台灣中西部粉土質砂土液化行為分析

Abstract

台灣中西部地表有厚層之粉土質細砂，在1999年集集地震時發生廣泛之液化行為。現有簡易法(simplified procedure)液化潛能評估是以乾淨砂做為基準，對於含有細粒料（通過200號篩之材料）之砂土，需依據細料含量(fines content, FC)做修正。使用簡易法程序為此區域粉土質細砂做液化潛能評估常因FC修正與現地試驗方法之選擇而產生不同之結論。唯FC對室內試驗所得砂土不排水強度的影響至今尚未有明確的定論，砂土中所含細料對現地試驗結果之影響所知更為有限。FC的修正在簡易法中之必要性與正確性並無嚴謹之驗證。麥寮砂與員林砂都是台灣中西部具有代表性之砂土。作者使用交通大學過去十數年對麥寮砂所做一系列圓錐貫入(CPT)標定與動靜態三軸試驗結果，做有系統之分析比對。就麥寮砂而言，其剪力波速（Vs）與反覆阻抗比（CRR）會隨FC之增加而降低，FC對Vs-CRR關係並無明顯之影響。在簡易法架構下使用Vs做液化潛能評估時，其FC修正之必要性遠不如現有方法所述明顯。在FC為0與15%時，CPT可視為排水行為，圓錐貫入阻抗（qc）與CRR間之關係無明顯差異。當FC達到30%而CPT成為部分排水狀況時，qc-CRR關係才有明顯改變。此發現說明qc-CRR關係主要受CPT貫入時土層之排水性所影響，而非細料含量本身。唯以上結論是根據重模試體所做試驗而得，研究指出天然砂土受其組構之影響，與重模試體行為有明顯差異。為驗證先前發現是否適用於天然砂土，作者研發使用Laval sampler在原地溫度情況下於粉土細砂中取樣，然後在地表加以冰凍保存與室內試體準備及試驗程序之技術。Laval sample與地表冰凍之成本遠低於將土壤於地層內冰凍然後取樣程序。使用此一技術，作者在員林試驗站取得一系列Laval sample，並在現場進行CPT及剪力波速量測。使用Laval sample進行室內動靜態三軸試驗，並使用bender element為三軸試體進行剪力波速量測。此系列試驗結果首先驗證Laval sample低擾動之特性，其Vs與CRR值明顯高於相同狀態下重模試體所得之結果。與麥寮砂相同，員林砂之Vs與CRR會隨細料之增加而降低。天然及重模員林砂土之Vs-CRR關係與麥寮砂類似，也不受細料含量影響。Laval sample所量得CRR與現地qc之關係其趨勢與麥寮砂重模試體所得結果類似。但天然員林砂內之CPT似乎需要在更高細料含量下才顯現部分排水之行為。此一現象可能受天然沖積砂土中存在許多乾淨砂所組成之排水薄層影響，使得CPT所激發之孔隙水壓極易消散。因此在天然沖積土中CPT即使在FC極高的情況下也可能是排水的行為，而qc-CRR關係之修正必須依靠現地砂土透水性量測之結果來做決定。可能使用之透水性量測方法可以包括CPT孔隙水壓消散試驗。

關鍵字：砂土、細料含量、反覆阻抗比、Laval sample、液化潛能、圓錐貫入試驗、剪力波速

Central Western Taiwan is covered by a thick deposit of alluvial silty fine sand. The Chi Chi earthquake of 1999 triggered extensive soil liquefaction in this area. The simplified procedure commonly used for liquefaction potential assessment starts by considering the granular material as clean sand. For sand with fines (particles passing #200 sieve), an adjustment for the fines content (FC) is required. The back analysis of liquefaction potential using the simplified procedure for the silty sand in this region often results in different conclusions because of the fines content adjustment and in situ test method used for the analysis. There has been no consensus as to how the fines affect the undrained shear strength of silty sands. The knowledge on how fines affect the in situ test results is even more limited. The need and accuracy of FC adjustment in the simplified procedure have not been rigorously verified. Mai Liao sand and Yuan Lin sand can both be considered as typical sand in Central Western Taiwan. The author compiled results from a series of cone penetration tests (CPT) in a calibration chamber and monotonic/cyclic triaxial tests on Mai Liao sand (MLS) and made systematic analyses. For MLS the shear wave velocity (Vs) and cyclic resistance ratio (CRR) decrease with FC. FC does not have a significant effect on the Vs-CRR correlation. Under the framework of simplified procedure there does not appear to be a need for FC adjustment. For FC of 0 and 15%, CPT can be viewed as drained test, and there was no obvious difference in the correlation between cone tip resistance （qc） and CRR as FC changed from 0 to 15%. As FC reached 30%, CPT becomes partially drained, and the effect on qc-CRR correlation becomes significant. This finding shows that it is the drainage conditions not FC itself that affect the qc-CRR correlation. The above conclusion was made entirely from tests on remolded specimens. Research has indicated that due to discrepancies in soil fabrics, there can be significant differences in the behavior between natural and remolded soil specimens. The author used a modified Laval sampler to retrieve undisturbed silty sand samples under ambient temperature and preserved samples by freezing above ground. Techniques of cutting frozen Laval samples and thawing soil specimens in triaxial were developed. The cost of taking Laval sample and freezing above ground is much lower than ground freezing and coring. With this new procedure, the author took a series of Laval samples, performed CPT and field shear wave velocity measurements at a test site in Yuan Lin. Monotonic/cyclic triaxial tests with shear wave velocity measurements using bender elements were conducted on specimens trimmed from the Laval samples. The quality of the Laval samples was verified first. The Vs and CRR values from Laval samples were significantly higher than those of remolded specimens with the same density and stress states. Similar to MLS, the Vs and CRR decrease with FC for Yuan Lin sand (YLS). FC had no effect on the Vs-CRR correlation for both natural and remolded YLS. However, the FC had to be much higher what was learned from MLS for CPT in natural YLS to be partially drained. For CPT in natural alluvial soil, the drainage conditions are strongly influenced by the possible existence of thin layers of clean sand. The effects of drainage during CPT due to these thin permeable layers may not have a direct relationship with the fines content of the silty sand mass. In order to account for the drainage effects, the CPT pore pressure dissipation test may be used as a basis for the qc-CRR correlation adjustments.

Keywords：sand, fines content, cyclic resistance ratio, Laval sample, liquefaction potential, cone penetration test, shear wave velocity