposium for the interaction of the effects of munitions with structures, Potsdam, 17–20 September, 2013.

Ichikawa

Matsuzaki

Moustafa

. (2015) Seismic Behavior of Ultra High Performance Segmental Bridge Columns. Washington, DC: Transportation Research Board.

Kawashima

Ichikawa

ElGawady

. (2012) Effect of UFC segments for enhancing the seismic performance of bridge columns. 10 pp. Available at: http://www.iitk.ac.in/nicee/wcee/article/WCEE2012_0669.pdf

10.

Kim

T-H

Lee

H-M

Kim

Y-J

. (2010) Performance assessment of precast concrete segmental bridge columns with a shear resistant connecting structure. Engineering Structures 32: 1292–1303.

11.

Hao

(2015) Preliminary investigation of blast resistance capacity of segmented column using numerical method. In: International conference on protective structures (ICPS3), Newcastle, NSW, Australia, 3–6 February.

12.

Mashal

White

Palermo

(2013) Quasi-static cyclic tests of Emulative Precast Segmental Bridge Piers (E-PSBP). In: 2013 NZSEE conference, Wellington, New Zealand. Available at: http://www.nzsee.org.nz/db/2013/Poster_27.pdf

13.

Molotnikov

Estrin

Dyskin

. (2007) Percolation mechanism of failure of a planar assembly of interlocked osteomorphic elements. Engineering Fracture Mechanics 74: 1222–1232.

14.

Motaref

Saiidi

Sanders

(2013) Shake table studies of energy-dissipating segmental bridge columns. Journal of Bridge Engineering 19: 186–199.

15.

Nikbakht

Rashid

Hejazi

. (2015) Application of shape memory alloy bars in self-centring precast segmental columns as seismic resistance. Structure and Infrastructure Engineering 11: 297–309.

16.

Y-C

(2007) Precast Segmental Post-Tensioned Concrete Bridge Columns for Seismic Regions. Buffalo, NY: State University of New York at Buffalo.

17.

Y-C

Chiewanichakorn

Aref

. (2007) Seismic performance of segmental precast unbonded posttensioned concrete bridge columns. Journal of Structural Engineering 133: 1636–1647.

18.

Y-C

Oktavianus

Tsai

M-S

(2013) An emulative precast segmental concrete bridge column for seismic regions. Earthquake Spectra 29: 1441–1457.

19.

Ramamurthy

Nambiar

(2004) Accelerated masonry construction review and future prospects. Progress in Structural Engineering and Materials 6: 1–9.

20.

Shim

Chung

C-H

Kim

(2008) Experimental evaluation of seismic performance of precast segmental bridge piers with a circular solid section. Engineering Structures 30: 3782–3792.

21.

Sideris

Aref

Filiatrault

(2014a) Large-scale seismic testing of a hybrid sliding-rocking posttensioned segmental bridge system. Journal of Structural Engineering 140. Available at: http://ascelibrary.org/doi/abs/10.1061/(ASCE)ST.1943-541X.0000961

22.

Sideris

Aref

Filiatrault

(2014b) Quasi-static cyclic testing of a large-scale hybrid sliding-rocking segmental column with slip-dominant joints. Journal of Bridge Engineering 19. Available at: http://ascelibrary.org/doi/abs/10.1061/(ASCE)BE.1943-5592.0000605

23.

Sprinkel

(1985) Prefabricated bridge elements and systems. NCHRP Synthesis of Highway Practice. Available at: http://onlinepubs.trb.org/onlinepubs/nchrp/nchrp_syn_324.pdf

24.

Wang

Chang

. (2008) Large-scale seismic tests of tall concrete bridge columns with precast segmental construction. Earthquake Engineering & Structural Dynamics 37: 1449–1465.

25.

Wang

Wei

(2014) Seismic performance of precast hollow bridge piers with different construction details. Frontiers of Structural and Civil Engineering 8: 399–413.

26.

Woodson

Baylot

(1999) Structural Collapse: Quarter-Scale Model experiments (DTIC document-ADA369355, Defense Technical Information Center, August 1999). Available at: http://oai.dtic.mil/oai/oai?verb=getRecord&metadataPrefix=html&identifier=ADA369355

27.

Woodson

Baylot

(2000) Quarter-scale building/column experiments. Advanced Technology in Structural Engineering. Available at: http://ascelibrary.org/doi/abs/10.1061/40492(2000)99

28.

Zhang

Hao

(2016) Experimental investigation of the response of precast segmental columns with posttensioning tendon under impact loading. International Journal of Impact Engineering 95: 105–124.