published_or_final_version / abstract / toc / Civil Engineering / Doctoral / Doctor of Philosophy

Traditionally, void ratio, e has been used as a state variable for predicting the liquefaction behaviour of soils under the Critical State (Steady State) framework. Recent publications show that void ratio, e may not be a good parameter for characterizing sand with fines as the steady state, SS data points move downward in e-log(p) space up to certain fines content termed as threshold fines content, TFC. Thus, it was difficult to apply SS concept on sand with fines as a small variation of fines content may lead to different SS line. Many researchers proposed to used equivalent granular void ratio, e* as an alternative state variable (i.e. in lieu of void ratio, e) in attempt to obtain a narrow trend line for SS data points irrespective of fc provided fc  TFC. The e* is obtained from e. For the conversion from e to e*, one need a parameter b which presents the active fraction of fines in overall force structure of sand. However, predicting the b is problematic. Most, if not all, of the b reported were determined by case-specific back-analysis, that is, the b-value was selected so that the test results for a given sand-fines type could be correlated with the equivalent granular void ratio, e* irrespective of fines content. This thesis examines the factors that affecting the b value by examining published work on binary packing. This leads to a simple semi-empirical equation for predicting the value of b based onparticle size ratio,  and fines content, fc. Published data and experimental results on Sydney sand appears to be in support of the proposed equation. The single relation of SS data points in e*-log(p) space for sand with fines is referred as Equivalent Granular Steady State Line, EG-SSL. The EG-SSL is then used to define the equivalent granular state parameter,*. A good correlation observed between * and q-p, q- q responses in undrained shearing. The e* and * are also used to modified a state dependent constitutive model. Seven model input parameters are needed in addition four to critical state input parameters. These parameters are obtained from drained test. The model is used to predict q-pand q- q responses for flow, non-flow and limited flow behaviour for 0% to 30% fines contents. The model predictions are in good agreement with experimental results. The effect of fines types (in terms of plasticity and angularity) on the prediction equation of b are also examined with four different types of fines. A negligible effect of fines type on the prediction equation of b is observed. The link between monotonic and cyclic loading behaviour for sand with fines are also examined with emphasis on cyclic instability and strain hardening behaviour after quasi steady state, QSS for a range of fines contents (provided that fc < TFC). It is found that a single set of rules could be used to correlate monotonic and cyclic behaviour for a range of fines contents at same *.