Theoretical and Experimental Relationships between Stress Dilatancy and IDS Strength Comporents By KUM-HUNG HO A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF ‘THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 1971,To my wife, Nancy, for her con- stant support and encouragement without which this dissertation Would not have been completed.ACKNOWLEDGEMENTS The author gratefully expresses his appreciation to Dr. John H. Schmertmann, Chairman of his Supervisory Com- mittee,for the encouragement and guidance during the course of this study. ‘The writer is also indebted to Dr. M. W. Self and Dr. F, G, Martin for serving on his Supervisory Committee. This investigation was supported by National Science Foundation Grant No. GK-1908X which provided the research assistantship to make this work financially possible. Thanks are due Mr. William Whitehead who provided general assistance in the laboratory.ACKNOWLEDGEMENTS . . LIST OF TABLES . . . LIST OF FIGURES . . NOTATION. . 2 2. ABSTRACT... 1. mm. ur. Iv. INTRODUCTION . . TABLE REVIEW OF PREVIOUS Historical Background . OF CONTENTS WORK Methods of Separating Shear Resistance. Hvorslev's "True" Strength Parameters . . « Schmertmann's IDS Concept. . Rowe's Stress Dilatancy Theory . Other Methods . a)The Critical state concept b)Theory of Rate Process . ¢)Stress Loci Method Definition of Parameters Soil structure Bond Strength Frictional Strength. - THEORETICAL CONSIDERATIONS AND ROWE THEORIES . Schmertmann's IDS Concept . Rowe's Stress Dilatancy Theory. Experimental Plots. Definition of Strain Paths. Stress Dilatancy Method . IDS Method. . . OF SCHMERTMANN Need for Establishing Linearity TEST METHOD, EQUIPMENT AND EXPERIMENTAL RESULTS... . - Test Method... I, and D, Component swith Dilatancy Corrections | Equipmenf. . . Vertical Load Measurement. Vertical Deformation . Back Pressure Saturation . Volume Change Measurements iv Page Aid svi 28 +28 +33 +38 +40 +40 41 42 43 +43 44 + 46 - 46 - 46 + 48 + 50ve vr. Sample Preparation and Testing. ..... 2.4. Kaolinite Samples... . . Constant-s} series . . Constant-p’ series . . Constant-volume series. Glass Beads. . . Glass Beads with Hydrocal Correction of Measured Data. . Vertical Stress, 0}... ~~ Strength of filter strips | Rubber membrane correction. Weights of plunger and rod. Plunger friction... ... Horizontal Stress, of»... Vertical Deformation... - Volume Change Readings. . . . . Apparent membrane correction. Membrane penetration correctio: Temperature, . 2. ee ee ee ie: 5S n PROOF OF LINEARITY OF STRAIN PATHS... +. +--+ Typical Experimental Results. ......,-2--% Statistical Analysis and Inference... 1... +s Linear Regression and Correlation . | | pint Comparisons of Drained Tests and IDS Tests. with Volume Change Measurements... ...... + Analysis of Experimental Data... 1. ee ee eee Glass Beads... ee ee Kaolinite. 12.1.1 tll IDS Tests with Volume Change Measurements. Comparisons of Bond Strength I, and c¢ Kaolinite... . ee eee ee ee Glass Beads... . . eee ee Glass Beads with Hydrocal. 1 1) 2 1 Confidence Limits on Dona Strength and Correlation Coefficients. .... . Confidence Intervals for te and ep Confidence Limits for 9; afd og. + If versus Ss" Relationships... | Lingarity of Btrain Paths... .. : THEORETICAL RELATIONSHIPS BETWEEN IDS AND STRESS DILATANCY PARAMETERS... . -- + eee e+ Ip and og Parameters. . . « ce and If parameters. | |! ag and op Parameters. | |! ge and $f Parmeters . |. Simmary 5¢ ¢ and a Paroneter Relationships between $, dg and @g in Cohesionless materials on Goliparison of #<, 8 and ¢, Parameters. + Correlation of Gg anda... 2... 0 S +54 +56 +58 58 58 +59 359 +59 359 +62 90 +90 +92 +93 94 +94 +95 97 101Page ay Theoretical Computations of I, Cg, gand a Parameters... ke ee ee 108 Cohesionless materials...) ee 108 Cohesive soil. 2... ee ee ee ee ee odd Discussion... .. . . eee eee ee ee MD Bond Strength Ig andcp s+ 12 i lee. dl $ anda Parameters... 1. 222222220 Ii0 m2 VII. DISCUSSIONS AND CONCLUSIONS... ....+...+.. 114 Mobilised Stress-Strain Path... ......... 14 Comments on IDS and Stress Dilatancy Methods of Obtaining Bond Strength... .... . 116 Conclusions. . 1 eee ee ee ee ee ee eee HT Suggestions for Future Research. |) 222. 11. 1119 APPENDIX A PHYSICAL PROPERTIES OF KAOLINITE (EDGAR,PLORIDA) GLASS BEADS AND GLASS BEADS WITH HYDROCAL . - . ee ee ee ee ee ee ee + 120 APPENDIX B CORRECTIONS FOR EXPERIMENTAL DATA... . . 124 APPENDIX C ANALYSIS OF EXPERIMENTAL DATA. ..... - 129 Stresses and Strains and Component Separation . . . 129 Kaolinite... ...-. : tee ee 130 Drained Test Series on 0-C. Kaolinite, |... 130 Constant-o} Test Series on 0.C. Kaolinite . . . 134 Constant-o} Test Series on N.C, Kaolinite . . . 140 Constant-p! Test Series on 0.C. Kaolinite . . + 161 Constant-p' Test Series on N.C. Kaolinite . . . 167 Constant-Volume Test Series on 0.C. Kaolinite... ++ .eee wee aT Constant-Volume Test Series on N. ce. Kaolinite . - eee eee 183 Glass Beads. eee ee ee ee ee ee ee 189 Drained Test Series on Glass Beads | |. . |: 189 Constant-c} Test Series on Glass Beads ... . 192 Glass Beads with Hydrocal... .. sone ee 202 Constant-o} Test Series on Glass Beads with Hydrocal se eee ee ee ee ee ee 202 APPENDIX D a, AND TAN Gp VS STRAIN RELATIONSHIPS . . . 217 LIST OF REFERENCES © ee eee ee ee ew 225 BIOGRAPHICAL SKETCH. - se ee ee ee ee eee ee 230TABLE II-1 Iv-1 ve v-7 v-8 v-10 viI-1 viI-2 vI-3 vI-4 VI-5 vI-6 LIST OF TABLES Definitions of $ Angles Required Back Pressure For Achieving 1008 and 99% Saturation Interparticle Parameters cp and $¢ for Glass Beads Interparticle Parameters cg and $, for 0.C. Kaolinite Regression Coefficients of IDS and Stress Dilatancy Methods on N.C. Kaolinite Regression Coefficients of IDS and Stress Dilatancy Methods on 0.C. Kaolinite Regression Coefficients of IDS and Stress Dilatancy Methods on Glass Beads Regression Coefficients of IDS and Stress Dilatancy Methods on Glass Beads with Hydrocal Confidence Intervals for Ig and c¢ Lower and Upper Confidence Limits of Population Correlation Coefficients p; and p¢ Regression Coefficients for I* vs df" Relationships Lower and Upper Confidence Limits of Population Correlation Coefficient o* $_ and a, Values for Two Cases Computed $ and a Values for Glass Beads ‘Two Data Points for N.C. Kaolinite Computed Parameters for N.C. Kaolinite Comparisons of ce, Ig, Ig Values at Each Strain Effect of Soil Types on 8 vii PAGE 27 ag 75 16 79 80 82 83 85 86 87 88 97 109 10 no 12 113Table aA-1 A-2 Water Contents, Degrees of Saturation and Void Ratios of Drained Tests and Constant-o} IDS Tests on Kaolinite Water Contents, Degrees of Saturation and Void Ratios of Constant-p' and Constant-Volume IDS Test on Kaolinite Water Contents and void Ratios of Glass: Beads Water Contents and Void Ratios of Glass Beads with Hydrocal vidi Page 121 122 123 123LIST OF FIGURES FIGURES II-1 II-2 11-3 UI-1 111-2 11-3 qI-4 11-5 III-6 II-7 III-8 TI-9 Iv-1 IVv-2 IVv-3 Iv-4 Iv-5 I, and D, Components of Shear Resistance € Constant Structure Envelope $ Values as a Function of Porosity Constant Structure Envelope Showing Fitted Se- cant and Tangent to Obtain T, Components Separation of Shear Resistance into I, and De Components at Constant Structure and Strain vs Ds, Grain Size at Constant Effective ress Two Cohesionless Particles at Instant of Slip Stress Summation over ap Plane in Triaxial Sample Two Cohesive Particles at Instant of Slip Stress Dilatancy Relations for Loose and Dense Sand Definition of Strain Path by Stress Dila~ tancy Method Definition of Strain Path by IDS Method Separation of Shear Resistance at Constant Structure and Strain with Correction for Dilatancy Schematic Diagram of Triaxial Apparatus for IDS Test with Volume Change Measurements Effect of Filter Strips, Test Nos. 3-18-Nc2, 3-17-NC2 Bedding Correction of Top Cap, Ball Joint and Frictionless Platens Apparent Membrane Correction at Normally Con- solidated Condition ix PAGE 10 13 29 31 32 34 36 37 39 40 41 45 47 57 60 61FIGURE PAGE Iv-6 Membrane Penetration Correction for Glass Beads 63 v-1 Drained Test on 0.C. Kaolinite 66 v-2 Drained Test on Glass Beads 67 V-3 Constant-o{ IDS test on N.C. Kaolinite 68 v-4 Constant-o} IDS Test on Glass Beads 69 v-5 Constant-oj IDS Test on Glass Beads with Hydro cal 70 vV-6 I, vs Tf Plot 73 V-7 tt vs Sf Plot 73 VI-l Isolation of Interparticle Parameters, cg and $¢ 91 VI-2 Mohr Circles for Cohesionless Element 96 . VI-3 ¢ vs Strain Relationships for Glass Beads 98 ) VI-4 vs Strain Relationships for Glass Beads with Hydrocal 98 VI-5 ¢ vs Strain Relationships for Three Different Stress Paths on N.C. Kaolinite 99 VI-6 vs Strain Relationships for Three Different Stress Paths on 0.C. Kaolinite 100 VI-7 tan ay vs ag for Constant} Test Series on N. C. Kaolinite 102 ViI-8 tan a, vs a, for Constant-Volume Test Series on N.G, Kaolinite 103 VI-9 tana, vs a, for Constant-p' Test Series on N.C. Raolinite 104 vi-10 tan ap vs a, for Three Different Stress Paths on 0.8, Kaofinite 105 vi-ll tan a,, tan ¢,, 4, vs Strain Relationships for Glass Beads 106 VI-12 tana, vs a, for Glass Beads with Hydrocal 107 VII-1 Mobilised "saw-tooth' Stress Strain Path 114Pigure B-1 Effect of Filter Strips, Test Nos. 3-21-NC1 and 3-16-NC1 Effect of Filter Strips, Test Nos. 3-20-NC4 and 3-19-NC4 Apparent Membrane Correction at Normally Consolidated Condition Apparent Membrane Correction at Overconsolidated Condition a, and tan ap vs Strain, Constant-o} Test Series On N.C. Kaolinite Q, and tan Gp vs Strain, Constant-p' Test Series On N.C. Kaolinite a, and tan a, vs Strain, Constant— vélume Test Series on N.C. Kaolinite @, and tan ap vs Strain, Constant-of Test Series on 0.C. Kaolinite a, and tan Gp vs Strain, Constant-p' T8st Series 6n 0.C. Kaolinite a, and tan ap vs Strain, Constant- vOlume Test Series on 0.C. Kaolinite a, and tan Gp vs Strain, Constant-o} Tést Series On Glass Beads with Hydrocal, xi Page 125 126 127 128 218 219 220 221 222 223 224© Pre ee a NOTATION zero intercept of Rowe's stress path plot average area at point of contact of particles area of specimen after consolidation phase cross section area of piston initial cross section area of membrane cross section area of specimen cohesion term ini Coulomb's equation Hvorslev effective or 'true' cohesion cohesion corrected for dilatancy, interparticle cohesion Gibson's modified Hvorslev c, parameter zero cohesion intercept interparticle cohesion strained at minimum energy conditions of formation of soil confidence interval slope of constant structure envelope = tan ¢_ dilatancy factor = (1 + 2) effective stress Dependent component of mobilised shear resistance diameter of particle at which 50% of the soil is finer void ratio of specimen void ratio after consolidation phase final void ratio initial void ratio Young's modulus xii .energy ratio @ilatancy component in direct shear test specific gravity of soil solids height of spicimen stress history Independent-Dependent-strain maximum value of I, component in IDS test bond strength effective stress Independent component of mobilised shear resistance I, component correctéd for dilatancy correction factor for filter paper tan? (45 - $¢/2) tan? (45 + $¢/2) ratio of length of projected solid path to number of contact points in vertical direction ratio of length of projected solid path to number of contact points in horizontal direction slope of Roscoe's p vs q line porosity number of samples normal load at point of contact of particles oft2o$ TT mean principal stress vertical force shear stress horizontal force sample correlation coefficient for tf vs Tf! sample correlation coefficient for I, vs St relationship ® e xiiixt sample correlation coefficient for I¥ vs Tf! relationship radius of sample principal stress ratio = 01/9} uplift on piston degree of saturation degree of saturation after consolidation phase final degree of saturation initial degree of saturation sample standard deviation soil structure, geometric arrangement of soil particles thickness of membrane time to end of primary consolidation temperature Pore water pressure water content of specimen water content after consolidation phase final water content initial water content horizontal relative movement ox abscissa vertical height or movement or ordinate sample mean xivGreek Alphabet Rowe's structural parameter Schmertmann's soil structure parameter in con- stant structure envelope slope of I, vs Fj line and also parameter in con- ‘stant st@ucturé envelope zero intercept of linear model slope of line in linear model increment of major principal strain, defined as positive in compression increment of minor principal strain, defined as positive in compression = Oly - of,, difference in major principal effective stfess if’ rps test strain axial strain due to consolidation and undrained deformation error component volumetric strain, positive for volume increase strain in same direction asv:, positive for com- pression strain in same direction as 03, negative for hori- zontal decrease strain rate slope of ce vs of line rate of change of volumetric strain to vertical strain = de, e, Population mean population correlation coefficient for T# vs Tg" relationship(o,-03) population correlation coefficient for I, vs 5; relationship population correlation coefficient for If vs Gt" relationship t effective stress component derived from primary chemical bonds effective stress component derived from van der Waal's forces and electrostatic forces variance of error component o - u, where ' indicates effective stress effective consolidation pressure equivalent effective consolidation pressure pressure normal to failure plane hydrostatic cell pressure average normal effective stress at any strain between the two points of common Mohr circle tangency in an IDS test Sj, corrected for dilatancy major principal effective stress major principal effective stress at high pressure level major principal effective stress at low pressure level minor principal effective stress deviator stress = k(o}, + of), average major principal effective streds **% = (0, - 05), shear stress shear stress on failure plane average shear stress at any strain between two points of common Mohr circle tangency in an IDS test 4 corrected for dilatancy xvishear force at point of contact of particles angle of internal friction in Coulomb's equation interparticle friction angle sheared at constant ‘volume Hvorslev effective or ‘true’ angle of internal friction angle of internal friction corrected for dilatancy, interparticle friction Gibson's modified Hvorslev $, parameter friction angle from Roscoe's Critical State Concept Coulomb's $ after deduction of external work by dilation friction angle applied to a, plane xate of change of D, component of mobilised shear resistance © $_ corrected for dilatancy physical angle of friction between the minerals slope of Rowe's stress path plot xvii,Abstract of Dissertation Presented to the Graduate Council of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy THEORETICAL AND EXPERIMENTAL RELATIONSHIPS BETWEEN STRESS DILATANCY AND IDS STRENGTH COMPONENTS By Kum-Hung Ho August 1971 Chairman: Dr, John H. Schmertmann Major Department: Civil Engineering The object of this study is to establish the rela~ tionships between Schmertmann's IDS components of I,, tan %_ and ag, and Rowe's interparticle parameters of ce, tan $¢ and ay. Both investigators realised the importance of soil structure but took entirely different approaches to account for this effect. Both sets of parameters are claimed to be more fundamental than Hvorslev's 'true' para- meters of c, and $, and,moreover, they are obtained as a function of strain. The IDS test was extended to include measurement of volume changes during shear so that both sets of para- meters can be computed from the data on the same sample. The linear relationship between the I, component and 3j, the average normal effective stress at each strain,has been shown to be valid for the three types of materials inves- tigated, i.e. kaolinite, glass beads and glass beads with hydrocal. xviiiThe average shear stress and average normal effective stress corrected for dilatancy, tf and Uf", are also demonstrated to be linearly related at each strain for the three mater- ials with still higher positive correlation coefficients than the I, versus 3} correlation at each strain. Both the IDS and Stress Dilatancy Methods can be employed to determine 'bond strength’ I, and c, respectively, as a function of strain. Their difference lies in the effect of dilatancy, which is small. The choice of the method depends on the individual since each method has some advantages over the other. Introducing a dilatancy cor- xection into Schmertmann's method, giving I% and Of', is not recommended for the determination of bond strength because of the larger scatter. $) %er b, and ¢, friction angles are presented on the same Mohr diagram for cohesionless materials. The two a parameters show some correlation for the three materials tested. The application of small sample procedures in statistics provides a rational means of analysing and draw- ing conclusions from the data obtained from this work. xixCHAPTER I INTRODUCTION The concept of shear strength occupies an important position in the field of soil mechanics. The most frequent problem encountered by a soils engineer is that of stability, which requires knowledge of the shear strength of soils. The selection of the proper value of shear strength to suit a particular field problem is sometimes difficult, es- pecially for fine grained soils. In practice, the shear strength parameters at failure, ¢ and $ are determined from a series of test samples and these are then substituted into the expressions for bear- ing capacity or slope stability. The separation of shear resistance into components of ‘cohesion’ and ‘friction’ has recieved wide attention. Following Taylor (1948), energy corrections have been used for the analysis of shear data by many investigators (Penman, 1953; Gibson, 1953; Bishop, 1954; Newland and Allely, 1957; Roscoe, Schofield and Wroth, 1958; Bjerrum, 1961) and have been made only for failure or ultimate conditions. To better understand the mobilization of shear strength, Schmertmann and Osterberg (1960) introduced a new labora- tory technique and Rowe (1962) independently put forward a new mathematical theory to separate shear resistance intocomponents at any stage of deformation. These two entirely different approaches resulted in different parameters. A unifying concept is needed to relate the parameters obtained from these two methods. The object of this dissertation is to determine the relationships between the two sets of parameters obtained by Schmertmann's IDS Concept and Rowe's Stress Dilatancy Theory so that greater understanding into the mobilization of shear strength with strain will result. The experimental part of this study is confined to triaxial compression tests on three types of materials, i.e. kaolinite (cohesive), glass beads (cohesionless) and glass beads with hydrocal (artificial bonds). Three different stress paths of constant-c,', p' and volume were imposed on the extruded kaolinite while constant-o,' IDS tests were nade on the other two materials.CHAPTER II REVIEW OF PREVIOUS WORK Historical Background Although Coulomb had introduced his shear strength equation almost 200 years ago (1776), the present-day con- cept of shear strength is still built around his original form of equation where the maximum shear resistance was given by te = +o, tan $ Q-1) where c and ¢ are the apparent cohesion and angle of shear~ ing resistance respectively and og, the pressure normal to the failure plane. Mohr, in 1882, proposed a more general criterion of failure based on shear stress. He assumed that at failure the shear stress, Te and the normal stress on the failure plane, dgrare uniquely related Te, = £log) (2-2) This relation is represented on the shear stress ver- sus normal stress diagram by the envelope of all possible Mohr circles of stress for the material. when the rela- tion of the Mohr criterion is linear, it becomes identical with the Coulomb criterion, i.e. te =o +c, tan 6 (2-3) This is referred to as the Mohr-Coulomb failure criterion. In 1923, Terzaghi showed that effective stress, o',controls the deformation behaviour of saturated soils and it is equal to the total stress, o,minus the pore water pressure,u, ots o-u (2-4) The Mohr-Coulomb failure criterion for saturated soils is therefore unique when expressed in terms of effective stresses. te = 6" + (Gen) tan ¢! (2-5) This equation remains one of the most controversial in soil mechanics in the sense that the parameters c and ¢ are not unique. Methods of Separating Shear Resistance Hvorslev's 'True'Strength Parameters The triaxial test is most commonly used ir research work and in routine testing for the determination of effec- tive stress parameters. The standard types of triaxial tests are the undrained or UU test, the consolidated-un- drained test (CD) with pore pressure measurements and drained test. The undrained test in saturated clay gives a horizontal Mohr envelope where no volume or intergranu- lar stress changes take place. This yields the undrained shear strength for the ¢ = 0 method of analysis. Effective stress parameters are obtained by comparing samples having different intergranular stresses and volume changes at failure (or void ratio). Such conventionally determined strength parameters vary, among other factors, with stresshistory (normally or overconsolidated) and drainage condi- tions at the time of test. Since the series of tests de~ fining the envelope are of different void ratio and also different soil structures, their strength components are also different. ‘Thus the effective strength parameters ob- tained conventionally are empirical factors and not funda- mental properties of the soil material. Working with remolded clays, Hvorslev in 1937 came a little closer to obtaining the true cohesion and friction of a soil. He compared normally consolidated and overcon~ solidated samples having the same water content (void ratio) at failure. He obtained linear shear strength versus nor~ mal stress at failure lines whose slopes are equal and their inclination gives the effective angle of internal fxiction, $,. The zero intercept on the shear strength axis e° yields the effective cohesion component, Ce, for the parti- cular void ratio. Different cg values were obtained for different void ratios and it was found that they vary linearly with the equivalent consolidation pressure, o,', for clays remolded at high water contents. Bjerrum (1954) found that for clays remolded at low water contents, a zero cohesion intercept, corexists. ” Se = Cy + Koh (2-6) where k is the slope of the cg vs o{ line. Thus, on the basis of Hvorslev's work, Coulomb's equa- tion becomes Te = oy + OF tan ¢, (2-7)where cy is a function of the water content and be was approximately constant and! = normal effective stress at failure. Gibson (1952) proposed a failure criterion based upon energy considerations in direct shear tests. He retained the form of Hvorslev's equation but the meaning of the fun- damental parameters is changed. The equation becomes Te = &y + oF tang, + oF Se (2-8) where cg and 4, are the modified parameters, g¢ the dila- tancy component. Scott (1943) commented that Gibson's use of direct shear tests instead of triaxial tests was because of the diffi- culty of calculating or measuring the work done against the confining pressure from the gradient of the motion of the particles perpendicular to the failure plane, He refers to Gibson's dilatancy component as ‘geometric factor' which for any one soil will be reflected in the variation of void ratio and the grain orientations of the soil. Trollope et _al.(1965) suggested that the open, 3-dimen- sional random structure of aggregate units of crganic molecules of brown coal may be related to the cpen struc- tural sensitive clays which have a 3-dimensional card-house structure. The major difference between brown coals and sensitive clays is that the structure of brown coal units is more fibrous than platy as in clays and the existence of primary chemical bonds between elements of brown coal. They proposed a generalised effective stress equationwhereby the cohesion term may be considered as of a fric- tional nature and shear strength is given by T= (p, + Pg + Of) tan be (2-9) where = effective stress component derived from primary chemical bonds which is zero for clays. = effective stress component derived from van der Waals' forces and elec- trostatic forces. 's = similar to friction angle of Hvorslev and is constant for a given structural arrangement of particles. It should be noted that tan $, includes a geometric component as well as a coefficient of interpartical fric~ tion. Schmertmann's IDS Concept Schmertmann was the first to realise that any funda- mental cohesion and friction components must vary with strain, and to devise a test procedure to investigate such variation. The development of the IDS curve-hopping technique (Schmert- mann and Osterberg, 1960) in the separation of I,and D.com- ponents from shear resistance is one phase of research into the stress-strain-time response of cohesive soils dealing with stress-strain but with the time element held constant {method of testing, strain rate, specimen storege time, and any time dependent effects). The formulation of the effective stress principle im- plies a unique relationship between t, and ot. Actually shear strength, based on present knowledge (1971) of soilbehaviour is a function of a large number of factors: tg = fle re,€ 0"), TH, 8,C,A) (2-10) void ratio strain strain rate effective stress soil structure temperature stress history degree of saturation condition of formation of soil = rate of change of volumetric strain to vertical strain where ~oumananmo 1 Though the list of factors affecting shear strength looks formidable, fortunately not all of the factors are independent. Moreover, most of the factors are variable during deformation. For example, soil structure, $ is a function of H, C, e, o' and €. Void ratio is a function of o', H, C, and S. With the IDS technique, a given type of soil is sheared under constant temperature and strain rate at an identical soil structure where the void ratio change is negligible after a small variation in effective stress, the shearing resistance at any strain can be expressed as t= £(0',A) (2-11) Realising that Hvorslev's parameters were obtained at failure by comparing specimens at equal e (but not neces~ sarily of the same structure), Schmertmann invented his curve-hopping technique to separate shear resistance into I, and De components at any strain. Schmertmann (19633, 1963») found favourable agreement between (tan de)na, with Hvorslev's tan ¢, for remolded kaolinite, an undisturbed and remolded glacial lake clay and undisturbed Leda clay and suggested that I, and D, components are generalisations oflvorslev's parameters, It may be noted that at large strains where the comparisons of Hvorslev and Schmertmann components were made, the rate of volume changes of the samples would likely be small or negligible. The separation of shear resistance at a certain strain requires the fitting of a tangent to two Mohr circles hav- ing the same structure. The zero tangent intercept on the shear stress axis gives the I, component while the slope, tan 9,,is a measure of the shear resistance sensitivity to a change in effective stress at the particular soil struc- ture. The D_ component is the difference between the total shear resistance and I, I, = Gf tan 6, (2-12) where tangential points to the two Mohr circles. is the average normal effective stress at the ot Figure II-1 I, and D, Components of Shear Résistance10 Schmertmann and Osterberg (1960) had shown that the T, component (then called cohesion, c,) reaches a maximum at low strains (1% or less) while $, increases with strain, For remolded soils, the maximum value of I,, 1, had been found to be quite stable, being independent of the minera- logy, time for primary and secondary consolidation, rate of strain suitable for IDS test, previous creep behaviour, stress history and type of pore fluid. However, I, was found to be linearly related to the average normal effective stress! This confirmation was obtained for remolded and a few undisturbed clays as well as reassembled silts and sands. Thus I,, must have contained part of the shear strength due to curvature of the stress envelope. constant structure envelope Figure II-2 Constant Structure Envelope1 Based on the linear relationship between I, and Gi, a hypothetical Mohr envelope at constant structure was de- rived from the differential eguation a ~I5-B,0; Ste = Tete Smt (2-13) asp giving the solution as TEE To + asthBgayM (Sy) (2-14) where Ig = bond strength ag = soil structure parameter varying with strain,stress ratio, density, etc. 8, = soil constant varying seemingly with grain size only (value increas- ing with decreasing grain size) Tests to date (Schmertmann, 19662) indicate that gq varies between limits of 0.15 and 0.22. For cohesive soils, bond strength I, contributes to Im in addition to the effect of envelope curvature, It is also assumed that maximum bond strength occurs at Iq. Topshoj (1870) concluded that bond strength of remolded kaolinite does not differ signi- ficantly for both triaxial compression and extension con- stant-c} IDS tests. More detail considerations of the IDS Concept will be presented in Chapter III. Rowe's Stress Dilatancy Theory Rowe (1962) was not satisfied with the traditional approach to the problem of shear strength of soils involving the use of the Mohr-Coulomb Theory and the implications carried with it in respect to the manner of failure and slip12 planes. He considered Coulomb's ¢ which includes both the frictional and dilatancy components of strength cannot pro- perly define the behaviour of the material and came up with a theory that considers the effect of friction and dila~ tancy separately. The basis of his Stress Dilatancy Theory is derived from the study of ideal assemblies of rods and uniform spheres under deviatoric stress systems. He considers the medium as discrete particles rather than as a continuum and derives relationship for stress and strain in terms of pack- ing geometry, a,and 4, the physical angle of friction OR, between the particle surfaces. He assumed that the form of law which applies to any uniform packing is also applicable to a mixture of packings. By minimising the ratio of energy input to energy output in a triaxial test, Rowe obtained the stress dilatancy rela~ tion for cohesionless soils. ——— = tan? (45+9,,/2) (2-15) os (144) are the major and minor principal stresses respectively where dis the rate of change of volumetric strain to vertical strain For any state of packing other than very dense, the parti- cles undergo rearrangement prior to failure involving fur- ther energy expended. To fit the above equation, ¢, is replaced by ¢¢ where ¢¢ is greater than ¢,. The upper13 limit of @¢ is reached when at large strains, no volume change takes place and $5 = 4, The relation between ¢,, cv" $g and} is given by Rowe (196?) for medium fine sand. 38 Porosity, n(%) Figure II~3 ¢ Values as a Function of Porosity (After Rowe, 1962) In the derivation of his theory, Rowe [1262) made the following assumptions: i) Volume change due to ambient or mean principal stress change, associated with particle slides or elastic compression at contact point, is neglected. ii) Shearing resistance is given by sliding of parti- cles at their contact points and that rolling of particles is neglected. iii) The measured volume change applies to the whole volume of the sample before any formation of slip planes. Rowe (L962) also extended his theory to clays by intro- ducing the interparticle cohesion term, cg,resulting in the following expression:14 ca (14,) = 93 tan*(45+¢¢/2) + 2c¢g tan(45+b¢/2) (2-16) Rowe et al. (1963) found that the stress dilatancy per- formance of two clays, West water clay (12% cley fraction) and Derwent clay (50% clay fraction) was similer to that_— 9 ze of sands. The cg term was very small (1.7 psi) /even when the clay had been consolidated anisotropi¢ally and had allowed secondary consolidation to take place. on the practical application of his theory, Rowe (1963) introduced a method of earth pressure and slope stability calculation based on the effective interparticle friction angle,$g,and the degree of interlocking of the soil struc- ture, Gp. The improvements of this proposed method are i) It gives a better insight into the mechanism of shear strength of soils where the effect of dilatancy is taken directly into account and not taken as a correction to shear strength. i) The stress dilatancy equation is valid at any stage of deformation including the peak and residual states. For the special case of no volume change, the solution becomes identical to the well known solution based on Coulomb. Horne (1965a) in his series of papers further extended the stress dilatancy work of Rowe. He assumed that the assembly consists of rotund, rigid and cohesionless parti- cles and elastic, plastic deformation, crushing and crack- ing are ignored. Individual particles acting es 'rollers' between groups of particles do not occur significantly in a deforming assembly.a5 Horne (1965b) introduced the concept of mean-projected- solid-path (m.p.s.p.) to describe the changing state of anisotropy existing in a deforming assembly through the triaxial test. The degree of aniosotropy denoted by m,/m, is the ratio of the m.p.s.p. in the axial (m,) and radial (ms) directions. This ratio increases to a maximum close to the peak stress ratio and then decreases until the condi- tion of no further volume change causes the degree of anioso- tropy to become stabilised so that m,/m, = (1/4) tan(45+0gy/2) (2-17) where $9, = interparticle friction angle sheared at con- ev stant volume. Horne (1969), in his third paper of the series, ob- tained a theoretical prediction of ¢,, based on $,, and showed that his prediction is more accurate than the expres- sions of Caquot (1%34) and Bishop (154) given by tan boy = (1/2) tan o, (2-18) sin ¢cy = (25 tan $,,)/(10+3 tan 9,,) _ (2-19) respectively. Although Newland and Allely (1957) had proposed a dilatancy theory before Rowe, their theory had been rela~ tively ignored. King and Dickin (1970) stated the reasons for this unpopularity. i) Their theory was not based on rigorous analysis and their experimental results were not so reliable. ii)Their theory was intended to apply at peak stress ratio only while Rowe's theory is applicable at all stages of deformation up to formation of slip plane.16 King and Dickin extended Newland and Allely's theory to be applicable at all stress ratios and showed that the stress and volume change equations of the two theories are the same, their difference occuring in the final as- sumptions made in the interpretation of the results. Rowe postulated that for incremental deformation to occur, the ratio of the work done on the assembly by the najor princi- pal stress to that done on the minor principal stresses should be a minimum,while Newland and Allely adhered to the Mohr-Coulomb principal of failure. The theoretical relationships of the two parameters $, anda are obtained as follows: tan ¢¢ (Newland & Allely) _ 2 YD (2-20) tan $¢ (Rowe) D+1 t Newland & Allel; 1 an a_(Newlani ely) _ (o-2 tan a (Rowe) yD where D = (1 +2), Rowe (1971) pointed out that Newland and Allely's assumption of a = (45 + $/2) is invalid for soils which change in volume during shear. It is only applicable for stress paths in which \ = 0, Rowe (1952) showed that for medium dense sand his o> compared favourably with those of Newland and Allely at high porosities, their differences increasing with decreas- ing porosities. This was confirmed by King and Dickin (2979),wy Rowe et al. (1964) derived stress dilatancy equations for triaxial compression tests, triaxial extension tests and direct shear tests. ‘They showed theoretically the differences between Coulomb ¢, $, after Taylor and Bishop x energy correction and Rowe's $g. It was pointed out that oy is the equivalent Coulomb $ after deduction of the external work done by dilation and $, is obtained after further reduction of internal work done by dilation from dp: The external done in dilation is the energy spent in overcoming cell pressure while the internal work done in dilation is the energy required to rearrange the parti- cles during sliding friction. Barden and Khayatt (1966) further confirmed the validity of the Stress Dilatancy Theory by imposing different stress paths during triaxial compression and extension tests where mean stress was either increasing or decreasing. The stress ratio o}/o}, denoted R, increased during all the tests. El-Sohby (1964) showed that for paths of increasing R elastic (recoverable) strains are small compered with slip strains (irrecoverable) and therefore may be neglected. However, data are lacking for stress paths of decreasing R where the elastic deformation may possibly be as important as slip strains. Barden et_al-(1969) confirmed that the elastic com ponent of shear strain is negligible in the case of first loading and rebound shear paths. They showed that isolation of elastic from total strain is not possible using loading and unloading cycles because of marked changes in structural18 arrangement of particles. Employing constant R tests and cycling between chosen values of mean stress p', they at- tempted to separate total strain into elastic end slip com- ponents on the assumption that unloading defines the elas- tics. Lee (2965) put the Stress Dilatancy Theory on a firmer footing by extending the confirming experimental data to a material (feldspar) with an extremely high value of inter~ particle friction of 37°. Previous studies (Rowe 1992) had established the validity of the stress dilatancy equation for granular materials with 9, values varying from 7°(steel balls) to 26° (graded sand). Summarising, Rowe's Stress Dilatancy Theory has been confirmed for a wide range of cohesionless materials with friction angles of ¢, varying from 7° to 377. As for cohe~ sive materials, only two clays had been investigated (Rowe et_al.1963) besides the extruded kaolinite and glass beads with hydrocal used in this study. All the above work were triaxial compression tests with a few extension tests (Barden and Khayatt, 1966). Other Methods Other methods that may be used to separate shear resis- tance into components include a) The Critical state Concept b) Theory of Rate Process c) Stress Loci Method19 a) The Critical, State Concept Though the Critical State Concept invented by Roscoe et_al. (1955) was never intended to separate shear resis- tance into components, Rowe (1944) demonstrated the corre- lation of his ¢~ and ag parameters with Roscoe et al!s M and $pg values for cohesionless soils. Though there is some common meeting ground between the two theories, Rowe pointed out that the fundamental difference lies in his particulate mechanics approach and Roscoe's continuum ap- proach to the shear strength problem. b) Theory of Rate Process Mitchell (2264), using the theory of rate process, derived an expression for shear resistance of a soil as a function of the frictional (includes dilation) and cohe- sive properties, effective stress, soil structure, rate of strain and temperature at any stage of deformation. He assuued solid particle to particle contacts. This approach, however, requires much improvement and modification before soil behaviour can be completely under- stood within the framework of the theory because of the many assumptions and speculations involved. Nevertheless, its potential appears great. ¢) Stress Loci Method Yong and Vey (i5%2) presented a graphical method for the determination of effective stress parameters by stress20 loci. ‘They claimed that the method is applicable to normally consolidated as well as sensitive clays. Definition of Parameters Soil Structure The concept of soil 'structure' or the arrangement of the soil particles was put forward by Terzaghi in 1925 to explain the great variations of void ratio and behaviour of soils. Casagrande (199%) proposed that undisturbed sensi- tive clays have flocculated structures and a compound honey~ comb structure for silt grains. Present-day terminology of soil structure is taken to include the oriertation and distribution of particles as well as the electrochemical forces between adjacent soil particles. (Lambe and Whitman, 1969) The particulate nature of soils distinguishes itself from that of solid mechanics and fluid mechanics in that the soil particles are not strongly bonded together in the way that the crystals of a metal are and are therefore relative~ ly free to move with respect to one another. Thus the re- sistance to shear deformation of a soil mass is controlled by the interactions between adjacent particles, or soil structure. The sliding and probably rolling of adjacent particles results in deformation and therefore soil struc- ture varies with strain, besides other factors as effec- tive stress, void ratio, etc. The development of Schmertmann's IDS Concept on curve-21 hopping was based on the assumption that two effective stress conditions can be imposed on a soil at approximately con- stant structure and at the same strain and thereby per- nit separation of the shear resistance into I, and D, com- ponents. The complex nature of the arrangement of parti- cles has defied many researchers to express soil structure quantitatively. Rowe (1942) defined tan ap as "the ratio of sliding contacts of an assembly in the principal directions." Schmertmann (1966a) called a, in his ‘constant structure envelope’ a structural parameter but gave no physical ex~ planation of its meaning. Mitchell (1944) defined S as a "structural factor represented by the number of interparti- cle contacts per unit cross-sectional area." Horne (1955) introduced the concept of mean-projected-soil-path and pro- posed the ratio m,/m,(see page 15) as a measure of the aniso- tropy of an assembly of particles. This ratio is equiva~ lent to Rowe's tan ap. Bond Strength It is well recognised that a portion of the shear re~ sistance of a cohesive soil is due to cohesion or 'sticking together’ of adjacent particles without externally applied normal pressure. The nature of such bonding has been the subject of considerable research, supplemented by extensive investigation in the fields of colloid science (Kruyt, i992; Rosenquist, 1952). It is agreed that bonding between22 particles may be the result of van der Waals’ forces of at~ traction, Coulombic forces of electrostatic attraction and repulsion and in some cases by hydrogen or potassium bonds or from cementation by organic or inorganic compounds. Though formulas showing variation of attraction and repulsive forces with distance and system characteristics for the simplified case of parallel platelets have been developed by colloid chemists (Kruyt, 1952), these are of limited quantitative value to the soils engineer who has to deal with random arrangements of irregular particles with variable spacings. As yet there is still no reliable ex- pression for determining the bond strength of a cohesive soil. Lambe (1,960) presented a theoretical consideration of the mechanism of shear resistance in cohesive soils with an equation relating the various pressures acting across a shear surface between adjacent soil particles. This equa~ tion is helpful in understanding qualitatively soil volume change and shear strength characteristics because,at the present state of knowledge, several of the terms cannot be evaluated. Since Hvorslev's work, there has been some speculation that ce and $— may be descriptive of the actual internal mechanisms of shear resistance and cg was thought to be an indication of the magnitude of the bonding between the par- ticles. Schmertmann ii943c) introduced a method to determine bond strength from the zero intercept of a I, vs Gj plot.23 He defined bond strength as the shear strength at zero ef- fective stress on the plane of envelope tangency, at the same structure. Schmertmann (19¢6b) showed that cg con- sists of the effect of constant structure envelope in ad- dition to bond strength. The constants « in equation (2-6) and 8, in equation (2-14) are both believed to be a measure of envelope curvature. Rowe's interparticle cohe- sion,c;, may be thought of as shear strength at zero ef- fective stress or biud strength with dilatancy effect re- mewed. Christensen and Wu (1964) defined bond strength as the force required to produce relative displacement between two particles at a contact point. Contact refers to a point at which a bond is formed between two particles and does not necessarily mean that the particles are in solid- to-solid contact. ‘The mechanism of shear resistance was pictured as slippage occurs first at contacts with weak bonds under a given stress. The breaking of these weak bonds transfers load to other existing stronger bonds or new bonds formed as a result of strain. This concept of shear deformation and creep is similar to that propounded by Bowe (1962). Applying the Rate Process Theory, Christensen and Wu (2964) postulated that the activation energy required to move a particular unit from one minimum energy position to the next minimum energy position represents the bond strength of the contact. This bond strength would include the) 24 dilatant component as well as the adhesion and friction at the contacts. As expected, they found that bond strength of remolded specimens is lower than undisturbed ones and that the low initial bond strength of remolded clays in- crease with deformation as a result of stronger bonds. Mitchell et al. (i909), in their study of stress- strain-time response of soils, also made use of the acti- vation energy in Rate Process Theory to determine the num- ber of interparticle bonds under a variety of conditions. However, they assumed solid-to-solid contact between par- ticles with different number of bonds per contact and the individual bonds are of equal strength. These assumptions differ from those of Christensen and wu. “Mitchell st al. found that the number of bonds per square cm. is linearly proportional to the effective consol- idation pressure. They also pustulated that the nunber of souree of true volhwsion. They showed that strength and the number of bonds are linearly related irrespective of whether the material is undisturbed or remolded, and normally con- solidated or overconsolidated San Francisco Bay mud, wet or dry illite and dry medium fine sand. The experimental evi- dence cannot be said to prove the validity of the theory which still has to stand the test of time because of the many assumptions involved. In summary, the definitions of bond strength given by Mitchell et al. and Christensen and Wu are similar to25 Schmertmann's definition in that it includes the effect of dilatancy. Rowe's definition of interparticle cohesion, cg, differs from the above in that it is free of dilatancy effect. Frictional Strength The effective frictional component is generally de- fined as the remaining part of shear resistance, depen- dent upon the magnitude of effective stress. Associated with this component is the angle of friction expressed as the ratio of shear stress to normal effective stress. Coulomb's $ angle is the slope of the envelope which compares samples at different void ratios and effective stresses at failure, It had been known that c and $ vary with the types of shear tests. Following Hvorslev's work, the friction angle ¢, was thought to be representative of the true mechanism of friction of soils at failure but it was shown by Gibson (195%) and later by Rowe (1963) to con- tain the effect of dilatancy. Schmertmann and Osterberg (1960) in developing the IDS technique of separating shear resistance into components as a function of strain defined tan $, as : “The angle of internal friction,#,, at any strain, is the angle whose tangent is the ratio of the change in shear stress to change in normal intergranular (effective) stress occuring on the plane of Mohr envelope tangency at that strain, dur- ing a stress change occuring without significant change in26 soil structure." This slope is now seen to be slope of imaginery constant structure envelope (See Figure II-2), tan ¢, = 4,. $,, therefore, does not take into account the effect of dilatancy. Rowe (62) introduced the following three friction angles (Figure II-3) in his Stress Dilatancy Theory: %, is the physical angle of friction between minerals, o¢ is 6, plus the effect of rearrangement of particles or interlocking but with dilatancy effect removed, Soy 48 6, Plus the effect of particle rearrangement for samples sheared at constant volume. Rowe (1964) suggested a friction angle $,, which resulted from the ‘boundary energy correction' proposed by Roscoe ) etal (1958). The following table summarises the various definitions of the angle of friction:27 Definitions | Corrected for | not Corrected| Remarks Coulomb ¢ none £,e,S, at failure dilatancy Bishop 9, |e, dilatancy | e,S any strain (external) dilatancy (internal) Hvorsiev , e =ySy at failure dilatancy Schmextmann Se e,dilatancy | any strain dentan™' (a) Rowe ¢ e,dilatancy | S,e any strain (internal & external) | Roscoe §,, | €,dilatancy | S,dilatancy | any strain (external) (internal) Table II-1 Definitions of $ Angles To date, the IDS Concept and the Stress Dilatancy Theory are the only two available laboratory methods of determining bond strength. Since both methods were developed to measure the fundamental components of shear resistance as a function of strain, there should exist a theoretical relationship between the two sets of para- meters. To achieve this goal, one must examine in great~ er detail the underlying assumptions and considerations of the two theories.CHAPTER III ‘THEORETICAL CONSIDERATIONS OF SCHMERTMANN AND ROWE THEORIES Schmertmann's IDS Concept ‘The IDS Concept is based on the fact that by keeping all the variables listed in expression (2-10) constant ex- cept effective stress o', then at a certain strain, the shear strength will be a function of effective stress at that particular soil structure. The change between two levels of effective stress is so chosen that the soil struc- ture is not significantly altered and can be considered con- stant at that strain. The I, and D, components are dependent on the magni- tude of the jump selected, the larger the jump the smaller will be the computed I, value (Figure III-1). A study of the effect of the magnitude of the jump for constant-¢' IDS tests on computed I, values was made by Schmertmann (1966a) who considered that the high level of o! was fixed and the jump was then decreased to a lower level of gi. Con- sidering the jump of Ag {=20% oh as the standard, he compared I, values for extruded kaolinite, Enid clay and sandy kao- linite with curve-hopping ratio of ao}/o!, varying from 10% to 60% resulting in almost identical theoretical and exper- imental (I,/I,20%) ratios varying from 1.1 to 0.6 respec- tively. He confirmed the use of the curve-hopping ratio 2829 Figure III-1 Constant Structure Envelope Show~ ing Fitted Secant and Tangent to Obtain I, Components30 of 208 in the constant-c} IDS test as a compromise between conflicting requirements. A small increment will minimise the effect of soil structural changes and also decrease the difference between the tangent and the secant slope of the constant structure envelope resulting in more accurate separation of the strength components. A larger increment will enable the line of common tangency to be drawn with better precision. The same curve-hopping ratio of 20% was also recommended for constant-c} or constant pore pressure versions of IDS tests. For constant-volume IDS tests, an increase in volume of 0.25 @ was employed to obtain the lower stress-strain curve. For the kaolinite, this will result in a drop of of varying from about 18% to 10% for strains ranging from 1% to 5%. The error involved in the computation of the T, component with the assumption that the fitted secant is the correct tangent to the envelope, is about 1.5% too low when compared to the theoretical con- stant structure envelope with I, = 0. (Figure ITI-1) The IDS test is neither a drained nor undrained test in the conventional sense. Because of the imposed stress conditions, the sample is not completely free draining and volume changes occur with changes between two levels of effective stress. The validity of the one-sample curve~ hopping technique had been shown for nine soils, seven be~ ing remolded and two undisturbed (Schmertmann, 1962). The applicability of the one-specimen test eliminates one of the fundamental variables inherent in testing a natural soil deposit.31 IX7ZA\ 2 2 ‘h Figure III-2 separation of shear Resistance into I, and D, Components at Constant structure and Strain From the geometry of the two Mohr circles at the same strain, the I_ and D_ components as well as the average nor- e e mal stress at the plane of tangency, 3; are conputed from the following: (6,-0,)_ ~ (61-05) $e = sin (3-1) 2olnma{ ,)~E (4-05) 4 (0,-04) 9} a u% [tesyter dating) ¢ feuding oreo) | t (3-2) De = Fy tan o, (3-3) Ip = TenDe (3-4) where subscripts h and ! denote high and low effective stress conditions. Subscripts 1 and 3 denote major and minor principal stresses respectively. Schmertmann (1966a) demonstrated the stability of the32 value which, however, is found to be a function of the magnitude of effective stress. The linear relationship between I, and effective stresses,existing even for sands, in- m dicates that I, must include some other component besides the bond strength between particles, The introduction of the concept of the imaginary constant-structure envelope helps to explain this paradox. The curvature of the con- stant-structure envelope results in a contribution to the Tg component and it was found that in the remolded clays and sands investigated (Schmertmann, 1966a), almost all the I, values result from envelope curvature. In, besides being effective stress dependent, was also found to be affected by the size of the soil particles, Im decreasing with increasing average grain size defined by D,,. Schmert~ nann's work included tests in assorted glass beads. He proposed that it is the radius of particles at their contact points rather than the particle size that is significant. 036 kaolinite 8 3 = FE +0002 +001 01 el 1.0 Ds,(mm) - Log Scale = (After_Schmertmann 1966a) Figure III-3 I, vs Ds, Grain Size at Constant Effective Stress33 Rowe's Stress Dilatancy Theory Terzaghi (1920) wrote about the fallacy of Coulomb's equation, "The fundamental assumptions of the traditional earth pressure theories cannot, in fact, stand even super- ficial examination. The fundamental error was introduced by Coulomb, who purposely ignored the fact that sand con~ sists of individual grains and who dealt with the sand as if it were a homogeneous mass with certain mechanical prop- erties.. Coulomb's idea proved very useful as a working hypothesis for the solution of one special problem of the earth pressure theory, but it developed into an obstacle against further progress as soon as its hypothetical charac- ter come to be forgotten by Coulomb's successors. “The way out of the difficulty lies in dropping the old fundamental principles and starting again from the elemen- tary fact that the sand consists of individual grains." Rowe (1962) indeed made a fresh start by considering Ule particulate nature of the particles. starting with two particles in limiting sliding equilibrium, he derived a stress dilatancy relation between stress ratio,o{/o}, the physical angle of friction between minerals ,$,, the ratio of changes in volumetric strain to vertical strain, dey/de, and the angle of interlocking, ag.34 Instantaneous. (Gy Slip, ~Z Resultant Figure III-4 Two Cohesionless Particles at Instant of Slip The two particles having a physical angle of friction bys slide at an angle of 8 to the vertical under the action of forces P and Q acting in the vertical and horizontal dir- ections respectively. From geometry, P/Q = tan(B+d,,) (3-5) dx/dy = tan 8 (3-6) therefore (P ay) /(Q dx), = tan(8+¢,)/tan B= EQ (3-7) where Ep is called the energy ratio. In a triaxial compression test, summing forces verti- cally and horizontally for a random assembly of particles having n sliding contacts, mh Pldy) _ , ode, _ tan(B+dy) _ y 2D Q(ax) 20 4de 5 tan B R (3-8) where all contacts are at minimum energy condition. For35 a given energy input, the minimum energy ratio is given by aE, /aB= 0 whence B = (45 - $y, /2) + EL = tan? a5 + 4,/2) (3-9) 2osde For small strains, tac = -de, + 2dey (3-10) ' de “te at SY) tan (4544/2) @-1) oy oe, This is the stress dilatancy equation for a cohesion- less soil deforming under the minimum energy condition. The sign convention adopted is positive for volume increase and positive for vertical compression. The above equation gives the lower limit of interpar- ticle friction angle, ¢,,when all contacts are sliding at critical angle 8. This condition occurs for very dense pack- ing of particles. For densities other than very dense, $,, is replaced by ¢, in equation (3-11) and at large strains when the sample is shearing at constant volume, ¢, ie re- placed by ¢, where ¢¢ lies between the lower and upper limits of %, and ¢,, respectively. Whatever $_ may be, the defor~ mation is always conceived of as occuring such that the ratio of energy absorbed in friction to energy supplied is a mini- nun. Rowe (1962) introduced the op parameter to describe the soil structure or geometry of packing of the particles. He defined ag as the ratio of the number of contacts per unit area in the principal directions. Referring to Figure III-5,36 Le Q Sa yi b ta ag| oR al b) Figure III-5 Stress Summation over Gp Plane in Triaxial Sample oib P/Q = —*"_ = tan(8+4,) (3-12) ojbtana, i= tanaptan (45+ ¢/2) (3-13) % Combining with equation (3-11) ac, tan ap = (1+ —¥ ) tan (45+¢¢/2) G-14) ae. : The ¢, parameter is a variable quantity and depends on the magnitude of the stress and stress system because the geometry of packing is altered with stress and strain. ‘the a, lines are not slip planes although there is no great harm in considering them as a kind of slip line. It should be noted that 4, can be equated to (4546/2) only if $, = ¢ at the critical void ratio state which is yy.37 Rowe et al. (1963) extended the Stress Dilatancy Theory to include cohesive soils by assuming that shear force be- tween points of contact of adjacent particles follows the law T = c,A +N tan 6, (3-15) where A average area of contact N = normal load of contact Cer, = interparticle cohesion and friction. Introducing the interparticle cohesion into the two sliding cohesionless particles in Figure III-4 and resol~- ving forces, B44, 2 Q + cyy tan Slip Plane. ae Resultant b) Figure III-6 ‘wo Cohesive Particles at Instant of Slip oly sec? 8 P/Q = tan(d,+8 + (3-16) Q(l-tan $, tan 8) ax/dy = tan 8 (3-6) dividing, Pay _tan(gyts) | cuy seo? 8 dx tan Bg Q tan g(l-tang, tang) (3-17) ty38 Again, if there are n particle contacts, zp PAdy)_| ofde, = band, +8) zr} Q(dx) 2a$de, tan B a a (3-18) ob tan agsin 8 cos B(1-tan $, tan 8) 2 ® u Minimising Eg yields 3 = (45-$,/2) which together with the small strain equation (3-10) substituted in equation (3-18) gives o ——1— = 33 tan? (454+6),/2) + 2c, tan (4549/2) ey (3-19) which is the stress dilatancy relation for a cohesive soil. structural 4, parameter for a cohesive soil is given by (3-14). Experimental Plots For cohesionless soils, the stress dilatancy relation is usually displayed with the stress ratio, R as the ordinate de, and (1+ ae ) = Das the abscissa with the slope of the 1 graph represented by tan*(45+$,/2) = Kpe-39 Figure III-7 Stress Dilatancy Rela- tions for Loose and Dense Sand For dense sands at low strains, the stress dilatancy relationship follows the minimum energy line when $4 1 but as strain increases resulting in remolding or rearrange- ment of the particles, ¢, increases until the critical state when there is no volume change, 9 9= boy: For a loose sand, ¢¢ will always be greater than ¢, and it increases as strain increases until at the peak stress ratio state, demboy- ‘The difference between oe and Coulomb's ¢ is attributed to the work done in overcoming the cell pressure as well as the rearrangement of the particles in dilation. In the separation of the interparticle components cg and $¢ Of a cohesive soil as a function of strain, it is conven- ient to plot stress paths with (oJ/(1+,)+0}) as abscissa and {o}/(4)-o$} as the ordinate. The parameters are then com- puted from (see Figure ITI-8,b)40 sin $, = tan y (3-20) cg = (a/2)sechs (3-21) For cohesionless soils, the stress dilatancy theory has been substantially verified for a wide spectrun of materi- als ranging from steel balls ($, = 9°), 0.25 mm diameter glass ballontini (¢, = 17°), sands (oy, = 23° and 26°), silt ($, = 31.5°) to feldspar (4, = 37°). On the other hand, only two clays had been tested and they were found to behave similarly as sands with zero or very small value of cg. Definition of Strain Paths The meanings of strain paths for the two methods are explained as follows: Stress Dilatancy Method 1.0 €, (8) Figure III-8 Definition of Strain Path by Stress Dilatancy Method41 Figure III-8 a) shows the stress-strain curves of a series of three triaxial tests conducted at different cell pressures. At a vertical strain of, say 1%, the principal stresses of the three tests were computed and the results a3) plot were then presented ona G2l-c 4) vs @ (Figure III-8 b) ). The line joining the three stress points at the same strain is then known as the strain path for the Stress Dilatancy Method. IDS _Methoa b) - Figure III-9 Definition of Strain Path by IDS Method42 In the IDS method, a series of three IDS tests were per- formed at different confining pressures (Figure III-9 a)). The I, components are then determined at the same vertical strain, say 1t, for the three tests and the values plotted against the corresponding Gj. The line joining these (1g,5{) points at the same strain is then known as the strain path for the IDS Method. Need for Establishing Linearity Before the relationship between the two theories can be established, it is necessary to prove that the ‘strain paths’ are linear for both methods of determining bond strength. This is because that the Stress Dilatancy Theory is derived on the assumption of a linear stress law between the particle contacts (equation (3-15)) while the IDS Con- cept does not assume any such linear stress-strain behaviour. Secondly, linearity of strain paths in both methods will simplify the relationships between the two sets of para~ meters.CHAPTER IV TEST METHOD, EQUIPMENT AND EXPERIMENTAL PROGRAM . Test Method In order to extablish a link-up between the parameters of the IDS and Stress Dilatancy Methods, it is convenient if a method of test can be developed so that both sets of para- meters can be computed from the experimental data obtained from the same specimen. Most of IDS tests performed to date on remolded and undisturbed cohesive and cohesionless soils have been the ‘standard’ constant-o' version IDS compression tests and it was established that the one-specimen curve-hopping tech- nique gives essentially two separate stress-strain curves at different effective stress levels. In the derivation of the Stress Dilatancy Theory, it was assumed that interparticle slips are caused by increas- ing stress ratios, R, where total strains are considered to be slip strains with negligible elastic strains. Most investigations made to date to verify the Stress Dilatancy Theory had been based on drained tests (Rowe, 1962; Rowe et _al., 1963; Lee, 1966). Barden and Khayatt (1966) em- ployed loading and unloading stress paths in compression and extension tests on a fine sand to investigate the condi- tions under which the upper ¢¢- (gy) and lower $¢,(9,) 4344 ~ values are reached. The types of IDS tests made in this investigation con- sist of constant-c!, oj, p', and constant-volune tests which yield increasing R values. All the tests result in increas- ing p' while constant-c} tests give decreasing p' with strain. It is thus postulated that the data obtained from IDS tests with volume change measurements can be used to compute Rowe's interparticle parameters cp and ¢¢ as well as the I, and D, components. To date it has not been claimed that the cy and $¢ parameters are unique nor has the effect of stress paths on these parameters been investigated. I, and D, Components with Dilatancy Corrections Using the IDS tests with volume change measurements, the IDS components corrected for dilatancy can be calculated from the following equations: og = sin? (4-1) of a (ga) > (aa), * eine? h & pity (3) + (=) — Gesin 4p) (2) t 1+ h 1th ls, 2 1+, + 2) - oh, = 33 a (ce ® ‘ (4-2)45 (4-3) If = ri - De (4-4) where the superscript * denotes values after correction for dilatancy. It should be noted that the maximum shear stress of each circle falls on a 45° line from its uncorrected po- sition after dilatancy correction. Figure IV-1 Separation of Shear Resistance at Con- tant Structure and Strain with Corrce- tion for Dilatancy When the rates of volumetric strain to vertical strain for both the 'high' and 'low' Mohr circles are equal, i.e. =X, then it can be shown analytically that (4-5) t (4-6) where v= Sv46 Equipment The 'Geonor' triaxial equipment used in this research was developed by Norwegian Geotechnical Institute and has been described by Andresen and Simons (1960). The follow- ing modifications were made to the triaxial equipment to give the necessary precision required for this work. Vertical Load Measurement Strain gage load cells are used to measure the piston load instead of the conventional proving rings for quicker response in IDS work. For cohesive soils, a 100-1b capacity Thwing-Albert load cell with a sensitivity of 0.0075301 Kg/microinch/inch was used, while for cohesionless soils, a 200-1b capacity load cell with a sensitivity of .0151515kg/ nmicroinch/inch was employed. These load cells deflect much less than the proving rings although this factor would not have affected the accuracy of the vertical defcrmation measure~ ibed ments because of the desiyn of the system which is des in the following paragraph. Vertical Deformation To eliminate the effects of cell expansion or con- traction from the variation in cell pressure as well as the deformation of the load cell, the vertical deformation of the specimen is measured by the relative movement between the top of the piston and the bottom of the cell. A dial gage with an accuracy of 0,002 mm per division deflectiona7 squewernsesy e6ueyD OUNTOA UGTA 389% SdI x03 snqezeddy TeyxeTzL 30 wezbeTa oTJeusyS Z-AT omnbta WAISKS WALSAS WELSAS MOLYOIGNI SONWHO BWOTOA munssi@ud wove ‘TOULNOD TUNSSHIa TIAO Azeqtydeo sseqto ‘a°r wu T soyescuey | J qteg SzROzOH eanssezg ue}SsU0D Torq0D KOZOS joa il kznoton i dousta Teta uoTyeutozaq Teorszen, zoqeoTDur a 1190 uyez3s |Q_" 0 peor48 is used. With this set-up, corrections that still have to be made are bedding of the ball joint, compression of the top cap, grease and rubber discs or frictionless platens. Back Pressure Saturation To ensure the sample is 100% saturated before and during the shearing process, back pressure saturation technique was employed. This back pressure is supplied by the Bishop self- compensating mercury pot system which has the advantage over the Norwegian constant pressure system of no piston friction. The mercury pot has an accuracy of 1/15 psi pressure though the Bourdon gage can only be estimated to 1/4 psi. Lowe and Johnson (1960) gave an expression for calcu- lating the back pressure required to achieve any degree of saturation by increasing the pore pressure under the condi- tions of constant overall soil volume and increasing water content. 1-.98: ou = qhtgf 9885 ) (en . 1- 98s, where s, = initial degree of saturation Sp = degree of saturation desired The following table shows the back pressure in kg/cm* required to achieve 100 and 99% saturation from an initial saturation of s,.49 Si Aur 00 Aus s 100 0 - 98 1.05 6.35 95 2.53 1.34 90 5.06 3.09 88 6.05 3.73 Table IV-1 Required Back Pressure For Achieving 100% and 99% Saturation All the kaolinite samples had an initial saturation of over 98% and most of them are close to 100% after consol- idation. All kaolinite samples tested at an effective cell pressure of less than 4.0 kg/cm? had a back pressure of 4.0 kg/em*. Samples tested at 4.0 kg/cm? effective cell pressure had a back pressure of 2.9 ke/em*. Even this back pressure is hiyh enough to ensure total saturation of the kaolinite samples. Since the glass beads were boiled and then placed under water and vibrated in a mold, the possibility of air trap- ped in the sample is very low. However, a back pressure of 2.0 kg/om? was used further to ensure the glass beads were totally saturated. The glass beads with hydrocal samples were installed dry in the triaxial cell and then saturated by vacuum to 88% saturation. A back pressure of 4.0 kg/cm? was used50 producing a saturation of 99.2%. A higher back pressure was not possible because the back pressure plus the maxi- mum consolidating pressure of 3.0 kg/cm’ give a total cell pressure of 7.0 kg/cm? which is the design capacity of the Geonor triaxial cell. ‘The back pressure was applied to all the samples after consolidation and then allowed to come to equilibrium (usually overnight) before the start of the shearing pro- cess. Volume Change Measurements The measurement of volume change is as important as the vertical deformation readings because of the incremental strain ratios used in the Stress Dilatancy Theory. The accuracy of the volume change measurements have to be com- patible with the vertical deformation readings and at the same time, the system has to have the capacity to measure the large volume changes that take place. The final design consists of a 46-in.-long glass capil- lary with an internal diameter of 1.0 mm connected to two pairs of Circle Seal valves which, by closing one pair and opening the other pair, reverse the movement of the air bubble in the glass capillary without changing the move- ment of the water into or out of the sample. A bubble dis~ placement of one inch is equivalent to a volume change of 0.0238 cc and the movement can be read to the nearest 0.05 inch. For a sample having a height of 8,00 cm and a volume51 of 80.00 cc, the ratio of the smallest units of measurable volumetric strain to vertical strain equals volumetric strain = (0255205) ) 80 = 1.19 vertical strain The above ratio indicates that both strains are deter- minable with similar precision. The volume change of a sample during the shearing pro- cess is measured by the flow of water into and out of the sample, while the pore water pressure is kept constant. The importance of keeping a constant pore water pressure throughout the test is to ensure the amount of dissolved air and the volume of air bubbles, if any, to be constant. The reader will see that extreme precision is required to determine volumetric strain with the accuracy required for the Stress Dilatancy calculations. Sample Preparation and Testing Three types of materials were used: Kaolinite Glass beads (no. 140 to 200 sizes) Glass beads with hydrocal (10:1 ratio by weight)52 Kaolinite Samples This pure commercial clay and its characteristics have been described by Topshoj (1970). It consists of over 99 8 of kaolinite clay minerals and has a specific gra- vity of 2.609. The kaolinite powder was first mixed with distilled water to a predetermined water content of about 40 per cent and then cured overnight. Tt was then passed through an ‘vac-Aire' extruder five times to distribute the water evenly before it was cut to lengths of 10 om, wrapped in wax paper an@ dipped into warm wax several times to prevent evapora~ tion. The samples were numbered in ascending order as they came out of the machine. Water content samples were taken every five samples to check the uniformity of the batch. Each batch produces 50 to 55 samples and three batches were made in 1969, All samples have three wool drains drawn through pre- punched holes in a triangular pattern with a 0.18 cm needle. Three O-rings were used both around the top cap and the ped- estal to prevent cell water seeping into the sample. Pre- vious tests (Schmertmann 1960) have indicated that as long as the duration of the cell pressure does not exceed two days, the effect of moisture migration is relatively minor. The sample is then hydrostatically.amsolidated at pre~ selected pressures for about a day, although the time for the end of primary consolidation, tio; is generally reached between 50 to 110 minutes for cell pressures between 1.0 and53 4.0 kg/cm2, using the Casagrande graphical method to deter~ mine t,,,- The advantage of using an open burette during consolidation is to allow air bubbles trapped between the rubber membranes and the sample to escape. After consoli-~ dation is complete, a back pressure is applied through the volume change system and allowed to come to equilibrium before the subsequent compression phase of the test. The following three types of IDS tests were then performed on normally consolidated and overconsolidated kaolinite. Constant-o{ series To perform a constant-of IDS test, one can control the pore water pressure with a constant cell pressure cx con- trol the cell pressure with a constant pore pressure. The latter method was chosen because of the necessity of having a constant back pressure for reliable volume change measure- ments. However, this variation of cell pressure results in expansion and contraction of the triaxial cell and also variation of the uplift on the piston load. ‘The system for obtaining vertical deformation has been designed to nullity this effect by measuring the relative movement between the piston and the immovable cell base. ‘The variation in up- lift is accounted for in.the following formula for compu~ tation of deviator stres. (ey05)q = SRTRaI =O ap (4-2) As54 where = initial cell pressure 944 = cell pressure at strain i Ap = area of piston Ag = area of sample at strain i Ry = initial load cell reading = load cell reading at strain i Constant-p' series The method of testing for this series is similar to the constant-3, series by varying the cell pressure with con- stant back pressure. This series was suggested by one of Rowe's assumptions that the volume change due to mean stress change is neglected. Constant-volume series The correlation between the two methods of component separation depends mainly on 2, the rate of volumetric strain to vertical strain. When no volume change takes place, the term becomes zero and the Ig and De components vemain unchanged. . The three types of tests involved different dilatancy rates during the shearing process and hence different energy input into the soil sample. The constant volume tests re- quire the highest cell pressure of the three types of tests to maintain the sample volume constant. All three test types employ a back pragsure of 4.0 kg/cm? except samples tested at a cell pressure of 4.0 kg/cm? when the back