In situ tests II Dilatometer tests

Requirements and limitations

Dilatometer tests have been used in Sweden for a number of years, particularly for estimation of compression moduli in silty and sandy soils, and experience is generally good. A standard for the test was recommended by the Swedish Geotechnical Society in 1995.

The dilatometer can be pushed down to approximately the same depths as in CPT tests using the same drill rig. In some soils, the dilatometer has a tendency to become stuck during the time for the test because of increasing rod friction. This can be remedied by using a friction reducer or, in a simpler way, by lifting the dilatometer a few millimetres before it is pushed down to the next test level. The dilatometer can be forced down through stiffer layers by blows, but this should be avoided as far as possible because of the risk of disturbance of the soil, leading to less relevant results, and of damage to the equipment.

Inte1pretation

In the evaluation of the test results, the initial pore pressure and the in situ effective vertical stress are used. Both the soil classification and many of the evaluated parameters are thus significantly affected by the in situ pore pressures. Unless negative pore pressures in the upper part of the soil profile are measured and accounted for, the classification will yield an excessively coarse soil with exces­

sive over consolidation.

In sands and silts, the commonly used interpretation methods proposed by Marchetti (1980) yield compression moduli which can be applied directly in settlement calculations. The corresponding method proposed by Marchetti for estimation of a compression modulus in clay is not always used in Sweden. Before a compression modulus is evaluated in clay, a check should be made to ensure that the test results indicate a sufficient overconsolidation in the soil, so that the stresses also after application of the additional load remain within the preconsolidated range. If this is the case, the moduli evaluated according to Marchetti or the empirical relations between undrained shear strength and modulus of elasticity, which are commonly used in rough estimates of settlements in overconsolidated clays in Sweden, may be applied. The undrained shear strength is then

SGI Report No 54

166

calculated from the results of the dilatometer tests, normally using the formula c ₁₁=(p_{1 - CJhOJ}

I

10.3. The relation between undrained shear strength and modulus is then selected on the basis of the dilatometer test results in such a way that

E (or M₀) "" 250·c₁₁ for clay _(1D(corr) < 0. 35) E (or M₀⁾ ""500·c ₁₁ for silty clay (0.35 < ID(corr) < 0.6)

E (or M₀)"" lO00·c ₁₁ for clayey silt _(ID(corr);::: 0, 6)

The latter estimation of the modulus is less sensitive to a disturbance of the soil than the Marchetti method, (Larsson 1989).

In normally consolidated and only slightly overconsolidated clays and clayey silts, the compression modulus cannot be evaluated from the dilatometer test or any other undrained or only partly drained test. In such soils, undisturbed samples should be taken and tested by oedometer tests in the laboratory. In layered soils, the results of the dilatometer tests clearly identify the layers in which this procedure should be applied. The results of the oedometer tests in the fine-grained soil layers and the dilatometer tests in the coarser layers are then used together in the settlement calculation.

In difficult soil profiles with alternating relatively thin layers of coarser and more fine-grained soils, or with coarser objects embedded in the fine-grained soil, the pushing-in of probes or the dilatometer may result in a considerable remoulding of the soil at a number of levels in the soil profile. The evaluated moduli then become very scattered with high values in layers with "undisturbed" soil and very low values in the remoulded layers. In this case, the problem cannot normally be readily remedied by sampling and laboratory tests, because similar disturbance effects may be expected during ordinary sampling operations. However, it is normally possible to separate the unusually disturbed test levels and to improve the interpretation of the dilatometer tests results to yield useful estimations of the properties also in such soils. This procedure starts with an examination of the overconsolidation ratio versus depth is studied. The overconsolidation ratio in a soil profile should not vary randomly and a combination of an unusually low value of the material index and a significantly lower overconsolidation ratio than in surrounding layers may be considered as a firm indication of a remoulded soil during the test. If the general trend for the overconsolidation ratio, except for the

Investigations and load tets in silty soils 167

remoulded layers, indicates a certain overconsolidation, the modulus in the disturbed layers may be estimated from the empirical relations with the undrained shear strength described above. This generally yields an estimation on the safe side, i.e. somewhat too low moduli, because the estimated value of undrained shear strength is also affected by the disturbance, although normally to a limited degree, and the evaluated material index is too low.

As an example, the results from the test site at Mjardevi can be studied. This is a very difficult profile, in which both attempts to take undisturbed samples and to perform pressuremeter tests largely failed. The initially estimated moduli showed a wide scatter with values falling to almost zero in many layers, Fig. 5.4 a. The examination of the evaluated material indices and overconsolidation ratios showed that the values at 2.6-3.6, 5.2, 7.0-7.2, 8.2-8.6 and 9.4 metres depth should be considered to represent remoulded soil, Fig. 5 .4 b. The general applicability of the empirical relations between modulus and undrained shear strength could then be checked by applying them to all levels at which the soil had been classified as clay or silt, and comparing them to moduli estimated in the ordinary way from the tests in apparently intact soil. In this comparison, the reference values were taken as single values or averages from several boreholes. The comparison, Fig. 5.4 c, shows that the values are generally of the same size. The empirically estimated values in the obviously disturbed zones follow the general trend, but are somewhat on the low side.

This confirmation enabled a new evaluation of the tests in the borehole, shown in Fig. 5.4 d, which yielded an average modulus of about 10 MPa for the upper ten metres of the soil profile and a harmonic mean value of 5 MPa. The fact that the modulus was not overestimated by this procedure was further confirmed by the oedometer tests in the single level at 2.5 metres depth where relatively undisturbed samples could be taken.

SGI Report No 54

168

Modulus M, MPa

Probably remoulded soil Clay Silt Sand

0.01 0.1 10

Fig. S.4 Example of estimation of moduli from dilatometer tests in a difficult.

soil profile using results from the test field in Mjardevi.

a) Moduli evaluated by the ordinary procedure.

b) Separation of obviously remoulded levels.

investigations and load tets in silty soils 169

2

0 Modulus estimated from evaluated undrained Cl Iii

Ill shear strength and empirical relation

10

Fig. 5.4 cont. Example of estimation of moduli from dilatometer tests in a difficult soil profile using results from the test field in Mjardevi.

c) Check of the applicability of the empirical relations.

d) Estimated moduli.

170 SGI Report No 54

Ill Pressuremeter tests

In this project, only tests with equipment and procedures according to the Menard method have been performed. The test results obtained in this way are heavily dependent on the quality of the pre-drilled test holes. The pre-drilling was performed with a special screw auger with a hollow stem and hollow drill rods. In this way, air or a bentonite slmTy could be let in at the bottom of the screw during its withdrawal and thereby prevent suction and collapse of the hole. The bentonite slurry was used to stabilise the holes below the ground water level. This equipment and technique have previously been used in sands with good results. In the current investigation in silts, very good holes and test results were obtained only above the free ground water levels, particularly at Vatthammar where the negative pore pressures were high.

For tests below the free ground water level, the quality of the holes was generally poor in spite of repeated attempts to vary and improve the pre-drilling technique.

The factor least affected by the quality of the hole is normally the limit pressure, pu but in the heterogeneous soil in Mjardevi, it was often impossible to evaluate even this parameter because the disturbance of the hole was so great that the expansion capacity of the probe became insufficient.

The moduli evaluated according to Menard are very sensitive to the quality of the holes. For the plate load tests in the current project, the depths of influence were limited and it was therefore possible to obtain useful results also at Vagverket, where the free ground water level was relatively high. In general, however, it must be concluded that with the present techniques for making the test holes, good pressuremeter tests according to the Menard procedure can in silts only be expected above the free ground water level. The tests were also interpreted according to Briaud ( 1995) and this procedure also appears to be sensitive to the hole quality.

In contrast to most other in situ tests, the pressuremeter measures a complete stress strain curve up to failure. With new interpretation methods, test procedures and equipment, it is also possible to study, for example, the growth in elastic stiffness with increasing pressure level. This in tum opens the way for using new, much needed calculation methods in which the change in modulus with shear strain and pressure level can be taken into account. A general improvement in both testing equipment and installation technique in silt is therefore to be desired.

According to the recommendations by Briaud, a calculation of settlements should be based on averages or selected values from several test levels and boreholes.

Investigations and load tets in silty soils 171

Experience from the present project supports this recommendation.

The results of pressuremeter tests in soils with negative pore pressures and the evaluated parameters are largely dependent on the suction in the pore water and the related "false cohesion" prevailing during the tests. Ifthese results are to be applied in design, it has to be ascertained that negative pore pressures of the same size will prevail throughout the life-time of the design. Otherwise, the effects of the negative pore pressures during the tests have to be estimated and deducted.

Ill Field vane tests

In Sweden, the undrained shear strength is normally measured by field vane shear tests. In silts, this test can hardly be expected to be undrained. In the normal test procedure, the vane is inserted to the test level and after waiting 2 to 5 minutes, the vane is rotated at a rate expected to produce failure after about another 3 minutes.

Considering that almost full excess pore pressure dissipation in a CPT test normally takes less than 5 minutes, a normal field vane test in silt cannot be expected to be an undrained test.

Silt often exhibits a strain hardening behaviour, which entails that a distinct failure stress is often difficult to evaluate. In dense and medium dense silt, the shear strength becomes so large that smaller vanes than usual have to be used and still the measuring capacity of the ordinary recording instruments normally used for clays often becomes insufficient.

Results from the layers with more clayey silt and other investigations, (Borgesson 1981), show that the results from field vane tests may still be found to yield fairly good estimations of the undrained shear strength in some silty soils.

1111 Pore pressure measurements

Pore pressure measurements in silt should normally be performed in closed systems because of the possibility of rapid fluctuations. The pore pressure conditions in a silt profile are often not hydrostatic and the possibility of obtaining a general picture of the pore pressure distribution can be greatly aided by CPT-tests with frequent stops allowing full dissipation of generated excess pore pressures.

Measurement of negative pore pressures can also be performed with closed piezometers of the BAT type, but measurements over long periods of time often require regular re-saturation of the filter tips. A special technique for this has been developed at Chalmers University of Technology (Forsgren and Oberg 1997).

SGI Report No 54

172

Measurement with BAT piezometers in silt also requires special precautions in terms of sealing all connections to prevent silt particles from seeping in and to keep the readout assembly and leads lowered down to the tip clean. Otherwise, a sufficient amount of silt will soon accumulate in the tubes down to the filter tip to prevent the hypodermic needle in the readout equipment from penetrating the sealing membrane and coming into contact with the pressure chamber in the tip.

The pore pressure conditions should preferably be studied continuously for a fairly long period of time together with the local precipitation. Measurements should normally also be performed at several depths in the profile. Studies of the pore pressure conditions during snow melting and thawing should also be made and, if possible, coupled to information on normal and extreme snow and frost conditions at the site during such periods. The latter information is particularly important when negative pore pressures in the soil are utilised in design calculations or estimation of stability for natural slopes.

11111 Seismic cone tests

The seismic cone test, (Campanella et al 1986), has been shown to be a rational way of determining the initial shear modulus at small strain, G₀. The method has previously been thoroughly tested in Swedish clays and found to yield good and repeatable results, (Larsson and Mulabdic' 1991).

In the related study of the general formulation for G ₀, also other tests were included. Among these were results from low-plastic silty clays/clayey silts at three test sites used at the Norwegian Institute of Technology in Trondheim reported by Westerlund (1978). From these results, it was found that the relation proposed by Hardin (1978)

625 · OCRk' · ( p'·Pa

)°

⁵

G o =

-03 +07-e² where OCR

=

overconsolidation ratio

p'

=

mean effective stress (cr'₁+cr'₂+cr'₃) / 3, kPa pa

=

reference pressure

=

98.1 kPa

k'

=

a factor related to the plasticity index of the soil e

=

void ratio

gave a good estimate of G_Oin low-plastic soils. Also the results in the tests in the current investigation performed in more silty soil support this finding.

Investigations and load tets in silty soils 173

The exponentk' decreases with the plasticity index and is close to zero in silty soils.

The effect of the overconsolidation ratio can therefore often be disregarded in silt.

• Pore pressure dissipation tests

The method of stopping the penetration process in CPT tests at a number of levels and studying the dissipation of the excess pore pressure with time in order to obtain an estimate of the consolidation properties in the soil has been used for about 20 years. Much research has been aimed at finding methods of evaluating the coefficient of consolidation cv1z· These methods have found very limited use in Sweden, partly because of uncertainties in the evaluation of the results with employment of very large and rough empirical correction factors to yield useful parameters. The time required for a sufficient degree of pore pressure dissipation in the highly compressible and low permeable soft Swedish clays is also so large that the method is not rational either from a technical or economic point of view.

However, the method can be used to great advantage for other purposes. It can thus be used for rough estimates of the general drainage conditions in various layers in the soil profile. A coarse division of the drainage conditions with respect to normal loading conditions, that is when a load is applied gradually over a time span of at least a week, can be made from short dissipation tests lasting no more than 5 minutes each, (Larsson and Mulabdic' 1991). The soil can thus be classified as

free-draining if less than 20 % of the generated excess pore pressure remains after 5 minutes stop in penetration

semi-draining if between 20 and 60 % of the generated excess pore pressure remains after 5 minutes stop in penetration

non-draining if more than 60 % of the generated excess pore pressure remains after 5 minutes stop in penetration.

This information is useful for estimation of whether drained or undrained param­

eters, or both, should be considered relevant for estimation of bearing capacity, whether there is a need for vertical drains etc.

Short dissipation tests in more permeable layers are also useful for determination of the in situ pore water pressure. In such layers, the full dissipation, which can be checked by the measured pore pressure dissipation-time curve, often takes place within seconds. This procedure enables a rapid and rational determination of the pore pressure profile in the ground, which is often not hydrostatic. The established

SGI Report No 54

174

profile is representative only of the conditions during the time for the test, but it is often sufficient for interpretation of the CPT test itself. It often also significantly reduces the required number of piezometers in the long term observations and helps in selecting their locations.

In document LARSSON ROLF (Page 168-177)