Abstract. Keywords: Liquefaction, Silt, Non-Plastic Fines, Fines Correction, Simplified Procedure

Transcription

1 Abstract It has been understood since the 1960's that the presence of silt and clay particles will in some manner affect the resistance of a sand to liquefaction. However, a review of studies published in the late literature shows that no clear conclusions can be drawn as to in what manner altering the fines content affects the liquefaction resistance of a sand under cyclic loading. This is particularly true for soils containing non-plastic, i.e. silty fines. An attempt of clarification of the effects of non-plastic fines on the liquefaction susceptibility of sandy soils is presented here in light of a recent study conducted by Martin and Polito (2001), and a recent conceptual framework proposed by Thevanayagam (2000). The assessment of triggering potential is also focused in this document with a review of the latest updates of in situ liquefaction hazard evaluation methods. Since the influence of nonplastic fines on liquefaction susceptibility is the main subject of this work, a detailed presentation of the correction factors, often used in simplified liquefaction evaluation methods to take into account the affect of fines content on the soil, is made for each in situ test presented. Finally, the influence of the presence of nonplastic fines might have on the methods for liquefaction analysis for sands currently used in engineering practice is analyzed and some suggestions will be made on how to evaluate liquefaction potential in these cases, based on the latest studies on the matter. Keywords: Liquefaction, Silt, Non-Plastic Fines, Fines Correction, Simplified Procedure I

2 Resumo É sabido deste a década de 1960 que a presença de partículas de silte ou argila afectam de algum modo a resistência de uma areia à liquefacção. No entanto, uma revisão dos estudos publicados na literatura mostra que não há um consenso sobre de que forma a percentagem de finos afecta a resistência à liquefacção de uma areia submetida a um carregamento cíclico. Isto é particularmente verdade para solos com finos não-plásticos do tipo silte. Uma tentativa de clarificação sobre os efeitos de finos não-plásticos na susceptibilidade à liquefacção de solos arenosos, é apresentada neste trabalho com base num recente estudo feito por Martin e Polito (2001), e por um recente modelo conceptual desenvolvido por Thevanayagam (2000). A avaliação do potencial de liquefacção é também focada neste trabalho com uma revisão dos últimos avanços nos métodos de avaliação da susceptibilidade de liquefacção com base em ensaios in situ. Uma vez que a influência de finos não-plásticos na susceptibilidade à liquefacção é o tema principal deste trabalho, uma apresentação detalhada dos factores de correcção, actualmente utilizados para ter em conta o efeito da percentagem de finos no solo, é feita para cada ensaio in situ apresentado. Finalmente, analisa-se a influência que finos não-plásticos em areia pode ter nos métodos para análise da liquefacção usados actualmente na prática de engenharia e apresentam-se sugestões de como avaliar o potencial de liquefacção nesses casos, baseadas nos últimos estudos sobre o tema. Palavras-chaves: Liquefacção, Silte, Finos Não-Plásticos, Correcção de Finos, Método Simplificado II

3 TABLE OF CONTENTS LIST OF TABLES... V LIST OF FIGURES... V LIST OF SYMBOLS... VII CHAPTER ONE: INTRODUCTION Scope of this Document Outline of this Document...2 CHAPTER TWO: LITERATURE REVIEW Liquefaction Susceptibility Historical Criteria Geologic Criteria Compositional Criteria State Criteria Critical Void Ratio Steady State of Deformation State Parameter Initiation of Liquefaction Flow Liquefaction Surface Influence of Excess Pore Pressure Characterization of Earthquake Loading Characterization of Liquefaction Resistance Evaluation of Initiation of Liquefaction...20 CHAPTER THREE: ASSESSMENT OF TRIGGERING POTENTIAL Standard Penetration Test (SPT) Previous SPT-based Correlations New SPT-based Correlations Adjustments for Fines Content Magnitude-Correlated Duration Weighting Additional Adjustments for Effective Overburden Stress Use of the Proposed SPT-based Correlations Cone Penetration Test (CPT) Previous CPT-based Correlations New CPT-based Correlations Normalization of CPT Tip and Sleeve for Effective Overburden Stress Thin Layer Correction Adjustment for Fines Content SPT vs CPT V S -Based Triggering Correlations...44 III

4 CHAPTER FOUR: SOILS WITH NON-PLASTIC FINES Literature Review -The Effects of Fines Content on Liquefaction Resistance The Effects of Non-Plastic Fines Content The Effects of Non-Plastic Fines Cyclic Resistance Changes in Soil Specific Relative Density with Increasing Silt Content Decreasing Cyclic Resistance with Increasing Silt Content Decreasing and then Increasing Cyclic Resistance with Increasing Silt Content Cyclic Resistance with Constant Sand Skeleton Void Ratio Increasing Cyclic Resistance with Increasing Silt Content Interpretation of Results Conclusions...61 CHAPTER FIVE: ASSESSMENT OF TRIGGERING POTENTIAL OF SOILS WITH NON- PLASTIC FINES Liquefiable Soil Types In Situ Testing Soils Below the Limiting Silt Content SPT-based Correlations CPT-based Correlations Soils Above the Limiting Silt Content Parametric Analysis...72 CHAPTER SIX: CONCLUSIONS...76 REFERENCES...78 APPENDIX...81 IV

5 LIST OF TABLES Table 1 - Liquefaction susceptibility of silty and clayey sands (Andrews and Martin, 2000)...63 Table 2 Medium Grain Diameter for mixtures of Yatesville sand with silt...73 Table 3 Parametric Analysis Results for the CPT correlations...74 Table 4 - Parametric Analysis Results for the SPT correlations (converted from the CPT)...74 Table Appendix 1 Recommended Corrections for SPT Equipment Energy and Procedures...85 LIST OF FIGURES Fig. 2.1 Void Ratio variation with confining stress...4 Fig. 2.2 Undrained tests in saturated sand...5 Fig. 2.3 Relationship between limiting epicentral distance of sites at which liquefaction has been observed and moment magnitude for shallow earthquakes...6 Fig. 2.4 Modified Chinese Criteria (After Wang (1979) and Seed and Idriss (1982))...7 Fig. 2.5 Stress-strain and stress-void ratio curves for loose and dense sands at the same effective confining pressure...8 Fig. 2.6 Use of the CVR line as a boundary between loose contractive states and dense dilative states..9 Fig. 2.7 Liquefaction, Limited Liquefaction, and Dilation in monotonic leading tests...9 Fig. 2.8 State criteria for flow liquefaction susceptibility...10 Fig. 2.9 State parameter...12 Fig Response of isotropically consolidated specimen of loose, saturated sand: (a) stress-strain curve; (b) effective stress path; (c) excess pore pressure; (d) effective confining pressure...13 Fig Response of five specimens isotropically consolidated to the same initial void ratio at different initial effective confining pressures...14 Fig Orientation of the flow liquefaction surface in stress path space...15 Fig Variation of flow liquefaction surface inclination with initial principal effective stress ratio for constant void ratio...15 Fig Zone of susceptibility to flow liquefaction...16 Fig Zone of susceptibility to cyclic mobility...16 Fig Typical irregular time history of shear stress...18 Fig Cyclic stresses required to produce initial liquefaction and 20% axial strain in isotropically consolidated Sacramento River Sand triaxial specimens Fig Process by which zone of liquefaction is identified in the cyclic stress approach...20 Fig Process by which the zone of liquefaction is identified in the cyclic strain approach...20 Fig Correlation between equivalent uniform cyclic stress ratio and SPT N 1,60-Value for events of Magnitude M w=7,5 for varying slit contents, with adjustments at low cyclic stress ratio as recommended by NCEER Working Group (Modified from Seed et al., 1984)...23 Fig Probabilistic Correlation for evaluation of liquefaction potential Seed et al. (2003)...24 Fig Results from response analysis for 2,153 combinations of site conditions and ground motions, superimposed with heavier lines showing the earlier recommendations of Seed and Idriss...25 Fig Probabilistic SPT-based liquefaction triggering correlation...28 Fig Deterministic SPT-based liquefaction triggering correlation with adjustments for fines content shown...29 Fig Recommendations for magnitude-correlated duration weighting factor, with recommendations from Current Studies...31 Fig Magnitude-Correlated Duration weighting factor as a function of N 1, Fig K σ values for σ v>2atm...32 Fig Values of K σ as a function of SPT N-Values for effective vertical stresses of less than 3atm (After Harder and Boulanger, 1997)...33 Fig Probabilistic SPT-based Liquefaction triggering correlation (For Mw=7.5 and σ v=1.0atm)...34 Fig Deterministic SPT-based Liquefaction triggering correlation (For Mw=7.5 and σ v=1.0atm) with adjustments for fines content shown...35 Fig CPT-based liquefaction triggering correlation for clean sands and Fines Correction as proposed by Robertson and Wride (1998)...36 Fig Soil Behaviour Type Index, I c...37 Fig Recommended CPT Tip Normalization exponents, and approximate soil characterization framework (After Olsen & Mitchell, 1995)...38 V

6 Fig CPT Tip and Sleeve resistance normalization exponents, and previous recommendations of Olsen & Mitchell (1995)...39 Fig.3.16 Thin Layer correction for CPT Tip resistance and earlier NCEER working group recommendations...39 Fig.3.17 CPT Tip resistance modification for Fines Content and Character as function of qc,1 and Rf, compared with the earlier recommendations of Robertson and Wride (1997)...40 Fig.3.18 Comparison between the Recommended New CPT-based Correlation and previous relationships proposed by Suzuki et al. (1995) and Robertson and Wride (1998) [Mw=7.5, σ v=1atm]...41 Fig.3.19 Results SPT and CPT in two adjacent borings...42 Fig.3.20 Relationship between q c/n and soil type (from Burland and Burbidge, 1985, with permission Institution of Civil Engineers)...43 Fig.3.21 Conversion of SPT N-values to CPT q c-values using Median Grain Diameter...44 Fig.3.22 V s-based liquefaction triggering correlation (Andrus & Stokoe, 2000)...45 Fig.4.1 Variation in index void ratios and soil specific relative density for Yatesville sand specimens prepared to a constant gross void ratio of Fig.4.2 Variation in cyclic resistance and soil specific relative density for Monterey sand specimens prepared to a constant gross void ratio of Fig.4.3 Effect of fines content n liquefaction resistance of sand-fines mixtures...52 Fig.4.4 Number of cycles to initial liquefaction versus cyclic stress ratio for Yatesville sand and silt at 50% soil specific relative density...52 Fig.4.5 Comparison of variations in normalized cyclic resistance between data from the current being presented and published studies...53 Fig.4.6 Comparison of variations in normalized cyclic resistance between data from the current being presented and published studies...54 Fig.4.7 Variation in index void ratios and gross void ratio for Monterey sand specimens prepared to a constant sand skeleton void ratio of Fig.4.8 Variation in cyclic resistance for Monterey sand specimens prepared to a constant sand skeleton void ratio of Fig.4.9 Variation in cyclic resistance for Yatesville sand specimens prepared to constant sand skeleton void ratios...56 Fig.4.10 Variation in cyclic resistance with soil specific relative density for Yatesville sand specimens prepared to constant sand skeleton void ratios...57 Fig.4.11 Increase in normalized cyclic resistance with increasing silt content...58 Fig.4.12 Results of the study presented before...59 Fig.4.13 Phase diagram of microstructure and intergranular matrix for the conceptual framework (Thevanayagam 2000)...60 Fig.5.1 Recommendations Regarding Assessment of Liquefiable soil types...64 Fig.5.2 Cyclic resistance curves based on NCEER (1997) and correction with the term γ...67 Fig.5.3 Proposed correlations between cyclic resistance ratio (CRR) 7.5 and normalized cone resistance qc1 for sand with 0, 5 and 10% nonplastic silt...68 Fig.5.4 Cyclic Resistance (CRR) 7.5 versus skeleton void ratio (e s) for clean and silty sands (after consolidation)...69 Fig.5.5 Variation in cyclic resistance above and below the limiting silt content...70 Fig.5.6 Variation in cyclic resistance with silt content for Yatesville sand specimens adjusted to 25% soil specific relative density...71 Fig.5.7 Approximate Soil Characterization Framework (Olsen and Mitchell, 1995)...72 Fig.5.8 Values chosen for the parametric analysis of the correlations...73 Fig. Appendix 1 R d Results for Various Bins Superimposed with the Predictions (Mean and Mean ±1σ) Based on Bin Mean Values of V s,40 ft, M w and a máx...82 Fig. Appendix 2 - R d Results for Various Bins Superimposed with the Predictions (Mean and Mean ±1σ) Based on Bin Mean Values of V s,40 ft, M w and a máx...83 Fig. Appendix 3 - R d Results for Various Bins Superimposed with the Predictions (Mean and Mean ±1σ) Based on Bin Mean Values of V s,40 ft, M w and a máx...84 Fig. Appendix 4 Recommended CR Values (rod length from point of hammer impact to tip of sampler).86 Fig. Appendix 5 Grain size distributions for component soils...87 VI

7 LIST OF SYMBOLS η σ u e LL w e c ε q p S su ψ e SS G a máx g σ v r d CSR Stress Ratio Stress Pore Pressure Gross Void Ratio Liquid Limit Natural Water Content Critical Void Ratio Deformation Deviator Stress Mean Stress Steady-State Strength State Parameter Void Ratio at the Steady-State Line Shear Modulus Peak Ground Acceleration Gravitational Acceleration Vertical Stress Factor for the Calculation of the Horizontal Seismic Coefficient Cyclic Stress Ratio τ cyclic Cyclic Shear Stress N 1 V s N 60 V s,40ft M w d FC Q F q c f s D 50 e s PI Number of cycles Shear Wave Velocity SPT blowcount value (blows/feet) Average Shear Wave Velocity over the top 40 feet of a site Earthquake Magnitude Depth Fines Content Normalized CPT Penetration Resistance Normalized CPT Friction Ratio CPT Tip Resistance CPT Sleeve Resistance Medium Grain Diameter Sand Skeleton Void Ratio Plasticity Index VII

8 CHAPTER ONE: INTRODUCTION Derived from the Latin verb liquefacere, meaning to melt, to dissolve, or to weaken, liquefaction is the term commonly used to describe the sudden, dramatic strength loss which sometimes occurs in sands during seismic loading. While most frequently associated with cohesionless soils and dynamic loadings, it has been reported in many types of soils under both dynamic and static loadings. The term liquefaction has historically been used in conjunction with a variety of phenomena that involve soil deformations caused by monotonic, transient, or repeated disturbance of saturated cohesionless soils under undrained conditions. The liquefaction of sands during earthquakes has occurred throughout recorded history, and certainly before that, however it was not until the early 1960 s that scientific research into the subject began in earnest. Since the 1964 Anchorage (Alaska) and Nigata (Japan) earthquakes, great strides have been made in understanding the mechanisms behind liquefaction and the conditions that make soils susceptible to it. 1.1 Scope of this Document It has been understood since the 1960's that the presence of silt and clay particles will in some manner affect the resistance of a sand to liquefaction. However, a review of studies published in the late literature shows that no clear conclusions can be drawn as to in what manner altering the fines content affects the liquefaction resistance of a sand under cyclic loading. This is particularly true for soils containing non-plastic, i.e. silty, fines. Numerous laboratory studies have been performed, and have produced what appear to be conflicting results. Studies have reported that increasing the silt content in a sand will increase the liquefaction resistance of the sand, decrease the liquefaction resistance of the sand, or decrease the liquefaction resistance until some limiting silt content is reached, and then increase its resistance. Additionally, several studies have shown that the liquefaction resistance of a silty sand is more closely related to its sand skeleton void ratio than to its silt content. A clarification of the effects of non-plastic fines on the liquefaction susceptibility of sandy soils will be presented here in light of a recent study conducted by Martin and Polito (2001) and a recent conceptual framework proposed by Thevanayagam (2000). The assessment of triggering potential is also focused in this document. A review of the latest updates of in situ liquefaction hazard evaluation methods will be made. Since the influence of non-plastic fines on liquefaction susceptibility is the main subject of this work a detailed presentation of the correction factors, often used in simplified liquefaction evaluation methods to take into account the affect of fines content on the soil, will be made for each in situ test presented. 1

9 Finally, the effect of the behaviour of sand with nonplastic fines might have on the methods for liquefaction analysis currently used in engineering practice will be analyzed, and some suggestions will be made on how to evaluate liquefaction potential in these cases, based on the latest studies on the matter. 1.2 Outline of this Document Following this introductory section, Chapter Two of this document presents a literature review of what is known to date about the liquefaction phenomena (state-of-the-art), in general, concerning mostly mechanics and assessment of liquefaction potential. Next, Chapter Three is focused on the latest updates of in situ liquefaction hazard evaluation methods where a detailed presentation is made regarding fines correction factors for the in situ tests presented. Chapter Four presents the findings of a number of studies on the effects of nonplastic fines content on the liquefaction of sandy soils as well as a conceptual framework for the understanding of soil behaviour. Chapter 5 presents, based on recent studies, how to assess liquefaction triggering potential in situ for soils with nonplastic fines, using the SPT or the CPT tests. Finally, Chapter Six presents the conclusions of this document as well as suggestions where further study is required in this area. 2

10 CHAPTER TWO: LITERATURE REVIEW The term liquefaction has historically been used in conjunction with a variety of phenomena that involve soil deformations caused by monotonic, transient, or repeated disturbance of saturated cohesionless soils under undrained conditions. There is a known tendency for dry cohesionless soils to densify under loading. However, if the soil is saturated, rapid loading (such as earthquakes) occurs under undrained conditions and this tendency will cause excess pore pressure to increase and, consequently, effective stresses to decrease. Excess pore pressure results from the impeded drainage caused by plastic volumetric strain that arises quickly enough so that the pore fluid cannot escape as fast as the plastic strain accumulates. Liquefaction phenomena that result from this process can be divided into two main groups: Flow Liquefaction and Cyclic Mobility. Flow liquefaction occurs much less frequently than cyclic mobility but its effects are usually far more severe. Cyclic mobility, on the other hand, can occur under a much broader range of soil and site conditions than flow liquefaction. Flow liquefaction produces the most dramatic effects of all the liquefaction related phenomena tremendous instabilities know as flow failures. Flow liquefaction can occur when the static shear stress of a soil mass (shear stress required for static equilibrium) is greater than the static shear resistance of the soil in its liquefied state, which produces large permanent deformations. If the loading is monotonic, only the stress ratio (η ) matters in triggering flow liquefaction, whereas if the loading is cyclic, it is the cyclically induced excess pore pressure that may be sufficient to cause outright soil failure under the imposed loadings, producing flow failure. Flow failure is characterized by the sudden nature of its origin, the speed it develops and the large distances liquefied materials move. In contrast to flow liquefaction, cyclic mobility occurs when the static shear stress is less than the shear strength of the liquefied soil but the cyclic shear stress is large enough that the steady-state strength is exceeded momentarily. The deformations develop incrementally during earthquake shaking and are driven both by cyclic and static shear stresses (lateral spreading). The difference between static and cyclic induced liquefaction is the way plastic strains are generated. The case of cyclic induced liquefaction, the plastic volumetric strains arise through densification which tends to pack the soil particles close together. Cyclic induced densification affects any soil from loose to dense sands any even overconsolidated clays, it is only a question of to what extend. In contrast to static induced liquefaction, in cyclic mobility the zone of maximum excess pore pressure generation may not be the loosest soil, but rather the soil that was in the most stressed location. A special case of cyclic mobility is level-ground liquefaction that can occur when cyclic loading is enough to produce high excess pore pressures. Levelground liquefaction failures are caused by the upward flow of water that occurs when seismically induced excess pore pressure dissipate. Since there are no horizontal shear stresses that could drive lateral deformations, level-ground liquefaction can produce large movements known as ground oscillation (excessive vertical settlement and consequent flooding 3

11 of low-lying land). Level-ground liquefaction may occur well after ground shaking as ceased depending on the time length required to reach hydraulic equilibrium which makes it highly unpredictable. To explain the liquefaction phenomena Seed (1976) suggested a mechanism considering an element of soil A (Fig. 2.1) anisotropically consolidated under cyclic shear stress with horizontal predominance. Fig. 2.1 also represents the variation of void ratio with confining pressure. In point A, the initial stresses result in a decrease of volume in the specimen (just like it would happen when considering a drained situation). In the undrained situation, that effect is restricted by the water (incompressible) resulting in an increase in pore pressure and an equal decrease in the effective stresses resulting in the specimen not exhibiting any volume change (point A C). If the specimen is loose sand, true liquefaction would occur which causes endless deformations of the soil with very low residual strength. On the other hand, if the specimen is dense, the consecutive cycles of loading will cause the soil initially to contract but then quickly dilate, ultimately dissipating pore pressures and ceasing the deformations initiated with the contraction for the soil reaches equilibrium where it can resist to the loading cycles without deforming. This behaviour was then termed as initial liquefaction with limited deformation (also known as limited liquefaction). Seed does not, however, explicitly observe that this behaviour can only occur when volumetric strains surpass the shear strains in the beginning of the loading. Otherwise the soil would initially exhibit dilative behaviour which would not allow liquefaction to be triggered. Fig. 2.1 Void Ratio variation with confining stress 4

12 To better distinguish flow liquefaction from cyclic mobility a constitutive analysis is required based on triaxial testing with a specimen of saturated sand. The steady-state line represents the steady-state of deformation of the soil, which is the state where soil flows continuously under constant shear stress and constant effective confining pressure at constant volume and constant velocity (Fig. 2.2). Fig. 2.2 Undrained tests in saturated sand Flow liquefaction is the result of a rupture in an undrained specimen of loose sand (contractive) beginning in point C and ending at point A (no volume change undrained condition). The loading can be monotonic or cyclic, both will cause pore pressure to increase. In this graphic is also represented point Q which refers to a quick sand state, where the soil has lost all its shear resistance, with no tendency to dilate or contract since the grains of sand are no longer in contact. Considering a specimen of saturated dense sand under monotonic loading starting in point D, the trajectory will be from left to right until it reaches the steady-state line. On the other hand, if the loading is cyclic the trajectory will be from right to left since there is no volume variation and there is an increase pore pressure resulting in a decrease of effective stress. This increase in pore pressure results from plastic volumetric strains through densification brought on by the cyclic stresses which tend to pack the soil particles closer together. Depending on the intensity of the cyclic loading, among other factors, the pore pressure increase can be of such magnitude that point B might be reached, effectively nullifying effective stresses. Large deformations may occur resulting in what is known as cyclic mobility. However, one important point must be made about cyclic mobility, there is no sustained zero effective stress although a transient zero effective stress can occur, in other words, there is no requirement of zero effective stresses for liquefaction to occur. 5

13 Lab testing evidence also show void ratio redistribution in the sample, increasing on the top and decreasing at the base, resulting in the horizontal line between points D and B. This non-uniform strain leads to thinning of the upper portion of the specimen causing considerable uncertainty in the application of cyclic triaxial test results to field conditions. 2.1 Liquefaction Susceptibility The first step in a liquefaction hazard evaluation is the quantification of liquefaction susceptibility of the site. If the soil is not susceptible to liquefaction, liquefaction hazards do not exist. However, if the soil is susceptible, there are several criteria by which liquefaction susceptibility can be analyzed and there are different evaluation approaches for flow liquefaction and cyclic mobility. These include historical, geologic, compositional and state criteria. Geologic, compositional and state criteria must be met for a soil to be susceptible to liquefaction, otherwise the soil is non-susceptible Historical Criteria Liquefaction often occurs at the same location where soil and groundwater conditions have remained unchanged after an earthquake. Thus it is possible to map liquefaction susceptibility given a number of instances where there is historical evidence of liquefaction. Post-earthquake field investigations have shown that liquefaction effects usually occur within a particular distance of the seismic source. The distance which liquefaction can be expected increases almost exponentially with increasing magnitude. Fig. 2.3 Relationship between limiting epicentral distance of sites at which liquefaction has been observed and moment magnitude for shallow earthquakes 6

14 2.1.2 Geologic Criteria Soils with uniform grain size distributions and deposited in loose states are highly susceptible to liquefaction. Consequently, fluvial deposits and aeolian deposits, when saturated, are likely to be susceptible to liquefaction. The susceptibility of older soil deposits to liquefaction is generally lower than that of newer deposits. Human-made deposits such as loose fills placed without compaction, are very likely to be susceptible to liquefaction. The stability of hydraulic fill dams and mine tailings piles where soil particles are loosely deposited by settling through water, are also susceptible to liquefaction. Liquefaction occurs only in saturated soils, so the depth of groundwater greatly influences liquefaction susceptibility since it decreases for increasing groundwater depth. The effects of liquefaction are most commonly observed at sites where groundwater is within a few meters of the ground surface Compositional Criteria Compositional characteristics (size, shape and gradation), associated with high volume change potential, tend to be associated with high liquefaction susceptibility. For many years it was thought that liquefaction-related phenomena was limited to sands since finer-grained soils were considered incapable of generating the high pore pressures with which liquefaction is normally associated and coarser-grained soils were considered too permeable. However, liquefaction of nonplastic silts has been observed indicating that plasticity characteristics rather than grain size alone influence liquefaction susceptibility in fine-grained soils. Clays, however, remain nonsusceptible to liquefaction. This topic will be better analysed further ahead. The Modified Chinese Criteria (Wang (1979), Seed and Idriss (1982)), represent the most widely used criteria for defining potentially liquefiable soils over the last two decades. These criteria consider that fine (cohesive) soils that plot above the A-line are considered to be of the potentially liquefiable type and charactering if: Fraction finer than mm 15% Liquid Limit, LL 35% Natural water content 0.9 LL Liquidity index 0.75 Fig. 2.4 Modified Chinese Criteria (After Wang (1979) and Seed and Idriss (1982)) 7

15 At the other end of the grain size spectrum, liquefaction of gravels has already been observed in the field. When pore pressure dissipation is impeded by the presence of impermeable layers so that truly undrained conditions exist, gravely soils can also be susceptible to liquefaction. The shape of particles also influences liquefaction susceptibility since soils with rounded particle shapes are known to densify more easily than soils with angular grains. Particle rounding frequently occurs in the fluvial and alluvial environments where loosely deposited saturated soils are frequently found. As for soil gradation, field evidence indicates that well-graded soils are generally less susceptible to liquefaction than poorly graded soils since most liquefaction failures involved uniformly graded soils State Criteria Even considering that the soil meets all the preceding criteria for liquefaction susceptibility it does not necessarily mean that it is susceptible to liquefaction, it greatly depends on the initial state of the soil since it is strongly influenced by both relative density and initial stress conditions. These liquefaction criteria are different for flow liquefaction and cyclic mobility unlike those presented earlier Critical Void Ratio Casagrande (1936) in his pioneering work on the shear strength of soils observed that initially loose specimens contracted (densified) during shearing and initially dense specimens first contracted but then very quickly began to dilate. At large strains all specimens approached the same density and continued to shear with constant shearing resistance. The void ratio corresponding to this constant density was termed the critical void ratio, e c. Fig. 2.5 Stress-strain and stress-void ratio curves for loose and dense sands at the same effective confining pressure 8

16 At the time, since the CVR line marked the boundary between contractive and dilative behaviour, it was considered to mark the boundary between soils susceptible to liquefaction and non-liquefiable soils. Fig. 2.6 Use of the CVR line as a boundary between loose contractive states and dense dilative states Steady State of Deformation Castro (1969) performed static and cyclic triaxial tests on isotropically consolidated specimens and several static tests on anisotropically consolidated specimens. Three different types of stress-strain behaviour were observed for anisotropically consolidated specimens. Very loose specimens (Specimen A) exhibited a peak undrained strength at a small shear strain and then collapsed to flow rapidly to large strains at low effective confining pressure and low largestrain strength Flow liquefaction. Dense specimens (Specimen B) initially began to contract but then dilated until a high constant effective confining pressure and large strength was reached Non-susceptible to flow liquefaction. Fig. 2.7 Liquefaction, Limited Liquefaction, and Dilation in monotonic leading tests 9

17 At intermediate densities (Specimen C) the exceedance of peak strength at low strain was followed by a limited period of strain-softening behaviour which ended with the start of dilation at intermediate strains Limited liquefaction. The point where the soil turns from contractive to dilative behaviour was termed the phase transformation point (Ishihara 1975). This testing showed a unique relationship between void ratio and effective confining pressure at large strains that, graphically, are plotted below and are roughly parallel to the CVR line obtained from drained-controlled tests. The state in which the soil flowed continuously under constant shear stress and constant effective confining pressure at constant volume and constant velocity was defined as the steady state of deformation. Since the steady state of deformation is reached only at large strains (initial conditions of the soil such as fabric, stress and strain history, and loading conditions have been obscured) the effective confining pressure of the soil in the steady state of deformation was considered to depend only on the density of the soil. The locus of points describing the relationship between void ratio and effective confining pressure in the steady state of deformation is called the steady-state line SSL. The SSL can also be expressed in terms of the steady-state strength S su. Fig. 2.8 State criteria for flow liquefaction susceptibility Under given loading conditions any sand will reach a unique combination of effective confining pressures, shear strength and density at large strains described graphically by a SSL. The position of the SSL is most strongly influenced by grain size and grain shape characteristics. The behaviour of sand is strongly related to its position relative to the SSL allowing the SSL to be useful for identifying the conditions under which a particular soil may or may not be susceptible to flow liquefaction: - Soils whose state plots below the SSL are not susceptible to flow liquefaction - A soil whose state lies above the SSL will be susceptible to flow liquefaction only if the static shear stress exceeds its steady state strength (residual strength) Cyclic mobility, on the other hand, can occur in soils whose state plot above or below the SSL, which means it can occur in both loose and dense soils, and when the static shear stress is less than the shear strength of the liquefied soil as long as the cyclic shear stress is large enough that the steady-state strength is exceeded momentarily. 10

18 Finally, a comment on the steady-state line and the critical state line used in the critical state soil mechanics (Modified Cam Clay model, for instance) for normally consolidated sand and clay. There has been a lot of discussion between specialists whether these two lines were the same. The main disagreement was that the SSL is obtained for soft sands (contractive) in a triaxial undrained controlled loading while the critical state line is generally obtained when testing dense sands (dilative) in drained loading with controlled deformation. According to Been, Jefferies and Hachey (1991) after examining the results of an extensive triaxial testing program, drained and undrained, they came to the conclusion that, for practical purposes, the two lines are equivalent and independent of the strain trajectory State Parameter The steady-state line approach illustrates the limited applicability of absolute measures of density, such as void ratios and relative density, for characterization of a potentially liquefiable soil since an element of soil with a particular void ratio can be susceptible to soil liquefaction or not, depending on the confining pressure. Using concepts of critical-state soil mechanics, the behaviour of a cohesionless soil should be more closely related to the proximity of its initial state to the steady-state line than to absolute measures of density. Using this logic a state parameter was defined ψ = e e ss (Eq. 1) where e ss is the void ratio of the steady-state line at the effective confining pressure of interest. When state parameter is positive, the soil exhibits contractive behaviour, therefore, the undrained strength will be less tan the drained strength, and the soil may be susceptible to flow liquefaction. When it is negative, dilative behaviour will occur, therefore, the undrained strength will be greater than the drained strength, and the soil is not susceptible to flow liquefaction. The reason why ψ is used rather than the void ratio or relative density directly is because high confining stress levels tend to suppress dilatancy and the definition of state must take into account the stress level. It is the magnitude of dilation that determines strength, not the void ratio or density at which dilation occurs. The concept of state parameter is very useful and the possibility of determining its value from in situ tests is appealing. The accuracy with which the state parameter can be determined depends on the accuracy by which the position of the SSL can be determined. An approach based on the state parameter was developed by Been and Jefferies (2007) but since the influence of fines cannot be observed, for the state parameter is independent of fines content of the soil according to Been and Jefferies (2007), this approach will not be discussed here. 11

19 Fig. 2.9 State parameter Although ψ is a controlling influence on the behaviour of sand, it does not provide a complete description of sand behaviour. This can be observed in the test results from 2 samples, with the same void ratio and subjected to equal confining pressures, where neither the deviator stress curves nor the volumetric strain curves are similar. Arthur and Menzies (1972) showed the importance of initial fabric to sand behaviour, with differences of over 200% in axial strain to reach a given stress ratio for a sand at the same void ratio and stress level. For cyclic loading, the conclusion that ψ is the most important variable of the behaviour of sand, and that fabric is a second order effect, is not necessarily appropriate. For example, data from Nemat- Nasser and Tobita (1982) show that for a state parameter change of 0.07, the shear stress for liquefaction in 10 cycles reduces 32%. 2.2 Initiation of Liquefaction Even if a soil deposit is susceptible to liquefaction it does not mean, however, that liquefaction will be triggered if an earthquake or any other disturbances occur. Cyclic mobility is an earthquake related phenomena, but flow liquefaction can be triggered in a variety of ways (pile driving, geophysical exploration, blasting and vibrations from a passing train, for example). For liquefaction to be triggered it requires a strong enough disturbance and it is the evaluation of the nature of that disturbance that is one of the most critical parts of liquefaction hazard evaluation. Understanding the initiation of liquefaction requires identification of the state of the soil when liquefaction is triggered. The conventional view is that an undrained strength (or its equivalent) is appropriate and can be used in the stability analysis. The soil behaviour, however, continues to be the result of effective stresses and previously established properties from drained tests apply. The use of drained properties is crucial because, in a field situation, there will be little drainage in the short term, but complete drainage in the long term. To base stability analysis in undrained tests calibrations/models alone is potentially misleading. 12

20 The initiation of flow liquefaction is most easily seen with a sample of soil being subjected to monotonic increasing stresses. Consider an isotropically consolidated specimen of very loose saturated sand in undrained stress-controlled triaxial compression (initial state point A in Fig. 2.10). The initial state of the specimen is well above the SSL so the sand will exhibit contractive behaviour (positive state parameter). When undrained shearing begins, the specimen will develop shearing resistance up to a peak (point B), at relatively small strains, at which point it becomes unstable. As shearing continues pore pressure increases dramatically resulting in a decrease in effective stresses as well as decreasing shear resistance. The result is the collapse of the specimen (flow liquefaction) since the static shear stresses required for equilibrium were greater than the available shear strength of the liquefied soil (point C). Fig Response of isotropically consolidated specimen of loose, saturated sand: (a) stress-strain curve; (b) effective stress path; (c) excess pore pressure; (d) effective confining pressure 13

21 2.2.1 Flow Liquefaction Surface Now consider the response of a series of triaxial tests with the same initial void ratio but at different effective confining pressures. Observing Fig. 2.11, points C, D and E are clearly above the SSL so they will exhibit contractive behaviour upon shearing, and very similar stress paths to that observed in the previous example. As for specimens A and B they are located below the SSL and will exhibit dilative behaviour upon shearing. Fig Response of five specimens isotropically consolidated to the same initial void ratio at different initial effective confining pressures 14

22 For specimens C, D and E liquefaction is triggered at the peak of each stress path. The limited liquefaction exhibited by specimen C, is significant for cases in which the static shear stress increases (monotonic loading). Graphically, the line that unites the peak of the various stress paths is termed the flow liquefaction surface, FLS, which represents the stress conditions at the initiation of liquefaction. The FLS marks the boundary between stable and unstable states in undrained shear. If the stress conditions in an element of soil reach the FLS under undrained conditions by monotonic loading, flow liquefaction will be triggered and shearing resistance reduced to the steady-state of deformation. As for cyclic loading, whether liquefaction is initiated precisely at the FLS, as monotonic loading, is not currently known for certain but, as an idealization, it can be considered that it is. Since flow liquefaction cannot occur if the stress path is below the Steady- State point, the FLS is truncated at that level. Fig Orientation of the flow liquefaction surface in stress path space If the initial shear stresses are high, the FLS may be close to the initial stress point in which case liquefaction only requires a small disturbance to be triggered, and therefore explain some case histories that have been attributed to spontaneous liquefaction. Such soils may represent a high liquefaction hazard. The liquefaction resistance will be greater the farther way from the FLS the initial stress conditions are. It has also been observed that samples with high initial shear stress, compared to soils with lower initial shear stress, at the same void ratio, the FLS is steeper. Fig Variation of flow liquefaction surface inclination with initial principal effective stress ratio for constant void ratio 15

23 Flow liquefaction develops in two stages. The first stage, which takes place at low levels of strain, involves the generation of sufficient pore pressure to move the stress path from its initial position to the FLS after which the soil becomes unstable. The second stage involves strain-softening, as well as excess pore pressure generation, driven by the shear stresses required for static equilibrium which are known as driving stresses. Large strains develop in this second stage as the stress path moves from the FLS to the SSL. Once the soil reaches the FLS under undrained, stress-controlled conditions, the second stage is inevitable Influence of Excess Pore Pressure The generation of excess pore pressure is the key to the initiation of liquefaction, both cyclic mobility and flow liquefaction, although the different phenomena can require different levels of pore pressure to occur. As stated before, flow liquefaction can be initiated by cyclic loading only when the shear stress required for static equilibrium is greater than the steady-state strength. Therefore, initial states that plot in the shaded area, if liquefaction is triggered (moving the effective stress path from its initial state to the FLS), it is flow liquefaction that will most likely occur. Fig Zone of susceptibility to flow liquefaction Cyclic mobility on the other hand, can develop when the static shear stress is smaller than the steady-state shear strength and both in loose and dense soils, from very low to very high confining pressures (including states that plot above and below the SSL). Fig Zone of susceptibility to cyclic mobility 16

24 To evaluate the initiation of liquefaction, a number of approaches have been developed being the most renowned ones the cyclic stress approach and the cyclic strain approach. The main difference between the two is that while one is based on the evaluation of cyclic shear stresses, the other is based on cyclic shear strains, required to produce liquefaction. Comparing the two approaches, the cyclic shear strain approach appears to be more directly related to pore pressure generation since it is the tendency to density under loading when saturated that generates pore pressures, which are only indirectly related to cyclic shear stresses (through the shear modulus, G). However, cyclic strains are considerably more difficult to predict accurately than cyclic stresses due to the difficulty in estimating the shear modulus since it varies with different shear strains. The cyclic stress approach is conceptually very simple: by comparing the earthquakeinduced loading, expressed in terms of cyclic shear stresses, with the liquefaction resistance of the soil, also expressed in terms of cyclic shear stresses, the locations where the loading exceeds the resistance is where liquefaction is likely to occur. The main body of the approach is to correctly characterize both the earthquake loading and the liquefaction resistance of soils. 2.3 Characterization of Earthquake Loading Earthquake loading generates high pore pressures in saturated sands, because the loading is too fast to allow the soil to dissipate the water. The question is how much excess pore pressure is enough to trigger liquefaction in a soil susceptible to liquefaction. In the cyclic stress approach, it is considered that pore pressure generation is due mostly to cyclic shear stresses. The level of excess pore pressure required to initiate liquefaction is related to both amplitude and duration of earthquake cyclic loading. In order to characterize earthquake loading, it is necessary to perform laboratory testing on specimens replicating earthquakes and then, resorting to ground response analyses, produce time histories with transient, irregular characteristics of actual earthquake motions. However, the laboratory data from which liquefaction resistance can be estimated are usually obtained from tests in which the cyclic shear stresses have uniform amplitudes, which doesn t correspond to real earthquake motion. To compare the two a conversion of an irregular time history of shear stress to an equivalent series of uniform stress cycles must be made. The most usual form of converting is the one proposed by Seed et al. (1975) a τ τ 65 g σ máx cyclic = 0.65 máx = 0. vrd (Eq. 2) 17

25 Fig Typical irregular time history of shear stress 2.4 Characterization of Liquefaction Resistance The liquefaction resistance of a soil depends on how close the initial state of the soil is to the state corresponding to failure and on the nature of the loading required to move it from the initial state to the failure state. As has been said before, the failure state for flow liquefaction is the FLS but when it comes to cyclic mobility, its initiation isn t so easily described. Cyclic mobility is generally considered to occur when pore pressures increase enough to produce phenomena like ground oscillation and/or lateral spreading. For the characterization of liquefaction resistance there are two different approaches, those based on laboratory tests and methods based on in situ tests. In laboratory testing, it is very important to be able to define the point where liquefaction failure occurs. It is usually considered to be when initial liquefaction or some limiting strain amplitude is reached. The relationship between density, cyclic stress amplitude and number of cycles to liquefaction failure can be expressed graphically by cyclic stress curves (Fig. 2.17). When normalized by the initial overburden pressure (vertical effective pressure for a cyclic simple shear test or initial confining pressure for a cyclic triaxial test) it produces a cyclic stress τ cyclic ' v0 ratio ( CSR = ). σ Fig Cyclic stresses required to produce initial liquefaction and 20% axial strain in isotropically consolidated Sacramento River Sand triaxial specimens. 18

26 Until recently, liquefaction resistance was only evaluated resorting to laboratory testing. However, recent work on the topic showed that other factors besides initial density and stress conditions influence liquefaction resistance of a soil. These factors include soil fabric, history of prior seismic straining (a specimen that has been subjected to prior seismic straining has greater liquefaction resistance that another specimen, with the same density, that hasn t) and length of time under sustained pressure (liquefaction resistance increases with the specified length of time). These soil characteristics are destroyed in the process of sampling making it impossible to test important liquefaction resistance factors in specimens with laboratory testing. Because of these factors, the characterization of liquefaction resistance is now mainly based on in situ test results. For the characterization of liquefaction resistance based on in situ testing, a number of tests are used where the most common are the SPT (Standard Penetration Test), CPT (Cone Penetration Test) and shear wave velocity measurements in the soil deposit. For this approach, the use of liquefaction case histories is important to characterize liquefaction resistance in terms of measured in situ parameters. It is usual to consider the in situ parameters that describe density and pore pressure generation characteristics as liquefaction resistance parameters and cyclic stress ratio as the loading. Comparing the two most common tests, SPT and CPT, the SPT is the most commonly used worldwide for liquefaction resistance characterization and has the largest case history database of any in situ test. However, the CPT is becoming a more common test for liquefaction resistance characterization since it is able to detect thin layers since it provides continuous record of penetration resistance (unlike the SPT) of potentially liquefiable soils that may exist. There is a known correlation between SPT and CPT resistances, by supplementing these data it is possible (within certain limits) to expand the database for both tests, especially for the CPT. The relationship between cyclic stress ratio (CSR) required to trigger liquefaction and the liquefaction resistance based on situ tests can be described graphically (most results are based on historical criteria), where earthquake magnitude and fine content play a very important part. 19

27 2.5 Evaluation of Initiation of Liquefaction After characterizing both the cyclic loading imposed by an earthquake and the liquefaction resistance of the soil, the evaluation of liquefaction potential can be assessed. By comparing the different cyclic loadings and corresponding soil resistance, one can expect liquefaction at depths where the loading exceeds the resistance, both in terms of cyclic shear stress or strain. A safety factor is usually calculated to describe the liquefaction potential: cyclic shear stress/strain required to trigger liquefaction FS = (Eq. 3) equivalent cyclic shear/strain stress induced by earthquake motion Fig Process by which zone of liquefaction is identified in the cyclic stress approach Fig Process by which the zone of liquefaction is identified in the cyclic strain approach 20

28 In liquefaction evaluation and study, the undrained monotonic behaviour of sands has received much attention, especially loose sands, since dense (dilatant) sands exhibit much greater shear strength undrained than drained. However, far more soils than a limited range of sands can exhibit static liquefaction. The range of experience includes rather coarse uniform sands through to nearly pure silt sized soils and combinations between these gradational limits. From a practical point of view, the undrained behaviour of loose sands isn t of much interest since soils that fail in undrained monotonic shear pose too great a risk of catastrophic failure that soil improvement is mandatory in engineering practice. The true engineering problem is to identify when a soil is sufficiently dense that treatment is not required. Engineers must assume that there is always the possibility of some strain in any structure, (movements in underlying weaker clays); the aim is to ensure that the soil is sufficiently dense that there is very little strength drop as a result of increase pore pressures. Since the strength drop is caused by pore pressures generated by volumetric compression, if the sand is dense enough that shear dilation is far greater than volumetric compression, there will be no strength drop. This occurs at about ψ in triaxial compression and at about ψ 0.05 in simple shear tests which correspond to the critical density defined by Lindenberg and Koning (1981) at which cumulative volumetric strain at peak strength is zero. As a minimum criterion of density, all evidence points to ψ since sands that are denser do not show a phase transformation, they dilate continuously under monotonic loading. Of course the actual value will depend on a number of factors (loading conditions, sand properties, etc.) but it s a good starting point. 21

29 CHAPTER THREE: ASSESSMENT OF TRIGGERING POTENTIAL There are two general types of approaches available for the assessment of liquefaction triggering potential which are laboratory testing of undisturbed samples and the use of empirical relationships based on correlation of observed field behaviour. The use of laboratory testing is complicated by difficulties associated with sample disturbance during both sampling and reconsolidation. It is also difficult and expensive to perform high-quality cyclic shear testing and also the fact that cyclic triaxial testing poorly represents the loading conditions of principal interest for most seismic problems. Accordingly, the use of in situ index testing is the dominant approach in common engineering practice. The three in situ test methods that have now reached a level of sufficient maturity as to represent viable tools for this purpose are the Standard Penetration Test (SPT), the Cone Penetration Test (CPT) and the measurement of in-situ shear wave velocity (V s ). Most available correlations for liquefaction potential assessment are based on field observations and SPT or CPT data for liquefied and nonliquefied sites. Liquefaction occurrence may not be easily detected on the ground surface and penetration resistance data is usually obtained from sites that have undergone considerable shaking. Estimating soil state from penetration tests is the backbone of liquefaction assessments. Since it is extremely difficult to obtain samples of cohesionless soils in anything like an undisturbed condition, engineering of sands and silts is highly dependent on penetration tests. Penetration testing almost always meant the SPT but during the last twenty years the CPT has been progressively replacing the SPT. Regardless of the penetration method used, it is the soil response to an enforced displacement that is measured but it is the soil properties that are sought. It is the inversion of penetration data that provides the engineering design parameters. This chapter will focus in the latest updates of in situ liquefaction hazard evaluation methods. Since the influence of non-plastic fines on liquefaction susceptibility is the main subject of this work a detailed presentation of the corrections factors, often used in simplified liquefaction evaluation methods to take into account the affect of fines content on the soil will, be made for each in situ test presented. 22

30 3.1 Standard Penetration Test (SPT) The use of SPT as a tool for evaluation of liquefaction potential first began to evolve after a pair of devastating earthquakes in 1964, the Alaska earthquake and the Niigata earthquake where both produce significant liquefaction-related damage Previous SPT-based Correlations There are numerous SPT-based correlations but one of the most widely accepted and used is the deterministic relationship proposed by Seed et al. (1984, 1985) showed in Fig.3.1. Fig Correlation between equivalent uniform cyclic stress ratio and SPT N 1,60 -Value for events of Magnitude M w =7,5 for varying slit contents, with adjustments at low cyclic stress ratio as recommended by NCEER Working Group (Modified from Seed et al., 1984) This familiar relationship is based on comparison between SPT N-values, corrected for both effective overburden pressure and energy (reference condition approach), equipment and procedural factors affecting SPT testing vs intensity of cyclic loading, expressed as magnitudeweighted equivalent uniform cyclic stress ratio. This relationship is also a function of fines content since it was observed early in the evaluation of field case histories that soil type mattered. 23

31 Although it is the most used in modern engineering practice, this relationship is dated and does not make use of any field case histories since Also, it is particularly lacking in data from cases of high peak ground acceleration levels (CSR > 0.25). The two most common critics put forward about this correlation is the weak theoretical basis of it, since it is basically founded on historical data, and the fact that this correlation has no formal probabilistic basis, giving no insight regarding either uncertainty or probability of liquefaction. A number of researchers have developed similar correlations, but formally probabilistic-based, in recent years to better assess triggering potential New SPT-based Correlations The next correlations were presented by Seed et al. (2003) in a recent state-of-the-art paper regarding the assessment of triggering potential. Fig.3.2 shows this correlation, with contours of probability of liquefaction plotted for P L =5, 20, 50, 80 and 95%. This correlation provides greatly reduced overall uncertainty being the principal uncertainty in the engineer s ability to assess a suitable CSR and representative N 1, 60 values for design cases. The development of this correlation is founded in an expanded database of field performance case histories, improved knowledge of factors affecting the SPT, improved methods for accessing site-specific ground motions, screening of field data case histories for quality data only and the use of powerful probabilistic tools. The resulting relationships provided greatly reduced uncertainty as well as help to deal with corollary issues that have long been difficult and controversial such as magnitudecorrelated duration weighting factors, adjustments for fines content and corrections for effective overburden stress. Fig Probabilistic Correlation for evaluation of liquefaction potential Seed et al. (2003) 24

32 A major improvement is the better estimation of in situ CSR within the critical stratum for each of the case field histories. Previous studies used the simplified method of Seed and Idriss (1971). CSR a σ máx v peak = rd (Eq. 4) g σ ' v Fig.3.3. The original rd values proposed by Seed and Idriss (1971) are the heavy lines shown in Fig Results from response analysis for 2,153 combinations of site conditions and ground motions, superimposed with heavier lines showing the earlier recommendations of Seed and Idriss The numerous light grey lines show the results of 2,153 seismic site response analyses performed to assess the variation of r d over ranges of site conditions and ground motions excitation characteristics. The recognition that r d is nonlinearly dependent upon a suite of factors led to studies by Cetin and Seed (2000) to develop improved correlations for estimation of r d. Cetin and Seed (2000, 2003) propose a new, empirical basis for estimation of r d as a function of depth, earthquake magnitude, intensity of shaking and site stiffness. The values of r d from the 2,153 site response analysis performed as part of the study sub-divided into 12 bins as a function of peak ground acceleration (a máx ), site stiffness (V s,40ft ), earthquake magnitude (M w ) and depth (d) can be seen in the Appendix of this document [V s, 40ft is the average shear wave velocity over the top 40 feet of a site]. 25

33 (Eq. 5) However, in situ CSR (and r d ) can transition irregularly within a specific soil profile, especially near sharp transitions between soft and stiff strata making the best means for CSR in situ estimation to directly calculate it by means of appropriate seismic response analyses. The new correlation is a significant improvement since all prior correlations had been based on the use of the simplified r d of Seed and Idriss (1971) for back analysis of field performance case histories and were, as a result of that, unconservatively biased relative to actual case-specific seismic response analysis. The new correlations, on the other hand, can be safely used in conjunction with project-specific dynamic response analyses without introducing bias. In the new correlations proposed, the in situ stress ratio (CSR) is taken as the equivalent uniform CSR equal to 65% of the peak CSR. CSR = (Eq. 6) eq CSR peak Now considering the SPT N 1,60 values. The values employed were truncated mean values within the critical stratum. Measured N-values were corrected for overburden, energy, equipment, and procedural effects to N 1,60 values. 26

34 N-values were corrected for overburden pressure according to the reference condition approach. This approach is most commonly used in penetration data from the SPT and CPT. It basically consists in mapping penetration data to a reference stress level commonly to be taken as σ = 100kPa( 1tsf or 1atmosphere). Penetration resistance mapped back to this ref reference level are usually subscripted by a 1 and, for the SPT, N 60 becomes N 1,60 while for CPT, q c becomes q c,1. The common practice mapping of the SPT, the one that is considered in the EC8, is usually quoted in the form: N1 = C N (Eq. 7) N with C N being the mapping function. Nowadays, after a comparative study from Liao and Whitman (1986) suggested a reasonable average relationship was the simple, now widely used, equation: C N σ ' = σ v ref n (Eq. 8) The value of n is considered to be 0.5 for the SPT. The resulting N 1 values were then further corrected for energy, equipment, and procedural effects to fully standardized the value as N 1,60 = N1. C R. CS. CB. C E (Eq. 9) C R = correction for short rod length C S = correction for non-standardized sampler configuration C B = correction for borehole diameter C E = correction for hammer energy efficiency These corrections correspond to those recommended by the NCEER Working Group (NCEER, 1997; Youd et al., 2001) presented in the Appendix. Fig.3.4 shows the proposed probabilistic relationship between duration-corrected equivalent uniform cyclic stress ratio (CSR eq ), and fines corrected penetration resistances (N 1,60,CS ) with the correlations as all field data shown normalized to an effective overburden pressure of 0.65atm. This relationship must be normalized to σ v = 0.65 atm since it would be rendered unconservative if normalized to σ v = 1 atm. It should be noted, however, that this unconservatism is minimized if the correlations are applied at shallow depths. 27

35 Fig Probabilistic SPT-based liquefaction triggering correlation By simply observing Fig.3.4 one might assume that the clean sand line of Seed et al. (1984) corresponds to P 50%. This is not the case since the clean sand line was based on L biased values of CSR as a result of biased r d at shallow depths (as was discussed earlier). At the time, it was Seed s intent that the recommend (1984) boundary line should represent about 10 to 15% probability of liquefaction although it does correspond to approximately P 10 to 30% except at very high CSR were there was little data at the time. L Also shown in Fig.3.4 is the boundary curve proposed by Yoshimi et al. (1994) which is arguably unconservative biased at very low densities (low N-values) as these loose samples densified during laboratory thawing and reconsolidation. Improvement of the new correlation at high CSR values is due to the availability of significant new data (at high CSR) from recent earthquakes that had not been available in

36 Adjustments for Fines Content As a basis for illustration of the correlation s regressed correction for the effects of fines content the boundary curve for P = 15% will be used as shown in Fig.3.5. L Fig Deterministic SPT-based liquefaction triggering correlation with adjustments for fines content shown In this figure, both the correlations as well as the mean values (CSR and N1,60) of the filed case history data are shown not corrected for fines. The N-values in this correlation, N 1,60, are corrected for fines content as N 1 (Eq. 10),60, CS = N1, 60 C FINES The fines correction is equal to 1.0 for fines contents of FC 5%, and reaches a maximum (limiting) value for FC 35%. The maximum fines correction results in an increase of N-values of about +6blows/ft. (at FC 35% and high CSR) which is somewhat smaller than the earlier maximum correction of 9.5 blows/ft proposed by Seed et al. (1984). 29

37 The regressed relationship for C FINES is C FINES = ( FC) FC N Lim: FC 5% and FC 35% (Eq. 11) 1,60 where FC = percent fines content (percent by dry weight finer than 0.074mm), expressed as an integer (e.g. 15% fines is expressed as 15), and N 1,60 is in units of blows/ft. However, this correction does not make any distinction between plastic and non-plastic fines which is erroneous and potentially dangerous Magnitude-Correlated Duration Weighting The correction of equivalent uniform stress ratio is made to represent the equivalent CSR for a typical duration of an average event of M w =7,5. This duration weighting factor has been somewhat controversial. It is usually done by means of a magnitude-corrected duration weighting factor (DWF M ) as CSR CSReq, M = M, (Eq. 12) DWF eq M = 7,5 = M where the values for DWF M are presented in Fig

38 Fig Recommendations for magnitude-correlated duration weighting factor, with recommendations from Current Studies Fig Magnitude-Correlated Duration weighting factor as a function of N 1,60 The values for DWF M presented in the recent state-of-the-art paper by Seed et al. (2003) propose a different correlation for obtaining the DWF M based on new data and more powerful probabilistic tools. Also, the correlation is tested against varying densities, which correspond to different N 1,60 -values, proving that the dependence on density, is relatively minor. 31

39 Additional Adjustments for Effective Overburden Stress The additional effects of reduction of normalized liquefaction resistance with increased effective initial overburden stress have been demonstrated by means of laboratory testing. It is a manifestation of critical state types of behaviour since soils become less dilatant at increased effective stress. These effects were not considered in earlier adjustments since the case history database did not have data for higher effective overburden stress. To apply this correction, NCEER Working Group recommends to correct the normalized resistance to liquefaction at an initial effective overburden stress of 1atm (CSR liq,1atm ) as CSR liq = CSRliq 1atm K σ, (Eq. 13) where K σ can be estimated as f 1 ( σ ' ) Kσ = (Eq. 14) v and f 0.6 to 0.8 as N 1,60,CS varies from 1 to 40 blows/ft. Fig K σ values for σ v >2atm For sites with sloping ground conditions the presence of non-zero shear stresses affects liquefaction resistance. Routinely expresses as the ratio of static, driving shear stress acting on a horizontal plane divided by the effective vertical stress acting on that plane as (α) τ hv α = (Eq. 15) σ ' v 32

40 Increasing levels of static driving shear stresses can have an increasing effect on the vulnerability of the soil to cyclic generation of pore pressures. In these cases, a factor is usually applied to scale the equivalent uniform cyclic stress ratio required to trigger liquefaction as CSR liq, α > 0 = CSRliq, α = 0 Kα (Eq. 16) where values for Kα can be estimated from Fig.3.9. Fig Values of K σ as a function of SPT N-Values for effective vertical stresses of less than 3atm (After Harder and Boulanger, 1997) 33

41 Use of the Proposed SPT-based Correlations These proposed SPT-based correlations can be used directly by using (Eq. 17 or in parts using the method described in this section. (Eq. 17) Fig Probabilistic SPT-based Liquefaction triggering correlation (For Mw=7.5 and σ v =1.0atm) 34

42 For deterministic evaluation of liquefaction resistance, largely compatible with the intent of the earlier relationship proposed by Seed et al. (1984), the same steps can be undertaken (except for the fines adjustment). The recommendations of Fig.3.11 correspond approximately to P 15% except for CSR >0.4 since there is no conclusive field data. L Fig Deterministic SPT-based Liquefaction triggering correlation (For Mw=7.5 and σ v =1.0atm) with adjustments for fines content shown 35

43 3.2 Cone Penetration Test (CPT) The CPT is another in situ test commonly used to assess liquefaction triggering potential. According to some authors, SPT-based correlations have been better defined, and have provided lesser levels of uncertainty than other methods. However, most agree that CPT is near parity with the SPT, and newly developed CPT-based correlations are significantly better and more accurate. CPT-based correlations have been based on much less numerous and not so well defined earthquake field case histories than SPT-based correlations which makes it important to complement and augment CPT-based correlations. A common question is which test is best to assess liquefaction potential. The best answer is that the two tests used in conjunction provide much more and better data Previous CPT-based Correlations The most commonly used and known CPT-based correlation is the Robertson and Wride (1998) correlation which is also recommended by the NCEER Working Group. Fig.3.12 shows the boundary line triggering curve of Robertson and Wride for clean sandy soils. The adjustment for fines content is based on combinations of sleeve friction ratios and tip resistances so that the boundary curve is adjusted based on a composite parameter I C. I C is a measure of the distance (radius) from a point above and to the left of the plot of normalized tip resistance (q c,1 ) and normalized friction ratio (F) as indicated in Fig Fig CPT-based liquefaction triggering correlation for clean sands and Fines Correction as proposed by Robertson and Wride (1998) 36

44 The correction for fines content and plasticity according to Robertson and Wride correlations is by means of a parameter K C as q c, 1,mod = qc, 1 K c (Eq. 18) where K C is a nonlinear function of I C, and ranges from K C = 1.0 at I C = 1.64 to a maximum of K C =3.5 at I C = 2.6. It is further recommended that the fines correction K C be taken as 1.0 in the shaded zone within Area A. Fig Soil Behaviour Type Index, I c By comparing Robertson and Wride fines correlations to the SPT-based correlation by Seed et al. (2003), it is slightly unconservative for clean sands at high CSR and it is very unconservative for soils of increasing fines content and plasticity. However, Robertson and Wride had access to a relatively small field case history database, and so their correlation represents a value interim contribution as development of new correlations taking advantage of the wealth of new earthquake field case history data now available now proceed New CPT-based Correlations The next correlations were presented by Seed et al. (2003) in a recent state-of-the-art paper regarding the assessment of triggering potential. This approach follows the same guidelines as the one proposed for SPT so it shares many of the same strengths. The key feature of these following correlations is the significant database expansion in recent years. 37

45 Normalization of CPT Tip and Sleeve for Effective Overburden Stress Approaches have differed significantly here. Olsen and Mitchell (1995) presented the most comprehensive set of recommendations in this regard which are presented in Fig.3.14 f s where Friction Ratio is taken as R f = 100 [%]. q c Fig Recommended CPT Tip Normalization exponents, and approximate soil characterization framework (After Olsen & Mitchell, 1995) CPT tip resistance and sleeve resistance can be normalized to q c,1 and f s,1 as q = C q c, 1 q c C q P a = σ ' v c f = C s, 1 f f s C f P a = σ ' v s (Eq. 19) The recommended normalization exponents (s) and (c) are shown is Fig.3.15 as a function of normalized tip resistance and friction ratio. 38

46 Fig CPT Tip and Sleeve resistance normalization exponents, and previous recommendations of Olsen & Mitchell (1995) Thin Layer Correction This is another source of potential uncertainty to measured CPT tip resistances for finite stiff layers. The effects of initial penetration into a stronger (e.g., less cohesive, potentially liquefiable) layer prior to achieving sufficient penetration into the layer to develop a fully developed tip resistance can result in a reduced tip resistance reading, with a similar reduction occurring as the cone approaches and exits the bottom of a stronger layer. The correction of CPT tip resistances for this effect can be estimated as q cb, corrected = qcb, thin = Cthin qcb (Eq. 20) where C thin is as shown in Fig It is however recommended that values of C thin greater than 1.8 not be used for engineering applications because of the intrinsic uncertainty of this adjustment. Fig.3.16 Thin Layer correction for CPT Tip resistance and earlier NCEER working group recommendations 39

47 Adjustment for Fines Content CPT tip resistances are normalized for the frictional effects of apparent fines content and character. Values of q c,1 are adjusted as q q q c, 1,mod = c, 1 + c (Eq. 21) where q c is a function of q c,1, R f, and I C, as shown in Fig.3.17 also with the fines correction factors recommended by Robertson and Wride (1997). These contours also provide for much smaller adjustments of q c,1 for fines content and character than did the curves from Robertson and Wride that were already said to be unconservative. Fig.3.17 CPT Tip resistance modification for Fines Content and Character as function of qc,1 and Rf, compared with the earlier recommendations of Robertson and Wride (1997) Finally, Fig.3.18 shows the comparison between this recent CPT-based correlation and the previous from Robertson and Wride (1997) and Suzuki et al. (1995). For the clean sand all three correlation are very similar but when it comes to the fines-corrected lines, both Suzuki et al. and this recent correlation by Seed et al. shows much smaller adjustments for fines than the one from Robertson and Wride although Suzuki et al. line does not extend to high friction sleeve ratios (Rf > 1.0). 40

48 Fig.3.18 Comparison between the Recommended New CPT-based Correlation and previous relationships proposed by Suzuki et al. (1995) and Robertson and Wride (1998) [Mw=7.5, σ v =1atm] 3.3 SPT vs CPT The SPT is the most commonly used in situ test. It has been used for more than 75 years and has the largest database of penetration data worldwide. Its obvious advantages are the very simple procedure, almost all soil types can be tested and rugged equipment. What is usually neglected by engineers is the lack of repeatability of the test even when considering the same equipment and adjacent borings. Even after improvements to the SPT by mechanizing the hammer system to control the energy delivered, the repeatability of the test can only be considered of 60% which is considerably low for engineering purposes. To illustrate the poor repeatability of the SPT even with energy correction, observe the results of two borings, only 2m apart, in sand fill and alluvial sands in Canada. The N 60 values vary more than ± 25% making it impossible to even attempt any to base any engineering on a test that is so inaccurate. 41

49 Fig.3.19 Results SPT and CPT in two adjacent borings The SPT can only be performed at vertical spacings of about 75cm or more which can result in the test completely missing a thin, but potentially important, liquefiable strata between test depths. The CPT has been around nearly as long as the SPT. The main difference is that the CPT evolved considerably since its original form with the growth of the offshore industry making the CPT the reference test for geotechnical engineering offshore. There were numerous improvements but the most important in engineering purposes is the interesting combination of a continuous data record with excellent repeatability and accuracy, all at relatively low cost. The standard deviation is about 2% of full-scale output. The fact that CPT is fully continuous means it misses almost nothing. Even for strata too thin to characterize, the CPT provides some indicators of potentially problematic materials if one examines q c and f s traces carefully. CPT also offers advantages with regard to cost and efficiency since no borehole is required. The two main reasons why CPT has not been used extensively for liquefaction assessment are that the test does not provide a sample for soil classification and grain size analyses and the limited amount of CPT-based field data pertaining to liquefaction potential was available. However, the number of field case histories with CPT data has increased significantly. In round numbers, the SPT, even with energy measurements and subsequent correction, is four to five times less repeatable than the CPT. An argument sometimes put forward in favour of the SPT is the possibility of obtaining a geological sample of the stratum which the CPT is unable. However, for engineering purposes, it s not that important to have geological information of the stratum if its mechanical properties are known, which is what the 42

50 CPT is able to do and with much more accuracy than the SPT although, some times, a good compositional knowledge of the soil is required and so a geological sample is needed. To sum up, there s no contest; choose either the CPT or both tests. Note, however, that this does not mean that the SPT database is useless. As was said before, it is possible to use SPT-based experience when only CPT soundings have been carried out. The general form of the relationship is simply q c = αn 60 (Eq. 22) Fig.3.20 Relationship between q c /N and soil type (from Burland and Burbidge, 1985, with permission Institution of Civil Engineers) 43

51 Other relationships have been developed to relate CPT to SPT results. Fig.3.21 presents additional SPT-CPT conversions including data presented by Seed and De Alba (1986), Robertson and Campanella (1985), and additional data from field investigations conducted by Youd and Bennett (1983), Bennett (1989), Bennett (1990), and Kayen et al. (1992) for a median grain diameter D 50 (mm). Fig.3.21 Conversion of SPT N-values to CPT q c -Values using Median Grain Diameter 3.4 V S -Based Triggering Correlations This method of assessing liquefaction susceptibility is still under development but it shows great potential since shear wave velocity can be measured with non-intrusive methods and can be used in all types of soils. SPT and CPT have difficulties in coarse soils since the tests can be obstructed by coarse soil particles. It is also a very useful tool for a rapid screening a site. The V S -based correlations, however, still fail to properly correlate with liquefaction resistance as does penetration resistance since V S is a low small-strain measurement unlike liquefaction and penetration resistance. Also, the relationship between V S and the CSR required for liquefaction varies significantly with the geologic age of deposits in question. Finally, the overburden correction for V S is somewhat controversial and uncertain, and also the observations that shear wave velocity of sand is insensitive to factors as soil fabric, overconsolidation ratio, prior cyclic straining, which are known to influence liquefaction resistance, suggests that shear wave velocity measurements alone might not be sufficient to evaluate the liquefaction potential of all soil deposits (Jamiolkowsky and LoPresti, 1992; Verdugo, 1992). V S correlations for resistance to triggering of liquefaction are best employed conservatively and to be supplemented with other methods or just as a preliminary screening tool. 44

52 At this time the best V S -based correlation is the one of Andrus and Stokoe (2000) presented in Fig.3.22 where the V S is corrected for overburden pressure according to the reference condition approach. Measured shear wave velocities can be normalized to a standard effective overburden pressure of 1ton/ft 2 (96kPa) by 1 n 1 ( ' v0 ) v s = v s σ (Eq. 23) where σ v0 is in tons/ft 2 and n has been taken as 3 (Tokimatsu et al., 1991) or 4 (Finn, 1991; Kayen et al., 1992). Fig.3.22 V s -based liquefaction triggering correlation (Andrus & Stokoe, 2000) 45

53 CHAPTER FOUR: SOILS WITH NON-PLASTIC FINES Since the 1960 s it is known that the presence of silt and clay particles will in some manner affect the resistance of a sand to liquefaction. However, when reviewing the studies published in the literature they show that no clear conclusions can be drawn as to in what manner altering the fines content affects the liquefaction resistance of a sand under cyclic loading. This is particularly true for soils containing non-plastic fines (silts). Plastic fines, however, depending on the level of plasticity, are generally considered to increase the cyclic resistance of sand. After a review of the literature it is possible to observe this trend of increasing cyclic resistance with increasing plasticity and fines content, when considering fines with some plasticity, is verified by numerous authors. The plasticity index of the fines fraction has been recognized as an important factor in the liquefaction susceptibility of silty sand (Ishihara and Koseki 1989; Ishihara 1993). These authors found that sands containing fines with PI<10 do not exhibit a significant increase if any of liquefaction resistance with respect to that of clean sands. According to Ishihara (1985), tailing silty sands can be as liquefiable as clean sands due to the origin and nonplastic nature of the fines. There are numerous laboratory studies that have produced conflicting results. Some report that increasing the silt content of a sand will increase liquefaction resistance of the sand, others that it decreases liquefaction resistance of the sand and even some report that it decreases liquefaction resistance until some limiting silt content is reached and then increase its resistance. Additionally, several studies have shown that the liquefaction resistance of a silty sand is more closely related to its sand skeleton void ratio than to its silt content. The presence of plastic or clayey fines is generally considered to decrease the liquefaction susceptibility of a soil. Numerous studies have shown that soils with more than 10 or 15 percent fines do not liquefy during earthquakes. During the 1970 s, Engineers in the Peoples Republic of Chine (PRC) developed a set of criteria in their building codes, commonly referred to as the Chinese Criteria, which deem certain soils as non-liquefiable due to their plastic nature. Also, the manner in which pore pressure generation varies as the quantity and type of fine-grained material in a sand increases is an important aspect of liquefaction susceptibility assessment. Finally, it is important to examine in what manner conclusions from the previous analyses will affect the way in which working engineers perform simplified liquefaction analyses. 46

54 4.1 Literature Review -The Effects of Fines Content on Liquefaction Resistance Both clean sands and sands containing fines have been shown to liquefy in the field (Mogami and Kubo (1953); Robertson and Campenella (1985); and Holzer et al. (1989)) and in the laboratory (Lee and Seed (1967a); Chang et al. (1982); and Koester (1994)). Also, nonplastic silts, most notably mine tailings, have also been found to be susceptible to liquefaction (Dobry and Alvarez (1967); Okusa et al. (1980); and Garga and McKay (1984)). A review o the literature, however, shows conflicting evidence as to the effect which non-plastic fines have on the liquefaction resistance or cyclic strength of a sand The Effects of Non-Plastic Fines Content There is no clear consensus in the literature as to the effect which increasing nonplastic fines content has upon the liquefaction resistance of a sand. Both field and laboratory studies have been performed and some of the results of these studies are conflicting. Field studies following major earthquakes have produced conflicting evidence as to the effects of silt on the liquefaction resistance of sands. Based upon case histories of actual soil behaviour during earthquakes, there is evidence that soils with greater fines contents are less likely to liquefy in a seismic event. Okashi (1970) observed that during the 1964 Nigata earthquake in Japan, sands were more likely to liquefy if they had fines content of less than 10 percent. Fei (1991) reports that for the 1976 Tangshan earthquake in China the liquefaction resistance of silty soils increased with increasing fines content. Also, Tokimatsu and Yoshimi (1983) found in a study of 17 worldwide earthquakes that 50 percent of the liquefied soil had fines contents of less than 5 percent. They also found that sands with fines content greater that 10 percent had a greater liquefaction resistance than clean sand at the same blowcount. Other filed reports show the opposite results. Tronsco and Verdugo (1985) report that mine tailings dams constructed of soils with higher silt contents are more likely to liquefy than similar dams constructed of sands with lower silt contents. Chang, Yeh, and Kaufman (1982) note that case studies reveal that most liquefaction resulting from earthquakes has occurred in silty sands and sandy silts. When considering laboratory testing there continues to be conflicting evidence about the effect of non-plastic fines in the cyclic resistance of sands. Several investigators have found that the cyclic resistance of a sandy soil increases with increasing silt content. For specimens prepared to a constant gross void ratio, Chang et al. (1982) found that after a small initial drop, cyclic resistance increased dramatically with increasing silt content. The cyclic resistance increased nearly linearly with silt content until a silt content of 60 percent was reached, increasing to a cyclic resistance between 50 and 60 percent greater than that of the clean sand. Similarly, Dezfulian (1982) reported a trend of increasing cyclic resistance with increasing silt content. 47

55 Numerous authors have reported a decrease in cyclic resistance with increasing silt content. Shen et al. (1977), Tronsco and Verdugo (1985), and Vaid (1994) have all reported this trend for specimens prepared either to a constant gross ratio or a constant dry density. The decreases in cyclic resistance were marked, decreasing as much as 60 percent from their clean sand values for an increase in silt content of 30 percent (Tronsco and Verdugo, 1985). Rather than a simple decrease in cyclic resistance with increasing fines contents, several investigators have reported that the cyclic resistance of the sand first decreased as the fines content increased and then increased after crossing some threshold fines content. Koester (1994) and Law and Ling (1992) found that for specimens prepared to a constant gross void ratio, as silt content increased the cyclic resistance of the soil decreased until some limiting silt content was reached, at which point the cyclic resistance began increasing. Koester (1994) reported a decrease in cyclic resistance to less than one-quarter of the clean sand cyclic resistance at a silt content of 20 percent, followed by an increase in cyclic resistance to 32 percent of the clean sand value at a silt content of 60 percent. Unlike Chang et al. (1982), and Dezfulian (1982), neither of these other studies reported increases in cyclic resistance to levels greater than those determined for the clean sand. Several studies have shown that cyclic resistance is more closely related to sand skeleton void ratio than it is to gross void ratio, gross relative density or fines content. Finn, Ledbetter, and Wu (1994) found that at the same gross void ratio, the cyclic strength of a sand decreases with increasing fines content. They also found that at the same sand skeleton void ratio, cyclic strength remains constant with increasing fine content, as long as the fines can be accommodated in the void spaces created by the sand skeleton. Clearly, based upon the conflicting evidence presented in the literature, the fines content of a sandy soil does not alone provide a definitive measure of its liquefaction potential. 48

56 4.2 The Effects of Non-Plastic Fines Cyclic Resistance During the past 40 years, development in the understanding of liquefaction of clean sands under seismic loads has been intensely studied and a sound understanding of its mechanics and the parameters which affect it has been developed. Unfortunately, the understanding of the liquefaction of sands containing fine-grained materials is less complete. A review of the literature shows that there is no clear consensus as to what effect an increase in non-plastic silt content has upon the liquefaction resistance of a sand. There are numerous laboratory studies that have produced conflicting results. Some report that increasing the silt content of a sand will increase liquefaction resistance of the sand, others that it decreases liquefaction resistance of the sand and even some report that it decreases liquefaction resistance until some limiting silt content is reached and then increase its resistance. Additionally, several studies have shown that the liquefaction resistance of a silty sand is more closely related to its sand skeleton void ratio than to its silt content. A recent study was conducted by Martin and Polito (2001) were nearly 300 cyclic triaxial tests of sands with non-plastic silt contents ranging from 0 to 100 percent were performed, in an attempt to clarify the discrepancies found in the literature and to find a single parameter which controls the liquefaction resistance of silty sands and sandy silts. These tests were performed and evaluated using gross void ratio (gross relative density) and sand skeleton void ratio as measures of density. The gross void ratio of a specimen is simply the ratio of the volume of the voids in the specimen to the volume of the solids, and is a function of the dry density of the specimen and specific gravity of the soil, whereas the sand skeleton void ratio, is the void ratio that would exist in the soil if all of the silt and clay particles were removed, leaving only the sand grains to form the soil skeleton. The results of that study were used to explain the behaviours found in the literature. First, the trend of decreasing and then increasing cyclic resistance with constant sand skeleton void ratio will be examined. Next, the findings of constant or increasing cyclic resistance with constant sand skeleton void ratio will be examined. Lastly, the trend of increasing cyclic resistance with increasing silt content will be discussed. 49

57 4.2.1 Changes in Soil Specific Relative Density with Increasing Silt Content The maximum and minimum index densities of a sand and silt mixture, which are an indication of the range of densities it can achieve, vary with silt content. When the silt content of a sand is increased, the maximum and minimum void ratios initially decrease as the soil becomes more well-graded. When the soil reaches the limiting silt content, the maximum and minimum void ratios reach their lowest values. As the silt content of the soil continues to increase, the maximum and minimum void ratios increase as the grain size distributions of the soil becomes more uniform. The soil specific relative density is the relative density of the specimen based upon its gross void ratio and the maximum and minimum index void ratios for that particular mixture of sand and silt. This behaviour is shown in Fig.4.1 for the Yatesville sand for specimens prepared to a gross preconsolidation void ratio of Fig.4.1 Variation in index void ratios and soil specific relative density for Yatesville sand specimens prepared to a constant gross void ratio of

58 Because a specimen s ability to resist liquefaction is directly related to its soil specific relative density, a decrease in soil specific relative density produces a decrease in cyclic resistance and an increase in soil specific relative density leads to an increase in cyclic resistance. This pattern is shown for the Monterey base sand in Fig.4.2. Fig.4.2 Variation in cyclic resistance and soil specific relative density for Monterey sand specimens prepared to a constant gross void ratio of 0.68 It should be noted that the lowest cyclic resistance does not correspond precisely with the limiting silt content or the lowest soil specific relative density. This occurs because, as the silt content of the soil is increased, the soil transforms from a sand-controlled matrix to a siltcontrolled matrix. However, in the range of silt contents immediately above the limiting silt content, the sand grains are still closely situated and exert an influence upon one another. 51

59 Another example of this behaviour of sand with increasing silt content, for a constant value of global void ratio, was observed in an experimental study by Xenaki and Athanasopoulos (2003) which the results can be seen in Fig.4.3. Fig.4.3 Effect of fines content in liquefaction resistance of sand-fines mixtures The existence of a critical value of fines content is clearly shown. As stated before, this critical value separates the range of fines content values where the liquefaction resistance decreases with an increase in fines content from the range in which the behaviour is reversed. There is, however, an inherent difference in cyclic resistance between sand and silt (Fig.4.4). This difference results in a decrease in cyclic resistance as silt content increases, although the soil specific relative density is actually increasing. Fig.4.4 Number of cycles to initial liquefaction versus cyclic stress ratio for Yatesville sand and silt at 50% soil specific relative density 52

60 4.2.2 Decreasing Cyclic Resistance with Increasing Silt Content Numerous authors have reported a decrease in cyclic resistance with increasing silt content. Shen et al. (1977), Tronsco and Verdugo (1985), and Vaid (1994) have all reported this trend for specimens prepared either to a constant gross void ratio or a constant dry density. This decrease in cyclic resistance was observed to be as much as 60 percent compared to their clean sand values for an increase of 30 percent (Tronsco and Verdugo, 1985). The decrease of cyclic resistance with increasing silt content can be explained in terms of the soil specific relative density. The specimens used in the study presented before were all prepared to constant gross void ratios or densities, and were at silt contents that were likely below the limiting silt content. Because the specimens were prepared to constant gross void ratios, their soil specific relative densities decreased with increasing silt contents, leading to a decrease in cyclic resistance. Since all the specimens were only tested at silt contents below their limiting silt contents, the increase in cyclic resistance which would have occurred at the higher silt contents was not detected. Fig.4.5 presents the normalized cyclic resistances for both the Monterey and Yatesville sands prepared to constant gross voids ratio at silt contents below their limiting silt contents. These curves all show similar trends, confirming that this behaviour can be explained in terms of the soil specific relative density. Fig.4.5 Comparison of variations in normalized cyclic resistance between data from the current being presented and published studies 53

61 4.2.3 Decreasing and then Increasing Cyclic Resistance with Increasing Silt Content Rather than a simple decrease in cyclic resistance with increasing fines content, several investigators have reported a decrease and then an increase in cyclic resistance with increasing silt content. Koester (1994) reported a decrease in cyclic resistance at a silt content of 20 percent, then an increase in cyclic resistance to 32 percent of the clean sand value at a silt content of 60 percent. Unlike Chang et al. (1982), and Dezfulian (1982), neither Koester (1994) nor Law or Ling (1992) reported increases in cyclic resistance to levels greater than those of the clean sand. The decrease and then increase of cyclic resistance with increasing silt content can also be explained in terms of the soil specific relative density. The specimens tested in the study were all prepared to constant gross void ratios, creating soil specific relative densities that vary with varying silt content. As the silt content of the specimens increases, their soil specific density first decreases and then increases as the maximum and minimum index void ratios vary. This variation in soil specific relative density creates the reported variations in cyclic resistance that can be seen in Fig.4.6 for both the Monterey and Yatesville sand specimens as well as published studies. Fig.4.6 Comparison of variations in normalized cyclic resistance between data from the current being presented and published studies 54

62 4.2.4 Cyclic Resistance with Constant Sand Skeleton Void Ratio Several studies have shown that cyclic resistance is more closely related to sand skeleton void ratio than it is to gross void ratio, gross relative density, or fines content. Tronsco and Verdugo (1985) found that at a constant sand skeleton void ratio, the cyclic resistance of a sand is constant with increasing silt content. Shen et al. (1977), Kuerbis et al. (1988), and Vaid (1994) however, have shown that at a constant sand skeleton void ratio, the cyclic resistance of a sand does not remain constant but increases with increasing silt content. Sand skeleton void ratio is the void ratio that would exist in the soil if all of the silt and clay particles were removed, leaving only the sand grains to form the soil skeleton. Finn et al. (1994) stated that the relationship between cyclic resistance and sand skeleton void ratio is only applicable to soils which have sufficient room in the voids created by the sand skeleton to accommodate all of the silt present, in other words, soils below their limiting silt contents. When above the limiting silt content, like it was shown earlier, the cyclic resistance is controlled by the silt fraction and is independent of sand skeleton. Shown in Fig.4.7 is the variation for the Monterey sand, below the limiting silt content and at a constant sand skeleton void ratio of 0.75, of the maximum and minimum index void ratios as silt is added, not because the sand skeleton is altered, but because the voids in the sand skeleton are being filled resulting in a decrease of void ratios. Fig.4.7 Variation in index void ratios and gross void ratio for Monterey sand specimens prepared to a constant sand skeleton void ratio of

63 Since all three different void ratios decrease at approximately the same rate over this range of silt contents, the soil specific relative density remains nearly constant for a given sand skeleton void ratio, which in turn produces nearly constant cyclic resistance (Fig.4.8). Fig.4.8 Variation in cyclic resistance for Monterey sand specimens prepared to a constant sand skeleton void ratio of 0.75 This shows that for soils with constant sand skeleton void ratios produce constant soil specific relative densities, the cyclic resistance remains constant which matches the behaviour reported by Tronsco and Verdugo (1985). Not all soils however, exhibit a constant cyclic resistance for a constant sand skeleton void ratio. Shen et al. (1977), Kuerbis et al. (1988), and Vaid (1994) have shown that for specimens prepared to a constant sand skeleton void ratio, the cyclic resistance of a sand does not remain constant but increases with increasing silt content (Fig.4.9). Fig.4.9 Variation in cyclic resistance for Yatesville sand specimens prepared to constant sand skeleton void ratios 56

64 The increase in cyclic resistance measured for these soils at a constant sand skeleton void ratio occurs because as their silt content increases, their maximum and minimum index void ratios decrease less rapidly than their gross void ratios. As a result, although the sand skeleton void ratio remains constant, the soil specific relative density increases with increasing silt content. This increase in soil specific relative density causes in turn an increase in cyclic resistance. These specimens of Yatesville sand, despite having different sand skeleton void ratios and silt contents exhibit the same variation in cyclic resistance when examined in terms of their soil specific relative density. Despite the different results of the Monterey sand and the Yatesville sand in the relation between sand skeleton void ratio, with increasing silt content, and cyclic resistance, whether the cyclic resistance remains constant or increases when the sand skeleton void ratio is held constant, depends on the rate at which the maximum and minimum index void ratios change relative to the gross void ratio as silt content increases (Fig.4.10). Fig.4.10 Variation in cyclic resistance with soil specific relative density for Yatesville sand specimens prepared to constant sand skeleton void ratios 57

65 4.2.5 Increasing Cyclic Resistance with Increasing Silt Content Several investigators have found that cyclic resistance increases with increasing silt content. For specimens prepared to a constant gross void ratio, Chang et al. (1982) and Dezfulian (1982), both found that after a small initial drop, cyclic resistance increased dramatically with increasing silt content. However, the silt used in these studies was slightly plastic with a plasticity index of 5 and the results can be seen in Fig Fig.4.11 Increase in normalized cyclic resistance with increasing silt content The normalized cyclic resistance increased nearly linearly with silt content until a silt content of 60 percent is reached, increasing to a cyclic resistance between 50 and 60 percent greater than that of the clean sand. In this recent study no evidence was found to either support or explain these large increases in cyclic resistance with increasing silt content. One clue to the reason for this behaviour lies in the fact that it was only reported for studies that used silts with some plasticity. However, there was a recent study by Carraro, Bandini and Salgado (2003) where undrained triaxial tests were performed on Ottawa sand with 0, 5, 10 and 15 percent non plastic fines over a wide range of relative densities for each gradation. It was observed that between 0 to 5 percent fines content, the cyclic resistance of sand increases slightly for relative densities ranging from 40 to 70 percent and then decreasing with increasing silt content. 58

66 4.2.6 Interpretation of Results After the presentation of the results there is incoherence with the behaviour of sand with a percent of silt above and below the limiting silt content. Below the limiting silt content, when compared at the same gross void ratio, e, liquefaction resistance decreases with an increase in fines content up to the threshold value (limiting silt content). However, when compared at the same sand skeleton void ratio, e s ( increases the liquefaction resistance. e s e + = 1 fc fc, fc = FC 100 ), an increase in fines content Fig.4.12 Results of the study presented before Above the limiting silt content, when compared at the same gross void ratio, liquefaction resistance increases with an increase in fines content but, when at the same sand skeleton void ratio, an increase in fines content decreases the liquefaction resistance. These behaviours can be seen schematically in Fig.4.12 where the specimens A, B, C, D and E correspond to an increase in silt content for a constant value of global void ratio and specimens F, G, H and I correspond to an increase in silt content for a constant value of intergranular void ratio. The threshold value of fines content for the soil tested is approximately equal to 44 percent. The interpretation of the behaviour depicted in the figure can be based on the conceptual framework by Thevanayagam (2000). 59

67 Fig.4.13 Phase diagram of microstructure and intergranular matrix for the conceptual framework (Thevanayagam 2000) In Fig.4.13 four limiting cases of microstructure and the relevant roles of coarse and fine grains of the above conceptual framework are described. In case (i) the finer grain content (FC) it is less the threshold value (FC th ) and the finer grains are fully confined within the void spaces between the coarser grains. The mechanical behaviour in this case is primarily by the coarser grain contacts. In cases (ii) and (iii) the fines content is still less than the threshold value whereas the value of the gross void ratio has increased. The intercoarser grain contacts in these cases still play a significant role. However, some of the finer grains (termed separating fines) become active participants in the internal force chain, while the remaining fines (termed confined fines) fill the voids between the coarser grains. Finally, in case (iv) the value of fines content exceeds the threshold value and the finer grains begin to play a rather important role, while the role of coarser grains begins to diminish. 60

68 4.2.7 Conclusions Several conclusions regarding the effects of non-plastic fines on the liquefaction susceptibility of sandy soils can be drawn from this recent study in conjunction with the conceptual framework by Thevanayagam (2000). Three distinct behavioural patterns were found for the cyclic resistance of soils composed of sand and non-plastic silt. The type of behaviour is determined by whether there is sufficient room in the voids created by the sand skeleton to contain the silt present without disturbing the sand structure. This silt content has been called the limiting content and occurs between 30 and 40 percent for the sands in the study. If the silt content of the soil is below the limiting silt content, there is sufficient room in the voids created by the sand skeleton to contain the silt, and the soil can be described as having silt contained in a sand matrix. The cyclic resistance of the soil is then controlled by the soil specific relative density of the specimen, where the soil specific relative density is calculated using the gross void ratio of the specimen and the maximum and minimum index void ratios for that particular mixture of sand and silt. The cyclic resistance for these soils is independent of silt content and increases with increasing soil specific relative density. The limiting silt content is usually between 15 to 40% fines content for a sand. If the silt content is greater than the limiting silt content, the specimen s structure consists predominately of sand grains suspended within a silt matrix with little sand grain to sand grain contact. Above the limiting silt content, the amount of sand present in the soil and its soil specific relative density have little effect on its cyclic resistance. The cyclic resistance for these soils is controlled by the silt fraction void ratio of the soil. Decreasing the silt fraction void ratio increases the soil s cyclic resistance. There is a transition zone consisting of soils with silt contents at or slightly above the limiting silt content. This zone occurs as a result of the structure of the soil changing from a predominantly sand controlled fabric to a predominately silt controlled fabric as silt content increases. Soils in this zone possess cyclic resistances intermediate to those with greater or lesser silt contents. The trend of increasing cyclic resistance with increasing silt content which has been reported in the literature does not appear to occur in non-plastic silts and is likely due to the plasticity of the fines used in those studies. 61

69 CHAPTER FIVE: ASSESSMENT OF TRIGGERING POTENTIAL OF SOILS WITH NON-PLASTIC FINES The first step in engineering assessment of the potential for triggering or initiation of soil liquefaction is the determination of whether or not soils are susceptible to liquefaction. The Modified Chinese Criteria (Wang (1979), and Seed and Idriss (1982)) represent the most widely used criteria for defining potentially liquefiable soils over the last two decades. However, in recent earthquakes there was significant liquefaction-type damage where the soils responsible appeared to be more cohesive than would be expected based on the Chinese Criteria. It is recommended that the Modified Chinese Criteria be relegated to history and that we move forward to broader consideration of potentially liquefiable soil types. This approach will be described here based on Andrews and Martin (2000). The affect the behaviour of sand with nonplastic fines might have on the methods for liquefaction analysis currently used in engineering practice will now be analyzed. Considering the results from the study here presented, the cyclic resistance of sands with non-plastic fines can be divided into two main categories depending whether the silt content is above or below the limiting silt content. The cyclic resistance of soils below the limiting silt content is controlled by the soil specific relative density, while the cyclic resistance of soils above the limiting silt content is controlled by the silt fraction void ratio of the soil. A small parametric analysis will be conducted in order to evaluate the current and proposed liquefaction assessment correlations for the SPT and CPT tests considering the approximate soil characterization framework based on normalized cone penetration resistance and friction ratio by Olsen and Mitchell (1995). 62

70 5.1 Liquefiable Soil Types Andrews and Martin (2000) re-evaluated the liquefaction field case histories from the database of Wang (1979), as well as a number of subsequent earthquakes, and have transposed the Modified Chinese Criteria to U.S. conventions (with clay sizes defined as those less than about 0.002mm). Their findings are largely summarized in Table 1. Liquid Limit 1 < 32 Liquid Limit 32 Further Studies Required Clay Content 2 < 10% Susceptible Further Studies Required (Considering plastic non-clay sized grains such as Mica) Clay Content 10% (Considering non-plastic clay sized grains such as mine and quarry tailings)) Notes: 1. Liquid Limit determined by Casagrande-type percussion apparatus 2. Clay defined as grains finer than 0.002mm Not Susceptible Table 1 - Liquefaction susceptibility of silty and clayey sands (Andrews and Martin, 2000) In the new field performance cases it is often difficult to reliably discern whether or not soils with cohesive fines liquefied. Soils with large fines contents do not generally exude excess pore pressure rapidly, and so are less prone to produce soil boil ejecta than are cleaner cohesionless soils. As a result, soils with significant fines have been sampled and then subjected to cyclic testing in the laboratory by a number of researchers as was exposed in Chapter Four. Sandy soils, and silty soils of very low plasticity, tend to experience triggering of cyclically induced soil liquefaction at relatively low shear strains (typically on the order of 3 to 6 percent), and the loss of strength can be severe. 63

71 Fig.5.1 Recommendations Regarding Assessment of Liquefiable soil types Fig.5.1 represents interim recommendations regarding liquefiability of soils with significant fines content. This may evolve further, based on work in progress but it is a good summary of what we know to date. For soils with sufficient fines content that the fines separate the coarser particles and control overall behaviour: Soils within Zone A are considered potentially susceptible to classic cyclically induced liquefaction Soils within Zone B may be liquefiable Soils within Zone C (not within Zones A or B) are not generally susceptible to classic cyclic liquefaction, but should be checked for potential sensitivity By observing the characteristics of each zone it is easily observed that Zone B can motivate doubt about the evaluation of soils that are within that zone since they fall into a transition range. These soils are also, in many cases, not well suited to evaluation based on conventional in situ penetration-based liquefaction hazard assessment methods. These types of soils usually are amenable to reasonably undisturbed sampling, however, and so can be tested in the laboratory. It should be remembered to check for sensitivity of these cohesive soils as well as for potential cyclic liquefiability. Appropriate sampling and testing protocols of soils of Zone B are not yet well established and further research is needed here. As was described in Chapter Four, for the soils with sufficient fines to separate the coarser particles, it is the fines that control the potential for cyclically induced liquefaction. In those cases, cyclically-induced liquefaction appears to occur primarily in soils where these fines are either nonplastic or are low plasticity silts and/or silty clays ( PI 12% and LL 37% ) and with high water content relative to their Liquid Limit ( w c > 0. 85LL ). In fact, low plasticity or 64

72 non-plastic silts and silty sands can be among the most dangerous of liquefiable soils, as they not only can cyclic liquefy but they also hold their water well and dissipate excess pore pressures slowly due to their low permeabilities. Finally, two additional conditions necessary for potential liquefiability are saturation (or at least near-saturation) and rapid (largely undrained) loading. As can be seen from the criteria here presented, sand with nonplastic fines are susceptible to classically induced liquefaction until the liquid limit of 37 percent, and may be liquefiable if its liquid limit is between 37 and 47 percent. 5.2 In Situ Testing The cyclic resistance of the soil is determined as function of either SPT blowcount or CPT tip resistance and fines content. These parameters are plotted together in the form of the familiar charts that divide in liquefiable and non-liquefiable zones shown in Chapter Three. The cyclic resistance of the soil is then determined by entering the chart at the corrected penetration resistance, going to the appropriate curve based on the fines content of the coil and then reading the corresponding cyclic resistance off the vertical axis. Some charts do not have multiple curves for the various fines contents but employ correction factors for fines content Soils Below the Limiting Silt Content The cyclic resistance of soils with non-plastic fines below the limiting silt content has been shown to be solely a function of the soil specific relative density and to be independent of silt content. Since cyclic resistance is independent of silt content, the differences which occur in the cyclic resistance versus penetration resistance curves with fines content is due solely to the differences in penetration resistance caused by the presence of the silt. Thus the separate curves for different fines content are a result of the differences in penetration resistance brought about by the varying silt content, not a difference in liquefaction resistance. The presence of silt has long been known to lower the penetration resistance of a loose sand because it prevents full drainage of the soil. In order to correct for this lowering of the penetration resistance, some correction factor must be applied to the penetration resistance in order to make the use of the clean sands curve applicable. Because this correction has been shown to be correcting only for penetration resistance and not for cyclic resistance, the appropriate fines correction factors will cause the cyclic resistance curves for different fines contents to collapse on top of the clean sands curve. This, of course, is only the case when the soil under investigation has a limiting silt content of greater than the non-plastic fines content for which the cyclic resistance curve was intended. 65

73 SPT-based Correlations The correction that was proposed in this recent study, considering the behaviour of soils with non-plastic fines below the limiting silt content, is to adjust the NCEER curves so that they become coincidental with the clean sands curve when the fines corrections are applied. The new termγ is to be used in conjunction with the NCEER α and β factors. It is both a function of fines content and the corrected SPT blow count, N 1,60. ( ) α + β ( N + γ 1 60 = 1) 60 N CS (Eq. 24) Where: α and β as for NCEER γ = (( N ) 4)( FC ) NCEER Youd and Idriss (1997) α = 0 α = exp 1.76 α = 5.0 β = 1.0 β = β = [ (190/ FC )] 1.5 [ ( FC /1000)] For FC < 5% For 5% < FC < 35% For FC > 35% For FC < 5% For 5% < FC < 35% For FC > 35% (Eq. 25) 66

74 Applying this correction factor to the standard curves the cyclic resistance curves for the 15 and 35 percent fines have converged onto the clean sand curve (Fig.5.2). This indicates that, when applied to soils with silt contents below the limiting silt content, an identical cyclic resistance will be determined regardless of the silt content and is only a function of their soil specific relative density. Fig.5.2 Cyclic resistance curves based on NCEER (1997) and correction with the term γ CPT-based Correlations The procedure for performing the simplified method based upon CPT results involves correcting the penetration resistance for fines based upon the soil behaviour type index, I C, proposed by Robertson and Wride (1998). It is important to note that, in the development of these correlations of the simplified method, the nature and plasticity of the fines present in the soil were not taken into consideration in a consistent manner. Also, the occurrence of liquefaction at a site was judged by the investigators(s) from the appearance of sand boils, settlement and/or damage of overlying structures, or lateral ground spreading. The nonoccurrence of liquefaction was assumed by the lack of the aforementioned liquefaction evidence. Liquefaction occurrence may not be easily detected on the ground surface, and penetration resistance data are usually obtained from sites that have already undergone considerable shaking. Since prior cyclic loading can increase the liquefaction resistance of soils (Finn et al. 1971; Ishihara and Okada 1978; Seed et al. 1977, 1988) and thus affect the relationship between penetration resistance and cyclic resistance. Seed et al. (1988) found, from cone penetration tests on large-scale triaxial specimens, that prior cyclic history also increases the cone-sleeve friction, even though no significant effect on the tip resistance was observed in these tests. 67

75 The correlation here presented was proposed by Carraro, Bandini and Salgado (2003) in a recent paper on the effects of nonplastic fines in liquefaction initiation. This correlation was developed following an approach targeted at avoiding the subjectivity inherent in the observation of liquefaction in the field and the effects of prior shaking. A (CRR) 7.5 q c1 correlation is obtained by combining the results from the penetration resistance analysis of Salgado et al. (1997) using CONPOINT and a considerably large set of laboratory test results. It was found in this study that, for a given relative density, sand with nonplastic silt up to 10 percent has a slightly higher cyclic resistance than clean sand. However, cone resistance increases at a faster rate with fines content than cyclic resistance, and the proposed liquefaction resistance curves for 5 and 10 percent silt content are located to the right of the clean sand curve when they are usually to the left (Fig.5.3). This gradual shift to the right of the CRR-qc1 curves may be explained by observations made by Salgado et al. (2000). These authors found, for the same materials used in the present study, that sands with low silt content and a fabric in which the sand particles are mostly or completely in contact are more dilative than clean sand. For a given relative density, dilative silty sands subjected to cyclic loading show higher resistance to liquefaction than clean sand. Cone resistance is increased not only by dilatancy, but also by the critical-state friction angle, which consistently increases with the addition of fines. Fig.5.3 Proposed correlations between cyclic resistance ratio (CRR) 7.5 and normalized cone resistance qc1 for sand with 0, 5 and 10% nonplastic silt 68

76 The results of the cyclic triaxial tests show that, for a given relative density, the cyclic resistance of sand increases slightly when small amounts of nonplastic silts are added as was stated in Chapter 4. The cyclic resistance of sand with 5 percent silt content increases approximately 25 percent with respect to the resistance of clean sand and this can be seen in Fig.5.4 Cyclic Resistance (CRR) 7.5 versus skeleton void ratio (e s ) for clean and silty sands (after consolidation) Further study is needed in this topic since not all silts result in an increase in cone resistance as can be seen in Olsen and Mitchell (1995) approximate soil characterization framework based on normalized cone penetration resistance and friction ratio. If the use of the clean sands curve would be conservative or not in that case is still an uncertainty. The presence of nonplastic fines does not always have a positive effect of the liquefaction resistance of a sand. To assume that, for a sand with nonplastic fines below the limiting silt content, the cyclic resistance is the same as for the clean sand can be considered a conservative approach to the matter considering the conclusions presented in Chapter 4 (framework proposed by Thevanayagam, 2000). 69

77 5.2.2 Soils Above the Limiting Silt Content For soils above the limiting silt content, a large decrease in cyclic resistance is observed. Fig.5.5 shows the cyclic resistance curves for Monterey sand with 25 percent silt and Monterey sand with 35 percent silt adjusted to a soil specific relative density of 50 percent. These silt contents represent conditions just above and below the limiting silt content of 32 percent calculated for this sand. The decrease is approximately of 300 percent from 0.24 to 0.08 when evaluated at 10 cycles. Fig.5.5 Variation in cyclic resistance above and below the limiting silt content 70

78 If the 35 percent fines curve or the correction factors are applied to soils with limiting silt contents below 35 percent, the predicted cyclic resistance may be grossly over predicted. This represents a very dangerous situation that may also occur in soils with fines content below 35 percent, as long as they are above their limiting silt content. Fig.5.6 Variation in cyclic resistance with silt content for Yatesville sand specimens adjusted to 25% soil specific relative density Currently there are no correlations specific for sands with silt above the limiting silt content but, according to a recent paper of Boulanger and Idriss (2006), non-plastic silt exhibits a very similar behaviour to that of clean sand when subject to cyclic loading. This similarity in behaviours might indicate that the clean sands curve can be used to assess the liquefaction potential for sands with silt above the limiting silt content but the corrections for fines content must be investigated before being used. The reason for this need for investigation is due to the fact that it is the silt fraction void of the soil that controls behaviour so it would be the sand to fill the voids. However, the behaviour of silt with sand in the voids may not the similar to that of sand with silt in the voids since, according to Martin and Polito (2001), the amount of sand present has little effect in the soil s cyclic resistance. Also, the difference in penetration resistance might not be enough to illustrate the strength loss due to the fines content being above the limiting silt content of the soil. 71

79 5.2.3 Parametric Analysis In order to test the correlations presented before in this work a parametric analysis was performed. This analysis was based in values from Olsen and Mitchell (1995) approximate soil characterization framework based on normalized cone penetration resistance and friction ratio (Fig.5.7). Fig.5.7 Approximate Soil Characterization Framework (Olsen and Mitchell, 1995) Values of the Friction Ratio and the Normalized Cone Resistance for Loose and Medium Dense lines where determined from these charts for the lines of Fines < 5%, Fines = 10 to 15% and Fines = 40 a 60% as can be observed in Fig.5.8. No intermediate value of fines content between the 10 to 15 percent and the 40 to 60% was analysed for it is complicated to evaluate if the soil is above or below the limiting silt content since Martin and Polito (2001) showed that, for most soils, the limiting silt content is within that interval. The values for the analysis were chosen in order to try to obtain cone resistance values for soils with varying silt content but with the same soil specific relative density. 72

80 Fig.5.8 Values chosen for the parametric analysis of the correlations For this analysis it was considered that when the fines content were varying from 40 to 60 percent that it was above the limiting silt content for that soil, the fines varying from 10 to 15 percent to be below the limiting silt content and when the fines were below 5 percent it was considered to be clean sand. After obtaining the values of cone penetration resistance and friction ratio a correlation was used to obtain the equivalent N 60 for the SPT, as a function of the medium grain diameter, which was presented before in this work. This was done in order to compare the CRR obtained from the CPT charts with the charts of the SPT presented in Chapter Three. In order to obtain the medium grain size diameter, a sample of Yatesville, already shown in this work, was considered and values of D 50 were obtained from Martin and Polito (2001) for different values fines content. These values are shown in the following table and the grain size distribution can be seen in the Appendix. Silt Percent [%] Medium Grain Diameter, D 50 [mm] 0,17 0,17 0,17 0,06 Table 2 Medium Grain Diameter for mixtures of Yatesville sand with silt Afterwards, a comparison was made between the CRR from the general procedure to those of specific for nonplastic fines content, but only for the soils with fines content below the limiting silt content, since for silt content greater than the limiting silt content there are no specific correlations. The correlations specific for soils with nonplastic fines, below the limiting silt content, analysed were the ones presented by Martin and Polito (2001) for the SPT, and the one proposed by Carraro, Bandini and Salgado (2003) for the CPT. 73

81 Soil Characteristics Loose Clean Sand Loose Silty Sand Loose Sandy Silt Medium- Clean Dense Sand Medium- Silty Dense Sand Medium- Sandy Dense Silt FC Rf [%] [%] q c q c,1 q c q c,1,mod [atm] [MPa] [MPa] [MPa] CPT - CRR Current (5% Liquefaction Recommended Probability) Out of Chart Non-Applicable Out of Chart Non-Applicable Table 3 Parametric Analysis Results for the CPT correlations SPT - CRR Current Soil Characteristics N 1,60 C FINES N 1,60,CS (5% Liquefaction Probability) Recommended Loose Clean Sand Loose Silty Sand Loose Sandy Silt Non-Applicable Medium-Dense Clean Sand Medium-Dense Silty Sand Medium-Dense Sandy Silt Non-Applicable Table 4 - Parametric Analysis Results for the SPT correlations (converted from the CPT) As was stated before, the Clean Sand and Silty Sand are considered to have silt content below the limiting silt content. The results of these soils in the parametric analysis will now be discussed. Starting with the CPT correlations, it can be seen that both for the Loose and Medium-Dense sand with silt below the limiting silt content, the recommended and the current correlations have a nearly perfect correspondence. However, the proposed correlation for the CPT by Carraro, Bandini and Salgado (2003) is based on an increase in cone resistance with increasing silt content. Since this parametric analysis was based on the approximate soil characterization framework by Olsen and Mitchell (1995), so that type of behaviour could not be analysed. This means that the CRR determined by the proposed correlation is conservative in this analysis since it considers a decrease in cone resistance with increasing fines content. This 74

82 can be observed from the results of the analysis were the CRR for the proposed approach are either equal to that of the current approach ( Clean sand ), or so low that can not even be observed in the chart of the proposed correlation. However, the current correlations used will be unconservative when considering a soil for which the cone resistance increases with increasing silt content. The SPT correlations are in very good agreement although it must be pointed out that, conceptually, the current approach is still wrong. Increasing cyclic resistance with increasing silt content without taking plasticity into consideration is erroneous as was shown by Martin and Polito (2001). Also, if the framework by Salgado et al. (2000) can be transposed to the SPT, since both the CPT and SPT measured penetration resistance, the current approach will be unconservative if increasing fines content increases SPT blowcount. Now discussing the Sandy Silt results from the parametric analysis one can clearly observe that the great decrease in cyclic resistance predicted by Martin and Polito (2001), when above the limiting silt content, does not occur when considering the current approach. This occurs because the current approach considers an increase in cyclic resistance with increasing silt content without considering plasticity. However, the current approach cannot be deemed unconservative since there is no data about how great is the difference between cyclic resistance between sand and silt. Since a similarity in behaviours between nonplastic silt and sand is known and, according to Martin and Polito (2001), when a sand has a silt content greater than the limiting silt content, it is the silt fraction void of the soil that controls behaviour, the clean sand curves might be considered adequate for liquefaction assessment. However, this hypothesis must be investigated as well as the correction factors to apply when the silt content reaches the limiting silt content. Again, this analysis was made considering a decrease in cone resistance with increasing silt content. If that does not occur, for soils with silt content above the limiting silt content, the current approach would be deemed unconservative. 75

83 CHAPTER SIX: CONCLUSIONS Since the 1960 s it is known that the presence of silt and clay particles will in some manner affect the resistance of a sand to liquefaction. Also, the plasticity index of the fines fraction has been recognized as an important factor in the liquefaction susceptibility of silty sand (Ishihara and Koseki 1989; Ishihara 1993). However, clean sands have been the main subject for research. In result of that there is a less understanding of the liquefaction of sand containing nonplastic fine-grained material (fines) than there is of clean sands. Based on the framework proposed by Martin and Polito (2001) in conjunction with the conceptual framework by Thevanayagam (2000), it is possible to isolate two factors, which govern the liquefaction behaviour of silty sands and sandy silts. Three distinct behavioural patterns were found for the cyclic resistance of soils composed of sand and non-plastic silt. The type of behaviour is determined by whether there is sufficient room in the voids created by the sand skeleton to contain the silt present without disturbing the sand structure. This silt content has been called the limiting content. When the silt content of a soil is below its limiting silt content, there is sufficient room in the voids created by the sand skeleton to contain the silt and the soil can be described as having silt contained in a sand matrix. The cyclic resistance of the soil is then controlled by the soil specific relative density of the specimen, where the soil specific relative density is calculated using the gross void ratio of the specimen and the maximum and minimum index void ratio for the particular mixture of sand and silt. Increasing the soil specific relative density increases the soil s cyclic resistance. The limiting silt content is usually between 15 to 40% fines content for a sand. If the silt content is greater than the limiting silt content, the specimen s structure consists predominately of sand grains suspended within silt matrix with little sand grain to sand grain contact. Above the limiting silt content, the amounts of sand present in the soil and its soil specific relative density have little effect on its cyclic resistance. The cyclic resistance for these soils is then controlled by the silt fraction void ratio of the soil. Decreasing the silt fraction void ratio increases the soil s cyclic resistance. There is a transition zone consisting of soils with silt contents at or slightly above the limiting silt content. This zone occurs as a result of the structure of the soil changing from a predominately sand controlled fabric to a predominately silt controlled fabric as silt content increases. 76

84 The liquefaction potential assessment in situ has been greatly developed in the last years for clean sands. The most common tests for evaluation liquefaction triggering potential in situ are the SPT and the CPT, both penetration tests since they describe cyclic resistance more accurately than other tests. The charts developed for these tests in order to evaluate the cyclic resistance of the soil are essentially based on historical data and are being constantly updated with new data from recent earthquakes. Among other factors, these charts have different correlations between penetration resistance and cyclic resistance for different fines content. These correlations are also based on historical data but most lack the plasticity of the fines with which the different correlations of the chart were developed. This lack of distinction for different plasticity of the fines can render the correlations unconservative when considering nonplastic fines in the sand voids considering the influence of nonplastic fines in the cyclic resistance of sand here presented. After reviewing the literature for correlations specific for sand with nonplastic fines, and conducting a small parametric analysis for sand with different amounts of fines content some conclusions can be made. When the silt content of a soil is below its limiting silt content, the current correlations provide a good approach to the cyclic resistance that is expected of the soil since it does not differ much from the clean sand cyclic resistance. The decrease observed in the parametric analysis is due to the fact that the analysis is based a decrease in penetration resistance with increasing silt content. If the silt content is greater than the limiting silt content, no definitive conclusions can be made for further investigation on this topic is required. The great loss of cyclic resistance predicted by Polito could not be observed in the analysis made and, since currently there are no correlations specific for sands with silt above the limiting silt content, the accuracy of the current approach could not be evaluated. Also, the difference in penetration resistance might not be enough to illustrate the strength loss due to the fines content being above the limiting silt content of the soil. However, the similarity in behaviour between sand and silt, shown by a recent study presented before in this work, might indicate that the clean sands curve can be adequate for the evaluation of liquefaction resistance for sands with silt greater than the limiting silt content. This correlation might be possible since Martin and Polito (2001) showed that it is the silt controlling the behaviour of the soils when considering silt contents of that magnitude and that the presence of sand does not influence the behaviour in a significant manner. The findings in the study conducted by Salgado et al. (2000) should be investigated further. This need for investigation is due to fact that the findings of that study suggest that increasing silt content of a soil increases its penetration resistance. This finding would render all current correlations unconservative when considering silt that presents this behaviour. This trend as been shown to be present in the CPT and should also be verified for SPT. The proposed correlation presented by Carraro, Bandini and Salgado (2003) for the CPT was chosen based on its conservative nature on a topic which clearly needs further investigation. 77

85 REFERENCES Boulanger, R. W., and Idriss, I. M. (2004). Evaluating the potential for liquefaction or cyclic failure of silts and clays. Report UCD/CGM-04/01, Center for Geotechnical Modelling, University of California, Davis, CA, 130 pp. Boulanger, R. W., and Idriss, I. M. (2006). "Liquefaction susceptibility criteria for silts and clays." Journal of Geotechnical and Geoenvironmental Engineering, ASCE, 132(11), Boulanger, R. W., and Idriss, I. M. (2007). "Evaluation of cyclic softening in silts and clays." Journal of Geotechnical and Geoenvironmental Engineering, ASCE, 133(6), Bray, J.D., and Sancio, R.B. (2006). Assessment of the liquefaction susceptibility of finegrained soils. Journal of Geotechnical and Geoenvironmental Engineering, ASCE, 132: doi: /(asce) (2007)133:6(641). Erten, D., Maher, M.H. (1995). Liquefaction Potential of Silty Soils. Soil Dynamics and Earthquake Engineering, Proc. VII Int. Conf., A.S. Cakmark and C.A. Brebbia, eds., Elsevier Science, London, Folque, J. (1980). Liquefacção de Solos Arenosos. GEOTECNIA, nº29, J. Laue and J. Buchheister, (2004). Condition indicators for liquefaction susceptibility with focus on silty soils. International Conference on Cyclic Behaviour of Soils and Liquefaction Phenomena, Th. Triantafyllidis, A.A. Balkema Publishers, Bochum, Germany. Jefferies, M., and Been, K. (2006), Soil Liquefaction: A Critical State Approach, Taylor & Francis, 479 p. Jorge Luis Cardenas Guillen. Estudo de Modelos Constitutivos para Previsão da Liquefacção em Solos sob Carregamento Monotônico f. Dissertação (Mestrado em Engenharia Civil) - Pontifícia Universidade Católica do Rio de Janeiro, Coordenação de Aperfeiçoamento de Pessoal de Nível Superior. Orientador: Sergio Augusto Barreto da Fontoura. Kramer, S.L. (1996). Geotechnical Earthquake Engineering, Prentice Hall, Upper Saddle River, NJ, 653 p. 78

86 Lai, Sheng-Yao, Chang, Wen-Jong (2006). "Logistic Regression Model for Evaluating Soil Liquefaction Probability Using CPT Data" Journal of Geotechnical and Geoenvironmental Engineering, ASCE, 132(6), Moss, R. E. S. (2003). CPT-Based Probabilistic Assessment of Seismic Soil Liquefaction Initiation, Ph.D. dissertation, University of California Berkeley Moss, R.E.S., Seed,.R.B., Kayen, R.E., Stewart, J.P., Kiureghian, A. Der., Cetin, K.O. (2006). CPT-based Probabilistic and Deterministic Assessment of In Situ Seismic Soil Liquefaction Potential. Journal of Geotechnical and Geoenvironmental Engineering, ASCE, 132(8), Polito, C. P. (1999). The effects of non-plastic and plastic fines on the liquefaction resistance of sandy soils. Ph.D. thesis, Virginia Polytechnic Institute and State University, Dec. Polito, C. P., and Martin, I., J. R. (2001). "Effects of Nonplastic Fines on the Liquefaction Resistance of Sands." Journal of Geotechnical and Geoenvironmental Engineering, 127(5), Prakash, S., Puri, V.K. (1999). Liquefaction of Silts and Silt-Clay Mixtures. Geotechnical and Geoenvironmental Engineering, 125(8), Robertson, P.K. (2008). Discussion of Liquefaction potential of silts from CPTu. Canadian Geotechnical Journal, 45: doi: /t Robertson, P.K., Wride, C.E. (1998). Evaluating cyclic liquefaction potential using the Cone Penetration Test Canadian Geotechnical Journal, 35: Salgado, R., Bandini, P., Carraro, J.A.H. (2003). Liquefaction resistance of Clean Nonplastic Silty sands Based on Cone Penetration Resistance. Journal of Geotechnical and Geoenvironmental Engineering, ASCE, 129(11), Seed, H.B., Cefin, K.O., Moss, R.E.S., Kammerer, A.M., Wu, J., Pestana, J.M., Reimer, M.F., Sancio, R.B., Brag, J.D., Kayen, R.E., and Faris, A. (2003). Recent Advances in soil liquefaction engineering: A unified and consistent framework. Keynote Presentation, 26 th Annual ASCE Los Angeles Geotechnical Spring Seminar, Long Beach, Calif., 30 April Report , Earthquake Engineering Research Institute, Berkeley, Calif. 71 pp. Stark, Timothy D., Olson, Scott M. (1995). "Liquefaction resistance using CPT and field case histories." Journal of Geotechnical and Geoenvironmental Engineering, ASCE, 121 (12),

87 Xenaki, V.C., Athanasopoulos, G.A. (2001). Discussion of Effects of Nonplastic Fines on the Liquefaction Resistance of Sands. Journal of Geotechnical and Geoenvironmental Engineering, ASCE, 127(5), Y. Nabeshima, M.A. El Mesmary and T. Matsui (2002). Effect of Non-Plastic Fines on Liquefaction Characteristics of Sandy Soils under Cyclic Loading, Proceedings of The Twelfth (2002) International Offshore and Polar Engineering Conference Kitakyushu, Japan, May 26-31, Youd, T.L. (1996). Preliminary Report From the NCEER Workshop on Evaluation of Liquefaction Resistance of Soils: Proceedings, 4th Caltrans Seismic Research Workshop, Sacramento, California. July 9, Salgado, R., P., Carraro, (2004). Mechanical Behaviour of Non-Textbook Soils. Joint Transportation Research Program, Purdue University, Project No. C-36-50X 80

88 APPENDIX 81

89 Fig. Appendix 1 R d Results for Various Bins Superimposed with the Predictions (Mean and Mean ±1σ) Based on Bin Mean Values of V s,40 ft, M w and a máx 82

90 Fig. Appendix 2 - R d Results for Various Bins Superimposed with the Predictions (Mean and Mean ±1σ) Based on Bin Mean Values of V s,40 ft, M w and a máx 83

91 Fig. Appendix 3 - R d Results for Various Bins Superimposed with the Predictions (Mean and Mean ±1σ) Based on Bin Mean Values of V s,40 ft, M w and a máx 84