Abstract :
[en] Cyclic loading on civil structures can lead to a reduction of strength of the used materials. A literature study showed that, in contrast to steel structures and material engineering, there are no design codes or standards for fatigue of foundations and the surrounding ground masses in terms of shear strength reduction. Scientific efforts to study the fatigue behaviour of geomaterials are mainly focused on strain accumulation, while the reduction of shear strength of geomaterials has not been fully investigated. It has to be mentioned that a number of laboratory investigation have been done and some models have been already proposed for strain accumulation and pore pressure increase which can lead to liquefaction.
Laboratory triaxial tests have been performed in order to evaluate the fatigue of soils and rocks by comparing the shear strength parameters obtained in cyclic triaxial tests with the static one. Correlations of fatigue with both, the number of cycles and cyclic stress ratio have been given. In order to apply cyclic movements in a triaxial apparatus, a machine setup and configuration was made. A special program was written in LabVIEW to control the applied stresses and the speed of loading, which allowed simulating the natural loading frequencies. Matlab scripts were also written to reduce the time required for the data processing.
Both cohesive and cohesionless geomaterials were tested: artificial gypsum and mortar as cohesive geomaterials, and sedimentary limestone, and different sands, as cohesionless and low-cohesive natural materials. The artificial gypsum, mortar and natural limestone exhibit mostly brittle behaviour, where the crumbled limestone and other sand typical ductile one. All the sands as well as the crumbled limestone were slightly densified before testing therefore; they can be treated as dense sands. The UCS for the crumbled limestone is 0.17 MPa and standard error of estimate σest = 0.021 MPa, where for mortar UCS = 9.11 MPa with σest = 0.18 MPa and for gypsum UCS = 6.02 MPa with standard deviation = 0.53.
All triaxial tests were conducted on dry samples in the natural state, without presence of water (no pore pressure). The range of the confining pressure was between 0 MPa and 0.5 MPa. The cyclic tests carried out were typical multiple loading tests with constant displacement ratio up to a certain stress level. The frequency was kept low to allow for precise application of cyclic load and accurate readings. What is more, the frequency of the cyclic loading corresponds to the natural loading of waves and winds. The number of applied cycles was from few cycles up to few hundred thousand (max number of applied cycles was 370 000).
Due to the complex behaviour of materials and high scatter of the results, many tests were required. Two different strategies were used to investigate fatigue of geomaterials:
1) the remaining shear strength curve; after a given number of cycles, a final single load test was done until failure in order to measure the remaining shear strength of the sample.
2) the typical S-N curve (Wöhler curves); there is simply a constant loading until failure.
The remaining shear strength (or strength reduction) curve has been compared with the standard S-N curve, and is found to be very similar because the cyclic stress ratio has little influence. The cyclic loading on geomaterials, being an assemblage of different sizes and shapes of grains with voids etc., showed different types of effects. Cohesionless materials show a shear strength increase during the cyclic loading, while cohesive ones show a shear strength decrease. For the cohesive materials the assumption was made that the friction angle remains constant; so, the fatigue of geomaterials can be seen as a reduction of the cohesion. In this way, the fatigue of a cohesive geomaterial can be described by a remaining cohesion.
The imperfections in the artificial gypsum have a significant impact on the results of the (especially cyclic) strength tests. Therefore another man made materials was used – a mixture of sand and cement (mortar). As the first static test results were very promising, mortar was used in further tests. The cyclic tests, however, presented similar, high scatter of results as for artificial gypsum. An unexpected observation for both materials was a lack of dependency of the remaining shear strength on the cyclic stress ratio.
The strain-stress relationship in cyclic loading shows that the fatigue life of the geomaterials can be divided into three stages, just as for creep. The last phase with a fast increase in plastic strains could be an indicator of an incoming failure. The accumulation of strains and increase of internal energy could be good indicators too, but no strong correlation, has been found.
Similar to the shear strength, the stiffness changes during cyclic loading; for cohesive materials the stiffness increase, while for cohesionless it decreases. This could help to predict the remaining shear strength of a geomaterial by using a non-destructive method.
