Abstract
Keywords
Introduction
The common material glass, of which transparency is the greatest feature, remains one of the most important architectural materials when devising functional and artistic architectural spaces, from the exteriors of buildings to the interior since the dawn of modern architect. More recently, glass, originally a non-structural component, has been applied as a structure component. Examples include the type of tempered glass guard and handrail system studied here, typically installed for transparency purposes along streets outside of buildings and inside buildings for the same purpose.
Glass has a critical vulnerability, that is, its failure mode by rapid brittle fractures, which can occur without warnings. Despite the improved performance of tempered glass, its brittleness, a fundamental weakness of the material itself, remains unimproved. The guard and handrail system studied here is a statically determinate structure with a cantilever which is only fixed at the bottom, on the stairs and slabs, for instance. It cannot be expected to promote greater bearing capacities through a redistribution of force. Moreover, when the connection has geometric discontinuity, known as a notch, it is inevitably accompanied by concentrated stress which triggers the occurrence of the cracks. The glass guard and handrail system with these unfavorable conditions (i.e. material brittleness, low redundancy of the structural system, and a stress concentration at a connection) should be regulated strictly so that a high structural safety grade can be ensured.
Various factors which affect the structural safety of guard and handrail systems including the tempered glass as a component have been studies by researchers and practical structural engineers. Most of them concentrate on several human ergonomic factors related to graspability and the height of the handrails. Maki et al.1,2 investigated the influence of the handrail height on the capacity of stairway users through an experimental study. Dusenberry et al. 3 addressed the influence of the size and shape of the handrails on the grasping force through various experiments and analyses. Schneider et al. 4 conducted an experimental study on the empirical characteristics of the resistance of annealed and tempered glass against surface scratches. Dembele et al. 5 studied on important parameters for the thermal breakage of the window glass in room fires condition. Wang et al. 6 analyzed the bearing capacity of tempered glass panel in pint supported glass facades against in-plane load. Beden et al. 7 investigated design methods, existing research, current issues, and trends for structural glass facades under extreme loading. However, the structural performance of an appropriate connection method, especially for a guard and handrail system without a reinforced frame, has not yet been addressed.
Therefore, this article summarizes the structural performance evaluation results of a tempered glass guard and handrail system as a cantilever structure located in stairways and indoor and outdoor walkways. After manufacturing full-scale specimens to simulate the actual constructed situation of a structure and the connection condition, the serviceability and safety for the out-of-plane loading are investigated and design recommendations are proposed for practical structural engineers.
Experimental program
An experiment was done to evaluate the structural performance of both the tempered glass and the connection used in the guard and handrail system. The experimental key parameters are the connection method of the tempered glass, the shape of the tempered glass, and the experimental loading protocol. A bolted connection for the handrail and a plaster-filled connection on both sides of the tempered glass, both widely applied connection methods in the manufacturing of the guard and handrail system, were investigated here (see Figure 1(a) and (b)). Tempered glass samples rectangular in shape and with a parallel shape in accordance with the two connection methods were adopted. Both a monotonic loading protocol and a cyclic loading protocol using manufactured thin steel plates with various weights were established. In this experiment, a large loading amount was unnecessary, but a loading device with a very small margin of an error was required. The loading experiments were conducted using thin steel plates in this study, as the actuator considered in the initial process in this study has significant measurement error of ±1 kN. The monotonic loading experiment forced a uniformly distributed load and a concentrated load onto the edge and the corner of the specimen, respectively. The cyclic loading experiment repeatedly and alternately forced a concentrated load in both directions, that is, at the center of the edge and the corner of the specimen.
View of the constructed guard and handrail system with the bolted or the plaster-filled connection: (a) bolted connection and (b) plaster-filled connection.
Specimen fabrication
Eight full-scale specimens were fabricated as a connection condition equal to a realistic situation was simulated (see Table 1). Specimens with the bolted connection were connected and pierced with five bolt holes at the designated locations (see Figures 2(a), (c), and 3(a)). For the specimens with the plaster-filled connection, they were filled with plaster after inserting the tempered glass in a manufactured jig, after which they were closed with a sealant (see Figures 2(b) and 3(b)).
Experimental key parameters of the specimens.
Specimen name: Shape of the tempered glass—Connection—Loading type (REC: RECtangular shape; B: Bolted; U: monotonic Uniform loading; C: monotonic Concentrated loading at corner point; P: Plaster-filled; PA: PArallelogram; CM: Cyclic concentrated loading at the Mid-point; CC: Cyclic concentrated loading at the Corner point).
Details of the guard and handrail specimens: (a) bolted specimen with the rectangular tempered glass, (b) plaster-filled specimen with the rectangular tempered glass, and (c) bolted specimen with the parallelogram tempered glass.
Two connection methods of the specimens: (a) bolted connection and (b) plaster-filled connection.
As shown in Figure 2(a)–(c), specimens for the monotonic loading experiment were installed in the horizontal direction to evaluate the flexural capacity. The plaster-filled connections were reinforced with stiffeners. In order to prevent the formation of a gap between the tempered glass and the filler during the curing period, the tops of these samples were enhanced using ribs. In order to prevent out-of-plane deformation, the jig on which the specimen was established was reinforced by stiffeners. Three differently shaped jigs fabricated depending on the connection method were installed onto the planned specimens, respectively. The jig and the reaction frame were connected by bolts. The upper steel plate of the bolted specimen protruded in order to join the tempered glass and the bolt for the handrail. An aluminum spacer and a plastic bushing were used between the upper steel plate as an actual stairway frame and the tempered glass and between the bolt for the handrail and the tempered glass, respectively, to prevent the tempered glass from becoming damaged by the jig and the bolts (see Figure 3(a)). In addition, the specimens with the plaster-filled connection used eight holes which had been drilled on both sides of the jig in order to fix in the jig to the removable reaction frame (see Figure 3(b)).
For the cyclic loading assessment, two specimens with plaster-filled connection, predicted to show the greatest flexural capacity levels among the monotonic loading specimens, were fabricated.
Material properties of the tempered glass and others
The tempered glass is created by rapidly cooling a heat-treated plate glass. It is a type of safely glass with enhanced strength and good heat resistance but equal in terms of transparency to an ordinary plate glass. The impact resistance and the static resistance of the tempered glass are five times and three to four times greater than those of the plate glass, respectively. As it has high strength, its fracture rate is low. Even when damaged by a limited impact, it is safe with less risk of injury, as it breaks into small particles. In addition, while ordinary plate glass can fracture if the temperature difference exceeds 60°C, tempered glass has strong heat resistance such that it can withstand temperature changes up to approximately 200°C. The elastic modulus of the tempered glass used is 7.5 × 104 MPa. 8 In addition, the failure probability of the tempered glass is known to be significantly lower than that of general structural members.
The typical field materials for the others including a connector blot, an aluminum spacer, a plastic bushing, a filled plaster, and a sealant were used.
Test setup, loading, and measurement plan
The reaction force of the specimen was supported with a reaction frame. For the monotonic loading experiment, two-stage removable reaction frames were positioned considering the loading device and the vertical displacement caused by it. The specimen was then installed in the horizontal direction. For the loading, the fixed end of a rod connected to the tempered glass utilized a bolted connection (hinge) (see Figure 4(a) and (b)).
Test setup for the monotonic loading experiment: (a) uniformly distributed loading experiment and (b) concentrated loading experiment.
As shown in Figure 5, for the cyclic loading experiment, the specimen was directly bolted to the reaction frame and two-stage removable reaction frames were set on both sides of the specimen. For installation in the vertical direction in the reaction frame directly, two holes each with a diameter of 32 mm on both sides of the jig were drilled. The jig, equal to specimen height, was then fabricated and installed. The weights used for the cyclic loading experiment on the outer parts of both reaction frames were placed to connect with the chain block of the crane on the top of the reaction frame in order to ensure a safe experiment and fixed using wire connected to the jig. The edge of the jig was curved in order to minimize friction, such as that caused by the rollers. In addition, to prevent the yielding of the jig, its center was reinforced with stiffeners. The top surface of the jig was treated with a lubricant in order to minimize the friction with the wire used for the cyclic loading. A hinge connection between the specimen and the weights was manufactured and set using steel wire and a jig. This load carrying system was used to apply all the loading weight to the edge of the tempered glass.
Test setup for the cyclic loading experiment.
The displacement of the specimen was measured using a LVDT (linear variable differential transformer). The acquired data were recorded and saved through a UCAM 70A device. For the monotonic loading experiments, conducted separately according to the use of a uniformly distributed load and a concentrated load, displacement meters were installed. For the uniformly distributed loading experiment, three displacement meters were installed at the center and on both ends (see Figure 4(a)). For the concentrated loading experiment, two displacement meters were located on both ends (see Figure 4(b)). In the case of the cyclic loading experiments, as shown in Figure 5, two displacement meters on the front side and a displacement meter on the back side were installed.
For the monotonic loading experiments, to charge the self-weight of the tempered glass and force an additional load, the device established in the tempered glass is acted on as the initial load. The additional load was applied as a force onto the fracture point with lighter weights used around the fracture point initially, followed by heavy weights (see Figure 6(a)). The uniformly distributed load was constantly applied in the vertical direction from the edge of the tempered glass. The concentrated load was designed to be applied at the corner, which is the most unfavorable point for an external load.
Pseudo-static loading protocol: (a) monotonic loading protocol and (b) cyclic loading protocol.
The cyclic loading experiments were conducted as both sides of weights of 2.45 kN, which is roughly identical to a proof load level mentioned in the next section, were connected to the specimen. A crane was located on top of the reaction frame, and both sides of the chain block of the crane repeatedly moved up and down (see Figure 6(b)). These components were set at the center and the corner, respectively.
Finally, the tempered glass is painted with colored lacquer to effectively observe the structural response of the tempered glass.
Performance evaluation criteria
In the initial step of this study, it was difficult to select an experimental method after a review of the literature and the guidelines related to the guard and handrail system of the type used in this study. For this reason, AC2739 was referenced; it is considered in the structural specifications of buildings, such as IBC 2015 10 , IRC 2015, 11 and others. AC273 suggests a limited load for guard and handrail systems, an allowable displacement and others. The experiment conducted here therefore considered these items.
The proof load to assess the required bearing capacity of the structures according to AC273 is defined as the load level multiplied by a load factor (or a safety factor) of 2.5 of the service load of the guard and handrail system in the relevant specifications (IBC 2015, IRC 2015, ASCE 7-16 12 and others). The size of a service load of the tempered glass in the specifications is regulated to press over a concentrated load of 200 lb (= 0.89 kN) at the point expected to be most unfavorable for the bearing capacity and over a uniform distribution of 50 plf (= 0.73 kN/m). Therefore, in this study, the proof load was set to 500 lb (= 2.22 kN) for the concentrated load and 125 plf (= 1.82 kN/m) for the uniformly distributed load. It was evaluated by a pseudo-static monotonic loading experiment or a cyclic loading experiment.
For the allowable displacement of the guard and handrail system, AC273 and ASTM D7032-17 13 cover only a concentrated load. ASTM D7032-17 prescribes the limitation of ((guard height / 24) + (effective guard length / 96)) at the midpoint of the top rail and (effective guard height / 12) at the end point when it reaches the service load. However, AC273 regulates that the allowable deflection of the top rail shall be exceeded (effective guard length / 96) if the post is sufficiently laterally braced. In this study, the limitation of ASTM was adopted since specimens were installed without the lateral brace.
Moreover, AC174 14 related to the loading time suggests 10 s to 5 min, and a sustainable load of more than 1 min is defined as the proof load of the structures. However, the experiments were conducted with a loading time of 30 min due to the inevitable process time of the loading procedure in this case. Thus, the experimental results may be analyzed conservatively.
Structural performance evaluation
Given that glass is a very brittle material, it is instantaneously destroyed during the loading process. Therefore, minimizing factors that affect the structural performance evaluation of the specimen due to a shock which may occur during the loading experiment was important. In this case, a safety film was attached to the surface of the tempered glass to simulate actual field construction conditions. Even if the tempered glass suddenly breaks when the capacity of the specimen reaches the fracture load, this measure will prevent the splintering of glass when it shatters and falls off, as a plate glass will do. Therefore, this was done to avoid injuries which can be caused by the fracturing of the specimen. The experiments were conducted with various safety devices installed.
Failure mode
Monotonic loading guard and handrail specimen
A REC-B-U specimen and a PA-B-U specimen for which the rectangular tempered glass and the parallelogram tempered glass were connected by bolts, respectively, were used to evaluate the structural performance under the uniformly distributed load onto the edge of each specimen. Specimens with the bolted connection showed the similar behavior in accordance with the loading method. The REC-B-U specimen and the PA-B-U specimen were suddenly destroyed due to the stress concentration phenomenon transferred to the bolt row located at the front of the bolt connection. This can be observed from the fracture pattern around the bolt connection after the destruction of the tempered glass (see Figure 7). The safety film attached to the tempered glass was very effective at preventing the scattering of glass splinters when the tempered glass was destroyed (see Figures 8–13, and 15).
Fracture mode of the bolted connection due to the stress concentration phenomenon.
Fracture mode of the REC-B-U specimen.
Fracture mode of the PA-B-U specimen.
Fracture mode of the REC-B-C specimen.
Fracture mode of the PA-B-C specimen.
Fracture mode of the REC-P-U specimen.
Fracture mode of the REC-P-C specimen.
A REC-B-C specimen and a PA-B-C specimen for which the rectangular tempered glass and the parallelogram tempered glass were connected by bolts, respectively, were used to investigate the structural performance under the concentrated load at the most vulnerable corner as an external load. As in the experimental case of the uniformly distributed load, the stress concentration phenomenon was spread with the bolt row was arranged such that it was closest to the front face from the loading position, causing these experimental results to show a sudden brittle fracture mode in the bolt connection. For the REC-B-C specimen, the connection was completely removed, unlike in the uniformly distributed loading specimens (see Figure 10). For the PA-B-C specimen, because the attached safety film prevented the scattering of glass splinters, it prevented the tempered glass from becoming completely separated from the connection (see Figure 11).
A REC-P-U specimen with the rectangular tempered glass and the plaster-filled connection was used to investigate the structural performance under the uniformly distributed load. Overall, the experimental results indicated that the specimen with plaster-filled connection had the best performance regardless of the shape of the tempered glass adopted in this study. After the final fracture stage, as with the REC-B-U specimen, the connection of the REC-P-U specimen was partially solidarized. Moreover, its shape was not significantly distorted if the safety film attached to the bottom surface of the tempered glass was damaged (see Figure 12).
Specimens with bolted connections were destroyed by the excessive stress concentration phenomenon occurring at the area of the connection. However, because the REC-P-U specimen minimizes the stress concentration phenomenon induced by the bolt connection filling the plaster throughout the connection around the tempered glass, it was found that the entire width of the connection resists the loading force. In other words, the guard and handrail system with the plaster-filled connection may exhibit sufficiently the potential performance of the structure with an excessive nominal bearing capacity embedded in the tempered glass.
A REC-P-C specimen with the rectangular tempered glass and the plaster-filled connection was used to evaluate the structural performance under a concentrated load at the most vulnerable corner. This connection was entirely destroyed (see Figure 13). It presented a problem in which a part of the loading jig pushed the tempered glass as the tempered glass was also over-deformed when the considerable weight of the loading frame installed at the corner of the tempered glass was added. These sequences show directly the progress to the fracture of the specimen and how it induced the stress concentration phenomenon in the portion connected to the tempered glass with the jig.
Cyclic loading guard and handrail specimen
A REC-P-CM specimen with the rectangular tempered glass and the plaster-filled connection was used to investigate the structural performance under a cyclic concentrated load at the center of the edge of the tempered glass. To identify the status of the connection during the experimental processes, the experiment was conducted after removing the sealant on the plaster. In this experiment, additional secondary moment is generated by the lateral displacement and the axial force action, as the self-weight of the specimen and part of the jig together apply axial force and because the loading force substitutes for the lateral load toward the left and the right of the specimen. Thus, a conservative interpretation of this experimental result is possible.
Because this specimen resisted even cyclic loading of more than 120 times, it was concluded that this experiment establishes that sufficient rigidity exists. Figure 14 shows the damaged connection after the end of the experiment. The longitudinal filled plaster along the tempered glass was broken into small pieces by the end of the experiment. It was also found that both the filler of the connection and the tempered glass have a crack width of approximately 10 mm. Typically, the guarantee period of the tempered glass is 10 years; hence, it is expected that the long-term cyclic loading would induce a continuously progressing crack in the depth direction even if a difference in the degree of damage in the plaster exists in the connection of the specimen. In other words, the guard and handrail system with plaster-filled connection may appear to be a problem that cannot investigate the degree of damage of the connection. Therefore, the long-term safety of the guard and handrail system with the plaster-filled connection established at various points in human traffic patterns may not be ensured. However, if an unexpected problem does not occur during the construction process, the connection as well as the entire structure of the guard and handrail system will likely exhibit sufficient structural performance. This is why, when an impact load equal to the experimental situation of the guard and handrail system here but existing in an ordinary building does not arise repeatedly and continuously, as in this experiment, the extreme resisting status of the specimen is assumed.
Damaged connection for the REC-P-CM specimen.
A REC-P-CC specimen with the rectangular tempered glass and the plaster-filled connection was used to investigate the structural performance under a cyclic concentrated load at the corner of the edge of the tempered glass. After removing the sealant, this experiment and one with a REC-P-CM specimen were carried out. The specimen was destroyed at a moment that cyclic loading of 61 times was completed (see Figure 15). However, as described above, because the possibility that an impact load of more than 60 times in an actual situation with an actual constructed guard and handrail system generates repeatedly and continuously is extremely insignificant, such a system will also satisfy the required performance level for the specimen.
Fractured status for the REC-P-CC specimen.
Figure 16 shows the failure mode in the longitudinal direction for the connection of the REC-P-CC specimen. The filled plaster was damaged approximately 10 mm in the oblique direction from the outside to the inside and from both sides of the tempered glass toward the depth direction. This is why this specimen, unlike the REC-P-CM specimen, was fractured.
Fractured connection for the REC-P-CC specimen.
Performance evaluation according to the required resistance capacity and allowable displacement
Monotonic loading guard and handrail specimen
To assess the structural performance of the guard and handrail system, a response corresponding to the initial load as determined to calculate the self-weight of the tempered glass and the jig for loading was considered as the initial deflection. As mentioned above, AC273 recommends the allowable displacement for the service load as well as the maximum proof load. In this study, the performance of the guard and handrail system was evaluated per this report.
Figure 17 shows the load and displacement relationship of the monotonic loading specimens. They overall present typical linear elastic and brittle behavior. Regardless of the shape of the tempered glass, the specimens with the plaster-filled connection showed superior performance compared to the specimen with the bolted connection. As described in the previous section, the specimens tested under a uniformly distributed load show excellent performance in comparison with those tested under a concentrated load, under which a brittle failure arose due to the stress concentration phenomenon in the bolt row located at the front side of the bolted connection, as the entire width in this case resists the external load.
Load and displacement relationship for the monotonic loading specimens.
Table 2 summarizes the results of the structural performance evaluation in terms of the proof load of the monotonic loading specimens per AC273. The fracture load of the REC-B-U specimen is 2.5 kN and its fracture displacement is 54.62 mm. The safety factor, which is the fracture load of the specimen divided by the service load, is 1.69. The fracture load and the fracture displacement of the REC-B-C specimen are 1.79 kN and 91.88 mm, respectively. As the safety factor is 2.01, it is relatively high compared to the uniformly distributed loading specimen.
Structural performance evaluation in terms of the proof load of monotonic loading specimens per AC273.
Specimen name: Shape of the tempered glass—Connection—Loading type (REC: RECtangular shape; B: Bolted; U: monotonic Uniform loading; C: monotonic Concentrated loading at corner point; P: Plaster-filled; PA: PArallelogram).
Safety factor: fracture load / service load.
The fracture load of the REC-P-U specimen is 4.8 kN and its fracture displacement 113.57 mm. The safety factor is 3.65. The fracture load of the REC-P-C specimen is 3.06 kN and its fracture displacement 132.63 mm. As its safety factor is 3.43, it is evaluated as being lower than the uniformly distributed loading specimen. However, as mentioned above, the specimen with the plaster-filled connection was destroyed by the unexpected contact between the tempered glass and the experimental jig. Considering the final result of this experiment in conjunction with the experimental results of the specimen with the bolted connection with regard to this problem, it may be that the specimen with the plaster-filled connection may have undergone a somewhat higher fracture load and larger displacement corresponding to this load than those derived in the experimental result.
The fracture load of the PA-B-U specimen is 3.22 kN and its fracture displacement is 77.87 mm. It has a safety factor of 1.87, a value that is higher than that of the REC-B-U specimen. The fracture load of the PA-B-C specimen is 2.07 kN and its fracture displacement is 92.83 mm. As its safety factor is 2.32, the safety of this specimen was highly assessed in comparison with the uniformly distributed loading case.
Considering the safety factors of all specimens, although the proof load of the specimens under the uniformly distributed load is established higher compared to the concentrated load, the safety factor of the specimens under the concentrated load is higher because the specifications related to a guard and handrail system require a higher performance level for a uniformly distributed load. Regardless of the loading method, the specimens with bolted connections have similar values. It was found that the specimens with the plaster-filled connection are secured with regard to safety by approximately 1.7–2.2 times compared to the specimens with the bolted connections. In other words, this implies that the safety of specimens with bolted connections should be ensured through structural reinforcements, as they are all inferior compared to the required safety factor per the specification (= 2.5).
Table 3 summarizes the results of the structural performance evaluation in terms of the allowable displacement for the concentrated load per AC273. The limitations of the allowable displacement for the REC-B-C specimen, the REC-P-C specimen, and the PA-B-C specimen are 91.25, 93.75, and 92.93 mm, respectively. The displacement for the service load in these cases occurred at 40, 40, and 35 mm, respectively. It was therefore found that they present excellent performance.
Structural performance evaluation in terms of the allowable displacement of the service load of monotonic loading specimens per AC273.
Specimen name: Shape of the tempered glass—Connection—Loading type (REC: RECtangular shape; B: Bolted; C: monotonic Concentrated loading at corner point; P: Plaster-filled; PA: PArallelogram).
Cyclic loading guard and handrail specimen
Figures 18 and 19 present the load–displacement relationship of the REC-P-CM specimen and the REC-P-CC specimen, respectively. Their maximum displacements are 87.3 and 125.8 mm, respectively. Investigating the displacement corresponding to the service load (= 2.45 kN) obtained from the monotonic loading experiment on the specimen with identical connections, the displacement for the uniformly distributed load is approximately 56 mm and that for a concentrated load is about 107 mm. Although it is impossible simply to compare this given the different conditions with regard to the experiment, large displacement arises due to the P-delta effect and the loading velocity as caused by the large weights at once unlike the monotonic specimens.
Load and displacement relationship for the REC-P-CM specimen.
Load and displacement relationship for the REC-P-CC specimen.
However, despite the cyclic loading of these proof load levels, the degradation due to rigidity in this connection is slight. For example, the degradation due to rigidity of the REC-P-CM specimen and the REC-P-CC specimen is only 26.58% and 16.79%, respectively. Therefore, this specimen satisfies the performance level requirements for cyclic loading.
Summary and conclusions
A full-scale experimental study of monotonic loading and cyclic loading in order to evaluate the structural performance capabilities of a guard and handrail system with tempered glass without any structural reinforced framework was conducted here. The main results of this study are summarized below.
For monotonic loading, all specimens showed typical linear elastic and brittle behavior. As the connection of this structural system is a brittle system without energy dissipation by ductile behavior, it needs to be designed and constructed to have a relatively high safety margin because there is no advance warning through inelastic behavior up to the point of fracture. Considering that this system is a statically determinate cantilever structure not expected to redistribute the force, this applies to this system even more so.
It was observed that fractures of the specimens with bolted connections are generated by the stress concentration phenomenon at the opening of the bolt connection located along the stress-transferring path closest to the loading point. The ensured safety factors of these specimens are 1.69–2.32 according to the stress concentration phenomenon, never reaching the stipulated level of 2.5. When adopting the bolted connection, it is necessary to devise an advanced bolted arrangement or to improve the upper cap to relieve this stress concentration phenomenon.
The specimens with the plaster-filled connection exhibited excellent performance, with a safety factor ranging from 3.43 to 3.65. Because a continuous connection is established through the plaster-filling, this explains why the stress concentration phenomenon is drastically mitigated, unlike with the bolted connection.
As the lateral displacement under a service load observed in the experiment ranged from 30 to 45 mm, the specimens satisfied enough the allowable displacement of 90 mm per AC273.
Cyclic loading experiments were conducted in order to identify the strength of the plaster-filled connection or whether or not its rigidity is mitigated. For cyclic loading at the center, there were little degradation of the rigidity or the strength was up to as many as 120 cycles. However, delamination of the contact surface was observed in accordance with the accumulation of the pressed deformation of the plaster for the connection. Even for cyclic loading in a corner under an extremely disadvantageous loading condition, the fracture of the connection occurred during the 61st cycle. Overall, the specimens with the plaster-filled connection were found to retain an excellent bearing capacity even under cyclic loading.
Full-scale experimental information on guard and handrail systems with glass is very rare. The results of this study can serve as a reference to increase the understanding of practical structural engineers of the characteristics of brittle behavior and the risk and improvement possibilities during the design and the construction of these systems in the future.
Footnotes
Declaration of conflicting interests
Funding
ORCID iD
