Experimental Tests of Laminated Glass with Embedded Steel Mesh Subjected to In-Plane Loading

This study used a commercial soda–lime–silicate (SLS) float glass, which is the most popular glass type used in building construction. Architectural elements made of glass mounted with point fixings usually require the application of heat-treated glass due to the high stress concentrations around the holes [26]. For this reason, fully tempered glass was used to produce samples.

An ethylene-vinyl acetate (EVA) interlayer was used in the lamination process of the glass panels. EVA film is increasingly used as an alternative to the polyvinyl butyral (PVB) interlayer. This is due to its low lamination temperature and the fact that no autoclave is required in the manufacturing process, making it a more cost-effective alternative to the standard PVB interlayer [27]. In addition, EVA shows better resistance to moisture (it is not hygroscopic) and is less likely to delaminate than PVB, so it is ideal for wet conditions, for example, external balustrades and glass canopies. In addition, it should be mentioned that EVA film shows a lower tensile strength than PVB. Its strength is between 9.5–10.0 MPa, while, in the case of PVB foil, it is 20.8 MPa [28].

The reinforcement between the two layers of EVA Clear foil was made of EN 1.4301 woven steel mesh [29]. It is an austenitic, chlorine-nickel, low-carbon steel.

All of the steel elements of the test stand were made of steel grade S355 [30]. The 20 mm diameter bolts were made of grade 8.8 steel, resulting in a shear capacity of 117 kN [31]; this is much higher than the tensile strength of the samples assessed in preliminary studies, which were made before main tests.

In order to reduce the stress concentrations in glass due to the direct contact with steel elements, 2.5 mm thick aluminium spacers were used [32].

Laminated glass elements were used in the experimental campaign (Figure 1). The dimensions of the samples were 190 × 300 mm² and consisted of two panes made of fully tempered glass (8, 10, and 12 mm in thickness) and two layers of EVA Clear film (1.52 mm in thickness). A steel woven mesh was placed between two layers of EVA Clear film before the lamination process. In each of the samples, three holes with a diameter of 26 mm were made using a waterjet technique. Two of them were at the bottom of the sample, while the third was at the top. Such a design of the hole locations was intentional, to force the sample failure in the upper hole. The exact location of the holes can be seen in Figure 1.

Figure 1.

Shift of components panes was found in some samples, see Figure 2. This feature occurred in the production process. The shift affected the position of the holes relative to each other and their non-collinearity which may influence the in-plane load transfer of both panes in the experiments. The standard [33] describes the displacement tolerance of component panes in laminated glass relative to each other, amounting to a maximum of 2 mm for elements with a length not exceeding 1000 mm.

Figure 2.

The tests were carried out on reference samples and reinforced samples. The reference samples without reinforcement (REF) only consisted of glass panes and interlayers, while in the case of reinforced samples (MESH), a steel woven mesh was placed between two interlayers over the entire surface of the sample, before the lamination process. The steel mesh consisted of wires (0.35 mm in diameter) at a spacing of 1×1 mm². In both cases, the thickness of the interlayer (3.04 mm) was the same. A visual comparison of the reference and reinforced samples is presented in Figure 3. The study considered three thicknesses of glass pane (8, 10 and 12 mm), divided into two subtypes – reference and reinforced samples. This led to 6 test series of 6 samples each, for a total of 36 samples (Table 1).

Figure 3.

The configuration of the tensile strength experiment is shown in Figure 4. The experimental setup consisted of two handles and four mounting plates. The handles were made of two sheets, with dimensions 60 × 140 mm², and a hole of 24 mm diameter. In the lower part of the handle, there was a pin for attaching the handle to the testing machine. The lower part of the sample was fastened to the holders with two fixing plates of dimension 130 × 150 mm². The upper fixing plates had dimensions of 60 × 150 mm². The appropriate number of holes necessary for mounting the test sample was made in the lower and upper mounting plates. Five 20 mm diameter screws were used to assemble the sample. The upper handle and the upper mounting plates were used to introduce displacement during the experiments. The tensile strength test of the laminated glass was carried out in a displacement-controlled testing machine with a displacement speed of 10 mm/min. The tests were performed at room temperature and a relative humidity of 50%.

Figure 4.

During the tests force and displacement of the cross-head were monitored. Load cell and cross-head displacement signals were recorded and stored in an external data acquisition device. In addition, a linear differential transformer (LVDT) with a measuring length of ±12 mm was mounted to the upper handle, which was used to measure the local displacement of the tested samples. In the sample preparation phase, the value of the relative shift was controlled. All signals were acquired using an external data acquisition device with 50 Hz data acquisition. Pictures of the test stand are shown in Figure 5.

Figure 5.

Figure 6 shows exemplary results for a selected series (G12 series, glass thickness 12 mm). The graph compares force-displacement diagrams for a reference sample (REF) and a sample with embedded steel woven mesh (MESH). The values on the horizontal axis were derived from the cross-head displacement. The results of other test series were analogous in terms of global behaviour.

Figure 6.

The behaviour of the samples during testing can be divided into three main phases: elastic, force drop and post-breakage. In the first phase, the relationship between the load and the displacement is almost linear and refers to the perfectly linear-elastic response of the composite materials to the load. As a result of the uniformly increasing load, the stress in the glass increases until a limit stress value is reached, and the glass fractures in a brittle manner. During testing, the stiffness of the sample in the elastic phase was the same for reference and reinforced samples. Due to the arrangement of the holes (see Section 2.2), the origin of the fracture is located at the top hole.

Following the glass failure, there is a sudden drop in the force due to the almost complete loss of tensile stiffness of fully tempered glass due to fracture (second phase). After this stage, progressive degradation of the sample occurs due to further cross-head movement (third phase). The post-breakage behaviour differs for the reference (REF) and reinforced samples (MESH). The latter shows increasing load due to the presence of the reinforcing mesh, while in the case of the reference sample, the load is almost constant and relates almost completely to the stiffness of the inter-layer. Fully tempered glass breaks into many small pieces and bears tensile loads only for a short time with the interlayer, contributing to the load transfer. Because of this, the force does not drop to zero. Then, due to the separation of the cracked glass fragments, the interlayer is the only component of the laminate that allows the sample to carry a tensile load. It was noticed that samples with embedded steel mesh retained higher values of residual force after glass breakage.

In the second phase, two modes of laminated glass failure were observed. The first mode occurred when two panes broke simultaneously (Figure 7a). The second mode involved non-simultaneous breakage of the panes. The fracture of a single pane (Figure 7b) caused a force drop and then a force increase to the value at which the second glass pane failed. The reason for the two failure modes is the initial, relative shift of the glass panes in some samples which was revealed before testing (see point 2.2). In the case of the first model, the force was successfully introduced to both panes, while the second mode is the result of the inequality of forces transmitted by the panes and the impossibility of their equalization due to the low stiffness of the interlayer and the aluminium spacers.

Figure 7.

In the third, post-breakage stage, the samples exhibited behaviour where the force increased to a certain level and then slowly decreased until the interlayer was degraded entirely and the upper screw ultimately passed through the sample (Figures 7c and 7d). For all specimens, the load increased after the force drop was observed. The specimens reinforced with embedded steel woven mesh showed higher post-breakage capacity. The post-breakage capacity is assumed to be the ratio of the maximum force after glass fracture F_cr,max and the maximum force in the elastic phase F_el,max. Such behaviour of the reinforced samples indicates the woven steel mesh's beneficial effect on the element's load-bearing capacity in the post-critical phase. The steel mesh activates and redistributes local stresses in the interlayer over a larger area allowing for carrying additional load.

Figure 8 shows schematic representations of the global behaviour of the samples during the experiments. Figure 8a shows the first failure mode, where two glass panes break simultaneously, while Figure 8b shows the second failure mode (progressive glass breakage). The notation in the index “el” refers to the first phase of the test (elastic phase), and the designation “cr” refers to the second phase (cracked phase).

Figure 8.

The average values of forces and displacements at characteristic points in the global force-displacement behaviour are listed in Table 2 along with the corresponding standard deviations. During the tests, 20 out of 36 samples failed according to the first failure mode, which is 56% of the total number of samples. The remaining 44% comprise 16 samples which fractured according to the second failure mode.

The first value obtained from the tests was the force at which the glass fractured (F_el,max). Since in the elastic phase, the effect of the interlayer on the glass load capacity is negligible, the results from the reference and reinforced samples can be considered together. In the case of samples made of thicker glasses, an increase in force at glass breakage was observed. In the first mode (simultaneous fracture of both glass panes), the average force F_el,max,T was recorded for samples G8, G10 and G12 at 11.48±3.32, 15.70±5.45 and 19.65±4.40 kN, respectively (average variation of 29%). The local displacement disregarding the deformation of the test system u(F_el,max,T) obtained from the local LVDT measurements was 0.95±0.34; 1.41±0.42 and 1.20±0.38 mm for samples G8, G10 and G12, respectively. Dependence u(F_el,max,T) for various glass thicknesses seems nonlinear, however, it should be noted there is a relatively large standard deviation for the mean values. The reason for this may also be the LVDT measurement error. In the second mode (progressive breakage of panes), the values of force F_el,max,M1 was obtained at 6.81±2.24, 13.52±4.01 and 23.37±9.35 kN (average variation 34%) and u(F_el,max,M1) was 0.95±0.36, 1.06±0.30 and 1.88± 0.13 mm for samples G8, G10 and G12, respectively. In turn, the values of forces F_el,max,M2 were obtained as 12.00±2.72, 17.94±7.95 and 29.00±6.15 kN (average variation 29%) and u(F_el,max,M2) were 2.06±0.48, 1.85±0.59 and 2.40±0.27 mm for samples G8, G10 and G12, respectively. The average variation of F_el,max amounted to 34, 29 and 29%, respectively. This may seem relatively high for the first and second failure modes, but these values are often found for materials as brittle as glass [34].

It was discovered that when there was a progressive breakage of glass panes (the second failure mode), the F_el,max,M2 force for the failure of the second pane was higher than the F_el,max,T force (the failure model of both panes breaking at the same time). This is because when one pane is fractured, the toughening compression energy was released and the reinforcing mesh was activated through the interlayer. Therefore, the force F_el,max,M2 is the combination of the pane capacity and partial load capacity of the reinforcing mesh.

In the fracture phase, the average force F_cr,min was recorded for reference samples G8, G10 and G12 at 0.06±0.01, 0.08±0.02 and 0.06±0.01kN, respectively (average variation of 20%). For reinforced samples, the values were 2.52±0.34, 2.32±0,46 and 2.15±0.45 kN, respectively (with an average variation of 18%).

For the maximum force in the post-breakage phase F_cr,max, the values for the reference samples were 1.24±0, 1.17±0.13 and 1.23±0.13 kN (average variation of 10%) for samples G8, G10 and G12 respectively. In turn, the reinforced samples (G8, G10 and G12) obtained values 4.80±0.22, 4.92±0.22 and 4.67±0.61 kN (average variation of 7%).

The maximum displacement at sample failure u_cr,ult was obtained as 96.34±2.02, 98.29±10.04 and 99.68±11.73 mm (average variation of 8%) for G8, G10 and G12 respectively. For reinforced samples with glass thicknesses of G8, G10 and G12, the values of 87.20±10.04 85.43±12.04 89.55±8.28 mm (average variation of 12%).

The effect of strengthening the laminated glass with a steel mesh is revealed as soon as the glass is fractured. Figure 9 shows the percentage decrease in force after glass breakage from F_el,max to F_cr,min. The analysed values are presented in two separate graphs. Figure 9a shows results for the first failure mode (F_el,maxT), and Figure 9b shows the results for the second failure mode (F_el,max,M2). Regarding a force drop for reference samples (REF), the values in the graph were basically the same for all glass thicknesses (average 99.50±0.16 %). In the case of reinforced samples (MESH), a difference in the results can be seen in favour of samples with a lower glass thickness. The reason for the variation in the results of the reinforced samples may be that the presence of a steel mesh in the interlayer increases its resulting stiffness while reducing sudden deformations after glass failure. Reduced deformations of the glass laminate have a positive effect on the post-breakage stiffness, which depends on the mechanical properties of the interlayer, as well as the ratio of interlayer thickness to glass thickness [35].

Figure 9.

Figure 10a shows the average values of F_cr,max obtained in the post-critical phase for all tested samples. The reference samples (REF) values are almost identical for all glass thicknesses and amount to an average of 1.21±0.13 kN (without distinction due to glass thickness). In the case of reinforced samples (MESH), the average values are 4.80±0.41 kN, which corresponds to an almost threefold increase in relation to the reference samples. This proves that, in the REF samples, tensile force is only transferred through the interlayer, without the participation of glass. In the case of reinforced samples, the interlayer and the steel woven mesh resist the load.

Figure 10.

Based on the values presented in Table 2, the effect of reinforcement on the samples’ post-breakage load capacity was determined. Figure 10b shows the ratio of globally averaged F_cr,max / F_el,max ratio which can be defined as a residual force index. Based on the residual force index, it is possible to determine the potential of reinforcing with a steel mesh on the load capacity of the element in the post-breakage phase. Reinforced samples showed an increase in residual force index over reference samples of 287, 321 and 279% for the series G8, G10 and G12, respectively.

The lack of a linear relationship in the index may be due to the high standard deviations of the F_el,max force. The results from Figure 10b were globally averaged across glass thicknesses and normalized to the results of the reference (non-reinforced) samples. After normalization, it was found that placing a steel woven mesh between the layers of the EVA film increased the laminate's post-breakage load capacity by on average almost 300% compared to the reference samples.

Figure 11 shows the average displacement u_cr,ult at which the samples completely fail (see Figures 7c and 7d). The graph shows that the reference samples failed at the average cross-head displacement of 98.21±9.37 mm, while the reinforced samples failed at 87.40±10.37 mm. The reason for the difference is not clear. It may be associated with a change in the arrangement of cracks by the use of the reinforcing mesh or be a result of considering combined deformations of the sample and test stand since the displacement was controlled by the cross-head displacement of the machine.

Figure 11.