Numerical data from different sources, such as mechanical and classification laboratory tests, in situ investigations and continuous monitoring, are often used in many fields of applied geology. Despite of this, a careful control of the quality of this kind of data is often missing; as a consequence, they cannot be confidentially used to implement models and to solve problems in applied geology. The uncertainty on data can significantly influence model results in many situations; the study of accuracy, repeatability and reproducibility of test methods is therefore of basic interest in applied geology, and geothecnics. The direct shear test (DST) is very popular for the laboratory testing of soils owing to its simplicity. When performing the test, the main problem is to obtain similar results when repeating the same test under the same conditions. DST tests were conducted on fine-grained soils taken from a quarry located along the Paglia alluvial plane (Orvieto, TR). The soils were compacted by using the CBR test equipment, which allows the specimen’s water content and the dry unit weight to stay constant. Four direct shear tests were performed with the same sequence of normal stress (100, 150, 200, 250 kPa), with the same method on identical test items (ASTM D 3080-98), in the same laboratory by the same operator and using the same equipment. The specimens tested showed an initial low degree of saturation (Sr = 70%). During consolidation the specimens absorbed water and reached a high saturation degree (up to 95%): the shear tests were executed in drained conditions. Standard deviation on mechanical parameters was calculated (c’= 19.2 ± 3.3 kPa and φ’= 27° ± 1°) and an analysis of the all possible combinations of c’ e φ’ was made obtaining 256 values. A frequency analysis on c’and φ’ showed that values accepted by standard deviation analysis belong to the class with higher frequency. The repeatability of grain size analysis of soils was also tested, highlighting the effects on soil classification and, as a consequence, on its proper use for engineering purposes. Six specimens of selected soils were first analysed by sieve (CNR UNI 2234) and hydrometer tests (ASTM D 2217, D 422-63) separately; then ten specimens were analysed by means of full grain size analysis. Results have shown that, as for the sieve analysis, the worst controlled stage is the sampling, which can lead to select poorly representative samples. As for the hydrometer analysis, the main problem seems to be that the operator can choose to test the bulk sample or the fraction finer than 0.075 mm, on the basis of his personal analysis. This choice introduces a bias error the magnitude of which can vary by changing operator. It seems that the laboratory standards would better establish to perform the hydrometer test only on the fraction finer than 0.075 mm. As for the full grain size analysis, the tricky stage is the washing, which leads to the separation of the fraction finer than 0.075 mm. This procedure is particularly difficult for samples with high percentages of fine sand and for high amounts of material. The repeatability of wax procedure to determine the shrinkage limit (ASTM D 4943-95) of a clayey soil was finally analysed. In the geotechnical practice the wax method replaced the mercury procedure because, besides being safer, is less time consuming and influenced by laboratory conditions and manual skills of the operator. Nevertheless some stages of the standardized wax procedure are not completely set, and the repeatability of the method has not been tested. Results have shown that the worst controlled test stages are the coating, the quality of which depends on the temperature of the molten wax, and the tying of the dry pat with the thread. However, the statistical analyses have shown a good repeatability, which can be improved by a better control of the testing time and the methodical implementation of each stage of the procedure.
File in questo prodotto:
Non ci sono file associati a questo prodotto.