TDR量測夯實土壤乾密度與含水量之可行性評估

Abstract

由於輾壓土壤乾密度與含水量控制其強度與水密性，對以輾壓土壤為主之大壩與土堤等工程影響甚鉅，傳統以烘乾法與砂錐法獲得土壤乾密度與含水量之方式耗時繁瑣，而核子式儀器檢測迅速但具輻射污染疑慮。非核子式的快速檢測方法為碾壓土壤檢測的發展趨勢，ASTM規範近年推出以介電度為基礎的時域反射法（Time Domain Reflectometry, TDR）(ASTM D6780-05)，此規範之現地評估與可能潛在問題已有文獻探討(Lin et al.(2012))，後續的改良朝向結合導電度的量測法及最新提出的正規化反射係數法(ASTM D6780-12)。然而，這些方法之可行性與潛在問題尚未有文獻探討。因此，本研究主要進行室內試驗，並針對不同量測方法，配合貝氏分析探討其量測參數之相依性與其方法之不確定性，並了解其量測之準確度。 研究成果顯示導電度法與正規化反射係數法在含水量與乾密度的量測具有一定程度的相依性，但導電度法因其與土壤含水量及乾密度之數學關係與介電度相似，當兩者斜率相近時將造成不合理結果；正規化反射係數法之反算數學結構較穩定，能提供較具唯一性的估計且不確定性小於導電度法，亦可避免非常不合理估計。若經個別土壤的率定，正規化反射係數法大都能維持密度15%以內及含水量3%以內，準確度尚無法完全達到工程應用的需求，此外，若考慮土壤種類的變異、水質導電度的變異及土壤經乾溼循環影響，將出現更無法接受之誤差。穩定準確的含水量與乾密度非核子式量測技術仍須進一步的研究。

The moisture content and dry density are crucial parameters for quality control of compacted soils. Conventional measurement methods are time consuming while the nuclear method although efficient is potentially hazardous. The trend of compaction quality control is towards the development of fast non-nuclear method. The TDR method designated ASTM D6780(2005) has recently been introduced primarily based on dielectric constant. The experimental evaluations and potential problems have been investigated (Lin et al., 2012). Subsequent improvements of the TDR method have been made by including electrical conductivity or reflection magnitude (ASTM D6780-12) also extracted from the TDR measurement. However, the feasibility and evaluation of these improvements has not been thoroughly investigated. The objective of this study is to conduct laboratory tests and utilize the Bayesian analysis to investigate the uncertainty associated with them. Bayesian analysis shows that the estimated water content and density are dependent to some degree by both methods. But the electrical conductivity has a relationship with water content and density similar to the dielectric constant. Unreasonable results may be resulted in when the two relationship slopes are close to each other. The normalized reflection coefficient method has a more stable mathematical inversion structure. It gives more reasonable results with better uniqueness and less uncertainty. When a soil is calibrated specifically, the error of density estimation is mostly within 15% and the water content within 3%. The corresponding errors are still greater than what are normally expected in engineering practice. Furthermore, when the effects of soil type, water salinity, and wetting and drying cycle are taken into account, larger unacceptable errors may be resulted in. Non-nuclear methods for more accurate and stable measurements of soil water content and density require further study.