Laboratory tests

Ill Bulk density

The determination of bulk density is a straightforward operation, particularly when the "undisturbed" samples are brought into the laboratory in filled sample tubes. It is then only necessary to weigh the tubes and subtract their weight when empty. However, in order to obtain a representative value, it is often necessary to weigh all the sample tubes. Because of the natural layering of the soil, the bulk density can vary considerably over short depth intervals and the results of single tests on small specimens can be misleading.

Ill Grain size distribution

Determination of grain size distribution should be performed with discretion.

Because of natural layering, the grain size distribution can vary considerably also over small depth intervals. The soil should therefore first be inspected with regard to stratification and possible layers of different nature. These layers should then, if possible, be separated and the grain size distributions determined separately.

Grain size distributions determined on a mixture of the soil layers occurring, such as disturbed samples taken with a coarser sampling method, can be very mislead­

ing.

Ill Natural water content

The natural water content is significantly affected by the grain size distribution of the soil. It is therefore necessary to specify whether the water content refers to a specific sub-layer of a certain nature or if it refers to an average of the bulk soil volume. Average values may be required to calculate the degree of saturation together with the bulk density. Values specific for a certain sub-layer may be required to estimate the liquidity or consistency indices of the particular layer.

Ill Atterberg limits

The liquid limit of the soil is preferably determined by the fall cone method. In silt, there is a risk of post sinking of the cone after a momentary stop. This should be observed and the penetration of the cone is normally read off 5 seconds after the release of the cone. The same problems with layering of the soil as for grain size distribution and natural water content apply for both liquid limit and plastic limit.

Ill Classification

In Sweden, a classification system based on grain size distribution is used, (Karlsson and Hansbo 1984 ). Internationally, the Unified Soil Classification System, (ASTM 1992), or some similar system based on consistency limits is often

Investigations and load tets in silty soils 153

used to separate clays and silts. Thorough studies at SGI have shown that the A­

line used in the USCS system differs from what is found in Swedish soils, (Karlsson and Hansbo 1984). This has also been demonstrated in the current project. On the other hand, the composition of the clay minerals in Swedish fine­

grained soils is fairly homogeneous with mainly illite and other low to medium active clay minerals. In soils with more active clay minerals, a considerably lower content of clay size particles than in the Swedish rules for classification may change the characteristic behaviour of the soil to that of a clay. Classification of the soil should therefore be made according to rules which have been shown to apply to the local soil conditions.

Because of the possible layering of the soil, the classification should preferably be made by visual inspection of undisturbed samples and, if possible, classification tests on the separate sub-layers.

Ill Capillarity

Capillarity is an important parameter in silt, where negative pore pressures due to the capillary suction can create a considerable so called "false cohesion". To a great extent, capillarity also governs he risk of frost heave in the soil. Capillarity or, more elaborately, the water retention curve, can be determined in a number of ways. The most common capillary meters used in Swedish laboratories are the tube capillary meter and the Beskow suction capillary meter. The tube capillary meter is an open tube filled with the soil material, usually gravel, and with its lower end inserted in water. The capillary rise with time is then studied. The method is primarily used to test the suitability of the material to cut off the capillary rise of water. The suction capillary meter is used to study what capillary suction the soil can withstand without air breaking through. The method can only be applied for suctions up to 25 kPa and is primarily used to evaluate the frost heave susceptibility of the soil.

In agricultural soil laboratories, the pF apparatus has for a long time been used to determine the pF curve, showing the relation between suction and water content in the soil. This apparatus is primarily designed to measure suctions in the very high range exerted by plant roots in the soil. In connection with investigations in non-saturated soils for geotechnical purposes, a number of devices and measuring techniques have been developed (e.g. Znidarcic et al. 1991, Tremblay 1996 and Oberg 1997). In many of these methods, a soil specimen is placed on a high-air­

entry disc inside a pressure cell. The soil sample is then subjected to an increasing air pressure, whereby the water in the sample, but not the air, can escape through the disc. By measuring the expelled water, the water content in the specimen can

SGI Report No 54

154

then be plotted versus the applied air pressure, the latter being assumed to correspond directly to a matrix suction in the pore water in the soil.

In the current investigation, a very simple such device was designed at SGI. The original intention was only to design an apparatus corresponding to the suction capillary meter but with increasing air pressure on top of the soil specimen in order to increase the measuring range to about 100 kPa. It later turned out that it was very simple also to measure the expelled water instead of just visually observing the breakthrough of air, and this facility was subsequently incorporated, Fig. 5.1.

3.5 3

~ 2.5 .l!!

~ 2

j 1.5 ai c.

><

w 0.5

0 IHHHHIHIHl...,..,.t!!!.:..:_ _- - - + - - - + - - - 1

0 50 100 150 200

Applied pressure, kPa

Fig. 5.1 a) Capillary meter designed at SGI.

b) Results of a test in the apparatus.

Investigations and load tets in silty soils 155

The results obtained on remoulded and re-compacted samples in this apparatus may differ somewhat from the results obtained in more elaborate apparatuses, in which "undisturbed" soil specimens may be tested. The results from both types of tests, however, are significant for the particular specimens only. They may thus be regarded as index tests giving indications of possible suctions in the ground. In the profile in situ, the soil normally consists of several layers with different properties and the resulting pore water suction may depend on the interaction between these layers. In order to estimate these conditions correctly, the pore pressures should be measured in the field. Tests on samples in the current investigations yielded capillarities which varied between 40 and 60 kPa ( 4 to 6 m water head) for samples from Vatthammar and 25 to 70 kPa for samples from Vagverket. For samples from Kil, capillarities of more than 100 kPa were measured.

II Oedometer tests

Oedometer tests are normally carried out only on specimen from levels where the in situ tests and the classification have shown that the clay content is high. On such levels, the in-situ tests may have been performed under undrained or only partly drained conditions and samples of high quality can be taken and tested in the oedometer to yield more reliable and relevant results, particularly when CRS tests are used. These levels can be identified clearly from the results of CPT tests and dilatometer tests.

It is possible to perform oedometer tests also on other levels with coarser soil, but the extent of such tests required to yield a satisfactory result and the related cost cannot normally be justified. In such soils, in situ testing is a more rational way of determining the compressibility of the soil.

II Triaxial tests

Triaxial tests are performed to study the behaviour of the soil in more detail. The tests enable evaluation of a number of parameters required in more advanced numerical calculation methods and also studies of the soil behaviour under more complex loading conditions, such as dynamic loading. In the tests, it is also possible to study the soil behaviour under truly undrained or drained conditions and, in more elaborate tests, at various degrees of saturation and related matrix suction in the soil.

The test is not used very widely in Sweden for routine calculations concerning strength and settlement problems in silt. Only in clayey silt and silty clay, where these properties cannot readily be determined by in situ tests, are ordinary triaxial

SGI Report No 54

156

tests performed to determine the drained shear strength parameters. In these soils, undrained tests may also be performed to study the truly undrained behaviour and possible effects of anisotropy. Triaxial tests may also be employed to study the behaviour of proposed future fills and embankments of silty soils.

In the current project, a number of triaxial tests have been performed, some of them in conjunction with other projects in which there has been a direct need for such tests. These tests showed, for example, that the truly undrained shear strength measured in triaxial tests could be very different from the corresponding values measured by vane tests in the field. The triaxial tests also showed that the frequently used technique of running the tests with a backpressure of 200 kPa or more is questionable in silts which exhibit a tendency to dilatant behaviour in undrained shear. The resulting decrease in pore pressure then often becomes so large that in natural conditions in the field it would correspond to pressures lower than both vacuum and the capillarity of the soil. There is therefore a risk that such a test in the laboratory will yield strength values that cannot be mobilised in the field. Tests consolidated for and started at the actual in situ stresses appear to yield more realistic strength values.

5.2 FIELD TESTS