| 摘要: | 摘要
過去許多學者提出SPT - CPT 相關性方法，使工程師在利用不同的關係式求取現場參數更為方便。然而，這些關係式僅根據數據進回歸而得，並無力學上的依據。本文提出一種新方法，分別從現有的 SPT 與 CPT 力學公式中，選定一對相同的輸出的公式作相等，並推導出理論關係式。此理論關係式為建立SPT - CPT 相關性的前期知識，然後利用現地調查之SPT - CPT 數據對進行校準，最終產生新的關係式。
本文希望建立正規化SPT-N（(N1)60,cs）及正規化CPT-qc（qt1N,cs）關係式可用於土壤液化評估上，因此對砂性土進行探討。首先，根據三個砂性土壤相關參數(反覆液化阻抗比、有效摩擦角及相對密度)公式，分別利用數值及解析方式得到基於力學性質的(N1)60,cs及正規化 qt1N,cs 之關係。繼之，利用大量相鄰很近的砂性土 SPT - CPT 數據對作為觀測資料，以最大似然估計（Maximum likelihood estimation, MLE）更新關係式，然後根據貝氏訊息準則（Bayesian information criterion, BIC）之加權平均組合三個關係式，最後提出的新(N1)60,cs – qt1N,cs 關係式，與現有關係式進行比較。;Abstract
Many scholars proposed the correlations between the SPT-N and the CPT-qc, so that the engineers could obtain the in-situ soil parameters easily through different correlations. However, these correlations were based on the data driving without considering the mechanical-based model. This paper proposes a new approach that starts with a theoretical baseline derived from a selected pair of existing transformation models that use SPT and CPT, respectively, as the input for the same output. This theoretical baseline, considered a prior knowledge of the intended SPT-CPT correlation, is then calibrated with the collected data to yield the final correlation. We are focused on developing the correlation of the corrected clean sand equivalence of SPT-N ((N1)60,cs) value and CPT-qc (qt1N,cs) for sandy soil.
Firstly, this paper obtained three theoretical equivalence correlations of (N1)60,cs and qt1N,cs based on the equations of cyclic resistance ratio, effective friction angle, and relative density, respectively. To illustrate the proposed approach, we created a high-quality database from a large number of side-by-side SPT - CPT pairs. Then, the updated correlations could be obtained via the maximum likelihood estimation (MLE) that adjusted the coefficients of the correlations by observing the in-situ SPT - CPT pairs. Finally, the three (N1)60,cs – qt1N,cs correlation models
are combined through a weighted average process based on the Bayesian information criterion (BIC). The final (N1)60,cs – qt1N,cs correlation is then proposed, and a comparison with an existing model is made. |
