LARSSON ROLF

(1)

STATENS GEOTEKNISKA INSTITUT SWEDISH GEOTECHNICAL INSTITUTE

Investigations and load tests in silty soils

Results from a series of investigations in silty soils in Sweden

ROLF LARSSON

LINKOPING 1997

(2)

(3)

~ STATENS GEOTEKNISKA INSTITUT

SGI SWEDISH GEOTECHNICAL INSTITUTE

v

Rapport

No54

Report

Investigations and load tests in silty soils

Results from a series of investigations in silty soils in Sweden

ROLF LARSSON

This project was partly financed jointly by the Swedish Council for Building Research (BFR), Grant No. 930591-70 and the Swedish Geotechnical Institute.

LINKOPING 1997

(4)

Report

Order

ISSN ISRN SGI project no Edition

Printer

Swedish Geotechnical Institute SE-581 93 Linkoping

SGI Literature Service Phone: 013-20 18 04 Fax: 013-20 19 09 E-mail: info@geotek.se

Internet: http://www.sgi.geotek.se

0348-0755 SGI-R--97/54--SE

19409453

SOO

Roland Offset AB, Linkoping, July 1997

(5)

Preface

'""rbis report deals with the results of a research project concerning investigations

.l

and evaluation of properties in silty soils. The investigations in this part of the project have been preceded by a comprehensive literature survey reported in SGI Report 49 "Silt - Geotechnical Properties and their Determination". Thereafter, a series of investigations has been performed at different locations with silty soils and the results from different types of sounding tests have been compared to each other and to results from laboratory tests concerning properties and classification of the soils. In three locations, a number of in situ and laboratory tests have also been performed and compared to the results from large scale field loading tests.

The report contains comments and recommendations on the use of the various types of investigation methods together with special precautions that should be observed in silts because of the special properties and conditions encountered in this type of soil. It also contains comments and recommendations regarding commonly used methods for estimation of properties from the different types of test results and the calculation of bearing capacity of shallow foundations and settlements in silt. Most of these methods have been elaborated for use under normal conditions in sands and the findings in this report do not necessarily reflect their usefulness in these conditions.

The report is intended for geotechnical engineers, both designers and researchers, and others who are involved in geotechnical investigations and design in silty soils.

The research project has been financed jointly by the Swedish Council for Building Research (BFR), Grant No. 930591-0, and the Swedish Geotechnical Institute.

The author wishes to express his thanks to all those who have participated and co

operated in the project. Special thanks go to Per U5fling and his colleagues at the Swedish National Road Administration and to Mats Larsson and Lars-Goran Ivers at KM-Geokonsult AB for their invaluable help in locating and establishing the test fields in the Borlange area, and also to the owners, Mr. Lars-Ake Mattsson and

Investigations and load tets in silty soils 3

(6)

Vasakronan AB, for their kind permission to use the land. Special thanks also go to Anna-Lena Oberg and her colleagues at Chalmers University of Technology for their co-operation in the tests concerning the variability of the ground water situation in Vatthammar. Many other colleagues, both at the Swedish Geotechni

cal Institute and other institutions, have also been involved and contributed to the project to various extents. Finally, the successful execution of the project is largely a result of the skill and dedication of the staff at the Institute's division for Field and Measuring Techniques, often exerted under long working days and harsh climatic conditions.

Linkoping, December 1996 Rolf Larsson

SGI Report No 54 4

(7)

Readers guide to this report

Why this report is special

This book gives a broad review of the usefulness of different investigation and calculation methods when applied to silty soils. It focuses on the practical application and use of the most rational ways of solving the engineering problems encountered. It describes the possible sources of error when using different methods and also the possible hazards in design arising through the extensive influence of the ground water conditions, which during the lifetime of the structure may be very different from the conditions at the time of site investigation. The report also describes problems encountered during the execution of the test programme, which are very common also in practical foundation works but which are rarely reported.

The goal of this report

The goal of this report is to provide recommendations for the methods that should be used in investigations in silts, the special precautions that should be observed when using the methods and the way in which the results should be interpreted. The report also contains recommendations for the way in which bearing capacity and settlements for shallow foundations should be calculated in this type of soil.

Who should read this report and why

This report will give the reader an insight into the advantages and shortcomings of the investigation and calculation methods normally employed in silty soils.

The report will be useful for geotechnical engineers planning investigations for selecting the most appropriate methods, giving instructions on how they should be carried out, specifying what supplementary investigations and observations should be made and stating how the results should be interpreted. It is useful for field engineers understanding how the methods work in this type of soil, what precautions have to be taken and why it is important to follow different special

(8)

procedures. It also gives guidance to designers when selecting calculation methods, parameters and critical conditions during the design life. It gives an insight into the problems that may occur during execution of the foundation works and the precautions that need to be taken in that context.

Finally, the report gives clients for geotechnical investigations and design an insight into the relevance of different methods and procedures, which enables a better understanding of the quality of different procedures and why certain precautions have to be taken.

How this report is organised

The report starts with a summary setting out the main results and recommenda

tions. It then continues with a detailed description of the investigations and tests performed and the results from these. The results of the investigations and the practical experiences and observations in connection with these are summarised in a special chapter, where also more detailed recommendations are given for the investigations that should be performed and the special precautions to be observed.

A description of the various methods for calculation of settlements and bearing capacity of shallow foundation then follows. Finally, the results obtained with these methods are compared to the results obtained in the large scale loading tests in the field and more detailed recommendations are given for the methods and precautions that should be employed.

SGI Report No 54 6

(9)

Empirical methods Calculations based on stress distribution according to theory of elasticity and moduli Calculation of settlements of footings based on Menard type pressuremeter tests The Briaud method of calculating settlement and bearing capacity from results of pressuremeter tests. Other approaches in order to take the variation in the moduli into account 6.3 Calculation of bearing capacity ... 211

The general bearing capacity equation Empirical methods based on sounding test results Calculation based on pressuremeter test results Calculation of bearing capacity with respect to maximum settlement criteria 7. COMPARISON BETWEEN PREDICTED AND MEASURED SETTLEMENTS ... 223

7.1 Mjardevi ... 223

7 .2 Vagverket ... 226

7.3 Vatthammar ... 232

· Effect of negative pore pressure · Summing-up of the results of the settlement predictions at Vatthammar 7.4 Summary of the findings in the comparisons of settlement predictions ... 239

8. COMPARISONS BETWEEN PREDICTED AND MEASURED BEARING CAPACITY ... 241

8.1 Vagverket ... 241

8.2 Vatthammar ... 245

REFERENCES ... 251

SGI Report No 54 10

(13)

Summary

Effect of ground water conditions

The results from the present investigation have illustrated the paramount impor

tance of the ground water conditions for the engineering properties of silt. The ground water conditions may vary from high free ground water levels and artesian ground water pressures to low-lying free ground water levels with a considerable matrix suction in the soil above. In areas with deep silt deposits with alternating plains and deep ravines because of erosion by streams and rivers, the conditions may be highly variable. The conditions may also be highly variable during different seasons, with significantly higher ground water levels and pore pres

sures, particularly during the thawing and snow melting season in spring. Because of the often high degree of saturation in silt also above the free ground water level, the ground water conditions can react rapidly on precipitation and even limited amounts of rainfall can cause significant changes in the ground water conditions also during the other seasons.

The ground water conditions during the time for the field investigations have a significant effect on the results and the evaluation should take this into consider

ation. Also the results of in situ loading tests, such as pressuremeter tests, screw plate tests and plate loading tests, are heavily dependent on the prevailing pore pressure conditions in the soil during the tests and have to be evaluated according

ly. This entails that, at low-lying ground water tables, also the matrix suction in the soil above has to be measured. Furthermore, the bearing capacity of foundations in silt is also heavily dependent on the groundwater conditions prevailing through

out the lifetime of the construction. The pore pressure conditions therefore have to be measured at the time for the field investigations and for a sufficient time to enable a prediction of the extreme maximum pore pressures during the lifetime of the construction. This should be done with a frequency adapted in such a way that also the effect of periods with heavy rainfall can be studied. Construction in silt often entails excavation down to frost free depths and precautions should be taken

(14)

so that any ground water conditions occurring during this operation can be dealt with in a satisfactory way.

Site Investigations

Site investigations in order to determine the statigraphy of the soil should preferably be made by CPT tests with simultaneous measurement of the generated pore pressure, (also called piezocone tests or CPTu tests). In very stiff soils, very deep profiles and profiles with embedded coarse or stiff layers or larger objects, CPT tests may have to be supplemented by dynamic probing tests in which the probe can be advanced by blows. On the other hand, the latter type of test has a much poorer resolution and is not very useful in very loose silt. The results of the CPT tests are significantly affected by the in situ pore pressure conditions and these have to be taken into account. Also a matrix suction has a large effect on the results and unless this is taken into account the soil becomes classified as coarser and stiffer than is actually the case. When the pore pressure conditions are properly considered, the classification proposed in SGI Information No.15 (Larsson 1992) appears generally to function well but the results need to be inspected carefully in order to avoid mistakes.

The results of weight sounding tests in silt were found to be generally unreliable.

This is probably mainly related to the effects of the ground water conditions, for which these results cannot be corrected.

In all site investigations in areas where artesian ground water pressures may exist, the need for sealing of test holes and sampling holes must be considered. This should preferably be established very early in the investigation process by measuring the stabilised pore pressures in pore pressure dissipation tests in deeper and more permeable layers during the initial CPT tests.

Determination of shear strength

The drained shear strength properties in silt in terms of a friction angle can be estimated from CPT tests, provided that the test has been performed under drained conditions. This can be controlled by the pore pressure readings which must then register no or only very small generated excess pore pressures. When significant excess pore pressures are registered, the Senneset and Janbu ( 1984) method can be applied. However, it should then. be considered that the friction angle thus evaluated primarily relates to the friction angle at constant volume, which may often be used in evaluation of the stability of natural slopes but which is too high

SGI Report No 54

12

(15)

for evaluation of bearing capacity of foundations on loose silt where the soil is compressed. In the latter case, it is better to use empirical values based on soil type and relative stiffness as presented by Bergdahl et al (1993) and the Swedish National Road Administration (1994).

The undrained shear strength, when applicable, is best determined by triaxial tests in the laboratory. Field vane tests may also be used but the risk of significant disturbance is great, particularly in varved and layered soils, and the tests may not be relevant for undrained conditions. The relevance of undrained tests and the test method in the particular soil can be estimated by dissipation tests during the CPT soundings.

Determination of compressibility

The compressibility of silt may be estimated from CPT tests, provided that checks have been made to ensure that the tests have been drained, and in stiffer silt from dynamic probing tests using the empirical relations presented by Bergdahl et al (1993) and the Swedish National Road Administration (1994). For CPT tests the modulus can be estimated from E

=

4.3q/⁹³and for dynamic probing tests the relation may be written E 2.8N_20!HJA;°-91 . The modulus estimated in this way tends to become too low at shallow depths in crusts and stiff soils.

A more reliable estimate of the compression modulus is obtained by dilatometer tests. The dilatometer can be used in a large number of soil types and, with a special procedure described in the report, it is also possible to sort out and correct values in layers where the penetration causes excessive disturbance, such as varved clayey silt and alternating thin layers of silt and clay. This reduces the need for supplementary undisturbed sampling and oedometer tests to layers of normally consolidated or only slightly overconslidated clay. Similar to the CPT test, the soil classification from the dilatometer tests requires the pore pressures and possible matrix suction to be measured and accounted for in order to yield good results.

When this requirement is fulfilled, the classification proposed in SGI Information No. 10 (Larsson 1990), appears to function well.

Pressuremeter tests may also be used to determine the compressibility of the soil.

However, it is difficult to create a test cavity of good quality at levels below the free ground water level by predrilling a hole. Above the ground water level, the results appear to be heavily dependent on the matrix suction and their relevance has to be examined.

(16)

The initial shear modulus G₀, also called the dynamic shear modulus, can be determined by seismic cone tests or by the empirical relations presented by Hardin (1978).

Sampling and laboratory tests

It is possible to obtain relatively undisturbed samples in silt by Swedish standard piston sampling. The samples then retain enough of their stucture to be tested for undrained shear strength in triaxial tests and for compressibility in oedometer tests.

The latter should preferably be performed as CRS tests. However, the sampling operations and the amount of oedometer testing required to obtain a good picture of the compressibility in the soil profile are so extensive that for rational reasons the oedometer tests should be restricted to such layers where the results of the in situ tests are not relevant, i.e. mainly layers of normally consolidated or only slightly overconslidated clay.

Standard piston sampling is also preferred when the samples are taken for verification of the soil classification and there is less demand for high quality samples. In Sweden, the soil in silt profiles is often more or less varved and layered and this and the related geotechnical implications are often largely missed out when coarser sampling methods are employed, in which the soil is more or less remoulded.

Both the classification and the performance of classification tests in this often varved and layered soil require certain consideration. It should also be observed that in silt, the Swedish classification based on grain size distribution may differ considerably from other classification systems which are based on results from consistency limit tests.

Calculation of settlements

Calculation of settlements can be made by the use of theory of elasticity and moduli. The results of the present investigations indicate that the distribution of settlements with depth agrees fairly well with the distribution calculated in this way. The only notable exception is that the measurements indicate that no significant settlements occur below a relative depth of 2b - 2.5b under a square plate with width b, while the theory of elasticity indicates that some small settlements will occur below this level. The moduli are best determined by dilatometer tests, possibly supplemented by oedometer tests in more clayey layers.

Moduli estimated from CPT tests and dynamic probing tests may also be used with certain reservations. The settlements can also be calculated from results of

SGI Report No 54

14

(17)

pressuremeter tests using the Menard method. All of these settlement calculations normally refer to 10-year settlements.

The settlements calculated in this way in principle yield a straight line relation between load and settlement. However, in reality the load-settlement relation is continuously curved with a continuously decreasing modulus with increasing load. The calculated settlements therefore mainly correspond to the real settle

ments at a certain relative settlement which was found to be about 0.014b. For smaller relative settlements, the aforementioned methods yield too large calculat

ed values and at larger settlements the calculated values become too small. This can be rectified by using the proposed method which takes the curved load-settlement relation into account.

Another way of calculating the continuously curved relation between load and settlement is to use the method proposed by Briaud (1995) for results from pressuremeter tests. The method has been used for the results in the present investigation with good results but a modified T-factor for tranferring pressurem

eter pressure to footing pressure has to be applied, (Larsson 1997), and some uncertainty regarding the time function remains. Care must also be taken when a matrix suction plays a significant role for the test results and the bearing capacity.

The DeBeer and Schmertmann methods often employed for calculation of settle

ments in sand yield very conservative results if they are applied in silt, i.e. the calculated setttlements become much too large.

Calculation of bearing capacity

The ultimate bearing capacity of foundations on silt at failure can be calculated by the general bearing capacity equation. In these calculations, the effect of a matrix suction has a great importance. However, if it cannot be ascertained that a matrix suction will prevail throughout the lifetime of the construction, these effects should rather not be taken into account.

The ultimate bearing capacity can also be calculated from results from pressurem

eter tests but it should be observed that in case the pressuremeter tests were conducted at levels where a matrix suction was prevailing at the time for the test, the results only apply for conditions with a similar matrix suction.

The bearing capacity at a given limiting settlement can be calculated using the proposed method of calculating settlements with stress distribution according to

(18)

theory of elasticity and moduli taking the curved load-settlement relation into account. It can also be calculated from pressuremeter test results using the method proposed by Briaud (1995) with the modifications proposed by Larsson (1997).

However, also in this case the restrictions related to matrix suctions apply, particularly if a higher bearing capacity is utilized than that calculated by the general bearing capacity equation without regard to the matrix suction.

SGI Report No 54

16

(19)

Chapter I.

Introduction

Purpose and background of the investigation

Silty soils occur frequently in Sweden and entail a considerable number of special geotechnical problems concerning field investigations, sampling and laboratory testing, frost susceptibility, ground water conditions, excavations, handling and compaction of soil masses, bearing capacity and settlements, among other things.

In this project, questions regarding field investigations, sampling and laboratory testing, ground water conditions, bearing capacity of shallow foundations and settlements in natural soils below footings and fills are addressed.

In spite of being a common type of soil, the number of special investigations in silts is limited and the number of special investigation and calculation methods few.

Instead, methods originally intended for clays or sands are normally used, in combination or individually, depending on what is considered most appropriate in the particular case. Empirical methods for estimation of parameters and the methods elaborated for calculation of bearing capacity and settlements on the basis ofresults from sounding tests in sands are often cautiously modified with respect to silt content in the sand. However, in most cases the originators have never claimed that they should be applicable to pure silts or even clayey silts, and the relevance of the results is very uncertain.

In Sweden, earlier practice when investigating silt deposits has to a great extent consisted of performing weight sounding tests and/or dynamic probing tests to estimate the denseness of the soil and taking disturbed samples to verify the soil type. From the results, design friction angles and a modulus of elasticity have been estimated on the basis of empirical experience. In fine and clayey silts, vane tests in the field and oedometer tests on undisturbed samples in the laboratory have often been performed as supplements. Other sounding methods such as total pressure sounding, in situ tests such as screw plate tests and Menard-type pressuremeter tests, and laboratory tests such as direct simple shear tests have also been employed to some extent.

(20)

Furthermore, deposits of silty soils are often inhomogeneous, with alternating layers of coarser soils and clayey silt or clay, which often necessiates different investigation methods for the different layers.

In recent years, new investigation methods, such as the piezocone tests (CPT test) and the dilatometer test, have appeared. In particular, the dilatometer has already proved to be a useful tool for investigating the compressibility of silty soils and has gained a certain acceptance in the current practice. A certain development of the older methods has also taken place and some new interpretation methods and calculation methods have been presented. However, many of these interpretation and calculation methods have, like the older methods, been intended mainly for clean sands.

The present investigation has therefore been aimed at investigating the useful

ness of the different investigation methods in silty soil deposits and determining the possible modifications to the existing interpretation methods that should be performed. The aim has also been to investigate the present methods of calculating bearing capacity of shallow foundations and settlements in this type of soil. The final goal has been to find a recommendation as to which investigation method or combination of methods should be used in different types of deposits, the special precautions that should be observed in connection with investigations in this type of soil, how different soil properties should be evaluated and how bearing capacity and settlements for shallow foundations should be calculated.

All references to evaluated properties, soil classifications etc. from field and laboratory tests in this report refer to established methods used in Swedish practice unless otherwise stated.

The test and interpretation methods commonly used in Sweden are described in the following publications:

SGI Report No 54

18

(21)

Test method Test Inter- Publication procedure nretation

Weight sounding X X Bergdahl (1984). Geotekniska undersokningar i falt. SGI

test Information No. 2

X Bergdahl et al. (1993). Plattgrundlaggning. Svensk Byggtjiinst.

X TC 16 of the Intemational Geoteclmical Society (1988), Reference test procedures CPT-SPT-DP-WST. SGI Information No.7.

Dynamic probing X X Bergdahl (1984). Geotekniska undersokningar i falt. SGI

test type HfA Information No. 2

X Ber!!dahl et al. (1993). Plattgrundliiggning. Svensk Bw2tjiinst.

CPT-test X Swedish Geotechnica] Society SGF (1993). Recommended

standard for CPT-tests.

X X Larsson (1992 ). The CPT-test. SGI Information No. 15.

Field vane shear X X Swedish Geoteclmical Society SGF (1993). Recommended

test standard for field vane shear tests.

Seismic cone test X Campanella et a] (1986). Seismic cone penetration test. Proc. In Situ 86. ASCE.

X X Larsson and Mulabdic (1991). Shear moduli in Scandinavian Clavs. SGI Report No. 40.

Dilatometer test X Swedish Geotechnical Society SGF (1993). Recommended standard for dilatometer tests.

X X Larsson (1990). Dilatometerforsok. SGI Info1111ation No. 10.

Pressuremeter test X X Baguelin et al. (1978). The Pressuremeter and Foundation Engineering. Trans. Tech. Publications.

Pore pressure X Tremblay (1990). Matning av grundvattenniva och portryck. SGI

measurements Information No. 11

Plate load tests X X Bergdahl (1984). Geotekniska undersokningar i fiilt. SGI Information No. 2

X X Ber!!dahl et al. (1993). PlattgrundJaggning. Svensk Bvggtianst.

Sampling X Bergdahl (1984). Geotekniska undersokningar i falt. SGI

Information No. 2

Classification X X Karlsson and Hansbo (1984). Soil Classification. Swedish Council ofBuildine Research, T21:1982.

Bulk densitv X X Swedish standard SS 02 7114

Water content X X Swedish standard SS 02 7116

Liauid limit X X Swedish standard SS 02 7118

Plastic limit X X Swedish standard SS 02 7118

Sedimentation test X X Swedish standard SS 02 71 24

Oedometer test X X Swedish standard SS 02 71 29

with incremental loading

Oedometer test X X Swedish standard SS 02 71 26

with constant rate of deformation

Scope of the investigation

The project was started with a thorough review of the existing geotechnical literature on silty soils. The contents were synthesised and reported in SGI Report No. 49 "Silt - Geotechnical Properties and their Determination", (Larsson 1995).

The following part of the project reported here should comprise extensive investigations with all available and possibly relevant methods and large scale loading tests in three locations covering a wide range of conditions in silty soils.

(22)

A series of investigations was then started in order to find suitable locations for test fields where different in situ tests as well as large scale loading tests could be performed and compared. An inventory of available older investigations was made and a number of possible locations were selected. One of these locations, with very heterogeneous silty soils, is located in Linkoping close to the Institute and full scale loading tests had already been performed in connection with preloading of the soil before the construction of buildings. In this area, only supplementary investiga

tions with new test methods were required.

The new investigations in the other locations proved to illustrate the lack of precision of the older investigation methods. Site after site was eliminated because unexpected soil types or layers that had previously been undetected or misinter

preted were encountered. Although the results have been very useful in evaluating the different investigation methods employed, the search for suitable test sites had to be performed in another way.

The problem was solved with help from the Swedish Road Administration in Borlange and the consulting firm KM-Geokonsult AB, both of which had access to more detailed investigations in the Borlange area where it was ascertained that the soil conditions were suitable. Two locations with very different conditions were selected, one with an approximately 15 m thick deposit of loose fine to medium silt with some layers of clayey silt and a free ground water level about 2 m below the ground level, and one with a deposit of dense medium to coarse silt at least 10 m thick with a free ground water level about 18 m below the ground surface. Quite homogeneous profiles with silt are rare and also the latter profile contains thin layers of more fine-grained soil. Both deposits overlie coarser soil extending to great depths, but these layers do not affect the results from the loading tests.

The first profile later proved to have artesian water pressure in the bottom layers and this and especially the free ground water level varied strongly with season and rainfall. This created considerable problems at excavation and installation of the plates for the load tests, which were to be placed at the normal foundation depth below the frost limit and the weathered zone. The pore pressures have therefore been recorded for long periods together with meteorological and hydrological observations, and special precautions had to be taken to control the ground water situation at installation of the plates. The results from the investigations and the loading tests in the second profile proved to be heavily dependent on the prevailing negative pore pressures in the ground. The measurements of the pore pressure profile at the investigations and during the loading tests at this site were later

SGI Report No 54

20

(23)

supplemented by a special study of the possible effects of thawing and very heavy rain. In this study, the ground was soaked by filling water into a shallow excavation covering a large area. The pore pressure profile was then monitored by frequent automatic readings of installed pore pressure gauges for a period of time. This study was conducted by Anna-Lena Oberg at Chalmers University of Technology as part of her studies of the effect of negative pore pressures on stability conditions in silty soils.

The large scale loading tests in Borlange were conducted as two series of plate loading tests. At each location, three load tests were performed on square plates with dimensions 0.5 x 0.5, 1.0 x 1.0 and 2.0 x 2.0 m.

The results of the tests have then been compared to bearing capacities and settlements calculated with the currently available interpretation and calculation methods. As stated before, many of these methods were not originally intended for use in silts and the purpose was then to examine the possibility of extending their use also to this type of soil.

(24)

Chapter 2. Investigations

I

nvestigations have mainly been performed at five sites, Fig. 2.1. Two of these were discarded after the preliminary supplementary investigations, which were aimed at confirming the suitability of the sites for comparative studies of full scale loading tests and predictions based on ordinary geotechnical investigations. Two other sites were also considered, based on results from available investigations, but were discarded after a visual inspection on site. In one case, this was due to a very uneven terrain combined with large trees and boulders on the surface, which would have entailed a considerable cost to remove and excavate. In the other case, the intended test site was considered to be too close to a residential area. Extensive programs of investigations and load tests have been carried out in three main test locations.

Kil

The preliminary supplementary investigations consisted of CPT tests and the taking of "undisturbed samples" with a Swedish standard piston sampler in order to verify the stratigraphy and the estimated type of soil in the various layers. In the first profile in Kil, which was found in connection to planning of a new road, the available investigations, consisting mainly of weight sounding tests and disturbed sampling, had indicated a fairly homogeneous deposit of dense to very dense silt.

Supplementary dynamic probing tests had shown that the thickness of the deposit was about 50 m. The silt had been judged to be somewhat clayey close to the ground surface and to contain alternating layers of coarse and medium silt further down. However, the supplementary investigations showed that the upper 6.5 m in the profile consisted of medium stiff partly silty clay. Below this level, alternating layers of medium dense sand and medium stiff clay were found down to 10 m depth, where dense sand was encountered. The free ground water level was located at 12 m depth below the ground level and the influence of the resulting negative pore pressures in the upper part of the profile appears to be the main reason for the original misinterpretation.

SGI Report No 54

22

(25)

Fig. 2.1. Location of test sites.

(26)

Branas

In the second profile in Branas, which was close to a bridge over a river founded on piles in silt, research had previously been carried out concerning the growth in bearing capacity of driven friction piles with time, (Astedt et al. 1992), and it was ascertained that the profile contained silt to great depths. Results from various sounding tests and dilatometer tests as well as certain laboratory triaxial tests were also available. However, the supplementary investigations showed that the upper 4 m of the profile consisted of sand which had not been pointed out and was probably considered insignificant in the previous investigations. The free ground water level was regulated by the river and was about 1.5 m below the ground surface. It would thus have been very difficult to perform plate loading tests at the site in such a way that the results reflected the properties of the underlying silt.

Mjardevi

At the third location, Mjardevi, which is close to the Institute in Linkoping, investigations have been performed at various times for a number of years in connection with development of the area. The first investigations consisted of static total pressure sounding, dynamic probing tests and disturbed sampling. The results showed very heterogeneous profiles with mainly silty soils containing alternating layers of silt and stiff clay and with infusions of lenses and coarser objects. The infusions of coarser objects increased with depth and rotation often had to be applied to advance the static total pressure sounding. In spite of this, the penetration depth of the static total pressure soundings was often limited and normally varied between 5 and 15 m. The dynamic probing tests were in most cases stopped above 20 m depth even if they could be advanced further since a number of initial tests to greater depths had resulted in breakage and loss of the sounding rods.

An attempt was made to estimate the compressibility of the soil by Menard type pressuremeter tests. However, the ground water level was about 1 m below the ground surface and the difficulties in obtaining proper pre-drilled holes proved to be very great. The results were therefore considered unreliable. Because of this and the heterogeneity of the soil in the area, which entailed a great risk for uneven settlements, it was decided to preload the ground where the buildings were to be constructed. This was done by means of earth fills covering the whole area of the projected buildings and high enough to provide a load that corresponded to that of the future buildings plus a surcharge. The fills were instrumented with settlement gauges and horizontal settlement hoses. The settlements were followed during load application and for some time afterwards until the fills were removed and the

SGI Report No 54

24

(27)

buildings constructed. The method proved to be very successful and has since then been utilised for most of the area, even if the instrumentation is usually omitted.

The wisdom of this approach was later illustrated in a case where higher buildings were constructed in the area and the foundations had to be made on piles. The site investigation demanded dynamic probing tests to estimate the piling depth and resulted in heavy losses of drilling equipment. In the following piling operation with precast concrete piles, a large number of the driven piles were lost.

In connection with the introduction of the dilatometer test in Sweden, a new series of investigations was performed in the area. It then proved possible to penetrate and test to approximately the same depths as for the static total pressure sounding, and it was also possible to use the results to select levels at which fairly undisturbed samples could be taken with the standard piston sampler. The tests and measure

ments at Mjardevi have been used in practical design and gathered in SGI files, but have not been published before.

In connection with the present project, a series of CPT tests and weight sounding tests has been performed. Also these tests reached about the same levels as the previous static total pressure soundings and the dilatometer tests.

Vagverket

The fourth test site is located just outside the head office of the Swedish National Road Administration (Vagverket) in Borlange between the buildings and a nearby ravine with a brook. The location of the test site originates from the investigations at the construction of the buildings in the late 70's. Further investigations have been performed close to the present test site in connection to a research project concerning stress conditions and movements in and close to natural slopes, (Andersson et al. 1991). The previous investigations mainly consisted of weight sounding tests, dilatometer tests, pore pressure observations and undisturbed sampling. These investigations had shown that the soil in the profile consisted of loose silt with a more significant layer of clayey silt/silty clay at about 5 m depth and some less pronounced clayey layers between 8 and 15 m below the ground surface. Below about 15 m depth, there is a thick sand layer which is estimated to extend to at least 40 m below the ground surf ace. The pore pressures had been observed on 6 occasions during about half a year from December to July. The situation had then been found to be fairly stable with a free ground water level about 2.3 m below the ground surface and gradually increasing artesian water pressures below this level down to the sand layer 15 m below the ground surface,

(28)

where the water pressure corresponded to a hydrostatic water pressure from the ground surf ace. The most significant deviation had occurred in April, which would coincide with the snow melting and thawing period in the spring, when the free ground water level was about 1 m higher. The corresponding pore pressure increase further down in the profile decreased gradually to only about 2 kPa in the bottom sand layer. The results from the dilatometer tests indicated that the soil is slightly overconsolidated at the top and becomes normally consolidated with depth.

The investigations in the current project comprised CPT tests at four points, additional dilatometer tests at two points, field vane tests at two points, Menard type pressuremeter tests in two pre-drilled holes (plus a few tests in an additional hole), one seismic CPT sounding, pore pressure dissipation tests in one CPT test, pore pressure observations at different levels at a number of points and "undis

turbed" piston sampling at two points down to 10 m depth. Dynamic probing was not considered to be relevant in this soft soil.

The samples have been investigated in the laboratory concerning classification, bulk density, water content, liquid limit, plastic limit and grain size distribution.

The compressibility has been investigated in oedometer tests, both incrementally loaded tests and automatic constant rate of strain tests. The shear strength parameters have been tested in triaxial tests at SGI and also at the Norwegian Geotechnical Institute in connection with another research project concerning the dynamic behaviour of silt, using samples taken in the current investigation at Vagverket.

Certain additional investigations and measurements, mainly concerning the var

iability of the ground water conditions, later became necessary because of problems encountered at the installation of the plates for the load tests.

Vatthammar

The last site, Vatthammar, is located at Stora Tuna about 5 km south-east of Borlange. This site had been investigated in connection with a proposed new railway crossing for a local road. These previous investigations comprised weight sounding tests, CPT tests and disturbed sampling. They indicated that the soil consisted of very stiff silt down to at least 10 m, that the soil profile was thicker than 15 m and that the free ground water level was located below 11 m depth.

SGI Report No 54

26

(29)

The new investigations comprised three CPT tests, one seismic CPT test, two dilatometer tests, one borehole with Menard-type pressuremeter tests, one dynam

ic probing test, pore pressure measurements with the CPT-equipment in the deeper soil layers and measurement of negative pore pressures by BAT-piezometers at a number of levels in the upper part of the soil profile. "Undisturbed" samples were taken in two holes down to 10 m depth and were investigated in the laboratory concerning classification, bulk density, water content, liquid limit, plastic limit and grain size distribution. The compressibility has been investigated in oedom

eter tests, both incrementally loaded tests and automatic constant rate of strain tests. The shear strength parameters have been tested in triaxial tests at SGI.

After the following plate load tests, a special investigation was performed by Anna-Lena Oberg at Chalmers University of Technology concerning the possible variations in the pore pressure profile at soaking of the top of the profile. In these investigations, the pore water pressures in the upper part were measured by continuously monitored BAT piezometers. Supplementary to these measure

ments, the water retention curves of the soil in the profile were determined in the laboratory at Chalmers University of Technology and the capillarity of the soil was measured in a new type of capillarity meter developed at SGI. Tests with the latter equipment were also performed on the soil in the first profile in Kil.

The two test fields in the Borlange area cover a large part of the range of conditions in natural fairly homogeneous silt deposits in Sweden, from a loose, only slightly overconsolidated deposit of partly clayey silt with a high free ground water level and artesian pore water pressures to a medium stiff silt with a deep free ground water level and negative pore water pressures in a large part of the profile. There are deposits with coarser silts, but these are closer to sand and may be expected to behave in a similar way. There are also stiffer silts, mainly in the form of silt moraine, but these are outside the scope of the present project which has been restricted to sedimentary silt deposits.

It would have been desirable to perform more types of tests, particularly in the last two test fields, but this was not possible within the present project. One such test is the screw plate test, which has previously been found to be useful in silts.

However, the previous models of this equipment have been too weak and vulnerable, and no operational equipment was available. Development of a new, more robust type of equipment is reported to be under way at the Norwegian Institute of Technology in Trondheim, but this was not ready in time for the cmTent project. Another type of equipment, which it would have been desirable to test, is

(30)

the self-boring pressuremeter. At present, there is only one operational unit of this type in Sweden and unfortunately it was not possible to have it brought to Borliinge at the time for the investigations. A further type of pressuremeter, the ^TEXAMtype, also allows more elaborate measurements and interpretations to be made. Howev

er, this type of pressuremeter was not introduced in Sweden before the investiga

tions in the present project were finished.

SGI Report No 54

28

(31)

Chapter 3. Principle of the plate loading tests

r-r,ie

plate loading tests were performed in order to check the applicability of

.l

different methods of estimating bearing capacity and settlements for shallow foundations in silt. The tests were performed in series of tests on square plates with dimensions selected in such a way that ultimate bearing capacity failure was expected to be reached for at least the smallest plates and with such variations that the effect of the stresses reaching down to various depths could be studied. The largest plates also had dimensions similar to an ordinary foundation.

Foundations in Sweden are made so deep that they remain unaffected by frost action, which normally means a foundation level somewhere between 1.1 and 2.5 m below the ground surface, unless special precautions are taken. In the plate loading tests, it was also desired to lay the plates on top of fairly homogeneous soil below the desiccated and stiffer crust in order to facilitate the interpretation of the results. In ordinary foundations, a part or all of the excavation for the foundation is often back-filled after the foundation work has been completed.

In the current case, load tests with basically the same equipment should be performed at two sites with very different soil conditions. Much of the loading equipment is standard equipment used for this kind of tests, which sets limits for possible dimensions and maximum loads. The loading equipment consisted of a system of loading beams, ground anchors and hydraulic jacks and pumps. The main beams consisted of two 17 m long steel profiles, which could be bolted together. Such beams are kept in depots spread over the country by the Swedish National Rail Administration to be at hand in case they should be required for temporary emergency repairs of the railway lines and can often be made available for short term loading tests. The ends of these beams are placed on a support of wooden rafts of the type normally used to support excavators on soft ground. In this case, they are used to provide a firm and level base and the number of rafts at the two ends can also be adjusted to place the beams in a horizontal position. The required reaction force can be provided by deadweights or by ground anchors. In

(32)

this case, it was considered unsuitable to use deadweights, especially at the site with the soft soil, and a system of four ground anchors was installed at each site.

The ground anchors consisted of Swellex type expander bodies which were lowered in pre-drilled holes down to firm soil strata and then expanded by pumping in cement grout under high pressure. The pre-drilled holes at the site with the high ground water level were supported by a bentonite suspension. The ground anchors were placed in pairs, one on each side of the pair of beams, and fitted with tie rods extending well above the beams. Shorter beams were placed across the ends of the long beams and these were tied down with a pre-stress in the tie rods in order to fix the system.

The dimensions of the reaction system made it possible to install three plates measuring 0.5 x 0.5 m, 1 x 1 m and 2 x 2 m in a row along the beam without significant interference in terms of the same soil mass being affected by the different load tests. The distances between the plates and the loading sequence were also adjusted in such a way that the influence from a preceding load test on the results from the following load tests was minimised. In this way, the smallest plate was placed at one end of the row, the largest plate was placed about in middle but somewhat closer to the smallest plate and the intermediate plate was placed at the other end. The plates were then tested in a sequence with the smallest plate first and the largest plate last. The same plate dimensions were used at both sites.

However, since the preliminary calculations showed that it would be uncertain if failure could be reached even for the smallest plate at the site with the stiffer soil, most of the back-filling had to be omitted at this site.

The load on the plates was provided by hydraulic pumps and jacks. According to Swedish practice, the load in tests on friction soils is normally applied in steps with a duration long enough to enable a study of the creep rate of the deformations (Bergdahl et al. 1993). For cohesive soils, a longer duration is required to allow for full dissipation of excess pore pressures and related consolidation if the drained properties are to be studied. A minimum of ten steps is normally used to enable evaluation ofboth load-settlement curves and failure loads. In tests on friction soils and in undrained tests on clays, a duration for each load step of about 8 minutes is normally considered sufficient. In the present series of tests, the tests at the site with the stiffer and coarser silt and the very deep ground water level were performed in steps with 16 minutes duration. At the site with the softer and more clayey silt and with a high ground water level, preliminary calculations showed that longer duration was required in order to ensure full pore pressure dissipation.

The durations of the load steps in these tests were therefore selected to be 2, 3 and

SGI Report No 54

30

(33)

5 hours respectively for the three plates and the pore pressure in the soil below the plate was measured during the tests in order to verify that full excess pore pressure dissipation was achieved.

Different systems for application of both constant static loads, cyclic loads and dynamic loads by hydraulic jacks have been developed at the Institute. Because of interference with other ongoing load test projects, two different systems were used, one manually regulated for the short duration tests and one electronically operated in the more long term tests. The load was measured and regulated by means of an electronic load cell.

(34)

Chapter 4. Investigations and load tests at the three main test locations

4.1 MJARDEVI, LINKOPING Test field

The test field is located in western Linkoping,just across the road from the Swedish Geotechnical Institute and Linkoping University. The area has been developed in the late 80's and the 90's in connection with the construction of Mjardevi Science Park. The soil profile is dominated by silty soils but is very heterogeneous with alternating layers of silt and clay and many infusions of clay lenses and coarser objects, the latter increasing with depth. The weathered crust in the upper part of the profile is about 2 m thick and this and a somewhat softer layer about 1 m thick lying just below consist of more clayey soil. The ground water level is located about 1 m below the ground surface and may vary with season from the ground surface to 1.8 m below. The pore water pressure is approximately hydrostatic from the ground water level. The silty soil profile is estimated to be 16 to 17 m thick and is followed by coarser soil. Only the upper parts of the profile down to about 20 m depth have been investigated, the depth of the various investigations depending on the ability of the various methods to penetrate this type of soil.

First investigation

The soil conditions at the test site, which is relatively large and comprises half a block with five buildings, Fig. 4.1.1, were first investigated in 1988 before construction of the buildings, (Ottoson and Bergdahl 1988). The first investiga

tions were aimed at determining the type of soil, its stiffness and the possibilities for foundation on spread footings, and also to investigate the depth of penetration in dynamic probing and thereby the necessary length of driven piles if these were required. The investigations were performed with a local non-standard type of static total pressure sounding, dynamic probing according to the Swedish HfA method, pore pressure measurements at two levels and disturbed sampling by screw auger. The non-standard static total pressure sounding method basically uses the same principle as the ordinary Swedish static total pressure sounding

SGI Report No 54

32

(35)

" _C

100 m

Fig. 4.1.1 Plan of buildings in the test area.

method, i. e. cp 22 mm rods, a total pushing force of 10 kN and rotation of the rods with the maximum force applied when this force alone is insufficient to penetrate the soil. The difference is that the ordinary square tip with 1000 mm²cross section and a slip coupling, which enables separation of tip resistance and rod friction, is replaced by a twisted cp 25 mm screw-shaped tip normally used in the weight sounding test. The Swedish HfA dynamic probing test uses a 63.5 kg hammer with a free fall of 0.5 m and a cp 45 mm conical tip with apex angle 90° and a mantle length of 90 mm. The penetration resistance is registered as the number of blows required for every 0.2 m penetration. The rod friction is estimated by measuring the torque required to rotate the rods and the net penetration resistance is then calculated. From experience, this method yields approximately the same results as theSPT-test, N20 HfA. Net "" N_{30 SPT'}(Bergdahl and Ottosson 1988).

(36)

According to the results from these penetration tests, the silty soil generally appears to be very loose down to 3.5 m depth. It then becomes varyingly loose to medium stiff down to about 10 m depth and then mainly loose at further depths down to stop in penetration. The variation in the results was considerable, particularly below 3.5 m depth, Figs. 4.1.2 - 3. Penetration stop was obtained at widely varying levels. The static total pressure soundings stopped at depths varying from 5 to 20 m, in all cases but one at less than 15 m, and the cause was generally judged to be a large object such as a cobble or boulder. The dynamic probing tests generally penetrated somewhat further, but initial attempts to penetrate past what were believed to be large objects down to a firm bottom soon

Total penetration force, kN

5 10

5

^' '

^\_I

I II

.,.,

Rotation E

.s:::.

c.

_Ql ¹⁰

0

15

- Estimated average curve

Fig. 4.1.2 Variation in results from static total pressure soundings at Mjardevi.

SGI Report No 54

34

(37)

:, < (D V> rt ~- o· :::, V> p.) :::, 0.. 0 p.) 0.. ~ r;: 3·

~-

~ Q, V, v.l (J\

111.:::111.::: HfA j_ t 1

106 543 120 m + tip lost

IJ I '+ +0.00 HfA +0.00 HfA +0.00 /IIEIIIE Ill.Ell/= IIIEIIIE IIIE'IIIE 111=111:= 111.=111.=

L

59

rr

52 63

' 6

' ' I I I I I 60/0 1020304050 Blows/0.2 m 178 140

r

1.)-LI I' I I I so;o 1020304050123 1020304050 Blows/0.2 m Blows/0.2 m m rod + tip lost 1020304050 Blows/0.2 Fig. 4.1.3 Results of dynamic probing tests at Mjardevi. 7 m rods

(38)

resulted in breakage of the rods and loss of the tips together with a certain number of rods. In all cases, the encounteling of large objects was judged to be the reason for the stop in penetration.

The classification of the disturbed samples showed that the soil in the upper 7 .5 m consists mainly of silt but that it contains numerous thinner layers of clay. The dry crust on top, particularly its upper part, consists of clay. Also the softest layer just below the crust contained so much clay that it appeared possible to take relatively undisturbed samples with the Swedish standard piston sampler. This was attempt

ed in three boreholes where samples were taken at 2.5 m depth and in two holes attempts were also made to take undisturbed samples at deeper levels where more clayey soil had been found.

The latter samples provided more precise information on the stratification of the soil and showed that also the soil in the more clayey parts of the profile below the dry crust consists of varved/layered partly clayey silt and alternating varves/

layers of clay and silt. The clay also appears as small lenses in the silt. The unit weight of the soil is about 1.95 t/m³and the natural water content varies between 20 and 42 % depending on the clay content. A number of specimens were prepared for oedometer tests on the most clayey parts of the samples and constant rate of strain oedometer tests and incrementally loaded tests were performed, both types in accordance with the relevant Swedish standard. Indications of a preconsolida

tion pressure could be observed only from the results of two of the constant rate of strain tests. Both specimens were from 2.5 m depth and the indicated precon

solidation pressures were 110 and 204 kPa respectively. The preconsolidation pressure can vary rapidly with depth just below the dry crust and the penetration test results indicate that the minimum strength value may be found about half a metre further down. Nevertheless, the results from the oedometer tests indicate a certain overconsolidation in the soil. The evaluated moduli from the oedometer tests show that the minimum value of the oedometer modulus would be around 2.7 MPa.

Pressuremeter tests

An attempt was then made to obtain better values of the compressibility of the soil, which would also be more representative for a larger volume of the heterogeneous mass, by using pressuremeter tests. The pressuremeter tests were performed with Menard type equipment and the pre-drilled holes were made by use of a so-called

"bentonite screw". The bentonite screw is a screw auger with a hollow stem and hollow rods through which a bentonite suspension is pumped down when the screw

SGI Report No 54

36

LARSSON ROLF

STATENS GEOTEKNISKA INSTITUT SWEDISH GEOTECHNICAL INSTITUTE

Investigations and load tests in silty soils

Results from a series of investigations in silty soils in Sweden

~ STATENS GEOTEKNISKA INSTITUT

SGI SWEDISH GEOTECHNICAL INSTITUTE

v

Rapport

No54

Report

Investigations and load tests in silty soils

Results from a series of investigations in silty soils in Sweden

Preface

.l

Readers guide to this report

Contents

Summary

=

Chapter I.

Introduction

Chapter 2.

Investigations

I

Chapter 3.

Principle of the plate loading tests

r-r,ie

.l

Chapter 4.

Investigations and load tests at the three main test locations

" C

' '

.,.,

c.

~-

L

rr

r

" _C

^' '