چكيده به لاتين
For years, the effect of non-plastic silt on the liquefaction behavior of silty sands has been a challenging and controversial topic among researchers. The extensive damage and significant losses caused by liquefaction in soils containing non-plastic fines in recent years have highlighted the necessity for a better understanding of the liquefaction behavior of silty sands and the need to clarify existing ambiguities. On the other hand, the energy method, as an alternative to the stress method for evaluating liquefaction, has advantages and merits that can be utilized to assess the liquefaction behavior of silty sands. The energy method is based on the premise that there is a strong relationship between the dissipated energy and the generated excess pore water pressure. In the energy method, liquefaction resistance is expressed as energy capacity (Wliq), which is the amount of dissipated energy until liquefaction initiation (zero effective stress or reaching a defined strain) and is independent of the stress path.
In this study, to investigate the liquefaction behavior of sand-silt mixtures, a comprehensive laboratory testing program was conducted on various mixtures of Firoozkooh sand and silt with a wide range of fines content, relative density, and effective stress. This program included performing basic tests to determine the physical characteristics of the Firoozkooh mixtures, 39 undrained cyclic triaxial tests under strain-controlled conditions, and 38 undrained monotonic triaxial tests. Additionally, several test results from previous studies were collected and reanalyzed.
Accordingly, a predictive model for the excess pore water pressure ratio (ru) was developed based on dissipated energy. This model has a simple functional form and is applicable to various sands and sand–silt mixtures with nonplastic fines contents, and it can be easily implemented in site response analysis for energy-based liquefaction potential evaluation. Convincingly, the accuracy of the proposed model was verifed using the results of a series of centrifuge tests, reported by others, and the recorded data of wildlife downhole array site during the Superstition Hills 1987 earthquake. A notable feature of this model is that the factor of safety (FS=W⁄Wliq ) can be determined based on an acceptable design ru value.
The results showed that relative density (Dr) could be used as a proper parameter to define soil density state for predicting W_liq of clean sands and sand‒silt mixtures with fines content greater than the threshold fines content (FCth). However, due to the complex role of fines in the microstructure of sand-silt mixtures, a general relationship between energy capacity and fines content was not observed for mixtures with fines content less than FCth. Therefore, the concept of equivalent granular void ratio (e*) was used to capture the coarse‒fine interactions in such mixtures. It was also found that the fines contribution factor (b), which is the fraction of fines participating in load transfer, is dependent on Dr, as well as particle size disparity ratio (χ) and FC, neglected in previous studies. Consequently, a new model was proposed for the prediction of the fines contribution factor (b), showing that it can effectively provide a unique relationship between e* and Wliq for all mixtures of specified sand and silt where FC ≤ FCth.
Monotonic test results indicated that the equivalent intergranular void ratio (e*) can be used to evaluate the monotonic stress-strain behavior of various mixtures of a specific sand and silt with different fines contents, not only at the critical state but also at other states along the stress path. The average b value obtained from monotonic tests for Firoozkooh mixtures matched the average b value derived from cyclic test results. Additionally, at a constant relative density, the collapse potential (CP) increases with an increase in fines content. In other words, with increasing of fines content, the soil's tendency towards contractive or softening behavior increases, making it more susceptible to collapse.
Finally, based on statistical analysis of several data sets, an approximate lower boundary of 25% and an upper boundary of 35% for the threshold fines content were observed. For simplicity, the threshold fines content can be considered 30%, or with consideration for dispersing fines content (FCdis), it can be set at 35%.
