Svensk Djupstabilisering Swedish Deep Stabilization Research Centre
Report 16 Strength of stabilised soils – A laboratory study on clays and organic soils stabilised with different types of binder
Doctoral Thesis Helen Åhnberg Svensk Djupstabilisering
Svensk Djupstabilisering (SD) är ett centrum för forskning och utveckling inom djupstabilisering med kalk-cementpelare. Verksamheten syftar till att initiera och bedriva en branschsamordnad forsknings- och utvecklingsverksamhet, som ger säker- hetsmässiga, funktionsmässiga och ekonomiska vinster som tillgodoser svenska intressen hos samhället och industrin. Verksamheten baseras på en FoU-plan för åren 1996 – 2004. Medlemmar är myndigheter, kalk- och cementleverantörer, entreprenö- rer, konsulter, forskningsinstitut och högskolor.
Verksamheten finansieras av medlemmarna samt genom anslag från Byggforsknings- rådet/Formas, Svenska byggbranschens utvecklingsfond och Kommunikationsforsk- ningsberedningen.
Svensk Djupstabilisering har sitt säte vid Statens geotekniska institut (SGI) och leds av en styrgrupp med representanter för medlemmarna.
Ytterligare upplysningar om verksamheten lämnas av SD:s projektledare Göran Holm, tel: 013–20 18 61, 070–521 09 39, fax: 013–20 19 14, e-post: [email protected], internet: www.swedgeo.se.
Swedish Deep Stabilization Research Centre
The Swedish Deep Stabilization Research Centre coordinates research and development activities in deep stabilization of soft soils with lime-cement columns. A joint research programme based on the needs stated by the authorities and the industry is being conducted during the period 1996 – 2004. Members of the Centre include authorities, lime and cement manufacturers, contractors, consultants, research institutes and universities.
The work of the Swedish Deep Stabilization Research Centre is financed by its members and by research grants.
The Swedish Deep Stabilization Research Centre is located at the Swedish Geotechnical Institute and has a Steering Committee with representatives choosen from among its members.
Further information on the Swedish Deep Stabilization Research Centre can be obtained from the Project Manager, Mr G Holm, tel: +46 13 20 18 61, +46 70 521 09 39, fax: +46 13 20 19 14, e-mail: [email protected], internet: www.swedgeo.se. Department of Construction Sciences Structural Mechanics
ISRN LUTVDG/TVSM--06/1020--SE (1-194) ISBN 978-91-628-6790-4 ISSN 0281-6679
STRENGTH OF STABILISED SOILS A Laboratory Study on Clays and Organic Soils Stabilised with Different Types of Binder
Doctoral Thesis by HELEN ÅHNBERG
Copyright © Helen Åhnberg, 2006. Printed by KFS i Lund AB, Lund, Sweden, April 2006.
For information, address: Department of Construction Sciences, LTH, Lund University, Box 118, SE-221 00 Lund, Sweden.
Preface
The work presented in this thesis deals with the strength properties of soft soils stabilised with different binders. The study was initiated by the Swedish Deep Stabilization Research Centre (SD). The work has been carried out partly within the SD research programme and partly within other research projects at the Swedish Geotechnical Institute (SGI). My supervisor was Professor Jan Hartlén. I would like to express my gratitude for his enthusiastic attitude and advice throughout my studies. I’m also grateful to my co-supervisor, Per-Evert Bengtsson, SGI, for early on appreciating my intentions in the studies and for supporting me in carrying them out.
Financial support was provided by the SGI and the SD. Financial contributions have also been provided by Cementa AB and Nordkalk AB. These sources of funding, as well as permission of the participants within the EuroSoilStab project under the Brite EuRam contract BRPR-CT97-0351, to present data from that project, are gratefully acknowledged.
A number of people have supported me in various ways during the course of these studies, and I would like to express my special thanks to them. • Ola Antehag, SGI, for keeping excellent track of, and performing tests on, numerous samples of stabilised soils in the laboratory. I would also like to thank Inga-Maj Kaller, SGI, Lovisa Moritz, the Swedish Road Administration (formerly SGI), Christina Berglund, SGI, and Gunnar Westberg, SGI for carrying out parts of the tests. • Rolf Larsson, SGI, Miriam Smith, SGI, Bo Berggren, SGI, Torbjörn Edstam, SGI, Leif Eriksson, SGI, Martin Holmén, SGI, Bo Westerberg, SGI and Sven-Erik Johansson, Cementa AB, for critically reviewing parts of or the whole of the manuscript. • My superiors at SGI for their support and encouragement throughout this work. • My colleagues at SGI and LTH for friendly assistance and valuable discussions.
Finally, but not least, I would like to thank all the members of my family for their endless support and encouragement, also in my studies.
Lund, March 2006
Helen Åhnberg
Abstract
Stabilisation of soft soils by deep mixing with binders is the most frequently used method of ground improvement in Sweden today, and is increasingly being used internationally. The most common binders employed are cement and lime, but a variety of other binders may also be used for stabilisation of soils. For further development of the deep mixing method, there is a need for more extensive research on the undrained and drained strength properties and behaviour of soils stabilised with various types of binders. The overall objective of the study presented in this thesis was to improve the understanding of some of the important aspects of the strength behaviour of stabilised soils. For this purpose, a series of laboratory tests was performed on four soils stabilised with different types of binders. The soils included two types of clay and two types of organic soil. Cement, lime, blastfurnace slag and fly ash in different combinations, alone and together with a number of admixtures, were used as binders.
The laboratory testing was focused on the strength properties. However, a number of other basic properties of importance for the understanding of the strength behaviour of stabilised soils were also investigated, such as the density, water content, degree of saturation, permeability and the compression properties. Unconfined compression tests and triaxial tests were used to investigate the strength of the various samples stabilised in the laboratory. The unconfined compressive tests were performed mainly in order to study the effects of different binders on the increase in strength with time after stabilisation. The triaxial tests, conducted both as drained and undrained tests, were performed mainly to study the strength behaviour under various drainage and stress conditions. The influence of different testing procedures, such as different back pressure and rate of strain, was also studied. Both active and passive undrained triaxial tests were performed in order to investigate the strength anisotropy of the stabilised soils.
The results provided illustrative examples of the effects of different binders on the increase in strength of stabilised soils, which may be linked to the chemical
Strength of stabilised soft soil v Abstract reactions taking place after mixing. In addition to the composition of the soil and the binder, a number of other factors also affect the strength of stabilised soil. The results show that the drainage conditions and the stresses acting on stabilised soil may have a considerable influence on the measured strength. A stress dependence of the undrained strength as well as the drained strength is evident from the test results. A quasi-preconsolidation pressure, which is governed not only by the stresses applied to the stabilised soil, but also by the cementation taking place during curing, is shown to influence the strength behaviour. An improved strength model is proposed that describes the strength behaviour in the same stress plane commonly used for natural unstabilised soils.
It was concluded that although the type of binder may strongly affect the rate of strength increase and the final strength, the general strength behaviour is the same for soils stabilised by the most common binders. The strength and deformation properties of stabilised soils are similar to those of cemented and overconsolidated natural soils. The same set of parameters used to describe the strength of natural soils can also be used for stabilised soils.
vi Strength of stabilised soft soil Sammanfattning
Djupstabilisering av jord genom inblandning av bindemedel är den metod för jordförstärkning som oftast används vid byggande på och i lösa jordar i Sverige idag, framförallt i samband med väg- och järnvägsbyggande men även för förstärkning av t.ex. slänter och schakter och för grundläggning av ledningar och byggnader. Metoden, som utvecklades parallellt i Sverige och Japan under 70-talet, har under senare år kommit att tillämpas i ökad omfattning i ett allt större antal länder. De bindemedel som används för stabilisering är vanligtvis kalk och cement, var för sig eller blandade i olika proportioner, men även andra typer av bindemedel kan användas för stabilisering av jord. För vidareutveckling av metoden har det under senare år funnits ett behov av att öka kunskapen om hållfastheten hos stabiliserad jord vid användning av fler typer av bindemedel än de som idag vanligen används. Den utförda studien har syftat till att öka kunskapen om stabiliserad jords egenskaper, med inriktning på hur olika typer av bindemedel inverkar på hållfasthetstillväxten efter inblandning och hur stabiliserade jord uppträder ur hållfasthetssynpunkt vid olika spännings- och dräneringsförhållanden. Serier av laboratorieförsök har utförts på fyra jordar som stabiliserats med olika typer av bindemedel. De studerade jordarna är två oorganiska leror och två organiska jordar, varav en gyttja och en torv. De bindemedel som användes vid laboratorieförsöken var främst olika kombinationer av cement, kalk, masugnsslagg och flygaska. För vissa kombinationer av bindemedel användes också tillsatsmedel i form av aluminatcement, gips, silikastoft, kalciumklorid, rökgasstoft från kalktillverk- ning samt vattenglas.
Laboratorieundersökningarna har huvudsakligen fokuserat på undersökning av hållfasthetsegenskaper. För att öka förståelsen för stabiliserad jords beteende efter inblandning har också studerats ett antal andra egenskaper som har inverkan på hållfastheten. Dessa är densitet, vattenkvot, vattenmättnadsgrad, permeabilitet och kompressionsegenskaper. Undersökning av de stabiliserade jordarnas hållfasthet utfördes genom enaxliga tryckförsök och genom olika typer av triaxialförsök. De enaxliga tryckförsöken utnyttjades främst för att mäta hållfasthetstillväxten med tiden efter inblandning av olika typer av
Strength of stabilised soft soil vii Sammanfattning stabiliseringsmedel. Triaxialförsöken ger möjlighet till mer avancerad styrning av pålagda spänningar och mätning av respons hos provet under försöken. Dessa försök utnyttjades i första hand för att studera hållfasthetsegenskaper och beteende hos de stabiliserade jordarna vid olika spännings- och dräneringsförhållanden. Inverkan av metodik för triaxialförsök studerades inledningsvis genom användning av olika mottryck, dvs. applicerat portryck, för att vattenmätta proverna i olika grad, och olika deformationshastighet under försöken. Både aktiva och passiva triaxialförsök utfördes för att studera i vilken utsträckning hållfasthetsegenskaperna hos stabiliserad jord är anisotropa.
Resultaten av försöken visar hur olika typer av bindemedel inverkar på hållfasthetstillväxten hos stabiliserad jord. Resultaten visar även på hur vattenkvot, vattenmättnadsgrad, densitet och permeabilitet hos jord förändras efter inblandning av stabiliseringsmedel. Ökningen i hållfasthet med tiden kan kopplas till mängd och typ av reaktionsprodukter som bildas efter inblandning av bindemedel i jorden. Hållfastheten hos stabiliserad jord påverkas dock inte bara av typ av jord och bindemedel utan också av ett antal andra faktorer. Resultaten av studien visar att bl. a. rådande spännings- och dräneringsförhållanden före och under belastning har inverkan på uppmätt hållfasthet. Ett spänningsberoende kan påvisas för hållfastheten hos stabiliserad jord, under såväl odränerade som dränerade förhållanden. Vidare visar resultaten på inverkan på hållfasthetsegenskaperna av ett kvasi- förkonsolideringstryck, vilket är en funktion av både tidigare rådande effektivspänningar och den cementering som sker i den stabiliserade jorden. Detta kvasi-förkonsolideringstryck kan vara olika stort i olika riktningar beroende på vilka spänningsförhållanden som råder under härdning. En materialmodell föreslås vilken beskriver den stabiliserade jordens hållfasthetsbeteende baserat på inverkan av kvasi-förkonsolideringstrycket, vilket har likheter också med beteendet hos naturlig, ostabiliserad jord.
En övergripande slutsats är att typen av bindemedel kan ha stor inverkan på den hållfasthetstillväxt som sker efter inblandning. Typen av bindemedel har däremot normalt mindre betydelse för den stabiliserade jordens allmänna egenskaper. Spännings-töjningssamband styrs främst av den hållfasthetsnivå som uppnåtts snarare än av typen av medel som blandats in. Beteendet hos stabiliserad jord liknar i mycket det hos cementerade och överkonsoliderade naturliga jordar. Samma typ av parametrar som används för beskrivning av hållfasthet hos naturlig jord kan därmed användas också för stabiliserad jord.
viii Strength of stabilised soft soil Papers
This thesis is based on the following papers, which will be referred to in the text by their Roman numerals. The papers are appended at the end of the thesis.
I. Åhnberg H.., Johansson S.-E., Pihl H. and Carlsson T., 2003. “Stabilising effects of different binders in some Swedish soils”. Ground Improvement, Vol. 7, No. 1, pp. 9-23.
II. Åhnberg H. and Johansson S.-E., 2005. “Increase in strength with time in soils stabilised with different types of binder in relation to the type and amount of reaction products”. Proceedings International Conference on Deep Mixing, Stockholm 2005, Vol. 1.1, pp. 195-202.
III. Åhnberg H., Bengtsson P.-E. and Holm G., 2001. “Effect of initial loading on the strength of stabilised peat”. Ground Improvement, Vol. 5, No. 1, pp. 35-40.
IV. Åhnberg H., 2003. “Measured permeabilities in stabilised Swedish soils”. Proceedings 3rd International Conference on Grouting and Ground Treatment, New Orleans 2003, pp. 622-633.
V. Åhnberg H., 2004. “Effects of back pressure and strain rate used in triaxial testing of stabilised organic soils and clays”. Geotechnical Testing Journal, Vol. 27, No. 3, pp. 250-259.
VI. Åhnberg H., 2006. “Consolidation stress effects on the strength of stabilised Swedish soils”. Ground Improvement, Vol. 10, No. 1, pp. 1-13.
VII. Åhnberg H., 2006. “On yield stresses and the influence of curing stresses on stress paths and strength measured in triaxial testing of stabilised soils”. Submitted for publication in the Canadian Geotechnical Journal.
Strength of stabilised soft soil ix Papers
Division of work between authors
The author of this thesis performed the studies and wrote, in full, the papers I, II and III with the valuable support of the co-authors.
x Strength of stabilised soft soil Symbols and abbreviations
a constant, expressing undrained shear strength aq constant, expressing undrained strength a non-evaporable water content of hydration product b exponent
Cq undrained strength intercept c´ effective cohesion intercept cu undrained shear strength
IL liquidity index
IP plasticity index k permeability, hydraulic conductivity, coefficient of permeability ksoil permeability of natural soil kstab permeability of stabilised soil k initial permeability of stabilised soil stab i
K0 coefficient of earth pressure at rest
K0nc coefficient of earth pressure for normally consolidated soil
K0qp ratio of horizontal to vertical effective yield stresses of stabilised soil OCR overconsolidation ratio q deviator stress q28 unconfined compressive strength after 28 days qdr drained compressive strength qfailure deviator stress at failure qmax maximum deviator stress qt compressive strength after t days qu undrained compressive strength qu(σ´3c=0) undrained compressive strength at zero effective cell pressure
Strength of stabilised soft soil xi Symbols and abbreviations
quc unconfined compressive strength qyield deviator stress at yield in drained tests
Sr degree of saturation s´ mean effective stress in stress plane t shear stress in stress plane t time u pore water pressure w0 water content of unstabilised soil before mixing wL liquid limit wN natural water content wstab water content of stabilised soil wP plastic limit
α undrained strength increase factor ε1 major principal strain φ´ effective friction angle ρ soil density of natural soil σ ´1 major principal effective stress σ ´1c major principal effective consolidation stress σ ´3 minor principal effective stress σ ´3c minor principal effective consolidation stress σ ´c preconsolidation pressure; effective consolidation stress σ ´qp quasi-preconsolidation pressure σ ´qp0 quasi-preconsolidation pressure with no external stress during curing σ ´v vertical effective stress σ ´vc vertical effective consolidation stress τ fu undrained shear strength τ fd drained shear strength
A Al2O3 (notation used in cement chemistry) ASTM American Society for Testing and Materials BET Brunauer-Emmett-Teller, method for determination of specific surface areas
xii Strength of stabilised soft soil Symbols and abbreviations b.p. back pressure C CaO (notation used in cement chemistry) c cement CAD Anisotropically Consolidated Drained test CAU Anisotropically Consolidated Undrained test CEM II/A-LL Portland-limestone cement, 6-20% limestone CEN Comité Européen de Normalisation, European Committee for Standardization CL calcium lime CRS constant rate of strain EC European Commission ETC European Technical Committee
F Fe2O3 (notation used in cement chemistry) f fly ash
H H2O (notation used in cement chemistry) H1-H10 degree of humification, von Post classification system l lime Q quicklime R high early strength (rapid), cement strength class
S SiO2 (notation used in cement chemistry) s slag s.p. stress path SD Swedish Deep Stabilization Research Centre SGI Swedish Geotechnical Institute
Strength of stabilised soft soil xiii Symbols and abbreviations
xiv Strength of stabilised soft soil Contents
1 INTRODUCTION...... 11 1.1 BACKGROUND...... 1 1.2 OBJECTIVES...... 3 1.3 OUTLINE OF THIS THESIS...... 4
2 SCOPE OF THE STUDY...... 55 2.1 GENERAL APPROACH...... 5 2.2 EXTENT OF THE STUDIES ...... 6 2.3 MATERIALS AND TESTING METHODS...... 6 2.3.1 Types of soil ...... 6 2.3.2 Binders and other additions...... 9 2.3.3 Testing methods...... 10
3 OUTLINE OF THE PAPEPAPERSRS ...... 1313 3.1 GENERAL...... 13 3.2 BRIEF SUMMARIES OF THE PAPERS ...... 13
4 EFFECTS OF BINDERS ON SOIL PROPERTIES - FIELD OF RESEARCH...... 1717 4.1 GENERAL EFFECTS ...... 17 4.1.1 Binder-soil reactions ...... 17 4.1.2 Changes in basic geotechnical properties ...... 21 4.1.3 Changes in permeability ...... 24 4.1.4 Changes in compressibility and strength ...... 26 4.2 INCREASE IN STRENGTH...... 29 4.2.1 Effects of different binders on the strength increase...... 30 4.2.2 Influence of initial loading...... 36 4.3 UNDRAINED AND DRAINED STRENGTH BEHAVIOUR ...... 39
Strength of stabilised soft soil Contents
4.3.1 Influence of testing methodology ...... 39 4.3.2 Stress dependency...... 44 4.3.3 Adaptation of a strength model – stress paths and influence of stresses during curing ...... 51
5 COCCONONNCNCLCLLUUSSIOIONNSSAS ANDAANNDD FUFURTHFURTHERTHERERRRE RRERESEESSEASEEARARCRCHRCCHH...... 61 BOOKMARK NOT DEFINED. 5.1 CONCLUSIONS ...... 61 5.2 FURTHER RESEARCH...... 65
REFERENCES ...... 6767 REFERENCES IN SUMMARY ...... 67 REFERENCES EXCLUSIVELY IN THE PAPERS...... 76
PAPERS II---VIIVII
StreStrengthngth of stabilised soft soil 1 Introduction
1.1 BACKGROUND
Improvement of soft soil by deep mixing, a method originally developed in Sweden and Japan more than thirty years ago, is becoming well established in an increasing number of countries. Today, it is the most commonly used method for soil improvement in Sweden. To improve the strength and stiffness of the soil, binders of various types are mixed into the soil forming columns of stabilised soil down to depths of, typically in Sweden, ten to twenty metres. In highly organic soils, whole blocks of soil may be stabilised down to depths of typically three to five metres. In rare cases, such as in large offshore applications in Japan, deep mixing has been performed down to depths of about seventy metres (Kitazume & Terashi, 2002). About two million metres of columns of stabilised soil are produced annually (2003) in Sweden. The diameters of the columns typically range from 0.5 to 0.8 metres.
In the mid 1970′s when deep mixing was first used in practice in Sweden, only lime in the form of quicklime was used, whereas today, a mixture of lime and cement is the clearly dominating binder. Figure 1.1 shows the different types of binder used for deep mixing in Sweden. Other binders are also used, although on a small scale. In Figure 1.1, “other binders” mainly includes slag in combination with cement, primarily for the stabilisation of organic soils. However, the use of other binders is likely to increase in the years to come. Other types of binder are increasingly being used internationally, primarily for shallow stabilisation of capping layers and sub-bases, but also for deep soil stabilisation. Besides slag, other industrial by-products, such as different types of ash, may be of interest. Apart from possible environmental benefits of using industrial by-products, there may be economic as well as technical reasons for incorporating a larger number of binders.
Strength of stabilised soft soil 1 CHAPTER 1
100%
80%
60% Lime Lime- cement
40% Other Percent of total use 20% binders Cement 0% 1980 1990 2000 Year
Figure 1.1 Binders used for deep mixing in Sweden.
An understanding of the different properties and behaviour of stabilised soil is of vital importance for the design of deep mixing in soft soils. The existing design models for deep mixing used in Sweden (Carlsten, 2000) are largely the same as those originally developed for the “lime-column method” (Broms & Boman, 1977; Broms, 1984), the original designation still commonly used for the method in the Nordic countries. These models are primarily based on the assumption of undrained conditions in the columns and thus mainly undrained strength parameters are employed during design. Furthermore, assumptions regarding the properties of the stabilised soil have largely been based on empirical knowledge gained from early experience with lime stabilisation of soft clay.
In recent years, there has been a need to adapt the design methods to take into account the effective stress state during loading, as well as the use of a larger variety of binders in the stabilisation of different types of soil. To further develop the design methods in these respects, more detailed investigations of the properties of stabilised soil are needed.
2 Strength of stabilised soft soil Introduction
The strength of the stabilised soil is an important property. Although a number of investigations have been performed regarding different aspects of the strength of stabilised soils (e.g. Terashi et al., 1979; Kawasaki et al., 1984; Kivelö, 1998; Porbaha et al., 2000; Åhnberg et al., 1995), there was a need for further studies, particularly concerning the effects of different types of binder on undrained as well as drained strength. However, not only the type of soil and binder but also the mixing procedure (e.g. Larsson, 2003; Yoshizava et al., 1996) and the curing conditions (e.g. stress conditions, moisture content and temperature) affect the strength of stabilised soil (e.g. Babasaki et al., 1996; Åhnberg, 1996). Furthermore, the strength properties of stabilised soil may change considerably with time, mainly due to different chemical reactions taking place, but also external influences, such as the effect of an embankment or foundation loading, or changes in the surrounding soil and ground water conditions. Studies of the influence of these factors are called for in order to improve the models describing the strength behaviour of stabilised soil as a basis for safer and more cost-effective designs of soil improvement by deep mixing.
1.2 OBJECTIVES
The overall objective of the research presented in the thesis was to study the strength properties of stabilised soil as a function of soil conditions and type of binder.
• The general strength behaviour and similarities/differences in strength properties were to be investigated in the laboratory and further clarified for soft soils stabilised with various binders used for deep mixing. The possible influence of the stress conditions during curing was also addressed in this respect.
• The investigations were also intended to include the evaluation of other properties of importance for an understanding of the strength behaviour, such as water content, density, degree of saturation, compressibility and permeability.
Strength of stabilised soft soil 3 CHAPTER 1
1.3 OUTLINE OF THIS THESIS
This thesis is presented in the form of seven previously published papers by the author, with a synthesising summary. After the introductory chapter, the scope of the research is presented in Chapter 2, giving a description of the approach used for studying the strength behaviour of stabilised soils, the extent of the study, and the materials and methods used in the research. In Chapter 3, brief accounts of the seven papers are presented, giving an overview and introduction to the discussions in the subsequent chapter on the subjects focused on in each paper. The field of research is summarised and some of the more important results obtained in the study are presented and discussed in Chapter 4. This chapter is divided into two parts. The first, introductory, part describes chemical reactions and changes in the properties of the soil after stabilisation that are of relevance for the measured strength behaviour. The second part focuses in more detail on the strength achieved in different soils, and the strength behaviour of stabilised soils under different curing and testing conditions. Conclusions and suggestions for further research are presented in Chapter 5.
The information gathered during the literature survey that was performed as part of this research is not presented in a separate section, but integrated with related results and discussed where relevant.
Papers I-VII are presented, in full, in the last part of the thesis.
4 Strength of stabilised soft soil 2 Scope of the study
2.1 GENERAL APPROACH
Mixing various binders into a soil will bring about significant changes in most of the soil properties. The strength properties of stabilised soil are affected by several different factors. The factors regarded as being important in this research were the type and quantity of binder, the type of soil and the curing conditions, with focus on the stress conditions.
A hypothesis in studying the strength behaviour was that the stabilised soils would exhibit strength and deformation properties similar to cemented and overconsolidated natural soils, making it possible to describe the strength and deformation properties with the same set of parameters as those normally used for unstabilised, natural soils. A further hypothesis was that the general behaviour of the stabilised soil would be related primarily to the attained strength level, as opposed to the type of binder.
The investigations comprised laboratory testing of soft soils stabilised with a number of binders and admixtures. Complementary data from other investigations presented in the literature were also used in the analysis of stabilised soil behaviour. Assessments of adequate methods of strength testing were included in the research. The tests were mostly performed within a series of projects entitled “Stabilised soil properties” organised by the Swedish Deep Stabilization Research Centre (SD) (SGI, 1997). Tests performed in two other projects at the Swedish Geotechnical Institute (SGI) concerning the properties of stabilised organic soil were also used to evaluate strength properties: the EC-funded (Brite EuRam) project “EuroSoilStab” (e.g. Åhnberg & Holm, 1999; Hebib & Farrell, 1999; den Haan, 1998; Lahtinen et al., 1999; Cortellazzo & Cola, 1999), and the internal SGI project, “Triaxial testing procedures for stabilised organic soils” (SGI, 1999).
Strength of stabilised soft soil 5 CHAPTER 2
2.2 EXTENT OF THE STUDIES
The tests performed were all laboratory tests. Comparisons made with the field behaviour of stabilised soils were based on experience gained by the author in earlier projects, or on results published in the literature. The laboratory tests performed on stabilised soil samples were all physical tests, i.e. chemical testing of reaction products was not included.
The soils used in the tests were all soft Swedish soils, i.e. clays and organic soils, gyttja and peat. Stabilised silt or coarser materials, such as sand, till or gravel, were not included in the testing programme. Although stabilisation of these soils may be of interest in other contexts, such as in the improvement of sub-bases (e.g. Sherwood, 1993; Lindh, 2004), it is of less relevance for deep mixing applications.
The binders and admixtures used in the tests were all readily available from industry. The binders were used either separately or blended in various combinations. No further chemical refinement of the materials or “odd laboratory brews” were developed for the tests. In all, about 2000 samples of stabilised soil were prepared for testing in the laboratory.
2.3 MATERIALS AND TESTING METHODS
The materials and testing methods employed in the different parts of the research are described in detail in the individual papers. An overview of the testing programme is presented here to facilitate comparisons of the tests referred to in the separate papers, and to provide a background to the test results presented in subsequent chapters.
2.3.1 Types of soil
Four types of soil were included in the laboratory tests. The soils were chosen to be of different character and included two types of soft clay, one of gyttja, and one of peat. The geographical origins of the clays and the organic soils are shown in Figure 2.1. The clays have fairly similar geotechnical properties, but are from different geological deposits. According to earlier experience from
6 Strength of stabilised soft soil Scope of the study these areas, it was expected that their strength increase would be affected differently when mixed with various binders. One of the clays is a post-glacial clay from an area near Linköping in the eastern part of Sweden. It was deposited in alternating lake and brackish sea water. The other is a marine post-glacial clay containing occasional shells, from a site near Löftabro on the west coast of Sweden. Both clays are high-plastic, with liquid limits of approximately 70%. The dominating clay mineral in both clays is illite, which is also the most common clay mineral in Sweden. The gyttja used in the study was from Holma Mosse, in the eastern part of Sweden, and the peat from Dömle Mosse in the province of Värmland, in the western part of Sweden. The Holma gyttja is a clayey gyttja with an organic content of 10%. The peat has a very high water content and is classified as H5-H7 according to the von Post scale (von Post, 1924, e.g. Landva & Pheeney, 1980; Hartlén & Wolski, 1996). Results from laboratory characterisation of the test soils are presented in Table 2.1 and Table 2.2.
Dömle Holma peat gyttja Linköping Löftabro clay clay
100 km
Figure 2.1 Locations of the test soils.
Strength of stabilised soft soil 7 CHAPTER 2
Table 2.1 Properties of the four soils.
LinköpingLinköping Löftabro Holma Dömle clay clay gyttja peat Depth (m) 3-6 2-5 3-5 2-5 Bulk density (t/m3) 1.55 1.52 1.23 0.951 Specific gravity (t/m3) 2.72 2.73 2.36 1.44 WWaterater content (%) 78 89 220 2000 Plastic limit (%) 24 23 64 480 Liquid limit2 (%) 70 66 170 1300 Undrained sheshearar strength2 (kPa) 15 8 5 … Sensitivity 20 25 10 … Organic content (%) 1.0 1.0 10.3 97 Clay content (%))637280… Clay mineralogy Illitic Illitic Illitic … Chloride content (%) 0.01 0.38 <0.01 0.02 Sulphide content (%) 0.05 0.18 0.2 … Carbonate contentcontent ( %) 1.1/1.03 8.4/<0.53 0.6/…3 … pH 7.7 8.7 8.5 4.3 1 The sample in the laboratory was not fully saturated. 2 Determined by the fall cone test (ETC5, 1998). 3 CaCO3/MgCO3
Table 2.2 Chemical composition and specific surface of the clays and the gyttja.
LinköpingLinköping Löftabro Holma clay clay gyttja CaO ((total)total) ( %) 1.75 4.62 1.47
SiO2 (%) 66.5 59.2 56.9
Al2O3 (%) 16.1 16.2 14.1
Fe2O3 (%) 7.23 7.27 6.56 MgO ( %) 2.48 2.79 2.59
K2OO((%) 4.34 4.54 3.68
Na2O (%) 1.82 2.17 1.64 1 SO3 (%) 0.23 1.17 1.80 Specific surface2 (m2/kg) 16 200 15 600 23 100 1 Total sulphur expressed as SO3. 2 Determined by the BET method in accordance with Brunauer et al. (1938) (e.g. Taylor 1997).
8 Strength of stabilised soft soil Scope of the study
The two clays were used for tests performed in all seven parts of the research, as described in the different papers. Tests performed on stabilised gyttja are described in four papers, Papers I, II, IV and V, and tests on stabilised peat in two papers, Papers III and V. References are also made to the organic soils in the other papers, regarding results from similar tests in earlier studies.
2.3.2 Binders and other additions
The type and number of binders used for stabilising the selected soils were varied in the studies. In total, five types of cement, five types of lime, three types of ash and one type each of slag, silica fume, gypsum, kiln dust, water glass and salt were used in the laboratory test series. Only dry binders were used. A broad selection of binders and admixtures was used in screening tests, described in Paper I. The effects of most of these binders were also studied regarding the permeability of the stabilised samples, which is discussed in Paper IV. The bulk of the research, however, focused on the fundamental strength behaviour, and the effects of stress conditions on soil stabilised with different combinations and quantities of mainly one type of cement, lime, slag and fly ash. The types of cement and lime used were those most commonly used for soil stabilisation in Sweden; a Portland-limestone cement, designated CEM II/A-LL 42.5 R (CEN Standard, 2000) and a quicklime, CL90-Q (CEN Standard, 2001). The slag was a ground, granulated blastfurnace slag and the fly ash was a pozzolanic ash, Class F (ASTM, 1994), from pulverised coal combustion residues. The composition of the four main binders used is given in Table 2.3.
The quantities of binders used for stabilisation varied in the different parts of the research and for the different soils. For the clays, the binder quantity used corresponded to 50 to 200 kg/m3 of soil. For the gyttja and the peat, the binder quantities ranged from 70 to 150 kg/m3 and from 100 to 300 kg/m3, respectively. The binder quantity most frequently used in the clays and gyttja was 100 kg/m3 and in the peat 200 kg/m3.
Strength of stabilised soft soil 9 CHAPTER 2
Table 2.3 Composition and specific surface of the main binders.
2 3 CaO SiO2 Al2O3 Fe2O3 MgOMgO K2O Na2O SO3 Spec. surface (%) (%) (%) (%) (%) (%) (%) (%) (m2/kg) Lime 93.01 1.4 0.6 0.3 1.0 <0.1 <0.1 <0.1 660 Cement 61.4 19.9 3.6 2.6 2.8 1.0 0.2 3.3 470 Slag 32.1 35.2 13.6 0.2 16.8 0.6 0.6 1.8 480 Fly ash 5.9 54.4 30.5 5.5 1.8 1.2 0.5 0.5 470 1 Available CaO 92%. 2 Total sulphur expressed as SO3. 3 Determined by the Blaine method (CEN Standard, 1989).
2.3.3 Testing methods
Prior to stabilisation, the natural soils were characterised with respect to their physical and chemical properties, see Table 2.1. In preparing the laboratory samples, the soil was first thoroughly homogenised, and then the dry binders were mixed into the soil for five minutes. In the standard procedure, the stabilised soil mixtures were filled into plastic tubes, which were then sealed and stored for various periods of time. The stabilised clay and gyttja samples were stored in climate-controlled rooms at 7 °C. The stabilised peat samples were placed into plastic tubes and stored in water-filled storage boxes at room temperature, enabling absorption of water during the curing process. A small load, corresponding to that normally used in the field when stabilising peat, was applied to the majority of the stabilised peat samples. The laboratory samples were prepared according to common procedures for stabilising soil in Sweden (Carlsten & Ekström 1995; Carlsten, 2000; EuroSoilStab, 2002). In order to study the influence of curing stress on the strength, as described in Paper VII, a number of stabilised clay samples were stored in special cells or in storage boxes, with access to water, at room temperature.
The curing time varied from one day to slightly more than two years. The majority of laboratory tests were performed on the samples 28 days after mixing.
Laboratory testing was focused on the increase in strength and the strength behaviour of stabilised soil. The strength of the stabilised samples was measured by unconfined compression tests and/or triaxial tests. The
10 Strength of stabilised soft soil Scope of the study unconfined compression tests, which are simple and relatively quick tests (Swedish Standard, 1992a), were used to compare the effects of different binder compositions and other factors on the strength increase of the stabilised soils. The triaxial tests, which are somewhat more complex and time consuming, were used for more detailed studies of the strength behaviour by measurements of changes in stresses and pore pressure, as well as strain, during loading. To gain a better understanding of the effects of binders on the behaviour of the stabilised soil, a number of other tests were also performed. These included the determination of basic parameters such as the density (Swedish Standards, 1989a; 1989b), water content and consistency limits (Swedish Standards, 1989c; 1990a; 1990b) of the stabilised soils. A number of permeability tests were performed in flexible-wall permeameters, mainly in accordance with the Nordtest testing procedures (Sjögren et al., 1994). Furthermore, the compressibility was investigated in a series of oedometer tests performed both as CRS oedometer tests and as incrementally loaded tests in accordance with Swedish Standards (1991 and 1992b, respectively). In the CRS oedometer tests, also the permeability of the stabilised samples was measured. To obtain comparable reference data and to illustrate the effect of stabilisation on the strength and deformation behaviour, triaxial tests and oedometer tests were also performed on unstabilised, natural soil.
Strength of stabilised soft soil 11 CHAPTER 2
12 Strength of stabilised soft soil 3 Outline of the papers
3.1 GENERAL
The papers in this thesis are based on the results of several investigations performed by the author. The papers are not appended in strict chronological order but rather in an order of related subjects, from more general effects of binders on the strength increase and other basic properties of stabilised soil, as described in Papers I-IV, to a more detailed description of the strength behaviour in Papers V-VII.
3.2 BRIEF SUMMARIES OF THE PAPERS
Paper I: Stabilising effects of different binders in some Swedish soils The first paper, Paper I, addresses the effects of a wide range of binders and admixtures used in stabilising clay and gyttja. A series of common geotechnical laboratory tests, mainly unconfined compression tests and determinations of water content and density, were performed on samples of stabilised soils at different time intervals up to one year after mixing. The aim was not to identify the “best” binder, but to elucidate the differences and similarities in strength increase between various combinations of binder and soil. Whereas some binders were found to be robust, producing good results in most types of soil, others produced very good results in certain soils but poor results in others. Comparisons are also made with results from some previous investigations using different binders.
Strength of stabilised soft soil 13 CHAPTER 3
Paper II: Increase in strength with time in soils stabilised with different types of binder in relation to the type and amount of reaction products In the first paper, the focus was on the change in strength and other physical properties after stabilisation, whereas the chemical composition and typical reactions of the binders were discussed more briefly. In Paper II the chemical processes that take place after mixing the binders with the soils are discussed in greater detail. Since different compounds produced by the reactions of different binders in the soil are likely to influence the increase in strength to a significant degree, rough estimates of the type and amount of the reaction products were made for the most common binders, in order to gain a clearer understanding of the processes involved. In Paper II, the long-term strength increase measured in the main test series described in Paper I is discussed further, and the relation between strength and chemical reaction products is demonstrated.
Paper III: Effect of initial loading on the strength of stabilised peat Paper III focuses on the stabilisation of peat, highlighting the effects on the increase in strength resulting from preloading performed shortly after stabilisation. When stabilising peat in the laboratory, the procedure of mixing and preparing samples of stabilised soil is similar to that for the other types of soil. However, the curing conditions as a rule are different, mainly in that a load is applied on top of the stabilised peat samples shortly after mixing. This is in accordance with the normal procedure for stabilisation in the field, where a fill of half a metre to one metre is placed on top of the stabilised area as soon as possible after mixing in order to obtain a denser and more homogeneous stabilised mass. The strength increase was studied for a number of samples stored with and without a load during curing. The results show that initial loading greatly influences the strength of stabilised peat.
Paper IV: Measured permeabilities in stabilised Swedish soils Paper IV presents work performed to evaluate another significant property affecting the soil behaviour, namely the permeability of stabilised soil. The permeability is of importance not only when evaluating the consolidation processes, or for designing seepage barriers in stabilised soil, but it also affects the changes in the effective stress and the strength increase that occur with time after loading. A number of permeability tests were performed on stabilised soil samples in order to study the effects of different types of binder on the permeability and the extent to which the permeability in stabilised soils
14 Strength of stabilised soft soil Outline of the papers changes with time. Paper IV presents the results of the permeability tests performed as part of this work, and compares them with results obtained from investigations performed by others. Relations between measured permeability, measured strength and change in water content are proposed and discussed.
Paper V: Effects of back pressure and strain rate used in triaxial testing of stabilised organic soils and clays Paper V is the first of the three papers that present results of the triaxial tests on stabilised soils. In this paper, the influence of different methods of testing is discussed. Triaxial tests, which are useful tools for investigating the strength properties under undrained and drained conditions at different stress levels, should normally be run under conditions resembling those in the field as closely as possible. However, conditions in the field may change with time and in widely varying ways, making simulation test series laborious and expensive. In selecting a simpler test procedure, it is important to be aware of the effects of factors that are not simulated in the tests. A number of tests were performed on stabilised clay, gyttja and peat using low and high back pressures and different strain rates in order to study how these factors affect the results and the importance of these effects. The results showed that, in particular, the back pressure can greatly affect the measured strength of stabilised soils.
Paper VI: Consolidation stress effects on the strength of stabilised Swedish soils The effect of different confining stresses on the strength of stabilised soil is discussed in Paper VI. A series of triaxial compression tests and also oedometer tests was performed on samples of clay stabilised with the main binders. The results showed that both the drained and the undrained strength were dependent on the consolidation stresses during testing. However, both the prevailing stress and strength levels affected the extent of this stress dependence. Relationships for describing the variation in undrained strength as well as drained strength with increasing consolidation stress are proposed in Paper VI. The concept of quasi-preconsolidation pressure is proposed to improve the modelling of the strength of stabilised soils.
Strength of stabilised soft soil 15 CHAPTER 3
Paper VII: On yield stresses and influence of curing stresses on the stress paths and strength measured in triaxial testing of stabilised soils In Paper VI, the results from the triaxial tests were presented solely in the form of strengths and strains at failure. In Paper VII, the stress paths observed during loading are discussed, and the role of the quasi-preconsolidation pressure is further analysed. Series of triaxial tests and oedometer tests were performed on the stabilised clays cured with and without the application of external stresses. The results showed that both cementing processes and curing stresses influence the quasi-preconsolidation pressures and, in turn, the stress paths and strength of stabilised soil. The applicability of a strength model commonly used for natural soils is demonstrated and the influence of the degree of “overconsolidation” on the stress-strain behaviour is elucidated.
16 Strength of stabilised soft soil 4 Effects of binders on soil properties - Field of research
4.1 GENERAL EFFECTS
Mixing binders into a soil will affect the fundamental properties of the natural, unstabilised soil to varying extents. The strength and deformation properties are typically those of main interest when designing the extent of stabilisation required for ground improvement. However, in assessing strength and deformation parameters from various tests, it is also important to consider the changes in other properties and their possible influence on the behaviour of the stabilised soil. Such additional considerations include effects of changes in, for example, water content, degree of saturation and permeability. Furthermore, a basic understanding of the types of chemical reaction that take place and the compounds formed when using different binders is essential in analysing the rate and type of changes in properties that may develop. This also applies to the durability of stabilised soils. However, chemical durability aspects have not been included in this work.
The field of research is summarised in this chapter, including some of the more important results presented in the papers. A number of properties of the stabilised soil are addressed with focus on the strength properties.
4.1.1 Binder-soil reactions
The chemical reactions involved in the hydration of different types of cement or lime have been described and discussed thoroughly in many papers and
Strength of stabilised soft soil 17 CHAPTER 4 textbooks (e.g. Taylor, 1997; Boynton, 1980). The various chemical processes involved in soil stabilisation using a variety of binders have also been described in the literature, (e.g. Chew et al., 2004; Janz & Johansson, 2001; TRB, 1987; Saitoh et al., 1985; Ingles & Metcalf, 1972; Ruff & Ho, 1966), although the focus has mainly been on the two most common binders, cement and lime.
The reactions generated when mixing various binders with soil vary by process, intensity and duration, but in general, exhibit many similar characteristics. Figure 4.1 presents a rough outline of the chemical processes taking place and the main reaction products formed when mixing common binders such as quicklime, cement, slag and fly ash into a soil, as described in Paper II. The abbreviations most often used for various compounds in the hydration processes are given in the lower part of the figure. As the binder is mixed into the soil, hydration takes place, although slag may need an activator from another binder to start this process. Some reactions may involve cementation starting up directly, while others may lead to further reactions with the soil and its minerals.
The reaction products formed are of somewhat different types. When using quicklime, which contains large amounts of calcium oxide (denoted C), hydration will occur as the lime comes into contact with the pore water in the soil, resulting in the formation of calcium hydroxide (denoted CH). Some of this calcium hydroxide will be adsorbed onto the soil particles. Ion exchange will take place and the soil will be modified into a somewhat drier and coarser structure due to the slaking process and flocculation of the clay particles that take place (e.g. Boardman et al., 2001; Saitoh et al., 1985). The calcium hydroxide not consumed in this process is free to react with the silica (denoted S) and alumina (denoted A) contained in minerals present in the soil. These reactions, termed pozzolanic reactions, will result in the formation of calcium aluminate silicate hydroxide (CASH1), calcium silicate hydroxide (CSH1) and/or calcium aluminate hydroxide (CAH1) (e.g. TRB, 1987). When using cement, primarily CSH is produced, but also, although to a much lesser degree than for lime, pozzolanic reaction products, containing silica and alumina from the soil, are produced. Ground granulated blastfurnace slag, which is a latent hydraulic cement, will react in much the same way as ordinary cement, and lead to the formation of similar hydration products. The fly ash acts mainly as a pozzolanic material, reacting with the calcium hydroxide added to or generated by hydration.
1 The compounds CSH, CAH and CASH here denote compositions of C, S, A and H in non-specific proportions.
18 Strength of stabilised soft soil Effects of binders on soil properties - Field of researchresearch
C - CaO H - H2O
A - Al2O3 S - SiO2 F - Fe2O3 AAFFm/t - AF mono- and tri-phases (e.g.(e.g. ettringite)
Figure 4.1 Rough outline of the principal chemical reactions and reaction products formed by different types of binders in a soil. After Paper II.
In addition to the materials described above, there are a number of admixtures or mineral additions that may be used for soil stabilisation, most of which are commonly used for improving the hardening or setting properties of concrete. However, these products do not normally substantially change the principal types of reaction products and bonds formed (Taylor, 1997). An exception is the addition of gypsum, which may result in a significant formation of ettringite, which contributes to the strength by forming needle-shaped, “reinforcing”, compounds.
The various binders can be characterised with respect to possible type and rate of reactions by looking at their content of CaO, Al2O3 and SiO2. In general, the reactivity increases with total content of CaO + Al2O3+ SiO2 of the binders (Taylor, 1997). However, the effects of both major and minor components are complex. For example, the reactivity of MgO in slag is quantitatively equivalent to CaO up to a certain content (Taylor, 1997). In Figure 4.2, typical relative proportions of CaO, Al2O3 and SiO2 of various binders and other possible additions, often used with concrete, are shown in a three-phase
Strength of stabilised soft soil 19 CHAPTER 4 diagram, together with the compositions of the materials used in this study. Lime, which contains a high proportion of CaO, has a high potential for forming large amounts of reaction products when mixed with soil. However, pozzolanic reactions with soil are normally relatively slow, due to the restricted accessibility of the silica and alumina contained in the soil. In fly ash, the silica and alumina are more easily accessible for reactions with binders that contain calcium hydroxide. The reactions forming CSH upon hydration of cement involve minerals contained in the binder itself and are thus, as a rule, more rapid than pozzolanic reactions with the soil. In slag, the ratio of CaO to SiO2 is significantly lower than in cement and, as a result, the build-up of reaction products is normally slower than for cement (Taylor, 1997). In calcium aluminate cements, the high content of alumina will cause much faster reactions than those of Portland cements.
Figure 4.2 Relative percentages of CaO, SiO2 and Al2O3 in the materials used in this study (see legend) and typical ranges of similar materials of interest in concrete production.
20 Strength of stabilised soft soil Effects of binders on soil properties - Field of researchresearch
The reactions taking place during hydration generate heat. In general, the slaking of quicklime will produce the largest amount of heat. The total amount of heat generated by soil stabilisation with cement will normally be less than half of that from stabilisation with equivalent quantities of quicklime (e.g. Åhnberg et al., 1989). Composite binders of cement and lime can be expected to generate amounts of heat between those of lime and cement, while those of cement-slag and cement-fly ash may generate an amount of heat closer to that of pure cement (Pihl & Kuusipuro, 2004). An increase in temperature will generally increase the rate of the reactions that occur. In the field there will be a significant increase in temperature in the deep-mixed columns, which in turn will affect the rate of the chemical processes taking place (e.g. Åhnberg & Holm, 1987). In normal laboratory testing, on the other hand, the effects of heat generation are minor, since the samples are small and are stored at a constant temperature.
Besides the effects of admixtures and temperature on the rates of chemical reactions, these may also be affected by various substances in the soil having retarding or accelerating effects, some of which, but not all, are well known (e.g. Taylor, 1997; Kitazume & Terashi, 2002; Åhnberg & Pihl, 1998; Paper II).
4.1.2 Changes in basic geotechnical properties
A certain increase in the bulk density and decrease in water content can be expected when binders are mixed with soft soils. These changes normally lead to an increase in strength, as well as a decrease in compressibility and normally, with time, also a decrease in permeability. However, there is no unique relation between the water content or density and the latter properties in stabilised soils, as relationships vary with soil as well as with type of binder.
The slaking of quicklime will involve a certain degree of expansion upon hydration (e.g. Boynton, 1980), whereas the hydration of other binders, as a rule, only causes minor changes in volume (e.g. Taylor, 1997). In the field, the slaking of quicklime may increase the total horizontal stress somewhat and, depending on the soil conditions and geometry of the adjacent ground surface, cause displacement or consolidation of the surrounding soil. In the laboratory, when mixing binders with clay at normal quantities, the effects of the hydration process on the density are limited. The density is as a rule somewhat lower than the theoretical value, depending on the extent to which
Strength of stabilised soft soil 21 CHAPTER 4 inclusions of air pockets can be avoided in preparing the test samples. When using dry binders, the density can be expected to be roughly the same or slightly higher than that of the natural, unstabilised soil. However, density changes in peat during stabilisation may be significant. The addition of binder at quantities up to 300 kg/m3 in the highly organic soils studied here was found to increase the density by up to about 20% (Paper III). The initial loading normally applied on stabilised peat shortly after mixing will increase the density even further.
The decrease in water content typically observed in soils after stabilisation is caused by the introduction of dry solid particles into the soil, as well as by some water being bound by chemical reaction products during the hydration process. The water content is also affected to a small extent by the drying/evaporation of water during mixing. Most of the decrease in water content occurs during the first week after stabilisation, but a continued decrease is normally observed up to at least one month after mixing. The water content of the stabilised soil can roughly be calculated (Paper I) as
w ρ N − ax soil w +1 w = N (4.1) stab 1 ρ + (1+ a)x soil w + N 1
ρ 3 where soil is the bulk density of unstabilised soil (t/m ); wN is the natural water content of the unstabilised soil (in decimal number); x is the amount of dry binder added to the soil (t/m3); and a is the content of non-evaporable water of the hydration product with respect to dry binder weight (in decimal number). For the main binders cement, slag and lime used in this work, a is of the order of 0.2 to 0.3, whereas the fly ash has only marginal effects on the total amount of non-evaporable water.
The behaviour of the stabilised soil is affected not only by its water content, but also by its liquidity or consistency index. As in natural soils, a decrease in water content and thus liquidity index, is accompanied by an increase in strength (Paper I). Both the liquid limit and the plastic limit will also change to varying degrees, largely depending on the type of soil (e.g. Dumbleton, 1962; Brandl, 1981; Sivapullaiah et al., 2000). The plastic limit normally increases after stabilisation. The liquid limit of Swedish soft clays typically also
22 Strength of stabilised soft soil Effects of binders on soil properties - Field of researchresearch increases after stabilisation, but in some high-plastic and organic soils, it decreases (Paper I). Examples of changes in water content, w, liquid limit, wL, and plastic limit, wP measured in stabilised soils are shown in Figure 4.3.
Adding dry binders to a soil will result in a stabilised soil that is not fully saturated. This is the case for laboratory prepared samples, which are normally stored without access to water, and partly also for stabilised soil in the field. The degree of saturation in the field will change to varying degrees with time depending on the pore pressure conditions and the permeability and water absorption ability of the stabilised soil. Since incomplete saturation affects the pore pressure build-up, a decrease in the degree of saturation will lead to an increase in the undrained strength. This effect is discussed further in Section 4.3.1. The degree of saturation of laboratory samples may be of the order of approximately 96-98% in a peat of high water content when stabilised with a binder quantity of 200 kg/m3, and 93-95% in stabilised high-plastic clays when using a binder quantity of 100 kg/m3 (Paper V).
Figure 4.3 Examples of range of measured w, wL and wP in a clay and a gyttja before and after mixing with different binders. Binder quantity 100 kg/m3. After Paper I.
Strength of stabilised soft soil 23 CHAPTER 4
4.1.3 Changes in permeability
The permeability1 of the stabilised soil is often of interest for several reasons. It may affect the pore pressure response at loading and, depending on the rate of loading, influence the extent to which undrained or drained conditions govern the strength behaviour. Furthermore, the permeability will affect possible leakage of binder substances from the stabilised soil, as well as the risk of changes in groundwater conditions, e.g. due to lowering of artesian groundwater pressures. Other important effects, although not directly related to strength, are the influence on the rate of consolidation after construction and the potential impact from leaching of trace elements on the environment.
Earlier investigations have reported both increases (e.g. Bengtsson & Åhnberg, 1995; Baker, 2000) and decreases in permeability (e.g. Terashi & Tanaka, 1983; Kawasaki et al., 1984) or both depending on the type of binder used (e.g. Åhnberg et al., 1995). However, these diverging results need not be considered questionable. Depending on the effect of the binder and the time of curing, the macro-structure of the stabilised soil mass and stress conditions, the permeability may be higher or lower than that of the soil before stabilisation (Paper IV). The permeability may thus be very different under field conditions compared to that measured in the laboratory. In fairly homogeneous stabilised soils, i.e. samples prepared in the laboratory, the change in permeability can be described by an initial change, an increase or decrease, followed by a decrease with time.
Initial flocculation and other structural changes will cause a change in the permeability immediately after stabilisation. As a rule, a certain increase in the permeability can be measured, but the addition of large amounts of binders to highly organic soils, and any compaction performed, may cause an initial decrease in permeability. There will then be a decrease in permeability with time, influenced by the continuing formation of different reaction products in the stabilised soil. The rate and extent of this growth will depend on the type and amount of binder and the type of soil, and will also be influenced by any retarding substances present in the soil.
The initial change in permeability is linked to a change in void ratio, and can also be roughly related to the change in water content after mixing and
1 The term permeability is used here (denoted k, with the unit m/s), since it is the most commonly used term for this property in the field of soil mechanics. Another, more precise, term is the hydraulic conductivity or the coefficient of permeability.
24 Strength of stabilised soft soil Effects of binders on soil properties - Field of researchresearch possible compaction. After the initial change, the decrease in the permeability can be related to the increase in strength, which is an indirect measure of the growth of reaction products. The cementation process leads to about the same relative decrease in permeability with increase in strength. Measured differences in permeability when using different types of binders may thus be largely related to differences in the strength of the stabilised soils. Figure 4.4 shows relationships between initial changes in water content and permeability after mixing and further change in permeability with increasing strength as measured in laboratory tests. The figure summarises data from soils ranging from clay compacted at an optimum water content of only slightly higher than 20%, (Brandl, 1999) to stabilised peat having an initial water content around 2000%. Most of the peat samples were subjected to a vertical pressure of 18 kPa during curing, but some were cured with no load. The change in permeability can be roughly estimated from the change in water content and the strength (Paper IV) according to
w 6⋅ −0.004q k w uc stab ≈ 0.043⋅e 0 k (4.2) soil
100.00 100.00
Compacted clay (a) lime 1-7.5% (b) (Brandl 1999)
10.00 10.00
soil 1.00 1.00
/ k - 0,0039x soil 27,65e /k
stab. k i −0.0017q k stab ⋅10 uc stab, initial stab, 0.10 0.10 k k soil
6,0048x 0,043e
0.01 0.01
Stabilised peat initial load 18 kPa 0,0502e - 0,0045x 0.001 0.001 0.0 0.2 0.4 0.6 0.8 1.0 0 500 1000 1500 w/w0 quc, kPa
Figure 4.4. Changes in permeability after stabilisation. a) Initial change in permeability vs. change in water content. b) Change in permeability vs. strength. The shaded area represents results from investigations on various stabilised soils. After Paper IV.
Strength of stabilised soft soil 25 CHAPTER 4
where w and kstab are the water content and permeability of the stabilised soil and w0 and ksoil are those of the unstabilised soil, respectively, and quc is the unconfined compressive strength (kPa). If not measured, the change in water content can be estimated based on the type and amount of binder used, see Section 4.1.2. The results of the study indicate that the permeability decreases roughly by a factor of ten for about every 600 kPa increase in strength.
Various types of testing performed in columns in the field have shown large variations in the results (e.g. Nygren & Wellander, 1991; Bergwall & Falksund, 1996; Baker, 2000), but as a rule higher permeability is found in situ than in the laboratory. This is probably the effect of a more inhomogeneous macro-structure in the columns with a more uneven distribution of binder, and the occurrence of fissures or micro-cracks in the material. The permeability measured in the laboratory may be regarded as a lower limit for the stabilised soil.
4.1.4 Changes in compressibility and strength
Stabilisation of soil involves a decrease in compressibility and an increase in strength. In general, the stiffness of the stabilised soil will increase more than the strength. Figure 4.5 outlines the decrease in strain at failure with increasing unconfined compressive strength (Paper I). The relatively distinct change from large to limited, almost constant, failure strain, mirrors the transition from contractive to dilative behaviour as the strength of the stabilised soils increases. More ductile behaviour with larger strains at failure can be observed for stabilised peat compared with other soft, stabilised soils. This is believed to be caused by a reinforcing effect of the peat fibres.
The strain at failure may vary not only with the strength achieved, but also with confining stress (Paper VI, VII). Figure 4.6 shows schematically the variation in strain with normalised shear stress in undrained and drained triaxial tests. The shaded areas represent data measured for samples tested at about the same σ´ /σ´ ratio, where σ´ is the applied vertical consolidation σ vc qp vc stress and ´qp is the quasi-preconsolidation pressure, i.e. the vertical yield stress dependent not only on the previous consolidation stress but also on the cementation of the material. Under drained conditions, there is a significant increase in strain at failure with increasing confining stress. This behaviour may be attributed to the materials being consolidated to a state closer to the
26 Strength of stabilised soft soil Effects of binders on soil properties - Field of researchresearch
Figure 4.5. Changes in strain at failure with increase in unconfined compressive strength. After Paper I.
quasi-preconsolidation pressure. Effects of this quasi-preconsolidation pressure are discussed in more detail in Sections 4.3.2 and 4.3.3.
In undrained tests, there is no distinct increase in average failure strain with increasing confining stress. After an initial rapid increase in shear stress up to failure, the stress-strain curve levels off and only a limited change in shear stress is normally observed at continued deformation after failure. This almost ideal-plastic behaviour indicates that under undrained conditions, the full strength of the stabilised soil and the unstabilised soil in the field may be mobilised together. However, under drained conditions, highly “overconsolidated” stabilised soils, i.e. stabilised soils of high strength subjected to low confining stresses, will exhibit more brittle behaviour with a significant reduction in strength after failure at small strains. The full strengths
Strength of stabilised soft soil 27 CHAPTER 4
(a) (b)
Figure 4.6. Variation in normalised shear stress with axial strain in (a) undrained and (b) drained triaxial compression tests on stabilised Linköping and Löftabro clay. Results for unstabilised Linköping clay are shown for comparison. After Paper VII.
of the unstabilised soil and the stabilised soil can then not be expected to be mobilised at the same time.
In the oedometer case, where no horizontal displacement takes place, the compression modulus, M, will increase together with the strength. The quasi- preconsolidation pressure may be estimated roughly as a factor of about 1.3 times the unconfined compressive strength. The range for this factor has been reported to be between 1.2 and 1.9 (Terashi et al., 1980 in Kitazume & Terashi, 2002; Porbaha et al., 2000; Kwan & Buazza, 2002; Paper VI). This relation is to some extent affected by the overconsolidation ratio of the stabilised soil, i.e. the higher the overconsolidation the higher the factor. Various ways of estimating different compression parameters of stabilised soils have been suggested (e.g. Terashi et al., 1980 in Kitazume & Terashi, 2002; Baker, 2000; Lorenzo & Bergado, 2002; Alén et al., 2005), including the vertical yield stress as related to the void ratio (Tremblay et al., 2001; Rotta et al., 2003). Figure 4.7 shows a schematic variation of the compression modulus with increasing stress, expressed in accordance with the model normally used for calculations of settlements of natural soft soils in Sweden (e.g. Larsson et al., 1997). After passing a yield stress, the modulus reaches a minimum value. Thereafter, it increases with further increase in stress and, as an effect of the breakdown of the cementation forces with increasing stress level, is then
28 Strength of stabilised soft soil Effects of binders on soil properties - Field of researchresearch
' σ i high cementation effect Mi
1
M' i
M' M low min 1 cementation effect Compression modulus
uns tabilized s oil
Effective vertical stress
Figure 4.7 Schematic variation in compression modulus according to oedometer tests. The compression modulus presented is the tangent modulus. From Åhnberg (1996).
governed by a modulus number M′ of the same magnitude as that of the unstabilised soil (Åhnberg, 1996).
4.2 INCREASE IN STRENGTH
The increase in strength with time after stabilisation is governed by a number of factors. The type of binder will normally have a significant impact on the results, although the effect may vary considerably depending on the type of soil. Other factors affecting the increase in strength are the amount of binder, the mixing effort, the temperature and the stresses during curing (e.g. Babasaki et al., 1996; Åhnberg, 1996). Another factor of importance, especially in stabilising peat, is the effect of an initial loading of the stabilised soil.
Strength of stabilised soft soil 29 CHAPTER 4
4.2.1 Effects of different binders on the strength increase
The optimal binder to be used for stabilisation of a soil will vary depending on the desired strength in the short-term as well as the long-term perspective, for both drained and undrained conditions. Robustness and good durability are also important factors in cases of varying soil conditions or risks of the stabilised soil being subjected to aggressive ground water with high mobility. The effects of different binders on the strength increase are in such cases more appropriately described in terms of general strength levels and rate of strength increase, without defining optimal binders and exact strengths for all possible cases.
The effect of different types of binders on the strength increase of soft soils varies considerably (Papers I & II). Figure 4.8 shows examples of the increase in unconfined compressive strength, for the two clays stabilised with various binders (Paper II). During the first three months after mixing, the most rapid strength increase as a rule occurs in samples stabilised with binders containing cement. In general, after about three months, the strength of these samples levels off to a fairly constant value or shows clearly decreased rates of strength gain.
3000 3000 (a) (b)
, kPa sl
2500 , kPa 2500 uc uc q q
2000 2000
cl 1500 1500 l cs cs c c 1000 1000 cl
cf cf 500 500 sl Unconfined compressive strength Unconfined compressive strength l 0 0 0 200 400 600 800 0 200 400 600 800 Time, days Time, days
Figure 4.8 Examples of measured strength with time after mixing for (a) Löftabro clay and (b) Linköping clay with cement , lime and various composite binders (50:50). c = cement, l = lime, s = slag, f = fly ash. Binder quantity 100 kg/m3. After Paper II.
30 Strength of stabilised soft soil Effects of binders on soil properties - Field of researchresearch
Samples containing lime usually exhibit a pronounced long-term increase in strength, although this may vary considerably with the type of soil, as illustrated in Figure 4.8. Figure 4.9 shows the measured variation in unconfined compressive strength, for different categories of binders arranged in order of obtained 28-day strength. Cement alone or together with slag or fly ash, with a cement percentage of 50% or higher, gave the highest short- term strength and the lowest spread in measured strength between the soils. Lime alone or in combination with cement, with a lime percentage of 50% or higher, or with slag, gave the lowest short-term strength and the largest spread in measured strength between the soils. It was also observed that after one year, roughly the same strength was achieved with a large number of the different types of binder combinations. This finding has been reported in also
Figure 4.9 Examples of approximate range of measured strength at different times after mixing for clays and gyttja stabilised with various combinations of binders. Binder quantity 100 kg/m3. After Paper I.
Strength of stabilised soft soil 31 CHAPTER 4 previous investigations (e.g. Åhnberg et al., 1995). Lime alone or in combination with slag or cement gave 7-day strengths of 0.1-1.0 times that at 28 days, and 1-year strengths of 1.3-13 times that at 28 days; cement, cement- slag and cement-fly ash, gave corresponding ratios of 0.4-0.7 and 1.2-2.8, respectively. With pure cement, the early strength ratio was the same, but the range of 1-year strengths was narrower, at 1.4-1.8 times that at 28 days. The increase in unconfined compressive strength for the cement-stabilised soils, which were cured at 7 °C from 7 to 800 days, can be approximately described by the expression
q t ≈ 0.3 ⋅ ln t (4.3) q 28
where t is the time (days), and qt and q28 are the unconfined compressive strengths after t days and 28 days, respectively, see Figure 4.10. Equation (4.3)
3 Linköping clay Löftabro clay Holma gyttja y = 0.305ln(x) - 0.04 R2 = 0,94 Trend Nagaraj et al. (1996)
2 28 q /
t Horpibulsuk et al. (2003) q
1
0 0 200 400 600 800 1000 Time after mixing t , days
Figure 4.10 Relative increase in unconfined compressive strength with time for cement-stabilised clays and gyttja.
32 Strength of stabilised soft soil Effects of binders on soil properties - Field of researchresearch is similar to relationships reported previously for cement-stabilised soils (e.g. Nagaraj et al., 1996; Porbaha et al., 2000; Horpibulsuk et al., 2003).
The variation in strength increase with time is linked to differences in the chemical reactions taking place, as described in Section 4.1.1 (Paper II). The reaction products formed by the various binders (see Figure 4.1) are of different types and take different times to form. Assuming complete hydration and sufficient alumina and silica in the soil, it is possible to calculate the approximate amount of reaction products that would ideally form bonds in the soil when using different binders. In Figure 4.11, estimated variations in the amount of reaction products that can be produced by common types of binders are presented. In Figure 4.11, the blue, or pale shaded, bars represent the more rapid cement reactions and the yellow, or unshaded, bars represent the more long-term pozzolanic reactions with the soil. The red, or darker shaded, bars, for combinations of cement and fly ash, represent the pozzolanic reactions that may occur with the fly ash itself, since silica and alumina normally are more readily available in the fly ash than in the soil.
The amount of reaction products that may be formed gives an approximate indication of the effects of different binders on the rate and level of strength increase that can be expected. For comparison, Figure 4.11 also shows the strength measured in the three soils tested in the present study. The mean strength values after 28 days are fairly proportional to the theoretical amount of short-term reaction products formed. After one year, the strength is considerably higher in several cases. However, the variation in strength for different soils is large for binders that are intended to react mainly with minerals in the soil.
Besides the main reaction products formed, other factors may influence the rate of strength increase. The presence of retarding or accelerating chemical ingredients in the soil may lead to deviating increases in strength. The water content and the composition of the pore water may also affect the solubility of the pozzolans in the soil (Buchwald et al., 2003), and thus the rate of pozzolanic reactions with time. Additionally, although the different reaction products may be of the same type, their density and thus the achieved strength may vary. The use of combinations of binders may have an extra positive or negative effect on the strength increase, with one binder altering the conditions for the binding processes of the other component.
Strength of stabilised soft soil 33 CHAPTER 4
400 4000 Binder quantity 100 kg/m3 Max. pozzolanic soil reactions, CASH Max. pozz. binder reactions, CASH Cement reactions, CSH
quc, mean, 28 days 300 3000 , kPa q , 1 year uc
uc, min-max q
200 2000 Reaction products, kg products, Reaction
100 1000 Unconfined compressive strength compressive Unconfined
0 0 Lime cl (50:50) sl (50:50) Cement cs (50:50) cf (75:25) cf (50:50) Stabilising agent
Figure 4.11 Estimates of the amount of reaction products contributing to the strength of stabilised soils (bars) together with measured strength in three soils one month ( ) and the range one year ( ) after mixing. c = cement, l = lime, s = slag, f = fly ash. (From Paper II.)
Admixtures or other additions at low proportions may be used to improve the effects of the main binder. The effect of these additions will be expected to vary depending on the type of soil and type of main binder used. In this study, none of the different additions proved to have a generally positive effect on the strength of stabilised soils (Paper I). When considering the use of admixtures or mineral additions, trial tests on the soil of interest are thus important. Salt in the form of CaCl2 is a well-known accelerator of cement in concrete, although it may not lead to a higher final strength (Taylor, 1997). This admixture may have positive effects in lacustrine clays like the Linköping clay together with cement or lime, but has shown negative effects together with other types of binders. The addition of alumina cements may have a positive effect with time. However, the long-term stability must be clarified (Robson, 1962). Silica fume may have positive effects together with cement or cement-lime, but has shown negative effects with other binders. Water glass,
34 Strength of stabilised soft soil Effects of binders on soil properties - Field of researchresearch i.e. sodium silicate, was found to have mostly negative effects in clays and gyttja. Kiln dust from lime production gave about the same strength when replacing lime. Gypsum may have a wide range of effects, both negative and positive effects, and in all cases the effects may change direction with time. When using gypsum, the formation of ettringite involves expansion of the material. If early reactions are unhindered this need not create a problem. However, if delayed there may be a breakdown of bonds already formed (Taylor, 1997). Furthermore, for ettringite to remain stable, it is important that the pH remains at a high value, not lower than about 10-11 (e.g. Kujala, 1983; Högberg, 1983), and that the temperature does not become too high (Esrig, 1999).
In the tests carried out in this study, as most often in laboratory tests, the temperature during curing was kept constant. In deep mixing in the field, the temperature of soils stabilised with binders containing quicklime can be expected to be higher than those containing only cement, slag and fly ash in any combination. A higher temperature will speed up the chemical processes involved, resulting in a faster increase in strength with time (e.g. Åhnberg & Holm, 1987; Åhnberg et al., 1989). A decrease in water content, which always occurs when using dry binders, will in general cause an increase in strength. In the laboratory the water content will remain low or decrease further, whereas it may increase with time in the field due to saturation. An exception to the increase in strength with decreasing water content may occur, primarily in the field, when the initial water content is very low and adding large quantities of dry binder affects mixing and hydration negatively (Axelsson et al., 1996; Eriksson et al., 2005). Adding water, separately or in the form of a wet binder, i.e. a binder slurry, may then lead to a higher strength. Varying the mixing intensity (Larsson, 2003) or binder quantity may also influence the resulting strength, to different degrees depending on the type of binder. Normally, cement-containing binders are more sensitive to the mixing work and also to the binder quantity than binders dominated by lime (e.g. Åhnberg et al., 1995).
The strength increase in the field will normally not be quite the same as for samples stabilised in the laboratory. However, the principal effects of different binders can be expected to be fairly similar. Provided that sufficient binder quantities are used, a significant long-term strength increase can also be expected in the field. Continued increase in strength measured long after installation has been reported for lime-cement columns at a number of sites (e.g. Edstam et al., 2004; Löfroth, 2005), as well as for cement columns in a number of investigations (e.g. Hayashi et al., 2003).
Strength of stabilised soft soil 35 CHAPTER 4
4.2.2 Influence of initial loading
The application of an initial load shortly after mixing will affect the strength of the stabilised soil. Especially when stabilising peat, an initial surcharge, or preloading, in the field has been regarded necessary in order to create a more homogeneous stabilised mass of peat. In addition, the fill serves as a trafficable bed for the continued stabilisation of adjacent areas. This initial loading may also considerably improve the strength of the stabilised soil (Paper III). Figure 4.12 shows examples of increase in strength in samples stabilised with different binders in the laboratory for initial loads up to 18 kPa, which corresponds to about one metre of fill. A preload of 18 kPa may increase the strength of stabilised samples up to several times.
It is not primarily the magnitude of the initial load itself that governs the increase in strength, but the amount of compression resulting from loading.
Figure 4.12 Examples of measured unconfined compressive strength in stabilised peat samples with preloads of 0, 9, and 18 kPa. Binder quantity 200 kg/m3. After Paper III.
36 Strength of stabilised soft soil Effects of binders on soil properties - Field of researchresearch
The compression that occurs under preloading will reduce the distance between the binder grains and the particles in the peat and facilitate the build- up of bonds by the outer hydration products formed by the binders. The compression caused by preloading will increase with the size of the load, with decreasing initial density of the stabilised soil, and decreasing time lapse between mixing and loading. Furthermore, the compression will increase in the laboratory with decreasing sample height or similarly, with the distance to permeable soil layers or the ground surface in the field (Hayashi et al., 2005; Paper III).
Depending on the amount of binder added to the soil, the density of the soil will increase, and the void ratio and the water content, which is more easily and commonly determined than the void ratio1, will decrease to various extents after mixing. There will then be a further increase and decrease, respectively, in these properties, with increasing compression caused by the initial loading. The water content of the stabilised soil, which is affected by both the initial compression and the binder quantity, may be correlated with the strength. Figure 4.13 shows measured strength versus the water content of various types of stabilised peat samples after 28 days. The relationship between strength and water content is similar for various groups of stabilised samples. However, the relationship between water content and strength can be expected to change with time in different ways depending on the type of binder.
In the field, for cases in which a preload is applied to stabilised soil (e.g. peat), the drainage paths are much longer than in laboratory samples and the consolidation rate becomes correspondingly slower. Furthermore, the curing of the stabilised soil will, after a short time, increase the strength to such an extent that further compression after loading is prevented and the total compression will be reduced. To achieve a high strength it is therefore important to apply the load as soon as possible after mixing. In this study, loading delays of 4 and 24 hours reduced the strength of cement-slag stabilised peat samples by about 30% and 75%, respectively, compared with a delay of 45 minutes.
If the stabilised soil is overstressed at loading, causing large shear strains to occur, there will be a breakage of bonds (e.g. Hammond, 1981: Åhnberg et al., 1996; Steensen-Bach et al., 1996). Recovery will take place to some extent
1 Since stabilised soils as a rule are not completely saturated, the water content is a not fully correct, but still a practical measure of the void ratio.
Strength of stabilised soft soil 37 CHAPTER 4
1200 Stabilised peat cement cement-slag 28 days cement-slag-silica fume 1000 cement-slag-gypsum
800
cement-lime 600 cement-fly ash (kPa) cement-peat ash uc q
400
200 cement-gypsum cement-fly ash-gypsum
0 0 100 200 300 400 500 w (%)
Figure 4.13 Measured unconfined compressive strength versus water content in peat samples stabilised with different binders, initial load and loading delay. Binder quantity 100-300 kg/m3. Preload 9-18 kPa. Loading delay 0.75-24 hours. After Paper III.
by autogenous healing. However, the recovery in strength will decrease with increasing curing time before rupture occurs (e.g. Thompson & Dempsey, 1969; Hammond, 1981).
Apart from the effects of initial loading, the stress conditions in situ after stabilisation and during continued loading of the stabilised soil also affect the strength of the stabilised soil. This is discussed further in Section 4.3.3.
38 Strength of stabilised soft soil Effects of binders on soil properties - Field of researchresearch
4.3 UNDRAINED AND DRAINED STRENGTH BEHAVIOUR
The strength of the stabilised soil will vary depending on the stress conditions and the drainage conditions. In the previous section, the effects of various binders on strength levels and rates of strength increase were described. Examples of measured strengths were presented in terms of the results of unconfined compression tests, which are useful tests for investigating the approximate strength levels achieved in stabilised soils and for comparing the effects of various binder combinations on the strength increase. In order to fully describe the strength behaviour of a stabilised soil, the influence of stresses and pore pressure on the drained and undrained strengths should be investigated by performing various triaxial tests.
4.3.1 Influence of testing methodology
Triaxial tests can be performed in different ways. Several factors, such as the choice of back pressure, i.e. pore pressure applied during consolidation and testing, and the rate of strain during testing will affect the results of the tests (Paper V). Relatively low back pressures may be chosen to simulate conditions resembling those prevailing in the field shortly after stabilisation. High back pressures may be used in order to obtain strength parameters relevant to the more long-term case, where water from the surrounding soil enters the column and increases the degree of saturation. A high back pressure is needed to saturate the samples as much as possible and thus ensure more accurate measurements of pore pressures in the samples. A high back pressure is normally also required in order to enable measurements of the decrease in pore pressure generated in stabilised soils exhibiting dilatant behaviour. This is often the case for stabilised soils of medium to high strength. To ensure that any air trapped in the sample dissolves in the pore water, a back pressure of at least 400 kPa is normally called for in stabilised soils, where the degree of saturation after mixing is often as low as about 93-95% (Paper V). A back pressure of this magnitude is also required for natural soils at initial degrees of saturation down to about 92% (e.g. Lowe & Johnsson, 1960; Bishop & Henkel, 1962).
During the consolidation stage, resaturation occurs as water is pressed into the sample and air is dissolved in the pore water by the applied back pressure. The calculated degree of saturation of the stabilised clays and organic soils in the
Strength of stabilised soft soil 39 CHAPTER 4 study was between 93 and 98% prior to consolidation. In the stabilised clays, the degree of saturation increased to about 95-96% when using a back pressure of 20 kPa and to 97-99% for a back pressure of 400 kPa. Figure 4.14 shows examples of the effects of different back pressures on the degree of saturation of the different types of stabilised soil in the study. Complete saturation of clays stabilised with dry binders is normally neither possible to achieve with common triaxial equipment nor relevant, considering normal pore pressure conditions in the field.
The choice of back pressure affects the results of undrained tests. A higher failure strength is measured at low back pressures than at high back pressures (Paper V). The unsaturated materials initially respond as in a partly drained
1.05 Stabilised peat, 200 kg/m3 Stabilised gyttja, 70 kg/m3 Stabilised clays, 100 kg/m3
50 kPa 1.00 20 kPa
Back pressure
~400 kPa afterconsolidation r 0.95 S ~50 kPa ~20 kPa
0.90 0.90 0.95 1.00
Sr before consolidation
Figure 4.14 Calculated degrees of saturation, Sr, in stabilised Dömle peat, Holma gyttja and Linköping and Löftabro clay samples before and after consolidation at different back pressures. After Paper V.
40 Strength of stabilised soft soil Effects of binders on soil properties - Field of researchresearch test in the sense that excess pore water pressures are partly suppressed by compression or expansion of the entrapped air. Different stress paths may also be followed up to failure in samples that have approximately the same failure strength but have different degrees of saturation. The less saturated specimens initially follow a stress path closer to that in a drained test. As compression continues, the pore pressure generation increases, causing the stress paths to approach those measured in undrained tests on more fully saturated specimens. Examples of this behaviour are shown in Figure 4.15, where stress paths in tests on the stabilised peat are shown. The ratios between measured strengths at lower back pressure levels of 20-50 kPa and those at a higher back pressure level of 400 kPa varied between 1.0 and 1.6. The differences in measured strength increase with stress level and the strength of the stabilised soil. With increasing stiffness of the stabilised soil skeleton, the effect of small quantities of gas in the pores becomes greater, resulting in larger differences between unsaturated and almost completely saturated samples.
Stabilised peat τ fd (drained tests) 3 200 kg/m (c'=137 kPa 400 φ '=32 o ) )/2, kPa 3 ' σ σ σ σ - 1 ' 200 σ σ σ σ (
Undrained test, b.p.=400 kPa Undrained test, b.p.=50 kPa Drained test, b.p.=400 kPa Drained test, b.p.=50 kPa 0 0 200 400 600 σ σ (σ'1+σ'3)/2, kPa
Figure 4.15 Examples of registered stress paths in stabilised peat when using a high and a low back pressure (b.p.).
Strength of stabilised soft soil 41 CHAPTER 4
In drained triaxial tests, the specimens are sheared at a low enough rate of strain to allow full pore pressure equalisation to take place. There may thus be a limited difference in response when applying a low back pressure compared to a high one. However, a smearing effect due to an increase in water content, as well as possible wetting of drier clusters of stabilised soil, may lead to a somewhat lower strength. In this study (Paper V), the results of the drained tests showed no significant influence of different back pressures on the measured strength, which indicates that an adequate rate of strain had been chosen for these tests.
The rate of strain at which the tests are run may otherwise greatly affect the results of drained triaxial tests. At too high a strain rate, pore pressure equalisation may be too slow to prevent pore pressure build-up in the middle of the samples, leading to partly undrained behaviour. As a result, in tests on contractive stabilised soils, i.e. materials that decrease in volume during testing, a considerably lower “drained” strength may be measured. Contractive stabilised soils are commonly those with low strength that are tested at high consolidation stresses. In dilative stabilised soils, i.e. materials that expand during testing, negative pore pressures will instead build up in the middle of the sample and a higher strength will be registered if too high a testing rate is used. Dilation commonly occurs in stabilised soils of high or medium strength tested at low consolidation stresses. In the present work this behaviour was observed in the tests on the stabilised gyttja when using the higher strain rate of 0.2%/min, compared with the normal rate of 0.02%/min (Paper V). Similar behaviour has also been reported for unstabilised overconsolidated clay (e.g. Bishop & Henkel, 1962) and compacted silt (Roy & Sarathi, 1976).
In undrained tests, an increased strain rate may yield somewhat higher values of strength due to decreased pore pressure generation. A higher strength is also normal in unstabilised soils (e.g. Sheahan et al., 1996; Bjerrum, 1972; Torstensson, 1977; Karlsson, 1963). However, in this study, changes in the rate of strain from 0.002 to 0.2%/min had a limited influence, about 10% or less, on the measured strength in undrained tests on stabilised organic soils (Paper V).
The strengths measured in undrained triaxial tests normally differ from those of unconfined compression tests, particularly at high consolidation stresses. Unconfined compression tests only register the strength at zero total confining stress. There may also be a difference in measured strength at low consolidation stresses in undrained triaxial tests (Paper VI), as well as a
42 Strength of stabilised soft soil Effects of binders on soil properties - Field of researchresearch difference in stress-strain behaviour (Paper VII), compared with that in unconfined compression tests. In stabilised soils of relatively low strength, the undrained triaxial strength at low effective cell pressures and the strength measured in unconfined compression tests are usually of approximately the same magnitude. However, as the strength of the materials increases, the strength determined by unconfined compression tests as a rule becomes higher than that determined by undrained triaxial tests at very low effective confining stress. Figure 4.16 shows examples of differences in compressive strength evaluated from unconfined compression tests and those from undrained triaxial tests for effective confining stresses corresponding to zero. The latter values were estimated by extrapolating the triaxial test results. The differences were found to be small for strengths up to approximately 300 kPa. At higher strengths, the unconfined compression tests yielded higher strength values than those measured in the triaxial tests.
400 Linköping and Löftabro clay Dömle peat (Åhnberg et al., 2000) Holma gyttja (Åhnberg et al., 2000) 300 = ∆ triaxial strength (unsaturated - saturated) , kPa
=0 kPa) 200 3c ' σ σ σ σ ( triax - q 100 unconf. compr. q 0 0 200 400 600 800 1000
-100 σ q triax( '3c=0 kPa), kPa
Figure 4.16 Difference between compressive strength evaluated from unconfined compression tests and undrained triaxial tests; and from triaxial tests at low and high back pressure, i.e. unsaturated – “saturated”. After Papers V and VI.
Strength of stabilised soft soil 43 CHAPTER 4
The differences in results at higher strength levels are related to several factors. The triaxial and unconfined compression tests were performed on specimens with different degrees of saturation at different rates of strain. The specimens tested in the unconfined compression tests are not subjected to any back pressure prior to testing. For comparison, the differences in measured strength in the triaxial tests performed in this study with low and high back pressures, i.e. in unsaturated and “saturated” specimens (Paper V), are also displayed in Figure 4.16. For a lower degree of saturation, as in unconfined compression tests or triaxial tests with low back pressures, a lower pore water pressure is generated and a higher strength is measured. As the strength level increases, the stiffness of the material also increases, increasing the influence of small amounts of air in the pores. In this study, the differences in strength were somewhat smaller for the stabilised organic soils having a higher initial degree of saturation than the stabilised clays (Paper VI).
Besides different degrees of saturation, the results from the two types of tests are normally affected by differences in the strain rates used. Unconfined compression tests are typically performed considerably faster than triaxial tests. However, according to the effects of strain rates measured in the triaxial tests in this study, this may cause a limited difference in strength, of the order of about 10% or less.
4.3.2 Stress dependency
Both the undrained and the drained strength of stabilised soils exhibit a stress dependency (Paper VI). Approximations of the undrained strength, qu, as a function of the consolidation stress can be made with simple linear expressions. However, the mean increase in strength per unit increase in consolidation stress varies with the magnitude of the stress interval considered and also with the soil type. Another approach, similar to that often applied to natural clays can also be used to describe the stress dependence of the σ undrained strength in stabilised soils. A quasi-preconsolidation pressure, ´qp, exists at which yield occurs, although the stabilised soil has not been subjected to this pressure earlier. The quasi-preconsolidation pressure is governed both by previous stresses and the cementation of the soil (Paper VII). The occurrence of vertical yield stress has been demonstrated earlier in oedometer tests and triaxial tests on stabilised soils (e.g. Tatsouka & Kabayashi, 1983; Balasubramaniam & Buensuceso, 1989; Åhnberg et al., 1995; Lee et al., 2004). In the present study, the undrained strength was shown to be
44 Strength of stabilised soft soil Effects of binders on soil properties - Field of researchresearch dependent on the quasi-preconsolidation pressure and the corresponding overconsolidation ratio of the stabilised soils (Papers VI and VII). The variation in undrained strength of “overconsolidated” samples can be described by the following relationship
b §σ ′ · q = a ⋅σ ′ ⋅¨ qp ¸ u q 1c ¨ σ ′ ¸ (4.4) © 1c ¹
σ where ´1c is the consolidation stress applied in the tests, and aq and b are constants (Paper VI). At very low consolidation stresses, a minimum value of σ qu of approximately 0.6 ´qp may be applied. In this study, aq and b were approximately 1.0 and 0.9, respectively for the stabilised clays. Figure 4.17 shows the variation in undrained strength of the stabilised clays normalised to
2 Stabilised clays Löftabro clay Linköping clay 1.5 qp ´ σ σ 1 / u q
0.5
0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
σ ´1c/σ ´qp
Figure 4.17 Measured undrained strength in stabilised Löftabro and Linköping clay normalised to quasi-preconsolidation pressure at different degrees of overconsolidation. After Paper VI.
Strength of stabilised soft soil 45 CHAPTER 4 the quasi-preconsolidation pressure with the degree of overconsolidation σ σ expressed in terms of ´1c/ ´qp. There was no indication of any differences in the general behaviour with the different types of binders used. For specimens consolidated to vertical stresses higher than the quasi-preconsolidation σ pressures, as measured in oedometer tests, the normalised strengths, qu/ ´1c, were approximately the same, with a mean value slightly lower than one. These specimens may be regarded as being approximately “normally consolidated”, having an overconsolidation ratio close to unity, and included specimens with low strengths that were cured for very short periods of time, specimens stabilised with mainly ineffective binders, or specimens of normal strength tested at high stress levels. A general strength model and the influence of different stress and strength boundaries on the strength behaviour are discussed further in Section 4.3.3.
Fewer tests were performed on stabilised organic soils in this study but similar patterns were observed. The stress dependence appeared to be about the same for the stabilised peat as for the stabilised clays, while the results for the stabilised gyttja indicated somewhat higher values of the factors aq and b,of the order of 1.2 and 0.9, respectively.
The variation in undrained strength with the overconsolidation ratio of the stabilised soils is similar to that of the undrained shear strength of natural σ b soils, cu, which in most general models is described by cu = a ´vOCR or similar expressions where OCR is the overconsolidation ratio of the soil, (e.g. Ladd & Foott, 1974; Jamiolkowski et al., 1985). The value of b for the stabilised clays was slightly higher than that of natural soils, approximately 0.9 compared with the common value of 0.8 for clays (e.g. Mayne, 1988; Larsson & Åhnberg, 2005) and 0.85 for clay till (Steenfelt & Foged, 1992; Larsson,
2000). The value of the factor aq, used to express the compressive strength, in analogy to a used for the shear strength, is generally higher than would be expected for a corresponding compressive strength of natural soil, about 1.0- 1.2 compared with about 0.66 for clay (e.g. Larsson & Åhnberg, 2005), 0.8 for clay till (Christensen et al., 1992; Larsson, 2000) and 1.0 for highly organic soils (Larsson, 1990). At very low strength levels, as in the case of the lime-stabilised Linköping clay, the values of aq may be closer to those expected for natural clay.
The magnitude of the stress dependence measured in the study is similar to that reported in earlier investigations (e.g. Okumura, 1996; Balasubramaniam & Buensuceso, 1989; Åhnberg et al., 1996). However, large variations in
46 Strength of stabilised soft soil Effects of binders on soil properties - Field of researchresearch results, with significantly lower, as well as slightly increased, strength with decreasing confining stresses, have been reported in the literature. The significant scatter in the results may reflect differences in the mode of sample preparation as well as testing procedure.
The triaxial tests used to study the stress dependence of stabilised soils were all performed on specimens that were saturated by high back pressures. A lower degree of saturation in stabilised soils can be expected to result in a greater stress dependence of the undrained strength (see previous section), as is also the case for a partly saturated soil (e.g. Fredlund & Rahardjo, 1993). High “undrained friction angles”, as this stress dependency somewhat misleadingly is often denoted, have thus been reported from tests performed on unsaturated samples of soils stabilised with dry binders (e.g. Kivelö, 1998; Axelsson, 2000).
The drained strength of stabilised soils exhibits a clear stress dependency (Papers VI and VII). This stress dependency is greater than that of the undrained strength. Figure 4.18 shows the drained strength and the undrained strength at varying degrees of “overconsolidation” for the stabilised clays in this study. Contrary to natural soils, the drained strength is generally about the same as or higher than the undrained strength also in highly “overconsolidated” stabilised soils. At low confining stresses, the drained and undrained shear strengths are often approximately equal. In this study, the agreement was found to be fairly good at medium strength levels, whereas at lower strength levels the extrapolated drained strength at zero effective confining stress was, in most cases, lower than the undrained strength. In tests performed at higher consolidation stresses than the quasi-preconsolidation pressure evaluated from the oedometer tests, the samples were assumed to be close to normally consolidated. The variation in normalised undrained and σ σ drained strength with increasing ´1c/ ´qp ratio for these tests is indicated by dashed lines in Figure 4.18. The drained tests referred to in the figure were performed at K0= 0.8. Higher or lower values of K0 will give higher and lower values, respectively, of the drained strength, see Section 4.3.3 and, e.g., Figure 4.25.
In this study, a fairly constant relationship between the drained strength and the effective consolidation stress was obtained for the different materials at the investigated strength levels. The mean effective friction angles φ´ evaluated were about 33° in both of the stabilised clays, and 33° and 31° in the stabilised peat and gyttja respectively. The values of φ´ for all the test series in the study varied independently of the type of binder from 26° to 38°. A tendency
Strength of stabilised soft soil 47 CHAPTER 4
Figure 4.18 Measured undrained and drained strength of stabilised Löftabro and Linköping clay normalised to quasi-preconsolidation pressures at different degrees of overconsolidation. The estimated influence of K0 is indicated by the shaded area.
towards curvature in some of the strength envelopes indicates a certain decrease in friction angle and increase in cohesion intercept with increasing stress level. This curvature is probably caused by decreasing dilatancy with increasing stress level.
The cohesion intercept, c´, varies greatly with time and type of binder. It increases with time in a way similar to the undrained strength. The variation of c´ with qu(σ´3c=0) observed in this study is shown in Figure 4.19. The value of c´ was approximately 0.25 times the undrained compressive strength at zero effective cell pressure, qu(σ´3c=0), as estimated from extrapolation of the results from the triaxial tests. The ratio of c´ to q varied between 0.12 and 0.33 u(σ´3c=0) for all the test series. The lower ratios were measured in stabilised soils of relatively low strength, whereas higher ratios were measured in stabilised soils
48 Strength of stabilised soft soil Effects of binders on soil properties - Field of researchresearch
250
y = 0.2482x 200 R2 = 0.9209
150 c', kPa 100
Linköping clay 50 Löftabro clay Dömle peat Holma gyttja 0 0 200 400 600 800 1000
qu(σ'3c=0 kPa), kPa
Figure 4.19 Evaluated cohesion intercept, c´, vs. undrained compressive strength. After Paper VI
of medium and high strength. The higher ratios are probably affected by the increase in c´ caused by dilatancy effects at higher strength levels.
Results from drained triaxial compression tests on stabilised soils are less frequently reported in the literature than results from undrained tests. Figure 4.20 shows effective friction angles, φ´, evaluated from the drained tests in the present study together with data from earlier investigations. Results reported from drained tests performed on samples of generally higher strength than those in the present study (Tatsouka & Kobayashi, 1983; Kawasaki et al., 1984; Åhnberg et al., 1996) indicate both slightly increasing and decreasing effective friction angles with increasing strength. The strength ratios, c´/qu(σ´3=0), evaluated from the drained tests in the present study, together with those assessed from test results in earlier investigations, are shown in Figure 4.21. The strength ratio levels off towards a value of the order of 0.15-0.2 for high- strength samples. The stress dependence at high strength levels requires
Strength of stabilised soft soil 49 CHAPTER 4
0.5 Present study - clays and organic soils Tatsouka & Kobayashi (1983) - Tokyo clay Kawasaki et al. (1984) - Japanese clay 0.4 Balasubramaniam & Buensuceso (1989) - Bangkok clay Present study Åhnberg et al. (1996) - Mellösa clay Åhnberg et al. (1996) - Lövstad gyttja Åhnberg et al. (1996) - Kil silt
0.3 =0 kPa) =0 3c ' σ σ σ σ ( u
0.2 c' / q
0.1
0 0 500 1000 1500 2000 2500 3000 σ qu( '3c=0 kPa), kPa
Figure 4.20 Evaluated effective friction angle vs. strength of stabilised soil in this study together with results from earlier investigations. Binder contents 5-300% (of dry weight). After Paper VI.
60
50 Present study
40
o 30 ', φ φ φ φ
20 Present study - clays and organic soils Tatsouka & Kobayashi (1983) - Tokyo clay Kawasaki et al. (1984) - Japanese clay 10 Balasubramaniam & Buensuceso (1989) - Bangkok clay Åhnberg et al. (1996) - Mellösa clay Åhnberg et al. (1996) - Lövstad gyttja Åhnberg et al. (1996) - Kil silt 0 0 500 1000 1500 2000 2500 3000 σ qu(σ'3c=0 kPa), kPa
Figure 4.21 Strength ratio vs. strength of stabilised soil in this study together with corresponding results from earlier investigations. Binder contents 5-300% (of dry weight). After Paper VI.
50 Strength of stabilised soft soil Effects of binders on soil properties - Field of researchresearch further testing over a greater strength range. Results from earlier investigations indicate a slightly increasing friction angle with decreasing water content (Balasubramaniam & Buensuceso, 1989; Åhnberg et al., 1995). However, this effect was not obvious from the results in the present study. A tendency towards a slight increase in the friction angle with decreasing plasticity index was, however, observed. This is in accordance with the effects of plasticity index on effective friction angles reported for natural soils by, e.g., Kenney (1959) and Terzaghi & Peck (1967).
4.3.3 Adaptation of a strength model – stress paths and influence of stresses during curing
In previous sections, it has been shown that the quasi-preconsolidation pressure is strongly linked to the strength of the stabilised soil and has a considerable influence on its deformation behaviour, see Sections 4.3.2 and 4.1.4, respectively. The effects of the quasi-preconsolidation pressure on the strength behaviour of stabilised soils can be further illustrated by studying the undrained stress paths and the stress-strain responses. Studies of yield stresses and their effect on undrained stress paths (e.g. Kasama et al., 2000; Uddin & Buesuceso, 2002; Balasubramaniam et al., 2005) and strength (e.g. Horpibulsuk et al., 2004) in triaxial strength testing are few, and little has been published concerning the effects of different binders. In the present study, results from series of active and passive triaxial tests served to describe the strength behaviour in this respect (Paper VII).
In Figure 4.22 examples are shown of measured effective stress paths in undrained tests together with failure and yield stresses evaluated from drained tests on stabilised clay, presented in a s´:t effective stress plane, where s´ = σ σ σ σ ( ´1+ ´3)/2 and t = ( ´1- ´3)/2. The test results include different binders and quantities, and times after mixing. No external stresses were applied to the samples during curing. The measured strength varied considerably, reflecting soft to stiff stabilised soils. Lines are drawn indicating the evaluated effective strength parameters friction angle, φ´, and cohesion intercept, c´, for the σ different mixtures. Lines representing quasi-preconsolidation pressures, ´qp, evaluated from the oedometer tests, are also indicated in the diagrams.
At consolidation stresses in triaxial tests higher than, or of the same order as, the quasi-preconsolidation pressure, relatively high positive pore pressures are generated during undrained testing. This is seen in a more distinct turn in the
Strength of stabilised soft soil 51 CHAPTER 4
500 Linköping clay Lime 100 kg/m3 CAU CAD
s.p. q max q yield 400 1 day 28 days 1 year 300 )/2, kPa 3 ' qmax (drained triaxial tests) c'=5-13 kPa σ σ σ σ
- o
1 φ ' 200 '=29-33 σ σ σ σ (
s.p. (undrained triaxial tests) 100 qyield (drained triaxial tests)
σ'qp (oedometer tests) 0 0 100 200 300 400 500 600 700
(σ'1+σ'3)/2, kPa 500 Linköping clay Cement 100 kg/m3
CAU CAD s.p. 400 q max q yield 1 day 28 days 1 year c'=63-140 kPa 300 φ'=31-35o )/2, kPa )/2, 3 ' σ σ σ σ - 1 ' 200 σ σ σ σ (
100
0 0 100 200 300 400 500 600 700 σ σ (σ'1+σ'3)/2, kPa 500 Linköping clay 28 days CAU CAD 400 s.p. q max q yield Slag-lime Cement-lime Cement-fly ash Cement-slag 300 c'=12-125 kPa φ'=35-37o )/2, kPa 3 ' σ σ σ σ - 1 ' 200 σ σ σ σ (
100
0 0 100 200 300 400 500 600 700
(σ'1+σ'3)/2, kPa
Figure 4.22 Examples of measured stress paths (s.p.) in the s´:t stress plane for stabilised clay. From Paper VII.
52 Strength of stabilised soft soil Effects of binders on soil properties - Field of researchresearch effective stress paths towards the effective stress failure line in the drained tests, and failure in the undrained tests occurs close to the point where this line is reached. In tests where the consolidation stresses are well below the quasi- preconsolidation pressure, the undrained stress paths approach the drained failure line more gradually, and failure occurs at a point below or around the state where the effective vertical stress equals the quasi-preconsolidation pressure. In the various test series in this study, most of the quasi- preconsolidation pressures were evaluated from single oedometer tests and, as an effect of the scatter in results, varying deviations from this behaviour can also be observed.
When stress paths in drained tests pass a point at which the effective stress corresponds approximately to the quasi-preconsolidation pressure, a distinct break in the stress-strain curves can often be observed. The strains measured up to this “breakpoint” are low, only a few tenths of a percent, whereas failure in these cases often occurs at considerably larger strains. This yielding becomes less distinct as the strength of the samples increases and occurs then, if at all, at stresses and strains closer to failure. In the present study, the variation in stress at which a first yielding was observed was quite large, indicating that these breakpoints can only be taken as an indication of the approximate level of the quasi-preconsolidation pressure.
There are several model used for describing the behaviour of natural soils. A strength model developed for natural clays (Larsson, 1977) can be adapted to describe the strength behaviour of stabilised soils. In this strength and yielding model, limit state curves are schematically given as four segments, two of which correspond to the failure strength lines in compression and extension, σ σ σ σ and two that correspond to ´v= ´vc and ´h= K0nc ´vc. The model describes the limit state of natural soils for the triaxial case, showing the effects of anisotropy in a simple way linked directly to normally determined soil parameters. The adaptation of this model to stabilised soils is demonstrated in Figure 4.23, where results are shown from active and passive tests performed on samples of one type of stabilised Linköping clay. For these samples, prepared and stored in the normal way with no external stresses, an approximately isotropic yield surface can be adopted, described by an apparent
K0nc value close to unity. For comparison, stress paths, effective stress failure lines, and yield surfaces of the natural, unstabilised clay are also shown.
The effects of stabilisation are clearly demonstrated by the significant increase in cohesion intercept and the much higher quasi-preconsolidation pressure
Strength of stabilised soft soil 53 CHAPTER 4
Figure 4.23 Measured stress paths (s.p.) in the s´:t stress plane in active and passive triaxial tests on stabilised and unstabilised Linköping clay. From Paper VII.
than that of the natural soil, whereas a change in effective friction angle is less evident. For stabilised soils, variations in effective friction angles of the same order as those observed in the study (see previous section), in general have limited influence on the drained strength compared with the increase in cohesion intercept, except at high effective confining stresses. The increase in c´ is closely related to the simultaneous increase in undrained strength, as shown in the previous section. An exception to the normal increase in c´ for stabilised soils could be observed when using lime in the Linköping clay, see Figure 4.22. In these samples practically no change in cohesion intercept or effective friction angle with time was observed. This lack of any difference between one day and one year was also observed in the unconfined compression tests on this clay mixed with lime, see Figure 4.8b.
The strength behaviour may be further illustrated by normalising the stress paths with respect to the quasi-preconsolidation pressure evaluated from the oedometer test. This normalisation provides a congruent picture of the stress paths for the different types of stabilised soils included in the study. Figure
54 Strength of stabilised soft soil Effects of binders on soil properties - Field of researchresearch
4.24 shows the normalised results for the active tests on the two clays in this study. In cases where the consolidation stress applied in the triaxial testing was σ higher than the value of ´qp evaluated from the oedometer tests, the specimens were treated as being normally consolidated. The stress paths of these specimens are indicated by dotted lines, and failure by open circles in the figure. The results of the tests on mixtures of lime and Linköping clay, which barely showed any signs of having been stabilised and thus clearly deviated from the rest of the materials with regard to the magnitude of the normalised cohesion, are not included in the figure. Normalisation of the stress paths for this material indicated behaviour closer to that of natural clay.
Based on the results of both active and passive tests in this study, Figure 4.25 shows generalised stress paths of stabilised clays that have been cured without being subjected to external stresses. Considering the results of all the tests on the different types of stabilised clays in the present study, the mean drained
1.5 CAU CAD
s.p q failure q max q yield Stab. Linköping clay Stab. Löftabro clay Stab. Linköping clay Stab. Löftabro clay 1 qp ' σ σ σ σ σ c'=(0.14-0.16) 'qp
)/2/ o
3 φ
' '=33-34 σ σ σ σ - 1 ' σ σ σ σ (
0.5
σ σ "oc" ( 'vc < 'qp) σ σ "nc" ( 'vc d 'qp)
0 0 0.5 1 1.5 2
(σ'1+σ'3)/2/σ'qp
Figure 4.24 Measured stress paths normalised to the quasi-preconsolidation pressures in the s´:t stress plane in triaxial tests on stabilised Löftabro and Linköping clays. After Paper VII.
Strength of stabilised soft soil 55 CHAPTER 4 strength could be approximately described by a cohesion intercept of 0.15σ´ ≈ σ σ σ qp and a friction angle of 33°, giving qd 0.25 ´qp+ 0.54( ´1 + ´3). Under undrained conditions, the strength is affected by the magnitude of the consolidation stress in relation to the quasi-preconsolidation pressure. The undrained strength of the stabilised clays in the study varied from σ σ approximately 0.6 ´qp for highly overconsolidated samples to 1.0 ´qp for “normally consolidated” samples.
The specimens that prior to testing had been consolidated at stresses close to or higher than the quasi-preconsolidation pressure measured in the oedometer σ tests, indicate a somewhat higher yield stress, of up to 1.1 to 1.2 ´qp. These results are probably affected to some degree by cementation effects during consolidation. The evaluation of the quasi-preconsolidation pressure from the oedometer tests on the stabilised soils also affects the results. The quasi- preconsolidation pressures were evaluated in the standard way for natural, unstabilised soils. No further analyses of measurements or evaluation of
1
Stabilised clays Failure line σ c'=(0.14-0.16) 'qp φ'=33-34o 0.5 qp ' σ σ
σ σ σ ' h =0 )/2/ h ' σ σ σ σ - v ' σ σ σ σ Yield surface (
0 00.511.52
σ Undrained tests ' v =0 Drained tests
-0.5
(σ'v+σ'h)/2/σ'qp
Figure 4.25. Stress paths normalised to the quasi-preconsolidation pressure in the s´:t stress plane in triaxial tests on stabilised soils cured under isotropic stress conditions.
56 Strength of stabilised soft soil Effects of binders on soil properties - Field of researchresearch preconsolidation pressure from oedometer tests were performed in the present study.
When the specimens are consolidated at low confining stresses, the stress paths in undrained tests may approach zero effective horizontal stress and come close to the tension cut-off line. Here, the stabilised specimens may fracture unless a tensile strength can be mobilised. In highly “overconsolidated” specimens, negative pore pressures as a rule start to develop as the stress paths approach the failure line. In the present study, the pore pressures generated in σ σ the stabilised soils varied between approximately -0.1 ´qp and 0.5-0.6 ´qp for highly “overconsolidated” and “normally consolidated” specimens, respectively. In the drained tests, the volumetric strains for these specimens also showed that dilation starts at points somewhat below the indicated failure line. This gives a slightly rounded shape to the yield surface close to the origin, in contrast to the straight lines given by the model.
For undrained conditions, the stabilised soils exhibit almost ideal-plastic behaviour after failure, without significant reduction in strength, see Figure 4.6. For drained conditions in highly overconsolidated specimens, the stabilised soils exhibit brittle behaviour with a significant reduction in strength after failure, which occurs at a very small strain. No further studies of the strength behaviour at large strains were made in this study.
In the passive tests, the specimens designated normally consolidated exhibited a horizontal yield surface corresponding to a somewhat lower apparent K0nc value than the others. However, the effect of cementation, which is of isotropic nature, has a significant impact on the results and constrains the change in apparent K0nc to a higher value than that used during consolidation in the triaxial tests.
In this study, the same effects of a quasi-preconsolidation pressure were observed in the stabilised organic soils as in stabilised clays. However, the σ mean ratio of c´/ ´qp for the stabilised organic soils was slightly higher than that for the stabilised clay. Whether an initial loading of the order of 18 kPa (see Section 2.3.3), which corresponded to only about 10% of the quasi- preconsolidation pressure, has any anisotropic effect on the yield stresses, could not be interpreted from the results.
Subjecting the stabilised samples to stresses during curing results in an increase in strength as well as an increase in quasi-preconsolidation pressure. Applying
Strength of stabilised soft soil 57 CHAPTER 4 the curing stresses shortly after mixing causes compression of the sample and allows the cementation processes to proceed with the particles arranged closer together. In the present study, a number of tests were performed on samples subjected to external stresses during curing. It was clear that apart from the quasi-preconsolidation pressure being affected to a large degree by cementation effects, the stresses acting during curing also have an impact on the strength behaviour of the stabilised soil. The increase in c´ was σ approximately proportional to the increase in ´qp, whereas the effective friction angle was about the same for samples cured with and without external stresses. Since both the curing stress and cementation effects affect the yield surface, the ratio between horizontal and vertical yield stresses, K0qp, will not be the same as the K0 value used during curing. However, the K0 value used during curing does have an influence. When a K0 of 0.5 was used, an increase in horizontal yield stress was observed corresponding to half the increase in vertical quasi-preconsolidation pressure. Figure 4.26 shows the general effective stress paths followed in samples subjected to anisotropic external
1
Stabilised clays σ c'=(0.14-0.16) 'qp φ'=33-34o
0.5 qp ' σ σ σ σ )/2/ h ' σ σ σ σ - v ' σ σ σ σ (
0 00.511.52
K 0qp σ ' qp
Undrained tests Drained tests
-0.5 σ σ σ (σ'v+σ'h)/2/σ'qp
Figure 4.26 Stress paths normalised to the quasi-preconsolidation pressure in the s´:t stress plane in triaxial tests on stabilised soils cured under anisotropic stress conditions.
58 Strength of stabilised soft soil Effects of binders on soil properties - Field of researchresearch
curing stress. The effect of the K0 value used during curing on the value of K0qp for the yield stresses can be expressed approximately as
′ ′ 1+ K ∆σ qp σ qp K = 0 0 (4.5) 0qp ′ ′ 1+ ∆σ qp σ qp 0
′ σ qp where 0 is the quasi-preconsolidation pressure for a sample not subjected to external stresses during curing.
Since the effects of increased stresses are related to the compression of the stabilised material, increased stresses caused by a load, e.g. from an embankment, should preferably be applied in a way that produces a certain compression of the material shortly after stabilisation. This is in accordance with results observed from initial loading of stabilised peat, see Section 4.2.2.
Strength of stabilised soft soil 59 CHAPTER 4
60 Strength of stabilised soft soil 5 Conclusions and further research
5.1 CONCLUSIONS
A large number of binders can be used for the stabilisation of soils. In this study, a variety of binders and binder combinations were used to stabilise, i.e. increase the strength and stiffness of, four very soft soils, two clays and two organic soils. These soils were stabilised with good results with most, but not all, of the binders. The composition of the soil may have a considerable influence on the results, and it was concluded that the optimum combination of binders for a certain application will vary with the type of soil.
To interpret the results obtained from tests of stabilisation effects, it is essential to understand the basic reactions normally taking place in soil after mixing with binders. This study showed that the increase in strength with time, both in short-term and long-term perspectives, can in general be linked to the type and quantity of various reaction products that are generated by the chemical processes taking place during stabilisation.
When stabilising soils with binders, the properties will change to different extents depending on the type and amount of binder used and the type of soil. However, the results presented in this thesis show that the ways in which the soil reacts with a binder are largely similar, regardless of the type of binder used. Effects of stabilisation with dry binders in the laboratory are that the water content typically decreases, the density increases or remains about the same, and the degree of saturation decreases. The permeability may initially increase or decrease as a result of the introduction of binder particles into the
Strength of stabilised soft soil 61 CHAPTER 5 soil, flocculation of soil particles and any compaction applied to the stabilised soil. The permeability thereafter decreases with time, i.e. with increased cementation. A rough estimate of the change in permeability of homogeneous, stabilised soils can be made on the basis of the change in water content and the strength of the stabilised soil.
The strength obtained in a stabilised soil depends not only on the specific combination of soil and binder, and the quantity of binder, but also on a number of other factors, such as the degree of saturation, the stress conditions and the drainage conditions. The effects of these factors will vary according to the application, but must be taken into consideration in the design of deep mixing. In the laboratory, tests should preferably be performed so as to simulate the conditions expected in the field. However, the conditions in the field may vary considerably, calling for rather laborious testing programmes if all the expected conditions are to be covered. In designing more rational testing programmes, it is therefore important to consider the influence of factors that may vary in the field. In this study, several examples of typical strength behaviour were shown in this respect.
The drainage conditions, the effective consolidation stresses, the back pressure and strain rate may vary depending on the method used for strength testing. The results of the present work show that triaxial tests on soft soils stabilised with normal quantities of dry binders should preferably be performed at a back pressure of at least about 400 kPa and a strain rate of the order of about 0.02%/min or slower, to yield results relevant under almost fully saturated conditions. This recommended strain rate refers in particular to drained tests. For soil stabilised by dry binders, a significant difference in the measured undrained shear strength is observed when using different levels of back pressure. Under unsaturated conditions, the stabilised soils behave as if partly drained since excess pore water pressure can be partly equalised by compression or expansion of the entrapped air. In the drained tests, on the other hand, no significant differences were observed with different back pressures. Different rates of strain in the undrained triaxial tests indicate a limited influence on the measured strength, when the rate was varied from 0.2 to 0.002%/min. In the drained tests, varying the rate of strain resulted in differences in measured strength in tests performed at 0.2%/min compared with 0.02 and 0.002%/min. This was assumed to be caused by a build-up of excess pore pressure (positive or negative) with the highest rate, thus entailing not fully drained conditions.
62 Strength of stabilised soft soil Conclusions and further research
For low or medium stabilised soil strengths, the results of unconfined compression tests, performed without any further saturation, indicate about the same undrained strength as triaxial compression tests at low confining stresses on samples saturated by high back pressures. At higher strength levels, however, the unconfined compression tests indicated higher strength values than the triaxial tests. In cases when the degree of saturation can be expected to increase with time, unconfined compression tests may thus yield misleadingly high results.
The results showed that although the type of binder used in stabilising the soils may strongly affect the rate of strength increase and final strength, the general strength behaviour is the same for the most common binders. Furthermore, the same set of parameters that is used to describe the strength of natural soils can also be used for stabilised soils. However, it should be observed that specific binders or soil types can have specific effects on the behaviour, e.g. fibrous materials and needle-shaped reaction products may render more ductile behaviour.
σ A quasi-preconsolidation pressure, ´qp, can be observed in the triaxial tests as well as in the oedometer tests. It was shown that both the cementation process involved and the stresses applied during curing determine this quasi- preconsolidation pressure. The influence of the quasi-preconsolidation pressure was observed in active as well as passive triaxial tests. It affects the stress paths and is closely related to the strength of the stabilised soils. The stabilised soils behave in an overconsolidated manner when the consolidation stress is significantly lower than the quasi-preconsolidation pressure, and in a normally consolidated manner when the consolidation stress is of the order of σ 0.8 to 1.0 ´qp.
The undrained and drained strengths of stabilised materials are stress dependent in a way similar to that of natural soils. The stress dependence of the undrained strength can be described by taking into account a degree of quasi-overconsolidation giving a curved strength-stress relation in the “overconsolidated range”. For stabilised clays, the results indicated a curved σ σ σ 0.88 relation in accordance with approximately qu = 1.0 ´1c ( ´qp/ ´1c) . The undrained compressive strength of the stabilised clays varied from σ σ approximately 0.6 ´qp for highly overconsolidated samples to 1.0 ´qp for normally consolidated samples. The measured drained strength parameters of the four stabilised soils in the present study varied to a fairly large extent. The mean cohesion intercept, c´, could be expressed as approximately 0.23qu (or
Strength of stabilised soft soil 63 CHAPTER 5
φ ° 0.46cu). The mean effective friction angle, ´, was 31-33 for the four stabilised soils, independent of the type of binder. The variation in drained strength of the different types of stabilised clays could approximately be σ expressed by a mean cohesion intercept of 0.15 ´qp and a mean friction angle of 33°.
A common yielding model for natural clays (Larsson, 1977) was also found suitable for describing the behaviour observed in the tests on stabilised soils. The yield surface can be described schematically by the failure lines in σ σ σ Κ σ compression and extension and by ´v= ´qp and ´h= 0qp ´qp. An isotropic yield surface can be adopted for samples cured in the normal way, without external stresses. Stress applied to a stabilised soil shortly after mixing will compress the material, causing cementation to take place with particles arranged closer together, resulting in increased strength and increased quasi- preconsolidation pressure. The results of the study showed that an initial load during curing can greatly improve the strength of stabilised soil. An even early loading in the field is thus beneficial, especially in improving the homogeneity and strength of stabilised peat. A K0 value lower than unity of the stresses applied during curing results in a ratio of the horizontal-to-vertical yield stresses, K0qpc, lower than 1.0, but higher than the applied stress ratio.
Consistent patterns linked to the degree of overconsolidation can also be observed in the stress-strain relations. In undrained tests on a stabilised soil subjected to a confining effective stress, there is only limited further change in deviator stress with continued straining after failure. This is in contrast to the stress-strain response most often measured in unconfined compression tests, where a drastic reduction in strength normally occurs at large strains. Not taking into account the effects of confining stresses in situ may thus lead to miscalculation of stabilised–unstabilised soil interaction. In drained tests on stabilised soils consolidated at stresses well below the quasi-preconsolidation pressure, failure occurs at small strains, and a significant reduction in shear stress with strain after failure can be observed. Stabilised soils that are close to “normally consolidated”, on the other hand, may not exhibit failure until after large strains.
64 Strength of stabilised soft soil Conclusions and further research
5.2 FURTHER RESEARCH
Very large differences in stabilising effect were found between the two clays studied in the investigations. Detailed studies of the influence of the chemical composition of the soils on the strength increase were not included in this study. Further studies of the impact of different soil parameters should be performed to improve the estimates of the chemical reactivity of different soil types.
In the present study, four types of soil were investigated. In order to expand the applicability of the general behaviour of stabilised soil described in this thesis, the strength behaviour of an increased number of stabilised soils, e.g. silty soils, soils of different mineralogy and highly organic gyttja, should also be investigated.
The stress dependence at high compressive strength levels, about 800 kPa and above, requires additional testing in order to model the behaviour over a larger strength range. Further studies of the influence of curing stresses on the strength behaviour of samples cured at different strength levels should also be performed in order to allow more general conclusions to be drawn.
The behaviour of partly saturated stabilised soils should be investigated further. In the present study, differences in results between saturated and unsaturated samples were investigated by a number of tests, but attention was focused on the behaviour of saturated stabilised soils. This was considered to be a necessary first step in studying stabilised strength behaviour. In the next step, partly saturated stabilised soils should be studied further, regarding both changes in degree of saturation to be expected in situ and the influence of this on the strength behaviour.
The properties of stabilised soils using different types of binder should also be studied in the field. Correlations between laboratory and field properties should be studied. Comparative studies of field and laboratory strength behaviour should be made in order to improve the basis for the prediction of strength in situ. Further studies are also required concerning the permeability of columns in the field, as the permeability measured in the laboratory can only be regarded as a lower limit. Large-scale tests performed on columns in the field have shown somewhat higher or even considerably higher values than those measured in the probably more homogeneous samples in the laboratory.
Strength of stabilised soft soil 65 CHAPTER 5
Design models for deep mixing should be further developed to incorporate the concepts of an effective stress state, a quasi-preconsolidation pressure and a stress dependence of undrained as well as drained strength. Furthermore, substantial long-term increases in strength as well as significant changes in other important properties, such as permeability and compressibility, were observed for many of the binder combinations. A time factor should thus be incorporated into the design model in order to better account for the construction and lifetime behaviour of a foundation on stabilised soil.
66 Strength of stabilised soft soil References
REFERENCES IN SUMMARY
Alén, C., Baker, S., Bengtsson, P.-E. & Sällfors, G. (2005). Lime/cement column stabilised soil – A new model for settlement calculation. Proceedings International Conference on Deep Mixing, Stockholm 2005. Vol. 1.2, pp. 205-212. ASTM standard (2000). ASTM standard C 618 – 00. Standard specification for coal fly ash and raw or calcined natural pozzolan for use as a mineral admixture in Portland cement concrete. Axelsson, K., Johansson S-E. & Andersson R. (1996). Stabilisation of organic soils. Feasibility study. Swedish Deep Stabilisation Research Centre, Report No. 3. (In Swedish) Axelsson, M. (2000). Deep stabilization with lime cement columns. Methods for quality control in the field. Licentiate Thesis, Royal Institute of Technology. Division of Soil and Rock Mechanics. Also in Swedish Deep Stabilization Research Centre, Report No. 8, 163 p. (In Swedish) Babasaki, R., Terashi, M., Suzuki, T., Maekawa, A., Kawamura, M. & Fukazawa, E. (1996). JGS TC Report: Factors influencing the strength of improved soil. Proc. 2nd Int. Conference on Ground Improvement Geosystems – IS Tokyo’96. Grouting and Deep Mixing, Tokyo, 1996. Vol. 2, pp. 913- 918. Baker, S. (2000). Deformation behavior of lime/cement column stabilized clay. Ph.D. Thesis, Department. of Geotechnical Engineering, Chalmers University of Technology, Gothenburg. Balasubramaniam, A.S. & Buensuceso, B.R. (1989). On the overconsolidated behavior of lime treated soft clay. Proc. XII International Conference on Soil Mechanics and Foundation Engineering. Rio de Janeiro 1989, pp. 1335- 1338. Balasubramaniam, A.S., Buensuceso, B., Oh, E.Y.N., Bolton, M., Bergado, D.T. & Lorenzo, G. (2005). Strength degradation and critical state seeking
Strength of stabilised soft soil 67 References
behaviour of lime treated soft clay. Proc. International Conference on Deep Mixing – Best Practice and Recent Advances, Deep Mixing’05, Stockholm 2005, pp. 35-40. Bengtsson, P-E. & Åhnberg, H. (1995). Settlement calculations for lime column stabilisations. Road E18 at Fiskvik kanal. SGI project 1-261/86. (In Swedish) Bergwall, M. & Falksund, M. (1996). Consolidation of lime cement columns – permeability and stiffness. Master’s thesis 1996:1, Department of Geotechnical Engineering, Chalmers University of Technology, Gothenburg. Bishop, A.W. & Henkel, D.J., 1962, The measurements of soil properties in the triaxial tests. Second edition, Edward Arnold Ltd, London. Bjerrum, L. (1972). Embankments on soft ground. Proc. Specialty Conference on Performance of Earth and Earth-Supported Structures, Lafayette 1972, Vol. II, pp. 1-54. Boardman, D. L, Glendinning, S. & Rogers, C. D. F. (2001). Development of stabilisation and solidification in lime-clay mixes. Géotechnique Vol. 50, No. 6, pp. 533-543. Boynton, R. (1980). Chemistry and technology of lime and limestone. John Wiley & Sons Inc., New York1980. Brandl, H. (1981). Alteration of soil parameters by stabilization with lime. Proc. Xth International Conference on Soil Mechanics and Foundation Engineering. Stockholm 1981. Vol. 3, pp. 587-594. Brandl, H. (1999). Long-term behaviour of soils stabilised with lime and with cement. Proc. Geotechnics for Developing Africa, Durban 1999, pp. 219-232. Broms, B., 1984. Stabilization of soil with lime columns. Design Handbook, 3rd Edition, Lime Column AB. Broms, B. & Boman, P., 1977. Stabilization of soil with lime columns, Design Handbook, 1st Edition, Division of Soil & Rock Mechanics, Royal Institute of Technology, Stockholm. Brunauer, S., Emmett, P.H. & Teller, E. (1938). Adsorption of gasses in multimolecular layers. Journal of American Chemical Society 60:309- 319. Buchwald, A., Kaps, C. & Hohmann, M. (2003). Alkali-activated binders and pozzolan cement binders – compete binder reactions or two sides of the same story? Proc. 11th International Congress on the Chemistry of Cement. Durban 2003, pp. 1238-1246.
68 Strength of stabilised soft soil References
Carlsten, P. & Ekström, J. (1995). Lime and lime cement columns. Guide for project planning, construction and inspection. Swedish Geotechnical Society Report 4:95E. Carlsten, P. (2000). Lime and lime cement columns. Guide for project planning, construction and inspection. Swedish Geotechnical Society Report 2:2000. (In Swedish) CEN standard (1989). CEN standard EN196-6:1989. Methods of testing cement – Part 6: Determination of fineness. European Committee for standardization (CEN), Brussels. CEN standard (2000). CEN standard EN197-1:2000. Cement – Part 1: Composition, specifications and conformity criteria for common cements. European Committee for standardization (CEN), Brussels. CEN standard (2001). CEN standard EN459-1:2001. Building lime – Part I: Definitions, specifications and conformity criteria. European Committee for standardization (CEN), Brussels. Chew, S.H., Kamruzzaman, A.H.M. & Lee, E.H. (2004). Physicochemical and engineering behavior of cement treated clays Journal of Geotechnical and Geoenvironmental Engineering, Vol. 130, No. 7, July l, 2004. pp. 696- 706. Christensen, J., Schjønning, E. & Foged, N. (1992). Comparison of clay till strength parameters using British Standard and Danish test procedures. Proc. Nordic Geotechnical Meeting, NGM-92, Aalborg, 1992, Vol. 1/3, pp. 69-74. Cortellazzo, G. & Cola, S. (1999). Geotechnical characteristics of two Italian peats stabilized with binders. Proc. of the International Conference on Dry Mix Methods for Deep Soil Stabilization. Stockholm 1999. den Haan, E. (1998). Cement based stabilizers for Dutch organic soils. Proc. Problematic soils, Sendai 1998. pp. 53-56. Yanagisawa, Moroto & Mitachi (eds). Balkema, Rotterdam. Dumbleton, M.J. (1962). Investigations to assess the potentialities of lime stabilization in the United Kingdom. Road Research Technical Paper No. 64. Edstam, T., Ekström, J. Hallingberg, A. & Nilsson, L. (2004). Control of lime-cement columns in the valley of the river Göta-älv. Proc. Nordic Geotechnical Meeting, NGM 2004, Ystad, 2004, Vol. 1, pp. D-43-54. (In Swedish) Eriksson, H., Gunter, J. & Ruin, M. (2005). MDM combines the advantages of dry and wet mixing. Proc. International Conference on Deep Mixing, Stockholm 2005, Vol. 1.2, pp. 509-520.
Strength of stabilised soft soil 69 References
Esrig, M.I. (1999). Keynote lecture: Properties of binders and stabilized soils. Proc. of the International Conference on Dry Mix Methods for Deep Soil Stabilization, Stockholm 1999, pp. 67-72. ETC5 (1998). Recommendations of the ISSMGE for geotechnical laboratory testing. Beuth Verlag, Berlin. EuroSoilStab (2002). Development of design and construction methods to stabilise soft organic soils. Design guide soft soil stabilisation. CT97-0351. Project No. BE-96-3177, Industrial & Materials Technologies Programme (Brite- EuRam III), European Commission. Fredlund, D. & Rahardjo, H. (1993). Soil mechanics for unsaturated soils. John Wiley & Sons, Inc, New York. Hammond, A.A. (1981). Measurement of autogenous healing in stabilized soils. Proc. Xth International Conference on Soil Mechanics and Foundation Engineering. Stockholm 1981. Vol. 3, pp. 671-676. Hartlén, J. & Wolski, W. (1996). Embankments on organic soils. Eds: Hartlén, J., & Wolski, W.. Developments in geotechnical engineering, 80. Elsevier, Amsterdam. Hayashi, H., Nishikawa, J., Kanta Ohishi, K. & Terashi, M. (2003). Field observation of long-term strength of cement treated soil. Proc. 3rd International Conference on Grouting and Ground Treatment, New Orleans 2003, pp. 598-609. Hayashi, Y., Suszuki, A., Kitazono, Y. & Teraoka, K. (2005). Consolidation characteristics of the cement-treated soil in a fresh state. Proc. International Conference on Deep Mixing, Stockholm 2005. Vol. 1.1, pp. 63-67. Hebib, S. & Farrell, E. (1999). Some experience of stabilising Irish organic soils. Proc. International Conference on Dry Mix Methods for Deep Soil Stabilization. Stockholm 1999. Horpibulsuk, S., Miura, N. & Nagara, T. S. (2003). Assessment of strength development in cement-admixed high water content clays with Abrams’ law as a basis. Géotechnique, Vol. 53, No. 4, pp. 439-444. Horpibulsuk S., Miura, N. & Bergado, D. T. (2004). Undrained shear behavior of cement admixed clay at high water content . Journal of Geotechnical and Geoenvironmental Engineering, Vol. 130, No. 10, pp. 1096-1105. Högberg, E. (1983). Durability. Nordic Seminar on Deep Stabilisation. Nordforsk, Dipoli 1983. (In Swedish) Ingles, O.G. & Metcalf, J.B. (1972). Soil stabilization. Butterworths Pty. Ltd, Australia.
70 Strength of stabilised soft soil References
Jamiolkowski, M., Ladd, C.C., Germaine, J.T. and Lancelotta, R. (1985). New developments in field and laboratory testing of soils. Theme Lecture. Proc. 11th International Conference on Soil Mechanics and Foundation Engineering, San Francisco 1985,Vol. 1, pp. 57–153. Janz, M. & Johansson, S.E. (2001). /English translation 2002./ The function of different binding agents in deep stabilization. Swedish Deep Stabilisation Research Centre, Report No. 9. Johansson, S.E. (1997). Cement. Concrete handbook, Materials. 2nd Edition. AB Svensk Byggtjänst, Stockholm 1997, pp. 33-67. (In Swedish) Karlsson, R. (1963). On cohesive soils and their flow properties. Swedish Geotechnical Institute, Reprints and preliminary reports, No. 5. Kasama K., Ochiai, H. & Yasufuku, N. (2000). On the stress-strain behaviour of lightly cemented clay based on an extended critical state concept. Soils and Foundations, Vol. 40, No.5, pp. 37-47. Kawasaki, T., Saitoh, S., Suzuki, Y.& Babasaki, R. (1984). Deep mixing method using cement slurry as hardening agent. Seminar on Soil Improvement and Construction Techniques in Soft Ground. Singapore 10-11 January 1984, pp. 17-38. Kenney, C. (1959). Geotechnical properties of glacial lake clays. Discussion. Journal of Soil Mechanics and Foundation Division, ASCE, Vol. 85, SM3. pp. 67-79. Kitazume, M. & Terashi, M. (2002). The Deep Mixing Method – Principle, design and construction. Ed. Coastal Development Institute of Technology (CDIT), Japan. A.A. Balkema Publishers, Tokyo. Kivelö, M. (1998). Stabilization of embankments on soft soil with lime/cement columns. Ph.D. Thesis, Deptartment of Civil and Environmental Engineering, Royal Institute of Technology, Stockholm. Kujala, K. (1983). The use of mineral additions together with lime deep stabilisation of soils. Proc. Nordic Seminar on Deep Stabilisation. Nordforsk, Dipoli 1983. (In Swedish) Kwan, S. & Bouazza A. (2002). Deformation properties of a treated soft silty clay. Proc. 4th International Conference on Ground Improvement Techniques: Kuala Lumpur 2002, pp. 457-463. Ladd, C.C. & Foott, R. (1974). New Design Procedure for Stability of Soft Clays. ASCE, Journal of the Geotechnical Engineering Division, Vol. 100, No. GT7, pp. 763-786. Lahtinen, P., Jyrävä, H. & Kuusipuro, K. (1999). Development of binders for organic soils. Proc. of the International Conference on Dry Mix Methods for Deep Soil Stabilization. Stockholm 1999.
Strength of stabilised soft soil 71 References
Landva , A.O. & Pheeney, P.E. (1980). Peat fabric and structure. Canadian Geotechnical Journal, Vol. 17, No. 3, pp.416-435. Larsson, R. (1977). Basic behaviour of Scandinavian soft clays. Swedish Geotechnical Institute, Report No. 4, Linköping. Larsson, R. (1990). Behaviour of organic clay and gyttja. Swedish Geotechnical Institute, Report No. 38, Linköping. Larsson, R. (2000). Investigations and load tests in clay till. Swedish Geotechnical Institute, Report No. 59. Larsson, R., Bengtsson, P.-E. & Eriksson, L. (1997). Prediction of settlements of embankments on soft, fine-grained soils – calculation of settlements and their course with time. Swedish Geotechnical Institute, Information No. 13E. Larsson, R. & Åhnberg, H. (2005). On the evaluation of undrained shear strength and preconsolidation pressure from common field tests in clay.... Canadian Geotechnical Journal, Vol. 42, No. 4, pp. 1221-1231. Larsson, S. (2003). Mixing processes for ground improvement by Deep Mixing. Doctoral Thesis. Division of Soil and Rock Mechanics, Royal Institute of Technology, Stockholm. Lindh, P. (2004). Compaction- and strength properties of stabilised and unstabilised fine-grained tills. Doctoral Thesis. Division of Soil Mechanics and Foundation Engineering, Lund Institute of Technology, Lund University. Lee, K. Chan D. & Lam K. (2004). Constitutive model for cement treated clay in a critical state frame work. Soils and Foundations, Vol. 44, No.3, pp. 69-77. Lorenzo, G. & Bergado D. (2002). Fundamental parameters of cement- admixed clay—New approach. Journal of Geotechnical and Geoenvironmental Engineering, Vol. 130, No. 10, pp. 1042-1050. Lowe, J. & Johnson, T.C. (1960). Use of back pressure to increase degree of saturation of triaxial test specimens. Proc. Research Conference on Shear Strength of Cohesive Soils. Boulder 1960. pp. 819-836. Löfroth, H. (2005). Properties of 10-year-old lime-cement columns. Proc. International Conference on Deep Mixing, Stockholm 2005, Vol. 1.1, pp 509-127. Mayne, P. W. (1988). Determining OCR in clays from laboratory strength. Proceedings ASCE, Journal of Geotechnical Engineering, Vol. 114, No. GT1, pp. 76–92. Nagaraj, T.S., Miura, N., Yaligar P.P. & Yamadera A. (1996) Predicting strength development by cement admixture based on water content. Proc.
72 Strength of stabilised soft soil References
2nd International Conference on Ground Improvement Geosystems – IS Tokyo’96. Grouting and Deep Mixing, Tokyo, 1996. Vol. 1, pp. 431-436. Nygren, M. & Welander, A.-S. (1991). Estimates of properties of lime-cement columns. Master’s thesis 91/5, Royal Institute of Technology, Stockholm. (In Swedish) Okumura, T. (1996) Deep Mixing Method of Japan. Proc. 2nd International Conference on Ground Improvement Geosystems – IS Tokyo’96. Grouting and Deep Mixing, Tokyo, 1996, Vol. 2, pp. 879-887. Pihl, H. & Kuusipuro, K. (2004). Temperature increase at hydration of binders. Internal work report. Nordkalk AB Porhaba, A., Shibuya, S. & Kishida, T. (2000). State of the art in deep mixing technology. Part III geomaterial characterization. Ground Improvement, Vol. 4, No. 3, pp. 91-110. Robson T.D. (1962). High-alumina cements and concretes. Contractors Record Ltd, London 1962. Rotta, G.V. Consoli, N.C., Prietto, P.D.M., Coop, M.R. & Graham. J. (2003). Isotropic yielding in an artificially cemented soil cured under stress. Géotechnique, Vol. 53, No. 5, pp. 493-501. Roy, N. & Sarathi, P. (1976). Strain rate behavior of compacted silt. Journal of the Geotechnical Engineering Division, ASCE, Vol. 102, No. GT4, pp. 347-360. Ruff, C.G. & Ho, C. (1966). Time-temperature strength-reaction product relationships in lime-bentonite-water mixtures. Highway Research Record No. 139, Highway Research Board. pp. 42-60. Saitoh, S., Suzuki, Y. & Shirai, K. (1985). Hardening of soil improved by the deep mixing method. Proc. 11th International Conference on Soil Mechanics and Foundation Engineering, Vol. 3. pp. 1745-1748. SGI (1997). SD SJE 1.1 Influence of type of binder, SGI project 1-9710-504, SD SJE 1.2 Quantity of type of binder, SGI project 1-9712-646, SD SJE 1.5 Influence of consolidation stresses, SGI project 1-9712-680, SD SJE 2 Long term properties, SGI project 1-9712-682. (In Swedish) SGI (1999). Procedures for triaxial testing of stabilised organic soils, SGI project 1-9906-382. Sheahan, T.C., Ladd, C.C. & Germaine, J.T. (1996). Rate-dependent undrained shear behavior of saturated clay. Journal of Geotechnical Engineering, ASCE, Vol. 122, No. 2, pp. 99-108. Sherwood, P.T. (1993). Soil stabilisation with cement and lime. State-of-the-art review. Transportation Research Laboratory, HMSO London.
Strength of stabilised soft soil 73 References
Sivapullaiah, P.V., Sridharan A. & Bhaskar Raju K.V. (2000). Role of amount and type of clay in the lime stabilization of soils. Ground Improvement, Vol. 4, pp. 37-45. Sjögren, M., Carlsten, P. & Elander, P. (1994). Determination of the permeability of residual products and soil, in situ and in the laboratory. Nordtest Technical Report 254. Steenfelt, J. & Foged, N. (1992). Clay till strength – SHANSEP and CSSM. 11. Proc. Nordic Geotechnical Meeting, NGM-92, Aalborg, 1992, Vol. 1/3, pp. 81-86. Steensen-Bach, J. O., Bengtsson, P.-E. & Rogbeck, Y. (1996) Large scale triaxial tests on samples from lime-cement columns. Proc. XIIth Nordic Geotechnical Conference NGM-96, Reykjavik 1996, Vol. 1, pp. 135-146. Swedish Standard (1989a). SS 02 71 14, Geotechnical tests – Bulk density. Swedish Standards Institute, SIS Förlag AB, Stockholm (In Swedish) Swedish Standard (1989b). SS 02 71 15, Geotechnical tests – Grain density and specific gravity. Swedish Standards Institute, SIS Förlag AB, Stockholm (In Swedish) Swedish Standard (1989c). SS 02 71 16, Geotechnical tests – Water content and degree of saturation. Swedish Standards Institute, SIS Förlag AB, Stockholm (In Swedish) Swedish Standard (1990a). SS 02 71 20, Geotechnical tests – Cone liquid limit. Swedish Standards Institute, SIS Förlag AB, Stockholm (In Swedish) Swedish Standard (1990a). SS 02 71 21, Geotechnical tests – Plastic limit. Swedish Standards Institute, SIS Förlag AB, Stockholm (In Swedish) Swedish Standard (1991). SS 02 71 26, Geotechnical tests - Compression properties - Oedometer test, CRS-test - Cohesive soil. Swedish Standards Institute, SIS Förlag AB, Stockholm (In Swedish) Swedish Standard (1992a). SS 02 71 28, Geotechnical tests – Shear strength – Unconfined compression tests – Cohesive soil. Swedish Standards Institute, SIS Förlag AB, Stockholm (In Swedish) Swedish Standard (1992b). SS 02 71 29, Geotechnical tests - Compression characteristics - Oedometer test incremental loading - Cohesive soil. Swedish Standards Institute, SIS Förlag AB, Stockholm (In Swedish) Tatsuoka, F. & Kobayashi, A. (1983). Triaxial strength characteristics of cement-treated soft clay. Proc. 8th European Conference on Soil Mechanics and Foundation Engineering, Helsinki 1983, Vol. 1, pp. 421-426. Taylor, H.F.W. (1997). Cement chemistry. Thomas Telford, London 1997.
74 Strength of stabilised soft soil References
Terashi, M. & Tanaka, H. (1983). Settlement analysis for deep mixing methods. Proc. 8th European Conference on Soil Mechanics and Foundation Engineering, Helsinki 1983, pp. 955-960. Terashi, M., Tanaka, H. & Okumura, T. (1979). Engineering properties of lime-treated marine soils and D. M. Method. Proc. 6th Regional Conference on Soil Mechanics and Foundation Engineering. Singapore 1979, Vol. 1, pp. 191-194. Terashi, M., Tanaka, H., Mitsumoto, T., Niidome, Y. & Honma, S. (1980). Fundamental properties of lime and cement treated soils (2nd report). Report of the Port and Harbour Research Institute, 19(1), pp. 33-62. (In Japanese/English summary). Terzaghi, K. & Peck, R.B. (1967). Soil mechanics in engineering practice, 2nd Edition, John Wiley & Sons, Inc., New York. Thompson, M.R. and Dempsey, B.T. (1969). Autogenous healing of lime-soil mixtures. Highway Research Board. Highway Research Record No.263., pp. 1-7. Torstensson, B-A. (1977). Time-dependent effects in the field vane test. Proc. International Symposium on Soft Clay. Bangkok 1977, pp. 387-397. TRB (1987). Lime stabilisation. Reactions, properties, design, and construction. State-of-the-Art Report 5. Transportation Research Board, National Research Council, Washington, D.C. Tremblay, H., Leroueil, S. & Locat, J. (2001). Mechanical improvement and vertical yield stress prediction of clayey soils from eastern Canada treated with lime or cement. Canadian Geotechnical Journal, Vol. 38, pp. 567-579. Uddin K. & Buensuceso B. R. (2002). Lime treated clay: Salient Engineering Properties and a Conceptual Model. Soils and Foundations, Vol. 42, No.5, pp. 79-89. von Post, L., 1924, Das genitische system der organogenen bildungen Schwedens. Mémoires Nomenclat. et Classific. Sols, Comité int. Pédologie, 4 Commission No. 22, pp. 287-304. Yoshizawa, H., Okumura, R, Hosoya, Y., Sumi, M. & Yamada, T. (1996). JGS TC Report: Factors affecting the quality of treated soil during execution of DMM . Proc. 2nd International Conference on Ground Improvement Geosystems – IS Tokyo’96. Grouting and Deep Mixing, Tokyo 1996, Vol. 2, pp. 931-937. Åhnberg, H. (1996). Stress dependent parameters of cement and lime stabilised soils. Proc. 2nd International Conference on Ground Improvement Geosystems – IS Tokyo’96. Grouting and Deep Mixing, Tokyo 1996, Vol. 1, pp. 387-392.
Strength of stabilised soft soil 75 References
Åhnberg, H. & Holm, G. (1987). On the influence of curing temperature on the strength of lime and cement stabilised soils. Swedish Geotechnical Institute, Report No. 30, pp. 93-146. (In Swedish) Åhnberg, H. & Holm, G. (1999). Stabilisation of some Swedish organic soils with different types of binder. Proc. International Conference on Dry Mix Methods for Deep Soil Stabilization, Stockholm 1999, pp. 101-108. Åhnberg, H. & Pihl, H. (1998). Effects of different quicklimes on clays – a preliminary study. Proc. 2nd International Conference on Ground Improvement Techniques, Singapore 1998, pp. 47-54. Åhnberg, H., Bengtsson, P-E. & Holm, G. (1989). Prediction of strength of lime columns. Proc. XII International Conference on Soil Mechanics and Foundation Engineering. Rio de Janeiro 1989. Åhnberg, H., Johansson, S-E., Retelius, A., Ljungkrantz, C., Holmqvist, L. & Holm, G. (1995). Cement and lime for deep stabilisation of soil. Swedish Geotechnical Institute, Report No. 48. (In Swedish)
REFERENCES EXCLUSIVELY IN THE PAPERS
ASTM standard (1990). ASTM standard D 5084– 90. Standard test method for measurement of hydraulic conductivity of saturated porous materials using flexible wall permeameter. ASTM standard (1994). ASTM standard C 618 – 94a. Standard specification for coal fly ash and raw or calcined natural pozzolan for use as a mineral admixture in Portland cement concrete. ASTM standard (1995). ASTM standard C 989 – 94a. Standard specification for ground granulated blast furnace slag for use in concrete and mortars. Berre, T. (1997). Methods for consolidated triaxial compression tests on water saturated soils. European Technical Committee, Document No. ETC5-F1.97, pp. 75–86. Black, D.K. & Lee, K.L. (1973). Saturating laboratory samples by a back pressure. Journal of the Soil Mechanics and Foundation Div., ASCE, Vol. 99, No. SM1, pp. 75-93. Broms, B. (1999). Can lime/cement columns be used in Singapore and Southeast Asia? 3rd GRC Lecture, Singapore, 19 Nov. 1999. Broms, B. (2000). Lime and lime/cement columns. Summary and visions. Keynote lectures. Proc. Nordic Geotechnical Meeting NGM-2000 and 4th GIGS, Helsinki 2000, pp. 43-93.
76 Strength of stabilised soft soil References
Bruce, D., Bruce, M.E. & DiMillio, A.F. (1999). Dry mix methods: A brief overview of international practice. Proc. International Conference on Dry Deep mixing Methods for deep Soil Stabilization, Stockholm 1999, pp. 15- 25. Casagrande, A. & Wilson, S.D. (1951). Effect of rate of loading on the strength of clays and shales at constant water content. Géotechnique, Vol. 2, No. 3, pp. 251-263. Consoli, N.C., Rotta, G.V. & Prietto, P.D.M. (2000). Influence of curing under stress on the triaxial response of cemented soils. Géotechnique, Vol. 50, No. 1, pp. 99-105. DeGroot, D.J. & Sheahan, T.C. (1995). Laboratory methods for determining engineering properties of overconsolidated clays. Engineering properties and practice in overconsolidated clays,. Transportation Research Record, No. 1479, pp. 17-25. Hartlén J. (1974). Strength properties and bearing capacity of clay till in southern Sweden. PhD Thesis, Department of Geotechnical Engineering, Chalmers University of Technology, Gothenburg. (In Swedish) Hebib, S. & Farrell, E. (2000). EuroSoilStab, Task 2: Laboratory and numerical analyses for field tests. Trinity College, Dublin. Internal Report, EuroSoilStab project. Hoikkala S., Leppänen M. and Lahtinen P. (1996). Mass stabilization of peat in road construction. Proc. Nordic Geotechnical Meeting NGM, Reykjavik 1996, pp. 391-395. Holm, G. & Åhnberg, H. (1987). Use of lime-fly ash for deep stabilisation of soil. Swedish Geotechnical Institute, Report No. 30. (In Swedish) Holm, G. (2003). State of practice in dry deep mixing methods. Proc. 3rd International Conference on Grouting and Ground Treatment, New Orleans 2003, pp. 145-163. Holmén, M. (2003). Advanced triaxial testing of clay till. Licentiate Dissertation, Division of Soil Mechanics and Foundation Engineering. Lund Institute of Technology, Lund University. Håkansson K., Johansson S.-E., Bengtsson P.-E., Holm G. &. Åhnberg H. (2000). Task 2.2 and 7. Results of chemical analysis of soils and stabilised soils – Test sites Dömle Mosse and Holma Mosse, Sweden. National Report, Part 2. Internal Report, EuroSoilStab project, SGI 3-9705-239 and 3-9708-395. Högberg, E. (1979). The role of lime in the stabilisation process. Seminar on the lime column method in practice. Soil and Rock Department, Royal University of Technology, Stockholm.
Strength of stabilised soft soil 77 References
Ishihara K., Tatsuoka F. & Yasuda S. (1975). Undrained deformation and liquefaction in sand under cyclic stress. Soils and Foundations, Vol. 15, No. 1, pp. 29-44. Jelisic N. & Leppänen M. (1999). Mass stabilization of peat in road and railway construction. Proc. International Conference on Dry Mix Methods for Deep Soil Stabilization. Stockholm 1999, pp. 59-64. Jelisic N. (1999). Mass stabilisation. Licentiate Thesis. Lund Institute of Technology, Lund university, Lund. (In Swedish) Karlsson, R. (1981). Consistency limits. Performance and interpretation of laboratory investigations, part 6. Byggforskningsrådet D9:1981, Stockholm. Kezdi, A. (1979). Stabilised earth roads. Elsevier 1979. Klemm, W. & Adams, L. (1990). An investigation of the formation of carboaluminates. Carbonate additions to cement. ASTM, Special Technical Publications STP 1064, pp. 60-72. Kohata, Y., Maekawa, H., Muramuto, K., Yajima, J. & Babasaki, R. (1996). JGS-TC Report: Deformation and strength properties of DM cement- treated soils. Proc. 2nd International Conference on Ground Improvement Geosystems, Grouting and Deep Mixing, Tokyo 1996, Vol. 2, pp. 905-911. Lacasse, S. & Berre, T. (1988). Triaxial testing methods for soils, Advanced triaxial testing of soil and rock, ASTM Special Technical Publications STP 977, pp. 264-289. Lade, P. & Hernandez, S. (1977). Membrane penetration effects in undrained tests. Journal of the Geotechnical Engineering Division, Vol. 103, No. GT2, pp. 109-125. Lambe, W. (1951). Soil testing for engineers. John Wiley & Sons, Inc., New York. Landva, A.O. & La Rochelle, P. (1983). Compressibility and shear characteristics of Radforth Peats. Testing of Peats and Organic Soils, ASTM Special Technical Publications STP 820, pp. 157-191. Larsson, R. (1980). Undrained shear strength in stability calculation of embankments and foundations on soft clays. Canadian Geotechnical Journal, Vol. 17, No. 4, pp. 591–602. Lee, K.L., Morrison, R.A. & Haley, S.C. (1969). A note on the pore pressure parameter B. Proc. 7th International. Conference on Soil Mechanics and Foundation Engineering, Mexico City 1969, Vol. 1, pp. 231-238. Leroueil, S. & Barbosa, P.S. de A. (2000). Combined effect of fabric, bonding and partial saturation on yielding of soils. Proc. Asian Conference on Unsaturated Soils, Singapore, pp. 527-532.
78 Strength of stabilised soft soil References
Locat, J., Tremblay, H. & Leroueil, S. (1996). Mechanical and hydraulic behaviour of a soft inorganic clay treated with lime. Canadian Geotechnical Journal Vol. 33, No. 4, pp. 654-669. Mitchell, J.K. (1976). Fundamentals of soil behavior. John Wiley & Sons Inc., 1976. Poussette K., Mácsik J., Jacobsson A., Andersson R. and Lahtinen P. (1999). Peat soil samples stabilised in laboratory – Experiences from manufacturing and testing. Proc. International Conference on Dry Mix Methods for Deep Soil Stabilization, Stockholm 1999, pp. 85-92. Richardson, A.M. & Whitman, R.V. (1963). Effect of strain-rate upon undrained shear resistance of a saturated remoulded fat clay. Géotechnique, Vol. 13, No. 4, pp. 310-324. Sivapullaiah P.V., Prashanth J.P. & Sridharan A. (1998a). Delay in compaction and importance of the lime fixation point on the strength and compaction characteristics of soil. Ground Improvement Vol. 2, No. 1, pp. 27-32. Sivapullaiah P.V., Prashanth J.P. & Sridharan A. (1998b). Effect of delay between mixing and compaction on strength and compaction parameters of fly ash. Geotechnical Engineering Bulletin, Vol. 7, No. 4., pp. 277-285. Skempton, A.W. (1954). The pore-pressure coefficients A and B. Géotechnique, Vol. 4, No. 4, pp.143-147. Suzuki, Y. (1982). Deep chemical mixing using cement as hardening agent. Symposium on Soil and Rock Improvement, Bangkok 1982, pp. B.1.1-1.24. Sweeney D.A., Wong D.K.H. and Fredlund D.G. (1988). Effect of lime on highly plastic clay with special emphasis on ageing. Transportation Research Record No. 1190. Taylor, D.W. (1948). Fundamentals of Soil Mechanics, John Wiley & Sons, New York. Terashi, M. (2003). The state of practice in deep mixing methods. Proc. 3rd International Conference on Grouting and Ground Treatment, New Orleans 2003, pp. 25-49. Terashi, M., Tanaka, H., Mitsumoto, T., Honma, S. & Ohashi, T. (1983): Fundamentals of lime and cement treated soils (3rd report). Report of the Port and Harbour Research Institute. Vol. 22, No. 1, pp. 69 – 96. (In Japanese) Wood, D. M. (1990). Soil behaviour and critical state soil mechanics. Cambridge University Press, Cambridge 1990. Yamamoto, T., Suzuki, M., Okabayashi, S., Fujino, H., Taguchi, T. & Fujimoto, T. (2002). Strength and deformation characteristics of cement-
Strength of stabilised soft soil 79 References
stabilised soil cured under an overburden pressure. Proc. 4th International Conference on Ground Improvement Techniques, Kuala Lumpur 2002, pp. 761-766. Yin, J.-H. (2001). Properties and behaviour of raw sludge mixed with PFA and lime. Geotechnical Testing journal (ASTM), Vol. 24, No. 3, pp. 299- 307. Zhou, Cheng, Yin, J.-H. and Ming, J.-P. (2002) Bearing capacity and settlement of weak fly ash ground improved using lime-fly sash or stone columns. Canadian Geotechnical Journal, Vol. 39, No. 3, pp. 585-596. Zhu, F., Clark, J.I. & Paulin, M.J. (1995). Factors affecting at-rest lateral stress in artificially cemented sands. Canadian Geotechnical Journal, Vol. 32, No. 2, pp.195-203. Åhnberg, H. (2003). Measured permeabilities in stabilised Swedish soils. Proc. 3rd International Conference on Grouting and Ground Treatment, New Orleans 2003, pp. 622-633. Åhnberg, H. (2004). Effects of back pressure and strain rate used in triaxial testing of stabilised organic soils and clays. Geotechnical Testing Journal. Vol. 27, No. 3, pp 250-259. Åhnberg, H. (2006). Effects of consolidation stresses on the strength of some stabilised Swedish soils. Ground Improvement, Vol. 10, No. 1, pp. 1-13. Åhnberg, H., Bengtsson, P.-E. & Holm, G. (2000). Task 2.2 and 7. Laboratory tests of mixtures – Test sites Dömle Mosse and Holma Mosse, Sweden. National Report, Part 1. Internal Report, EuroSoilStab project, SGI 3-9705-239 and 3-9708-395. Åhnberg, H., Bengtsson, P-E. & Holm G. (2001). Effect of initial loading on the strength of stabilised peat. Ground Improvement, Vol. 5, No. 1, pp. 35- 40. Åhnberg, H., Johansson, S.-E., Pihl, H. & Carlsson, T. (2003). Stabilising effects of different binders in some Swedish soils. Ground Improvement, Vol. 7, No. 1, pp 9-23.
80 Strength of stabilised soft soil Paper I Reprinted with permission from Ground Improvement, Thomas Telford Ltd, 1 Heron Quay, London E14 4JD, UK. From Vol. 7, No.1.
250 Gyttia: Holma
w L
200 w P
w 150 : % w Clay 1: Linköping 100 w L
w 50 w P
w w w P L 0 0 50 100 150 200 250 300 350 400 Time: days
2000 7 days Clay 1, 28–91 days 100 kg/m3 28 days Clay 2, 28–91 days cl sl 91 days l 365 days Gyttja 1, 28–91 days 1500
c cs cf
1000 /2: kPa uc q
500
0 l c c 2 c 3 llb l2 l4 llb4 c005 l50/50 l75/25 s50/50 cf50/50 cf75/25 cl50/50 cl75/25 sl50/50sl75/25 cs50/50c 2 cs25/75 cpf50/50 cpf75/25 c 2 c 2 slf33/33/33 Stabilising agent
2000 Clay 1-7d 3 Clay 1-28d 100 kg/m Clay 1-91d Clay 1-365d Clay 2-7 days Clay 2-28 days Clay 2-91 days 1500 Clay 2-365 days Gy-7 days Gy-28 days Gy-91 days Gy-365 days
: kPa : 1000
uc 1 year q
91 days
500 28 days
7 days
0 l c 2 c l4 llb c005 s50/50 l75/25 l50/50 2 cf75/25 cl75/25 2 sl75/25cl50/50cf50/50 sl50/50 cs50/50c 2 c cpf75/25 cpf50/50c cs25/75 slf33/33/33 Stabilising agent 1500 Present investigation 28 days Dömle peat-200 kg/m3 - EuroSoilStab Dömle peat-200 kg/m3 (series II)- EuroSoilStab Dömle gyttja-200 kg/m3 - EuroSoilStab
1000
Clay 1, clay 2 and gyttja 1 - 100 kg/m2
: kPa :
uc q
500
0 c l c 2 l4 llb c005 s50/50 l75/25 /75/25 l50/50 2 cf75/25 cl75/25 2 sl75/25 cl50/50cf50/50 sl50/50 cs50/50c 2 c c 2 cpf50/50c cs25/75 slf33/33/33 Stabilising agent
Paper II Reprinted with permission from the Swedish Deep Stabilization Research Centre. From Proceedings of the International Conference on Deep Mixing, Stockholm 2005, Vol. 1.1. Increase in strength with time in soils stabilised with different types of binder in relation to the type and amount of reaction products
Åhnberg, H. Swedish Geotechnical Institute, SE-581 93 Linkoping, Sweden [email protected],
Johansson, S.-E. Cementa AB, Box 30022, SE-200 61 Malmö, Sweden [email protected]
ABSTRACT: For design of deep mixing, the strength of different mixtures of soils and binders is normally investigated at various time intervals after mixing. However, limited design time often leads to tests being performed on only one or two occasions. The full strength development is thus not usually determined or taken into account. The variation in strength increase when using different types of binder has been studied for three stabilised soils up to two years after mixing in the laboratory. The study shows that there can be a considerable increase in strength also long after mixing. The results provide increased insight into the effect of different binders on the increase in strength of stabilised soils.
1 INTRODUCTION have been made in this study, in order to gain a clearer The deep mixing method has been used to improve the understanding of the processes involved. In this paper, data strength and compression properties of soft soils for nearly on the increase in soil strength when using different types thirty years. The method is well established in foremost of binders are presented and a relation between strength and Japan and the Nordic countries and has recently been chemical reaction products is demonstrated. introduced in an increasing number of countries (Bruce, The binders used in the present study were various 1999; Terashi, 2003; Holm, 2003). An increasing number combinations of cement, lime, slag and fly ash. The tests of binders is also being employed, although the most were part of a series of investigations concerning the common binders used are still lime and cement. For design, properties of soft soils stabilised with different types of series of laboratory tests on stabilised samples of different binder (Åhnberg et al., 2003; Åhnberg, 2003, 2004). composition are normally carried out to provide a basis for the final choice of type and quantity of binder to be used. However, the number of tests is normally limited due to 2 SCOPE OF THE INVESTIGATION restraints on time and cost. Most often, the increase in strength with time is not fully investigated and time effects 2.1 Types of soil are not incorporated into the design calculations. This may Three types of soft soil were chosen for the investigation. lead to uneconomic solutions for a construction and Two of the soils were soft clays of types that had previously underestimation of its long-term behaviour. been shown to give diverse stabilising results in earlier The increase in strength with time for various stabilised investigations. The clays have fairly similar geotechnical soils has been investigated in a large number of projects, properties according to the routine investigations, but are although most of these investigations are limited to a few from different geological deposits. One of them is a post- types of binder. Rough relations between strength and time glacial clay from an area near Linköping in the eastern part have been reported, mainly for the common binders lime of Sweden. It was deposited during periods of and cement (e.g. Porbaha et al., 2000; Kitazume & Terashi, sedimentation in alternating lake and brackish sea water. 2002; Horpibulsuk et al., 2003). Questions still remain, The other clay is a marine post-glacial clay containing however, regarding the stabilisation effect of the different occasional shells, from a site near Löftabro on the Swedish types of binders commonly discussed today. Further studies west coast. The Löftabro clay has a somewhat higher water of these are necessary in order to improve the basis for the content and lower undrained shear strength than the assessment of strength increase with time in design. Linköping clay. Both clays are high-plastic with liquid In order to study the stabilising effect of different limits of approximately 70% and organic contents of binders, unconfined compression tests have been performed approximately 1%. The third soil used in the investigation on samples of stabilised clay and organic soil up to 2 years was an organic soil, a gyttja from Holma Mosse in the after mixing in the laboratory. Since the different eastern part of Sweden. The stabilisation effects of different compounds produced by different binders in the soil are binders in this soil have been studied previously in the EC- likely to influence the increase in strength to a large degree, funded project EuroSoilStab (Åhnberg et al., 2000). The rough estimates of the type and amount of these bonding Holma gyttja is a clayey gyttja with a water content of
Deep Mixing '05 195 220% and a liquid limit of 170%. The organic content is Table 1. Properties of the test soils. 10%. Results from laboratory tests of the three soils are presented in Table 1. Linköping Löftabro Holma clay clay gyttja 2.2 Types of binder Depth (m) 3-6 2-5 3-5 In the present investigation, two single binders and four Density (t/m3) 1.55 1.52 1.23 composite binders were used. These were pure cement (c), Specific gravity (t/m3) 2.72 2.73 2.36 pure lime (l), cement–lime (cl), cement–slag (cs), cement– Plastic limit (%) 24 23 64 fly ash (cf) and slag–lime (sl). All the composite binders Water content (%) 78 89 220 were used in proportions of 50:50 (by weight), and cement– Liquid limit1 (%) 70 66 170 fly ash was also studied at the proportions 75:25. All the Undr. shear strength1 (kPa) 15 8 5 binders were used in a total quantity corresponding to 100 Sensitivity1 20 25 10 3 kg per m soil. Organic content (%) 1.0 1.0 10 The cement used in the investigations was the currently Clay content (%) 63 72 80 dominating standard cement in Sweden, a Portland- Chloride content (%) 0.01 0.38 <0.01 limestone cement designated CEM II/A-LL 42.5 R (CEN, Sulphide content (%) 0.05 0.18 0.20 2000). The lime used was the quicklime normally used for pH 7.7 8.7 8.5 deep stabilisation, CL90-Q (CEN, 2001). The slag was a 1Determined by the fall cone test (ETC5, 1998) ground, granulated blast furnace slag with a glass content of 99%. This type of binder has so far been used only to a very limited extent for deep stabilisation of soils in Sweden. The control test. Two and a half years after mixing, the fly ash used was pure coal fly ash from pulverised coal remaining spare sample of several of the test series was also combustion. It was of Class F (ASTM, 2000), having tested. pozzolanic properties. Fly ash has not yet been used for The unconfined compression tests were performed at a deep stabilisation in practice in Sweden, but a number of rate of strain of 1.5%/min on specimens with a height to lime–fly ash test columns have been installed in a research diameter ratio of 2. Prior to strength testing, the specimens project (Holm & Åhnberg, 1987). The composition of the were cut and smoothed to form parallel end surfaces. The different binders is given in Table 2. end platens were lubricated with grease in order to minimise friction at the end surfaces. 2.3 Preparation of samples Samples were prepared by first mixing the binder with the soil for five minutes. The mixtures were then placed in 3 RESULTS AND DISCUSSION layers in ∅50 mm plastic tubes using a momentary compaction pressure of 100 kPa. The tubes were filled to a 3.1 Measured strength increase height of approximately 150 mm. Six to seven samples, The results of the tests showed that the increase in strength including one spare sample, were prepared from each batch with time of the different types of stabilised soils varied of soil-binder mixture. The tubes were sealed with rubber considerably, both with type of binder and with type of soil. lids and stored in a climate-controlled room at 8°C. This This is illustrated in Figure 1, where the results are shown mode of preparation and storage is common procedure in as the mean values obtained from all the unconfined Sweden (Carlsten & Ekström, 1995). compression tests performed on samples stabilised with cement, lime and composite binders in proportion 50:50. 2.4 Testing During first three months after mixing, the most rapid To investigate the increase in strength of the samples, strength increase occurred in the samples stabilised with unconfined compression tests were performed 7, 28, 91 and binders containing cement. In general, the strength of these 364 days after mixing. Duplicate tests were performed 28 samples thereafter levelled off to fairly constant values or and 91 days after mixing, whereas only single tests were showed clearly reduced rates of long-term increase, performed after one week and one year. However, in cases whereas those containing lime exhibited a pronounced of clearly deviating results, a spare sample was used for a
Table 2. Chemical composition of the binders.
CaO SiO2 Al2O3 Fe2O3 MgO K2O Na2O SO3 (%) (%) (%) (%) (%) (%) (%) (%) Lime 93.01 1.4 0.6 0.3 1.0 <0.1 <0.1 <0.1 Cement 61.4 19.9 3.6 2.6 2.8 1.0 0.2 3.3 Slag 32.1 35.2 13.6 0.2 16.8 0.6 0.6 1.8 Fly ash 5.9 54.4 30.5 5.5 1.8 1.2 0.5 0.5 1Available CaO, 92%
196 Deep Mixing '05 3000 3000
Löftabro clay Linköping clay
100 kg/m 3 sl 3 2500 2500 100 kg/m , kPa , kPa uc uc q q h 2000 2000
cl 1500 1500 l
cs cs c c 1000 1000
cl
cf Unconfined compressive strength cf Unconfined compressive strengt compressive Unconfined 500 500 sl
l 0 0 0 100 200 300 400 500 600 700 800 900 0 100 200 300 400 500 600 700 800 900 Time, days Time, days 3000 Holma gyttja 100 kg/m 3 Lime 2500 Cement
, kPa Cement–lime, 50/50 uc
q Cement–fly ash, 50/50 2000 Cement–slag, 50/50 cl Slag–lime, 50/50 sl
1500
1000 c
cs
Unconfined compressive strength strength Unconfined compressive 500 cf Figure 1. Measured compressive strength with time after mixing with cement, lime and composite binders 50:50. 0 0 100 200 300 400 500 600 700 800 900 Time, days
13 Linköping clay-7days long-term increase in strength. The strength increase at 12 92days 360days different times after mixing is shown in Figure 2, where 11 Löftabro clay, 7days 92days the 7-, 91- and 364-day strengths for the different binder 10 360days Holma gyttja, 7days 9 92days Holma gyttja combinations are normalised to the 28-day strength, a 360days testing time often used for the evaluation of suitable 8 binders in design. In samples stabilised with cement, 7 uc-28days cement–slag and cement–fly ash the strength after 7 days /q 6 Löftabro clay uc was about 0.4 to 0.7 times that of 28 days. One year after q 5 mixing with these binders, the strength had increased to 4 3 between 1.2 and 2.8 times that of 28 days. The measured Linköping clay strength in cement–slag stabilised Löftabro clay and 2 Holma gyttja after one year was lower than that after 1 three months. However, partly jagged stress-strain curves 0 c cf50/50 cs50/50 cl50/50 sl50/50 l were obtained in the tests on these samples, indicating Stabilising agent the existence of cracks or other inhomogeneities in the Figure 2. Compressive strength measured at 7, 91 specimens. The increase in strength of the cement- and 364 days relative to the strength at 28-day vs stabilised samples was of the same order as, or slightly type of binder.
Deep Mixing '05 197 higher than, that reported previously for cement Table 3. Abbreviations commonly used in binder stabilised soils (Porbaha et al., 2000; Horpibulsuk et al., chemistry. 2003). With binders containing lime, the long-term increase in strength was more pronounced, although it Abbreviation Chemical product varied considerably with type of soil. Using lime alone or C CaO together with slag or cement gave 7-day strengths of 0.1- H H O 1.0 and 1-year strengths of 1.3-13 times that at 28 days. 2 A Al2O3 Although in many cases there is a distinct break in the S SiO strength vs time curve at one to three months after 2 F Fe2O3 mixing, it is clear from the results that there may be a AFm,t AF mono and tri phases (e.g. ettringite) considerable increase in strength a long time after this. From the variation in short-term and long-term strength increase, it is evident that the various binders act in alumina (A) contained in minerals present in the soil. different ways. The bondings created by the binders are These pozzolanic reactions, which are normally relatively not simply of a different type or strength, they also take slow, due to a restricted accessibility of silica and different times to form. alumina in the soil, may result in the formation of calcium aluminate silicate hydroxide (CASH), calcium 3.2 Estimates of bondings formed by hydration silicate hydroxide (CSH) and/or calcium aluminate processes and pozzolanic reactions influencing hydroxide (CAH) (e.g. TRB, 1987). strength Cement, on the other hand which, apart from C also The chemical reactions involved in the hydration of contains a considerable amount of S, will primarily form different types of cement or lime have been described the C-S-H gel at hydration. Besides CSH, a certain and discussed thoroughly in many papers and textbooks amount of CH is also generated. As in the case of lime, (e.g. Taylor, 1997; Boynton, 1980). The chemical this CH may react with minerals in the soil contributing processes involved in soil stabilisation have also been to an increase in long-term strength by also forming described (e.g. Chew et al., 2004; TRB, 1987; Saitoh et CASH, CSH and/or CAH. Other cement reaction al., 1985; Ingles & Metcalf, 1972, Ruff & Ho, 1966), products which may contribute to the increase in although descriptions of the effects of binders other than strength, although to a lesser degree, are various lime and cement are more scarce in the literature. In aluminate ferrite tri phases, e.g. ettringite, and mono order to improve the basis for the evaluation of the phases (AFt/AFm). effects of different types of binders and possible strength Slag, which is a latent hydraulic cement, will react in increase with time from strength tests performed on much the same way as ordinary cement and lead to the stabilised soil, relations to chemical reactions and formation of similar hydration products. However, since reaction products can be investigated. In analysing the slag as a rule has to be activated by adding some form of test results, rough assessments of the reaction products alkali (Taylor, 1997) it is normally used in combination were made with respect to type and quantity of binder. with other binders, e.g. cement, in order to be effectively Figure 3 shows a rough outline of the principal activated by the CH generated by the latter. The build-up chemical processes taking place and the resulting of reaction products is normally much slower with slag reaction products when mixing common binders with the than with ordinary cement. However, the long-term soil. In the figure, abbreviations are used for the reaction strength of hydrated slag in combination with cement is products. In describing the chemical reactions taking normally higher than that of hydrated cement alone. In place in the hydration processes of cementitious binders, slag, the ratio of C to S is lower than in cement. The a number of abbreviations are normally used for various chemical composition of slag corresponds to that of a compounds. The most common of these abbreviations are cement containing a higher amount of belite, C2S, and shown in Table 3. When mixing binders such as those less alite, C3S, the two principal reactive components in used in the present study with soil, similar types of ordinary cement, the former normally generating a higher bonding are formed. However, the chemical processes long-term strength following hydration. and the main type of reaction products differ somewhat The fly ash acts mainly as a pozzolanic material, i.e. depending on the type of binder used. the silica and alumina react with any CH added to or When mixing quicklime, which contains large formed in the soil after mixing. It is not treated as a amounts of calcium oxide, (C), into the soil, hydration reactive material in itself. However, the silica and will occur as this lime comes into contact with the pore alumina in fly ash are often more easily accessible for water in the soil, resulting in the formation of calcium reactions with any CH added through the binders, hydroxide (CH). As the CH is generated, some of it will compared with the same minerals in the soil. The be adsorbed to the soil particles. Ion exchange with reaction products generated are much the same as those primarily Na+ and K+ ions in the soil will take place, of soils containing silica and alumina, i.e. mainly CASH, leading to modification of the soil into a somewhat drier, CSH and/or CAH. coarser structure due to the slaking processes and a In order to further understand the way in which the flocculation of the clay particles. The CH not consumed various reaction products generated from different types in this process is free to react with the silica (S) and of binder affect the increase in strength with time after
198 Deep Mixing '05 Binders Soil Stabilised soil
Ground Clay Clay CASH,(CSH),(CAH) Quicklime C water CH particles CH minerals
Adsorption, Pozzolanic CSH,(AFm/t) Cement C,S,A,F CH ion CH reactions exchange CASH,(CSH),(CAH)
- - Hydration H
O
reactions OH
Alkaline Slag S,C,A Hydration CSH,(CAH),(CASH) activa- CH tion CH CH
Pozzolanic Fly ash S,A,C CASH,(CSH),(CAH) CH reactions
Figure 3. Rough outline of the principal chemical reactions and reaction products formed by different types of binders. The abbreviations are explained in Table 3. mixing, rough estimates were made of the amount of belite in the cement. Estimates of the quantities of bondings being formed. The amount of reaction products hydration products formed, mainly in the form of CSH generated when adding a quantity of binder and CH, were made according to: corresponding to 100 kg/m3 to the soil was assessed based mainly on the mole weights of the principal 100 kg cement + 30 kg H → 98 kg CSH +32 kg CH (3) chemical elements involved. When using quicklime, the amount of CH generated by slaking can be calculated as The CH may react with the minerals in the soil. follows: Assuming that principally strätlingite is formed, this would result in about 90 kg CASH. The total amount of → 100 kg CaO + 32.1 kg H2O 132.1 kg Ca(OH)2 (1) reaction products formed by the cement can thus be estimated to be about 188 kg. The binder combination The lime used in the study had an available C content of cement–lime (50:50) can be estimated to result in a 92%. Adding 100 kg of this binder would thus generate quantity of reaction products between those of pure about 122 kg CH. Assuming roughly that this CH in turn cement and pure lime. reacts fully with the silica and alumina present in the soil Slag will typically generate about the same types of and that then primarily CASH, expressed in the form of reaction product as cement. However, the water content strätlingite, C2ASH8, is formed, the quantity of the latter is generally lower and the paste somewhat denser than can be calculated according to: that of hydrated Portland cement. The quantity of CH formed by slag is low. When used together with cement, 100 kg CH + 109 kg (A+S) + 73 kg H→ 282 kg CASH (2) the quantity of CH produced has been reported to be even A quantity of 100 kg/m3 of the quicklime may thus lower than if slag had not taken part in the reactions produce approximately 343 kg of reaction products per (Taylor, 1997). The composite binder cement–slag m3 of soil. (50:50) was roughly estimated to generate reaction For hydrated cements, the chemically bound water products according to: content is typically about 28% (Betonghandbok, 1997) to 32% (Taylor, 1997). Apart from the main constituent 100 kg cement–slag + 25 kg H → 110 kg CSH + 15 kg CH (4) C3S2H3, hydrated cement typically contains about 25% CH (e.g. Betonghandbok, 1997; Taylor, 1997). The A total quantity of 152 kg of reaction products was quantity of CH formed varies with the ratio of alite to calculated assuming that the CH in turn reacts with the soil minerals and forms CASH. Using slag together with
Deep Mixing '05 199 lime, was estimated to lead to about 230 kg of reaction example, the amount of contained, unreacted CH present products. in the hydrated cement may affect the porosity of the The last binder used was fly ash, which contains paste. large amounts of S and A but only limited amounts of C, in combination with cement. When fly ash and cement 3.3 Achieved strength in relation to estimated are used in equal proportions, the CH formed following bondings formed by hydration and pozzolanic hydration was calculated to lead to about 56 kg CASH reactions resulting from the fly ash. Since the amount of CH In analysing the effect of the type of soil on the available is limited, excess S and A will remain in the fly strength measured in the unconfined compression tests in ash. A larger amount of cement in relation to fly ash the present study, several similarities were observed. The (75:25), which could utilise more of the S and A present strength levels were approximately the same in the in the fly ash, was also tested in a number of samples. Löftabro clay and the Holma gyttja, and to some extent The estimated quantity of reaction products resulting from reaction with the fly ash and also minerals in the also in the Linköping clay for binders containing cement soil was then about 69 kg CASH. A shortage of A in the alone or in combination with slag or fly ash. However, fly ash may well lead to the formation of larger quantities the effect of lime in the Linköping clay was very limited. of CSH instead of CASH. However, for comparison The test results clearly indicate that for stabilisation with between the different binders, the calculations were cement or other binders not leading solely to pozzolanic consistently performed based on formation of CASH. reactions with minerals contained in the soil itself, all The rough estimates of amounts of reaction products three soils can be regarded as having normal, high formed by the various binders are summarised in Table 4. potential for stabilisation, without any substantial In comparing the principal reaction products generated retarding substances in the pore water. Lime and other by the different binders, it can be seen that there are large binders involving a large degree of pozzolanic reaction differences in total quantities as well as the mechanism of also showed high potential for stabilisation in two of the formation, whether by the more short-term cement soils, but not for the Linköping clay. It can be noted that reactions or by the long-term pozzolanic reactions. This although the Holma gyttja contained about 10% organic is likely to affect the increase in strength with time in the material, its potential for stabilisation seems not to be various stabilised samples. lower than that of the others. Although lime alone, in In general, the formation of CSH following hydration of accordance with normal practice for organic soils, was cement is a faster process than the pozzolanic reactions not used in this soil, the results when using the combined resulting from adding lime or cement. However, it should binders cement–lime and slag–lime were similar to those be kept in mind that in the stabilisation of soils, a number obtained by these binders in the Löftabro clay. Using of factors may influence the rate of strength increase. lime alone in the Holma gyttja would have meant that the Possible retarding or accelerating chemical elements in CH had not been able to react in full to form CASH, the soil, as well as variations in temperature or stresses because of a shortage of alumina in this organic soil. This may lead to widely deviating increases in strength. The in turn would have resulted in a larger proportion of CSH amount of water and the type and concentration of alkalis instead of CASH or CAH being formed and thus a lower in the pore water are factors that may affect the solubility total amount of reaction products. of the potential pozzolanic elements of the soil Figure 4 shows the calculated quantities of reaction (Buchwald et al., 2003), which in turn will affect the rate products formed by the different types of binders, drawn of pozzolanic reactions with time. Furthermore, it should as bars with the scale on the left axis. The bars are be observed that although the different reaction products divided indicating products formed by cement reactions, may be of the same type, the denseness, and thus also the pozzolanic binder reactions and pozzolanic soil reactions. strength, may vary depending on different factors. For For comparison, the strength measured 28 days and 1 year after mixing are plotted in the same diagram with the scale on the right hand axis. The full line in the Table 4. Estimated maximum quantities of the main diagram shows the mean strength for all three soils after reaction products formed from 100 kg of each 3 28 days, i.e. relatively soon after stabilisation, while the binders per m of soil. two dash-dotted lines show the relatively large variation in long-term strength resulting from the different binders Binder Cement Pozzolanic Total after one year. reactions reactions The diagram shows that the variation in short-term (kg) (kg) (kg) strength after 28 days for the different binders roughly Lime - 345 345 mirrors the variation in the estimated quantities of the Cement 100 90 190 Cement–slag 110 40 150 cement hydration products in terms of CSH. An Slag–lime 60 170 230 exception is slag–lime, for which the measured strength Cement–lime 50 215 265 was somewhat lower than that of the others in relation to Cement–fly ash (50:50) 50 55 105 the calculated amount of hydration products. This is in Cement–fly ash (75:25) 75 70 145 accordance with the slower strength increase commonly
200 Deep Mixing '05 strength increase with time (Åhnberg & Holm, 1987; 400 4000 Åhnberg et al., 1989). Other factors that must be taken Binder quantity 100 kg/m3 max. pozzolanic soil reactions, CASH max. pozz. binder reactions, CASH into account are the stress level after mixing and whether cement reactions, CSH further loading is carried out. This may influence the quc, mean, 28 days 300 3000 , kPa density and in turn the increase in strength of the q , 1 year uc uc, min-max q stabilised soil. The use of combinations of binders may
gth have an extra positive or negative effect on the strength of the stabilised soil, by way of one binder altering the oducts, kg oducts, 200 2000 conditions for the bonding processes of the other, e.g. through a change in density, water content, workability
Reaction pr or temperature. Furthermore, in estimating the strength increase, the possible influence of retarding substances or 100 1000 other factors such as varying mixing work or binder Unconfined compressive stren quantity should always be considered. While the strength increase will not be the same in the 0 0 field as in samples stabilised in the laboratory, the Lime cl (50:50) sl (50:50) Cement cs (50:50) cf (75:25) cf (50:50) Stabilising agent principal effects of different binders can be expected to be fairly similar. Provided that sufficient binder Figur 4. Estimates of the amount of reaction quantities are used, a significant long-term strength products contributing to the strength of the increase can be expected also in the field, especially for stabilised soils (bars) together with the strength binders containing some amount of lime. Continued measured in the three soils one month and one year increases in strength measured also long after installation after mixing (curves). have thus been reported for lime–cement columns at a number of sites (e.g. Edstam et al., 2004) as well as for cement columns in a number of investigations (e.g. ordinary cement hydration, cf. comment in previous Hayashi et al., 2003). section. After one year, the strength had continued to Although different types of chemical reactions are increase in varying degrees for the different stabilised obviously involved when mixing soils with various soils. The increase to a large extent reflects the possible common binders, this has been shown not to affect the amount of pozzolanic reactions taking place. Slag is an fundamental behaviour of the stabilised soils to any exception also after one year, foremost in combination significant extent (e.g. Åhnberg, 2003, 2004). The type with lime, which exhibits a more pronounced long-term and amount of binder chosen can strongly affect the strength increase compared to the other binders. For all strength achieved at a certain time after mixing, and the binders except cement and cement–fly ash (50:50), there increase in strength with time should be taken into is a relatively large variation in strength depending on the account in design. However, the behaviour of the type of soil, which to a large degree is related to the stabilised soil at a specific strength level will be exceptionally slow progress of pozzolanic reactions in approximately the same independent of the type of binder the Linköping clay. used.
3.3 Considerations when predicting strength in design 4 CONCLUSIONS The observed pattern of the increase in strength for the Strength tests have been performed on samples consisting same type of soil when stabilised with different types of of different types of binder mixed into three types of soil, binders, may serve as basis for assessments of strength giving increased insight into the effect of the different increase with time. The approach used in the present binders on the increase in strength with time in stabilised study, i.e. relating the increase in strength to possible soils. reaction products, has provided an increased − The use of cement, lime, slag and fly ash may give understanding of the processes governing the varying increase in strength with time in stabilised development of cementing bonds in soil. soils depending on the combination of binders chosen. However, the results can only be used for rough − The increase in strength was found to be roughly estimates of differences in strength increase for different related to the type and quantity of possible reaction binders in otherwise similar conditions. It should be products. noted that in the tests, the temperature during curing was − Substantial long-term increases in strength were the same for all samples. In deep mixing in the field, the observed for several of the binder combinations. temperature of soils stabilised with binders containing The results of this study highlight the importance of quick lime can be expected to be higher than those taking into account a time factor in estimating the containing only cement, slag or fly ash in any strength of stabilised soils in design. In order to better combination. A higher temperature will speed up the account for the life-time behaviour of a foundation on chemical processes involved, resulting in a faster rate of
Deep Mixing '05 201 stabilised soil, a time factor should be incorporated into Holm, G. & Åhnberg, H. (1987). Use of lime-fly ash for the design models for soil improvement by deep mixing. deep stabilisation of soil. Swedish Geotechnical Institute, Report No. 30. (In Swedish) Horpibulsuk, S., Miura, N. & Nagara, T. S. (2003) Assessment of strength development in cement- 5 ACKNOWLEDGEMENTS admixed high water content clays with Abrams’ law as Financial support for this study was provided by the a basis. Géotechnique, Vol. 53, No. 4, pp. 439-444. Swedish Geotechnical Institute (SGI) and the Swedish Ingles, O.G. & Metcalf, J.B. (1972). Soil stabilization. Deep Stabilization Research Centre (SD). The special Butterworths Pty. Ltd. contributions of the SD participants Cementa AB, Kitazume, M. & Terashi, M. (2002). The Deep Mixing Nordkalk AB and Merox AB are highly appreciated. Method – Principle, Design and Construction. Ed. Coastal Developmnent Institute of Technology (CDIT), Japan. A.A. Balkema Publishers Tokyo. 6 REFERENCES Ruff, C.G. & Ho, C. (1966). Time-temperature strength- ASTM (2000). ASTM standard C 618 – 00. Standard reaction product relationships in lime-bentonite-water specification for coal fly ash and raw or calcined mixtures. Highway Research Record No. 139, Highway natural pozzolan for use as a mineral admixture in Research Board. pp. 42-60 Portland cement concrete. Saitoh, S., Suzuki, Y. & Shirai, K. (1985). Hardening of Betonghandbok (1997). Betonghandbok Material. Utgåva soil improved by the deep mixing method. Proc. 11th 2. AB Svensk Byggtjänst, Stockholm. Int. Conference on Soil Mechanics and Foundation Bruce, D., Bruce, M.E. & DiMillio, A.F. (1999). Dry mix Engineering, Vol. 3. pp. 1745-1748. methods: A brief overview of international practice. Taylor, H.F.W. (1997). Cement chemistry. Thomas Proc. Int. Conference on Dry Deep mixing Methods for Telford, London 1997. deep Soil Stabilization, Stockholm 1999. pp15-25. Terashi, M. (2003). The state of practice in deep mixing Boynton, R. (1980). Chemistry and technology of lime methods. 3rd Int. Conference on Grouting & Ground and limestone. John Wiley & Sons Inc., Treatment, New Orleans 2003, pp 25-49. Buchwald, A., Kaps,, C. & Hohmann, M. (2003). Alkali- TRB (1987). Lime stabilisation. Reactions, properties, activated binders and pozzolan cement binders – design, and construction. State-of-the-Art Report 5. compete binder reactions or two sides of the same th Transportation Research Board, National Research story? Proc. 11 International Congress on the Council, Washington, D.C. Chemistry of Cement. Durban 2003. pp 1238-1246 Porhaba, A., Shibuya, S. & Kishida, T. (2000). State of Carlsten, P. & Ekström, J. (1995). Lime and lime cement the art in deep mixing technology. Part III Geomaterial columns. Guide for project planning, construction and characterization. Ground Improvement (2000) Vol. 4, inspection. Swedish Geotechnical Society, Report No. 3, pp. 91-110. 4:1995E. Åhnberg, H. (2003). Measured permeabilities in CEN (2000). CEN standard EN197-1:2000. “Cement – stabilised Swedish soils. Proc. 3rd Int. Conference on Part 1: Composition, specifications and conformity Grouting & Ground Treatment, New Orleans 2003, pp criteria for common cements”. European Committee for 622-633. standardization (CEN), Brussels. Åhnberg, H. (2004). Effects of back pressure and rate of CEN (2001). CEN standard EN459-1:2001. “Building strain in triaxial testing of some stabilised Swedish lime – Part I: Definitions, specifications and conformity soils. Geotechnical Testing Journal. Vol. 27, No. 3. pp criteria”. European Committee for standardization 250-259 (CEN), Brussels. Åhnberg, H., Bengtsson, P-E. & Holm, G. (1989). Chew, S.H., Kamruzzaman, A. H. M. & Lee, F. H. Prediction of strength of lime columns. XII (2004). Physicochemical and engineering behavior of International Conference on Soil Mechanics and cement treated clays. Journal of Geotechnical and Geoenvironmental Engineering. Vol. 130, No. 7, pp. Foundation Engineering. Rio de Janeiro 1989. pp. 696-706. 1327-1330 Edstam, T., Ekström, J., Hallingberg A. & Nilsson, L. Åhnberg, H., Bengtsson, P.-E. & Holm, G. (2000). Task (2004). Provning av kalkcementpelare i Göta älvdalen. 2.2 and 7. Laboratory tests of mixtures – Test sites Proc. XIV Nordic Geotechnical Meeting, Ystad 2004, Dömle Mosse and Holma Mosse, Sweden. National pp. D-43-54. report, part 1. Internal Report. EuroSoilStab project, ETC5 (1998). Recommendations of the ISSMGE for SGI 3-9705-239 and 3-9708-395. geotechnical laboratory testing. Beuth Verlag, Berlin. Åhnberg, H. & Holm, G. (1987). On the influence of Hayashi, H., Nishikawa, J., Kanta Ohishi, K. & Terashi, curing temperature on the shear strength of lime and M. (2003). Field observation of long-term strength of cement stabilised soil. Swedish Geotechnical Institute, cement treated soil. Proc. 3rd Int. Conference on Report No. 30. pp. 93-146 (In Swedish) Grouting & Ground Treatment, New Orleans 2003, pp Åhnberg, H., Johansson, S.-E., Pihl, H. & Carlsson, T. 598-609. (2003). Stabilising effects of different binders in some Holm, G. (2003). State of practice in dry deep mixing Swedish soils. Ground Improvement, Vol. 7, No. 1, pp methods. 3rd Int. Conference on Grouting & Ground 9-23. Treatment, New Orleans 2003, pp 145-163.
202 Deep Mixing '05 Paper III Reprinted, with permission from Ground Improvement, Thomas Telford Ltd, 1 Heron Quay, London E14 4JD. From Vol. 5, No. 1.
900 cl80/20, 7d Peat 1—Dömle 1 year 3 800 cl80/20, 28d (200 kg/m ) cl80/20, 91d 91 cl80/20, 360d days 700 cs50/50, 7d Cement–slag 600 cs50/50, 28d cs50/50, 91d 28 days 500 cs50/50, 360d cf50/50, 28d : kPa : uc q 400 7 300 days 200 Cement–lime (7–360 days) 100 Cement–fly ash 0 0 5 1015 20 Load: kPa
Paper IV Reprinted with permission from the American Society of Civil Engineers, ASCE. From Geotechnical Special Publication No. 120, Proceedings of the 3rd International Conference on Grouting and Ground Treatment, New Orleans 2003, Vol. 1. Measured permeabilities in stabilised Swedish soils
Helen Åhnberg1
Abstract In connection with a large Swedish research programme concerning different aspects of deep stabilisation of soft soils, a number of permeability tests have been performed on certain soft soils stabilised with different type of binder in the laboratory. The permeabilities of the stabilised samples were measured in permeameter tests and in CRS oedometer tests. Although exhibiting a fairly large scatter in results, the tests showed that the permeability changed with time and with the amount of binder. They also showed that the permeability to some extent varied with the type of binder used. Comparisons made with other permeability investigations on stabilised soils have shown both agreement and deviations from the results in the present study. It was found that changes in permeability in the stabilised material to a large extent could be linked to the increase in strength and decrease in water content.
Introduction In deep stabilisation of soft soils, the permeability of the stabilised soil may be important from several aspects. The permeability is used in estimating the consolidation rate with time after construction, as well as in evaluating the risks of changes in groundwater conditions due to lowering of artesian groundwater pressures, possible leakage of binder substances from columns or impact of trace elements on the environment. Earlier laboratory investigations of the change in permeability after stabilisation of clays have been reported to show increases (e.g. Bengtsson and Åhnberg, 1995; Baker, 2000) or decreases (e.g. Terashi and Tanaka, 1983; Kawasaki et al., 1984), or both increases or decreases depending on the type of binder used (e.g. Åhnberg et al., 1995). In all soils, permeability decreases with decreasing water content, or void ratio. In stabilised soils, permeability has been reported to decrease with increasing amount of binder and with decreasing water content of the stabilised soil, Figure 1. Furthermore, permeability has been reported to decrease with time after stabilisation, Figure 2. Studies have also been performed on the permeability of a stabilised soil ------1Swedish Geotechnical Institute, SE-581 93 Linköping, Sweden; [email protected]
622 GROUTING AND GROUND TREATMENT 623
Figure 1. Change in permeability with cement content (aw) and water content (wa) of a stabilised marine clay. From Terashi and Tanaka (1983).
Figure 2. Influence of amount and type of additive and time of curing on the permeability of compacted stabilised clay. From Brandl (1999). coupled to the void ratio (Locat et al., 1996), which showed that for a constant void ratio, permeability increased after stabilisation with small amounts of lime, whereas it decreased with larger amounts of lime. A research project has been initiated and financed by the Swedish Deep Stabilization Research Centre (SD) in which permeability tests have been performed on soils stabilised with different types of binder. Results from series of unconfined compression tests performed on these soils after stabilisation with the various types 624 GROUTING AND GROUND TREATMENT of binder have been reported by Åhnberg et al. (2002). The main objectives of the permeability tests have been to study the effects of different types of binder on permeability and the extent to which permeability changes with time after stabilisation. Results from measurements on fairly homogeneous stabilised soil, i.e. samples tested in the laboratory, have indicated that the change in permeability can be described by an initial change followed by a decrease with time. Changes in permeability with time are likely to be influenced by the continuing formation of different hydration products in the stabilised soils. One measure of this growth is the strength of the stabilised soils. In order to study the extent to which the change in permeability is linked to the strength of a stabilised soil, the type of soil and the type of binder, comparisons have also been made with earlier investigations in stabilised Swedish soils performed in laboratory and in situ. The paper presents results from the permeability measurements performed in the project as well as from other investigations of the permeability of stabilised soil. Observed relations between measured permeability, measured strength and change in water content are also presented and discussed.
Scope of the tests Type of soils. Two types of clay and one type of organic soil, gyttja, were used in the tests. All three soils were taken from the test sites used in the programme of the Swedish Deep Stabilization Research Centre. The gyttja was taken from 3 – 5 m depth at the test site at Holma Mosse near the east coast of southern Sweden. It is a clayey gyttja with an organic content of 10 % and a water content of 220 %. The two clays had fairly similar geotechnical properties according to the routine investigations, but came from two different geological deposits. Clay 1 is a post- glacial clay from Linköping in eastern Sweden. Clay 2 was taken at Löftabro on the Swedish west coast. The latter is a marine post-glacial clay containing occasional shells. The Löftabro clay has a somewhat higher water content and lower undrained shear strength than the Linköping clay. Both clays are high-plastic with liquid limits around 70 % and organic contents around 1 %. Results from classification and routine tests of the properties of the three different soils are shown in Table 1. Type of binder. In the tests, the traditional binders cement and quicklime, which are most commonly used in Sweden, and binders based on slag or fly ash were used. Altogether, three different types of cement, six types of lime, one type of slag and two types of ash were used in the project (Åhnberg et al., 2002). The main type of cement was the commonly used cement in Sweden, a Portland- limestone cement designated CEM II/A-LL 42.5 R (CEN standard, 2000). The main type of lime was the quicklime normally used in deep stabilisation, CL90-Q (CEN standard, 1998). The slag used in the tests was a ground granulated blast furnace slag. The main ash used was a fly ash from pulverised coal combustion. Several other materials were used in proportions of 2 - 33 % in different combinations with the main binders. These materials were silica fume, alumina GROUTING AND GROUND TREATMENT 625
Table 1. Properties of the test soils. Holma Mosse Linköping Löftabro Gyttja Clay1 Clay2 Depth (m) 3-5 3-6 2-5 Density (t/m3) 1.23 1.55 1.52 Specific gravity (t/m3) 2.36 2.72 2.73 Plastic limit (%) 64 24 23 Water content (%) 220 78 89 Liquid limit (%) 170 70 66 Organic content (%) 10.3 1.0 1.0 Clay content (%) 80 63 72 Permeability (m/s) 1·10-9 5·10-10 1.1·10-9 Undr. shear strength (kPa) 5 15 8 Sensitivity 10 20 25 cement, calcium chloride and gypsum. Certain tests were also performed with the addition of a secondary product from lime production, kiln dust, with lime or water glass as admixture. A total binder quantity corresponding to 100 kg/m3 was used in most of the tests, although some tests were also performed with 50 kg/m3 or 150 kg/m3.
Preparation of samples. Preparation and storage of the samples were performed in accordance with the common procedure used in Sweden (Carlsten and Ekström, 1995). The binders were mixed with the soils for five minutes. The clay mixtures were gradually filled into ∅50 mm plastic tubes using a compaction pressure of 100 kPa. The gyttja mixtures were placed by hand in ∅50 mm plastic tubes. The mixtures in the tubes were given a height of about 150 mm. The tubes were sealed with rubber lids and stored in a climate controlled room at a temperature of 8 °C.
Testing. Permeability tests were performed 28 days and one year after mixing. The tests were performed in flexible-wall permeameters mainly in accordance with the Nordtest test procedures (Sjögren et al., 1994). The permeameters are of common triaxial cell type and in order to prevent waterflow along the rubber membranes the tests are performed at effective confining stresses of at least 20 kPa. The permeability tests were performed on ∅50 mm specimens with a height of 50 mm. The tests were performed without back pressure for saturation of the specimens, since only a limited number of permeameter cells with this facility were available for the test series. Hydraulic gradients of up to 40 - 60 were used in the tests. This is slightly higher than is normally recommended (ASTM Standard, 1990; Sjögren et al., 1994). Previous experience in testing stabilised soils had shown difficulties in recording stable measurements of permeability when using lower gradients. The main part of the tests started with a gradient of 40. Measurements were made during a period of one month, after which the gradient was increased to 60 before being lowered again to 40, both conditions prevailing for one week each. This procedure was used in 626 GROUTING AND GROUND TREATMENT order to check for possible clogging or piping phenomena in the samples during testing. The permeability of stabilised samples was also measured in a number of CRS tests according to Swedish Standard (1991) 1 day, 28 days and one year after mixing.
Test results Influence of type of binder. The majority of permeability tests were performed on Linköping clay stabilised with different types of binder. Figure 3 shows permeabilities measured 28 days after mixing the clay with the different types of binder. As observed also in earlier investigations on stabilised soils (Åhnberg et al., 1995), the permeability of lime stabilised samples was in general higher than that of samples stabilised with cement-based binders. When quicklime, CaO, is mixed into soil, hydration of the binder takes place as the quicklime comes into contact with the pore water and calcium hydroxide. Ca(OH)2 is then formed. This is followed by ion exchange with the soil particles, resulting in flocculation of the clay and a coarser structure of the stabilised material (e.g. Högberg, 1979; Dumbleton, 1962) compared to the original soil. Also cement- based binders produce a certain amount of calcium hydroxide and allow ion exchange to take place, but to a lesser extent. This could partly explain the differences in measured permeability between soils with lime-based binders and those with cement-based binders. However, the amount of calcium hydroxide consumed in these processes in the various types of stabilised soil is uncertain.
1.0E-07 Linköping clay
Cement-based binders 1.0E-08 Lime-based binders
1.0E-09 unstab. soil
Hydraulic conductivity, m/s conductivity, Hydraulic 1.0E-10
28 days 100 kg/m3 1.0E-11
l 5 c 0 0 4 0 2 0 50 / c3 /2 1 5 l c lsb / ,5 8 /2 /5 3/33 c005 7 0/ l95/5 0 0 i90/10l50 3 i9 a f50/ 8 5 s cl50/503/ /4 sa9 s c c g s ls 3 5 cpf50/50 c c c c 7, 4 c2lk clwa Type of additive
Figure 3. Measured permeability vs type of binder in stabilised Linköping clay. Abbr. l=lime, c,c2,c3=cement, s=slag, f=fly ash, pf=PFBC ash, si=silica fume, al=alumina cement, sa=salt, g=gypsum, wa=water glass, k=kiln dust, lsb=soft-burned lime, l4=coarse lime. GROUTING AND GROUND TREATMENT 627
Measured permeability after stabilisation. The measured permeabilities were mainly the same or slightly lower than those of the unstabilised soils. However, also increases in permeability were measured, particularly in the lime stabilised Linköping clay. Figure 4 shows the ranges of measured permeability in the stabilised soils. The figure includes ranges of permeability measured in different types of stabilised organic soil within the EC-funded (BriteEuRam) project “EuroSoilStab” (e.g. Åhnberg and Holm, 1999). The organic soils in that study were a peat with high water content, 2000 %, and three gyttjas with water contents of 220 - 370 % and organic contents of 10 - 11 %. One of these was the same type of gyttja, Holma gyttja, as that used in the present project. The types of binder used were generally of the same type as those used in the present project. It can be observed that the range of measured permeability in the stabilised soils covered about a ten-fold difference, or more, between the highest and lowest estimates. It could further be observed that, although the range of measured permeability was larger in the lime stabilised samples, there was no indication of any distinct difference in level between these samples and those stabilised with cement- based binders.
Change in permeability versus strength. After the initial effects of ion exchange leading to flocculation and changes in structure, the cementation processes will dominate with hydration of hydraulic binders as well as pozzolanic reactions with the soil or other pozzolanic constituents in the binders. These processes will involve a continuous formation of different reaction products in free spaces between the particles in the soil. The voids in the soil will thereby be filled to an increasing extent
1,0E-06 cement-based binders, 28 days cement-based binders, 1 year lime-based binders, 28 days 1,0E-07 lime-based binders, 1 year ESS Peat
1,0E-08 ESS Gyttja
1,0E-09 Permeability stabilised soil, m/s 1,0E-10
1,0E-11 1,00E-10 1,00E-09 1,00E-08 1,00E-07 1,00E-06 Permeability unstabilised soil, m/s
Figure 4. Range of measured permeability in stabilised samples vs that of the natural soils. Supplementary data from EuroSoilStab project (Åhnberg et al., 2000). 628 GROUTING AND GROUND TREATMENT with time. The rate and extent of this growth will depend on the type and amount of binder and the type of soil, and are also influenced by possible retarding substances in the soil. The strength of the stabilised material provides one measure of the growth of cementing products. Figure 5 shows how the ratio between permeability before and after stabilisation varied with the measured unconfined compressive strength of the stabilised soils. The figure also includes results from CRS tests performed one day, 28 days and one year after stabilisation. These tests were performed on specimens which had not been subjected to an initial saturation period corresponding to that in the permeameter tests. They could therefore be expected to yield somewhat lower permeability values. However, no significant difference in results was observed. The measured permeability decreased in general with increasing strength of the stabilised soils. There was a fairly large scatter in the results, particularly in the samples tested after one year, where also indications of increases in permeability were observed. The homogeneity of the samples was not altogether satisfactory at this stage of the project and there was a certain amount of air pockets in the samples. This would have affected the scatter in the results. The procedure for preparation was subsequently modified to produce more homogeneous samples (Åhnberg et al., 2002). As the samples gradually increased in strength, they also became more brittle and the occurrence of microcracks may have increased, possibly being enhanced during the cutting and preparation of samples.
Figure 5. Ratio of measured hydraulic conductivity before and after stabilisation vs the strength of the stabilised soils. GROUTING AND GROUND TREATMENT 629
Discussion Change in permeability in the laboratory compared to field conditions. The results of the tests in the present project can be compared to those from other tests performed on stabilised soils in Sweden, both in the laboratory and in the field. Figure 6 shows results from other investigations performed on lime, lime-cement and cement stabilised soils in Sweden. The strength of the stabilised soils has been determined in the laboratory by unconfined compression tests and in the field by lime column penetrometer tests. The shear strengths determined in the field have been converted to compressive strength through quc=2·cu. Exceptions are the Mellösa columns, where only permeability tests and no field tests of strength were performed. Instead, the strength of the Mellösa columns has been estimated roughly from empirical relations between laboratory and field strength in Sweden (Åhnberg et al., 1995). No significant difference in measured permeability were observed between laboratory prepared samples and columns in situ or between different types of binder. The measured permeabilities are considerably lower than those indicated in analyses of settlement processes (Bengtsson and Holm, 1984) and those normally used in calculating consolidation rates in design (Carlsten, 2000). The permeability tests in the field were performed as falling head tests in open piezometers at different depths in the columns. The results thus indicate a high homogeneity of the column material, or that the possible influence of a more permeable macrostructure cannot be detected in this type of test. However, other investigations of permeability of stabilised columns in the field by methods such as packer tests and pressure- permeameter tests have been reported to yield permeabilities 10 to 100 times higher
100,00
10,00 soil
/k 1,00 stab.
k 1-28 days
1 year 0,10 Present project Present project Fiskvik clay, 90-180d (Bengtsson & Åhnberg, 1995) field tests - Fiskvik clay, 90d Mellösa clay, 14-200d (Åhnberg et al., 1995) field tests- Mellösa clay, 14 - 77d Lövstad gyttja clay, 14-200d (Åhnberg et al., 1995) field tests- Lövstad gyttja 14 - 91d 0,01 0 200 400 600 800 1000 1200 1400 quc, kPa
Figure 6. Changes in permeability after stabilisation measured in other investigations vs the strength of the stabilised soils. Laboratory samples; Columns in the field. 630 GROUTING AND GROUND TREATMENT
(Baker, 2000) and 100 - 1000 times higher (Nygren and Wellander, 1991) respectively than that of the natural soil. These tests may reflect better the permeability of the macrostructure, but the results may also be affected by factors such as leakage along the probe. Permeability tests in the laboratory on samples from columns taken at the field sites have been reported to yield values 1 - 90 times higher than those of the natural soils (Baker, 2000). Uneven distribution and mixing of the binders and development of micro-cracks and fissures in the columns affect the permeability in the field. A more porous zone or open voids left at the centre of the column after withdrawal of the installation bar may also affect permeability. A 20-fold difference in permeability has been reported between samples taken at the central zone and those taken halfway between the centre and the periphery of lime-cement columns (Bergwall and Falksund, 1996). The different types of investigations show large variations in results. However, it should be noted that comparisons are made between results from different test methods used for measuring permeability. Each of these has its merits and drawbacks, and would normally exhibit large variations in results. Further studies of the permeability of columns in the field are needed. These should include different types of permeability measurements at a number of sites, as well as measurements and back-analyses of consolidation processes in stabilised soils subjected to loading e.g. from embankments of different heights. There are then a number of factors affecting the consolidation rate that have to be taken into account. Many of these are not easily measured, e.g. load distribution between columns and soil in the stabilised area, load distribution to the surrounding soil, generated pore pressures in columns and soil from installation and loading, suction of water to the partly unsaturated columns, and continuing changes in strength and modulus with time. Furthermore, the permeability of both natural and stabilised soil is affected by changes in the stress situation and may change as the consolidation proceeds. Continued compression of the stabilised soil will involve a reduction in void ratio and thus a decrease in permeability with time. However, depending on the type of loading, generation of large shear strains may lead to the development of microcracks, involving an overall increase in permeability.
Change in permeability linked to strength and water content. The initial change in permeability is linked to the change in void ratio and can roughly be related also to the change in water content resulting from stabilisation. The permeability of the stabilised soil then decreases with increasing strength. Changes in permeability as measured in laboratory tests are shown in Figure 7b. The figure includes data from soils ranging from clay compacted at an optimum water content of 23% (Brandl, 1999) to a gyttja and a peat in the EuroSoilStab project, having initial water contents of 370% and 2000% respectively. Most of the peat samples were subjected to a vertical pressure of 18 kPa while curing, but some were cured without such a load. This pressure corresponds to the initial loading with about one metre of fill normally applied to stabilised peat in the field. The general effects of this initial loading have been described by Åhnberg et al., (2001). GROUTING AND GROUND TREATMENT 631
The results in Figure 7b indicate that the decrease in permeability with increasing strength is approximately the same in the different types of stabilised soil. The change in the ratio between the permeability of homogeneous natural soils to that of the unstabilised soils, with increasing strength can be expressed as k k stab stab − q ≈ i ⋅ e 0.004 uc (1) ksoil ksoil where kstab is the permeability of the stabilised soil, ksoil is the permeability of the k /k q natural soil, stab i soil is the initial change in permeability and uc is the unconfined compressive strength in kPa. Alternatively, the change in permeability with increasing strength can be expressed as k ≈ k − ⋅ q log stab log stabi 0.0017 uc (2)
In Figure 7a, the initial change in permeability of the different types of material, evaluated by extrapolation of the measured relation between change in permeability and strength, is shown together with measured change in water content after stabilisation. The results clearly indicate that the initial change in permeability can be expressed as a function of the change in water content. The approximate relation shown is
w k 6⋅ stabi w ≈ 0.043⋅e 0 (3) k soil
100,00 100,00 Present project Present proj Fiskvik clay, 90-180d (Bengtsson & Åhnberg, 1995) Dömle Peat, 18kPa Mellösa clay, 14-200d (Åhnberg et al., 1995) Dömle Peat, 0kPa Compacted clay, lime 1-7.5 % (Brandl (1999) Dömle gyttja Compacted clay, cem. 2.5-7.5 % (Brandl (1999) Lövstad gyttja 10,00 10,00 Lövstad gyttja, 14-200d (Åhnberg et al., 1995) Mellösa clay Dömle gyttja, 200kg/m3 Fiskvik clay Dömle peat, 100-200kg/m3, load = 0 kPa Compacted clay Dömle peat, 100-200kg/m3, load = 18 kPa
soil 1,00 1,00 / k
soil -0,0039x y = 27,65e /k stab. k
stab, initial stab, 0,10 0,10 k
y = 0,043e6,0048x
0,01 0,01
-0,0045x y = 0,0502e 0,001 0,001 0,0 0,2 0,4 0,6 0,8 1,0 0 500 1000 1500 2000 w/w0 quc, kPa a) b)
Figure 7. Changes in permeability after stabilisation. a) initial change in permeability vs change in water content. b) change in permeability vs strength. 632 GROUTING AND GROUND TREATMENT
After insertion in Eq. 1, the change in permeability can be expressed as
w 6⋅ −0.004q k stab w uc ≈ 0.043⋅e 0 (4) k soil where w is the water content of the stabilised soil and w0 is that of the natural soil. The water content w can be estimated on the basis of the bulk density and water content of the natural soil, the amount of binder and the non-evaporable water content of the hydration product (Åhnberg et al., 2002).
Conclusions There are several factors influencing the permeability of a stabilised soil. The tests performed in the project and those reported previously in a number of other studies show both agreement and differences in results. • Measured differences in permeability when using different types of binders may largely be related to different strengths of the stabilised soils. • Stabilised clay most often exhibits an initial increase in permeability after stabilisation, followed by a decrease in permeability with time. The initial change is influenced by the introduction of binder particles into the soil, flocculation of soil particles and any compaction applied to the stabilised soil. The change with time is believed to be caused mainly by different cementation processes. • A rough estimate of the change in permeability of homogeneous stabilised soils can be made on the basis of the change in water content and the strength of the stabilised soil as shown in Figure 7 and Eq. 4. • The permeability measured in the laboratory can be regarded as a lower limit. Large scale tests performed on columns in the field have shown somewhat higher or even considerably higher values compared to the probably more homogeneous samples in the laboratory. Further studies are needed concerning the permeability of columns in the field and its change during the cementation and consolidation processes.
Acknowledgements Financial support for the tests has been provided by the Swedish Geotechnical Institute (SGI) and the Swedish Deep Stabilization Research Centre (SD). The financial contributions of the participants Cementa AB, Partek Nordkalk AB, Merox AB, SGI and SD are gratefully acknowledged. The permission of the participants in the European Community project EuroSoilStab under contract Brite EuRam BRPR- CT97-0351 to present data from this project is also gratefully acknowledged.
References Åhnberg H., Bengtsson P-E. and Holm G. (2000). Task 2.2 and 7. Laboratory tests of mixtures – Test sites Dömle Mosse and Holma Mosse, Sweden. National report, part 1. Internal Report. EuroSoilStab project, SGI 3-9705-239 and 3-9708-395. GROUTING AND GROUND TREATMENT 633
Åhnberg H., Bengtsson P-E. and Holm G. (2001). Effect of initial loading on the strength of stabilised peat. Ground Improvement. Vol 5, No 1, pp. 35 – 40. Åhnberg H. and Holm G. (1999). Stabilisation of some Swedish organic soils with different types of binder. Proc. of the International Conference on Dry Mix Methods for Deep Soil Stabilization, Stockholm 1999, pp. 101-108. Åhnberg H., Johansson S-E., Pihl H. and Carlsson T. (2002). Stabilising effects of different binders in some Swedish soils. SGI project 1-9710-504 (to be published). Åhnberg, H., Johansson, S-E., Retelius, A., Ljungkrantz, C., Holmqvist, L., Holm, G. (1995). Cement and lime for deep stabilisation of soil. Swedish Geotechnical Institute, Report No. 48. (In Swedish) ASTM Standard (1990). ASTM Standard D 5084– 90. Standard test method for measurement of hydraulic conductivity of saturated porous materials using flexible wall permeameter. Baker, S. (2000). Deformation behavior of lime/cement column stabilized clay. Ph.D. thesis, Dept. of Geotechn. Eng., Chalmers Univ. of Techn., Gothenburg. Bengtsson, P-E. and Åhnberg, H. (1995). Settlement calculations for lime column stabilisations. Road E18 at Fiskvik kanal. SGI project 1-261/86. (In Swedish) Bergwall, M. and Falksund, M. (1996). Consolidation of lime cement columns – permeability and stiffness. Master’s thesis 1996:1, Dept. of Geotechn. Eng., Chalmers Univ. of Techn., Gothenburg. Brandl, H. (1999). Long-term behaviour of soils stabilised with lime and with cement. Proc. Geotechnics for Developing Africa, Durban 1999, pp. 219-232. Carlsten, P. (2000). Lime and lime cement columns. Guide for project planning, construction and inspection. Swedish Geotechn. Soc. Report 2:2000. (In Swedish). Carlsten, P. and Ekström, J. (1995). Lime and lime cement columns. Swedish Geotechn. Soc. Report 4:1995E. CEN Standard (1998). European Standard pr-EN459-1:1998. Building lime – Part I: Definitions, specifications and conformity criteria. CEN Standard (2000). European Standard pr-EN197-1:2000. Cement – Part 1: Composition, specifications and conformity criteria for common cements. Dumbleton, M.J. (1962). Investigations to assess the potentialities of lime stabilization in the United Kingdom. Road Research Technical Paper No. 64. Högberg, E. (1979). The role of lime in the stabilisation process. Sem. on The lime column method in practice. Soil and Rock Dept., Royal Univ. of Techn., Stockholm. Kawasaki, T., Saitoh, S., Suzuki, Y. and Babasaki, T. (1984). Deep mixing method using cement slurry as hardening agent. Seminar on soil improvement and construction techniques in soft ground, Singapore1984. Locat, J., Tremblay, H. and Leroueil, S. (1996). Mechanical and hydraulic behaviour of a soft inorganic clay treated with lime. Can. Geot. J. Vol. 33, No. 4, pp. 654-669. Nygren, M. and Welander, A.-S. (1991). Estimates of properties of lime-cement columns. Master’s thesis 91/5, Royal Univ. of Techn., Stockholm. (In Swedish) Sjögren, M., Carlsten, P. and Elander, P. (1994). Determination of the permeability of residual products and soil, in situ and in the laboratory. Nordtest Techn. Rep. 254. Swedish Standard (1991). SS 02 71 26, Geotechnical Tests- Oedometer tests with constant rate of deformation – CRS-tests. SBSI, Stockholm. (In Swedish) Terashi, M. and Tanaka, H. (1983). Settlement analysis for deep mixing methods. Proc. 8th Eur. Conf. on Soil Mech. and Found. Eng., Helsinki 1983, pp. 955-960. Paper V Reprinted, with permission from the Geotechnical Testing Journal, Vol. 27, No. 3, copyright ASTM International, 100 Barr Harbor Drive, West Conshohocken, PA 19428, USA.
Paper VI Reprinted with permission from Ground Improvement, Thomas Telford Ltd, 1 Heron Quay, London E14 4JD. From Vol. 10, No. 1.
50 50
40 40
30 30 : degrees : degrees ¢ ¢ ö ö 20 20 ö¢cs for natural soft and stiff clays, shales, clay minerals (Terzaghi and Peck, 1967)
Present project, Linköping clay Present project, Löftabro clay c50200, Linköping clay l100, 1 day1 year 10 Holma gyttja, Åhnberg et al. (2000) 10 cl100 cs100 Dömle peat, Åhnberg et al. (2000) cf100 sl100 l50200, Löftabro clay c100 Åhnberg et al. (1995) - Kil silt, Mellösa clay, Lövstad gyttja cl70, Holma gyttja cs70100 Balasubramaniam and Buensuceso (1989) - Bangkok clay cp70 0 0 0 200 400 0 50 100
wstab: % IP, stab: % (a) (b)
20 Linköping clay 20 Linköping clay ó¢ 5 20 kPa 3c ó¢ 5 20 kPa ó¢ 5 80 kPa 3c 3c ó¢ 5 80 kPa ó¢ 5 160 kPa 3c 3c ó¢ 5 160 kPa ó¢ 5 240 kPa 3c 3c ó¢ 5 240 kPa Linköping clay 3c 15 15 Linköping clay ó¢3c 5 20 kPa ó¢3c 5 20 kPa ó¢3c 5 80 kPa ó¢3c 5 80 kPa ó¢3c 5 160 kPa ó¢3c 5 160 kPa quc, min-max ó¢3c 5 240 kPa quc, min-max : % 10 : % 10
failure ó¢ 5 160 kPa 3c failure å å
ó¢3c 5 80 kPa 5 5 ó¢3c 5 240 kPa
ó¢3c 5 20 kPa
( ) 0 0 0 500 1000 0 500 1000 1500 q: kPa q: kPa (a) (b)
Paper VII Submitted for publication to the Canadian Geotechnical Journal, National Research Council of Canada, Ottawa, ON K1A 0R6, Canada. On yield stresses and the influence of curing stresses on stress paths and strength measured in triaxial testing of stabilised soils
Helen Åhnberg
Abstract: Studies on the behaviour of stabilised soils under different loading conditions are essential to identify which parameters are relevant in the design of deep mixing. An investigation has been carried out in which a series of triaxial tests have been performed on soils stabilised with different types of binders. The tests were performed with the purpose of demonstrating the existence of quasi-preconsolidation pressures, i.e. vertical yield stresses that are not primarily linked to previous consolidation pressures but to the cementation taking place, and the effects of these on the strength behaviour of stabilised soil. The effect of stresses applied during curing has also been studied. Drained triaxial compression tests, as well as undrained triaxial compression and extension tests, were performed on two stabilised clays. The binders used were cement, lime, slag and fly ash in different combinations. Comparisons have also been made with results from previous tests on two organic soils stabilised with much the same types of binder. The results show that both the cementation processes involved and the stresses applied during curing affect the quasi-preconsolidation pressure. This pressure is strongly linked to the strength of the stabilised soil and has a considerable influence on its deformation behaviour. A model is proposed which describes the strength behaviour in the same effective stress plane that is commonly used for natural clays.
Introduction
Improvement in the strength and stiffness properties of soft soils by deep mixing is becoming a common method in many parts of the world. A large number of investigations have been performed to study the effect of foremost cement and lime mixed into various soils, in the design stage, as well as for research purposes, (e.g. Broms 2000; Kitazume and Terashi 2002). The test method commonly used in these investigations is the unconfined compression test. This is a simple, economical test useful for comparing the effects of different types of binder and other factors influencing the strength of stabilised soil (e.g. Kawasaki et al. 1984; Babasaki et al. 1997; Yin 2001; Zhou et al. 2002; Åhnberg et al. 2003). However, for detailed studies of stabilised soil behaviour and the assessment of changes in effective stresses and strength, as well as in stiffness, triaxial tests are more suitable. To improve the models used to describe stabilised soil behaviour, more studies using triaxial tests are needed on soils stabilised with various types of binder, under undrained as well as drained conditions. The occurrence of vertical yield stresses has been demonstrated in oedometer tests and in triaxial tests on stabilised soils, (e.g. Tatsouka and Kabayashi 1983; Balasubramaniam and Buensuceso 1989; Åhnberg et al. 1995; Lee et al. 2004). Research on yield stresses and their effect on the compressibility of stabilised soils has been reported in several papers (e.g. Tremblay et al. 2001). In recent years, some studies have been performed also concerning the effect of yield stresses on undrained stress paths (e.g. Kasama et al. 2000; Uddin and Buesuceso 2002; Balasubramaniam et al. 2005) and
Åhnberg, H., On yield stresses and the influence of curing stresses… 1 strength (e.g. Horpibulsuk et al. 2004; Åhnberg 2006;) in triaxial strength testing. However, investigations on yield stresses on stabilised soils and their influence on strength behaviour reported in the literature (in English) are still few, and very little has been published concerning the effects of different binders. Although these yield stresses, or “quasi-preconsolidation pressures”, are affected to a large degree by cementation effects, the stresses acting during curing can also be expected to play a role in the strength behaviour of stabilised soil. However, studies performed on stabilised soils cured under different vertical stresses (e.g. Yamamoto et al. 2002; Åhnberg et al. 2001) and on artificially cemented soils cured under various confining stresses (e.g. Consoli et al. 2000; Rotta et al. 2003; Zhu et al. 1995) are relatively few, and have mostly involved testing of compression properties. The latter area of research has primarily involved testing of sands artificially “cemented” by adding different types of cement. To verify that cementing processes and curing stresses may influence the quasi- preconsolidation pressure, and in turn the stress paths and strength of stabilised soil, and to study the extent to which this occurs, a series of triaxial tests have been performed on two stabilised clays cured with and without the application of external stresses. In addition to the triaxial tests, a number of oedometer tests were performed on the same types of material to enable comparisons of the strength behaviour with the quasi- preconsolidation pressures evaluated from the latter tests. The tests were performed within a series of projects (e.g. Åhnberg et al. 2003; Åhnberg 2006) organised by the Swedish Deep Stabilisation Research Centre (SD) concerning the properties of stabilised soils. To further confirm the role of cementation on the development of a quasi- preconsolidation pressure and its effect on the strength behaviour, comparisons were made with results from a previous series of triaxial tests on two organic soils (Åhnberg et al. 2000) where similar types of binder were used. In this paper, the effect of quasi- preconsolidation pressures and the influence of curing stresses on the stress paths and strengths of soils stabilised with different types of binder are discussed and the applicability of a strength model commonly used for natural soils is demonstrated.
Scope of the investigation
Types of soil Two types of clay were included in the investigation. They were soft clays of types that have been shown to give different stabilising results in previous investigations (Åhnberg et al. 2003). The clays have fairly similar geotechnical properties according to routine classification tests, but are from different geological deposits. The first is a post- glacial clay from an area near Linköping in the eastern part of Sweden. It was deposited in alternating lake and brackish sea water. The second is a marine post-glacial clay containing occasional shells, from a site near Löftabro on the west coast of Sweden. Both clays are high-plastic with liquid limits of approximately 70% and organic contents of approximately 1%. Results from laboratory testing of the two clays are presented in Table 1.
Types of binder Two single binders and four composite binders were used for stabilisation of the clays. Only dry binders were used. They were pure cement, pure lime, cement–lime,
Åhnberg, H., On yield stresses and the influence of curing stresses… 2 Table 1. Properties of the two clays.
Linköping Löftabro clay clay Depth (m) 3-6 2-5 Density (t/m3) 1.55 1.52 Specific gravity (t/m3) 2.72 2.73 Water content (%) 78 89 Plastic limit (%) 24 23 Liquid limit1 (%) 70 66 Undrained shear strength1 (kPa) 15 8 Sensitivity 20 25 Organic content (%) 1.0 1.0 Clay content (%) 63 72 Chloride content (%) 0.01 0.38 Sulphide content (%) 0.05 0.18 pH 7.7 8.7 1Determined by the fall cone test (ETC5, 1998)
cement–slag, cement–fly ash and slag–lime. All these binders were used in the Linköping clay, while only pure cement and pure lime were used in the Löftabro clay. All composite binders were used in proportions of 50%:50% (by weight). All binders were added at quantities quantity corresponding to 100 kg per m3 of soil. In addition, a number of tests were performed with 50 kg/m3 and 200 kg/m3 cement and 70 kg/m3 cement-lime in the Linköping clay. Pure lime, which in earlier tests has been shown to be rather ineffective in this clay, was used only at a quantity of 100 kg/m3. In the Löftabro clay, only 100 kg/m3 cement was used, whereas the lime was used in quantities of 50, 100 and 200 kg/m3 (see Table 2). The binders chosen are those commonly used for soil stabilisation. The lime used was a quicklime, designated CL90-Q (CEN Standard 2001); the cement was a Portland- limestone cement, CEM II/A-LL 42.5 R (CEN Standard 2000); the slag was a ground, granulated blast furnace slag, and the fly ash came from pulverised coal combustion residues.
Preparation and curing of stabilised samples Samples were prepared by first mixing the binders with the soils for five minutes. The mixtures were then filled in layers, of approximately 30 mm, into ∅50 mm plastic tubes using a momentary vertical pressure of 100 kPa for each layer. This pressure is used not for compaction but for producing homogeneous samples with a minimum of air pockets. The tubes were filled to a height of approximately 150 mm. A number of the samples prepared in this way were then sealed with rubber lids and stored in a climate-controlled room at 8°C. This mode of preparation and storage is in accordance with the common procedure for testing of laboratory samples of stabilised soils in Sweden (Carlsten and Ekström 1995; Carlsten 2000).
Åhnberg, H., On yield stresses and the influence of curing stresses… 3 To study the effect of curing under various stresses, a number of samples stabilised with cement–lime were stored and cured in triaxial cells. Directly after preparation of the samples, they were trimmed to lengths of 110 to 120 mm and mounted in the cells for curing under different stresses. Figure 1 shows the set-up of the eight cells used for this purpose. Four samples at a time were mounted, K0-consolidated and then cured, under the same stresses, for 28 days. Different consolidation stresses could be used for any two series run in parallel. The confining stress, the pore pressure and a vertical load corresponding to a consolidation stress ratio K0 of 0.5 were applied in steps, concurrently for the four samples. This procedure was finalised approximately 1.5 hours after mixing. The effective confining stresses used were 20 kPa and 60 kPa, the vertical effective stresses 40 kPa and 120 kPa respectively, and the pore pressure 20 kPa. The samples were cured at room temperature (approximately 20°C). For comparison, a number of samples were stored for 28 days without being subjected to external stresses but with access to water through filter stones placed at the bottom and on top of the samples. These samples, contained in sample tubes perforated at the top end, were placed in boxes filled with water to a level just above the top of the upper filter stones. The boxes were kept in the same room as the curing triaxial cells.
Testing A series of consolidated undrained (CAU) and drained (CAD) triaxial tests were performed at different times after mixing. Stabilised clay specimens containing composite binders and pure cement or lime in quantities corresponding to 50 and 200 kg/m3 were all tested in four weeks after mixing, while those containing 100 kg/m3 pure
Fig. 1. Set-up of triaxial cells for curing of stabilised clays subjected to different stresses.
Åhnberg, H., On yield stresses and the influence of curing stresses… 4 cement or lime were tested in one day, four weeks and one year, or slightly later, after mixing. An outline of the triaxial testing programme is presented in Table 2. The specimens tested after four weeks or more were prepared from the same batch. A maximum of five tests were performed in a week, giving a variation in specimen age of 28–32 days for the tests after four weeks and the same span for those after about one year. Since the strength shortly after mixing is normally highly dependent on time, the specimens tested after one day were all prepared from different batches. Most of the tests were performed as single tests, but a number of duplicate and triplicate tests were also performed. The triaxial tests were performed on specimens with a height-to-diameter ratio of 2. Prior to testing, the specimens were cut and smoothed to provide flat, parallel end surfaces. Polished stainless steel platens together with greased rubber membranes were used to minimise friction at the end surfaces. Filter paper strips were mounted on the sides of the specimens and extended to reach annular filters at the back side of the end platens. This set-up approach has been recommended by other authors (Lacasse and Berre 1988; deGroot and Sheahan 1995) and has been found to work well in tests performed earlier at the Swedish Geotechnical Institute (SGI) laboratory (Larsson 2000). The back pressure used in the tests was 400 kPa. This back pressure level was chosen with the intent of obtaining full saturation. The degree of saturation of the different stabilised clay specimens before testing was approximately 93–95%. In order to ensure that all air enclosed in the specimen shall dissolve in the pore water, a back pressure of at least 400 kPa is required for an initial degree of saturation down to approximately 92% (e.g. Lowe and Johnson 1960; Bishop and Henkel 1962). The specimens were consolidated for 7.5 hours before the start of the shear test. This consolidation time was somewhat longer than necessary according to estimates of the permeability of the materials and earlier experience, but was chosen in order to account for possible variation in composition and to fit into an effective testing schedule. A K0 value of 0.8
Table 2. Triaxial testing programme. Soil Binder Quantity Curing Curing Triaxial tests time conditions kg/m3 days Linköping cement 50 28 no load, 8°C CAU, CAD compression clay 100 1, 28, 360 200 28 lime 100 1, 28, 360 cement-lime 70 28 ext. stress, 20°C CAU, CAD compression CAU extension no load, 20°C CAU, CAD compression CAU extension 100 no load, 8°C CAU, CAD compression cement-slag cement-fly ash slag-lime Löftabro cement 1, 28, 360 clay lime 50 28 100 1, 28, 360 200 28
Åhnberg, H., On yield stresses and the influence of curing stresses… 5 was used for the majority of the tests; a stress ratio chosen from empirical experience of in situ measurements. To investigate in greater detail the influence of curing stresses, a number of tests were also performed with a K0 value of 0.5. The maximum rate of strain suitable in drained triaxial tests on saturated soils (Berre, 1997) was estimated to be 0.03–0.08%/min in the different types of stabilised soil. A slightly lower rate of 0.02%/min was chosen in order to account for possible variation in composition. The same rate of strain was used in the undrained triaxial compression tests. A number of undrained passive tests were also performed using a constant rate of horizontal stress increase. In addition to the triaxial tests, a series of oedometer tests were performed on the same types of material as those used in the triaxial tests. The oedometer tests were performed to assess the yield stresses in the stabilised soils. They were performed both as CRS oedometer tests and as incrementally loaded tests in accordance with Swedish Standards (1991 and 1992, respectively).
Test results and discussion
Measured effective stress paths in the various types of stabilised soils The strength of the stabilised soils varied considerably for the different types of samples. Depending on the type of clay, the type and quantity of binder, the curing stress, time after mixing and testing conditions, compressive strengths of approximately 50 kPa to 1500 kPa were measured. This variation in strength enabled studies of the effective stress response under loading for specimens representing soft to stiff stabilised soils. In Fig. 2a and b the measured effective stress paths of the stabilised Linköping clay and Löftabro clay are shown respectively in the s´:t effective stress plane, where sƍ = (σƍ1+σƍ3)/2 and t = (σƍ1-σƍ3)/2. The test results are shown in separate diagrams to facilitate comparison of results representing different binders, quantities and time after mixing. The results are shown as effective stress paths measured in the undrained triaxial tests and as failure and yield stresses evaluated from the drained triaxial tests. Lines are shown representing the evaluated effective strength parameters, friction angle, φƍ, and cohesion intercept, cƍ, for the different mixtures. Lines representing quasi- preconsolidation pressures, σƍqp, evaluated from the oedometer tests, i.e. vertical yield stresses, are also indicated in the diagrams. The results show that the stabilised soils behaved in general in an overconsolidated, or normally consolidated, manner when the consolidation stresses, σƍ1c, used in the triaxial tests was lower or higher than the quasi-preconsolidation pressure, respectively. At consolidation stresses higher than, or of the same order, as the quasi-preconsolidation pressure, relatively high positive pore pressures are generated. This is seen in a distinct turn in the undrained effective stress paths towards the failure envelope of the drained tests, where failure occurs close to the point where this line is reached. In tests where the consolidation stresses were well below the quasi-preconsolidation pressure, the stress paths approached the failure envelop line more gradually and failure occurred at a point below or around the state where the effective vertical stress equals the quasi- preconsolidation pressure. This stress dependency of the undrained compressive strength, as well as comparisons between results of unconfined compression tests and the triaxial tests, has been described in a previous study (Åhnberg 2006). However, it should be noted that most of the quasi-preconsolidation pressures were evaluated from single oedometer tests and, as an effect of the scatter in results, varying deviations from
Åhnberg, H., On yield stresses and the influence of curing stresses… 6 Fig. 2. Measured stress paths in the s´:t stress plane for (a.) stabilised Linköping clay and (b.) stabilised Löftabro clay. (s.p. =stress path)
(a) 500 800 Linköping clay Lime 100 kg/m3 Linköping clay Cement 28 days
CAU CAD CAU CAD s.p. q q max yield s.p. q max q yield 400 3 1 day 600 50 kg/m 28 days 100 kg/m3 1 year 200 kg/m3 300 c'=54-230 kPa φ'=28-38o
)/2, kPa 400 )/2, kPa 3 3 ' ' qmax (drained triaxial tests) c'=5-13 kPa σ σ σ σ σ σ σ σ -
- o 1
1 φ '=29-33 ' ' 200 σ σ σ σ σ σ σ σ ( (
s.p. (undrained triaxial tests) 200 100 qyield (drained triaxial tests) ( ) σ'qp (oedometer tests) 0 0 0 100 200 300 400 500 600 700 0 200 400 600 800 1000 σ σ σ σ (σ'1+σ'3)/2, kPa ( '1+ '3)/2, kPa 500 500 3 Linköping clay Cement 100 kg/m Linköping clay 28 days
CAU CAD CAU CAD s.p. q q 400 max yield 400 s.p. q max q yield 1 day Slag-lime 28 days Cement-lime 1 year Cement-fly ash c'=63-140 kPa Cement-slag 300 φ'=31-35o 300 c'=12-125 kPa φ'=35-37o )/2, kPa )/2, )/2, kPa 3 3 ' ' σ σ σ σ σ σ σ σ - - 1 1 ' 200 ' 200 σ σ σ σ σ σ σ σ ( (
100 100
0 0 0 100 200 300 400 500 600 700 0 100 200 300 400 500 600 700 σ σ (σ'1+σ'3)/2, kPa (σ'1+σ'3)/2, kPa
(b)
500 500 Löftabro clay Lime 28 days Löftabro clay Lime 100 kg/m3
CAU CAD CAU CAD s.p. q q s.p. q max q yield c'=15-20 kPa max yield 400 3 o 400 50 kg/m φ'=34-36 1 day 100 kg/m3 28 days 200 kg/m3 1 year 300 300
q (drained triaxial tests) c'=15-175 kPa max )/2, kPa )/2, kPa o 3 3
' φ ' '=30-34 σ σ σ σ σ σ σ σ - - 1 1 ' ' 200 200 σ σ σ σ σ σ σ σ ( (
s.p. (undrained triaxial tests)
100 qyield (drained triaxial tests) 100
σ'qp (oedometer tests)
0 0 0 100 200 300 400 500 600 700 0 100 200 300 400 500 600 700 σ σ σ σ (σ'1+σ'3)/2, kPa (σ'1+σ'3)/2, kPa 500 Löftabro clay Cement 100 kg/m3 CAU CAD s.p. q q 400 max yield 1 day 28 days c'=85 kPa 1 year φ'=32o 300 )/2, kPa 3 ' σ σ σ σ - 1
' 200 σ σ σ σ (
100
0 0 100 200 300 400 500 600 700 σ σ (σ'1+σ'3)/2, kPa
Åhnberg, H., On yield stresses and the influence of curing stresses… 7 this behaviour can be observed. Furthermore, assessment of preconsolidation pressures was possible only for values lower than or about 600 kPa due to limitations in the maximum applied stress. No significant differences in evaluated effective friction angles were observed when using different types of binder. The evaluated effective strength parameters, cƍ and φƍ, varied from 5 to 175 kPa and from 30° to 36°, respectively in the Linköping clay, and from 15 to 230 kPa and 28° to 38° respectively, in the Löftabro clay. The evaluation of effective strength parameters for each series is based on the measured strengths of a limited number of specimens, two to four, with varying degree of scatter in the results. However, the variation in effective friction angle can be regarded as minor in comparison with that of the cohesion intercept. The cohesion intercept varied considerably, displaying a clear increase with time and with binder quantity. This increase in cƍ is closely related to the simultaneous increase in undrained strength (Åhnberg 2006). A significant exception is the lime-stabilised Linköping clay. The use of lime as a binder had a very small stabilising effect in this clay. No further improvement was measured between one day and one year. This lack of effect has been observed in other investigations on this clay (Åhnberg et al. 2003). Another exception from the general pattern of an increase in cƍ with increasing time and binder quantity, is the lime-stabilised Löftabro clay, which showed no difference in strength for binder quantities corresponding to 50, 100 or 200 kg/m3. However, this is common when using lime as binder. Effects of larger quantities of lime are, as a rule, observed mainly as a more accentuated long-term strength increase. As the stress paths in the drained tests approached or passed a point where the effective stresses corresponded approximately to the quasi-preconsolidation pressures, a distinct break in the stress-strain curves could be observed in several of the tests, see Fig. 3. This yielding should, however, not be confused with the yield point sometimes used
Fig. 3. Examples of specimens exhibiting distinct yielding (σƍ3c=80-240 kPa), and one that does not (σƍ3c=20 kPa), in drained triaxial tests.
σ ' = 20 kPa Löftabro clay 3c σ ' = 80 kPa Lime 100kg/m3 3c σ 28 days ' 3c = 160 kPa σ ' 3c = 240 kPa
500 , kPa 3 ' σ σ σ σ - 1 ' σ σ σ σ
0 0 5 10 15 20
ε1, %
Åhnberg, H., On yield stresses and the influence of curing stresses… 8 to define failure. The measured strains at this breakpoint are low, only a few tenths of a percent, whereas failure in these cases was observed at considerably greater strains. This behaviour was observed in practically all drained tests conducted on samples of relatively low strength. At the lowest effective confining stress, 20 kPa, this first yield point occurred for strength levels up to about 100 kPa. In stronger specimens no distinct yielding was observed at the lowest confining stresses. The yielding becomes less distinct as the strength level increases and occurs, if at all, at stresses and strains closer to failure. The stresses at which a first yielding occurred are indicated in the diagrams in Fig. 2 by symbols of reduced size compared with those representing maximum stresses. The variation in these yield stresses observed in the triaxial tests in relation to the quasi- preconsolidation pressures determined by the oedometer tests was quite large, indicating that the stresses at first yield evaluated in drained triaxial tests should only be taken as an indication of the approximate level of the quasi-preconsolidation pressure. When further analysing the stress paths, a phase transformation point (Ishihara et al. 1975), at which the stress path in overconsolidated samples changes its direction to more or less follow that of the critical state line, could be observed in a number of undrained tests. This point, could be alternatively determined as the point at which the change in pore pressure is zero (e.g. Holmén 2003), which has been shown to be a useful way of determining the critical state line in different natural soils. In the stabilised soils, the points of zero change in pore pressure indicated a yield surface of a significantly different shape. The method was thus found not to be useful for stabilised soils. The yielding model proposed by Larsson (1977) for natural clays was found to be applicable to the stabilised soils. In this model the limit state curves are schematically given as four segments, two of which correspond to the failure strength envelopes in compression and extension, and two of which correspond to σƍv= σƍvc and σƍh= K0nc·σƍvc. A similar yield model relevant for clays of varying microstructure and also unsaturated clays was proposed by Leroueil and Barbosa (2000), in which the horizontal yield line corresponds to σƍh= K*·σƍp, where K* is a function of the effective friction angle, φƍnc, normally yielding a slightly higher value than K0nc in natural clays. However, the latter function for natural soils was found not to be applicable to stabilised materials. The use of the former model is demonstrated in Fig. 4, where results are shown from active and passive tests performed on Linköping clay stabilised with lime-cement at a quantity of 70 kg/m3. The evaluated effective friction angle in the active tests is of the same order as the values displayed in Fig. 2. For the passive tests, the effective stress parameters indicated in the diagram are similar to those evaluated from the active tests. The influence of a quasi-preconsolidation pressure can also be observed in the passive tests. For these samples, which had been stored in sample tubes with no external stresses, an approximately isotropic yield surface could be adopted, described by an apparent K0nc close to unity. For comparison, stress paths and failure envelopes of the natural, unstabilised, clay are also shown in the figure. The effects of lime-cement stabilisation can be observed clearly in the formation of a significant cohesion intercept and a much higher quasi-preconsolidation pressure than that of the natural soil, whereas a change in effective friction angle is less evident.
Stress paths correlated to the quasi-preconsolidation pressure From the results it is evident that the quasi-preconsolidation pressure affects the stress paths to a high degree and is closely associated with the undrained strength of the stabilised soils. To study this further, the stress paths can be normalised with respect to
Åhnberg, H., On yield stresses and the influence of curing stresses… 9 Fig. 4. Measured stress paths in the s´:t stress plane in active and passive triaxial tests on stabilised and unstabilised Linköping clay. (s.p. = stress path)
the quasi-preconsolidation pressures evaluated from the oedometer tests. This normalisation provided a congruent picture of the stress paths for the different types of stabilised soils. Figure 5 shows the stress paths normalised to the evaluated quasi- preconsolidation pressures for all the active tests and for the active and passive tests performed on the mixtures of cement-lime stabilised Linköping clay using binder quantities of 70 kg/m3. The results from lime stabilised Linköping clay are here omitted as hardly any effect of the binder was observed, and this clearly deviated with regard to the magnitude of the normalised cohesion intercept (cf. Fig. 2a). The normalisation of the stress paths for this material indicated behaviour closer to that of natural clay. In cases where the consolidation stresses applied in the triaxial testing were higher than the value of σƍqp evaluated from the oedometer tests, the specimens were considered as being approximately normally consolidated. The stress paths of these specimens are shown by dotted lines and failure by open circles in the figures. The results shown in Fig. 5 for the different samples all fall into a common pattern, with a cƍ of approximately 0.12-0.16·σƍqp and a φƍ of approximately 33°-34°. The undrained compressive strength, i.e. the deviator stress at failure, varies from approximately 0.6·σƍqp to 1.0·σƍqp for highly overconsolidated samples to normally consolidated samples, respectively. In Fig. 5b it can be seen that in the passive tests, the “normally consolidated” specimens, i.e. those having been consolidated at stresses higher than the quasi-preconsolidation pressure measured in the oedometer tests, indicate a horizontal yield surface corresponding to a somewhat lower K0nc value than the others. These specimens had been consolidated at a K0 of 0.5, which affects the results. However, the effect of the cementation has a greater impact on the results and being of isotropic nature constrains the change in apparent K0nc to about 0.85.
Åhnberg, H., On yield stresses and the influence of curing stresses… 10 Fig. 5. Measured stress paths normalised to the quasi-preconsolidation pressures in the s´:t stress plane in (a.) triaxial compression tests on stabilised Linköping clay (lime stabilised samples omitted) and Löftabro clay and (b.) triaxial compression and extension tests on cement-lime stabilised Linköping clay. (a) 1.5 CAU CAD
s.p q failure q max q yield Stab. Linköping clay Stab. Löftabro clay Stab. Linköping clay Stab. Löftabro clay 1 qp ' σ σ σ σ σ c'=(0.14-0.16) 'qp
)/2/ o
3 φ
' '=33-34 σ σ σ σ - 1 ' σ σ σ σ (
0.5
σ ' 3 =0 σ σ "oc" ( 'vc < 'qp) σ σ "nc" ( 'vc ¡Ãσ'qp) d 'qp)
0 00.511.52 σ σ σ (σ'1+σ'3)/2/σ'qp
(b) 1 Linköping clay Cement-lime c'=0.12σ'qp CAU CAD φ'=34o s.p. q failure q failure q yield
0.5 qp ' σ σ σ σ )/2/ h ' σ σ σ σ - "oc" (σ'vc < σ'qp) v ' σ σ σ σ σ σσ ( "nc" ( 'vc ¡Ãd ''qpqp)) 0 00.511.52
σ ' v =0
-0.5
(σ'v+σ'h)/2/σ'qp
Åhnberg, H., On yield stresses and the influence of curing stresses… 11 When the specimens had been consolidated at low confining stresses, the stress paths approached and came close to the zero minor principal effective stress line. At this line, the stabilised specimens may fracture unless a tensile strength can be mobilised. In these highly “overconsolidated” specimens, negative pore pressures as a rule started to develop as the stress paths approached the failure envelope. Also the volumetric strains for these specimens showed that dilation started at points somewhat below the indicated failure line. This behaviour results in a slightly rounded shape of the yield surface close to origin, in contrast to the straight lines in the model. The same effect of a quasi-preconsolidation pressure as in the stabilised clays could also be observed in previously performed tests on stabilised organic soils. Figure 6 shows stress paths measured in triaxial tests on gyttja and peat stabilised with similar types of binder as in the present project, although at a higher quantity. These soils had been tested with respect to the stabilisation effects of different binders in connection with the EC-funded project EuroSoilStab (Åhnberg et al. 2000). They were a gyttja from Holma Mosse in Östergötland, in the eastern part of Sweden, and a peat from Dömle Mosse in Värmland, in the western part of Sweden. The Holma gyttja is a clayey gyttja with a water content of 220% and a liquid limit of 170%. The organic content is 10%. The natural peat has a high water content of about 2000% and is classified as H5–H7 according to the von Post scale (von Post 1924, e.g. Landva and Pheeney 1980). The effective friction angles were, however, roughly about the same as those evaluated from the tests on the stabilised clay. The stabilised peat samples had been stored in sample tubes subjected to a load of 18 kPa applied on top of the samples shortly after mixing, simulating preloading corresponding to about one metre of fill in the field. Two extension triaxial tests were performed on stabilised peat samples. However, whether the
Fig. 6. Measured stress paths normalised to the quasi-preconsolidation pressures in the s´:t stress plane for the stabilised organic soils.
1 Dömle peat Cement-lime,Cement-slag Stab. peat Holma gyttja 28 days-1 year Stab. gyttja
c'=(0.22-0.24)σ'qp φ'=31-34o
0.5 qp ' σ σ σ σ )/2/ h ' σ σ σ σ - v ' σ σ σ σ ( 0 00.511.52
CAU CAD
q failure q failure Stab. peat Stab. gyttja -0.5
(σ'v+σ'h)/2/σ'qp
Åhnberg, H., On yield stresses and the influence of curing stresses… 12 initial loading, which was relatively small, corresponding to only about 10% of the quasi-preconsolidation pressure, had any anisotropic effect on the yield stresses could not be interpreted from the results. Normalisation of the stresses in the different stress planes provided a useful tool for studying the stress response of the different types of stabilised soils. In spite of some scatter in the measurements of, foremost the quasi-preconsolidation pressures, but also the strength for individual test series, a congruent picture of the strength behaviour can be observed. Considering the results of all the tests, the drained strength of the different types of stabilised clays could be approximately described by a mean cohesion intercept of 0.15·σƍqp and a mean friction angle of 33°, giving qd ≈ 0.25·σƍqp+ 0.54·(σƍ1 + σƍ3). It may be possible to use this relation in preliminary estimates of drained strength in stabilised clays of similar types. In undrained conditions, the strength is closely related to the magnitude of the consolidation stresses and the quasi-preconsolidation pressures. The variation in undrained compressive strength in stabilised soils has been described by 0.88 the approximate relation qu = 1.0·σƍvc·OCR in previous studies (Åhnberg, 2006). The overconsolidation ratio, OCR, is here the relation σƍqp/σƍvc.
Effects of curing stresses on the strength Subjecting the stabilised samples to stresses during curing resulted in an increase in strength and an increase in quasi-preconsolidation pressure. Figure 7a shows the measured stress paths in undrained triaxial tests performed on the cement-lime-stabilised samples cured for 28 days, with effective vertical curing stresses of 0, 40 and 120 kPa, and corresponding effective horizontal curing stresses of 0, 20 and 60 kPa, respectively. The consolidation stresses during curing were all below the quasi-preconsolidation pressure of about 210 kPa measured on samples not subjected to any curing stresses. However, since the curing stresses were applied shortly after mixing this caused compression of the sample, in turn enabling the cementation processes to take place with the soil and binder particles arranged closer together. This resulted in an increase in quasi-preconsolidation pressure compared with those in samples cured with no external stress applied, of about 38% and 90% after 28 days for the samples cured at effective vertical stress of 40 and 120 kPa, respectively. The increase in cƍ was approximately proportional to the increase in σƍqp, whereas the effective friction angle was about the same for samples cured with and without external stresses. Since the curing stresses are well below the isotropic yield locus of samples cured without stresses, the K0 of 0.5 used during curing did not result in a corresponding ratio for vertical and horizontal yield stresses. However, the K0 used during curing did have an influence, which was observed in that the increase in horizontal yield stress due to curing stresses was only half the increase in vertical quasi-preconsolidation pressure. The ratio between vertical and horizontal yield stresses, K0qp, was approximately 0.86 and 0.76 for the samples cured at effective vertical stresses of 40 kPa and 120 kPa, respectively. The reduced value of K0qp could be expressed approximately as
1+ K ∆σ ′ σ ′ [1] K = 0 qp qp0 0qp 1+ ∆σ ′ σ ′ qp qp0 where σ ′ is the quasi-preconsolidation pressure when the sample is not subjected to qp0 stresses during curing.
Åhnberg, H., On yield stresses and the influence of curing stresses… 13 Fig. 7. Measured effective stress paths (a.) and stress paths normalised to the quasi- preconsolidation pressure (b.) in the s´:t stress plane for stabilised Linköping clay subjected to different curing stresses after mixing. (a)
300 Linköping clay Cement-lime 3 c'=25-63 kPa 70 kg/m o 200 φ'=33-34
100 )/2, kPa h ' σ σ σ σ - v ' σ σ
σ σ 0 ( 0 100 200 300 400 500 600
Curing stress (K0=0.5)
-100 σ'1c= 0 kPa
σ'1c= 40 kPa σ σ'1c= 120 kPa
-200
(σ'v+σ'h)/2, kPa
(b) 1 Linköping clay σ c'=0.14 'qp Cement-lime φ'=33o 70 kg/m3
0.5 qp ' σ σ σ σ )/2/ h ' σ σ σ σ σ σ - "oc" ( 'vc < 'qp) v ' σ σ σ σ "nc" (σ' σ' ) ( vc d qp 0 00.511.52
K 0qp = 0.76, 0.86, 1.0
-0.5
(σ'v+σ'h)/2/σ'qp
Åhnberg, H., On yield stresses and the influence of curing stresses… 14 The results obtained from samples cured at different stresses provided useful information on how stresses at certain depths in situ may affect the strength behaviour compared with that of stabilised soils at shallow depths, or samples of stabilised soils stored in the normal way without external stresses in the laboratory. Further tests of this kind for different strength levels and for different types of soil are needed in order for more general conclusions to be drawn regarding the influence of curing stresses on the strength behaviour. However, the effects of increased stresses can be assumed to be related mainly to compression of the stabilised material. This implies that increased stresses caused by a load, e.g. from an embankment, should preferably be applied in a way that produces a certain compression of the material shortly after stabilisation. If the load is applied with a certain delay, the stabilised soil may have gained sufficient strength to prevent significant compression for a limited load and any desired further strength increase linked to this. This effect of delay in loading has been demonstrated for stabilised peat in earlier investigations (Åhnberg et al. 2001). It is also recommended practice in Sweden to apply a load increment as soon as possible to all types of stabilisation under embankments (e.g. Carlsten 2000).
Changes in stress with strain The stress-strain relation was studied with regard to changes in stresses at larger strains and for assessment of the joint response of stabilised soils and surrounding natural soils. To study the changes in stress with strain during loading for all the different types of samples, normalisation of the stresses with respect to the quasi- preconsolidation pressure was performed. Figure 8 shows examples of the measured variation in normalised shear stress with strain in the undrained and drained tests. The curves shown in Fig. 8 are from measurements performed with external strain gauges and the strains presented are thus only approximate. However, the general pattern is about the same as that obtained by internal measurements. In a large number of the tests,
Fig. 8. Examples of measured stress vs. strain in (a.) undrained and (b.) drained triaxial compression tests on stabilised Linköping clay. (a) (b) 1.5 1.5 σ σ' = 20 kPa Linköping clay '3c = 20 kPa Linköping clay 3c σ σ'3c = 80 kPa '3c = 80 kPa 3 3 σ 100 kg/m σ'3c = 160 kPa 100 kg/m '3c = 160 kPa σ σ'3c = 240 kPa '3c = 240 kPa
1 1 Lime
Cement kPa qp ' 'qp, kPa σ σ σ σ σ σ σ σ /2/ /2/ u
Cement d
q Lime q 0.5 0.5
Unstabilised clay Unstabilised clay
0 0 0 5 10 15 20 0 5 10 15 20 ε ε ε1, % 1, %
Åhnberg, H., On yield stresses and the influence of curing stresses… 15 local measurements of the axial strains were also performed. These measurements were, however, limited to strains smaller than about 10%. Although the results represent a large variation in strength depending on the type of binder etc., similar patterns could be observed in the stress-strain response of the different specimens. In the undrained triaxial tests, after an initial rapid increase in shear stress up to failure, the shear stress levelled off and only a limited change was measured at continued straining after failure. In the drained triaxial tests, on the other hand, a large variation was observed in the stress-strain response. Failure occurred at various values of strain. In cases where failure occurred at small strains, a significant reduction in shear stress with increasing strain after failure was observed. In general, this applied to specimens that in the tests had been consolidated at stresses well below the quasi- preconsolidation pressure, whereas those exhibiting failure at large strains were closer to “normally consolidated”. The variation in normalised stress with strain measured internally is shown schematically in Fig. 9, together with normalised pore pressures and volumetric strains measured in the undrained and drained triaxial tests, respectively. For comparison, results are also shown from triaxial tests performed on unstabilised clay. The shaded ranges represent data measured for samples tested at about the same σƍvc/σƍqp ratio. The influence of this “degree of overconsolidation” on the stress-strain relations, pore pressures and volumetric changes can be clearly seen in the figure. At low σƍvc/σƍqp ratios, the specimens behaved in a dilatant manner, involving increases in volume in the drained tests and the development of negative pore pressures in the undrained tests, as the stresses approached failure. The pore pressures generated in the stabilised soils varied between approximately -0.1σƍqp and 0.5-0.6σƍqp for highly “overconsolidated” and “normally consolidated” specimens, respectively. This corresponded to negative pore pressures of 10-20% and to positive pore pressures of 50-60% of the compressive strength, respectively. The normalised shear stresses and pore pressures generated were significantly higher in the stabilised soils than in the natural, unstabilised soil. The triaxial tests on the Linköping clay were performed at a σƍv/σƍvc ratio of 0.5, which is only slightly lower than the ratio in situ for the natural soil. In Fig. 9b results are also shown from a drained triaxial test performed on the clay at an elevated consolidation stress representing normally consolidated conditions. The strains in the natural clays were measured solely with external gauges. The results indicated that under undrained conditions, the strength of the stabilised soil and the unstabilised soil may be mobilised concurrently in the materials. In addition to the failure strains of the stabilised soil and the unstabilised soil being about the same in many cases, both materials exhibited near ideal-plastic behaviour without a significant reduction in strength after failure. On the other hand, under drained conditions in highly overconsolidated specimens, the stabilised soils exhibited brittle behaviour with a significant reduction in strength after failure, which occurred at a very small strain. The normalisation of the results further elucidated the response of stabilised soils when subjected to loading. A view often expressed in the design of deep mixing, is that the stabilised soil and the natural soil respond in fundamentally different ways, and only a very limited part of the soil strength, if any, can be activated when the composite system of stabilised and unstabilised soil is subjected to loading in situ. This view is probably based primarily on the results from unconfined compression tests, where a distinct peak in the stress-strain curve followed by a considerable reduction in strength is
Åhnberg, H., On yield stresses and the influence of curing stresses… 16 Fig. 9. Schematic variation in normalised shear stress, normalised pore pressure and relative volumetric change vs. axial strain in (a) undrained and (b) drained triaxial compression tests on stabilised Linköping and Löftabro clay. Results on unstabilised Linköping clay are shown for comparison. (a) (b) 1 1.5 σ σ Linköping clay Linköping clay '1c/ 'qp Löftabro clay Löftabro clay
0.9-1.0 σ' /σ' 1c qp 1
0.9-1.0 qp qp ' ' 0.6-0.7 σ σ σ σ σ σ 0.5 0.6-0.7 σ σ
/2/ 0.4-0.5 /2/ u d
q 0.2-0.3 q 0.4-0.5
0.2-0.3 0.03-0.1 0.5 0.5 0.1-0.2 0.5
1.0 0.03-0.1 Unstabilised Linköping clay Unstabilised Linköping clay 0 0 05100510 ε ε , % ε1, % 1 1 -5 Linköping clay Linköping clay σ σ Löftabro clay Löftabro clay '1c/ 'qp σ σ σ'1c/σ'qp 0.03-0.1 0.9-1.0 0510 0.5 0 0.1-0.2
0.7-0.8 qp
' 0.3-0.4 σ σ σ σ 0.4-0.5 , % v u/ ε ε ε ε 0.3-0.4 0.5 0.6-0.7 0 0.2-0.3 0.1-0.2 5 0.03-0.1 0510 0.9-1.0 1.0 Unstabilised Linköping clay Unstabilised Linköping clay -0.5 10 ε ε ε1, % 1, %
often observed. The unconfined compression test may be regarded as an undrained test. However, at failure cracks start to develop and the lack of a confining support then results in a major loss in strength. In deep mixing, the confining stresses prevailing in the field are likely to bring about more tenacious behaviour with a more limited reduction in the undrained strength with strain in the stabilised soil, as well as in the natural soil. However, under drained conditions in stabilised soil of high strength and at low confining stresses, the failure strains may be low, and the reduction in strength at larger strains more pronounced, resulting in full strength not being mobilised simultaneously in the stabilised and unstabilised soil.
Conclusions The results of the investigation performed on the two clays stabilised with different types of binder showed that triaxial tests and oedometer tests are useful tools for studying the strength behaviour of the materials.
Åhnberg, H., On yield stresses and the influence of curing stresses… 17 • Quasi-preconsolidation pressures, σƍqp, governed by cementation effects as well as by previous curing stresses, were observed in the triaxial tests as well as in the oedometer tests. The influence of this quasi-preconsolidation pressure was observed in compression as well as extension triaxial tests. The quasi-preconsolidation pressure affects the stress paths and is closely associated to the strength of the stabilised soils. The results show that the stabilised soils behave in an overconsolidated manner when the consolidation stresses are significantly lower than the quasi-preconsolidation pressures, and in a normally consolidated manner when the consolidation stresses are of the order of 0.8 to 1.0·σƍqp. No tendencies towards differences in the general strength behaviour when using different types of binder were observed. • A common yielding model for natural clays (Larsson 1977) was also found suitable for describing the behaviour observed in the tests on stabilised soils. The yield surface is schematically described by the failure strength envelopes in compression and extension and by σƍv= σƍqp and σƍh= Κ0ncσƍqp. An isotropic yield surface could be adopted for the samples cured in the normal way, without being subjected to external stresses. • Normalisation of the deviator stress with respect to the quasi-preconsolidation pressure was effective for studying the behaviour of different types of stabilised soils. The undrained compressive strength varied from approximately 0.6·σƍqp to 1.0·σƍqp for highly overconsolidated samples and normally consolidated samples, respectively, while the variation in drained strength of the different types of stabilised clays could be expressed by a mean cohesion intercept of 0.15·σƍqp and a mean friction angle of 33°. • Stresses applied shortly after mixing will compress the stabilised soil and result in increased strength and increased quasi-preconsolidation pressure. A K0 value lower than unity applied during curing results in a ratio of the horizontal-to-vertical yield stresses, K0nc, lower than that of samples cured without stresses. • Although the results represent a large variation in strength depending on the types of soil, binder, binder quantity and curing time, consistent patterns could be observed in the stress-strain relations. The behaviour is linked to the degree of overconsolidation. In undrained tests there is only limited further change in deviator stress at continued loading after failure. In drained tests on specimens consolidated at stresses well below the quasi-preconsolidation pressure, failure occurs at small strains and a significant reduction in shear stress with strain after failure is observed. Stabilised samples that are close to “normally consolidated” exhibit failure at large strains.
Acknowledgements
Financial support for this study was provided by the Swedish Geotechnical Institute (SGI) and the Swedish Deep Stabilization Research Centre (SD). These contributions, as well as permission of the participants within the European Community project EuroSoilStab under the Brite EuRam contract BRPR-CT97-0351, to present data from this project, are gratefully acknowledged.
Åhnberg, H., On yield stresses and the influence of curing stresses… 18 References
Åhnberg, H. 2004. Effects of back pressure and strain rate used in triaxial testing of stabilised organic soils and clays. Geotechnical Testing Journal. Vol. 27, No. 3. pp. 250-259. Åhnberg, H. 2006. Effects of consolidation stresses on the strength of some stabilised Swedish soils. Ground Improvement, Vol. 10, No. 1, pp. 1-13. Åhnberg, H., Johansson, S-E., Retelius, A., Ljungkrantz, C., Holmqvist, L., Holm, G. 1995. Cement and lime for deep stabilisation of soil. Swedish Geotechnical Institute, Report No. 48. 213 p. (In Swedish). Åhnberg H., Bengtsson P-E. and Holm G. 2000. Task 2.2 and 7. Laboratory tests of mixtures – Test sites Dömle Mosse and Holma Mosse, Sweden. National report, part 1. Internal Report. EuroSoilStab project, SGI 3-9705-239 and 3-9708-395. 35 p. Åhnberg, H., Bengtsson, P-E. and Holm G. 2001. Effect of initial loading on the strength of stabilised peat. Ground Improvement, Vol 5, No 1, pp. 35 – 40. Åhnberg H., Johansson S-E., Pihl H. and Carlsson T. 2003. Stabilising effects of different binders in some Swedish soils. Ground Improvement, Vol. 7, No. 1, pp. 9- 23. Babasaki, R., Terashi, M., Suzuki, T., Maekawa, A., Kawamura, M. and Fukazawa, E. 1996. JGS TC Report: Factors influencing the strength of improved soil. In Proceedings of the 2nd Int. Conference on Ground Improvement Geosystems – IS Tokyo’96. Grouting and Deep Mixing, Tokyo, 1996. Vol. 2, pp. 913-918. Balasubramaniam, A.S. and Buensuceso, B.R. 1989. On the overconsolidated behavior of lime treated soft clay. In Proceedings of the XII International Conference on Soil Mechanics and Foundation Engineering. Rio de Janeiro 1989. pp. 1335-1338. Balasubramaniam, A.S., Buessucesco, B., Oh, E.Y.N., Bolton, M., Bergado, D.T. and Lorenzo, G. Strength degradation and critical state seeking behaviour of lime treated sofl clay. In Proceedings of the International Conference on Deep Mixing – Best Practice and Recent Advances, Deep Mixing’05, Stockholm 2005, pp. 35-40. Berre, T. 1997. Methods for consolidated triaxial compression tests on water saturated soils. European Technical Committee, Document number ETC5-F1.97, pp.75–86. Bishop, A.W. and Henkel, D.J. 1962. The measurements of soil properties in the triaxial tests. Second edition. Edward Arnold Ltd, London 1962. Broms, B. 2000. Lime and lime/cement columns. Summary and visions. Keynote lectures NGM-2000 and 4th GIGS, Helsinki 2000, pp. 43-93. Carlsten, P. and Ekström, J. 1995. Lime and lime cement columns. Guide for project planning, construction and inspection. Swedish Geotechnical Society Report 4:95E. Carlsten, P. 2000. Lime and lime cement columns. Guide for project planning, construction and inspection. Swedish Geotechnical Society Report 2:2000. (In Swedish) CEN standard 2000. CEN standard EN197-1:2000. Cement – Part 1: Composition, specifications and conformity criteria for common cements. European Committee for standardization (CEN), Brussels. CEN standard 2001. CEN standard EN459-1:2001. Building lime – Part I: Definitions, specifications and conformity criteria. European Committee for standardization (CEN), Brussels. Consoli, N.C., Rotta, G.V. and Prietto, P.D. M. 2000. Influence of curing under stress on the triaxial response of cemented soils. Géotechnique, Vol. 50, No. 1, pp. 99-105.
Åhnberg, H., On yield stresses and the influence of curing stresses… 19 deGroot, D.J. and Sheahan, T.C. 1995. Laboratory methods for determining engineering properties of overconsolidated clays. Engineering properties and practice in overconsolidated clays. Transportation Research Record, No. 1479, pp. 17-25. ETC5, 1998, Recommendations of the ISSMGE for geotechnical laboratory testing. Beuth Verlag, Berlin. Holmén, M. 2003. Advanced triaxial testing of clay till. Licentiate Dissertation. Soil Mechanics and Foundation Engineering. Lunds Institute of Technology, Lund University. Horpibulsuk S., Miura, N. and Bergado, D. T. 2004. Undrained shear behavior of cement admixed clay at high water content . Journal of Geotechnical and Geoenvironmental Engineering, Vol. 130, No. 10, pp. 1096-1105. Ishihara K. Tatsuoka F. And Yasuda S. 1975. Undrained deformation an liquefaction in sand under cyclic stress. Soils and Foundations, Vol. 15, No. 1. Kasama K., Ochiai, H. and Yasufuku, N. 2000. On the Stress-Strain Behaviour of Lightly Cemented Clay Based on an Extended Critical State Concept. Soils and Foundations, Vol. 40, No.5, pp. 37-47. Kawasaki, T., Saitoh, S., Suzuki, Y. and Babasaki, R. 1984. Deep mixing method using cement slurry as hardening agent. In Proceedings of the Seminar on Soil Improvement and Construction Techniques in Soft Ground, Singapore 1984. pp. 17- 38. Kitazume, M. and Terashi, M. 2002. The Deep Mixing Method – Principle, Design and Construction. Ed. Coastal Development Institute of Technology (CDIT), Japan. A.A. Balkema Publishers, Tokyo. 123 pp. Lacasse, S. and Berre, T. 1988. Triaxial testing methods for soils. Advanced triaxial testing of soil and rock. ASTM Special Technical Publications STP 977, pp. 264-289. Landva , A.O. and Pheeney, P.E., 1980, “Peat fabric and structure”. Canadian Geotechnical Journal, Vol. 17, No. 3, pp.416-435. Larsson, R. 1977. Basic behaviour of Scandinavian soft clays. Swedish Geotechnical Institute, Report No. 4, Linköping. Larsson, R. 2000. Investigations and load tests in clay till. Swedish Geotechnical Institute, Report No. 59, Linköping. Lee, K. Chan D. and Lam K. 2004. Constitutive Model for Cement Treated Clay in a Critical State Frame Work. Soils and Foundations, Vol. 44, No.3, pp. 69-77. Leroueil, S. and Barbosa, P.S. de A. 2000. Combined effect of fabric, bonding and partial saturation on yielding of soils. In Proceedings of the Asian Conference on Unsaturated Soils, Singapore, pp. 527-532. Lowe, J. and Johnson, T.C. 1960. Use of back pressure to increase degree of saturation of triaxial test specimens. In Proceedings of the Research Conference on Shear Strength of Cohesive Soils. Boulder 1960. pp. 819-836. von Post, L., 1924, “Das genitische system der organogenen bildungen Schwedens”. Mémoires Nomenclat. et Classific. Sols, Comité int. Pédologie, 4 Commission No 22. pp. 287-304. Rotta, G.V. Consoli, N.C., Prietto, P.D.M., Coop, M.R. and Graham. J. 2003. Isotropic yielding in an artificially cemented soil cured under stress. Géotechnique, Vol. 53, No. 5, pp. 493-501. Swedish Standard 1991. SS 02 71 26, Geotechnical tests - Compression properties - Oedometer test, CRS-test - Cohesive soil. Swedish Standards Institute, SIS Förlag AB, Stockholm. (In Swedish.)
Åhnberg, H., On yield stresses and the influence of curing stresses… 20 Swedish Standard 1992. SS 02 71 29, Geotechnical tests - Compression characteristics - Oedometer test incremental loading - Cohesive soil. Swedish Standards Institute, SIS Förlag AB, Stockholm. (In Swedish). Tatsuoka, F. and Kobayashi, A. 1983. Triaxial strength characteristics of cement-treated soft clay. In Proceedings of the 8th European Conference on Soil Mechanics and Foundation Engineering, Helsinki 1983, Vol. 1, pp. 421-426. Tremblay, H., Leroueil, S. and Locat, J. 2001. Mechanical improvement and vertical yield stress prediction of clayey soils from eastern Canada treated with lime or cement. Canadian Geotechnical Journal, Vol. 38, pp. 567-579. Uddin K. and Buensuceso B. R. 2002. Lime Treated Clay: Salient Engineering Properties and a Conceptual Model. Soils and Foundations, Vol. 42, No.5, pp. 79-89. Yamamoto, T., Suzuki, M., Okabayashi, S., Fujino, H., Taguchi, T. and Fujimoto, T. 2002. Strength and deformation characteristics of cement-stabilised soil cured under an overburden pressure. In Proceedings of the 4th International Conference on Ground Improvement Techniques, Kuala Lumpur 2002. pp. 761-766. Yin, J.-H. (2001). Properties and behaviour of raw sludge mixed with PFA and lime. Geotechnical Testing journal (ASTM), Vol. 24, No. 3, pp. 299-307. Zhou, Cheng, Yin, J.-H. and Ming, J.-P. (2002) Bearing capacity and settlement of weak fly ash ground improved using lime-fly sash or stone columns. Canadian Geotechnical Journal, Vol. 39, No. 3, pp. 585-596. Zhu, F., Clark, J.I. and Paulin, M.J. 1995. Factors affecting at-rest lateral stress in artificially cemented sands. Canadian Geotechnical Journal, Vol. 32, No. 2, pp. 195- 203.
Åhnberg, H., On yield stresses and the influence of curing stresses… 21
Rapport
1. Erfarenhetsbank för kalk-cementpelare. 1997 Torbjörn Edstam 2. Kalktypens inverkan på stabiliseringsresultatet. En förstudie. 1997 Helen Åhnberg & Håkan Pihl 3. Stabilisering av organisk jord med 2000 cement- och puzzolanreaktioner Karin Axelsson, Sven-Erik Johansson & Ronny Andersson 4. Provbank på kalk/cementpelarförstärkt gyttja och 1999 sulfidhaltig lera i Norrala Rolf Larsson 5. Masstabilisering 2000 Nenad Jelisic 6. Blandningsmekanismer och blandningsprocesser 2000 – med tillämpning på pelarstabilisering Stefan Larsson 7. Deformation Behaviour of Lime/Cement Column Stabilized Clay 2000 Sadek Baker 8. Djupstabilisering med kalkcementpelare 2001 – metoder för produktionsmässig kvalitetskontroll i fält Morgan Axelsson 9. Olika bindemedels funktion vid djupstabilisering 2001 Mårten Janz & Sven-Erik Johansson 10. Mitigation of Track and Ground Vibrations by High Speed Trains 2002 at Ledsgård, Sweden Göran Holm, Bo Andréasson, Per-Evert Bengtsson, Anders Bodare & Håkan Eriksson 11. Miljöeffektbedömning (LCA) för markstabilisering 2003 Tomas Rydberg & Ronny Andersson
12. Mixing Processes for Ground Improvement by Deep Mixing 2004 Stefan Larsson 13. Proceedings of the International Conference on Deep Mixing 2005 – Best Practice and Recent Advances, Deep Mixing´05 Stockholm, May 23 – 25, 2005 14. Stabilisering av torv i laboratorium 2006 – Metodbeskrivning & Underlagsrapport Martin Holmén 15. Provbankar Riksväg 45/Nordlänken. 2006 Bankar på kalkcementpelarförstärkt jord – Beräkningsmodell för sättningar Claes Alén, Göran Sällfors, Per-Evert Bengtsson & Sadek Baker Svensk Djupstabilisering Swedish Deep Stabilization Research Centre c/o SGI, SE-581 93 Linköping Phone: +46 13 20 18 61, Fax: +46 13 20 19 14 Internet: www.swedgeo.se/sd