本篇論文利用邏輯迴歸分析結果及訊息理論(information theory)，推導了一新的SPT-N液化評估準則，及液化機率與損害評估模式。文中首先利用Seed形式、T-Y形式及NJRA形式之邏輯迴歸分析結果，藉由訊息理論，評估不同邏輯迴歸模式之可信度及相對權重，建立以三形式液化機率等值線（對應於最大綜合預測成功率）之權重平均為抗液化強度曲線(CRR)之SPT-N液化評估準則，簡稱為MaxSRP法。該法綜合預測成功率達87%，整體表現較現行國內常用之簡化法要好，且兼具簡潔與較大適用範圍之特點（可適用於(N1)60, cs>30之範圍），並搭配抗液化安全係數FS對應之液化機率P(下標 L)、土壤參數折減係數D(下標 E)，以及液化潛能指數IL與液化損害程度之推估方法及評量方式，整個流程前後連貫且一致，並反映了九二一地震之本土液化震害經驗，對液化損害評估及抗液化工程設計提供了一具機率意涵之量化工具。最後，本文也繪製了一以機率為基準之地盤改良設計圖表，可作為抗液化工程設計依循之參考。 In this paper, a new SPT-based simplified method, MaxSRP, for assessing the liquefaction potential, probability and associated damages is established. The assessment model is developed using the method of information theory and the results of logistic regression. The performance of the model is better than that of other simplified methods such as Seed method because of the higher success rate of prediction (87%), wider range of application, consistency of evaluation process for damages evaluation, and inclusion of the experiences from the 921 Chi-Chi earthquake. In this model, relationships between the calculated factors of safety (FS) and probability of liquefaction (P(subscript L)) are calibrated using the logistic mapping approach and Bayesian mapping approach. Damage indices, including reduction factor of soil (D(subscript E)), liquefaction potential index (I(subscript L)), are also revised and proposed for the new simplified method. A design chart for soil improvement based on liquefaction probability is also prepared. The newly developed model provides a method to assess liquefaction probability by factors of safety easily, and forms a basis for risk-based evaluation of liquefaction potential and damages quantitatively and consistently. Applying these concepts and results can be beneficial for the engineering community.
關聯:
中國土木水利工程學刊=Journal of the Chinese Institute of Civil & Hydraulic Engineering 20(3)，頁301-314
