Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Towards comprehensive interpretation of the state parameter from cone penetration testing in cohesionless… Ghafghazi, Mohsen 2011

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

Item Metadata

Download

Media
24-ubc_2011_spring_ghafghazi_mohsen.pdf [ 3.57MB ]
Metadata
JSON: 24-1.0063028.json
JSON-LD: 24-1.0063028-ld.json
RDF/XML (Pretty): 24-1.0063028-rdf.xml
RDF/JSON: 24-1.0063028-rdf.json
Turtle: 24-1.0063028-turtle.txt
N-Triples: 24-1.0063028-rdf-ntriples.txt
Original Record: 24-1.0063028-source.json
Full Text
24-1.0063028-fulltext.txt
Citation
24-1.0063028.ris

Full Text

  TOWARDS COMPREHENSIVE INTERPRETATION OF THE STATE PARAMETER FROM CONE PENETRATION TESTING IN COHESIONLESS SOILS    by  MOHSEN GHAFGHAZI   B.Sc., Isfahan University of Technology, 2002 M.Sc., Sharif University of Technology, 2004    A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF   DOCTOR OF PHILOSOPHY  in  THE FACULTY OF GRADUATE STUDIES  (Civil Engineering)    THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)    April 2011 © Mohsen Ghafghazi, 2011 Abstract The Cone Penetration Test (CPT) is widely used for determining in-situ properties of soil because of its continuous data measurement and repeatability at relatively low cost. The test is even more attractive in cohesionless soils such as sands, silts and most tailings due to difficulties associated with retrieving undisturbed samples in such soils. Behaviour of soils is highly dependent on both density and stress level. The state parameter is widely accepted to represent the soil behaviour encompassing both density and stress effects. Hence, determining the in-situ state parameter from CPT is of great practical interest. The CPT was analysed using a large strain spherical cavity expansion finite element code using a critical state soil model (NorSand) capable of accounting for both elasticity and plastic compressibility. The constitutive model was calibrated through triaxial tests on reconstituted samples. The state parameter was then interpreted from CPT tip resistance, and the results were verified against an extensive database of calibration chamber tests. The efficiency of the method was further investigated by analysing two well documented case histories confirming that consistent results could be obtained from different in-situ testing methods using the proposed framework. Consequently, cumbersome and expensive testing methods can be substituted by a combination of triaxial testing and finite element analysis producing soil specific correlations. One of the difficulties in analysing the cone penetration problem is the less researched effect of high stresses developing around the cone on the behaviour of the soil. A hypothesis was put forward on the particle breakage process at the particle level and its implications for the behaviour of sands at higher stress levels were discussed. A series of triaxial tests were performed, focusing on the effects of particle breakage on the location of the critical state line. The hypothesis was used to explain the observed behaviour. Particle breakage was shown to    ii cause additional compression and a parallel downward shift in the critical state line. The magnitude of the shift was linked to the amount of breakage and it was argued that significant breakage starts after the capacity for compression due to sliding and rolling is exhausted.                        iii Preface The research topics presented in the thesis were chosen jointly by Dr. D.A. Shuttle and me. I was responsible for performing all the numerical analyses, triaxial compression tests on Fraser river sand and data reduction work associated with the research. Initial versions of all thesis chapters were drafted by me, reviewed by Dr. D.A. Shuttle (and also Dr. Roberto Olivera for Chapter 7), and based on comments received I revised the thesis contents prior to submission. Dr. Roberto Olivera also provided guidance on the laboratory testing procedures described in Appendix A. The following is the list of publications presented in this thesis: Ghafghazi M., and Shuttle D.A. 2006. Accurate Determination of the Critical State Friction Angle from Triaxial Tests. Proceedings of the 59th Canadian Geotechnical Conference, Vancouver, 278-284, (Chapter 2) . Ghafghazi M., and Shuttle D.A. 2008. Evaluation of Soil State from SBP and CPT: A Case History. Canadian Geotechnical Journal, (45)6: 824-844, (Chapter 5) . Ghafghazi M., and Shuttle D.A. 2008. Interpretation of Sand State from Cone Penetration Resistance. Géotechnique, 58(8): 623–634, (Chapter 4) . Ghafghazi M., and Shuttle D.A. 2009. Confidence and Accuracy in Determination of the Critical State Friction Angle. Soils and Foundations, 49(3): 391-395, (Chapter 3) . Ghafghazi M., and Shuttle D.A. 2010. Interpretation of the In-situ Density from Seismic CPT in Fraser River Sand. Proceedings of the 2nd International Symposium on Cone Penetration Testing, Huntington Beach, California, (Chapter 6) . Ghafghazi M., Shuttle D.A., and Olivera R.R. 2011. Particle Breakage and the Critical State of Sand. submitted, (Chapter 7) .    iv Table of Contents Abstract .........................................................................................................................................ii Preface ..........................................................................................................................................iv Table of Contents..........................................................................................................................v List of Tables............................................................................................................................. viii List of Figures ..............................................................................................................................ix List of Symbols and Abbreviations ......................................................................................... xiii Acknowledgements ....................................................................................................................xvi Dedication..................................................................................................................................xvii Chapter 1. Introduction and Background ..................................................................................1 1.1. Introduction ..................................................................................................................1 1.2. Verification and Validation .........................................................................................8 1.3. The NorSand Constitutive Law ..................................................................................9 1.4. Literature Review.......................................................................................................14 1.4.1. Cone Penetration Test........................................................................................14 1.4.2. Interpretation of the State Parameter from CPT Tip Resistance..................16 1.4.3. Calibration Chamber Testing ...........................................................................19 1.4.4. Analytical and Numerical CPT Interpretation Methods................................26 1.4.5. Particle Breakage ...............................................................................................42 1.5. Overview of the Proposed Research .........................................................................46 1.6. References ...................................................................................................................63 Chapter 2. Determination of the Critical State Friction Angle from Triaxial Tests ............78 2.1. Introduction ................................................................................................................78 2.2. Definition of the Critical State ..................................................................................79 2.3. Stress-Dilatancy Definitions ......................................................................................81 2.4. Determination of the Critical State Friction Ratio..................................................82 2.4.1. End of Test (ET) Method ...................................................................................82 2.4.2. Maximum Contraction (MC) Method ..............................................................83 2.4.3. Bishop Method (BM)..........................................................................................83 2.4.4. Stress-Dilatancy (SD) Method ...........................................................................84 2.5. Obtaining the Critical State Stress Ratio of Ticino Sand .......................................85 2.5.1. Ticino Data Processing.......................................................................................86 2.5.2. Discussion of Results for Ticino Sand...............................................................87 2.6. Obtaining the Critical State Stress Ratio of Erksak Sand......................................88 2.6.1. Erksak Data Processing .....................................................................................88 2.6.2. Discussion of Results for Erksak Sand .............................................................89 2.7. Summary .....................................................................................................................90 2.8. References ...................................................................................................................96 Chapter 3. Confidence and Accuracy in Determination of  Critical State Friction Angle...98 3.1. Introduction ................................................................................................................98 3.2. Triaxial Database .....................................................................................................100 3.3. Methodology Used for c Determination................................................................101 3.4. Determination of c from Limited Number of Triaxial Tests ..............................103 3.5. Validation against Independent Triaxial Database...............................................105 3.6. Summary and Conclusions ......................................................................................107    v 3.7. References .................................................................................................................113 Chapter 4. Interpretation of Sand State from CPT Tip Resistance .....................................114 4.1. Introduction ..............................................................................................................114 4.2. Background to CPT Interpretation in Sand ..........................................................119 4.3. Materials and Testing ..............................................................................................121 4.4. Constitutive Model ...................................................................................................122 4.5. Calibration of NorSand............................................................................................124 4.6. Spherical Cavity Expansion Analysis .....................................................................125 4.7. Numerical Formulation ...........................................................................................126 4.8. Comparison to Calibration Chamber Data ...........................................................127 4.9. Inverse form for Interpretation of CPT .................................................................128 4.10. Discussion ..............................................................................................................130 4.11. Summary and Conclusion....................................................................................133 4.12. References .............................................................................................................145 Chapter 5. Evaluation of Sand State from SBP and CPT: A Case History.........................151 5.1. Introduction ..............................................................................................................151 5.2. Tarsiut P-45 Case History .......................................................................................157 5.3. Erksak Sand ..............................................................................................................159 5.4. Laboratory Tests ......................................................................................................160 5.5. In-situ Tests...............................................................................................................161 5.5.1. CPT ....................................................................................................................161 5.5.2. SBP.....................................................................................................................162 5.5.3. Elastic Stiffness from Geophysical Tests........................................................163 5.6. Calibration Chamber ...............................................................................................163 5.7. Modelling Erksak Sand Behaviour.........................................................................164 5.7.1. NorSand.............................................................................................................164 5.7.2. Calibration to Erksak 355/3.0 and Erksak 330/0.7 Sand..............................167 5.8. Evaluation of In-situ CPT Data ..............................................................................169 5.8.1. Methodology......................................................................................................169 5.8.2. State Parameter from CPT Tests....................................................................171 5.8.3. Inversion Parameters .......................................................................................172 5.9. Evaluation of SBP Data ...........................................................................................175 5.9.1. Methodology......................................................................................................175 5.9.2. State Parameter from SBP Tests ....................................................................176 5.10. Discussion ..............................................................................................................177 5.11. Conclusion .............................................................................................................179 5.12. References .............................................................................................................198 Chapter 6. Interpretation of the Sand State from CPT in Fraser River Sand: A Case History .......................................................................................................................................205 6.1. Introduction ..............................................................................................................205 6.2. Site Investigation Program ......................................................................................207 6.3. Material and Testing ................................................................................................208 6.4. Methodology..............................................................................................................208 6.5. Constitutive Modeling ..............................................................................................211 6.6. Spherical Cavity Expansion Analysis .....................................................................212 6.7. Inverse Form for Interpretation of CPT ................................................................213    vi 6.8. Analysis and Results.................................................................................................214 6.9. Discussion and Conclusions .....................................................................................215 6.10. References .............................................................................................................227 Chapter 7. Particle Breakage and the Critical State of Sand................................................230 7.1. Introduction ..............................................................................................................230 7.2. Hypothesis .................................................................................................................233 7.3. Fraser River Sand ....................................................................................................239 7.4. Testing Program .......................................................................................................239 7.5. Results........................................................................................................................241 7.6. Discussion ..................................................................................................................245 7.6.1. Influence of Breakage on CSL ........................................................................245 7.6.2. The Onset of Breakage.....................................................................................246 7.6.3. The Influence of Breakage on Behaviour.......................................................248 7.6.4. The Influence of Breakage on CPT.................................................................250 7.7. Summary and Conclusions ......................................................................................252 7.8. References .................................................................................................................264 Chapter 8. Summary and Conclusion .....................................................................................268 8.1. Summary ...................................................................................................................268 8.2. Major Topics of Research........................................................................................270 8.2.1. Determination of the Critical State Friction Angle from Triaxial Tests - Summary of Findings .......................................................................................................272 8.2.2. Confidence and Accuracy in Determination of Critical State Friction Angle - Summary of Findings .......................................................................................................273 8.2.3. Interpretation of Sand State from CPT Tip Resistance - Summary of Findings ............................................................................................................................274 8.2.4. Evaluation of Sand State from SBP and CPT: A Case History - Summary of Findings ............................................................................................................................276 8.2.5. Interpretation of the Sand State from CPT in Fraser River Sand: A Case History - Summary of Findings.......................................................................................278 8.2.6. Particle Breakage and the Critical State of Sand - Summary of Findings .281 8.3. Contributions ............................................................................................................282 8.4. Limitations ................................................................................................................283 8.5. Future Studies...........................................................................................................285 8.6. References .................................................................................................................287 Appendix A. Triaxial Tests on Fraser River Sand Procedures and Results.......................289 A.1. Testing Equipment ...................................................................................................290 A.2. Test Procedures ........................................................................................................291 A.3. Repeatability of Test Results ...................................................................................293 A.4. Test Results ...............................................................................................................294 A.5. References .................................................................................................................308     vii List of Tables Table 1- 1 Summary of NorSand (Jefferies and Shuttle, 2005) .......................................53 Table 1- 2 NorSand soil parameters and values for sands (Jefferies and Shuttle, 2005)54 Table 1- 3 Current calibration chambers in the world, expanded after Ghionna and Jamiolkowski (1991) ........................................................................................................55 Table 1- 4 Boundary conditions in conventional calibration chamber and simulator tests (Huang and Hsu, 2005)....................................................................................................56 Table 1- 5 Summary of B values (Equation 1-13) proposed for different sands by Cudmani and Osinov (2001) ............................................................................................56  Table 2 - 1 Mtc and N* parameters (from the Stress-Dilatancy method) for nine wet pluviated triaxial tests on dense Erksak 330/0.7..............................................................91  Table 3 - 1 Confidence level, | c|, versus number of triaxial tests .............................109 Table 3 - 2 Confidence levels for 6 tests from the independent Been et al. (1991) dataset ........................................................................................................................................109 Table 3 - 3 Summary of initial conditions for Been et al. (1991) triaxial tests..............109  Table 4 - 1 Index properties of studied sands and tailings ............................................135 Table 4 - 2 Summary of drained triaxial compression tests used in calibration of NorSand .........................................................................................................................136 Table 4 - 3 Summary of calibration chamber tests on normally consolidated soils......136 Table 4 - 4 NorSand parameters for nine CPT calibration chamber sands ..................137 Table 4 - 5 Equations for ksph and msph as functions of G/p′0 for database soils ...........138 Table 4 - 6 Summary of 0 obtained for nine database soils ......................................138  Table 5 - 1 Index properties of Erksak 330/0.7 and Erksak 355/3.0 sands ...................181 Table 5 - 2 Summary of triaxial tests on Erksak 330/0.7 and Erksak 355/3.0 sands.....181 Table 5 - 3 Drained triaxial tests on moist tamped Erksak 355/3.0 sand......................181 Table 5 - 4 Summary of CPT calibration chamber tests (after Been et al., 1987b) .......182 Table 5 - 5 NorSand calibration to Erksak sand ...........................................................182 Table 5 - 6 Summary of data used in estimation of in-situ state from CPT and SBP tests ........................................................................................................................................183  Table 6 - 1 NorSand calibration parameters for Fraser river sand ..............................220 Table 6 - 2 0 interpretation summary...........................................................................220  Table 7 - 1 Summary of testing program at lower stress level ......................................254 Table 7 - 2 Summary of testing program at higher stress level and tests performed on pre-sheared samples ......................................................................................................255    viii List of Figures Figure 1-1 Definition of state parameter   and overconsolidation ratio R (after Jefferies and Shuttle, 2005)..............................................................................................57 Figure 1-2 Illustration of NorSand yield surfaces and limiting stress ratios (Jefferies and Shuttle, 2005) ...................................................................................................................58 Figure 1-3 The standard electrical cone penetrometer (ASTM D5778)..........................59 Figure 1-4 Contours of state parameter (Robertson, 2009) ............................................60 Figure 1-5 Normalisation factors for calibration chamber size and boundary conditions (Been et al., 1987) ............................................................................................................60 Figure 1-6 Schematic view of the Eulerean approach to the cone penetration problem (van den Berg et al., 1996)...............................................................................................61 Figure 1-7 Steady-State behaviour and boundary conditions (Yu et al., 2000) ..............62  Figure 2 - 1 Stress ratio and volumetric strain versus deviatoric strain for Ticino sand (test 09-CID-D169) ..........................................................................................................92 Figure 2 - 2 Application of Bishop's method to Ticino sand ...........................................92 Figure 2 - 3 Stress-dilatancy plot ( D ), around peak ................................................93 Figure 2 - 4 Stress-dilatancy plot ( D ) for Ticino sand (test 09-CID-D169) ............93 Figure 2 - 5 Stress-dilatancy plot ( D ) for Ticino sand (test 09-CID-D169), comparison of methods ....................................................................................................94 Figure 2 - 6 Stress-dilatancy plot ( D ) for Ticino sand .............................................94 Figure 2 - 7 Stress-dilatancy ( maxmin D ) plot from triaxial data on Erksak 330/0.7 sand ..................................................................................................................................95 Figure 2 - 8 Stress-dilatancy plot ( D ) for Erksak 330/0.7 sand (test CID-G667)....95 Figure 2 - 9 Stress-dilatancy ( maxmin D ) plot from triaxial data on Erksak 330/0.7 sand  (data after Vaid and Sasitharan, 1992)..................................................................95  Figure 3 - 1 Stress-Dilatancy plot for test D-667 (Been et al., 1991) on Erksak sand..110 Figure 3 - 2max vs. Dmin of 34 triaxial compression tests on Erksak sand (after Vaid and Sasitharan, 1992) ...........................................................................................................110 Figure 3 - 3 Frequency of error in c for 5, 10, 15 and 24 tests....................................111 Figure 3 - 4 Error in c vs. number of triaxial tests at different confidence levels .......111 Figure 3 - 5 Error in calculation of c obtained from 9 tests with different combinations of loose, medium and dense tests ...................................................................................112 Figure 3 - 6 max vs. Dmin of 9 triaxial compression tests on Erksak sand (after Been et al., 1991) ........................................................................................................................112  Figure 4 - 1 Stress-dilatancy plot ( - D) to obtain Mtc and N* (Ticino 4, C264) .........139 Figure 4 - 2 CSL determination for Ticino 4 .................................................................139 Figure 4 - 3 Comparison of CSL for all nine sands.......................................................140    ix Figure 4 - 4 NorSand fits to loose and dense Ticino 4 triaxial data (C262, C264).......140 Figure 4 - 5 Behaviour of an element close to the cavity during spherical expansion..141 Figure 4 - 6 Qcc and Qsph vs. 0 for Ticino 4 sand.........................................................142 Figure 4 - 7 Qcc vs. Qsph for Ticino 4 sand.....................................................................142 Figure 4 - 8 Qcc vs. Qsph for all nine database sands.....................................................143 Figure 4 - 9 Effect of G/p′0 on ksph and msph for Ticino 4 sand ......................................143 Figure 4 - 100 from Equation 4-5 vs. 0 measured in the chamber............................144  Figure 5 - 1 a) Aerial photo of Molikpaq; b) Schematic cross section of Molikpaq at Tarsiut P-45 (Jefferies and Been, 2006) ........................................................................184 Figure 5 - 2 Tarsiut location showing proximity of Erksak borrow (after Jefferies et al., 1985) ..............................................................................................................................184 Figure 5 - 3 Photograph of washed Erksak 355 sand particles (Been et al., 1987b).....185 Figure 5 - 4 Particle size distribution of Erksak sand in hopper of dredge prior to placement (after Goldby et al., 1986) ............................................................................185 Figure 5 - 5 Section through the core showing the 1985 in-situ testing program and the results of gradation tests (Golder Associates Ltd., 1986)..............................................186 Figure 5 - 6 Grain size distribution of different Erksak sand gradations: Erksak 355/3.0 and 330/0.7 (after Been et al., 1987b, 1991) ..................................................................187 Figure 5 - 7 CSL locations for Erksak 330/0.7 and Erksak 355/3.0..............................188 Figure 5 - 8 Peak drained triaxial compression strength of dense Erksak sand samples ........................................................................................................................................188 Figure 5 - 9 a) Cone tip resistance and SBP depths; b) Friction ratio; c) Pore water pressure u2; d) material index Ic after Been and Jefferies (1992); e) Horizontal effective stress from SBP ..............................................................................................................189 Figure 5 - 10 Shear modulus measurements from SBP unload-reload loops and geophysical testing (data from Golder Associates Ltd., 1986) ......................................190 Figure 5 - 11 Initial state plot for CPT calibration tests in Erksak 355/3.0 sand .........190 Figure 5 - 12 Stress-dilatancy plot ( - D) for Erksak 330/0.7 sand (test CID_G666).191 Figure 5 - 13 Plot of dilatancy at peak, Dmin, versus the state parameter at the image condition i at peak to determine the value of tc..........................................................192 Figure 5 - 14 Fitting NorSand to triaxial tests on Erksak 355/3.0, Moist tamped ........192 Figure 5 - 15 Fitting NorSand to triaxial tests on Erksak 330/0.7 ................................193 Figure 5 - 16 CPT in calibration chamber versus spherical cavity expansion: Erksak sand ................................................................................................................................194 Figure 5 - 17 Computed effect of soil rigidity Ir on CPT calibration coefficients in hydraulically placed Erksak 355/3.0 sand.....................................................................194 Figure 5 - 18 Procedure followed to obtain the normalised cone tip resistance Q from CPT data at depth of the adjacent SBP..........................................................................195 Figure 5 - 19 Procedure followed to obtain the initial state parameter 0 from SBP data ........................................................................................................................................195 Figure 5 - 20 IFM calibration of SBP to determine range of 0 and K0 (SBP corrected for finite geometry effects) .............................................................................................196 Figure 5 - 21 Comparison of back calculated from CPT and SBP (numbers on the figure indicate test depth) ..............................................................................................197     x Figure 6 - 1 a) Massey site in the Fraser river delta; b) Layout of the Large Diameter Laval Sampler (LDS), the frozen core, and the CPTs (M9401 to M9406) relevant to the work at Massey site (after Wride and Robertson, 1997)................................................221 Figure 6 - 2 Gradation curves of Fraser river sand: UBC and Massey samples..........221 Figure 6 - 3 Flowchart for Ghafghazi and Shuttle (2008) method................................222 Figure 6 - 4 Critical State Loci for UBC and Massey samples; end points of drained and undrained tests are plotted with different signs assigned to tests that were contracting, dilating or had not volume change at the end of the test ...............................................222 Figure 6 - 5 Variation of the hardening parameter H with the initial state parameter for UBC and Massey samples..............................................................................................223 Figure 6 - 6 a) Qsph vs.  0 for range of 0pGI r  ; b) msph and ksph vs. normalised shear modulus .....................................................................................................................223 rI Figure 6 - 7 CPT data M9401 to M9406 a) Tip resistance b) Sleeve friction c) Pore pressure d) Friction ratio...............................................................................................224 Figure 6 - 8 Upper and lower bound CPT response and state parameter interpretation for the target zone: a) Tip resistance b) Normalised tip resistance c) State parameter Interpretation (Ghafghazi and Shuttle, 2008) with ±0.04 and ±0.07 error margins d) Alternative methods of interpretation: Konrad (1997), Been et al. (1987), and Plewes et al. (1992)........................................................................................................................224 Figure 6 - 9 Shear modulus measurements from seismic CPTs, the average values and the target depth range ....................................................................................................225 Figure 6 - 10 Comparison of Critical State Loci for UBC, Massey and U of A samples and the direct void ratio measurements.........................................................................226  Figure 7 - 1 Full stress range CSL in e  log p' space (after Russell and Khalili, 2004) and schematic undrained triaxial test ............................................................................256 Figure 7 - 2 Microscopic picture of FRS grains............................................................256 Figure 7 - 3 Void ratio versus p for virgin FRS tests summarised in Table 7-1 (for figure clarity only start and end points are plotted for cyclic tests).........................................257 Figure 7 - 4 Gradation curve for samples of FRS before and after shearing................258 Figure 7 - 5 e  log p' plot for all tests on sieve sample #1 in Table 7-2 (obtained from parent sample CIU-D 1300 kPa) ...................................................................................258 Figure 7 - 6 e  log p' plot for test CID-L 1600 kPa (sieve sample #2) ........................259 Figure 7 - 7 e  log p' plot for all tests on sieve sample #3 in Table 7-2 (obtained from CID-L 1400 kPa)............................................................................................................259 Figure 7 - 8 e  log p' plot for all tests on sieve sample #5 in Table 7-2 (obtained from CID-L 1400 kPa-2) ........................................................................................................260 Figure 7 - 9 e  log p' plot for test CID-D 600 kPa (sieve sample #4)..........................260 Figure 7 - 10 e  log p' plot for all tests on sieve sample #6 in Table 7-2 (obtained from CID-D 600 kPa-peak) ....................................................................................................261 Figure 7 - 11 e  log p' plot for all tests on sieve sample #7 in Table 7-2 (obtained from CID-D 1000 kPa-peak) ..................................................................................................261 Figure 7 - 12 e  log p' plot for all tests on sieve sample #8 in Table 7-2 (obtained from CID-D 1000 kPa) ...........................................................................................................262    xi Figure 7 - 13 Variation in  vs. fines content after shearing of FRS under drained and undrained conditions (Sieve test numbers are shown beside data points) ....................262 Figure 7 - 14  vs. the change in fines content after shearing at high pressures: comparison between FRS and Kurnell sand (Russell and Khalili, 2004)......................263  Figure A - 1 Triaxial test loading frame, pressure pumps and cell set-up ....................295 Figure A - 2 Split mould device .....................................................................................296 Figure A - 3 Triaxial base with split mould and membrane ..........................................297 Figure A - 4 Adding water for sample preparation .......................................................298 Figure A - 5 Sample preparation in layers ....................................................................299 Figure A - 6 Sample compaction....................................................................................300 Figure A - 7 Prepared sample before cell assembly......................................................301 Figure A - 8 Deviator stress and volumetric strain versus axial strain for three pairs of tests starting from identical density and stress states ....................................................302 Figure A - 9 Void ratio versus mean effective stress for three pairs of tests starting from identical density and stress states ..................................................................................303 Figure A - 10 Deviator stress, pore pressure and volumetric strain plotted against the axial strain for tests aimed at determining the CSL for virgin FRS, summarised in Table 7-1 ..................................................................................................................................304 Figure A - 11 Deviator stress, pore pressure and volumetric strain plotted against the axial strain for tests aimed at determining the effects of particle breakage on FRS, summarised in Table 7-2................................................................................................307             xii List of Symbols and Abbreviations a [-] Net area ratio cA  [L2] Projected cone area sA  [L2] Friction sleeve area B [-] Skempton’s pore pressure parameter, 3  uB Bq [-] CPTu excess pore pressure ratio,  0 02 vt q q uuB   CPT [-] Cone penetration test CSL [-] Critical state locus D  [-] Total dilatancy, qv   pD  [-] Plastic dilatancy, pq p v   rD  [-] Relative density, minmax max ee eeDr   e [-] Void ratio ec [-] Void ratio at the critical state for the current mean stress mine  [-] Minimum void ratio, ASTM D4254 maxe  [-] Maximum void ratio, ASTM D4253 Fr [-] Normalised friction ratio, 100 0  vt s r q fF  fs [FL-2] Friction sleeve stress measurement sF  [F] Sleeve friction resistance force G [FL-2] Elastic shear modulus Gs [-] Specific gravity H [-] Dimensionless plastic hardening modulus for loading Ic [-] Soil classification index from Been and Jefferies (1992)        22 log3.15.111log3 rqc FBQI  Ir [-] Dimensionless elastic shear rigidity, pG  k [-] Soil-specific coefficient in equation [1-4] K0 [-] In-situ lateral effective stress ratio, vh ''  m [-] Soil-specific coefficient in equation [1-4] M [-] Value of ratio  at the critical state (varies with lode angle) Mi [-] Absolute value of (M-) used in the flow rule Mtc [-] Value of ratio  at the critical state under triaxial compression conditions n  [-] Stress exponent *, NN  [-] Volumetric coupling parameter p  [FL-2] Mean total stress ,   3/321  p    xiii p  [FL-2] Mean effective stress ,   3/'''' 321  p 0p  [FL -2] In-situ mean total stress 0p  [FL-2] In-situ mean effective stress ,   3/'''' 0000 zyxp   ap  [FL -2] Atmospheric pressure ip  [FL-2] Mean effective stress at the image state q  [FL-2] Deviatoric stress invariant, 2 1))()()(( 21321 2 322 12 212 1  q Q  [-] Dimensionless CPT resistance based on mean stress required by mechanics, 00 '/)( ppqQ t  qc [FL-2] Tip resistance cQ  [F] Cone tip resistance force qt [FL-2] Corrected tip resistance,   21 uaqq ct  tnQ  [-] Dimensionless CPT resistance based on vertical stress required by mechanics,   nvaavttn ppqQ 00 /)(   r  [-] Chamber size correction factor R  [-] Overconsolidation ratio, currentppR ''max SBP [-] Self-bored pressuremeter test SD [-] Standard deviation us  [FL-2] Undrained shear strength u [FL-2] Excess pore pressure u0 [FL-2] In-situ pore pressure u2 [FL-2] Pore pressure measured by CPT during sounding (at shoulder location) tc  [-] Proportionality coefficient, relating minimum dilation to state be  [-] Reduction in void ratio caused by particle breakage sre  [-] Change in void ratio caused by sliding and rolling of particles  e [-] Strain v  [-] Volumetric strain, , e superscript indicates elastic eeeev 321   p v  [-] Volumetric strain, , p superscript indicates plastic  ppppv 321   q  [-] Shear strain measure work conjugate with q , dot superscript denotes rate,  321 )cos(sinsin2)cos(sin 3331    q   [-] Effective friction angle c  [-] Constant volume (critical state) friction angle  [-] Dimensionless shear measure as ratio of stress invariants, '/ pq L  [-] Limiting stress ratio  [-] ‘Altitude’ of CSL, defined at 1 kPa 10  [-] Slope of CSL, defined on base 10 e,  [-] Slope of CSL on natural logarithmic scale    xiv    d] [-] Poisson’s ratio 3[Ra Lode angle, sin( 321 /'''5.13)3 q    [FL-2] Total stress    [FL-2] Effective stress 0h   [FL-2] In-situ effective horizontal stress 0v  [FL-2] In-situ total vertical stress 0v   [FL-2] In-situ effective vertical stress   [-] State parameter, cee  0  [-] Initial state parameter    xv Acknowledgements The research work presented in this thesis was funded by the Natural Sciences and Engineering Research Council (NSERC) of Canada, MITACS and Golder Associates; their support is greatly appreciated. I wish to thank my supervisor Dr. Dawn Shuttle for her great encouragement, support and guidance and Dr. Michael Jefferies for his valuable inputs and comments. I also wish to thank Dr. Nemkumar Banthia for his support and guidance.    xvi    xvii Dedication To my parents and my wife whose love and support made this thesis a wonderful experience. Chapter 1.    Introduction and Background    1.1. Introduction  In-situ testing techniques are widely used to characterise the engineering properties of soils in engineering practice. The Cone Penetration Test (CPT) has been widely used for this purpose because of its continuous data measurement and repeatability at relatively low cost. The test is even more attractive in cohesionless soils such as sands, silts and most tailings due to difficulties associated with retrieving undisturbed samples in such soils. The behaviour of cohesionless soils strongly depends on their density as well as the stress level. While relative density, Dr, is an almost universally used density index for sand, it is easy to show that it can be misleading (e.g. Tavenas, 1973). Alternatives to Dr that capture both the effect of void ratio and the effect of mean stress such as Bolton (1986) relative density index and the state parameter,  (Been and Jefferies, 1985) can better represent soil behaviour. The difficulty with any penetration test, however, is that the state measure of interest (e.g. Dr, )    1 is not what is measured. Instead it is calculated from the penetration resistance; a process usually referred to as interpretation. This interpretation involves solution of an inverse boundary value problem to obtain mechanical properties of the soil from test measurements. But the large deformations associated with the CPT, along with the nonlinear behaviour of the soil and complicated boundary conditions, make this analysis an extremely difficult task, and the solution non-unique. The interpretation framework is also difficult to establish. No simple closed-form solution for  or Dr from CPT has been developed; and, nobody - to date - has provided a full numerical simulation of drained penetration that matches calibration data, although several have tried (de Borst and Vermeer, 1982; Willson et al., 1988; van den Berg, 1994; Huang et al., 2004; and Ahmadi et al. 2005). Two different directions have emerged to estimate soil state from CPT data: calibration chamber tests and analytical treatments. Ideally, the adequacy and accuracy of interpretation methods should be verified against direct measurements of the in-situ density, such as undisturbed sampling. However sampling is particularly difficult in cohesionless soils with in-situ ground freezing being the only widely accepted technique; but such methods are prohibitively expensive and time consuming (Hoffman et al., 2000). Another problem with developing interpretation methods based on direct measurement of in-situ density is the unknown state of stress in the ground. The relation is affected by the horizontal stress (h0) in addition to the readily determined in-situ vertical effective stress v0 (Baldi et al., 1986; Been et al. 1987; Schnaid, 1990), so that the horizontal geostatic stress ratio K0 becomes important to accurate determination of  or Dr. The need for controlled test conditions is not unique to this problem and is a common reason for resorting to laboratory testing. Calibration chamber testing has been the standard method adopted in the literature for developing interpretation methods for CPT.    2 Calibration chambers are circular steel tanks, typically about a metre in diameter and similar height. Sand is deposited at a known density and consolidated to the desired stress state within the tank. A cone penetration test is then performed along the vertical axis of the sample. Each test provides a tip resistance qc for the given value of density and stress of the sample. A large number of tests, covering the range of densities and stresses of interest, provide the relation between qc, in-situ effective stresses (1,2, 3), and the density (usually expressed as  or Dr) for the tested material. The in-situ state 0 (the subscript “0” denotes the initial value) is then obtained from the CPT by comparison of field CPT qc measurements at the estimated in- situ stresses, to the qc0 relation determined in the calibration chamber. Calibration chamber testing has been successfully used for calibrating CPT results to the state parameter (e.g. Been et al., 1987). However, the difference between the fabric and age of sample and the in-situ conditions still remains one of the major problems of extending calibration chamber results to in-situ CPT calibration. Moreover, this method is very time consuming and expensive; consequently, only a limited number of soils have been tested in calibration chambers and these are mostly at the research level. The qc0 relation differs from one soil to another so that, although the form remains common among soils, the coefficients involved are particular to the calibrated soil. Different theoretical solutions have arisen as alternatives to calibration chamber testing for interpretation of the Cone Penetration Test. These methods have the advantage of being relatively easy and cheap to perform while providing insight into the parameters affecting the interpretation. This insight is a product of the theoretical methods incorporating some intrinsic material characteristic and thus relating them to the interpretation scheme. Consequently, theoretical solutions have the potential to account for the basic differences    3 between calibration chamber tests and in-situ conditions such as fabric and age. Extending the argument to the effects of other material characteristics on the interpretation suggests that sufficiently advanced solutions should be able to provide material specific correlations for soils that have not been tested in calibration chambers. A variety of methods have been used for modelling the deep penetration problem (mainly CPT and pile driving). The earliest methods approached the deep penetration phenomenon as a bearing capacity problem (e.g. Meyerhof, 1951; Vesic, 1963; and Sokolovskii, 1965). These bearing capacity idealisations assumed perfect plasticity and did not consider the deformations caused by the cone penetration. The finite element method has been used for the analysis of the deep penetration problem for almost three decades (e.g. de Borst and Vermeer, 1982). Early efforts made significant simplifying assumptions to deal with the complexities of the problem; for example, prescribed deformation patterns were applied to replicate the penetration (e.g. Cividini and Gioda, 1988) or small strain formulation was used (e.g. Griffiths, 1982). With the advances in computational power, capabilities such as large strain formulation, adaptive mesh refinement and contact interface have been incorporated into the numerical analyses. More recent efforts (e.g. Susila and Hryciw, 2003; and Huang et al., 2004) have captured the geometric aspects of the problem with minimal simplifying assumptions. Although significant progress has been made in capturing the geometrical aspects of cone penetration, less attention has been given to realistic modelling of soil behaviour. Elastic- perfectly plastic models with different flow rules are dominant for representing frictional soil (e.g. Huang et al., 2004; and Susila and Hryciw, 2003). Simple critical state models such as Modified Cam Clay have been used by some researchers (e.g. Sheng et al., 2005), but these    4 models have certain deficiencies in modelling density dependent behaviour of sand. Since the cone penetration is a deformation controlled process, it is necessary to simulate the dilation behaviour of the soil to an acceptable extent; a more sophisticated model is necessary to do so. Such a model should be able to simultaneously capture the behaviour of the part of the domain that is critical, dilating, or contracting; with the exact behaviour being a function of position. Despite the obvious advances in modelling the problem in both sand and clay, it seems that the objective of determining soil properties from CPT has been lost in most efforts adopting finite element modelling; most attention has been directed towards modelling the problem itself assuming that the soil parameters are known. Ahmadi (2000) commented that no comprehensive comparison between experimental data and finite element results had been made. Most works are limited to comparing a few tests on a single soil and are primarily focused on predicting the resistance profile, while the steady value of resistance is what matters in uniform deposits. Between the bearing capacity methods and full 3-D numerical modelling of the complete CPT penetration using a realistic soil model, lie a variety of simplified approaches such as spherical cavity expansion (e.g. Bishop et al., 1945; Yu and Houlsby, 1991; Salgado, 1993; and Shuttle and Jefferies, 1998) and steady state analyses (e.g. Baligh, 1985; and van den Berg, 1994). These methods make simplifying assumptions which induce certain limitations to their application for interpretation purposes. However, it is fair to say that they have contributed much to the understanding of the factors affecting the deep penetration problem as well as correlating in-situ parameters to the penetration resistance. The cavity expansion is an especially attractive approach because of its relative ease of implementation and application even in comparison to steady state analyses. This makes it    5 useful to most geotechnical engineering projects requiring a practical CPT interpretation framework. The reduction of the problem from its actual geometry to spherical symmetry necessarily results in applying only one stress invariant on the boundaries – as opposed to different vertical and horizontal stress components applied to the 3-D problem. However, this may not be a deficiency in itself if it is shown that the stress invariant applied at the boundaries can normalise any stress effects on qc. Many researchers have pointed out the relation between qc , the soil density (Dr or ) and the mean effective stress, p (Clayton et al., 1985; Baldi et al., 1986, and Been et al., 1987). The data obtained with sands in calibration chamber testing show that there is negligible influence of K0 , provided that the mean stress is used in normalising the data (Houlsby and Hitchman, 1989). In this thesis the cavity expansion analogy is employed along with a constitutive model capable of capturing the density dependent behaviour of cohesionless soils to determine the state parameter from the CPT tip resistance qc. Special attention is directed to   Independent calibration of the numerical model to material behaviour through laboratory element tests on reconstituted samples  Thorough verification and validation of the analysis against an extensive database of calibration chamber tests and in-situ measurements  Associating confidence levels in estimating the state parameter to different levels of accuracy  The cone penetration was analysed as the expansion of a spherical cavity using a large strain    6 finite element code incorporating a critical state soil model (NorSand) capable of accounting for both elasticity and plastic compressibility. The constitutive model was calibrated through triaxial tests on remoulded samples. The state parameter was interpreted from CPT tip resistance, and the results were verified against a database of calibration chamber tests on natural sands and tailings, in addition to the laboratory standard sands. The efficiency of the method was further investigated by analysing two well documented case histories confirming that consistent results could be obtained from different in-situ testing methods using the proposed analysis technique. Consequently, cumbersome large scale testing methods, such as calibration chamber testing, can be replaced by a combination of triaxial testing and finite element analysis to produce soil specific interpretations. There are two fundamental limitations in capturing the true behaviour of the material around the cone in the current analysis: The assumption that the elastic moduli remain unchanged during penetration and ignoring the particle breakage phenomenon. For a certain material, the elastic moduli are known to be mainly affected by the stress level and void ratio, and to a lesser degree by other characteristics such as fabric. The penetration of the cone significantly increases the stresses in its vicinity while reducing the void ratio. Both of these will increase the elastic moduli around the cone resulting in a stiffer response. However, this effect has not been incorporated into the model due to numerical stability issues arising with elasticity approaching rigidity. Another difficulty in understanding the cone penetration problem is the less researched effect of high stresses developing around the cone on the behaviour of the soil. Such stresses cause breakage in the particles increasing the compressibility of the soil. Russell and Khalili (2002) provided one of the earliest attempts at incorporating the effect of particle breakage into CPT    7 interpretation, by assuming a steepening of the Critical State Locus (CSL) with increasing mean stress, into a cavity expansion analysis using a critical state based model. The effect of this change was a reduction in the limiting cavity expansion pressure, inferring denser soil states for identical CPT tip resistances. However, the idea of a steepening CSL is theoretically questionable and fails to explain some of the behaviour observed in laboratory experiments. While the changes in elasticity are relatively well understood and the problem is now reduced to a purely numerical one, the modelling of the particle breakage is still in its infancy especially within the critical state soil mechanics. Thus an attempt was made at further developing the state of knowledge on the subject. A simple hypothesis on particulate behaviour of the material is proposed to understand the movement of the Critical State Locus due to shearing. The hypothesis was tested using a series of triaxial compression tests on a local sand.  1.2. Verification and Validation  Any theoretical framework has to be verified and validated against an acceptable amount of data before being used as a reliable interpretation technique. This makes a method only as accurate as the experimental data that it has been verified against. But more importantly, there will be a limit to the extent that a method can be generalised for application to materials different from those included in the verification process. This limit is set by the number of material characteristics that can affect the interpretation and are properly accounted for in the theoretical framework. For example, assume that elasticity is known to play a role in the interpretation of the state    8 parameter from CPT tip resistance, and a method has been verified for a number of materials with a certain range of elastic moduli. The validity of the application of this method to materials with elastic moduli outside of that range will depend on the capability of the method to adequately capture the effect of elasticity on the interpretation. Validation immediately raises the issue of determining ‘ground truth’. This situation has arisen in other aspects of in-situ testing of soil, and it is usual to evaluate the soil properties determined by one test method against those determined by another method in evaluating reliability of the two methods, for example, work at ‘national test sites’ by Nash et al. (1992) and Woods et al. (1994). In case of in-situ state of cohesionless materials, direct measurements of the void ratio were done in a few cases using ground freezing techniques (e.g. Wride et al., 2000; Plewes et al., 1994). Such cases would provide ideal validation cases for a theoretical method verified against a database of calibration chamber tests.  1.3. The NorSand Constitutive Law  The constitutive model adopted, NorSand (Jefferies, 1993), is an isotropically hardening - isotropically softening generalised critical state model that captures a wide range of particulate soil behaviour. The primary attribute of NorSand for this work is that it dilates in a realistic manner, thereby allowing the effect of volume change on the CPT resistance to be well represented. NorSand can be regarded as a super-set of the well-known Cam Clay model (Schofield and Wroth, 1968), with Cam Clay being obtained as a special case of NorSand by appropriate choice of the soil properties and initial conditions. NorSand has been well documented in the literature.  The version used herein is that for monotonic loading of general    9 3-D stress states with constant principal stress direction as described in Jefferies and Shuttle (2002) but with one modification to improve accuracy (described below); a good overview of this version of the model is provided in Jefferies and Shuttle (2005). The critical state, which is the condition at which soils deform continuously and indefinitely at constant volume, is used as the reference framework. The void ratio at the critical state depends on the mean effective stress ce p , and various relations have been proposed for this. These different relations are details that do not affect the overall approach, and preference for one relation over another is given to fit the behaviour of a particular soil. For most purposes the familiar pec  ln  idealisation is both simple and sufficiently accurate.  is the intercept of CSL, defined at 1 kPa and is its slope on a logarithmic scale. An infinity of Normal Consolidation Loci (NCL) for cohesionless soils (Ishihara et al., 1975) forces two parameters to characterise the state of a cohesionless soil:   and R. The state parameter   is a measure of the location of an individual NCL in  space. The overconsolidation ratio R represents the proximity of a state point to its yield surface when measured along the mean effective stress axis. Note in particular that R and pe   are not alternate identities as implied by Cam Clay or its variants; R and   represent measures of different things as illustrated in Figure 1-1. The critical state is also a relation between mean and shear stress, and this has been extensively investigated. To a high precision and high stress levels, the critical state is fitted by cc pMq  (where the subscript ‘c’ denotes critical conditions and M is a soil property;  is the deviator stress invariant at the critical state and should not be confused with its application as the CPT cone resistance in the rest of the thesis). More precisely, M varies with the proportion of cq    10 intermediate principal stress (usually given as the lode angle ) and the soil property is conventionally defined under triaxial compression conditions as with the subscript ‘tc’ being used to indicate this. is related to the critical state friction angle tcM tcM c  through Equation 1-1:  c c tcM   sin3 sin6           [Eq. 1 - 1]  NorSand is a plasticity model for soil. As such, and in common with other plasticity models, it comprises three items: (1) a yield surface; (2) a flow rule; and, (3) a hardening law. These three aspects of the NorSand model are described below, with the equations of the model summarised in Table 1-1. The NorSand yield surface has the familiar bullet-like shape of the classical Cam Clay model but with one important difference – there is an internal cap so that the soil cannot unload to very low mean stresses without yielding. This internal cap is taken as a flat plane, and its location depends on the soil’s current state parameter. Figure 1-2 illustrates the NorSand yield surface for two cases, a very loose soil and a very dense soil. As illustrated on Figure 1-2, the location of the internal cap controls the limiting stress ratio L  that the soil can sustain; what is sometimes called the Hvorslev Surface is seen to be a hardening limit for the true yield surface. This internal cap generalises the ideas of Drucker et al. (1957), the key insight of which is that the familiar Mohr Coulomb strength envelope is not a yield surface, and gives realistic dilatancy with normality. Yield surfaces have a ‘bullet’ shape that intersects the Mohr Coulomb strength envelope at the critical state.    11 The plastic dilatancy  is determined from the idealised stress-dilatancy relation that underlies the model, but as there are three strain rates this is insufficient to determine each of them. The intermediate principal strain rate is therefore interpolated depending on the lode angle. This interpolation approach is somewhat unusual for plasticity models, but is taken to ensure consistency with the work dissipation postulate that is the basis of the model. pD The third aspect of the model is the hardening law, which describes how the yield surface increases or decreases in size with plastic straining. The size of the yield surface is controlled by what is termed the image stress  and which forms the object of the hardening law. It is called the image stress because it represents a situation in which one of the two conditions for the critical state (zero volume change) is met, and the meaning is readily apparent from Figure 1-2. ip NorSand has isotropic hardening which expands or contracts the yield surface, as required by the hardening law, while retaining its shape. The position of the internal cap evolves with the changing state parameter. Whether the yield surface hardens or softens depends on two things: the current state parameter and the direction of loading. Loading past the internal cap always softens the yield surface, as does principal stress rotation. However, the key determinant in the general hardening is the state parameter. As illustrated on Figure 1-2, the critical state does not usually intersect the yield surface (this is the largest single difference between NorSand and Cam Clay). This divergence of yield surface from critical state is used as the basis of the hardening law, and the hardening law acts to move the yield surface towards the critical state under the action of plastic shear strain – which directly captures the essence of critical state principles. Elasticity in soils is arguably a more complicated aspect than plasticity. Various elastic models can be used depending on the trade off between complexity and accuracy. Pestana and Whittle    12 (1995) looked at elastic models that accounted for the effect of void ratio in some detail with Jefferies and Been (2000) suggesting a further refinement. However, for most situations involving soil modelling this sophistication is unwarranted. On one hand the soil fabric affects modulus and thus elastic modulus should be measured in-situ. On the other hand, plastic strains often dominate. For most purposes a simple constant shear rigidity  and constant Poisson’s ratio is a sufficient representation of elasticity and this approach is used here. This is equivalent to the constant κ idealisation of the classic CSSM models. rI NorSand is a sparse model. For the variant presented here, in which the CSL has been taken as the approximate semi-log form and with the simplest representation of elasticity, there are eight model parameters. These are summarised on Table 1-2 with typical ranges in values for sands indicated. How to determine the model parameters is presented later in the chapters and in detail in Jefferies and Shuttle (2005); examples of calibrated parameter sets are given there. All the parameters are dimensionless, although   has a reference stress level associated with it. Most of the parameters are familiar including rtc IM ,,,  and ν. Only three parameters may be unfamiliar. The parameter tc  is the slope of the trend line for minimum dilatancy versus the state parameter at minimum dilatancy. In the original version of NorSand this trend was thought to be a model constant. But, further data have shown that tc  varies somewhat from soil to soil and could also be a function of soil fabric. The reference condition is taken as triaxial compression because dilatancy is itself a function of the lode angle. A feature of the original version of NorSand was a volumetric coupling parameter N for stress- dilatancy. Subsequently, Jefferies and Shuttle (2002) suggested that N could be eliminated from the model since Ntc1 (based on average values of a large quantity of triaxial tests on different    13 sands). However, individual soils demonstrate a variety of Ntc values and some accuracy is sacrificed by following this suggestion. Adding N back to the model neither increases the complexity of the model, nor constitutes additional effort in calibration of the model, as N is obtained with Mtc from the stress-dilatancy plot and reintroduction of this parameter resulted in better replication of the soil behaviour. For the current work N was obtained as the slope of the post peak stress-dilatancy plots, rather than using the Bishop (1971) methodology and is termed N* for clarity. The plastic hardening modulus H can in principle be a function of soil fabric, and data to date suggests that it is often a function of  . There is some evidence that H is proportional to 1/(λ−κ) which is what might be anticipated from NorSand’s similarity to to Cam Clay. On the other hand, such a linkage should also be anticipated to be affected by soil fabric.  1.4. Literature Review  1.4.1. Cone Penetration Test  Since being developed in early 1930s as a geological tool for stratigraphic purposes, the cone penetration test has attracted attention from the geomechanics perspective. The main advantage of the CPT is the interesting combination of a continuous data record with excellent repeatability and accuracy at relatively low cost. The first cone penetrometers were made in 1932 by P. Barentsen, an engineer at the Department of Public Works in Netherlands (Lunne et al., 1997). The first electric cone penetrometer was most likely developed in Berlin during the Second World War (Lunne et al., 1997). Much of the    14 subsequent developments occurred in the Netherlands (Broms and Flodin, 1988). The work was continued in Canada by Campanella and his students (Campanella et al., 1983; Robertson and Campanella, 1983a,b, 1986; Robertson et al., 1986; Robertson and Wride, 1998; and Robertson, 2009), and by many other workers worldwide. The cone penetration test is conducted by pushing a penetrometer with a conical tip attached to the end of a series of rods into the ground at a constant rate; continuous measurements are made of the resistance of the tool to penetration. The standard electronic cone penetrometer (ASTM D5778) has a conical point with 60° apex angle and a projected cone base of 10 cm2 (see Figure 1-3). The cone penetrometer is advanced through soil at a constant rate of 20 mm/sec. The cone tip resistance (Qc) and sleeve friction resistance (Fs) are the basic readings of CPT results (both in force units). The excess pore water pressure (u) can be obtained through a piezometer on or behind the cone tip. It is common to have the piezo-element 5 mm above the cone and the pore pressure recorded at this location is denoted by u2 . Nowadays a cone penetrometer can contain many sensors and measurements of electrical conductivity, inclination, temperature and shear wave velocity can also be performed during cone penetration. Readings from the electronic measuring devices are usually recorded automatically in a computer during penetration. Qc divided by the projected area of the cone, Ac, produces the tip resistance qc. qc is corrected for the effect of the net area ratio (a) through Equation 1-2    21 uaqq ct           [Eq. 1 - 2]  qt and qc are essentially the same in drained penetration (as in cohesionless soils) due to low pore water pressures around the cone. Fs divided by the surface area of the friction sleeve, As,    15 produces sleeve resistance (fs). The normalised friction ratio  is defined as a percentage with rF  100 0  vt s r q fF           [Eq. 1 - 3]  Since fs is measured about 10 cm behind the cone tip (for a typical ASTM D5778 cone), the value of fs and qt in Equation 1-3 should not be the values recorded at the same time. Instead, the values recorded at the same depth should be used; i.e. fs should be shifted to account for the distance between the cone tip and the centre of the friction sleeve.  1.4.2. Interpretation of the State Parameter from CPT Tip Resistance  The CPT provides three measurements; the tip resistance, the sleeve friction, and the pore pressure. A combination of the three is usually used in a qualitative way for soil classification (i.e. Been and Jefferies, 1992; Robertson and Wride, 1998). The CPT in sand provides just two outputs; the tip resistance and the sleeve friction as the penetration is drained so the pore pressure transducer simply measures the in-situ pore pressure. Once a soil type has been identified it is common practice to correlate certain soil parameters with one of the measurements. The shear strength parameters (e.g. su, friction angle and density indices) are usually estimated through the tip resistance. The state measure used in this work is the state parameter,  (Been and Jefferies, 1985). Because  is used as an internal state variable in the numerical model, the subscript ‘0’ is used    16 to denote the in-situ (or initial) value of 0 under geostatic conditions to be consistent with the original usage by Been and Jefferies. Initial work with determining 0 from CPT data comprised triaxial testing of sands for which chamber test data was available to define the CSL of each sand, and then processing the chamber test data to develop dimensionless relations (Been et al., 1986, 1987) of the form:   0exp mkQ           [Eq. 1 - 4]  Where Q is the tip resistance normalised by the initial mean total and effective stresses, p0 and p′0:  0 0 'p pqQ t             [Eq. 1 - 5]  Been et al. (1987) suggested that the two coefficients k and m in Equation 1-4 depend on the compressibility of the soil through the slope of the critical state line . They used a database of five different clean sands to develop their correlations for obtaining the state parameter from CPT tip resistance. Plewes et al. (1992) looked at a database of clean sands as well as silty clays and suggested that 10 (on logarithm base 10) can be approximated by the normalised friction ratio  (written as a decimal) according to Equation 1-6: rF     17 1010 rF            [Eq. 1 - 6]  Following the idea of Been et al. (1987) for correlating k and m to Plewes et al. (1992) provided generalised equations that can be applied to both sands and clays:  1085.03 tcMk           [Eq. 1 - 7] 103.139.11 m           [Eq. 1 - 8]  Equation 1-7 implies that k and m in turn the interpreted state parameter are functions of the critical state friction ratio . tcM Konrad (1997) adopted Equation 1-4 and suggested that the in-situ state parameter can be determined by normalising the state parameter with respect to emin and emax. He also developed a correlation for correcting the stress level effect based on data from Ticino sand. He suggested that in the absence of material specific correlations, this stress level correlation can be used for other materials. Robertson (2009) developed contours of state parameter on the material type behaviour charts of Robertson and Wride (1998) based on laboratory and field investigation data. The contours are shown on Figure 1-4 where normalised tip resistance,  is plotted against normalised friction ratio, .  is defined by tnQ rF tnQ     nvaavttn ppqQ 00            [Eq. 1 - 9]     18 n is a stress exponent that varies from 0.3 to 0.9 for most coarse grained soils.  is a reference pressure assumed to be equal to 100 kPa and is used to preserve the non-dimensionality of the normalizations. ap Combinations of measurements from independent in-situ tests have also been adopted for obtaining the state parameter in sands. Yu et al. (1996) used the ratio of measured cone tip resistance and pressuremeter limit pressure to calculate the peak friction angle and the state parameter. Schnaid and Yu (2007) used the ratio of elastic stiffness to the cone tip resistance to estimate the same parameters.  1.4.3. Calibration Chamber Testing  A calibration chamber test involves preparing a large sand specimen in the laboratory, consolidating it to a desired stress level, and then performing a CPT test under given boundary conditions. Since the entire experiment is conducted in the laboratory, the test quality can be readily controlled. The large sand specimen, with uniform deposition and known engineering properties, provides reference values for the interpretation and thus calibration of the in-situ test method. A cone penetration test is then performed along the axis of the sample. Each CPT test provides one value of qc corresponding to the chamber’s initial density and stress state. A number of tests, covering the range of densities and stresses of interest, provide a relation between qc, density and stresses. To interpret the properties of a soil, the measurements of CPT in the field are compared with the measurements for the same soil in calibration chambers. Chamber tests have been in use for over 40 years. The first advanced calibration chamber (including measurement of boundary stresses and strains) was built in 1969 at the Country    19 Roads Board (CRB) in Australia (Holden, 1991). Nowadays, the chamber tests differ in a number of ways such as dimensions, nature and form of control of boundaries, deposition procedure, and capability to handle saturated specimens. In 1991 Ghionna and Jamiolkowski provided a list of 19 calibration chamber tests reported in the literature. More calibration chambers have been built since then (e.g. Hsu and Huang, 1998; Ajalloeian and Yu, 1998; and Tan et al., 2003). Table 1-3 is an extension of a compilation of calibration chambers by Ghionna and Jamiolkowski (1991) to encompass some of the recent works. Holden (1991) summarised the advantages of the calibration chamber as follows:  1. The lateral boundaries are flexible. It is possible to produce the normally consolidated (NC) or overconsolidated (OC) specimens under K0 conditions. 2. The boundary stresses are known and controlled. 3. The chamber specimen can be large enough to perform a full scale CPT. 4. By means of pluvial methods, the chamber specimens are uniform and reproducible.  Been et al. (1988) and Ghionna and Jamiolkowski (1991) reviewed the problems associated with calibration chamber testing. Sample age, limited number of sands tested, and sample size and boundary effects are among the important limitations of calibration chamber testing. Calibration chamber tests are performed on specimens of freshly reconstituted sands. The fabric of these samples of sand will likely be different from that of the natural soil deposit in situ. The structure and ageing effects have been reported to have a significant effect on measured cone resistance (Schmertmann, 1991). Most calibration chamber tests have been performed on uniform, clean predominantly silica sands. Natural sand deposits are seldom as uniform, and generally contain    20 some amount of fines that may significantly influence their behaviour. In addition, many relevant engineering problems are linked to crushable and compressible materials such as carbonate sands which may also be slightly cemented. Strictly speaking, the correlations are only valid for sands of similar gradation, particle shape, and mineralogy to the sands the correlations are based upon. To extend calibration chamber research to more realistic soils such as silty, crushable, and cemented sands is a desirable aim for future work.  1.4.3.1. Boundary effects  It is well known that the value of penetration resistance measured during the test is influenced by the conditions at the sample boundaries (Parkin and Lunne, 1982; Baldi et al. 1982; Bellotti et al., 1982; Been et al., 1987; Sisson, 1990; Mayne and Kulhawy, 1991; Salgado, 1993; Zohrabi et al., 1995; Huang and Hsu, 2005). As summarised in Table 1-4, five different boundary conditions have been used in calibration chambers. The differences between the five types of boundary conditions are in the type of stress or displacement boundary conditions imposed on the top, bottom, and circumferential surfaces of the sample. Been et al. (1988) based on a limited amount of data, indicated that boundary conditions on top and bottom of the chamber specimen had little effect on CPT results. Boundary effects are more apparent in dense sand than in loose sand. The influence also decreases with the compressibility of sand. For loose sand, chamber results are relatively independent of boundary conditions, even when the ratio of chamber to cone diameter (D/dc) is as low as 21; for dense sand, all calibration chamber results are affected by boundary conditions, even for D/dc of 60 or greater (Parkin, 1988).    21 BC1 and BC4, both corresponding to a constant lateral stress during cone penetration, generate approximately the same penetration resistance values, all other factors being equal. Similarly, BC2 and BC3 generate comparable penetration resistance values under the same conditions. None of these four boundary conditions perfectly represent free-field conditions, so qc in the calibration chamber is generally different from qc in the field. The difference between chamber and field qc values decreases as the ratio of chamber to cone diameter (D/dc) increases. Whether the chamber qc values are bigger or smaller than the in-situ values also depends on the type of boundary conditions. Under BC1 conditions, qc is always lower than in the field, because a constant lateral stress during penetration underestimates the lateral stresses that will develop during penetration in the field. Ideal BC3 conditions, with a perfectly rigid lateral wall, would always impose boundary effects that lead to higher qc values than those measured in the field. Additionally, a rigid wall does not offer the same degree of control during the sample preparation, consolidation, and penetration stages of the test. In recent years, there have been attempts to simulate real soil response at boundaries of a calibration chamber specimen. The idea is that the lateral boundary deforms in response to the cone penetration test conducted within the soil specimen. The boundary deformation should then induce variation of stress on the lateral boundary due to the reaction of soil from beyond the lateral boundary, if the soil extended to infinity as in field conditions. Such chambers are called servo-controlled calibration chambers. The servo-controlled systems reported by Ghionna and Jamiolkowski (1991) and Foray (1991) allowed the stress to vary but remain uniform throughout the lateral boundary. Hsu and Huang (1998) developed a servo-controlled (BC5) calibration chamber that allowed the boundary response to be independently controlled at    22 different depths. However, there is little information about the efficiency of these methods and most available data still come from tests with the first four boundary conditions.  1.4.3.2. Correction and Normalisation of Boundary Conditions Effect  Recognising the boundary effects on calibration chamber data, researchers have developed correlations and graphs to correct for them. Looking at calibration chamber data on Ticino sand at D/dc=34.2, Baldi et al. (1982) reported a series of chamber size correction factors, r defined as  ccc fieldc q q r , ,            [Eq. 1 - 10]  where qc,cc is the experimental value of tip resistance observed in the calibration chamber and qc,field is the tip resistance expected to be measured in the field for the same sand with the same relative density and the same in-situ stresses as in the chamber. They suggested that the value of r increases with density and the over consolidation ratio. Lunne and Christophersen (1983), based on chamber test results using Hokksund sand, suggested that for a chamber to cone diameter ratio of 50, the difference between the tip resistance obtained in the chamber and the field is small. Jamiolkowski  et al. (1985) proposed a formula to relate the tip resistance obtained in the chamber to the tip resistance obtained in the field based on data from Ticino and Hokksund sands under BC1 and BC3 type boundary conditions.    23       60 30%2.01 , , r ccc fieldc D q q        [Eq. 1 - 11]  Equation 1-11 implies that for loose sand with a relative density of 30%, the experimental result of the chamber and the field are similar and no calibration size effect should be considered. However, as the relative density of sand in the chamber increases, the size effect will become larger (e.g. 1.13 for relative density of 70%). Been et al. (1986) used previously published data to study the chamber size effects on normalised cone resistance. They used data from Hokksund and Ticino sands with different boundary conditions to normalise the cone resistance to that of a chamber size with D/dc of 50. Figure 1-5 illustrates the variation of the Normalisation factors in different boundary conditions versus the state parameter. It should be noted that the factors are derived based on a particular design of chamber and for a limited number of sands. Hence they should be cautiously applied to other soils and chamber test designs. They acknowledged that a D/dc of 50 may not be analogous to the field and is only used as a base for comparing different data. Mayne and Kulhawy (1991) assumed that, regardless of the relative density and stress state, a chamber diameter to cone diameter ratio of 70 is sufficient to achieve the "free field" condition. This means that in order to prevent boundary effects for the standard cone diameter of 3.57 cm, a 2.5 m diameter chamber is required. They proposed a relation to correct the cone resistance for boundary effects that includes the chamber over cone diameter ratio and the relative density. This relation is based on some of the 24 sets of data of cone penetration test results performed in conventional calibration chambers.     24    200% , , 70 1/ rD c ccc fieldc dD q q              [Eq. 1 - 12]  In a 1.2 m wide chamber the equation returns correction factors of 1.12 for Dr = 30% and 1.30 for Dr = 70%. Cudmani and Osinov (2001) adopted the form of the Mayne and Kulhawy (1991) correlation and generalised it as      BI c c ccc fieldc D dD AdD q q / 0, , * / /               [Eq. 1 - 13]  where (D/dc)0 is the ratio at which the influence of the chamber size vanishes, and A and B are constants to be determined.  is a normalised state parameter varying between 0 and 1 and being negative for very loose states. They suggest A=0 and (D/dc)0=60 and suggest the values in Table 1-5 for different sands based on previously published experimental data. * DI Salgado (1993) proposed an analytical scheme to correct the boundary effects based on cylindrical cavity expansion considering the dilatancy effects. Salgado et al. (1997) too used the spherical cavity expansion and concluded that the tip resistance is a function of relative density, stress state, and intrinsic parameters of the sand. This work was then used in Salgado et al. (1998) to obtain correction factors for chamber tests and was compared to experimental data. They proposed charts for obtaining correction factors that were dependent on relative density, stress state, and intrinsic parameters of the sand. In summary, different researchers have tried to determine the ratio of chamber to cone diameter at which boundary conditions have little effect on the cone resistance; Values of D/dc between    25 50 to 200 have been suggested. Since the correction factors are sensitive to the type of sand being tested as well as the boundary conditions, they should be used cautiously. The ideal approach is to apply factors developed for each sand under a certain set of boundary conditions, to the same sand and boundary conditions; but such information is not available in most cases. Since the differences between the various boundary effect corrections is usually about an order of magnitude smaller than the random variability in the calibration chamber data, for most practical purposes application of any reasonable correlation is expected to sufficiently reduce the error caused by the boundary effects. The correction factor suggested by Been et al. (1987) has been used in this work, and indicates correction factors between 0.9 and 1.5 to the tip resistance qc. This correction generally falls midrange between those of Jamiolkowski et al. (1985) and Mayne and Kulhawy (1991). Using Been et al. (1987) correlation provides a uniform treatment of the boundary effects amongst all the materials.  1.4.4. Analytical and Numerical CPT Interpretation Methods  A range of analytical and numerical methods have been used to analyse the deep penetration problem, some focusing on interpreting in-situ parameters from Cone Penetration Testing. These methods include the bearing capacity theory, the steady cone penetration and strain path methods, the cavity expansion theory, finite element method and other analysis techniques such as finite difference and discrete element methods. Yu and Mitchell (1998) provided a general overview of different methods used for analysis of the cone penetration. These methods are reviewed in this section to the extent that they relate to this work.    26 1.4.4.1. Bearing Capacity Theory  One of the first methods used for the analysis of cone penetration was to treat it as a bearing capacity problem. The cone resistance is assumed to be equal to the collapse load of a deep circular foundation in soil. Two analytical approaches, namely limit equilibrium and slip-line analysis, have been used to determine the cone resistance. The limit equilibrium method adopts the concept of Terzaghi's (1943) bearing capacity theory. Under an assumed failure mechanism, qc is computed based on global equilibrium of the soil mass which is treated as a rigid body (e.g., Meyerhof, 1951; Durgunoglu and Mitchell, 1975). In the slip-line analysis, the stress field in the plastic zone generated by cone penetration is simulated as a network of slip-lines. A yield criterion (such as the Mohr-Coulomb or Tresca) is combined with the equations of equilibrium to give a set of differential equations of plastic equilibrium in the soil mass. The value of qc is determined by the distributed contact pressure acting on the cone face (e.g., Sokolovskii, 1965; de Simone and Golia, 1988; Koumoto, 1988). Although the stress field obtained from the slip- line method satisfies both the yield criterion and the equilibrium condition inside the slip-line network region, the stress distribution outside this region is not defined. Yu and Mitchell (1998) summarised the bearing capacity solutions for cone resistance in both cohesive and cohesionless soils. A major advantage of this approach is its relative simplicity. It can be easily accepted by many engineers who are already familiar with bearing capacity calculations. However, the relative simplicity of the approach also typically extends to the use of very simple, and unrealistic, idealisations of soil behaviour. In addition, the bearing capacity approach for the analysis of cone penetration has serious limitations, some of which are described by Yu and Mitchell:    27  In bearing capacity analysis, deformations of the soil are neglected, and this means the dependence of the cone resistance on soil stiffness and compressibility, as observed in laboratory and field testing, cannot be predicted.  The bearing capacity approach ignores the influence of the cone penetration process on the stress states around the shaft. In particular, the horizontal stress tends to increase around the cone shaft above the cone, and the influence of this change on the cone resistance is not considered in bearing capacity analysis.  Slip-line analysis is more rigorous than the limit equilibrium method, as it satisfies both the equilibrium equations and the yield criterion everywhere within the slip-line network. The limit equilibrium method only satisfies the global equilibrium.  Shear surfaces assumed are usually not observed accompanying deep cone penetration.  1.4.4.2. Steady Cone Penetration and Strain Path Method  Although the initial insertion of the cone into the ground is a transient process, the penetration in a uniform material is a steady process. In steady state analysis of the penetration it is assumed that the cone penetration has proceeded for a long enough length that steady-state conditions have been reached. In other words, it is assumed that an observer situated on the cone penetrometer observes steady-state conditions in the soil passing the penetrometer during the penetration process. According to Yu et al. (2000) this approximation requires that the soil deposit is an infinite, isotropic, homogeneous mass with a known initial stress state. The assumption of homogeneity in the vertical direction removes the possibility of material property,    28 initial stress, or strength dependence with depth within the zone of influence of the cone penetrometer. This assumption limits the applicability of these methods in modelling the behaviour of layered soil. Another requirement is that the cone penetrometer is inserted into the ground with a constant velocity. This requirement is usually fulfilled as constant velocity is a requirement of ASTM D5778. One of the first methods for conducting steady-state analysis of cone penetration in soil was the strain path method (Baligh, 1985). This method, which has long been used in modelling steady state cone penetration of incompressible soils such as clays (Baligh, 1985; Acar and Tumay, 1986; Huang, 1989; Gill and Lehane, 2000), is based on the assumption that the soil deformation is completely decoupled from the soil strength parameters. Sagaseta et al. (1998) extended the method to analyse shallow penetration problems. A combination of a strain path and the finite element method have been proposed by Teh and Houlsby (1991). In the development of the method, soil deformations were estimated using velocity fields from potential theory (where the material properties correspond to an ideal, incompressible and inviscid fluid). Using this approximation, different penetrometer geometries can be modelled through combinations of sources and sinks or other surface mapping techniques such as the boundary element method. For simple penetrometer geometries, the velocity and strain rate components can be obtained from closed-form expressions, while the displacements and strains are solved by numerical integration along the streamlines. Once the strain paths of individual soil elements are known, the material constitutive equations can be used to derive the effective soil stresses. For undrained penetration in clays, the constitutive equations can be represented using either total stresses (relating shear stresses and shear induced pore pressures to shear strains) or effective stress models. In either case, due to incompressibility, there is one stress    29 component (either the excess pore pressure or the mean stress) that cannot be obtained from the stress–strain relations and must be solved from the equilibrium equations. In general, it is not possible to match all of the equilibrium equations with a single unknown function (i.e. field of u or p). A unique solution for this stress component is only possible if the strain field (derived for an inviscid fluid) is exact. The main idea of Baligh’s strain path method is that the soil deformations and strains caused by the CPT are independent of soil strength and stiffness parameters and can then be estimated, for example, from irrotational flow of an ideal fluid. This assumption was based on experimental observations on deep penetration in clay. As shown in Ladanyi (1963) and Collins and Yu (1996), for 1-D problems such as cavity expansion in undrained clay, the soil deformations and strains can be determined from the incompressibility condition without considering stresses and strength. In other words, for 1-D cavity expansion problems in undrained clays, the strains are completely independent of soil strength parameters. Considering that cavity expansion theory has long been used with reasonable accuracy to model much of the soil behaviour during deep penetration, Baligh (1985) logically assumed that the soil deformation caused by cone penetration may also be determined with reasonable accuracy purely from an incompressibility requirement. It has been found, however, that the resulting stresses derived from this approach may not satisfy all the equilibrium equations. This is because for 3-D problems, such as cone penetration in undrained clays, the soil deformation is not completely decoupled from the soil strength parameters, although the coupling is believed to be quite weak for undrained problems (Yu et al., 2000). In order to account for this problem, Baligh (1985) and Houlsby et al. (1985) proposed an iterative procedure between the assumed velocity field (equivalent to the strain path) and the stresses obtained from the method in order to reduce the imbalance in the    30 equilibrium equation. However, as reported by Whittle (1992) this procedure cannot remove all the errors that exist in the strain path stress solutions. The application of the strain path method is mainly restricted to undrained clays. The application to cohesionless soils is almost impossible as the initial estimate of flow field for frictional-dilatant soils is very difficult to obtain. The reason is that the method is based on the assumption that the soil deformation is completely decoupled from the soil strength parameters. While the coupling is believed to be quite weak for undrained problems, it is much stronger in drained behaviour of dilatant soils. Other limitations of the strain path method are that the roughness of cone and shaft cannot be modelled, and that this method is not applicable to cone penetration in layered deposits. Van den Berg (1994) used a Eulerean framework, which is a special case of the Arbitrary Lagrangian Eulerean (ALE) method, to model penetration in both sands and clays. In this method the movement of the element nodes and the material points is decoupled. In other words, the material can flow through the mesh. In van den Berg’s work, the mesh is fixed, so the problem of extensive element distortion due to large displacements is solved. Simply speaking, the analysis is performed using a split algorithm; first, implicitly a Lagrangian step is calculated and, subsequently convection is taken into account explicitly in a remap loop over the nodes. The basic idea is the introduction of continuous stress and strain fields by interpolation of nodal point stresses and strains. So the next step starts with the mesh nodes at the initial location, but with an updated stress and strain field. This approach has long been applied to fluid mechanics problems, but is relatively new to geotechnical literature. Since in the model the constitutive behaviour of the material is coupled to the fixed elements, the method is capable of giving reasonable results only for homogeneous materials throughout    31 the complete finite element discretisation. In order to model layered soil, a tracking algorithm was proposed enabling the material properties to stream through the mesh as well. In other words, the material particles’ movement through the fixed finite element mesh is captured by tracking their constitutive behaviour. Van den Berg et al. (1996) mentioned that this algorithm may cause sudden imbalances in the numerical iteration procedure. This implies that in practice the difference between both sets of material parameters is limited, depending on the element size, the stress conditions and the total number of elements. Van den Berg (1994) used the Tresca yield criterion to model the behaviour of clay and both Drucker-Prager and Mohr-Coulomb yield criteria with hardening-softening capability and a non-associated flow rule for modelling behaviour of sand. The standard cone was modelled with a completely smooth surface. As illustrated in Figure 1-6, cone penetration was modelled by moving the lower boundary upwards, while keeping the cone in its original place. There were no changing boundary conditions, or element distortions associated with the problem because of the approach taken. Although chamber tests have been performed and reported in van den Berg (1994), no direct comparison of numerical results with measured values of cone penetration outputs was done. Instead, the focus of the testing was on the deformation field around the cone. However, the numerical approach of this work was notable for its time and the attention paid to modelling layered soil (both numerically and experimentally) is very important in the sense that even today very few works have been published in this regard. The finite element approach used by Yu et al. (2000) focused on the penetration process at a particular instant in time and used the spatial variation of the stress state instead of time integration of finite element matrices to obtain the solution at a certain point in the domain (e.g.    32 point P in Figure 1-7). Thus, the state of stress at point P is obtained by an integration process considering all points below P until the initial stress state of undisturbed soil is reached. This process converges, as the finite-element grid is refined, in the same way that a time marching scheme converges as the time steps are made progressively smaller. The dependency of the cone factor on soil stiffness, shaft roughness, in-situ state, and OCR were evaluated. The results were found to be similar to that of strain path method. The finite-element results on soil strain paths confirm the validity of the basic assumption in strain path method and that is, for undrained clay (i.e., incompressible materials), the coupling between soil deformation induced by steady cone penetration and soil strength or stiffness properties is very weak. It is therefore not surprising to note that the finite-element results (in particular, the dependence of cone factor on stiffness index G/Su) are similar to those of the strain path method. Advantages of the finite element procedure over the strain path method for steady state analysis include:   Equilibrium equations are fully accounted for in the finite-element method so that errors due to equilibrium imbalance can be minimised.  Cone geometry can be properly modelled in the finite-element approach.  Cone and shaft roughness can be accounted for through contact elements.  The finite-element approach can be easily modified for application to frictional- dilatant soils.       33 1.4.4.3. Cavity Expansion Theory  Many simplified theoretical treatments have used spherical (and occasionally cylindrical) cavity expansion as analogues to the cone penetration test, essentially the same approach as used in conventional design of end bearing capacity of piles. The spherical cavity expansion analogy idealises the CPT as a cavity in a uniform medium under an isotropic stress state, with the internal pressure of the cavity initially equal to the in-situ mean effective stress p0. The cavity is expanded by monotonically increasing its radius until a limiting (constant) pressure is obtained, this being the pressure of interest. This idealisation greatly simplifies the analysis because the symmetry allows only radial displacements and in turn this permits a one-dimensional description of the problem. The initial work by Bishop et al. (1945) and Hill (1950) addressed incompressible materials with associated flow rules, corresponding to the familiar and simple idealisation of the undrained behaviour of clay. Chadwick (1959) derived the pressure-expansion relation for a Mohr-Coulomb material with an associated flow rule. Vesic (1972) gave the general solution to the cavity expansion problem in an elastic-perfectly plastic material with a Mohr-Coulomb material with a non-associated flow rule. Vesic (1977) suggested that the bearing capacity of the soil around the tip of a pile can be estimated from the limit pressure required to expand a spherical cavity. The interesting further development of cavity expansion theory for non- associated flow has been considered by a number of workers including Carter et al. (1986) and Yu and Houlsby (1991). The central assumption of these studies has been that both the friction and dilation angles remain constant during shearing. Salgado (1993) and Salgado et al. (1997) advanced the work in a different direction by invoking a comprehensive database of well    34 documented calibration chamber tests into the analysis. Their work too used constant dilation and friction angles. Although constant dilation and friction angles led to analytical or semi- analytical solutions for cavity expansion, the idealisation has the fundamental deficiency that soil does not behave in such a manner – the nuances of soil behaviour being of first-order significance to realistic modelling of cavity expansion in soil. Collins et al. (1992) provided the first realistic model for cavity expansion in soil using a state parameter based numerical analysis. Their results showed that the relation between normalised CPT tip resistance and 0 was still affected by the stress level (as suggested by Sladen, 1989) and depended on material properties of the sand. Yu et al. (1996) used the same approach and combined it with cylindrical cavity expansion analysis of accompanying pressuremeter tests to interpret the state parameter as well as the friction angle from a combination of CPT tip resistance and pressuremeter limit pressures. Shuttle and Jefferies (1998) used a general work hardening/softening critical state model (NorSand) to evaluate changes in CPT calibration in terms of material properties that can be measured in situ (including elasticity G) or determined in routine triaxial testing of reconstituted samples (including the critical state parameters . They showed that CPT behaviour in Hilton Mines sand, currently one of the most unusual of the published calibrations, could be predicted based on Ticino sand data by allowing for the changes in fundamental soil properties between Hilton Mines and Ticino. The analogy of the cone penetration to expansion of a sphere in soil is based on the existence of a limit cavity expansion pressure; although it has long been recognised that there is an inherent mismatch between spherical cavity expansion analysis results and those of calibration chamber tests. The magnitude of the mismatch is enlarged with the application of realistic models that    35 capture the variable nature of strength and dilation angle. Their predecessors likely masked the mismatch by assuming constant friction and dilation angles. The difference between the spherical cavity analogy and the actual CPT geometry is usually accounted for by a shape factor (scaling the spherical results to the true in-situ resistance). Shuttle and Jefferies (1998) showed that there is a one to one correspondence between the normalised tip resistance obtained from spherical cavity expansion and that obtained in chamber testing on both Ticino and Hilton Mines sand. Cudmani and Osinov (2001) suggested a common shape factor for five different sands that was a function of the relative density corrected  for stress level effects. It is interesting that although particle breakage is known (e.g. De Beer, 1963; Yang et al., 2010) to occur at the high stresses usually induced by CPT in cohesionless soils, only a few of the preceding analyses explicitly model the effects of particle breakage on the limiting cavity pressure. Russell and Khalili (2002) considered this issue by incorporating one effect of particle breakage, a steepening of Critical State Locus with increasing mean stress, into a cavity expansion analysis using a critical state based model. The result was a reduction in the limiting cavity expansion pressure, inferring an increase in the shape factor required to match the chamber penetration resistance.  1.4.4.4. Finite Element Method  CPT and pile driving analyses are interchangeably considered in reviewing the research done on finite element modelling of the deep penetration problem. The early finite element models (de Borst and Vermeer, 1982; and Griffiths, 1982) considered the problem in undrained clay using elasto-perfectly plastic models. The problem of deep penetration in sand was first dealt    36 with by Willson (1985) where sand was modelled as a Mohr-Coulomb material. The first large strain analysis of the penetration process was performed by Cividini and Gioda (1988). Cividini and Gioda's work was also important in the sense that it was the first attempt to clarify the similarity of cavity expansion to penetration in soil through a finite element model. Another advantage of their work was that they pointed out the importance of a varying boundary condition at the tip of the cone and tried to address it by discretising the penetration process. Advances were made in accounting for different aspects of the deep penetration problem focusing on undrained penetration in clay (Kiousis et al., 1988; Mabsout et al.,1995; Abu- Farsakh et al., 1998; Voyiadjisa and Kim, 2003; Lu et al., 2004; and Wei et al., 2005). More sophisticated models (e.g. modified Cam-Clay) were used and the interface friction between the cone (or concrete in case of piles) and the soil were accounted for. The reason to use a ‘good’ constitutive model for modelling the cone penetration problem in sand is that, in general, part of the domain will be deforming at constant volume, other parts dilating, and other parts contractive; with the exact behaviour being a function of position relative to the cone tip. The works of Susila and Hryciw (2003), Sheng et al. (2004) and Huang et al. (2004) made significant advances in capturing the geometric aspects of the cone penetration problem in sand, but failed to satisfactorily capture the material behaviour. Susila and Hryciw (2003) modelled the cone penetration in normally consolidated sand using an updated Lagrangian large strain formulation and auto-adaptive remeshing technique; this technique was utilised for handling the very large distortion of the mesh surrounding the cone tip. The Drucker-Prager soil model with a non-associated flow rule was used to simulate the soil behaviour. Their analysis showed that the distribution of sleeve resistance along the cone sleeve is not uniform. It is significantly lower close to the cone tip and becomes uniform higher up    37 along the sleeve. Cone tip resistance was plotted against friction angle and vertical effective stress and a chart was proposed to obtain the friction angle from cone resistance. It was also shown that finite element results are in good agreement with previously published cavity expansion solutions and average values from calibration chamber test data. Sheng et al. (2004) modelled pile penetration in sand using a large strain finite element scheme. The pile was modelled with a conical tip instead of the flat end pile geometry. The pile had a tip angle of 60˚ which made it identical to CPT. A Modified Cam Clay soil model was used to simulate the behaviour of fine Silica sand and the pile was assumed to be rigid. Application of the Modified Cam Clay model to sands has serious limitations and does not capture their stress- dilatancy controlled behaviour. The computed total and shaft resistances during pile installation were first compared with measured values from centrifuge tests, which indicated that the total resistance was well predicted by the finite element model, but the shaft resistance was not. The computed stress paths indicate that both the mean and deviator stresses first increase when the pile is above or at the level of the observation point in the soil, and then decrease once the pile has moved below the observation point. They concluded that when the soil is represented by the modified Cam clay model, a thin soil layer of one pile radius immediately around the pile, extending from the ground surface to a distance of one pile radius above the pile tip, is under elasto-plastic expansion. Just outside this expansion (softening) zone, a compression zone of a U form was observed. Huang et al. (2004) modelled the cone penetration test using a large strain (updated Lagrangian) finite element scheme. The penetrometer deformation was assumed to be negligible, so the cone was treated as a rigid body. Mohr-Coulomb soil model with a non-associated flow rule was used    38 to simulate the soil behaviour. The interface between the cone and the soil was modelled using a friction contact interface. Parametric studies were performed to investigate the influence of the pressure level p0, the shear modulus G, the soil internal friction angle , and the dilation angle. Effects of these parameters on the cone resistance and the plastic zone around the cone were studied in detail and the results were compared to those of cavity expansion and bearing capacity theories. Parametric studies showed that the cone resistance is influenced more by deformation parameters (such as shear modulus and dilation angle) than by shear strength parameters (such as the friction angle). This finding was in agreement with the general understanding of the cone penetration as a deformation controlled process. Foster et al. (2005) performed a 3-D modelling of cone penetration into a cubic medium. The cone was modelled as an elastic shell while the soil was modelled using a model with 14 parameters. The model seems to take into account the critical state concept and assumes a constant elastic shear modulus. Another assumption of the model is that the plastic strain is proportional to the total strain. A master and slave surface concept was used with a frictional contact model to simulate the interface of cone and soil. Data from Norfolk sandy loam and Decatur clay loam were used to verify the performance of the model. Results of simulation were compared to experimental data for a limited number of tests plotted as cone factor versus displacement up to about 200 mm of penetration. Foster et al. (2005) used a realistic soil model for modelling behaviour of both sand and clay, but the approach of their work being mostly towards computational aspects of the problem and not the geotechnical aspects, makes it difficult to be used by the geotechnical community. Moreover, like many other works, it was not well verified with experimental data.    39 1.4.4.5. Other Methods  In addition to the methods previously discussed, there have been a few efforts to use other numerical techniques such as discrete element (Cundall and Strack, 1979) and finite difference methods to model the cone penetration problem. Huang and Ma (1994) were the first to apply discrete elements method to the cone penetration problem. Tannant and Wang (2002) used the method to model penetration of a wedge into oil sands. Iqbal (2004) used a 2-D discrete element method code to model CPT in a coarse grained soil with rigid single sized particles. He reports acceptable agreement between the results of the method and experimental and other numerical methods. The discrete element method, being a relatively new method in geomechanics, has not been verified against experimental data as widely as other numerical methods have. Computational effort required for the method is higher than that of continuum methods and this has restricted the application of the model to problems with a limited number of particles or problems dealing with coarse granular materials. All the works mentioned were done using 2-D codes; in other words they are all solving the problem of penetration of a wedge into a stock of cylinders. The method can be very helpful in understanding the penetration mechanism (especially with the more recent availability of 3-D codes) and issues such as fabric effect and changes of fabric during penetration about which other methods such as finite element method have failed to yield useful results. Ahmadi et al. (2005) used a commercial finite difference package (FLAC) to model the cone penetration test in sands. They used the Mohr-Coulomb elasto-plastic model with stress- dependent elastic moduli and friction and dilation angles to account for the high stresses    40 developed around the cone. The standard cone was modelled and it was assumed to be infinitely rigid. Large strain effects were accounted for by updating the mesh after each step. In order to tackle the changing boundary conditions created by penetration along the axis of symmetry, the analysis was carried out for an axisymmetric body with a pre-existing small hole along its axis of symmetry. The penetration was modelled using prescribed displacements. The main criterion considered in estimating the prescribed displacements was producing a deformation pattern in the analysis that was consistent with a penetrating cone and similar to that of the experiments. Generalising such a deformation field to all sands may cause major errors. As discussed in reviewing the steady state methods, different deformation fields will be generated in different cohesionless soils with different densities. This is the main reason methods such as the strain path method are not applicable to cohesionless soils. An important aspect of their work was that the results of the analysis were compared with experimental data obtained in chamber tests; previously published data on Ticino sand were used for this purpose. The ratio between the vertical and horizontal prescribed displacements at the cone boundary that yielded the best match between the analysis and the experimental data was determined by performing a parametric study; a ratio of 0.85 yielded the best fit. This adjustment makes the comparison to the experimental data a means of calibration, rather than verification. Ahmadi and Robertson (2005) extended the analysis to layered soils including two layers of different sands or sand and clay and thin layers of sand embedded in soft clay. They discussed the effects of thin layers on cone resistance and proposed a correction factor.    41 The application of the finite difference method to the Cone Penetration Test used by Ahmadi and coworkers is essentially identical to finite element method used by others. Hence their work can be viewed along with the finite element analyses performed by others.  1.4.5. Particle Breakage  The significance of particle breakage on the behaviour of granular materials has been recognised since the earliest days of soil mechanics (Bridgman, 1918). He looked at the problem from the point of view of geology and the existence of cavities in rocks at depth. De Beer (1963) looked at the scaling effect in using cone penetration data to design of pile foundations and studied the compression during cone penetration into a dense sand reaching pressures of up to 35 MPa. He suggested that the amount of breakage was affected by the characteristics of the material and was greatly intensified by introduction of large shear strains. Construction of large earth dams brought the shearing behaviour of granular materials at high stresses to the spotlight again in the 60s and 70s; interest was directed to the changes in strength, compressibility and permeability under high pressures. Bishop (1966) looked at the behaviour of a number of sands under a range of stresses of up to 10 MPa and concluded that the particle breakage is to a large extent caused by shearing rather than consolidation. He also suggested that during shearing the gradation shifts towards a well graded soil such as that observed in natural glacial tills and by doing so the soil can sustain a higher level of stress by increasing the number of inter-particle contacts carrying the stresses. Lee and Farhoomand (1967) investigated the effect of high stress levels on compressibility and particle breakage of a number of soils ranging from gravels to fine sands. They showed that    42 breakage starts from the finer part of the gradation curve. Uniform and angular soils demonstrate more breakage than well graded and round grained soils. They pointed out that particle breakage induces additional compressibility and showed that higher deviator stresses result in more breakage. Vesic and Clough (1968) looked at the behaviour of sands under a wide range of pressures and identified three ranges: very low pressures where dilatancy controls the behaviour and the breakage is negligible, higher pressures where breakage becomes more pronounced and suppresses dilatancy effects, and very high pressures where all effects of initial density vanish and sand behaves like a linear elastic material. Billam (1971) related the particle breakage to the tensile strength of particles. He also suggested that particle breakage increases the axial strain to failure by “reducing the rate at which the material can accept additional stress”. Different variants of the gradation curve have been used to quantify particle breakage. Miura and O-Hara (1979) suggested the increase in the surface area of the soil sample as a measure for the amount of particle breakage. They also suggested that the amount of particle breakage is a function of the plastic work done and proposed a relation between this plastic work and increase in the surface area. The idea of a breakage potential function and a total breakage function was introduced by Hardin (1985) based on the area between the gradation curves before and after breakage and investigated the effects of different factors such as initial particle distribution, particle shape, the amount of shearing and the stress path, initial void ratio, particle hardness and the presence of water. He concluded that the amount of breakage is increased by an increase in shear stress, initial density and particle angularity.    43 The effect of high pressures on the critical state of cohesionless soils has become of interest with the emergence of the constitutive models that are based on critical state soil mechanics. According to critical state theory, once a volume of soil reaches the critical state, it is expected to continue to shear at constant stress and constant volume. A two or three part linear CSL in  space has been generally accepted (Been et al., 1991; Konrad, 1998; Russell and Khalili, 2004) for the full range of mean effective stress. 'log pe  That the changing gradation caused by the breakage imposes a downward shift on the CSL was first suggested by Daouadji et al. (2001). Muir Wood (2007) developed this idea, suggesting that during shearing at higher stresses the CSL moves down towards a final location associated with a fractal gradation (McDowell et al., 1996). A third dimension was proposed to the e  log p' space called the “grading state index”; a parameter between 0 and 1 which identifies soil’s state on a scale between uniform and fractal gradations. Muir Wood and Maeda (2008) showed, using a discrete element analysis, that the effect of particle breakage on the CSL location in e  log p' space is essentially a parallel downward shift as a function of the grading state index. A question pertinent to proper modelling of the particle breakage phenomenon is the existence of a final stable gradation. It is always possible to imagine a gradation where all voids are filled with progressively smaller and smaller particles (Fuller and Thompson, 1907) as emphasised by the works of McDowell and co-workers. Such a “fractal” gradation would be linear on a log-log plot of particle size and proportion finer than that size. Been et al. (1991), Konrad (1998), and Russell and Khalili (2004) by adopting the three part CSL framework implicitly assumed that a continual constant volume state will be achieved once the tests approach the second part of the CSL. The data presented by Russell and Khalili (2004) suggest that for higher stress levels (above 1 MPa) the loose tests do not reach a constant volume and continue to contract. Lade and    44 Yamamuro (1996) made the same observation on tests presented in Yamamuro and Lade (1996) and concluded that the critical state conditions can only be achieved at low pressures (the first part of CSL) or at extremely high pressures where particle breakage has ceased (the third part of CSL). An experimental study on a granitic soil by Lee and Coop (1995) suggested that the amount of particle breakage at the critical state is path independent and solely a function of the value of p on the CSL. In order to investigate whether the critical state can be achieved at higher stress levels, Luzzani and Coop (2002) used ring shear tests to take samples of a carbonate sand to more than 700% shear strain. They concluded that “the volumetric compression appeared to be directly related to the particle breakage, and the rate of increase of both reduced with increasing shear strain apparently tending towards constant values. It appears therefore that a constant volume will only be reached when the breakage ceases, but this will only be at strains about an order of magnitude greater than those achieved in our [their] work”. In a later attempt, Coop et al. (2004) used ring shear tests to take samples of a carbonate sand to up to 100,000% shear strain. They concluded that particle breakage continues to very large strains beyond those reached in triaxial tests, but a constant gradation is reached at very large strains. This constant gradation is dependent not only on the stress level but also on the uniformity and particle size of the original gradation. There have been a number of attempts at modelling the consequences of particle breakage in the past ten years (e.g. Daouadji et al., 2001; Indraratna and Salim, 2002; Russell and Khalili, 2002; Einav, 2007; and Muir Wood, 2007). Russell and Khalili (2002) incorporated the effect of particle breakage into a bounding surface plasticity model, by assuming a steepening of the CSL with increasing mean stress. The more plausible idea of a shifting CSL with the change in    45 material during breakage (Daouadji et al., 2001; and Muir Wood, 2007) has yet to be validated against experimental data.  1.5. Overview of the Proposed Research  Looking at the body of the work reported to date on interpretation of the in-situ density of cohesionless soils through CPT reveals that significant progress has been made in realistic modelling of the cone penetration process, and in performing calibration chamber tests. However, none of the works to date offer a reliable framework for interpreting in-situ density of the soil. The numerical models have usually failed to adequately capture the material behaviour. In addition, not enough attention has been paid to clear and independent determination of material parameters. Finally, systematic verification and validation of the analyses is rather scarce in the literature. Calibration chamber tests provide correlations between soil properties and CPT data but they are prohibitively expensive, limiting them to a small number of very large projects. The issue of different fabric and age between reconstituted chamber samples and in-situ conditions and variability of in-situ soils further limits the application of calibration chamber testing for engineering projects. In this research, an analytical framework feasible for most engineering programs is laid out for analysing the Cone Penetration Test in cohesionless soils. The framework has three important aspects: Independent calculation of soil parameters, realistic constitutive modeling of soil behaviour, and comprehensive verification and validation of the results.    46 A central feature of the framework is the application of a critical state constitutive model that adequately captures the behaviour of cohesionless materials over a range of material properties and states. Another important aspect of the analysis is independent and thorough calibration of the constitutive model to material behaviour using element tests. The influence of number and quality of the element tests on estimated material properties is investigated. Spherical cavity expansion analysis is chosen for modelling the cone penetration problem. This method has the advantage of being numerically stable and fast and is thus suitable for modelling a large number of tests. A unique shape function is identified for converting the spherical cavity expansion results to CPT tip resistance. Special attention has been directed towards verification of the analytical method against experimental data. A comprehensive database of calibration chamber tests from the literature is processed and used to verify the framework. The method is also validated against in-situ measurements of soil density by comparison to other testing methods (the Self-Bored Pressuremeter testing in this case) as well as direct measurements through ground freezing techniques. The confidence and accuracy in the interpretation results and ways of improving them are discussed. An important phenomenon pertinent to the cone penetration in cohesionless materials is particle breakage. However, it has been generally overlooked in the analyses of the CPT. One of the reasons has been a lack of understanding of the basics of the phenomenon. An extensive laboratory testing program is performed and the results are used to support a hypothesis for treating particle breakage within the critical state soil mechanics. The work provides a foundation for implementing the particle breakage in the analysis of the Cone Penetration Test.     47 The content is organised in the following order:  1. Chapter one (Introduction and Background) The problem of CPT interpretation in cohesionless soils is explained, different aspects of the problem and different solutions available in the literature are reviewed. A gap is identified in linking the experimental and analytical solutions, as well as theoretical understanding of the particle breakage phenomenon, which is pertinent to CPT interpretation. The need for a comprehensive theoretical framework verified against different experimental data is justified. The scope and structure of the dissertation is presented.  2. Chapter two (Determination of the Critical State Friction Angle from Triaxial Tests) The critical state friction angle is one of the parameters affecting the material behaviour and one of the most important ones influencing CPT resistance. Different methods of estimating this parameter from triaxial compression tests are compared and discussed.  3. Chapter three (Confidence and Accuracy in Determination of Critical State Friction Angle) A statistical investigation is performed to identify the effect of the number and quality of triaxial compression tests used to calculate critical state friction angle on the accuracy of the estimated parameter. Recommendations are made on the number and distribution of the tests necessary to achieve different levels of accuracy.  4. Chapter four (Interpretation of Sand State from CPT Tip Resistance) There is an inherent mismatch between measured CPT tip resistance and the limit cavity pressure obtained from    48 analysis. Calibration chamber data are presented for nine soils, comprising a range of soil types and material properties, for which triaxial testing is also available. Cavity expansion analysis is performed and a unique and unbiased relation between limit cavity expansion pressure and calibration chamber normalised tip resistance is identified, defined as the shape function. This approach recovered values of soil state with a precision close to that of the published calibration chamber data.  5. Chapter five (Evaluation of Soil State from SBP and CPT: A Case History) Validation of the interpretation framework requires knowledge of the ground truth. Comparing the data obtained from different in-situ testing methods performed in close proximity is one common way of assessing the reliability of either method. The chapter compares interpretations of the in-situ state parameter obtained from CPT and SBP testing performed off-shore in a uniform hydraulic fill. Extensive calibration chamber tests and high quality triaxial compression tests are available to calibrate the constitutive model. The SBP and CPT tests are independently analysed, considering the effects of fabric. The predictions of in-situ state for the fill from the CPT are close to those derived from the SBP. Although not proof of accuracy (validation) of either test, since ground truth is not known, the results lend support to the adequacy of the interpretation methodology used for both. Further improvements are discussed.  6. Chapter six (Interpretation of the In-situ Density from Seismic CPT in Fraser River Sand) The ideal validation for an interpretation framework would come from a comparison of the predicted state parameter with direct measurements of the in-situ state parameter. Such    49 direct measurements are usually obtained from void ratio measurements on undisturbed samples. This chapter describes an application of the CPT interpretation framework presented in Chapter 4 to obtain the state parameter from CPT tip resistances in the CANLEX (CANadian Liquefaction EXperiment) dataset. The CANLEX database provides a rare combination of reliable in-situ measurements and laboratory element testing on undisturbed and reconstituted samples, allowing for direct evaluation of the capability of the method in obtaining ground truth (validation). The effects of differing gradations and soil fabrics have been captured and reflected in the resultant state parameter interpretation. Accuracy is evaluated by comparison to in-situ density measurements and comparison to other published interpretation methods.  7. Chapter seven (Particle Breakage and the Critical State of Sand) Particle breakage has important implications for soil behaviour during shearing. In particular the change in a soil’s gradation during particle breakage changes its critical state, which is a fundamental input to the majority of advanced constitutive models. It is a phenomenon pertinent to the analysis of CPT due to the high stresses that develop at the cone tip during cone penetration, and can directly affect the interpretation process. This chapter provides a hypothesis on the particle breakage process at the particle level and expands its implications for the critical state of sands at higher stress levels. Experimental data are presented and the hypothesis is used to explain the observed behaviour. The data support the proposition that breakage shifts the CSL down in the void ratio-mean effective stress space without changing its slope. It is suggested that the void ratio shift in the CSL is equal to the reduction in void ratio due to breakage. It is also proposed that significant particle breakage in a particulate material does    50 not occur unless two conditions are satisfied: Capability of the materials for contraction merely by sliding and rolling of the particles is exhausted; and, a stress level threshold is surpassed. Finally it is speculated that particle breakage does not directly affect the evolution of soil state towards the critical state and is merely a factor working alongside dilatancy imposing additional compressibility on the soil.  8. Chapter eight (Summary and Conclusion) A summary of key findings and conclusions is presented along with the limitations of the methods and results. Potential future studies to further advance the knowledge in this field are also identified.  9. Appendix A The testing procedures and equipment are explained in detail and the laboratory data obtained as part of this research are presented.  10. Appendix B A bibliography of the manuscripts included in this thesis is presented.  This document has been prepared in accordance with UBC’s formatting principles for manuscript-based theses. A manuscript-based thesis is constructed around one or more related manuscripts. As a result, some information such as background information, numerical procedures and the constitutive model, and material descriptions is repeated in various chapters. The introduction provides background and context to the manuscripts. In this thesis, chapters two to seven are the individual manuscripts; it is a UBC requirement that at least one of them is appropriate for publication as a journal paper. The writer of the thesis can be either the sole author or a senior co-author of the manuscripts. The place of publication for the manuscripts is    51 provided as a footnote at the start of each chapter and a list of all the publications is provided in Appendix B. The concluding chapter summarises the main findings from the individual chapters in the context of interpretation of the in-situ state parameter in cohesionless soil using the CPT.    52 Table 1- 1 Summary of NorSand (Jefferies and Shuttle, 2005) Internal Model Parameters )ln( ppii     where  cee   tcitci MNMM *1  Critical State  pec  ln   AND  )(21 MNMCc MMM   where         sin3/61cos/33  tcMC MM     and       32 2 232 2 9 23 327 sin4 3sin9 83 327 tctc tc MNMN MN MM M MM M      Yield Surface and Internal Cap  ii p p M   ln1     with      tcitci Mp p       exp max   Hardening Rule On yield surface: p q i tc itc itc i i i p p Mp p M MH p p                   exp 2   On internal cap: p q tci i i i M MH p p  ,7.2     Stress Dilatancy  and  Plastic Strain Rate Ratios  ip MD    and itcipptc MMDD /, iteippte MMDD /,  define p tc p tc tc D D z 26 32 ,3     and  p te p te te D Dz 23 62 ,3                  ) 2 903cos(,3,3,3 1 3    tetctc zzz   define  3/)cos3(sin  a , 3/sin2 b , 3/)cos3(sin  c              )1/())1(1( 1 3 1 2 ppp bDcDaD           Elasticity p GIr    with     )21(3 )1(2     GK      53 Table 1- 2 NorSand soil parameters and values for sands (Jefferies and Shuttle, 2005) Parameter Typical Range Remark CSL  0.9 –1.4 Intercept of CSL, defined at 1 kPa  0.01 – 0.07 Slope of CSL, defined on base e Plasticity Value of ratio pq /  at the critical state.  Triaxial compression used as reference condition Mtc 1.2 – 1.5 N* 0 – 0.5 Volumetric coupling parameter H 50 – 500 Plastic hardening modulus for loading, often f() tc 2.5 – 4.5 Relates minimum dilatancy to i Elasticity Ir 100-800  Dimensionless shear rigidity  0.1 – 0.3 Poisson’s ratio, 0.2 commonly adopted    54 Table 1- 3 Current calibration chambers in the world, expanded after Ghionna and Jamiolkowski (1991) Boundary Conditions Calibration chamber sample diameter (m) sample height (m) Radial Bottom Top Reference Soil Type Cornell University, U.S.A.  2.10 2.90 Flexible Rigid Rigid Ghionna & Jamiolkowski, 1991 Country Roads Bureau, Australia 0.76 0.91 Flexible Flexible Rigid Ghionna & Jamiolkowski, 1991 ENEL-CRIS, Milan, Italy 1.20 1.50 Flexible Flexible Rigid Ghionna & Jamiolkowski, 1991 Ticino sand Golder Associates, Calgary 1.40 1.00 Flexible Rigid Flexible Ghionna & Jamiolkowski, 1991 Erksak sand, Syncrude Tailings ISMES, Bergamo, Italy 1.20 1.50 Flexible Flexible Rigid Ghionna & Jamiolkowski, 1991 Ticino sand Louisiana State University 0.55 0.82 Flexible Rigid Rigid Kim, 1999 67% fine sand and 33% Kaolin Monash University, Australia 1.20 1.80 Flexible Flexible Rigid Ghionna & Jamiolkowski, 1991 NCTU, Taiwan 0.79 1.60 Flexible* Rigid Flexible Hsu & Huang, 1998 Da Nang sand NCTU, Taiwan 0.51 0.76 Flexible Rigid Rigid Hsu & Huang, 1999 Mai-Liao sand North Carolina State University 0.94 1.00 Flexible Rigid Rigid Ghionna & Jamiolkowski, 1991 Norwegian Geotechnical Institute 1.20 1.50 Flexible Flexible Rigid Ghionna & Jamiolkowski, 1991 Oxford University, England 0.90 1.10 Flexible Flexible Rigid Ghionna & Jamiolkowski, 1991 Tokyo University, Japan 0.90 1.10 Flexible Rigid Rigid Ghionna & Jamiolkowski, 1991 UFRJ-COPPE,Rio 1.20 1.50 Flexible Flexible Rigid Ghionna & Jamiolkowski, 1991 University of California, Berkeley 0.76 0.80 Flexible Rigid Rigid Ghionna & Jamiolkowski, 1991 Monterey sand Clarkson University,USA 0.51 0.76 Flexible Rigid Rigid Ghionna & Jamiolkowski, 1991 University of Florida, Gainesville 1.20 1.20 Flexible Flexible Rigid Ghionna & Jamiolkowski, 1991 Reid Bedford, Ottawa and Hilton Mines tailings University of Grenoble, France 1.20 1.50 Flexible Flexible Flexible Ghionna & Jamiolkowski, 1991 University of Houston 0.76 2.54 Flexible Flexible Flexible Ghionna & Jamiolkowski, 1991 University of Newcastle, Australia 1.00 1.00 Flexible Flexible Rigid Ajalloeian & Yu, 1998 University of Oklahoma 0.61 0.45-1.42 Flexible Flexible Rigid Tan et el., 2003  Minco silt University of Sheffield, U.K.  0.79 1.00 Flexible Rigid Flexible Ghionna & Jamiolkowski, 1991 University of Texas at Austin 0.88 0.85 Flexible Rigid Flexible Ghionna & Jamiolkowski, 1991 Virginia Tech, Blacksburg 1.50 1.50 Flexible Rigid Rigid Ghionna & Jamiolkowski, 1991 Yatesville silty sand Waterways Experiment Station 0.80-3.00 variable Flexible Rigid Rigid Peterson & Arulmoli, 1991   * Can be controlled by a computer to simulate field conditions.    55 Table 1- 4 Boundary conditions in conventional calibration chamber and simulator tests (Huang and Hsu, 2005) Top and bottom boundary Lateral boundary Boundary Conditions   Stress Strain Stress Strain BC1 Constant - Constant - BC2 - 0 - 0 BC3 Constant - - 0 BC4 - 0 Constant - BC5 Constant - Servo-controlled    Table 1- 5 Summary of B values (Equation 1-13) proposed for different sands by Cudmani and Osinov (2001) SAND Ticino Monterey L. Buzzard Hokksund Toyoura B 1.3 2.3 3.5 4.0 4.0   56  Figure 1-1 Definition of state parameter   and overconsolidation ratio R (after Jefferies and Shuttle, 2005)  57  a) Very loose sand  b) Very dense sand Figure 1-2 Illustration of NorSand yield surfaces and limiting stress ratios (Jefferies and Shuttle, 2005)  58  Figure 1-3 The standard electrical cone penetrometer (ASTM D5778)   59  Figure 1-4 Contours of state parameter (Robertson, 2009)   Figure 1-5 Normalisation factors for calibration chamber size and boundary conditions (Been et al., 1987)    60    Figure 1-6 Schematic view of the Eulerean approach to the cone penetration problem (van den Berg et al., 1996)  61  Figure 1-7 Steady-State behaviour and boundary conditions (Yu et al., 2000)   62 1.6. References  Abu-Farsakh M.Y., Voyiadjis G.Z., and Tumay M.T. 1998. Numerical Analysis of the Miniature Piezocone Penetration Tests (PCPT) in Cohesive Soils. International Journal for Numerical and Analytical Methods in Geomechanics, 22: 791-818. Acar Y.B., and Tumay M.T. 1986. Strain Field around Cones in Steady Penetration. Journal of Geotechnical Engineering, 112(2): 207-213. Ahmadi M.M, and Robertson P.K. 2005. Thin-Layer Effects on the CPT qc Measurement. Canadian Geotechnical Journal, 42: 1302–1317. Ahmadi M.M. 2000. Analysis of Cone Tip Resistance in Sand. PhD thesis, University of British Columbia, Vancouver, Canada. Ahmadi M.M., Byrne P.M., and Campanella R.G. 2005. Cone Tip Resistance in Sand: Modeling, Verification, and Applications. Canadian Geotechnical Journal, 42: 977–993. Ajalloeian R., and Yu H.S. 1998. Chamber Studies of the Effects of Pressuremeter Geometry on Test Results in Sand. Géotechnique, 48(5): 621-636. ASTM D5778 - 07 . Standard Test Method for Electronic Friction Cone and Piezocone Penetration Testing of Soils. Published : November 2007. Baldi G., Bellotti R., Ghionna V., Jamiolkowski M., and Pasqualini E. 1982. Design Parameters for Sands from CPT. Proceedings of the 2nd European Symposium on Penetration Testing, Amsterdam, 2: 425-432. Baldi G., Bellotti R., Ghionna V., Jamiolkowski M., and Pasqualini E. 1986. Interpolation of CPTs and CPTUs part 2, Drained-Penetration of Sands. Proceedings of the 4th  63 International Geotechnical Seminar. Field instrumentations and in situ measurements, Singapore, 143-156. Baligh M.M., 1985. Strain Path Method. Journal of Geotechnical Engineering, ASCE, 111: 1108–1136. Been K., and Jefferies M.G. 1985. A State Parameter for Sands. Géotechnique, 35(2): 99-112. Been K., and Jefferies M.G. 1992. Towards Systematic CPT Interpretation. Proceedings of Wroth Memorial Symposium, Thomas Telford, London, 121-134. Been K., Crooks J.H.A., and Rothenburg L. 1988. A Critical Appraisal of CPT Calibration Chamber Tests. Proceedings of the 1st International Symposium on Penetration Testing (ISOPT), 2: 651-660. Been K., Crooks J.H.A., Becker, D.E., and Jefferies M.G. 1986. The Cone Penetration Test in Sands. Part I: State Parameter Interpretation. Géotechnique, 36(2): 239-249. Been K., Jefferies M.G., Crooks J.H.A., and Rothenburg L. 1987. The Cone Penetration Test in Sands. Part II: General Inference of State. Géotechnique, 37(3): 285-299. Been K., Jefferies M.G., and Hachey J.E. 1991. The Critical State of Sands. Géotechnique, 41(3): 365-381. de Beer E.E. 1963. The Scale Effect in the Transposition of the Results of Deep-Sounding Tests on the Ultimate Bearing Capacity of Piles and Caisson Foundations. Géotechnique, 13(1): 39-75. Bellotti R., Bizzi G., and Ghionna V. 1982. Design, Construction, and Use of a Calibration Chamber. Proceedings ESOPT II, Balkema, Amsterdam, Netherlands, 2: 439-446. van den Berg P. 1994. Analysis of Soil Penetration. PhD Thesis, Delft University of Technology, ISBN 90-407-1004-X.  64 van den Berg P., de Borst R., and Huetink H. 1996. An Eulerean Finite Element Model for Penetration in Layered Soil. International Journal of Numerical and Analytical Methods in Geomechanics, 20: 865-886. Billam J. 1971. Some Aspects of the Behaviour of Granular Materials at High Pressures. Proceedings of the Roscoe Memorial Symposium, R.H. Parry (ed.), Henley-on-Thames: Foulis, 69-80. Bishop A.W. 1966. Strength of Soils as Engineering Materials. 6th Rankine Lecture, Géotechnique, 16: 89-130. Bishop R.F., Hill R., and Mott N.F. 1945. Theory of Indentation and Hardness Tests. Proceedings of the Physical Society, 57: 147–159. Bolton M.D. 1986. The Strength and Dilatancy of Sands. Géotechnique 36(1): 65-78. de Borst R., and Vermeer P.A. 1982. Finite Element Analysis of Static Penetration Tests. Proceedings of the 2nd European Symposium on Penetration Testing, Amsterdam, Netherlands, 457-562. Bridgman P.W. 1918. The Failure of Cavities in Crystals and Rocks under Pressure. American Journal of Science, 45: 243-268. Broms B.B. and Flodin N. 1988. History of Soil Penetration Testing. Proceedings of the International Symposium on Penetration testing, Penetration Testing 1988, ISOPT-1, Orlando, Vol. 1, J. de Ruiter (ed.), Balkema, Rotterdam, 157-220. Campanella R.G., Robertson P.K., and Gillespie D. 1983. Cone Penetration Testing in Deltaic Soils. Canadian Geotechnical Journal, 20:23-35. Carter J.P., Booker J.R., and Yeung S.K. 1986. Cavity Expansion in Cohesive Frictional Soils. Géotechnique, 36(3): 349–358.  65 Chadwick P. 1959. Quasi-Static Expansion of Spherical Cavity in Metals and Ideal Soils. The Quarterly Journal of Mechanics and Applied Mathematics, 12(1): 52-71. Cividini A., and Gioda G. 1988. A Simplified Analysis of Pile Penetration. Proceedings of the 6th International Conference on Numerical Methods in Geomechanics, Balkema, Rotterdam, Netherlands, 1043-1049. Clayton C.R.I., Hababa M.B., and Simons N.E. 1985. Dynamic Penetration Resistance and the Prediction of the Compressibility of a Fine-Grained Sand – A Laboratory Study. Géotechnique, 35: 19-31. Collins I. F., and Yu H. S. 1996. Undrained Cavity Expansions in Critical State Soils. International Journal of Numerical and Analytical Methods in Geomechanics, 20(7): 489–516. Coop M.R., Sorensen K.K., Bodas Freitas T., and Georgoutsos G. 2004. Particle Breakage during Shearing of a Carbonate Sand. Géotechnique, 54(3): 157-163. Cudmani R., and Osinov V.A. 2001. The Cavity Expansion Problem for the Interpretation of Cone Penetration and Pressuremeter Tests. Canadian Geotechnical Journal, 38: 622-638. Cundall P.A., Strack O.D.L. 1979. A Discrete Numerical Model for Granular Assemblies. Géotechnique, 29: 47-65. Daouadji A., Hicher P. Y., and Rahma A. 2001. An Elastoplastic Model for Granular Materials Taking into Account Grain Breakage. European Journal of Mechanics - A/Solids, 20: 113–137. Drucker D.C., Gibson R.E., and Henkel D.J. 1957. Soil Mechanics and Work Hardening Theories of Plasticity. Transactions of the American Society of Civil Engineers, 122, ASCE, Paper No. 2864, 338–346.  66 Durgunoglu H.T., and Mitchell J.K. 1975. Static Penetration Resistance of Soils, I-Analysis, II- Evaluation of Theory and Implications for Practice. Proceedings of the Specialty Conference in In-Situ Measurements of Soil Properties, ASCE, Raleigh, N.C., 1: 151- 189. Einav I. 2007. Breakage Mechanics – Part II: Modelling Granular Materials. Journal of the Mechanics and Physics of Solids, 55: 1298-1320. Foray P. 1991. Scale and Boundary Effects on Calibration Chamber Pile Tests. Proceedings of the 1st International Symposium on Calibration Chamber Testing, Potsdam, New York, 147-160. Foster W.A., Johnson C.E., Chiroux R.C., and Way T.R. 2005. Finite Element Simulation of Cone Penetration. Applied Mathematics and Computation, 162: 735-749. Fuller W., and Thompson S.E. 1907. The laws of proportioning concrete. Transactions of the American Society of Civil Engineers. Paper no 1053: 67-143. Ghionna V.N., and Jamiolkowski M. 1991. A Critical Appraisal of Calibration of Calibration Chamber Testing of Sands. Proceedings of the 1st International Symposium on Calibration Chamber Testing, ISOCCT 1, Potsdam, New York, 13-39. Gill D.R., and Lehane B.M. 2000. Extending the Strain Path Method Analogy for Modelling Penetrometer Installation. International Journal of Numerical and Analytical Methods in Geomechanics, 24: 477-489. Griffiths D.V. 1982. Elasto-Plastic Analysis of Deep Foundations in Cohesive Soil. International Journal of Numerical and Analytical Methods in Geomechanics, 6: 211- 218.  67 Hardin B.O. 1985. Crushing of Soil Particles. ASCE, Journal of Geotechnical Engineering, 111(10): 1177-1192. Hill R. 1950. The Mathematical Theory of Plasticity. Oxford University Press, Oxford. Hofmann B.A., Sego D.C., and Robertson P.K. 2000. In Situ Ground Freezing to Obtain Undisturbed Samples of Loose Sand. Journal of Geotechnical and Geoenvironmental Engineering, ASCE, 126(11): 979-989. Holden J.C. 1991. History of the First Six CRB Calibration Chambers. Proceedings of the 1st International Symposium on Calibration Chamber Testing, ISOCCT 1, Potsdam, New York, 1–12. Houlsby G.T., Wheeler A.A., and Norbury J. 1985. Analysis of Undrained Cone Penetration as a Steady Flow Problem. Proceedings of the 5th International Conference on Numerical Methods in Geomechanics, 4: 1767–1773. Houlsby G.T., and Hitchman R. 1988. Calibration Chamber Tests of a Cone Penetrometer in Sand. Géotechnique, 38(1): 39–44. Houlsby G.T., and Hitchman R. 1989. Calibration Chamber Tests of a Cone Penetrometer in Sand. Discussion. Géotechnique, 39(4): 727–731. Hsu H.H., and Huang A.B. 1998. Development of an Axisymmetric Field Simulator for Cone Penetration Tests in Sand. Geotechnical Testing Journal, 21(4): 348-355. Hsu H.H., and Huang A.B. 1999. Development Calibration of Cone Penetration Test in Sand. Proceedings of the National Science Council. ROC(A), 23(5): 579-590. Huang W., Sheng D., Sloan S.W., and Yu H.S. 2004. Finite Element Analysis of Cone Penetration in Cohesionless Soil. Computers and Geotechnics, (31): 517–528.  68 Huang A.B. 1989. Strain-Path Analyses for Arbitrary Three-Dimensional Penetrometers. International Journal of Numerical and Analytical Methods in Geomechanics, 13: 551- 564. Huang A.B., and Hsu H.H. 2004. Advanced Calibration Chambers for Cone Penetration Testing in Cohesionless Soils. Proceedings of ISC-2 on Geotechnical and Geophysical Site Characterization, Viana da Fonseca and Mayne (eds.), 147-166. Huang A.B., and Hsu H.H. 2005. Cone Penetration Tests under Simulated Field Conditions. Géotechnique, 55(5): 345-354. Huang A.B., and Ma M.Y. 1994. An Analytical Study of Cone Penetration Tests in Granular Material. Canadian Geotechnical Journal, 31(1): 91–103. Indraratna B., and Salim W. 2002. Modelling of Particle Breakage of Coarse Aggregates Incorporating Strength and Dilatancy. Proceedings of the Institution of Civil Engineers, Geotechnical Engineering, 144(4): 243-252. Iqbal M.S. 2004. Discrete Element Modeling of Cone Penetration Testing in Coarse Grain Soils. M.S. thesis, Civil and Environmental Engineering, University of Alberta, Edmonton. Ishihara K., Tatsuoka F., and Yasuda S. 1975. Undrained Deformation and Liquefaction of Sand Under Cyclic Stresses. Soil and Foundations, 15(1): 29-44. Jamiolkowski M., Ladd C.C., Germaine J.T., and Lancellotta R. 1985. New Developments in Field and Laboratory Testing of Soils. Proceedings of the 11th ICSMFE, San Francisco, CA, 1: 54-154. Jefferies M.G. 1993. Nor-sand: a Simple Critical State Model for Sand. Géotechnique, 43(1): 91-103.  69 Jefferies M.G., and Been K. 2000. Implications for Critical State Theory from Isotropic Compression of Sand. Géotechnique, 50(4): 419-429. Jefferies M.G., and Shuttle D.A. 2005. NorSand: Features, Calibration and Use. Proceedings of the Specialty Conference on Soil Constitutive Models: Evaluation, Selection, and Calibration, ASCE Geotechnical Special Publication 128: 204-236. Jefferies M.G., and Shuttle D.A. 2002. Dilatancy in General Cambridge-Type Models. Géotechnique, 52(9): 625-638. Kim D. 1999. Numerical Simulation and Experimental Verification of Cone Penetration Rate and Anisotropy in Cohesive Soil. PhD Thesis, Department of Civil Engineering, Louisiana State University. Kiousis P.D., Voyiadjisa G.Z., and Tumay M.T. 1988. A Large Strain Theory and its Application in the Analysis of the Cone Penetration Mechanism. International Journal of Numerical and Analytical Methods in Geomechanics, 12: 45-60. Konrad J.-M. 1997. In Situ Sand State from CPT: Evaluation of a Unified Approach at Two CANLEX Sites. Canadian Geotechnical Journal. 34(1): 120–130. Konrad J.-M. 1998. Sand State from Cone Penetrometer Tests: A Framework Considering Grain Crushing Stress. Géotechnique, 48(2): 201–215. Koumoto T. 1988. Ultimate Bearing Capacity of Cones in Sand. Proceedings of the 1st International Symposium on Penetration Testing, ISOPT-1, Orlando, Florida, De Ruiter (ed.), Balkema, Rotterdam, 809-813. Ladanyi B. 1963. Expansion of a Cavity in a Saturated Clay Medium. Journal of the Soil Mechanics and Foundations Division, ASCE, 89(4): 127–161.  70 Lade P.V., and Yamamuro J.A. 1996. Undrained Sand Behaviour in Axisymmetric Tests at High Pressures. Journal of Geotechnical Engineering, 122(2): 120-129. Lee I.K., and Coop M.R. 1995. Intrinsic Behaviour of a Decomposed Granite Soil. Géotechnique, 45(1): 117-130. Lee K.L., and Farhoomand I. 1967. Compressibility and Crushing of Granular Soil in Anisotropic Triaxial Compression. Canadian Geotechnical Journal, 4(1): 68-86. Lu Q., Randolph M.F., Hu Y., and Bugarski I.C. 2004. A Numerical Study of Cone Penetration in Clay. Géotechnique, 54(4): 257–267. Lunne T., and Christophersen H.P. 1983. Interpretation of Cone Penetration Data for Offshore Sands. Proceedings of the 15th Annual Offshore Technology Conference, Houston, Texas, 181-192. Lunne T., Robertson P.K., and Powell J.J.M. 1997. Cone Penetration Testing in Geotechnical Practice. Routledge, Blackie Academic & Professional. London. Luzzani L., and Coop M.R. 2002. On the Relationship between Particle Breakage and the Critical State of Sands. Soils and Foundations, 42(2): 71-82. Mabsout M.E., Reese L.C., and Tassoulas J.L. 1995. Study of Pile Driving by Finite-Element Method. Journal of Geotechnical Engineering, ASCE, 121(7): 535-543. Matsuoka H., and Nakai T. 1974. Stress-Deformation and Strength Characteristics of Soil under Three Different Principal Stresses. Trans. JSCE, 6: 108-109. Mayne P.W., and Kulhawy F.H. 1991. Calibration Chamber Database and Boundary Effects Correction for CPT Data. Proceedings of the 1st International Symposium on Calibration Chamber Testing, ISOCCT 1, Potsdam, New York, 257-264.  71 McDowell G.R., Bolton M.D., and Robertson D. 1996. The Fractal Crushing of Granular Materials. Journal of the Mechanics and Physics of Solids, 44(12): 2079-2101. Meyerhof G.G. 1951. The Ultimate Bearing Capacity of Foundations. Géotechnique, 2(4): 301-332. Miura N., and O-Hara S. 1979. Particle-Crushing of a Decomposed Granite Soil under Shear Stresses. Soils and Foundations, 19(3): 1-14. Muir Wood D. 2007. The Magic of Sands. The 20th Bjerrum lecture presented in Oslo, 25 November 2005, Canadian Geotechnical Journal, 44: 1329-1350. Muir Wood D., and Maeda K. 2008. Changing Grading of Soil: Effect on Critical States. Acta Geotechnica, 3: 3-14. Nash D.F.T., Powell J.J.M., and Lloyd I.M. 1992. Initial Investigations of the Soft Clay Test Site at Bothkennar. Géotechnique, 42: 163-181. Parkin A., and Lunne T. 1982. Boundary Effects in the Laboratory Calibration of a Cone Penetrometer in Sand. Proceedings of ESOPT II, Balkema, Amsterdam, The Netherlands, 2: 761-768. Parkin A.K. 1988. The Calibration of Cone Penetrometers. Proceedings of the 1st International Symposium on Penetration Testing, ISOPT-1, Orlando, Florida, Vol.1, pp.221-243. Pestana J.M., and Whittle A.J. 1995. Compression Model for Cohesionless Soils. Géotechnique, 45: 611-631. Peterson R.W., and Arulmoli K. 1991. Overview of a Large Stress Chamber System. Proceedings of the 1st International Symposium on Calibration Chamber Testing, Potsdam, New York, 329-338. Plewes H.D., Davies M.P., and Jefferies M.G. 1992. CPT Based Screening Procedure for Evaluating Liquefaction Susceptibility. Proceedings of the 45th Canadian Geotechnical  72 Conference, Toronto, Ont., 26-28 Oct. 1992, Canadian Geotechnical Society, Alliston, Ont., 4: 1-9. Plewes H.D., Pillai V.S., Morgan M.R., and Kilpatrick B.L. 1994. In Situ Sampling, Density Measurements, and Testing of Foundation Soils at Duncan Dam. Canadian Geotechnical Journal, 31: 927–938. Robertson P.K. 2009. Interpretation of Cone Penetration Tests – a Unified Approach. Canadian Geotechnical Journal, 46(11): 1337-1355. Robertson P.K., and Campanella R.G. 1983a. Interpretation of Cone Penetration Tests. Part I: Sand. Canadian Geotechnical Journal, 20(4): 718-733. Robertson P.K., and Campanella R.G. 1983b. Interpretation of Cone Penetration Tests. Part II: Clay. Canadian Geotechnical Journal, 20(4): 734-745. Robertson P.K., and Campanella R.G. 1986. Guidelines for the Use and Application of the CPT and CPTU. Soil Mechanics series No. 105, Department of Civil Engineering, The University of British Columbia, 187 p. Robertson P.K., and Campanella R.G., Gillespie D., and Greig J. 1986. Use of Piezometer Cone Data. In Use of In Situ Tests in Geotechnical Engineering, Proceedings of In Situ ’86, a Specialty Conference sponsored by the Geotechnical Engineering Division of the American Society of Civil Engineers, S.P. Clemence (ed.), American Society of Civil Engineers, Geotechnical special publication No. 6, 1263-1280. Robertson P.K., and Wride C.E. 1998. Evaluating Cyclic Liquefaction Potential Using the Cone Penetration Test. Canadian Geotechnical Journal, 35(3): 442-459.  73 Russell A.R., and Khalili N. 2002. Drained Cavity Expansion in Sands Exhibiting Particle Crushing.  International Journal of Numerical and Analytical Methods in Geomechanics, 26(4): 323–340. Russell A.R., and Khalili N. 2004. A Bounding Surface Plasticity Model for Sands Exhibiting Particle Crushing. Canadian Geotechnical Journal, 41(6): 1179-1192. Sagaseta C., Whittle A.J., Santagata M., 1998. Deformation Analysis of Shallow Penetration in Clay. International Journal of Numerical and Analytical Methods in Geomechanics, 21(10): 687 – 719. Salgado R. 1993. Analysis of Penetration Resistance in Sands. PhD thesis, Department of Civil Engineering, University of California at Berkeley. Salgado R., Mitchell J.K., and Jamiolkowski M. 1997. Cavity Expansion and Penetration Resistance in Sand. Journal of Geotechnical and Geoenvironmental Engineering, 123(4): 344-354. Salgado R., Mitchell J.K., and Jamiolkowski M. 1998. Calibration Chamber Size Effects on Penetration Resistance in Sand. Journal of Geotechnical and Geoenvironmental Engineering., ASCE, 124(9): 878-888. Schmertmann J.H. 1991. The Mechanical Ageing of Soils. Journal of Geotechnical Engineering, ASCE, 117(12): 1288-1330. Schnaid F. 1990. A Study of the Cone-Pressuremeter Test in Sand. PhD thesis, Magdalen College, University of Oxford, Oxford. Schnaid F., and Yu H.S. 2007. Interpretation of the seismic cone test in granular soils. Géotechnique, 57(3): 265-272. Schofield A.N., and Wroth C.P. 1968.  Critical State Soil Mechanics. McGraw-Hill.  74 Sheng D., Eigenbrod K.D., and Wriggers P. 2005. Finite Element Analysis of Pile Installation Using Large-Slip Frictional Contact. Computers and Geotechnics, 32: 17–26. Shuttle D.A., and Jefferies M.G. 1998. Dimensionless and Unbiased CPT Interpretation in Sand. International Journal of Numerical and Analytical Methods in Geomechanics, 22: 351-391. de Simone P., and Golia G. 1988. Theoretical Analysis of the Cone Penetration Test in Sands. Proceedings, 1st International Symposium on Penetration Testing, ISOPT-1, Orlando, Florida, de Ruiter (ed.), Balkema, Rotterdam, 729-753. Sisson R.C. 1990. Lateral Stresses on Displacement Penetrometers. PhD thesis, University of California at Berkeley. Sladen J.A. 1989. Problems with Interpretation of Sand State from Cone Penetration Test. Géotechnique, 39(2): 323-332. Sokolovskii V.Y. 1965. Statics of Granular Media. Pergamon Press, Tarrytown, N.Y. Susila E., and Hryciw R.D. 2003. Large Displacement FEM Modelling of the Cone Penetration Test (CPT) in Normally Consolidated Sand. International Journal for Numerical and Analytical Methods in Geomechanics, 27: 585-602. Tan N.K., Miller G.A., and Muraleetharan K.K. 2003. Preliminary Laboratory Calibration of Cone Penetration in Unsaturated Silt. Proceedings of the 12th Pan American Conference on Soil Mechanics and Geotechnical Engineering, Cambridge, Massachusetts, 1: 391-396. Tannant D.D., and Wang C. 2002. PFC Model of Wedge Penetration into Oil Sands. Discrete Element Methods: Numerical Modeling of Discontinua. Proceedings of the Third International Conference, Santa Fe, September 2002, Reston, Virginia, ASCE, 311-316. Tavenas F.A. 1973. Difficulties in the Use of Relative Density as a Soil Parameter. Evaluation of Relative Density and its Role in Geotechnical Projects Involving Cohesionless Soils,  75 American Society for Testing and Materials, Selig E.T., and Ladd R.S. (eds.), Philadelphia, ASTM Special Technical Publication 523: 478-483. Teh C.I., and Houlsby G.T. 1991. An Analytical Study of the Cone Penetration Test in Clay. Géotechnique, 41: 17-34. Terzaghi K. 1943. Theoretical Soil Mechanics. John Wiley & Sons, New York, N.Y. Vesic A.S. 1963. Bearing Capacity of Deep Foundations in Sand. Highway Research Board, Record No. 39, Stresses in Soils and Layered System, 112-153. Vesic A.S. 1972. Expansion of cavities in infinite soil mass, Journal of Soil Mechanics and Foundations Division, ASCE, 98(SM3): 265-290. Vesic A.S. 1977. Design of pile foundations. National Research Council, Washington, D.C. Vesic A.S., and Clough G.W. 1968. Behavior of Granular Materials under High Stresses. Journal of the Soil Mechanics and Foundations Division, American Society of Civil Engineers, 94 (SM3): 661-688. Voyiadjisa G.Z., and Kim D. 2003. Finite Element Analysis of the Piezocone Test in Cohesive Soils Using an Elastoplastic–Viscoplastic Model and Updated Lagrangian Formulation. International Journal of Plasticity, 19: 253–280. Wei L., Abu-Farsakh M.Y., and Tumay M.T. 2005. Finite-Element Analysis of Inclined Piezocone Penetration Test in Clays. International Journal of Geomechanics, ASCE, 5(3): 167-178. Whittle A.J. 1992. Constitutive Modelling for Deep Penetration Problems in Clay. Proceedings of the 3rd International Conference on Computational Plasticity: Fundamentals and Applications, 2: 883–894. Willson S.M. 1985. Finite Element Analysis of Piles and Penetrometers. PhD thesis, University of Manchester, UK.  76 Willson S.M., Ims B.W., and Smith I.M. 1988. Finite Element Analysis Of Cone Penetration. Penetration testing in the UK, Thomas Telford, London, 157-159. Woods R.D., Benoît J., and deAlba P. 1994. National Geotechnical Experimentation Sites. Geotechnical News, 12(1): 39-44. Wride (Fear) C.E., Hofmann B.A., Sego D.C., Plewes H.D., Konrad J.-M, Biggar K.W., Robertson P.K., and Monahan P.A. 2000. Ground Sampling Program at the CANLEX Test Sites. Canadian Geotechnical Journal, 37(3): 530–542. Yamamuro J.A., and Lade P.V. 1996. Undrained Sand Behaviour in Axisymmetric Tests at High Pressures. Journal of Geotechnical Engineering, 122(2): 109-119. Yang Z.X., Jardine R.J., Zhu B.T., Foray P., and Tsuha C.H.C. 2010. Sand Grain Crushing and Interface Shearing During Displacement Pile Installation in Sand. Géotechnique, 60(6): 469-482. Yu H.S., and Houlsby G.T. 1991. Finite Cavity Expansion in Dilatant Soils: Loading Analysis. Géotechnique, 41(2): 173–183. Yu H.S., and Mitchell J.K. 1998. Analysis of Cone Resistance: Review of Methods. Journal of Geotechnical and Geoenvironmental Engineering, ASCE, 124(2): 140-149. Yu H.S., Schnaid F., and Collins I.F. 1996. Analysis of Cone Pressuremeter Tests in Sands. ASCE, Journal of Geotechnical Engineering, 122(8): 623-632. Yu H.S., Herrmann L.R., and Boulanger R.W. 2000. Analysis of Steady Cone Penetration in Clay. ASCE, Journal of Geotechnical and Geoenvironmental Engineering, 126(7): 594-605. Zohrabi M., Sibbald A. and Fairfield C.A. 1995. Boundary Conditions and Chamber Size Effects on CPT. Proceedings of the International Symposium on Cone Penetration Testing, Linkoping, Sweden, 2: 363–368.  77 Chapter 2.    Determination of the Critical State Friction Angle from Triaxial Tests2    2.1. Introduction  The idea that soils sheared to very high values of strain will eventually reach a constant void ratio, , and friction angle, c, is well established in soil mechanics. This constant volume state, usually termed the critical state, is also inherent in many, if not most, advanced constitutive soil models, including the well known critical state soil models Cam Clay and Modified Cam Clay. ce Geotechnical predictions are sensitive to the assumed shear strength parameters. Hence it is important that geotechnical engineers have access to good parameter estimates. However, despite the importance of the critical state to understanding soil behaviour, measurement of c  2 A version of this chapter has been published. Ghafghazi M., and Shuttle D.A. 2006. Accurate Determination of the Critical State Friction Angle from Triaxial Tests. Proceedings of the 59th Canadian Geotechnical Conference, Vancouver, 278-284.  78 remains problematic. This is particularly the case in engineering practice, where typically only a limited number of soil tests are available. This chapter describes four methods from the literature to obtain the critical state friction ratio in triaxial conditions. The accuracy of each of these methods is discussed based on previously published data for Ticino sand taking into consideration the limited number of tests typically available in practice. Then the two most promising methods are used to determine shear strength parameters for Erksak sand, to investigate the repeatability of the methods. Finally recommendations are made on how to acquire the most accurate parameters from a limited amount of data.  2.2. Definition of the Critical State  The critical state (also called steady state) was defined by Roscoe et al. (1958) as the state at which a soil "continues to deform at constant shear stress and constant void ratio". Poulos (1981) gives a more precise definition as "the steady state of deformation for any mass of particles is that state in which the mass is continuously deforming at constant volume, constant normal effective stress, constant shear stress and constant velocity". Writing these definitions in a logical form we have:    qvvppqeC    00,, 0         [Eq. 2 - 1]  Where C( ) is the function defining the CSL, e is the void ratio, q is the deviatoric stress invariant, p′ is the mean effective stress and v and q are volumetric and deviatoric strain  79 invariants respectively. The above equation includes two important conditions: first, the volumetric strain rate must be zero; second, the rate of change of this strain rate must also be zero. Hence, both dilatancy and rate of change of dilatancy must be zero during shearing at the critical state. There are no strain rate terms in C( ), making the CSL identical to the steady state of Poulos (1981). Constant mean stress is invoked in the equation to avoid a less easily understood definition for the situation in which mean stress is increased while the soil is continuously sheared at the critical state (Jefferies, 1993). These definitions can be used to infer the existence of a unique Critical State Locus (not necessarily linear) in  space. This line is the locus of end points of state paths ( ) for different tests sheared to large strain values. Many authors have presented data on the existence of a unique Critical State Locus (e.g. Castro, 1969; Been and Jefferies, 1985; Been et al., 1991; Vaid and Sasitharan, 1992; Garga and Sedano, 2002). pe  log pe  log Equation 2-1 also requires the existence of a unique critical state locus in p  ace. Since the models are cast in terms of stress invariants, and the relation between these invariants needs to be expressed for the critical state, it is convenient to replace the critical state friction angle with a parameter which is directly related to the stress invariants. The convention is to introduce a critical stress ratio, M, such that at the critical state: q  sp  pMq            [Eq. 2 - 2]  Constant M does not imply a constant friction angle and experimental data suggest that constant M yields unrealistic friction angles in general 3-D stress space (e.g. Bishop, 1966; Wanatowski and Chu, 2006). A constant friction angle may be obtained by treating M as a function of the  80 intermediate principal stress, represented by the lode angle, . Friction angle is also likely not constant with  (Jefferies and Shuttle, 2002). In this work triaxial compression conditions are taken as the reference case for which soil properties are determined. Thus, Mtc becomes the soil property (where subscript 'tc' denotes triaxial compression), and M( ) is evaluated in terms of this property. For known stress conditions the friction angle is directly related to stress ratio and the two parameters can be applied interchangeably; for example, in triaxial compression we have  c c tcM   sin3 sin6           [Eq. 2 - 3]  This chapter only considers triaxial compression conditions.  2.3. Stress-Dilatancy Definitions  The stress-dilatancy plot, which consists of the stress ratio pq  versus plastic dilatancy,  is very useful in understanding the behaviour of soils. When plotting data from triaxial tests, to reduce the noise in the raw data a central-difference approach is employed to compute . That is, p q p v pD   / pD  p jq p jq p jv p jvp jD 1,1, 1,1,                [Eq. 2 - 4]   81 where j subscript means that the value of parameter at jth data point is considered. Stress-dilatancy is a concept regarding plastic strain rates, and thus, Equation 2-4 is not itself sufficient to reduce test data, as there is an elastic component of strain rate. The plastic dilatancy is estimated as         Gqq KppD jjjqjq jjjvjvpq p vp j 3111,1, 111,1,                 [Eq. 2 - 5]  where K is the elastic bulk modulus and G the elastic shear modulus.  2.4. Determination of the Critical State Friction Ratio  The critical state of soils is usually achieved at very large strains. The widely used triaxial testing apparatus typically cannot achieve the strains required to get to the critical state, and after the peak stress is reached, shear banding may occur making the measured stress ratio unreliable. This has resulted in many researchers attempting to obtain critical state shear strength properties using peak and pre-peak triaxial data.  2.4.1. End of Test (ET) Method  A direct way of obtaining Mtc from soil samples is to plot the stress ratio, pq  , versus deviatoric strain,q ,for a drained test sheared up to about 20% of strain. The final stress ratio of such a test is taken as Mtc for that test. Figure 2-1 shows the stress ratio versus deviatoric strain  82 from a drained triaxial test on a dilatant specimen of Ticino sand. Mtc obtained using this method is denoted by (Mtc)ET in this work; where the ET subscript stands for "end of test". The major problem with this method is that the sample has likely not reached the critical state. We are also dealing with triaxial data obtained at relatively large strains, where localisation may occur. As shown in Figure 2-1, at the end of the test neither the stress ratio nor the volumetric strain has reached a constant value. Hence this dense sample has not yet reached the critical state.  2.4.2. Maximum Contraction (MC) Method  Maximum contraction during shearing of a soil sample is identified as the point where the sample reaches its minimum volume as illustrated in Figure 2-1. At this point only one of the conditions of critical state, 0v  in Equation 2-1 is satisfied. Negussey et al. (1988) suggested obtaining the critical state friction angle c from mobilised friction angle at maximum contraction based on data including ring shear tests on Ottawa sand, two tailings sand, granular copper, lead shot and glass beads. Mtc parameter obtained from this method will be denoted by (Mtc)MC in the following; where MC subscript stands for "maximum contraction".  2.4.3. Bishop Method (BM)  Bishop (1972) suggested a method of obtaining the critical state friction angle using the results of drained tests on dense samples at varying densities. For each test the value of peak dilatancy  83 Dmin at peak strength peak  is computed, making use of the fact that at peak strength the elastic strain increment is zero. In elasticity strain increments are linearly related to stresses. At peak the stresses change from increasing (elastic compression) to decreasing (elastic expansion), and correspondingly the elastic strain rate changes from positive to negative giving an instantaneous zero elastic strain increment. With zero elastic component, the total and plastic strain increments are identical. This method relies on the experimental observation that dilatancy, and hence peak friction angle, increase with density. And for a sand that reaches the critical state directly, i.e. without any dilation, the peak stress ratio corresponds with the critical state where tcM and . Mtc can then be determined by extrapolation to zero dilatancy. For convenience, we plot max instead of 0min D peak  and the corresponding dilation rate (peak dilatancy is  because of the compression-positive convention). Figure 2-2 shows the application of this method in 0minD minmax D  space applied to Ticino sand. The slope of this line, ( N1 ), has been used by some researchers as a material parameter (Nova 1982; Jefferies, 1993); N is called the volumetric coupling parameter.  2.4.4. Stress-Dilatancy (SD) Method  By definition the stable stress ratio at which no volume change occurs, 0D  remains there with continuing shearing, 0D e stress ratio at the critical state. Most triaxial tests do not reach the critical state within the strain limitations of the apparatus. As discussed earlier, localisation may occur post peak making the measured stress ratio unreliable. But it is possible , and  is th to infer Mtc portion of a stress ratio versus dilatancy plot to zero dilatancy (the vertical ax ,  by extrapolating the post-peak is where 0D ).  84 The post-peak (hook) portion of the graph is usually linear, and this extrapolation to the critical state has been done by assuming th ineat a l ar trend continues to the critical state. The line drawn n may happen as soon as the sample strains beyond peak strength e more accurate the extrapolation will be. In the case where the ritical State Stress Ratio of Ticino Sand ular sand. A summary of its roperties and the test program used here is available in Been et al. (1987) and Bellotti et al. through the stress-dilatancy plot in Figure 2-3 illustrates how this method is applied to a triaxial test on a dense sample. Post peak data should be used with significant caution. From an experimental viewpoint, shear banding and localisatio towards critical state. At larger strains (above 15%) other problems such as a higher effect of membrane penetration, tilting or bulging of the sample can occur; hence the data should be looked at very cautiously. In this method the closer the test gets to the critical state, without developing shear bands, or other sources of error, th dilatancy path deviates from the linear trend approaching the end of the test, it is recommended that the initial part of the hook gets a higher weight in determining the location of the extrapolation line, and the second part be discarded. In this work the slope of this line is identified as ( *1 N ).  2.5. Obtaining the C  Ticino sand is a medium to coarse predominantly quartz sub-ang p (1996). Data presented here are from tests performed in 1987 at Golder Associates' Calgary laboratory. All samples were dry pluviated and taken from bulk samples known as Ticino 04, 08 and 09.  85 2.5.1. Ticino Data Processing  For the elastic properties of Ticino sand, Shuttle and Jefferies (1998) suggest using  48.0 5.6  p 3.1     ape G          [Eq. 2 - 6]  where pa is a reference pressure equal to 100 kPa. The equation is obtained from bender element sting data presented by Bellotti et al. (1996). For test 09-CID-D169 with p'=300 kPa and te e=0.686 at the beginning of the test, we can calculate pDD   for the whole test, which is plotted against stress ratio as shown in Figure 2-4. Theoretically, the difference between total and plastic dilat pDDancy (  ) is equal to zero at the peak. One important observation from Figure 2-3 is that the dif rence between the total and pproxim MC to 1.445 for ET with BM and SD yielding similar fe mphasiplastic dilatancy ( pDD  ) is negligible post-peak. Hence, as the e s is on the latter part of the stress-dilatancy plot, it is possible to use total dilatancy, D, instead of plastic dilatancy, DP, as a reasonable a ation. This is convenient because elastic modulus is not measured using bender elements in all triaxial tests. Total dilatancy (D) has been used instead of plastic dilatancy (DP) in the rest of this work. Figure 2-5 shows a comparison between Mtc obtained using the four methods described earlier; the values of Mtc range from 1.25 for values of 1.33 and 1.345 respectively.    86 2.5.2. Discussion of Results for Ticino Sand )ET =1.445 appears to overestimate Mtc for the ense specimen used here; the sample is not at the critical state at the end of the test, as expected  critical state.  The strength from the end of test method, (Mtc d for a relatively dense sample. Even for loose sand reaching the critical state within the strain limits of the triaxial test is problematic. That test 09-CID-D169 is not at the critical state as confirmed in Figure 2-5, where the dilation rate is not zero at the end of the test. The MC method is based on the assumption that the post-peak hook in the stress-dilatancy plot corresponds to the maximum contraction point when the sample reaches the Although this is assumed by many stress-dilatancy rules, including that of Cam-Clay and Nova, it is not a requirement of critical state theory. None of the triaxial data considered for this work demonstrates such behaviour and, as illustrated in Figure 2-5, this method predicts a much lower shear strength, (Mtc)PT =1.25, than the Stress Dilatancy (SD) and Bishop (BM) methods. The large difference between (Mtc)MC and the critical state strength predicted by the SD and BM methods, combined with lack of a conceptual rationale as to why the maximum contraction and the critical state should generally be coincident, casts doubt on the accuracy of Mtc estimated from the maximum contraction point. (Mtc)SD = 1.345 is very close to (Mtc)BM = 1.33 and the difference is smaller than the 03.0 resolution of Mtc determination suggested by Jefferies and Been (2006). In order to show how h the peak points. That is, the second part of stress- BM results compare to the results of obtaining Mtc from the stress-dilatancy plot, the results are plotted in the same space in Figure 2-6. It can be seen that the two tests for which the entire stress-dilatancy path is plotted are coincident with the line passing throug  87 dilatancy path (the hook) lies on this line. Consequently, the Bishop Method yields the same results as the stress-dilatancy method; we have (Mtc)SD = (Mtc)BM  and *NN  . Note that *NN   is necessary for these two methods to yield equal Mtc values; if *NN   then the hook part of stress-dilatancy plot will leave Bishop's line after the peak and intersect the 0D  axis at a different Mtc. As the stress-dilatancy plots for all of the ten tests show a relatively good agreement with Bishop's method, we can use Bishop's Mtc=1.33 and N=0.4 for Ticino sand with confidence.  2.6. Obtaining the Critical State Stress Ratio of Erksak Sand inantly quartz, sub-rounded and. A complete description of this sand and the test program undertaken is available in Been own not to yield accurate Mtc values. Hence, only the BM and D methods are applied to Erksak data and discussed here.  Erksak sand is a uniformly graded, medium to fine grained predom s et al. (1991). The gradation referred to here has 0.7% fines content and D50=330 m. Nine tests on wet pluviated samples are used here. These tests are presented in Table 2-1.  2.6.1. Erksak Data Processing  The ET and MC methods were sh S maxmin D  pairs are plotted in Figure 2-7 in order to locate Bishop's line. Another line can be drawn passing through six points with a very good resolution and ignoring the three points which are off the line. This results in Mtc=1.10 and N=0.065; this N value is lower than the expected range of 0.2-0.4 casting doubt about the reliability of this chosen line. The best fit  88 trend-line through all nine tests is also plotted in Figure 2-7. This results in Mtc=1.25 and N=0.41, which are more common values for these two parameters. Still there is quite a scatter around this line (R2=0.79) and it does not seem reasonable to merely rely on this line to obtain the critical state parameters. This example shows that Bishop's method xtrapolation for l tests on dense samples would be considered a xurious set of data in most engineering, and even academic, projects. Hence it is very is very sensitive to selection of data and the approach taken in drawing the line. To obtain an accurate critical state stress ratio from a limited number of tests, we now consider the stress dilatancy plot for the entire stress paths. Figure 2-8 shows how the e test CID-G667, where the earlier portion of the hook is preferentially weighted, has been done. The same process was applied to all nine tests, and the M and N* parameters obtained are reported in Table 2-1. The average values of Mtc=1.28 and N*=0.40 are very close to those obtained from the Bishop approach using the best fit line and including all data points.  2.6.2. Discussion of Results for Erksak Sand  It is of interest to note that this set of nine triaxia lu important to be able to estimate the critical state friction angle (or stress ratio) with acceptable accuracy relying on this number of tests. In this case, we have the advantage of having results of a research project (Vaid and Sasitharan, 1992) on the same sand, allowing comparison of the two proposed Bishop's lines as well as the average SD values. Mtc=1.276 and N=0.41 are obtained from Vaid and Sasitharan (see Figure 2-9); these numbers are in good agreement with those obtained from fitting the best line passing through nine data points in BM and the average values obtained from SD method.  89 It is interesting to note that for test CID-G667 (and many other tests) although the maxmin D coincides with Bishop’s line, the dilatancy path leaves this line and estimates a hig and N= btaining critical state friction angle were summarised and the advantages and isadvantages of each method were investigated using previously published data from drained itical state and were found to give unrealistic estimates of Mtc. esting . This makes Stress-Dilatancy her Mtc than that of Bishop method; this also implies *NN  . However, as shown in Table 2-1, the average values of Mtc and N* are only slightly higher than those estimated from Bishop's method. Based on the above discussion Mtc=1.27 0.40 are reasonable estimates for Erksak sand based on both the Stress-Dilatancy and Bishop's methods.  2.7. Summary  Four methods of o d triaxial tests on two sands. Two of the methods "end of test" and "maximum contraction" (ET and MC) are inconsistent with the definition of the cr While Bishop method is the reference method for assessing the critical state friction angle in most cases and especially where a large number of tests are available, most triaxial t programs include only a limited number of tests. With only a small number of tests available, the Bishop method alone is sensitive to any outlying data points. An advantage of the Stress-Dilatancy method is that it yields an estimate of Mtc for every single test, which can then be compared with that of Bishop's method method especially helpful when dealing with a small number of tests. It is therefore recommended that the whole stress-dilatancy path of tests be plotted and used in conjunction with the Bishop approach.  90 Table 2 - 1 Mtc and N* parameters (from the Stress-Dilatancy method) for nine wet pluviated Test name P'0(kPa) OCR e triaxial tests on dense Erksak 330/0.7 Mtc N* CID-G661 140 1.0 0.676 1.28 0.36 CID-G662 60 1.0 0.595 1.26 0.36 CID-G663 300 1.0 0.601 1.30 0.70 CID-G664 300 1.0 0.570 1.23 0.33 CID-G665 130 1.0 0.610 1.18 0.30 CID-G666 60 1.0 0.637 1.24 0.20 CID-G667 130 1.0 0.527 1.38 0.41 CID-G761 250 4.0 0.589 1.32 0.45 CID-G762 250 4.0 0.491 1.32 0.50 Average      1.28 0.40              91 0. 0. 0. 0. 0. 1. 1. 1. 5 0 25 q  /p ') 0 2 4 6 8 0 2 4 1.6 1.8 0 10 15 2 (%)  tc =( q -6 -5 -4 -3 -2 -1 0 1  v( % ) M M MC ET Figure 2 - 1 Stress ratio and volumetric strain versus deviatoric strain for Ticino sand (test 09-CID-D169)  1.2 1.3 1.4 1.5 1.6 1.7 1.8 -0.8 -0.6 -0.4 -0.2 0 Dmin  m ax TICINO-09 TICINO-08 TICINO-04 Ticino 08,09-M=1.33,N=0.4 Ticino 04-M=1.26,N=0.35 tc tcM N1 1 Bishop's Line  m ax   Figure 2 - 2 Application of Bishop's method to Ticino sand  92 1.10 1.20 1.30 1.40 1.50 1.60 -0.4 -0.2 0.0 0.2 Dilatancy (q /p )  SDtcM *1 N 1   Total Dilatancy Plastic Dilatancy D-Dp  Figure 2 - 3 Stress-dilatancy plot ( D ), around peak   0.00 0.50 1.00 /p ) 1.50 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 Dilatancy (q  Figure 2-3 Dilatancy Plastic Dilatancy D-Dp 1.2  Figure 2 - 4 Stress-dilatancy plot ( D ) for Ticino sand (test 09-CID-D169)  93 1.10 1.20 1.30 1.40 1.50 1.60 -0.4 -0.2 0.0 0.2Dilatancy  94 (q /p )   445.1ETtcM   25.1MCtcM   345.1SDtcM   33.1BMtcM maxmin ,D Total Dilatancy  igure 2 - 5 Stress-dilatancy plot ( DF ) for Ticino sand (test 09-CID-D169), comparison of ethods  m  1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 -0.8 -0.6 -0.4 -0.2 0 0.2  maxmin ,D ,   m ax  Peaks CID_D165 Dilatancy CID_D169 Figure 2 - 6 Stress-dilatancy plot (  D ) for Ticino sand 1.0 1.1 1.2 1.3 1.4 1.5 1.6 -0.6 -0.4 -0.2 0 Dmin  m ax  maxmin DFigure 2 - 7 Stress-dilatancy ( ) plot from triaxial data on Erksak 330/0.7 sand 1.1 1.2 1.3 1.4 1.5 1.6 -0.6 -0.4 -0.2 0 Dilatancy (q /p )  Figure 2 - 8 Stress-dilatancy plot ( D ) for Erksak 330/0.7 sand (test CID-G667) 1.2 1.3 1.4 1.5 1.6 1.7 1.8 -0.8 -0.6 -0.4 -0.2 0 Dmin  m ax  Figure 2 - 9 Stress-dilatancy ( maxmin D ) plot from triaxial data on Erksak 330/0.7 sand (data after Vaid and Sasitharan, 1992) *1 N  SDtcM Bishop’s line Three tests ignored  in identifying  Bishop's line Best fit line considering all data points  ( q/ p)  1  95 2.8. References  Been K., and Jefferies M.G. 1985. A State Parameter for Sands. Géotechnique, 35(2): 99-112. Been K., Jefferies M.G., Crooks J.H.A., and Rothenberg L. 1987. The Cone Penetration Test in Sands: Part II, General Inference of State. Géotechnique, 37(3): 285-299. Been K., Jefferies M.G., and Hachey J.E. 1991. The Critical State of Sands. Géotechnique, 41(3): 365-381. Bellotti R., Jamiolkowski M., Lo Presti D.C.F., and O'Neil D.A. 1996. Anisotropy of Small Bishop A.W. 1966. Strength of Soils as Engineering Materials. 6th Rankine Lecture. Géotechnique, 16: 89-130. Bishop A.W. 1972. Shear strength parameters ndisturbed and remolded soil specimens. In Proceedings, Roscoe Memorial Symposium, Cambridge University. Edited by R.H.G. Parry. G.T. Foulis & Co. Ltd., Yeovil, U.K., 3–139. Bolton M.D. 1986. Strength and Dilatancy of Sands. Géotechnique, 36(1): 65-78. Castro G. 1969. Liquefaction of Sands. PhD. Thesis, Harvard University, Cambridge, Mass. (Harvard Soil Mechanics Series 81). Garga V.K., and Sedano J.A.I. 2002. Steady State Strength of Sands in a Constant Volume Ring Shear Apparatus. Geotechnical Testing Journal, 25(4). Jefferies M.G. 1993. Nor-sand: a Simple Critical State Model for Sand. Geotechnicque, 43(1): 91-103. Jefferies M.G., and Shuttle D.A. 2002. Dilatancy in General Cambridge-Type Models. -638. Strain Stiffness in Ticino Sand. Géotechnique, 46 (1): 115-131. for u Géotechnique, 52: 625  96 Jeffe es M.G., andri  Been K. 2006. Soil Liquefaction: A Critical State Approach.  Taylor & Francis (Abingdon & New York).  ISBN 0-419-16170-8. s. Neguss ction Angle of Poulos S.J. 1981. The Steady State of Deformation. Journal of Geotechnical Engineering, Roscoe of Soils. Géotechnique, Rowe atancy Relation for Static Equilibrium of an Assembly of Shuttle ethods in Geomechanics, 22: 351- 9(2): 1-9. Klotz E.U., Coop M.R. 2002. On the Identification of Critical State Lines for Sand Geotechnical Testing Journal, 25(3). ey D., Wijewickreme W.K.D., and Vaid, Y.P. 1988. Constant Volume Fri Granular Materials. Canadian Geotechnical Journal, 25: 50-55. American Society of Civil Engineers, 17(G75): 553-562.  K.H., Schofield A.N., and Wroth C.P. 1958. On the Yielding 8(1): 22-53. P.W. 1962. The Stress Dil Particles in Contact. Proceedings of the Royal Society of London, A 269: 500-527.  D.A., and Jefferies M.G. 1998. Dimensionless and Unbiased CPT Interpretation in Sand. International Journal for Numerical and Analytical M 391. Vaid Y.P., and Sasitharan S. 1992. The Strength and Dilatancy of Sand. Canadian Geotechnical Journal, 29: 522-526. Wanatowski D., and Chu J. 2006. Stress-Strain Behavior of a Granular Fill Measured by a New Plane-Strain Apparatus. Geotechnical Testing Journal, 2  97 Chapter 3.   Confidence and Accuracy in Determination of Crit    Geotechnical predictions are sensitive to the shear strength parameters and so it is important for strengt al state (or constant volume) friction angle, c, a soil property that Critica  high values of strain will eventually reach a constant void ratio, ec, termed the critical void ratio, and constant friction angle, c, termed the critical friction angle. Well established in soil mechanics since the late 1950s, critical state soil mechanics provides a basis for understanding the soil behaviour, with   ical State Friction Angle3  3.1. Introduction geotechnical engineers to have access to good parameter estimates. One widely used shear h parameter is the critic varies over a surprisingly wide range depending on soil mineralogy amongst other factors. l state soil mechanics is based on the idea that soils sheared to very  3 A version of this chapter has been published. Ghafghazi M., and Shuttle D.A. 2009. Confidence and accuracy in determination of the critical state friction angle. Soils and Foundations, 49(3): 391-395.  98 many soil constitutive models based on this concept. Equally, the simplest engineering problems ate of the critical state friction angle, even if a critical state framework is not explicitly adopted, as c is central to stress dilatancy (which is e micromechanical process governing cohesionless soil behaviour). For example Bolton 986) relative density index provides a simplified but useful engineering relation which allows c is known. Despite the importance of c, the literature contains little guidance on how many laboratory tests uracy. This is particularly problematic in engineering practice, where typically only a limited number of soil tests are possible due to budget limitations. This situation is exacerbated by the lack of a consensus on the most accurate method of obtaining c from standard laboratory tests.  ch xtensive triaxial testing program from the literature to determine c using ishop’s method (1972), this being a standard method for determining c. The database chosen  dealing with shear strength of soils require an estim th (1 for estimating the peak friction angle peak as a function of the relative density Dr if are required to determine c within the desired acc This apter uses an e B is unusual in the sense that it comprises a large number of tests on a single material. By performing a statistical analysis of the database, guidance is obtained on the number, density range and pressure range of triaxial tests required to reach a specified level of accuracy and confidence in c. Finally, the performance of the proposed methodology for estimating the accuracy of c determination is tested against a smaller independent set of triaxial tests on the same sand, performed in a commercial testing laboratory.     99 3.2. Triaxial Database  The material used for this study was Erksak sand, a sand used in several offshore construction projects in the Beaufort Sea (Jefferies et al., 1985). Erksak sand is a medium (D50 = 0.34 mm), uniform (Uc = 1.8), mainly quartz sand with sub-rounded particles. Testing of this sand was independently reported by Vaid and Sasitharan (1992) and Been et al. (1991). The majority of the work in this chapter focuses on 34 drained triaxial compression tests performed in the University of British Columbia’s (UBC) soils laboratory, as reported by Vaid and Sasitharan (1992). Samples were prepared using the water pluviation technique at three different relative densities; 26% (13 tests), 56% (13 tests) and 70% (8 tests) representing loose, medium dense and dense conditions. A smaller set of drained triaxial data was reported by Been Calgary commercial testing laboratory. The Been et al. tests were also prepared using water pluviation and had relative densities ranging between 20% and 74% with five loose, two medium and two dense samples. The terms loose, medium dense and dense refer to initial relative density values of 15-35%, 35-65%, and 65-85% respectively. It should be mentioned that even the loosest samples tested showed “dense” sand behaviour; i.e. having a clear peak strength and dilating towards the critical state. This lack of pressure dependence on soil behaviour is a pitfall of using Dr as the primary density index. However, despite this limitation Dr is used as the density index in this chapter due to its simplicity, and more importantly its widespread adoption in engineering practice.   et al. (1991), and comprised 9 triaxial compression tests performed in Golder Associates’  100 3.3. Methodology Used for c Determination ompression tests. c c and dv/d1. However, the alternative identities  and D are preferred because these variables are, theoretically, linearly related (Schofield and Wroth, 1968; Nova, 1982). The dilatancy, D, is  The critical state friction angle c appears to be constant for a particular soil under triaxial compression conditions (e.g. Rowe, 1962; Negussey et al., 1988). However, there is no consensus on the most accurate method of determining its value. Ghafghazi and Shuttle (2006) (chapter 2) discussed four different methods of determining c from triaxial c The first method, termed End of Test, assumes the measured friction angle at the end of the triaxial test is equal to c. End of Test is simple, but most samples do not reach the critical state within the strain limits of the triaxial test and post-peak localisation could also render the measurements unreliable. The second method, Maximum Contraction, incorrectly assumes that the friction angle at maximum contraction is equivalent to   and so provides poor   values. The third method, Stress-Dilatancy, involves extrapolating a plot of the post-peak stress ratio versus dilatancy data to zero. This provides good predictions in the absence of localisation immediately post peak. However it is difficult to automate and involves user interpretation. The final method considered was Bishop’s Method (1972), which has the advantage of yielding theoretically correct answers while being easily automated for any dataset of triaxial compression tests. This study uses Bishop’s method. Bishop’s method of obtaining the critical state friction angle is based on the idea that, for constant mean stress, the peak friction angle of a soil increases with increasing density, and that a purely contractive soil will reach peak strength at critical state. Bishop used the parameters   101 defined as the ratio of rate of volume change to the rate of change in shear strain invariant. In athematical form: m  q vdD            [Eq. 3 - 1] 2 d  where v [=1 + 23] and q [= /3 (1 - 3)] are the triaxial volumetric and deviatoric strain invariants respectively. The dilatancy at peak strength is negative because of the compression positive convention of soil mechanics. Mobilised stresses are represented by the stress ratio, :  p           [Eq. 3 - 2]  where q [= ′1 - ′3] and p′ [= (′1 + 2′3)/3] are the triaxial deviatoric stress and mean effective stress invariants respectively and ′1 and (′2 =′3) are the three principal effective stresses. For known stress conditions the friction angle is directly related to the stress ratio at the critical state and the two parameters can be applied interchangeably; for example, in triaxial compression we have q       tcc M M 6 3  where Mtc is the stress ratio at the critical state under triaxial compression conditions.  tcarcsin          [Eq. 3 - 3]  102 With Bishop’s method, a series of triaxial tests at differing densities are carried out. As illustrated in Figure 3-1, for each test the mobilised stress ratio, , is plotted against dilatancy, , and the peak point of the plot is chosen to represent the test on a plot of max vs. Dmin . Mtc is determi tests, with Mtc being the intercept at zero dilatancy, as shown in Figure 3-2. For a soil sample that reaches the critical state directly, i.e. without dilation,   (or equivalently  ) corresponds with the critical state. In this work it is assumed that the best fit regression line through all 34 Vaid and Sasitharan tests provides the "correct" answer. The trendline shown in Figure 3-2 yields Mtc = 1.276 (or c = 31.75°) with a coefficient of determination, R2= 0.955.  .4.  Determination of c from Limited Number of Triaxial Tests uld ch is most accurate when the tests are spread over a range of Dmin; using ndom tests would result in some unrepresentatively poor estimates of c. Hence it was assumed that every combination of tests used to determine  include at least one loose, one medium dense and one dense sample (i.e. one test at Dr = 26%, 56% and 70% respectively for the Vaid and Sasitharan dataset). Additionally, no repetition was allowed in this procedure so D ned by extrapolation of a linear regression through the Dmin- max for all the 34 triaxial max peak 3  To determine accuracy in c determination from fewer triaxial tests the simplest method wo be to randomly sample the required number of tests from the database of 34 tests. However, this approach was felt to be unrealistic as most testing programs include a range of soil densities. Also, Bishop’s approa ra c that no test could be sampled twice in any realisation.  103 Adopting this statistical methodology led to c being calculated with 3, 4, 5,…, 34 tests. Typically 3 to 20 tests are presented here: 20 tests being considered an upper bound for most practical testing programs. Each combination of loose, medium dense and dense tests comprising the total number of required tests was sampled 300 times. For example, 4 tests can comprise either 1 loose, 1 medium, 2 dense or 1 loose, 2 medium, 1 dense or 2 loose, 1 medium, 1 dense tests. After each of these three possible combinations was randomly realised 300 times, Mtc was again calculated from linear regression through the realised data points. Values of Mtc xpected, with an increasing number f tests the accuracy of the c prediction improves. However, Figure 3-3 also indicates that there is bias is also evident in e determined Mtc data and may be related to the use of a real dataset. The magnitude of this were then converted to c using Equation 3-3. The precision of the c calculation, plotted as the percentage of tests falling within 0.1 degree bins, is shown in Figure 3-3 for 5, 10, 15 and 24 tests. As e o is a slight bias towards over-prediction of c for all numbers of tests. Th th bias is small: for 10 tests the bias in the results from quoting the average absolute error of 0.36 degrees, rather than the average over and under-prediction errors of 0.39 and 0.33 degrees respectively, is only 0.03. Hence the bias is ignored in the remainder of the work. Figure 3-4 plots |c| versus number of tests for a range of confidence levels. At each confidence level the error rapidly decreases from 3 tests to around 8 tests. Thereafter a slower, almost linear, enhancement in accuracy is indicated. At the 85% confidence level this corresponds to a 1.31 degree accuracy for 3 tests to better than 0.75 degrees accuracy with 8 tests. The same information is provided in numerical form in Table 3-1 at 75%, 80%, 85%, 90%, and 95% confidence levels for 3 to 20 tests.  104 As discussed earlier, Figure 3-4 is obtained us ng one loose, medium dense and dense test in each realisation with the remaining tests randomly distributed. From a practical perspective it is of interest to determine whether improved accuracy could be obtained by further specifying the initial densities of the samples. Figure 3-5 plots the magnitude of the average and standard deviation absolute error for all possible combinations of 9 tests including at least one loose, one medium and one dense test. The combinations are organised with the largest absolute error on the left hand side of the figure. The results show that of the 28 possible i  combinations of tests, the 10 combinations with Table 3-1. Hence these confidence levels could be considered an upper bound on likely commercial accuracy. The general applicability of the confidence levels is tested using nine good quality commercial tests reported by Been et al. (1991). The full suite of nine tests shown in Figure 3-6 indicates M  = 1.254, corresponding to  = 31.24° and suggesting an error  the largest error have only one from each grouping. Moreover, the three combinations with 7 tests from one density grouping are among the worst 5 groups. Conversely, the average error is approximately half where the tests are better distributed between the three groups. Hence it is advisable that the program be designed to distribute tests equally between loose, medium and dense samples.  3.5. Validation against Independent Triaxial Database  A high quality university testing program was used to compute the confidence levels shown in tc c compared to the original database of c = 0.51° (associated with 67 % level of confidence).  105 To directly compare with the predicted confidence levels it is necessary to consider possible subsets of these data. If six tests are considered, there are 70 possible combinations of tests which include at least one loose, medium dense and dense test. Therefore from Table 3-1 we would expect 52 of the combinations to predict c within an accuracy of 0.70º at 75% ccuracy. Additionally, comparison of Figure 3-2 with Figure 3-6 indicates that ll of the “dense” tests for the Golder dataset are significantly less dilatant (Dmin > -0.55) than ll but two samples). This problem ould be reduced by maximising the range of Dmin in Figure 3-6 by ensuring that for each   confidence. The Been et al. data suggest a lower accuracy of only ±1.22º at 75% confidence (see Table 3-2). Some reasons for the lower accuracy may be observed from Figure 3-6. The accuracy predicted from the Vaid and Sasitharan data is predicated on having one test from each density range. Although this criterion has been enforced for the Been et al. dataset as well, the effect of stress level (see Table 3-3) has resulted in one of the two medium dense tests being less dilative than all of the loose tests. Therefore the range of Dmin between density ranges is reduced. Although this problem also exists for the large dataset, the few medium dense tests in the small dataset increase the ina a the dense samples from the larger dataset (Dmin < -0.55 for a c confining pressure at least one loose and one dense test are undertaken. This is equivalent to defining a range of state parameters (Been and Jefferies, 1985),  [= e – ec], which accounts for the stress level effect on dilatancy, in addition to the void ratio accounted for in Dr.     106 3.6. Summary and Conclusions  A statistical evaluation of a drained triaxial compression test database was performed to determine accuracy and confidence level in determining the critical state friction angle. Critical state friction angle was obtained from a dataset comprising 34 triaxial tests using the methodology proposed by Bishop (1972); c being obtained using linear regression. It was risingly) the rrors from the commercial dataset were slightly larger, the academic database provided a asonable upper bound on likely achievable accuracy.  conclusion, although soil type and gradation might be expected to affect sample uniformity uring reconstitution and hence influence the repeatability (and hence accuracy) of the triaxial assumed that the correct c was obtained if all 34 tests were included in the analysis. In determining the accuracy of smaller realisations of the dataset it was assumed that any test program will include at least one loose (Dr = 26 %), one medium dense (Dr = 56 %) and one dense (Dr = 70 %) sample tested under drained triaxial compression conditions. Results were presented as error in c versus number of tests for confidence levels of 75%, 85%, 90% and 95%. As the number of tests increased from 3 to 8, a sharp increase in accuracy was observed at all confidence levels. Hence it is recommended that any commercial testing program for evaluation of the critical state friction angle includes at least 6 tests (6 tests yielding an accuracy of 1.0º from university quality data with 90% confidence). For academic purposes, where accuracy of 0.5º with 90% confidence may be needed, more than 20 tests may be required. Although the presented results were developed using one comprehensive academic testing program, their application to commercial data was encouraging. Although (unsurp e re In d  107 testing program, distributing the triaxial tests over a wide range of initial Dr, or ideally initial , hould provide greater accuracy in c for fewer tests. s  108 Table 3 - 1 Confidence level, | c|, versus number of triaxial tests of tests 3 4 5 6 7 8 9 10 Number 11 12 13 14 15 16 17 18 19 20 95 1.67 1.42 1.28 1.16 1.06 0.98 0.93 0.89 0.86 0.85 0.82 0.81 0.80 0.76 0.73 0.69 0.66 0.62 90 1.46 1.22 1.09 1.00 0.88 0.84 0.80 0.77 0.75 0.73 0.71 0.70 0.69 0.67 0.64 0.61 0.57 0.54 85 1.31 1.08 0.97 0.87 0.79 0.75 0.71 0.70 0.67 0.65 0.64 0.63 0.62 0.60 0.58 0.55 0.52 0.49 80 1.18 0.98 0.88 0.77 0.71 0.69 0.65 0.63 0.61 0.59 0.58 0.57 0.57 0.55 0.53 0.50 0.48 0.45 C on fid en ce  L ev el  (% ) 75 1.06 0.87 0.79 0.70 0.64 0.62 0.59 0.58 0.55 0.54 0.52 0.52 0.52 0.50 0.48 0.46 0.44 0.41    Table 3 - 2 Confidence levels for 6 tests from the independent Been et al. (1991) dataset Confidence Level (%) 75 80 85 90 95 Error |c| (degrees) 1.22 1.29 1.39 1.96 2.00    Table 3 - 3 Summary of initial conditions for Been et al. (1991) triaxial tests Test Name Void ratio e p' (kPa) Dr (%) 0 Dmin CID G666 0.707 60 20.4 -0.055 -0.28 CID G665 0.687 130 29.2 -0.063 -0.295 CID G661 0.676 140 34.1 -0.073 -0.250 CID G662 0.675 60 34.5 -0.087 -0.381 CID G663 0.671 300 36.3 -0.066 -0.337 CID G7611 0.649 250 46.0 -0.091 -0.250 CID G664 0.630 300 54.4 -0.107 -0.400 CID G7621 0.601 250 67.3 -0.139 -0.450 CID G667 0.587 130 73.5 -0.163 -0.551 1 The two tests G761 and G762 are performed on samples with over-consolidation ratio = 4      109 0.0 0.2 0.4 0.6 0.8 1.0 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 D = d   / d  1.2 1.4 1.6 1.8 D min , max =  q  / p' v q  Figure 3 - 1 Stress-Dilatancy plot for test D-667 (Been et al., 1991) on Erksak sand   1.1 1.2 1.3 1.5 1.6 1.7 1.8 -0.6 -0.4 -0.2 D min  m ax 1.4 -0.8 Loose tests (Dr = 26%) Medium dense tests (Dr = 56%) Dense tests (Dr = 70%) 0 M tc 6  F  vs.  34 tria ompr  tes ksak sand (after Vaid and Sasitharan, 1992)    = 1.27 igure 3 - 2max  Dmin of xial c ession ts on Er  110 05 10 15 20 25 30 35 -1 .8 -1 .6 -1 .4 -1 .2 -1 .0 -0 .8 -0 .6 -0 .4 -0 .2 0. 0 0. 2 0. 4 0. 6 0. 8 1. 0 1. 2 1. 4 1. 6 1. 8   c  (degrees) Fr eq ue nc y (% ) 24 tests 15 tests 10 tests 5 tests  Figure 3 - 3 Frequency of error in c for 5, 10, 15 and 24 tests   0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1 3 5 7 9 11 13 15 17 19 21 23 25 No. of triaxial tests |   95% 90% 85% 75%  c|  Figure 3 - 4 Error in c vs. number of triaxial tests at different confidence levels        111 00.1 0.2 0.3 0.4 0.5 0.6 0.7 1, 7, 1 1, 6, 2 6, 1, 2 7, 1, 1 1, 1, 7 1, 5, 3 1, 4, 4 5, 1, 3 6, 2, 1 1, 2, 6 5, 2, 2 1, 3, 5 2, 6, 1 2, 5, 2 5, 3, 1 4, 1, 4 2, 4, 3 4, 4, 1 4, 2, 3 3, 5, 1 2, 3, 4 4, 3, 2 3, 4, 2 2, 2, 5 3, 3, 3 3, 2, 4 3, 1, 5 2, 1, 5 Combinations of loose, medium and dense tests   c average stdev  Figure 3 - 5 Error in calculation of c obtained from 9 tests with different combinations of loose, medium and dense tests    1.1 1.2 1.3 1.4 1.5 1.6 1.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0 D min  m ax Loose tests (15%<Dr<35%) Medium dense tests (35%<Dr<65%) Dense tests (65%<Dr<85%)  M tc  = 1.254  igure 3 - 6 max vs. Dmin of 9 triaxial compression tests on Erksak sand (after Been et al., 1991) F  112 3.7. References  Been K., and Jefferies M.G. 1985. A state Parameter for Sands. Géotechnique, 35(2): 99-112. Been K., Jefferies M.G., and Hachey J.E. 1991. The Critical State of Sands. Géotechnique, 41(3): 365-381. Bishop A.W. 1972. Shear strength parameters for undisturbed and remolded soil specimens. In Proceedings, Roscoe Memorial Symposium, Cambridge University. Edited by R.H.G. Parry. G.T. Foulis & Co. Ltd., Yeovil, U.K., 3–139. Jefferies M.G., Stewart H.R., Thomson R.A.A., and Rogers B.T. 1985. Molikpaq Deployment at Tarsiut P-45. Proceeding of ASCE Specialty Conference on Civil Engineering in the D.A. 2006. Accurate Determination of the Critical State Friction Angle from Triaxial Tests. Proceedings of the 59th Canadian Geotechnical Conference, Vancouver, 278-284. Negussey D., Wijewickreme W.K.D., and Vaid Y.P. 1988. Constant Volume Friction Angle of Granular Materials. Canadian Geotechnical Journal, 25(1): 50-55. Nova R. 1982. A Constitutive Model under Monotonic and Cyclic Loading. Proceedings of Soil Mechanics – Transient and Cyclic Loads, Pande and Zienkiewicz (eds.), Wiley, 343-373. Rowe P.W. 1962. The Stress Dilatancy Relation for Static Equilibrium of an Assembly of Particles in Contact. Proceedings of the Royal Society of London, Series A, 269: 500-527. Schofield A.N., and Wroth C.P. 1968. Critical State Soil Mechanics. McGraw-Hill, London. Download: http://www.geotechnique.info/ Arctic Offshore, San Francisco, CA, 1-27. hafghazi M., and Shuttle G  . Vaid Y.P., and Sasitharan S., 1992. The Strength and Dilatancy of Sand. Canadian Geotechnical Journal, 29: 522-526.  113 Chapter 4.   Interpretation of Sand State from CPT   4.1. The behaviour of cohesionless soils (sands, s that it can be m Jefferie is very  sampling – an “undisturbed” cohesionless    Tip Resistance4  Introduction  ilts, tailings) strongly depends on their density. While relative density Dr is an almost universally used density index for sand, it is easy to show isleading (e.g. Tavenas, 1973). An alternative to Dr that captures both the effect of void ratio and the effect of mean stress on soil behaviour is the state parameter,  (Been and s, 1985). Different soils, or the same soil at different stress levels, display similar behaviour at the same value of  However, determining the in-situ  (or Dr) in the laboratory  difficult because of density changes during sample is essentially impossible to obtain in normal engineering practice. Penetration tests have  4 A version of this chapter has been published. Ghafghazi M., and Shuttle D.A. 2008. Interpretation of sand state from cone penetration resistance. Géotechnique, 58(8): 623–634.  114 thus become routine practice for testing cohesionless soils, with the modern electronic CPT easurement, and excellent repeatability and accuracy at relatively low cost. he difficulty with any penetration test, however, is that the state measure of interest (e.g. Dr, )  not measured. Instead it is deduced from the penetration resistance; a process usually referred roblem to obtain mechanical properties of the soil from test measurements. But the large deformations viour of the soil and complicated boundary conditions, make this analysis an extremely difficult task, and renders the solution non-unique. The interpretation framework is also difficult to establish. No simple closed-form solution for  or Dr has been developed from CPT; and, nobody - to date - has provided a full numerical simulation of drained penetration that matches calibration data, although several have tried (de Borst and Vermeer, 1982; Willson et al., 1988; van den Berg, 1994; Huang et al., 2004; and Ahmadi et al. 2005). Two different directions have emerged to deduce soil state from CPT data: calibration chamber tests and simplified theoretical treatments. Calibration chambers are circular steel tanks typically about a metre in diameter and similar height. Sand is deposited at a known density and consolidated to the desired stress state within the tank. A cone penetration test is then performed along the vertical axis of the reconstituted specimen exactly as in natural ground. Each test provides a tip resistance qc for the given value c 1 2 3 density (more usually expressed as  or Dr) for the tested material. The in-situ state 0is then offering continuous data m T is to as interpretation. This interpretation involves solution of an inverse boundary value p associated with the CPT, along with the nonlinear beha of density and applied stress. A large number of tests, covering the range of densities and stresses of interest, provide the relation between q , effective stresses ( , ,  ), and the  115 obtained from the CPT by comparison of field CPT qc measurements at the estimated in-situ stresses, to the qcp00 relation determined in the calibration chamber. The first CPT calibration chamber was built in 1969 at the Country Roads Board (CRB) in jects, it is not practical to with both of these approaches was that no mechanistic framework was offered. Australia (Holden, 1991). Chamber tests are now reported in the literature with differing dimensions, nature and form of control of boundaries, method of specimen reconstitution, and capability to handle saturated specimens. While improvements have been made to reconstitute more consistent specimens, most effort has been devoted to minimising and quantifying the influence of sample boundaries on the penetration resistance (e.g. Bellotti et al., 1982; Jamiolkowski et al., 1985; Been et al. 1987a; Mayne and Kulhawy, 1991; and Salgado, 1998). Chamber ratios, defined as the ratio of the chamber to cone diameter (D/d) of up to 200 have been proposed as the minimum ratios required to eliminate boundary effects on measured CPT tip resistance (Salgado et al., 1998). A large body of calibration chamber data is available for clean sands, through decades of testing. These data show that the calibration is soil specific. This soil specific calibration is also affected by the stress history and, possibly, fabric. But, for most pro conduct calibration chamber tests for the soil of interest at the different conditions expected to be encountered in the field. Thus, the case for an analytical method of CPT interpretation that can be calibrated for a specific soil based on easier laboratory tests seems compelling. Robertson and Campanella (1983) first suggested that the difference between the CPT calibrations in various sands might be understood in terms of the intrinsic ‘compressibility’. However, the compressibility measure was not defined. Subsequently, Been et al. (1987a) invoked the slope of the critical state line as the compressibility index. The basic problem  116 Many simplified theoretical treatments have used spherical cavity expansion as an analogue of the penetration test on which to develop a mechanistic framework, essentially the same s the fundamental deficiency that soil does not behave in ardening,  used a general approach as used in conventional design of end bearing capacity of piles. The initial work by Bishop et al. (1945) and Hill (1950) addressed incompressible materials with associated flow rules, corresponding to the familiar and simple idealisation of the undrained behaviour of clay. Chadwick (1959) derived the pressure-expansion relation for a Mohr-Coulomb material with an associated flow rule, and further development of cavity expansion theory for non-associated flow has been considered by a number of workers including Carter et al. (1986) and Yu and Houlsby (1991). The central assumption of these studies has been that both the friction and dilation angles remain constant during shearing. Although constant dilation and friction angles lead to analytical or semi-analytical solutions for cavity expansion, the idealisation ha such a manner – the nuances of soil behaviour being of first-order significance to realistic modelling of cavity expansion in soil. Realistic simulation of cavity expansion requires a stress- strain model that captures the evolution of dilatancy with accumulated strain and stress, as it is this behaviour that distinguishes one sand from another. The consequences of realistic modelling of dilatancy are magnified by the confinement of the cavity expansion. In more comprehensive evaluations ‘good’ soil models have been used allowing for plastic h strain softening, etc. Collins et al. (1992) provided the first realistic model for cavity expansion in soil using a state parameter based numerical analysis. Their results showed that the relation between normalised CPT tip resistance and 0 was still affected by the stress level (as suggested by Sladen, 1989) and depended on material properties of the sand. Shuttle and Jefferies (1998)  117 work hardening/softening critical state model to evaluate changes in CPT calibration in terms of critical state parameters  and  that can be determined in routine triaxial testing of reconstituted specimens. These authors showed that CPT behaviour in Hilton Mines sand, currently the most unusual of the published calibrations, could be predicted based on Ticino sand data by allowing for the changes in fundamental soil properties between Hilton Mines and Ticino. The analogy of the cone penetration to expansion of a sphere in soil is based on the existence of a limit cavity expansion pressure, although it has long been recognised that there is an inherent mismatch between spherical cavity expansion analysis results and those of calibration chamber tests. The magnitude of the mismatch is enlarged with the application of realistic models that enetration, none of the preceding analyses explicitly model the effects of capture the variable nature of strength and dilation angle. Their predecessors likely masked the mismatch by assuming constant friction and dilation angles. The difference between the spherical cavity analogy and the actual CPT geometry is usually accounted for by a shape factor. Shuttle and Jefferies (1998) showed that there is a one to one correspondence between the normalised tip resistance obtained from spherical cavity expansion and that obtained in chamber testing on Ticino sand. Cudmani and Osinov (2001) suggested a common shape factor for five different sands that is a function of the relative density corrected for stress level effects. It is interesting that although particle breakage is known to occur at the high stresses usually induced during cone p particle breakage on the limiting cavity pressure. Russell and Khalili (2002) incorporated one effect of particle breakage, a steepening of Critical State Locus (CSL) with increasing mean stress, into a cavity expansion analysis using a critical state based model. The effect of this change was a reduction in the limiting cavity expansion pressure, inferring an increase in the  118 shape factor required to match the chamber penetration resistance. Investigation into particle breakage is still in its infancy, with issues of changing stress-dilatancy due to energy dissipation during breakage yet to be resolved, and hence this effect is not explicitly modelled in this work. In this chapter spherical cavity expansion is applied to a database of nine soils including laboratory standard and natural sands, as well as relatively clean sand-size tailings for which both chamber testing and triaxial compression data are available in the literature. The soil behaviour is captured by the isotropically hardening - isotropically softening generalised critical state model, NorSand (Jefferies, 1993; Jefferies and Shuttle, 2002; and Jefferies and Shuttle, 2005), independently calibrated to triaxial tests. A unique relation between cavity expansion and calibration chamber normalised tip resistances, termed the shape function is identified for all database sands. This shape function is used to predict the calibration chamber results from spherical cavity expansion simulations with the same level of accuracy as observed in published experimental data. A framework for interpreting the state parameter from CPT tip resistance is then presented that eliminates the need to perform calibration chamber tests to obtain soil specific qcp00 correlations. Calibrations to triaxial testing determine material parameters that are then used in numerical analysis to predict the in-situ state, without sacrificing accuracy, and without resorting to expensive and difficult calibration chamber tests.  4.2. Background to CPT Interpretation in Sand  The CPT in sand provides just two outputs; the tip resistance and the sleeve friction (the penetration is drained so the pore pressure transducer simply measures the in-situ pore pressure). This is why most CPT data are interpreted in terms of one parameter alone, commonly relative  119 density, but sometimes peak friction angle. The state measure used in this work is the state parameter,  (Been and Jefferies, 1985). Because  is used as an internal state variable in the numerical model, the subscript ‘0’ is used to denote the in-situ (or initial) value of 0 under geostatic conditions to be consistent with the original usage by Been and Jefferies. Initial work on determining 0 from CPT data comprised triaxial testing of sands for which chamber test data were available to define the CSL of each sand, and then processing the chamber test data to develop dimensionless relations (Been et al., 1986, 1987a) of the form:  Q = k exp( - m0 )         [Eq. 4 - 1]  Where Q is the tip resistance normalised by the initial mean effective stress, p′0:  0'p pq   The two coefficients k and m in Equation 4-1 differ from one sand to another. Since all analytical methods of interpreting 0 from CPT results mus 0Q c           [Eq. 4 - 2] t be verified against vailable calibration chamber data, the scatter in experimental results limits the accuracy of any  measure reproducibility of calibration hamber results by repeating tests on specimens with exactly the same density and stress a interpretation method. Experimentalists have tried to c conditions. These efforts have resulted in  25% variability in measured Q in the most recent works (e.g. Hsu, 1999). However, the majority of available data suggest that  50% accuracy in measured repeatability of Q is ‘good’ quality data. Of course translating the accuracy in Q to the  120 accuracy in 0 involves interpretation, but a rough estimate based on average k and m values for Ticino 4 sand (Shuttle and Jefferies, 1998) would suggest that  0.05 accuracy in 0 is about the best accuracy that can be expected in predicting any particular test in the available calibration chamber data.  4.3. Materials and Testing  The nine soils used in this research were chosen based on availability of reliable triaxial and boratory standard sands, Hokksund, Ottawa, Ticino and Toyoura sands; and two mine tailings, erties of these oils as reported by different researchers. The database includes median grain sizes of 160 to 1130 m and rounded to angular grain shapes. Most calibration chamber tests are performed on dry specimens. There was little or no fine grained material in the soils, resulting in completely rained penetration even in cases where water was present in the chamber. on reconstituted calibration chamber test data. They include two natural sands, Da Nang and Erksak sands; four la Syncrude oil tailings and Hilton Mines sand. Table 4-1 contains the index prop s d Triaxial compression tests were used to determine soil properties based specimens. Table 4-2 summarises the drained triaxial testing used to calibrate NorSand. In cases where undrained tests were available they were only used to determine the CSL in pe  log space. Previously published calibration chamber tests performed between 1974 and 1998, and summarised in Table 4-3, have been used to illustrate the utility of this approach. Although some over-consolidated chamber tests were available, this work focuses on the normally consolidated tests.  121 rs between 0.9 and 1.5 to the tip resistance qc. This correction generally falls idrange between those of Jamiolkowski et al. (1985) and Mayne and Kulhawy (1991). .4. Constitutive Model soil behaviour. The y. Subsequently, Jefferies and Shuttle (2002) suggested that N could be eliminated from onstitutes additional effort in calibration of the model, as N is obtained with m  4  The reason to use a ‘good’ sand model for cavity expansion analysis is that, in general, part of the soil domain will be critical, other parts strongly dilating, and other parts contractive; with the exact behaviour being a function of the soil’s position relative to the cone tip. Accordingly, it is helpful to look at the computed behaviour of the sand before examining the computed limiting pressures as a function of initial state. The constitutive model adopted is NorSand, an isotropically hardening - isotropically softening generalised critical state model that captures a wide range of particulate version used is that for general 3-D stress states with constant principal stress direction as described in Jefferies and Shuttle (2002) but with a further extension to improve accuracy. A feature of the original version of NorSand was a volumetric coupling parameter N for stress- dilatanc the model since N1 (based on average values of a large quantity of triaxial tests on different sands). However, individual soils demonstrate a variety of N values and some accuracy is given up by following this suggestion. The N model parameter neither increases the complexity of the model, nor c  122 Mtc from the stress-dilatancy plot and reintroduction of this parameter resulted in better replication of the soil behaviour. For the current work ‘N’ was obtained from the post-peak stress-dilatancy plot (see Figure 4-1), rather than the slope of the line passing through peak points of the stress-dilatancy plot, and hence is referred to as N* in the remaining text. Like most elasto-plastic models, NorSand has four basic aspects: a yield surface; a work rdening law; a plastic flow rule (stress-dilatancy); and, elasticity. Many material parameters in  Cam Clay. Three parameters describe the critical state, these eing the critical friction ratio tc (the tc denoting the reference triaxial compression condition  the yield surface from the CSL and hence the constant during the test. Although G is treated as a soil property, the sensitivity of the modulus to soil fabric argues in favour of elastic modulus perhaps being viewed as a further state r 0 ha NorSand will be familiar from b and allowing M to vary with lode angle, ) and  which describe the CSL for the semi-log idealisation used. There are two parameters relating to the plastic behaviour: tc and H. The property tc scales the maximum dilatancy to . In some ways this is similar to specifying a relation between peak strength and . The dimensionless hardening modulus H is required because the state parameter approach de-couples slope of the CSL no longer acts as a plastic compliance. A finite elastic shear modulus G is also adopted. The rigid-plastic idealisation in shear used by Cam Clay is both unrealistic, and inconvenient for use with numerical methods. Various forms of elasticity can be used within the same framework. For the current work both elastic parameters G and  are assumed to be measure rather than as a property. To preserve the dimensionless nature of NorSand, it is convenient to specify G in terms of the dimensionless rigidity I  (= G/p′ ). Despite compressibility having an important influence on CPT resistance (Robertson and Campanella, 1983), similar to most constitutive models, compressibility is not represented  123 within NorSand by a single material property. Both plastic and elastic compressibility play an important part in the limiting spherical cavity pressure. Plastic compressibility is incorporated in NorSand primarily through the parameters  (also used as a compressibility metric by Been et al., 1987a) and H. Elastic compressibility is equally important to the cavity resistance, although  * often neglected in the CPT literature, and is specified in NorSand as the rigidity Ir.  4.5. Calibration of NorSand  The aim in calibrating NorSand to a particular soil is to obtain a consistent set of material parameters that adequately represents that soil’s behaviour over the range of pressures and states. Hence the goodness of fit for any individual test is, by necessity, compromised to obtain a better representation of the overall soil response. This calibration process is summarised below; the reader is directed to Jefferies and Shuttle (2005) for a more detailed discussion of the calibration procedures. Determining the soil’s critical state is independent of the NorSand model, as any critical state model will have the same parameter set. For the semi-log idealisation of the CSL, a best-fit line is put through the critical states judged from high strain results on loose samples as shown in Figure 4-2 for Ticino 4 sand. Figure 4-3 compares all the CSLs for the nine investigated soils. As illustrated in Figure 4-1, N  and tc are obtained from the slope and intercept of the post peak portion of the stress-dilatancy plot for each triaxial test (Ghafghazi and Shuttle, 2006). Average values are then chosen to represent the overall soil behaviour. The plasticity parameter  was also derived from  -D plots, this time plotting dilation at peak stress ratio from Figure 4-1 against the state parameter at the image condition at the same point  124 d NorSand arameters for all nine soils. n from the literature, elastic shear modulus was obtained in a ariety of ways (see Table 4-4). Bender element measurements in the triaxial samples were used p As the sand properties were take v to measure G for Hokksund, Ottawa, Syncrude, Ticino and Toyoura sands. Consolidation measurements of unload-reload were used to infer elasticity for Erksak sand. No measurement of elasticity was made for Hilton mines tailings and Da Nang sand. Da Nang sand elasticity was estimated by curve-fitting to the triaxial tests. For Hilton mines, G was assumed to be similar to Toyoura sand.  4.6. Spherical Cavity Expansion Analysis  The spherical cavity expansion analogy idealises the CPT as a cavity in an infinite uniform medium under an isotropic stress state, with the internal pressure of the cavity initially equal to the in-situ mean effective stress 0p . The cavity is expanded by monotonically increasing its radius until a limiting (constant) pressure is obtained, this being the pressure of interest. This idealisation greatly simplifies the analysis because the spherical symmetry allows only radial displacements and in turn this permits a one-dimensional description of the problem. The corresponding stresses are a radial and two equal hoop stresses.  125 The spherical cavity finite element code used by Shuttle and Jefferies (1998), modified to incorporate changes to the NorSand model, is used in this study. The underlying code and finite element mesh remained the same, hence retaining the verified large displacement performance of the code. The one-dimensional problem was analysed using 70 elements, with a single stress sampling location at the centre of each element. The problem has no intrinsic measure of scale and so, for convenience, the original cavity radius was set to unity. The outer boundary was set cal Formulation o viously depends on the initial state of the sand. o observe the key aspects of this behaviour, cavity expansion from a dense initial state may be on up the envelope (which is as a zero displacement node at a distance of 500, chosen to avoid artefacts of boundary conditions. To capture the rapid variation of stress close to the cone, the element spacing was set to logarithmically increase with distance from the cavity. Similarly, a second-order numerical difference was used to extrapolate the element stresses to the cavity wall to accurately capture the high stress gradients present in the vicinity of the cavity.  4.7. Numeri  The s il response during spherical expansion ob T considered. And, because computation starts from “at rest” conditions, the behaviour of an element close to the cavity wall is typical because there is no intrinsic scale in the analysis. Figure 4-5a illustrates the stress path of an element of Ticino 4 sand initially at 74 kPa effective stress and with an initial void ratio of 0.596, corresponding to 0 = -0.285 based on the CSL from Table 4-4. Even though spherical cavity expansion is often referred to as being analogous to triaxial conditions, the stress path really amounts to a rapid excursion from isotropic conditions to the Mohr-Coulomb envelope followed by dilati  126 curved because it is a function of the state parameter which decreases as the sand dilates to critical conditions at a shear strain of more than 50%). The state path of the sand is shown in Figure 4-5b. Note that the sand moves up the CSL as the critical state is approached because the cavity has not reached its limiting expansion pressure at the time this particular element achieves criticality, and so the element tracks down the CSL as the confining stress increases.  4.8. Comparison to Calibration Chamber Data  In the current work the soil parameters for every calibration chamber test may be calculated knowing p′0 and e, and by inference 0. Each chamber test is modelled and the limit cavity ressure is calculated at 500% strain. The cavity pressure, being analogous to CPT qc, is then al resses using Equation 4-2 to obtain Qsph. The results are then lotted versus 0 as illustrated in Figure 4-6 for Ticino 4 sand. In order to compare the results p norm ised by the initial mean st p with those obtained in calibration chamber testing, normalised tip resistance Qcc for the Ticino 4 calibration chamber tests are also plotted on the same figure. It can be seen that Qcc results are significantly higher than Qsph. This mismatch is observed for every soil in the dataset and is quantified using a shape function. While the ratio between normalised tip resistances (Qcc / Qsph ) is a function of 0, plotting Qcc versus Qsph in Figure 4-7 suggests that for Ticino sand there is a one on one correlation between them (termed the shape function) irrespective of 0 or the stress level. Figure 4-7 shows that 91% of the data fall within the two dashed boundary lines at ±70% of the average trendline defined by Equation 4-3:  127   7.17.0 sphcc QQ           [Eq. 4 - 3]  This accuracy is close to the repeatability of Q typically achieved in calibration chamber testing itself. Hence a large part of the scatter in Figure 4-7 stems from the scatter in calibration chamber data itself, which is the maximum accuracy achievable by any analytical approach. Plotting Qcc versus Qsph for all of the database soils in Figure 4-8 shows that the shape function efined by Equation 4-3 appears independent of the soil type for all of the nine normally restingly, all these soils fall within the oundary lines drawn for Ticino 4 sand, with essentially the same overall confidence level of d consolidated clean sand-size soils considered. Inte b 89%, suggesting that no extra error has been introduced by comparing different soils.  4.9. Inverse form for Interpretation of CPT  Shuttle and Jefferies (1998) showed that Equation 4-1 may be used to recover 0 from calibration chamber data provided k and m are functions of soil characteristics and the stress level. Using a spherical cavity expansion analysis, we can summarise the effect of different soil characteristics as NorSand parameters in the following equation:  ksph = f1(G/p′0, Mtc, N, H, ,,)       [Eq. 4 - 4a] msph = f2(G/p′0, Mtc, N, H, ,,)        [Eq. 4 - 4b]  where ksph and msph are introduced as the spherical cavity expansion equivalents to k and m in Equation 4-1.  128 All the NorSand articular 0 all the variables in Equation 4-4 take a single value derived from Table 4-4 except ple om Equation 4-5 which is directly deduced from writing Equation 4-1 for spherical cavity  parameters in Equation 4-4 are constants, or known functions of e and p′.  At a p for G/p′0 (the stress level effect) as illustrated for Ticino 4 sand in Figure 4-9. Hence the trend of ksph and msph with  may be determined for a single value of G/p′0 from Qsph using Equation 4-4. This analysis is then repeated with different values of G/p′0.  By plotting the computed ksph and msph as functions of G/p′0, as illustrated for Ticino 4 sand in Figure 4-9, the complete functions are recovered. Values of ksph and msph as functions of G/p′0 are given for all nine investigated sands in Table 4-5. For each calibration chamber test, having Qsph determined from Equation 4-3, and ksph and msph from the equations in Table 4-5, one can infer the initial state parameter of the chamber sam fr expansion:  sph sph m Q   ln 0 sphk           [Eq. 4 - 5]  The inferred 0 is plotted versus the one obtained by experimentally measuring the void ratio in ese error margins, as well as the level of confidence that the method can predict 0 within ber tests investigated in this work, 66 predictions produced a 0 Figure 4-10. Two boundary lines are drawn beside the line of equivalency in Figure 4-10, presenting ±0.04 and ±0.07 error margins. Table 4-6 summarises the number of tests with 0 calculated within th both error margins. From 301 calibration cham  129 error of more than 0.04 and only 24 tests predicted 0 of more than ±0.07, equivalent to confidence levels of 78% and 92% respectively. However, the method did not produce equal accuracy for all sands.  Ottawa, Toyoura 160 and Syncrude Oil Tailings showed excellent prediction. Da Nang and Ticino yielded an accuracy close to the average. Lower accuracy was obtained with Hokksund, Hilton Mines and Erksak sands. Hokksund calibration chamber data was collected from different sources and had the greatest scatter in Q - 0 among the database soils. As a result of this scatter in the raw data this sand yields a poor 0 prediction. The second “problem” sand, Hilton Mines, has long been known for ny of the sands considered here. The presence of fines can make sample uniformity  achieve. This can in turn reduce the repeatability of the tests and affect the accuracy of the predictions. Eliminating these three soils from the database would improve predictions to 84.3% because detailed simulation of the CPT presents a substantial challenge in mechanics. The present work adopted an intermediate step, using the analogy of limit pressure in spherical cavity expansion to CPT tip resistance. Cavity expansion allows fast numerical computation of being the most compressible among those tested in calibration chambers. The poor 0 prediction for Erksak sand is less well understood, but may have resulted from the consolidation unload- reload loops under-predicting the elastic stiffness. In addition, Erksak sand had the highest fines content of a much more difficult to  confidence of 0 less than ±0.04 and 97.2% less than ±0.07.  4.10. Discussion  The CPT is an important and widely used test, with particular relevance for cohesionless soils. However, theoretical understanding of the test has not matched its widespread use. In part this is  130 limit pressure, with a large strain approach and using sophisticated soil models. Of course, a central question then follows: how close is the cavity expansion analogy to the real CPT. This issue was dealt with by introducing a shape function; defined by Equation 4-3. This shape function incorporates the spherical to true geometry scaling, as well as model simplifications including neglecting particle breakage and use of constant elasticity. A central finding of the  a century. lastic hardening, which will have a model-dependent parameter scaling it, is central to any alisation – so, although a different parameter to H occurs with the various odels, these hardening parameters have a similar role and can be expected to affect the present work is that this shape function is the same for all nine soils evaluated, from which we can infer that the shape function is unique since the sands involved cover a reasonable range of properties. The finding of a unique shape function should not be influenced by the constitutive model adopted (e.g. Cudmani and Osinov, 2001), although the state parameter is central. Advanced plasticity models for soil with density independent parameters are, to date, almost invariably based on the state parameter (e.g. Manzari and Dafalias, 1997; Wan and Guo, 1998; Li et al., 1999; Gajo and Wood, 1999). Their properties are also similar – for example, the semi-log idealisation for the CSL is verging on ubiquitous and leads to  being common. Similarly, the concept of a critical friction ratio (angle) has been uncontroversial for at least half P work hardening ide m computed results as much as H does. As for elasticity, an isotropic shear modulus will be the default starting point for any elasto-plastic model. A unique function means that spherical cavity solutions using a ‘good’ soil model (i.e. one whose parameters are density independent) can provide the framework for estimating 0 from the CPT qc data for any cohesionless soil. But this misses an important opportunity. Although  131 more practical than calibration chamber testing, the dependence of the inversion from qc to 0 on soil properties is inconvenient since properties change from one stratum to the next and it is clearly desirable to avoid a suite of triaxial tests for each soil type in a CPT profile. There is an opportunity to minimise supporting laboratory testing by using additional information measured by the CPT – for example, friction sleeve data are widely used as a basis to infer soil type. Plewes et al. (1992) suggested that at least  might be estimated this way. Looking to the future, the success of the spherical cavity analogy presented here should be viewed as the starting point for proper understanding of the CPT. Although difficult, detailed modelling of actual CPT 0 0 0 are those penetration in NorSand (or similar theoretical soil) will allow understanding of how H, etc. can be related to sleeve friction – markedly improving the available precision in inferring 0 in practical situations. Turning to accuracy in estimating  , the spherical approach is shown to provide excellent resolution for six of the nine sands investigated. The question then becomes, how can resolution be improved for all soils? Hokksund sand appears to provide poor predictions due to the variability in its calibration chamber database, rather than any systematic failing in the approach. This calibration chamber variability is illustrated on Figure 4-6 for Ticino 4 where the recorded CPT resistance for Tests E139 and I169 differ by more than a factor of 2, but the tests have virtually identical initial conditions in terms of stress state, geostatic stress ratio (K0), and void ratio. Hilton Mines tailings sand also shows no systematic bias in error over the range of  . Looking at Table 4-6 one can see that the soils yielding the best accuracy in with bender elements measurements of the elastic shear modulus G (i.e. Ottawa, Syncrude, and Toyoura). High quality G measurement has shown that G can easily vary by  50% because of sample preparation all else equal (Bellotti et al., 1982). Hence the lack of repeatability seen in  132 the chamber test data could credibly be caused by changes in G from one calibration chamber experiment to the next, even under ostensibly similar situations. Shuttle and Jefferies (1998) suggest a factor of  50% on G will produce about   0.03 on 0, which is about half the residual uncertainty. Therefore it is strongly recommended that G be measured to eliminate an unnecessary error in evaluation of the in-situ state parameter. It is well known that both fabric and ageing affect the soil response. Hence while critical state parameters are independent of fabric, and can be determined from triaxial testing, and G may be measured in-situ using the seismic CPT; hardening modulus H will likely vary between the lab and the field. Despite this, inspection of Table 4-2 and Table 4-3 suggests that differing reconstitution methods in calibration chamber and triaxial tests did not significantly reduce the accuracy of the inversion for 0. Hence one might expect reasonable state predictions in-situ. t  4.11. Summary and Conclusion  A framework for evaluation of in-situ soil state parameter 0 from CPT qc data has been developed based on a spherical cavity expansion analogue. Central to the approach is a shape function that relates the normalised spherical and calibration chamber limiting pressure. A database of nine normally consolidated soils, including laboratory standard and natural sands as well as tailings materials, was compiled for which both calibration chamber tests and triaxial ests were available. The NorSand critical state soil model was calibrated for each of the nine soils using the triaxial data. Spherical cavity expansion analysis was then performed for 301 normally consolidated calibration chamber tests and the results compared with those obtained experimentally. All nine soils followed the same trend with a scatter equivalent to that observed  133  134 iaxial compression tests) the soil 1 hamber tests and the resulting states compared to those measured in the lab. The method % confidence and to within ±0.07 with 92% confidence. wo sources of error were explicitly identified; inherent variability in the calibration chamber Ticino for the most intensely tested calibration chamber sand, Ticino. A unique and consistent shape function between the normalised experimental CPT and spherical cavity limit pressures was identified. The shape function was shown to be insensitive to soil type and density. Estimation of accuracy of the proposed framework is constrained by the repeatability seen in the calibration chamber data. For known soil properties (from tr specific k and m relation recovers qc to near perfect accuracy, and without bias. Individual calibration chamber results cannot be recovered to this accuracy, but that appears a consequence of the intrinsic variability in the calibration chamber data. Consequently, it was shown that the normalised cone resistance can be analytically reproduced with an accuracy of ±70% to 89% confidence. The spherical analysis results were then used to back calculate 0 using Equation 4-5 for all 30 c predicted 0 to within ±0.04 with 78 T testing and elastic modulus. Clearly no analytical method can predict 0 with accuracy higher than the repeatability of the calibration chamber data set itself, and this inherent variability between calibration chamber tests likely accounts for about half the observed error. For sand the estimated error in interpreted 0 due to variations in G was estimated as up to ±0.03. Hence, it is recommended that G be measured in-situ to eliminate an unnecessary source of error. Table 4 - 1 Index pro ie s d d  t g                        Sand/Sou   Property Da Nan , Erksak /3 n , 7 to ne in n , 7 kk : n , 7 aw n , 7 ncr li d o 7 cin en ., 87 ci ld so s p 8 y 0: or te al 91 pert s of tudie  san s and ailin s rce g: Hsu 1999 355 .0: Bee  et al. 198  b Hil n Mi s tail gs: Bee  et al. 198 a Ho sund Bee  et al. 198  a Ott Bee 198 a: et al.  a Sy ude Oil Tai ngs: Gol er Ass ciates, 198 a Ti o 4: Be  et al 19  a Ti no 9: Go er As ciate Re ort, 19 7b To oura 16 Fi avan et ., 19 Mineralogy Quar o ou f rt r nor ou f rt ar e s sc , a, e ds ar e rtz Quar al ou m s scret ve d p ar Trace of ca s  b mi t ci nd Main ua 3% c tz, min r am nts o che Qua tz, mi am nts o che Qu tz, som feld par, mu ovite mic heavy min rals Fel par, Qu tz, som  mica Qua tz Sm l am nt of bitu en a di e gra l size lum s Qu tz Mi As umed to e si lar to tha of Ti no 4 sa ly Q rtz, hert Median grain size D50: m 0    0 0113 355 200 390 530 207 53 530 16 Effective grain size D10: m   680 180 110 210 350 - 360 - - Uniformity coefficient  5 1.84 2.2 2.0 2.0 1.7 1.9 1.6 - 1. Percentage passing no. 200 sieve 0 3-6 2.5 0 0 3.5 0 0 0 Specific gravity of particle 2 6 7 652.61 2.65 3.0 2.70 2.6 2.64 2.70 2.6  2. Average sphericity 2 3 9 - 0.75 0.7 0.77 0.8 - 0.7 - - Average roundness 3 5 8 - - 0.2 0.32 0.5 - 0.3 - - Grain description - ul angular - n gu - angular unded gu - angular, stl cubical b- n b-sub ang ar to Sub rou ded An lar Sub Ro An lar to sub mo y Su rou ded - Su angular Maximum voids ratio emax 08 63 5 9 98 0 977 0.8  0.9  1.0 0.91 0.7 0.8  0.89 0.9  0. Minimum voids ratio emin 15 25 2 9 44 0 605 0.5  0.5  0.6 0.55 0.4 0.5  0.60 0.6  0.  135 Table 4 - 2 Summary of drained triaxial compression tests used in calibration of NorSand Sand Reference Number of tests Sample prepara Range of void ratio Range of mean stress p0' (kPa) tion D H ed 0  50-400 a Nang su, 1999 15 Wet pluviat .56-0.66 Erksak 355/3.0 Been et al., 1987 b 4 Moist tamped 0.54-0.63 50-400 Hi  M s G er o es, 1985 6 Moist tamped 0.71-0.87 80-1000 lton ine old  Ass ciat H N ech. Inst.,1982 15 Unknown 0.51-0.72 34-233 okksund orwegian Geot Ot G 1985 6 Moist tamped 0  50-220 tawa older Associates,  .53-0.74 Sy Ta Gol 5 ated 0. 0-500 a ncrude Oil der Associates, 1987 Wet pluvi 53-0.71 10 ilings Ti Gol ssociates, 1986 5 Moist t ed 0.65-0.85 100-200 cino 4 der A amp T G ociates, 1987b 5 ed 0  50-300 icino 9 older Ass  Dry pluviat .64-0.81 T  G ociates, 1989 ed 0  33-1000 oyoura 160 older Ass  14 Wet pluviat .61-0.84     Table 4 - 3 Summary of calibration chamber tests on normally consolidated soils San Reference No. of BC1 Sat. Chamber ratio D le void ratio e p0' kPa Chamber descrip. d Samp prep. tests /d D Hsu, 1 44.2 pluv. 0.56 0.73 - 150 Hsu & Huang, 1998 a Nang 999 38 BC5 Dry 22.1, Dry - 33 E 35 Been et al., 1987b 14 BC4 Sat. 38 Mo  0.53- 0.66 25- 0 Been et al., 1987b rksa 5/3 k .0 ist ampedt  30 H Mines Harm 20 B BC3 Sat. pluv. 0.65 0.95 - 180 Laier et al., 1975 ilton an, 1976 C1, Dry, 34.2 Dry - 30 Hokksund arkin et al ; Bald ;  a 51 BC1, BC3 Dry 33 0.58- 0.87 25- 370 Bellotti et al. ,1982; Chapman, 1974  P 1980 1986 1987 ., i et al.,  Been et al.,  .6-48 Dry pluv. Ottawa arman, 1976 30 BC1, BC2 Dry, Sat. 34.2 Dry plu 0.54- 0.73 30- 200 Laier et al., 1975  H v. Sy Oi Ta ncrud l ilings e  Gol A de ssoc 8 BC4 Sat., 38 We tamped 0.55 0.70 - 0 Been et al., 1987 b r iates, 1987a t pluv. oist -  30 40moist M Ticino 4 Baldi et ., 6 90 BC1, B B B Dry 33.6 Dry plu 0.50- 0.90 50- 0 Bellotti et al., 1982  al 198 v. C2, C3 C4  50 ,  Ticino 9 Golder Associates, 1987b 9 BC4 38 - 0.63- 0.88 30- 5 Been et al., 1987 b -  31 Toy 160 o Iwasaki et al., 1988; Fioravante et al., 1991 39 B B B 2 1 Dry pluv. 0.64 0.85 - 7 Iwasaki et al., 1988; Bellotti et al., 1982 ura C1, C2, C3  D S ry at , . 2.1- 20 -  30 17 1 B BC vol  oun 3, um dar con e ba y conditio C1, constant stress side and base restraints; BC2, constant  side and base restraints; stant u ide restraint, const se 4, constan  side restraint, constant se re in lgado et al., 1998); estraint, co olume base restraint. ns: B me s t (Sa  vol t s nst ume tress ant v  vol stra ant  BC  str 5, ess ser  ba vo-co  res ntro tra lled si int; BC de r 136 Table 4 - 4 NorSand parameters for nine CPT calibration chamber sands Elasti CSL  city1Plasticity P ter  e Mtc N* arame H Ir    Da Nan 0.99 7 25 0.35 x 20. g 0.031 1. 40-2200 4.3 300-  20000 From fitting to tria ial data E 0.8 34 0.019 1.25 0.32 rksak 355/3.0 100 -4000 3.8 50.0  *min  refpe 195     p e After Jefferies and Been 6) 55  at compressibility becomes zero. 0.2 (200 *e =0.3  is the void min etric ratio  which volum Hilton ines 1.315 0.0738 1.40 0.20 50 3.5 Equal to Toyoura 160 0.2 M Hokksund 0.934 0.0235 1.00 0.40 138+5130+ 22  2 4.0 50 0 54.0 0'720   p 30.  f After es l (199 0.2 1e  ti et a   rep  L Pro . 2) O .0122 5 80ttawa 0.754 0 1.24 0.4 1 -4000 4.0  29.5 e 48.0 0   23 1 66.2    p p e Afte erts  a (199 0.2 1.8  ref   e r Rob 5) on et l. Syncrude, Oil T 1.27 43-2960 + 192702  0.28 ailings 0.890 0.0283 5.8   48.0 0.0 0 11.31 1 64.2      ref pK e After Cunning 995) 228.8 e 0.  e 26   p   et al.  (1 T 0  icino 4 0.986 0.0243 1.27 0.40 70 -200 3.0 52.0 0   p p 3.1 650     refe After Shuttle & Jefferies (199 20. 8) Ticino 9 1.33 0.40 60-3500 3.8 Equ ici0.970 0.0217 al to T no 4 0.2  T ra 160 1.28   oyou 0.983 0.019 0.41 100 4.4   47.00  0 2  1 17.2878   p p e e .2  ref After Chaudhary et al. (2004) 1 p  is equal to 100 kPa or equivalent ref  137 Table 4 - 5 Equations for ksph and msph as functions of G/p′0 for database soils ksph m   Sand sph Da Nang 5  + 0  G 4.  0.465 /p0) .52 .0007 /p0 50 +  ln(G Erksak 355/3 (G  .0 6.42 + 0.58 ln /p0) 0.86 + 0.79 ln(G/p 0) Hilton Mines 3.19 + 0.90 ln(G/p0) 2.68 + 0.081 ln(G/ ) p0 Hokksund 15.0 – 1.31 ln(G/p0) -5.37 + 1.78 ln(G/p0) Ottawa  + (G p 1.26 + 0.733 l G/p8.05 0.66 ln / 0) n( 0) Syncrude Oil  1 – ( 5 p0) Tailings 11.2  0.51 ln G/p0) -2.62 + 1.7  ln(G/ Ticino 4 5.27 + 0.56 ln(G/p0) 2.62 + 0.316 ln(G/p0) Ticino 9 8.74 + 0.0006 G/p0 2.07 + 0.59 ln(G/p0) Toyoura 160 10.47 – 0.10 ln(G/p0) 2.15 + 0.50 ln(G/p0)     4 - 6 a 0 n nine base Sand Total of tests Number of t 0 > 0.04 Number of  0.07 Confidence Level that 0 < 0.04 Level that 0 < 0.07 Table Summ ry of   obtai ed for data soils Number ests with  tests with 0 >  Confidence Da Nang 38 11 2 71.1 % 94.7 % Erksak 355/3.0 14 6 2 .7 % 57.1 % 85 Hilton Mines 20 11 5  75.0 % 45.0 % Hokksund 5  70.6 % 78.4 % 1 15 11 Ottawa 30 1 0 96.7 % 100.0 % Syncrude Oil Tailings 8 1 0 87.5 % 100.0 % Ticino 4 90 16 3 82.2 % 96.7 % Ticino 9 9 3 1 66.7 % 88.9 % Toyoura 160 41 2 0 95.1 % 100.0 % Sum 301  66 24  % 78.1 % 92.0     138 1.6 1.5 0.6 0.7 0.9 1.0 3 0.0  = d  v / d * 0.8 1.4 1-N M tc 1 1. 1.2 1.1 q  / p' = -0.4 0.4 D D  q min  Figure 4 - 1 Stress-dilatancy plot ( - D o obtain M nd N* (Ticino 4, C264)  -0.2 0.2 ) t tc a CSL 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95 Initial isotropic state 10 100 1000 p' (kPa) e Critical State  Figure 4 - 2 CSL determination for Ticino 4 139 Da Nang Erksak 355/3.0 Hilton Mines Hokksund Ottawa Syncrude Tailings Ticino 04 Ticino 09 Toyoura 0.60 0.70 0.80 0.90 1.00 1.10 10 100 1000 p' (kPa ) e  Figure 4 - 3 Comparison of CSL for all nine sands  0 100 200 300 400 0 1 2 3 vo lu m et ric  s tra in , v  (% ) Model Test 0 100 200 300 400 kP a ) 500 0 5 10 15 20 25 axial strain: % de vi at or  s tre ss , q  ( Test Model 500 kP a ) 600 0 5 10 15 20 25 axial strain: % de vi at or  s tre ss , q  ( Test Model -4 -3 -2 -1 0 1 vo lu m et ric  s tra in , v  (% )  Model Test  Figure 4 - 4 NorSand fits to loose and dense Ticino 4 triaxial data (C262, C264) 140 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 p' (MPa ) q ( M Pa ) q=Mp'  a) Stress path  CSL 0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.01 0.1 1 10 p' (MPa ) vo id  ra tio , e  b) Void ratio versus log of mean effective stress  Figure 4 - 5 Behaviour of an element close to the cavity during spherical expansion   141 110 100 1000 -0.35 -0.30 -0.25 -0.20 -0.15 -0.10 -0.05 0.00 0.05  0 Q cc  , Q sp h E139 Calibration Chamber Testing Spherical Cavity Expansion Analysis I169  Figure 4 - 6 Qcc and Qsph vs. 0 for Ticino 4 sand  10 100 1000 1 10 100 Q sph Q cc +70 % -70 % Q cc = 0.7Q sph 1.7  Figure 4 - 7 Qcc vs. Qsph for Ticino 4 sand  142 10 100 1000 1 10 100 Q sph Q cc Da Nang Erksak 355/3.0 Hilton mines Hokksund Ottawa Syncrude Ticino 04 Ticino 09 Toyoura 160 +70 % -70 %  Figure 4 - 8 Qcc vs. Qsp for all nine database sands  h  m sph  = 2.62 + 0.316 ln (G / p 0'  ) k sph  = 5.27+ 0.56 ln (G / p 0'  ) 4 6 8 10 100 1000 10000 G / p 0' m sp h , k sp h  Figure 4 - 9 Effect of G/p′  on k  and m  for Ticino 4 sand 0 sph sph  143 -0.50 -0.40 -0.30 -0.20 -0.10 0.00 0.10 -0.40-0.30-0.20-0.100.000.10  0 measured  0  c al cu la te d Da Nang Erksak 355/3.0 Hilton mines Hokksund Ottawa Syncrude Ticino 04 Ticino 09 Toyoura 160 +0.07 -0.07 +0.04 -0.04  Figure 4 - 100 from Equation 4-5 vs. 0 measured in the chamber 144 4.12. References  Baldi G., Bellotti R., Ghionna V.N., Jamiolkowski M., and Pasqualini E. 1986. Interpretation of CPTs and CPTUs, 2nd Part: Drained Penetration of Sands. Field Instrumentation and In- situ Measurements. Proceedings of the 4th International Geotechnical Seminar, Singapore, Nanyang Technological Institute, 143-156. Been K., Jefferies M.G., Crooks J.H.A., and Rothenburg L. 1987a. The Cone Penetration Test in Sands. Part II: General Inference of State. Géotechnique, 37(3): 285-299. Been K., Lingnau B.E., Crooks J.H.A., and Leach B. 1987b. Cone Penetration Test Calibration for Erksak (Beaufort Sea) Sand. Canadian Geotechnical Journal, 24: 601-610. Been K., and Jefferies M.G. 1985. A State Parameter for Sands. Géotechnique, 35(2): 99-112. Been K., Crooks J.H.A., Becker D.E., and Jefferies M.G. 1986. The Cone Penetration Test in Sands: Part I, State Parameter Interpretation. Géotechnique, 36(2): 239-249. n and Use of a Calibration Chamber. Proceedings of the 2nd European Symposium on Penetration Testing, Amsterdam, 2: 439-446. van den Berg P. 1994. Analysis of Soil Penetration. Delft University of Technology, PhD Thesis, ISBN 90-407-1004-X. Bishop R.F., Hill R., and Mott N.F. 1945. Theory of Indentation and Hardness Tests. Proceedings  of Physics Society, 57: 147–159. de Borst R., and Vermeer P.A. 1982. Finite Element Analysis of Static Penetration Tests. Proceedings of the 2nd European Symposium on Penetration Testing, Amsterdam, 2: 457-562. Bellotti R., Bizzi G., and Ghionna V. 1982. Design, Constructio 145 Carter J.P., Booker  J.R., and Yeung S.K. 1986. Cavity Expansion in Cohesive Frictional Soils. Géotechnique, 36(3): 349–358. Chapm m, 2: 59-65.  and Triaxial Collins d Loading n Cudma pretation of Cunnin ls. Canadian Geotechnical Journal, 32: 848-858.  or Numerical and Analytical Methods in Chadwick P. 1959. Quasi-Static Expansion of Spherical Cavity in Metals and Ideal Soils. The Quarterly Journal of Mechanics and Applied Mathematics, 12(1): 52-71. an G.A. 1974. A Calibration Chamber for Field Test Equipment. Proceedings of the 1st European Symposium on Penetration Testing, Stockhol Chaudhary S.K., Kuwano J., and Hayano Y. 2004. Measurement of Quasi-Elastic Stiffness Parameters of Dense Toyoura Sand in Hollow Cylinder Apparatus Apparatus with Bender Elements. Geotechnical Testing Journal, 27(1): 1-13.  I.F., Pender M.J., and Yan W. 1992. Cavity Expansion in Sands under Draine Conditions. International Journal for Numerical and Analytical Methods i Geomechanics, 16: 3-23. ni R., and Osinov V.A. 2001. The Cavity Expansion Problem for the Inter Cone Penetration and Pressuremeter Tests. Canadian Geotechnical Journal, 38: 622-638. g J.C., Robertson P.K., and Sego D.C. 1995. Shear Wave Velocity to Evaluate In Situ State of Cohesionless Soi Fioravante V., Jamiolkowski M., Tanizawa F. and Tatsuka F. 1991. Results of CPT's in Toyoura Quartz Sand. Calibration chamber testing. Huang A.B. (ed.), Elsevier, 135-146. Gajo A., and Muir Wood D.A. 1999. Kinematic Hardening Constitutive Model for Sands: The Multiaxial Formulation. International Journal f Geomechanics, 23(9): 925-965. 146 Ghafghazi M., and Shuttle D.A. 2006. Accurate Determination of the Critical State Friction Angle from Triaxial Tests. Proceedings of the 59th Canadian Geotechnical Conference, Golder Golder  Syncrude Tailings. Golder Golder orida, MSc. thesis. Holden Hsu H. s. PhD thesis, Hsu H. or Cone Huang  Computers and Geotechnics, 31: 517–528. Iwasaki k., Tanizawa F., Zhou S., and Tatsuka F. 1988. Cone Resistance and Liquefaction Strength of Sand. Proceedings of the 1st International Symposium on Penetration Vancouver, 278-284.  Associates. 1985. Report No. 852-2041. Golder Associates. 1986. Report No. 852-2042.  Associates. 1987a. Cone Penetrometer Calibration Chamber Tests on Report No. 872-2402.  Associates. 1987b. Report No. 862-2801.  Associates. 1989. Report No. 892-2021. Harman D.E. 1976. A Statistical Study of Static Cone Bearing Capacity, Vertical Effective Stress and Relative Density of Dry and Saturated Fine Sands in a Large Triaxial Testing Chamber. University of Fl Hill R. 1950. The Mathematical Theory of Plasticity. Oxford University Press, Oxford.  J.C. 1991. History of the First Six CRB Calibration Chambers. Proceedings of the 1st International Symposium on Calibration Chamber Testing, Potsdam, New York, 1-12. H. 1999. Cone Penetration Tests in Sand under Simulated Field Condition Department of Civil Engineering, National Chiao Tung University, Hsin Chu, Taiwan. H., and Huang A.B. 1998. Development of an Axisymmetric Field Simulator f Penetration Tests in Sand. Geotechnical Testing Journal, 21(4): 348-355. W., Sheng D., Sloan S.W., and Yu H.S. 2004. Finite Element Analysis of Cone Penetration in Cohesionless Soil. 147 Testing, Orlando, 20-24 March 1988, de Ruiter. J., and Balekma A.A. (eds.), Rotterdam, 2: 785-791. Jamiolkowski M., Ladd C.C., Germaine J.T., and Lancellotta R. 1985. New Developments in roceedings of the 11th ICSMFE, San Francisco, nce on Soil Constitutive Models: Evaluation, Selection, and ublication, 128: 204-236. atancy in General Cambridge-Type Models. Jefferie aylor & Laier J in 9-611. eotechnical Field and Laboratory Testing of Soils. P CA. Jefferies M.G., and Shuttle D.A. 2005. NorSand: Features, Calibration and Use. Proceedings of the Specialty Confere Calibration. ASCE Geotechnical Special P Jefferies M.G., and Shuttle D.A. 2002. Dil Géotechnique, 52(9): 625-638. s M.G. 1993. Nor-Sand: a Simple Critical State Model for Sand. Géotechnique, 43(1): 91-103. Jefferies M.G., and Been K. 2006. Soil Liquefaction: A Critical State Approach. T Francis, Abingdon & New York, ISBN 0-419-16170-8. .E., Schmertmann J.H., and Schaub J.H. 1975. Effect of Finite Pressuremeter Length Dry Sand. Proceedings of the Conference on Insitu Measurement of Soil Properties. Raleigh, ASCE, New York, 241-259. Li X.S., Dafalias Y.F., and Wang Z.L. 1999. State-Dependent Dilatancy in Critical-State Constitutive Modelling of Sand. Canadian Geotechnical Journal, 36(4): 59 Lo Presti D.C.F., Pedroni D.C.F., and Crippa V. 1992. Maximum Dry Density in Cohesionless Soil By Pluviation and by ASTM D 4253-83: A Comparative Study. G Testing Journal, 15(2): 180-189. 148 Manza e, 47(2): 255-272. ber Testing, ISOCCT 1, Potsdam, New York, 257-264. Parkin tibility. Proceedings of the 45th Canadian Geotechnical Robertson P.K., and Campanella R.G. 1983. Interpretation of Cone Penetration Tests. Part I: Robert 1995. Shear-Wave Velocity to Russell A.R., and Khalili N. 2002. Drained Cavity Expansion in Sands Exhibiting Particle ri M.T., and Dafalias Y.F. 1997. A Critical State Two-Surface Plasticity Model for Sands. Géotechniqu Mayne P.W., and Kulhawy F.H. 1991. Calibration Chamber Database and Boundary Effects Correction for CPT Data. Proceedings of the 1st International Symposium on Calibration Cham Norwegian Geotechnical Institute 1982. Report No. 842-2007. A., Holden J., Aamot K., Last N., and Lunne T. 1980. Laboratory Investigation of CPTs in Sand. Norwegian Geotechnical Institute, Report 52-18-9. Plewes H.D., Davies M.P., and Jefferies M.G. 1992. CPT Based Screening Procedure for Evaluating Liquefaction Suscep Conference, Toronto. Sand.  Canadian Geotechnical Journal, 20(4): 718-733. son P.K., Sasitharan S., Cunning J.C., and Sego D.C. Evaluate In-Situ State of Ottawa Sand. ASCE Journal of Geotechnical Engineering, 121(3): 262-273. Crushing. International Journal for Numerical and Analytical Methods in Geomechanics, 26(4): 323–340. Salgado R., Mitchell J.K., and Jamiolkowski M. 1997. Cavity Expansion and Penetration Resistance in Sand. Journal of Geotechnical and Geoenvironmental Engineering, 123(4): 344-354. 149 Salgado R., Mitchell J.K., and Jamiolkowski M. 1998. Calibration Chamber Size Effects on Penetration Resistance in Sand. Journal of Geotechnical and Geo-environmental Shuttle n Geomechanics, 22: 351-   Testing and Materials, Selig E.T., and Ladd R.S. (eds.), Wan R l for Granular Soils: Modified Willson g in the UK, Thomas Telford, London, 157-159. Engineering, ASCE, 124(9): 878-888.  D.A., and Jefferies M.G. 1998. Dimensionless and Unbiased CPT Interpretation in Sand. International Journal for Numerical and Analytical Methods i 391. Sladen J.A. 1989. Problems with Interpretation of Sand State from Cone Penetration Test. Géotechnique, 39(2): 323-332. Tavenas F.A. 1973. Difficulties in the Use of Relative Density as a Soil Parameter. Evaluation of Relative Density and its Role in Geotechnical Projects Involving Cohesionless Soils, American Society for Philadelphia, ASTM Special Technical Publication 523: 478-483. .G., and Guo P.J. 1998. A Simple Constitutive Mode Stress-Dilatancy Approach. Computers and Geotechnics, 22(2): 109-33.  S.M., Ims B.W., and Smith I.M. 1988. Finite Element Analysis of Cone Penetration. Penetration Testin Yu H.S., and Houlsby G.T. 1991. Finite Cavity Expansion in Dilatant Soils: Loading Analysis. Géotechnique, 41(2): 173–183. 150 Chapter 5.   Evaluation of Sand State from SBP and CPT: A C     is an alm Dr can be misleading void ratios upon which function of mean stress. An alternative to Dr that captures both the effect of void ratio and the effect of mean stress on soil behaviour is the state parameter,  (Been and Jefferies, 1985). Different soils, or the same soil at different stress levels, display similar behaviour at the same value of    ase History5 5.1. Introduction The behaviour of cohesionless soils depends strongly on their density. While relative density Dr ost universally used density index for sand, it is easy to show that (e.g. Tavenas, 1973). Apart from the lack of accuracy in identifying minimum and maximum  Dr depends, it is well understood that behaviour of soils is also a  5 A version of this chapter has been published. Ghafghazi M., and Shuttle D.A. 2008. Evaluation of Soil State from SBP and CPT: A Case History. Canadian Geotechnical Journal, (45)6: 824-844. 151 Regardless of whether  or Dr is used as the characterisation parameter, outside of a few search situations it is impractical to obtain undisturbed samples of cohesionless soils. Engineering of cohesionless soils must be based on a combination of true properties determined om disturbed or reconstituted samples, with in-situ measurements to determine the value of the ate measure adopted. PT and CPT have similarities, measuring a resistance (N or qc respectively) to an imposed displacement of c CPT emerged as the appropriate penetration test for the offshore oil industry in the 1970’s, has been further enhanced over the subsequent decades with additional transducer channels (induced pore pressure in particular), in-tool correction for thermal drift, in-tool A-to-D conversion, and even wireless data transmission. Electronic CPT equipment is now displacing earlier mechanical CPT and SPT everywhere, this modern version offering a continuous data record, excellent repeatability, excellent accuracy, and relatively low cost. This chapter is based on modern electronic CPT data as described in ASTM D5778. The difficulty with any penetration test, however, is that the state measure of interest (e.g. Dr, ) is not measured directly; instead the chosen state measure is calculated from the tip resistance qc, a process usually referred to as interpretation. Interpretation based on mechanics involves the solution of an inverse boundary value problem to obtain mechanical properties of the soil from test results. But the large deformations associated with these penetration problems, along with the nonlinear behaviour of the soil and complicated boundary conditions make the analysis an extremely difficult task, and the solution non-unique. The interpretation framework is also difficult to establish. No simple closed-form solution for  or Dr from CPT (nor the more complex boundary conditions of SPT) has been developed; and, nobody - to date - has provided re fr st For most situations penetration tests are the basic in-situ reference tests. Both S the tool. The modern electroni 152 a full numerical simulation of drained penetration that matches calibration data, although several have tried (e.g. De Borst and Vermeer, 1982; Willson et al., 1988; Van den Berg, 1994; Huang et al., 2004; and Ahmadi et al., 2005). Two different directions have emerged to estimate soil state from CPT data: correlations established through calibration chamber tests and simplified theoretical treatments. Calibration chambers are circular steel tanks typically about a metre in diameter and similar height. Sand is deposited at a known density and consolidated to the desired stress state within the tank, and a cone penetration test is then performed along the vertical axis of the sample exactly as in natural ground. Each test provides a qc for the given value of density and stress of the sample. A large number of tests, covering the range of densities and stresses of interest, provide the relation between qc, in-situ effective stresses (1,2, 3), and the density (more usually expressed as  or Dr) for the tested material. The in-situ state is then obtained from the CPT by comparison of field CPT qc measurements, at the estimated in-situ stresses, to the qc relation determined in the calibration chamber (and similarly for Dr). There is a considerable history to calibrating the CPT in large chambers, with chambers increasing in size over time and becoming less subject to corrections for finite chamber size (boundary effects). The first advanced calibration chamber was built in 1969 at the Country Roads Board (CRB) in Australia (Holden, 1991). Chamber tests are now reported in the literature with differing dimensions, nature and form of control of boundaries, deposition procedure, and capability to handle saturated specimens. Ghionna and Jamiolkowski (1991) provided a list of 16 calibration chamber tests in the literature. More calibration chambers have been built since (e.g. Peterson and Arumoli, 1991; Hsu and Huang, 1998; Ajalloeian and Yu, 1998; and Tan et al., 2003). 153 Although chamber tests data appear to provide unarguable calibration for the CPT, this substantial body of experimental data is not sufficient in general. In addition to soil fabric and ageing which are known to complicate correlation of laboratory response to in-situ behaviour, there are two difficulties specific to CPT interpretation. First, the qc relation differs from one soil to another so that, although the form remains common among soils, the coefficients n effort beyond soil models have been used allowing for plastic hardening, strain involved are particular to the calibrated soil. Second, the relation is controlled by the mean effective stress p, not the readily determined in-situ vertical effective stress v, so that the horizontal geostatic stress ratio K0 becomes important to accurate determination of  or Dr. These two issues have been addressed below. In the case of calibration coefficients changing from one soil to another, simplified theoretical modelling has been used to develop interpolations between different calibrations since it is impractical to calibrate the CPT for soils encountered in routine practice. A single chamber test involves preparing a uniform sample of between one and two tonnes at uniform density, and a calibration of the CPT in a chamber for one soil involves many chamber tests – a that affordable by all but a few large projects. In addition, even large projects must account for the soil being non-uniform in gradation, with place to place variations even within a defined geological stratum. Many simplified theoretical treatments have used spherical cavity expansion as an analogue of the penetration test, essentially the same approach as used in conventional design of end bearing capacity of piles. Robertson and Campanella (1983) first suggested that the difference between the CPT calibrations in various sands might be understood in terms of an undefined ‘compressibility’ of the sand involved. In more comprehensive evaluations of factors causing differing calibration, ‘good’ 154 softening, etc. Shuttle and Jefferies (1998) used a general work hardening/softening critical state model to evaluate changes in CPT calibration in terms of critical state parameters (the critical state friction ratio and slope of the critical state line in pe  log  space, respectively) that can be determined in routine triaxial testing of reconstituted samples. They showed that CPT behaviour in Hilton Mines sand, currently the most unusual of the published calibrations, could be predicted based on Ticino sand data by allowing for the changes in fundamental soil properties between Hilton Mine and Ticino sand. However, this work is more in the nature of verification than validation since it does not involve independent evaluation of ‘ground truth’ in-situ at prototype scale – and stress conditions are perfectly known. thus there is a validation question of just how well  can practically be Turning to the issue of stress conditions, current practice appears to involve widespread neglect of K0, despite the chamber test calibrations showing that K0 is important. Jefferies et al. (1987) drew attention to the situation, pointing out the error caused by neglecting K0 was credibly as large as the error caused by uncertainty in the effect of different sands on the CPT response. Although limited research has been done on using the hoop stress in the friction sleeve to infer K0 (Huntsman et al., 1986; Jefferies et al., 1987) this approach is rarely used and has its own significant uncertainties. Clearly there is an issue with neglecting K0, for example by using v rather than p, and estimated from the CPT under normal circumstances. Validation immediately raises the issue of determining ‘ground truth’. This situation has arisen in other aspects of in-situ testing of soil, and it is usual to evaluate the soil properties determined by one test method against those determined by another method in evaluating reliability of the two methods (for example, work at ‘national test sites’ Nash et al., 1992 and Woods et al., 1994). This is the approach followed in this chapter. 155 Four alternative methods are found in the literature for density or state determination: the Self- Bored Pressuremeter test (SBP), flat plate Dilatometer Test (DMT) (Konrad, 1988), geophysically based density measurements (Plewes et al., 1988; Plewes et al., 1994; Hofmann et al., 2000), and shear wave velocity using geophysical methods (Cunning et al., 1995). Of these, the last two estimate void ratio, not  or Dr, and independent measurements of CSL (or emin and emax) and in-situ stress are also needed, making the predictions less accurate for characterising an in-situ deposit. The DMT is a difficult test to evaluate theoretically or numerically, and there is less calibration experience with it in comparison to the CPT – as such, it offers little in the way of a complementary test. In contrast the SBP is readily amenable to theoretical and ticity and included the effects of large tate parameter. numerical analysis. The attractiveness of the SBP in the present context is that it can easily be simulated using very advanced soil models. As such, SBP data provides a reasonably independent evaluation of  estimated from the CPT. Many researchers have used Self-Bored Pressuremeter tests to determine soil density. The conventional approach to evaluation of SBP data follows the methodology developed by Hughes et al. (1977) in which the cylindrical cavity expansion analysis was adopted assuming a non- associated Mohr-Coulomb model with no elasticity. This type of analysis uses dilation angle as the index of material behaviour, a behaviour that is intimately linked to . Carter et al. (1986) provided an improved solution that accounts for elas strains. A different derivation using the same idealisation as Carter et al. was given five years later by Yu and Houlsby (1991). Numerical determination of the state parameter from the Self- Bored Pressuremeter during loading was first undertaken by Yu (1994). His numerical analysis concluded that there is a unique linear correlation between the slope of the pressuremeter expansion curve on logarithmic scale and the initial s 156 Yu et al. (1996) proposed an approach for obtaining the initial state parameter from the ratio of CPT tip resistance and the limit cavity pressure of SBP test. The major limitation of this approach is that SBP tests are rarely taken far enough to reach the limit pressure, rendering the method difficult to apply to most data. An alternative approach to interpreting SBP data is to fit the entire SBP load displacement curve. The mechanical parameters of the soil, including density, can then be determined for the chosen soil model (e.g. Shuttle and Jefferies, 2000; Shuttle, 2006). In this chapter, finite element cavity expansion analysis with a critical state constitutive model has been used to obtain the state parameter from previously published CPT and SBP data obtained as part of a site investigation program on a hydraulically placed, clean quartz sand in Beaufort sea. The work is based on independent calibration of the constitutive model with triaxial tests. The state of the soil is then calculated from CPT data using calibration chamber ded drilling. These islands testing results interpreted through numerical analysis. Soil state is also independently obtained from SBP data by fitting an analytical pressure-strain curve to the measured data. Effects of elasticity, ageing and fabric on the results are considered and reliability of the two test methods in estimating soil state is evaluated.  5.2. Tarsiut P-45 Case History  Exploration for hydrocarbons in the Canadian Arctic in the 70s and 80s was largely based on construction of artificial islands using sandfill. Islands were used rather than drill-ships because of the short open water season, and the presence of moving ice during the winter. Islands had sufficient mass to resist the forces of moving ice and thus allow exten 157 became quite large over time, were constructed of rather uniform hydraulically placed sand, and had extensive engineering and quality assurance. Although two decades or more old, the testing was often at the state of the art and much of the data is of good quality even by current standards. Most importantly, not only are there literature contributions on the work (e.g. Stewart et al., 1983; Jefferies et al., 1985; Hicks and Smith, 1986; Hicks and Smith, 1988; Been et al., 1991; Jefferies et al., 1988; and Hicks and Onisiphorou, 2005) but the original data (as raw digital test data) supporting these contributions are in the public domain (www.golder.com/liq). a drilling barge, and hence named as if it were a ship). igure 5-1a shows a view of this structure while it was operating in moving winter ice, with n through the structure. Because the Molikpaq was designed r use at multiple sites of varying water depth, it had to be founded on a sandfill platform. This These data provide: several CPT/SBP tests in close proximity, together with supporting laboratory tests; in-situ geophysical tests were used to determine elastic properties of the sandfill; and, most importantly, calibration of the CPT in a state-of-the-art calibration chamber using the construction sand. Together, this data set provides a unique opportunity to evaluate  inferred from CPT with  from SBP in a full scale case history – and is the subject of this chapter. The island from which the data used in this chapter was obtained, was not a true island, but was a caisson with an inner core of hydraulically placed sand. This particular caisson was known as the Molikpaq (it was classified as F Figure 5-1b showing a cross sectio fo platform, referred to as a berm, enabled the caisson to be configured for a constant set-down depth across a range of water depths. Foundation failure beneath the berm was avoided by excavating any weak clay at the seabed, using a dredge, to create what was referred to as a subcut. With the subcut excavated, infilled with sand and the berm raised, the Molikpaq was 158 ballasted down onto the berm. Then, the hollow core of the Molikpaq was filled with sand. The subcut infill, the berm, and the core were constructed of the same sand. For the case history used in this chapter, the initial deployment of the Molikpaq was at the Tarsuit P-45 location (see Figure 5-2) in 1984, and the construction sand came from the Erksak borrow pit (location also shown on Figure 5-2). All aspects of construction with sand involved hydraulic filling, using a trailer-hopper dredge, without any mechanical densification/compaction. However, the deposition methods differed between the subcut infill/berm and the core, resulting in very different in-situ densities being obtained by the different construction methods. Details of the construction methods and their achieved densities were given in Jefferies et al (1988) and can be summarised as follows. The infill and the berm were constructed by predominantly bottom dump placement from the dredge. In this method, the dredge is manoeuvred over the target and valves opened in the base of the hull, dropping 6,000 tonnes of sand into place in a few minutes. The core was filled by slurrying the sand in the hopper of the dredge and then pumping it through a floating pipeline across to the Molikpaq and discharging the slurry from a spigot in the centre of the core at sea level, a process referred to as spigot placement. The bottom dump placement produced sand with penetration resistance about double those achieved by spigotting. The focus of this chapter is on the core as it is looser, and hence of more interest from a stability viewpoint.  5.3. Erksak Sand  Erksak sand is a predominantly quartz, sub-rounded sand; Figure 5-3 illustrates the grain shape (fines content removed). The fines content of the sand in the borrow pit is about 10%, but most 159  the  a subsequent Molikpaq deployment that used the same sand (see Been et al. 991) for details). However, this volume of material was nothing like sufficient for the testing and a truckload of Erksak sand was shipped from the field to the sting laboratory. Once thoroughly mixed, this large sample had a 355/3.0 gradation; a slightly  hopper. Figure 5-4 shows histograms of mean grain size and fines content of the sand after it was loaded into the hopper of the dredge, the hopper being sampled using grab samples as well as a vibrocore drill. This variability is to be expected, as the borrow area from which the sand was extracted was some 1 km wide by 2 km long and with the dredge taking a pass along the pit with the draghead pulling sand from the seabed as the dredge slowly traversed the area. Figure 5-5 shows a cross section through the core showing the 1985 investigation boreholes and the median grain size and fines content measured on samples in the core zone. The D50 of the as- placed sand lies between 224 and 480 microns, with an average of 330 microns. Silt contents in- situ are in the range 1 to 4%. Although soil behaviour is sensitive to fines content, the range observed at this site is considered narrow enough to treat the soil as a uniform material.  5.4. Laboratory Tests  Laboratory testing of Erksak sand concentrated on two gradations. Much of the testing to determine mechanical properties, and in particular the reliability of the critical state determination, was on 330/0.76 sand, this material being a composite of the various borehole samples from (1 calibration chamber te different gradation from the composite laboratory sample. Grain size distribution curves for  6 The notation means D50=330 m and 0.7% of the soil passes #200 sieve. 160 these samples of Erksak sand are shown in Figure 5-6. Index properties are summarised in Table 5-1. Both gradations of Erksak sand were tested in triaxial compression; Table 5-2 summarises this testing. Testing of the 330/0.7 gradation was extensive, and is documented in Been et al. (1991). Table 5-3 presents the comparable information for the 355/3.0 tests. Figure 5-7 shows the CSL determined for each gradation of Erksak sand while Figure 5-8 plots peak friction angle versus a rried out in 1984 in the two weeks following placement and levelling f the sand core. Subsequently, this data was supplemented by another 5 CPTs some six months CPT’s were within about 1 to 2 m of each other. As illustrated in Figure 5-5 CPT test MACRES initial state parameter for the dense drained tests.  5.5. In-situ Tests  5.5.1. CPT  A tot l of 33 CPTs were ca o later in 1985 to investigate the effect of “ageing”. The combined data set has been made available to researchers and has been the basis of extensive research into characterizing soil state variability and the consequence of that variability (e.g. Popescu et al, 1998). Although the site was a manmade one with rigorous quality control and a uniform material and deposition method, considerable variability was observed amongst different CPT tests. For example the tip resistance of MACRES 01, 02 and 03 (locations shown in Figure 5-5) were different by a factor of 2 at any given depth. However, based on the assessment of the site data by the original investigators (Golder Associates, 1986) the variability was reduced when the 161 d in determining the characteristic qc associated with the SBP data. . 84 i ation program included three SBP profiles in the core zones; these data were ot available to the authors and have not been used in this paper. In 1985, 16 SBP tests were Figure 5-9 shows MACRES 02 CPT data including tip resistance qc, friction ratio F, pore pressure u2 (measured at the ‘shoulder’ location) indicating a water table elevation at 3.5 m depth, and material index Ic. The depths of the SBP tests are also shown on the tip resistance plot. Horizontal effective stresses inferred from SBP results (discussed later) are also presented on this figure.  5.5.2 SBP  The 19 nvestig n conducted in boring BM1 (see Figure 5-5) seven of which were used to evaluate soil state in the core zone; the rest being obviously disturbed, for example due to a large mismatch between the individual arm displacements (Jefferies and Been, 2006). The tests were carried out in undensified hydraulically placed Erksak sand, which was approximately six months old at the time of testing. A correction was applied to account for the effect of membrane on the horizontal stress measurement. This membrane correction was developed during the fieldwork as a function of the hoop strain, R; the engineers who carried out the testing reported that a pressure correction of dP = 18 + R (%) should be used, giving values in the range 18-30 kPa. The effective horizontal stress measured by the pressuremeter was calculated by subtracting the in-situ pore water pressure from the total pressure measured with the water table at 3.5 m depth. 162 e finite length of the ite the apparent reliability of the measurements, the strain amplitude over which UR was measured varied between SBP tests. For consistency with the CPT interpretation, in the uent analyses the elastic shear modulus from geophysical tests, G, has been used. e trend of the shear modulus G may be approximated by the relation l., 1987b) at pressuremeter (discussed below), is illustrated in Figure 5-10 and shows a consistently increasing trend with depth. GUR indicates values between 10 and 90 MPa with an average of 50 MPa. The range of GUR and its dependence on mean effective stress is generally consistent with the correlation suggested by Jefferies and Been (2006) from laboratory measurements. However, desp G subseq  SBP  5.5.3. Elastic Stiffness from Geophysical Tests  Cross-hole and down-hole seismic velocity profiles were developed as part of the geophysical testing program. The shear wave velocities based on incremental shear wave velocity measured by Vertical Seismic Profile in Hole I 02 (not shown on Figure 5-5) are plotted against depth and mean effective stress (using the average K0 of 0.66 obtained from the SBP tests’ lower range) in Figure 5-10. Th G = 1.9 p 0.8 where G is given in MPa and p  in kPa.  5.6. Calibration Chamber  A feature of this case history is that CPT calibration chamber data are available for the sand used in large scale construction. This calibration testing involved 14 tests (Been et a 163 relatively dense initial state parameters of -0.069 to -0.229 (see Table 5-4 and Figure 5-11). The calibration chamber at Golder Associates’ Calgary laboratory can accommodate a soil sample up to 1.0 m in height with a diameter of 1.4 m (chamber to standard cone diameter ratio of 38). The top and bottom boundaries are rigid and the lateral boundaries impose constant stress on the soil sample. CPT chamber samples were prepared by moist tamping, which is similar to that used for triaxial sample preparation for all Erksak 355/3.0 and some of Erksak 330.0.7 samples but with some modifications because of the very large sample size. Successive layers were added and tamped until the desired sample height was obtained. The sample was weighed, sample dimensions recorded, and overall average void ratio calculated. A complete description f the chamber testing program and the material is presented in Been et al. (1987b). .7. Modelling Erksak Sand Behaviour . soil behaviour. The version used is that for general 3-D stress states with constant principal o  5  The approach to evaluating the state parameter from SBP data involves formal modelling of the pressure-hoop strain relation. Similarly, for the CPT, the relation between qc and  in the calibration chamber has been established by modelling. As the same constitutive model is used in both situations, this model is overviewed and then calibrated for Erksak sand before discussing its application to each of the in-situ tests.  5.7.1 NorSand  The constitutive model adopted is NorSand (Jefferies, 1993), an isotropically hardening - isotropically softening generalised critical state model that captures a wide range of particulate 164 stress direction as described in Jefferies and Shuttle (2002) but with a further extension to improve accuracy as described below. NorSand can be regarded as a super-set of the well- known Cam Clay model (Schofield and Wroth, 1968), with Cam Clay being obtained as a special case of NorSand by appropriate choice of the material parameters and initial conditions. A feature of the original version of NorSand was a volumetric coupling parameter N for stress- dilatancy. Subsequently, Jefferies and Shuttle (2002) suggested that N could be eliminated from the model since N1 (based on average values of a large quantity of triaxial tests on different sands). However, individual soils demonstrate a variety of Nvalues and some accuracy is given up by following this suggestion. The N model parameter neither increases the complexity of the model, nor constitutes additional effort in calibration of the model, as N is obtained with tc from the stress-dilatancy plot (see Figure 5-12) and reintroduction of this parameter resulted r the current work a minor modification of the riginal N, termed N* for clarity, was obtained as the slope of the post peak stress-dilatancy  from plane strain (Jefferies and Shuttle, atically varied triaxial paths from compression through to extension (Jefferies  All NorSand parameters apart from elasticity are invariant with soil r stress level. The differing soil behaviours simulated are entirely controlled by the M in better replication of the soil behaviour. Fo o plots. A central feature of NorSand is the use of the state parameter  as an internal rate variable, which allows the model to simulate soil behaviour ranging from liquefaction of loose soil when loaded undrained through to extreme dilation of very dense samples. This particular version of NorSand has been validated for stress paths ranging 2002) to system and Shuttle, 2005a,b). density o state measures. 165 As a critical state model, NorSand is based on the premise that soil’s state tends to critical conditions as shear strain accumulates. Two modelling choices are involved to determine the level of accuracy desired in the representation of the CSL; and, to capture the influence of intermediate principal stress on M. For many soils, and much engineering, a semi-logarithmic approximation of the critical state locus is sufficient and that is used here. Regarding M, it has been known since Bishop (1966) that constant M was inconsistent with the no-tension behaviour of particulate materials. The analysis here adopts the empirical rule for M proposed by Jefferies and Shuttle (2002), which takes tc as the basic soil property ( is called the Lode angle, and describes the proportion of the 3-D shear stress invariant attributable to the intermediate principal stress). NorSand has four basic aspects: a yield surface; a work hardening law; a plastic flow rule (stress-dilatancy); and, elasticity. NorSand is a sparse model, with the fixed principal stress direction variant having eight material parameters, seven of which are dimensionless. Many soil properties in NorSand will be familiar from Cam Clay. Three properties describe the critical ealistic, and inconvenient for use with numerical methods. Within NorSand the physically real “finite” elastic shear modulus G is adopted. Various forms of elasticity can be state, these being the critical friction ratio tc and  which describe the CSL for the semi-log idealisation used. There are two additional properties relating to the plastic behaviour: tc and H. The property tc scales the maximum dilatancy to . In some ways this is similar to specifying a relation between peak strength and . The dimensionless hardening modulus H is required because the state parameter approach de-couples the yield surface from the CSL and hence the slope of the CSL no longer acts as a plastic compliance. The rigid-plastic idealisation in shear used by Cam Clay theoretically requires that G is infinite. This is both unr 166 used within the same framework. For the purpose of the current research although the initial elastic shear modulus is assumed to depend on the void ratio and stress level, both G and  are taken as constant during the test.  5.7.2. Calibration to Erksak 355/3.0 and Erksak 330/0.7 Sand  The aim in calibrating NorSand to a particular soil is not to fit one test as elegantly as possible. Rather, the objective is to obtain a consistent set of material parameters that are able to represent the behaviour of the sand over all available tests and covering the range of pressures and states. Hence the goodness of fit for any individual test is, by necessity, compromised to obtain a better representation of the overall soil response. This calibration process is summarised below; the reader is directed to Jefferies and Shuttle (2005a) for a more detailed discussion of the calibration procedures. The critical state of the soil is not really a NorSand calibration, as any critical state model will have the same parameter set. For the semi-log idealisation of the CSL, a best-fit line is put through the critical states judged from high strain results on loose samples. A variety of methods are available to estimate tc (Ghafghazi and Shuttle, 2006). As illustrated in Figure 5-12, N* and tc are respectively obtained as the slope and intercept of the post peak portion of the stress- dilatancy plot for each triaxial test. Average values were then chosen to represent the overall soil behaviour. The plasticity parameter, tc , was derived by plotting dilation at peak against the state parameter at the image condition (see Figure 1-2) i at peak, with tc being the slope of the trend in Figure 5-13. 167 The last plasticity parameter, the hardening parameter H, is fabric dependent and can only be determined by iteratively fitting the NorSand model to triaxial data. In this iterative fitting, the parameters established as just described were kept constant, and H was determined by modelling e drained triaxial tests and selecting the parameter set that gave the best overall visual fit osen from each sand and ample reconstitution method. Pa. This range matches the values measured in the field (10-90 MPa ( n method has no effect on the critical state parameters, it does affect the plastic th across all the tests. Figures 5-14 and 5-15 show the fits for two tests ch s Elastic shear modulus G is also expected to be affected by fabric (and age of deposit). Jefferies and Been (2006) suggest a correlation between G, void ratio and stress level of the sample based on the unloading portion of a series of isotropic compression tests on Erksak 330/0.7 samples. No distinguishable difference was observed between moist tamped and wet pluviated samples, so the same relation of G against depth is applied to both Erksak sample types (Table 5-5). For the range of void ratio and mean effective stress at this site the correlation yields G values between 40 and 100 M from SBP and 40-120 MPa from geophysical measurements), and suggests that ageing had not significantly affected the stiffness of Erksak sand at the time of testing. Table 5-5 summarises the calibration of NorSand for Erksak 355/3.0 moist tamped samples. The calibration of NorSand to Erksak 330/0.7 sand was documented in Jefferies and Shuttle 2005a). A difference between the two calibrations is that two sample reconstitution methods, moist tamping and water pluviation, were used in the case of 330/0.7 sand. Although the reconstitutio hardening modulus, with the moist tamped sand being stiffer than water pluviated sand, thus yielding a higher H value for identical initial state conditions as reflected in Table 5-5.  168 5.8. Evaluation of In-situ CPT Data  5.8.1. Methodology  The original framework for the relation between penetration resistance and state was developed empirically (Been et al., 1986, 1987a). Triaxial testing was used to determine CSLs of the various sands for which chamber test data existed, allowing the calibration data base to be transformed from void ratio measurements to state parameter. The effect of stress level on the CPT was removed by transforming the CPT data to dimensionless form using:  p pqQ c            [Eq. 5 - 1]  where qc is the tip resistance for drained penetration (corrected for chamber size) and p and p   are the mean total and effective stresses respectively. Plotting chamber test data in this normalised form supported a simple dimensionless relation for any sand:   0exp mkQ           [Eq. 5 - 2]  where the two coefficients k and m in Equation 5-2 differ from one sand to another. In the original work, it was suggested that both k and m were functions of the slope of the CSL, , but that no other soil properties were identified as being involved in the relation between Q and 0. 169 Equa ion 5-2 is readily inverted to give t 0 provided that k and m are known from calibration tudies (or otherwise): s    m kQln 0           [Eq. 5 - 3]  of the credible nge in the parameter sought. Sladen’s identification of stress bias in a dimensionless formulation resulted in other workers turning to numerical analysis to resolve the conundrum. Collins et al. (1992) reported a drained avity expansion analysis using a state parameter based model. Their results showed that there o sand and found that e computed effect of stress level was half the experimental scatter in the calibration chamber data, but clearly plemented NorSand in a finite element spherical cavity expansion analysis and demonstrated constant Ir, Equation 5-2 fits all simulations regardless of the confining stress. The issue arises though, that Ir is not constant in real sands. Doubling the stress level produces less than a  Sladen (1989) tested Equation 5-3 for bias using the extensive Ticino sand calibration data and found a systematic bias in Equation 5-3 with stress level despite the dimensionless formulation. Potential errors as much as  0 = ±0.2 were suggested, in the order of 50% ra c was an effect of stress level on the state parameter approach, consistent with the suggestion of Sladen, and further that the relation between Q and 0 depended on material properties of the sand. Jefferies and Been (1995) compared Collins et al. results for Ticin th there was an unknown bias. Subsequently, Shuttle and Jefferies (1998) im that Equation 5-2 was an accurate approximation of the numerical results for fixed soil properties. However, both k and m were found to be strong functions of G/p0 (termed Ir), and it is this aspect that led to an apparent stress level bias within a dimensionless formulation. For 170 doubling in the shear modulus, making Ir reduce as the stress level increases. It is this reduction in Ir that leads to apparent stress level bias because m and k also reduce with reducing Ir. he result of exploring the stress level ‘bias’ identified by Sladen is that it is not a real bias at all. Rather, an im evaluation of  from CPT data requires detailed knowledge of the shear modulus profile Ir has to be determined for the soil being tested. r entally and numerical simulations are essential. Parametric 1 r 2 3 4 5 6 7 r 8 9 10 11 12  1 12   T portant parameter group, Ir, had been omitted from the framework. Accurate through the deposit, and, the trend of k and m with Understanding the relation between k and m and I  illustrates the limitations of physical testing – there is no independent control over shear modulus in the calibration chamber, so the relation cannot be determined experim simulations provide data to develop influence functions for each of the material parameters, leading to the approximation (Shuttle and Jefferies, 1998):  k = (f ( I ) f (M) f (N) f (H) f () f ())Z      [Eq. 5 - 4a] m = 1.45 f ( I ) f (M) f (N) f (H) f () f ()      [Eq. 5 - 4b] where the functions f  - f  are simple algebraic expressions and Z is a scaling factor accounting for the inherent mismatch between cone penetration and spherical cavity expansion processes. 5.8.2. State Parameter from CPT Tests  The spherical cavity finite element code used by Shuttle and Jefferies (1998) was used in this study. The analysis code and finite element mesh remained the same, thus retaining the verified 171 large displacement performance of the code. In the current work for the calibration chamber tests, all moist tamped Erksak 355/3.0 soil parameters are either constant or known functions of been well understood that there is an inherent mismatch between spherical cavity          [Eq. 5 - 5] e ample preparation method does affect stress-strain behaviour. This creates a dilemma for onstituting samples, how can real construction rocesses be simulated in the laboratory, and how can the soil fabric developed in one situation  0 , p or e (see the fourth column in Table 5-5). The shear modulus is obtained from the correlation suggested by Jefferies and Been (2006), which corresponds to that of a newly prepared sample. Each chamber test is modelled as having the initial state, horizontal and vertical stresses and the Q obtained is plotted against that measured in the test in Figure 5-16. It has expansion analysis results and those of calibration chamber tests. This difference stems from the assumptions made in the analogy of cone penetration and spherical cavity expansion, and is usually accounted for by a scaling factor. The scaling function7 obtained for this NorSand representation of Erksak sand, and as shown in Figure 5-16, is:    7.147.0 sphcc QQ  5.8.3. Inversion Parameters  The analysis developed is for Erksak 355/3.0 sand placed by moist tamping. However, this was not the in-situ placement method for the core, and laboratory observations clearly indicat s engineering analysis. If forced to rely on rec p  different scaling function is applied. The two functions are numerically close as the scaling from Ticino sand is shown to be applicable to all the database sands as discussed later in chapter 4. 7 The analysis presented in this chapter was performed and published before those presented in chapter 4. Hence a 172 be checked against another?  Perhaps a closer analogy to the full scale hydraulic fill placement is water pluviation, and therefore a new parameter set is generated by adopting the plastic ardening parameter H from tests on wet pluviated samples of Erksak 330/0.7 while keeping all the other parameters from the Erksak 355/3.0 calibration. A summary of this parameter set is presented in the fifth column of Table 5-5. The elastic shear modulus for the in-situ CPT is obtained from down-hole seismic shear wa h ve caling factor obtained from calibration chamber data. Combining Equations 5-2 and 5-5 we can write       [Eq. 5 - 6a]          [Eq. 5 - 6b] parameters and results are summarised in Table 5-6. velocity measurements. This type of data is now frequently obtained with cone penetration and represents the in-situ soil elasticity. With every parameter known as a constant or a function of 0, the CPT inversion parameters k and m versus Ir trends are developed for the in-situ soil conditions using the spherical cavity expansion analysis (kspherical , mspherical). The kspherical and mspherical values are then scaled using the s  k  7.147.0 sphericalk  m sphericalm 7.1  and obtain k and m vs Ir, the results being shown in Figure 5-17. The in-situ normalised cone tip resistance is calculated for the centre point of each SBP test from the procedure shown in Figure 5-18. Knowing the mean effective stress and G , we can calculate k and m for each depth. 0 is then conveniently calculated from Equation 5-3. All the 173 The state parameter at the location of each SBP was estimated from the CPT data allowing for variable penetration resistance along the length of the pressuremeter, and variable K0 as described below. The CPT penetration resistance is plotted in Figure 5-9, and the depths of the SBP tests are indicated. The best estimate of the qc value at the CPT is taken as that at the mid-point of the  e variation in qc. he best estimate values of Ir shown in Table 5-6 are calculated using the best estimates of seismic shear mod in K0 is presented nal variation in the normalised tip resistance Q (presented 0.  SBP, with the upper and lower bounds for qc being taken as the least and greatest value, respectively, over the length of the SBP. These values are tabulated as error margins in Table 5-6. A main source of uncertainty in both CPT and SBP interpretation is the ambiguity in horizontal stress. In this project we have the luxury of good quality pressuremeter data to interpret a range of h and hence p  for each test depth (as explained in the next section). The variation in p' then results in a variation in G, Ir and Q leading to a range of 0 predicted when superimposed with th T ulus, G, with the best estimates of p'. The variation caused by the uncertainty  as error margins for G. Variations in penetration resistance qc along the length of the pressuremeter cause an additio in Table 5-6) again resulting in uncertainty in the state parameter     174 5.9. Evaluation of SBP Data  5.9.1. Methodology ed until an acceptable fit is obtained in a procedure similar to that of Jefferies (1988) and eter length. The original cavity radius was set equal to unity; the  be thought because there ppears to be a very simple relation between the cavity expansion pressure under the cylindrical ssumption and that under a finite pressuremeter geometry. This aspect was explored xperimentally by Ajalloeian and Yu (1998), who carried out experiments in a calibration hamber using several pressuremeter geometries and several sand densities. If the cavity xpansion pressure under cylindrical symmetry is denoted as  and the cavity expansion  The SBP tests were analysed using Iterative Forward Modelling (IFM). IFM of a SBP involves estimating the parameter set for the test, computing the pressure-hoop strain curve, and comparing that curve with the measured one. The parameter set is varied and the procedure repeat subsequently adopted by Cunha (1994) and Shuttle and Jefferies (1995). IFM has the advantage of using all the available information while preventing inconsistent parameter combinations which may occur if the parameter selection is uncoupled. The IFM work was based on the expanding cylindrical cavity idealisation of the pressuremeter, corrected for finite pressurem outer boundary was set as a zero displacement node at a distance of 800, and the logarithmic element spacing was used. The formulation assumes that the pressuremeter is infinitely long and so the soil around it is in a state of both plane strain and axial symmetry. A complete description of the formulation is presented in Shuttle (2006). The idealisation of cylindrical symmetry is not as restrictive as might a a e c Pe 175 pressure for a finite SBP with a length (L) to diameter (D) ratio of L/D = 6 (typical of many ommercial devices) as P6, Shuttle (2006) used the experimental data to show that these two c variables are related by   5.06 1.01 RPP            [Eq. 5 - 7]  where R is the cavity hoop strain in percent at the pressuremeter-soil interface. Other SBP geometries have similar relations using different coefficients. Hence, it is convenient to rely on this result and use the simplification of cylindrical cavity expansion (Shuttle, 2006).  5.9.2. State Parameter from SBP Tests  In analysing the SBP data, sand parameters are known from the calibration summarised in Table 5-5, column 4 and 5, and with elastic shear modulus obtained from geophysical tests. Only two parameters remain to be determined by IFM: the initial horizontal stress h   and the state parameter 0 . Despite the data being of good quality, disturbance is observed at the start of ma aking determination of ny of the tests, m h   from the lift-off pressure un  Therefore reliable. both h   and 0  have been determined using IFM, with the fits constrained to pass through the higher strain SBP loading data where the effect of disturbance is lower. In addition, an upper limit of unity on K0 was assumed as, although the sand was pumped at approximately 5000 t/hour via a 36 inch diameter central spigot using 8 MW and 15 MW pumps and higher values of K0 could be anticipated, the sand was not overconsolidated. This methodology is  summarised in Figure 5-19 and resulted in a non-unique, although constrained, predicted range 176 for h   and 0 . The vertical effective stresses and K0 values estimated for each SBP test are summarised in Table 5-6 and the credible range of “best fits” are plotted in Figure 5-20. The quality of the fits to SBP curves is not equal for all tests. For some tests, including SBP 14,  w nly ible to fit the last part of the curve by the model and K0 and it as o poss 0 were not very wel nstr d. re meter was expanded to less than 3% strain and the inferred l co ssure aine Conversely, visually an excellent fit is achieved for SBP 10, but this p 0  is likely unrealistically l o situ state parameter of a sand deposit using two iff t in  methods so as to obtain an indication of the reliability with which state observed in calibration chamber tests performed under controlled conditions in the laboratory. dense.  5.10. Discussion  The go f this study was to evaluate the in-a erend -situ parameter can be determined. Two in-situ methods were used, the CPT and the SBP, each having differing constraints, attributes, and uncertainties. Data from the CPT and SBP were evaluated independently, and plotted in Figure 5-21 for comparison. Six of the seven evaluations (the exception being test 10 which is discussed below) indicate good correspondence between estimates for 0 estimating the state parameter with a difference of 0 to 0.05. This is considered a very good level of consistency when compared to the common scatter In comparison to CPT, SBP tends to estimate more negative values of 0 by approximately 0.02 indicating a denser deposit. Although small, this estimate is non-conservative relative to the CPT interpretation; so the results should be cautiously applied to engineering problems. 177  test is be treated with extreme caution when it comes to ining the state parameter. The irony is that test 10 predicts too dense a state while in other ter than the numerical model results. Hence if we assume that test 10 would have followed the same trend as others, it should have become stiffer towards realisable from full numerical simulation of a real CPT geometry, and in which results are also computed for the friction sleeve. Evaluation of SBP data in sand is at an early stage compared to the wealth of experience with the CPT. For example, the correction applied for limited pressuremeter length is purely done on the densest material in the borehole, the SBP estimate seems unrealistic especially as no densification effort has been undertaken at the site. Looking at Table 5-6, one would realise that this test was continued to the lowest strain (2.8%) amongst all the SBP tests done at the site. Other tests are taken to at least 5.5% and up to 10.5%. Thus, it could be speculated that any SBP test taken to strains less than about 5% should determ tests the low strain data is generally sof the end resulting in even denser 0 predictions. A striking feature of analysing these two different types of tests is the relative ease. The CPT is very easy to do in the field, but requires considerable care in estimating the state parameter. Conversely, with SBP it is challenging to get good data in sand but rather straightforward to analyse with the IFM approach. The CPT interpretation would be aided by routine use of the seismic cone to provide the relevant shear modulus data. However, the biggest step for a range of soil types may be to use the data from the friction sleeve to indicate the appropriate soil compressibility behaviour. Plewes et al. (1992) initiated this approach based on field data, but enormous insights should be 178 experimental and constant for different soil densities. Since the mismatch between CPT and BP states increases with density (more negative 0), and geometric corrections such as the one for limited boundaries of the CPT calibration chamber increase with density, it may be that the present finite geometry correction is too approximate. But this will diminish in importance in time when the IFM method runs fast enough to use real pressuremeter geometry in the finite element mesh of the numerical model. The difference between 0 interpreted by the two methods can also be magnified by an error in evaluating K0 from SBP data, which is sensitive to disturbance. Obtaining a unique fit becomes more difficult because increased initial stress has a similar effect to decreased initial state. Further work is warranted on exploring the effect of disturbance, and in particular whether the contraction stage of the pressuremeter test migh S t usefully constrain the estimate of h0 as is  results appeared more consistent, and the state parameter readily done in the case of clays (Shuttle and Jefferies, 1996).  5.11. Conclusion  State of the art methods were used to interpret the state parameter from in-situ SBP and CPT in a hydraulically placed clean quartz sand in Beaufort Sea. Calibration chamber test results have been modified for fabric effects to replicate in-situ soil conditions and used as reference. SBP and CPT interpretation methods were used to estimate the in-situ state parameter with an acceptable consistency as compared to common scatter observed in laboratory testing performed under controlled conditions. CPT values were more reasonable with respect to site conditions. However, since CPT calibration chamber testing has been used as a reference, and no independent measurement of the in-situ 179 One should keep in mind that although the CPT appears to produce a more reasonable and consistent state parameter interpretation for the site investigated, performing an adjacent pressuremeter test is necessary to obtain the horizontal stress required for the CPT interpretation formulation. In other words, the accuracy of CPT inferred state parameter relies on SBP measurement of horizontal stress levels. SBP results can then be used to verify state parameters obtained from CPT testing through the nalysis method employed in this research. However, any SBP test taken to strains of less than e treated with extreme caution in determining the state parameter. or the site investigated in this program, which was six months old at the time of testing, ageing a about 5 % should b F did not appear to have a significant effect and could be ignored in the interpretation. Conversely, fabric effects were deemed important and were accounted for in making use of laboratory measurements for site interpretation.   180 Table 5 - 1 Index properties of Erksak 330/0.7 and Erksak 355/3.0 sands      Property Been et al. (1991)             Sand/Source Erksak 330/0.7 Erksak 355/3.0 Been et al. (1987b ) Mineralogy Quartz, minor amounts of Quartz 73%, Feldspar 22%, Other 5% chert Median grain size D50: m 330 355 Effective grain size D10: m 190 180 Uniformity coefficient 1.8 2.2 Percentage passing no. 200 sieve 0.7 3-6 Specific gravity of particle 2.66 2.65 Average sphericity - 0.75 Grain description Subrounded Subrounded Maximum void ratio e  0.753 0.963 max Minimum void ratio emin 0.527 0.525               Sand/Source Erksak 330/0.7 Erksak 355/3.0  Table 5 - 2 Summary of triaxial tests on Erksak 330/0.7 and Erksak 355/3.0 sands       Been et al. (1991) Been et al. (1987b ) Number of undrained tests 40 5 Number of drained tests 16 5 Range of void ratio tested in drained triaxial apparatus 0.53-0.82 0.54-0.63 Range of mean effective stress apparatus applied in drained triaxial 60-1000 (kPa) 50-400 (kPa)   Table 5 - 3 Drained triaxial tests on moist tamped Erksak 355/3.0 sand   Initial conditions Test Conditions1 End of test Test Void p'    0 Drainage Stress Steady '3   p'   q   Void c .ratio kPa Path state 2 kPa kPa kPa Ratio 3 deg CID L1 0.628 100 -0.119 D C Mdil 100 185 255 0.689 33.4 CID L2 0.615 400 -0.105 D C Mdil 400 680 840 0.659 29.1 CID L3 0.709 195 -0.025 D C Con 195 300 313 0.705 25.4 CID D1 0.551 100 -0.196 D C Yes 100 190 268 0.658 32.8 CID C166 0.536 50 -0.224 D C Dil 50 86 109 0.590 32.3 1 U, Undrained conditions; D, drained conditions; L, load-controlled compression; C, triaxial compression. 2 Yes = critical state apparently reached; Dil/Con = sample still dilating or contracting at end of test; Mdil means small amount of dilation with sample close to critical state at end of test 3 Void ratios corrected for membrane penetration  181 182 987b) TEST 'h k 0 q Q Table 5 - 4 Summary of CPT calibration chamber tests (after Been et al., 1 e 'v o p' c  Boundary     kPa kPa CC-3A 100  1.00  100.0  -0.117 kPa   M 6.7  1.00  63 Pa  Correction 0.63  100 CC 100  1.00  -0.167  14.1  05  144 CC 89  0.70  -0.216  26.2  .15  290 CC 4  0.70  -0.069  12.4  .00  49 C 4  0.70  -0.109  18.6  00  74 4  0.69  -0.149  30.4  .00  122 CC 214  0.70  -0.169  31.5  05  133 CC-08 266  0.71  3 0  -0.126  29.7  00  97 CC-09 44  0.70 -0.200  10.5  229 CC- 131  0.70    -0.169  27.8    193 CC- 126  0.70    -0.210  31.2    240 C-12 0.60  180  126  0.70  144.0  -0.140  12.9  1.00  88 C-18 0.64  30  22  0.73  25.0  -0.133  1.9  1.00  65 C-19 0.53  63  45  0.71  51.0  -0.229  11.5  1.20  260 -3B 0.58  100  100.0  1. -05 0.53  127 -6A 0.66  306  21 102.0 245.0 1 1 C-6 21 CC-6C 0.58  309  21 B 0.62  307  245.0  1. 246.0  1 -07 0.56  307  245.0  1.  0.60  374  02. 1. 0.56  63    50.0  1.11 10 0.57  188  150.0 1.05 11 0.53  180  144.0 1.12 C C C  Table 5 - 5 NorSand calibration to Erk meter Erksak Erksak 0/0.7 oist ped) 355/3.0 (moist tam itu Soil: Erksak 355/3.0 (Hydraulic- ally placed) Rem sak sand Para Erksak In-s  ped) ark 330/0.7 33 (wet (m pluviated) tam Critical State  0.828 0.834 0.834 Void ratio of CSL at 1 kPa 0.828 e 0.0160 0.0160 0.019 0.019 Slope of CSL, defined on base e Mtc 1.27 1.27 1.25 1.25 Critical state friction ratio Plasticity Plastic hardening modulus for 043070   0400100   0400100   043070   H loading tc 5.40 3 imum dilatancy to again dete r xia re  c itio     Relates min  5.40 .80 3.80 rmined unde tria l comp ssion ond ns N 0. .40 0.32 lum co gfor stress-dilatancy rule 40 0 0.32 Vo etric uplin  parameter Elasticity G (kPa)    '' refpp  50.0 minee  195   A . fter eq 3.37 of Jefferies and Been, 2006     50.0'' refpp mine 195 e ft  A er eq. 3.37 of Jefferies and Been, 2006     50.0min'' pp 195 ref e   Af .   in-si mic ve  is th tiv ress a ref is ere tr ua 00 kP i v tio  e ter eq 3.37 of velocity min Jefferies and Been, (Figure 5- 10) which the volumetric compressibility becomes zero. 2006 8.01 0 p 90  p' stfrom tu p'seis shear wa e initial mean  in kP  effec e  a ref nce s ess eq l to 1 a 35.0 5e  s the oid ra  at   0.20 0.20 0.20 0.20 Poisson’s ratio, assumed value (measured values typically found to be in range 0.10- 0.25) Table 5 - 6 Summary of data used in estimation of in-situ state from CPT and SBP T e s t  Depth (m)  note 1  tests  v (kPa) note 2 K0 (---)   (kPa) c (MP ( G (MP in SBP test BP p q a) Q ---) a) k (---) m (---)  from   CPT M str axi ain mu  ( m %) 0 m Sfro  1 2.7 85 ± 0.15 45.0 ± 5.0 2.6 ± 0.3  7.3 .9 19.59 11.18 -0.11 ± 0.01 8.4 -0 0.02 9 10.7 77 ± 0.23 114.5 ± 20.7 9.2 ± 0.7 82.6 ± 21.2 1 19 0.04 10 11.5 143 0.85 ± 0.15 128.8 ± 14.3 14 5.0 114.0 ± 12.8 92.5 ± 8.2 19.45 -0 -0 0.04 11 16. 80 ± 0.20 163. 25 2.3 ± 5.0  43.0 9.42 10.82 -0.11 ± 0.06 10.5 -0. 0.03 13 18 80 ± 0.20 180. 27 8.0 ± 2.0  18.2 9.41 0 8 0. 0.03 14 20. 77 ± 0.24 190. 35 12.7 ± 2.0  23.5 9 0.04 15 21.0 96 ± 0.04 229.8 ± 6.3 12.4 ± 3.0 52.6 ± 14.5 7. 19.38 10.72 -0.09 ± 0.03 8.0 -0 0.00 50 0. 135 0. 64.3 ± 39 84.  ± 3.5 ± 12.2 .06 ± 7 ± .26 ± 12 ± 11 ± 13 ± .11 ±   8 1 1 1 .47 10. 10 92 .88 -0.13 .1 ± 0. 6 ± 0. 02 01 5. 2. 5 8 -0.1 .7 ± 2 .2 0 189 0. 209 0. 226 0. 236 0. 9 ± 8 ± 8 ± .2 .8 .5 1   79 45 69 9 ± .2 ± .1 ± 112.1 ± 13. 121.3 ± 15.0 126.5 ± 18.9 14 2 ± 3.2 10. 10. 79 78 - .07 11 ± 0. ± 0. 04 04 .0 1 - .40 -0. 9. -0. Notes:   (1) Depths are quoted to the strain arm measurement axis of the SBP. (2) Water Table estimated at 3.5 m from surface from adjacent CPTu.  183  a)    b) Figu P-4     re 5 - (Jeff  1 a) kpaq; b) Schematic cross section of Molikpaq at Tarsiut 5 eries an Aeria d B l p e ho en to , 2  o 00 f M 6) oli  Figu  r  Tarsi  proximity of Erksak borrow (after Jefferies et al., 1985) e 5 - 2 ut location showing  184  Figure 5 - 3 Photograph of washed Erksak 355 sand particles (Been et al., 1987b)    0 D 50 ( m ) 200 240 280 320 360 50 100 150 200 250 300 350 400 Fr eq ue nc y  0 0.0 1.0 2.0 3.0 4.0 50 100 150 200 250 300 350 400 450 500 Fines (%) Fr eq ue nc y  igure 5 - 4 Particle size distribution of Erksak sand in hopper of dredge prior to placement fter Goldby et al., 1986) F (a  185  Figure 5 - 5 Section through the core showing the 1985 in-situ testing program and the result gradation tests (Golder Ass s of ociates Ltd., 1986)  186 020 40 60 80 100 0.010.1110 Diameter (mm) Pe rc en t p as si ng Calibration chamber sample Erksak 355/3.0 Triaxial sample Erksak 330/0.7 Triaxial sample  Figure 5 - 6 Grain size distribution of different Erksak sand gradations: Erksak 355/3.0 and 330/0.7 (after Been et al., 1987b, 1991)      187 0.55 0.60 0.65 0.70 0.75 0.80 0.85 1 10 100 1000 10000 p' Vo id  ra tio , e CSL Erksak 330/0.7 CSL Erksak 355/3.0   = 0.834  e  = 0.019   = 0.828  e  = 0.016  Figure 5 - 7 CSL locations for Erksak 330/0.7 and Erksak 355/3.0  20 25 30 35 40 45 -0.25 -0.20 -0.15 -0.10 -0.05 0.00 0 '  p ea k Erksak 330/0.7 Erksak 355/3.0  Figure 5 - 8 Peak drained triaxial compression strength of dense Erksak sand samples  188  189 0 5 10 15 20 25 1 2 3 I c 30 2.7 21.0 20.0 18.2 16.2 11.5 10.7 0 5 10 15 20 25 0 5 10 15 20 25 30 Tip resistance, q c (MPa ) D ep th  (m ) 30 C O R E 0 5 10 15 20 25 -0.1 0.1 0.3 0.5 Pore pressure, u 2  (MPa ) 0 5 10 15 20 25 0 1 2 Friction ratio, F r (%) 227 173 167 151 122 104 42 0 5 10 15 20 25 30 0 50 100 150 200 250 'h  (kPa) t si lty  s an d to  s an dy  s il 3030  a)   b)     c)        d)               e) Figure 5 - 9 a) Cone tip resistance and SBP depths; b) Friction ratio; c) Pore water pressure u2; d) material index Ic after Been and Jefferies (1992); e) Horizontal effective stress from SBP t  BE R M sa nd  to  s ilt y sa nd sa nd y si lt to  c la ye y si l gr av el ly sa nd  to  s an d 05 10 15 20 25 0 50 100 150 200 Elastic shear modulus, G  (MPa ) de pt h be lo w  s ur fa ce  (m ) 0 50 100 150 200 M ea n ef fe ct iv e st re ss   p ' ( kP a ) SBP unload-reload Geophysical testing Loose Core Dense Berm G  = 1.9 p '0.8  Figure 5 - 10 Shear modulus measurements from SBP unload-reload loops and geophysical ng ata fr  Gol  Associates Ltd., 1986) testi (d om der  0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 CSL Erksak 355/3.0 1 10 100 1000 10000 p' 0 Vo id  ra tio , e Calibration chamber samples  Figure 5 - 11 Initial state plot for CPT calibration tests in Erksak 355/3.0 sand  190 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5 -0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 D min, max 1-N * 1 M tc =1.27 St re ss  ra tio   = q/ p ' Image condition D = 0 Dilatancy D = d  v / d  q igure 5 - 12 Stress-dilatancy plot ( - D) for Erksak 330/0.7 sand (test CID_G666)  F  191 -1.2 -1.0 -0.8 -0.6 -0.4 -0.2 0.0 -0.25 -0.20 -0.15 -0.10 -0.05 0.00  i,peak D m in Erksak 355 / 3.0 Erksak 330 / 0.7  = 3.80   = 5.40  Figure 5 - 13 Plot of dilatancy at pe k, Dmin, versus the state parameter at the image condition t peak to determine the value of tc a i a -9 -8 -7 -6 -5 -4 -3 -2 -1 0 1 vo lu m et ric  st ra in  (% ) 0 50 100 150 200 250 300 350 400 0 5 10 15 20 25 axial strain (%) de vi at or  st re ss , q  (k Pa ) Norsand Test Norsand Test ψ 0 = -0.196 p '0 = 100 kPa -5 -4 -3 -2 -1 0 1 vo lu m et ric  st ra in  (% ) 0 50 100 150 200 250 300 350 0 5 10 15 20 25 axial strain (%) de vi at or  st re ss , q  (k Pa ) Norsand Test Norsand Test ψ 0 = -0.119 p '0 = 100 kPa  a) Test CID D1  b) Test CID L1 Figure 5 - 14 Fitting NorSand to triaxial tests on Erksak 355/3.0, Moist tamped   192 -6 -5 -4 -3 -2 -1 0 1 vo lu m et ric  st ra in  (% ) 0 100 200 300 400 500 600 700 800 900 1000 0 5 10 15 20 25 axial strain (%) de vi at or  st re ss , q  (k Pa ) Norsand Test Norsand Test ψ 0 = -0.106 p '0 = 300 kPa  a) Test CID G664, Wet pluviated -8 -7 -6 -5 -4 -3 -2 -1 0 1 vo lu m et ric  st ra in  (% ) 0 50 100 150 200 250 300 350 400 450 500 0 5 10 15 20 25 axial strain (%) de vi at or  st re ss , q  (k Pa ) Norsand Test Norsand Test ψ 0 = -0.153 p '0 = 130 kPa  b) Test CID G667, Wet pluviated 0 1 2 3 4 vo lu m et ric  st ra in  (% ) 0 200 400 600 800 0 1 2 3 4 5 vo lu m et ric  st ra in  (% ) 0 50 100 150 200 250 300 350 1000 kP a) 1200 0 5 10 15 20 25 axial strain (%) de vi at or  st re ss , q  ( Norsand 400) 450 0 5 10 15 20 25 axial strain (%) de vi at or  st re ss , q  (k Pa Test Test Norsand Test ψ 0 = 0.036 p '0 = 500 kPa  c) Test CID G682, Moist tamped Norsand Norsand Test ψ 0 = 0.065 p '0 = 195 kPa  d) Test CID G685, Moist tamped Figure 5 - 15 Fitting NorSand to triaxial tests on Erksak 330/0.7  193 Q cc = 0.47Q sph 1.7 10 100 1000 10 100 Q sph  from spherical cavity expansion analysis C PT  Q cc  in  c al ib . c ha m be r  Figure 5 - 16 CPT in calibration chamber versus spherical cavity expansion: Erksak sand   m  = 1.535 + 1.421 ln(G /p '0) k  = 15.225 + 0.643 ln(G /p '0) 5 7 9 11 13 15 17 19 21 23 25 100 1000 10000 G/p 0 ' C oe ffi ci en ts  in  E q.  [5 -2 ] : m ,k  n coefficients in hydraulically placed Erksak 355/3.0 sand Figure 5 - 17 Computed effect of soil rigidity Ir on CPT calibratio  194  Estimate cone tip resistance qc  from the CPT log Calculate         p pqQ c   Estimate             v                  for the relevant depth Obtain          h                From SBP data Correct for thin layer  effects if necessaryObtain the mean effective stress             p    '    Figure 5 - 18 Procedure followed to obtain the normalised cone tip resistance Q from CPT data at depth of the adjacent SBP      Figure 5 - 19 Procedure followed to obtain the initial state parameter 0 from SBP data     195 050 100 150 200 250 0.0 2.0 4.0 6.0 8.0 10.0 12.0 Cavity strain (%) C av ity  p re ss ur e (k Pa ) Test 01 @ 2.7 (m)   a) 0 200 400 600 800 0.0 2.0 4.0 6.0 8.0 10.0 12.0 Cavity strain (%) C av ity  p re ss ur e (k Pa ) Test 09 @ 10.7 (m)   b) 0 200 400 0 200 400 600 800 600  (k Pa ) 800 0.0 2.0 4.0 6.0 8.0 10.0 12.0 Cavity strain (%) 1000  k Pa ) 1200 0.0 2.0 4.0 6.0 8.0 10.0 12.0 Cavity strain (%) C av ity  p re ss ur e ( C av ity  p re ss ur e Test 10 @ 11.5 (m) Test 11 @ 16.2 (m)    c)  d) 0 200 400 600 800 1000 1200 1400 0.0 2.0 4.0 6.0 8.0 10.0 12.0 Cavity strain (%) C av ity  p re ss ur e (k Pa ) Test 13 @ 18.2 (m)   e) 0 200 400 600 800 1000 1200 1400 0.0 2.0 4.0 6.0 8.0 10.0 12.0 Cavity strain (%) C av ity  p re ss ur e ( k Pa ) Test 14 @ 20.0 (m)   f) 0 200 400 600 1200 1400 800su re 1000 0.0 2.0 4.0 6.0 8.0 10.0 12.0 Cavity strain (%) C av ity  p re s  ( kP a) Test 15 @ 21.0 (m)   g) Figure 5 - 20 IFM calibration of SBP to determine range of  0 and K0 (SBP corrected for finite geometry effects)  196 -0.30 -0.25 -0.20 -0.15 -0.10 -0.05 0.00 -0.30 -0.25 -0.20 -0.15 -0.10 -0.05 0.00 0 from CPT  0  fr om  S BP 10.7m 11.5m 2.7m 18.2m 21m 20m 16.2  Figure 5 - 21 Comparison of back calculated from CPT and SBP (numbers on the figure indicate test depth)    197 5.12. References  Ahmadi M.M., and Robertson P.K. 2005. Thin-Layer Effects on the CPT qc Measurement. Canadian Geotechnical Journal, 42: 1302–1317. Ajalloeian R., and Yu H. S. 1998. Chamber Studies of the Effects of Pressuremeter Geometry on Test Results in Sand. Géotechnique, 48(5): 621–636. ASTM D5778 - 07 . Standard Test Method for Electronic Friction Cone and Piezocone Penetration Testing of Soils. Published : November 2007. Been K., and Jefferies M.G. 1985. A State Parameter for Sands. Géotechnique, 35(2): 99-112. roth Memorial Symposium, Thomas Telford, London, 121-134. Been K., Crooks J.H.A., Becker, D. E., and Jefferies M.G. 1986. The Cone Penetration Test in Sands: Part I, State Parameter Interpretation. Géotechnique, 36(2): 239-249. Been K., Jefferies M.G., Crooks J.H.A., and Rothenberg L. 1987a. The Cone Penetration Test in Sands: Part II, General Inference of State. Géotechnique, 37(3): 285-299. Been K., Lingnau B.E., Crooks J.H.A., and Leach B. 1987b. Cone Penetration Test Calibration for Erksak (Beaufort Sea) Sand. Canadian  Geotechnical Journal, 24: 601-610. Been K., Jefferies M.G., and Hachey J. 1991. The Critical State of Sands. Géotechnique, 41(3): 365-381. Bishop A.W. 1966. Strength of Soils as Engineering Materials. 6th Rankine Lecture, Géotechnique, 16: 89-130. Carter J.P., Booker J.R., and Yeung S.K. 1986. Cavity Expansion in Cohesive Frictional Soils. Géotechnique, 36(3): 349-358. Been K., and Jefferies M.G. 1992. Towards Systematic CPT Interpretation. Proceedings W  198 Collins I. F., Pender M.J., and Yan W. 1992. Cavity Expansion in Sands under Drained Loading Conditions. International Journal for Numerical and Analytical Methods in Cunha uremeter Tests in Sand. PhD thesis, Cunnin ve Velocity to Evaluate in Situ de Bor  of Static Penetration Tests. , Ghafgh f the Critical State Friction ymposium on Goldby h to Fill dian Conference on Marine Geotechnical Engineering, St. John's, Golder a) Ltd. 1986. Report on Review of In Situ Testing Tarsuit P- Geomechanics, 16: 3–23. R.P. 1994. Interpretation of Selfboring Press Department of Civil Engineering, the University of British Columbia. g J.C., Robertson P.K., and Sego D.C. 1995. Shear Wa State of Cohesionless Soils. Canadian Geotechnical Journal, 32(5): 848-858. st R., and Vermeer P.A. 1982. Finite Element Analysis Proceedings of the 2nd European Symposium on Penetration Testing, Amsterdam Netherlands, 457-562. azi M., and Shuttle D.A. 2006. Accurate Determination o Angle from Triaxial Tests. Proceedings of the 59th Canadian Geotechnical Conference, Vancouver, 278-284. Ghionna V.N., and Jamiolkowski M. 1991. A Critical Appraisal of Calibration of Calibration Chamber Testing of Sands. Proceedings of the 1st International S Calibration Chamber Testing, ISOCCT 1, Potsdam, New York, 13-39.  H.M., Crooks J.H.A., Harper J.R., and Stuckert B. 1986. Novel Approac Material Quality Assessment near Real Time Grading of Dredged Sand. Proceedings of 3rd Cana Newfoundland, Canada, 1: 409-427.  Associates (Western Canad 45 Research Project, April 1986.  199 Hicks M. A., and Smith I. M., 1986. Influence of Rate of Pore Pressure Generation on the Stress-Strain Behaviour of Soils. International Journal for Numerical Methods in Engineering,  22(3): 597-621. ineering, 126(11): 979-989. Hsu H.H., and Huang A.B. 1998. Development of an Axisymmetric Field Simulator for Cone Huang 528. Huntsm n. Proceedings of In Situ '86, Use of In Situ Tests Hicks M.A., and Smith I.M. 1988. Class A Prediction of Arctic Caisson Performance. Géotechnique, 38(4): 589-612. Hicks M.A., and Onisiphorou C. 2005. Stochastic Evaluation of Static Liquefaction in a Predominantly Dilative Sand Fill. Géotechnique, 55(2): 123-133. Hofmann B.A., Sego D.C., and Robertson P.K. 2000. In Situ Ground Freezing to Obtain Undisturbed Samples of Loose Sand. ASCE, Journal of Geotechnical and Geoenvironmental Eng Holden J.C. 1991. History of the First Six CRB Calibration Chambers. Proceedings of 1st International Symposium on Calibration Chamber Testing, Potsdam, New York, 1-12. Penetration Tests in Sand. Geotechnical Testing Journal, 21(4): 348-355. W., Sheng D., Sloan S.W., and Yu H.S. 2004. Finite Element Analysis of Cone Penetration in Cohesionless Soil. Computers and Geotechnics, 31: 517– Hughes J.M.O., Wroth C.P., and Windle D. 1977. Pressuremeter Tests in Sands. Géotechnique, 27(4): 455-477. an S.R., Mitchell J.K., Klejbuk L.W. Jr., and Shinde S.B. 1986. Lateral Stress Measurement during Cone Penetratio in Geotechnical Engineering, Blacksburg, VA, USA. ASCE Geotechnical Special Publication 6: 617-634.  200 Jefferies M.G. 1988. Determination of horizontal geostatic stress in clay with self-bored pressuremeter. Canadian Geotechnical Journal, 25(3): 559-573. .Jefferies M.G. 1993. Nor-Sand: a Simple Critical State Model for Sand. Géotechnique, 43(1): Jefferies M.G., and Been K., 1995. Cone Factors in Sand. Proceedings of CPT '95, International Jefferie  Calibration and Use. Proceedings of d Advances in Jefferie . Taylor & Jefferies M.G., Stewart H.R., Thomson R.A.A., and Rogers B.T. 1985. Molikpaq Deployment at Jefferie idation in Some Beaufort Sea Clays. Canadian Geotechnical Journal, 24(3): 342-356. 91-103. Symposium on Cone Penetration Testing, Swedish Geotechnical Society, Linkoping, Sweden, Oct. 4-5 1995: 187-193. Jefferies M.G., and Shuttle D.A. 2002. Dilatancy in General Cambridge-Type Models. Géotechnique, 52(9): 625-638. s M.G., and Shuttle D.A. 2005a. NorSand: Features, Specialty Conference on Soil Constitutive Models: Evaluation, Selection, and Calibration. ASCE Geotechnical Special Publication, 128: 204-236. Jefferies M.G., and Shuttle D.A. 2005b. NorSand: Calibration and Use. Proceedings of the 11th Conference of the International Association for Computer Methods an Geomechanics, Turin, 345-352. s M.G., and Been K. 2006. Soil Liquefaction: a Critical State Approach Francis (Abingdon & New York). ISBN 0-419-16170-8. Tarsiut P-45. Proceedings of ASCE Specialty Conference on Civil Engineering in the Arctic Offshore, San Francisco, Ca., 1-27. s M.G., Crooks J.H.A., Becker D.E., and Hill P.R. 1987. Independence of Geostatic Stress from Overconsol  201 Jefferies M.G., Rogers B.T., Stewart H.R., Shinde S., James D., and Williams-Fitzpatrick S. 1988. Island Construction in the Canadian Beaufort Sea. Proceedings of specialty  Nash D.F.T., Powell J.J.M., and Lloyd I.M. 1992. Initial Investigations of the Soft Clay Test Peterso 91. Overview of a Large Stress Chamber System. Plewes ensity Logging in Plewes Eva th Canadian Geotechnical Plewes 94. In Situ Sampling, Density Popesc 997. Effects of spatial variability on soil Robert ical Journal, 20: 718-733. conference on Hydraulic Fill Structures, ASCE Geotechnical Special Publication, 21: 816-883. Konrad J.M. 1988. Interpretation of Flat Plate Dilatometer Tests in Sands in Terms of the State Parameter. Géotechnique, 38(2): 263-277. Site at Bothkennar. Géotechnique, 42(2): 163-181. n R.W., and Arulmoli K. 19 Proceedings of First International Symposium on Calibration Chamber Testing, Potsdam, New York, 329-338.  H.D., McRoberts E.C., and Chan W.K. 1988. Downhole Nuclear D Sand Tailings. ASCE Geotechnical Special Publication, 21: 290-309.  H.D., Davies M.P., and Jefferies M.G. 1992. CPT Based Screening Procedure for luating Liquefaction Susceptibility. Proceedings of the 45 Conference, Toronto.  H.D., Pillai V.S., Morgan M.R., and Kilpatrick B.L. 19 Measurements, and Testing of Foundation Soils at Duncan Dam. Canadian Geotechnical Journal, 31: 927–938. u R., Prevost J.H., and Deodatis G. 1 liquefaction; some design recommendations.  Géotechnique, 47(5): 1019-1036. son P.K., and Campanella R.G. 1983. Interpretation of Cone Penetration Tests. Part I: Sand. Canadian Geotechn  202 Schofield A.N., and Wroth C.P. 1968. Critical State Soil Mechanics. McGraw Hill, London.  D.A. 2006. Can the effect of sand fabric on plastic hardening be determined using self- bored pressuremeter? Canadian Geotechnical Journal, 43: 659-673. Shuttle Shuttle D.A. and Jefferies M.G. 1995. Reliable parameters from imperfect SBP tests in clays, Shuttle rameters from Imperfect SBP Tests in Clays.  (ed.), published January 1996 by Shuttle alytical Methods in Geomechanics, 22: 351- Shuttle ompaction Grout , 104: 48-64 Stewar and Goldby H. M. 1983. Berm Construction for the Gulf Canada ICE Conference on Advances in Site Investigation Practice, London, March 1995.  D.A., and Jefferies M.G. 1996. Reliable Pa Advances in Site Investigation Practice: Proceedings of the International Conference Held in London on 30-31 March 1995, C. Craig Thomas Telford, ISBN: 0727725130, 571-585,  D.A., and Jefferies M.G. 1998. Dimensionless and Unbiased CPT Interpretation in Sand. International Journal of Numerical and An 391.  D.A. and Jefferies M.G. 2000. Prediction and Validation of C Effectiveness. Proceedings of Geo-Denver 2000, Advances in Grouting and Ground Modification, August 5-8 2000, ASCE Geotechnical Special Publication, Krizek R.J., and Sharp K. (eds.) Sladen J.A. 1989. Problems with Interpretation of Sand State from Cone Penetration Test. Géotechnique, 39(2): 323-332. t H.R., Jefferies M.G., Mobile Arctic Caisson. Proceedings of Annual Offshore Technology Conference, 2: 339-346.  203  204  of usetts, 1: 391-396. Soils, van de Woods R.D., Benoît J., and deAlba P. 1994. National Geotechnical Experimentation Sites. Yu H.S Yu H.S., and Houlsby G.T. 1991. Finite Cavity Expansion in Dilatant Soils: Loading Analysis. Yu H.S., Schnaid F., and Collins I.F. 1996. Analysis of Cone Pressuremeter Tests in Sands. Tan N.K., Miller G.A., and Muraleetharan K.K. 2003. Preliminary Laboratory Calibration Cone Penetration in Unsaturated Silt. Proceedings, 12th Pan American Conference on Soil Mechanics and Geotechnical Engineering, Cambridge, Massach Tavenas F.A. 1973. Difficulties in the Use of Relative Density as a Soil Parameter. Evaluation of Relative Density and its Role in Geotechnical Projects Involving Cohesionless American Society for Testing and Materials, Selig E.T., and Ladd R.S. (eds.), Philadelphia, ASTM Special Technical Publication 523: 478-483. n Berg P. 1994. Analysis of Soil Penetration. PhD Thesis, Delft University of Technology, ISBN 90-407-1004-X. Willson S.M., Ims B.W., and Smith I.M. 1988. Finite Element Analysis of Cone Penetration. Penetration Testing in the UK, Thomas Telford, London, 157-159. Geotechnical News, 12 (1): 39-44. . 1994. State Parameter from Self-Boring Pressuremeter Tests in Sand. ASCE, Journal of Geotechnical Engineering, 120(12): 2118-2135. Géotechnique, 41(2): 173-183. ASCE Journal of Geotechnical Engineering, 122(8): 623-632.  Chapter 6.   Interpretation of the Sand State from CPT in  6.1. With the recent advances in the analysis and de es used in geomechanics, accurate community. In the case of cohesi includes knowledge of in-situ The behaviour of cohesionless soils strongly de Dr, is an almost universally used density index for sand, it is easily shown that Dr can be misleading (e.g. Tavenas, 1973). An alternative to Dr that captures both the effect of void ratio and mean stress on soil behaviour is the state parameter,  (Been and Jefferies, 1985). The state   Fraser River Sand: A Case History8   Introduction  sign techniqu interpretation of ‘ground truth’ has become of even greater significance to the geotechnical onless soils, ‘ground truth’ gradation, density, fabric and stress state, and the spatial variability of these parameters. pends on their density. While relative density,  8 A version of this chapter has been published. Ghafghazi M., and Shuttle D.A. 2010. Interpretation of the In-situ Density from Seismic CPT in Fraser River Sand. Proceedings of the 2nd International Symposium on Cone Penetration Testing, Huntington Beach, California.  205  parameter is defined as the difference between the current void ratio of the soil and its critical oid ratio at the same mean effective stress. However, determining the in-situ  (or Dr) is very difficult because of density changes during sampling – an “undisturbed” sample is essentially possible in normal engineering practice. Penetration tests have thus become the norm for sting cohesionless soils, with the modern electronic CPT offering continuous data The difficulty with any penetration test, however, is that the state measure of interest is not e; a process usually referred to as interpretation. This interpretation involves solving an inverse boundary value problem to obtain mechanical properties from the measurements. Ghafghazi and Shuttle (2008) (presented in chapter 4) analysed a database of nine soils, including laboratory standard and natural sands, as well as relatively clean sand-size tailings, for which both chamber testing and triaxial compression data were available in the literature. This methodology offers a framework for interpreting the state parameter from CPT tip resistance. The interpretation framework is applied to the Massey Tunnel site, an extensively investigated site in Fraser river delta in British Columbia, Canada. The effect of soil fabric on the interpretation results has been considered by adjusting the calibration parameters with respect to tests on undisturbed samples. The accuracy of the method is evaluated by comparison to in-situ density measurements and compared to other methods of interpreting the state parameter from CPT.  v im te measurement with excellent repeatability and accuracy at relatively low cost. measured. Instead it is calculated from the penetration resistanc    206  6.2. Site Investigation Program  The Canadian geotechnical community completed a major collaborative research project between 1993 and 1997 entitled the CANadian Liquefaction EXperiment (CANLEX). The project was divided into different phases. One of the sites investigated in Phase II was located south of Massey Tunnel, connecting Richmond and Delta (Figure 6-1a) in British Columbia. ial tio of 0.96 with standard deviation (SD) of 0.05. sat and dry were accordingly calculated s 18.2 The full program was reported in five companion papers (Robertson et al., 2000a, b; Wride et al., 2000a, b; and Byrne et al., 2000). Complete summary reports with all data were published in a five volume series; the data for the Massey site reported in this chapter are extracted from volume 4 (Wride and Robertson, 1997). The site characterisation program was targeted at depths of 8 to 13 m and included two standard penetration tests (SPT), six seismic cone penetration tests (CPT), three boreholes with Self- Bored Pressuremeter tests (SBP), and two geophysical logs. Ground freezing and sampling was carried out, providing undisturbed cores which were trimmed into samples used in both triax and simple shear testing. Samples were also obtained using the Laval Large Diameter Sampler. Figure 6-1b shows the locations of the tests relevant to the current work. The water table was measured at 2.3 m depth. Based on SBP tests Wride and Robertson (1997) suggested the coefficient of lateral earth pressure at rest of K0 = 0.5 for the target depth range; this evaluation was adopted here. The frozen and Laval large diameter samples yield an average void ra  3mkN and 13.4 3mkN respectively for computing the vertical stresses. a    207  6.3. Material and Testing  Fraser River Sand (FRS) is a uniform, angular to sub-angular with low to medium sphericity medium grained clean alluvial sand widely spread in the Fraser river delta. For Massey samples emin (ASTM D4254) and emax (ASTM D4253) are reported as 0.677 and 1.056 and  2.68  Gs  . Laboratory testing of the Massey site samples included testing reconstituted and undisturbed reduced to 0.8% to  S sample fines content has been t s situ (or initial) alue of  under geostatic conditions. ones to evaluate the soil response to both undrained monotonic and cyclic loading. Since this method requires drained triaxial compression tests over a range of stresses and densities, which were not available in the CANLEX database, a second set of data on a batch of FRS entitled the “UBC sample” has been used. The FR produce a clean sand for testing. emin and emax are reported as 0.627 and 0.989 and  2.719  Gs  by Shozen (1991). emin was measured according to ASTM D2049 while emax was reported as the initial deposition void ratio in the loosest state. Figure 6-2 illustrates the gradation curves of the two FRS samples used in this work.  6.4. Methodology  The CPT in sand provides just two outputs; tip resistance (q ) and sleeve friction (f ); the penetration is drained so the pore pressure transducer simply measures the in-situ pore pressure. The state measure used in this work is the state parameter, . Because  is used as an internal state variable in the numerical model, the subscript ‘0’ is used to denote the in- v  208  Initial work with determining 0 from CPT data comprised triaxial testing of sands for which hamber test data was available to define the critical state locus (CSL) of each sand, and then c processing the chamber test data to develop dimensionless relations (Been et al., 1987) of the form:   0exp mkQ           [Eq. 6 - 1]  where Q is the normalised tip resistance defined as:  0 0 'p pqt  where p0 is the initial mean total stress, and p′0 is the initial mean effective stress. The two coefficients k and m in Equation 6-1 differ from one sand to another. Q           [Eq. 6 - 2]  ethods of interpreting 0 from CPT must be verified against available ata, the scatter in experimental results limits the interpretation method’s ccuracy. Experimentalists have tried to measure reproducibility of calibration chamber results lues suggests that  0.05 is about the best accuracy that can be expected in prediction of any particular test in the available calibration chamber data. As all analytical m calibration chamber d a by repeating tests on samples with the same density and stress conditions. These efforts have resulted in  25% error in measured Q in recent works (Hsu, 1999). However, the majority of available data suggest that 50% accuracy in measured Q is ‘good’ quality data. Of course, translating accuracy in Q to accuracy in 0 includes some sort of interpretation, but a rough estimation based on average k and m va  209  Ghafghazi and Shuttle (2008) analysed a total of 301 calibration chamber tests and achieved an error of less than 0.04 with 78% confidence level and less than ±0.07 with 92% confidence level. Their results were improved to 84% and 97% respectively for cases where elasticity was measured using bender elements. This is deemed to be an excellent accuracy achievable in an nalytical method. This method is used here for evaluating the state parameter at the Massey site. he method involves two parallel tasks: the normalisation and processing of CPT tip resistance our and calculation of material specific orrelations. The correlations are then used to calculate the state parameter. Figure 6-3 presents a summary of the method and the required steps. Normalising and processing the CPT tip resistance data is done by estimating the stress state long the depth of interest. The total vertical and effective stresses can be calculated by ter tests. The mean stress a T data, together with identification of soil behavi c a estimating dry and saturated soil densities. The coefficient of earth pressure at rest (K0) should be estimated or measured from in-situ tests such as SBP or dilatome can then be calculated and the normalised tip resistance Q can be calculated from Equation 6-2. Central to the method is the application of a shape function which converts the calibration chamber test normalised tip resistance Qcc to its spherical cavity expansion analysis equivalent Qsph. Replacing the calibration chamber test normalised tip resistance Qcc with its analogue in- situ normalised tip resistance Q and inverting Equation 4-3 (Ghafghazi and Shuttle, 2008) we can write:  59.0 7.0  QQsph          [Eq. 6 - 3]    210  6.5. Constitutive Modeling  The behaviour of the material is captured through constitutive model calibration to drained triaxial compression tests. The calibrated model is then used in a spherical cavity analysis to calculate Qsph . The reason to use a ‘good’ sand model for cavity expansion analysis is that, in general, part of ide range of particulate soil behaviour. The  n tests (one test with unload-reload loops was not used). The elastic shear modulus as identified through bender element tests. Details of the testing are presented in detail in chapter 7 and appendix A. The calibration param Table 6-1. the domain will be critical, parts dilating, and other parts contractive; with the exact behaviour being a function of position relative to the cone tip. The constitutive model adopted is NorSand (Jefferies, 1993; Jefferies and Shuttle, 2002), an isotropically hardening - isotropically softening generalised critical state model that captures a w reader is directed to Jefferies and Shuttle (2005) for a detailed discussion of the calibration procedures. The CSL was identified based on the end points of 9 consolidated drained (CID) and 7 consolidated undrained (CIU) “UBC sample” triaxial compression tests as illustrated in Figure 6-4. The samples were prepared by moist tamping, and lubricated end platens were used to reduce the end friction. As not all of the tests reached the critical state, whether the sample was dilating or contracting was reported to help define the CSL. Corresponding data to determine the CSL of the “Massey” samples are also shown on Figure 6-4. The remaining model calibration was based on 8 of the drained “UBC sample” triaxial compressio w eters for the UBC sample are presented in  211  To ac ommodate the differencec  between the material tested at UBC and the material found at e site, model parameters that could be estimated using the Massey sample data replaced the  to identify the critical state line ( th “UBC” parameters in the analysis. The Massey sample data included 22 undrained triaxial tests and one drained triaxial compression test on undisturbed (frozen core) samples. The undrained tests were used e, ) (see Figure 6-4), and the critical state friction ratio (Mtc). This allowed the calibration parameters to be modified to match altering material behaviour caused by changing gradation. An adjustment to the hardening parameter H, which is affected by soil fabric, was made based on the drained test on the Massey sample (see Figure 6-5). The calibration parameters for the Massey sample are also presented in Table 6-1. It is worth noting that the CSL in 'log pe   is bilinear at lower stresses for both UBC and Massey samples. As discussed earlier in the thesis, the shape of the CSL is an arbitrary choice and does not affect the fundamentals of the model or the results. In this case a linear or a power fective stress . The cavity is monotonically expanded by increasing its radius until law (Jefferies and Shuttle, 2011) relation could replace the familiar semi log representation. One should also notice that the measurements of the mean effective stress at low stresses become very sensitive to the accuracy of the testing equipment. At higher stresses (800-900 kPa for FRS), particle breakage controls the status of the CSL as discussed later in chapter 7. In this chapter the semi-log representation of the CSL is considered sufficient.  6.6. Spherical Cavity Expansion Analysis  The spherical cavity expansion analogy idealises the CPT as a cavity in an infinite uniform medium under an isotropic stress state, with the internal pressure of the cavity initially equal to the mean ef 0p  212  a limiting (constant) pressure is obtained. This idealisation greatly simplifies the analysis because the spherical symmetry allows only radial displacements, in turn permitting a one- dimensional description of the problem. The corresponding stresses are a radial and two equal hoop stresses. The spherical cavity finite element code developed by Shuttle and Jefferies (1998) is used in this study. The code and finite element mesh remained the same, hence retaining the verified large displacement performance of it.  6.7. Inverse Form for Interpretation of CPT  Shuttle and Jefferies (1998) showed t tion 6-1 may be used to recover 0 from CPT data provided that k and m are functions of soil characteristics and the stress level. Using a spherical cavity expansion analysis, we can summarise the effect of different soil characteristics in the form of NorSand parameters in the following equation:  ksph = f1(G /p′0, Mtc, N*, H, tc, e, )       [Eq. 6 - 4a] msph= f2(G /p′0, Mtc, N*, H, tc, e, )       [Eq. 6 - 4b]  hat Equa ll the NorSand parameters in Equation 6-4 are constants, or known functions of 0. Hence at a take a single value except for G / p′0 (the stress vel effect) which is usually a function of both void ratio and stress level. This makes Qsph a A particular 0, all the variables in Equation 6-4 le function of the stress level at a particular 0 (Figure 6-6a). ksph and msph can then be determined as functions of  G / p′0 (Figure 6-6b) in Equation 6-5.   213   0maxln336.082.9 pGksph         [Eq. 6 - 5a]  ln49.077.1 pGm         [Eq. 6 - 5b] Q Q k m from Equation 6-5, provided that the shear modulus G is independently available. The in- 0maxsph  Having the normalised tip resistance , sph can be determined from Equation 6-3, and sph and sph situ state parameter can then be calculated from Equation 6-6 which is directly deduced from writing Equation 6-1 for spherical cavity expansion:  sph sphk  ln sph m Q  0          [Eq. 6 - 6]  Combining Equations 6-1, and 6-3 and following the procedure in Figure 6-6, one can directly lculate the state parameter for the in-situ Massey site Fraser river sand from Equation 6-7 as ca  )pG0.83ln(+3.01 )pG2.28ln(+33.05 ln      Q 0max 0max 0          [Eq. 6 - 7]  Diameter Sampler coring (shown as LDS n Figure 6-1b) was also located on the perimeter of the layout. Although the target zone was  6.8. Analysis and Results  Testing at the Massey site included a Frozen sampling core surrounded by six CPTs at a radius of approximately 5 m (Figure 6-1b). The Laval Large o  214  identified as a fairly uniform l measurements fell within a range bet e  8 to 9 m depth and relative uniformity between 9 and 13 m. Parts of the logs obtained from tests M9401 and M9402 were ignored between 12 and m due to their discrepancy with the trend established by the rest of the tests. The range of tip resistance measured within the target zones of the CPT selected for analysis, is plotted in Figure 6-8a. The normalised tip resistance Q is also plotted in Figure 6-8b as a range. ough downhole seismic measurements for all CPTs. The results were shown in the form of elastic shear modulus G in Figure 6-9 with the range of interest and the aver any given depth. onsidering  a range of 0 was calculated using Equation 6-7 (Figure 6-8c). Ghafghazi and Shuttle (2008) margins of error of ±0.04 and ±0.07 are also plotted. al. (1987) and Plewes et al. (1992) are plotted in Figure 6-8d. To be able to directly compare methods, the average and standard deviation (SD) of 0 data obtained from each method are summarised in Table 6-2. sions  ayer, and the tests were relatively close, the CPT tip resistance ween approximately 4 to 8 MPa. The tests plotted in Figur 6-7 suggested lower tip resistance values from 13 The shear wave velocity was determined thr age values plotted. The analyses were done using the average values at  5.00 KC The results obtained from the methods (see section 1.4.2) proposed by Konrad (1997), Been et  6.9. Discussion and Conclu  The samples trimmed from the cores obtained from ground freezing and Laval large diameter sampler are assumed to represent the real in-situ void ratio. An average void ratio of 0.96 with SD of 0.05 is obtained, translating into a state parameter of -0.055 with the same SD based on  215   = 1.17 and e = 0.035 obtained for Massey sample. The wide scatter in the measured void ratios, reflected in the large SD, covers a range of 01.0155.0 0    . The scatter more likely stemmed from the ground sampling techniques, rather than being a characteristic of the ground, as widely ranging void ratios were measured in samples from adjacent points of the core. While the variation in the measured void ratios represents a wide range of sand behaviour, ranging from loose to dense, the geology of the site and all in-situ testing (including CPTs) imply a relatively uniform deposit associated with a narrower range of ground density and sand UBC and Massey samples are plotted in Figure 6-10 along with that of series of re. The gradation of the ccepting the notion that the behaviour of the material is controlled by the state parameter (or d ratio (or relative density Dr) one can see that any direct easurement of void ratio, no matter how accurate (as in this case history), can lead to behaviour. This paradoxical observation in methods of obtaining the in-situ density from undisturbed samples of sands calls for a more cautious treatment of the results, and emphasises the need for better interpretation techniques for tests such as CPT. The CSLs of tests on a slightly different gradation of Fraser river sand at University of Alberta (Chillarige, 1995). The in-situ measurements of void ratio are also plotted in this figu “U of A sample” is also plotted in Figure 6-2. Although the three samples are of the same origin and have similar gradations, the variability in their CSLs is significant. This variability translates into a difference of 0.05 to 0.10 in the state parameter at the stress levels associated with Massey site. A other similar indexes) and not the voi m significant misrepresentations of the field behaviour. This misrepresentation will occur due to variations of the CSL, if the CSL is not rigorously determined for every variation of the material  216  gradation encountered. This can add significantly to the cost and difficulty of estimating the in- situ behaviour of the material by direct measurement . On the contrary, the framework presented here has the advantage of directly providing the state parameter based on conveniently available CPT data. Once the CSL is established for a representative sample for capturing the essence of material behaviour, variations of the CSL due to local and limited material variability will have a second degree effect on the accuracy of the method, as discussed in detail by Shuttle and Jefferies (1998) in investigating the sensitivity of the interpretation to changes in  and e. As illustrated in Figure 6-8c, 98% of the undisturbed sampling measurement points fall within the ± 0.07 error margins of the Ghafghazi and Shuttle (2008) m s of the void ratio ethod and 70.5% within the ± 0.04 error margins. 20.5%, fall within the upper bound and lower bound lines, representing a zone of ideal accuracy for the information used in this work. The confidence levels are well comparable with those observed in validation of the method against calibration chamber testing results. The accuracy offered by the ± 0.07 error margins covers a wide range of possible state parameters (typically 05.030.0 0   ) when combined with the variation in the original CPT data represented by upper and lower bound envelopes. However, the range is very similar to that covered by the frozen and LDS samples, suggesting that this method is as capable as the most expensive and cumbersome of ground sampling techniques for determining the soil’s in-situ density. The average 0 of -0.067 is also very close to that measured by ground sampling. The difference is close to ±0.01; the ground sampling technique error margin given by Wride and Robertson (1997).  217  As shown in Table 6-2, amongst all interpretation methods, the Ghafghazi and Shuttle (2008) method provides the closest estimation for 0 of Fraser river sand. Although the method appears more difficult than the others, the difficulty only lies in the analysis and modeling effort which is achievable in a matter of hours. With the exception of Plewes et al. (1992), all interpretation methods presented in Table 6-2 require knowledge of the Critical State Locus in pe  log space. Performing a number of triaxial compression tests is the easiest and most common way of estimating the CSL. The only additional requirement of the Ghafghazi and Shuttle (2008) method is for these tests to be performed under drained conditions; a requirement that does not pose any additional laboratory testing effort. Plewes et al. (1992) correlated the slope of the critical state line in pe  log  space to the CPT friction ratio based on experimental results, hence eliminating the need to experimentally obtain the CSL. However, the method does require laboratory testing to measure the critical state friction angle (or the analogous Mtc). Plewes et al. (1992) suggested using Mtc = 1.2 ( 30 ) c for all soils, advising that doing so would cause less than 10% error in the estimated 0. Likely due to the high Mtc of Fraser river sand, adopting the Mtc = 1.2 approximation changes the 0 prediction by about 25% in these analyses, resulting in a less accurate prediction that puts the 0 measured Mtc results in a less negative (looser) state parameter that is closer to that implied by ground sampling techniques. echanical aspects of soil behaviour and returns the most discrepancy in estimated state parameter. The Been et al. (1987) method accounts for material compressibility estimated   in line with the other methods (see the last two columns of Table 6-2). Use of the The success of these methods in obtaining the in-situ state parameter appears to be directly related to the level of material behaviour taken into consideration. Konrad (1997) does not account for any m  218  through the slope of the CSL. The Plewes et al. (1992) method adds the effect of the critical state friction angle to their framework resulting in an even better estimation. The Ghafghazi and Shuttle (2008) analytical procedure also accounts for both compressibility and friction angle, and importantly adds elasticity, as well as stress level, dilatancy, and fabric. An overall comparison of the model parameters for Fraser river sand to those of other sands presented in Ghafghazi and Shuttle (2008) suggests that the most important factor that makes the other methods systematically biased towards a more negative state parameter in Fraser River sand is its high critical state friction angle. This is further confirmed by the fact that a correct Mtc value in the Plewes et al. (1992) method results in a better interpretation.  219  Table 6 - 1 NorSand calibration parameters for Fraser river sand  Critical State Plasticity Elasticity Parameter  e Mtc H tc N* Ir = G/p′0  UBC (moist tamped) 1.22 0.060 1.45 80-3100 3.2 0.45   42.00 1 289.14       p p e e † 0.2 2  r Massey 1.17 0.035 1.49 110-3100 3.2 0.45 Seismic CPT(undisturbed) (650< Ir < 800) 0.2 † pr is a reference pressure of 100 kPa. The correlation is obtained from bender element testing on some of the samples.    0 Interpretation Undisturbed Current Konrad Been et al. Plewes et  † Plewes et  Table 6 - 2  interpretation summary method sampling work (1997) (1987) al. (1992)  al. (1992) average 0 -0.055 -0.067 -0.114 -0.092 -0.089 -0.071 SD 0.050 0.028 0.032 0.021 0.038 0.038 † Mtc = 1.20 is used in the formula  220   a) b) re 6 - 1 a) y site in raser elta; b) Layout of the Large D ter Laval re, and the CPTs (M9401 to M9406) relevant to the work at Massey site (after Wride and Robertson, 1997)   Figu Masse  the F river d iame Sampler (LDS), the frozen co 0 20 40 60 80 100 0.010.11 Diameter (mm ) Pe rc en t p as si ng UBC sample, Shozen (1991) Massey sample, Robertson et al. (2000a) U of A sample, Chillarige (1995)  Figure 6 - 2 Gradation curves of Fraser river sand: UBC and Massey samples   221  Perform drained triaxial compression tests Calibrate the numerical model (NorSand) Perform spherical cavity expansion analysis & obtain ksph & msph vs. G / 0p  correlation Calculate 0 using Equation 6-6 Estimate the coef. of lateral earth pressure at rest (k0) Normalise CPT tip resistance (qt) to obtain Q from Equation 6-2 Calculate the total & effective mean stresses 0p  & 0p Calculate Qsph using Eq ion 6-3  Figure 6 - 3 Flowchart for Ghafghazi and Shuttle (2008) method uat 0.7 0.8 0.9 1.0 1.1 10 100 1000 p' (kPa) vo id  r at io , e Massey, No Volume Change Massey, Contracting Massey, Dilating UBC, No Volume Change UBC, Contracting UBC, Dilating Massey sample CSL UBC sample CSL CSL curvature  Figure 6 - 4 Critical State Loci for UBC and Massey samples; end points of drained and ndrained tests are plotted with different signs assigned to tests that were contracting, dilating or had not volume change at the end of the test u  222  0 50 100 150 200 -0.3 -0.25 -0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15  0 H Massey Sample UBC Sample H = 80-310 0 H = 110-310 0  Figure 6 - 5 Variation of the hardening parameter H with the initial state parameter for UBC and Massey samples   m sph  = 1.77 + 0.49 ln(G  / p 0' ) k sph  = 9.82 + 0.336 ln(G  / p 0' ) 4 6 8 10 12 14 200 400 600 800 1000 1200 1400 G  / p 0' m sp h ,  k sp h  Q sph  = 11.80e -4.69 0 Q sph  = 12.10e -5.08 0 Q sph  = 12.17e -5.23 0 10 15 20 25 30 -0.25 -0.2 -0.15 -0.1 -0.05 0 0.05  0 Q sp h G/p'0=1200 G/p'0=800 G/p'0=400     a)    b) Figure 6 - 6 a) Qsph vs.  0 for range of 0pGI r  ; b) msph and ksph vs. normalised shear modulus  rI  223  0 5 10 15 20 0 5 10 15 20 25 Tip resistance, q t  (MPa) D ep th  ( m ) M9401 M9402 M9403 M9404 M9405 M9406 ta rg et  z on e 0.00 0.05 0.10 0.15 0.20 Sleeve friction, f s (MPa ) -0.1 0.0 0.1 0.2 0.3 Pore pressure, u 2 (MPa ) 0 1 2 Friction ratio, F r (%) 3 The portion of data excluded from the analysis         a)                                   b)                     c)       d) Figure 6 - 7 CPT data M9401 to M9406 a) Tip resistance b) Sleeve friction c) Pore pressure d) riction ratio F   8 9 10 11 12 13 2 4 6 8 10 Tip resistance, q t (MPa) D ep th  ( m ) -0.20 -0.15 -0.10 -0.05 0.00 0.05 state parameter,  0 -0.20 -0.15 -0.10 -0.05 0.00 0.05 state parameter,  0 50 70 90 110 130 150 Normalised tip resistance, Q Frozen samples Laval Samples 0.040.04 0.07 Frozen samples Laval Samples Konrad 1997 Been et al. 1987 Plewes et al. 1992 0.07          a)              b)     c)          d) Figure 6 - 8 Upper and lower bound CPT response and state parameter interpretation for the target zone: a) Tip resistance b) Normalised tip resistance c) State parameter Interpretation (Ghafghazi and Shuttle, 2008) with ±0.04 and ±0.07 error margins d) Alternative methods of interpretation: Konrad (1997), Been et al. (1987), and Plewes et al. (1992)  224  G 6 8 10 12 14 16 18 20 20 40 60 80 100 max  (MPa) D ep th  (m ) M9401 M9402 M9403 M9404 M9405 M9406 average ta rg et  z on e  igure 6 - 9 Shear modulus profile derived from shear wave velocity measurements from seismic CPTs, the average values and the target depth range F  225  0.7 0.8 0.9 1.0 1.1 10 100 1000 p' (kPa) vo id  r at io , e Massey sample CSL UBC sample CSL U of A sample CSL Frozen Samples Laval Samples CSL curvature  Figure 6 - 10 Comparison of Critical State Loci for UBC, Massey and U of A samples and the             direct void ratio measurements    226  6.10. References  ASTM D2049-69 Test Method for Relative Density of Cohesionless Soils (Withdrawn 1983). ASTM D4253 - 00(2006) Standard Test Methods for Maximum Index Density and Unit Weight of Soils Using a Vibratory Table. ASTM D4254 - 00(2006)e1 Standard Test Methods for Minimum Index Density and Unit Weight of Soils and Calculation of Relative Density. Been K., and Jefferies M.G. 1985. A State Parameter for Sands. Géotechnique, 35(2): 99-112. Been K., Jefferies M.G., Crooks J.H.A. and Rothenburg L. 1987. The Cone Penetration Test in Sands. Part II: General Inference of State. Géotechnique, 37(3): 285-299. Byrne, P.M. et al. 2000. CANLEX Full-Scale Experiment and Modelling. Canadian hillarige A.V. 1995. Liquefaction and Seabed Instability in Fraser River Delta. PhD Thesis, Department of Civil Engineering, University of Alberta, Canada. hafghazi M. and Shuttle D.A. 2008. Interpretation of Sand State from Cone Penetration Resistance. Géotechnique, 58(8): 623–634. su H.H. 1999. Cone Penetration Tests in Sand under Simulated Field Conditions. PhD thesis, Department of Civil Engineering, National Chiao Tung University, Hsin Chu, Taiwan. efferies M.G. 1993. Nor-sand: A Simple Critical State Model for Sand. Géotechnique, 43(1): 91-103. efferies M.G. and Shuttle D.A. 2002. Dilatancy in General Cambridge-Type Models. Géotechnique, 52(9): 625-638. Geotechnical Journal, 37: 543–562. C G H J J  227  Jefferi s M.G. and Se huttle D.A. 2005. NorSand: Features, Calibration and Use. Proceedings of the specialty conference on Soil Constitutive Models: Evaluation, Selection, and ational Conference on Earthquake Geotechnical Konrad n of a Unified Approach at Two  Geotechnical Robert B.R., Atukorala U., Biggar K.W., Byrne P.M.,  Konrad J.-M., Küpper wart R.A., Watts B.D., Woeller D.J., Youd lla R.G., Cathro D.C., Chan D.H., Czajewski K., Finn W.D.L., Gu W.H., an P.A., Morgenstern N.R., Phillips R., Piché R., Plewes H.D., Scott D., Sego D.C., Sobkowicz J., Stewart R.A., Watts B.D., Woeller D.J., Youd Calibration. ASCE Geotechnical Special Publication, 128: 204-236. Jefferies M.G. and Shuttle D.A. 2011. Understanding Liquefaction through Applied Mechanics. Proceedings of the 5th Intern Engineering, Santiago, Chile, January 10 – 13, 2011.  J.M. 1997. In Situ Sand State from CPT: Evaluatio CANLEX Sites. Canadian Geotechnical Journal, 34: 120–130. Plewes H.D., Davies M.P., and Jefferies M.G. 1992. CPT Based Screening Procedure for Evaluating Liquefaction Susceptibility. Proceedings of the 45th Canadian Conference, Toronto. son P.K., Wride (Fear) C.E., List Campanella R.G., Cathro D.C., Chan D.H., Czajewski K., Finn W.D.L., Gu W.H., Hammamji Y., Hofmann B.A., Howie J.A., Hughes J., Imrie A.S., A., Law T., Lord E.R.F., Monahan P.A., Morgenstern N.R., Phillips R., Piché R., Plewes H.D., Scott D., Sego D.C., Sobkowicz J., Ste T.L., and Zavodni Z. 2000a. The Canadian Liquefaction experiment: an Overview. Canadian Geotechnical Journal, 37: 499–504. Robertson P.K., Wride (Fear) C.E., List B.R., Atukorala U., Biggar K.W., Byrne P.M., Campane Hammamji Y., Hofmann B.A., Howie J.A., Hughes J., Imrie A.S., Konrad J.-M., Küpper A., Law T., Lord E.R.F., Monah  228  T.L., and Zavodni Z. 2000b. The CANLEX Project: Summary and Conclusions. Canadian Geotechnical Journal, 37: 563–591.  T. 1991. Deformation under the Constant Stress State and its EffeShozen ct on Stress-Strain Shuttle biased CPT Interpretation in Sand.  Testing and Materials, Selig E.T., and Ladd R.S. (eds.), Wride Wride rnal, 37: 505–529. Behaviour of Fraser River Sand. MASc. thesis, Department of Civil Engineering, University of British Columbia, Vancouver, Canada.  D.A., and Jefferies M.G. 1998. Dimensionless and Un International Journal for Numerical and Analytical Methods in Geomechanics, 22: 351- 391. Tavenas F.A. 1973. Difficulties in the Use of Relative Density as a Soil Parameter. Evaluation of Relative Density and its Role in Geotechnical Projects Involving Cohesionless Soils, American Society for Philadelphia, ASTM Special Technical Publication 523: 478-483. (Fear) C.E., Hofmann B.A., Sego D.C., Plewes H.D., Konrad J.-M., Biggar K.W., Robertson P.K., and Monahan P.A. 2000a. Ground Sampling Program at the CANLEX Test Sites. Canadian Geotechnical Journal, 37: 530–542. (Fear) C.E., Robertson P.K., Biggar K.W., Campanella R.G., Hofmann B.A., Hughes J.M.O., Küpper A., and Woeller D.J. 2000b. Interpretation of In Situ Test Results from the CANLEX Sites. Canadian Geotechnical Jou Wride C.E., and Robertson P.K. 1997. The Canadian Liquefaction Experiment, Phase II Data Review Report, Massey and Kidd Sites, Fraser river delta, June 1997. ISBN 0-921095- 49-X. Wride C.E. and Robertson P.K. (eds.), BiTech Publishers Ltd., Richmond, BC, Canada.  229  Chap    Particle Breakage and the Critical State of Sand   7.1.  constan howeve ontinuous change to the soil’s ambigu change s such as pile driving breakag framew  ter 7. 9  Introduction In critical state soil mechanics shearing drives samples of particulate soils towards a state of t volume at a constant mean effective stress, termed the critical state. At high pressures r, the particles tend to undergo breakage resulting in a c gradation. The breakage also imposes additional compressibility on the soil giving rise to ities around the definition of the critical state condition as it relates to zero volume . In practice, high pressures often occur in deep penetration problem and cone penetration testing, and at the base of large earth-fill dams. Hence the effect of particle e becomes important for rigorous modelling of such problems within the critical state ork.  9 A version of this chapter has been submitted for publication. Ghafghazi M., Shuttle D.A., and Olivera R.R. 2010. Particle Breakage and the Critical State of Sand.  230  A two or three part linear Critical State Locus (CSL) in e  log p' space (where e is the void tio and p′ is the mean effective stress), similar to the one shown in Figure 7-1, has generally been accepted for the full range of p. This agrees with the three zones of behaviour identified y Vesic and Clough (1968): very low pressures where dilatancy controls behaviour and reakage is negligible, higher pressures where breakage becomes more pronounced and y vanish and sand behaves like an elastic material. his second stress range (typically 1 MPa < p' < 30 MPa), which more or less covers the higher nd of stresses (all stresses are assumed to be effective in this chapter) of interest in geotechnical roblems, has been less studied in comparison to the lower stress range. To date there is no n  a unique CSL exists for this stress range, and how it is affected by the ontinuous gradation change due to particle breakage. This work focuses on the critical state at articles of exactly the same mineralogical combination. Hypothetically this is a viable proposition knowing that particle breakage drives the sample towards a more well-graded distribution which will establish more inter-particle contacts (Bishop, 1966) and thus reduce the ra b b suppresses dilatancy effects, and very high pressures where all effects of initial densit T e p conse sus on whether c the lower end of the second stress range (1 MPa to 3 MPa) where breakage becomes gradually dominant over dilatancy. It is well accepted that shearing at high pressure changes the soil’s gradation, which in turn contributes to volumetric compression (e.g. Lee and Farhoomand, 1967). The critical state is associated with a state of constant volume despite continued shearing. It is hence expected that for the sample to reach the critical state, a stable gradation should be reached for a specific stress level (Luzzani and Coop, 2002). This implies that such gradation would be more stable than the original one, and can sustain a higher level of stress without further breakage despite being formed by p  231  stress concentrations at particle level; however experimental evidence demonstrating the formation of such stable gradation related to a particular stress level is rather scarce in the literature. Been et al. (1991), Konrad (1998), and Russell and Khalili (2004) all adopted the three part CSL framework as illustrated in Figure 7-1 and assumed that a continual constant volume state will be achieved once the tests approach the second part of the CSL. The data presented by Russell and Khalili (2004) (Figure 9 of their paper) suggest that for higher stress levels (above 1 MPa) itical state is path independent and solely a function of the value of p the loose tests do not reach a constant volume and continue to contract. Lade and Yamamuro (1996) made the same observation on tests presented in Yamamuro and Lade (1996) and concluded that the critical state conditions can only be achieved at low pressures (the first part of CSL) or at extremely high pressures where particle breakage has ceased (the third part of CSL). An experimental study on a granitic soil by Lee and Coop (1995) suggested that the amount of particle breakage at the cr on the CSL. In a later attempt to investigate whether the critical state can be achieved at higher stress levels, Coop et al. (2004) used ring shear tests to take samples of a carbonate sand to up to 100,000% shear strain. They concluded that particle breakage continues to very large strains beyond those reached in triaxial tests, but a constant gradation is reached at very large strains. This constant gradation is dependent not only on the stress level but also on the uniformity and particle size of the original gradation. Theoretically, the change in gradation cannot continue indefinitely; it is always possible to imagine a gradation where all voids are filled with progressively smaller and smaller particles as proposed by McDowell et al. (1996). Such “fractal” gradation would be linear on a log-log plot  232  of particle size and proportion finer than that size. This condition of zero void space in the soil structure has been experimentally obtained by the early work of Bridgman (1918) for very high stresses falling on the third part of the CSL. But, for the range of stresses, strains and breakage  and fractal gradations. Muir Wood and Maeda (2008) showed, using a discrete element of interest to this work, and much of geotechnical engineering, it appears safe to assume that breakage does not completely stop at higher stresses. Thus the idea of a CSL defined as a state of zero “total” volume change becomes irrelevant for the higher stress ranges. Daouadji et al. (2001) first suggested that the changing gradation caused by the breakage imposes a downward shift on the CSL. Muir Wood (2007) developed this idea, suggesting that during shearing at higher stresses the CSL moves down towards a final location associated with the fractal gradation. He proposed a third dimension to the e  log p' space called the “grading state index”; a parameter between 0 and 1 which identifies soil’s state on a scale between uniform analysis, that the effect of particle breakage on the CSL location in e  log p' space is essentially a parallel downward shift as a function of the grading state index. The current work builds upon this idea by proposing a simple conceptual framework to explain the movement of the CSL due to breakage. Triaxial testing on samples of a uniformly-graded natural sand before and after breakage is then presented to illustrate the utility of the framework.  7.2. Hypothesis  Basic difficulties with the proposition of a three part CSL, can be most easily illustrated using the state parameter concept (Been and Jefferies, 1985); the state parameter becomes ill- conditioned for the second part of the CSL (Jefferies and Been, 2000). But more importantly, its  233  evolution towards the three part CSL becomes logically questionable for stress paths that undergo a reduction in p. For the test shown in Figure 7-1, the sample starts at an initial state parameter, 0, which is defined as the difference between the initial void ratio and the CSL void ratio at the same initial p. If that sample is now taken to the critical state, for example under undrained conditions, then the implication is that the sample has gone from a condition associated with a certain amount of breakage, to one with less, or no breakage. This is in i. The capability of the material for contraction merely by sliding and rolling of the es is exhausted. ii. A stress threshold is surpassed. contradiction to the fact that breakage is a “damage effect” which logically cannot be “reversed” by further shearing. An alternative hypothesis is proposed here which explains how the breakage affects the CSL in e  log p' space, when the breakage starts and how it contributes to the soil’s compressibility. The new hypothesis is expanded from two assumptions:  1. For small amounts of particle breakage, the finer particles generated by the breakage process do not contribute to the soil’s load carrying skeleton; and the breakage does not affect the overall characteristics of the particles forming the soil’s load carrying skeleton. 2. Significant particle breakage in a particulate material does not occur unless two conditions are concurrently satisfied: particl  The first statement is fairly intuitive provided only a relatively small number of soil particles break during shearing. It is known that breakage starts with the smallest load carrying particles  234  in the soil (Lee and Farhoomand, 1967; McDowell and Daniell, 2001). As long as the ratio of the particles that have undergone breakage remains small, the finer particles generated by the breakage do not contribute to the soil’s load carrying skeleton; instead, they fall into the void space. The idea that particle breakage does not commence until all sliding and rolling compressibility is suppressed is easy to hypothesise. Rolling is likely the first prevailing mechanism as it requires the least amount of energy to mobilise; once rolling is suppressed by an increase in confinement resulting from increased stress and reduced void space, sliding starts to dominate le for dense samples it is followed by a ndency for dilation. At this stage, if the stresses are large enough to break the particles, dev credibility of the assumptions. The most fundamental implication of the first assumption is its effect on th void ratio ec. N rgoes a finite amount of breakage, the particles chipped off bigger one ction in its void ratio (we limit the discussion to rained conditions considering the undrained condition a boundary constraint). This reduction in (Skinner, 1969). Breakage does not start while the particles have the opportunity to avoid loading by falling into the voids by merely rolling or sliding (compression). In loose samples the end of compression coincides with the critical state, whi te breakage starts. The finer particles produced in the process are now capable of sliding and rolling into the voids and thus further reducing the sample’s volume. Inferences can be drawn from the above hypothesised behaviour. These inferences are eloped here and then used to explain the observed experimental results to lend support to the e CSL in e  log p' space. A sample is expected to reach the critical state at a certain ow if the sample unde s during shearing will cause a redu d void ratio due to breakage is denoted as eb. For small amounts of breakage the load bearing  235  skeleton will still reach the critical state at the expected “skeleton void ratio”; but the sample’s overall void ratio will be lower by eb. The locus of the critical state void ratios (CSL) of the soil skeleton is a function of the current stress level as identified by the intercept  and the slope 10 (the CSL shape is an arbitrary   For the second assumption, let us consider the expected behaviour for different initial states. For a loose sample starting at high stresses the CSL remains fairly unchanged as the soil particles preferentially roll and slide as the applied pressure increases, until the sample approaches the critical state. As the critical state is approached, the capacity for contraction decreases and particle breakage starts to shift the CSL downwards, and the sample follows it by further reduction in volume. For a dense sample starting at higher stresses, the CSL remains unchanged until the initial choice and does not affect the argument). Conversely the eb generated is not related to the current stress level, instead being a function of the amount of breakage. Hence if the amount of breakage remains constant, the effect of breakage for the whole sample would be a parallel shift in the CSL (). Equation 7-1 immediately follows from this argument:  e [Eq. 7 - 1] b contraction phase (related to sliding and rolling) ends. As contraction changes towards dilation, breakage starts, moving the CSL down and increasing the total contraction. Hence the contraction may be prolonged more than that expected in the absence of breakage, and some breakage may be observed before the overall contraction is replaced by dilation.  236  Breakage continues throughout the dilation phase and eventually the volume reduction caused by breakage, and the volume increase due to dilation, balance (Chandler, 1985; Baharom and  on the state parameter nd how it changes with regard to a moving CSL. The state parameter is defined as  Stallebrass, 1998). This transient constant volume stage may be called the “apparent critical state”. Usually researchers have stopped shearing at this stage considering the sample to be at critical state (e.g. Russell and Khalili, 2004; Yamamuro and Lade, 1996). However, this may not necessarily be the case; breakage may continue beyond this point further reducing the sample’s volume. More insight into the breakage phenomenon can be gained by focusing a cee            [Eq. 7 - 2]  where ec is the critical void ratio for the current (with breakage) CSL and at the current mean effective stress. Considering purely shear effects by holding mean stress constant, differentiating both sides of Equation 7-2 results in  cee           [Eq. 7 - 3]  If 10  is constant, differentiating the CSL (  pec  log10 ) gives   ce           [Eq. 7 - 4]   237  Combining Equation 7-1 with Equations 7-3 and 7-4 we can write  bee           [Eq. 7 - 5]   Separating the change in void ratio into two parts  ee  bsr e          [Eq. 7 - 6] where  is the change in void ratio caused by sliding and rolling. From Equations 7-5 and 7-6 we obtain   sre sre           [Eq. 7 - 7]  Equation 7-7 suggests that the breakage, and the change in the void ratio it induces, do not affect how the state parameter evolves. This implies that (while the initiation, and probably rate of breakage depend on the dilatancy caused by sliding and rolling of the particles) the sample moves towards the critical state irrespective of the breakage. The change in void ratio caused by the breakage is  changes the gradation due to breakage, thus ence of breakage. As breakage is associated with dditional reduction in void volume (compressibility), reaching a stable critical state requires breakage  to liminate the breakage from a sample undergoing shearing at high stresses. For such a sample,  merely superimposed on that controlled by sliding and rolling. Shearing of a sample at high stress constantly changing the CSL as defined in the abs a to cease. So to understand how the CSL evolves with breakage it is necessary e  238  this would only be possible if the particles became instantaneously unbreakable at the moment f interest during shearing. Since this is physically impossible, a compromise was to reduce the stresses to a le ple hich has already undergone particle breakage. This is the rationale behind the testing program 7.3. Fraser River Sand The tested material, Fraser River Sand (FRS), is an alluvial deposit widely spread in the Fraser river delta in the Lower Mainland of British Columbia, Canada. The gradation used in this research contains around 0.8% fines content and has D50 and D10 of 0.271 mm and 0.161 mm respectively. FRS is a uniform, angular to sub-angular with low to medium sphericity, medium grained cl  and emax are Gs = 2.719 by Shozen (1991). The value of emin was measured max o vel where no significant breakage happens and determine the CSL for a sam w presented here.   ean sand. Figure 7-2 shows a microscopic picture of FRS grains. emin reported as 0.627 and 0.989 and according to ASTM D2049 while e was reported as the initial deposition void ratio in the loosest state. The average mineral composition based on a petrographic examination is 25% quartz, 19% feldspar, 35% metamorphic rocks, 16% granites and 5% miscellaneous detritus.  7.4. Testing Program  The test program included 28 drained and 11 undrained triaxial tests using lubricated end platens to reduce stress non-uniformity within the samples. All samples were 142 mm in height and 71 mm in diameter and were prepared using the moist tamping technique. The samples were  239  flushed with CO2 and de-aired water and back pressurised until a B value of 0.95 or greater was obtained. They were then consolidated and sheared until a steady state was reached, apparent shear localisation was observed, or the equipment limitations were met. The strain controlled shearing was applied at a constant rate of 5% per hour. Special attention was paid to accurate measurement of the void ratios considering its ignificance to the arguments made. At the end of the shearing phase the drainage valves were d. oved from the cell and the cell base, membrane and cap were dried efore putting the setting in a freezer for 24 hours. The frozen sample was then extracted with o investigate the effect of particle breakage on CSL, an additional 8 samples (7 drained,  at higher levels of stress and dry sieved at the end of each test. Two amples (corresponding to sieve tests #1 and #2) were sieved before shearing and found to have s close  The sample was rem b extreme care making sure that no water or grains were lost during the process. This technique (Sladen and Handford, 1987) effectively eliminated loss of water during sample extraction, enabling accurate determination of water contents. A repeatability of 0.01 or better was obtained for three pairs of tests that were targeted to start from identical conditions. Corrections were applied for membrane penetration (Vaid and Negussey, 1984) and membrane force (Kuerbis and Vaid, 1990). The initial stage of the testing program was aimed at measuring the critical state locus of FRS at low stress levels expected to be uninfluenced by particle breakage. Table 7-1 summarises these 16 lower stress tests. T 1 undrained) were sheared s fines contents of 0.7% and 0.9%. Table 7-2 provides details of these 8 tests referred to as the “parent tests”, identified in the table by bold font. Information provided includes initial and final  240  stress level and void ratio, status of the sample with regards to the critical state at the end of the test, and the percentage fines following shearing. Six samples showed an increase in their fines content following shearing. They were used to prepare samples for testing at lower stress levels to investigate the change in CSL due to rained) have reached a steady constant value. 0.138 (e = 0.06) in the typical semi-log space idealisation. Below  200 kPa, CIU-L 390 kPa, CID-L 100 kPa, CID-M 400 kPa and CID-L 190 kPa UR. Another breakage. Table 7-2 also provides details of the subsequent 15 tests performed on pre-sheared samples at lower stress levels.  7.5. Results  Figure A-10a shows the deviator stress and volumetric strain plotted against axial strain for the drained tests presented in Table 7-1. Figure A-10b shows the deviator stress and the pore pressure plotted against the axial strain for the undrained tests presented in Table 7-1. A sample is considered at the critical state if the stress ratio and the volumetric strain (drained) or pore pressure (und To obtain the critical state parameters of FRS all the tests presented in Table 7-1 are plotted in Figure 7-3 as e log p' state paths. Between 100 kPa and 900 kPa the CSL is approximated with = 1.22 and 10 = p = 100 kPa the semi-log CSL idealisation does not match the data well and the CSL is plotted with a dashed line to illustrate uncertainty. As discussed in the previous chapter a bilinear semi- logarithmic representation of the CSL is also observed for other gradations of FRS and is not considered to be associated with a real change of behaviour. The dataset provides five tests that ended on the proposed CSL within a measured void ratio difference of 0.01 or less: CIU-M  241  three tests (CID-L 300 kPa, CIU-D 200 kPa and CID-D 200 kPa UR) delineate the same CSL with an error margin of ±0.05 or better. The location of the CSL is further assured by the fact  important to the current work which focuses on a higher stress nge. Above a particle ‘breakage threshold’ stress level particle breakage may occur. The tion of this threshold is shown on Figure 7-3, and is related to soil mineralogy, ngularity, etc. Yamamuro et al. (1996) also showed an increase in particle breakage with -11a to A-11c. Two of the parent  that every single test below the proposed CSL was dilating and every single test above it was contracting at the end of their shearing phase. Any lowering of the CSL would undermine this basic principle of the critical state theory. The CSL below 100 kPa is not ra approximate loca a reducing void ratio, suggesting the approximate threshold indicated on Figure 7-3 is in reality non-vertical. The exact location and nature of this breakage threshold is not important to the arguments presented, so the approximate threshold shown on Figure 7-3 is considered sufficient. The deviator stress, pore pressure and volumetric strain have been plotted against axial strain for samples subjected to higher levels of stress in Figures A samples (#1 and #2) were sieved before testing. At the end of each parent test the sample was sieved and then retested in the triaxial apparatus at lower levels of stress as summarised in Table 7-2 and shown in Figures A-11d to A-11f. Figure 7-4 illustrates the gradation curves for the sieve samples #1 and #2 before shearing and samples #1, #2 and #3 after. The remaining sieve tests were not plotted on this figure for clarity; their fines contents (FC) are reported in Table 7-2. Additional sieve tests were performed after shearing the pre-sheared samples whenever breakage was suspected to have occurred (tests starting at p' = 600 kPa or higher).  242  Post shearing fines content in the 8 samples ranged between 0.8% and 7.8%. Although more detailed indexes for quantifying particle breakage are available (e.g. Hardin, 1985; Miura and O-Hara, 1979), for simplicity the fines content has been used here to quantify particle breakage. The tests performed on pre-sheared samples have been plotted in e  log p' space along with their “parent” test (shown in bold) in Figures 7-5 to 7-12 in order to identify the effect of ple was then mixed with about 5% of its weight virgin sand (giving 6.3 % fines content) to produce enough material for the following tests. The mix was then tested loose at 200 kPa and 600 kPa mean particle breakage on the CSL. In Figure 7-5 the samples with an initial fines content of 4.1%, obtained by previously shearing sample CIU-D 1300 kPa to p = 2600 kPa and q = 3900 kPa, have been tested under drained and undrained conditions. The two undrained tests starting at p = 200 kPa reached the critical state in the curved, low stress, zone of the CSL (see Figure 7-3) so these tests were discarded in determining the CSL. Hence the inferred location of CSL is biased towards Test “CIU-M #1 400 kPa”. The parent test CID-L 1600 kPa (Figure 7-6) on a virgin sample, was planned to be sheared drained near the equipment’s pressure limit. Shortly after starting the test, the cell pump controlling the cell pressure failed and the cell pressure started decreasing. The test was continued until the sample failed. The sieve test performed post-shearing suggested no particle breakage had occurred and so no further tests were done on this sample. The parent test shown on Figure 7-7 was performed on a loose drained sample consolidated to 1400 kPa and sheared to 43% axial strain. As expected for such large axial strains, towards the end of the test excessive bulging and shear localisation was observed, along with a drop in the measured deviator stress. At the end of the test the sample had 7.0% fines. This sam  243  effective stresses respectively. The CSL was inferred based on the test at 200 kPa. Test CID-M #3 600 kPa was sieved and an increase in the fines content to 7.8% observed. To determine the change in CSL due to this further breakage, another test was conducted at   s sheared to a smaller strain (10%) and post test sieving showed nt after shearing was 3.2%.  phase. In Figure 7-12 150 kPa suggesting a further drop in CSL. The parent test in Figure 7-8 is similar to that in Figure 7-7; the difference being that the #5 sample shown in Figure 7-8 wa a lower fines content of 1.9%. Two tests at 100 kPa and 300 kPa were performed on the sheared sample, which clearly showed a parallel shift in the CSL. The sample was then used in a test starting at 600 kPa. In this test, crossing the CSL, particle breakage was detected as indicated by the increase in fines content to 4.3%. Test CID-D 600 kPa (Figure 7-9) was performed on a dense sample starting at p=600 kPa. The fines conte A dense sample at p=600 kPa was sheared drained up to its peak strength as shown in Figure 7-10. Sieving detected no particle breakage. To confirm the absence of any effects on the CSL, one test at 100 kPa was performed on the sample which confirmed the original CSL. Figures 7-11 and 7-12 plot tests from two nearly identical dense samples consolidated to just above 1000 kPa. The test in Figure 7-11 was stopped before the dilation the test was taken to larger strains until dilation ended and the sample started to contract again (see Figure A-11b). The test in Figure 7-11, which was stopped early, showed very little increase in its fines content. Two subsequent tests conducted on this sample indicated only a small CSL shift. Conversely, the more highly sheared sample (Figure 7-12) showed a significant fines increase, and a larger shift in the CSL.  244  The results presented in Figures 7-8, 7-11 and 7-12 suggest that for both loose and dense samples the effect of particle breakage on the location of CSL in pe  log plot is a downward shift without a change in the slope of the line, consistent with the proposed varying  and constant 10. Postulating a parallel shift in CSL, it is possible to approximately locate the CSL  by sliding and rolling of the particles. The observation with only one test at lower stresses using pre-sheared samples; the condition being that this test should have reached the critical state at the end of shearing. This approach was used to identify the CSL for the data presented in Figures 7-5 and 7-7.  7.6. Discussion  The critical state has been largely defined and tested in the absence of significant breakage of particles, when behaviour is controlled made at higher levels of stress has been that samples reach the critical state at lower void ratios than those expected by extrapolating the - usually but not necessarily - linear CSL in pe  log space. This additional compressibility is understood to be caused by the breakage of soil s the CSL down (reducing ) without changing its slope (constant 10). This is in agreement with Muir Wood and Maeda (2008) who found that particles resulting in generation of finer particles that fill the voids. The approach taken in this work is to distinguish between this additional compressibility caused by the breakage of soil particles and that caused by sliding and rolling (strictly speaking of plastic compressibility).  7.6.1. Influence of Breakage on CSL  These data suggest that particle breakage move  245  the change in 10 is negligible in comparison to the change in . Looking at the tests on drained loose samples in Figures 7-7 and 7-8 it can be observed that a CSL extrapolation to higher stresses passes through the end point of the “parent” test. The end point of “CID-L 1400 kPa” is lower than that obtained from the extrapolation of the CSL drawn through the end point of “CID-M #3 200 kPa”; this test had 5% virgin soil added to the sample that has caused the CSL to move up relative to the CSL of CID-L 1400 kPa. The two consecutive tests again suggest that the test end point undergoing particle breakage (“CID-M #3 600 kPa”) lies on an extrapolation f the CSL obtained from the test at the lower stress (“CID-L #3 150 kPa”). en loose samples undergoing particle breakage are coincident with the CSL btained from subsequent tests at lower stresses. Hence it can be inferred that samples .6.2. The Onset of Breakage et us now investigate when the breakage starts by considering test CID-L 1600 kPa (Figure o The d points of o undergoing particle breakage at higher stresses are at their CSL from a sliding and rolling point of view. In other words, the reduction in volume is entirely caused by the breakage phenomenon. So a sample being sheared at higher stresses is actually on its CSL defined as if the breakage stopped right away, but the line is continuously moving down parallel to the original CSL because of continuous breakage. This is the idea expressed by Equation 7-1 and shown in Figure 7-8. The data presented suggests that Equation 7-1 holds true some time after the sample has passed the original CSL.  7  L 7-6) which was isotropically consolidated to a high confining pressure. The sample was sheared to 0.1% strain before failure of the cell pump occurred; shearing then continued until the sample  246  is is a rather strong  e (fines content increased to 3.2%). The tests starting t 1000 kPa showed the same trend, suggesting that virtually all particle breakage occurs after is exhausted. This is consistent with Hyodo et al. (1999) ndings for undrained tests on sands, that particle breakage accelerates after the phase hint supporting the second assumption on which the conceptual model is based. The drained tests on dense samples shown in Figures 7-9 to 7-12 start from 600 kPa and 1000 kPa and were taken to different levels of strain in order to investigate the initiation of breakage. Test “CID-D 600 kPa-peak” was stopped around its peak strength when only some slight dilation had occurred after the sample had undergone the initial contraction (see Figure A-11b) expected from a dense sample. The consequent gradation test showed no increase in the fines content and the test performed at 100 kPa confirmed no change in the CSL. In contrast, test “CID-D 600 kPa”, which was sheared until the dilation was completely suppressed, showed a considerable amount of particle breakag a sliding and rolling compressibility fi transformation point. This observation can also be explained by assumption 2; that breakage does not occur during the initial contraction phase in shearing of dense samples.  247  The proposed mechanism can be used to explain the behaviour seen in test “CID-L 1600 kPa” (and similarly the undrained test shown in Figure 7-1). Although starting at the highest mean effective stress of all tests, the sample showed no breakage in the sieve test performed  dilating, some afterwards because the sample contracted by sliding and rolling during the test and reached the CSL at a stress level lower than the ‘particle breakage threshold’.  7.6.3. The Influence of Breakage on Behaviour  Although test “CID-D 1000 kPa-peak” was stopped before the sample started minor breakage was recorded; fines content increased from 0.8% to 1.0% and a small drop in the CSL was registered (Figure 7-11). It is possible to explain this using Equation 7-6: towards the end of the contraction phase and at the beginning of the dilation phase, esr becomes very small while breakage starts to kick in ( 0 be ). The resulting measured total e remains negative. Depending on the stress level, density, and the material, this effect can completely suppress the dilation phase and cause the sample to behave like a loose sample (Vesic and Clough, 1968) despite starting under the CSL. A similar mechanism may occur during dilation when the sample is approaching the standard ‘sliding/rolling’ critical state. At this stage of the test the breakage may balance dilation, creating a state of zero volume change or “apparent critical state”. The proposed hypothesis suggests that provided that breakage can continue with further shearing, compression will resume. This was observed in “CID-D 1000 kPa” on Figure A-11b as a resumption of contraction beyond 20% strain. For dense tests like “CID-D 1000 kPa”, sheared to the “apparent  248  critical state” and beyond, the CSL is expected to follow the sample as it moves towards lower void ratios. Although the tests presented in Figure 7-12 show that the CSL of the pre-sheared samples is significantly lower than the original CSL, the CSL of the pre-sheared samples is still well above the end point of the parent test. This could be due to localisation occurring post peak in dense amples, affecting both the volume changes and amount of breakage measured in the parent test. red void ratio at the end of the test than  would have if the entire sample had dilated to the critical state. The same mechanism applies is reached ( s With localisation the whole sample has a smaller measu it to particle breakage (Luzzani and Coop, 2002) resulting in a sample which is not only non- uniform in straining but also in gradation and amount of breakage. The pre-sheared sample is thus a mixture of materials with different degrees of breakage resulting in a higher CSL. This makes correlating the CSL with the sample’s state at high stresses more difficult for dense samples. Not considering localisation, it is expected for Equation 7-1 to similarly apply to dense samples once shearing has been continued far enough to initiate breakage. However, since in dense samples dilation occurs before the critical state 0 sre ) it is not possible to directly 1600 kPa” with no breakage). The results of the current work do not suggest that no breakage measure eb to verify the applicability of the equation. The testing program presented here was not aimed at determining the exact nature of the stress level required to cause particle breakage (assumption 2.ii). But it appears that a breakage threshold around p′ = 900 kPa marks the onset of measurable breakage for Fraser river sand. It is well accepted that more particle breakage occurs during shearing than in isotropic compression (e.g. Bishop, 1966; Ueng and Chen, 2000), as also is observed here (test “CID-L  249  occurs in compression; likely just higher stress levels are required. Nevertheless, it is probable that the basics of the conceptual model are applicable to isotopic compression. ore support to the ideas presented.  chapters, it is interesting that although ted Although determining the CSL for samples undergoing breakage requires additional tests to be performed on sheared samples, the correlation between the shift in the CSL and the fines content produced by breakage is rather promising.  is plotted against the fines content in Figure 7-13 showing a direct relation between the two. And once this relation is established the CSL can be estimated for other samples by performing a sieve test. The variation in , (), is plotted against the change in fines content for Fraser river sand as well as Kurnell sand (Russell and Khalili, 2002) in Figure 7-14. The kurnell sand data is plotted by applying the conceptual framework presented here to the triaxial compression and gradation test data presented by Russell and Khalili (2002). Despite the differences between the materials and the testing performed, the data presented in Figure 7-14 follow strikingly similar trends lending m  7.6.4. The Influence of Breakage on CPT  In the context of the CPT analyses presented in earlier particle breakage has been known (e.g. De Beer, 1963; Yang et al., 2010) to occur at the high stresses often induced by CPT, only a few of the preceding analyses have explicitly modelled the effects of particle breakage on CPT interpretation results. The hypothesis and data presen here allows additional insight into the effects of breakage on CPT interpretation. Russell and Khalili (2002) considered breakage by incorporating a steepening CSL with increasing mean stress into a cavity expansion analysis using a critical state based model. The  250  fundamental issues with this approach were discussed earlier, but this approach was nevertheless able to partially address the effect of the additional compressibility caused by  its consequences on specific aspects of lili (2002), a declining will likely result in a duction in the limiting cavity expansion pressure. However, breakage as an energy dissipation ism gth of the materials undergoing breakage. The ffect will likely be analogous to an increase in Mtc . The increase is likely small as the energy breakage on CPT analysis results. As expected, the result was a reduction in the limiting cavity expansion pressure, inferring an increase in the shape function required to match the chamber penetration resistance. Although the framework adopted in this work for interpreting CPT does not account for the effects of breakage, it provides a basis for investigating behaviour through the critical state parameters , 10, and Mtc. The hypothesis and the data suggest a constant 10 and a declining  due to breakage in zones around the cone where the stresses exceed the breakage threshold. The confined conditions around the cone and the large strain nature of deep penetration likely guarantee that the other requirement for the triggering of breakage (i.e. exhaustion of the capacity for sliding and rolling) be satisfied around the cone. Similar to the findings of Russell and Kha re mechan  can also add a component to the stren e dissipated in breakage is suggested to be around 10% of the energy dissipated in friction (McDowell et al., 2002). The added compressibility (declining ) will have a far more pronounced effect on cone penetration as a deformation controlled process. However, as mentioned in chapter 6, an increase in Mtc can still have significant implications for the interpretation of the state parameter from CPT. One difficulty in incorporating the breakage phenomenon in the analysis of CPT is the inherent mismatch between the stress levels achieved in analysis and in reality, as reflected by the shape  251  function. Since breakage is governed by the stress level and in particular a stress threshold, if the numerical model is working at a lower stress range, it may not detect, or underestimate the amount of breakage and its consequences. In the CPT interpretation framework adopted and developed in this work, the effects of particle breakage are implied in the shape function. By explicitly incorporating particle breakage into the L in framework one can expect further improvement in the accuracy in estimating the state parameter from the tip resistance. However, the current accuracy of the framework appears to be more constrained by the repeatability of the experimental data against which it is validated than by its simplifying assumptions, including its treatment of particle breakage.  7.7. Summary and Conclusions  The current work proposed a simple hypothesis to understand Critical State Locus movement due to shearing a particulate soil at stresses high enough to produce particle breakage. The conceptual model was tested using a series of triaxial compression tests on Fraser river sand, reaching p levels of up to 3 MPa. The associated increase in the fines content of sheared samples ranged from zero to nearly 7%. The main observations from the hypothesis and the associated testing performed are: Measurable breakage only starts after the soil’s contraction capacity is exhausted. As postulated by Daouadji et al. (2001) and Muir Wood and Maeda (2008), breakage causes a downward parallel shift in the location of the CS  pe  log space. The magnitude of this CSL shift is equal to the void ratio reduction due to breakage (  be ) and is directly correlated with the increase in fines content.  252  These observations contradict the widely adopted three part line used to model the effect of particle breakage on the CSL and it is recommended that future work idealises the CSL as a series of parallel loci each associated with a certain level of particle breakage. The experimental observations also imply that since  be , the state parameter is independent of breakage. The volume reduction caused by the breakage may simply be superimposed on the volume change controlled by stress-dilatancy. Hence it may be possible to uncouple the critical state framework and the breakage phenomenon. The experimental data provided in this chapter are encouraging in the support they provide for the proposed hypothesis. More testing is needed to further investigate the generality of the ideas resented here and other aspects of breakage. p  253   254   Initial conditions End of Test Remarks  Table 7 - 1 Summary of testing program at lower stress level Test name* p' e 0** Status*** 1 p' q e  (kPa) (%) (kPa) (kPa) CIU-L 100 kPa 97.7 1.053 0.108 Con. 1.6 11.0 16.7 1.053 CIU-L 200 kPa 201.9 0.963 0.061 CS 14.0 9.4 15.5 0.963 CIU-M 200 kPa 200.2 0.897 -0.005 Dil. 10.2 184. 5 0.897   6 273. CIU-D 200 kPa 196.4 0.820 -0.083 Dil. 5.1 334. 7 0.820   6 504. CIU-L 300 kPa 301.0 0.966 0.088 CS 16.0 24.4 0.966   19.0 CIU-D 390 kPa 388.4 0.906 0.044 Dil. 24.3 186.0 266.4 0.906 CIU-M 400 kPa 393.3 0.832 -0.030 Dil. 19.6 613.6 896.5 0.832 CID-D 50 kPa 50.3 0.753 -0.232 Dil. 25.1 92.1 126.7 0.878 CID-L 100 kPa 102.1 0.948 0.005 CS 34.8 193.7 274.2 0.911 void ratio calculated from reconstitution density due to loss of water during retrieval CID-D 115 kPa 113.9 0.668 -0.268 Dil. 21.7 226.0 336.9 0.794 CID-L 300 kPa 302.9 1.005 0.128 Con. 29.1 585.9 847.1 0.868 CID-D 410 kPa 409.6 0.634 -0.225 Dil. 24.3 811.1 1196.6 0.728 CID-D 515 kPa 514.5 0.689 -0.156 CS 28.4 984.3 1405.1 0.742 CID-L 600 kPa 603.3 0.857 0.021 Con. 31.7 1134.9 1594.0 0.772 Some particle breaka expected ge CID-L 190 kPa UR 190.0 0.902 -0.004 CS 23.6 356.0 496.1 0.865 CID-D 200 kPa UR 198.0 0.730 -0.173 Dil. 26.2 399.2 601.3 0.804 Un/Re-load cycles applied * CIU, consolidated undrained test; CID, consolidated drained test; L, loose; M, medium dense; and D dense sample. ** The state parameter is obtained with respect to the original CSL defined in Figure 7-3. *** Con., contracting; CS, no change in volume or pore pressure i.e. critical state; and Dil., dilating.    Table 7 - 2 Summary of testing program at higher stress level an m   Initial conditions End of d tests  Tes  p t erformed on pr  e-sheared   sa  ples Test name P′ (kPa) e  Status 1 (%) p′ (kP P q a) e F (% C ) Si T eve est Remarks a) (k CIU-D 1300 kPa* 1288.1 0.620 -** CS 16.8 2584 9 e.4 38 4.3 0.620 4.1 #1 Pr -testing fines content is 0.7 % CID-D #1 100 kPa 102.2 0.698 -0.120 Dil. 1.9 209. 78 31 .1 0.774 CIU-L #1 200 kPa 200.9 0.915 0.138 CS 18.7 5.7 8.      1 0.915 CIU-M #1 200 kPa 203.9 0.868 0.092 Con. 9.6 35. 1 0.868    5 5 .1 CIU-M #1 400 kPa 394.0 0.820 0.084 CS 24.9 149. 34 20 .9 0.820 CID-L 1600 kPa 1601.4 0.867 - Con. 8.3 545. 4 0.843 0.8 #2 l an arbitrary stress path e Ce l pump failed, but test continued down Pr -testing fines content is 0.9% 5 74 .3 CID-D 1400 kPa 1396.2 0.839 - Con. 10.7 2718 5.7 39 0.5 0.600 7.0 #3 CID-M #3 200 kPa 204.6 0.824 0.044 CS 43.0 392. 0 About 5% of the sample weight virgin cing2 56 .9 0.746   FR FC S w  to 6 as a .3% dd  ed to the sample, redu CID-M #3 600 kPa 614.9 0.780 0.065 CS 25.9 1193 6 0.657 7.8 #3a  .0 17 0.6 CID-L #3 150 kPa 155.9 0.831 0.059 CS 28.4 274. 3 0.739    4 36 .5 CID-D 1400 kPa-2 1403.5 0.849 - Con. 39.4 2325 4 s.5 27 4.7 0.736 1.9 #5 Te t stopped early CID-L #5 100 kPa 103.6 0.999 0.087 CS 9.8 186. 9 88   5 24 .3 0. 3 CID-L #5 300 kPa 303.617 0.9381 0.091 CS 35.0 563. 0 0.809    5 78 .4 CID-L #5 600 kPa 604.103 0.8904 0.085 CS 32.3 1188 6 o in th0.729 4.3 #5a Prtest blems with cell pressure early e .4 17 2.4 CID-D 600 kPa 603.8 0.752 - CS 31.8 1134 9 0.74 3.2 #4  .3 15 4.8 8 CID-D 600 kPa-peak 578.3 0.701 - Dil. 33.7 1321 5 s.0 21 0.2 0.694 0.8 #6 Te t stopped early CID-L #6 100 kPa 104.4 1.004 0.063 CS 5.8 189. 57 25 .3 0.902 CID-D 1000 kPa-peak 1003.4 0.666 - Con. 31.3 2019 2 0.643 1.0 #7 Test stopped early .6 30 7.3 CID-L #7 100 kPa 105.2 1.005 0.074 CS 23.1 187. 62 24 .0 0.898 CID-L #7 300 kPa 302.7 0.954 0.086 CS 38.8 570. 26 80 .4 0.823 CID-D 1000 kPa 1014.6 0.663 - Con. 36.3 1880 3 0.65 #8  .8 26 3.7 7 3.1 CID-L #8 100 kPa 104.1 0.982 0.095 CS 6.0 187. 06 25 .2 0.854 CID-L #8 300 kPa 304.1 0.924 0.102 CS 34.4 559. 7655 .6 0.788 * Tests shown in bold provide the “parent” samples for the following tests after b  siev or gradation. ** The state parameter not stated for “parent” samples as CSL changes during test  to p a eing  due ed f article bre kage.  255  0.0 0.2 0.4 0. 0.8 1.0 10 0 ,000 at io , e e f 6 10 1,000 10,000 100 vo id  r m an e fective stress, p' (kPa) Behavi o led by dilatanc in  rolling) our c y (slid ntrol g & Particle o domin ta  brea ant o kag ver e bec  dila ming ncy 0 All v stic- oids Ela like b ur  fill eh ed, avio  Undrained Shearing   7 stre a  C  i  ) and at n x te Fig sch   ure em  - 1 Full ss r nge SL n e  log p' space (after Russell and Khalili, 2004 ic undrai ed tria ial st  Figure  u f S 7 - 2 Microscopic pict re o  FR  grains  256  0.60 0.70 0.80 0.90 1.00 1.10 1 10 100 1000 10000 CSL curvature A pp ro xi m at e B re ak ag e th re sh ol d mean effective stress, p' (kPa) vo id  r at io , e  Start of Test End of Test, No Volume Change End of Test, Contracting End of Test, Dilating Undrained Tests Drained Tests CSL  igure 7 - 3 Void ratio versus p  for virgin FRS tests summarised in Table 7-1 (for figure clarity nly start and end points are plotted for cyclic tests)  F  o   257  0 20 40 60 80 100 0.010.1110 Diameter (mm ) Pe rc en t p as si ng Sample #1, before shearing Sample #2, before shearing Sample #1, after shearing Sample #2, after shearing Sample #3, after shearing  Figure 7 - 4 Gradation curve for samples of FRS before and after shearing 0.55 0.65 0.75 0.85 0.95 1.05 1 10 100 1,000 10,000 vo id  ra tio , e mean effective stress, p' (kPa) B re ak ag e th re sh ol d Original CSL CSL CIU-D 1300 kPa CIU-M #1 400 kPa CIU-M #1 200 kPa CIU-L #1 200 kPa CID-D #1 100 kPa Start of Test End of Test on CSL End of Test NOT on CSL  Figure 7 - 5 e  log p' plot for all tests on sieve sample #1 in Table 7-2 (obtained from parent sample CIU-D 1300 kPa)  258  0.55 0.65 0.75 0.85 0.95 1.05 100 1,000 10,000 vo id  ra tio , e mean effective stress, p' (kPa) Original CSL CID-L 1600 kPa Start of Test End of Test B re ak ag e th re sh ol d  Figure 7 - 6 e  log p' plot for test CID-L 1600 kPa (sieve sample #2) 0.55 0.65 0.75 0.85 0.95 1.05 Original CSL CSL for sample #3 CSL for sample #3a CID-L 1400 kPa CID-M #3 200 kPa CID-M #3 600 kPa CID-L #3 150 kPa Start of Test End of Test on CSL End of Test NOT on CSL 100 1,000 10,000 vo id  ra tio , e mean effective stress, p' (kPa) Br ea ka ge  th re sh ol d  Figure 7 - 7 e  log p' plot for all tests on sieve sample #3 in Table 7-2 (obtained from CID-L 1400 kPa)  259  0.55 0.65 0.75 0.85 0.95 1.05 100 1,000 10,000 vo id  ra tio , e mean effective stress, p' (kPa) Original CSL CSL CSL for sample #5a CID-L 1400 kPa-2 CID-L #5 100 kPa CID-L #5 300 kPa CID-L #5 600 kPa Start of Test End of Test on CSL End of Test NOT on CSL B re ak ag e th re sh ol d eb=  Figure 7 - 8 e  log p' plot for all tests on sieve sample #5 in Table 7-2 (obtained from CID-L 1400 kPa-2) 0.55 0.65 0.75 0.85 0.95 1.05 Original CSL CID-D 600 kPa Start of Test End of Test on CSL 100 1,000 10,000 mean effective stress, p' (kPa) vo id  ra tio , e B re ak ag e th re sh ol d  Figure 7 - 9 e  log p' plot for test CID-D 600 kPa (sieve sample #4)  260  0.55 0.65 0.75 0.85 0.95 1.05 100 1,000 10,000 vo id  ra tio , e mean effective stress, p' (kPa) Original CSL CID-D 600 kPa-peak CID-L #6 100 kPa Start of Test End of Test on CSL End of Test NOT on CSL B re ak ag e th re sh ol d  Figure 7 - 10 e  log p' plot for all tests on sieve sample #6 in Table 7-2 (obtained from CID- 600 kPa-peak) D  0.55 0.65 0.75 0.85 0.95 1.05 100 1,000 10,000 vo id  ra tio , e mean effective stress, p' (kPa) Original CSL CSL CID-D 1000 kPa-peak CID-L #7 100 kPa CID-L #7 300 kPa Start of Test End of Test on CSL End of Test NOT on CSL Br ea ka ge  th re sh ol d  Figure 7 - 11 e  log p' plot for all tests on sieve sample #7 in Table 1000 kPa-peak) 7-2 (obtained from CID-D  261  0.55 0.65 0.75 0.85 0.95 1.05 100 1,000 10,000 vo id  ra tio , e mean effective stress, p' (kPa) Original CSL CSL CID-D 1000 kPa CID-L #8 100 kPa CID-L #8 300 kPa Start of Test End of Test on CSL Br ea ka ge  th re sh ol d   Figure 7 - 12 e  log p' plot for all tests on sieve sample #8 in Table 7-2 (obtained from CID-D 1000 kPa)  1.06 1.08 1.10 1.12 1.14 1.16 1.18 1.20 1.22 1.24 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 FC (%)  Drained Tests Undrained Tests Virgin, #2, #6 #7 #5 #1 #3 #8 #3a #5a  Figure 7 - 13 Variation in  vs. fines content after shearing of FRS under drained and undrained conditions (Sieve test numbers are shown beside data points)  262   263 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 18.0 FC (%)   e b Fraser River Sand Kurnell Sand  Figure 7 - 14  vs. the change in fines content after shearing at high pressures: comparison between FRS and Kurnell sand (Russell and Khalili, 2004)       7.8. References  Baharom B. and, Stallebrass S.E. 1998. A Constitutive Model Combining the Microscopic and Macroscopic Behaviour of Sands in Shear and Volumetric Deformation. Proceedings of the 4th European Conference on Numerical Methods in Geotechnical Engineering, Udine, 1998. Springer-Verlag Wien, New York, 263–273. Been K., Jefferies M.G., and Hachey J. 1991. The Critical State of Sands. Géotechnique, 41(3): 365–381. Been K., and Jefferies M.G. 1985. A State Parameter for Sands. Géotechnique, 35(2): 99-112. de Beer E.E. 1963. The Scale Effect in the Transposition of the Results of Deep-Sounding Tests on the Ultimate Bearing Capacity of Piles and Caisson Foundations. Géotechnique, ishop A.W. 1966. Strength of Soils as Engineering Materials. 6th Rankine Lecture. Géotechnique, 16: 89-130. Bridgman P.W. 1918. The Failure of Cavities in Crystals and Rocks under Pressure. American Journal of Science, 45: 243-268.  Drucker’s Postulate, Suitable for Granular Materials. Journal of the Mechanics and Physics of Solids, 33: 215–226. Coop M.R., Sorensen K.K., Bodas Freitas T., and Georgoutsos G. 2004. Particle Breakage During Shearing of a Carbonate Sand. Géotechnique, 54(3): 157-163. Daouadji A., Hicher P. Y., and Rahma A. 2001. An Elastoplastic Model for Granular Materials Taking into Account Grain Breakage. European Journal of Mechanics - A/Solids, 20: 113–137. 13(1): 39-75. B Chandler H.W. 1985. A Plasticity Theory without  264  Hard  B.O. 1985. in Crushing of Soil Particles. ASCE, Journal of Geotechnical Engineering, 111(10): 1177-1192. Jefferie tate Theory from Isotropic Konrad J.M. 1998. Sand State from Cone Penetrometer Tests: A Framework Considering Grain Lade P amuro J.A. 1996. Undrained Sand Behaviour in Axisymmetric Tests at Lee I. 5. Intrinsic Behaviour of a Decomposed Granite Soil. Lee K Compressibility and Crushing of Granular Soil in Luzzan akage and the McDow hing of Granular McDow Hyodo M., Aramaki N., Nakata Y., and Inoue S. 1999. Particle Crushing and Undrained Shear Behaviour of Sand. Proceedings of the 9th International Offshore and Polar Engineering Conference, Brest, France, May 30-June 4, 1999. ISBN 1-880653-39-7. s M.G., and Been K. 2000. Implications for Critical S Compression of Sand. Géotechnique, 50(4): 419-429. Crushing Stress. Géotechnique, 48(2):201–215. Kuerbis R.H., and Vaid Y.P. 1990. Corrections for Membrane Strength in the Triaxial Test. Geotechnical Testing Journal, 13(4): 361-369. .V., and Yam High Pressures. Journal of Geotechnical Engineering, 122(2): 120-129. K., and Coop M.R. 199 Géotechnique, 45(1): 117-130. .L., and Farhoomand I. 1967. Anisotropic Triaxial Compression. Canadian Geotechnical Journal, 4(1): 68-86. i L., and Coop M.R. 2002. On the Relationship between Particle Bre Critical State of Sands. Soils and Foundations, 42(2): 71-82. ell G.R., Bolton M.D., and Robertson D. 1996. The Fractal Crus Materials. Journal of the Mechanics and Physics of Solids, 44(12): 2079-2101. ell G.R., and Daniell C.M. 2001. Fractal Compression of Soil. Géotechnique, 51(2): 173-176.  265  McDowell G.R., Nakata Y., and Hyodo M. 2002. On the plastic hardening of sand. Géotechnique, 52(5): 349-358. Muir W ted in Oslo, 25 Muir W l: Effect on Critical States. Acta Russell xpansion in Sands Exhibiting Particle nd. MASc. thesis, Department of Civil Engineering, Skinne ength of Sladen  laboratory testing of very Ueng T Shear of Miura N., and O-Hara S. 1979. Particle-Crushing of a Decomposed Granite Soil under Shear Stresses. Soils and Foundations, 19(3): 1-14. ood, D. 2007. The Magic of Sands – The 20th Bjerrum lecture presen November 2005. Canadian Geotechnical Journal, 44: 1329-1350. ood D., Maeda K. 2008. Changing Grading of Soi Geotechnica, 3: 3-14.  A.R., and Khalili N. 2002. Drained Cavity E Crushing.  International Journal of Numerical and Analytical Methods in Geomechanics, 26(4): 323–340. Russell A.R., and Khalili N. 2004. A Bounding Surface Plasticity Model for Sands Exhibiting Particle Crushing. Canadian Geotechnical Journal, 41(6): 1179-1192. Shozen T. 1991. Deformation under the Constant Stress State and its Effect on Stress-Strain Behaviour of Fraser River Sa University of British Columbia. r A.E. 1969. A Note on the Influence of Interparticle Friction on the Shearing Str a Random Assembly of Spherical Particles. Géotechnique, 19: 150-157.  J.A., and Handford G. 1987. A potential systematic error in loose sands. Canadian Geotechnical Journal, 24: 462-466. .-S., and Chen T.-J. 2000. Energy Aspects of Particle Breakage in Drained Sands. Géotechnique, 50(1):65-72.  266  Vaid Y.P., and Negussey D. 1984. A Critical Assessment of Membrane Penetration in the Triaxial Test. Geotechnical Testing Journal, 7(2): 70-76. s Division, ASCE, 94 (SM3): 661-688. ): 109-119. f Geotechnical Engineering, ASCE, 122(2): 147-154. Vesic A.S., and Clough G.W. 1968. Behavior of Granular Materials under High Stresses. Journal of the Soil Mechanics and Foundation Yamamuro J.A., and Lade P.V. 1996. Undrained Sand Behaviour in Axisymmetric Tests at High Pressures. Journal of Geotechnical Engineering, ASCE, 122(2 Yamamuro, J.A., Bopp, P.A. and Lade, P.V. 1996. One dimensional compression of sands at high pressures. Journal o Yang Z.X., Jardine R.J., Zhu B.T., Foray P., and Tsuha C.H.C. 2010. Sand Grain Crushing and Interface Shearing During Displacement Pile Installation in Sand. Géotechnique, 60(6): 469-482.  267  Chapter 8.   Summary and Conclusion   8.1.  The Cone Penetration Test (CPT) has been widely used for evaluating the in-situ conditions of soils because of its continuous data measurement and repeatability at relatively low cost. The test is even more attractive in cohesionless soils such as sands, silts and tailings due to difficulties associated with retrieving undisturbed samples in them. The behaviour of cohesionless soils strongly depends on their density as well as stress level.  An index that captures both the effects of void ratio and mean stress on soil behaviour is the state parameter,  (Been and Jefferies, 1985). Determining the state parameter from CPT data is done through a process referred to as interpretation. Both experimental and analytical approaches have been used to develop interpretation techniques. Analytical solutions have certain advantages but they have to be verified and validated against experimental data. Most analytical solutions to date are either too   Summary  268  complicated to be applicable to real engineering projects, or fail to capture the essential . Even those that appear competent in dealing with the cone penetration problem have not been verified and validated against a sufficiently wide range xperimental data. Such verification and validation is necessary to provide confidence in both e accuracy and the range of ground conditions to which any analytical framework can be  the state parameter from CPT tip resistance, qc as laid out, verified, validated, and discussed. The cone penetration was analysed as the xpansion of a spherical cavity with a large strain finite element code using a critical state soil odel (NorSand) capable of accounting for both elasticity and plastic compressibility. simple enough to be used in real engineering projects, and still sufficiently etailed to capture the essential characteristics of the problem. The framework relies on ed, and the confidence and lts could be obtained from different in-situ testing characteristics of the problem e th applied in practice. An analytical framework for the interpretation of w e m The framework is d material-specific and independent calibration of the numerical model through triaxial tests. Special attention was paid to the critical state friction angle, c as one of the parameters having a pronounced effect on the interpretation results. Different methods of determining the state parameter were discuss accuracy in its determination was quantified. The efficiency of the method was further investigated by analysing two well documented case histories confirming that consistent resu methods using the proposed analysis technique. Consequently, cumbersome large scale testing methods such as Calibration Chamber (CC) testing can be substituted by a combination of triaxial testing and finite element analysis producing soil specific correlations.  269  One of the difficulties in understanding the cone penetration problem is the less researched effect of high stresses developing around the cone on the behaviour of the soil. A series of triaxial tests on Fraser river sand were performed focusing on the effects of particle breakage on the location of the CSL. Particle breakage was shown to cause additional compression in the soil and a parallel shift in the Critical State Locus (CSL).  8.2. Major Topics of Research  1. Four different methods of obtaining the critical state friction angle, c from drained triaxial tests were examined through two independent and extensive databases of triaxial nd confidence level in determining the critical vels of accuracy in determining the critical state consistent shape function between the normalised measured CPT tip resistance and spherical cavity testing on clean sand. The purpose was to identify the most suitable method of determining the critical friction angle for use in calibrating the NorSand constitutive model to laboratory test data. 2. A statistical evaluation of a drained triaxial compression test database on Erksak sand was performed to determine accuracy a state friction angle. Recommendations were made on the number and distribution of the tests necessary to achieve different le friction angle. 3. A framework for evaluating in-situ soil state parameter, 0 from CPT tip resistance, qc was developed based on a spherical cavity expansion analogy. A database of nine normally consolidated soils, including laboratory standard and natural sands as well as tailings materials, was used to verify the framework. A unique and  270  expansion limit pressures was identified. The spherical analysis results were then used to back calculate 0 for 301 calibration chamber tests and the resulting states were compared to those measured in the lab. 4. In order to validate the proposed framework for obtaining 0 from qc data, cylindrical and spherical cavity expansion analyses were used to interpret the state parameter from in-situ Self-Bored Pressuremeter (SBP) and CPT in an hydraulically placed clean quartz 5. The developed framework for determining 0 was applied to a site in the Fraser river o occur at stress levels similar to those often generated sts on Fraser river sand. sand in the Beaufort Sea. delta in British Columbia, Canada, that has been extensively investigated as part of the CANadian Liquefaction EXperiment (CANLEX) project. The void ratios obtained from the cores obtained from ground freezing and Laval large diameter sampler were used to validate the approach. 6. Particle Breakage is known t during CPT penetration, although this phenomenon is rarely considered as part of CPT interpretation. The final aspect of this research was to develop a simple hypothesis explaining the material behaviour during the shearing a particulate soil at stresses high enough to produce particle breakage. The hypothesis was tested using a series of triaxial compression te  The following sections (i.e. sections 8.2.1 to 8.2.6) summarise the conclusions derived from the work in this research with respect to the items above.    271  8.2.1. Determination of the Critical State Friction Angle from Triaxial Tests - Summary of Findings  Four m equival disadva triaxial The "m definiti test" m critical for localisation, the “end of test” method provided poor estimates of Mtc. Two m (1972) minimu which While limitati ude only a limited number of tests. The Bishop ethod, although very simple, is sensitive to any outlying data points (associated with poor pler to discard outlying values making it specially helpful when dealing with a small number of tests. The Stress-Dilatancy method sually has better resolution if high-scan-rate data is available for processing; an issue of less ethods reported in the literature for obtaining the critical state friction angle, c (or the ent critical state friction angle Mtc) were summarised and the advantages and ntages of each method were investigated using previously published data from drained  tests on two sands. aximum contraction", although widely reported in the literature, is inconsistent with the on of the critical state and was found to give unrealistic estimates of Mtc. The "end of ethod was also found to be problematic. The large strain levels required to reach the  state were generally not achieved in these triaxial tests and, combined with the potential ethods are recommended for obtaining reliable values of c: The 40 year old Bishop method which uses an extrapolation of the peak friction angle (or stress ratio) versus m dilatancy to the zero dilatancy axis; and an application of the stress-dilatancy method involves extrapolation of the complete stress-dilatancy path to zero dilatancy. both methods did provide reliable values of c, they had differing advantages and ons. Most triaxial testing programs incl m quality tests). Conversely, an advantage of the Stress-Dilatancy method is that it yields an estimate of Mtc for every single test, so it is much sim e u  272  importance when using the Bishop method. However, the use of post peak data in the Stress- Dilatancy method requires cautious application of it due to the possibility of shear banding and tress localisation post peak. It is therefore recommended that the whole stress-dilatancy path of  Accuracy in Determination of Critical State Friction Angle -  proposed by Bishop ted under s tests be plotted and used in conjunction with the Bishop approach; obtaining the benefits of both methods.  8.2.2. Confidence and Summary of Findings  A statistical evaluation of a drained triaxial compression test database was performed to determine accuracy and confidence level in determining c. The critical state friction angle was obtained from a dataset comprising 34 triaxial tests using the methodology (1972); c being obtained using linear regression. It was assumed that the correct c was obtained if all 34 tests were included in the analysis. In determining the accuracy of smaller realisations of the dataset it was assumed that any test program will include at least one loose (Dr = 26 %), one medium dense (Dr = 56 %) and one dense (Dr = 70 %) sample tes drained triaxial compression conditions. Results were presented as error in c versus number of tests for confidence levels of 75%, 85%, 90% and 95%. As the number of tests increased from 3 to 8 a large increase in accuracy was observed at all confidence levels. Hence it is recommended that any commercial testing program for evaluation of c includes at least 6 tests (6 tests yielding an accuracy of 1.0º from university quality data with 90% confidence). For academic purposes, where accuracy of 0.5º with 90% confidence may be needed, more than 20 tests may be required.  273  Although the presented results were developed using one comprehensive academic testing program, their application to commercial data was encouraging. Although (unsurprisingly) the errors from the commercial dataset were slightly larger, the academic database provided a reasonable upper bound on likely achievable accuracy. In conclusion, although soil type and gradation might be expected to affect sample uniformity uring reconstitution and hence influence the repeatability (and hence accuracy) of the triaxial should pro c for fewer tests. d for which both calibration chamber tests and triaxial ical cavity limit pressures was d testing program, distributing the triaxial tests over a wide range of initial Dr, or ideally initial , vide greater accuracy in  8.2.3. Interpretation of Sand State from CPT Tip Resistance - Summary of Findings  A framework for evaluation of in-situ soil state parameter 0 from CPT tip resistance qc data was developed based on a spherical cavity expansion analogy. Central to the approach is a shape function that relates the calculated normalised spherical cavity expansion pressure and the measured calibration chamber tip resistance. A database of nine normally consolidated soils, including laboratory standard and natural sands as well as tailings materials, was selecte tests were available. The NorSand critical state soil model was calibrated for each of the nine soils using the triaxial data. Spherical cavity expansion analysis was then performed on 301 normally consolidated calibration chamber tests and the results compared to those obtained experimentally. All nine soils followed the same trend with a scatter equivalent to that observed for the most intensely tested calibration chamber sand, Ticino. A unique and consistent shape function between the normalised experimental CPT and spher  274  identified. Estimation of accuracy of the proposed framework is constrained by the repeatability seen in the calibration chamber data. For known soil properties (from triaxial compression tests) the soil specific k and m relations recover the qc trend line to near perfect accuracy, and without bias. e to that observed in recovering the state arameter in calibration chamber data itself. d e resulting states compared to those measured in the laboratory. The method predicted 0 to lf. In some of the most extensively investigated Individual calibration chamber results cannot be recovered to this accuracy, but this appears to be a consequence of the intrinsic variability in the calibration chamber data. For the nine soils, it was shown that the normalised cone resistance can be analytically reproduced with an accuracy of ±70% with 89% confidence. This accuracy is clos p The spherical analysis results were then used to back calculate 0 for all 301 chamber tests an th within ±0.04 with 78% confidence and to within ±0.07 with 92% confidence. Two sources of error were explicitly identified; inherent variability in the calibration chamber testing and the soil’s elastic modulus. Clearly no analytical method can predict 0 with accuracy higher than the repeatability of the dataset itse sets of data (calibration chamber tests on Ticino sand) the tip resistance can vary by a factor of two, despite ostensibly identical initial conditions. This inherent variability between calibration chamber tests likely accounts for about half the observed error. Part of the lack of repeatability seen in the chamber test data could credibly be caused by changes in the elastic modulus G. For Ticino sand the estimated error in interpreted 0 due to a  50% variation in G was estimated to be up to ±0.03. Hence, it is recommended that G be measured in-situ to eliminate an unnecessary source of error. Where G is adequately measured, the accuracy of the state parameter estimation is virtually the same as that of costly calibration chamber tests.  275  8.2.4. Evaluation of Sand State from SBP and CPT: A Case History - Summary of ted independently, and plotted for comparison. Six n chamber tests st taken to Findings  The interpretation framework was used to interpret the in-situ state parameter from CPT in a hydraulically placed clean quartz sand in Beaufort Sea. An analogous technique was used for obtaining 0 from in-situ SBP tests in an adjacent borehole. Calibration chamber test data available for the site were modified by the means of the analytical method for fabric effects to replicate in-situ soil conditions and used as reference. Data from the CPT and SBP tests were evalua of the seven evaluations indicate good correspondence between estimates for 0 estimating the state parameter with a difference of 0 to 0.05. This is considered a very good level of consistency when compared to the common scatter observed in calibratio performed under controlled conditions in laboratory. In comparison to CPT, SBP tends to estimate more negative (denser) values of 0 by approximately 0.02, indicating a denser deposit. Although small, this bias in the estimate is non- conservative relative to the CPT interpretation; so the results should be cautiously applied to engineering problems. There was one problematic SBP test, for which the SBP curve fitting estimated a 0 value of -0.26 compared to 0 = -0.16 from CPT data at the same depth. Although the CPT data confirms that the test was done on the densest material in the borehole, the SBP estimate seems unrealistic, especially as no densification effort had been undertaken at the site. However, this test was continued to the lowest strain (2.8%) amongst all the SBP tests done at the site. Other tests were taken to 5.5% to 10.5% strain. Thus, it could be speculated that any SBP te  276  strains of less than about 5% should be treated with extreme caution when it comes to determining the state parameter. PT results appeared more consistent, and the state parameter values were more reasonable with ity amongst other characteristics) in which pted CPT interpretation. An adjacent pressuremeter test was used to e C respect to site conditions. However, since CPT calibration chamber testing was used as the reference, and no independent measurement of the in-situ state is available, it is not possible to say whether the CPT or SBP gave the “correct” answer; but the CPT and SBP results were close enough to be considered “in general agreement” with each other and the expected site conditions. The mismatch between the state parameters obtained from SBP and CPT may be due to the geometry correction applied to the SBP. The correction is applied to account for the difference in cylindrical cavity expansion analysis that assumes infinite pressuremeter length and the real pressuremeter geometry. The current correlations, based on limited experimental data, do not account for the possible effects the material (its dens the test is being done may have on the correction. One should keep in mind that although the CPT appears to produce a more reasonable and consistent state parameter interpretation for the site investigated, an estimate of horizontal stress is required for the ado obtain the horizontal stress at this site, hence the accuracy of CPT inferred state param ter relies in part on SBP measurement.     277  8.2.5. Interpretation of the Sand State from CPT in Fraser River Sand: A Case History - Summary of Findings to in-situ density measurements and compared to other methods of interpreting the sed in developing the framework. The results calculated for the range of CPT resistances, representing ideal ccuracy for the information used in this work. irect measurements of the void ratio yielded an average void ratio of 0.96 with standard eviation (SD) of 0.05, translating into an average state parameter of -0.055 with the same SD. The wide scatter in the measured void ratios, reflected in the large SD, covers a range of  The interpretation framework presented in Chapter 4 was applied to the Massey Tunnel site, an extensively investigated site in Fraser river delta in British Columbia, Canada. The effect of soil fabric on the interpretation results was considered by adjusting the calibration parameters with respect to tests on undisturbed samples. The accuracy of the method was evaluated by comparison state parameter from CPT. The samples trimmed from the cores obtained from ground freezing and Laval large diameter sampler were considered to represent the real in-situ void ratio. The state parameter was obtained using the framework developed in Chapter 4 for the range of CPT tip resistances measured in the site. ± 0.07 and ± 0.04 error margins were identified to be able to compare the confidence level of the interpretation to the data u were then compared to the state parameters calculated from the in-situ void ratio measurements. 98% of the undisturbed sampling measurement fell within the ± 0.07 error margins and 70.5% within the ± 0.04 error margins. These confidence levels compare very well with the 92% and 78% observed in the calibration chamber data used in developing the framework. 20.5% of the data were coincident with 0 a D d  278  01.0155.0 0   . Part of the scatter likely stems from the ground sampling techniques, rather  a characteristic of the ground, than being as widely ranging void ratios were measured in amples from adjacent points of the coring. While the variation in the measured void ratios CPT. s represents a wide range of sand behaviour, ranging from loose to dense, the geology of the site and all in-situ testing (including CPTs) imply a relatively uniform deposit associated with a narrower range of ground density and sand behaviour. This paradoxical observation in methods of obtaining the in-situ density from undisturbed samples of sands calls for a more cautious treatment of the experimental results and emphasises the need for better interpretation techniques for tests such as The accuracy offered by the ± 0.07 error margins covers a wide range of possible state parameters (typically 05.030.0 0   ) when combined with the range of the original CPT data. However, the range is very similar to that covered by the frozen and Laval large diameter samples, suggesting that this method is as capable as the most expensive and cumbersome of ground sampling techniques for determining the soil’s in-situ density. The average   of -0.067 differs from the -0.055 average measured by ground sampling by -0.012. This difference is close to ±0.01; the ground sampling technique error margin given by Wride and Robertson (1997). The results were compared to those obtained from Been et al. (1987), Plewes et al. (1992) and Konrad (1997) methods. Amongst all interpretation methods the one presented in Chapter 4 provided the closest estimation of   for Fraser river sand. Although the method appears to be more difficult than the others, the difficulty lies 0 0 only in the analysis and modeling effort which is achievable in a matter of hours. With the exception of Plewes et al. (1992), all interpretation methods require knowledge of the CSL in pe  log  space. Performing a number of triaxial compression tests is the easiest and most common way of estimating CSL. The only additional  279  requirement of the present method is for these tests to be performed under drained conditions; a requirement that does not pose any additional laboratory testing effort. Plewes et al. (1992) correlated the slope of the critical state line in pe  log  space to the CPT friction ratio based on experimental results, hence eliminating the need to experimentally obtain the CSL. However, the method does require laboratory testing to measure c (or the analogous Mtc). Plewes et al. (1992) suggested using Mtc = 1.2 ( 30c ) for all soils, advising that doing so would cause less than 10% error in the estimated 0. However, using Mtc = 1.2 for Fraser river sand puts the estimated   in line with the other methods (with an average   of -0.089), resulting in a more negative (denser) state param 0 0 eter than that implied by ground sampling 0 tter 0 e state p techniques. The success of these methods in obtaining  appears to be directly related to the level of material behaviour taken into consideration. Konrad (1997) does not account for any mechanical aspects of soil behaviour and returns the most discrepancy in estimated state parameter (with an average 0 of -0.114). The Been et al. (1987) method accounts for material compressibility through the slope of the CSL (with an average 0 of -0.092). The Plewes et al. (1992) method adds the effect of the critical state friction angle to their framework resulting in an even be estimation (with an average   of -0.071). The framework presented in this thesis also accounts for both compressibility and friction angle, and importantly adds elasticity, as well as stress level, dilatancy, and fabric. An overall comparison of the model parameters for Fraser river sand to those of other sands presented in Chapter 4 suggests that the most important factor that makes the other methods systematically biased towards a more negativ arameter in Fraser river sand is its high  280  critical state friction angle. This is further confirmed by the fact that a correct Mtc value in the Plewes et al. (1992) method results in a markedly better interpretation.  8.2.6. Particle Breakage and the Critical State of Sand - Summary of Findings  The current work involved using laboratory testing to assess a hypothesis for particle breakage; of importance to CPT due to the high pressures encountered at the cone tip. The hypothesis explained the CSL movement due to shearing a particulate soil at stresses high enough to produce particle breakage. This conceptual model was tested using a series of 39 triaxial compression tests on Fraser river sand, reaching p levels of up to 3 MPa. The associated increase in the fines content of sheared samples ranged from zero to nearly 7%. The main aspects of the hypothesis that are confirmed by laboratory testing were:   Measurable breakage only starts after the soil’s contraction capacity is exhausted.  Breakage causes a downward parallel shift in the location of the CSL in pe  log space.  The magnitude of this CSL shift is equal to the void ratio reduction due to breakage ) and is directly correlated to the increase in fines content.  These observations on the parallel shift in the CSL contradict the widely adopted three part line used to model the effect of particle breakage on the CSL (e.g. Russell and Khalili, 2004). It is recommended that future work idealises the CSL as a series of parallel loci each associated with a certain level of particle breakage. (  be  281  The hypothesis also implies that since  be , the state parameter is independent of breakage. The volume reduction caused by the breakage may simply be superimposed on the olume change controlled by stress-dilatancy. Hence for small amounts of breakage the critical am for soil mains valid even in the presence of particle breakage; particle breakage can be treated as an .3. Contributions  The m e in the rese c  1. A correlation was provided between the number of triaxial tests required to achieve  extensive database of calibration chamber tests, and validated against two well documented case histories. It was shown that the confidence v state fr ework found in many, and arguably most, advanced constitutive models re uncoupled phenomenon that changes . The work presented here lays out a conceptual framework for the idea of parallel CSLs (Daouadji et al., 2001) and provides experimental evidence for it. The simplicity of the hypothesis and its implications on the evolution of the state parameter towards the critical state under particle breakage provides a basis for modelling of the phenomenon within a critical state soil mechanics context.  8 ain contributions of this dissertation to the current state of knowledg  fields ar hed are: certain accuracy and confidence levels in determining the critical state (constant volume) friction angle in laboratory. 2. An analytical framework for estimating the in-situ state parameter of sands was verified against an unprecedentedly  282  and accuracy achieved through the framework is comparable to the best and most expensive laboratory and field methods. 3. A simple hypothesis was put forward to explain the effects of particle breakage on the critical state and behaviour of sand. A significant number of triaxial tests were performed for the first time to investigate the matter. The hypothesis was demonstrated to successfully explain the behaviour observed in the tests. ata. Application of this work to other situations should take account of uch limitations. The following specific limitations can be pointed out with regards to different  for a clean quartz sand. Soil type and gradation are expected to affect sample uniformity during reconstitution and hence influence the repeatability (and hence 2. y is not applicable to finer grained materials and/or higher  8.4. Limitations  The results presented in this thesis were obtained based on a series of assumptions and verified against a limited set of d s portions of the work:  1. The confidence and accuracy levels in determining the critical state friction angle were obtained accuracy) of the triaxial testing program. The framework for interpretation of the state parameter was developed for drained penetration. The methodolog rates of penetration where development of excess pore water pressure around the cone is expected.  283  3. The constitutive model used to represent soil, Norsand, is applicable to granular soils. Hence the methodology presented does not apply to cohesive or cemented materials. 5. The framework is verified and validated for an extensive database of calibration chamber situ measurements of the state parameter. However, all the materials comprising the database are more or less sand size materials with fines contents not effects of particle breakage on the critical state of the sands was limited to one suite of testing on Fraser river sand. Further testing investigating the effects of particle breakage  4. An important assumption of the cavity expansion analysis is a uniform homogeneous medium around the expanding cavity. The current framework is therefore not capable of capturing the effects of phenomena such as closely layered soils and soils with significantly anisotropic stress-strain behaviour. tests and in- exceeding 6%. The extension of its use to other cohesionless materials such as silts and fines rich sands and materials with significantly different minerologies (e.g. Calcareous sands) should be sufficiently examined as such data become available. 6. The experimental work done to investigate the is needed to confirm the generality of the results. Testing with a range of material types is also needed to extend the theoretical framework to link the reduction in  during particle breakage to the encountered stress-strain path. 7. The stress and strain levels achieved in were limited by the constraints of the triaxial testing apparatus. Hence the amount of particle breakage was limited and it did not become clear how far the observed trends would continue as more breakage occurs under higher stress and strain levels.   284  8.5. Future Studies  The work presented in this thesis focused on interpreting the state parameter from CPT tip resistance. The critical state friction angle and the particle breakage phenomenon were further investigated as factors with great influence on the interpretation. The following work is recommended to be undertaken as part of future research on this subject:  1. 2.  shear strain and 3.  approximation. A more realistic  . Investigating the effect of lateral stress on CPT tip resistance and its implications. This issue has been previously investigated, but the addition of new chamber testing using Studying the effect of over-consolidation ratio on the proposed CPT interpretation methodology. The current work only considered normally consolidated soils, but over- consolidated soils are common in practice and calibration chamber data are available with over-consolidated samples. Further laboratory testing to tie the amount of particle breakage to the stress level. And investigating how the stress path taken can affect particle breakage. Such research can provide the experimental evidence for a new framework for modeling the particle breakage phenomenon. It will be then possible to investigate the effects of including particle breakage in CPT interpretation. Modelling the CPT using an axisymmetric geometric geometric approximation enables investigation of the effect of sleeve friction on interpretation of the state parameter. Accurate geometry also allows investigation of the effect of soil layering on CPT resistance. The understanding gained from these simulations could be used directly, or incorporated into the current framework. 4  285  more advanced chambers and application of more sophisticated constitutive models can illuminate the previous discussions on this subject. itively expensive and umbersome calibration chamber tests to a few triaxial tests on reconstituted samples, followed nical problems such as liquefaction, more projects will find present   In the absence of material-specific data, careful consideration should be given when extrapolating the methods and data presented in this thesis to other cases involving different materials. The analysis presented here reduces the need for prohib c by a limited amount of numerical analysis. The geotechnical community is moving away from the use of rule of thumb correlations for interpreting CPT data except for problems with insignificant consequences (Robertson, 2009). As requirements are expanding to require analysis of the consequences of geotech  the resources to justify the additional efforts required for adopting methods such as those ed here.  286  8.6.  een K., and Jefferies M.G. 1985. A State Parameter for Sands. Géotechnique, 35(2): 99-112.  Sites. Canadian Geotechnical Journal. 34: 120–130. uir Wood D. 2007. The Magic of Sands. The 20th Bjerrum lecture presented in Oslo, 25 November 2005, Canadian Geotechnical Journal, 44: 1329-1350. Plewes H.D., Davies M.P., and Jefferies M.G. 1992. CPT Based Screening Procedure for Evaluating Liquefaction Susceptibility. Proceedings of the 45th Canadian Geotechnical Conference, Toronto. Robertson P.K. 2009. Interpretation of Cone Penetration Tests – A Unified Approach. Canadian Geotechnical Journal, 46: 1337-1355. Russell A.R., and Khalili N. 2004. A Bounding Surface Plasticity Model for Sands Exhibiting Particle Crushing. Canadian Geotechnical Journal, 41(6): 1179-1192. References B Been K., Jefferies M.G., Crooks J.H.A., and Rothenburg L. 1987. The Cone Penetration Test in Sands. Part II: General Inference of State. Géotechnique 37(3): 285-299. Bishop A.W. 1972. Shear strength parameters for undisturbed and remolded soil specimens. In Proceedings, Roscoe Memorial Symposium, Cambridge University. Edited by R.H.G. Parry. G.T. Foulis & Co. Ltd., Yeovil, U.K., 3–139. Daouadji A., Hicher P. Y., and Rahma A. 2001. An Elastoplastic Model for Granular Materials Taking into Account Grain Breakage. European Journal of Mechanics - A/Solids, 20: 113–137. Konrad J.-M. 1997. In Situ Sand State from CPT: Evaluation of a Unified Approach at Two CANLEX M  287  Wrid C.E. and Roe bertson P.K. 1997. The Canadian Liquefaction Experiment, Phase II Data Review Report, Massey and Kidd Sites, Fraser River Delta, June 1997. ISBN 0-921095- ,  49-X. Wride C.E. and Robertson P.K. (eds.), BiTech Publishers Ltd., Richmond, BC Canada.  288                 Appendix A  Triaxial Tests on Fraser River Sand Procedures and Results            289  A.1. Testing Equipment riaxial tests were carried out using a computer-controlled triaxial system capable of consolidating and shearing soil specimens along user specified stress paths. Figure A-1 presents a picture of the equipment and the triaxial set se up. The system consisted of a load frame, two og data acquisition system all onnected and controlled by a computer via interface modules. The pressure/volume actuators generated and controlled pressures up to 2100 kPa (300 psi) and flow rates ranging from 25 ml/min to 0.000025 l/min. The system also included one load cell, one deformation sensor, three pressure s mperature sensor. Load cells with 500, 2000 and 10,000 lbs (227, 907 and 4536 kg respectively) capacities were available for obtaining the optimum accuracy fr d a range of -68 kPa to 2100 kPa (-10 psi to 300 psi) and were used in the pressure lume control pumps; the third sensor had a range of -68 kPa to 1380 kPa (-10 psi to 200 psi) and was used for pore pressure measurements. Linear displacement transducers had a range of 76.2 mm (3.0 in). All electronic measurement devices were connected to the loading frame and then into the laboratory PC; data was logged electronically recording 200 measurement every hour giving a typical data set of 600 measurements in a test taken to 15% axial strain. The triaxial cell was equipped with ‘frictionless’ end platens. These platens had a diameter of 74 mm. and the samples were prepared with a split mould (Figure A-2) with a diameter of 71.5 mm. The split mould height was 142.8 mm for a sample height to diameter ratio of 2.    T t- pressure/volume actuators (flow pumps) and a high resolution anal c m ensors and one te om the tests. Two of the pressure sensors ha /vo  290  A.2. Test Procedures  The vacuum split mould was attached to the triaxial cell base with a membrane connected to the bottom platen (Figure A-3). The top of the membrane was folded over the split mould and the membrane was held open with a vacuum pressure. Samples were prepared by mixing distilled water with the soil (Figure A-4) to a moisture content of 5% with a weight precision of 0.01 g. Six layer mass portions were determined using the undercompaction method proposed by Ladd  by a smear of igh-vacuum silicon-based grease. (1978). The six mass portions of the moist soil were then placed in the mould (Figure A-5) and compacted with the tamping rod (Figure A-6) in six equal thickness lifts (each 1/6 the total sample height in the mould). The method was employed to ensure that each layer, and thus the entire sample, is prepared to the target density and void ratio. Prior to the next lift being placed, the surface of the previous lift was scarified to a depth of 1 mm to 2 mm. After tamping of the sixth lift, the top cap platen was placed on the sample, the membrane turned up and the o-rings applied to seal the membrane to the top cap (Figure A-7). Then the vacuum was switched from the split mould to the sample through the drainage port. At this point a vacuum pressure of about 20 kPa was applied. While under the vacuum pressure, sample height and diameter were measured to a precision of ±0.01 mm. Each dimension was measured at four different locations and the average values were recorded as sample dimensions. Triaxial tests were carried out using lubricated enlarged end platens aimed at reducing stress non-uniformities and minimizing the end restrain. The procedure developed by Tatsuoka et al. (1984) was used by using two rubber membranes (0.21 mm thickness) separated h  291  The triaxial cell was assembled following the measurement of the sample dimensions. The cell as filled with water and a 20 kPa seating cell pressure was incrementally applied to the sample investigate the particle breakage ple and the loading cap were gently wiped and put into a freezer for a minimum 24 hours. After the specimen was frozen, the membrane, top and bottom w while reducing the vacuum on the sample. Then, CO2 gas was slowly percolated through the sample for a period of two to four hours. Following flushing with CO2, de-aired water was allowed to flow through the sample under less than 2 m driving head for 8 to 12 hours. Sample dimensions during saturation were monitored by the displacement of the top cap and chamber volume changes. The sample pore pressure lines were then connected to the pressure/volume actuators and back pressure was applied. The back pressure saturation was carried out by increasing the cell pressure and measuring the response in pore pressure. The cell pressure was increased in increments until a B ≥0.95 for drained tests or B ≥0.98 for undrained tests was achieved. After completing the saturation phase, the sample was consolidated to the required pressure in incremental steps. The volume change during the consolidation process was continuously monitored. The sample was then sheared at a constant rate of 5% per hour. The tests aimed at locating the CSL were prepared to a loose state and sheared until a steady state of constant volume was achieved. Dense samples were sheared until clear shear banding was detected in the sample. Some of the tests were stopped at lower strain levels to occurred at intermediate states of shearing as explained in Chapter 7. At the end of the shearing phase the drainage valves were closed keeping and all pore water in the sample. The cell’s confining fluid was drained and after removing the chamber sleeve, the cell pedestal together with the sam  292  platens were carefully removed and any soil grains remaining on them carefully brushed into the sample pan. The sample was then dried in a laboratory oven for determining the final water content. This water content was used to calculate the void ratio at the end of shearing. This technique (Sladen and Handford, 1987) effectively eliminated loss of water during sample extraction, enabling accurate determination of void ratio at the end of the test. It also allowed for ing and isotropic consolidation with the accuracy provided by the nce in experimental observations is greatly enhanced by the repeatability of the nd strains as illustrated in determining the void ratio to the end of the back-pressurisation phase by back tracking the void ratio changes during shear flow pumps. Corrections were applied for membrane penetration (Vaid and Negussey, 1984) and membrane force (Kuerbis and Vaid, 1990).  A.3. Repeatability of Test Results  The confide results. Repeatability is achieved by consistent reproduction of the void ratio, fabric (through reconstitution method), repeatability of the procedures, duplication of the stress paths, and the measurement accuracy. A repeatability of 0.01 or better in void ratio measurement was achieved for three pairs of tests that were targeted to start from identical conditions; and, there was corresponding near-perfect repeatability in measurement of stresses a Figures A.8 and A.9.     293  A.4. Test Results  Figure A.10a shows the deviator stress and volumetric strain plotted against axial strain for the drained tests presented in Table 7-1. Figure A.10b shows the deviator stress and the pore pressure plotted against the axial strain for the undrained tests presented in Table 7-1. The deviator stress, pore pressure and volumetric strain are plotted against axial strain for samples subjected to higher levels of stress in Figures A-11a to A-11c . At the end of each parent test the sample was sieved and then retested in the triaxial apparatus at lower levels of stress as summarised in Table 7-2 and shown in Figures A-11d to A-11f .        294  Figure A - 1 Triaxial test loading frame, pressure pumps and cell set-up   295  Figure A - 2 Split mould device  296  Figure A - 3 Triaxial base with split mould and membrane   297  Figure A - 4 Adding water for sample preparation  298  Figure A - 5 Sample preparation in layers   299  Figure A - 6 Sample compaction   300  Figure A - 7 Prepared sample before cell assembly    301  0 500 1000 1500 2000 2500 3000 3500 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0  1 (%) q  ( k Pa ) CID-D #1 100 kPa CID-D #1 100 kPa Repeated CID-D 1000 kPa CID-D 1000 kPa-peak CID-L 1400 kPa CID-L 1400 kPa-2  -4 -2 0 2 4 6 8 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0  1 (%)  v (% )  Figure A - 8 Deviator stress and volumetric strain versus axial strain for three pairs of tests starting from identical density and stress states    302   303 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95 10 100 1000 10000 mean effective stress, p' (kPa) vo id  r at io , e  CID-D #1 100 kPa CID-D #1 100 kPa Repeated CID-D 1000 kPa CID-D 1000 kPa-peak CID-L 1400 kPa CID-L 1400 kPa-2  Figure A - 9 Void ratio versus mean effective stress for three pairs of tests starting from identical density and stress states     304 0 200 400 600 800 1000 1200 1400 1600 1800 0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 1 (%) q  ( k P a )  0 100 200 300 400 500 600 700 800 900 0.0 5.0 10.0 15.0 20.0 25.0 1 (%) q  ( k P a )  -8.0 -6.0 -4.0 -2.0 0.0 2.0 4.0 6.0 8.0  v  ( % ) CID-D 50 kPa CID-L 100 kPa CID-D 115 kPa CID-L 300 kPa CID-D 410 kPa CID-D 515 kPa CID-L 600 kPa CID-L 190 kPa UR CID-D 200 kPa UR 0 50 100 150 200 250 300 350 0.0 5.0 10.0 15.0 20.0 25.0 1 (%) u  ( k P a ) CIU-L 100 kPa CIU-L 200 kPa CIU-M 200 kPa CIU-D 200 kPa CIU-L 300 kPa CIU-L 390 kPa CIU-M 400 kPa Figure A - 10 Deviator stress, pore pressure and volumetric strain plot t the axial strain for tests aimed at determining the CSL for virgin FRS, summarised in Table 7-1  a) b) ted agains  0 500 1000 1500 2000 2500 3000 3500 4000 4500 0.0 5.0 10.0 15.0 20.0 25.0 1 (%) q  ( k P a )  0 500 1000 1500 2000 2500 3000 3500 0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 1 (%) q  ( k P a ) -100 0 100 200 300 400 500 600 700 0.0 5.0 10.0 15.0 20.0 25.0 1 (%) u  ( k P a ) CIU-D 1300 kPa 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 1 (%)  v  ( % ) CID-L 1600 kPa CID-D 600 kPa CID-D 600 kPa-peak CID-D 1000 kPa-peak CID-D 1000 kPa b) a)  305  0 500 1000 1500 2000 2500 3000 3500 4000 4500 0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 40.0 45.0 1 (%) q  ( k P a ) 0 200 400 600 800 1000 1200 1400 1600 1800 0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 40.0 1 (%) q  ( k P a )  -6.0 -4.0 -2.0 0.0 2.0 4.0 6.0 8.0 0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 40.0 1 (%)  v  ( % ) CID-D #1 100 kPa CID-D #1 100 kPa CID-L #3 150 kPa CID-M #3 600 kPa CID-M #3 200 kPa  0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 40.0 45.0 1 (%)  v  ( % ) CID-L 1400 kPa CID-L 1400 kPa-2 c)  d)  306   307 0 50 100 150 200 250 0.0 5.0 10.0 15.0 20.0 25.0 1 (%) q  ( k P a )  0 200 400 600 800 1000 1200 1400 1600 1800 2000 0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 40.0 1 (%) q  ( k P a )  0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 40.0 1 (%)  v  ( % ) CID-L #5 100 kPa CID-L #5 300 kPa CID-L #5 600 kPa CID-L #6 100 kPa CID-L #7 100 kPa CID-L #7 300 kPa CID-L #8 100 kPa CID-L #8 300 kPa f) 0 50 100 150 200 250 300 350 0.0 5.0 10.0 15.0 20.0 25.0 1 (%) u  ( k P a ) CIU-L #1 200 kPa CIU-M #1 200 kPa CIU-M #1 400 kPa Figure A - 11 Deviator stress, pore pressure and volumetric strain plotted against the axial strain for tests aimed at determining the effects of particle breakage on FRS, summarised in Table 7-2 e)  A.5. References  Kuerbis R.H., and Vaid Y.P. 1990. Corrections for Membrane Strength in the Triaxial Test. Geotechnical Testing Journal, 13(4): 361-369. Ladd R.S. 1978. Preparing test specimens using undercompaction. Geotechnical Test Journal, 1(1): 16-23. Negussey D., Wijewickreme W.K.D., and Vaid, Y.P. 1988. Constant Volume Friction Angle of Granular Materials. Canadian Geotechnical Journal, 25: 50-55. Sladen J.A., and Handford G. 1987. A potential systematic error in laboratory testing of very loose sands. Canadian Geotechnical Journal, 24: 462-466. Tatsuoka F., Molenkamp F., Torii T., and Hino T. 1984. Behavior of lubrication layers of platens in element tests. Soils and Foundations, 24(1): 113-128.       ing  308

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share