As with the previous discussion on salt scaling, all of the information presented on this topic is explained in detail in my Ph. D. thesis. Perhaps this point is most relevant when discussing the empirical evidence for the occurrence of the glue spall mechanism. Because it is not possible to present all the evidence here, and keep the attention of my visitors. Therefore, If you are interested in learning more about the damage mechanics of salt scaling, I urge you to familiarize yourself with the theoretical arguments outlined in Chapter 3 of my Ph. D. Thesis, as well as all of the emipirical evidence for the occurrence of the glue-spall mechanism in Chapter 7 of my Ph. D. thesis. If you are really bored, my Ph. D. thesis is nearly 400 pages, and you are welcome to request the whole thing! Here we will discuss direct observations of ice cracking, and the occurrence of edge cracking in a cementitious surface from the glue-spall stress, σgs. In addition, we will investigate results from the warping experiment that provide evidence for the glue-spall mechanism, and a secondary mechanism known as ice wedging.
Figure 1 - Digital image from a freezing experiment with pure water at -18°C.
The picture shows that there is no sign of cracking in the pure ice.
Figure 2 - Digital images from a freezing experiment with
3% NaCl solution at -18°C.
The pictures show cracks in the ice. Comparison of the two images illustrates crack propagation.
Experimental observations with an optical microscope and a cold stage support the conclusions of the viscoelastic stress calculation, which indicated that pure ice will not crack, moderately concentrated salt solutions form brine ice that does crack, and more highly concentrated solutions do not gain strength in the temperature range of interest. Figures 1 and 2 show photos from freezing experiments, where the solution of interest was held in a glass cup. The surface of the cup was sandblasted to promote adhesion. As you can see in Figure 1 the pure water ice shows no signs of cracking, while cracking is clearly observed with ice formed from a 3% NaCl soluition (Figure 2). In addition, comparison of the two images in Figure 2 illustrates crack propagation. Solutions with a solute concentration of 5% or higher exhibited no interesting behavior.
Figure 3 - Photo of a cement plate with a square lattice drawn on the surface with vacuum grease.
The freezing experiments support the assertion that the pessimum concentration arises from the effect of brine pockets on the mechanical properties of ice. Despite the optomistic predictions of the fracture mechanics analysis, it is still necessary to show that cracks in the ice layer will indeed result in surface cracking on the cement surface. To do so we mimic cracking in a pure ice layer by compartmentalizing water on the surface of a cement plate in a warping experiment (Figure 3). After the sample in Figure 3 was frozen with water in the compartments, the area under the partitions was investigated for surface cracks from the glue-spall stress at the perimeter of the ice islands. It was determined that no cracking occurred under the ice islands, while cracking was observed to circumscribe the islands, and run along the edges of the sample. Moreover, a companion experiment was performed where two sets of 4 adjacent compartments where left dry, and it was concluded that no cracking occurred under the partitions between dry compartments, while cracking did occur under the partitions adjacent to wet compartments. Images of the perimeter cracking are shown in Figure 4, which shows that all cracks are concentrated under the vacuum grease partition.
Figure 4 - a) Photo of the area under an ice island, and the
corresponding vacuum grease partitions. The picture shows
cracks that circumscribe the ice island, and the removal of surface latence due to the shear stress at the interface. b) Similar photo
showing cracks under the grease partitions,
including a crack that runs the entire length of the sample, along the top edge.
Figure 5 - Side view of crack that intersects the edge of the sample in Figure 4b. The photo shows that the crack bifurcates
below the interface, which is exactly what fracture mechanics predicts.
Figure 5 shows the side view of a crack that intersects the sample edge in Figure 4b. The picture shows that the crack bifurcates below the interface, as predicted by the fracture mechanics analysis. In addition, to the qualitative evidence of cracking, the mechanical response of the plate was monitored using our warping apparatus. The apparatus monitors the deflection of a simply supported plate when a pool is frozen on the surface. The mechanical response of a sample with compartmentalized water is shown below. We see that intially there is negative deflection which corresponds to bimaterial bending from the thermal expansion mismatch. However, this bending is suddenly reversed, giving way to a positive deflection. The positive deflection is due to ice wedging open surface cracks from the glue spall stress. The glue spall stress is proportional to the thickness of the ice islands. Accordingly, by varying the thickness of the ice islands we find that the stress at which the deflection is reversed, or in other words when glue-spall cracking occurs, is on the order of 1.5 - 2 MPa, in very good agreement with the stress necessary to result in cracking in the cement surface.
Figure 6 - Mechanical response of cement plate when compartmentalized water (Figure 3) is frozen on the surface.
Failure of the sample, for which the mechanical response is shown in figure 6, occurs by ice wedging. When pure water is on the surface of the sample ice wedging dominates the fracture process. Accordingly, all of the cracks are through cracks that are either parallel or perpindicular to the sample edge. The former is due to the fact that ice wedging drives crack penetration, while the latter is due to the fact that the glue-spall cracks form parallel to the vacuum grease partitions, and the ice is compartmentalized by a square lattice. The difference between the damage morpohology realized in this experiment, and an identical experiment with a 3% NaCl solution is discussed on the next page, along with the ice wedging mechanism.