UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Interpretation of selfboring pressuremeter tests in sand Cunha, Renato Pinto da 1994

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_1994-953292.pdf [ 8.19MB ]
Metadata
JSON: 831-1.0050418.json
JSON-LD: 831-1.0050418-ld.json
RDF/XML (Pretty): 831-1.0050418-rdf.xml
RDF/JSON: 831-1.0050418-rdf.json
Turtle: 831-1.0050418-turtle.txt
N-Triples: 831-1.0050418-rdf-ntriples.txt
Original Record: 831-1.0050418-source.json
Full Text
831-1.0050418-fulltext.txt
Citation
831-1.0050418.ris

Full Text

iNTERPRETATION OF SELFBORING PRESSUREMETER TESTS iN SANDBYRENATO PINTO DA CUNHABSc., Universidade Federal do Rio de Janeiro, 1986M.Sc., Universidade Federal do Rio de Janeiro, 1988A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFDOCTOR OF PHILOSOPHYinTHE FACULTY OF GRADUATE STUDIESDepartment of Civil EngineeringWe accept this thesis as conformingto the required standardTHE TJNWERSITY OF BRITISH COLUMBIAOctober 1994©Renato Pinto da Cunha, 1994In presenting this thesis in partial fulfillment of therequirements for an advanced degree at the University of BritishColumbia, I agree that the Library shall make it freely availablefor reference and study. I further agree that permission forextensive copying of this thesis for scholarly purposes may begranted by the head of my department or by his or herrepresentatives. It is understood that copying or publication ofthis thesis for financial gain shall not be allowed without mywritten permission.Department of C(tIL.The University of British ColumbiaVancouver, CanadaDate it 1&1UABSTRACTThis thesis addresses the analytical interpretation of selfboring pressuremeter testing curves insands. Emphasis is placed on the development of a new approach to analyze the data and hence to derivereliable predictions of the basic soil parameters, namely the friction angle, the lateral stress and the shearmodulus.The new methodology of interpretation relies on a “curve fitting” technique to match theexperimental and idealized (model) curves, from which a set of fundamental soil parameters are derived.These parameters are linked to each other in the framework of the cavity expansion model adopted.Some of the elasto-plastic models currently available are adopted for use under the new methodology ofinterpretation. A new model that extends the rheological equations of Hughes et al, 1977 is alsodeveloped. Pressuremeter tests under controlled conditions are analyzed in order to verify the basicassumptions of the chosen models. Some of the best calibration chamber data from the University ofCambridge (Fahey, 1986) and from the Italian ENEL-CRIS laboratory (Bellotti et a!, 1987) are used forthis purpose. Once the reliability of the chosen models is established, the new methodology ofinterpretation is applied to field pressuremeter data. Several high quality tests carried out by the writer ina granular site in Vancouver, Canada are analyzed. The results of both field and chamber tests confirmthe reliability of the new interpretation approach proposed here.The new interpretation approach also provides the engineer with a technique to numerically quantifythe disturbance of the testing curve. Using the new disturbance criterion ranges for “undisturbed”,“disturbed” and “highly disturbed” testing curves are proposed. This criterion aided in the establishmentof the insertion procedure of the UBC selfboring pressuremeter, allowing optimization of the insertiontechnique and minimization of soil disturbance during selfboring.It is believed that the contribution given in this thesis aids pressuremeter practitioners to designmore economical engineering works based on reliable soil parameters derived from the selfboringpressuremeter test. Simplicity and reliability are the essential features of the proposed methodologies ofinsertion, testing and interpretation described herein.111TABLE OF CONTENTSABSTRACT.iTABLE OF CONTENTSLIST OF TABLESLIST OF FIGURES vilLIST OF SYMBOLSACKNOWLEDGMENTS xiiDEDICATIONCHAPTER 1.0 INTRODUCTION 1CHAPTER 2.0 ANALYSIS OF TESTING CURVES IN SANDS 52.1 1NTRODUCTION 52.2 CONSTITUT1VE MODELS FOR PRESSUREMETER ANALYSIS 52.2.1 General 52.2.2 Basic Considerations of an Elastic Model 92.2.3 Basic Considerations of an Elasto-Plastic Model without Volume Change 122.2.4Unloathng 152.2.5 Review of Elasto-Plastic Models with Volume Change 172.2.6 New Cavity Expansion Model 292.2.6.1 Introduction 292.2.6.2 Basis of the New Model 302.2.6.3 Derivation ofElastic Strains in the Plastic Zone 302.2.7 Comparison Between Cavity Expansion Models 402.3 TRADITIONAL INTERPRETATION METHODOLOGIES iN SAND 442.3.1 Introduction 442.3.2 Friction Angle 452.3.3 Horizontal Stress 482.3.4 Shear Modulus 522.3.4.1 Unload Reload Shear Modulus 522.3.4.2 Initial Shear Modulus 652.3.5 Conclusions of Section 2.3 692.4 NEW INTERPRETATION METHODOLOGY FOR PRESSUREMETERS 702.4.1 Basis of the Curve Fitting Technique 702.4.2 Link Between Gi and Gur 752.4.3 Modulus Reduction Curve 802.5 VERIFICATION OF CONSTITUTWE MODELS WITH CHAMBER DATA 832.5.1 Tests with Leighton Buzzard Sand 832.5.2 End Effects 942.5.3 Tests with Ticino Sand 99iv2.6 PARTICULAR ASPECTS OF THE FITTiNG TECHNIQUE iN SANDS.1082.6.1 Strain Range of Curve Match 1082.6.1.1 High Quality Testing Curves 1082.6.1.2 Disturbed Testing Curves 1202.6.2 Sensitivity Analysis 1242.7 SUMIvIARY AND CONCLUSIONS 131CHAPTER 3.0 INSERTION AND TESTING PROCEDURE FOR THE UBC SBPM ... 1373.1 INTRODUCTION 1373.2 FIELD TESTING PROGRAMME 1373.3 UBC SBPM EQUIPMENT CHARACTERISTICS 1423.3.1 Steel Lantern Characteristics 1433.3.2 Effects of the Lanterns on Gut 1473.3.3 Method of Installation 1493.4 REVIEW OF INSERTION PROCEDURES 1513.4.1 Key Insertion Parameters in Clays 1513.4.2 Key Insertion Parameters in Sands 1533.5 NUMERICAL QUANTIFICATION OF DISTURBANCE 1553.6 INSERTION PROCEDURE FOR THE UBC SBPM 1593.6.1 Shoe Plugging 1623.6.2 Jetting Rod Position 1653.6.3 Jetting System 1703.6.4 Dimensional Differences Along Shaft 1713.6.5 Recommended UBC SBPM Insertion Procedure 1763.7 TESTING PROCEDURE 1773.7.1 Rate of Inflation 1773.7.2 Holding Time Prior to Unload Reload Loops 1783.7.3 Recommended UBC SBPM Testing Procedure 1813.8 SUMMARY AND CONCLUSIONS 184CHAPTER 4.0 TEST RESULTS AT THE UBC SITE AT LAING BRIDGE , 1874.1 INTRODUCTION 1874.2 GENERAL CHARACTERISTICS OF THE LANG BRIDGE SITE 1874.2.1 Geology 1874.2.2 Location and Features 1894.2.3 Stratigraphy 1914.2.4 Maximum Shear Modulus 1944.2.5 Laboratory Testing Results 1964.2.5.1 Soil Strength 1964.2.5.2 Soil Classification 1964.3 INTERPRETATION OF UBC SBPM DATA 1974.3.1 Testing Curves 1974.3.2 Curve Fitting Interpretation 2014.3.2.1 Friction Angle 2014.3.2.2 Coefficient of Earth Pressure at Rest 2084.3.2.3 Shear Modulus 2124.3.2.4 Modulus Reduction Curve 215V4.4 CONCLUSIONS 220CHAPTER 5.0 SUMMARY AND CONCLUSIONS 2225.1 SUMIvIARY 2225.2 METHODOLOGY OF INSERTION 2245.3 METhODOLOGY OF TESTING 2265.4 METHODOLOGY OF INTERPRETATION 2265.5 FINAL REMARKS 2285.6 SUGGESTIONS FOR FUTURE RESEARCH 2285.6.1 Equipment Development 2295.6.2 Interpretation of SBPM Data 229BIBLIOGRAPHY 230APPENDIX A DESCRIPTION OF THE UBC SBPM 245A. 1 INTRODUCTION 245A.2 PUSHiNG AND PUMPING UNITS 245A.3 TESTING UNIT 247A.3.1 Data Acquisition System and Related Sensors 247A.3 .2 UBC Selfboring Pressuremeter 250APPENDIX B CALIBRATION OF THE UBC SBPM 254B.1 INTRODUCTION 254B.2 STRAiN ARMS 254B.3 PRESSURE TRANSDUCERS 256B.4 MEMBRANE RESISTANCE AND LANTERN COMPLIANCE 257B.4. 1 Compliance Correction of Unload Reload Loops 262B.5 SUMMARY 265APPENDIX C RESULTS OF THE TRIAXIAL TESTS 266C.l INTRODUCTION 266C.2 TESTS WITH THE RECONSTITUTED SAMPLES 268C.3 TESTS WITH THE “UNDISTURBED” SAMPLES 271C.4 CONSTANT VOLUME FRICTION ANGLE 277C.5 SUMMARY 278‘ILLIST OF TABLESTable 2.1: Comparison of Cavity Expansion Models for Interpretation of Pressuremeter TestsmSands 20Table 2.2: Comparison of Friction Angle Predicted by Different Traditional InterpretationMethodologies 46Table 2.3: Calibration Chamber Testing Results on Leighton Buzzard Sand (Modified afterFahey, 1986) 85Table 2.4: Fitting Results on Chamber Tests with Leighton Buzzard Sand 90Table 2.5: Calibration Chamber Testing Results on Tiemo Sand (Modified after Bellotti et a!, 1987) ... 101Table 2.6: Recommended Fitting Range of SBPM Tests in Sands 125Table 2.7: Sensitivity of Fitting Results for Increase or Decrease in one of the Input Parameters 130Table 2.8: Sensitivity of Fitting Results for (a) Constant Volume Friction Angle, (b) Poisson’sCoefficient 132Table 3.1: Testing Programme at the Laing Bridge Site 140Table 3.2: Characteristic of the Pressuremeter Tests Performed at the Site 141Table 3.3: General Characteristics of the Lanterns Used 146Table 3.4: Successful SBPM Drilling Variables in Sand 154Table 3.5: Establishment of the Optimum Jetting Rod and Nozzle Type 168Table 3.6: Influence of Creep Strain on Unload Reload Modulus 182Table 4.1: Curve Fitting Results in Laing Bridge Site 205Table 4.2: Assessment of Coefficient of Earth Pressure by Empirical Formulae 210Table 4.3: Strain and Stress Levels Related to Gur and Gi 214Table BA: General Calibration Characteristics of the UBC SBPM Sensors 255Table Cl: Results of the Triaxial Testing Programme with “Undisturbed” Samples 276viiLIST OF FIGURESFigure 2.1: Indirect Approach for Analysis of Selfboring Pressuremeter Results 6Figure 2.2: Orientation of Stresses and Strains of a Soil Element at a Radial Distance r 8Figure 2.3: Idealized Pressuremeter Stress Path and Elastic Plus Plastic Zones Developed Around theCavity 13Figure 2.4: Stress Distribution Around a Cylindrical Cavity in Sand (Modified after Howie, 1991) 16Figure 2.5: Effect of Unloading under a Yield Surface (Modified after Hughes, 1982) 18Figure 2.6: Stress-Strain and Volumetric Strain Shear Strain Curves for (a)Simple Shear Test Results(Stroud, 1971) and (b) Idealized by Hughes et a!, 1 977(after Bellotti et al, 1987) 22Figure 2.7: Particular Stage of Cavity Expansion in Sand 32Figure 2.8: Elastic Radial Displacement at the Cavity Wall for a Particular Stage of Expansion 35Figure 2.9: Elastic Displacement at the Wall due to Compressive Volume Change of a Soil Element 38Figure 2.10: Comparison of Predictions given by the Numerical and Analytical Solutions 41Figure 2.11: Comparison of Pressuremeter Curves Predicted by Different Models 43Figure 2.12: Comparison of Measured and Applied Lateral Stress in Calibration Chamber(afterBellottietal, 1987) 50Figure 2.13: Concept of Elastic Shear Stiffness as Obtained from Unload Reload Loops (ModifiedafterBruzzietal, 1986) 53Figure 2.14: Effect of Stress Level on Unload Reload Shear Modulus 55Figure 2.15: Plastic and Elastic Zones Prior and After the Loop Stage 56Figure 2.16: Effect of Strain Level on Unload Reload Shear Modulus 60Figure 2.17: Variation of Shear Modulus with Shear Strain for Sands (after Idriss, 1990) 61Figure 2.18: Determination of Initial Modulus from SBPM Curve 66Figure 2.19: Variation of Soil Stiffness in Elastic Zone for a Particular Stage of the Cavity Expansion.. 68Figure 2.20: Basis of the Curve Fitting Technique 72Figure 2.21: Hyperbolic Soil Response at the Elastic Zone in terms of(a) Stress-Strain Curve and(b) Modulus Reduction Curve 76Figure 2.22: Curve Matching without allowance for Finite Boundaiy Correction - Fahey, 1986 Tests .... 86Figure 2.23: Curve Matching on Fahey, 1986 Tests: Hughes et al, 1977 Model 87Figure 2.24: Curve Matching on Fahey, 1986 Tests: Carter et a!, 1986 Model 88Figure 2.25: Curve Matching on Fahey, 1986 Tests: New Cavity Expansion Model 89Figure 2.26: Curve Matching on Fahey, 1986 Tests: Limits for Gi 93Figure 2.27: (a) Curve Matching on Jewel! et a!, 1980 Test B5(b) Elastic and Plastic Zones Inside the Calibration Chamber 96Figure 2.28: Curve Matching on Bellotti et al, 1987 Test 228: Influence of Chamber to Probe Ratio 98Figure 2.29: Curve Matching on Bellotti et a!, 1987 Tests: Ideal Installation Tests 102Figure 2.30: Curve Matching on Bellotti et al, 1987 Tests: Selfbored Tests 103Figure 2.31: Predicted Results of the Curve Matching: Friction Angle and Lateral Stress 104Figure 2.32: Predicted Results of the Curve Matching: Shear Modulus 105Figure 2,33: Influence of Range of Fitting on Results: Undisturbed Test(a) Curve Fitting at Distinct Ranges, (b) Variation of Parameters 110Figure 2.34: Schematic of Plastic Zones Developed at Cavity Strains of 10 and 20 % 112Figure 2.35: Idealized and Typical Experimental Stress Strain Curves of Sand During Shear 114Figure 2.36: Effect of Continuous Shearing on Measured SBPM Response 115Figure 2.37: Soil Elements Around the Probe Chosen for the Numerical Analysis FollowingManassero, 1989 117Figure 2.38: Results of Numerical Analysis for Test SBPO9: 5.3m. (a) Volumetric Strain againstCavity Strain for Soil Elements, (b) Relation between Plastic and Critical State ZonesDeveloped around the Probe 119vii’Figure 2.39: Influence of Range of Fitting on Results: Disturbed Test(a) Curve Fitting at Distinct Ranges, (b) Variation of Parameters 121Figure 2.40: Effect of Disturbance on Measured SBPM Response 123Figure 2.41: Curve Fitting of Test SBPO9 (5.3 m) with Carter et al, 1986 Model 127Figure 2.42: Sensitivity Analysis of the Curve Fitting: (a) Friction Angle, (b) Shear Modulus(c) Lateral Stress 128Figure 3.1: Location of In Situ Tests at Lamg Bridge Site 139Figure 3.2: Steel Lanterns with Overlapping Steel Strips and Camkometer Lantern 145Figure 3.3: Lantern Effects on Gur 150Figure 3.4: Disturbance Evaluation of SBPM Testing Curve 157Figure 3.5: Practical Numerical Assessment of Disturbance 160Figure 3.6: Typical Variables Obtained with the Sounding File of the UBC SBPM 161Figure 3.7: Typical Comparison of Curves from Tests with Different Shoe Pluggings 164Figure 3.8: Relationship Between Disturbance and Penetration Resistance 166Figure 3.9: Comparison of Coefficients of Disturbance for Different Testing Soundings 169Figure 3.10: Characteristics and Effects Caused by the Oversized Lantern Retainers 173Figure 3.11: Effects of Dimensional Differences Along Shaft 175Figure 3.12: Influence of Rate of Inflation on Pore Pressure Development 179Figure 3.13: Influence of Rate of Inflation on Testing Curve 180Figure 3.14: Proposed Testing Procedure for Selfboring Pressuremeter Tests 183Figure 4.1: (a) Laing Bridge Site Location, (b) Test Area of this Site (Modified after Sully, 1991) 188Figure 4.2: Geological Cross Section of the Fraser Delta (after Blunden, 1975) 190Figure 4.3: Section 1-1 of Laing Bridge Site 192Figure 4.4: Typical Piezocone Profile in Laing Bridge Site 193Figure 4.5: Profiles of Shear Wave Velocity and Maximum Shear Modulus 195Figure 4.6: Typical Pressure Expansion Curves Measured by the UBC SBPM at Laing Bridge 198Figure 4.7: Typical Displacements Measured during the Expansion Stage of the UBC SBPM 200Figure 4.8: Final Curve Matching with Field Data: (a) Depth 5.3 m, (b) Depth 6.3 m 202Figure 4.9: Final Curve Matching with Field Data: (a) Depth 8.3 in, (b) Depth 10.3 m 203Figure 4.10: Final Curve Matching with Field Data: Depth 13.3 m 204Figure 4.11: Comparison of Curve Fitting Results: Friction Angles 207Figure 4.12: Comparison of Curve Fitting Results: Shear Modulus 213Figure 4.13: Determination of Modulus Reduction Ratio for Different Amplitudes of Shear Strain 216Figure 4.14: Comparison between the Predicted Gi from Curve Fitting and the Seismic Modulus G0.... 217Figure 4.15: (a) Hyperbolic Stress Strain Curve for Depth 8.3 m (SBP 19)(b) Secant Shear Modulus Ratio versus Shear Strain 219Figure A. I: Pushing and Pumping Units ofthe UBC SBPM System 246Figure A.2: Testing Unit of the UBC SBPM System 248Figure A.3: UBC Selfboring Pressuremeter Sectional Assembly 252Figure B. 1: Membrane Resistance from Expansion in Air 258Figure B.2: Lantern Compliance from Expansion Inside Split Cylinder 261Figure B.3: Compliance Correction of Unload Reload Loops: (a) Compliance Test, (b) Determinationof Gsys with Basis on Power Equation 264Figure C.1: Schematic of Triaxial Testing Apparatus 267Figure C.2: Typical Stress Strain Response of the Reconstituted Samples: Constant C. Pressure 269Figure C.3: Typical Stress Strain Response of the Reconstituted Samples: Constant Relative Density.. 270Figure C.4: Typical Stress Strain Response in the Tnaxial Test: “Undisturbed” Samples 273Figure C.5: Determination of the Peak Friction Angle from the Triaxial Tests: “Undisturbed” Samples 274Figure C.6: Determination of the Constant Volume Friction Angle of the Lamg Bridge Sand 279ixLIST OF SYMBOLSb Intermediate stress parameterD DiameterDr Relative densityD10 Diameter at which 10 % of soil is finerD50 Diameter at which 50% of soil is finerE Young modulusf Empirical factorG Shear modulus or stiffness of the sandGi Shear modulus of the elastic sand, as obtained by the interpretation of the SBPM dataGp Shear modulus of the zone of soil encompassed by the plastic zoneGur Unload reload shear modulusGurc Unload reload shear modulus corrected for stress levelGmax Maximum shear modulus predicted with the pressuremeter GurGmeasured Measured (not corrected) shear modulus from the unload reload loopGsys System shear modulus, defined by Fahey and Jewell, 1990 to correct the GurMaximum shear modulus of the sandg Green straing Radial green straingo Circumferential green strainH HeightK Coefficient of lateral pressureK Upper bound of the constant K from Rowe’s stress dilatancy equationK0 Coefficient of earth pressure at restKa Coefficient of active earth pressureK0 Coefficient of passive earth pressureKg Modulus numberL LengthN Principal stress ration Constant from Rowe’s stress dilatancy equationP’ Stress component of diagram of Lambe and Whitman, 1979Effective internal pressure at the start of unload reload ioopPa Atmospheric pressureP Effective pressure at the cavity wall of expanding cavityQ Stress component of diagram of Lambe and Whitman, 1979Q Cone bearing resistance corrected for pore pressurer radiusr radius of the cavity wallr radius of the cavity wall at a deformed positionr0 initial radius of the cavity wail or the pressuremeterur Radial displacementCircumferential displacementu Displacement at the cavity wallu Radial displacement at the cavity wallUrew Elastic radial displacement at the cavity wallPlastic radial displacement at the cavity wallu Displacement at the elasto-plastic boundaryDisplacement at the inner soil ring caused by the external ringxlire Elastic radial displacement at the soil ringU0 Hydrostatic ground water pressureU1 Dynamic pore pressure measured at the face of the piezoconeU2 Dynamic pore pressure measured behind the tip of the piezoconeU3 Dynamic pore pressure measured behind the friction sleeve of the piezoconeV VolumeV Shear wave velocityw Water contenta StressEffective lateral stress of the soilEffective radial stressEffective circumferential stressa or a Effective vertical stressam Average normal effective stress in the horizontal planeav Average normal stress within the plastic zone that existed prior to the loop stageEffective radial stress at the cavity wall or cavity pressureLimit effective radial stress to start yielding of the sandLimit effective radial stress to start yielding of the sand at the cavity wallDeviator stress of triaxial testEffective confining pressure of triaxial testsMajor principal effective stress of triaxial testa’2 Intermediate principal effective stress of triaxial testMinor principal effective stress of triaxial testEffective boundary stress of calibration chamber testsB StrainEr . . . . . ..... Radial strainCircumferential strainVertical strain80w Circumferential strain at the cavity wall or cavity strainLimit circumferential strain to start yielding of the sandCircumferential strain at the start of the holding phaseAlmansi strain(Yr Alniansi radial strainAlmansi circumferential strainAlmansi circumferential strain at the cavity wall8 Dilation angleoIL, Direction of plane of anisotropy from Oda, 1981 tests13 Empirical parameterv Coefficient of Poissony Shear strainyf Limit shear strain to start yielding of the sandYav Average shear strain amplitude imposed by the loop in the surrounding soilp Deformed radius or soil densityt Shear stressTm Average shear stress in the horizontal planeLimit shear stress to start yielding of the sandTmax Maximum shear stress of the sand in the horizontal planeFriction angleConstant volume friction anglexAxially symmetric friction anglePS Plane strain (converted) friction angleAmplitude of shear strain at the cavity wallIncrement in volumeIncrement in effective horizontal stressIncrement in effective vertical stressdP Increment of effective internal pressure above hdP Increment of effective internal pressure during unload stagedr Increment of radiusdO Increment of angled10 Initial length of a small liner elementdl Length of this element in the deformed stateda Increment of effective radial stressda0 Increment of effective circumferential stressda Increment of effective vertical stressIncrement of effective radial stress at the cavity walld Increment of volumetric straindEa Increment of axial straindE9 Increment of circumferential strain at the cavity wall during unload stageds Increment of radial strain at the cavity wallIncrement of circumferential strain at the cavity walld’y Increment of shear strainACRONYMS:CD Coefficient of disturbanceCJ Central jetting systemCPTU PiezoconeCVL CivilENEL-CRIS Italian National Electricity Board-Hydraulic and Structural Research CenterFDP sounding.. .Testing sounding in which FDPM tests were carried outFDPM Full displacement pressuremeterLL Liquidity limitOCR Overconsolidation ratioOLR Oversized lantern retainerP1 Plastic indexP1 Limit PressurePL Plastic LimitSBP sounding . . .Testing sounding in which SBPM tests were carried outSBPM Selfboring pressuremeterSC tests Calibration chamber tests with smooth boundaries and constant boundary pressureSCPT Seismic coneSH Shower Head systemSPT Standard penetration testST1 Stationary piston sampler from Rocktest Inc.TOLR Tapered oversized lantern retainerUBC University of British ColumbiaxiiACKNOWLEDGMENTSTo the Lord, for the guidance of my life and destinyTo my parents Aldo da Cunha Rosa and Acilea Pinto da Cunha, for their love and support throughoutmy lifeTo my wife Daniela Abrahami Pinto da Cunha and daughter Bianca Abrahami Pinto da Cunha, fortheir support and strong encouragementTo the Brazilian Council of Research and Development (CNPq) for their generous 4 years scholarshipTo my advisor Dr. R.G.C. Campanella, for his valuable assistance and guidanceTo Dr. J,M.O. Hughes, for his enthusiastic assistance, encouragement and numerous discussions ofboth theoretical and practical aspects of the pressuremeter testingTo Drs. Byrne, Vaid and Fannin for their valuable discussions throughout my years at UBC andcritical review of the contents of this thesisTo Dr. J.A.R. Ortigão for his encouragement in the initial stages of the research and valuabletechnical discussionsTo Dr. Pat Stewart for the critical review, friendship and English correction of my thesisTo the colleges Matt Kokan, Thandara Murthy, John Sully, Mike Davies, Tim Boyd and Jody Everardfor their field help as well numerous technical discussionsTo the technical (and talented) staff of the Department of Civil Engineering for their support andpatient, with special attention to Art Brookes, Jim Greig, Thomas, Scott Jackson, Harald Schremp, PatSheenan and Flora LewTo Dr. De Souza Coutinho, for the permission of use of the cavity expansion model “Anabela”To all those people who directly or indirectly contributed and helped me in the development of thisthesisXIIIDEDICATIONThis thesis is dedicated to my wfe Daniela Abrahami Pinto da Cunhq, for her emotionalsuppor4 understanding andpatient throughout this endeavor1CHAPTER 1.0 INTRODUCTIONSite investigation and the assessment of the characteristics of the soil are important aspects of thegeotechnical design process. In situ testing has become increasingly important in geotechnical engineeringas a complement as well as a substitute for laboratory tests. The demand for in situ testing has developedwith a growing awareness that field sampling and laboratory testing may have problems of disturbance.This can be principally the case in granular soils, where the absence of cohesion demands an extremelyhigh (and costly) technology to retrieve truly representative samples from the site.On the other hand, in situ tests have a much more complex boundary condition than those imposed bylaboratory test devices. This considerably complicates the interpretation of in situ data as the stress andstrain conditions of the soil surrounding any of the in situ probes are difficult (if not impossible) to define.This major difficulty leads to the large array of empirical relationships that are applied to derive basic soilparameters from in situ measured values. Since the empirical relationships mainly rely on correlations withlaboratory tests, sample disturbance effects from to these tests will be inherently imposed on the empiricallyderived in situ soil predictions. Moreover, the empirical relationships are valid for a particular type of soiland testing conditions, that do not universally apply for all the existing natural deposits.In situ methods are basically divided into two general groups, namely logging and specific methods(Campanella and Robertson, 1982). The logging methods were primarily developed for stratigraphicprofiling determinations. They are fast and relatively economical in comparison to the specific in situmethods. The best example is the piezocone test. The specific in situ methods were developed in order toextend the particular knowledge of some soil property at specific locations defined by the logging tools. Ingeneral these methods are much more specialized and often slower to execute than the logging methods.Although the specific methods can also rely on empirical relationships to define the soil parameters, theyhave the potential to define the soil properties in tenns of a specific stress-strain model.It was in the context of the specific in situ methods that the pressuremeter was developed. Similarly,like all other specific in situ tools, the pressuremeter does not directly measure any soil parameter, butrather the pressure and the change in volume or radius of an expanding cylindricai membrane. However,the major attraction of the pressuremeter test is the fact that it constitutes a simple boundary value problem2in soil mechanics. It can be theoretically modeled by the expansion of an infinitely long cylindrical cavity,where boundary conditions are well defined and controlled. This offers the possibility of the simultaneousderivation of both in situ deformation and strength parameters when applying any of the several availablecavity expansion theories.The original concept of lowering a balloon like device down a borehole and inflating it to measuredeformation properties dates from 1930 or 1931 (Baguelin et al, 1978). The first reference to such devicewas given by Kogler, 1933 who developed a simple probe with length of 125 cm and diameter of 10 cm.This first device consisted of a long sausage shaped bladder which stretched between two metal discs. Thediscs were held apart at a fixed distance by a steel rod which formed the backbone of the device. It waslowered into a predrilled hole and gas inflated. The impact of KOgler’s invention in the geotechnical areawas insignificant, even though he was able to use the equipment to record pressure volume change curvesthat are similar to those obtained nowadays with more sophisticated equipments.It was only in the 50’s that the pressuremeter was developed and started to be used in real engineeringterms. Without knowledge of Kogler’s work, a civil engineering student at the Univ. of filinois calledMenard developed a pressuremeter in 1955 (Menard, 1955). In less than 3 years the Menard pressuremeterstarted to be produced by Menard’s own firm and used as a consulting tool in France. At that time Menardwas able to benefit from technological and analytical interpretation advance that was not available inKOgle?s time. Like Kogler’s original probe, the Menard pressuremeter was designed to be inserted inprebored holes. Due to the high disturbance generated at the cavity wall by the preboring process, resultsof Menard pressuremeter tests are generally interpreted by empirical rather than analytical methods. Rulesfor the design of piles under vertical and horizontal load were developed by Menard and his co-workers inthe early sixties. Experience grew until nowadays in France (see Briaud, 1986, Baguelin et al, 1978 andBaguelin, 1982), and there exists a large database of experimentation and observation.Since Menard’s pioneering work there have been major developments in the pressuremeter, especiallyin the 70’s and 80’s. These developments can be subdivided into areas of pressure and strain measurement,probe insertion, analytical and numerical interpretation, and new types of pressuremeters (Wroth, 1982).The selfboring pressuremeter (SBPM) was devised to eliminate the disturbance problems caused by the soilpreboring adopted by the Menard pressuremeter. The first SBPM was developed and used at the Saint3Brieuc Laboratoire des Ponts et Chaussées in 1967 (Baguelin et al, 1978), and also called the PressiomêtreAutoforeur, or PAF. The French work about this pressuremeter was only published in 1972 by Baguelin etal. At about the same time the development of the British SBPM was taking place at the University ofCambridge. The first publication about the British research was by Hughes, 1973. The initial Englishname chosen was Camkometer (Cambridge K0 meter), although it is nowadays called selfboringpressuremeter.The advantage of the selfboring pressuremeter is the fact that this probe can ideally be inserted intothe soil without disturbance, thus avoiding any stress and density changes on the original groundconditions. With this characteristic the test results can be properly analyzed in the light of cavity expansiontheories, where the initial conditions are well established. However, Hughes in 1973 demonstrated in thelaboratory with the X-ray technique that even under “perfect’ selfboring conditions a disturbance of at least0.5 % of the pressuremeter radius can be induced in sands. hi real seilboring cases an even higherdisturbance percentage can be expected. This disturbance undoubtedly hampers the major premise andpotential advantage of the use of selfboring pressuremeters, at least in sands.Since its original development the selfboring pressuremeter test has been subjected to close scrutiny bya number of researchers. It has been found that the results of this test are extremely sensitive to installationtechniques, test procedures and methodologies of interpretation, which is not surprising in view of thesophisticated nature of the pressuremeter probe (Mair and Wood, 1987). The degree of success inobtaining reliable predictions of soil parameters from the selfboring pressuremeter is critically dependent onthe combined influence of the methodology of interpretation and the disturbance built in the testing curve.An urgent need for the establishment of less sensitive interpretation methodologies is required.The emphasis of this thesis is placed on the methodology of interpretation of the selfboringpressuremeter testing data in sands. The current methodology of analysis of the pressuremeter testingcurves is reviewed and a new and less sensitive interpretation approach is proposed. The reliability of thederived soil parameters, and the sensitivity of this new approach, are respectively addressed and based onpressuremeter tests in controlled environments and parametric studies. The discussion of the mostadvanced existing cavity expansion models that can be applied in this new approach are also presented,together with the development of a proposed new cavity expansion model.4The complementary information that is obtained with the new interpretation methodology is thenumerical quantification of the disturbance present in the testing curve. With the aid of this newmethodology of disturbance assessment it is also possible to show the quantitative effects caused by thevariables that influence the testing curve of the UBC SBPM. This discussion considers the large databaseof field testing trials, in which equipment, insertion and site related variables were changed. With thesespecific findings the UBC SBPM insertion procedure was established. As will be demonstrated, thesuggested procedure not only aid in the standardization of the use of this complex in situ tool, but also leadsto “optimum” conditions of insertion where the likelihood of disturbance is considerably reduced.With the adoption of the suggested insertion procedure it was possible to obtain high quality testingcurves. The interpretation of these curves using the new interpretation methodology led to predictions ofthe basic soil parameters of the sand (friction angle, shear modulus and lateral stress). The discussion onthe reliability of such predictions is also explored herein.5CHAPTER 2.0 INTERPRETATION OF PRESSUREMETER DATA iN SANDS2.1 INTRODUCTIONThe objective of this chapter is to examine the methods and models used to determine soil parametersfrom seithoring (SBPM) pressuremeter data in sands. It describes the assumptions of the existing cavityexpansion models developed for the analytical and numerical interpretation of the testing curve of theSBPM test, and introduces a new cavity expansion model. The chapter also introduces the new analyticalinterpretation methodology proposed in this thesis to derive meaningful soil parameters from the SBPM. Inorder to assess the need of a new interpretation methodology the reliability of the soil parameters derivedwith the traditional interpretation methodologies is initially reviewed. These parameters are the frictionangle 4, the horizontal stress Oh, and the shear modulus G ofthe sand.2.2 CONSTITUTIVE MODELS FOR PRESSUREMETER ANALYSISPressuremeter testing results can be used in geotechnical design by two distinct approaches, theindirect and the direct one. The direct approach, adopted in the empirical relationships of Menard, uses thepressuremeter data to empirically derive the bearing capacity, settlement, etc. of engineering works. Theindirect approach, depicted in Figure 2.1, uses interpretation methods that allow the evaluation of theparameters that describe the basic material characteristics of the sand. This approach requires a model todescribe the expansion process in the sand. This thesis will adopt only the indirect approach to predict soilparameters from pressuremeter testing results.This section introduces the cavity expansion models developed to solve the complex boundary valueproblem of a pressuremeter test in sand.2.2.1 GeneralThe pressuremeter can be simply modeled as the expansion of an infinitely long cylinder. The soil isassumed to be homogeneous and isotropic. The radial (r) and circumferential (8) directions are considered,respectively, the major and minor principal directions of stress and strain. The initial stress condition isassumed to be isotropic in the horizontal plane, i.e., the lateral stress (Oh) is initially equal to the radial (Or)and circumferential (Os) stresses.6hdr/ruCircumferential StrainGroundCAVITY EXPANSION MODELEngineering Soil PprametgiFriction Angle, Honzontal Stmss, Sheer ModulusPressuremeter Test— Pressuremeter Results—.. Pressuremeter Test InterpretationGrounddrtoCotoC>CoC.)—Figure 2.1: Indirect Approach for Analysis of Selfboring Pressuremeter Results7Measurements during a test are made of cavity (internal) pressure and cavity volume or strain. Thesystem is at the rest condition when the internal cavity pressure P equals the horizontal stress ah, and liftoff of the arms occurs. A pressure increase dP above this horizontal stress causes expansion of the cavityin the radial direction. Under this condition, a particle located initially at a distance r to the center of thecavity will displace to a new deformed radius p so that:p=r+ur (2.1)where ur is the radial displacement.The surrounding soil will be subjected to plane strain deformation if the length of the pressuremeter ismuch greater than its radius. In this case no deformation will occur in the vertical direction. Figure 2.2shows the orientations of stresses and strains considered here, Elongation is considered positive, whereascontraction negative, as assumed by Baguelin et al, 1978 for the following mathematical equations.The strain acting in the radial and circumferential directions can be determined from the displacementfield. For small deformations Cauchy (E) definition of strain can be used, since Cauchy’s definitioninvolves just the first derivatives of the displacement vector with respect to the Cartesian coordinates.Using the orientation of strains given by Figure 2.2 the following compatibility equations can bederived:(r+u)O—rOu (2.2)rt3 rdu-dr (2.3)dr drwhere E0 is the circumferential strain and Er the radial strain.For large strains, Green (g) or Almansi (cx) definitions of strain are used. In this case:1d12_d12 1g= 2 d102°=((1+8)2—1) (2.4)1d12_d12 1 1=—(1— ) (2.5)2 d12 2 (1+8)28Radial and Tangential Stressescrr+(dcYr/dr).drRadial and Tangential displacementsue + (due/e).oFigure 2.2: Orientation of Stresses and Strains of a Soil Element at a Radial Distance r9where: d10 = initial length of a small linear elementdl = length of this element in the deformed stateand 8 is the Cauchy strain (“(dl-cllo)/dl0)The relationship given by Equations 2.4 and 2.5 allows the determination of circumferential strains (goand c) based on the Cauchy circumferential strain 80, as well as radial strains (ga- and a) based on theCauchy radial strain Er.2.2.2 Basic Considerations of an Elastic ModelFor the initial deformation stage of the cavity the soil is assumed to behave elastically. From thetheory of elasticity the principal stress changes dar (radial), do0 (circumferential) and do (vertical) can befound from Hooke’s law (rheological equations), as presented in Baguelin et al, 1978:ECr =dtr —u(dcr+dcY) (2.6)—E.e =d—u(dcY +dc7) (2.7)—E.e =du —u(d0 +dY) (2.8)where: E is the Young modulusv the Poisson’s coefficient. Assuming plane strain conditions= 0 (2.9)therefore dY2 V(dCYr + dJ0) (2.10)Substituting (2.10) into (2.6) to (2.8) and combining with (2.2) and (2.3):—E= thY—dr (2.11)(1—v)(1+U) dr r 1—v-E u 1)_________— do (2.12)(1—v)(1+v) r 1—1)Solving for dOr and do0 we obtain:—E(1—V) (dii v ii= I_r+ (2.13)r (1—2v)(1+v)dr 1—Un—E(1—U) (u V diithY =—i-I (2.14)0 (1—2l))(l+V)’r 1—1) r )10To determine the state of stress an equation of equilibrium is required to relate radial and tangentialstresses. This equation must satisfy the boundary conditions of the problem. The equilibrium equation forthe cylindrical expanding cavity is obtained by considering the equilibrium in the radial direction.Considering a small element in the radial plane (Figure 2.2) we obtain:0 do d____c.r.8+2.Y.dr—=0.(r+dr)(fY+ T.dr),or —-+ (2.15)2 dr dr rCompressive stresses are taken as positive by Baguelin et al, 1978, Substituting (2.13) and (2.14) into(2.15) results in the following differential equation:d2u dur2 r +r—-—u =0 (2.16)dr2 drwhich has solution in the form of ur A.r + BIr. For a large r, Ur —*0, therefore A= 0.At the cavity wall u ur = BIr, therefore B = u .r andU .I r2U = W W = (2.17)r rwhere: 6w is the circumferential strain at the wall.Combining (2.2) and (2.3) with (2.13) and (2.14) we obtain:2 2£ C .r=2(2.18), Cr = — 2 (2.19)r rand:E c (2.20)r h r h (1+1)) 2= + d0 =— E r 2 (2.21)(1+1)) r2Based on Equations 2.20 and 2.21 it is observed that all (elastic) soil elements around the cavity willfollow a unique stress path during the cavity expansion. Moreover, every increase of radial stress producesan equal decrease of tangential stress (dar=-dae, Lamé, 1852). Similarly, the circumferential strain will be11equal and opposite in sign to the tangential strain, i.e. s0 = r and no volume change takes place duringexpansion.Noting that the shear modulus (G) of the soil is given by E12 (l+v) it follows that Equations 2.20 and2.21 can be changed to:e .r2+ dO = + 2G 2w (2.22)re .r2c 3 +do y —2G (2.23)o h kThese equations will control the stress distribution in the elastic soil for pre yielding conditions. At thecavity wall r = r, and the boundary stress-strain relation becomes:CYh +2G.e (2.24)=o —2G.e0 (2.25)h wAs commented before, plane strain conditions are assumed. Since 0r + a0 = 20h = constant (i.e. radialand circumferential stresses change by equal and opposite amounts from the in situ lateral stress), theaverage normal stress 0m ( or (a+ae)I2) in the horizontal plane remains constant. Only the shear stress tm(equal to (ar-ao)/2) in the horizontal plane will vary from element to element. Therefore, the expansion ofthe cavity in the elastic medium is a pure shearing process at constant a.Rearranging Equation 2.24 it is possible to obtain the following expression for the shear modulus ofthe elastic sand:1c5—cY ldPG=— r h (2.26)2 (e) 2dewhere:dP is the increment of cavity pressure above Oh andds the corresponding cavity strain amplitude.Therefore the shear modulus is obtained by measuring the initial slope of the SBPM testing curve. Ifthe soil responds (linearly) elastically, as assumed, this slope will be a straight line.12The elastic model is, however, restricted to small deformations. Once the applied shear stress in thesand equals the available shearing strength, failure (or yield) occurs. As soon as yielding starts to begenerated at the cavity wall the elastic model no longer applies.2.2.3 Basic Considerations of an Elasto-Plastic Model without Volume ChangeAs the pressure is further increased the soil starts to yield at the cavity wall. An annulus of failed soilcommences to develop, extending from the deformed cavity radius r to an elasto-plastic boundary withradius r1,, as schematically shown in Figure 2.3(a). This boundary will expand as the pressure in the cavityincreases. In the zone beyond this boundary the shear stress has not reached the failure value, hence thesoil responds elastically. The elasto-plastic boundary, therefore, divides the soil into two distinct zones, the“plastic” and the “elastic” zones.The Mohr Coulomb failure criterion is assumed to rule the onset of yielding. This fIülure criterion isgiven by the following equation:1—sinØ—=N (2.27)1+sinØAt the elasto plastic boundary both elastic and plastic conditions apply. Therefore:c58 Y —d 1—sinh (2.28)O• an+dor 1+sm4and hencedCI = sin, (2.29)The limiting cavity pressure at wall at which yielding commences is given by:= h +d(Y = (1+sin ) (2.30)where: Gf is the limiting effective radial stress to start yielding of the soil.The shear stress of any element of soil at the onset of yielding (as element B of Figure 2.3(a)) is:tf—ah sine (2.31)where: tf is the limiting shear stress to start yielding of the soil.Since(2.32)13ElementI +s4)0(m.s1n4)!G(r=infinity)/VA(A) ELASTIC AND PLASTIC ZONES DEVELOPED AROUND THE CAVITYh(I+Sifl4)Motir Coulomb Failure Line(B) STRESS PATH FOR DRAINED LOADINGFigure 2.3: Idealized Pressuremeter Stress Path and Elastic Plus Plastic Zones Developed Around the Cavitythe shear strain amplitude of any element at the limiting (yield) conditions is:ELASTIC ZONEPLASTIC ZONEABRadial Stress Shear StrainAcyr = Nar-w=Isotropic Line0Th (YO14f YSfflØ3’ (2.33)G Gwhere: y is the limit shear strain to start yielding of the soil.Since the elasto plastic boundary represents the limiting (yield) condition of the soil surrounding theprobe as expansion proceeds, the elements of soil at this interface (like element B of Figure 2.3(a)) willalways be under levels of shear strain amplitude y.As the internal pressure of the cavity increases the elements of soil inside the plastic zone are subjectedto levels of shear stress above tf. The stress path of the soil elements inside this zone will follow the MohrCoulomb failure line. Figure 2.3(b) presents the idealized stress path of elements of soil around the probefor a particular expansion stage. As failure is mobilized at the cavity wall the stress path moves upwardswith a constant stress ratio N (as the path shown from points B to C). At this time, different elements ofsoil will be at different positions along this unique path.Within the annulus of soil under the plastic regime the equilibrium equation must be satisfied,therefore substitution of Equation 2.27 into 2.15 leads to the following result:du 7(1-N)=0 (2.34)dr rUsing the boundary conditions at a generic expansion stage after the development of the plastic zone,we can adopt °r = a at r = r (cavity wall). Substituting into Equation 2.34 and integrating from thecavity radius to a generic radius r it is obtained:1n—---=—(1—N)In-- (2.35)0 rrw wor within the plastic zone (for r < r, where r is the radius of the elasto-plastic boundary):I r (1-N)r0rwII (2.36)\. rjSimilarly, the circumferential stress is given by:15( (1-N)= N.0 (2.37)This solution demonstrates that the rate of decrease of stress with distance within the plastic zonedepends on the slope of the Mohr’s envelope (or the friction angle of the sand).The variation of radial and circumferential stress along the radius for a particular stage of the plasticradius r, (or the effective internal pressure of the cavity a) is schematically shown in Figure 2.4. Thesame soil elements considered in Figure 2.3(a) are represented in the former figure. It is noticed that thedistributions of stresses inside the plastic zone follows a different path than those given inside the elasticzone, since the equations that rule each case are different. It is also noticed inside the plastic zone that thedifference between the two stress distributions (Or and 00), in other words the shear stress, is variable.Using the Equations 2.36 and 2.37 it is possible to obtain the value of the average normal stress Gm atany soil element in the plastic zone, for a particular stage of the cavity expansion. Therefore:y +01 1r (1-N)0 = =—(1+N)0 I —-- I (2.38)1 rwj2 2Hence the average normal stress increases in the plastic zone as expansion proceeds.In summary, based on elasto-plastic considerations, the initial rise of the pressuremeter testing curvefollows a straight line with slope 2G. Once yielding starts at the cavity wall, and propagates further withthe development of a plastic zone, the testing curve becomes non linear. Within the elastic zone the averagestress remains constant and only shearing takes place. Elements of soil at the elasto-plastic boundary willbe under a unique condition of shear stress and shear strain level. Within the plastic zone both the averagenormal stress and the shear stress increase during the expansion process.2.2.4 UnloadingIf the cavity pressure is reduced in the course of a pressuremeter test, then the soil around the cavitywill initially behave elastically. An immediate transformation of the whole plastic zone into elastic occursin the start of unloading, as the stress paths of the soil elements will fall below the yield surface of the sand.This simple concept was put forward by Hughes, 1982 and Wroth, 1982 based on the ideal behavior ofsoils introduced by Roscoe et al, 1958.16PC PlasticZone0r = cTb(1+sin 1)ElasticZonecTINçr=R rC B AFigurc 2.4: Stress Distribution Around a Cylindrical Cavity in Sand (Modified after Howie, 1991)17Figure 2.5 presents a typical response during the loading and unloading stages of the pressuremeter,and the corresponding stress path of an element of soil at the cavity wall. The initial stresses at the probeand at all the soil surrounding the cavity are given by the effective horizontal stress 0h. As the pressure inthe cavity wall is increased beyond Oh lift off occurs, and the pressuremeter produces a linear testingresponse. Once the cavity stress reaches the limiting value (given by Equation 2.30) yield starts at the wall,and the stress path moves out at a constant stress ratio (path 1-2). The stress path of soil elements atdistinct radii r from the cavity wall will lag behind that shown in Figure 2.5, as the level of stresses willdecay with radius (see Figure 2.4). As the stress path of the soil element at the wall moves along the failureline, so does the current yield surface of this element. At point 2 of Figure 2.5 the current yield surface isshown by a dotted line. At this stage both the current yield surface and the stress state will coincide. Thesame will happen for other soil elements around the cavity. As the cavity pressure is reduced the stressstate of all the soil elements around the probe will fail below the (respective) current yield surfaces.Therefore, these elements will behave elastically if the unloading is sufficiently small to avoid plasticstrains. On the basis of the above concept, the behavior during an unload reload loop of the pressuremetertest can be expected to be elastic.Wroth, 1982 pointed out that on the continuation of the unloading process reverse failure may occur,with the circumferential stress as the major principal stress. Assuming that the elastic response is onlymobilized for stress paths between the Mohr Coulomb “passive” and “active” failure lines, he derived anequation to predict the maximum decrease of effective cavity pressure for the elastic response of thesurrounding sand. This maximum pressure decrease shall be observed when performing unload reloadloops to measure the shear modulus (Gur) of the sand.2.2.5 Review of Elasto-Plastic Models with Volume ChangeAnalytical models for the simulation of cavity expansion in sands are based on the compatibilityconditions between strains and displacements and on the equilibrium equation of the medium that envelopesthe cavity. These basic conditions are the same for all models.h the previous subsection the cavity expansion solution for a purely elastic ffictional medium wasgiven. It was also demonstrated that on the onset of yielding a plastic zone develops. Within this zone highlevels of shear stress are imposed in the sand.18Failure Linea)aWar=N0 57/jTESTING CURVE STRESS PATH AT WALLCavity StrainFigure 2.5: Effect of Unloading under a Yield Surface (Modified after Hughes, 1982)19Sands may dilate or increase in volume during shear. The phenomenon known as dilatancy hasalready been illustrated by many laboratory tests on sand (Rowe, 1962, Stroud, 1971, Vaid et al, 1980) andhas been shown to have an important influence on the behavior of medium dense to dense granularmaterials during pressuremeter expansion (Baguelin et al, 1978). Hence, in order to derive thedisplacement field within the plastic zone an assumption regarding the volume change of the sand duringshear must be made. Table 2.1 shows all the models developed so far for the analysis of pressuremetertests in sands, in which the dilatancy behavior of this material is considered.The first solution that incorporated volume changes for cavity expansion in sands was the solutionderived by Ladanyi, 1963. His model assumed the soil as an elasto-plastic material with a constant rate ofvolume change during the Mohr Coulomb failure condition. The relationship between volume change andshear strain was given by a simple and unique relationship. The general solution, however, required a stepby step procedure and the volumetric strain at failure had to be interactively chosen to give a straight line ina log-log graph of cavity pressure versus strain. Although more complex, this model did not lead to asignificant improvement over the previous Gibson and Anderson, 1961 model, which was the first modelexclusively developed for sands. Gibson and Anderson’s model did not consider volume changes duringshear, and historically their main contribution was the drawing of the log-log graph of cavity pressureversus strain to predict the friction angle.Vesic, 1972 developed solutions for the spherical and cylindrical expansion in sand. A linear elasticplastic behavior with volume change was considered. His theory was based on laboratory tests (planestrain or triaxial compression) to determine the volume change relationship in the plastic zone. The basicapproach was to estimate the volume change in such a way that, together with the limit pressure, it wouldallow the derivation of the peak friction angle. The major drawback of this model was the necessity oflaboratory tests to define the shear induced volume change of the soil.In 1975 Wroth and Windle expanded the finite difference method of Baguelin et al, 1972 (for clays,called the “subtangent” method) to account for volume changes in the soil, and hence was used in theevaluation of tests in sands. The complete stress path, stress-strain and strength for an element of soil atthe cavity wall could be obtained, if the constant rate between 8 and volume change was known.MODELSTRESS-STRAINAPPROACHADVANTAGESDISADVANTAGESUSEFULNESSINSANDSLadanyi,1963Linearelasticperfectlyplastic.StepbystepnumericalSimplerelationshipbetweenCumbersometouse.FlowrulemayCVolumechangebyfixedflowprocedurevolumechangeandshearbeunrealisticruleStrarnVesic,1972Linearelasticperfectlyplastic.LimitpressureoffieldcurveSolutionsforsphericalandNeedslaboratorytestsandinsituCVolumechangefromcylindricalexpansioninsanddensitylaboratorytestsWrothandWindle,Notassumed.NumericalsolutionbyfiniteAllows theknowledgeoftheNeedsflowruleconstant.C1975Volumechangebylinearflowdifferencetechniquestress-strainbehavioroftheNumericalinstabilityrulesandHughesetal.977Linearelasticperfectlyplastic.Log-logplotofpressureEasytosolvenianuallyorviaNocompressionduringshear.AVolumechangebystressexpansioncurvecomputer.IncorporatesRowe’sConstantduringfailuredilatancystressdilatancyRobertsonandHughes,Linearelasticperfectlyplastic.Log-logofpressureexpansionAccountsforinitialEmpiricalcorrectionisapplied.B1986Volumechangebystresscurveandnomoan1compressivebehavioroflooseValidlbrloosesandsdilatancysandsCarter etal,1986Linearelasticperfectlyplastic.NotexchisivelydevelopedforIncorporateselasticstrainsinNotpossibletouseintoalog-logAVolumechangebystresspressurenseteranalysis,buttoplasticzone.Easytosolveviaapproach. Smallstrainsolutiondilatancyderivelimitpressureinpilecomputer.problemsHoulsbyetal,1986Linearelasticperfectlyplastic.OptimizationroutineusingDisturbanceisminimizedinContractionofpressuremetermayCVolumechangebystressmodelandfieldcurvesthecontractionstageofthetestnotfollowtheassumptionsofthedilatancy.ContractionmodelmodelManassero, 1989NotassumedNumericalsolutionbyfiniteAllowstheknowledgeoftheNumericalinstabilityBVolumechangebystressdifferencetechniquestress-strainbehaviorof thedilatancysandDeSouzaCoutistho,Rigidnotperfectlyplastic.NumericalsolutionbyfiniteAllowstheknowledgeoftheNumericalinstabilityB1990Volumechangebystressdifferencetechniquestress-strainbehaviorofthedilatancysandJuranandNotassunsed.NumericalsolutionwiththeFlowruleincorporatesmostofNeedslaboratorytestsandinsituCMahmoodzadegan,Volumechangebystresstestingcurvethecharacteristicbehaviorofdensity1989dilatancysandsduringshearFerreira, 1992Hyperbolic.OptimizationroutineusingStress-strainidealizationisPreliminaryassessmentgaveAVolumeChangebystressmodelandfieldcurvesmorerealisticfor strainunreliableresultsforchamberdilatancyhardeningsandsdataA=FIIGHB=MODERATEC=LOWTable2.1:ComparisonofCavityExpansionModelsforInterpretationofPressuremeterTestsinSands21Hughes et a!, 1977 proposed the first model that incorporated all the main features of sand behaviorduring shear without the need of extensive laboratory input data. The model idealizes the material asbehaving in an elasto-plastic manner, where only volumetric plastic and shear strains due to dilation wouldbe in effect in the plastic zone. The soil behaves elastically until the peak stress ratio at the wall is reached.At the onset of yielding the Mohr Coulomb failure criteria is followed with only plastic dilative strains inthe emerging plastic zone. The volume changes occur throughout expansion at a constant dilation rate,leading to a constant mobilized friction angle. Hughes et al, 1977 based this sand behavior on simple sheartests of Stroud, 1971, as demonstrated in Figure 2.6, as well as on the stress dilatancy theory ofRowe, 1971. Stroud, 1971 demonstrated that for dense sands at low to moderate confming pressuresdilation starts from the beginning of the test (with a small contractive section) progressing in a close toconstant rate.The cylindrical cavity expansion model of Hughes et al, 1977 does not consider this initial contractivesection1,and relies on the stress dilatancy equation put forward by Rowe, 1971 to link the peak stress ratio(or friction angle) to the dilation rate of the material. This equation has been found to be valid to describethe sand behavior at large stress ratios (Barden and Khayatt, 1966, Rowe, 1971), and shown to be pathindependent (Tatsuoka, 1976), Rowe’s stress dilatancy equation is given by:1+smö (2.39)0 1—sin68where:is the dilation angle (Hansen, 1958)K, = (l+sin)I( 1 -sin)The term K depends on the constant volume friction angle (4) of the sand. This constitutes the onlylaboratory data input of this model. The stress dilatancy equation is used together with the equations thatgovern both the radial distribution of stresses and displacements within the plastic zone to predict thepressure expansion relationship at the cavity wall.‘Actually the theory incorporates a constant “c” representative of the initial contractive section. But, based on thetheoretical analysis with pressuremeter data in sands, these authors state that c is sufficiently small to be neglected./ 02VENSE SAND ciI Ca) RESULTS OF SIMPLE SHEAR TEST1•S1vVb) SIMPLIFIED MODEL1’tlvVFigure 2.6: Stress-Strain and Volumetric Strain Shear Strain Curves for (a) Simple Shear Test Results (Stroud,1971) and (b) Idealized by Hughes et al, 1977 (after Bellotti et al, 1987)t = (Yr -a0)/2, S = (ar +221S0.20.1CaLivV10 20yLOOSE SAND203010 4050SI — a023This relationship is given by:(n +1)I c I c( (1-N)—)1,---J (2.40)setting c 0 (suggested by authors) and rearranging, we find:(1—N)(8 (n+1)3ra II (2.41)R }where:= Circumferential plastic strain at the cavity wall.c = Intercept of the idealized relation between volumetric and shear strain of Figure 2.6.65= Limit circumferential strain at elasto-plastic boundary = ‘yp(2, as given by Equation 2.33.= = Effective internal pressure at cavity wall.Limit effective radial pressure to start yielding of the sand radial pressure at elasto-plasticboundary, or ah (1 + sin) as given by Equation 2.30.N = (1 - sin4)/(1 + sine) (2.42)n (1 - sin 8)1(1 + sin 8) (2.43)Taking logarithms of both sides of Equation 2.40 and solving with the stress dilata.ncy equation ofRowe, 1971 (Equation 2.39), it is possible to derive 2 equations that respectively relate 4 and 6 to the slopeof the pressuremeter testing curve in a log-log (cavity pressure x strain) graph (see for instance Figure 10of Hughes et al, 1977). This constitutes the common interpretation procedure used to derive or 8 fromthe pressuremeter testing curve in sand.Robertson and Hughes, 1986 argued that in loose sand the expansion of the pressuremeter up to 10 %will not strain the soil sufficiently for it to reach the maximum dilation rate, and hence “c” (Figure 2.6)would be of significant importance. Therefore, they constructed nomograms based on the drained simpleshear behavior of Ottawa sand to empirically correct Hughes et al, 1977 model for its usage in loosegranular materials. The empiricism built in this correction, however, may erase any eventual gain ofaccuracy for the derived 4,24Carter et al, 1986 extended the Hughes et al, 1977 model to incorporate elastic strains in the plasticzone. The same basic input variables of Hughes et al, 1977 were adopted, with the additional requirementfor Poisson’s coefficient of the sand. Carter et al, 1986 assumes that the total volumetric strains in theplastic zone are a combination of plastic dilational strains related to the shear strain and elasticcompression strains related to changes in the average normal stress in the horizontal plane. The model doesnot differentiate between shear moduli used in either plastic or elastic zones, ignoring the fact that the soilstiffhess will increase in the former zone due to increase in am.Similarly as in Hughes et al, 1977, the Carter et al, 1986 model is developed on the basis of Cauchystrains (small strain definition). The particular solution for the pressure versus expansion at the cavity wallis given by:-‘XX /((yl icy= C Al I +Bl +C (2.44)RRwhere:T —z 1+n (__2A= , B= , X= , T=211+ I, Z=1+n 1—17 1—17 n+17} n+7777=1—N, 2=(1—1))(N.n+1)—V(N+n)with the same aR and EOR as those ofthe Hughes et al, 1977 model.Carter et al, 1986 pointed out that the predicted pressure expansion curve of their model does notdiffer significantly from the prediction given by the Hughes et al, 1977 model, whenever high values ofrelative elastic stiffness (Gi/ah) are used and the expansion 50w is limited to 10 %. This is because the useof a high elastic stiffliess reduces the contribution of the elastic components of strain in the plastic zone.This model can not be used under the same log-log interpretation methodology given by Hughes et al, 1977to derive friction and dilation angles.In 1989 Manassero presented an analysis similar to the one developed by Wroth and Windle, 1975. Afinite difference technique was adopted to numerically derive the curve of mobilized stress ratio versusshear strain of an element of soil in the cavity wall, as well as the stress path in the P vs. Q diagram ofLambe and Whitman, 1979 (where P’= (ar+ae)12 and Q = (ar-ao)12) for this same element. The volumetricstrain-shear strain relationship is also obtained. This numerical technique simultaneously solves, at eachstep, the stress dilatancy equation of Rowe, 1971 and the differential equations of the equilibrium of25stresses and compatibility of strains around the cavity. It considers the non-linear nature of the volumechange during shear, so that tests in both loose and dense sands can be analyzed. The only inputparameters are the complete 0r versus s (testing curve) and j, as the model does not consider the initialpre yield linear elastic idealization, i.e. only volumetric dilational plastic strains are considered.The numerical solution presented by this author consisted of:ç(i)=A—B+C +D (2.45)where:A= r(j)[eow(i1)+Kacvr(j1)1 B=2[Or()(1 + Kacv)—3r(1 - 1)] ‘ 2[Or()(1 + Kacv)—(7r(1 1)]C C — Jr(j1)[8r(j—1)+(1 +Kacy)86w()12[Kacv.(Yr( —1)] ‘— 2[Kacv. 0r0 - 1)]With Kacv = l/K<,. The circumferential stress at the cavity wall is obtained by:o(i)Yde_‘1o (i) = r rw I (2.46)K “dcv)\.. 6wde dewhere: de_ = 1 + __! and de6w =1- —i.dy dythe symbols d6v and dy are the increment ofvolumetric and shear strain components at the cavity wall.They are found in the step by step numerical solution by:de =e(i)-e(i-1),dy= y(i)-y(i-l),withE(i)e(i)+e6(i),and,y6w(i)-e(i)Using the results from “ideal installation” pressuremeter tests in calibration chambers, Manassero,1989 was able to show that under optimum conditions it is possible to derive more acceptable values of 4)from his model than from Hughes et al, 1977, with or without Robertson and Hughes, 1986 correction. Hebasically compared the predicted SBPM 4)’s (corrected to axially symmetric conditions) to the results of 4)from drained triaxial tests with the calibration chamber sand. With the “selfboring installation” testingcurves similar results were obtained in relation to the other traditional methods, most probable due todisturbance effects.26Howie, 1991 in a preliminary analysis of this numerical approach noted instability problems and ahigh sensitivity to the noise and shape of the testing curve. This finding corroborates the experience ofother researchers when dealing with finite difference techniques to solve the cavity expansion problem (seeconmients of Denby, 1978 and Benoit, 1983 in regard to the “subtangent” method), and appears to be adrawback ofthese techniques in general.Following the same approach as Manassero, Dc Souza Coutinho, 1990 proposed another very similarfinite difference technique to derive the soil parameters with a pressuremeter test. His model constituted animprovement of Manassero’s work in the sense that large strains (Almansi) and an additional rheologicalhypothesis (e.g. equation) regarding the variation of volumetric strain at cavity wall are considered.This latter author demonstrated in 2 high quality field testing examples that similar (but slightlyabove) ‘s as those predicted by the Hughes et al, 1977 log-log approach could be obtained, with the addedadvantage that the stress path and the volumetric behavior of the sand could also be detenuined. This finitedifference methodology of analysis, however, incorporates the same instability and sensitivity featurescommented on before for Manassero, 1989 when applied to slightly disturbed testing curves.Using the concepts put forward by many of the above mentioned models, Juran andMahmoodzadegan, 1989 developed a complex model to simulate the cavity expansion in sands. Anelasto-plastic stress strain idealization with both contraction and dilation during shear was assumed. Thesoil was also considered as a homogeneous, isotropic and strain hardening material with a non associatedflow rule. These authors conducted laboratory “pressuremeter type” expansion tests (in a hollow cylindercell) on Fontainbleau sand specimens, aiming at both the simulation of the ideal SBPM test and evaluationof the capabilities of their model. A reasonably good agreement between experimental and idealizedpressure expansion curves was noticed, with higher values of the plane strain 4’s in comparison to theaxially symmetric values of the triaxial tests. The drawback of this model is the requirement of extensivelaboratory triaxial tests to furnish parameters for the flow rule of the model. Moreover, a step by stepdetermination of the slope of the loading pressuremeter curve is required in order to incrementally obtainthe shear stress-strain curve at the cavity wall. Therefore, the same instability problems noticed for themethod of Manassero, 1989 may exist in this case.27The most recent analytical model for the interpretation of drained selfboring pressuremeter results wasput forward by Ferreira, 1992. An hyperbolic constitutive law between the stress ratio (QIP’) and thecircumferential cavity strain with a linear relationship between dilative volume change and shear strain,was adopted to simulate the loading stage of the pressuremeter test. An additional relationship between thecavity strain and the radial coordinate r was included in the solution, following the equation proposed byWroth and Windle, 1975. The Mohr Coulomb failure criterion was assumed to define the relationshipbetween the ultimate shear and the nonnal stresses of the hyperbolic curve, and Rowe’s stress dilatancyequation was applied to establish a link between both the dilation and the ultimate hyperbolic friction angle.Although the hyperbolic model is able to closely capture the nonlinear nature of the stress strainbehavior of sheared sands, the preliminary assessment of this model using results from selfboringpressuremeter tests performed on a calibration chamber with Ticino sand (Bellotti et al, 1987) was notsatisfactory. The predicted plane strain friction angles and lateral stresses of the hyperbolic model wereunderestimated in relation to the reference values related by Bellotti et al, 1987 (plane strain 4) bylaboratory tests and 0h from the chamber boundary stresses). The predicted shear moduli wereconsiderably overestimated, being on average 4 times the low strain modulus of the tested sand (assessedby resonant column tests).Cavity contraction theory was also proposed to predict parameters from the sand. The basis of thecontraction theory is the fact that there is a consistency on the shape of the unload curve of SBPM tests insands. This consistency is evident when comparing tests by different pressuremcters in the same sand.Although the loading stages of the testing curves are different, the shape of the unload stages are similar. Itappears, therefore, that the unloading stage of the pressuremeter test is much less sensitive to any initialdisturbance caused by the installation of this probe.Houlsby et al, 1986 were the first authors to introduce the cavity contraction theory to analyze theunloading stage of SBPM tests in sands. Plane strain, cylindrical coordinates and an elasto-plasticbehavior with volume change ruled by Rowe’s stress dilatancy theory was adopted in a similar fashion asHughes et al, 1977. Despite the high potential of this model, preliminary interpretation of SBPM tests inUnited Kingdom revealed inconsistencies for the derived loading friction angle. In fact the resultsdemonstrated that conservative estimates of this variable are obtained by setting the unloading friction28angle to.This led Houlsby et al, 1986 to comment on the necessity of a large strain formulation toimprove their model. Withers, et a!, 1989 after the analysis of several pressuremeter tests in sandconcluded that simple cavity contraction models are not suitable to derive strength parameters from eitherfull displacement pressuremeter (FDPM) or SBPM tests. This is due to the extremely complicatedbehavior of the cavity during unloading.In summary, several different cavity expansion models exist and can be applied to the interpretation oftests in sands. Table 2.1 shows the basic fbatures and problems of all the aforementioned models. Modelsthat encompass the shear volume strain coupling characteristic of the granular material under shear have tobe considered in the SBPM interpretation. All the models require some input from laboratory tests, butsome of them (like Ladanyi, 1963, VesIc, 1972 and Juran and Mahmoodzadegan, 1989) require complexlaboratory variables to define the stress-strain-volume change behavior adopted in the rheologicalequations. This latter necessity constitutes a drawback since the problems inherent to the in situ densityestimation and sample disturbance may be present with granular materials.Some of the above models can be solved by an analytical manner (Hughes et al, 1977, Carter etal, 1986, Ferreira, 1992) where closed form solutions are used. Others are applied in a numerical finitedifference technique (Wroth and Windle, 1975, Manassero, 1989, De Souza Coutinho, 1990) that directlyuses the testing data points to obtain the shearing characteristics of the sand. This latter interpretationapproach gives more information regarding the stress-strain-volume change behavior of the sand than theinterpretation approach associated with the closed form solutions. The experience gathered so far tovalidate the finite difference approach is mainly derived from “ideal installation” SBPM tests in calibrationchambers. Little experience exists with the finite difference interpretation method, but preliminary analysesindicated that both numerical instability and high sensitivity to even small disturbance in the pressuremetercurve may constitute an impeding factor for its usage in practice.The model of Hughes et al, 1977 is most commonly used to interpret SBPM results in sands. It is asimple and easy to use model that incorporates the main (simplified) features of sand behavior in simpleshear. The Carter et al, 1986 model is a refined version of the Hughes et al, 1977 model, but is notcurrently used in the interpretation of pressuremeter results in sands. This is probably due to the fact that itcan not be incorporated into the traditional log-log approach to derive . Both models assume a constant29dilation rate on the onset of yielding, thus not taking into consideration the possible strain softening of thesand during the expansion process.The models that have the best potential for use in the new interpretation methodology are those whichsimulate the soil behavior in a simple but reasonably accurate manner, do not depend on extensivelaboratory input data, and ideally have closed form solutions that are easy to apply. At present, thesemodels are those put forward by Hughes et al, 1977 and Carter et al, 1986.2.2.6 New Cavity Expansion Model2.2.6.1 IntroductionUsing the Hughes et al, 1977 or any other model it is possible to predict the whole idealized testingcurve of the pressuremeter test (at the cavity wall). This is accomplished by combining Equations 2.24 and2.4 1(in the case of the Hughes et a!, 1977 model), respectively valid for the elastic and plastic zones. Byvarying the input cavity strain Ee in these equations it is possible to obtain the cavity pressure a at anystage of the expansion. It shall be observed, however, that:• For a given mobilized strain e (or y) below EOR (or y) the curve is only ruled by Equation 2.24. Inthis case the medium is entirely in the elastic range and a linear relation between a and 8O. is obtained.• For a (or y) above 80R (or ‘ye) the curve is given by Equation 2.41, as an annulus of yielding soil(plastic zone) will start to expand at the cavity boundary, in accordance with the assumptions of this model.This approach was not followed by Hughes et al in 1977 to simulate pressuremeter testing curves, dueto the extensive mechanical calculations needed and lack of personal computers at that time. Instead, the“log-log approach” was adopted to predict 4> and ö. Only in 1989, with the use of fast personal computers,Dr. J.M.O. Hughes (personal communication) was able to write the first computer program that simulatespressure expansion curves with the equation of the model that he developed 12 years before. The purposeof the 1989 aforementioned computer program was to use the Hughes et al, 1977 model under a newconcept of pressuremeter data interpretation, to be described in section 2.4 of this thesis.Using the aforementioned program it was possible for the writer to directly compare the predictionsgiven by Hughes et a!, 1977 and the Carter et al, 1986 models, with the same input parameters. A newcavity expansion model was developed after it was found that different pressure expansion curves wereobtained for the Hughes et al, 1977 and Carter et al, 1986 models, when a Poisson’s coefficient of 0.5 was30used in the Carter’s model. This finding may indicate some inconsistency in Carter’s model, in the sensethat elastic strains in the plastic zone are not fully erased in the limiting condition2of v.The new cavity expansion model was originally suggested by Dr. J.M.O. Hughes in 1993 (personalcommunication) for use in the interpretation of pressuremeter curves in sands. With his assistance it waspossible for the writer to further develop and implement this model into a computer program. Thisprogram predicts the pressure expansion response at the cavity wall (testing curve) for a given set of inputparameters.2.2.6.2 Basis of the New ModelThe new model relies on both the basic principles advanced in 2.2.1 and 2.2.2 and the rheologicalequations of Hughes et al, 1977. This model focused the implementation of the traditional Hughes etal, 1977 model with the concept of elastic strains in the plastic zone put forward by Carter et al, 1986. Thenew model, however, allows the adoption of different shear moduli in both elastic and plastic zones aroundthe probe.The basic assumptions of the new model are the same as those presented by Hughes et al, 1977, withthe incorporation of elastic strains in the plastic zone. In this zone the total volumetric strains are definedby the combination of dilative plastic strains, caused by plastic shear, and compressive elastic strainscaused by the increase in the average normal stress.The elastic strains are defined with the aid of the elasticity theory presented before, and can beestimated by either a numerical (interactive) or analytical approach.2.2.6.3 Derivation of Elastic Strains in the Plastic ZoneApproximate Numerical SolutionIn subsection 2.2.3 it was shown that a plastic zone develops once yielding starts at the cavity wall.This zone increases as the cavity pressure increases, leading to a simultaneous increase of both normal andshear stress in the horizontal plane. This was shown in Figures 2.3 and 2.4. The high values of averagenormal stress Gm can compress the sand inside the plastic zone imposing a compressive “elastic” volumetricstrain, in opposition to the dilative “plastic” volumetric strain induced by the increase in shear stress tm inthis same zone.2 limit corresponds to the case where no compressive volume changes take place, hence the compressiveelastic strains given by the increase in the stress level should be null.31Carter et al, 1986 developed a closed form solution to deal with this particular condition of the cavityexpansion process in sands, but as commented above some inconsistency appears to exist with this model.The following solution overcomes the apparent deficiency of Carter’s model, and allows for a variation ofelastic parameters in the plastic zone different from the elastic zone. In the following approach the elasticstrains are determined by an approximate numeric (interactive) method.Consider a particular stage of the cavity expansion process, where the initial cavity radius r0 isexpanded to a defonned radius r and a plastic zone of radius r (say 3 ro) is developed around the cavity.At this stage, depicted in Figure 2.7(a), the expansion takes place under an increment of effective internalcavity pressure da (or a-ah) above the initial horizontal stress of the sand. The radial displacementmeasured at the cavity wall, defined by u in Figure 2.7(b), will be the sum of all the plastic and elasticradial displacements that are induced within the plastic zone plus the elastic deformation of the surroundingelastic zone. The total plastic radial displacement at the cavity wall (u) is derived by the directapplication of Hughes et al, 1977 Equation 2.41 in the whole plastic zone, and represents the plasticdilation of this zone. This variable also includes the elastic deformation of the elastic zone. The totalelastic radial displacement measured at the wall due to elastic compression of the plastic zone (Urew) can beapproximately assessed by assuming that the zone of soil encompassed between r0 and r compresseselastically. The stresses inside the plastic zone are defined by the Mohr Coulomb failure condition.In the subsection 2.2.2 the elastic solution was presented for the expansion of the cavity in an infinitemedium. It was shown that the differential equation that rules the elastic distribution of displacements inthe medium surrounding the cavity (Equation 2.16) has a solution in the form of Ur = Ar +BIr.The total elastic radial displacement at the cavity wall, for the particular stage of cavity expansion ofFigure 2.7(a), is given by the substitution of the constants A and B into the above equation of Ur. At thisstage of expansion the boundary conditions are given by the conditions existing in both cavity (r ro) andelasto-plastic (r = r) boundaries. This latter boundary, on the other hand, is assumed to be rigid at thisparticular moment, such that the elastic radial displacement distribution varies from a high value at wall tozero at r.32(A) ELASTIC AND PLASTIC ZONES AROUND THE CAVITY00G)0)>0LUark .dr=vw=ureM+ tirpwElastic Distribution(Equation 2.24)(B) IDEALIZED RADIAL STRESS DISTRIBUTION INSIDE PLASTIC ZONE0 A QUARTER SECTION OF EXPANDING CAVITYrp=3roELASTIC ZONEPLASTIC ZONEruro/darw\‘ Elasto-PlasticBoundaryfo ru r=3roRadiusPlastic Distribution(Equation 2.36)ro ru r=3ro RadiusFigure 2.7: Particular Stage of Cavity Expansion in Sand33This assumption leads to the following equation:= Ar + BIrD 0, hence Ar -B/rn (2.47)at the cavity wall r = r0 = r r and the following equation is obtained:Ar+ BIr = A.r + B/re (2.48)and hence dujdr = A -BIr2 = A -BIr2 (2.49)Equations 2.48 and 2.49 can be substituted into the (linear) elastic rheological equation (Equation2.13) that combines Gr, Ur and dur/dr of the expanding cavity. Therefore:—(1—U) ( B_________I BdY El A—— I— El A +— I (2.50)(1—2U)(1+U) ç2) (1—2U)(1+U)rearranging terms and substituting E/(2(1+v)) for G yields:-2G I B 2.v.Bd’Y = IA——+ I (2.51)(1—2U) r2 2 ,)substituting now Equation 2.47 into 2.51 and rearranging again the terms it follows that:2d = (l)A[1+(12v)[]] (2.52)hence A— (1—21))2G 2r 2 +(1—2U)rU22rdçB=+ (1—21))22G(2.53)(2.54)Substituting A and B of Equations 2.53 and 2.54 into the equation of u leads to the derivation of theequation that predicts the total (from the plastic zone) elastic radial displacement at the cavity wall for thisexpansion stage. Therefore:dc3 r r (r2—r) 1u = (l—2v)l O U I (2.55)2G [ç+(1—2v)rJBy assuming that the soil encompassed between r0 and r behaves elastically, it is implicit that thedistribution of radial and circumferential stresses in this zone follow the elastic equations given in34section 2.2.2, i.e. Equations 2.24 and 2.25. As schematically shown in Figure 2.7(b) the “plastic” and“elastic” radial stress distributions are different between r and r, with exception of the inner boundary ofthe cavity (where r = rn). Thus, in order to “force” the derivation of a cavity Urew with a stress distributionthat closely resembles the plastic distribution depicted in Figure 2.7(b), the division of the zone of soilbetween r and r into a series of concentric rings shall be considered. Generalizing Equation 2.55 forseveral concentric rings between r and r, it follows that:dre(n)— 2 22G (r) +(1—2U)(r(fl_)‘where: ure(n) is the elastic radial displacement at a concentric ring with inner boundary at a radiusand outer boundary at a radius r(fl1).The variable daR() (or aR(fl) - ah) is the “plastic” increment of radial stress (above ah) in the innerboundary of the concentric rings. The general equation for aR(s) comes from Equation 2.36, as follows:(1—N)R() R(n-1) (2.57)For the purpose of determining the approximate (total) elastic radial displacement at the cavity wall,the outer boundary of each of the concentric rings is considered fixed. The elastic radial displacement ofthe inner boundary of each of the rings is calculated with both Equations 2.56 and 2.57 above. The (total)elastic displacement at the wall is the sum up of the elastic radial displacements induced in each of theconcentric rings.lii order to illustrate the above procedure let’s consider that, in the particular stage of the cavityexpansion of Figure 2.7(a), the zone of soil between r ( r0) and r is divided in only 2 concentric rings.This is schematically shown in Figure 2.8(a). The outer boundary of the external ring 1 (at theelasto-plastic boundary) does not displace. Equation 2.56 is applied by using the “plastic” distribution ofstress depicted for both rings in Figure 2.8(b), with the value of GR(1) from Equation 2.57. The value of theradial elastic displacement at the inner boundary of ring 1 (Ure(l)) is then determined, as presented in35rp=3roU)U)0)U)CuCuCu>C)0)w2/darw(B) DERIVATION OF RADIAL STRESS AND DISPLACEMENT AT EACH RINGci,QUARTER SECTION OF EXPANDING CAVITYroELASTIC ZONEElasto-PlasticBoundary3roRadiusro 2roH(A) DIVISION OF PLASTIC ZONEUrew= Ure(2 +U21(Elastic Displacement)ro 2ro 3ro RadiusFigure 2.8: Calculation of Elastic Radial Displacement at the Cavity Wall for a Particular Stage of Expansion36Figure 2.8(b). The same procedure is adopted for the ring 2. The outer boundary of ring 2 (at a radiusr = 2 ro) does not displace for the purpose of calculation of the radial elastic displacement (ure(2)) induced atthe inner boundary of this same ring. Equation 2.56 is again applied with the value of aR(2) defined byEquation 2.57. The total elastic radial displacement at the cavity wall will be the value of Ur2) added to anadditional displacement U*e(2), as shown in Figure 2.8(b).For strain compatibility reasons an additional displacement U*e(2) has to be considered at the cavitywall, due to the movement of the outside boundary (ring 1). To determine this additional displacement it isassumed, at this expansion stage, that the concentric (inner) rings move at constant volume. Therefore, forthe example of Figure 2.8 the displacement caused at the cavity wall by the movement of the external ring 1is given by:2. ic. r . U re( 1) 2. n. r . Ue(2), where r1 2r0 and r2 = r0r 2r* 1 0U e(2) —U 2 Ure(l) (2.58)r re(1) r2 0The final displacement at the cavity wall (at this stage of expansion) will be the value of Urew (Ure(2) +ue(2)) added to the value ofu, which is derived by the application of Hughes et al, 1977 Equation 2.41.For the example of Figure 2.8 this latter displacement is calculated in a single step by:(i-N)uTPW =%\ j (2.59)where: aR(2) is the effective radial stress at the cavity wall.The values of50R and 0R are respectively calculated by Equations 2.33 and 2.30.Once the value of u (= Urew + u) is obtained, the circumferential strain 80w at the cavity wall iscomputed by the use of Equation 2.2. This allows the derivation of a pair of testing coordinates (Eew, o)that represents the idealized pressure expansion response at the cavity wall when the plastic zone has aradius r= 3 r0.In Figure 2.8(b) it can be noticed that, with the incorporation of 2 rings between r0 and r, the idealized“elastic” radial stress distribution within the rings (points A-B, C-D) becomes indeed closer to the “plastic”radial stress distribution (points A-C-E). The higher the number of concentric rings between r0 and rp, thecloser the elastic stress distribution will be to a smooth function, approximating to the “plastic” stress37distribution. The division of the plastic zone into 30 concentric rings was used herein. The division of theplastic zone into 30 concentric rings is sufficient to get results within 2 % of the values obtained with 400(or more) concentric rings.As the cavity pressure increases, so does the size of the plastic zone. Therefore, the elastic componentof strain (as well as the total) at the cavity wall is recalculated for each increment ofr1>/r0. The computerprogram developed for the simulation of the cavity expansion with this numerical approach varies r to amaximum extension of 15 times r0, at which stage the cavity strains are usually greater than 10 %. Foreach r/r0 stage both the 6e and a are calculated, allowing the determination of the complete (idealized)testing curve of the SBPM for a given set of input parameters. This set is given by P, a, G, and v (asin the Carter et al, 1986 model) plus the shear modulus of the plastic zone. This latter modulus can behigher than the modulus of the elastic zone, due to stress level differences.Closed Form SolutionThe prediction of the elastic strains within the plastic zone can be also done with the aid of ananalytical solution.Figure 2.9 presents the stress and strain conditions at a particular stage of the cavity expansion. For aparticular soil element inside the plastic zone it is considered that the imposed (radial and tangential)stresses are given by Equations 2.36 and 2.37. These equations rule the distribution of the failure stresseswithin the plastic zone, with basis on the Mohr Coulomb failure condition.The imposed stress regime will cause an increase in the mean normal stress of the considered soilelement, leading to an overall decrease of volume. The (elastic) unit volume contraction of this element isgiven by:++e (2.60)where:= 0, assuming plane strain conditions.Combining Equations 2.6, 2.7 and 2.10 in the Equation 2.60 above it is possible to obtain:38ar+(dar/dr).drfpUrewaeFigure 2.9: Elastic Displacement at the Wall due to Compressive Volume Change of a Soil Element39e, =-[dOr(1—U2)—da6(v+ U2)}+{dO.o(1_u2)—do(u+ u2)J (2.61)where: dOr is the increment of effective radial stress imposed in the soil element above 0hdo0is the increment of effective tangential stress imposed in the element above 0hUsing Equations 2.36 and 2.37 in Equation 2.61 gives:= (1— 2 V)(N + l)(T) (2.62)where: E=2G(l+v) (2.63)da is the increment of effective radial stress at the cavity wall above 0hIn order to find the contribution of all the elements of soil encompassed between r0 and rp, hence thetotal elastic volume change within the plastic zone, it is necessary to integrate Equation 2.62 over the areaof this zone. Therefore:22r Tp —Totaled (1_2U2_U)(N+l)f I1 r.dr.dO (2.64)26 (l+v) j Jr0’or0ord0 r1_rN+1Total E = (1 —2 U). 2 Jr. “ (2.65)2G Nlbut the total volume change within this zone will be equivalent to:TotalE =2.Jr.r0.u (2.66)where: Urew is the total elastic displacement at the cavity wallhencerr N+1U=(l—2V)[ PrN ] (2.67)Comparison of SolutionsAs commented before, the new cavity expansion model adds the elastic component of cavitydisplacement to the plastic component, which is derived with the use of the Hughes et al, 1977 model.40The total elastic radial displacement at the cavity wall, at a given position of the elasto-plasticboundary (r1,/0), can be predicted by a numerical (interactive) or an analytical solution. In order tocompare the predictions by both solutions a numerical example was carried out. Figure 2.10 presents thepredicted testing curve obtained by each of the solutions outlined before. The same set of (elastic) inputparameters was used to generate the elastic radial displacements at the cavity wall. These parameters wereestimated as G =40 MPa and v 0.2.The effective internal pressure at the wall da was computed with Equation 2.36 for each position ofthe elasto-plastic boundary. The horizontal stress was assumed to be 100 kPa and the plastic radialdisplacements at the wall were computed with the Hughes et al, 1977 model. This latter model adopted afriction angle 4. of 45°, a constant volume friction angle 4. of 35° and a shear modulus of the elastic zoneof 20 MPa.The results presented in Figure 2.10 suggest that the numerical interactive solution matches the closedform solution with an acceptable accuracy. The agreement of the predicted testing curves would be evenbetter, if a higher number of concentric rings were used with the numerical solution (this solution adopted30 rings between r0 and rn). Nevertheless, for a practical purpose, any of the above approaches to computethe elastic strains in the plastic zone can be used.It is recognized that there are alternate methods of calculating the radial displacement at the cavitywall that may not agree with the methods described herein. This arises because a consistent elastic-plasticapproach was not used to calculate the displacements. However it is felt that the approach used suitablyaccounts for the elastic and plastic volume changes.in the following sections of this thesis the new cavity expansion model is defined as the model thatadopts the numerical (interactive) solution to compute the elastic strains in the plastic zone.2.2.7 Comparison Between Cavity Expansion ModelsUsing the computer program developed for the new cavity expansion model it was possible to comparethe idealized predictions of cavity pressure versus expansion of this model with the predictions given by411400-n200-0- Numerical (interactive) Solutioni 000 Closed Form Solution800-O 600- -Ca)400-a)>0 200-0)‘4--OBS: In the above predictions the plasticstrains were computed with Hughes et at, 1977model. These strains are the same for each case0— I I0 2 4 6 8 10Circumferentiai Strain (%)Figure 2.10: Comparison of Predictions given by the Numerical and Analytical Solutions to ComputeElastic Strains in the Plastic Zone42distinct rheological models. The same set of input parameters was chosen for the comparison between thenew, Hughes et a!, 1977 and Carter et a!, 1986 models. These parameters are given by the strength, thelateral stress and the stiffness of the sand.These cavity expansion models consider a stiffness “G” which governs the soil response in the“elastic” zone (see Figure 2.3). This stiffness is defined by Equation 2.26, and for ideal elastic soils willrepresent the slope of the initial stage of the testing curve. This stiffness is defined herein as “Gi”, or theinitial elastic shear modulus of the sand. Both the new and the Carter et a!, 1986 models take onconsideration the stiffliess “G” in order to predict the soil response in the “plastic” zone (see Figure 2.3).This latter stiffness is used in both models to compute the elastic strains within this zone. For Carter etal, 1986 model a unique value of “G” is defined for both elastic and plastic zones, whereas in the newmodel a distinction between shear moduli can be done. The shear modulus in the plastic zone is definedherein as “Gp”.The input soil parameters adopted for the comparison of the models are the same as those used togenerate the plots of Figure 2.10, i.e.: 4) 450, 350 0h 100 kPa, Gi = 20 MPa, Gp 2Gi 40 MPaand v = 0.2 as well as 0.5. The Hughes et al, 1977 model does not consider elastic strains in the plasticzone. This model adopted Gi only to compute the initial elastic displacement of the cavity wall caused bythe pure shearing process that takes place in the elastic zone. In the Carter et a!, 1986 model the samevalue of Gi was adopted in both elastic and plastic zones. The new model simulated the cavity expansionby considering (a) same G (= Gi) in both plastic and elastic zones, and (b) Gi in the elastic zone with Gp inthe plastic zone. A Poisson’s coefficient v varying from 0.5 to 0.2 was adopted in both the new and Carteret a!, 1986 models, in order to see its influence over the final testing curve.The relationship between Gi and Gp is based on the average results obtained by the writer with theinterpretation of field SBPM data, assuming Gp Gur (unload reload shear modulus from the testingloop). These results will be presented in Chapter 4.The comparison of the pressure expansion curves for all the models with the aforementionedparameters is presented in Figure 2.11. The following observations apply:1. With the incorporation of elastic strains in the plastic zone (v < 0.5) there is a softening effect overthe predicted testing curve. Compare curves A-B with A-E or A-F.200CLD(1)(I-)0431400Cavity Expansion ModelsH77 Hughes et al, 1977C86 = Carter et a!, 1986NM = New ModelB,CDE,FC1000800600400200A0C)cC>C)4-4-LU00000000DOH77C86,C86,NM,NM,NM,(A—B)v 0.5 (Gi in both zones= 0.2 (Ci in both zones= 0.5 A—C)ii = 0.2 Gp, Ci in each of the zones)0.2 Ci in both zones) (A—C)(A—D(A— F2 4 6Circumferentia’ Strain (%)Figure 2.11: Comparison of Pressuremeter Curves Predicted by Different Models442. The new and Hughes et al, 1977 models lead to similar results when the Poisson’s coefficient isequal to 0.5 (compare curves A-B and A-C), This is not observed for the model of Carter et al, 1986(compare curves A-B and A-D), and perhaps indicates that this model incorporates some other elastic(strain) variable in the plastic zone.3. With the new model the incorporation of the same shear modulus (Gi) in both elastic and plasticzones leads to a very soft pressure expansion curve (curve A-G). Considering a higher stiffness in theplastic zone (G Gp), it is possible to obtain a stiffer pressure expansion response at the cavity wall(curve A-E). Indeed, since in the plastic zone the level of stresses are higher than in the elastic zone, it ismore appropriate to use a higher stiffhess (than Gi) to compute the elastic strains in this former zone. Thisis so given the recognized stress level dependency of the shear modulus (to be addressed in thesubsection 2.3.4). The stiffliess in the plastic zone can be approximately measured when performing anunload reload loop during the test, since the testing loop is generally carried out at a stage where anexpanded plastic zone exists around the cavity. The determination of the unload reload shear modulus Gurfrom the testing curve is discussed in subsection 2.3.4.1.4. Surprisingly, Carter et al, 1986 and the new model gave identical testing curves when a Poisson’scoefficient equal to 0.2 was adopted, and a stiffness Gp was adopted in the new model.In summary, the incorporation of elastic strains in the plastic zone leads to softer (idealized) testingcurves. Both the new and Hughes et al, 1977 models converge when a Poisson’s coefficient of 0.5 isadopted in the former model. On the other hand, this is not observed for the Carter et al, 1986 model. Thenew model has the capability to incorporate a larger sand stiffness in the plastic zone to predict the elasticstrains. This stiffness can be approximately assessed by performing an unload reload loop during the test.2.3 TRADITIONAL iNTERPRETATION METHODOLOGIES IN SAND2.3.1 IntroductionWith the available theoretical models described in the previous section it is possible to derive thefriction angle, the shear modUlus and the lateral stress of the tested sand. The traditional interpretationmethodologies use the SBPM testing curve for this purpose.It is the objective of this section to briefly review these interpretation methodologies, addressing theirlimitations and the possible need of better alternative interpretation approaches.452.3.2 Friction AngleAs commented in subsection 2.2.5 the most common interpretation approach in sands for deriving thefriction angle is the use of Hughes Ct a!, 1977 model with the log-log plot of the testing curve. Robertsonand Hughes, 1986 nomogram can be additionally applied in loose granular materials. Finite differencetechniques were recently proposed for a more refined interpretation analysis of the SBPM testing curve.In order to assess the capability of the above interpretation methodologies, Bellotti et al, 1987conducted 47 SBPM tests in the Italian ENEL-CRIS calibration chamber using both Ticino and Hokksundsands. Pressuremeter tests were performed with the probe cast in place (ideal installation) as well as byselfboring into the sand. The English Canikometer with an L/D =6 was used. The samples were subjectedto 1D consolidation under K0 conditions and then unloaded, where K0 is the coefficient of earth pressure atrest. Both normally consolidated (NC) and overconsolidated (OC) sand specimens were tested. Triaxialcompression tests with the same sand under similar density and confinement conditions as those used in thechamber were performed in order to define baselines of 4. The triaxial ‘s were further corrected to bothstress level at fuilure (by the curved strength envelope equation of Baligh, 1976) and to plane strainconditions (by the empirical equation of Lade and Lee, 1976). Ring shear tests were used to derive theconstant volume ffiction angle required by the interpretation methodologies. The summary of theresults obtained by these authors is shown in Table 2.2, where the following observations apply:1. In general all the interpretation methodologies lead to average values of 4’ that are close to eachother, with some scatter (average ± 3.9°) in the predicted 4”s. The scatter of the ideal installation tests isconsiderably lower than the scatter found in the selfbored tests, suggesting that for the ideal installationtests the accuracy of the predicted 4”s by the log-log or the other interpretation methodologies is higher thanthe accuracy found with the selfbored tests. Indeed, Jewel! et a!, 1980 showed that Hughes et a!, 1977model can predict friction angles by the log-log approach that are extremely comparable to baseline valuesfrom laboratory simple shear apparatus. This is only the case if high quality (ideal installation) testingcurves are used. Eldridge, 1982 also demonstrated that in the absence of any disturbance the 4’ predicted bythe log-log approach of Hughes et al, 1977 is extremely reliable. He used a plane strain finite elementmesh, with a model that incorporates the shear volume coupling of the sand with a non linear incrementalCONDITIONREFERENCESBPMiNTERPRETATIONMETHODOLOGIESREFERENCE4)(Deg)(Deg)(A)(B)(C)Ideal Installation.OCR=143.5±1.540.1±2.443.1±2,241,6±3.2Bellottietal,1987IdealInstallation,OCR>144,7±2.437.2±3.740.9±2.637.5±4.1CalibrationChamberTests-CanikometerTicinoSand-Dr=401o71%Seitbored,OCR=141.9±0.639.9±3.642.8±3.0NASeltbored, OCR>143.5±3.443.7±9.646.1±7.742.8±8.5CALIBRATIONCHAMBERBorehole400339.1±1.139.8±7.043.1±5.1NABruzzietal,1986Borehole401737.8±1.330.3±3.436.4±2.2NAFieldTests-CanikometerBorehole505037,7±1.646.7±9.848.9±7.7NAPoRiverSand-LoosetoMediumDenseSandDepths6.2to23mFIELDTESTS(A) Hughesetal,1977modelwiththelog-loginterpretationapproach(B) RobertsonandHughes,1986empiricalcorrectionof(A)usinganomogram(C)Manassero.1989modelwithfinitedifferenceinterpretationapproachOBSERVATIONS:Data±Istandarddeviation.Reference(planestrain)4)ofBellottiCtal,1987fromLadeandLee,1976correlation,usingconventionalTriaxialcompressiontestingresults.Reference(axiallysymmetric) 4)ofBruzzi etal,1986fromDurgunogluandMitchell,1975theorLusingdatafrompiezoconetests.NANotAvailable.Table2.2:ComparisonofFrictionAnglePredictedbyDifferent Traditional InterpretationMethodologies47dependency of stiffness to confining pressure, to simulate the cavity expansion in sands. Finite elementsimulated pressure expansion curves were analyzed by the Hughes et al, 1977 log-log approach, predicting4)’s that were closer (within ± 10 %) to the input c’s of the finite element program.2. The scatter of 4)’s observed in the selfbored tests was mainly due to disturbance. For instance, themajor effect of disturbance on the traditional log-log approach of Hughes et al, 1977 is in regard to the nonlinearity of the log-log curve. Disturbance makes it difficult to define a unique slope in the log-log plot,hence a unique and reliable 4). The highly scattered results of Bellotti et a!, 1987 for the selfbored testssuggest that field pressuremeter curves, that invariably have some disturbance built in, are not suitable forthe prediction of 4) by any of the above traditional interpretation methodologies. On the other hand, if thefield curve is of extremely high quality, then the prediction of 4) will be accurate. According to Bellotti etal, 1987 none of the methods above provide a reliable estimate of 4) for sands from the SBPM, although themethod of Robertson and Hughes, 1986 produces the lesser amount of scatter.The results of Bruzzi et al, 1986 can be used to check the suitability of field SBPM curves for thedetermination of the friction angle. These authors carried out a similar program of SBPM tests to assessthe above interpretation methodologies. 53 SBPM tests were performed with the English Canikometer inthe aged granular deposit of Po River sand. The reference 4) values were derived from empiricalcorrelations with piezocone soundings (Durgunoglu and Mitchell, 1975 theory) which, according to theauthors, are only 1 to 20 lower than the 4)’s of triaxial tests. The ring shear was used in the SBPMinterpretation methodologies. The results of these authors are also shown on Table 2.2 and, similarly as inthe previous case, highly scattered values were obtained in all the boreholes tested. It is noticed in thistable that the scatter of the field results is of the same order of magnitude as the scatter obtained by Bellottiet al, 1987 with the OC selfbored chamber tests.According to Bruzzi et al, 1986 the Hughes et al, 1977 log-log interpretation approach leads to highlyscattered values of 4) that are generally too low but sometimes too high. The correction proposed byRobertson and Hughes, 1986 works in the right direction, but still does not yield consistent values of 4).Bruzzi et al, 1986 speculate that the major difficulties in predicting 4) from the SBPM are linked to factorslike the inadequacy of the constitutive relationships, the curvilinear strength envelope of sands and possiblythe finite length of the probe. On the contrary, they feel that the scattered 4) results are not caused by48disturbance effects, since their SBPM tests did not show signs of “large initial disturbance”, as, forexample, in Hughes et a!, 1977 and Robertson and Hughes, 1986.Based on the review of this subsection it is concluded that the 4) predicted by the current SBPMinterpretation methodologies is far from a reliable parameter if disturbed testing curves are interpreted. Onthe other hand, reasonably accurate results of 4) are obtained with high quality testing curves.For the disturbed curves, the difference of the results for different constitutive relationships was notobserved to be appreciably high, which suggests that the simplified assumptions of the traditionalrheological models are not the major factor responsible for the failure of the analysis to properly estimate arepeatable and reliable 4). Disturbance, numerical instability of the finite difference interpretation approachand the subjectivity built in the derivation of slopes from the log-log method are, without question, themajor variables that influence the current predictions of 4) from SBPM tests.There have been, however, improvements on the log-log approach of Hughes et al, 1977 to takeaccount of the interpretation of disturbed curves. Mair and Wood, 1987 suggested a trial and errorprocedure to chose a correct reference strain (related to Oh) to be used as datum in the log-log plot ofpressure versus strain, such that this plot would become a straight line over a larger strain range of the test.Fahey and Randolph, 1984 suggested a similar approach, but proposed as a first basis of datum strain theuse of the strain at which the cavity pressure is equal to the total vertical stress at the test position.Nevertheless, those corrective approaches are cumbersome to use and may not lead to a substantialincrease in the reliability of the interpretation of disturbed SBPM data.2.3.3 Horizontal StressThe lateral stress is a parameter of primary importance in soil mechanics and its evaluation hasalways been a matter of concern. The lateral stress is commonly used with the vertical stress to derive thein situ coefficient of earth pressure at rest (K0) ofthe deposit.The current approach followed by the SBPM test in sands is the direct measurement of the originalhorizontal stress of the deposit. For that, the only method available consists on the visual determination ofthe lift off pressure at each of the strain arms of the SBPM probe (Mair and Wood, 1987). The lift offpressure (P0) is defined by the pressure at which a “break” or substantial change of slope occurs in theearly stages of the testing curve. The lateral stress of the sand is commonly assigned to the average value49of the lift off pressures defined in each of the strain arms of the SBPM. The selfboring lateral stress ratio(K Oh/Ov) is commonly quoted as the value closest to K0. Therefore this value is currently used as areference basis for all other in situ estimations of lateral stress.Empirical relationships can be used as a basis to form opinions regarding the reliability of the SBPMah measurements. However, these empirical relationships do not always account for all the intrinsicvariables that affect the lateral stress in a natural environment. An alternative approach to evaluate SBPMah results is the interpretation of tests performed in a controlled environment, such as inside a calibrationchamber. Although the chamber environment can not fully simulate the soil conditions in situ, theboundary stresses are very well known.Therefore, the results of the chamber tests performed by Bellotti et al, 1987 were also used here toassess the capability of the SBPM to predict ah with lift off measurements. For that purpose, the averageP0 measured by the 3 strain arms of the Canikometer was compared to the boundary stress Ob of thechamber. For the ideal installation tests the lift off stresses were significantly different than the appliedchamber boundary stresses. The reasons cited by these authors were the stress concentration around therigid SBPM, the mechanical compliance of the strain arms and arching effects caused by the presence of anannulus of looser sand around the pressuremeter. The possibility of stress concentration was investigatedwith a unique ideal installation test where a rigid selfboring K0 cell was used. Apparently little or no stressconcentration was noticed. This led to a redesign of the strain arms of the Camkometer used by theseauthors. A new series of tests were then carried out with the new design, as indicated by the results of thetop plot of Figure 2.12. In this plot it is noted that a small difference of ± 10 to 15 % between P0 and theboundary stresses was found. It appears, therefore, that for high quality testing curves it is possible todefine accurate predictions of lateral stress with lift offmeasurements.The lower plot of Figure 2.12 shows the results for the selfbored tests of this same authors. As can beseen by this plot the measured average lift off pressures were lower than the applied chamber stresses.According to Bellotti et al, 1987 the lift offs were often close to the water pressure of the chamber,indicating that significant disturbance occurred during the probe installation, especially in the loose andmedium dense sands. The average ratio P0/Oh of the selfbored tests was 0.47 ± 0.28, indicating that even50LUDU)U)LU0LiLI0U-JLU0LUaLiiDU,Lii.3P0 (kg/cm)(b)MODIFIED STRAIN ARMSDEAL INSTALLATION ho o(I)U)LUII-(I)LULU0LU-JLU(-:ILU2.00 1 2 3APPLIED BOUNDARY STRESS, ho (kg/cm2)P0(kPa)MODIFIED STRAIN ARMSSELF- BORED INSTALLATION200= P0..00..100 200 300 400aho kPa)APPLIED BOUNDARY STRESSFigure 2.12: Comparison of Measured and Applied Lateral Stress in Calibration Chamber (after Bellotti etal, 1987)51under controlled conditions selfbored SBPM tests render lateral stress predictions that are highly scatteredand generally unreliable. This appears to be the case when lift offmeasurements are used.In natural sand deposits, however, more reliable predictions of 0h may be obtained. According toBellotti et al, 1987 the sand tested in the Italian chamber (i.e. freshly deposited, unaged, uncemented, cleansand) may create particularly unfavorable conditions with respect to the reliable assessment of the in situstress. Clarke and Wroth, 1985, stated that the level of knowledge of the selfboring process with thetraditional system in sands has evolved considerably in the last years, making it possible to obtain validvalues of lateral stress by improved drilling techniques in the field. Thus, the field results of Bruzzi eta!, 1986 can also be discussed here to check the suitability of SBPM curves for the 0h determination ingranular deposits.Bruzzi et al, 1986 were able to optimize the cutter setting by sequential selfboring trials, such that“reasonable” values of in situ horizontal stress were obtained (Lacasse et al, 1990). The “reasonable”estimates of the Camkometer were considerably scattered and 20 to 50 % higher than the author’s bestestimate of the in situ 0h with basis on the empirical formulas that relates K0 with OCR (OCR estimatedfrom oedometer tests on embedded silt layers of the deposit). The results obtained from tests using theFrench PAF-76 SBPM in the same Po River sand deposit yielded lower values than the best estimates of0h. Most of the scatter observed by Bruzzi et al, 1986 came from the same reason as the one found in thechamber tests of Bellotti et al, 1987: the disturbance of the test. The above results suggest that a smallamount of disturbance in the pressuremeter curve is enough to cause a large scatter of lateral stressespredicted by the visual inspection technique. This is valid for selfboring pressuremeter tests in eithercontrolled (chamber) or natural (field) environments.In summary, the lateral stress determined by the visual inspection method (lift off) is unreliable ifdisturbed data is analyzed. Scatter in the results is prone to occur even if slightly disturbed SBPM testingcurves are analyzed. Lift off pressures can be accurately used only with undisturbed curves. It is claimedthat by a proper optimization of the selfboring insertion technique undisturbed curves and reliable values ofOh can be obtained. The experience gathered by Bellotti et al, 1987 in the calibration chamber and Bruzziet al, 1986 in the field dismiss that, suggesting that the sensitiveness, of Po to any disturbance generatedduring insertion is so high that the visual inspection technique can not be efficiently used in selfbored tests.52Therefore, in view of the limited experience for the evaluation of the in situ stress in sands by thevisual inspection technique (Lacasse et al, 1990), the ability of the SBPM probe to accurately determine 0hhas not been fully proven yet.2.3.4 Shear ModulusThe SBPM is one of the most promising in situ tools that can be used to directly measure the stiffuessof the sand. The shear modulus of the pressuremeter can be currently evaluated by a interpretationprocedure that uses different parts of the testing curve, as is described next.2.3.4.1 Unload Reload Shear ModulusHughes, 1982 and Wroth, 1982 noted that the elastic modulus could be directly obtained byperforming loops during the expansion phase of the test, with the advantage that the modulus derived in thismanner would be much less influenced by disturbance (Mair and Wood, 1987). As discussed insubsection 2.2.4, within the framework of the theory of elasto-plasticity any unload of the expandingcavity wall brings the surrounding soil below the currently expanded yield surface. This is illustrated in thetop plot of Figure 2.13, which shows a typical section of the testing curve with an unload-reload loop BCD.It is noticed on the stress path (bottom plot) of this same figure that the elastic soil during the unload stageis only subjected to variation in the shear stress level, provided that the maximum unload criteria ofWroth, 1982 is followed.The slope of the previously executed loop BCD corresponds to the shear modulus of the elasticsurrounding medium, as derived in Equation 2.26. Therefore the unload reload modulus is given by:ldP (2,68)2 de0where: dP is the effective cavity stress difference applied during unloadingd60 the circumferential strain amplitude at the cavity wallHowever, the deformation parameter can not be considered to be a constant and inherent property ofthe soil, but rather a complex function of several variables (Hardin and Drnevich, 1972) that actsimultaneously during the penetration of the in situ probe. This is so because in real sands:1. The deformation parameter is a function of the level of mean stress that exists surrounding the insitu device, as well as stress history imposed prior to and during penetration.53C’)C’)U)ci)-cU)1)0)(U0-ci)U)(I)ci)3-(UCci)CPowB=Dd69Circumferential Strain (%)Mohr Coulomb Failure Line8=0sin4Domain(ar-ao)12Average Normal Effective Stress (r+ae)I2Figure 2.13: Concept of Elastic Shear Stifffiess as Obtained from Unload Reload Loops (Modified afterBruzzi et a!, 1986)A542. The deformation parameter behaves in a non linear manner in regard to the level of in situ shearstress and strain imposed during probe penetration.Effect of Stress Level:In order to illustrate the stress level effects in Our let’s assume that the only distinction between theelement of soils surrounding the cavity prior and during unloading relates to normal effective stress (am)differences. Figure 2.14(a) shows a hypothetical testing curve in which two unload reload loops wereperformed. Two distinct shear moduli (Our1 and Our2) were measured, with Gur2 greater than Our1.Figure 2.14(b) shows the stress path of the soil (at the cavity wall) during the loading and unloading stages.For the loop carried out in BCD the soil has been loaded elastically along the line A-A’ and then plasticallyalong A’-B. It is noticed that at each different position of the testing curve (Figure 2.14(a)), a differentlevel of average normal (am) and shear (tm) stress is mobilized at the wall (Figure 2.14(b)).Figure 2.15(a) shows the elastic and plastic zones developed prior (and after) the unload stage of theloop BCD. It is noticed in this figure that, prior to the loop BCD, the elasto-plastic boundary will be undera level of (shear and normal) stress similar to those of the point A’ of the testing curve (see Figure 2.14).One element of soil at an infinite radius will be under a level of stress similar to those of point A of thetesting curve. The cavity wall will be under a level of stress similar to those of point B of the testing curve.Upon unloading all the zone of soil surrounding the probe responds elastically, and the previouslyplastic zone is encompassed by the elastic zone (see Figure 2.15(a)). During unloading, the stress paths ofall the elements of soil between A and B will follow a similar path as the one presented in Figure 2.14(b)for the cavity wall (path B-C). In this latter figure it can be seen that after unload the cavity wall will beunder the same level of normal stress as before (equal to Omi) but the level of shear stress will decay fromtml to t.Each element of soil between A’ and B in Figure 2.15(a) will have a different level of normal stress amprior to the unload stage, for instance element A’ will be under amo whereas B will be under Oml (seeFigure 2.14(b)). This implies that each element of soil will have a different modulus 0 that differs fromelement to element as a function of °m• This modulus will be constant for each of the elements, as the soilis idealized to respond in a linear elastic manner with a unique modulus. Since the level of normal stress55C30ci)DCl)Cl)ciCuCci)CPoci)uJCircumferential Strain (%)(A) Hypothetical Testing Curve(ar-3e)12Cl)Cl)ci)Cl)Ici)-cCl)ci)Average Normal Effective Stress (cYr+ae)12(B) Stress State for Different Points of the Testing CurveE=GB=DGur2> Gur1Mohr Coulomb Failure LineE=GB=DtmlsinFAFigure 2.14: Effect of Stress Level on Unload Reload Shear Modulus56QUARTER SECTION OF EXPANDING CAVITYLoading Stago Unloading Stage(r=inf,nity) (r=infindy)AELASTICZONE ELASTICZONENPLASTIC ZONE \ A’Esto-PIastru Soindary ru-drrO ru rOru-drRadius Radius(A) ELASTIC AND PLASTIC ZONES FOR THE UNLOAD-RELOAD LOOP BCDQUARTER SECTION OF EXPANDING CAVITYLoading Stage Unloading StageA IrnfinityI A Irinf’t)ci,ELASTIC ZONEELASTIC ZONEPLASTIC ZONE Esto-PIastic\\• BoUndaryru+drlru drFrO ru+drl rO ru-drRadius Radius(A) ELASTIC AND PLASTIC ZONES FOR THE UNLOAD-RELOAD LOOP EFGFigure 2.15: Plastic and Elastic Zones Prior and After the Loop Stage57decreases from the cavity wall (element B of Figure 2.15(a)) to soil elements located at a greater radiusthan r, the shear modulus will also decrease throughout the soil surrounding the probe. Thus the slope ofthe loop (i.e. Gur1)will reflect an “average” stiffness of the whole material surrounding the probe duringthe unloading stage, and shall be assessed with the knowledge of the average level of normal stress thatexists within the plastic zone prior to this stage.If the expansion of the cavity continues, the plastic zone increases to a radius greater than the radiusobserved prior to the first unload-reload stage (BCD). Figure 2.15(b) shows the plastic and elastic zonesdeveloped prior to (and after) the unload stage of the loop EFG. As shown in Figure 2.14(b), the cavitywall element E will be under a much higher level of normal stress than before, as a is greater than °ml•This implies that each element of soil between A’ and E (in Figure 2.15(b)) will also be under a higherlevel than the level that each element had prior to the first unloading stage. Consequently, prior to thesecond loop the average level of normal stress within the plastic zone is higher than the level that existed (inthe plastic zone) prior to the first loop. Given again the stress level dependency of the modulus, it is to beexpected that the measured Gur for the second loop (Gur2)will be higher than the modulus of the first loop(Gur1). This implies that Gur will increase in the course of a expansion test, i.e. the relative position of theunload reload loop. This effect is substantiated by Clarke and Wroth, 1985, which noticed that by plottingGur against mean effective stress, rather than depth, the scatter of results decreases considerably.The above influence of the stress level in Gur led Robertson, 1982 and Bellotti et al, 1989 to proposeexpressions to derive the average stress level 0av surrounding the probe (in the plastic zone) that exists priorto the loop stage, and hence obtain a corrected for stress level modulus (Gur0) with the Janbu, 1963equation. Robertson, 1982 proposed the first approximation of the average mean octahedral effectivestress that acts in the plastic zone, assuming that:= 0.5 P (2.69)where is the effective cavity pressure at the start of unloading, as defined in Figure 2.13.Bellotti et al, 1989 developed an analytical equation, based on the cavity expansion in an elastoperfectly plastic sand, to derive the average normal stress of the plastic zone. Their equation wasdeveloped based on the integration of the normal effective stress of each element of soil encompassed by theplastic zone.58Their final equation is:1.pC drICY.—I mr=r z. P , where % (2.70)f drJrwhere r is the radius of the plastic zone prior to unloading stage, and r the respective cavity radius.In order to simplif’ the above equation these authors have scrutinized the chamber results of Bellotti etal, 1987. Since all the variables, including the plane strain of the tested sand is known for Bellotti’s 1987tests, a modified equation for 0av was suggested:av = +a(P h) (2.71)where cx is an empirical reduction factor equal to 0.2.The deficiency of this method is that by integrating along a radius the expression fails to incorporatethe larger volume of soil subjected to increase in stress as the test progresses. Besides, the incorporation ofan empirical parameter cx leads to an approximate (not accurate) detennination of the average stress in theplastic zone. Nevertheless, Howie, 1991 argues that given the unknown effects of disturbance, anisotropy,etc. over Gur no justification can be found to refine even more the above corrections.Once a reference stress is obtained it is possible to normalize Gur to Gurc, where Gurc is the modulusat the in situ normal stress (horizontal stress Oh). For granular materials the relationship between elasticmodulus and stress level can be expressed in a similar way to that suggested by Janbu, 1963:G—=K I—--I (2.72)Pawhere: Kg is the modulus number,n the modulus exponentPa the atmospheric pressure.The value of n is typically in the range of 0.4 to 0.8 (Bruzzi et al, 1986 and Bellotti et al, 1989) and isdependent on the strain amplitude over which the modulus is measured (Wroth et al, 1984). For all59purposes, it will be assumed as 0.5 in this thesis. Using the Equation 2.72 above, and noting that Gur is anaverage modulus which is a function of °av and the normalized GurC is referent ah, it is possible to derive:Gur c = Gur (2.73)av ,1The above discussion indicates that Gur reflects an average stiffliess that shall be related to somemeasure of the relevant stress level around the cavity for its proper use in design. Empirical andsemi-empirical methodologies to assess this stress level do exist in the literature.Effect of Strain Level:In order to illustrate the effects of the strain level on Gur let’s now assume that the only distinctionbetween the elements of soil surrounding the cavity prior and during unloading relates to the levels ofinduced shear strain. Figure 2.16 presents the same testing curve as discussed before, with twohypothetical unload reload loops. The second loop is carried out to a degree of cavity strain unload higherthan the degree of the first loop and, in a opposite fashion as before, the shear modulus of the second loopGur2 is lower than the modulus of the first loop Gur1. The loops have a non linear shape, suggesting thatmore than one shear modulus can be defined. In this case, Gur is defined with the slope of the upper (B, E)and lower (C, F) “cross over” points of the loop.The non-linear behavior of the loop occurs because real cohesionless soils have an elastic thresholdshear strain above which the behavior is non-linear. Research carried out by Dobri et al, 1980 with sandsdemonstrated that the maximum modulus G0 is mobilized for shear strains below i0 % to 6x103 %. Atthis level of induced shear strain the soil behaves in a linear elastic manner, as hypothesized by the linearequations of subsection 2.2.2. For higher strain amplitudes, like those mobilized during the unload reloadloops, the behavior is non linear and the modulus varies according to the applied cavity strain ds0, i.e. theinduced shear strain at the cavity wall and surrounding soil. Figure 2.17 shows the experimental modulusreduction curve of granular materials presented by Idriss, 1990, based on the range proposed by Seed andIdriss, 1970 defined after the compilation of several laboratory investigations of shear moduli for sandysoils. The higher the induced strain the lower is the shear modulus of the material in the non linear range ofCucL-U)U)a)I‘3-Cua)•1-Cw60P0Circumferential Strain (%)Figure 2.16: Effect of Strain Level on Unload Reload Shear Modulus611 .00.40.00.0001jo.90.80(I)I.o 06c.0.5NI.° 0.300.2I.0‘.— 0.10.001 0.01 0.1Shear Strain (%)7Figure 2.17: Variation of Shear Modulus with Shear Strain for Sands (after Idriss, 1990)62stress-strain behavior, and vice-versa. During the unload stage (B-B’-C) of the first loop BCD all the soilelements between A and B in Figure 2.15(a) will be strained by a different amount, hence leading to adifferential decrease of the modulus along the soil. As noted by Whittle et al, 1992 (and Equation 2.18) fora linear elastic isotropic soil the distribution of strain around the cavity varies inversely with the square ofthe radius, meaning that elements of soil at the cavity wall will be much more strained than elementslocated at a higher distance from the wall. This implies that the modulus decay at the cavity wall will bemuch higher than the decay at the adjacent outer elements, leading to a much softer soil response at thecavity wall. The soil response will gradually become stiffer as the position of the soil elements increase inrelation to the cavity radius. This is the opposite effect to that observed in the case of the stress level, butsimilarly as before the modulus measured at the cavity wall will reflect an “average” stiffness of the wholematerial surrounding the probe.Since during the unload process the loaded soil becomes softer as a whole, given the overall increaseof shear strain, the measured modulus at the wall will constantly decrease. This leads to the non-linearrounded shape of the unload stages of the loops BCD (B-B’-C) and EFG (E-E’-F). Upon reloading thesame straining effect that happened before occurs, but at the opposite direction. This leads to thenon-linear rounded shape of the reload stages of the loops BCD (C-C’-B) and EFG (F-F’-E), where thesecant cavity modulus G constantly decreases.Moreover, given the strain level dependency of the shear modulus of the sand, the higher the degree ofunload at the cavity wall the higher will be the straining induced at each of the elements of soil surroundingthe cavity. Consequently, the softer will be the general response of the medium and the lower is themeasured modulus. Figure 2.16 schematically shows that the modulus of the second loop Gur2 is lowerthan the modulus of the first loop Gur1. This effect was also observed by Whittle et al, 1992 with SBPMtests in London clay. These authors noticed that by increasing the loop amplitude there is a decrease in themeasured shear modulus.Thus in order to make use of the pressuremeter unload reload shear modulus Gur the assessment ofthe average level of shear strain which corresponds to this modulus is also required. The assignment of anaverage shear strain y for a particular testing loop is a function of the behavior of the soil elements around63the probe prior and during unloading. Empirical or analytical expressions can be used with this objective,as those respectively proposed by Robertson, 1982 and Bellotti et a!, 1989.Robertson, 1982 proposed that an average shear strain amplitude Yav equal to 0.5Ay, where Ay, is theshear strain amplitude of the loop cycle in the cavity wall (= 2dE), shall be assigned as the strainincrement relevant to Gur. Bellotti et al, 1989 defined the relevant strain increment of Our as the averageelastic shear strain induced in the surrounding zone of soil during the unload stage of the loop. The zone ofsoil considered is the zone that was previously occupied by the plastic zone prior to loop stage. Idealizingthe sand as an elasto-plastic medium they were able to obtain this average induced strain with an analyticalequation. This equation is dependent on the P, the plastic radius r, the 4 and 0h of the sand and on thedegree of cavity unload. However, similar to the case of stress level, with the use of the calibration chamberdata of Bellotti et a!, 1987 they were able to simplify their equation to:‘yav = 0.5.f3.Ay (2.74)where: 0.5 is a factor to produce the single amplitude of shear strainf3 is an empirical reduction factor equal to 0.5.Howie, 1991 suggested that an average strain level equal to 0.15 could be assigned to Gur. Hebased this suggestion on the same chamber results of Bellotti et al, 1987, since by plotting Bellotti’s datawith this strain definition a very reasonable agreement could be obtained with the Seed and Idriss, 1970general envelope for sands.Therefore, as discussed above, the Our also reflects the integrated effect of the soil deformation alongthe expanding cavity, and shall be related to some relevant measure of the average induced strain level forits use. Empirical and semi-empirical methodologies to assess the relevant strain level also exist inliterature. This strain is in general of the order of 101 %, with basis on the extensive data gathered in thecalibration chamber by Bellotti et al, 1987 and in the field by Bruzzi et al, 1986.Combined Effect of Stress and Strain Level:The combined effect of both stress and strain levels induced around the probe prior and during theunload reload stage shall be considered for the rational use of the pressuremeter Our.Bellotti et al, 1989 attempted to develop a methodology that corrected Our for stress as well as strainlevel, in order to link the high strain modulus of the pressuremeter to the low strain modulus of the soil.64With the stress level correction of Equations 2.71 and 2.73 it is possible to convert Gur to the modulusGurc, valid for a particular average nonnal stress level. Using the hyperbolic stress strain relation ofKondner, 1963 these authors were able to link Gurc to the low strain G0, or as defined in this thesis3 byGmax. Their hyperbolic equation is given by:__1= (2.75)G_ Gy__av‘CUsing the approach suggested by these authors it is also possible to link the Gur to the stiffnessrelevant to the design problem, where the average level of induced strain will differ from the average valuesimposed by the pressuremeter in the soil.Byrne et al, 1990 followed a more elegant direction for the problem of assigning stress and strainlevels to Gur in order to link this variable to Gm. These authors carried out a plane strain axisymmetricfinite element analysis in which both the stress and void ratio changes in the plastic zone were considered,as well as the non linear stress strain response of the sand during unloading. Stress ratio effects onwere included with the expression of Yu and Richart, 1984 and the unloading stress strain response wasmodeled with the hyperbolic equation of Kondner, 1963.The finite element analysis of these authors considered stress and strain changes in all the soil elementssurrounding the probe during unloading. The unloading was simulated in a number of small steps with thecumulative displacement at cavity wall being used at each stage to compute the cavity strain increment.The analysis output was presented in a chart format, such that for a given dPJP and P of the loop it ispossible to correct the measured Gur for both stress and strain amplitude, and hence predict the low strainmodulus Gm at the in situ normal stress level.The finite element analysis of Byrne et al, 1990 demonstrated that, as the expansion testing proceeds,it is possible to obtain a reasonably constant Gur if the size of the loops (dP) also increases. This is sobecause the opposite effects caused by the increase of strain amplitude and increase of stress level over Gurtend to cancel each other.Both Gmax and G are defined for the same strain level. 0m refers to pressuremeter predictions via Our, whereasG0 refers to the actual in situ maximum modulus measured by highly accurate techniques, such as geophysics tests.65Recently Fahey and Carter, 1993 expanded the pressuremeter capability of stiffliess derivation, byusing the pressuremeter unload reload loop into finite element simulations. A finite element analysis thatconsiders void ratio, stress ratio and stress level over Gm, in the plastic zone, and assumes a non linearhyperbolic variation of stiffness in the elastic zone, was developed to simulate the expansion and unloadreload testing stages. By matching numerical predictions with field data these authors were able tosimulate the degradation of the stiffliess ratio with the increase in strain amplitude for the tested sand.Thus, this methodology allows the derivation of the in situ modulus for any strain level, rather than theassessment of a unique Gurc modulus at some average strain level.The above discussion indicates that the unload reload modulus of the pressuremeter represents anindex to the average stiffness of the whole soil surrounding the cavity, in which each soil element is under aparticular stress and strain level.2.3.4.2 Initial Shear ModulusThe elastic shear modulus of the sand can be also obtained by the measurement of the initial slope ofthe testing curve. This modulus, previously defined as “Gi”, governs the soil response at the “elastic” zonewhere the sand is assumed to deform in a linear elastic manner.The top plot of Figure 2.18 demonstrates how this modulus is obtained in a typical testing curve. Themodulus Gi is defined with the use of Equation 2.26, which assumes the initial stress strain response of thesand as linear.However, given the previously discussed strain level dependency of the shear modulus, the initialstress strain response of the sand (as measured by the SBPM) should not be expected to be linear. Similarlyas the reload stage of the loop, during the initial (elastic) loading of the sand there will be an overallincrease of shear strain in the surrounding material, leading to a gradual decay of the soil responsemeasured at the cavity wall. Therefore, the stiffness measured at the cavity wall shall vary non-linearlyfrom a very high value, ideally equal to G0, to a low value equivalent to the stiffness at the limit (yielding)condition of the sand. The blow up of the initial section of the testing curve of Figure 2.18 indicates thatthe “actual” response of the sand does not follow the “idealized” response of the linear elastic theory.However, the current SBPM versions do not have the high accuracy required to measure the sand stiffnessCu0a)IU)Cl)a)CuCIa)CU-ICu11DU)U)a)1CuCa)4-CUIShear Strain (y = 26e) (%)66Failure at Cavity WallCircumferential Strain (%)Zoom of Initial Section of Testing CurveActual Response0TAPlastic DomainIdealized ResponseFigure 2.18: Determination of Initial Modulus from SBPM Curve67variation at the very low levels of strain mobilization (Hughes, 1993 personal communication). In otherwords, they can not precisely define the non-linear stress strain behavior of the sand showed in Figure 2.18for cavity shear strains below the failure (ye, which in general is in the order of 10’ %). Therefore, in thecurrent SBPM versions the initial slope approximates to a straight line, leading to an easy measurement ofGi if the testing curve is not disturbed.Given the non linear response of each of the soil elements around the probe, it is concluded that theassumption of a unique modulus for the whole elastic zone is an over simplification of the cavity expansionprocess. During the expansion process, say at point B of Figure 2.18, the induced strain amplitude in the(idealized) elastic zone will vary from a very high value at the elasto-plastic boundary, to a value close tozero at an infinity distance from the cavity. The shear modulus will consequently vary from a low value atthis boundary to a high value at infinity, equal to the maximum modulus G0. This variation isschematically shown in Figure 2.19. This same figure shows the idealized variation of soil stiffness (Gi)assumed by all the cavity expansion theories discussed in section 2.2. It is concluded that Gi does notrepresent the true stiffness of the sand, but reflects some index stiffliess that relates to the average (Yav)behavior of the soil elements at a variety of strain levels within this zone. Similarly as Gur, some relevantstrain level shall be assigned to Gi. In this thesis the index modulus Gi is assigned to the failure strain levelinduced at the elasto-piastic boundary (ye). This limiting strain can be easily computed with Equation 2.33once the soil parameters are known, and is in general in the order of 10’ %. It is believed that y- is in thesame order of magnitude as the average (unknown) strain level Yav induced in the elastic zone.In summary the present review indicates that the shear modulus of the sand can be predicted, viaSBPM, with the use of the unload reload loops as well as the initial slope of the testing curve. Theestimation of the sand stiffliess by the initial slope of the testing curve is not commonly done, due to thehigh sensitiveness of this section of the testing curve to disturbance. At present the only modulus derivedfrom the SBPM testing curves is the unload reload modulus Gur. Both moduli reflect some simplifiedweighted average” stifffiess of the sand elements around the probe, and are determined based on the“elastic” idealization of the medium. The advantage of Gi over Gur is the fact that the former modulus isalready measured at the in situ normal stress ah, thus not requiring any sort of correction for stress level.However, similarly as Gur, some relevant strain amplitude has to be assigned to Gi.68Cl)DD0ci)-cC/)ci)-#--ci)Eci)Cl)Cl)ci)0Elastic ZoneElasto-Plastic BoundaryMaximum Go ModulusIdealized Response‘Gi FromCavity Expansion Theoriesy =r = rpyavRadiusy= 0r= infinityFigure 2.19: Variation of Soil Stiffness in Elastic Zone for a Particular Stage of the Cavity Expansion692.3.5 Conclusions of Section 2.3All the points of this section indicate that there are difficulties with the interpretation approachescurrently in use with SBPM testing curves in sand. The general conclusions can be given by:• The best current interpretation approaches adopted with the existing cavity expansion models,namely log-log, nomogram or numerical solution, lead to highly scattered 4) results in disturbed curves.Disturbance during selfboring tends to considerably increase 4) predictions, and even in controlledenvironments a spread of results can be noticed. Although 4) is the least sensitive parameter in relation tothe disturbance of the curve (Robertson and Hughes, 1986), the current approaches are not reliable fordisturbed (selfbored) testing curves. This would be the case even if the constitutive relationships werehighly realistic to simulate the complex sand behavior around the cavity. Reliable predictions of 4) are onlypossible in high quality testing curves.• Lateral stress predictions by the SBPM (using the lift off stress) proved to be totally unreliable evenunder controlled conditions in the laboratory or under “optimum” insertion conditions in the field. A broadrange of lateral stress is in general the only information currently obtained from the interpretation ofselfbored testing results. Reliable results of0h are only obtained with ideal installation SBPM tests.• The shear modulus derived from the pressuremeter reflect an average weighted stiffness of the soilaround the probe, given the complex stress and strain gradients mobilized during the loading and unloadingstages of the field test. For instance the combined (and opposite) effects of increase in stress level andstrain amplitude prior and during the loop stage, lead to Gur values that are in the same order of magnitudeof the G0 measured by shear wave velocity techniques at equivalent depths. As will be commented inChapter 3, the ratio of Gur/G0 is close to 1 based on the data of Bellotti et al, 1987, Hughes andRobertson, 1984 and Bruzzi et al, 1986. However, average strain levels assigned to the pressuremetermoduli (Gur or Gi) are 2 to 3 times higher than the strain amplitude of G0. The pressuremeter moduli alsoreflect the deformation of the material in the horizontal direction.• All the basic soil parameters are determined in a “decoupled” manner, without any link betweenthem. 4), Oh and Gi are related to different interpretation methodologies that respond differentially to thedisturbance of the test.70From the above conclusions it is evident that research emphasis shall be placed on the reduction of thesensitivity of the predicted SBPM sand parameters to the disturbance generated prior to the test.Improvement of the predictions ofthis tool can be accomplished by the adoption of a new less (disturbance)sensitive interpretation methodology. The new interpretation methodology shall ideally derive all thedesired soil parameters in a “coupled” manner. The next section introduces this methodology.Another manner to improve the reliability of the predicted SBPM parameters is accomplished byreducing the disturbance generated during the insertion of this probe. Chapter 3 addresses this problemusing the results of the field testing programme carried out with the UBC SBPM probe.2.4 NEW INTERPRETATION METHODOLOGY FOR PRESSUREMETERS2.4.1 Basis of the Curve Fitting TechniqueThe prime objective of an in situ tool is to measure one or more parameters in situ, which can berelated to some material property. Most of the in situ tools rely on some sort of empirical correlation torelate what is measured to what is desired. This is so because the measured (in situ) parameter issimultaneously dependent on all the desired soil properties. Pressuremeter tests, on the other hand,represent an inverse boundary value problem in soil mechanics. From the pressuremeter test no directparticular soil parameter is measured, rather the pressure versus expansion data on the boundary of thecavity is obtained. Nevertheless, from these measurements of pressure and displacement there is a potentialto determine the basic soil parameters of the sand.This concept was initially put forward by Hughes, 1986 in a very simple manner. Suppose, forexample, that we wish to find the properties of a material which can be described by three independentparameters, A, B and C. To do this, we could have an in situ tool that would measure three values a, b andc that would be dependent on the original parameters A, B and C. If these measurements are independentof each other, then a = F (A, B, C), b G (A, B, C) and c = H (A, B, C) and the original soil parameterscould be found. Most of the in situ tools, however, can not provide independent values of a, b and c, butrather a unique variable z at the same depth. In contrast with any other tool, the pressuremeter is able toprovide an abundance of data a, b, c, ... n at each testing depth that can be independently connected to thematerial properties A, B and C.71Suppose, for instance, that an idealized set of soil measurements a, b, c, ... , n can be predicted at eachdepth based on a given set of original soil parameters A, B and C and some sort of constitutive relationship.In this case:a=P1=F(A,B,C, andD),b=P2=F(A,B,C andD2) .... n=P=F(A,B,C andD)Where P and D are respectively pressure and displacement at the boundary, and the function F isgiven by the equations of the cavity expansion model adopted.It will be possible, therefore, to compare the idealized prediction a, b, c ... n to the measured a, b, cn of the field test and derive some conclusions regarding the assumed A, B and C parameters of theconstitutive model. The better the rheological equations and assumptions of the model in relation to the realshearing phenomena, the closer will be the agreement of the idealized pressure expansion curve with respectto the field curve.As presented in the section 2.2 several cavity models can serve for this purpose. For undisturbed aswell as slightly disturbed testing data one has to rely on a interactive “curve fitting” analysis, to obtain fullconvergence of the idealized and field testing curves. The set of parameters A, B and C that leads to thecloser simulation of the measured testing curve represents the basic characteristics of the material. Thebetter the simulation of the testing curve, the higher is the reliability that one can place on the obtainedparameters. The fitting analysis has also the potential to be applied on disturbed testing data. In the case ofdisturbed curves the loading stage of the test shall be preferentially analyzed over its last part, which is ingeneral less influenced by the disturbance generated during selfbonng. This latter aspect is further exploredin section 2.6.The interactive analysis described above is here referred to as the “curve fitting technique”. The fittingtechnique is easily employed in any personal computer with fast processing as well as graphicalcapabilities, and basically consists of adjusting interactively the input parameters of the constitutive modeluntil a “match” is achieved between both field and model curves. By varying each of the possible sets ofdata, as schematically presented in Figure 2.20, it is possible to change the whole shape of the model curve.It can be noted that independent parameters A, B and C, represented in this figure as as,, Gi and aresimultaneously derived once the match is achieved.72Cu0ci)DCl)Cl)20CuCCwCurve Fitting Interpretation MethodologyCircumferential Strain (%)Variation of ParametersDuring the Curve Fitting(h1,Gi1,1)(ah3,Gi3,3)/Idealized Model CurveFigure 2.20: Basis of the Curve Fitting Technique73The fitting technique concept is not new in the pressuremeter technology, although it is almost ignorednowadays to derive soil parameters in clays or sands. In the case of clays the first mention to a techniquelike that was given by Arnold, 1981. He proposed the fit of the test data with an analytical relationshipbased on Kondner’s hyperbolic equation. Jefferies, 1988 was the first author to suggest an image-matchingprocedure to derive the soil parameters, due to advance in terms of data acquisition and processing systemsfor the pressuremeter in the 80’s. This author used this technique in the interpretation of SBPM results inthe Beaufort Shelf clay, and concluded that “once the model curve is fitted the numerical values areunarguable, at least within the validity of the model”. In sands Hughes, 1989 was the first author to adoptthe fitting technique with constitutive models for granular materials. This author showed with examples inlightly cemented sand and dense cobbly till that the fitting technique leads to reasonable results in materialswhich are almost impossible to sample and where the test data shows strong indications of disturbance.The success of the technique is based on the fact that the soil parameters are related to each other inthe context of the theoretical framework. This aspect represents the essence for a good quality of thegeotechnical parameters obtained by the pressuremeter. Previously the parameters were obtained withoutcoupling between them, leading to “loose” relationships between the variables and possible misleadingpredictions if one would attempt to use these parameters simultaneously in any classical elastic or plastictheory for soil mechanics. With the fitting technique there is a strong bond between the parameters, andtherefore the set of parameters, rather than each individual value, can be used in a more effective way tosimulate the soil in further design analyses. No one parameter has necessarily greater importance overanother, but it is the coupled set that must be used to describe the soil behavior. The theoretical cavityexpansion models to be used in this new interpretation methodology shall rely on few variables such thatthe match can be easily handled in practical cases. As demonstrated in section 2.2, the cylindrical cavityexpansion models that were chosen for that purpose rely on 4 or a maximum of 5 input variables. Theseare the basic four parameters of the sand (ah, Gi, 4. and 4) plus v (for some models).Matching between the idealized model curve and the pressuremeter testing curve was accomplished inthis thesis with the computer aided modeling programs (briefly described in section 2.2) written to simulatethe testing curve for different rheological models and input parameters. These programs allowed any of thefield testing curves to be piotted together with the idealized curves from any of the chosen cavity expansion74models. One simple method to match the curves involves the visual comparison between field and idealizedcurves, i.e. the set of input parameters is manually varied to accomplish the fit. Mathematically it is alsopossible to use standard statistical tools to accomplish the fit between field and idealized model curves.Programs like the Sigmaplot (Jandel Scientific, V.4.1) or the Kaleidagraph incorporate a nonlinearautomatic curve fitter (Marquardt-Levenberg algorithm in the case of Sigmaplot), that uses a leastsquare procedure to minimize the sum of the squares of the differences between the idealized and fieldcurves. These programs interactively guess and find out the required input parameters for the fit, but theyare not able to predict reliable parameters on disturbed data. This is so because other factors have to beconsidered for the curve fit with the disturbed testing curves (as will be discussed in subsection 2.6.1), andthe fitting can not be blindly accomplished without some input based on personal experience. For highquality testing curves either the visual or the statistical fitting approach lead to similar results. Thedifference between each of these approaches is the fact that with the use of the curve fitter software theinteraction process is done in a rather more efficient manner. In this thesis, however, only the visual fitbetween field testing and idealized model curves was adopted to predict the basic soil parameters of thesand.In order to establish the curve fit it is necessary, but not mandatory, to know the possible order ofmagnitude of the basic parameters adopted by the cavity expansion model, regardless if curve fittersoftware is used or not. The initial estimate of is done with any soil mechanics table once the generalcharacteristics of the tested sand are evaluated. Whenever possible, washed samples from the selfboringprocess shall be retrieved at particular depths for use in laboratory characterization tests. In addition tothat, piezocone sounding results can be used to infer the density characteristics of the studied granularprofile. The initial estimate of ah is done using the average value of the lift off stress measured by all thestrain arms of the pressuremeter. The estimate of the constant volume friction angle can be done with thetable proposed by Robertson and Hughes, 1986. The initial estimate of the pressuremeter modulus Gi canbe done with the measurement of the initial slope of the testing curve in the strain range below io %.However, for slightly disturbed or disturbed data it may be extremely difficult to define the initialestimates of Gi and ah, given the high sensitivity of these variables to the disturbance generated duringselfboring. A possible alternative to aid in the estimation of the initial magnitude of these variables, and75hence optimize the process of curve fitting, is the establishment of some sort of link between the inputvariables and some other pressuremeter variable that is less disturbance sensitive.In the last section it was stated that the initial slope currently measured in the pressuremeter testingcurve is equivalent to the secant shear modulus at a level of shear strain in the order of 1O %. This secantmodulus can be theoretically linked to the low strain modulus of the soil, related to the soil response at verylow levels of strain mobilization. In the absence of seismic shear wave measurements to define G0 theunload reload loop can be adopted. As commented in subsections 2.3.4 and 2.3.5 this modulus is readilyobtained in any standard pressuremeter testing curve, and in general appears to have a magnitude close tothe (low strain) modulus G0. Moreover, the methodologies proposed by Bellotti et al, 1989 or Byrne, et al1990, to correct the measured Gur for both stress and strain levels can be additionally used for a refinedestimation ofthe maximum shear modulus of the sand (G).Therefore, in order to optimize the process of curve fitting a link between Gi and the unload reloadmodulus was devised. An approximate theoretical development of such a link took in consideration thehyperbolic model put forward by Kondner, 1963, and was originally advanced by Dr. Hughes in 1992.2.4.2 Link Between Gi and GurThe shear stress strain behavior of the sand is recognized to be highly non linear with the stiffliessdepending on stress and strain levels, as discussed in the previous section. The strain level influence overthe shear modulus was schematically depicted in Figure 2.19, where the stiffness variation was presented inthe elastic zone at a particular stage of the expansion process. It is noticed in this figure that at the cavitywall close to failure moduli will exist, whereas at a large radius the soil modulus approach the low strainmodulus G0. The establishment of a link between this latter modulus and Gi leads to the knowledge of themodulus variation in the elastic zone, i.e. the determination of the non linear curve of this same figure.A better way to visualize the modulus variation of Figure 2.19 is presented in Figure 2.21(a). It isassumed here that the stress strain curve that rules the stiffness variation in the elastic zone follows thehyperbolic shape. This hyperbolic variation is a good approximation of the real shear behavior of granularmaterials over a variety of strain ranges, as successively demonstrated by Hardin and Drnevich, 1972,Ishihara, 1982, Bellotti et a!, 1989, Fahey, 1992, and others. The hyperbolic model was originally00Cl)Cl)U)0Cr)76Shear Strain, Y (%)1 .00(L0.00.0001 10Figure 2.21: Hyperbolic Soil Response at the Elastic Zone in terms of (a) Stress-Strain Curve and(b) Modulus Reduction Curve0.001 0.01 •0.1Shear Strain (%)77presented by Kondner, 1963 and used by Hardin and Drnevich, 1972 to show that the peak points of cyclicstress strain loops at successively higher stress amplitudes laid on an approximately hyperbolic “backbone”curve. In the hyperbolic model the parameters that are required to specify the structure of the whole stressstrain curve are the (low strain) maximum modulus G0 and the maximum shear stress tm. These variableshave to be defined with tests that take consider the in situ density and stress level conditions of the sandanalyzed.The hyperbolic curve of Kondner, 1963 is given by:—= 1 (2.76)7 1 7Gmaxrearranging Equation 2.76 and noting that e = G.y it is possible to derive:G__—=1— (2.77)GThe maximum shear stress that can be applied to an element of soil adjacent to the pressuremeter isderived by assuming pure elastic behavior prior to yield (no variation in the average normal stress up tofailure) and failure ruled by the Mohr Coulomb failure criteria (failure stress ratio ruled by the frictionangle of the sand). The maximum shear stress is given by:tmax ah tan 4) (2.78)Substituting into Equation 2.77 it is possible to obtain:G___—=1—- (2.79)GThis latter equation was also obtained by Fahey, 1990 in order to derive the secant shear moduli overany range of shear stress.In the elastic zone pure shear conditions will prevail. A unique and constant secant modulus Gi isdefined for this zone by the cavity expansion theories. As presented in Figure 2.21 (a) Gi is referenced to aknown strain amplitude ‘y±’, which is induced at the elasto-plastic boundary with a yield shear stress This78limiting yield stress, previously derived in section 2.2 (Equation 2.31) can be substituted together with Giinto Equation 2.79. This leads to the following equation:Gi hsm4 (2.80)G°høor GE = G0 (1-cos 4) (2.81)As noted before, the value of Gur obtained in standard pressuremeter tests can be used as a firstapproximation of the (low strain) maximum modulus G0. Therefore, assuming G0 Gur we obtain:GE Gur (1-cos 4) (2.82)Therefore, in slightly disturbed or disturbed data the fitting process can be optimized with the use ofEquations 2.81 or 2.82 above. These equations help in the establishment of the initial magnitude of Gi,based on a less disturbance sensitive parameter.The curve fitting process has to be carried out until full convergence of both curves is accomplishedfor most of the strain range of the field test. The initially guessed soil parameters will undoubtedly lead toan idealized curve that differs from the field curve. Therefore, these initial parameters have to be graduallychanged one by one ( and v are generally kept constant) until convergence takes place. With highquality testing curves the possible set of parameters is in fact very narrow, as the testing curve can be fittedby an ideal model curve in an almost unique manner. This is mainly valid for cavity expansion models thatrely on few parameters, as those discussed in section 2.2. In contrast to that, with models that containmany variables it is likely that a large array of combinations of the input variables suffice to provide fullconvergence of both idealized and field curves. Thus, with these models it may be difficult to define anarrow, or unique, combination of soil parameters that represents the response of the tested sand.Subjectiveness is often required on the part of the engineer during the curve fitting process, in whicheach of the input parameters has to be varied within physically acceptable boundaries. The higher thequality of the testing curve, the easier is the final derivation of the model parameters. In the case of slightlydisturbed or disturbed data it is useful to have some constraint between the input parameters in order toguide the engineer with respect to their variation during the fitting stage, and hence optimize the fittingprocess. Therefore the constraint imposed by Equations 2.81 or 2.82 can be also used to aid in theestablishment ofthe value of Gi at each step of the fitting stage, when the set of input parameters is chosen.79Basically Gi has to be varied simultaneously with 4), in order to keep its proportionality to G0 or Gur asexpressed by these equations.All the SBPM testing curves interpreted by the writer in Chapters 3 and 4 adopted the constraint ofEquation 2.81 during the curve match, regardless of the individual quality of each of the curves. Thevalues of G0 from downhole seismic shear wave measurements were used in the interpretation process,rather than the values of Gur, since they were afready known.Some final general observations have to be made in regard to the proposed interpretation technique:1. The use of Equations 2.81 and 2.82 for undisturbed or slightly disturbed data complements thefitting technique, establishing an additional constraint between the input soil variables. These equations aidin the establishment, of the initial value of Gi, as well as the establishment of its possible variation in thecourse of a fitting analysis. Basically, these equations reflect an approximation of what is experimentallyexpected in a granular material when sheared. This means that loose sands (with lower 4)’s) are expected tohave a lower stiffness Gi than dense sands (with higher 4)’s) if all other variables (G0, confining stress, etc.)are kept the same. Thus, it is logical that by setting a higher value of 4) during the curve fit a higher valueof Gi (as predicted by these equations) shall be also required, and vice-versa. Nevertheless, some care shallbe taken when using Equations 2.81 or 2.82 as they may not be universally applicable for all the exitinggranular deposits.2. The subjectiveness built in the fitting technique is an inherent characteristic of this methodology ofinterpretation. It will also exist when adopting highly refined finite element models to generate the idealizedpressuremeter curve, or when using the curve fitter algorithm of the commercially available sofiwareswritten for this specific purpose.3. Some of the soil parameters impose a higher variation in the idealized testing curve than theothers, This is due to the differential sensitivity of the cavity expansion model to each of the inputvariables. A better insight in this aspect is given later in order to evaluate the accuracy of the final results.Lastly, one must address questions relating to the consistency and simplicity desired for thesuggested interpretation methodology. When dealing with the solution of an inverse boundary valueproblem such as the pressuremeter, consistent and physically meaningful soil parameters shall be derived.According to Jefferies, 1988, when basic soil parameters are sought the solution via curve fitting technique80is in the great majority of the cases well conditioned and consistent. Mathematically many sets of coupledparameters can lead to an analytical equation that represents the experimental response. However,physically, just a small variation of the coupled set of parameters is acceptable because the analyticalresponse will considerably deviate from the experimental measurements. The larger the range of testingcurve available for match, and the higher the quality of this curve, the lower is the variation of the set ofparameters that allows the curve match. Simplicity of the methodology of analysis is a characteristic of theproposed approach. A simple interpretation technique is usually desired when the final objective is the useby engineering practitioners.In summary the proposed interpretation methodology leads to a coupled set of parameters that arerelated to each other by the framework of some constitutive theory. No parameter has a higher significancein relation to the others, but it is the coupled set that can be simultaneously used to predict the soil responseto the external action of any engineering work. The higher the capability of the model to simulate thetesting curve, the higher is the reliability of the parameters. With high quality testing data it is possible toobtain an almost unique set of input variables that allows the curve match. This is valid for the modelsdiscussed in section 2.2, since they rely on few parameters. With slightly disturbed or disturbed data it isalso possible to obtain meaningfhl soil parameters if a constraint is imposed between the parameters. Thisconstraint helps in the initial establishment, and subsequent variation, of the parameter Gi in the course ofthe fitting process. Thus, the constraint optimizes the curve fit as well as reduces the variability of the finalset of input parameters that allows the curve match.2.4.3 Modulus Reduction CurveOnce the fitted soil parameters for the tested depth are known it is possible to mathematicallygenerate the shear stress-shear strain monotonic “elastic” curve of the sand. This is the curve schematicallyshown in Figure 2.2 1(a). The importance of determining the non linear elastic response of the sand isrelated to the assessment of relevant secant moduli at the appropriate working strain levels of the designproblem. For instance, as demonstrated by Fahey et al, 1993, the knowledge of the in situ stress straincurve is extremely useful in deformation problems.As commented in subsection 2.3.4.1 Fahey and Carter, 1993 introduced a finite element methodologyto extend the information that is usually obtained from the unload reload loops of the pressuremeter, i.e. to81allow the derivation of the pressuremeter modulus at any strain level. This methodology, however, stillsuffers from some problems related to the finite element modeling of the pressuremeter expansion, asoutlined by this same authors. A possible approach to derive the modulus reduction curve of the sand is theuse of the hyperbolic equation of Koudner, 1963 with the fitting parameters.Using the information obtained by the curve fitting technique with any of the previously describedcavity expansion models, and Equations 2.76, 2.78 and 2.81 or 2.82, it is possible to derive:t 11 (2.83)‘1—(1—cos)+Gi tanThis equation allows the generation of the approximate spectrum of soil stiffness ratio (GIG0)variation with the induced level of shear strain, as schematically shown in Figure 2.21(b). This curve canbe derived for each testing depth, based on the predicted parameters of the fitting technique.The modulus reduction curve derived from Equation 2.83 is already assigned to the average normalstress level that operates at the depth of the tested sand. This is so because the interpreted (coupled) soilparameters of the fitting technique are related to the in situ density and normal stress levels. This is alsothe case for the maximum shear stress (tmaj. This is an important aspect of Equation 2.83 since, asexperimentally shown by Hardin and Drnevich, 1972 and Iwasaki et al, 1978, the modulus reduction curveof granular materials varies in accordance with the level of confining stress.The derivation of Equation 2.83 assumed that the behavior of the soil can be adequately representedby a hyperbolic curve. However, as observed by Hardin and Dmevich, 1972, the stress strain curves of thesoil are not truly hyperbolic. Nevertheless, these authors suggested that a hyperbolic form could still beobtained by imposing a distorted normalized strain scale. The distorted strain, called the hyperbolic shearstrain, is a function of empirical constants a and b. These empirical parameters determine the deviation ofthe stress strain relation of the soil from the hyperbolic shape, and are given by these authors in accordancewith the soil type (sands or clays) and presence of water (dry or saturated sand).Another approach was followed by Fahey and Carter, 1993 to allow the simple hyperbolic model tofit the observed shearing behavior of the sand.82These authors proposed a modified form of the hyperbolic Equation 2.77, as follows:/G 1lI (2.84)Gwhere: f and g are empirical parameters, introduced in an analogous manner to the parameters a and bof Hardin and Drnevich, 1972.Using the data of Teachavorasrnskun et al, 1991 from drained simple shear tests on hollow cylindricalspecimens of Toyora sand, Fahey and Carter, 1993 were able to estimate f and g as being respectively 0.98and 0.25.In this thesis only the original hyperbolic equation was adopted to predict the modulus reduction curveof the sand, at particular testing depths of the SBPM field programme carried out by the writer. This wasdone to simplify the prediction of this curve, by avoiding the incorporation of empirical variables a, b, forg of difficult estimation. The example of usage of Equation 2.83 is presented in Chapter 4.In summary the hyperbolic model used to establish a constraint between G0 (or Gur) and Gi alsoallows the establishment of the modulus reduction curve of the sand, if the remaining fitting parameters areknown at the testing depth. This curve is afready related to the in situ normal stress and will serve toestimate the sand stiffliess at a variety of strain levels. The most important aspect of the predicted modulusreduction curve is the fact that it is obtained based on a reference index modulus (Gi) in the strain range of10.1 %. This range is close to relevant strain levels of civil engineering works, which in accordance toseveral case histories presented by Burland, 1989 lies in the range of 10.2 to 10’ %.The curve fitting technique can be adopted with any of the cavity expansion models discussed before.In order to validate the assumptions built in each of these models, and hence impose a great amount ofcertainty on the predicted soil parameters, pressuremeter tests in controlled conditions have to be analyzed.This is so because under such conditions it is possible to establish baselines in terms of soil parameters forthe further comparison with the model predictions. With this objective calibration chamber testing resultswere analyzed with the new interpretation methodology, using the models discussed in subsection 2.2,7.This is shown next.832.5 VERIFICATION OF CONSTITUTIVE MODELS WITH CHAMBER DATA2.5.1 Tests with Leighton Buzzard SandFahey, 1986 described a series of 13 calibration chamber tests to investigate the interpretation with theHughes et al, 1977 theory. Given the high quality of Fahey’s testing results, and the particular conditionsof the tests, they are extremely useful here.The tests of Fahey, 1986 were performed with dry pluvially deposited Leighton Buzzard sand of thetype 14/25. All the tests were carried out with a small pressuremeter of L/D of 5, ideally installed (cast) inthe sample. For all the cases the constant volume friction angle was determined with the ring shear tests ofBudhu, 1979 and Cole, 1967 using the same sand. Reference peak friction angles were estimated based onthe simple shear tests of Stroud, 1971, also on Leighton Buzzard sand at similar density and mean normalstress conditions as those employed by Fahey, 1986. Unload reload loops were not performed byFahey, 1986 in his pressuremeter tests, and only the modulus Gi obtained by the measurement of the initialpressure expansion slope is related by this author. For additional details on the tests, sand parameters andcharacteristics ofthe chamber the reader is referred to Fahey, 1986.Tests with both flictionless top and bottom sections were used to simulate a perfect plane strainexpansion, without any of the possible end effects previously noticed by Jewell et al, 1980 in the Universityof Cambridge. A pressure controlled boundary with a constant boundary pressure during the testing stagewas also adopted in some of the chamber tests. According to Fahey, 1986 allowance for the influence ofthe finite dimension of the chamber (hence influence of the constant boundary pressure on the observedresults) had to be made, since the chamber radius was only 10 times the initial radius of the pressuremeterused. Using the basic principles of elasticity and the Mohr Coulomb failure criteria, Fahey 1986 was ableto demonstrate the influence of this finite boundary over the cavity pressure, and to derive theoreticalequations to adjust the cavity expansion models for this effect.In order to assess the capabilities of the selected cavity expansion models two of the high quality SCchamber testing series, with smooth end plates and constant boundary pressure, were adopted here. Thesetests are the ones denoted as SC7 and SC8. Both tests have similar void ratios and are normallyconsolidated. They differ solely with respect to the level of mean normal stress applied to the chamber prior84to the probe expansion. Table 2.3 presents a summary of the general conditions of the chosen tests afterthe sample consolidation as well as the reference soil parameters at each case.Tests SC7 and SC8 did not present three dimensional end effects, given the fact that the pressuremeterwas constrained by upper and lower smooth plates. Nevertheless these tests have to be analyzed in the lightof the analytical correction proposed by Fahey, 1986 for the finite boundary. Figure 2.22 presents thepossible curve fit with any cavity expansion model (in this example the Hughes et al, 1977 model) that isnot adjusted for the finite boundary of the chamber. It is noticed that both curves begin to divergesignificantly beyond a cavity strain of 1 %. For the infinite diameter (idealized) curve the pressurecontinues to increase, while for the finite diameter (experimental) curve the cavity pressure eventuallyreaches a maximum when the plastic zone approaches the outer boundary of the chamber. Therefore, themodels selected to be analyzed here were adjusted to include the finite boundary effect.Tests SC7 and SC8 were interpreted using the proposed methodology presented in the last section.Throughout the fitting analysis the lateral stress was set to a constant value equal to the lateral (boundary)stress of the chamber, presented in Table 2.3. The models adopted the constant volume friction angleexperimentally measured for this sand. For both the new and the Carter et al, 1986 models a value for thePoisson’s coefficient was required. Fahey, 1986 estimated this value as 0.25 in his analyses with the samedata. The same value for v was adopted here. For any of the selected models the simulation of the cavitypressure versus cavity strain curve was carried out up to the limit condition imposed by the chamber, i.e.up to the development of a plastic zone with the same size of the calibration chamber. This limit conditionis given byr1,/r0 equal to 10, where r is the radius of the plastic zone and r0 the initial pressuremeter radius.The plots showing the comparison between the experimental loading curves and the idealized modelcurves are presented through Figures 2.23 to 2.25. In Figure 2.23 the curve fit with the Hughes et a!, 1977model is shown for each of the adopted chamber tests. Figure 2.24 presents the fit with the Carter etal, 1986 model, whereas in Figure 2.25 the results with the new model are shown. The numbered points ineach of the plots represent the position of the elasto-plastic boundary in the course of the expansion test, asidealized by each of the models. It is noticed that in all the cases the expansion is carried out up to thelimiting condition of r1Jr0 of 10, as commented above. The soil parameters predicted for each case arepresented in Table 2.4.85ThST SAND VOID OCR o, a (Pcv1 Gi2TYPE RATIO (kPa) (kPa) (Deg) (MPa) (Deg)SC7 LB 14/25 0.5115±0.016 1 90 90 35 20.6 48-50SC8 LB 14/25 0.5115±0.016 1 45 90 35 25 48-50Tests SC7 and SC8 are tests in which the pressuremeter was ideally installed1 -Constant volume angle from Ring Shear tests of Bucthu, 1979 and Cole, 19672-Defined with the initial slope of the testing curve, for e < 0.5 %3-Plane strain friction angle from simple shear tests of Stroud, 1971, with Leighton Buzzard sandat a void ratio of 0.53 ± 0.005 and am = 90 kPaTable 2.3: Calibration Chamber Testing Results on Leighton Buzzard Sand (Modified after Fahey, 1986)86Test SC7 Hughes et al, 1977 Model1000-____Experimental Curve__Idealized Model CurveC -fTh- 600—600-(1)C -0=42deg400-iC -= 90 kPaGi = 30.5 MPaS 200-0— i I I I I I0 1 2 3 4 5 6 7Circumferential Strain (%)Test SC8 — Hughes et ci, 1977 Model1000-Experimental CurveIdealized Model CurveC -0-800-C600-C/iC0400-=41deg0C= 90 kPa= 29.5 MPa-200-CLf0— ‘ I I I I I0 1 2 3 4 6 6 7 8Circumferential Strain (%)Figure 2.22: Curve Matching without allowance for Finite Boundary Correction - Fahey, 1986 Tests87LiLfLi••••• Experimental Curveeo-o-o Idealized Model Curve87Test SC7 — Final Curve MatchingHughes et at, 1977 model800z500(/)(I)p400C-_3000C200—1000800700600Q)b00Cl)C!)400C-—30005200—1000Circumferential Strain (%)Experimental Curveo-o-e Idealized Model Curver/ro 65Test SC8 — Final Curve MatchingHughes et al, 1977 modelCircu mferentia6Strain (%)Figure 2.23: Curve Matching on Fahey, 1986 Tests: Hughes et al, 1977 Model••••• Experimental Curveeo-o-e Idealized Model Curve878891800700600a)EDSO0CoCo4000—3000200—100LU800700-‘600-: 500-CoCo400-—300-C200-100-Test SC7 — Final Curve MatchingCarter et al, 1986 model32 6Circumerentia Strain (%)5 7••••• Experimental Curvee-o-ee Idealized Model Curve8 10--6r/ro_je’4Test SC8 — Final Curve MatchingCarter et al, 1986 model7LU I 1 I0 1 2 3 4 5 6CirCumferentia’ Strain ()Figure 2.24: Curve Matching on Fahey, 1986 Tests: Carter et a!, 1986 Model8600‘,700w600500400300200100600e -o-e eExperimental Curveidealized60089Model Curve8 107r/ro5000)DU)C’)a)EL4003000c0) 2001000TestNewSC7—CavityFinal Curve MatchingExpansion modelCircumferential Strain (%).....o -o-eExperimental CurveIdealized Model Curver/ro8 9 1050)(I)(1)0)00C0)cTest SC8 —New CavityFinal Curve MatchingExpansion ModelCircu mferentialS 6Strain (%)7Figure 2.25: Curve Matching on Fahey, 1986 Tests: New Cavity Expansion Model90MODEL PARAMETERS CALIBRATION CHAMBER QUALITY OF CURVETESTS FITTING_______________SC7 SC8Hughes et al. 1977 4) (Deg) 45 43 Good to Very GoodGi (MPa) 35.1 32.290 90C7h (kPa)Carter et al, 1986 4) (Deg) 49.5 47 Very Good to ExcellentGi (MPa) 42 38.190 90h (kPa) 0.25 0.25VNew Cavity 4) (Deg) 49 46 Very Good to ExcellentExpansion Model Gi (MPa) 41.3 36.690 90ah (kPa) 71 71Gur (Mpa) 0.25 0.25V4)cv = 35° for all the cavity expansion modelsTable 2.4: Fitting Results on Chamber Tests with Leighton Buzzard Sand91The following observations can be drawn from all the results shown:1. In general the quality of matching between the curves is reasonably good for Hughes et al, 1977model with some discrepancy in both the early and latter stages of the loading points. With theconsideration of elastic strains in the plastic zone, as done by the new as well the Carter et al, 1986 models,the quality of the curve match increases. Similarly as in the example of subsection 2.2.7 close results interms of the shape of the idealized curve were obtained for these two models, besides of the differentapproaches adopted to derive the elastic strains in the plastic zone. This suggests again that similar resultsare achieved by either the new or the Carter et al, 1986 models when a Poisson’s coefficient in the range of0.2 is used.2. The ffiction angle derived by Hughes et al, 1977 model in both testing cases is smaller thanrespective values obtained by the other models. This is so due to the fact that the lack of elastic strains inthe plastic zone leads to a “more rigid” pressure expansion curve (see Figure 2.11) with the Hughes etal, 1977 model in comparison to the other models. Consequently Hughes et al, 1977 equations require alower 4 input to bring the analytical curve up to the experimental curve or a higher 4 decrease to bring thissame curve down to the experimental one. As expected, similar numerical results were obtained by bothCarter et al, 1986 and the new model. The plane strain ‘s predicted by these latter models agree with therange defined by Stroud, 1971 for Leighton Buzzard sand under the chamber density and confiningconditions.3. For all the cases the predicted Gi’s were above the moduli defined by the measurement of the initialslope of the experimental curve. A possible reason for that may be related to the density of the sandadjacent to the pressuremeter shaft. According to Jewell et al, 1980 during the raining process it is likelythat the suspended pressuremeter (ideal installation) has a local influence on the sand, thus producingaround its shaft a thin annulus of sand with random density variations. This variation biases the shearmodulus of the sample predicted by the initial slope of the testing curve. This is so given the highsensitivity of this initial section of the curve to the disturbance of the sample. According to Ferreira, 1992for ideal installation SBPM tests the initial slope of the experimental curve is unrealistically low.In order to obtain a definitive conclusion in regard to the accuracy of the magnitude of the predictedGi’s a parametric analysis with the test SC7 and the new cavity expansion model was carried out. The92optimum parameters for this case (presented in Table 2.4) were adopted, and Gi was allowed to vary from30 to 50 MPa. Figure 2.26 presents the predicted idealized curves as well as the experimental curve. It isnoticed that for shear moduli outside the specified range it is not possible to obtain an analytical curve closeenough to the experimental curve for a proper match. This suggests that the Gi’s of the elasto-plasticmodels used here indeed shall have a magnitude that differs from the magnitude of the moduli obtained withthe initial slope of the testing curve.Using Equation 2.81 it is possible to estimate the (low strain) maximum shear modulus of the sand atthe chamber density and confining conditions based on the predicted value of Gi and 4). Using theparameters obtained by the new cavity expansion model in the chamber test SC7, a value of G0 in the rangeof 120 MPa was obtained. As commented before the tests of Fahey, 1986 did not incorporate unloadreload ioop stages. However, a rough estimate of the possible Gur as well as Gm for this chamber test canbe obtained through the chamber test C2 of Jewell et al, 1980, since in both cases the same sand aidsimilar testing conditions were employed4. In Jewell’s case one unload reload loop stage was carried outduring the test (dP = 155 kPa, d80 = 0.11 %), leading to a Gur of 71 MPa. Using the methodologyproposed by Byrne et al, 1990 with the loop variables of Jewell’s C2 test it is possible to estimate the (lowstrain) modulus of the sand tested in Fahey’s SC7 test. A value of Gm in the range of 130 MPa isobtained, which is remarkably close to the value estimated with the use of Gi and Equation 2.81. Thisagain validates the proposed interpretation approach to derive the stiffness of the sand, as well as the othercoupled variables.In conclusion the above results place a high degree of confidence on the assumptions of the simpleelasto-plastic models compared in this thesis, as well as on the interpretation approach suggested in the lastsection. Based on two high quality chamber tests it appears that Hughes et a!, 1977 model gives aconservative estimate of the strength and stiffness of the sand. Carter et al, 1986 and the new cavityexpansion model lead to more accurate predictions of the plane strain 4) and Gi. Therefore, theincorporation of elastic strains in the plastic zone improves the quality of the analytical simulation of thepressuremeter test in sands. From a practical point of view the predictions of both the new and Carter eta!, 1986 models are identical, suggesting that both of them can be used.Jewell’s C2 test was carried out with a sample with CTm = 90 kPa and void ratio = 0.527.93Test SC7 — New Cavity Expansion Model900-o0-- - -Ii) /7,/BOO -1_I) /7,’400-!‘,—300- jio.1/a)200-C100••••• Experimental Curve- Idealized Model Curve: 01=30 MPa—— Idealized Model Curve: 01=50 MPa0- ‘ I0 1 2 3 4 B 6 7Circumferentia Strain (%)Figure 2.26: Curve Matching on Fahey, 1986 Tests: Limits for Gi94The calibration chamber tests adopted above are particularly useful to assess the suitability of planestrain solutions because they represent “special” tests in which the LID (end) effect of the pressuremeter isintentionally suppressed. In real cases, however, end effects may be of importance for the final results.According to Fahey and Carter, 1993, the incorporation of end effects on the testing results can lead toimproper soil predictions with the use of 11) cavity expansion models. 2D models solved via finite elementtechnique would be then recommended in this case, In order to assess the capability of the new cavityexpansion model for practical use additional chamber data is analyzed and presented next, together with abrief literature review of this topic.2.5.2 End EffectsThe influence of the length to diameter (LID) ratio of the probe is a subject of controversy.Suyaina et al, 1983 used X ray radiography techniques in model experiments to investigate the patternof expansion in monocell (LID = 10) and tricell (LID = 8.3) Menard type pressuremeters. Thedisplacement patterns showed a tendency to cylindrical expansion for both cases, which corroborates theuse of cylindrical cavity expansion models to interpret the testing curve. Yan, 1988 by modification of thefinite element program CONOIL of the University of British Columbia analyzed the cavity expansionproblem of the pressuremeter with both a 2D axisymmetric and a plane strain finite element mesh. Hefound that, in comparison to the plane strain predictions, the expansion of a pressuremeter with an L/D = 4and 12 leads to an overestimation of f respectively equal to 10 and 2 %. However, by carrying out asimilar analysis using as reference a field SBPM curve of Hughes and Robertson, 1985, for a granulardeposit of Vancouver, he concluded that “pressuremeters with LID = 6 can provide the field pressureexpansion curve that is close enough to the axisymmetrical plane strain condition”. More recently Yu andHoulsby, 1992 presented a finite element numerical analysis similar to the one described by Yan, 1988.These authors used an elastic perfectly plastic model following the Matsuoka, 1976 flow rule to accountfor dilatancy during cavity expansion. As in Yan’s case the Hughes et al, 1977 log-log approach was used,and an overestimation of 11 to 17 % of the plane strain 4 could be predicted from a pressuremeter withLID 6 and soil stifihess ratio (Gilah) respectively of 5.2 to 7.0.In order to understand in this subject, Salgado and Byrne, 1990 used the calibration chamber tests ofJewell et al, 1980 as a basis for comparison with their finite element predictions. They used an95elasto-plastic stress strain model coupled to the Matsuok&s flow rule with a 2D axisymmetric finiteelement mesh, hence simulating the geometry of the probe (LID = 6.2) as well as the chamber boundaryconditions. Analyses were also carried out using a plane strain axisymmetric domain, with finite or infiniteboundaries. The soil parameters for the model came from the simple shear tests of Stroud, 1971 and thepredicted finite element response at the cavity wall was directly compared to tests B3 and B5 of Jewel! eta!, 1980. Salgado and Byrne, 1990 results suggest that some error may be introduced by applying a planestrain model in the analysis of a finite length pressuremeter since, as observed by the (finite element)simulated displacement patterns inside the chamber, vertical movements will exist in the soil surroundingthe probe at the latter stages of expansion.In order to check the findings of these latter authors, the same reference chamber test of Jewel eta!, 1980 was interpreted here with the (plane strain) new cavity expansion model. The analysis was carriedout with the same methodology presented before up to a stage where the plastic zone radius, predicted bythe model, started to touch the outer boundary of the chamber (i.e. r1,/r0 = 11). Allowance was again madefor the finite boundary of the chamber via the Fahey, 1986 equations.Figure 2.27(a) presents the comparison between the experimental and the model curves. It is observedthat the quality of matching is reasonably good for values of r1,/r0 below 7. At rp./ro greater than 7 theidealized curve stays below and deviates from the experimental curve. The same trend was obtained bySalgado and Byrne, 1990 using the plane strain mesh (with finite boundary), but with a much largerdeviation. Nevertheless, the parameters obtained are representative of the Leighton Buzzard sand at thechamber density and confining conditions ( = 48°, Gi 36.4 MPa, Oh = boundary stress =90 kPa),The discrepancy observed forr1/r0 below 7 is partially caused by the disturbance of the soil around theprobe due to the raining process, as previously noted. The deviation of the curves after a r/ro of 7 may beexplained with the help of the Figure 2.27(b). This figure presents (in scale) the elastic and plastic zonesdeveloped for the limiting condition ofr1,/r0 = 7, at a mobilized cavity strain of 3 %. During the test asthe pressure increases in the pressuremeter a plastic zone of yielding material extends rapidly outward fromthe cavity wall, decreasing substantially the proportion of material outside this zone. It is this outsidematerial, or elastic zone, which supports the plastic zone. When the cavity strain has reached 3 % the96Experimental Curveoo-o- Idealized Model Curve611Test 65 — Final Curve MatchingNew Cavity Expansion ModelCirCumferential Strain (%)100000-8004000-200uJCDFigure 2.27: (a) Curve Matching on Jewel! et al, 1980 Test B5(b) Elastic and Plastic Zones Inside the Calibration Chamber2D0LUI()zI97plastic zone is considerably large in relation to the size of the chamber, 11 radius wide. Nevertheless evenallowing for the influence of the outer boundary via Fahey, 1986 equations, the difference between thecurves increased after this cavity strain. It is suspected that end or LID effects, not considered byFahey, 1986 correction, start to dominate after this stage due to the reduction of the elastic zone.It appears, therefore, that end effects are solely evident when the effects of the finite outer boundaryalso start to influence the expansion process, at least for pressuremeters with L/D above 6. This sameobservation was given by Fahey, 1986 based on the rough ended series of calibration chamber tests. Asstated by Fahey, 1986, end effects could not be noticed at the early stages of the chamber tests, but theybecame obvious when the effects of the finite outer boundary were also present.By carrying out tests in which the chamber to probe radius ratio is increased it may be possible todecrease the influence of the boundary as well as the influence of the finite length of the pressuremeter overthe results. In order to illustrate this statement the interpretation of another calibration chamber data ispresented. The Italian ENEL-CRJS chamber reported by Bellotti et al, 1987 for tests with Ticino sand islarger than the one used by Jewell et al, 1980 at Cambridge. The Italian chamber has a radius 14.6 timesthat of the probe. Similar to the Leighton Buzzard case, the tests were carried out with a pressuremeter ofL/D equal to 6. Figure 2.28(a) and (b) present the final curve matching using the chamber test 228. Thenew cavity expansion model was adopted in the interpretation, with (Figure 2.28(a)) and without(Figure 2.28(b)) the correction for finite boundary proposed by Fahey, 1986. The same set of inputparameters was used in both cases.For this particular analysis the larger size of the chamber, together with the characteristics of thetested sand, allowed the cavity strain to build up to over 10 % before the radius of the plastic zone hasexpanded to 7-8 times that of the pressuremeter. This means that throughout the expansion process theplastic zone did not occupy more than 50 % of the whole volume of soil inside the chamber, and hence theboundary effects will be lower than those existing in Jewefi’s tests. Indeed, if no allowance is made for theouter boundary then the new cavity expansion model can be used to predict the chamber test data withremarkable accuracy, as illustrated in Figure 2.28(b). This suggests that for Bellotti et a!, 1987 dataneither end nor boundary effects are high enough to hamper the analysis with plane strain models. The98(A)2000-____Experimental Curveo-o-a. Idealized Model Curve0rp/ro800-2/’4J 400-C Test 228 — Final Curve MatchingNew Cavity Expansion ModelModified for Boundary Conditions via Fahey, 1986W ‘ I I0 4 6 8 12Circumerentia Strain (%)(B)2000-Experimental Curve-e-o-e Idealized Model Curve1600-D—Cl) 4 - - -C/)L) -3 -,800-2,—’CL.C Test 228 — Final Curve Matching—New Cavity Expansion ModelNot Modified for Boundary Conditions via Fahey, 19860•••• I I0 4 6 8 10 12CirCumferential Strain (%)Figure 2.28: Curve Matching on Bellotti et al, 1987 Test 228: Influence of Chamber to Probe Ratio99same opinion is shared by Manassero, 1989. As will be presented in the next subsection, the parametersobtained in the analysis of the test 228 are also representative for Ticino sand.Extending the findings of Figure 2.28 to a hypothetical field test, where boundary effects do not exist,it may be concluded that the selfboring pressuremeter can be analyzed by using simple plane strain models.This is so provided that the cavity strains are limited to the usual testing values (< 10 %) and a probe withslenderness ratio of at least 6 is used.To further validate the capabilities of the new cavity expansion model more SBPM data from thecalibration chamber tests with Ticino sand is analyzed next. Given the findings of this subsection, the 3ideal installation and 2 selfbored tests chosen for interpretation are analyzed without any allowance forboundary effects.2.5.3 Tests with Ticino SandBellotti et al, 1987 carried out 47 SBPM tests in the Italian calibration chamber of ENEL-CRIS. Thetests were performed in both dry pluvially deposited Ticino and Hokksund sands, and are grouped inaccordance to the stress history of the samples as well as the mode of pressuremeter installation. Similar toFahey’s tests a pressure controlled boundary type chamber was used.The tests chosen to be interpreted using the proposed methodology are the tests 228 (presented before),222 and 234, representative of an ideal installation case, and tests 246 and 252, representative of theselfbored installation case. These tests were chosen since they cover a large range of relative densityvarying from 43 to 77 %, were performed with different types of Ticino sand (TS-4 to 6) and with effectivehorizontal boundary stresses ranging from 53 to 215 kPa. The overconsolidation of the samples also variedconsiderably, from 1 to 5.5. For additional details on the tests the reader is referred to Bellotti et al, 1987.For all the tests the maximum dynamic shear modulus G0 was determined by Lo Presti, 1987 viaresonant column tests on Ticino sand under similar conditions as those found in the chamber tests. Unloadreload loop stages were carried out for each of the tests and the values of Gur and Gur are also given byBellotti et al, 1987. The constant volume friction angle of this sand is 34°, as defined by the latter authorswith ring shear tests. The reference peak friction angles were measured for each sample by conventionaltriaxial tests. The axisymmetric angles were further converted to plane strain values with the empirical100relationship put forward by Lade and Lee, 1976. Table 2.5 presents a summary of the general conditionsof the chosen tests after the sample consolidation as well as the reference soil parameters in each case.The chosen tests were interpreted with the proposed interpretation methodology and the new cavityexpansion model. A value of Poisson’s ratio of 0.20 was adopted based on the estimate given by Bellotti etal, 1989 for this same sand. The constant volume friction angle adopted is the value obtainedexperimentally by Bellotti et al, 1987. As for the previous analyses, the simulation of the testing curve wascarried out up to the limit condition of the chamber, in this case up to a r1,/r0 of 14. All the basic soilparameters of the model (4, a and Gi) were allowed to vary during the fitting process. Both the Gur andG0, related by Bellotti et al, 1987 for each of the chamber tests, were used during the curve fitting. TheGur was used with the new model to allow the incorporation of elastic strains in the plastic zone. The G0was used with Equation 2.81 to establish the value of Gi at each step of the fitting process.The plots showing the comparison between experimental and idealized model curves are presented inFigures 2.29 and 2.30, respectively for the ideal and selfbored tests. The results of test 228 were alreadypresented in Figure 2.28. In general terms the curve matching is very good for the entire range of cavitystrains, although with a quality slightly inferior as observed before for the “perfect plane strain” tests ofFahey, 1986. It may be the case, therefore, that some small influence given by the finite boundary is stillpresent for the experimental results of Bellotti et al, 1987. Nevertheless, for practical purposes the fit isreasonably acceptable.It appears, however, that disturbance is present in some of the chamber tests, irrespective of the modeof pressuremeter installation. This particularly seems to be the case of tests 222 and 252, although to avery limited and small extent. No perceptible difference in the quality of the match is found with respect tothe different initial conditions existing in each sample. This perhaps suggests that the model can beuniversally applied within the range of density, stress level and stress history observed for these particularchamber tests. Indeed, provided that the sand has a dilatant response during shear, the basic assumptions ofthe model are met. For Bellotti et al, 1987 tests dilatant behavior during shear was the rule rather than theexception given the high density and low confining stress levels of most of the samples.Figures 2.31 and 2.32 present a comparison between the predicted soil parameters and the referencevalues reported by Bellotti et al, 1987 for the chamber tests.TESTSANDDrOCRO.k,ahOur2GurC0i4004)‘OBSERVATIONSTYPE(%)(kPa)(kPa)(Deg)(MPa)(MPa)(MPa)(MPa)(Deg)222TS-446.25.5111.895.163453.846.917.480.342.7IdealInstallation228TS-4771518215.823467.364.633.4139.748.2IdealInstallation234TS-476.15.3115.8103.993453.547.724.7102.848.9IdealInstallation246TS-543110252.973418.617.112.759.642.7SelfboringInstallation252TS-675110152.973434.628.818.174.950.3SelfbonngInstallationI -ConstantvolumeanglefromRingSheartestsofBellottietal,19872-First seriesofunloadreloadloops:Degreeof pressureunload20%,d80=0.06%3-CorrectedforstresslevelviaBellottietal,1987or1989equations4-Definedwiththeinitialslopeofthetestingcurve,for c0<0.5%5-LowstrainmodulusfromResonant Columntestsof LoPresti,19876-PlanestrainanglefromLadeandLee,1976correlation,usingconventionalTriaxialcompressiontestingresultsofBellottietal,1987Table2.3:CalibrationChamberTestingResultsonTicinoSand(ModifiedafterBellottietal,1987)102Experimental Curveo-o-e Idealized Model Curve98r065 -4--3--/Test 222 — Final Curve MatchingNew Cavity Expansion Model0 21000800-600-400-200-0100080060040020000ELa)(I)(I)a)0Ca)Cw0ELa)DU)(I)a)00Ca)I,Cw4 6Ci rc urn fe rent I a8 10Strain (%) 12Experimental Curvee-o-e Idealized Model Curve7r/roTest 234 — Final Curve MatchingNew Cavity Expansion ModelCircumerentiaI Strain (%)Figure 2.29: Curve Matching on Bellotti Ct al, 1987 Tests: Ideal Installation TestsLUci):3U)(I)ci)0700Experimental Curveo-o-eo Idealized Model Curve103400600O3ØØaE200ci)Clr/ro657489000Test 246 — Final Curve MatchingNew Cavity Expansion ModelCircumferential Strain (%)700c600Experimental Curveoo-o-e Idealized Model Curve119r/ro40087300aE200ci)-l00LU 0Test 252 — Final Curve MatchingNew Cavity Expansion Model2 4 6 8Circumferential Strain10(%)1Figure 2.30: Curve Matching on Bellotti et al, 1987 Tests: Selfbored Tests104GD /ci) //-0- /c /•540- // A/(1) / /+10%,”c30- / 7-/ // // —lOx/ /ci) / /7 /.2-c / /ci) 7/Ideal Installation-,••••• Selfbored InstaHation0— i I ‘ I10 20 30 40 60Reference Plane Strain P (Deg)-+10%/’ /200- / //O-// /-/ /160- /C/, /(/3 ,“ci) / /S./U)120. 7/ /7o / /7-ci) / .‘4..) / -o—J so- / /-o / /ci) //4-’ , -o 7/40-ci)W,,/ •.... Ideal Installation-• . . .. Selfbored Installatior0— I I I I I I I I I I I0 40 50 120 160 200Reference Boundary Stress ([<Pa)Figure 2.31: Predicted Results of the Curve Matching : Friction Angle and Lateral Stress105GurC (MPa)Figure 2.32: Predicted Results of the Curve Matching: Shear Modulus106The following observations can be drawn:1. Both the predicted plane strain friction angles and the lateral stresses compare well with thereference values for each test. The average error is less than 10 %, resulting in a high accuracy for themodel. Nevertheless, the predicted results are somehow underestimated. This may be caused by one (or thecombined effects) of the following reasons:• Disturbance of the chamber test: A small amount of disturbance that occurred during the sandpluviation or pressuremeter selfboring can impose a disturbed shape in the testing curve, affecting thematch.• Boundary effects: Given the findings of the last subsection the model was not corrected to accountfor boundary effects in the testing curve. However, it may be possible that such effects are still present to asmall extent on the chamber results.• Simplifications built in the cavity expansion model: Although most of the main features of theshearing behavior of the sand were accomplished by the new model, it still idealizes the medium based on asimplistic stress strain relationship.• Natural variability and sensitivity of the analysis: As commented before, the natural variability ofthe solution is an inherent characteristic of the fitting interpretation methodology. This variability can bedecreased with an increase of the quality of the testing curve, or the establishment of constraints betweenthe input variables in slightly disturbed or disturbed data.• Accuracy of the “reference” triaxial values: Dispersion of the results and experimental errorsinvolved with the testing methodologies adopted to derive the “reference” soil parameters have to beacknowledged. Moreover, laboratory testing that does not impose the same conditions as those that prevailduring the pressuremeter expansion also constitute a source of discrepancy for the comparisons. Bellotti etal, 1987 carried out conventional rather than plane strain triaxial tests to derive their friction angles. The“reference” values were defined after the application of the empirical relationship of Lade and Lee, 1976.Although this relationship is based on a large database it is still subject to the criticisms present with theuse of empirical equations to convert experimental variables.2. The predicted Gi is bounded by both the modulus defined by the initial slope of the experimentalcurve and by the unload reload (stress level corrected) modulus. The predicted Gi tends to be closer,107however, to the former variable. The reason for overestimation in relation to the stiffness measured by theinitial slope of the testing curve is the same as the one put forward for Fahey’s tests on subsection 2.5.1.The reason for underestimation in relation to Gure may be related to the different strain amplitudes that arerelated to each of these moduli. The important aspect is the thct that the magnitude of the predicted Gicompares extremely well with the G0 values related by Bellotti et al, 1987 in Table 2.5. As noted insubsection 2.3.4 the pressuremeter modulus Gi reflects the soil response at a shear strain amplitude in therange of 101 %, whereas the modulus G0 reflects the response at a shear strain level below iO %.Assuming the reduction curve proposed by Idriss, 1990 to be universally applied for all granular soils, withthe strain level differences above it shall be expected to have a Gi with a magnitude of 30 to 50 % the valueof G0 (see Figure 2.17). Using all the predicted Gi’s and the respective G0’s of Table 2.5 an averagemodulus ratio of 24 % is obtained for GiJG0. The slight difference for the magnitude expected based on thecurve of Idriss, 1990 can be accounted by factors such as soil cross anisotropy (Gi and G0 are modulirelated to different shearing directions) as well as difference in the tested sands.In conclusion, the proposed methodology of interpretation has the capability to simulate reasonablywell the pressuremeter loading curve when the new cavity expansion model is adopted. Provided that thesand has a dilatant behavior during shear, the new interpretation approach leads to accurate predictions offriction angle, lateral stress and shear modulus with either ideal installation (undisturbed) or selfbored(slightly disturbed) pressuremeter tests. In spite of the encouraging results of this section it shall beemphasized that more tests still need to be interpreted to confirm the reliability of the proposedinterpretation methodology, This is principally the case of sands that have different characteristics thanthose related here for Ticino and Leighton Buzzard sands, or sands that behave in a different manner duringshear.Factors like the strain range of curve match and the sensitivity of the new model to changes in theinput parameters have to be considered for the prediction of reliable results with the curve fitting technique.The following section addresses the problem of the interpretation of field curves, in which the abovevariables may play a more dominant role.1082.6 PARTICULAR ASPECTS OF THE FITTING TECHNIQUE IN SANDS2.6.1 Strain Range of Curve MatchFor undisturbed or disturbed data the strain range of curve match between field and model curves maybe of importance. In order to investigate this aspect SBPM tests carried out by the writer will beinterpreted and discussed herein. The SBPM tests were obtained with an extensive field testing programmein one of the UBC research sites near the city of Vancouver, in which the UBC SBPM was used. Thedetails of the testing programme as well as the characteristics of the adopted granular site are explored inthe next chapters.2.6.1.1 High Quality Testing CurvesTypical SBPM tests are expanded to (cavity) circumferential strains in the range of 10 %. Under thetraditional interpretation methodology only the last loading points obtained within this testing range areused to define the slope in the log-log graph, hence the friction angle. This is so because it is argued thatthe initial stages of expansion can be considerably affected by the disturbance generated prior to the test.Following this same reasoning, it may be also argued that if the approach followed by FDPM tests isadopted for the SBPM, with a testing stage carried out to a considerably high cavity strain, then it will bepossible to predict truly undisturbed soil parameters from the interpretation of the latest stages of theexperimental data.In order to assess this hypothesis a series of curve fitting interpretations was carried out here. Withthis purpose a high quality field curve expanded to a circumferential strain around 20 % was adopted forthe interpretation analyses. The fitting ranges chosen for the match of both experimental and idealized(model) curves varied from 0 to 5 %, 5 to 10 %, 10 to 15 % and above 15 %.The interpretation analysis was conducted with the new cavity expansion model, following theinterpretation methodology advocated in section 2.4. The adopted values of and v were respectively 34°and 0.25, since these values were experimentally obtained in drained triaxial tests with undisturbed samplesof this site (see Appendix C). The stiffness adopted in the new cavity expansion model (to incorporateelastic strains in the plastic zone) refers to the value of Gur measured during the unload reload stage of thefield test. The chosen test came from the testing sounding SBPO9, at a depth of 5.3 mm the UBC research109site. It is a high quality testing curve based on the visual quality assessment Cntefla put forward byRobertson, 1982. This testing curve is shown in the top plot of Figure 2.33.Figure 2.33 also shows the curve fit results at each of the chosen (cavity) strain ranges. The top plotpresents the match between both field and idealized model curves, whereas the bottom plot shows thepredicted parameters for each matching case. For this high quality curve the following comments apply:1. The predicted soil parameters are not unique and depend on the range of match adopted duringfitting. Nevertheless, for curve matches between 0 to 5 % or 5 to 10 % the same idealized curve sufficesto represent the measured experimental data. The same set of predicted soil parameters are obtained in thisrange. For curve matches above 10 % it is not possible to obtain a unique set of predicted parameters.The higher is the range adopted for match (above 10 %), the higher are the differences between thepredicted set of parameters and the parameters obtained with the match of the initial part of theexperimental curve.2. The predicted parameters from the curve match between 0 and 10 % ( = 43°, 0h = 33 kPa andGi = 11 MPa) are consistent with the expected values for this site, A deeper insight into the predicted sandparameters at this research site is given in Chapter 4 with the aid of reference laboratory test results andadditional curve fitting analyses. The quality of curve match in this range is excellent, which suggests thatthe simple model developed herein could “capture” the essential shearing behavior of this particular sand insitu.3. The predicted soil parameters for the curve match above 10 % do not seem to be realistic. Thepredicted effective lateral stresses appear to be extremely overestimated (consider, for instance, that theeffective vertical stress at this depth is around 60 kPa). The predicted friction angles appear to beextremely underestimated. This is so because friction values below the constant volume angle 4 arepredicted, suggesting a contractive behavior during shear rather than dilatant. A fully contractive behaviorwas not observed in the triaxial laboratory tests with the undisturbed samples, as commented in theAppendix C.The findings above suggest that for high quality SBPM testing curves meaningful parameters from thefitting technique can be solely obtained if the match is carried out with the initial loading points of theRange of MatchFigure 2.33: Influence of Range of Fitting on Results: Undisturbed Test(a) Curve Fitting at Distinct Ranges, (b) Variation of Parameters110Lciing Bridge Site — SBPO9 — Depth 5.3 mField CurveAnalytical: Match 0—5% and 5—10%—— Analytical: Match 10—15%— Analytical: Match 15—20%100!‘? (kPc)_!‘!_! CH(MPa) //______(Deg) ‘I‘ICl)U)U)Eaa00(1)-oU)C)-UU)a-/////60-n“ 0—5% 5—10% 10—15% 15—20%111testing curve, between 0 to 10 % (cavity strain). This finding is directly in opposition to the commonbelief that only the latest stages of the testing curve are useful for the interpretation analysis. Perhaps theinformation experimentally measured for cavity strains above 10 % is influenced by external factors thatare not considered in the new cavity expansion model. Two possible factors are prone to hamper theinterpretation analysis at the high strain levels of the SBPM test:1. End Effects.In the section 2.5 it was observed that simple plane strain solutions can be used to interpretundisturbed or slightly disturbed tests, provided that expansion is carried out to low (cavity)circumferential strains (below 10 %) and the SBPM has a slenderness ratio above 6. For expansionsabove this strain the size of the plastic zone starts to become very large, and the expansion deviates fromthe expansion of a cylinder.Figure 2.34 was prepared to compare the size of the plastic zone developed at the latest stages ofexpansion, Using the results of the curve fit between 0 to 10 % it was possible to predict the developmentofthe elasto-plastic boundary throughout the testing stage, more specifically during a cavity strain of 10 %and 20%.The UBC SBPM has an expanding section only 6 diameters (12 radii) long and so, as illustrated inthis figure, it is unlikely that a plane strain solution will be applicable in the latest stages of expansion. Atthese stages a compromise between the cylindrical and the spherical cavity expansion theory would have tobe developed (see the shape of the plastic zone in this figure) for the interpretation analysis.2. Strain Softening of the Sand.The typical shear behavior of the sand at a constant confining pressure was discussed by Vaid etal, 1980. Over a considerable range of strain, both initially loose and dense samples undergo volumeexpansion, and at very large shear strains tend to approach an ultimate strength and critical void ratio.This ultimate state is commonly referred to the critical state of the sand (Atkinson and Bransby, 1978), atwhich the sand shears with no change in stress or specific volume.Medium dense to dense sands present an initial small contractive section during shear, which isfollowed by a dilative behavior up to the critical state. The void ratio gradually increases to the critical112SBPM TESTING SOUNDING SBPO9DEPTH 53 mExpandable SectionUBC SBPM1PLASTIC ZONE ELASTIC ZONErp=12.roCD(cavity strain = 10%)— rp=17.ro___________(cavity strain = 20%)D(ro=012)Figure 2.34: Schematic of Plastic Zones Developed at Cavity Strains of 10 and 20 %113value and the mobilized angle of ffiction gradually decreases from the peak value, corresponding to themaximum dilation rate, to the lower bound value, corresponding to the critical state (null dilation rate).The above features were noticed by Lee and Seed, 1967 with drained triaxial tests in SacramentoRiver sand. Based on the results of these authors and the comments of Vaid et a!, 1980 it is possible tovisualize the typical (experimental) stress strain curve of medium dense to dense sands. This typical curveis shown in Figure 2.35. A similar stress-strain curves were obtained by Vaid et al, 1980 with drainedsimple shear tests in medium dense to dense Ottawa sand. The brittleness of these latter curves was,however, reduced. Figure 2.35 also shows the elasto-plastic representation idealized by the new modeldeveloped in this thesis. Note that the elasto-plastic model does not consider the strain softening effectobserved in the laboratory, but rather assumes the sand is dilating at the peak rate at all the stages ofexpansion.In the latest stages of the test the expansion takes place with the imposition of very high levels of shearstress and strain in the sand surrounding the cavity. Based on the experimental sand behavior relatedabove, it may be possible to speculate that in the latest stages of expansion an annulus of sand at criticalstate conditions will be developed between the cavity wall and the elasto plastic boundary.In this case the expansion process can be understood as the expansion of a two-layered system,composed of an inner layer shearing at constant volume conditions encompassed by an outer (plastic) layerwhere dilative volume change takes place. The response of this critical state annulus of sand has adominant effect on the measured testing response at the latest stages of expansion. As schematically shownin Figure 2.36 the pressuremeter curve deflects from the undisturbed model response, following a pathclose to the response defined by the critical state annulus. Therefore, in an opposite fashion to thatobserved for the initial stages of expansion, both experimental and model curves continuously deviate fortesting strains above 10 % (this is noticed in Figure 2.33 when comparing the field and the idealizedcurve matched between 0 to 10 %). If reliable parameters are sought with the curve fitting in the lateststages of expansion, then the cavity expansion model has to be modified to account for the strain softeningof the sand. Failure to do so results in the prediction of unrealistically high ‘s and low 4’s, such as thosepresented in Figure 2.33.0a)C.)I•1.-a)S114Peakalfa3 or arIOO = contantIDEALIZED MODELEXPERIMENTALPlasticShear StrainMedium Dense to Dense SandsIDEALIZED MODEL -7EXPERIMENTAL(contraction)Shear StrainFigure 2.35: Idealized and Typical Experimental Stress Strain Curves of Sand During Shear115Undisturbed Response (Model)Response of Disturbed AnnulusCavity StrainNatural SoilCl)Cl,a)I-.a>a)C-)FINAL STAGES OF EXPANSION: —10 % and Above(End Effects Not Considered)Figure 2.36: Effect of Continuous Shearing on Measured SBPM Response116In order to further investigate the above speculation, another analysis with the testing curve ofFigure 2.33 was carried out. The (plane strain) finite difference numerical methodology proposed byManassero, 1989 (see subsection 2.2.5) was chosen to be applied, since it is capable of predicting thevariation of stress ratio and volumetric strain with the mobilized shear strain in any element of sandsurrounding the expanding cavity. The numerical analysis was carried out with a Quickbasic programwritten specifically for this purpose. It is assumed (for the purpose of the numerical analysis) that endeffects did not exist in the latest stages of the expansion test. The following steps were taken:• The soil surrounding the cavity was subdivided into 1100 elements of thickness 0.1 mm. The outputin terms of volumetric strain, stress ratio, shear strain and cavity (circumferential) strain was specified foronly a few elements located at distinct radii around the cavity, as depicted in Figure 2.37. With this outputit was possible to infer the behavior of the plastic zone throughout the expansion process.• The finite difference technique of Manassero, 1989 was interactively applied for each of the 1100elements. For each pair of cavity 0r and uo it was possible to define the values of O and 6r at the wall(element 1) applying Manassero’s equations. Using the differential equations of stresses and strains for anexpanding cavity the values of Or and E at the adjacent soil element were then obtained. Applying againManassero’s equations in this new element it was possible to obtain another set of e and s,. This processcontinued up to the last soil element (1100), when another pair of experimentally measured o and 6 wasselected at the wall and the interaction with all the elements restarted.• Simultaneous to the derivation of 0r, 00, 80 and 8r, the values of stress ratio, volumetric strain andshear strain were also computed at each element during all the stages of the test. The program was writtenin such a manner that if for a particular soil element the mobilized stress ratio becomes lower than theconstant volume principal stress ratio (defined by K, or (1+sin4jI( 1-sin4)), then the stress ratio is set asconstant and equal to K(,,, for all the subsequent interactions. Moreover, a null dilation rate (daddy = 0) isalso imposed in this same element for all the subsequent interactions.• Once the whole testing curve was analyzed and the values of0r, G, 80 and 8r were known for each ofthe 1100 elements at all stages of expansion, the program wrote an ASCII file containing the stress-strainrelationships for the chosen soil elements.117PressuremeterAxis(D=74mm)SBPM TESTING SOUNDING SBPO9DEPTH: 5.3 mSOIL ELEMENTS CONSIDERED:183 366 550 732 915 1100BH dr=0.1 mmFigure 2.37: Soil Elements1989146.95Around the Probe Chosen for the Numerical Analysis Following Manassero,/R1 .37.05/ R183/ 5525— R366Radius (mm)/73.55R55091.95R732110.15R91 5128.45R1100118Figure 2.3 8(a) presents the (idealized) variations of volumetric strain against cavity circumferentialstrain for each of the elements selected. It shall be noticed that the critical state condition is mobilized atthe soil element when the curve reaches a horizontal level, characteristic of de = 0. The results of thisfigure indicate that, indeed, critical conditions are successively reached in each of the soil elements aroundthe cavity. The analyses indicate that the critical condition was achieved at the cavity wall for ancircumferential strain around 5 %. The annulus of soil at critical conditions expanded to a radius of55.2 mm (element 183) for a cavity strain around 7.5 %, and continued expanding afterwards. Thisannulus expanded up to a radius above 91.9 mm (element 550), or 2.5 r0, at the final expansion stage ofthis particular test.Figure 2.3 8(b) was prepared in order to compare the size of the annulus of sand in critical stateconditions in relation to the size of the plastic zone. Using the idealized model curve, fitted between 5 to10 % of this same experimental curve, it was possible to predict the development of the elasto-plasticboundary throughout the testing stage. This information, together with the results of Figure 2.3 8(a),allowed the estimation of the zone ofmaterial around the cavity under either dilative shearing (plastic zone)or critical (constant volume) conditions. These zones are plotted in this figure against the circumferentialstrains measured during the test. It can be noticed that the volume of soil under constant volume conditionsgradually increased in relation to the respective volume of soil under dilative (plastic) conditions. Forinstance, when the cavity strain was 10 % the critical state layer was equivalent to 7.7 % of the size of theplastic zone. This percentage increased to 11 % when the cavity strain reached 17 %.In conclusion, for sands the cavity expansion process shall not be carried out to cavity strains beyond10 %, where high gradients of shear stress and strain are imposed in the surrounding soil. These highgradients of stress and strain may induce the formation of an annulus of sand at critical (constant volume)conditions. This annulus grows simultaneously with the plastic zone, and at a faster rate. Thisphenomenon may partially explain the poor results for curve matches in the latest stages of the test, as thenew cavity expansion model is developed based on the cavity expansion at a constant (peak) dilation rate.End effects are also prone to happen at the latest stages of expansion, hampering the usage of plane strainsolutions for the interpretation ofthe SBPM curve.Manassero, 1989 Numerical Analysis — SBPO9 — 5.3 m119(A)0(Contraction)Cavity WallElement 183Element 366Element 550o-ee--ee Element 732*-**-*. Element 915øø- Element 11000c‘SA • *.., •w.—SS‘S ‘G.. *...5%S.. 0.•0-.‘ -.‘.‘‘S‘SS..- -o- - -e - -G - —(Dilation),-CaU5)Uci)E —1-D0>—2-0 5 10 15 20700-Defined via Curve Fitting: New Cavity Model (B)• Defined via Numerical Analysis: Manassero, 1989——.600 -E500->Elastic Zone400-0CDilative Shearing(Plastic Zone)200--o (Soil Element)Pressuremeter Radius0 i I I I I I I I I0 5 10 15 20Circumferential Strain at Cavity Wall (%)Figure 2.38: Results of Numerical Analysis for Test SBPO9: 5 .3m. (a) Volumetric Strain against CavityStrain for Soil Elements, (b) Relation between Plastic and Critical State Zones Developedaround the Probe120On the other hand, the example presented herein demonstrates that high quality SBPM tests in(medium dense to dense) sands can be reasonably well analyzed with the experimental informationcontained between 0 to around 10 % cavity strain.2.6.1.2 Disturbed Testing CurvesDisturbance affects the shape of the testing curve, and consequently shall also influence the final set ofpredicted soil parameters.In order to assess the likely influence of disturbance on the predicted parameters another series ofcurve fitting interpretations was carried out here. For this purpose a disturbed field curve was selected.The disturbed characteristics of the chosen curve are assessed based on the following evidence:1. The shape of the curve does not follow the “high quality standards” put forward byRobertson, 1982 with his visual quality assessment criteria. This is circumstantial evidence of thedisturbed characteristics of this field curve.2. The chosen curve caine from the testing sounding SBP19, at a depth of 5.3 m in the research site.This particular sounding consisted of 2 insertion trials at the same borehole. The first trial was carried outup to 5.7 m at a high penetration rate, resulting in the plugging of the cutting shoe to an extent of 70 % ofits sectional area. This invariably disturbed the surrounding sand up to 6 m deep. This is strongevidence of the disturbed characteristics of this particular field curve.As in the last subsection, the interpretation analysis was conducted with the new cavity expansionmodel following the interpretation methodology advocated in section 2.4. The same values of ,v aswell as Gur were adopted. However, a higher number of fitting ranges were selected for the interpretationanalyses. The fitting ranges chosen for the match of both experimental and model curves varied from 0 to3 %, 3 to 6 %, 4 to 7 %, 5 to 8 %, 6 to 9 %, 7 to 10 % and 9 to 10 %. Given the findings of the lastsubsection, the interpretation analysis was carried out up to a cavity strain of 10 %.Figure 2.39 shows the chosen field curve and the obtained results. The top plot presents the matchbetween both field and idealized model curves. For clarity, only 3 model curves are shown. The bottomplot shows the predicted parameters for each matching case.0aCl)ci)U)E0000Cl)ci)I,001)U-600s00121Laing Bridge Site — SBP19 — Depth 5.3 mField CurveAnalytical: Match 0—3%—— Analytical: Match 4—7—Analytical: Match 7—10%BCircumferential Strain (%) 13—6% 5—8% 7—10%1000h (kPa)Gi (MPc)A+AAA 4 (Deg)/.,/60-00—3% 4—7% 6—9% 9—10%Range of MatchFigure 2.39: Influence of Range of Fitting on Results: Disturbed Test(a) Curve Fitting at Distinct Ranges, (b) Variation of Parameters122For this disturbed testing curve the following comments apply:1. The quality of curve match is excellent in either the initial or in the latest stages of the field curve.This does not mean that the predicted soil parameters (for each of the fitting cases) are equally reliable.2. As for the high quality testing curves the parameters obtained are not unique, and depend on therange of match. Nevertheless, for each of the predicted parameters, the variation with the range of curvematch “levels off’ for matches at the latest stages of the test (close to 10 %). For curve matches in theinitial ranges of the test, from 0 to 5 % it is not possible to obtain a unique set of predicted parameters.3. The predicted parameters for curve matches between 5 % and 10 % (average 4) = 44.2°,37 kPa and Gi = 11.3 MPa) agree well with the parameters obtained with the fitting interpretation ofthe initial stages of the undisturbed curve (notice that both curves are related to the same testing depth).4. The predicted soil parameters for the curve match in the initial stages of the field test do not seem tobe realistic. This observation was also given with the results of the analysis carried out in the latest stages(beyond 10 %) of the undisturbed curve. With the disturbed curve, however, it appears that overestimatedfriction angles (above 500) and underestimated effective lateral stresses (below 20 kPa) were predicted.The findings above suggest that for disturbed SBPM testing curves meaningful parameters from thefitting technique can be solely obtained if the match is carried out with the latest stages (above 5 % andbelow 10 %) of the field curve. This is caused by the flict that disturbance affects the initial shape of thetesting curve, reducing its “roundness”. At the latest stages of this same curve, below a cavity strain ofapproximately 10 %, the effects of disturbance are decreased. For cavity strains beyond 10 % the effectsof disturbance on the testing curve may be even erased, but other factors start to dominate (as noted in thelast subsection) hampering the fitting interpretation analysis.The effects of disturbance on the field testing curve are visualized in Figure 2.40. When disturbanceis generated during the selfboring process an annulus of disturbed and loose soil is formed around theprobe. The diameter of this annulus is unknown and will depend on the degree of disturbance generatedprior to the testing stage. The response of this annulus of soil, schematically shown in Figure 2.40(a),influences the measured response of the test. The expansion process can be also understood as the cavityexpansion in a two-layered system, one looser close to the SBPM shaft and another denser around this firstci,I—DCoCo0ci)I—CoCo0>C.)123Natural SoilAUndisturbed Response (Model)Measured ResponseCavity Strain(A) INITIAL STAGES OF EXPANSION: 0 to — 5 %ADisturbed Annulusof Sand (Loose)Natural Soil(undisturbed)Cavity Strain(B) FINAL STAGES OF EXPANSION: —5 to -10 %Figure 2.40: Effect of Disturbance on Measured SBPM Response124layer. The SBPM testing curve will initially follow the path defmed by the looser (disturbed) annulus ofsand, therefore reducing its initial smooth “roundness”. As the plastic zone grows in the latest stages ofexpansion (beyond 5 %), the effects of the disturbed annulus on the measured response are continuouslydecreased. This is schematically shown in Figure 2.40(b), and results from the fact that a larger zone ofundisturbed soil starts to be encompassed by the expanding plastic zone. The measured cavity response atthe latest stages of expansion predominantly reflects the shearing response of this undisturbed zone of soil.In conclusion, the interpretation methodology advocated in this thesis allows the prediction of reliablesoil parameters in either undisturbed or disturbed data. This represents an advance in relation to thetraditional interpretation methodologies, that could only be applied in high quality SBPM curves.Nevertheless, the reliability of the predicted soil parameters may be expected to be directly proportional tothe quality of the testing curve. For high quality or slightly disturbed testing curves the reliability of thepredicted parameters is high. For disturbed curves the reliability of the parameters is somehow reduced.Table 2.6 presents the probable reliability of the predicted parameters from testing curves with differentdegrees of disturbance. The recommended fitting ranges for these curves are also given. As a general rule,for either undisturbed or disturbed data the curve fitting shall be carried out in the latest stages ofexpansion (between 5 % to 10 %).Table 2.6 is solely valid for SBPM testing curves in sands that are analyzed in accordance with theinterpretation methodology of this thesis. In order to efficiently use this table, knowledge of the quality ofthe testing curve is required. A subjective (visual) disturbance assessment, like the one proposed byRobertson, 1982, does not allow the clear distinction between “slightly” disturbed, disturbed or “highly”disturbed testing curves. Therefore, a numerical disturbance quantification is presented in Chapter 3 anddiscussed in the light of several field testing examples.2.6.2 Sensitivity AnalysisThe sensitivity of the new cavity expansion or the Carter et a!, 1986 models to changes in the inputvariables also dictates the reliability that one can place in each of the individual (predicted) soil parameters.The basic answers sought in this subsection are related to questions of the type:1. How does the fitting curve vary with the change of each of the input variables?2. Based on (1) what can we infer in regard to the accuracy ofeach of these variables?125QUALITY DEGREE RECOMMENDED RELIABILITYOF OF FITTING RANGE OFTESTING CURVE DISTURBANCE PREDICTED PARAMETERSUndisturbed or Low 0 to 10 % HighSlightly DisturbedDisturbed Medium 5 % to 10 % High to MediumHighly Disturbed High Close to 10 % Medium to Low (?)Table 2.6: Recommended Fitting Range of SBPM Tests in Sands1263. What is the influence of the estimation of either 4 or v for the curve matching?In order to assess the sensitivity of the model to changes in the input parameters a parametric analysiswas carried out. This analysis took into consideration the variation of ± 10 % of each of the inputparameters while keeping the remaining variables constant. The basic values selected to be varied werethose derived from the fitting interpretation of the high quality SBPM test of Figure 2.33. This specifictesting curve is presented again in Figure 2.41, together with the interpreted (“optimum”) values and theidealized model curve from the Carter et al, 1986 model. The optimum values were assessed with theinterpretation methodology advocated in this thesis.Figure 2.42(a), (b) and (c) present the idealized model curves obtained after a variation of ± 10 % ineach of the basic soil parameters 4, Gi and 0h, while keeping the others with the respective values presentedin Figure 2.41. For each curve the same values of and v, respectively of 34° and 0.25, were used. It isnoticed in Figure 2.42 that small variations in the value of can influence the model curve to a largerextent than the same variation produced by changes in either Gi or Oh. This implies that for this cavityexpansion model (or the new model) 4 is the less sensitive parameter for changes that eventually take placein the model curve during the fitting process, meaning that c is the parameter in which prediction a higherdegree of confidence can be placed. Both the 0h and Gi have similar sensitivities to possible changes in themodel curve, suggesting that for these variables the same amount of confidence shall be placed. Since thevariability of Gi and 0h for changes in the model curve during the match is higher than the variability of ,it can be concluded that the accuracy of the prediction of the former variables is lower than the accuracy ofthe prediction of . Nevertheless, the accuracy of any of the predicted soil parameters is still high, asdemonstrated in section 2.5.Suppose now that during the match one of the input variables is mistakenly adopted. This could leadto erroneous conclusions in regard to the remaining of the coupled set of variables if a good fit is obtained.lii order to simulate this situation, another parametric analysis was carried out. lii this case each of thebasic soil parameters of Carter et al, 1986 model (ist variable, either,a or Gi) was varied by a knownpercentage of the “optimum” value of Figure 2.41, while keeping the 2nd variable constant (equal to the“optimum” value) and letting the 3 variable to vary in order to produce the best curve match with the127Li500Lciing Bridge Site — SBPO9 — Depth 5.3 mField Curve-Idealized Model Cue3000400a)cn1-nU-)00U)200100Carter et al, 1986 Model1= 34 Deg, v 0.25Derived Parameters:= 44 Degrees= 35 kPa= 11.2 MPaCircumferential Strain (%)Figure 2.41: Curve Fitting of Test SBPO9 (5.3 m) with Carter Ct al, 1986 Modela500ci)(1)Cl)ci)0>>aC)1)C!)U)ci)0>>C0128Sensitivity Analysis of the Fitting TechniqueCarter et al, 1 986 Model—SBOg, 5.3 m4003002001000C500cD4OØ300ci)-200>a0a50040030020010002C I rc u m f e re n t i a6Strain ()Figure 2.42: Sensitivity Analysis of the Curve Fitting: (a) Friction Angle, (b) Shear Modulus(c) Lateral Stress129experimental data of this same figure. Table 2.7 indicates the respective percentage of increase or decreaseof the 3rd variable when the 1st variable is incorrectly assessed 20 % below or above the “optimum” value.The following observations can be given with this table:1. The percentage ofvariation of 4) to the “optimum” value is lower than 8 % when an error of ±20 %in either Gi or 0h happens. Moreover, when 4) is varied within ± 20 % of the “optimum” value a extremelyhigh (and similar) variation on either Gi or 0h is required to allow the fitting with the experimental curve.This is due to the differential sensitivities of 4), ah and Gi to the variations in the model curve. However,the quality of the curve match with an over or underestimated 4) is poor, being easily noticed during thevisual fitting procedure. The findings of this item suggest that the match in the final slope of theexperimental curve is extremely dominated by4), implying that if a high quality testing curve is adoptedduring the match then a highly accurate (plane strain) 4) is to be predicted.2. The percentage of variation of ah due to an error in (ii (while keeping 4) constant) is similar to thepercentage of variation of Gi given by an error in 0h. This implies that, if 4) is properly assessed during thematch, similar error (if one exists) is to be expected in respect to both Gi and ah, i.e., if for instance Gi isoverestimated by 20 % similar magnitude of underestimation shall be expected for ah. However, since inthe proposed interpretation methodology the constraint between Cli, G0 (or Gur) and 4) can be adoptedduring the curve match (Equations 2.81 or 2.82), it turns out that if 4) is properly assessed, then Gi will alsobe properly assessed. Consequently, given the observations of this item, ah will be well çstablished too.Either for Carter et al, 1986 or the new cavity expansion model the knowledge of the constant volumefriction angle (4)) is required. As experimentally demonstrated by Negussey et al, 1984 the constantvolume friction angle represents a constant material property, dependent on the sand mineralogy andindependent of particle size, confining pressure and density. It can be obtained with laboratory triaxialtests with undisturbed or reconstituted samples of the site, as done in this thesis and presented inAppendix C. In the absence of that, the table proposed by Robertson and Hughes, 1986 or the simple dryheap method of Comforth, 1973 with washed samples retrieved in the site can be adopted. Based on theexperience of the writer this latter approach leads to a as close as 1 to 20 of the value defined by thelaboratory tests.130PARAMETER VARIED TO A PARAMETER NOT VARIED PARAMETER VARIED TOKNOWN PERCENTAGE EQUAL TO OPTIMUM VALUE ACCOMPLISH BEST CURVE( 1 Variable) (2fld Variable) MATCH(3rd Variable)VARIABLE VALUE VARIABLE VALUE VARIABLE VALUE- 20 % 28 kPa 4) 440 Gi 14 MPa(+25%)’35kPa 4400h 8.9MPa42 kPa 440 Gi (- 20 % )Oh2% 4)Oh-2% 28kPa Gi 11.2MPa 4) 470(+6.8%)*35kPa Gi 11.2 MPaOh 41°42kPa Gi 11.2 MPa (-6.8%)(3h2O% 4)Gi - 20 % 9.0 MPa 0h 35 kPa 4) 4750(+7.9%)Gi 11.2MPa 35kPa0h 42°Gi + 20 % 13.4 MPa 35 kPa (.. 4.5 %)‘Oh 4)Gi-20% 9.OMPa 4) 440 Oh 43kPa *(+22 %)Gi 11.2MPa 4403OkPaGi+20% 13.4MPa (-14%)’__________4) Oh4)-20% 35.2° Gi 11.2MPa Oh 6lkPa(+ 74%)‘ 44° Gi 11.2MPa(j) 2OkPa52.8° Gi 11.2MPa (42%)*14)+20% Oh4)-20% 35.2° Oh 35kPa Gi 3OMPa(+ 167%)’44° 35kPa4) 0h 7.OMPa52.8° 35 kPa Gi (- 37 % )“4)+20% Oh* Variation in relation to the optimum value of this variable1-Poor Curve MatchTable 2.7: Sensitivity of Fitting Results for Increase or Decrease in one of the Input Parameters131As speculated by Ferreira, 1992, the associated error in the interpreted soil parameters is generallysmall when the value of is assumed during the curve fitting. This author additionally suggests that, inthe absence of laboratory data, an average value of 35° for 4 shall be adopted. In order to assess thisspeculation and hence evaluate the sensitivity of the model to changes in either the value of or v, anotherparametric (curve fitting) analysis was carried out with the same experimental curve presented inFigure 2.41. In this analysis the values of and v were allowed to respectively vary as much as ± 10 %and ±20 % of the “optimum” values related in this same figure. For each case a new set of predicted soilparameters was obtained by the fitting interpretation and compared to the “optimum” set derived with the“optimum” values of 34° and 0.25, experimentally defined for 4 and v.Table 2.8(a) and (b) indicate the respective variation in each of the predicted soil parameters given bythe variation of either or v. As demonstrated by the last row of each of the tables the average variationof the basic soil parameters was well below 10 %, suggesting that the model curve is almost insensitive tovariations in either or v. The reduced sensitivity of Gi in comparison to the sensitivity of0h comes fromthe use of Equation 2.81 during the curve match. Gi is not allowed to vary independently to variations in .Since 4 is the least sensitive variable of the model it turns out that the sensitivity of Gi to changes in eitheror v is also reduced. Nevertheless, the average variation of ah for this example was in the range of±6 % of the “optimum” value. This places a high degree of confidence on the speculations ofFerreira, 1992, suggesting that in the absence of laboratory baselines for cf or v “educated guesses” forthese parameters can be made.2.7 SUMMARY AND CONCLUSIONSThis chapter emphasized the analytical interpretation of selfboring pressuremeter testing curves insands. Emphasis was placed on the development of a new approach to analyze the data and derive reliablepredictions of the basic soil parameters, namely the friction angle, the lateral stress and the shear modulus.A comprehensive review of the existing cavity expansion models to simulate the pressuremeterexpansion in sands was presented. A new cavity expansion model that extends the theological equations putforward by Hughes et al, 1977 was devised to be used under the framework of a new interpretationmethodology. This new methodology relies on a curve fitting technique to match both experimental and132& VARIATION PREDICTED SOIL PARAMETERS(Deg) (%)ah(kPa) (MPa) (Deg)30.6 -10 33 11.1 43•734 0 35 11.2 4437.4 +10 37 11.5 44.5Average Variation: ±6 % ±2 % ± 1 %Observation: v 0.25 during the curve fitting(A) Variation of Predicted Results for Variation of Constant Volume Friction Anglev VARIATION PREDICTED SOIL PARAMETERS(%)cYh Gi(kPa) (MPa) (Deg)0.20 -20 37 11.2 440.25 0 35 11.2 440.30 +20 33 11.2 44Average Variation: ±6 % ±0 % ±0 %Observation: = 340 during the curve fitting(B) Variation of Predicted Results for Variation of Poisson CoefficientTable 2.8: Sensitivity of Fitting Results for (a) Constant Volume Friction Angle, (b) Poisson Coefficient133idealized curves, thus defining a “coupled” set of basic soil parameters. These parameters are linked toeach other by the constitutive model and can vary in accordance to some factors discussed in this chapter.The reliability of the Hughes et al, 1977, the Carter et al, 1986 as well as the new cavity expansionmodels was assessed with the use of a limited, but well established, calibration chamber database. Themajor aspects that can potentially influence the new interpretation methodology were addressed anddiscussed in the light of several analyses performed. These aspects are related to the possible influence ofend effects over the final results, the sensitivity of the analytical curve to changes in the input variables, thevariability of the solution for changes in either or v, and the influence of the strain range of curve matchfor the reliable prediction of the parameters.The main findings of this chapter can be given by:• The interpretation methodologies currently available for the analysis in sands lead to unreliablepredictions of the basic soil parameters in disturbed SBPM data. These methodologies are extremelysensitive to small amounts of disturbance in the testing curve, and should be solely appliedto high quality testing curves. The current interpretation approaches predict “uncoupled” soil parametersthat respond differentially to the disturbance of the test and are related to distinct stress-strain idealizations.• Simplistic cavity expansion models can be used to simulate the complex behavior that takes place inthe sand surrounding an expanding cavity, provided that realistic assumptions regarding the stress-strainand volume change of the sand are incorporated in the rheological equations. Models that idealize thestress-strain response of the sand with an elastic perfectly plastic representation, encompass the shearvolume coupling characteristic of the sand with a simple linear volumetric strain relationship, consider theelastic component of strain given by the increase in the average normal stress around the cavity, and do notextensively rely on input parameters from laboratory tests, are recommended for the SBPM interpretation.The new cavity expansion model developed in this thesis has all the above mentioned features and solvesthe cavity expansion problem in a step by step manner for each position of the elasto-plastic boundary.Contrary to Carter et al, 1986 model, the new model adopts different shear moduli in both the plastic andthe elastic zones. Nevertheless both models give similar results for Poisson’s coefficients in the range of0,2, suggesting that either one or another model can be efficiently used in the pressuremeter interpretationanalysis in sands.134The new interpretation methodology advocated in this thesis is simple to apply and can be easilyincorporated in any personal computer. When this new methodology is used in conjunction with the newcavity expansion model it has the capability to simulate reasonably well the pressuremeter loading curve ofcontrolled chamber tests. Reliable results with an accuracy within 10 % of the reference values wereobtained for either undisturbed or disturbed data. It shall be emphasized, however, that several factors areprone to happen during and before the interpretation process, leading to an increase or decrease of thereported accuracy. Factors like the disturbance of the test, the simplifications of the model and the strainrange of curve match do play an important role for the final reliability of the derived parameters.• Both the new and the Carter et al, 1986 cavity expansion models have the ability to reasonablysimulate the stress-strain-volume change behavior of medium dense to dense sands during shear. Theparameters predicted with the use of these models are extremely reliable if high quality SBPM testingcurves are analyzed. The reliability of the parameters with the interpretation of slightly disturbed ordisturbed curves is somehow reduced. The Hughes et al, 1977 model leads to conservative estimates of thestrength and the stiffness of the sand. This latter conclusion was also obtained by Yu, 1993.• Cylindrical cavity expansion theory can be adopted for the interpretation of SBPM tests that arecarried out to low cavity strains, below 10 %. This is valid for SBPM probes with slenderness ratiosabove 6. Plane strain solutions can also be adopted, provided that the pressuremeter is expanded in anunrestrained medium where boundary effects do not exist (and the aforementioned conditions apply).• The friction angle is the least sensitive and most reliable parameter obtained by the curve fittinganalysis. Given the fact that the curve match is extremely dominated by the friction angle it is unlikely thata 4 that differs more than 10 % of the “optimum” value can be obtained with a high quality testing curve.The error built into either Gi or Oh will be of similar magnitude, and close to the error built in 4. if the linkprovided by Equation 2.81 or 2.82 between this latter variable and Gi and G0 (or Gur) is adopted duringthe curve fitting process.• The final error in the predicted results given by the estimation of the constant volume friction angleor the Poisson’s coefficient is negligible if an “educated guess” of these variables (between ± 10 or 20 % ofthe “real” unknown values) is given.135• The strain range of match adopted between idealized and experimental curves during the fittingprocess has a fundamental weight for the accuracy of the final results. For undisturbed tests reliablepredictions of the soil parameters can be solely obtained with curve matches between cavity strains of 0 to10 %. Beyond this strain level the new cavity expansion model can not be used anymore, as the sandsurrounding the cavity may start to shear under critical conditions and the cavity expansion deviates fromthe idealized cylindrical form. For disturbed tests reasonably reliable results can be obtained by matchingboth field and idealized model curves in the latest stages of the test, between cavity strains of 5 % to10 %. The initial stages of the disturbed tests are considerably affected to render a reliable prediction ofthe sand parameters. As a general rule, however, it is concluded that for either undisturbed or disturbeddata the curve fitting shall be carried out in the latter stages of expansion, between cavity strains of 5 %to 10 %. For undisturbed tests, the idealized model curve obtained in this manner will also suffice torepresent the experimental behavior in the initial stages of expansion. For highly disturbed tests the curvematch shall be solely accomplished with the latest loading points of the field curve (strain ranges very closeto 10%).• Using a hyperbolic model to establish a link between the predicted pressuremeter modulus and theunload reload shear modulus of the pressuremeter test (or the low strain maximum shear modulus of thesand) it is possible to optimize the fitting process of slightly disturbed or disturbed data. The link betweenthese variables helps the establishment of the initial value of Gi, as well as its magnitude in the course ofthe fitting process. The hyperbolic model additionally serves to establish the idealized shear stress-shearstrain monotomc “elastic” curve of the sand, when used together with the final curve fitting parameters.This curve allows the prediction of the stiffness ratio of the sand (GIG0) for each level of induced shearstrain. This adds a new dimension to the pressuremeter predicted modulus, turning it relevant to anyengineering application where the average level of working strains differs from those imposed by thepressuremeter test.The interpretation methodology proposed in this thesis is easy to understand and simple to apply.Although simple, the new cavity expansion model can “capture” the most significant aspects of thecomplex pressuremeter expansion in medium dense to dense sands. The combined use of the newinterpretation methodology with the new model leads to the derivation of reliable and significant predictions136of the basic soil parameters from either undisturbed or disturbed data. Nevertheless, it is recommendedthat only high quality testing curves are used in the new interpretation analysis. This is so because thevariability of the curve fitting results is decreased and an easier match can be accomplished.In this regard the following chapter demonstrates how to enhance the quality of the testing curve,discussing the distinct variables that shall be considered for the optimization of the insertion procedure ofthe SBPM designed in the University of British Columbia (UBC).137CHAPTER 3.0 INSERTION AND TESTING PROCEDURE FOR THE UBC SBPM3.1 INTRODUCTIONIn this chapter emphasis is placed on the field determination of the UBC SBPM insertion and testingprocedure, with attention to the variables that affect the quality of the test. This chapter introduces the newdisturbance quantification criteria proposed in this thesis and discusses on the several variables that weregradually changed to enhance the quality of the pressuremeter testing data. With an understanding of theinfluence of each of these variables the recommended insertion procedure for the UBC SBPM is given.The conclusions of this chapter concentrate on what has been learned throughout the field testingprogramme to carry out selfboring pressuremeter tests in which disturbance is minimized.3.2 FIELD TESTING PROGRAMMEThe field testing programme proposed for this thesis required the use of a well known research site,where the soil characteristics were well documented. From the standard sites of the In Situ Testing Groupat UBC, a site called “Laing Bridge South” or “Laing Bridge” was selected. It is ideal not only because ofits proximity to UBC but also because it has been extensively studied in the past (Sully, 1991 andHowie, 1991). The Laing Bridge site is located in the Fraser Delta, near the city of Vancouver and at theInternational Airport. The geological characteristics of this delta and geotechnical details of the researchsite will be presented in the next chapter and only briefly discussed here.The Fraser Delta is characterized by the presence of sediments deposited in the Quatemary age viaalluvial process. The deposition process took place after the last glaciation under a variable sea level andhigh energy environment. The stratigraphy of the Laing Bridge site is basically comprised of a 2 to 3 m thinsurface layer of sandy silt underlain by a stratum of fine to medium sand of 15 to 20 m in thickness which,in turn, is underlain by a thick layer of normally consolidated clayey silt to silty clay. This last layerextends down to the Pleistocene till. The sand deposits of the Fraser Delta were not ice loaded.The testing programme targeted the 5 to 15 m depth range of this site, where the sand presents auniformly graded condition. At this depth interval the relative density (Dr) is extremely variable due to thevariability of the depositional conditions. hi general the Dr increases from 40 to 60 % based on the138piezocone interpreted results. Silt content is low in this range, below 5 %, and the sand is basicallycomposed of medium to line grain particles with a subangular to subrounded shape.The field testing programme consisted of 130 SBPM tests, 26 full displacement (FDPM) tests andother in situ tests. These latter additional in situ tests served to characterize the geotechnical features of thesite close to the pressuremeter soundings. They consisted of piezocone tests (CPT) with pore pressuresensors at different locations (face, behind the tip and behind the friction sleeve), the standard penetrationtest (SPT) with energy measurement, and downhole seismic piezocone tests (SCPT). The seismic conetests furnished this thesis with the evaluation of the (low strain) maximum shear modulus of the sand viadownhole shear wave measurements. Both disturbed and “undisturbed” samples (170 mm x 50 mm) wereobtained in the sand. These samples were retrieved with a stationary piston sampler denominated ST1from Rocktest Inc. The samples were used in triaxial tests, in order to obtain the peak friction angles ofthis sand as well as the values of and v required by some of the cavity expansion models. The results ofthese tests are given in Appendix C.Figure 3.1 and Table 3.1 present the location, the reference names and the general information of thein situ testing programme at the Laing Bridge site. At each test sounding a unique reference name (likeSBPO 1) is given, although in this sounding several pressuremeter tests were carried out at different depthsin the profile. As can be seen by Table 3.1, several soundings were performed throughout the experimentalstage of this thesis. This is because each of the test soundings had a particular purpose. In each soundinga unique combination of equipment, testing procedure and insertion characteristics was adopted, as detailedin Table 3.2. It shall be noted that the testing depths represent the values measured from surface to themiddle section of the expandable portion ofthe UBC SBPM..The initial tests from FDPO1 to SBPO5 were performed to familiarize the writer with both the dataacquisition system and the testing / insertion possibilities of this highly advanced probe. They also servedto evaluate and modify the recently developed UBC SBPM data acquisition system. The subsequent tests,from SBPO6 to SBP19, served to gradually improve both the design and field insertion procedures adoptedfor the UBC SBPM. This allowed the development of a customized insertion procedure, enhancingconsiderably the final quality of the derived testing curves. A new methodology was established to assess13902 10(m) / GATE ///7$ / // 10/ / .B2/ /// / / 0/ / / 11.// 20// 9// 2// cE24/ 22// L2FL1 P/6 / rA L3 12 -/ / ‘ •26/ / 14 24// 23// Hi : 15// 27c H// C// z// 181716 H5POSTFigure 3.1: Location of In Situ Tests at Lamg Bridge Site140SYMBOL I REFERENCE DATE TYPE OF iN SITU TEST1 FDPO 1 5-1-91 Full Displacement Pressuremeter2 FDPO2 5-17-91 Full Displacement Pressuremeter3 FDPO3 6-1-91 Full Displacement Pressuremeter4 SBPO 1 7-17-91 Selfboring Pressuremeter5 SBPO2 8-1-91 Seltboring Pressuremeter6 SBPO3 8-27-91 Seilboring Pressuremeter7 SBPO4 8-30-91 Selfboring Pressuremeter8 SBPO5 10-3-91 Selfboring Pressuremeter9 SBPO6 10-24-91 Seilboring Pressuremeter10 SBPO7 2-5-92 Selfboring Pressuremeter1 1 SBPO8 2-20-92 Seliboring Pressuremeter12 SBPO9 3-3 1-92 Selfboring Pressuremeter13 SBP1O 4-8-92 Sellbonng Pressuremeter14 SBPI 1 4-23-92 Selfboring Pressuremeter15 SBPI 2 5-12-92 Selfboring Pressuremeter16 SBP13 6-9-92 Selfboring Pressuremeter17 SBP14 6-30-92 Selfboring Pressuremeter18 SBPI5 7-9-92 Seilboring Pressuremeter19 SBP16 9-24-92 Se[tboring Pressuremeter20 SBP17 10-30-92 Seilboring Pressuremeter21 SBP18 1 1-25-92 Selfboring Pressuremeter22 SBP19 1 1-30-92 Seliboring Pressuremeter23 SBP2O 4-16-93 Sellbonng Pressuremeter24 SBP2 1 5-21-93 Seilboring Pressuremeter25 SBP22 5-27-93 Selfboring Pressuremeter26 SBP23 6-3-93 Selfboring Pressuremeter27a,2Th FDPO4, 05, 06 6-16-93 Full Displacement Pressuremeterand 27cPRESSUREMETER TEA1.A2 SCO1,SCO2 8-21-91 Dowiihole Seismic Cone-UBC#7ALB2 C02,C03 9-12-91 Piezocone Test-HOG#3-PPbhFSC DMTO 1 9-19-91 Marchetti Dilatometer TestD C04 9-26-9 1 Piezocone Test-HOG#3-PPFaceE1,E2 FVO1,SPTO1 10-10-9 1 Field Vane and SPT with Energy MeasurementF SCO3 10-17-91 Downhole Seismic Cone-UBC#9-PPFaceG C05 11-13-91 Piezocone Test-UBC#9-PPface, PPbhFSH1.H2 C06,C07 9-18-92 Piezocone Test-HOG#3, UBC#9-PPFace, PPbhFSand PPbhTipOTHER IN SITU TESTSOBSERVATIONS:Piezocone (CPT): Sectional Area 10 cm2,Diameter (D) = 35.6 mm. Details in Campanella and Robertson, 1981Pore Pressure Filter: Polypropylene, with thickness of 5 mmPore Pressure Measurements:PPFace (Ui) = Pore pressure sensor located at the face of the piezocone, at a L/D = 0.4PPbhTip (U2) = Pore pressure sensor located behind the tip of the piezocone, at a L/D = 0.9PPbhFS (U3) Pore pressure sensor located behind the friction sleeve, at a L/D 4.9Dowuhole Seismic Cone (SCPT): Sectional Area 10 cm2,Diameter (D) = 35.6 mm. Details in Campanella et al, 1986I Ll to L4 I ST1 I March 93 I Soil Sampling with ST1 SamplerSOIL SAMPLINGSTSTable 3.1: Testing Programme at the Laing Bridge SiteQ)0,E,— Q) 00oIH— 0 0 II‘Sc)—- I.) I) 001). S 8-C— N C)C.—* *— N C C)141L *C C C C C 1 ‘f C C C ‘r, N r C N N N C C C C N N N NNo.oc N‘_z Npa0o -— N C’ C N N C 1 C’ — t N N CN (:r- —:-:— —r-r--— C’, N flcE c,,r-, o—rCH ‘C 00-000 00\Q 00 000000V’/-;r-rNr- Nt-N‘ r‘? 00 ‘0 N N ‘C 0 ‘0’0 ‘0 ‘0i ,f“ “t N N ‘0 ‘f ‘ (N V ‘1 ‘1 N N(•‘,,(__N N N N ‘0 .J- — ‘- - ‘- — — —cNz * * * * * * *<.C)QC)QQL) QQQC)C)QQC)HH HWHHF C ‘fl C C C C C C C C C C ‘, C C C C( N N N C — — ( N00 , ,..., , - -, -,, QQQQLQUIDC’DQC)L)OQQQC)CCHH- CCCC— N N e r N 0000 C C C C N N C N N N Nh><‘i- - c C N . t Nkr’, ‘0 — N — N — — — — C N — — N — ‘r — — — N N’0, N N C C J C — f’ N C C 1 C’0 — —N ‘ c N ‘r 00 N — C (fl C’ N fl N — N N.’H— N..- N (‘) 4 A N 00 C — N r N 00 d.. C — N m ‘0,- C C C C C C C C C C C C N N N N C C CC0000000000000000000000000000000000000000000000I0000Ic-)142the influence of the jetting or equipment related variables in order to optimize the insertion procedure. Thisoptimization took in consideration the following key experimental variables:1. Different steel lanterns.2. Plugging of the cutting shoe.3. Different jetting rod positions.4. Different jetting systems.5. Dimensional differences along shaft.Testing soundings SBP2O to 23 were carried out to complement the database gathered to study theinfluence of the steel lantern on the unload reload shear modulus Gur. Full displacement pressuremetertests were performed from FDPO4 to 06 to assess the effects of variation of the rate of inflation on thederived testing curve.3.3 UBC SBPM EQUIPMENT CHARACTERISTICSDuring the last 13 years the department of Civil Engineering at the University of British Columbia hasbeen involved in research and development with the pressuremeter. In the early stages of this researchemphasis was placed on the full displacement pressuremeter test (Brown, 1985, 0’ Neil, 1985, Campanellaand Robertson, 1986, Hers, 1989 and Howie, 1991). The research work carried out with the fulldisplacement pressuremeter was performed with a pressuremeter probe that was an adaptation of theselfboring pressuremeter operated and developed by Dr. Hughes. The focus of the past research has beenmainly directed towards the equipment design and the testing methodologies of FDPM with a view todeveloping a cone-pressuremeter. A broad range of soil types have been used for this purpose using all theavailable UBC research sites in the Fraser Delta.With the build up of experience with the FDPM the research objectives of the In Situ Testing Group atUBC have shifted toward selfboring to obtain soil values for correlation to cone data. Early developmentsevaluated the selfboring installation, membrane protection and the development of a full automated dataprocessing / instrumentation system. This led to the construction of the UBC selfboring pressuremeter in1988 for Sully’s research (Sully, 1991).The equipment used for the tests of this thesis is a second UBC SBPM probe described by Cainpanellaet al, 1990, with additional modifications in the instrumentation system and the mechanical design. These143modifications led to a highly automated pressuremeter system, capable of full control of the expansion testand simultaneous measurement of several external variables during the insertion, testing and dissipationstages of the probe operation. A full description of the UBC SBPM system is presented in the Appendix A.This includes the detailed description ofthe UBC testing, pumping and pushing units.With the development of the UBC FDPM, and later the UBC SBPM, a comprehensive study of thetypes of steel “Chinese” lanterns (external shield) to protect the pressuremeter was initiated. This sectionpresents the main findings with respect to the use of such steel lanterns in the UBC SBPM.3.3.1 Steel Lantern CharacteristicsIn order to mitigate damage to the rubber membrane used in the expandable section of thepressuremeter, given the high frictional forces generated during the selfboring process, a stainless steellantern is required. It is in general made of flexible and curved steel strips that are longitudinally mountedand riveted or spot-welded together at the ends.Depending on the design ofthe lantern the steel strips can be:1. Overlapped and riveted together with a degree of overlap sufficient to reduce the existing gaps, thusdecreasing the soil ingress during the lantern deflation.2. Composed of butted strips bonded to a secondary rubber membrane, in order to avoid overlappingand soil ingress.Depending on the disposition and dimensions of the overlapping strips, and position and concentrationof welds along the longitudinal strips, different designs can be achieved. These differences in the designslead to differences in the lantern “membrane resistance” and system “compliance” corrections required toreduce the raw pressuremeter data (see calibration results in Appendix B).Butted steel strip lantern is the current standard protective sheath adopted by Cambridge In Situ, Ltd.for the English Camkometer. This lantern (defined here as “Camkometer” lantern) was reported by Faheyet al, 1988, and has the great advantage of preventing soil to ingress in between the strips. This addedadvantage, however, introduces a more sophisticated and difficult to manufacture design. It isconsequently much more expensive than the overlapped and welded steel lantern.Past research with the FDPM (0’ Neil, 1985, Brown, 1985) has indicated that depending on thelantern and membrane characteristics distinct corrections must be placed over the measured pressuremeter144results. Commercially available overlapping steel strip lanterns protecting rigid urethane membranes haveshown to produce both high and rate dependent membrane resistance corrections. This led to thedevelopment and construction of a IJBC lantern and membrane system for the UBC pressuremeter in weaksoils. A simple rubber membrane and overlapping steel strip lantern are extremely attractive, given the lessexpensive and quicker ways of replacing the lantern in case of accident in the field. Another reason toproceed with the development of the UBC steel lanterns was related to the maximum circumferential strainsthat were desired during the pressuremeter expansion. In the early stages of the UBC FDPM(0’ Neil, 1985, Brown, 1985) development, emphasis was placed in the expansion of the test up to acircumferential strain of 20 % in order to overcome the disturbance effects generated during insertion.Since the standard Camkometer lantern is designed to achieve a maximum strain of approximately 10 %(Hughes, 1991 personal communication), it did not satisfy the testing requirements of that time. The UBCsteel lanterns, on the other hand, could be designed to fulfill this requirement.Depending of the lantern design, the steel strips can be either clamped at both sides or screwed (orriveted) to a lantern retainer. The steel strips can have distinct dimensions and curvature, and can begrouped together by spot welding in order to increase the resistance of the lantern and its capability towithstand soil intrusion. The individual strips used in the steel lanterns of this research are commerciallyavailable 316 stainless steel strips manufactured by Rocktest Inc. for their pressuremeter. These stripshave an approximate thickness of 0.3 mm and can be obtained in distinct widths and lengths. Thecurvature of the individual strips used was lower than the curvature of the UBC SBPM probe. The stripswere secured in place by the clamping action of both bottom and upper lantern retainer rings.Unlike the UBC steel lanterns, the Camkometer lantern accounts for the effects of soil ingress by theadoption of an inner membrane bonded to the steel strip. With the use of 18 flat and narrow strips (12mm), it is possible to have a final lantern extremely well adjusted to the body of the pressuremeter. TheCamkometer lantern is screwed at both ends to steel rings. These steel rings have the freedom to movelongitudinally (as the membrane expands) by modifying the original lantern rings to be fully floating.Figure 3.2 presents the layout of the designs of the UBC steel lanterns, plus the layout of theCamkometer lantern. This figure is complemented by Table 3.3, were the general characteristics of each ofthe lanterns are presented. The main points of interest are:LANTERN TSPE SPOT WESDED STRPSB 1Strip ______________FSpot Weld8 7 TStrip(1 Major Longitudinal Gap)8 8 1___ ____ ____ _ __ _ __ _ __ _ ___StripF(2 Halves)8 1StripFB 10[Inner MembraneCONVENTIONS:• Spot Weldor Steel Strip Lunction or CapB Bottom of LanternT Top of LanternCBS:The bottom of the lantern is near the cutting shoeDrawings not to scaleFigure 3.2: Steel Lanterns Adopted in the UBC SBPM145146LANTERN STEEL STRIPS CHARACTERISTICS OF STRIPS(mm)LENGTH WIDTH1 540 16 Welded at both bottom and top to a collarI row of strips overlapped. Movement in between strips2 450 16 Welded at bottom to a lantern retainer3 rows of strips welded together3 470 24.5 3 rows of strips welded togetherMovement in between welded group of strips4 500 24.5 Welded at bottom and top by 2 lines of rivets2 rows of strips welded togetherMovement_at both_major longitudinal_gaps5 470 24.5 Welded at bottom and top by 2 lines of rivets3 rows_of strips welded together6 480 24.5 Same as lantern 5, with 3 sections instead of 2 halves7 480 24.5 Welded at both bottom and top1 row of strips overlappedMovement in between strips8 470 24.5 Welded at both bottom and top by 1 line of weld spots3 rows of strips welded togetherMovement at both major longitudinal_gaps9 480 17.5 3 rows of strips welded togetherMovement in between welded group of strips10 545 10 Canikometer type lantern18 flat steel strips bounded to an inner rubber membraneNo overlapping and extremely high flexibilityFully_floating_lantern_retainersTable 3.3: General Characteristics of the Lanterns Used1471. Lanterns 1 and 2 (not shown) were the original lanterns adopted in the initial design of the UBCFDPM. They have not been used in the present research.2. Lantern 3 has a similar steel strip configuration to Lantern 2, but has a wider strip which isclamped at both top and bottom by the lantern retainers.3. Lanterns 4 to 6 were soon discarded for any future use with the UBC SBPM. Lantern 4 proved tobe extremely weak for usage in sands, as it was easily destroyed in SBPO7. The calibration tests of lanterns5 to 6 indicated problems of differential expansion along the radius.4. Lantern 7 has a similar design to Lantern 1, but it is shorter in length and has wider strips. It isriveted at both bottom and top sections and has a major longitudinal gap for assemblage purposes.5. Lantern S has the same design as lantern 3 with 1 row of spot-welds at both bottom and topsections. It constitutes 2 halves that move independently. It has a much higher density of spot-welds thanlantern 3.6. Lantern 9 has exactly the same design as lantern 3, but it adopts a shorter width steel strip.7. Lantern 10 is the standard Canikometer lantern.In this thesis a study of the UBC SBPM lanterns was carried out with the analysis of results of severalfield tests in which specific lanterns (see Table 3.2) were tried out. The influence of the different lanternson the unload reload shear modulus Gur of the SBPM was investigated, as shown next.3.3.2 Effects of the Lanterns on GurSince Gur is much less affected by disturbance (Janiiolkowski et al, 1985) than any otherpressuremeter variable, it was possible to study the isolated effect of the lantern with results from differenttest soundings (each sounding had a different imposed disturbance, as will be shown later). All the Gurmoduli compared here were corrected for compliance effects, as recommended by Fahey and JewelI, 1990and described in Appendix B.Generally, Gur will decrease non linearly with increasing stress reduction of the unload ioop, orincreasing cavity strain amplitude (ds0) mobilized during unloading. The pressuremeter Gur modulus isalso affected by the average plane strain effective stress mobilized at the plastic zone, increasing with anincrease in the effective pressure at the cavity wall at the beginning ofthe unloading stage (Pj.148In order to isolate strain effects in the tests performed herein it was decided to compare only modulusvalues from loops with similar strain amplitudes. Additional care was taken to not exceed the maximumunloading pressure for each loop (as recommended by Wroth, 1982), thus avoiding reverse plastic failure.In this thesis loops with a degree of pressure unload (dP I P) of 40 % were used for the comparison. Thisled to loop cavity strain amplitudes between 0.1 and 0.3 %.Since the Gur is also pressure dependent only the loops with a P, close to 360 kPa have been used inthe comparison. This value was chosen on the basis of the pressure that was customarily reached at thecavity wall at the beginning of the holding phases before unload. For the testing program with lanterns 3 to9 the holding phase started with a E around 2 % whereas for lantern 10 it started at 3 %, in accordancewith the “command files” created (see Appendix A).A holding phase was adopted before the unload reload loop to reduce the creep influence over themeasured Gur. A minimum hold time of 8 mm. was adopted to yield a final target creep value of0.01 %/ prior to wiloading.It was decided to directly compare the slope of the loops, i.e. the uncorrected for in situ stress levelGur. The pressuremeter Gur modulus was also compared to the profile of (low strain) maximum shearmodulus G0 at the Laing Bridge site, as determined by the downhole seismic cone. The G0 modulus isequivalent to a shear strain amplitude of % with an equivalent mean effective stress equal to theoriginal in situ stresses. Since strain and stress differences of G0 and Gur will tend to give different modulifor the same depth tested the Gur/G0ratio is of interest.The results of Hughes and Robertson, 1984 for SBPM and FDPM tests in the McDonald’s Farm site,Vancouver, indicated ratios of Gur/G0from 0.5 to 1.05. These tests were carried out with loops with strainamplitudes in the range of 0.2 to 0.3 % and below 600 kPa. The G0 was obtained from downholeseismic cone tests. Bellotti et al, 1987 obtained GurIG0ratios from 0.2 to 0.7 for calibration chamber testswith Ticino sand at relative densities of 40 to 80 %. The unload reload loops had an average d80 of 0.1 %and P, below 800 kPa. The G0’s were obtained by resonant column tests on identically preparedspecimens. Bruzzi et al, 1986 obtained Gur/G0 between 0.5 and 0.7 in Po River site, Italy (Borehole4017), for pressuremeter loops with an average 0.08 % dse and P below 740 kPa. The G0 was measuredby cross hole seismic tests.149Thus, the available literature on this subject indicates that for sands with similar densities as LaingBridge, and pressuremeter tests with P, and ds0 in the range of those adopted here, a Gur/G0 ratio from 0.5to 1 (closer to 1) is to be expected.Figure 3.3 presents the Our results for different testing soundings, in accordance with the field testingprogram presented before. All steel lanterns, except the no. 10 (Canikometer lantern), gave GurIO0 ratioswell in excess of 1 suggesting the existence of a mechanical stiffness built into these lanterns. TheCanikometer lantern gave GurIG0 ratios between 0.5 to 0.9 in agreement with the above published values,valid for loops with similar characteristics as those carried out in this research.The high values of the measured Gur/G0 can be directly related to overlapping of strips, i.e., thegreater the overlap length (or area) the higher is the GurJG0. Because of the lateral stress a ffictionalresistance is developed at all overlaps and with sand ingress this ffiction can be further increased. Thehigher the ffiction, the stiffer the membrane will be since a relatively high internal pressure increase ordecrease is needed to initiate strain. Therefore, only lantern no. 10 (Canikometer lantern), which has nooverlapping can be expected to give accurate measurements of Our without a friction-stiffness component.This subsection discussed the influence of the adoption of the steel lanterns in the UBC SBPM. Theoverlap lantern friction had influence on the unload reload Gur modulus. This friction added to the effectgenerated by the compliance of the lanterns. The friction effect was responsible for very high Our valuesfor all steel lanterns. The UBC steel lanterns are generally not suitable for modulus evaluation via unloadreload ioops, unless the friction effect is somehow incorporated in the compliance calibrations of theselanterns. The Camkometer lantern (no. 10) gives reliable Our measurements as there is no friction effectadded to the compliance of this lantern. Therefore, the Canikometer lantern is the one recommended herefor usage with the UBC SBPM under optimum insertion procedures.3.3.3 Method of InstallationIt should be noted that in the UBC SBPM system selfboring is achieved by jetting rather than bycutting the soil (as used in traditional SBPM systems). The jetting in sand was introduced by Hughes etal, 1984, and uses high velocity jets of mud to break the granular soil during insertion. Jetting wasbasically devised to simplify the design of the selfboring equipment as well as to speed-up insertion andrterr, Effct ‘e,- 5Heer- M 8ridgeFigure 3.3: Lantern Effects on Gur150o 50 100 150 200 250 300 350 400 450 500I I I++0Averoqe VoIues Adopted for the Loops Selected:Pcu=360 RPO, Docree of Locding=4O% of Pcu000000 A0GD0274—5—6—7—10 —11 —12131415 —16171180CoTout fld LantnSoIlm Cone± Sapoa-LNr3>< SBPO5-I..NT3o.O S8P10-LNTSSBPI 1-LNT9SBPI2-LNT9* SBP13-LN9-------SBP1S-LNTIO----s---- SBP17-LNTIO. SBPtS-LNTIO151improve production. This is particularly important in offshore environments. Although it was conceivedalmost 10 years ago, jetting still represents a new concept of inserting a SBPM in the ground.The basic modification of the jetting system in relation to the traditional system is the substitution ofthe cutter bit by a roller bit (stiff clays) or a jetting nozzle (soft clays and sands) (Clough et al, 1990). Theholes in the jetting nozzle constrict the fluid flow, increasing considerably the jet velocity. These jets aregenerally directed upwards in the direction of the surface, so that washboring in front of the shoe does nothappen. The mud shall be puniped at sufficiently high pressure to break the soil during the jetting (orselfboring) process, and, as in the traditional system, a trial and error procedure is required to optimize theinsertion variables in a new environment.Since one of the objectives of this chapter is the definition of the UBC SBPM insertion procedure, i.e.,the optimum combination of drilling variables that speeds the insertion and minimizes disturbance, theknowledge of the previous experience gathered by other researchers in this respect is of interest. Thisexperience is reviewed next.3.4 REVIEW OF INSERTION PROCEDURESThe installation of the selfboring pressurerneter greatly affects the quality and reliability of the derivedparameters. The key to the insertion process is to balance the removal of soil in front and inside the cuttingshoe with the speed of advancement and other drilling variables, like mud pressure and velocity. Theoptimum insertion should not considerably alter the stress regime nor the density of the surrounding soil.3.4.1 Key Insertion Parameters in ClaysMost of the existing experience is related to selfboring in clays (Denby, 1978, Benoit, 1983,Atwood, 1990, Findlay, 1991 and others). With the “traditional” rotating cutter system the parameters thatare generally controlled are those that relate to the equipment design (dimensional tolerances, position,rotation and type of cutter bit), to the rig characteristics (pushing force), to the drilling equipment (fluidflow and pressure) as well as the rate of advance.The pushing force and mud pressure are in general controlled by the driller in charge of the selfboringoperation. If the rate of flow return suddenly decreases and the pressure increases it may be possible thatthe shoe got plugged. In sands the possibility of shoe plugging is less likely to occur than in clays (with152exception to dense to very dense sands) and a sudden increase of flow pressure may be indicative of anothereffect, most probably the increase of the density of the selfbored sand.Insertion should be ideally canied out at a constant rate, which requires varying the pushing force tosuit the variable soil conditions. The rate of penetration depends on the type of soil selfbored and thecharacteristics of the cutting or jetting systems. As a starting point Ghionna et al, 1981 suggested a valuenot greater than 120 cm/mm (in clays) to minimize soil disturbance.The position of the chopping bit is important in the sense as it can influence the stress regime of thesurroonding soil. If the cutter is set too far behind the cutting shoe then an increase of the level of stressesin the surrounding soil can occur, whereas if the cutter is set too close to the cutting shoe a reduction of thelevel of stresses occurs.Another important aspect is the diameter of the shoe, lantern and overall equipment in relation to theselfbored borehole. Discrepancies in diameter can arise due to equipment imperfections, differential rubbermembrane thickness or improper shoe sizes. Law and Eden, 1980 demonstrated that an oversized cuttingshoe (or any other section of the SBPM) creates a gap between the borehole and the pressuremeter probe,causing stress relief An undersized cutting shoe imposes a certain load to the surrounding soil prior to thepressurenieter test.The jetting system advocated by Hughes et al, 1984 for the SBPM insertion suffers from the sameoptimization needs as the “traditional” cutting system. The only difference is the fact that the actionproduced by the rotation of the cutter is replaced by the action of the high velocity jets from each orifice ofthe jetting nozzle. A comprehensive study program carried out in the University of New Hampshire(Atwood, 1990) with “traditional” and jetting systems demonstrated that the key optimization parametersfor jetting in clays are the same as those (related above) of the “traditional” system. The optimum drillingtime of the jetting system has been shown to be at least of three times faster than equivalent drilling times ofthe “traditional” system. The water flows were, however, five times higher than those of the “traditional”system.1533.4.2 Key Insertion Parameters in SandsIn sands the key insertion parameters are the same as those related above for clays, for both“traditional” and jetting systems. This is supported by the experience related by different researchers forSBPM insertion in sands, as presented in Table 3.4.For the “traditional” system typical penetration rates vary from 1.3 to 4.0 cm/mm. The cutter rotationcan vary from 40 to 400 rpm, and can be located as fir as 5 cm behind the shoe. A non uniquecombination of drilling variables appears to exist for selfboring in different sites. This is also the case ofthe few documented jetting insertions (Hughes, 1984 and Howie, 1991). For the jetting system typicalpenetration rates vary from 24 to 220 em/mm, with mud flows as high as 78 11mm. The jetting nozzle(center of orifices) is in general located between I to 2.5 cm behind the cutting shoe. Similar to the case ofclays, both the penetration rates and the mud flows adopted for the jetting system are considerably higherthan those required by the “traditional” system.The large variation related for each of the key insertion parameters of Table 3.4 reflects the variablesoil conditions (density, stress level, etc.) encountered by each of the authors during the selfboring process.It is concluded, therefore, that the insertion of the SBPM has to be optimized in each new granular deposit.The values related in Table 3.4 serve as a starting point for the establishment of the optimum combinationof the drilling variables.In sands, however, the frictional forces that develop between the soil and instrument can be very high.According to Fahey and Randolph, 1984 the optimization of the insertion procedure in sands is morecritical than in the case of clays, as the derived parameters (via traditional methods) are extremely sensitiveto the high frictional forces developed in the soil-probe interface, as well as the disturbance generated byimproper drilling variables.In summary, the above review shows that several variables have to be considered in the fieldoptimization of the SBPM insertion. Granular deposits are more prone to disturbance during the selfboringprocess than deposits composed of fine particles. This is caused by the fact that high frictional forcesdevelop in the soil-probe interface during selfboring in sands, and by the lack of cohesive bonding betweenthe granular particles. For clays there is a relatively large amount of experience to influence the154TEST LOCATION D50 Dr C / N’ MUD2 CIJITER RATE OF REFERENCE(mm) (%) POSITION FLOW TYPE ADVANCE(cm) (1/mm) (cmlmin)Pacifica Site, USA NA NA 0.6 4300 Spade Type 0.5-1.3 Bachus, 1983Lightly Cemented Sand (air)English CamkometerPo River Valley, Italy 0.2-0.4 NA 1.5-4.7 10-16 Helical Type 2.0-3.5 Bruzzi et al, 1986Slightly Aged Medium DenseSand, OCR 1 to 2English CamkometerNorwegian Dense Granular NA NA 3.0-5.0 NA Spade Type 3.0 Lacasse et al,Deposits 1990English CanikometerMcDonald’s Farm, Canada 0.1-0.6 40-60 NA NA Spade Type 2.0-4.0 Robertson, 1982Loose to Dense SandHughes PressuremeterTRADITIONAL SYSTEMMcDonald’s Farm, Canada 0.1-0.6 40-60 1.0-2.5 20-43 Jetting Nozzle 42-220 Hughes, 1984Loose to Dense SandHughes PressuremeterLulu Island, Canada 0.13 to 50-60 2.0 54-78 Jetting Nozzle 24-90 Howie, 1991Medium Dense Sand 0.17Hughes Pressuremeter1-Rotating cutter (traditional system) or jetting nozzle (jetting system)2-Mixture of water and bentonite mudJETTING SYSTEMTable 3.4: Successful SBPM Drilling Variables in Sand155establishment of guidelines with respect to the optimum drilling variables. Less experience exists forinsertion in sands, principally with the jetting system. For any of the cases a trial and error procedure hasto be established at each new site to properly define the optimum combination of drilling variables.The main effect of disturbance on the testing curve is the change of its shape in relation to the“undisturbed” shape. Since the “undisturbed” shape is unknown, it becomes difficult to access the degreeof disturbance present in the curve due to the improper insertion process. If the numerical degree ofdisturbance is known after each field trial with the SBPM it becomes easier to readjust the drillingvariables towards the “optimum” general combination. Several testing results originating from differentfield trials can be then successively assessed to indicate which combination of drilling variables minimizesthe disturbance of the testing curve. In order to do that a procedure to numerically quantify the disturbanceof the testing curve must exist. This procedure is introduced next.3.5 NUMERICAL QUANTIFICATION OF DISTURBANCEThe major benefit from an optimization routine carried out in the field is the faster selection of theoptimum drilling variables, which depends on the subjective opinion of the pressuremeter operator in regardto defining the “undisturbed” testing curve. As a basis for what can be defined as “undisturbed”, it iscommon to use the accumulated experience reported in literature. This experience indicates that highquality curves in both sands and clays are invariably characterized by a continuous and smooth curvilinearshape throughout the strain range of the test.Denby, 1978 and Benoit, 1983 observed that disturbance in pressuremeter tests in clays tends to“flatten out” the initial stage of the pressure expansion curve, leading to an almost linear shape.Disturbance can be present in the pressuremeter curve in different amounts, and hence different disturbedshapes will exist. Wroth, 1984 noticed that highly disturbed tests on sand can be easily determined by thevisual inspection of any pressuremeter operator. In this extreme case there is no dispute whether the curveis “undisturbed” or just “slightly” disturbed.Robertson, 1982 noted that there is no generally recognized criterion for the assessment of the qualityof SBPM testing curves. In general a subjective approach by the visual inspection of the testing curve isused. Based on his experience with the SBPM Robertson, 1982 tried to compile the basic requirementsthat would serve as a guide for the selection of “undisturbed” testing curves. These requirements are:1561. No inflection point near the beginning of the curve.2. All arms move in a consistent manner leading to lift offs that are close together in strain.3. The pore pressure measured before the test is close to the hydrostatic value.However, these requirements are broad and undoubtedly different pressuremeter operators will stillhave a different approach to classify the disturbance of the pressuremeter curve. The major drawback ofthe visual inspection routine is the lack of a recognized “undisturbed” reference curve. General statementsas those above can lead to different conclusions, because of each pressuremeter operator’s concept of whatshall be the smoothness and roundness of the “undisturbed” curve.Findlay, 1991 presented a methodology to overcome the pitfalls above and numerically quantify thedisturbance of pressuremeter tests in clays. He compared the shape of pressuremeter curves, obtained fromtests with various degrees of disturbance, to an empirical coefficient that was related to the initial slope ofthe testing curve. This coefficient was related to the slope developed within 1 to 5% cavity strain, since thefindings of Benoit, 1983 and Denby, 1978 demonstrated that disturbance is mainly concentrated in theinitial stage of the test. Findlay, 1991 defined typical disturbance values for “undisturbed” SBPM and(disturbed) push-in pressuremeter tests, in order to guide future expected coefficients in each case. Themajor drawback, however, was the fact that the obtained range of coefficients was site specific, andtherefore typical undisturbed values for one particular clay deposit couldn’t be used at a different site. Thisdrawback removes the universality of Findlay’s suggested approach, and may not lead to a much bettercriteria than the one previously used for clays and sands based on a visual inspection of the curve.On the other hand, with the concepts stated by Findlay, 1991 and the fitting technique advocated in theprevious chapter, one can formulate a disturbance criteria that could be universally applicable. With afitting technique the reference “undisturbed” curve is known, since it is defined by the idealized modelcurve. It should be noted that this curve will differ for each site and depth in accordance with the fitted soilparameters (ah, Gi, ‘p). The higher the disturbance the higher will be the influence in the initial stage of thefield curve, and hence the greater the deviation from the idealized (“undisturbed”) model curve. If thisinitial deviation, or differential area as presented in Figure 3.4 can be numerically quantified, than it shouldbe possible to quantify the disturbance of the test. Similar to the empirical coefficient of Findlay, a157c2Idealized Model CurveField Testing CurveArea is Proportional to DisturbanceCircumferential StrainFigure 3.4: Disturbance Evaluation of SBPM Testing Curve158numerical coefficient of disturbance (CD) was devised for this research, which measures the averagedeviation at 1 and 3 % circumferential strain for a testing curve. This range is based on the experience ofthe writer and literature findings regarding the zone most affected by disturbance. It is proposed that thenew CD parameter is defined by the following equation:+ Pci(3%)—Pc(3%)’1x 100 (%) (3.1)Pci (1%) Pci(3%) ) 2where:Pci is the idealized cavity pressure by the model at the circumferential strain given;Pc is the equivalent testing cavity pressure at similar strain.The CD value varies from 0 % (fully undisturbed) to 100 % (highly disturbed) for SBPM tests inwhich, in general, the idealized model curve stays above the field curve. Since CD is defined as anormalized ratio it can be used with any of the existing cavity expansion models discussed on Chapter 2.The use of a stress ratio, rather than a direct measurement of an area provides an analogous yet simplerevaluation of CD in the field and thus leads to a faster derivation. With a computer and the QuickBasicprogram mentioned in the last chapter, the CD can be easily obtained with the direct evaluation of the fittedcurve on the screen of the computer.It can be noted that CD is not an independent variable, but rather a variable that depends on the strainrange of curve matching. As a general rule, it is recommended that the curve fitting is performed between5 to 10 % circumferential strain for either undisturbed or disturbed data. Once the model curve ismatched a value of CD can be devised. It is suggested here that for CD’s below 10 % the field curve canbe considered of high quality (“low” disturbance). For CD’s from 10 to 30 % the field curve is of good tomedium quality (“medium” disturbance). For CD’s higher than 30 % the quality of the field curve is low(“high” disturbance). As presented in Table 2.6 the reliability of the predicted parameter is expected to beproportional to the quality of the curve, hence the numerical value of the coefficient of disturbance. ForCD’s lower than 10 % the reliability of the derived parameters is high. The reliability of “medium” quality159field curves may still be high, although reduced to some extent. The reliability of field curves with CD’sabove 30 % is probably low.An initial assessment of the usefulness of the numerical quantification of disturbance is given inFigure 3.5. This figure presents the calculated CD of one of the testing soundings (SBP 13) in which shoeplugging occurred to an extent of 25 % of the cross-sectional area of the cutting shoe. In the same plot isshown the plane strain 4) by both the traditional log-log approach of Hughes et a!, 1977, and the curvefitting methodology discussed in the last chapter (with Hughes et al, 1977 model), It can be seen thatdisturbance is not homogeneous throughout the profile, but it varies with the characteristics of the soiltested and insertion difficulties. The data suggests that all the 4) determinations by the log-log plot wereconsiderably influenced by the initial disturbance of the test, except at 8.4 and 10,5 m for which the CDwas only 24 % and the log-log 4) agreed with the curve fitted 4). The data also indicates that the curvefitting technique is less disturbance sensitive than the traditional log-log interpretation approach, as thevariability of the curve fitted 4)’s is low. This finding substantiates the comments above regarding thereliability of soil parameters predicted with slightly disturbed (“medium” disturbance, 10 % < CD <30 %)curves, and the usefulness of the CD coefficient.3.6 INSERTION PROCEDURE FOR THE UBC SBPMA mud flow and penetration rate that was close to the literature “optimum” values of Table 3.4 wasadopted. For this research an average mud flow of 20 1/mm (5 to 6 gpm) was adopted during insertion.This allowed an average penetration rate of 50 em/mm (0.8 cm/s) throughout the testing depths up to the10 Tonne capacity of the pushing rig. Since the pushing and pumping units shared the same source ofhydraulic power from the research truck (see Appendix A), after the attainment of the operational capacityof the rig “extra” hydraulic power had to be obtained at the expense of the pumping power of the system.In general, in the densest layers of the site (below 8 to 9 m depth) the rig capacity was reached. Figure 3.6presents a typical output of the “sounding” file of the pressuremeter. It demonstrates that beyond this depththere was a slight decrease in the flow rate employed, followed by a variation of the penetration rate aroundthe mean value of 50 cm/mm. This variation was not excessively high and all the testing soundings used inthe comparisons of this thesis suffered from similar problems.20I I60I — _i160Variabe40 8010-0V—• —I -.I./ /00000 CD %).-..-.. cP Deg), Curve Etting Methodology• — — Deg), Log—Log Methodology20 Intluence 0? LU on !rcaltIonc [-rictlon angie VCIUCtIOflSBP13 — 9JN92 — HUGHES ET AL, 1977 THEORY — LAING BRIDGEFigure 3.5: Practical Numerical Assessment of DisturbancePor p£ -p a w aPnoLrationResistdnce(EN)MudF1ov(1/mm):)iIL y•ii•JPresurIkP)0005010141500102030PennLiLionRdL01cm/mm)2040E,00141001PJ(flL]LuSi1L ySCliiUL)uLMudIui[JunueFigure3.6: TypicalVariablesObtainedwiththeSoundingFileof theU]3CSBPM162Sounding results like the ones of this figure were used in conjunction with the field characteristics ofthe test (shoe plugging, equipment characteristics, settings, etc.), and the CD defined above to drawconclusions regarding the insertion quality. This quality controls the amount of disturbance that is imposedin the pressuremeter testing curve.Based on the experience ofthe writer throughout the experimental testing stage, the insertion quality isa function of the characteristics of the SBPM probe (type ofjetting system and dimensional tolerances), thedrilling variables (position of the jetting rod, mud flow and pressure, rate of advance) and the unexpectedoccurrences in the field (as the plugging of the cutting shoe). The influence of each of these variables uponthe disturbance generated during selfboring will be assessed in the following subsections. This aids in thedevelopment of the best (insertion) procedure to reduce the likelihood of disturbance in the UBC SBPMresults. The best insertion procedure at the Laing Bridge site is the one that leads to testing curves in whichthe CD parameter is below 10 %.3.6.1 Shoe PluggingThe existence of two layers with differing soil characteristics (see Figure 3.6) in Laing Bridge sitewould require different jetting variables during insertion, i.e. distinct jetting flow velocities would be neededto selfbore the surficial silty sand to sandy silt material and the underlying clean sand strata with minimumdisturbance. This was not possible because the UBC SBPM pumping unit was not capable of producingthe high flows for optimum” insertion in flne-grained soils (see Clough and Denby, 1980, Lacasse andLunne, 1982 and Atwood, 1990). As such, partial plugging of the shoe with the surficial fine-grainedmaterial was invariably observed in many of the SBPM soundings. This section addresses the likely effectof partial plugging during selfboring, and how the insertion procedure can be varied to deal with thisproblem.The total plugging of the shoe can, as a first approximation, be visualized as the penetration of a fulldisplacement in situ device into the soil. Full displacement pressuremeters (or piezocones) have a solidcone apex of 60° in front and in the position of the cutting shoe.The penetration of a piezocone in a granular medium is a very complex phenomenon and difficult toanalytically model. When the tip of the cone passes through an element of sand, very high stresses aredeveloped, which is followed by a substantial unloading of stresses as the cone proceeds further and the soil163element passes the shoulder of the cone. Moreover, experimental results of Chong, 1988 indicate that looseand dense sands tend to dilate after the insertion of a piezocone.Hughes and Robertson, 1985 have qualitatively discussed the effects of the cone passage in sands andpresented arguments regarding the possible stress history of the elements of soil after the cone passage. Asthe cone approaches any element of soil the stress path of this element will move towards the failure line ofthe material, arid then will follow it. Effective radial stresses as high as several orders of magnitude greaterthan the original lateral stress can be hypothesized at this stage. The stress path will follow the failure line,pushing the yield surface until the soil element passes through the shoulder of the cone. At this stage thestress path follows an unloading path below the yield surface, and reverse failure (with the circumferentialstress as the major principal stress) is likely to occur. All the elements of soil will follow a similar stresspath, but elements that are located at a greater distance from the probe shaft will be stressed to a lowerextent. An annulus of soil with high residual stresses will exist beyond the soil elements close to the shaft,probably due to arching effects. Based on speculations of Withers et al, 1989 a zone of intense shearingand discontinuities may also exist between the probe shaft and the end of the induced plastic zone.Therefore, it can be expected that extremely complex stress, density and strain histories will beimposed in the soil in the event of total (presumably extending to partial) plugging of the cutting shoe. Thedisturbance mechanism in the SBPM is further complicated by the fact that plugging will not occur evenlyaround the shoe surface, and therefore the disturbance effects will not be evenly distributed in the soilaround the equipment.Shoe plugging effects were considered in the comparison of two test soundings, 5 m apart (SBP11 andSBP 12) in which the plug in the field, estimated by visual inspection of the shoe after withdrawal of theprobe, varied from 25 to 75 % of the total cross-sectional area of the cutting shoe. Similar equipment andjetting variables (rate of penetration and flow) were adopted for these two tests. Typical averaged (of the 6strain arms of the UBC SBPM) testing curves are compared in Figure 3.7 for one testing depth. It can benoted that the larger the plug the larger will be the lift off stress (indicating high residual stresses set up atthe soil-probe interface), as well as the CD parameter. The post lift off stage of the curve will reflect theoverall loose condition of the soil surrounding the SBPM, after the passage of the partially plugged SBPM.164DCCC!)Q)Testing Curves from Tests with Diff. Shoe Pluggings2000,-N180006001 40012001 000800040020000CcLiCircumterentia Strain (%)Figure 3.7: Typical Comparison of Curves from Tests with Different Shoe Pluggmgs165The influence of plugging in the original soil conditions can be also “picked up” by the total (tip plusfriction in the lantern) load measured by the load cell of the UBC SBPM. The results of Figure 3.8 suggestthat there is an increase of the penetration resistance with the increase of the percentage of shoe plug. Aswell, there is a correspondence between the increase in CD with the increase of penetration resistance.This subsection discussed the effect of shoe plugging on the results. It is suggested that the higher isthe percentage of shoe plugging, the higher will be the load imposed by the cutting shoe in the virginground during selfboring and hence disturbance (or CD). To reduce plugging of the shoe the followingprocedures were found to be effective:1. Since the plugged soil was always lightly plastic and fine, the insertion procedure in that materialwas changed to a slower rate of penetration ( 25 cm/mm) and sometimes a higher pumping rate( 30 11mm) (more close to values obtained by Atwood, 1990 and others in clay). The slower rate can benoted in the sounding profile of Figure 3.6.2. It was found necessary to prebore through the surface 1 to 3 m with a large dummy cone or with apush-in casing to by-pass the fines responsible for plugging.3. The jetting rod position was gradually optimized, as will be discussed next.3.6.2 Jetting Rod PositionThe jetting rod position and the mechanical design ofthe nozzle are important variables that have to beoptimized for the SBPM insertion process. They are used in conjunction with the specified mud flow andpenetration rate to reduce the amount of tip load in virgin ground during selfboring, which in turn reducesthe likelihood of soil plugging.Wroth, 1982, observed that the amount of disturbance that can be generated in the soil is stronglyrelated to the position of the cutter bit in traditional SBPM systems. If the cutter or the jetting nozzle(assuming that similar disturbance effects would operate in both selfboring systems) is placed too farbehind the edge of the cutting shoe then a temporary plug is fontied inside the shoe. In contrast, if thecutter or jetting nozzle is placed too close to the edge of the cutting shoe then washboring or stress reliefmay occur ahead of the shoe. A balance will exist between these 2 extreme positions and may be obtainedin a trial and error field optimization, as commented previously.C g 4.(rn)flur[131:3ci D ID 12I-(ccccIL cc flEZth——-I—----I-----I—cc)--cmcciI-i-0.)ccccamcc)amQTh31)1-om)--3120•(I)(.1) (1)1!)wwww—NJ1ccmmcccc3-1jc100uluimlu3-1UIpCc)CD -D j) CD 0 C) CD C) C) CDam ccccNJcccm cc167In this thesis the optimum setting of the jetting rod (or nozzle) of the central jetting system was refinedby the gradual optimization ofthe insertion routine, in order to reduce the amount of disturbance to a targetCD of 10 %. Since one of the major problems during selfboring at the Laing Bridge site was the pluggingof the cutting shoe, the optimum rod position assisted in the reduction of this occurrence.Table 3.5 and Figure 3.9 present the trial and error process to optimize the rod position and the designof the jetting nozzle. The optimum rod position was achieved simultaneously with the process of definitionof the insertion procedure for the UBC SBPM. Throughout this process, the gradual improvement of thenumber and diameter of the orifices of the jetting nozzles, and the gradual decrease of the jetting nozzledistance in relation to the edge of the cutting shoe helped to reduce the likelihood of shoe plugging.Optimum shoe plugging percentages of 0 % were obtained only after SBP17, as observed in Table 3.5.The optimum jetting rod position was obtained with the nozzle tip located at -5 mm (tip outside the shoe) inrelation to the cutting shoe. This leads to a distance of 18.4 mm between the centerline of the orifices of thejetting nozzle and the edge of the cutting shoe. This value is in between the values reported byHughes, 1984 and Howie, 1991 in Table 3.4.Figure 3.9 demonstrates that with the gradual optimization of the jetting rod position there was aconsiderable reduction of the amount of disturbance generated during selfboring. The quality of the testingcurves increased from SBPO1 to SBP19, such that in testing soundings SBP18 and 19 the majority of thecurves presented “low” disturbance characteristics based on the CD parameter. The most relevant aspectsof the comparison presented in Figure 3.9 are given by:1. The testing soundings considered herein are the ones in which the central jetting system forselfboring was used. The shower head system (both systems are presented in Appendix A), adopted inSBP14 and 15, effectively eliminated plugging yet still yielded pressuremeter curves with “medium”disturbance characteristics due to overcoring during insertion.2. All the testing soundings of this figure presented a variable effect of disturbance along depth. Mostof the peaks of disturbance tend to be aligned at similar depths for all testing soundings. One example isbetween depths 9 to 9.5 m. In this range the highest CD of SBP19 is obtained. A careful inspection of thesounding results of this testing sounding, expressed in Figure 3.6, reveals that at this depth the measuredTEST ROD POSITION’ NOZZLE SHOE PLUGGING2SOUNDING (mm) TYPE (%)SBPO1 50 A NASBPO2 35 A NASBPO3 30 A 50SBPO4 30 A NASBPO5 20 B 50SBPO6 20 B 50SBPO7 20 B 75SBPO8 20 C 50SBPO9 10 C 0SBPIO 10 C 50SBP1I 10 C 75SBP12 10 C 25SBPJ3 0 C 25SBP14 SH -- 0SBP15 SH -- 0SBP16 -5 C* 50SBP17 -5 C* 0SBP18 -5 C* 0SBP19 -5 C* 0OBSERVATIONS:SH = Shower Head Jetting System, NA = Not Available1-Distance from the edge of the cutting shoe to the tip of the jetting nozzle. Central jetting system2-Sectional area covered by plug material over the total sectional area of the cutting shoeNOZZLE TYPE:A= 8 holes of 3.55 mm (diameter) @21.3 mm from nozzle tipB=3 holes of 5.7 mm (diameter) @ 20.6 mm from nozzle tipC=4 holes of 4.0 mm (diameter) @ 23.4 mni from nozzle tipC” = Nozzle C with 4 steel vanes to reduce bending of the jetting rod during insertionTable 3.5: Establishment of the Optimum Jetting Rod and Nozzle Type16816910CD (%)20 30I I01234 5000000C DCCCpop)1’444I• .-.•.• _-. ._47-8-__c-cl)10-TESTING SOUNDINGS NOT SHOWN (REASONS):_____SBPO1—SBPO7: PM tests up to 6.4 m SBP1OSBPO8: PM tests influenced by sand SBP1 1entrapment inside lantern 7 SBP1 2SBP14—SBP15:Shower Head System SBP13SBP1 6SBP17CENTRAL JETTING SYSTEMHIGH11-I,12-13-14-N•LOW15-16-17-MEDIUMFigure 3.9: Comparison of Coefficient of Disturbance for Different Testing Soundings170penetration resistance presented a trough. This was most probably caused by the presence of a loose sandlayer at that depth. Thus, it may be speculated that the CD variability with depth is caused by the adoptionof a single combination of jetting variables along a randomly variable stratigraphic profile. The looser thelayer being selfbored, the higher will be the disturbance. The review presented in section 3.4 indicates thatit is difficult to specify a unique “optimum” combination of all jetting variables because of density, strengthand general variability of the soil conditions ofmost natural deposits with depth.This subsection emphasized the importance of the numerical assessment of disturbance to properlyevaluate the best combination of the jetting variables in the optimization process of the insertion procedure.The CD parameter, together with all the sounding measurements of the UBC SBPM system, allows a betterunderstanding of the disturbance process and the probable influence of some of the key variables in thiscontext.The next subsection will briefly focus on the results obtained with another jetting system, which wastried out in the test soundings SBP14 and 15.3.6.3 Jetting SystemAn alternative design for the jetting system was devised by Campanella et a!, 1990 in sands andJ. Benoit (stated in Findlay, 1991) in clays. In both cases a “shower head” (SH) system was designed. Thissystem jets the fluid through radial orifices at the inner wall of the cutting shoe. The center jetting rod isremoved and the jets are directed upwards to the center of the shoe. It has the attractive feature of allowingjetting in deposits with particles of slightly higher diameter than those that are usually selfbored by thetraditional central jetting system (CJ). Also, the SH system is very effective at eliminating blockage. Onthe other hand the position of the jets can not be easily adjusted, and the SBPM requires a newlymanufictured cutting head each time a new jet location is tried.Since its initial conception by Campanella et a!, 1990, no conclusive opinion regarding the usefulnessof this system was developed. Therefore, two test soundings (SBPI4 and 15) were carried out with the newshower head system. In both soundings similar field problems were noticed. as summarized below:1. During the initial stages of the seilboring process there was a large amount of sand flushed out ofthe borehole with the returning mud. This amount was much higher than the “typical” amount observed bythe writer in the nearby SBP13 (CJ system) borehole or any other testing sounding with the CJ system. It171was also noticed that the SH was totally clean after withdrawal of the probe from the ground. Bothobservations may indicate that overcoring occurred with this system.2. Continuous pumping at a very low rate ( 1.9 11mm or 0.5 gpm) during pressuremeter expansiontests was started from SBP 12 to reduce plugging of the orifices of the jetting nozzle. With the shower headit was difficult to keep the mud level at the surface constant (the mud level shall be kept greater than thewater level to help prevent borehole collapse), whereas with the CJ system the above rate proved sufficientto keep the mud level constant. With the SH a high average pumping rate of 3.5 times the previous citedvalue was required. This also suggests that the mud was being lost in front of the cutting shoe (due toovercoring), and therefore it was not returning back to the surface.The consequence of overcoring is the opening of a large void space in the zone surrounding thepressuremeter shaft. This space will eventually be filled by loose sand once the selfboring process isterminated and the soil around the probe starts to settle. Therefore, underestimation of the soil parametersmay occur with the interpretation of the pressuremeter curves using this jetting system.In conclusion, more research is still needed with this new and promising jetting system. The presentdesign and procedures (mud flow and rate of advance, as specified before) cause overconng in front of theshoe during selfboring, which is accompanied by a general “loosening” of the soil surrounding the SBPM.For the current research SH development was abandoned after SBP 15 because of the need for redesign andtechnician support, and emphasis was placed in the development of the insertion procedure with the CJsystem. It is believed, however, that with a modified design and a SBPM insertion with less flow and at afaster advance rate it would be possible to overcome the overcoring problem. For instance, in claysFindlay, 1991 demonstrated that the shower head can be as reliable as the traditional central jetting system.3.6.4 Dimensional Differences Along ShaftThe use of the Camkometer lantern (no. 10) introduced a possible source of disturbance generation.Since this lantern was not originally designed for the UBC SBPM unit, an adaptation of the clampingsystem had to be devised. This adaptation led to the design of oversized floating rings (lantern retainers) atthe ends, to hold the lantern. The objective of the present subsection is to discuss this modification and itslikely influence on the pressuremeter testing curve.172For this study four testing soundings were performed with the oversized lantern retainer. They wereSBP 16, 17, 18 and 19. SBP 16 and 17 were carried out with an oversized lantern retainer that had a largediameter difference with respect to the diameter of the lantern (1.5 mm) and a lesser difference to thediameter of the cutting shoe (0.5 mm). In SBP 18 and 19 a new lantern retainer was tried, in order toreduce the dimensional differences in relation to the remaining parts of the pressuremeter unit. In this casea tapered retainer, with diameter varying from 74 mm at the shoe connection (same diameter as the shoe) to74.5 mm at the lantern connection was devised. A lesser diameter difference of 0.5 mm with respect to thediameter of the lantern was obtained at the top of the retainer, because two rubber membranes were usedbeneath the Camkometer lantern. Figure 3.10(a) presents the configuration of the base of thepressuremeter unit in both conditions (SBP16, 17 and SBP18, 19).The dimensional tolerances along the shaft of the pressuremeter can be of extreme importance in theprocess of soil disturbance. Fahey and Randolph, 1984 carried out SBPM tests in a granular deposit inwhich oversized and undersized cutting shoes were used. In the oversized case stress relief occurred at thecavity wall, and a certain amount of strain was required to bring the cavity pressure to the original groundstress. In the case of the undersized shoe there was a combined effect of increase in lateral stress andfriction (during insertion) in the cavity wall. This led to final residual cavity stresses that were again belowthe original ground stresses. Such disturbance mechanisms hampered not only the derivation of 0h by thelift off visual technique, but also the derivation of all other soil variables using the traditional interpretationmethods.Similarly, the dimensional differences of SBP16 to 19 imposed disturbance in the surrounding soilduring SBPM insertion. When the oversized lantern retainer of SBP16 and 17 was used a small loadingand subsequent unloading of the soil in the vicinity of the pressuremeter shaft took place. This was causedby the initial outward soil displacement of 0.67 % of the diameter of the shoe, and the final inwarddisplacement of 2.0 % ofthis same diameter.In Figure 3.10(b) the idealized change of lateral stress coefficient (defined in terms of effective stressratio K) for various soil elements close to the shaft is schematically shown. Initially the soil is in anundisturbed state, represented by a coefficient of lateral stress equal to I(. With optimum SBPM insertion173OVEP7ZED LANTERN RETAINER(SBP 16— 17)73srøz74.5--74 -Dimension(mm)SoilElementTAPERED OVERSIZED LANTERN RETAINER(Double Rubber Membrane)(SBP 18— i)Dimensionm) Element=Camkometer Lantern RLantern Retainer S=Cutting Shoe(A) DIMENSION OP OVERSiZED LANTERN RETAINERmm 0.5 mmIn Si:osi0ve—OutwarosDspiacement(8) IDEALIZED CHANCE OF LATERAL STRESS COEFF. KIN THE VICINITY OF THE PRO8EFigure 3.10: Characteristics and Effects Caused by the Oversized Lantern Retainer RingsI74.5SoilK=ah/avANeqa’Jve—InwardsDisplacement(at rest)174conditions, this is the stress state of the soil element 1. With outwards soil movement there is an increaseof ah. Since the vertical stress does not change, there is a residual increase in the coefficient of lateralstress. As the pressuremeter moves down there is a high inwards soil movement close to the lantern surface.This is accompanied by a large decrease of0h and a large decrease in the coefficient of lateral stress K. Atthis stage the stress path of soil element 3 moves towards the active failure condition.In the case of SBP 18 and 19 a similar loading and unloading mechanism is imposed in the surroundingsoil. In this case, however, there is a gradual outwards displacement of the material up to 0.67 % of theshoe diameter, followed by a final inward soil displacement of similar magnitude. In Figure 3.10(b) theidealized variation of K in this condition is shown. It is hypothesized that soil elements 1’ and 2’ behavedin a somewhat similar fashion to 1 and 2. At the onset of unloading the stress path of element 3’ moves inthe direction of the active failure condition, and the lateral stress coefficient decreases to a value close to,but not equal to, the initial I( value. Although this simplified representation is instructive it is furthercomplicated by the friction that is developed along the pressuremeter shaft.The extent of the above disturbance effects surrounding the pressuremeter will depend on thesensitivity of the material to the loading-unloading generated. This sensitivity will be a function of theinitial density and confining characteristics at each depth. In Figure 3.11 the effects of dimensionaldifferences are shown for a typical set of tests. In this figure, the calculation of CD coefficient for SBP17and 19 were very different. The CD of 6 % suggests almost no disturbance for SBP 19 with tapered lanternretainers, while SBP17 had a CD of 40 % with unacceptable disturbance due to large diameter changes.This same conclusion can be seen for all depths in the profile of Figure 3.9 where CD vs. depth is shownfor SBP 17 and 19. Thus, the idealized concept shown in Figure 3.10 is validated for SBP 17, whereas inthe case of SBP19 the disturbance at the soil-probe interface was considerably low to affect the overallshape of the testing curve.This subsection presented the likely disturbance effects caused by the variation of the tolerances in thedimensions of the pressuremeter unit. Enlarged lantern retainers are required for clamping purposes withthe Camkometer lantern (no. 10). In the UBC SBPM the taper design and the small increase in diameter ofonly 0.7 % has essentially no effect on disturbance, as high quality testing curves (CD’s below 10 %)0DU)U)a-)a0ca-)I,cLiEffects of Dimensional Differences Along Shaft175SBP17, Depth 6.9 m, CD = 40 %SBP19, Depth 6.3 m, CD = 6 %——--‘10008002000——.4IiIjICircumferential Strain (%)Figure 3.11: Effects of Dimensional Differences Along Shaft176were obtained in SBPI8 and 19. Additionally there is an important advantage from the oversized lanternretainer since the lateral stress is slightly reduced, thus decreasing the shaft ffiction during insertion whichreduces the required pushing force (hence penetration resistance) and possible soil disturbance.3.6.5 Recommended UBC SBPM Insertion ProcedureThe insertion procedure developed for the UBC SBPM takes into consideration all the possiblevariables that could cause disturbance on the testing curve. Optimization of the procedure was obtainedduring several field trials, in which the disturbance caused by the key variables was assessed and graduallyminimized. The preceding subsections discussed the likely effects of such variables in the generation ofdisturbance. This subsection presents the recommended insertion procedure for the UBC SBPM.Based on what has been learned throughout all the test soundings, the following combination ofequipment characteristics and insertion variables are recommended for the UBC SBPM:1. Equiyment Characteristics:• Camkometer lantern (lantern 10) with 2 inner (Gooch or Alliance) rnbber membranes of 1 mmthickness each.• Central jetting system with nozzle C.• Tapered (0.5 mm) oversized lantern retainer rings.2. Insertion Variables:• Average mud flow rate of 20 1/mm (5-6 gpm).• Average penetration rate of 25 cm/mm in sandy silts and 50 cm/mm in sands.• Setting rod at -5 mm (centerline of orifices 18.4 mm behind the edge of cutting shoe).3. Ipportant Details:• The surficial grass shall be removed prior to the “dummy” cone insertion. This avoidsplugging ofthe orifices of the nozzle with grass inside the borehole.• Preboring of surface soil or pushing of a large diameter “dummy” cone and casing up to 3 m isrecommended to reduce the likelthood of shoe plugging.• Slow pump of mud during the test at 1.9 11mm (0.5 gpm) to avoid plugging of the jetting nozzle.a At the end of each pressuremeter test the mud shall be pumped at a rate of 29 11mm, for 30 s.This allows all the sand particles in suspension inside the borehole to be flushed out.177• 1 full expansion pressuremeter test at each depth (1 m mm. depth interval) is recommended.• Electronics/air cable shall be taped to the BW rod in a “spiral” manner, each 50 cm. A 3Mcloth duct tape (48 mm wide) is recommended.Insertion carried out in a consistent manner will allow a considerable improvement in the quality of thepressuremeter curves derived from the UBC SBPM system, when used in granular sites of the Fraser Delta.3.7 TESTiNG PROCEDUREThe testing procedure describes the manner in which the pressuremeter pressure expansion is carriedout in the field to allow the subsequent interpretation analysis of all the desired soil parameters. It wasshown in the last chapter that the fitting technique requires a testing curve expanded to a circwriferentialstrain of 10 %, with at least one unload reload loop. The recommended procedure will follow this basicrequirement. It is important to determine, however, the influence of the variation of the rate of inflation onthe derived testing curve (and pore pressure development), in order to obtain a practical value for theexpansion rate of the UBC SBPM. The practical hold time prior to the unload reload loops is alsodiscussed in this section.3.7.1 Rate of InflationAll the pressuremeter tests carried out for this thesis were done in stress controlled conditions, sincethis is the manner in which the UBC SBPM system is currently designed to operate.In order to investigate the influence of different expansion rates in the pressuremeter data a testingprogramme with 3 profiles of FDPM tests 1 m apart at the site was planned, as presented in Table 3.2 forFDPO4, 05 and 06. The rates of inflation adopted in each testing profile respectively varied from 0.5 to7.0 kPa/s, which encompass most of the rates previously adopted for the other soundings. Pressuremetertesting curves were compared at the same depth level from 4.3 to 6.3 m, in the clean sand layer. Resultsfor deeper depths were not compared as FDPM refusal occurred around 8 m in all testing profiles.The repeatability of the soil stratigraphic conditions at the 3 FDPM soundings was checked bycomparing the penetration resistances measured at the load cell of the pressuremeter. Similar to the conebearing resistance Q of conventional cones, this penetration resistance reflects the initial density andconfining conditions of the soil. For FDPO4, 05 and 06 an average difference of 2.5 kN (5 % of full scaleresistance) was noticed, thus indicating similar initial soil conditions at each FDPM profile.178The testing results for the 4.3 m depth, presented in Figures 3.12 and 3.13, are similar of the obtainedat the other depths. The following comments can be made:1. The excess pore pressures are essentially null for all the rates employed. The desired drained soilbehavior during expansion was therefore obtained for the rates adopted.2. Good agreement is noted between the shapes of the pressuremeter curves under distinct rates ofinflation.It seems, therefore, that for the rates adopted similar stress-strain curves were obtained. Thisconclusion agrees with the experimental observations of Jackson et al, 1980, that noticed in uniaxialcompression tests in sands that the loading rate has a relatively minor effect on the stress-strain responsefor loading times greater than 1 millisecond.Since a compromise between a low rate to always allow a fully drained expansion and a fast rate for aquick test is desired, an intermediate expansion rate is recommended for the UBC SBPM. As the besttesting results of SBP18 and 19 were obtained with a rate of inflation of 3.4 kPa/s, the recommended ratewill be in the range of 3 to 4 kPaJs.3.7.2 Holding Time Prior to Unload Reload LoopsCreep will influence the unload reload loops of the test. It was observed by Hughes, 1982 andHowie, 1991 that unloading in a stress controlled manner without dissipation of the creep strains will leadto unload reload loops with an initial rounded shape. A similar effect was observed by Whittle et al, 1992for ioops in clays. This occurs because at the start of the unload a high creep rate exists. The higher theinitial creep rate and amount ofpressure unload, the larger will be the initial roundness of the loop.Creep strains developed during this process will accumulate with the strains caused by the elasticshearing of the surrounding material, causing an error in the derived Gur modulus. Murthy, 1992 observedthat creep tends to decrease the unload reload modulus Gur. By performing several consecutive unloadreload loops he noticed that the soil modulus increased with the number of loops, and stabilized after aparticular amount of strain accumulation. On the other hand the creep rate decreased from loop to loop,179i----I Rata at irfiatian(kPa/a)a/;400= /20: II800U00 -zD// // /200-/ // /800U1/7—w/ I200 - 7//7”FDPO4,05,0SLaing Bridge2epth 4.3 mo2 4 8 10 12Figure 3.12: Influence of Rate of Inflation on Pore Pressure Development0C“—‘ 600C/)C/)40000u 200CUFDPO4—FDPO5—FDPO6—Lantern 7—Laing Bridge180800RATES OF INFLATION0.52.4——7.0Depth = 4.3 m(Clean Sand)kPa/skPa/skPa/s//rCircumferential StrainFigure 3.13: Influence of Rate of Inflation on Testing Curve181reaching a value below 0.01%/mm for the loop in which the Our stabilized. Howie, 1991 concludedthat pressure holding phases have to be adopted prior to unload reload loops if the modulus is to bemeasured in sands. He suggested that a final creep rate of 0.1 %/111j should happen in the end of thisholding phase.Based on the above considerations it is suggested that the holding phases have to be designed to yield amaximum creep rate of 0.01%/min before the start of the unloading stage. In this research an average8 mm. holding time was chosen prior to the loops, because it is fast enough for practical applications of thepressuremeter. This time has proven to yield creep rates in the range of 0.001 %/min prior to the loopstage.The influence of the above proposed rate on the measured Our modulus is demonstrated for one of thetest sounding of this research. It will be assumed here that creep takes place only during the unloadingstage of the loop, that the creep rate during this stage is constant and equal to the value at the end of theholding phase, and that creep strains can be simply calculated by the multiplication of the unloading timeversus this constant creep rate. Table 3.6 presents the basic information and creep strain estimation of theloops of tests in the sounding SBP17. These loops were carried out with a degree of pressure unloading(dP / P0) of 40 % at the recommended rate of 3.4 kPals. The “uncorrected” modulus in column 0 is thestandard value obtained after the compliance correction. The “corrected” modulus in column J is calculatedwith a “creep free” unloading strain (column I), based on the assumptions above. This very simplifiedexample illustrates that for final creep rates below the target value (see column D) the averageunderestimation of the measured Our will be in the range of 3 %. Using this same data it is simple todemonstrate that, if the final creep rates of column D were in the range of 0.1 %/n (recommended byHowie, 1991), the final underestimation of Our would be as high as 30 %.Based on the above findings a recommended testing procedure is proposed for the UBC SBPM.3.7.3 Recommended UBC SBPM Testing ProcedureThe recommended procedure follows the basic characteristics illustrated in Figure 3.14. The followingcomments apply:1. Inflation should be carried out in stress controlled manner with a rate between 3 to 4 kPalsthroughout the loading and unloading stages of the test.ABCDEFGHIJKDepthPcu8h1FinalCreep2UnloadTime3de94Gui5CreeStrain6d6c7Gur-cError9(m)(kPa)(%)Rate(%min)(mm)(%)(MPa)x10-(%/min)(%)(MPa)(%)4.92293.400.0060.6180.22327.513.7080.21928.142.25.92913.450.0130.7550.25731.599.8150.24733.204.86.92323.500.0090.6590.26426.565.9310.26326.680.47.92773.800.0070.7270.33723.005.0890.33123.492.08.92942.900.0050.8090.32426.794.0450,31927.301.89.94072.450.0060.9750.27742.615.8500.27143.812.711.93333.000.0100.9080.31731.629.0800.30732.893.812.74492.600.0100.9890.29143.279.8900.28145.244.3Average=0.008°/dmin1-Cavitystrainatthestartofthe8mm.holdingphase2-Creeprateattheendofholdingphaseandstartofunloading3-Timeelapsedfromupperunloadingpoint tolowestunloadingpointoftheloop4-Cavitystrainmobilizedduringunload-Includescreepstrain5-Definedas0.5dPu/dEO-containscreepstrainandiscompliancecorrected6-Circ.creepstrain=unloadtimeversusfinalcreeprate7-Correctedcavitystrainmobilizedduringunload-withoutcreepstrain,compliancecorrected8-Definedas0.5dP/deo-doesnotcontaincreepstrainandiscompliancecorrected9-Definedas(Gur-c-Gui)IGur-c]x100(%)SBP17-Rateofinflationanddeflation3.4kPalsLoopswithaveragedPufPcuof40%andholdingphasesof8rain.Average2.7±1.4%(±1std.dcv.)Table3.6:InfluenceofCreepStrainonUnloadReloadModulus183I- PcuC1G)-4-CwfCircumferential Strain (%)Loop CharacteristicsHolding Phase (b-c) of 8 to 10 mm.6eh 2 to 3 %Pcu variable in accordance with testing depthdPu/Pcu = 40 %Holding Phase ebcdPu10Figure 3.14: Proposed Testing Procedure for Selfboring Pressuremeter Tests1842. At a circumferential strain between 2 and 3 % the pressure should be held constant to allowdissipation of the creep strain rate to a value below 0.01 %/ A minimum holding time of 8 mm. isrecommended.3. An unload reload loop should be carried out. To ensure a well defined loop with enough data pointsthe degree of pressure unloading (dP I P) should be around 40 %.4. The probe should be finally expanded to a maximum circumferential strain of 10 %, followed by afinal unloading in order to allow the determination of the closing pressure.The above recommended procedure was used in the test soundings SBP18 and 19.3.8 SUMMARY AND CONCLUSIONSIn this chapter emphasis was placed on the field determination of the insertion procedure for the UI3CSBPM. Optimization of the key insertion variables led to the reduction of the disturbance imposed in thetesting curve. The identification and discussion of likely disturbance effects caused by each of the insertionvariables was presented. This helped to define the best insertion conditions that are recommended in theUBC SBPM insertion procedure. The aim of this procedure is to minimize disturbance during insertion.It is important to minimize or (if possible) eliminate all the possible sources of disturbance on thepressuremeter curve. Disturbance leads to a pronounced reduction of the initial roundness of the shape ofthe pressuremeter testing curve. Disturbance generated during seliboring is a complex combination of theinfluence of variables related to the jetting parameters adopted during insertion, to the equipment used, torandom field occurrences (such as shoe plugging), as well as to secondary unknown factors (vibration, nonverticality, etc.).Using the new interpretation methodology proposed in the last chapter a coefficient of disturbance CDwas developed. This allowed the numerical quantification of the disturbance present in the pressuremetercurves of this thesis, thus removing the subjectiveness that existed so far in the assessment of the quality ofsuch curves. Typical ranges of expected CD values, corresponding to different degrees of curve quality,were proposed. According to the experience of the writer, with CJYs lower than 10 % it is possible toobtain high quality SBPM curves in sands. The CD parameter has been shown to properly identify thedegradation of the quality of the pressuremeter curve caused partial shoe plugging, improper jetting rodpositions and dimensional tolerances.185The optimization of the insertion procedure of the UBC SBPM took into consideration theminimization of the CD parameter. A constant mud flow of 20 1/mm and penetration rate of 50 cm/mmwere adopted during these field trials. The main findings were:• Disturbance is not evenly generated along depth. The looser the layer the more likely it is to bedisturbed. It appears, therefore, that distinct combinations of jetting variables are necessary to selfbore anatural deposit. If a single combination is adopted, as in the present study, it may be expected that aresidual low disturbance will be imposed in some layers.• Partial plugging of the shoe invariably occurred in most of the SBPM tests of this research. Thiswas caused by the presence of a surficial silty sand to sandy silt layer in the Laing Bridge soil profile.Partial plugging imposes a complex stress, density and stress history in the soil surrounding the probe. Itleads to a large increase in both CD and pressuremeter penetration resistance. Plugging can be reduced oreven avoided by the insertion of a large dummy cone down to 3 m depth prior to selfboring, and by thedecrease in the rate of SBPM penetration in the fine and silty soils.• Tip loading, and partial plugging with the surficial fme-grained material, were also related to theposition of the jetting rod in relation to the cutting shoe. The optimum position of the rod was obtainedusing a trial and error procedure, in which the shoe plugging occurrence and the CD parameter were usedto assess the quality of the insertion. The optimum position was obtained in SBP18 when a CD lower than10 % was achieved. This took place with a simultaneous improvement of the design characteristics of theprobe. A continuous mud flow at a low rate of 1.9 1/mm (0.5 gpm) is required during the SBPM test inorder to prevent the plugging of the orifices of this nozzle.• A shower head jetting system proved to be unacceptable in its present design, since it leads tooverconng in front of the shoe. A redesign of this system should provide a very effective jetting alternative.• Inter-strip ffiction due to overlapping of the strips of the steel lanterns imposed constraints on theuse of these lanterns in the UBC SBPM. Only the Camkometer lantern is recommended for the UBCprobe, since this lantern gave the most consistent Gur values.• The details of the equipment design and tolerances have a significant impact on the final results.Disturbance effects are generated when SBPM tests are carried out with dimensional differences along theshaft (as is usually the case). The initial design adopted for the oversized lantern retainers led to a186substantial disturbance over the entire pressuremeter curve. With the modification of this initial design toreduce dimensional differences to less than 0.7 % (using tapered sections to gradually adjust dimensions)the disturbance effect was essentially removed from the testing curve when using the Camkometer lantern.With an understanding of the influence of the key insertion variables the recommended UBC SBPMinsertion procedure was proposed. This procedure was tried in testing soundings SBP18 and 19, leading topressuremeter testing curves with a disturbance coefficient CD below 10 % for most of the testing depths.Once insertion related factors are understood and the likely disturbance minimized, standardization ofthe test procedure is possible and recommended. Past experience with pressuremeter tests in sands indicatethat rate effects may be of importance for the derivation of the Gur modulus. Pressure holding phases of8 nun. were designed to reduce the creep rates to values below a target 0.01 %/n prior to the loop stage.A simplified analysis indicates that the Gur derived in this manner will be underestimated to an averageextent less than 3 %, at least for loops with a degree of pressure unloading of 40 %. Using FDPM testingresults from the Laing Bridge site it was noted that for expansion rates within 0.5 to 7 kPals a fully drainedtesting curve is obtained. No influence of the above variation of expansion rates on the stress-strain testingcurve was found. Based on these findings a recommended IJBC SBPM testing procedure was proposed.Standardization of the procedures adopted during insertion and testing of the SBPM in sands isessential for the derivation of repeatable results, The procedures suggested in this chapter will not onlystandardize the operation of this tool, but will also lead to high quality pressuremeter testing curves. Thesecurves will be used in the next chapter to demonstrate the methodology of interpretation of the basic soilparameters, using the sand at the Laing Bridge site.187CHAPTER 4.0 TEST RESULTS AT THE UBC SITE AT LAING BRIDGE4.1 INTRODUCTIONThis chapter demonstrates the application of the proposed interpretation methodology presented inChapter 2 on high quality field testing data. The tests were run in accordance with the recommendationsput forward in Chapter 3 concerning the insertion procedure for the UBC SBPM system. The field datawas gathered in an extensive field testing programme carried out by the writer at the Laing Bridge site.The detailed geological and geotechnical characteristics of this particular site are presented herein.The predicted soil parameters from the SBPM testing curves are compared to values obtained fromlaboratory as well as other in situ tests, allowing a brief discussion on the significance and reliability ofthese parameters.4.2 GENERAL CHARACTERISTICS OF THE LAING BRIDGE SITE4.2.1 GeologyThe Fraser River Delta is located at the western edge of the British Columbia Mainland, on the westcoast of Canada. This delta comprises a triangular area bounded on the west by the Strait of Georgia, onthe north by the north ann of the Fraser River, and on the south and southeast by the Cascade Mountains.The lowlands of this delta consist of flat-topped hills and plateaus which are separated by wide valleys(Clague et al, 1983). The Fraser Delta is occupied and subdivided by the Fraser River, as shown inFigure 4.1(a). The Fraser Delta serves as a model for a high energy, sand rich, estuarine system, as it hasbeen mainly built up by sand dominated river transport during spring and summer (Milliman, 1980). Giventhe complexities of the depositional process as shifting of the distributary channels, eustatic water levelfluctuations, etc., the sediments of this delta exhibit pronounced lateral and vertical variations in texture.The deposits of this delta are of the Quatemary age, with thickness that can vary up to 300 m.According to Blunden, 1975, the development of the Fraser Delta started 11000 years ago. This agecorresponds to the end of the last glaciation period of this area, when the ice retreat started to occur. Theweight of ice depressed the land to an extent so that its level became lower than the existing sea level. Asubmarine delta was then formed and ideal conditions for the accumulation of sediments were generated.188(B)4Stile,o 20 40.Figure 4.1: (a) Laing Bridge Site Location, (b) Test Area in this Site (Modified after Sully, 1991)189Underlying the Delta there is Tertiary bedrock overlain by Pleistocene deposits, which consist of aseries of thin seams of glacial till and glacial outwash from previous glacial ages. About 11000 years agolarge amounts of meitwater were funneled into the valley of the Fraser River, subjecting the area todeposition of fine sediments discharged into the sea by the Fraser River. As a result, the Fraser Riverfloodplain rapidly prograded westward. While these events were occurring the level of the sea continued tofall relative to the land (Clague et al, 1982). This period of low sea levels during the early Holocene stagewas followed by a marine transgression caused by an 11 m rise in the sea level. This relatively rapid sealevel rise continued until 5000 to 5500 years ago. During this period coarser sediments were depositedover the finer grained clays and silts already submerged in the submarine delta. The deposition of thesegranular sediments took place under a dynamically high and variable energy environment. Westwardprogradation of the Fraser Delta continued at a slow rate after 5000 years ago, due to low fluctuations ofthe sea level. During this period organic sedimentation commenced over nearly the entire eastern portion ofthe delta.The Fraser Delta continues to prograde westwards, although in a pattern of sedimentation which isslower than those from the past. The present rate of increase of the Fraser Delta varies from 2.5 to 8.5 mper year, depending on the depth of water and tide effects. The maximum tidal range is almost 5 m at themouth of the river, close to the Laing Bridge site, decreasing both landward and with increasing river flow(Ages and Woollard, 1976). The present geological profile of the Fraser Delta is shown in Figure 4.2.Since the Fraser Delta deposits were formed only after the last glaciation they have not beenmechanically overconsolidated by ice load. Thus, based on geological evidence they are normallyconsolidated. On the other hand post glacial events may have taken place. Extensive rework by channelmigration (Monahan et a!, 1992), and seismic liquefuction (Clague et al, 1992) have undoubtedly affectedthe original characteristics of these deposits.4.2.2 Location and FeaturesThe Laing Bridge site is located in an area of Sea Island close to Grant MeConachie Way, the ArthurLaing Bridge and the Vancouver International Airport. As shown in Figure 4.1(a) Sea Island lies betweenthe north and middle arms of the Fraser River...l2_’f A’Cou?K#. —190z.7?P _O1 ME%b. -‘0s Vk74L 74i% 6AfA7.’. ‘OKfZO.7AL 4if V Mjg.4 0Figure 4.2: Geological Cross Section of the Fraser Delta (after Blunden, 1975)191The test area in Laing Bridge site is presented in Figure 4.1(b). The topography is fairly flat with aslope of about 1% north eastwards. The site is grassed and is surrounded by a series of ditches and swales.The average groundwater level is about 0.5 to 2.5 m below ground surface, varying in accordance with thetidal regime of the nearby Fraser River. The site is bounded on the north by the South Approachembankment, constmcted in 1970 to connect the airport to downtown Vancouver (Bertok, 1987). It isbounded on the south by a major drainage ditch. The field test programme was performed in a triangulararea constrained by the major ditch and an existing gravel road (see Figure 4.1(b)). Along the side of thisarea the South Approach embankment has a variable height of 0 to 4 m.Fill was placed on the site during the construction of the South Approach embankment and theMcConachie overpass. This fill has a thickness of 1 to 1.5 m and overlays the original Fraser Deltasediments at this location.4.2.3 StratigraphyBased on the results of the in situ testing programme carried out by the writer, the general sitecharacteristics summarized by Bertok, 1987, and a field survey of the site, the geotechnical andtopographical proffle of section 1-1 from Figure 4.1(b) can be established. It is presented in Figure 4.3.The stratigraphic profile consists of 1.0 to 1.5 m of sandy fill, underlain by a sandy silt layer 1.5 to2.0 m thick. Below this layer there is a fine sand stratum with thickness varying from 15 to 20 m. The finesand is underlain by a clayey silt to silty clay layer that extends down to the Pleistocene till deposits.A better insight into the stratigraphic variations along the profile can be obtained with the testingresults of a logging tool like the piezocone (CPTU). Figure 4.4 presents a typical CPTU profile at the site,obtained from piezocone sounding (C06) with the pore pressure sensor located behind the tip. The close tozero differential pore pressures throughout the initial 20 meters of the profile is indicative of a drainedpenetration condition, typical of sands. This can be also seen by the close agreement of penetration porepressure (measured behind the tip, U2)with the hydrostatic pore pressure (U0)ofthe groundwater.The negative pore pressure values in the surficial sandy silt layer indicates a slightly overconsolidationeffect, as demonstrated by Robertson and Campanella, 1989 at other sites on the Fraser Delta. Theoverconsolidation of this layer was probably caused by desiccation. The clayey silt to silty clay below192Section 1-1Figure 4.3: Seetion 1-1 of Laing Bridge SiteHLu Lf) Li]: m H — —C — W—LiLi--- LLJ 0LL m —a-. i-i-LjLi_. ,5i-5-j z <. o71 —----fl c z I H m wzI z<Y (/)(( Li— DEzlU) ?5 0 I u193aHCL :J2CL I)CDo -H- .ECD0U—)inE___LiL j.¼(‘S5)LLiLiLiE0LiIr) Cr)(I) 71HCOCO19420 m of depth is normally consolidated based on consolidation tests (Bertok, 1987 and Le Clair, 1988) andinterpretation of piezocone data. The almost constant values of friction ratio for depths of 5 to 15 mindicates a fairly uniform and clean sand. Between approximately 15 m to 20 m a transition layer of sandinterbedded with fine lenses of silty material exists. This is readily seen by the high variation of the frictionratios and dynamic pore pressures measured by the CPTU. As noted in the last chapter the UBC SBPMtests were carried out in the uniform sand between 5 and 15 m ofthe profile.However, the sand of this deposit has a notably high variation in density, as can be seen by thevariability of the cone bearing resistances Q in Figure 4.4. This is caused by the variation of theenvironmental and energy associated conditions during the sand deposition, as noted before. Based on theBaldi et al, 1982 empirical relationship for relative density (Dr) determinations, valid for a mediumcompressible unaged quartz sand, this sand deposit can be depicted as having loose (40 %) to dense (60 %)characteristics within 5 to 15 m of depth. This density generally increases with depth.4.2.4 Maximum Shear ModulusThe results of downhole seismic piezocone (SCPT) tests were used to measure the (low strain)maximum shear modulus profile of this sand deposit. The seismic modulus of the SCPT (G0) is measuredat a strain amplitude of approximately i04 % and below (Campanella et al, 1986). Figure 4.5 presents theresults of tests SCO 1 and SCO2, performed 1.3 m apart in the field. The downhole shear wave velocity wascalculated using the cross correlation digital signal processing analysis presented in Campanella andStewart, 1990. This wave was generated by a swinging hammer hit on each side of the pads of the UBCresearch vehicle. G0 was derived with the use of the measured shear wave velocities, the density of the sand(from laboratory tests of Bertok, 1987) and the following equation (White, 1965):G0 =pV2 (4.1)where p is the soil density and V is the shear wave velocity.As expected, the stiffiess obtained by both SCPT testing profiles show a trend of variation similar tothe piezocone bearing results of Figure 4.4. As noted by Houlsby and Hitchman, 1988, Schnaid andHoulsby, 1991 and others, the cone bearing reflects the density and mean stress level of the sand prior tothe CPT insertion. Laboratory testing results published in the literature demonstrate that the shear waveE0a-pa ieIn15-- -_ i _-)HEd1 :JVH i’jL L../(rs / )0 40 80 120 168 200 240iaxisiurn Shear Modulu5(MPa)60 20 40 60 80 100 120 14019516Ij rF1 I F jI IiiLding BriqE SitEFigure 4.5: Profiles of Shear Wave Velocity and Maximum Shear Modulus196velocity in sands is affected by the same variables. G0 is very dependent on the void ratio of the sand(Hardin and Drnevich, 1972), on the stress levels that respectively act in the directions of shear wavepropagation and particle movement (Roesler, 1979) as well as on the ratio of such stress components (Yuand Richart, 1984). Hence, G0 reflects the maximum stiffness of the sand under the in situ mean effectivestress ((a + Oh)!2)and density regimes.The average of both G0 evaluations of Figure 4.5 was used in the interpretation of the pressuremeterdata of the Laing Bridge site.4.2.5 Laboratory Testing ResultsA laboratory testing programme was carried out with all the “undisturbed” and disturbed samplesretrieved from the site. It was directed towards the determination of the basic characteristics of the sand,including its classification. It also had the objective of evaluating the typical shearing behavior of the sand,and of obtaining the peak friction angle j at distinct depths.4.2.5.1 Soil StrengthThe peak friction angles and the stress-strain behavior of the sand were experimentally obtained withisotropically consolidated drained triaxial (Cm) tests, using reconstituted and undisturbed samples fromthe ST1 sampler. A total of 7 CID undisturbed and 9 reconstituted triaxial tests were performed underdistinct densities and stress level conditions, as detailed in Appendix C. A value of 340 and 0.25 wasrespectively obtained for 4 (constant volume friction angle) and v (Poisson’s ratio) from the triaxial tests.These values were used in the interpretation of the pressuremeter data ofthe Laing Bridge site.4.2.5.2 Soil ClassificationSoil classification, grain size analysis and Atterberg Limits were obtained with the disturbed samplesretrieved inside the split spoon sampler (ASTM D1586 and D1587). This sampler was employedsimultaneously with the standard penetration test (SPTO1) performed at each 1.5 m of the Laing Bridgeprofile. A total of 16 sieve analyses and 2 Atterberg limit tests were performed.The sand retrieved by the ST1 sampler also allowed minimum and maximum void ratiodeterminations, as well as the performance of two specific gravity tests. A visual petrographic analysis andone X Ray Powder Diffractometer test were kindly performed by Dr. Lee Grout of the UBC GeologicalSciences department.197The sieve analyses indicate that the granular deposit at Laing Bridge is a uniform sand with a smallamount of silt content, classified as SP to SP-SM using the Unified Classification of the soils. The X rayresults show that the main mineral of this granular deposit is quartz (67.3 %), followed by feldspar (14 %),anhydrite (14.7 %), chlorite (3.5 %) and miscellaneous (0.5 %). The quartz grains can be classified assubangular to subrounded in shape, with a sphericity of 0.77 in accordance with the chart ofRittenhouse, 1943. The gradation of the material is extremely uniform. The sand between 5 to 15 m isbasically composed of medium (average of 68 ± 8 % passing sieve #30) to fine (average of 17 ± 7 %passing sieve #60) grain particles, with an average D10 of 0.14 ± 0.02mm and D50 of 0.31 ± 0.05 mm. Thesurficial sandy silt contains a high amount of silt, varying from 60 % at 2 m deep to 2.6 % at 4,7 m. Theaverage fines content between 5 to 15 m in depth is below 5 %. Laboratory tests of Bertok, 1987 indicateda water content in this same sand varying between 20 to 30 %, with an average unit weight of 19.6 kN/m3.Minimum and maximum void ratio determinations of the granular samples retrieved from 5 to 15 m ofthe profile were carried out in accordance to the ASTM D2049-69. The minimum and maximum voidratios were respectively 0.51 and 0.84. The average calculated specific gravity of the sand was 2.67 inaccordance with the testing procedures suggested by Lambe, 1951. This value was used in conjunctionwith the minimum and maximum void ratios to define the initial and after consolidation relative densities ofall the samples tested in the triaxial cell.4.3 INTERPRETATION OF UBC SBPM DATA4.3.1 Testing CurvesThe UBC SBPM measures the pressure versus displacement (or strain, with the use of Equation 2.2)response at six distinct directions along the horizontal plane. These directions are given by 0, 60, 120, 180,240 and 300 degrees, respectively related to arms 1 to 6 around the center of the UBC probe. Arm 1 (at 0degrees) was in general directed towards the magnetic north at the Laing Bridge site.Therefore, with the UBC SBPM it is possible to obtain an array of pressure expansion (testing) curvesat the same depth. A typical example is presented in Figure 4.6 for the depth 8.3 m of SBP19. Using thisdata it is possible to obtain a set of sand parameters (4, oh, Gi) related to each of the distinct expansiondirections of the probe. Moreover, using the displacement measured in each of the strain arms at given0-(1)Cl)00CC19810006004002000wCircumferentia’ Strain (%)Figure 4.6: Typical Pressure Expansion Curves Measured by the UBC SBPM at Laing Bridge Site199levels of internal pressure it is possible to track the expansion of the cavity wall. Figure 4.7 presents thedisplacements measured in each of the arms during the testing expansion of Figure 4.6. For this particularexample the UBC SBPM tended to expand to a slightly lesser extent in the direction of the strain arms 1, 2and 3, than in the opposite direction (arms 4, 5 and 6). Perhaps the noted displacement differences arerelated to the influence of each of the following variables:1. Soil anisotropy: Soil anisotropy (fabric and lateral stress) along the horizontal plane is reflected bya differential behavior of the strain aims, hence the measured pressure expansion curve.2. Differential disturbance: It is unlikely that the (low) disturbance generated during the “optimum”insertion of the UBC probe will be evenly distributed along the horizontal direction. It is possible thatstress relief (or increase) is differentially imposed around the cavity. Note, for instance, the variability oflift off stresses measured by the testing curves of Figure 4.6.3. Translation of the center of the SBPM: If the soil is isotropic, and undisturbed, then the nonconcentric circles of Figure 4.7 represent the (displacement) response of a SBPM test with a constantlymoving axial center.It appears, therefore, that the average pressure expansion curve, rather than each individual curve, ismore appropriate in the curve fitting interpretation analysis. The averaging of the output of the strain armsof the SBPM helps in the compensation of errors due to the differential disturbance imposed in thehorizontal plane during insertion. It also dilutes the anisotropic soil response along the horizontal testingplane and assumes the soil is isotropic, which looks like a good assumption for this soil since the measureddisplacements give a cylindrical response in Figure 4.7. The greater the number of strain arms, the betterwill be the averaged response of the surrounding soil.Therefore, only the average testing curve at each depth was used to estimate the sand parameters ofthe Laing Bridge site. This approach for SBPM test data interpretation is commonly done in practice. Incommercial probes, like the one owed by Cambridge In Situ (Canikometer probe), the average output of the(three) strain measuring sensors is used in the interpretation process. The Menard pressuremeter adopts thevolume change of the expanding cavity to infer the average radial displacement of the cavity wall.The average testing curves adopted herein were obtained with the field testing programme carried outby the writer, as described in the last chapter. The best quality curves from both test soundings SBP 18 and200/ArE 3EN’. tnt. PEe5ure (kP)—D---NDuth 83 n (SBP19)Arrr. 3A EIII UOBSERVAT I ENS:i—Circles of displacenent are the best fit curvesthrough the S data points2-Radial displacenent divisions at each I noIArm 4Figure 4.7: Typical Displacements Measured during the Expansion Stage of the UBC SBPM201SBP19 were selected for this purpose, since the “optimum” insertion procedure was adopted for thesesoundings.4.3.2 Curve Fitting InterpretationThe most representative curves for this site are those for depths 5.3 m (SBP 18), 6.3 m, 8.3 m, 10.3 m,and 13.3 m (SBP 19), for which the CD was below 10 %. In each of the pressuremeter tests an unloadreload loop was carried out. This furnished the new cavity expansion model with the shear modulus Gur,required to define the elastic strains in the plastic zone surrounding the cavity.The methodology of interpretation of these curves followed recommendations put forward inChapter 2, with the use of the new cavity expansion model. Plots showing the comparison between themeasured experimental data and the curves fitted manually are presented in Figures 4.8 to 4.10, for each ofthe SBPM tests chosen. In general the curve fits are reasonably good, giving confidence in the derivedbasic soil parameters. Discrepancies found between the curves are due to a small amount of disturbancegenerated during insertion of the U]3C SBPM, as noted in Chapter 3 for SBP18 and 19.The final set of curve fitting parameters is shown in Table 4.1, together with the triaxial peak frictionangles, the seismic low strain G0 values and the unload reload mocluli of the pressuremeter tests. Thedisturbance coefficients defined after match are also shown. As expected they are low, but not null.In order to comment on the consistency of the derived set of parameters for the granular deposit ofLaing Bridge site, a comparison was made between the predicted parameters and the reference values. Asimilar approach was followed before in Chapter 2 with the chamber data, in order to validate theconsistency of both the new and the Carter et al, 1986 cavity expansion models. The comparison betweenpredicted (SBPM test interpretation) and reference (lab., etc.) geotechnical values is discussed next.4.3.2.1 Friction AngleIn order to have a basis of comparison for the predicted friction angles, the results of the laboratorytriaxial tests were used. The comparison of pressuremeter and triaxial friction angles is widely used inpractice (Robertson and Hughes, 1986, Bellotti et al, 1987, East et al, 1988, Manassero, 1989 andNewman et al, 1991), since triaxial tests are faster and easier to perform than other laboratory devices.Besides, most current engineering designs still rely on strength parameters derived from the triaxial test.202800-New Cavity Exponsion Model00600-ci):3(I)(1)400-00Cci)c 200-LUField Curve (A)idealized Model CueDepth 53 m (SBP18)I I I I2 4 6 8Circumferential Strain1(%) 122 4Circumferential Strain (%)Figure 4.8: Final Curve Matching with Field Data: (a) Depth 5.3 m, (b) Depth 6.3 m00a):5U)asG)0Ca)CbJ0ci):3a,(I)ci)00Cci)CuJNew Cavity Expansion ModelFigure 4.9: Final Curve Matching with Field Data: (a) Depth 8.3 m, (b) Depth 10.3 m2032 4Circumferential Strain (%)2041600140000- 12001000:3(I)(I)800O 600Cci)I.C 4009-w 20008Circumferential Strain (%)Figure 4.10: Final Curve Matching with Field Data: Depth 13.3 m205DEPTH 2 Gur3 GurC G0 (j)TRX(1)PSS CURVE FflTING RESULTS‘ (kPa) (MPa) (MPa) (MPa) (Deg) (Deg)(m)4) Gi K CD(Deg) (kPa) (MPa) (ahIcYV) (‘)5.3 63 16.9 11.1 40.2 38.7 41.5 43.5 37 11.0 0.6 7.36.3 73 30.6 20.7 42.6 39.1 41.6 48.5 70 14.4 0.9 6.08.3 93 37.9 26.2 54.4 40.5k 43.7’ 48.5 85 18.3 0.9 4.110.3 113 NA NA 72.9 38.2* 40.3* 46 86 22.2 0.7 .2.!_13.3 143 47.2 34.8 81.9 39.1 41.6 48.5 140 27.6 0.9 77Observation: * Value interpolated from experimental data, NA Not available.340 and v = 0.25 during curve fitting1-Distance from ground surface to the centre of the expandable section of the pressuremeter2-Effective vertical stress considering the (hydrostatic) ground water level during the day of the field test3-From the unload reload loops performed during the expansion stage: Degree of pressure unload =40 %4-Corrected for stress level via Bellotti et al, 1989 equations with soil parameters from the fitting analysis5-Low strain modulus from downhole seismic cone tests (SCOI and SCO2)6-Results of the triaxial testing programme with “undisturbed” samples at similar testing depths7-Converted plane strain friction angles derived using the Lade and Lee, 1976 empirical correlationTable 4.1: Curve Fitting Results in Laing Bridge Site206The triaxial peak friction angles adopted here were obtained with the “undisturbed” samples of theST1 sampler (details in Appendix C). The drained friction angles of the triaxial test (4) were furtherconverted into “plane strain” (4ic) angles by using the empirical relationship proposed by Lade andLee, 1976. This relationship is largely used when comparing friction angles of triaxial and pressuremeterdevices (used by all the authors cited above), and provides an initial estimate of the plane strain 4’ in theabsence of other tests.The comparison between plane strain friction angles predicted by the pressuremeter and axiallysymmetric triaxial values is presented on the top plot of Figure 4.11. It is noted that the pressuremeterangles are, on average, 8° above the peak triaxial values. These differences are accounted by two factors:1. The (pressuremeter) plane strain friction angles shall be indeed higher than the (triaxial) axiallysymmetric values:The lengthy review presented by Lee, 1970 and Ladd et a!, 1977 on this subject indicate that the planestrain friction angle can be as high as 6 to 8° degrees greater than the depending on the initial porosityof the sample. The greatest difference is associated with dense sands at low confining pressures (whichappears to be the present case), and the smaller differences are associated with either loose sands at allconfining pressures or dense sands at sufficiently high confining pressures to prevent dilation to occur.2. The stress paths imposed in the virgin sand by both triaxial compression and plane strainpressuremeter tests are different. This topic is discussed in the next comparison.The bottom plot of Figure 4.11 illustrates the comparison between predicted friction angles and“converted” plane strain values (4”) from the triaxial test. It can be noticed that the predictedpressuremeter angles are still higher than the “reference” values, although the overestimation is slightlyabove 10 %. Several combined reasons can be used to explain the differences, as follows:1. Simplifications built into the new cavity expansion model:Although it “captures” the essential behavior of medium dense to dense sands during shear, it stillidealizes the medium with a linearly elastic perfectly plastic representation.2. Differences in the modes of deformation (and stress paths) imposed by triaxial and pressuremetertesting devices:a)60-C -‘—‘20--Da) -I.C-)10C0I.(1)a)CC030-0cn0a)0-oa)0Figure 4.11: Comparison of Curve Fitting Results: Friction Angles40PS) Triaxial q TRX (Deg)207+ 10 % /‘—101AxiallyI I I20 30 40 60Symmetric Triaxial cP.JRX (Deg)60-208Wood and Wroth, 1977 observed that the mode of shear deformation imposed by pressuremeter testscorresponds to the deformation which could be imposed in a plane strain triaxial compression test with thesample in the horizontal position (the plane where a acts in the field would be the laboratory a2 plane).This differential mode of deformation imposes differences in the mechanical properties measured duringshear, as the mobilized stress ratio and the secant deformation modulus. This is due to the fact that naturalgranular deposits are composed of anisotropically assembled particles (cross anisotropic fabric), asdemonstrated by Oda, 1972.3. Failure mechanism of pressuremeter tests in sands:Fahey, 1986 noticed in X Ray radiographs of the chamber tests with Leighton Buzzard sand that noevidence of the development of shear bands could be seen, even for cavity strains greater than 30 %. Hespeculated that the pressuremeter is an almost perfect “pure shear” plane strain test, and the predictedparameters are indeed higher than those experimentally measured in testing devices that allow thedevelopment of planes of rupture (or progressive failure).The points discussed above suggest that the predicted ffiction angles of the pressuremeter test areindeed consistent with the geotechnical characteristics of the deposit (density, confining stress, mineralogy,etc.). The consistency of the results is directly related to the quality of the testing curves, as well thecapacity of the model to “capture” the essential shearing behavior of this sand. Indeed, the Laing Bridgesand exhibited a highly dilatant behavior when sheared in the triaxial cell (see Appendix C), whichconforms with the basic premise of the new cavity expansion model.Friction angles predicted by the pressuremeter are higher than those obtained by other testing devices,since the pressuremeter angles refer to a special shearing mechanism (under plane strain conditions) thatcan be solely simulated in the laboratory by cubic or plane strain triaxial tests.4.3.2.2 Coefficient of Earth Pressure at RestTable 4.1 presents the interpreted values of the coefficient of earth pressure at rest (K assumed as K0)for the granular deposit of Laing Bridge site. A range of K varying from 0.6 to 0.9 was obtained, reflectingthe natural variability of the sand strata with depth (see for instance the profile of Q expressed inFigure 4.4). According to Schmertmann, 1985 in only very few cases in nature are we likely to encounter209sands laid down in a manner as to result in an uniform lateral stress (or any other basic soil parameter)throughout the vertical profile.However, the predicted values of K from the proposed methodology are higher than the “expected” K0values of a normally consolidated uncemented (NC) sand. This can be better visualized with the K resultsexpressed in Table 4.2, calculated after using well established empirical relationships and the axiallysymmetric friction angles obtained by the triaxial tests. It can be noticed in this table that the predictedSBPM K’s are higher than the normally consolidated K0 values calculated with Jáky, 1944 equation.Assuming that K0 is uniquely related to overconsolidation ratio (OCR) and soil strength during onedimensional unloading, the equation proposed by Schmidt, 1966 can be also used to predictoverconsolidated K0 values. This exercise suggests that the predicted SBPM K’s for Laing Bridge site areequivalent to those obtained in mechanically overconsolidated granular samples subjected to OCR’s in therange of 2 to 4. Based on geological evidence, the deposits of the Fraser Delta did not suffer from any kindof mechanical preloading. Hence, other site related phenomena must have happened to yield the highlateral stresses predicted by the SBPM test interpretation. The high values of K0 in this site can beassociated to one or the combined effect of the following factors:1. Depositionnl characteristics of the Fraser Delta sediments:According to Monahan et al, 1993 the sands in this delta have been deposited under a high energyenvironment. Migration of the main channel of the Fraser River also took place extensively during thepast. This migration has been recorded, at least, over a 100 year period of time (Millinian, 1980, Clague etal, 1983), leading to reworking of the entire subaerial topset sands of the Fraser Delta (Monalian etal, 1992). The erosion and filling processes associated with this channel migration, allied with the highenergy of deposition of the granular particles, could have imposed an over consolidated characteristic tothis sand. It could also “lock in” high lateral stresses in the granular deposit.2. Increase of K0 with time:According to Sully, 1991 the IC in normally consolidated soils increases with time due to postdepositional history. As the soil is deposited the K values approach the NC K0 profile given by theempirical equation of Jáky, 1944. As the soil becomes progressively buried the lateral strain condition is210Equations (Jsed: Jáky, 1944 Schmidt, 1966Determination of K0:1-By Jáky, 1944 empirical equation for normally consolidated soil:K0= 1-sin2-By Schmidt, 1966 empirical equation for overconsolidated soil:K0(OC) = (1-sin ).OCR° where co = smDEPTH PREDICTED COEFFICIENTS OF EARTH PRESSURE K0(m) (Deg)OCR=1 OCR=2 OCR=3 OCR=44 42.3 0.32 0.52 0.68 0.835 38.7 0.37 0.58 0.74 0.896 39.1 0.36 0.57 0.74 0.897 43.5 0.31 0.50 0.66 0.819 37.6 0.38 0.60 0.76 0.9111 38.8 0.37 0.58 0.74 0.8914 39.1 0.36 0.57 0.74 0.89Table 4.2: Assessment of Coefficient of Earth Pressure by Empirical Formulae211modified and stress redistribution as a result of anisotropic hardening occurs, so that I( increases above theNC I( profile. The results presented by Graham and Jefferies, 1986 from several different hydraulic fillsconcur with the idea of K increasing with time, The analysis carried out by these authors on high qualitySBPM data indicates K values that are typically two to three times higher than the K0 predictions of Jáky’sequation. Since the fills were placed hydraulically, without compaction and preloading, they are, bydefinition, “normally consolidated”. These authors concluded that the horizontal in situ stresses in granulardeposits can be “much greater than most ofus would have expected”.3. Influence of past earthquakes in the Fraser Delta:Clague et al, 1992 presented conclusive evidence for seismic liquefaction and cyclic shearing of thegranular sediments of five sites close to the Vancouver metropolitan area. According to Robertson, 1982this past seismic activity was responsible for the generation of high “locked in” horizontal stresses in thesand of a site close to Laing Bridge, leading to K0 values that were almost double the NC I( valuespredicted by the Jáky, 1944 equation. Indeed cyclic shearing affects the lateral stress of the sand. Youdand Craven, 1975 studied the variation of the lateral stress coefficient of dry Ottawa sand when subjectedto a cyclic shear loading history in the simple shear apparatus. They concluded that during repeated shearstraining the coefficient of lateral stress increases with both shear strain amplitude and the number ofcycles. Therefore, prehistoric earthquakes in the Fraser Delta could have imposed a lateral stress regime inthe sand above the expected NC K profile of Jáky’s equation.Minor factors that could preload the sand in this site would be the stockpiling of earth material for theconstruction of the South Approach embankment, and the ground water level fluctuations (as much as2.5 m (Bertok, 1987)) due to the tidal influence of the nearby Fraser River.Conclusively it can be said that the measurement of lateral stresses in sands is a difficult task, and noreliable in situ methodology exists so ffir for that purpose. The interpretation of SBPM high quality testingresults based on the methodology advocated in this thesis constitute an initial step in this direction. It mustbe emphasized, however, that more experience is still required to validate the applicability of the predictedlateral stresses. The high accuracy obtained by the proposed methodology in relation to the prediction ofthe lateral stresses of the chamber tests (in Chapter 2) suggests that the values of I( established for theLaing Bridge site are consistent. It is the writer’s opinion that K0’s at the Laing Bridge site are indeed high,212caused by several of the combined effects of the possible overconsolidation and lateral stress increasemechanisms speculated above.4.3.2.3 Shear ModulusThe predicted initial shear moduli from the SBPM (Ui’s) were compared with other in situ shearmoduli. The unload reload modulus Our, its stress level corrected value Gur° and the seismic low strainmodulus G0 were adopted as a basis of comparison to Gi. As commented in Chapter 2 the SBPM Girepresents an index for the sand stiffness in the “elastic” zone surrounding the probe. Go was determinedwith seismic cone tests, as discussed before, whereas the pressuremeter Our was obtained in accordance tothe testing procedure recommended in Chapter 3. The stress level correction proposed by Bellotti etal, 1989 was adopted to convert the Our modulus to Gur”. This latter modulus is related to the originalaverage normal effective stress of the sand.The comparison between Ui and the pressuremeter unload reload moduli is presented in Figure 4.12.The top plot of this same figure shows the comparison between Gi and Gur, whereas the bottom plot showsthe comparison between Ui and Gur. For the particular characteristics of the unload reload loops carriedout in this thesis (degree of cavity pressure unload, cavity strain amplitude and there seems to exist areasonable constant ratio between predicted Ui’s, measured Our’s and calculated Gu? ‘s. In the case ofGur, the comparison of Figure 4.12 indicates that an average ratio of Our/Ui in the range of 2 was obtainedwhen considering all the testing depths.The much higher value of Gur in relation to Gi comes from the fact that the deformation parameter isinfluenced by the stress level of the test. As commented before, the initial modulus Gi is related to theaverage normal effective stress am of the elastic zone, hence the in situ 0h of the deposit. The unload reloadmodulus Our represents the stiffliess of the sand close to the cavity wall, hence it is related to the averagenormal effective stress a that existed in the plastic zone prior to the loop stage (see discussion inChapter 2). Table 4.3 presents the estimate of both levels of stress for each of the testing depths. It isnoticed in this table that the average stress level that existed in the plastic zone prior to the loop isapproximately double the average stress level in the elastic zone. Given the stress level dependency of themeasured modulus in sands, where higher moduli are measured for higher stress levels, it is expected thatthe measured Our’s will be considerably higher than the Gi’s.70-60-00‘‘40-C(f)30--DC) 20-0ci)10-213+ 10 % ////7,’ —‘7,’ ——‘— Gur = 2 GiI I I I10 20 30 40Unload Reload6 70Loop Our (MPa)Unload Reload Loop GurC (MPa)Figure 412: Comparison of Curve Fitting Results: Shear Modulus214TEST DEPTH UNLOAD RELOAD LOOP CURVE FITflNG(m) (Gur. Gur°) (Gi)_________Yav’ Oav2 Gm4ahx 10’ (%) (kPa) x 10 ( %) (kPa)SBP18 5.3 1.9 88.2 2.3 37SBP19 6.3 1.7 152.5 3.6 70SBP19 8.3 1.5 177.5 3.5 85SBP19 10.3 NA NA -- --SBP19 13.3 1.7 257.4 3.8 140Average: 1.7x10’% l69kPa 3.3x10’% 83kPa1-Average shear strain amplitude imposed by the loop cycle within the surrounding sand(Bellotti et al, 1989):Yav = 0.5.f3.A’ywhere:0.5 is a factor to produce a single amplitude of strain, as adopted by Seed and Idriss, 1970I is a empirical reduction factor based on the chamber data of Bellotti et al, 1987Ay is the shear strain amplitude of the loop cycle in the cavity wall 2 de2-Average normal effective stress within the plastic zone that existed prior to the loop stage(Bellotti et al, 1989):0av = Oh + cx(P - Oh)where:cz is a empirical reduction factor based on the chamber data of Bellotti et al, 1987P is the effective internal pressure at the start of the unload reload loop3-Calculated by Equation 2.334-Average nonnal effective stress in the horizontal plane = 0h predicted by the curve fittinganalysisTable 4.3: Strain and Stress Levels Related to Gur and Gi215The comparison between Gurc and Gi in Figure 4.12 indicates an average ratio of Gur7Gi in the rangeof 1.3 when all the testing depths are considered. Since both GurC and Gi are related to the same value ofaverage nonnal effective stress they should be, in principle, the same. The discrepancy between calculatedGur’ ‘s and predicted Gi’s may be given by the differences in the amplitude of shear strain that is related toeach of these moduli. Gur is related to the average shear strain amplitude Yav imposed by the loop cycle inthe surrounding sand, whereas Gi is assumed to be related to the limit shear strain amplitude ‘yf mobilized inthe elasto-plastic boundary. Using the average shear strain amplitude related to Gur’ and Gi in Table 4.3,it is possible to show in Figure 4.13 the expected values for the stiffness ratios Gurd/Go and G1JG0. Thesevalues are respectively 0.26 and 0.17. The ratio 0.26/0 17, or 1.5, represents the expected value ofGur°/Gi, assuming that the average attenuation curve of Idriss, 1990 is valid for this particular sand. Thisratio is close to the average experimental value of 1.3 presented in Figure 4.12, providing convincingevidence that the differences between Gu? and Gi are indeed due to strain level differences.The comparison between Gi and the seismic modulus G0 is presented in Figure 4.14. As noted before,the shear modulus detenriined from in situ downhole shear wave velocity measurements represents thestiffness of the sand at shear strain amplitudes in the range of 10’ %. According to the average attenuationcurve of Idriss, 1990 a ratio of GiJG0 equal to 0.17 was expected. The experimental results of Figure 4.14indicate a higher average ratio than the “expected” ratio based on Idriss’s curve. This may be partiallyrelated to the simplified manner at which some relevant strain amplitude is related to these moduli. Anotherreason may be the universality of Idriss’ proposed curve. Idriss’s curve may not be fully applicable tosands of different gradation or mineralogy than those used to generate his curve.The comparisons discussed above suggest that the predicted Gi of the pressuremeter test is areasonable representation of the response of the sand when sheared in the horizontal direction with a strainamplitude in the order of 10’ %.4.3.2.4 Modulus Reduction CurveAs commented in Chapter 2 using the information obtained by the fitting technique, together with thelow strain modulus (Ge) of the sand, it is possible to predict the shear stress-strain monotonic “elastic”curve of the tested sand. For that purpose, the hyperbolic model of Kondner, 1963 was adopted,c 0.4 -00.3 -0ci)0-CFigure 4.13: Determination of Modulus Reduction Ratio for Different Amplitudes of Shear Strain216‘0 1.0ci 0.8 -0 -ci)0-) -I.o 06-C-:0.5 =Modulus Reduction Curve(Idriss, 1 990).7x1 Q_1%)/GQ-0.0 - —0.0001I I I 11111 I I 11111 I I0.001 0.01 , 0.1Shear Strain (%)712179080CO_7Ø0600ffi(f)4Ø-Dci)C)-DcD20010030Seismic Cone C0 (MPa)Figure 4.14: Comparison between the Predicted Gi from Curve Fitting and the Seismic Modulus G0218allowing the development of a final hyperbolic equation (Equation 2.83) written in terms of the (fitted)basic parameters of the sand.This subsection briefly demonstrates the use of this equation, including how to predict the secant shearmodulus of the sand at a variety of shear strain levels. For that purpose it was adopted the pressuremetercurve with the lowest coefficient of disturbance. This curve is the one presented before for the depth 8.3 m,where a coefficient of disturbance of 4.1 % was obtained.Using Equation 2.83 with the curve fitting results (4), ah and Gi) of Table 4.1 for this depth it ispossible to draw the idealized monotonic stress strain response of the tested sand, from very small strains(1O %) to large strain levels (> 1 %). This curve is presented in Figure 4.15(a), where it can be noticed:1. The initial secant modulus of the curve, equivalent to the response at very low strain levels, leads toa shear modulus G equal to the G0 obtained at this same depth with the seismic cone results. This wasindeed expected, as the value of G0 was used to infer the final value of Gi predicted by the fitting technique(using Equation 2.83), in accordance to the interpretation approach suggested in Chapter 2.2. The strain level assigned to Gi in this figure is the value calculated in Table 4.3 with the fittingparameters and the Equation 2.33. This strain level represents the value above which failure starts to occurin the elasto perfectly plastic soil (y). Since the hyperbolic model considers that the soil fails only atinfinite strain (where t = t), it is possible to obtain the soil response for strain levels above y.Using the curve established in Figure 4.15(a) or Equation 2.83 it is possible to calculate the stiffnessratio of the sand (GIG0)for each level of induced shear strain (y), hence determine the complete curve thatdescribes the variation of the secant shear modulus with the strain level. Figure 4.15(b) presents thestiffness ratio attenuation curve for the tested sand. This curve ideally represents the pure shearstress-strain behavior of the sand at 8.3 m depth from very low to large strains, when sheared under aaverage normal stress level equal to a.In Figure 4.15(b) is also shown the singular value of stiffness ratio Gurd/Go as obtained with the datagathered by the unload reload loop performed during the testing stage. The agreement is good, besides ofthe semi-empirical approaches adopted to estimate the corrected (for stress level) Gur modulus and therelevant strain amplitude ‘lay. This suggests that the idealized modulus reduction curve may indeedresemble the soil response from very small to large strain levels.219Depth 8.3 m (SBP19)10090800S.—, 7060U) 50(1)ci)40o 30U)-c(1) 201001 .00.90.80.70.60.30.20.10.00.0001 0.001 0.01 . 0.1Shear Strain (%)Figure 4.15: (a) Hyperbolic Stress Strain Curve for Depth 8.3 m (SBP 19)(b) Secant Shear Modulus Ratio versus Shear StrainG/G0 (HyperboIc Equation)Gurc/Co (Unload Reload Loop)G0 = 544 MPa (Seismic Cone)220In conclusion, the procedure above constitutes a simple method to define the stress strain response inthe elastic zone. For that the proposed procedure uses fundamental (coupled) soil parameters derived froman elasto-plastic theory into a hyperbolic model. This model is assumed to represent the soil responsebeyond the elasto-plastic boundary. It is emphasized that the above procedure constitutes a rationalmanner (with closed form solutions) to extend the elastic shear modulus of the pressuremeter to a variety ofstrain levels, thus allowing this modulus to be applied to any engineering problem where the design loadinduces a variable pattern of strains in the soil.On the other hand, it is important to recognize that the deformation behavior of the soil is considerablyaffected by factors such as the consolidation history, the stress path and the stress system during shear, asdemonstrated by Vaid, 1985. Vaid showed that the hyperbolic approximation of Kondner, 1963, definedwith hyperbolic constants from the results of conventional triaxial tests, was not able to properly describethe undrained stress strain behavior of Haney clay under anisotropic consolidation history or other stresspaths during shear. Thus, separate hyperbolic representations for each particular testing condition had tobe defined. Extending such findings to the present case in sand, it is also concluded that the proposedprocedure must be used with care in cases where the boundary conditions of the analyzed engineering workdiffers considerably from those of the pressuremeter test.4.4 CONCLUSIONSA methodology to interpret pressuremeter testing curves in sand was presented in Chapter 2 and usedhere to derive the basic soil parameters of a typical granular deposit in the Fraser Delta, at the VancouverInternational Airport.The large amount of data obtained with the field trials carried out by the writer, together with thedevelopment of a new disturbance criteria, allowed the establishment of “optimum” conditions for insertionwith the UBC SBPM. The high quality testing curves adopted in this chapter are those derived under such“optimum” condition, in which disturbance was categorized and minimized.The measured friction angles by the pressuremeter test analysis were higher than the values obtainedby consolidated drained triaxial tests, most probably due to the particular shearing conditions imposed bythe pressuremeter. The SBPM plane strain 4. values of the tested sand were about 8° higher than thetriaxial values and varied from 43.5 to 48.5°, reflecting the variable density and confining conditions of the221different layers of sand in the granular strata. The interpreted lateral stresses predicted by the SBPM wereconsiderably higher than initially expected, leading to K’s that varied from 0.6 to 0.9 as a result of thecombined influence of several post depositional factors that took place at this site. The initial shearmodulus Gi varied from 11 to 27.6 MPa, for the same reasons observed above for 4). Nevertheless, a goodcorrespondence was observed between this modulus and the unload reload moduli (Gur, Gure), as well asthe low strain modulus (G0) from downhole shear wave velocity measurements, when stress level and strainamplitude effects are considered. It appears, therefore, that the pressuremeter modulus Gi obtained fromthe proposed curve fitting technique reflects the stress-strain response of the sand when sheared in thehorizontal plane under a shear strain level in the order of 101 %, and average normal stress equal to 0h.In summary, when the proposed interpretation methodology is used on high quality testing curvesobtained with the insertion and testing procedures advocated in Chapter 3, it is possible to derive aconsistent set of basic soil parameters. This set is coupled by the framework of the theoretical model andrepresents the strength, stiffness and lateral stress of the studied sand. The predicted strength and stiffnessare, however, related to the particular shearing mechanisms imposed by the pressuremeter in thesurrounding soil. The set of predicted parameters may be further used in a hyperbolic model to extend theapplicability of the pressuremeter modulus to a variety of strain levels. Additional validation of this latterapproach may be required before it is used in design.222CHAPTER 5.0 SuMMARY AND CONCLUSIONS5.1 SUMMARYThe emphasis of this thesis was placed on the establishment of a methodology of interpretation ofselfboring pressuremeter testing results in sands. Emphasis was also placed on the standardization of bothinsertion and testing procedures adopted for the UBC selfboring pressuremeter, when used in granulardeposits of the Fraser Delta.Recognizing the potential value that selfboring pressuremeter testing has to predict reliable soilparameters in soils that are difficult to sample, this thesis had the following objectives:1. Review the existing cavity expansion models developed to interpret the selfboring pressuremeterdata, considering the idealized assumptions made for the models to simulate the complex process of thecavity expansion in sands.2. Review the current interpretation methodologies applicable for predicting the basic soil parametersGh and Cl) from SBPM testing data in sands, discussing their possible limitations.3. Develop a new interpretation methodology to derive the basic soil parameters of the sand, witheither undisturbed or disturbed SBPM data. The new approach should lead to repeatable and reliableconclusions in the evaluation ofthe sand behavior in situ.4. Evaluate the reliability of some of the existing cavity expansion models, when applied together withthe new interpretation methodology to analyze SBPM testing data in sands.5. Develop an improved cavity expansion model to be used together with the new interpretationmethodology for the prediction of soil parameters from SBPM tests in sands.6. Detennine the “optimum” insertion procedure of the UI3C SBPM for its use in granular deposits ofthe Fraser Delta. This procedure shall minimize, to a large extent, the influence of disturbance on thetesting data.7. Develop a coefficient of disturbance to serve as an index to quantify the relative quality of theinsertion procedure.The main contribution with respect to the interpretation of SBPM testing curves was the developmentof a new methodology of analysis. This new methodology relies on a curve fitting technique to match the223idealized model curve to experimental testing curves, and simultaneously predict the basic parameters ofthe sand. It was used together with a newly developed cavity expansion model. The new cavity expansionmodel is based on the basic principles of cavity expansion in elasto-plastic frictional materials, and extendsthe rheological equations of Hughes et a!, 1977 to incorporate elastic strains in the expanding “plastic”zone. This model additionally differentiates between the required shear moduli of both “elastic” and“plastic” zones surrounding the probe. Initial validation of this model, as well as the new interpretationtechnique, was carried out with published SI3PM results from calibration chamber tests. These chambertests were carried out with different probes, installation procedures (“ideal” or selfbored), and differentsands at differing conditions of density, stress history and confining pressure. The remarkably goodagreement of the predictions (within 10 % of the baseline values) gives high confidence in the usage of thenew interpretation methodology in SBPM testing curves in medium dense to dense sands. The proposedinterpretation methodology can also be coupled to a hyperbolic model to establish the idealized shearstress-shear strain response of the tested sand. This model is useful to extend the applicability of thepredicted modulus Gi to strain levels that are more relevant for the design of civil engineering works.It has been determined that the quality of the soil parameters predicted by the proposed interpretationmethodology is also dependent on fhctors other than the initial conditions of the sand. The quality of thetesting curve (or disturbance built into this curve), the inherent sensitivity of the new cavity expansionmodel to changes in the input parameters, and the strain range of curve match are factors that have to beconsidered in the analysis of field SBPM curves. These factors were explored and discussed in detail inChapter 2, leading to general guidelines for the interpretation of SBPM testing data in sands.Disturbance is the major variable that reduces the quality of the testing curve, and hence, thereliability of the predicted sand parameters. In sands disturbance can be reduced to a large extent if anoptimization (trial and error) routine is carried out in the field with the insertion and equipment variables.This thesis also attempted to identify the most important variables that generate soil disturbance. This wasexplored in detail in Chapter 3. Using the new interpretation methodology a new approach to numericallyquantify the disturbance of the testing curve was created. This new disturbance criteria measures thedisturbance of the curve by a “coefficient of disturbance” (CD). The CD has been found to properly reflectthe degradation of the quality of the testing curve by partial plugging of the cutting shoe, improper jetting224rod positions and dimensional tolerances of the equipment. Using the CD parameter, numerical ranges for“undisturbed” (or close to), “disturbed” and “highly disturbed” testing curves were proposed. With theseranges it was possible to optimize the insertion procedure of the UBC SBPM in a particular granulardeposit of the Fraser Delta. Although the conclusions were developed for the Laing Bridge site they alsoapply to other sites of similar sand stratigraphy.This thesis also attempted to recommend a testing procedure for SBPM’s in sands. This proceduretook into consideration the findings of Chapter 2 (expansion up to 10 %, with at least 1 unload reload loopstage), the results of FDPM tests at different rates of inflation, and the assessment of the creep influence onloops with 8 mi holding phases.Using the recommended insertion procedure it was possible to obtain high quality SBPM testingcurves in the test site of the Fraser Delta. These field curves were used to further validate the proposedinterpretation methodology. The predicted sand parameters were compared to soil parameters fromlaboratory and other in situ tests, allowing a discussion of the significance of the pressuremeter predictions.The comparisons highlighted the fact that the predicted sand strength and stiffness are related to theparticular shearing mechanisms imposed by the SBPM in the surrounding medium.Simplicity, accuracy and reliability are the essential features of the proposed methodologies of thisthesis. It is believed that the infonnation contained herein will aid pressuremeter practitioners to designsafer and more economical civil engineering works.5.2 METHODOLOGY OF INSERTIONThe methodology of insertion of the SBPM is related to the procedures adopted in the field tominimize the extent of disturbance on the testing curves.This thesis identified and discussed the major insertion and equipment variables that have to beconsidered during the field optimization of the SBPM insertion. The recommended UBC SBPM insertionprocedure is presented in Chapter 3, and is based on the field experience gathered throughout the 23soundings performed at the Laing Bridge site. The recommended procedure considerably enhanced thequality of the SBPM testing curves, and serves to standardize the operation ofthis complex device.225The main findings with respect to the selfboring process are:I. Insertion of SBPM’s with a jetting system is viable, provided that a trial and error routine isadopted in the field to establish the optimum combination ofjetting variables.2. Disturbance can be expected to be differentially generated along the selfbored profile if a constantcombination of insertion variables is adopted. This is due to the differential sensitivity of the distinct sandlayers to the vibration, etc. that take place during the insertion of the SBPM. Loose layers are more proneto disturbance than dense layers. There seems to be a close relationship between the density of the layer (orthe pressuremeter penetration resistance) and the CD of the testing curve.3. It is possible to reduce the amount of disturbance in the testing curve. This requires the adoption ofa field optimization routine in which the CD parameter is used as a “benchmark” to guide the quality ofinsertion. The CD shall be mininaized in the field insertion routine to values below 10 %.The main findings with respect to the equipment adopted during the selfboring process are:1. The current design of the UBC “shower head” jetting system requires a redesign since it leads toovercoring of the sand during SBPM insertion. This system may also prove useful if insertion is carriedout at a faster rate with less mud flow.2. Steel lanterns composed of curved strips and overlapped together should not be used to shield theSBPM, unless the strips have the same curvature as the SBPM shaft. Interstrip friction generated by thelateral stress and the accumulation of granular particles are common occurrences with the use of suchlanterns. For the UBC SBPM only the Canakometer lantern (reported in Fahey et al, 1988), which hasbutted strips (no overlaps) and eliminates interstrip problems, is recommended.3. The dimensional tolerances of the equipment used to assemble the SBPM have a significant impacton the final testing results. Disturbance is generated in the surrounding sand during the selfboring processwhen the SBPM has dimensional differences along the shaft (as is usually the case). The disturbancegenerated by the UBC SBPM system can be essentially removed if tapered lantern retainers, designed toreduce dimensional differences to less than 0.7 % the diameter, are adopted to hold the Camkometerlantern. Besides, for the UBC design two 1 mm thick rubber membranes are also required beneath theCanakometer lantern.2265.3 METHODOLOGY OF TESTINGOnce the insertion related variables are understood and optimized it is possible to standardize theexpansion (testing) routine carried out in the field.The recommended UBC SBPM testing procedure was proposed in Chapter 3, based on the results ofseveral pressuremeter tests at the Laing Bridge site and the findings of Chapter 2. This procedure allowsthe establishment of the basic soil parameters with the proposed interpretation methodology, as well itprovides the new cavity expansion model with reliable input values of Our. The main findings are:1. The SBPM should be expanded to a maximum circumferential strain of 10 %, which is enough toyield the parameters of the sand via curve fitting technique. At least one unload reload loop stage should becarried out, after a (constant pressure) holding phase.2. Rate of inflation in between 0.5 to 7.0 kPals lead to similar stress-strain testing curves in sands.They also lead to fully drained testing curves. Therefore the inflation should be canied out in stresscontrolled manner with a rate of inflation between 3 to 4 kPals. This range represents a compromisebetween a low rate, required to allow fully drained expansion, and a fast rate, required for a quick test.Computer controlled SBPM tests are suggested in order to ensure the repeatability and the constant rate ofexpansion of the test.3. Pressure holding phases of (at least) 8 mm. are effective to reduce the final creep rate to a target0.001 %/min prior to the unload reload loop stage. A simplified analysis indicated that the Our derived inthis manner will be underestimated to an average extent of less than 3 %. This is valid for loops with adegree of pressure unload of 40 %.5.4 METHODOLOGY OF INTERPRETATIONThe methodology of interpretation is related to the procedures adopted to predict the basic soilparameters of the sand with the use of the SBPM testing curve.The main findings in this area are:1. The traditional methodologies currently available to interpret SBPM testing curves in sands do notlead to reasonable predictions of the basic sand parameters in disturbed SBPM test data. They can only beapplied to high quality testing curves.2272. The proposed interpretation methodology, together with the new cavity expansion model, has thecapability to simulate reasonably well the pressuremeter testing curve in medium dense to dense sands.Indeed, provided that the sand has a dilatant (shear induced volume increase) behavior during shear, thenew approach leads to reasonably accurate predictions of the sand parameters in both undisturbed (or closeto) and disturbed data.3. The proposed interpretation methodology leads to a set of “coupled” parameters that are linked toeach other by the framework of the chosen cavity expansion model. The set of parameters, rather than eachindividual value, can be used in a more reliable way to simulate the soil response in further design analyses.4. Matching between the field and the idealized model curve can be either visually accomplished, orwith the use of a “curve fitter” software program. The available software programs (Sigmaplot orKaieidagraph’) are useful for high quality testing curves, as they lead to similar results as those obtainedby the visual match of the curves. On the other hand, these software programs can not be adopted fordisturbed curves, as some experience related input may be required. This is because, with disturbedcurves, the “strain range” of curve match has a fundamental effect on the reliability of the final predictedparameters. For these curves (10 % < CD < 30 %) the curve match should be accomplished betweencavity strains of 5 to 10 % (to be visually defined), since disturbance will considerably affect theparameters obtained with the match in the initial stages (0 to 5 %) of the testing curve. For undisturbedor slightly disturbed curves (CD < 10 %) the curve match can be accomplished between cavity strains of 0to 10 %. Beyond this strain level any of the discussed models of Chapter 2 can not be adopted, as thesand surrounding the cavity starts to shear under critical state conditions and the cavity expansion deviatesfrom the idealized cylindrical fonn.5. End effects do not appear to influence the field testing results, provided that the cavity expansion iscarried out to low strains (below 10 %), and the pressuremeter has a slenderness ratio (LID) equal to (orgreater than) six.6. Both the proposed (new) and the Carter et al, 1986 models lead to similar results if the adoptedPoisson’s coefficient is in the range of 0.2. The Hughes et al, 1977 model tends to underestimate thestrength and the stiffness of the sand, due to the lack of incorporation of elastic strains in the expanding“plastic” zone.2287. The friction angle is the least sensitive (and most reliable) parameter obtained from the curve fittinganalysis. This is due to the fact that the curve match is dominated by this variable when the new (or theCarter et al, 1986) model is used. The errors in the assessment of Gi and ah are of similar magnitude.These errors are low (or null) if high quality testing curves are adopted, and the constraint imposed byEquation 2.81 (between the sand parameters) is adopted during the fitting process. Both the predictedstrength and stiffliess are intrinsically related to the particular mode of deformation imposed by the SBPMin the surrounding sand, and may be expected to differ from the parameters predicted in the same sand withother in situ or laboratory testing devices.8. Both the Poisson’s coefficient and the constant volume friction angle can be estimated in the curvefitting process. The final error of the predicted basic parameters of the sand is small (below 10 %), when“educated guesses” of v and are used.5.5 FINAL REMARKSThis thesis has shown that the parameters predicted by SBPM tests in sands are affected by manyfactors other than the shearing behavior of the soil. These factors are related to the quality of insertion, thetesting procedure and the interpretation approach. Standardization of the insertion stage of this probeconstitutes an initial step to reduce the differences due to different field procedures. This standardizationhas been applied to the operations of a particular system (as the UBC SBPM), in order to minimize thelikelihood of disturbance in the testing curve.The new interpretation methodology provides a framework to predict consistent values of the basicsand parameters. It also furnishes the pressuremeter practitioner with a technique to numerically quantifythe disturbance of the testing curve. This may prove useful in the optimization routine carried out in thefield with the insertion variables. At present, the proposed methodology constitutes the most fruitfulapproach to analyze SBPM results in sands, although more research still needs to be devoted in this area.5.6 SUGGESTIONS FOR FUTURE RESEARCHTwo complementary areas of research were identified with the SBPM test. One is specific to the UBCSBPM system, whereas the other can be applied to any SBPM. The first area would be the improvementof the current design of the UBC SBPM. The second area would focus on the further validation of the229proposed interpretation methodology, and its extension to sands that shear in a different manner than thesand studied here (like very loose to loose sands, cemented sands, etc.).5.6.1 Equipment DevelopmentImprovement of the pumping unit, in order to make this unit operate with an independent source ofhydraulic power. A high capacity pump capable of providing mud as high as 50 to 70 1/mm would berecommended. This is essential for jetting in clays.• Construction of a simulated laboratory test chamber with synthetic material, to calibrate the SBPM.• Improvement of the data acquisition system, such that strain controlled holding phases andexpansion tests would be possible.• Improvement of the effective (pore) pressure transducer, such that it can be connected to the outerlantern rather than the inner rubber membrane.• Improvement of the shower head jetting system. The redesign of this head and new field trials at afhster rate with less mud flow is suggested. This system may prove useful in deposits of coarse sands,where blockage ofthe cutting shoe may occur with the use of the central jetting system.5.6.2 Interpretation of SBPM Data• Validation of the proposed interpretation methodology in field tests where the soil baselines are wellestablished. Friction angles measured in either the cuboidal or the plane strain triaxial apparatus could beused as a reference for the predicted SBPM friction angles. Shear moduli from resonant column tests withundisturbed sand samples (consolidated to the lateral stresses predicted by the SBPM for the site) wouldserve as a basis of reference to the predicted SBPM shear moduli.• Verification ofthe capacity ofthe predicted SBPM soil parameters to simulate the behavior of actualengineering works. Field tests in areas where trial embankments will be constructed and monitored couldbe performed.• Development of a new cavity expansion model for the interpretation of SBPM testing results in sandsthat: (a) present a pronounced strain softening behavior during shear; (b) contract during shear.• Use of the SBPM data to study dilatancy of sands and apply to liquefuction analysis and triggermechanisms.• Use of cyclic SBPM tests to simulate and measure undrained residual strength of sands.230BIBLIOGRAPHYABBREVIATIONS:CGC = Canadian Geotechnical ConferenceCGJ = Canadian Geotechnical JournalECSMFE = European Conference on Soil Mechanics and Foundation EngineeringESOPT = European Symposium on Penetration TestingGTJ = Geotechnical Testing JournalICSMFE = International Conference of Soil Mechanics and Foundation EngineeringIJNAMG International Journal for Numerical and Analytical Methods m GeomechanicsISOPT = International Symposium on Penetration TestingISPMA International Symposium on the Pressuremeter and Its Marine ApplicationsISSRUT = International Symposium on Soil and Rock Investigation by In Situ TestingJ-GED = Journal of the Geotechnical Engineering DivisionJ-SMFD = Journal of the Soil Mechanics and Foundation DivisionOTC = Offshore Technology ConferencePCSMFE = Pan-American Conference of Soil Mechanics and Foundation EngineeringSCIMSP = Specialty Conference on In Situ Measurement of Soil PropertiesSF = Soils and FoundationsSMS = Soil Mechanics SeriesUSSSRIT = Updating Subsurface Sampling of Soils and Rocks and their In-Situ TestingAges, A. and Woollard, A. (1976). The tides in the Fraser estuary. Pacific Marine Science Report 76-5,Canada Department of Environment.Alpan, 1. (1967). The empirical evaluation of the coefficient K0 and KOR. SF, 7, No. 1, 31-40.Anderson, W.F. and Pyrah, I.C. (1989). Consolidation and creep effects in the PMT in clay. XII ICSMFE,Rio de Janeiro, Vol. 1, 153-156.Andrawes, K.Z. and El-Sohby, M.A. (1973). Factors affecting coefficient of earth pressure K0. ASCE,J-SMFD, 99, SM7, July, 527-539.Arnold, M. (1981). An empirical evaluation of pressuremeter test data. CGJ, 18, 455-459.Arthur, J.R.F. and Menzies, B.K. (1972). litherent anisotropy in a sand. Geotechnique, 22, No. 1, 115-128.Atkinson, J.H. and Bransby, P.L. (1978). The mechanics of soil-an introduction to critical state mechanics.Mc. Graw-Hill, London.Atkinson, J.H. and Richardson, D. (1985). Elasticity and normality in soil-experimental examinations.Geotechnique, 35, No. 4, 443-449.231Atkinson, J.H. and Salifors, G. (1991). Experimental determination of stress-strain-time characteristics inlaboratory and in situ tests. X ECSMFE, Vol. 3, Firenze, 9 15-956.Atwood, M.J. (1990). Investigation of jetting insertion procedures for rapidly deploying a self-boringpressuremeter in soft clays. M.Sc. Thesis, University ofNew Hampshire.Bachus, R.C. (1983). An investigation of the strength deformation response of naturally occurring lightlycemented sands. Ph.D. Thesis, Department of Civil Engineering, Stanford University.Baguelin, F. (1982). Rules of foundation design using self-boring pressuremeter test results. 1’ ISPMA,Paris, 347-360.Baguelin, F., Jézéquel, J.F. and Shields, D.H. (1978). The pressuremeter and foundation engineering. TransTech Publications, First Edition, Germany.Baguelin, F., Jézéquel, J.F., Le Mee, E. and Le Méhauté, A. (1972). Expansion of cylindrical probes incohesive soils. ASCE, J-SMFD, 98, SM1 1, November, 1129-1142.Baguelin, F., Jézequel, J.F., Lemée, E., Le Méhauté, A. (1972). Expansion of cylindrical probes incohesive soils. ASCE, J-SMFD, 98, SM1 1, November, 1129-1142.Baldi, G., Bellotti, R., Ghionna, V., Janiiolkowski, M. and Pasqualini, E. (1982). Design parameters forsands from CPT. 2nd ESOPT, Amsterdam, Vol. 2,425-438.Baldi, G., Bellotti, R., Ghionna, V.N., Jamiolkowski, M. and Lo Presti, D.C.F. (1989). Modulus of sandsfrom CPT’s and DMT’s. XII ICSMFE, Rio de Janeiro, Vol. 1, 165-170.Baligh, M.M. (1976). Cavity expansion in sands with curved envelopes. ASCE, J-GED, 102, GT1 1,November, 1131-1146.Barden, L. and Khayatt, A. (1966). Incremental strain rate ratios and strength of sand in the triaxial test.Geotechnique, 16, No. 4, 338-357.Been, K., Jefferies, M.G. and Hachey, J. (1991). The critical state of sands. Geotechnique 41, No. 3,365-381.Bellotti, R., Crippa, V., Ghionna, V.N., Jamiolkowski, M. and Robertson, P.K. (1987). Self-boringpressuremeter in pluvially deposited sands. Report to the U.S. Army, European Research Office.Bellotti, R., Ghionna, V., Jaxuiolkowski, M., Robertson, P.K. and Peterson, R.W. (1989). Interpretation ofmoduli from self-boring pressuremeter tests in sand. Geotechnique, 39, No. 2, 269-292.232Benoit, J. (1983). Analysis of self-boring pressuremeter tests in soft clay. Ph.D. Thesis, Department ofCivil Engineering, Stanford University.Benoit, J. and Clough, G.W. (1986). Self-boring pressuremeter tests in soft clay. ASCE, J-GED, 112,No. 1, January, 60-78.Bertok, J. (1987). Settlement of embankments and structures at Vancouver International Airport. CGJ, 24,72-80Bishop, A.W. (1961). Discussion on soil properties and their measurement. V ICSMFE, Vol. 3, 92-100.Bishop, A.W. (1966). The strength of soils as engineering materials. Geotechnique, 16, 91-128.Bishop, A.W. (1971). Shear strength parameter for undisturbed and remolded soil specimens. Proceedingsof the Roscoe Memorial Symposium, Cambridge University, 3-5 8.Blunden, RH. (1975). Urban geology of Richmond, British Columbia. Adventures in Earth Science Series,15, Canadian Geological FoundationBolton, M.D. (1986). The strength and dilatancy of sands. Geotechnique, 36, No. 1, 65-78.Borsetto, M., Imperato, L., Nova, R. and Peano, A. (1983). Effects of pressuremeters of finite length insoft clay. Proceedings ISSRIIT, Paris, Vol. 2, 211-215.Briaud, J.L. (1986). Pressuremeter and deep foundation design. 2” ISPMA, Texas A & M, 376-405.Brooker, E.W. and Ireland, H.O. (1965). Earth pressures at rest related to stress history. CGJ, 2, No. 1,February, 1-15.Brown, PT. (1985). Predicting laterally loaded pile capacity using the pressuremeter. M.Sc. Thesis,Department of Civil Engineering, University of British Columbia.Bruzzi, D., Ghionna, V., Janiiolkowski, M., Lancellotta, R. and Manfredini, 0. (1986). Self-boringpressuremeter in Po river sand. 2”’ ISPMA, Texas A & M, May, 57-74.Budhu, M. (1979). Simple shear deformation of sands. Ph.D. Thesis, University of CambridgeBurland, J.B. (1989). Ninth Laurits Bjerrum Memorial Lecture: small is beautiful-the stiffness of soil atsmall strains. CGJ, 26, 499-516.233Byrne, P.M., Salgado, F.M. and Howie, J.A. (1990). Relationship between the unload shear modulus frompressuremeter tests and the maximum shear modulus for sand. 3F ISPMA, Oxford Univ., 231-241.Campanella, R.G. and Robertson, P.K. (1981). Applied cone research. Proceedings of the Symposium oncone penetration testhg and experience, St. Louis, 343-362.Campanella, R.G. and Robertson, P.K. (1982). State of the art in in-situ testing of soils: Developmentssince 1978. California Conference on USSSRIT, Santa Barbara, 245-267.Campanella, R.G. and Robertson, P.K. (1986). Research and development of the UBC cone pressuremeter.3” Canadian Conference on Marine Geotechnical Engineering, Memorial University, Newfoundland.Campanella, R.G. and Robertson, P.K. (1988). Current status of the piezocone test. ISOPT-1, Orlando,93-116.Campanella, R.G. and Stewart, W.P. (1990). Seismic cone analysis using digital signal processing fordynamic site characterization. 43 CGC, Quebec.Campanella, R.G., Gillespie, D. and Robertson, P.K. (1982). Pore pressures during cone penetrationtesting. 2 ESOPT, Amsterdam, 507-512.Campanella, R.G., Robertson, P.K. and Gillespie, D. (1986). Seismic cone penetration test. In Situ’ 86,Geotechnical Special Pub. No. 6, ASCE, 116-130.Campanella, R.G., Stewart, W.P. and Jackson, R.S. (1990). Development of the UBC self-boringpressuremeter. 3’’ ISPMA, Oxford University, 65-72.Carter, J.P., Booker, J.R. and Yeung, S.K. (1986). Cavity expansion in cohesive frictional soils.Geotechnique, 36, No. 3, 349-358.Casagrande, A. (1936). Characteristics of cohesionless soils affecting the stability of slopes and earth fills.Journal ofthe Boston Society of Civil Engineers, January.Chong, M.K. (1988). Density changes of sand on cone penetration resistance. Proc. ISOPT 1, 707-7 14.Clague, J.J., Luternauer, J.L. and Hebda, R.J. (1983). Sedimentary environments and postglacial history ofthe Fraser Delta and lower Fraser Valley, British Columbia. Canadian Journal of Earth Sciences, 20,1314-1326.Clague, J.J., Naesgaard, E. and Sy, A. (1992). Liquefaction features on the Fraser delta: evidence forprehistoric earthquakes?. Canadian Journal of Earth Sciences, 29, 1734-1745.234Clarke, B.G. (1981). 1u situ testing of clays using the Cambridge self-boring pressuremeter. Ph.D. Thesis,Cambridge University.Clarke, B.G. (1992). The interpretation of self-boring pressuremeter tests to produce design parameters.Proceedings of the Wroth Memorial Symposium, Oxford University, 75-88.Clarke, B.G. and Wroth, C.P. (1985). Discussion on Effect of disturbance on parameters derived fromself-boring pressuremeter tests in sand. Geotechnique, 219-222.Clough, G.W. and Denby, G.M. (1980). Self-boring pressuremeter study of San Francisco bay mud.ASCE, J-GED, 106, GT1, January, 45-63.Clough, G.W., Briaud, J.L. and Hughes, J.M.O. (1990). The development of pressuremeter testing. 3idISPMA, Oxford University, 25-45.Cole, E.R.L. (1967). The behaviour of soils in the simple shear apparatus. Ph.D. Thesis, University ofCambridge.Cornforth, D.H. (1964). Some experiments on the influence of strain conditions on the strength of sand.Geotechnique, 14, No. 2, June, 143-167.Comforth, D.H. (1973). Prediction of drained strength of sand from relative density measurements. ASTMSpecialty Technical Publication 523, 28 1-303.t)aramola, 0. (1980). Effect of consolidation age on stiffness of sand. Geotechnique, 30, 213-216.Daramola, 0. (1980a). On estimating K0 for overconsolidated granular soils. Geotechnique, 30, 310-313.De Souza Coutinho, A.G.F. (1988). Theories for the interpretation of pressuremeter expansion tests.Memória No. 728, LNEC, Lisbon.De Souza Coutinho, A.G.F. (1990). Radial expansion of cylindrical cavities in sandy soils: application topressuremeter tests. CGJ, 67, 737-748.Denby, G.M. (1978). Self-boring pressuremeter study of the San Francisco bay mud. Ph.D. Thesis,Department of Civil Engineering, Stanford University.Denby, G.M. and Hughes, J.M.O. (1982). Horizontal stress interpretation of pressuremeter tests.California Conference on USSSRIT, Santa Barbara, 227-244.235Dobry, R., Powell, D.J., Yokel, F.Y. and Ladd, R.S, (1980). Liquefaction potential of saturated sand-thestiffness method. Proc. 7th World Conference Earthquake Engineering, Instanbul, 3, 25-32.Durgunoglu, H.T. and Mitchell, J.K. (1975). Static penetration resistance of soils, Il-Evaluation of theoryand implications for practice. Proceedings SCIMSP, Raleigh, 172-189.East, D.R., Cincilla, W.A., Hughes, J.M.O. and Benoit, J. (1988). The use of the electric piezocone formine tailings deposits. ISOPT-1, Orlando, 745-750.Everard, 1. (1992). In situ test methods at the Laing Bridge site. CVL 577 Final Report, Department ofCivil Engineering. University of British Columbia.Fahey, M. (1986). Expansion of a thick cylinder of sand: a laboratory simulation of the pressuremeter test.Geotechnique, 36, No. 3, 397-424.Fahey, M. (1988). The influence of creep on in situ pore pressure dissipation tests. ISOPT-1, Orlando.Fahey, M. (1990). Shear modulus of sand measured with the self-boring pressuremeter. Report GlOb,Geomechanics Group, The University ofWestern Australia.Fahey, M. (1992). Shear modulus of cohesionless soil: variation with stress and strain level. CGJ, 29,157-16 1.Fahey, M. and Carter, J.P. (1993). A finite element study of the pressuremeter test in sand using a nonlinear elastic plastic model. CGJ, 30, No. 2, April, 348-3 62.Fahey, M. and Jewel!, R. (1990). Effect of pressuremeter compliance on measurement of shear modulus.3rd ISPMA, Oxford University, 115-124.Fahey, M. and Randolph, M.F. (1984). Effect of disturbance on parameters derived from self-boringpressuremeter tests in sand. Geotechnique, 34, No. 1, 81-97.Fahey, M., Jewell, R.J. and Brown, T.A. (1988). A self-boring pressuremeter system. GTJ, 11, No. 3,187-194.Fahey, M., Robertson, P.K. and So!iman, A.A. (1993). Towards a rational method of predictingsettlements of spread footings on sand. Research Report G1086, The University ofWestern Australia.Ferreira, R.S. (1992). Interpretation of pressuremeter tests using a curve fitting technique. Ph.D. Thesis,Department of Civil Engineering. University of Alberta.236Findlay, R.C. (1991). Use of the 9-arm self-boring pressuremeter to measure horizontal in situ stress, stressanisotropy, and stress-strain behavior in soft clay. Ph.D. Thesis, University ofNew Hampshire.Ghionna, V., Janiiolkowski, M., Lacasse, S., Ladd, C.C., Lancellotta, R. and Lunne, T. (1983). Evaluationof selfboring pressuremeter. Proceedings of the ISSRIIT, Paris, Vol. 2, 397-402.Gibson, R.E. and Anderson, W.F. (1961). In situ measurement of soil properties with the pressuremeter.Civil Eng. Publication Works Rev., 56, No. 658, May, 615-618.Graham, J.P. and Jefferies, M.G. (1986). Some examples of in situ lateral stress determination in hydraulicfills using the self-boring pressuremeter. 39th CGC, Ontario, 191-199.Green, G.E. and Bishop, A.W. (1969). A note on the drained strength of sand under generalized strainconditions. Geotechnique, 19, 144-149.Hansen, B. (1958). Line ruptures regarded as narrow rupture zones; basic equation based on kinematicconsiderations. Proceedings of the Brussels Conference on Earth Pressure Problems, Vol. 1, 39-48.Hardin, B.O. and Drnevich, V.P. (1972). Shear modulus and damping in soils: design equations andcurves. ASCE, J-SMFD, 98, SM7, July, 667-691.Hardin, B.O. and Drnevich, V.P. (1972a). Shear modulus and damping in soils: measurement andparameters effects. ASCE, J-SMFD, 98, SM6, June, 603-624.Hedron, A.J. (1963). The behaviour of sand in one dimensional compression. Ph.D. Thesis, University ofIllinois.Hers, 1. (1989). The analysis and interpretation of the cone pressuremeter in cohesive soils. M.Sc. Thesis,Department of Civil Engineering. University of British Columbia.Hettler, A. and Vardoulakis, I. (1984). Behaviour of dry sand tested in a large triaxial apparatus.Geotechnique, 34, No. 2, 183-198.Holtz, R.D. and Kovacs, W.D. (1981). An introduction to geotechnical engineering. Prentice-Hall, Inc.Houlsby, G.T. and Hitchman, R. (1988). Calibration chamber tests of a cone penetrometer in sand.Geotechnique, 38, No. 1, 39-44.Houlsby, G.T., Clarke, B.G. and Wroth, C.P. (1986). Analysis of the unloading of a pressuremeter in sand.2rd ISPMA, Texas A & M, May, 245-262.237Howie, J,A. (1991). Factors affecting the interpretation and analysis of full-displacement pressuremetertests in sands. Ph.D. Thesis, Department of Civil Engineering, University of British Columbia.Hughes, J.M.O. (1973). An instrument for in situ measurement in soft clays. Ph.D. Thesis, University ofCambridge.Hughes, J.M.O. (1982). Interpretation of pressuremeter tests for the detennination of the elastic shearmodulus. California Conference on USSSRIT, Santa Barbara.Hughes, J.M.O. (1984). Pressuremeter results obtained using the Western Geosystems Inc. self-boringpressuremeter at McDonald farm. Report for the National Research Council of Canada, February.Hughes, J.M.O. (1986). Calibration requirements. In situ test instruments: pressuremeter. Workshop onGeotechnical In Situ Testing for the Canadian Offshore, Bedford Inst. of Oceanography, Dartmouth.Hughes, J.M.O. (1989). The pressuremeter: can useful data be obtained in granular material? Civil 577Notes, Department of Civil Engineering, University of British Columbia.Hughes, J.M.O. and Robertson, P.K. (1984). Full displacement pressuremeter testing in sand. SMSNo. 78, Department of Civil Engineering. University of British Columbia.Hughes, J.M.O. and Robertson, P.K. (1985). Full-displacement pressuremeter testing in sand. CGJ, 22,298-307.Hughes, J.M.O., Jefferies, M.G. and Mon-is, DL. (1984). Self-bored pressuremeter testing in the Articoffshore. 16th OTC, OTC 4676, Houston, 255-264.Hughes, J.M.O., Wroth, C.P. and Windle, D. (1977). Pressuremeter tests in sands. Geotechnique, 27,No. 4, 455-477.Hunt, R.E. (1986). Geotechnical engineering techniques and practices. McGraw-Hill, New York.Idriss, I.M. (1990). Response of soft soil sites during earthquakes. Proceedings H.B. Seed MemorialSymposium, Berkeley, 273-289.Ishihara, K. (1982). Evaluation of soil properties for use in earthquake response analysis. InternationalSymposium on Numerical Models in Geomechanics, Zurich.Iwasaki, T., Tatsuoka, F. and Takagi, Y. (1978). Shear moduli of sands under cyclic torsional shearloading. SF, 18, No. 1, March, 39-56,238Jackson, J.G.J., Ehrgott, J.Q. and Rohani, B. (1980). Loading rate effects on compressibility of sand.ASCE, J-GED, Vol. 106, GT8, August, 839-852.Jáky, 3. (1944). The coefficient of earth pressure at rest. Journal for Society of Hungarian Architects andEngineers, October, 355-358. (in Hungarian).Janiiolkowski, M., Ladd, C.C., Germaine, J.T. and Lancellotta, R. (1985). New developments in field andlaboratory testing of soils. Proceedings XI ICSMFE, San Francisco, Vol. 1, 57-153.Janbu, N. (1963). Soil compressibility as determined by oedometer and triaxial tests. 3d ECSMFE,Wiesbaden, 2, 19-24.Jardine, R.J., Potts, D.M., Fourie, A.B. and Burland, J.B. (1986). Studies of the influence of non linearstress-strain characteristics in soil-structure interaction. Geotechnique, 36, No. 3, 377-396.Jefferies, M.G. (1988). Determination of horizontal geostatic stress in clay with self-bored pressuremeter.CGJ, 25, 559-573.Jefferies, M.G., Crooks, J.H.A., Becker, D.E. and Hill, P.R. (1987). Independence of geostatic stress fromoverconsolidation in some Beaufort sea clays. CGJ, 24, 342-35 6. Plus Discussion on CGJ, 25,624-630, 1988.Jewell, R.J., Fahey, M. and Wroth, C.P. (1980). Laboratory studies of the pressuremeter test in sand.Geotechnique, 30, No. 4, 507-53 1.Juran, I. and Mahmoodzadegan, B. (1989). Interpretation procedure for pressuremeter tests in sand. ASCE,J-GED, 115, No. 11, November, 1617-1632.Kogler, F. (1933). Baugrundprufung im bohrloch. Der Bauingenieur, Berlin, 19-20.Kondner, R.L. (1963). Hyperbolic stress-strain response: cohesive soils. ASCE, J-SMFD, Vol. 89, SMI,February, 115-143Lacasse, S. and Lunne, T. (1982). In situ horizontal stress from pressuremeter tests. l ISPMA, Paris,187-207.Lacasse, S., D’Orazio, T.B. and Bandis, C. (1990). Interpretation of self-boring and push-in pressuremetertests. ISPMA, Oxford University, 273-285.Ladanyi, B. (1963). Evaluation of pressuremeter tests in granular soils. 2’ PCSMFE, 1, 3-20.239Ladd, C.C., Foott, R., Ishihara, K., Schiosser, F. and Poulos, H.G. (1977). Stress-deformation and strengthcharacteristics. State of Art Report, IX ICSMFE, Tokyo.Lade, P.V. (1988). Effects of voids and volume changes on the behaviour of frictional materials. LJNAMG,12, 35 1-370.Lade, P.V. and Lee, K,L. (1976). Engineering properties of soils. Report UCLA-ENG-7652, University ofCalifornia, School of Engineering and Applied Science.Laier, J.E., Schmertmann, J.H. and Schaub, J.H. (1975). Effect of finite pressuremeter length in dry sand.Proceedings SCIMSP, Raleigh, 241-259.Lambe, T.W. (1951). Soil testing for engineers. John Wiley and Sons, New York.Lambe, T.W. and Whitman, R.V. (1979). Soil Mechanics. John Wiley & Sons, Inc.Lambrechts, J.R. and Leonards, G.A. (1978). Effects of stress history on deformation of sand, ASCE,J-GED, 104, GT11, November, 1371-1387.Lamé, G. (1852). Leçons sur hi théorie mathématique de l’elasticité des corps solides. Bachelier, Paris,France.Law, K.T, and Eden, W.J. (1980). Influence of cutting shoe size in self-boring pressuremeter tests insensitive clays. CGJ, 17, 165-173.Lee, K.L. (1970). Comparison of plane strain and triaxial tests on sand. ASCE, J-SMFD, 96, SM3, May,901-923.Lee, K.L. and Seed, H.B. (1967). Drained strength characteristics of sands. ASCE, J-SMFD, 93, SM6,November, 117-14 1.Le Clair, D.G. (1988). Prediction of embankment performance using in situ testing. M.Sc. Thesis,Department of Civil Engineering, University of British Columbia.Lo Presti, D. (1987). Mechanical behaviour of Ticino sand from resonant column tests. Ph.D. Thesis,Politecmco di Torino.Mair, R.J. and Wood, D.M. (1987). Pressuremeter testing-methods and interpretation. CIRIA Report.240Manassero, M. (1989). Stress-strain relationships from drained self-boring pressuremeter tests in sands.Geotechnique, 39, No. 2, 293-307.Marchetti, S. (1980). In situ tests by flat dilatometer. ASCE, J-GED, 106, GT3, March, 299-321.Matsuoka, H. (1976). On the significance of the spatial mobilized plane. SF, 116, 91-100.Mayne, P.W. and Kulhawy, F.H. (1982). K0-OCR relationships in soil. ASCE, J-GED, 108, GT6, 851-872. and Discussions onJ-GED, 109, No. 6, 1983, 859-869.Ménard, L. (1955). Travail personnel sur le pressiomètre. Ecole Nationale des Ponts et Chaussées, Paris.Menard, L. (1957). Mesures in situ des propriétés physiques des sols. Annales des Ponts et Chaussées,Paris, 127, 3, 357-377.Milliman, J.D. (1980). Sedimentation in the Fraser River and its estuary, southwestern British Columbia(Canada). Estuarine and Coastal Marine Science, 10, 609-633.Milovic, D. (1992). Stresses and displacements for shallow foundations. Elsevier Editors.Monahan, P.A., Luternauer, J.L. and Barrie, J.V. (1992) A delta topset “sheet” sand and modernsedimentary processes in the Fraser River delta, British Columbia. Current Research, Part A,Geological Survey of Canada, Paper 93-lA.Murthy, R.T. (1992). Shear modulus and damping properties of sands from cyclic self-boringpressuremeter tests. M.Sc. Thesis, Department of Civil Engineering. University of British Columbia.Negussey, D. (1984). An experimental study of the small strain response of sand. Ph.D. Thesis,Department of Civil Engineering. University of British Columbia.Negussey, D. and Vaid, Y.P. (1990). Stress dilatancy of sand at small stress ratio states. SF, 30, No. 1,March, 155-166.Negussey, D., Wijewickreme, W.K.D. and Vaid, Y.P. (1988). Constant-volume friction angle of granularmaterials. CGJ, 25, 50-55.Newman, R.L., Chapman, T.J.P. and Simpson, B. (1991). Evaluation of pile behavior from pressuremetertests. X ECSMFE, Firenze, Vol.2,501-504.241O’Neill, B. (1985). An evaluation of the full displacement pressuretneter. M.Sc. Thesis, Department ofCivil Engineering. University of British Columbia.Oda, M. (1972). Initial fabrics and their relations to mechanical properties of granular material. SF, 12,No. 1, March, 17-36.0th, M. (1981). Anisotropic strength of cohesionless sands. ASCE, J-GED, 107, GT9, 1219-1231.Paikowsky, S.G. (1990). The mechanism of pile plugging in sand. Proceedings 22’ OTC, V.4, 593-604Prapaharan, S., Chameau, J.L., Altschaeffl, A.G. and Holtz, RD. (1990). Effect of disturbance onpressuremeter results in clays. ASCE, J-GED, Vol. 116, No. 1, 35-53.Rittenhouse, 0. (1943). A visual method of estimating two dimensional sphericity. Journal of SedimentaryPetrology, 13, No. 2, 79-8 1.Robertson, P.K. (1982). In-situ testing of soil with emphasis on its application to liquefaction assessment.Ph.D. Thesis, Department of Civil Engineering. University of British Columbia.Robertson, P.K. (1986). In situ testing and its application to foundation engineering. CGJ, 23, 573-594.Robertson, P.K. and Campanella, R.G. (1989). Guidelines for geotechnical design using CPT and CPTU.SMS No. 120, Department of Civil Engineering. University of British Columbia.Robertson, P.K. and Hughes, J.M.0. (1986). Determination of properties of sand from self-boringpressuremeter tests. 2 ISPMA, Texas A & M, May, 283-302.Roesler, S.K. (1979). Anisotropic shear modulus due to stress anisotropy. ASCE, J-GED, 105, GT7, July,871-880.Roscoe, K.H., Schofield, A.N. and Wroth, C.P. (1958). On the yielding of soils. Geotechnique, 8, No. 1,22-53.Rowe, P.W. (1962). The stress dilatancy relation for static equilibrium of an assembly of particles incontact. Proceedings of the Royal Soc. London, Series A, Vol. 269, 500-527.Rowe, P.W. (1969). The relation between the shear strength of sands in tnaxial compression, plane strainand direct shear. Geotechnique, 19, No. 1, 75-86.242Rowe, P.W. (1971). Theoretical meatung and observed values of deformation parameters for soil.Proceedings Roscoe Memorial Symposium, 143-194.Saglamer, A. (1975). Soil parameters affecting coefficient of earth pressure at rest of cohesionless soils.Instanbul Conference on Soil Mechanics and Foundation Engineering, Vol. 1, 9-16.Salgado, F.M. and Byrne, P.M. (1990). Finite element analysis of pressuremeter chamber tests in sand. 3’ISPMA, Oxford University, 209-219.Sasitharan, 5. (1989). Stress path dependency of dilatancy and stress-strain response of sand. M.Sc.Thesis, Department of Civil Engineering. University of British Columbia.Schmertmann, J. H. (1991). The mechanical aging of soils. ASCE. J-GED, Vol. 117, No. 9, 1288-1329.Schmertmann, J.H. (1970). Static cone to compute static settlement over sand. ASCE, J-SMFD, Vol. 96,SM3, May, 1011-1043.Schmertmann, J.H. (1975). Measurement of in situ shear strength. ASCE, Proceedings of the SpecialtyConference on In Situ Measurement of Soil Properties, Vol. 2, Raleigh, 57-138.Schmertmann, J.H. (1985). Measure and use of the insitu lateral stress. The Practice of FoundationEngineering-Osterberg Volume, Northwestern University, Department of Civil Engineering, 189-2 13.Schmidt, B. (1966). Discussion of earth pressure at rest related to stress history. CGJ, 3, No. 4, 239-242.Schnaid, F. and Houlsby, G.T. (1991). An assessment of chamber size effects in the calibration of in situtests in sand. Geotechnique, 41, No. 3, 437-445.Seed, H.B. and Idriss, I.M. (1970). Soil moduli and damping factors for dynamic response analyses.Report EERC 70-10, College of Engineering, University of California at Berkeley, December.Shibuya, S., Tatsuoka, F., Teachavorasinskun, S., Park, C. and Abe, F. (1991). Elastic properties ofgranular materials measured in the laboratory. X ECSMFE, Firenze, 163-166.Stroud, M.A. (1971). Sand at low stress levels in the simple shear apparatus. Ph.D. Thesis, University ofCambridge.Sully, J.P. (1991). Measurement of in situ lateral stress during full-displacement penetration tests. Ph.D.Thesis, Department of Civil Engineering. University of British Columbia.243Suyama, K., Ohya, S., Imai, T., Matsubara, M. and Nakayama, E. (1983). Proceedings of the ISSRITT,Paris, Vol. 2, 397-402.Tatsuoka, F. (1976). Stress dilatancy relations of anisotropic sands in three dimensional stress conditions.SF, 16, No. 2, 1-18.Tatsuoka, F., Iwasaki, T., Fukushima, S. and Sudo, H. (1979). Stress conditions and stress historiesaffecting shear modulus and damping of sand under cyclic loading. SF, 19, No. 2, 29-43.Teachavorasinskun, S., Shibuya, S. and Tatsuoka, F. (1991). Stiffness of sands in monotonic and cyclictorsional simple shear. ASTM Publication 27, Vol. 2, Boulder, 863-878.Terzaghi, K. and Peck, R.B. (1967). Soil mechanics in engineering practice. John Wiley and Sons, Inc.Vaid, Y.P. (1985). Effect of consolidation history and stress path on hyperbolic stress-strain relations.CGJ, 22, 172-176.Vaid, Y.P., Byrne, P.M. and Hughes, J.M.O. (1980). Dilation rate as a measure of liquethction resistanceof saturated granular materials. SMS No. 43, Department of Civil Engineering, UBC.VesIc, A.S. (1972). Expansion of cavities in infinite soil mass. ASCE, J-SMFD, 98, SM3, 265-290.VesIc, A.S. and Clough, G.W. (1968). Behavior of granular materials under high stresses. ASCE,J-SMFD, 94, SM3, May, 661-688.White, J.E. (1965). Seismic waves. McGraw-Hill, Inc., New York.Whittle, R.W., Dalton, J.C.P. and Hawkins, P.G. (1992). Shear modulus and strain excursion in thepressuremeter test. Proceedings of the Wroth Memorial Symposium, Oxford.Windle, D. and Wroth, C.P. (1975). Electrical resistivity method for determining volume changes thatoccur during a pressuremeter test. Proceedings SCIMSP, Raieigh, 497-510.Withers, N.J., Howie, J., Hughes, J.M.O. and Robertson, P.K. (1989). Performance and analysis of conepressuremeter tests in sands. Geotechnique, 39, No. 3,433454.Wong, R.K.S. and Arthur, J.R.F. (1985). Induced and inherent anisotropy in sand. Geotechnique, 35, No.4,471481.Wood, D.M. (1990). Strain-dependent moduli and pressuremeter tests, Geotechnique, 40, No. 3, 509-5 12.244Wood, D.M. and Wroth, C.P. (1977). Some laboratory experiments related to the results of pressuremetertests. Geotechnique, 27, No. 2, 18 1-201.Wroth, C.P. (1975). In situ measurement of initial stresses and deformation characteristics. Proceedings ofthe SCIMSP, Raleigh, Vol. 2, 181-230.Wroth, C.P. (1982). British experience with the self-boring pressuremeter. Proceedings of the ISPMA,143-164,Wroth, C.P. (1984). The interpretation of in situ soil tests. Geotechnique, 34, No. 4, 449-489.Wroth, C.P. and Windle, D. (1975). Analysis of the pressuremeter test allowing for volume change.Geotechnique, 25, No. 3, 598-604.Wroth, C.P., Randolph, M.F., Houlsby, G.T. and Fahey, M. (1979). A review of the engineering propertiesof soils with particular reference to the shear modulus. Soil Mechanics Report No. SM048/84,Cambridge University.Yan, L. (1986). Numerical studies of some aspects with pressuremeter tests and laterally loaded piles.M.Sc. Thesis, Department of Civil Engineering. University of British Columbia,Youd, T.L. and Craven, T.N. (1975). Lateral stress in sands during cyclic loading. ASCE, J-GED, 101,GT2, February, 217-221.Yu, P. and Richart, F.E. (1984). Stress ratio effects on shear modulus of dry sands. ASCE, J-GED, 110,No. 3, March, 33 1-345.Yu, H.S. (1993). A new procedure for obtaining design parameters from pressuremeter tests. AustralianCivil Engineering Transactions, Vol. CE34, No. 4, 353-359.245APPENDIX A DESCRIPTION OF THE UBC SBPMA.! INTRODUCTIONThis appendix briefly presents the basic characteristics of the UBC SBPM system used in this thesis.The UBC SBPM system consists of 3 main units: the testing, the pumping and the pushing units. Thetesting unit consists of the data acquisition system (and related sensors) required to control the probe, theair control system required to inflate the probe, and the pressuremeter probe itself. The pumping unit iscomposed of the pumping equipment (and related accessories) required to flush the drilling mud during theselfboring operation. The pushing unit is a hydraulic cross-head and other accessories housed in the UBCtesting vehicle.A.2 PUSHING AND PUMPING UNITSSelfboring of the UBC probe is achieved by the combined action of pushing and jetting. Thesupporting pushing and hydraulic control units are housed inside the UBC research vehicle. This in situtesting vehicle is detailed in Campanella and Robertson, 1981, and is depicted in Figure A.1(a). Theresearch truck has 4 built in hydraulic systems that operate independently in order to give versatility to thewhole pushing unit. Each of the systems is for a different task, such as the raising and leveling of the truck,the pushing and pulling of the in situ tools, or the clamping action of the steel rods. A variable volumepressure-compensated hydraulic pump is used to supply the hydraulic needs of all 4 independent systems.This hydraulic pump has an operational capacity of providing 6900 kPa (1000 psi) of pressure, whichsupplies the penetration head circuit (pushing rig) with a pushing capacity of 80 kN (10 tons). Thiscorresponds approximately to the dead weight reaction of the vehicle. Thrust in the pushing rig is providedby 2 double sided cylinders connected to the chuck head of the pushing frame. The maximum stroke of thepushing frame is 121.8 cm, however during the seilboring process a stroke of 100 cm is used. Thiscorresponds to the length of the steel BW rods used to connect the pressuremeter unit to the chuck headinside the truck.The internal hydraulic circuit of the research truck is additionally used to supply the pumping unit ofthe UBC SBPM system, as represented in the layout of Figure A. 1(b). Since pumping and pushingE LOUT OF Hr’DRALUC AND PUMP CONTPOL UNITMudpressuregauge Pressure246gaugeMudReservoirElectronicControlunitElectricDitch pumpVRODCableRESEARCH TRUCKSBPM(A) LAYOUT OF SBPM DRILLING AND PUSHING EQUIPMENT-cUSw iv eci0-JMudFlow DiverterElectric Pump289nWater*/E0U0DrumsFigure A. 1: Pushing and Pumping Units of the UBC SBPM System247occur simultaneously during the selfboring operation, the hydraulic power available in the truck has to beshared between the pushing and pumping hydraulic units.The connection between the truck and the outside pumping unit is done through 2 hydraulic lines.These lines feed a diaphragm type pump which is used to pump the mud during the selfboring operation.This pump is rated to provide a maximum flow of 48 1/mm (12.7 gpm) under an ideal pressure condition of3790 kPa (550 psi). A manual regulator is used to control the mud flow rate at the base of the SBPM.Water from a ditch or any other source at the site is collected through a Ponstar submersible pump to oneof the 170 1 (45 gallons) drums of the pumping unit. This drum serves to keep a supply of water that canbe readily used in any stage of the selfboring operation.This water is subsequently used to prepare the jetting mud, required by the selfboring operation. Thewater is transferred to another drum, where it is mixed with a “liquid avionic polymer” WDS-120. TheWDS-120 is a drilling additive from Westcoast Drilling Supplies Ltd. that serves to improve the flowcharacteristics of the fluid, to prevent collapse of the selfbored hole, as well as to enable the granularminerals displaced during the selfboring operation to reach surface. A ratio of 2 glasses (200 ml each) ofadditive per 170 1 of water was in general used in this thesis. The mixture of water and drilling additive, orjetting mud, is homogenized by diverting the flow from the mud pump back to the drum (see Figure A. 1(b))in a closed loop. This diverting process is done until a viscous and homogeneous fluid consistency isachieved.Once the jetting mud is prepared it is forced down to the pressuremeter unit through a 1.27 mm (1/2”)PVC tubing located inside the annular space of the steel BW rods. A swivel is used to adapt the BW rodsto the chuck head of the pushing rig, as well to establish a link between the PVC tubmgs and the pumpingunit. This swivel also houses a mud flow and mud pressure gauges.A.3 TESTING UNITA.3.1 Data Acquisition System and Related SensorsIn a similar way as the pumping unit the pressuremeter testing unit is connected to the facilitiesavailable inside the research truck. The layout of the whole testing unit is presented in Figure A.2.The UBC selfboring pressuremeter is equipped with 6 strain arm sensors, 2 pressure and 100LU)G)L)000-oU)U)CCoU)HoUUU)U)(flu)U--CUW0)E<•CC-o000200020)(-‘4(/)-o0)0U_00oo:1.)00UU)U)—Uwo0)00-oU)C0U)UUF—U)CDC--—Q0Q)UU-5OEo000<H-)rJ)rJ)J0DflSUDJjOJflS9OJ5IJoDnp9LIDJjMOIS249temperature transducer, 2 accelerometers and 1 load cell at the base of the cutting shoe. There are also twoanalog boards as well as a microcomputer board inside this probe. The function of these boards is relatedto amplification and conversion of the signals measured by the aforementioned sensors. Each of thesesensors have an instrumentation amplifier that gets the low level signal and amplifies it up to the conversionrange of the 12 bit A/D converter. The converter, a linear Technology LTC 1290, sends the digital dataupwards via a high speed serial interface RS 422 to the interface controller at surface. The RS 422 serialinterface is connected to a cable which is housed together with the air tube inside a 6.4 mm (outsidediameter) cable that is taped to the BW rods. The air tube is used to send pressurized air downhole duringthe expansion of the probe and is ultimately linked, via a pressure regulator, to an air source of 1860 kPa(270 psi) inside the truck.The interface controller contains a 12 bit D/A converter with a Motorola 68HC 11 microcontroller,which is programmed to receive the raw data from downhole and convert it into ASCII format inengineering units. This ASCII data is then sent via a ribbon serial cable to a RS232 port on the IBM PCcompatible 486 computer of the truck. The computer runs a data acquisition system developed at UBC toretrieve and store the pressuremeter data, as well as to set up and control the testing characteristics. Thedata acquisition system displays in real time all the values measured downhole, and presents a graphicaldisplay ofthe internal pressure versus circumferential strain of the test.Three testing modes are accomplished during selfboring and testing, namely the “testing”, “dissipation”and “sounding” modes. The data acquisition system and the interface controller “know” the present statusof the test (as defined above) by the connection of two switches between the pushing rig and the interfacecontroller. These are the “run” and the “load” switches of Figure A.2. These switches inform the controllerif the swivel is being pushed or if it is in a stopped position, thus allowing the data acquisition system tochange the mode of operation from “sounding” to “dissipation” modes. The variables recorded in“sounding” and “dissipation” modes are distinct and are stored in different files. The total (tip plus lateral)load, the inclination of two perpendicular planes to the probe vertical axis, the mud flow and pressure, thepore pressure, the temperature as well as the elapsed time are stored in the “sounding” file. The porepressure dissipation, the temperature and the time elapsed in the halting stages are stored in the“dissipation” file. As commented above, these ASCII files contain the measured variables already in250engineering units. The “testing” mode is started only by manual intervention during the halted stage of thesounding. The variables recorded in this mode are the circumferential strains measured by the 6 strainarms of the probe, the internal total and pore pressures, and the time elapsed during the pressuremeter test.Similar to the other modes, these variables are recorded in a “testing” file. The baselines of each of theaforementioned sensors are also included in the files generated by the data acquisition system.In order to keep track of the depth, principally during the “sounding” mode, an optical encoderadjusted to the hydraulic piston is connected to the interfuce controller. This encoder furnishes the dataacquisition system with depth values of the pressuremeter probe. The depth as well as any other measuredvariable can be read in continuous intervals of time as low as 1 second. The frequency of reading isselected by the engineer during the selfboring operation.The mud flow and mud pressure information is obtained through the transducers housed inside thejetting swivel. These transducers are also connected to the interface controller, as schematically shown inthe layout of Figure A.2.The interaction between the interface controller and the data acquisition system, with downholefeedback from the pressuremeter sensors, allows the knowledge of the pushing and pumping variables inthe course of a selfboring operation. This knowledge helps the standardization of the pressuremeterinsertion. Moreover it is also possible to fully standardize the testing variables adopted during expansionand deflation of the probe. The stress or strain levels at the commencement of the unload reload loop, thetime for holding prior to the unload reload loop, the maximum stress or strain levels, and the rate ofinflation, can be set up before the testing stage and controlled afterwards. The data acquisition systemallows the generation of a “command” file prior to each pressuremeter test, in which the basic testingcharacteristics (as related above) are specified by the engineer. The interface controller uses theinformation of the “command” file, plus the downhole information, to control the development of the test.A.3.2 UBC Sellbonng PressuremeterThe basic design of the UBC SBPM followed the general characteristics of the pressuremeter operatedby Dr. J. MO. Hughes. The UBC selfboring pressuremeter has an overall length of 143 cm and anexternal diameter of 74 mm, with an expandable section with length to diameter ratio of 6.251The UBC SBPM unit is shown in Figure A.3. The top of this figure shows the pressuremeter unitunder its fully assembled condition, On the center of this figure is depicted the pressuremeter unit withoutthe external components that shield the inner sensors and the expandable rubber membrane. This membraneencloses the strain arm heads and also supports the effective pressure transducer.The transducer is clamped and floats on the rubber membrane during the expansion stage. This unitserves to measure the dynamic pore pressures induced in the surrounding soil. A cylindrical porouspolypropylene filter 5.3 mm in diameter is encased on the top of the effective pressure transducer. Theinner chamber of the transducer is saturated in the field, prior to the placement of the porous filter. Thetraditional UBC saturation technique (see Robertson and Campanella, 1989) with glycerin is used. Theporous filter is pre-saturated in the UBC research laboratory by the application of an ultra sonic bath undervacuum. This technique led to a satisfactory sensor response during the calibration stages of the probe.The expandable inner rubber membrane also serves to prevent water ingress into the electroniccompartment inside the probe, and cause a short circuit of the boards. A compromise between arepeatable, elastic, flexible and at the same time resistant membrane had to be adopted for this expandablemembrane. The urethane membrane used in the past had a good ability to withstand high differential airpressures (as high as 5000 kPa), but in addition yielded high rate dependent effects (Howie, 1991) as wellas hysteretic behavior during loading and unloading in air (Hers, 1989). Given the relative low pressurerange used for the tests of this thesis (up to the maximum 1750 kPa of the air control system), a lowpuncture resistant rubber membrane was devised as a substitute for the urethane membranes. Recentresearch in this area (Campanella et al, 1990, Sully, 1991) indicated that a tubing of commerciallyavailable Gooch rubber membrane of 1 mm thickness could be adopted for the UBC SBPM. The testingresults of Campanella et al, 1990 and Sully, 1991 with the Gooch membrane indicated a bilinear envelope(expansion in air) with little or no hysteresis, as well as a high flexibility with very low correction for liftoff stresses. This membrane was used in the tests of this thesis up to 1992, when it became difficult to findGooch membranes for the UBC SBPM. A new rubber membrane from Alliance Rubber Company (also1 mm in thickness) was then selected for usage in the latter tests of this thesis.At the bottom of the pressuremeter unit two alternate jetting systems can be used for selfboring. TheDETAILS:(NottoScale)//CableProtector’////LanternRetaner//MudandCuttingsPortSteel(Chinese)Lantern/RubberMembraneRoaAdaptorCableandProbeConnectionCable(AirandElectronics)IH__/1LanternRetainerCuttingShoeStrainArms/Effective(Pore)Pressu:TransduceraEE0JettingRod010cmCentral JettingSystemShowerHeadSystem6ODegConicalTp(FDPM)FigureA.3:UBCSelfboringPressuremeterSectionalAssembly253initial design was made with the central jetting system, as shown in the lower left of Figure A.3. In thiscase the central rod is provided with 3 small vanes for stability, and a nozzle with several holes that jet themud backwards at an angle of 45°. The main advantage of this system is that the jetting holes can bepositioned at different levels inside the shoe, thus allowing a gradual optimization of the insertionprocedure. In this thesis 3 nozzles were designed and used. Nozzle A had 8 holes of 3.55 mm (diameter)located at 21.3 mm from the nozzle tip. Nozzle B had 3 holes of 5.7 mm located at 20.6 mm from thenozzle tip, and nozzle C had 4 holes of 4.0 mm located at 23.4 mm from the nozzle tip. The nozzlecharacteristics partially followed the design given by Howie, 1991 for jetting in sands.A shower head system has been also devised, as shown in the lower center of Figure A.3. In thissystem the mud passes to the outer wall and is jetted in the radial direction by a series of 12 channels(4 mm by 1.5 mm) located 40 mm from the end ofthe shoe. As before the direction of the jet is backwards,but at a lower inclination. The drawback of this system is that plugging can occur in the initial section ofthe shoe, where no jetting occurs. The internal spaces for both jetting systems limit the maximum particlesize that can be washed out to 10mm.The UBC selfboring pressuremeter can be converted into a full displacement device with the simplereplacement of the cutting shoe by a 60° conical tip. The conical tip has the same diameter as the cuttingshoe, and is shown inthe lower right of Figure A.3.Six cantilever beam type strain arms, made from beryllium copper, are located 60° apart in the middleof the expandable section. Each of the arms have 4 strain gauges. Two strain gauges are mounted on eachside of the beam, making up a fully active resistivity bridge. The arm is pivoted at one end and attached toa semi-spherical plexiglass dome on the other end. The plexiglass makes contact with the arm headunderneath the membrane. This simple but efficient design is a consequence of previous research done byHers, 1989. This author showed that the interaction between the membrane and the arm head can result inan apparent inwards movement prior to the lift off pressure, if the arm head is glued to the cantilever beam.The adopted arm configuration also circumvents the hysteresis problems observed with the typical designadopted for the Camkometer probe. This problem is well documented by Fahey and Jewell, 1990.The operating range of the strain arms of the UBC SBPM is 8 mm, which represents a maximumcircumferential strain of 22 %.254APPENDIX B CALIBRATION OF THE UBC SBPMB.1 INTRODUCTIONEquipment calibration constitutes an essential part of the pressuremeter testing procedure. It isperformed in order to obtain reliable and meaningful results without equipment related effects. It is alsoimportant to relate the measured signals to known engineering units. Transformation of the raw voltagesignals measured at the sensors of the pressuremeter (or at the swivel) to engineering ASCII results requiresa correlation between the variables. This correlation is defined by proper calibration stages performed witheach of the aforementioned sensors at regular intervals of time. The calibration also warns the user ofpossible discontinuities, voltage drifts, non linearity, hysteresis, resolution and accuracy errors of thesensors used.In this research a continuous calibration program with all the pressuremeter sensors was carried out atshort intervals of time. In general a full calibration stage was accomplished after every other day in thefield, or after some field problem (such as leakage of water into the internal probe circuitry and rupture ofthe air-electronics cable). This appendix briefly presents the typical calibration results obtained during thecalibration stages of the UBC selfboring pressuremeter.B.2 STRAIN ARMSThe strain arms were calibrated to furnish the relationship between the arm reading (in mm) and thecorresponding electronic output (in volts). The calibration was carried out without the rubber membrane,by placing a micrometer over the head of one of the arms while keeping the others in place. The arm beingcalibrated was pushed into a fully retracted condition (zero strain reference) and allowed to displaceoutwards while measuring with the micrometer. This calibration was done at 0.05 mm increments over theinitial 1 mm oftotal displacement, and thereafter at an increment of 0.5 mm.Table B. 1 presents the basic calibration characteristics of the strain arms. It is noticed that theaccuracy of the strain arms varied from 2 to 4.1 % of the full scale output (FSO), as expected givenslightly different arrangements of the resistivity bridges from arm to ann, as well as differential usageeffects over the gauges. The reported percentage values lead to a generally high accuracy of measurement255SENSOR OPERATING ACCURACY RESOLUTION SENSITIVITY TEMP. EFFECTSRANGE (%FSO) (%FSO/°C)Strain Anus1 0-8 nun 3.973 0.006% -0.496 V/mm -0.0122 0-8mm 3.862 0.006% -0.579 V/mm -0.0123 0 - 8 mm 2.971 0.006 % -0.557 V/mm -0.0284 0-8mm 4.169 0.006% -0.508 V/mm -0.0055 0 - 8 mm 2.404 0.006 % -0.555 V/mm -0.0066 0 - 8 mm 2.000 0.006 % -0.544 V/mm -0.006Internal 0 - 2976 kPa 0.53 0.70 kPa 0.00169 V/kPa -0.064PressureEff. (Pore) 0- 1316 kPa 0.35 0.10 kPa -0.00203 V/kPa -0.068PressureTip Load Cell 0 - 186 kN 0361 0.040 kN 0.0259 V/kN NAAccelerometer1 0 -20 deg 2.74 0.1 deg 0.110 V/deg NA2 0-20deg 1.37 0.ldeg 0.113V/deg NAMud Pressure 0 - 3448 kPa 0.043 13.78 kPa 0.0000945 V/kPa NA(Swivel)Mud Flow 0 -40 1/mm 0.374 0.378 11mm 0.007308 V/I/mm NA(Swivel)FSO = Full Scale OutputNA= Not AvailableTable B. 1: General Calibration Characteristics of the UBC SBPM Sensors256in the range of 0.1 to 0.3 mm, which closely tracks the arm movement from lift off to the full expansion ofthe membrane. The resolution of the arms, equal to 0.006 % (circumferential strain), did not vary fromcalibration to calibration. This is an intrinsic characteristic of the strain measuring system. Also due to thehigh value of resolution it was possible to obtain an accurate definition of the post lift off portion of thepressure expansion curves. The unload reload loops were also well delineated.B.3 PRESSURE TRANSDUCERSThe pressure transducers of the UBC SBPM were calibrated with the use of a portable digital pressureindicator Druck DPI6O 1.The (total) internal pressure sensor, a subminiature flat diaphragm sensor type from Sensotec(model F, made from stainless steel), was located in an underreamed section of the pressuremeter shaftbeneath the rubber membrane. It was calibrated by placing the expandable unit in a thick hollow steel tubeand inflating it, with the probe connected to the interface controller and data acquisition system. Table B. 1presents the general calibration characteristics of this sensor. A high resolution of 0.7 kPa and accuracy of0.53 % FSO were obtained. This high resolution and accuracy are important to precisely define the shapeof the unload reload loops, since a large and closely spaced number of data points are required.The effective (pore) pressure sensor of the pressuremeter, a differential diaphragm sensor type fromAsheroft (model K8, made from beryllium copper), was calibrated by placing the expandable unit of theprobe in a steel chamber specially designed for this purpose. The probe was connected to the interfacecontroller and data acquisition system, thus allowing a direct readout (in volts) of the air pressure appliedinside the calibration chamber. The air pressure was applied by connecting the chamber to the air supplyline of the UBC laboratory. The measured accuracy of this sensor was in the order of 0.35 % FSO asshown in Table B. 1. This value leads to an average error lower than 2.5 kPa in the pressure measured,indicating that small changes in pore pressure could be sensed by the probe (the pressure measured by thissensor is transformed into pore pressure by the data acquisition system, with the use of the measured totalinternal pressure). The resolution of the effective pressure sensor was also high, in the range of 0.1 kPa.The calibration of this sensor with a fully saturated filter element demonstrated that the adopted saturationtechnique was satisfactory for a fast and precise response ofthe sensor.257In addition to the pressure transducers of the pressuremeter unit all other measuring sensors werecalibrated in a routine basis, namely the (home made) load cell, the mud pressure and flow transducers andthe 2 accelerometers at the probe shaft. Typical calibration characteristics of all these sensors are alsopresented in Table B. 1. The measured accuracy and resolution are high enough to precisely define theinsertion variables of the UBC SBPM.B.4 MEMBRANE RESISTANCE AND LANTERN COMPLIANCEThe design of the steel lantern affects the corrections that have be applied over the experimentalresults. The calibration response of each of the designed steel lanterns, as well as the Canikometer lantern,is presented here. The design characteristics of these lanterns was shown in Chapter 3, with theCamkometer lantern defined as “lantern 10”.The membrane resistance is the force required to expand the membrane and the lantern in the absenceof external constraints. In the field test part of the measured force (or pressure) is related to the forcerequired to overcome the resistance offered by both the membrane and the lantern to expand. This forcehas to be removed from the overall value measured, in order to obtain the force applied to the soil.Membrane resistance is conventionally measured by expanding the probe in air in its fully assembledcondition. The pressure expansion curve obtained in this manner has two components, the lift off and thepost lift off component. The post lift off component is strain dependent, being related to the variableresistance offered by both the membrane and the lantern during the expansion process. The experimentallyobtained (post lift off) curve can be mathematically fitted by a polynomial equation. This equationprovides a smooth curve that is used inside the data reduction macro to correct the raw cavity pressure.The evaluation of the membrane resistance was done in the field with the expansion of the probe in airprior to the selfboring operation. A “command” file was created specially for this purpose. The calibrationtest was conducted with the expansion of the probe to a circumferential strain of 10 % at a rate of inflationof 3.4 kPa/s.The results for each of the designed steel lanterns are shown in Figure B. 1. The average of the outputof each of the six strain arms is shown for each lantern. This is so because only the average calibrationcurve was used in the correction of the raw pressuremeter data. The expansion with the rubber membranealone (without lantern) is also shown in this same figure.25836025DU)CcC40CAUBRATION RESULTS—SBPM—COMPARISON OF LANTERNSExpansion in Air—Membrane Resistance Evaluation10606 8Circumferential Strain (%)Figure B. 1: Membrane Resistance from Expansion in Air259The following observations can be drawn:1. The (post lift off) correction curve is non linear for all the steel lanterns tested, and approximatelylinear for the rubber membrane and the Camkometer standard lantern. The Camkometer lantern presents alow resistance to expansion because the displacement takes place on the outer rubber membrane, ratherthan in between the steel strips (as for the steel lanterns).2. The use of the inner Alliance rubber membrane has proven to yield a low and linear resistancecomponent on the total resistance to inflation. Therefore, the major component of the total resistance toinflation comes from the resistance imposed by the outer lantern.3. The different lanterns imposed distinct lift off resistances, that varied from 0.5 to 8 kPa. This rangeis considerably lower than the range of (effective) lift off pressures commonly measured in the tests at theLaing Bridge site. The commonly measured lift off pressures varied from 50 to 200 kPa, depending on thedepth and disturbance of the surrounding soil.4. The different lanterns also imposed distinct post lift off resistances. The measured resistance toinflation is related to the inter-strip friction, the end clamping effects, the amount of riveting and the rigidityof the whole lantern. Although differences of more than 20 kPa can be observed from one lantern toanother, in practical terms any of the lanterns can be indistinctly used. This is due to the fact that, duringthe field tests, a much higher range of post lift off pressures were measured. The maximum pressuremeasured in the field varied from 500 to 1500 kPa, respectively for tests 5 and 15 m deep in the site,whereas the maximum resistance imposed by the steel lanterns was below 40 kPa.The compliance correction is related to the inherent volume loss of the pressuremeter tubing system, aswell as the compression of the rubber membrane and the lantern. The steel lanterns are particularly proneto yield higher compliance strains than the standard Camkometer lantern, given the additionalcompressibility of the curved steel strips that are stacked together. Since these strips do not have the samecurvature radius as the shaft of the UBC pressuremeter, they “flatten out” with the increase of the externallateral pressure.The evaluation of the compliance strains is conventionally done with the probe inside a rigid thickwalled steel tube. The probe is placed vertically and expansion is carried out at the same inflation rate asused in the field. Given the negligible displacement of the inner wall of the rigid tube (Fahey and260Jewell, 1990 measured circumferential strains in the order of 0.005 %), the measured strain is directlyrelated to the compliance of the system (lantern plus rubber membrane). In this thesis the compliancecalibration tests were carried out with a “split cylinder”. This cylinder was composed of two thick steelhalves, that could be screwed around the expandable section of the probe.The compliance test results are presented in Figure B.2, where the following observations apply:1. The compliance of the steel lanterns is predominantly caused by the lantern compression rather thanthe inner rubber membrane compression, or the volume expansion of the air tubing. This is noticed whencomparing the compliance of the probes with and without lantern (only with the rubber membrane). Asimilar observation was given before in relation to the resistance to expansion in air.2. The measured compliance curves of all the steel lanterns are non linear. These curves are alsopressure dependent, as the “flattening out” of the steel strips will vary in accordance to the outside lateral(reaction) pressure applied by the split cylinder. It can be noticed that lanterns 8 and 9 presented muchhigher compliance strains than the remaining of the tested lanterns. The former lantern is composed of 2halves that induce a considerably amount of strain as they move independently, whereas the latter lantern iscomposed by 3 rows of steel strips stacked together. This design allows a high straining of the strips. Thecompliance strains measured for lantern 3 are lower than the compliance strains of lantern 9 at similarinternal pressures. Although both lanterns had similar designs, lantern 9 was built with steel strips oflower width (hence lower curvature radius) than the width of the strip adopted in lantern 3. This suggeststhat the lower is the width (or the curvature radius) of the steel strip, the higher is the “flattening out”effect.3. The Canikometer lantern yielded a linear compliance curve. The measured stress-strain response ischaracteristic of the linear elastic straining of the outer rubber membrane, since this lantern is not designedwith stacked steel strips. The steel strips of this lantern are bounded to the outer rubber membrane andhave the same curvature radius as the probe, which suggests that these strips did not “flatten out”.4. The compliance curve of lantern 7 varied considerably in the initial 0 to 600 kPa of internalpressure of the test. It appears that this variation was caused by the readjustment of the rubber membraneand the steel strips inside the split cylinder, rather than a direct straining of the individual strips.2612000ieoo016000%_‘1 4000ca)C12001000800CALIBRATION RESULTS—SBPM--COMPARISON OF LANTERNSExpansion in Split Cylinder—Compliance Evaluationa)L.DCl)U)L)60040020000CircumferentialFigure B.2: Lantern Compliance from Expansion Inside Split Cylinder262The compliance calibration tests demonstrated that high corrective factors have to be applied in theraw cavity strain when the steel lanterns are used. This is principally the case of lanterns 8 and 9, wherethe compliance strains were as high as 1.0 %. The Camkometer lantern and the steel lanterns 3 and 7yielded compliance strains with similar magnitude (at the same level of pressure). However, thecompliance curve of the steel lanterns presented a non linear shape.The compliance curves of Figure B.2 were also fitted by polynomial equations, in order to providesmooth correction curves for the data reduction macro (one curve per lantern type).The data reduction macro was written for usage with any spreadsheet generated by the programQuattro Pro (version 1.0). This macro is composed of 2 submacros, that serve to correct the raw datafor temperature and zero drift (first submacro), and membrane resistance and lantern compliance (secondsubmacro). The submacros are linked to each other, in the sense that once the data is corrected by the firstone it can be directly used into the second submacro to furnish the corrected values of strain and stress.The process of correction of the raw data is very simple. Once the raw pressuremeter data (“testingfile”) is loaded into the spreadsheet the first submacro is run. The circumferential strains of each of thestrain arms are corrected for the zero drift, and both the total internal and pore pressures are corrected withbasis on the baseline temperature. The second submacro is then run next. The total internal pressure issubtracted from the (post lift off) membrane resistance curve. The resulting pressure is further subtractedfrom the pore pressure measured during the test, in order to obtain the effective internal pressure. Theaverage circumferential strain, calculated with the strains of all the six anus, is subtracted from thecompliance strain. This furnishes the soil response at the cavity wall without compliance effects.The final corrected ASCII file contains 13 columns. The time (mm), the circumferential strain of eachof the six strain arms (%), the pore pressure (kPa), the effective internal pressure (kPa), the total internalpressure (kPa), the average circumferential strain (%), the strain rate (%/min) and the rate of inflation(kPals). The interpretation (via curve fitting technique) of the SBPM data was accomplished with theaverage curve of circumferential strain.B.4.1 Compliance Correction of Unload Reload LoopsThe correction to account for compliance strains has serious implications for the measured Gurmodulus, and shall be not disregarded (Houlsby and Schnaid, 1992). The methodology for compliance263correction of the loops followed the recommendations presented by Fahey and Jewell, 1990. These authorssuggested calibration tests inside the split cylinder with loops of the same rate of inflation and degree ofunloading as those in the field.The compliance correction of the unload reload loops of this thesis was obtained with the testingexpansion inside the split cylinder. These expansions were accomplished with each of the lanterns. Unloadreload loops were adopted during this calibration stage. The loops were carried out sequentially with theincrease of cavity pressure, as illustrated on the top plot of Figure B.3. The apparent strains developed ineach of these loops are a measure of the system compliance. The compliance of the system (rubbermembrane plus lantern) is expressed as an equivalent “system shear modulus” (Gsys), defined as half thegradient of the curve ofpressure versus the compliance strain.The corrected Our modulus is the difference of the measured modulus to the system modulus, inaccordance to the following expression:1 1 — 1 (B.1)Gur Gmeasured GsysDifferent lanterns yield different amounts of compliance strain during the loop stage, hence requiredistinct compliance corrections. The compliance correction is dependent on the internal pressure at thecommencement of the unload reload loop (P), and will be related to a particular degree of pressure unload.The bottom plot of Figure B.3 presents a typical relationship between Gsys and for the Camkometerlantern. This relationship is expressed by a power equation. With this power equation it is possible toevaluate the Gsys for any field loop in which the P is known and the degree of pressure unload is 40 %.With the value of Gsys and Gmeasured it is possible to obtain the corrected value of Our withEquation B. 1.For the other lanterns the dependence of Gsys on P was also observed. In general, the higher was theeffective pressure at the commencement of the loop the higher was the system shear modulus. Thisindicates that the compliance strains developed during the unload reload loop decrease (Gsys increases)with an increase of the outside pressure. The decrease of compliance strain with increase in pressure is0C0)CTest inside the Split CylinderLantern 10— dP/P = 40 — 1 Rubber Membrane2641200—1 00000-800CD:5U)C/)C)00C1)-JC60040020061200,—1000C0-0):5U)(1)0)ELOsys (MPa)Figure B.3: Compliance Correction of Unload Reload Loops: (a) Compliance Test, (b) Determination ofGsys with Basis on Power Equation265an expected characteristic of the steel (and Camkometer) lanterns. As the outside pressure increases themembrane and the protective steel strips of the lanterns become more deformed, and hence less prone todeflect under further pressure increases.The magnitude of the compliance correction has a fundamental influence on the shear modulus. Allthe Gur’s presented in this thesis were corrected in accordance with the above discussed procedure.B.5 SUMMARYIn this Appendix the general calibration characteristics of the UBC SBPM sensors were presented,together with the methodology of calibration adopted for this probe. Based on the previous discussion twomain points are highlighted:• Calibration is essential to remove any equipment related effect from the pressuremeter results, aswell as to furnish the data acquisition system with proportionality constants. High accuracy and resolutionof the sensors are extremely important for a precise delineation of the testing curve, principally at the stageswhere a high density of data points and shorter strain intervals are required.• Membrane resistance and compliance strains are expected to occur due to the characteristics of therubber membrane and the lanterns used. Corrective curves based on the fitting of a polynomial equationover the experimental calibration results are adopted to remove both effects from the raw testing data.Shear modulus from the unload reload loops also have to be corrected for compliance strains.266APPENDIX C RESULTS OF THE TRIAXTAL TESTSC.1 INTRODUCTIONIsotropically consolidated triaxial tests were performed with reconstituted and ‘undisturbed’ granularsamples retrieved from 4 to 14 m in the Laing Bridge site. The triaxial testing programme was carried outto characterize the stress strain behavior of this sand, as well as to furnish the peak and constant volumefriction angles.In this thesis simple consolidated drained triaxial tests were adopted. This is due to two basic reasons:It is a fast and well established laboratory testing technique and it could reasonably well accommodate thecylindrical samples of the ST1 sampler. The reconstituted samples were formed with the remolded sandcollected with this sampler.A standard strain controlled Wyhekam Farrance triaxial frame was adopted in the tnaxial tests. Thebasic features of the triaxial testing apparatus is presented in Figure C. 1. Volume changes during theshearing and consolidation stages were monitored by a burette connected to the base of the sample.Internal back pressure and external confining pressure were monitored with the aid of a pore pressuretransducer attached to the connections of the system. This transducer also enabled pore pressuremeasurements during the set up of the sample, saturation and shearing stages. Strains were recorded by adial gauge located inthe base of the moving frame. The precision of this gauge is 0.01 mm.During the shearing stage records of axial load, pore pressure, volume change and axial displacementwere taken at discrete time intervals. The reconstituted samples with dimensions of 35 mm x 70 mm(diameter x height) required manual readings each 20 s (up to 5 mm.), 30s (up to 10 mm.) and in increasingtime intervals thereafter, in order to ensure a high resolution in the initial stages of shear. In this case aresolution of 0.1 % was obtained with a shearing rate of 0.2286 mm/mm. A similar rate of readingintervals was adopted for the tests with the “undisturbed” samples. These samples had dimensions of 50mm x 100 mm and were sheared at a rate of 0.0762 mm/mm. The reduced shearing rate enabled a higherresolution of 0.03 % for these latter tests. On the other hand it increased the testing time to 4 hours.A “soft” ram contact with the top cap was used in the tests with the reconstituted samples. This267PPI fl/”flN irL U U I N UTRIAXIAL FRAME0 VALVESREGULATORSAIR SUPPLY & PRESSUREwjEwESAMPLEPOROUS STONEGAGEWYKEHAM FARRANCEFigure C. 1: Schematic of Triaxial Testing Apparatus268characteristic of the equipment has proven to accentuate the bedding errors at the initial stages of shearing.Therefore, for the subsequent tests with the “undisturbed samples a fixed top cap and ram connection wasadopted.C.2 TESTS WiTh THE RECONSTITUTED SAMPLESThe tests with the reconstituted samples were done to provide the writer with the typical stress strainbehavior of the Laing Bridge sand, sheared under known density and stress regimes. It also served toprovide peak friction angles under distinct sample conditions, that were used together with the anglesmeasured by the “undisturbed” tests to define the constant volume friction angle of this sand.The reconstituted samples were formed by using the water pluviation technique (see Negussey, 1984).This technique assures a good homogeneity of the testing specimen. With the careful control of the sampleheight, target relative densities were obtained. The densities chosen were 12, 24 and 45 %. Samples ateach of these densities were tested under confining pressures close to those that prevail in situ, namely 50,100 and 150 kPa. Specific details of the steps adopted in these triaxial tests can be found in the notes ofthe graduate course CVL 574 of the UBC Civil Engineering Department.Typical stress strain results of the tests with the reconstituted samples are shown in Figures C.2 andC.3. The following observations apply:1. The dilation rate, expressed by the inverse sine of the slope of the volume expansion curve (Hansen,1958), is greatly influenced by the density of the sample and to a lesser extent by the confining pressure. Itdecreases with a decrease in density or with an increase in confining pressure.2. The initial contractiveness of this sand, at all chosen densities, is followed by a dilation stage. Thisinitial contractiveness increases with the increase in confining pressure or decrease in relative density.3. Strain softening is apparent in most of the stress strain curves. After the peak friction angle,defined either in terms of maximum deviator stress or maximum stress ratio, there is a slightly tendency fora decrease in the dilation rate.Based on the experimental shear behavior of Sacramento River and Ottawa sand described by Lee andSeed, 1967, the results above conform to the expected shearing behavior of loose to dense sands under low269bb‘—‘40I,CU)(I)-UDCID Triaxial Tests on Reconstituted SamplesST1 Sampler Specimens from 5/10 m—L.Bridge32Axial Strain (%)1-5 040C20C)E- -0.00>+1.0Axial Strain (%)Figure C.2: Typical Stress Strain Response of the Reconstituted Samples: Constant Confming Pressure0Cl)()I.C)ED0>bb00cC/)C!)C)ICr)CD Triaxial Tests on Reconstituted SamplesST1 Sampler Specimens from 5/10 m—L.Bridge27043216 10Axial Strain (%)-4.5-3.5—2.5-0.66Axial10Strain2(%)Figure C.3: Typical Stress Strain Response of the Reconstituted Samples: Constant Relative Density271to medium confining pressures. According to this author the drained shearing resistance of sands appearsto be governed by three components: sliding friction, dilatancy and particle rearranging. At low pressures,dense sands dilate and exhibit a brittle type stress strain curve. Dilatancy is the major factor responsiblefor the significant increase in the friction angle at this level of stress. At medium pressures, the same densesand becomes less dilative and tends to exhibit a more plastic stress strain relationship.The granular deposit of the Laing Bridge site is characterized by loose to medium dense sand atrelatively low confining pressures (vertical effective pressures varying from 50 to 150 kPa in the testingrange of this thesis). Given the aforementioned results and observations it may be expected that this samesand, in situ, will shear under highly dilatant characteristics. A small initial contractive stage is expectedas well as a stress strain curve with some strain softening. This is verified next with the tests in which“undisturbed” samples were employed.C.3 TESTS WITH THE “UNDISTURDED” SAMPLESTests with the “undisturbed” samples were carried out to determine the peak friction angles of theLaing Bridge sand. The “undisturbed” samples are those in which a high quality of sampling was achievedin the field, and a good quality of trimming and assemblage in the triaxial cell was accomplished in the lab.It shall be noticed, however, that the retrieval percentage of the ST1 sampler was, in average, 60 %.Therefore, it may be expected that some disturbance is present in the “undisturbed” samples.Several steps were followed from the retrieval of the “undisturbed” samples to the measurement of thepeak ffiction angles. These steps are:1. Storage and preparation of the sample: After the field retrieval the samples were stored in thefreezer of the UBC Soil Mechanics laboratory. After 24 hours the samples were trimmed to the requireddimensions by using the diamond saw machine. The top and bottom parts were smoothed with a sandpaper file to allow parallelism, and hence to reduce the bedding errors. The samples were then refrozeninside the ST1 samplers and kept in the freezer until the testing day.2. Placement of the sample on the triaxial cell: On the testing day a sample is removed from thefreezer. It is left to thaw for 40 mm. until it can be extruded from the ST1 sampler without destruction ofthe edges. The sample is placed in the triaxial cell and is covered by a 0.03 mm thick rubber membrane.272This membrane is fixed to both pedestal and top cap by rubber bands. At this stage a vacuum of 100 mmofHg is introduced on the drainage line, enabling the final assemblage of the triaxial cell.3. Set up of the triaxial cell and thawing of the sample: The triaxial cell is set up in the triaxial frame.A small seating load of 50 kPa is applied to the sample through the quick connection at the top part of thecell. The drainage line is released ofvacuum and connected to the general deaired line of the triaxial frame.The sample is left to thaw for 4 hours with the drainage line closed.4. Saturation stage: The saturation stage is initiated after the sample is fully thawed. Saturation isachieved by applying back pressure and cell pressure so that a net confining pressure of 10 kPa can beimposed in the sample. The saturation stage took in general 12 hours, with both drainage and cell pressurelines open. After 12 hours the drainage line is closed and the saturation level is assessed by checking theratio AulAcell (known as B parameter), where Au is the increase of the excess pore pressure in the samplegenerated by the change of external pressure Acell. Typical values of B obtained for the “undisturbed”samples were above 0.95, for cell pressures greater than 210 kPa.5. Consolidation stage: The consolidation stage is started after the sample is saturated. The initialburette reading is taken and the cell pressure is increased to the required consolidation pressure. Thedrainage line is opened and the sample is left to consolidate for 15 to 30 mm. At this stage the sample isready to be sheared with the in situ isotropic stress. An isotropic consolidation pressure equivalent to theeffective lateral stress in the site was adopted. For the purpose of these tests, the field lateral stress wasbased on a K0 of 0.65, obtained by Robertson, 1982 with SBPM tests in a nearby site.6. Shearing stage: Once the sample is consolidated the initial values of pore pressure, externalpressure, axial force, burette volume and gage displacement are taken. The chronometer is started and thetest is run at the chosen shearing rate. Readings of each of the above variables are taken manually.For the present series of triaxial tests the ram friction, uplift force, weight of the rod and thickness ofthe membrane were considered in the data reduction. Details of the procedures for data reduction are alsofound in the notes of the course CVL 574 ofthe UBC Civil Engineering department.Typical results of the “undisturbed” tests are presented in Figures C.4 and C.5. It is noticed that thestress strain behavior, in terms of deviator stress and volumetric strain versus axial strain, follows the00Cl)C!)Cr)CID Iriaxial Tests on Undisturbed SamplesST1 Sampler Specimen from 11 m — L.Bridge300273200-4GBB-2b—0B0LIClC)U.)E0>-1.B6 10 12 14Axia Strain (%)Figure C.4: Typical Stress Strain Response in the Triaxial Test: “Undisturbed” Samplesbb0CCl)Cl)I,U’)3000200blOObCCID Triaxial Tests on Undisturbed SamplesST1 Sampler Specimens from 11 m — L.Bridge2027443216 10 2Axial Strain (%)Figure C.5: Determination of the Peak Friction Angle from the Triaxial Tests: “Undisturbed” Samples275expected trend devised from the results of the reconstituted samples. A small contractive stage followed bya highly dilatant stage is observed in the curve of volumetric versus axial strain. This tendency for dilationwas observed in all the tests, irrespective of the initial density of the samples. The curve of deviator stressversus axial strain presents a shape close to the curve of stress ratio versus axial strain. Indeed, in all thetests the peak deviator stress occurred at the same axial strain as the peak stress ratio. This happens due tothe fully drained characteristics of these tests. A small tendency to strain softening is observed in the lattershearing stages, which indicates the continuous decrease of the friction angle (or dilation rate) with theshearing of the sample.The peak friction angles were defined by using the maximum stress ratio failure criteria. For each testit was assumed that the Mohr Coulomb failure line passed through the origin of the Q x P’ stress diagramof Lambe and Whitman, 1979 (where P = (a’1+o3)/2 and Q = (a1-3)12). This hypothesis neglects anycohesion that may exist in the granular minerals of this sand. Based on this hypothesis, the bottom plot ofFigure C.5 demonstrates how the peak ffiction angle was calculated for the “undisturbed” samples.Table C.1 presents the final results of the triaxial testing programme carried out with the“undisturbed” samples. Peak friction angles varying from 37.6 to 4350 were obtained. This range ofvalues is slightly above “typical” values published in literature for poorly graded sands at loose to mediumdense conditions (see for instance tables of Holtz and Kovacs, 1981 or Hunt, 1986). Perhaps other factorsnot considered in the published tables are affecting the results reported in this thesis. According to Holtzand Kovacs, 1981 the peak friction angle may also be affected by grain shape, grain size distribution,mineralogy etc., that are not universally accounted for in the published tables.Table C. 1 also provides the values of Poisson’s coefficient obtained with the triaxial tests. ThePoisson’s coefficients were computed from the volume change data of the tests, in the axial strain range of101 %. In this strain range the stress strain behavior is almost linear and the theory of elasticity may beassumed to apply. Treating the soil as an ideally elastic isotropic material, and assuming that thevolumetric strain is caused by the sum of the three major principal strains, the following equation isobtained:276TEST DEPTH ‘ D 6a/ (( I)max V6 SAMPLE DENSITY7(m) (kPa) (%) (%) (Deg)4.0 35 44 5.14 5.14 - 42.3 0.24 Mediumdense5.0 40 43 5.86 4.34 38.7 0.28 Medium dense6.0 45 54 4.11 4.43 39.1 0.29 Medium dense7.0 50 72 6.25 5.43 43.5 0.35 Dense9.0 65 20 6.53 4.14 37.6 0.22 Loose11.0 80 42 6.38 4.36 38.8 0.30 Medium dense14.0 100 50 10.09 4.42 39.1 0.29 Medium denseAverage: 0.251-Effective confining pressure required to bring the sample to an isotropic state of stress similar to the lateral stressin the field. The field lateral stress is based on a K0 of 0.65, obtained by Robertson, 1982 with SBPM tests in anearby granular site.2-Relative density calculated from the sample volume and height. Also used here the minimum and maximum voidratios and the specific gravity value obtained in the preliminary classification tests.3-Axial strain at the failure. Failure defined with the peak stress ratio mobilized during shear.4-Peak stress ratio.5-Peak (axially symmetric) friction angle.6-Poisson’s coefficient calculated with the equation v = 0.5 (1-dc./dsa). The equation was used with increments ofvolumetric and axial strain in the strain range of I x BY’ %.7-Density classification with basis on Dr and the chart of Lade and Lee, 1976Table C. 1: Results of the Triaxial Testing Programme with “Undisturbed” Samples277i( de (Cl)2 de)where: v is the Poisson’s coefficientds the increment of volumetric straindEa the respective increment of axial strain.In this equation compressive strains are positive.The average value of v obtained by the triaxial tests was 0.25. This value is used in some of thecavity expansion theories discussed in Chapter 2, during the interpretation of the SJ3PM data of the LaingBridge site.C.4 CONSTANT VOLUME FRICTION ANGLEUnder large shearing strain levels the void ratio of the sample tends to approach the critical void ratioof the material (Casagrande, 1936). At this stage continuous deformation occurs under no volume change,and the dilation rate is null. The friction angle measured at this stage is defined as the constant volumefriction angle (4). The constant volume friction angle is an intrinsical property of the sand, and isrequired by all the cavity expansion models discussed in Chapter 2.As observed by Sasitharan, 1989 and Bishop, 1971 there is a unique relationship between the peakfriction angle and the maximum rate of dilatancy for a particular sand, which is independent on the stresspath, density, stress levels