*4.1. Results Obtained during Cyclic Loading*

Figure 5a shows the real progress of loading. The graph shows that the original load increment of 5 kN/m<sup>2</sup> was reduced by half. The reason for this was the formation of the first cracks in the substrate slab. During the following four cycles, the step-increase in strain did not occur (i.e., it corresponded to load), which is why the last loading during cycle No. 8 was increased by the original 5 kN/m2 to a total of 35 kN/m2. Figure 5b shows a graph depicting the dependence of AE counts on time (blue dots) and the dependence of the relative strain of the slab on time (red curve). The strain gauges were placed beneath the additional concrete layer, which is why the values represent tensile strain. This is an average value made up from data from three strain gauges. However, the record of the relative strain is not complete but ends at a point where one of the gauges stopped measuring; i.e., just before reaching the maximum load during the eighth cycle (the second strain gauge stopped measuring during the holding of the maximum load of the eighth cycle, the third worked until the end of the loading). The highest number of AE overshoots was recorded during the 2nd and 3rd loading cycle, during which the relative strain increased dramatically as well. It therefore appears that the concrete slab suffered the most substantial cracking during these cycles. During the other loading cycles, nothing significant in terms of acoustic emission had occurred.

**Figure 5.** Real progress of loading (**a**); AE counts and relative strain of the concrete slab (**b**).

The values of relative strain in the tiles were measured using three strain gauges glued onto their surface (see Figure 3); the values therefore represent compressive strain. The final values are an average of all three strain gauges. For the sake of clarity, graph axes in Figure 6 show the same range of values. System A, as opposed to the other systems, shows very small relative strain and very low acoustic emissions throughout the loading. A major increase in relative strain occurred only during the last cycle.

Systems B and C show a greater increase in relative strain during the third cycle. It is probable that the slab suffered cracks in these areas and that a part of the tile became detached from the concrete substrate. This is further confirmed by the increased AE counts, which most likely indicate a failure within the adhesive or a failure of the substrate/adhesive or tile/adhesive interface. Another more significant increase in relative strain only occurs during the eighth cycle.

System D showed no response to crack development in the concrete in terms of relative strain or AE counts. The reason may be its lower stiffness (i.e., high flexibility) that comes with its thickness of 3 mm—the tile followed the curvature of the slab's deflection and, unlike the 6-mm tiles, had a lower tendency to become detached. System D showed a major change during the sixth loading cycle where its AE counts saw a marked increase, and since the seventh cycle it also showed substantial strain.

The analysis of relative strain shows that the stress in system A (containing the membrane) had been markedly reduced since the third cycle. The difference between the system with the membrane and the ones without is visible since the third cycle and very pronounced since the seventh. The membrane reduced the stress transfer from the concrete substrate onto the tile down to a mere 40% of the relative strain of the other tiles (i.e., strain created in tiles that were glued directly onto the substrate). Even when the slab deflected to an extreme 66 mm, tile A suffered no damage. However, during the third cycle, increased AE counts were observed—a clear phenomenon occurred involving the membrane, possibly its elongation or partial damage to its fibres.

**Figure 6.** AE counts and development of relative strain in each tile (**A**) material system A, (**B**) material system B, (**C**) material system C and (**D**) material system D.

Figures 7 and 8 show the dependence of cumulative AE counts on permanent strain per every loading cycle. The loading was divided into three groups of cycles:


**Figure 7.** Dependence of AE counts on the development of permanent strain in the concrete slab.

**Figure 8.** Dependence of AE counts on the permanent strain in (**A**) material system A, (**B**) material system B, (**C**) material system C and (**D**) material system D.

The figures obtained from measuring the slab (Figure 7) clearly show that the first three cycles caused significant damage, accompanied by an extreme increase in cumulative AE counts. There is also a substantial decrease in AE counts, but also in permanent strain during later cycles after the reduction in the loading cycle. The final loading cycles again show an increase not only in AE counts, but also in permanent strain.

All graphs in Figure 8 have the same scale, except for system A where the scale was changed due to the low number of AE counts. System A, which contains the membrane, shows only a small increase in permanent strain and AE counts compared to the other systems. The membrane seems to have a dampening effect on the strain transfer from the substrate. A small inserted graph shows what the results of system A would have looked like if they were plotted to scale. The results in Figure 8 also show a striking similarity between systems B and C. Thanks to its elasticity, system D with the 3-mm tile stresses the adhesive to a smaller degree than systems with the 6 mm tile. The first six loading cycles did not damage the adhesive enough to affect the bond

The curves in Figure 8 were used in a calculation that determined, using linear regression, the slopes for each loading section, which were later compiled into a correlation matrix; see Table 3. It shows a clear correlation between the behaviour of the concrete slab and system D with the 3-mm slab; i.e., the fact, the system copied the slab's deflections during loading. It also shows an indirect correlation between the slab and systems B and C. This demonstrates the minimal difference between adhesive I and II. The influence of the textile membrane in system A is also visible, as it causes a certain degree of independence on the shape changes in the substrate.


**Table 3.** A correlation matrix of the groups of loading steps from Figure 8.

*4.2. Results Obtained after the Loading*

After all the cycles were completed, the surface of all the tiles was examined using acoustic tracing, where a percussive force was delivered by an impact hammer. The surface of the large-format tiles showed no cracks or defects that would be visible by the naked eye. Figure 9 shows a map of damage that was detected by acoustic tracing. System A, with the membrane, showed no damage to the tile or loss of bond strength. The other tiles that were attached directly showed a separation from the substrate around the cutouts. Systems B through D differed in the size of the unbonded areas. Figure 9 shows the percentages of the damaged areas compared to the total area of the tiles.

**Figure 9.** Areas and their percentages where acoustic tracing revealed a loss of bond strength after the eight loading cycles for system A, system B, system C and system D.

After the loading, the tiles were again tested using UPV in the same way as before the loading. Figure 10 shows a boxplot of the pulse velocities measured in each tile before and after loading. The statistical analysis of all the systems showed that the ultrasonic pulse velocity before and after the loading did not differ in a statistically significant way, which means that the internal structure of the tiles was not damaged. All the damage thus occurred in the bond or the slab.

**Figure 10.** Boxplots of UPV determined by surface measurement on each tile before and after the cyclic loading.
