Download files
Abstract
Liquefaction, a catastrophic failure phenomenon of saturated or nearly saturated granular materials, can be divided into either static or cyclic liquefaction depending on whether the loading condition is monotonic or cyclic. Static liquefaction, also referred to as static instability, is in fact undrained deviatoric strain softening behaviour where instability in the context of continuum mechanics, i.e.dσijdε<0, is manifested (where σij and εij are stress and strain tensors respectively and d represent infinitesimal increments). On the other hand, cyclic liquefaction can occur in the form of either cyclic instability or cyclic mobility. Cyclic instability is a deviatoric strain softening behaviour triggered by a series of cyclic stress pulses and followed the same line of argument as for static instability. Cyclic mobility can be categorised as a state of occurrence of 5% double amplitude (DA) axial strain and the effective stress path travel transiently to near zero in a load cycle, but the imposed peak (and trough) deviator stress can still be developed even a few loading cycles after the occurrence of cyclic mobility. Experimental results on Sydney sand with fines (particles passing 0.075 mm sieve) and coal ash, and different published databases for sands with fines as well as clean sands were used to investigate different aspects of above-mentioned forms of liquefaction giving particular emphasis on cyclic liquefaction.
Prior research investigations in linking static and cyclic instability mostly concentrated on clean sand and very limited studies in this regard could be found for sand with fines. In all of those investigations, the linkage between static and cyclic instability was examined by forming replicate test-pairs where a test-pair formed between undrained monotonic and cyclic loading test of one-way or two-way. Also, different characteristics lines have been proposed to define the triggering of cyclic instability. This thesis examines the linkage of static and cyclic instability by forming equivalent as well as replicate test-pairs. The concept of equivalent granular state parameter was proposed to capture the influence of fines content, ∫c. An equivalent test-pair was formed by different combinations of fines contents, ∫c (less than threshold fines contents, ∫thre), initial effective confining stress, p'0 and void ratio, e0, but having same equivalent granular state parameter, ψ∗(0) where the symbol 0 next to parameters is used to indicate prior to shearing. Experimental results showed that, cyclic instability, under one-way and two-way cyclic loading, triggered shortly after the cyclic ESP crossed the ηIS-zone as determined from a corresponding monotonic loading test, where, ηIS is the effective stress ratio (q/p׳) at peak of undrained monotonic ESP. Further, this finding is equally applicable for both equivalent and replicate test-pairs. The next step is to predict the triggering of cyclic instability without having a corresponding (equivalent or replicate) monotonic test. This was achieved via the concept of instability curve which is a single relationship between ηIS and ψ*(0) for a range of fines content less than the threshold value. Two instability curves (IC), namely ICcom and ICext as obtained in ηIS-ψ∗(0) space for Sydeny sand with fines were used for predicting cyclic instability of same soil, where subscript com and ext indicate the ICs obtained from data points of undrained monotonic compression and extension tests respectively. Test results showed that ICcom and ICext could be successfully used to predict triggering of cyclic instability in the corresponding stress space capturing the influence of ∫c (<∫thre), p'0 and initial static shear stress, q0. The lack of a unified framework in predicting different forms of cyclic liquefaction have been revealed in literature. Thus, published databases for three clean sands and four sands with fines were systematically analysed in context of critical state soil mechanics, CSSM to examine whether it is giving an indicative response. Analysed databases suggested that the occurred form of cyclic liquefaction (cyclic instability and cyclic mobility) could be predicted depending on the corresponding value of the state parameter, ψ(0), for clean sand; or ψ∗(0) for sand with fines. In the later case, the influence of ∫c (<∫thre) does not have to be separately accounted for as its already captured in the definition of ψ*(0). Then, these findings were further verified by conducting experimental investigation for Sydney sand with fines. During the experimental investigation, a new form of cyclic liquefaction was identified along with two others as mentioned above. This newly identified form of cyclic liquefaction was referred to as transition behaviour which manifested some (but not all) of the features of both cyclic instability and cyclic mobility. This study demonstrates that all forms of cyclic liquefaction can be predicted throughψ∗(0) capturing the effect of ∫c (<∫thre), p'0 and imposed q0. Therefore, a unified framework for predicting different mode of cyclic liquefaction under CSSM has been achieved.
In this thesis, cyclic liquefaction behaviour was also investigated for a coal ash which is predominantly fines. Preliminary experimental results on coal ash also showed three forms of cyclic liquefaction as observed for Sydney sand with fines.