13th International Conference on Fracture June 16–21, 2013, Beijing, China -4- The apparent strain distributions corresponding to load levels of 40, 55, 62, 62.7 and 64.5 kN respectively have been demonstrated respectively in Fig. 4. The apparent xxε at a given load level is inhomogeneous. The regions with apparent xxε value much larger than the average are highly localized and somewhat scattered at relatively lower load levels. For example, see Fig.4 (a) and (b). Such regions may be considered as locally split given the much larger than average values of the apparent xxε . With the increase in compressive load, the regions with apparent xxε much larger than the average become to join up and are distributed almost along a line inclined to the loading direction as shown in Fig. 4(d). Close to the rupture load, the regions with relatively large value of apparent xxε are distributed along a line and from the color pattern in Fig. 4(e) look like a macroscopic crack. The final rupture actually did propagate along that line. The apparent yyε behaves completely different. The area of interest as indicated in Fig. 1(a) is alternated with regions of small apparent yyε (nearly not deformed) and those of larger apparent yy ε . Deformation band structure forms. The number of bands increases with compressive load. The bands tend to be broken at the middle along a vertical line, possibly due to the damage leading to splitting. Distribution of the apparent engineering shear strain xy γ is also localized but much more scattered than that of apparent xxε . In most cases, the largest magnitude of the apparent shear strain is smaller than those of apparent xxε and apparent yyε . In fact, at relatively large compressive load, the apparent xxε has the largest magnitude among all the apparent strain components on the surface. It may be the result from the axial splitting and so the apparent strain xxε can be conveniently used for identification of cracking and damage in our experiments. The deformation band structure in the apparent yyε appears rather unexpected although compaction bands or shear bands have been observed in rocks with high porosity and usually under confinement [13, 14]. Because the porosity in our samples is very low (about 5%), the formation of deformation band can hardly be attributed to the sandstone compaction. Moreover, the banded structure evolves with compressive load even at levels larger than 30 kN, see Fig.4 (f-j) and no hardening behavior has been found in Fig.2 at those load levels. We have also excluded the possibility of artificial factors. Other algorithm has been adopted to calculate the apparent yy ε but with similar results despite the slight difference in magnitude (results now shown). Besides, the precision for displacement measurement is adequate given the high levels of loading applied in those results. The formation of band structure in yyε may affect the failure behavior of the present sandstones under uniaxial compression. Conventionally, axial splitting has been explained by formation and propagation of wing cracks that depends on the existence of a sliding crack or shear crack. However, no evidence for that has been detected. Before the final rupture via axial splitting, if there were sliding microcracks, the apparent shear strain would be very large and dominant over other apparent strain components at the crack faces. However, as mentioned above, the apparent xxε has the largest magnitude in those situations. The line along which the much-larger-than-average apparent strain xxε is distributed seems to be consistent with the shear plane predicted by Mohr-Coulomb strength criterion. But it cannot result from shear but rather from the coalescence of the relatively scattered and local splitting. With in mind that Bazant et al has suggested that under uniaxial compression splitting crack band may form which would subsequently result in the formation of column bundles [15], we have enlarged the displacement by fifty times so as to visualize the deformed profile of the area of interest. The results are demonstrated in Fig.5. Dash blue lines are plotted only guided eyes. If the material points distributed along a line with same x can be considered a column, then one may find that the area of interest deforms by compressing the columns and some of them buckled locally. By scrutinizing the results in Fig. 5, one may also notice that the buckled regions are correlated with
RkJQdWJsaXNoZXIy MjM0NDE=