Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Response of buried steel pipelines subjected to longitudinal and transverse ground movement Karimian, Seyed Abdolhamid 2006

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

Item Metadata

Download

Media
831-ubc_2006-199985.pdf [ 56.2MB ]
Metadata
JSON: 831-1.0063294.json
JSON-LD: 831-1.0063294-ld.json
RDF/XML (Pretty): 831-1.0063294-rdf.xml
RDF/JSON: 831-1.0063294-rdf.json
Turtle: 831-1.0063294-turtle.txt
N-Triples: 831-1.0063294-rdf-ntriples.txt
Original Record: 831-1.0063294-source.json
Full Text
831-1.0063294-fulltext.txt
Citation
831-1.0063294.ris

Full Text

RESPONSE OF BURIED STEEL PIPELINES SUBJECTED TO LONGITUDINAL AND TRANSVERSE GROUND MOVEMENT By SEYED ABDOLHAMID KARIMIAN B . S c , Univers i ty o f Tehran, Iran, 1999 M . S c , Univers i ty o f Tehran, Iran, 2002 A T H E S I S S U B M I T T E D I N P A R T I A L F U L L F I L L M E N T O F T H E R E Q U I R E M E N T S F O R T H E D E G R E E O F D O C T O R O F P H I L O S O P H Y in T H E F A C U L T Y O F G R A D U A T E S T U D I E S ( C i v i l Engineering) T H E U N I V E R S I T Y O F B R I T I S H C O L U M B I A September 2006 © Seyed Abdolhamid Karimian, 2006 ABSTRACT The performance of buried pipeline systems in areas subjected to ground deformations is an important engineering consideration, and there is a need for further research to advance the current fundamental understanding of this problem. A new full-scale physical modeling facility was developed to investigate soil-pipe interaction of relatively large diameter steel pipelines. The facility comprises a large soil box with the capacity to impose axial and lateral displacements on buried pipelines while simulating desired backfill and native soil configurations. Methods were developed to measure normal soil stresses on the pipe surface and observe movement of sand particles during testing. Numerical models, with parameters derived from geotechnical element testing and validated by physical modeling results, were used to conduct parametric studies of pipeline pullout response. The measured axial soil loads on pipes in loose sand were comparable with those from the commonly used ASCE-equation. Pipes in dense sand exhibited axial soil loads several-fold higher than those estimated from the same equation. The increase in normal soil stress on the pipe surface due to constrained dilation of sand in the shear zone was found to be the key reason for these high soil loads. For dilative soils, the equivalent lateral earth pressure coefficient (K) representing average normal stress distribution on the pipe is a more appropriate parameter for use in the ASCE-equation than the "at rest" lateral earth pressure coefficient (Ko). Based on numerical modeling, a series of charts and formulae were developed to obtain the value of K. The commonly used methods of wrapping pipelines with geosynthetic-layers were found to be generally effective in reducing axial soil loads. ii Measured lateral soil loads were generally lower than those estimated from existing guidelines. The outcomes from physical and numerical modeling were used to develop approaches to account for the effects of geometric, material, and interface parameters on the lateral soil loads. When in hard native soil, pipe buried in sand in a suitably wide trench with adequate horizontal distance from the trench boundary may effectively reduce the lateral soil resistance. In dual-geotextile-lined trench configurations, in addition to interface frictional characteristics, the relative stiffness of the native soil in comparison to the backfill and the ability of the backfill to move as a "cohesive block" become critical in reducing the lateral soil loads. iii TABLE OF CONTENTS A B S T R A C T ii T A B L E O F C O N T E N T S iv L I S T O F T A B L E S xii L I S T O F F I G U R E S xiii N O M E N C L A T U R E xxi A C K N O W L E D G E M E N T S xxiii 1 I N T R O D U C T I O N 1 1.1 B A C K G R O U N D 1 1.2 S O I L L O A D R E D U C T I O N M E T H O D S 6 1.3 O B J E C T I V E S O F T H E T H E S I S 8 1.4 S C O P E O F T H E T H E S I S 9 1.5 O R G A N I Z A T I O N 11 2 L I T E R A T U R E R E V I E W 13 2.1 R E S P O N S E O F BURIED PIPES S U B J E C T T O A X I A L L O A D I N G 14 2.1.1 Longitudinal soil loads on pipes 14 2.1.1.1 Tests on wrapped pipes 17 2.1.2 Relevance to shaft friction in axially loaded piles 18 2.1.3 Summary of key observations 20 2.2 R E S P O N S E O F BURIED PIPES S U B J E C T T O L A T E R A L L O A D I N G 21 2.2.1 Experimental research 22 2.2.2 Interpretation of experimental findings and numerical modeling 28 2.2.3 Summary of key observations 37 iv 2.3 P I P E L I N E DESIGN GUIDELINES 38 2.3.1 Estimation of axial soil loads 38 2.3.2 Estimation of lateral soil loads 38 2.4 C L O S U R E 3 9 3 P H Y S I C A L M O D E L I N G A S P E C T S 40 3.1 E X P E R I M E N T A L A P P A R A T U S 40 3.1.1 Testing chamber 42 3.1.2 Backfill material - handling, placement, and removal 45 3.1.3 Loading mechanism 46 3.1.4 Pipe specimen placement and actuator connections 50 3.1.5 Instrumentation 53 3.1.5.1 Force measurement 53 3.1.5.2 Actuator displacements 53 3.1.5.3 Pipe displacements 54 3.1.5.4 Displacement of geosynthetic 54 3.1.5.5 Pipe interface pressure measurements 54 3.1.5.6 Backfill soil pressure 56 3.1.5.7 Backfill density measurements 56 3.1.5.8 Data Acquisition 57 3.2 T E S T I N G P R O G R A M 57 3.2.1 Comments on the implementation of axial pullout test 60 3.2.1.1 Loading rate 61 3.2.1.2 Quality and loss of measurement data 63 3.2.1.3 Preparation and variations in axial tests on bare pipe (AB Tests) 64 3.2.1.4 Preparation and variation of axial tests on wrapped pipe 65 3.2.2 Comments on the implementation of transverse pullout tests 67 3.2.2.1 Loading rate 68 3.2.2.2 Preparation and variation of horizontal tests 69 3.2.2.3 Comments on configuration I tests (sand-sand) 70 3.2.2.4 Comments on configuration II tests (sand-lining-sand) 70 v 3.2.2.5 Comments of configuration III tests (sand-lining-hard boundary) 72 3.3 C H A R A C T E R I Z A T I O N O F M A T E R I A L PROPERTIES 73 3.3.1 Materials used in physical model testing 73 3.3.2 Laboratory element testing of materials: direct shear tests 75 3.3.2.1 Direct shear tests on Fraser River sand 75 3.3.2.2 Direct shear tests on Fraser River sand/sand-blasted steel interface 77 3.3.2.3 Direct shear tests on geotextile/geotextile interface 78 3.3.3 Laboratory testing: Triaxial tests 79 3.4 E X P E R I M E N T A L LIMITATIONS A N D ASSOCIATED ERRORS 83 3.4.1 Boundary constraints perpendicular to the direction of pipe movement (Sidewall friction) 84 3.4.2 Boundary constraints in the direction of pipe movement (front and rear wall effects) 87 3.4.2.1 Effect of boundary constraints in axial pullout tests 87 3.4.2.2 Effect of boundary constraints in lateral pullout test 88 3.4.3 Pulling system friction 89 3.4.4 Control of backfill density 90 3.5 S U M M A R Y O F T H E C H A P T E R 94 4 B U R I E D P I P E S S U B J E C T T O L O N G I T U D I N A L G R O U N D M O V E M E N T S 95 4.1 I N T R O D U C T I O N 95 4.2 A X I A L P U L L O U T TESTS O N B A R E PIPE BURIED IN DRY SAND 97 4.2.1 Axial load vs. displacement response 97 4.2.2 Soil pressures on pipe during specimen preparation and axial pullout 100 4.2.2.1 Soil pressures during specimen preparation 101 4.2.2.2 Soil pressures during axial pullout testing 104 4.2.3 Other visual observations 106 4.2.3.1 Surface deformations 106 vi 4.2.3.2 Deformations at the Soil/pipe Interface 107 4.3 A X I A L P U L L O U T TESTS O N S T E E L PIPE W R A P P E D W I T H G E O S Y N T H E T I C S BURIED IN DRY SAND I l l 4.3.1 Axial Load vs. Displacement Response I l l 4.3.1.1 Tests with two layers of woven-geotextile I l l 4.3.1.2 Test with one layer of woven-geotextile and one layer of bi-directional geonet 113 4.3.2 Displacement of geosynthetics during axial pullout testing 114 4.4 DISCUSSION O F T E S T R E S U L T S 117 4.4.1 Axial pullout tests on bare pipe buried in dry sand 117 4.4.1.1 Comparison of measured versus predicted axial soil loads for bare pipe 117 4.4.1.2 Role of Ko in observed differences between tests and predictions 121 4.4.1.3 Behaviour at interface shear zone 127 4.4.2 Observations and discussion on axial pullout tests,on pipe wrapped with geosynthetic layers 128 4.4.2.1 Effectiveness of geotextile-wrapped pipe in reducing soil loads 128 4.4.2.2 Effectiveness of pipes wrapped with geotextile/geonet in reducing soil loads 130 4.4.2.3 General comments 130 4.5 N U M E R I C A L M O D E L I N G 131 4.5.1 Numerical modeling of axial pullout consideration a vertical plane, normal to the pipe axis 132 4.5.1.1 Consideration for the development of model 132 4.5.1.2 Developing of numerical model 134 4.5.2 Thickness of shear zone and the degree of dilation at interface 135 4.5.3 Validation of the model and discussion of the results 138 4.5.3.1 Stresses under static conditions prior to pullout 138 4.5.3.2 Normal stresses on pipe during pullout 140 4.5.3.3 The effect of box size 144 vii 4.6 V A R I A T I O N O F K IN R E L A T I O N T O T H E L E V E L O F DILATION 145 4.6.1 Effect of burial depth on K value 147 4.6.2 Effect of pipe diameter on K value 148 4.6.3 Effect of soil parameters on K value 151 4.7 S U M M A R Y O F T H E C H A P T E R 153 5 B U R I E D P I P E S S U B J E C T T O T R A N S V E R S E G R O U N D M O V E M E N T S 158 5.1 I N T R O D U C T I O N 158 5.2 S U M M A R Y O F T E S T P A R A M E T E R S 159 5.3 L A T E R A L L O A D I N G RESPONSE O F PIPELINE BURIED IN COHESIONLESS SOIL 160 5.3.1 Lateral load-displacement response 160 5.3.2 Soil contact pressures on the pipe during lateral loading 163 5.3.3 Surface deformations and pipe movement 165 5.4 L A T E R A L L O A D RESPONSE O F PIPELINE BURIED IN T R E N C H -CONFIGURATIONS C O N S T R U C T E D IN " C O H E S I O N L E S S N A T I V E S O I L " 167 5.4.1 Load-displacement response 168 5.4.2 Movement of geotextile layers during lateral pulling 169 5.4.3 Surface deformations and pipe movement 170 5.5 L A T E R A L L O A D RESPONSE O F PIPELINE BURIED IN T R E N C H -CONFIGURATIONS C O N S T R U C T E D IN " H A R D N A T I V E S O I L " 172 5.5.1 Load displacement response 173 5.5.2 Soil contact pressures on the pipe during lateral loading 175 5.5.3 Movement of geotextile layers during lateral pulling 176 5.5.4 Surface deformation and pipe movement 177 5.6 DISCUSSION O F T E S T RESULTS 181 5.6.1 Pipeline buried in cohesionless soil 181 5.6.1.1 Comparisons with previous studies and current approaches 181 5.6.1.2 Interpretation based on measured soil stress on pipe 188 5.6.2 Pipeline buried in trench-configurations 192 viii 5.6.2.1 Evaluation of simplified mechanisms to estimate the sources of soil loads on pipe 192 5.6.2.2 Evaluation of measured soil pressures on pipe during lateral pulling tests 195 5.7 N U M E R I C A L M O D E L I N G 200 5.7.1 Development of mesh configurations 201 5.7.2 Parameters and constitutive models for numerical analysis 203 5.7.2.1 Modeling of pipe soil interface 204 5.7.2.2 Linear Elastic, Perfectly Plastic Mohr-Coulomb model 205 5.7.2.3 Hyperbolic Mohr-Coulomb Model 208 5.7.2.4 Modified hyperbolic model to account for density change during pulling of the pipe 215 5.7.3 Effect of size of the box 218 5.8 C O M P A R I S O N B E T W E E N N U M E R I C A L M O D E L R E S U L T S A N D E X P E R I M E N T A L R E S U L T S 220 5.9 E F F E C T O F D I F F E R E N T V A R I A B L E S O N N Q H V A L U E 223 5.9.1 Variation of horizontal bearing capacity factor (N q h) with overburden ratio and friction angle 224 5.9.2 Effect of pipe diameter 226 5.9.3 Effect of surface roughness 230 5.9.4 Effect of pipe and content weight on the soil loads 233 5.9.5 Effect of dilation angle 235 5.10 S O M E OBSERVATIONS R E G A R D I N G M O D E L L I N G O F PIPE IN T R E N C H 237 5.10.1 Modeling of pipe buried in dry sand "backfill" 237 5.10.2 Modeling of pipe with moist sand backfill 239 5.11 S U M M A R Y O F T H E C H A P T E R 242 6 S U M M A R Y A N D C O N C L U S I O N S 247 6.1 F U L L - S C A L E T E S T I N G F A C I L I T Y 249 6.2 FINDINGS F R O M PIPELINE RESPONSE DURING A X I A L P U L L O U T 250 6.2.1 Response of bare pipes 250 ix 6.2.2 Response of geosynthetic-wrapped pipes 252 6.3 FINDINGS F R O M P I P E L I N E RESPONSE DURING L A T E R A L P U L L I N G 252 6.3.1 Response of pipe buried in dry sand 252 6.3.2 Response of pipe buried in trench configuration with hard "native soil" 253 6.3.3 Response of pipe buried in trench configuration with surface covered with dual-geotextile layers 254 6.4 R E C O M M E N D A T I O N S F O R F U T U R E R E S E A R C H 255 R E F E R E N C E S 258 A A P P E N D I X A 274 A . l D I R E C T S H E A R TESTS O N M E D I U M DENSE F R A S E R R I V E R SAND 275 A . 2 D I R E C T S H E A R TESTS O N I N T E R F A C E O F M E D I U M D E N S E F R A S E R R I V E R SAND A N D S A N D - B L A S T E D S T E E L . . . . . 280 A . 3 D I R E C T S H E A R TESTS O N I N T E R F A C E O F 2 L A Y E R S O F F I L T E R W E A V E M I R A F I 700 W O V E N G E O T E X T I L E 285 A . 4 T R I A X I A L TESTS O N F R A S E R R I V E R SAND 288 B A P P E N D I X B 297 B. l R E S U L T S O F A X I A L P U L L O U T TESTS 298 B. 1.1 Axial load vs. displacement response, tests on bare pipe 299 B. 1.2 Pressure transducer measurements on bare pipe tests 302 B. 1.3 Pattern of sand particles movements around the pipe after pullout 304 B.l.4 Axial load vs. displacement, tests on wrapped pipe 306 B. 1.5 Displacement of geosynthetic layers wrapped around the pipe 308 B.2 R E S U L T S O F L A T E R A L P U L L I N G TESTS 310 B.2.1 Tests on pipe buried in cohesionless soil: lateral load vs. displacement response, 311 B.2.2 Tests on pipe buried in cohesionless soil: pressure transducers measurements (normal stress) 315 x B.2.3 Tests on pipe buried in cohesionless soil: observations after the test (surface deformation and pipe position) 316 B.2.4 Tests on pipe buried in trench configuration constructed in cohesionless native soil: lateral load vs. displacement response 318 B.2.5 Tests on pipe buried in trench configuration constructed in cohesionless native soil: Geotextile layers displacement 319 B.2.6 Tests on pipe buried in trench configuration constructed in cohesionless native soil: surface deformation and pipe position 320 B.2.7 Tests on pipe buried in trench configuration constructed in hard native soil: load vs. displacement response 321 B.2.8 Tests on pipe buried in trench configuration constructed in hard native soil: measurements of pressure transducers 324 B.2.9 Tests on pipe buried in trench configuration constructed in hard native soil: movement of geotextile layers 327 B.2.10 Tests on pipe buried in trench configuration constructed in hard native soil: surface deformation and pipe position 328 xi LIST OF TABLES Table 3-1: Identification of testing equipment 42 Table 3-2: Summary of performed tests and different measurements in each test 59 Table 3-3: List of performed axial pullout tests 60 Table 3-4: list of performed lateral pullout tests 69 Table 3-5: Summary of friction angles from laboratory direct shear testing of this study 79 Table 3-6: Summary of density and moisture content measurements 92 Table 4-1: Summary of parameters in axial pullout testing 96 Table 4-2: Comparison of K values back-calculated from axial load vs. soil pressure measurements 127 Table 5-1: Summary of parameters in lateral pulling tests 159 Table 5-2: Rf values based on Triaxial tests results 213 Table 5-3: Variation of soil loads due to pipe and content weight 234 Table 5-4: Effect of constant volume friction angle on the soil loads on pipe 236 xii LIST OF FIGURES Figure 1-1: Modes of pipeline restraints 2 Figure 1-2: Soil loads on pipeline(s) passing through an area subject to landsliding 4 Figure 1-3: Soil loads on pipeline crosses a strike slip fault with normal displacement component 4 Figure 1-4: Proposed method for reduction of axial soil loads on buried pipe 7 Figure 1-5: Introduced method for reduction of transverse soil loads on buried pipes 8 Figure 2-1: Prediction of horizontal bearing capacity on buried pipes; developed by Audibert and Nyman (1977) from method suggested by Hansen (1961) 30 Figure 2-2: Comparison between assumption of Hansen (1961) method (a) and Ovesen (1964) method (b); (Figure adopted from Trautmann and O'Rourke, 1983) 31 Figure 2-3: Prediction of horizontal bearing capacity on buried pipes; developed by Trautmann and O'Rourke (1983) from method suggested by Ovesen (1964) 32 Figure 2-4: Prediction of horizontal bearing capacity on buried pipes in deep embedment; developed by Yimsiri et al (2004) 35 Figure 2-5: Prediction of horizontal bearing capacity of buried pipes is dry and moist sand; developed by Turner (2004) 36 Figure 3-1: Plan view of testing equipment layout: Axial pullout tests (System 1) 41 Figure 3-2: Plan view of testing equipment layout: Lateral pullout tests (System 2) 41 Figure 3-3: Testing chamber 43 Figure 3-4: Testing configuration for lateral pullout (left) and axial pullout (right) tests 44 xiii Figure 3-5: Schematic figure of control system 50 Figure 3-6: Coupling system for axial pullout tests 51 Figure 3-7: Coupling system in lateral pullout tests 52 Figure 3-8: Pressure Transducers mounted on pipe 55 Figure 3-9: Displacement rate for Tests No. AB-2 and AB-3 61 Figure 3-10: Displacement rate for Tests No. AB-4, AB-5 and AB-6 62 Figure 3-11: Displacement rate of wrapped pipe tests 63 Figure 3-12: Schematic figure showing cigar-wrapping (top) and spiral-wrapping (bottom) methods for geotextile-wrapped pipes 66 Figure 3-13: Scaled schematic drawing of the pipe and trench position 68 Figure 3-14: procedure of constructing a trench in native soil 71 Figure 3-15: Grain size distribution for Fraser River sand, before and after the testing program 74 Figure 3-16: Results of Direct shear testing on Fraser River sand (initial density of 1575 kg/m3) 76 Figure 3-17: Triaxial rest setup for testing dry Fraser River sand under relatively low confining stress levels (< 50 kPa) 80 Figure 3-18: Peak friction angle calculated from triaxial test results 81 Figure 3-19: Calculation of initial elastic modulus using triaxial test results (test with D r = 75% and o' 3 = 50 kPa) 82 Figure 3-20: Initial elastic modulus calculated from triaxial tests results 83 Figure 3-21: Side walls covered with stainless steel sheets during lateral pullout tests 85 Figure 3-22: The schematic effect of side wall friction during lateral pullout tests (H/D=2.5 and D=457 mm, Tests No. LN-1 and LN-2) 86 Figure 3-23: Zones of displacements during lateral horizontal pipe pullout as identified by Audibert and Nyman (1977) and Trautmann and O'Rourke (1983) 88 Figure 3-24: Emptying the sand bags into the box 91 Figure 3-25: Density measurement distribution in axial pullout tests 93 Figure 3-26: Density measurement distribution in lateral pullout tests 93 xiv Figure 4-1: Load displacement response, Tests No. AB-3, AB-4 and AB-6, first loading 98 Figure 4-2: Load displacement response, Tests No. AB-2, AB-3, AB-4 and AB-6, during subsequent loadings after first loading/unloading 99 Figure 4-3: Load displacement response, Test No. AB-5 and reverse loading part of Test No. AB2 99 Figure 4-4: Radial position of pressure transducers 100 Figure 4-5: Measured soil stress normal to the pipe at transducer PT5 101 Figure 4-6: Measured soil stress normal to the pipe at transducer PT3 102 Figure 4-7: Measured soil stress normal to the pipe at transducers PT1 and PT2 103 Figure 4-8: Dimensionless normal stress on the pipe during Test No. AB-4 105 Figure 4-9: Dimensionless normal stress on the pipe during Test No. AB-5 105 Figure 4-10: Dimensionless normal stress on the pipe during Test No. AB-6 106 Figure 4-11: Forming strips of colored to observe particle movements 108 Figure 4-12: Colored sand patterns at the pipe centreline level prior to placement of remaining backfill during specimen preparation 108 Figure 4-13: Sand particles displacement after pipe pullout 109 Figure 4-14: Measurement of sand particles movement in Test No. AB-4 110 Figure 4-15: Load displacement response of Tests No. AGGC-1 , AGGC-2 , and AGGS-1 112 Figure 4-16: Load displacement response: Reverse loading of Tests No. AGGC-1 and AGGC-2 113 Figure 4-17: Load displacement response of Test No. ARGC-1 114 Figure 4-18: Geotextile layers displacement; Tests No. AGGC-1 and AGGC-2 115 Figure 4-19: Geosynthetic layers displacement; Test No. AGGS-1 115 Figure 4-20: Geosynthetic layers displacement; Test No. AGGS-1 116 Figure 4-21: Geonet material wrapped around the pipe in Test No. ARGC-1 117 Figure 4-22: Comparison of test results with predictions using A S C E (1984) formula - pipe in dense sand 119 Figure 4-23: Comparison of test results with predictions using A S C E (1984) formula - pipe in loose sand 120 xv Figure 4-24: Comparison of test results on pipe in dense sand during reverse loading with those on pipe in loose sand and predictions using A S C E (1984) formula , 120 Figure 4-25: Comparison of measured axial pullout loads and predicted loads using different K values 122 Figure 4-26: Normal stress on the pipe: average of Tests No. AB-4 and AB-6 (tests on dense sand) 123 Figure 4-27: Normalized soil stress G ' N on the pipe just prior to and during axial pullout in Tests No. AB-4 and AB-6 (Computed using stress measurements) 123 Figure 4-28: Normalized soil stress O ' N on the pipe: test AB-5 loose sand 125 Figure 4-29: Normalized normal stress on pipe during test AB-5 126 Figure 4-30: Comparison of load displacement curve of Tests No. AGGC-1 , AGGC-2, and AGGS-1 with predictions using A S C E (1984) formula and tests on bare pipe 129 Figure 4-31: Modeling the effect of dilation on soil loads on pipe 132 Figure 4-32: model geometry and mesh size for modeling of axial pullout tests 134 Figure 4-33: Particle movement at pipe springline level in axial pullout test on a PE pipe 137 Figure 4-34: (a) Horizontal and (b) vertical stresses contours prior to expanding the pipe 139 Figure 4-35: "at-rest" stress distribution around the pipe; comparison of numerical model results and pressure measurements via transducers 140 Figure 4-36: (a) Horizontal and (b) vertical stresses contours after 1 mm expansion of the pipe 142 Figure 4-37: Stress distribution around the pipe during pullout; comparison of numerical model results and pressure measurements 143 Figure 4-38: Horizontal and Vertical stresses contours after 0.92mm expansion of the pipe (box size: 6m W x 2.3m H) 144 Figure 4-39: Computed normal stress on the pipe for different model configurations 145 xvi Figure 4-40: Calculating of equivalent K value based on the numerical model results 146 Figure 4-41: Variation of K value with burial depth, D = 0.46 m 148 Figure 4-42: Variation of K value with burial depth, D = 0.23 m 149 Figure 4-43: Variation of K value with burial depth, D = 0.92 m 150 Figure 4-44: K value as a function of dilation effect and pipe size 150 Figure 4-45: Variation of K value with elastic modulus 151 Figure 4-46: Variation of K value with friction angle 153 Figure 4-47: Summary of axial pullout tests; Moving average and averaging of different tests are applied. Starting points of unloading / reloading parts are shifted 155 Figure 5-1: Load-displacement response - Tests No. LN-1 and LN-2 (H/D=1.92; D=457mm) 161 Figure 5-2: Load-displacement response, Tests No. LN-3, LN-4 and LN-7 (H/D=1.92; D=324mm) 162 Figure 5-3: Load-displacement response, Tests No. LN-1 to LN-7 163 Figure 5-4: Dimensionless stresses around the pipe during Test No. LN-7 (moving average is applied) 164 Figure 5-5: Surface deformation after Test No. LN-4 166 Figure 5-6: Surface deformation and pipe position after Tests No. LN-4 and LN-6 167 Figure 5-7: Typical dual-geotextile-lined trench-configuration for reduction of soil loads due to transverse ground movements 168 Figure 5-8: Load displacement response; Tests No. LNG-1 and LNG-2 169 Figure 5-9: Movement of geotextile layers during Tests No. LNG-1 and LNG-2 170 Figure 5-10: Surface deformation and pipe position after Tests No. LNG-2 and LN-2 171 Figure 5-11: Surface deformation after Test No. LN-4 172 Figure 5-12: Load displacement response; Tests No. LT-1 and LTG-1 173 Figure 5-13: Load displacement response; Tests No. LT-2, LT-3 and LTG-2 174 Figure 5-14: Dimensionless stress around the pipe during Test No. LT-1 (Moving averaging is applied) 176 xvii Figure 5-15: Geotextile layers movement in tests LTG-1 and LTG-2 177 Figure 5-16: Surface deformation and pipe position after pulling of the pipe: tests LT-1 and LTG-1 : 178 Figure 5-17: Surface deformation and pipe position after pulling of the pipe: tests LT-2 and LTG-2 179 Figure 5-18: Photo of surface deformation after pulling of the pipe: tests LTG-2 179 Figure 5-19: Comparison between the dimensionless peak lateral load ( F L ' ) obtained from the current study and published test results 183 Figure 5-20: Comparison of test results with ASCE (1984) / A L A (2001) Guidelines (H/D = 1.92; D =457 mm) 185 Figure 5-21: Comparison of test results with ASCE (1984) / A L A (2001) Guidelines (H/D = 1.92; D =324 mm) 186 Figure 5-22: Comparison of test results with ASCE (1984) / A L A (2001) Guidelines (H/D - 1.0; D = 324) 187 Figure 5-23: Comparison of test results with ASCE (1984) / ALA(2001) Guidelines (H/D = 2.75; D = 324 mm) 187 Figure 5-24: Comparison of N q r i from guidelines with F L ' from test results 188 Figure 5-25: Concept of calculating normalized horizontal soil stress component G ' N H at the locations PT2 and PT4 using stresses measured perpendicular to the pipe circumference 189 Figure 5-26: Average dimensionless horizontal stress component O ' N H at the locations of pressure transducers PT1, PT2, and PT3, during Test No. LN-7 190 Figure 5-27: Lateral soil stress distribution in front of the pipe in Test No. LN-7 191 Figure 5-28 Applied load on the pipe and passive wedge 193 Figure 5-29 Free body diagram of the passive wedge and pipe 195 Figure 5-30: Lateral soil stress based on pressure transducers reading in Test No. LT-1 197 Figure 5-31: Lateral soil stress based on pressure transducer reading in Tests No. L T - 2 & L T - 3 : 197 Figure 5-32: Lateral soil stress distribution in front of the pipe in Tests No. LT-1 to LT-3 198 xviii Figure 5-33: variation of average lateral soil stress in tests with trench configuration with pipe displacement 199 Figure 5-34: Lateral soil stress distribution in front of the pipe in tests with trench configuration 200 Figure 5-35: Mesh configuration for modeling Tests No. LN-1 and LN-2 202 Figure 5-36: Mesh configuration for modeling Tests No. LN-3, LN-4, and LN-7 202 Figure 5-37: Mesh configuration for modeling Tests No. LN-5 203 Figure 5-38: Mesh configuration for modeling Test No. LN-6 203 Figure 5-39: Initial elastic modulus calculated from triaxial tests results 206 Figure 5-40: Peak friction angle calculated from triaxial tests results 207 Figure 5-41: Simulate of Tests No. LN-1 and LN-2, using bilinear Mohr-Coulomb compared with test results 208 Figure 5-42: Numerical model result using hyperbolic Mohr-Coulomb model (with different failure ratios) versus test results (H/D=2.5, D=457 mm) 214 Figure 5-43: Numerical model result using modified hyperbolic Mohr-Coulomb model to account for densification; D = 324 mm and H/D = 1.92, Density = 1600 kg/m3 (D r=75%) 216 Figure 5-44: Numerical model result using modified hyperbolic Mohr-Coulomb model to account for densification; D = 324 mm and H/D = 1.92, Density = 1450 kg/m3 (D r=20%) 217 Figure 5-45: Density and friction angle contour for test in loose sand with initial density of 1450 kg/m 3 (D r = 20%) after 30 mm of pipe displacement (H/D=1.92, D=324 mm) 218 Figure 5-46: Soil loads on pipe in numerical modeling of box with different sizes 219 Figure 5-47: Minor principal stress contours after 30 mm of puling of the pipe, effect of box size 220 Figure 5-48: Comparison between numerical and experimental results; Configuration with H/D = 1.92 and D = 457 mm (Tests No. LN-1 and LN-2) 221 Figure 5-49: Comparison between numerical and experimental results; Configuration with H/D = 1.92 and D = 324 mm (Tests No. LN3, LN-4, and LN-7) 221 xix Figure 5-50: Comparison between numerical and experimental results; Configuration with H/D = 1 and D = 324 mm (Test No. LN-5) 222 Figure 5-51: Comparison between numerical and experimental results; Configuration with H/D = 2.75 and D = 324 mm (Test No. LN-6) 222 Figure 5-52: Prediction of dimensionless soil loads for different friction angle and overburden ratios 225 Figure 5-53: Variation of N q r i value with overburden ratio; Comparison of the results of current study with previous research 226 Figure 5-54: The effect of pipe diameter; Comparison of numerical model results with proposed equation 228 Figure 5-55: Comparison of the results of current study (numerical and experimental) with previous research after correction for the effect of pipe diameter 229 Figure 5-56: Numerical model results of test configuration (II) (H/D = 1.92 and D = 324 mm); Comparison of different surface roughness 230 Figure 5-57: Numerical model results of test configuration (III) (H/D = 1 and D = 324 mm); Comparison of different surface roughness 231 Figure 5-58: Numerical model results of test configuration (IV) (H/D = 2.75 and D = 324 mm); Comparison of different surface roughness 232 Figure 5-59: Suitability of the proposed equation for modification the effect of surface roughness; Comparison between proposed equation and numerical model results 233 Figure 5-60: Numerical model of pipe buried in trench configuration 238 Figure 5-61: Comparison on numerical model results and test results for pipe in trench 239 Figure 5-62: Displacement contours after 100 mm of pulling of the pipe, trench configuration with cohesive backfill material 241 Figure 5-63: The effect of apparent cohesion on soil loads on pipe, comparison between trench configuration and case with no trench 242 xx NOMENCLATURE C u coefficient of uniformity of sand D outside pipe diameter dso average grain size D r soil relative density F A axial force on pipe Ej initial elastic modulus E s e c secant elastic modulus E t a n tangent elastic modulus F A ' average interface shear force around the pipe normalized with respect to the vertical effective stress from the soil overburden at the centerline of the pipe F L horizontal lateral force on the pipe F L ' normalized lateral horizontal soil loads on the pipe with respect to the overburden stress at the centreline of the pipe and pipe diameter Gj initial shear modulus G s e c secant shear modulus Gtan tangent shear modulus H depth from the ground surface to the centerline of a buried pipeline K equivalent lateral earth pressure coefficient (other than at-rest condition) Ko coefficient of lateral earth pressure "at rest" L length of pipe test specimen xxi Nqh horizontal bearing capacity factor for sand = F L ' P a atmospheric pressure P u maximum horizontal soil resistance per unit length of pipe Rf failure ratio t pipe wall thickness Y measured horizontal pipe displacement Y ' dimensionless pipe lateral displacement (Y/D) Y u relative displacement between pipe and soil in horizontal direction necessary to develop Pu 5 interface friction angle 9 soil internal friction angle y total unit weight of soil y soil dilation angle Oh horizontal soil stress at pressure transducer location cr,' effective lateral stress ho <j'N normalized normal stress at pressure transducer location C ' N H horizontal soil stress at pressure transducer location normalized with respect to the vertical overburden effective stress at the transducer location (o"NH)eq equivalent uniform distribution of ONH a'n averaged pressures at a given transducer location a'v computed effective overburden stress at the pipe springline cr' effective vertical overburden stress vrj oV minor principal stress if shear stress at failure xxii ACKNOWLEDGEMENTS I would like to express my sincere appreciation to those without their support and assistance, the completion of this research would not have been possible. Above all, I would like to thank my supervisor, Dr. Dharma Wijewickreme for his unconditional support. He provided me the opportunity to freely choose my path, while sharing his invaluable experience and knowledge as well as his precious time. Without his patience and positive attitude, this project could not have been completed. Throughout my years in U B C , Dr. Carlos Ventura has always been a great support. His encouragement, confidence, and extensive knowledge helped me in many ways. I would like to thank Mr. Douglas Honegger for sharing his decades of experience in pipeline engineering. His endless efforts and precious feedback was a great help through the research program. Financial support provided by Pipeline Research Council International, Inc. (Contract PR-268-03111) is gratefully acknowledged. I would also like to thank Terasen Gas for providing pipe specimens. My thanks are also extended to technical staff of the Department of Civi l Engineering. These include Mr. Harald Shrempp, Doug Smith, Bi l l Leung, Doug Hudniuk, Max Nazar, Scott Jackson, and John Wong. During the research program, I was grateful to get assistance from some of my fellow graduate students and undergraduate students. I would like to thank Mr. Lalinda Weerasekara, Vahis Sofali, Tyler Lappin, Ziad Boustani, Andrew Chand, Mingi Song and Nima Nabavi. My appreciation goes also to Mr. Chris Anderson for sharing his experiences in full-scale testing of buried pipes. xxiii CHAPTER 1 INTRODUCTION 1.1 Background Buried pipeline systems (e.g., oil and gas pipelines, water distribution networks and sewage systems) form a key part of our global lifeline infrastructure, and any significant disruption to the performance of these systems often translates into undesirable impacts on businesses, economies, and/or the living conditions of citizens. Although sheltered from exposure to atmospheric elements or other above-ground hazards, one of the major risks to buried pipelines arises from landslides and earthquake-induced ground movements. Especially due to the continually increasing demand for energy, the performance of oil and gas pipelines is currently receiving particular attention in this regard. When a pipeline route passes through an area where permanent ground displacement hazard is identified, it is subjected to potentially high soil loads that may lead to unacceptably high strains in pipe sections. Although it is desirable to avoid areas subject to potential ground movement, due to land availability, environmental constraints, etc., it is sometimes inevitable that major pipelines need to be routed through such vulnerable areas. Permanent ground displacement might result from creeping ground, landslides, slope instability, and earthquake. Earthquakes may cause ground settlement due to 1 liquefaction, triggered landslide movements, lateral spreading, and surface fault movement. Based on the direction of differential displacement between soil and pipe, buried pipeline restraints against soil movement can be categorized into four different modes as shown in Figure 1-1: (a) vertical-uplift; (b) vertical-bearing; (c), horizontal-lateral.; and (d) longitudinal-axial (ASCE 1984). Any differential movement can be identified as one or a combination of these modes. Pipeline restraints may result in bending, shear, tension or compression in a given pipe section which, in turn, can lead to unacceptable strains and potential loss of pressure integrity of contents of the pipeline system. a: Vertical uplift c: Horizontal lateral Figure 1-1: Modes of pipeline restraints The oil and gas pipelines are usually buried with a soil cover of 0.6 to 1.5 m (2 to 5 feet). The overburden ratio (defined as H/D where D = diameter of pipe, H = soil cover above springline of pipe) for these types of pipelines varies in the range of 1 to 5. The critical soil loads on pipes with this level of soil cover typically arise from transverse, vertical bearing or axial loads imparted from lateral ground movements. Understanding of soil d: Longitudinal axial <>—0-2 loads during each of the pipe restraints provides the ability to design a pipeline system to withstand anticipated ground movements. Different modes of pipeline restraints occur in various geological circumstances and depend on the orientation of the pipeline with respect to the fault direction or landslide movement, as well as the direction of the fault slippage and ground deformation. Figure 1-2 presents an example of two conditions that result in mobilization of soil loads on a buried pipe. The modes of restraints, stress, and loads in pipe sections are functions of orientation of the pipe with respect to the ground movement. When the pipe direction is aligned with the direction of the landslide, the pipe is subjected to longitudinal-axial restraint and the different sections of the pipe will experience either tension or compression stresses. If the pipe direction is normal to the landslide direction, the soil mass applies lateral load to the pipe which results in bending moment and shear loads on different sections of the pipe. In the case of ground settlement (vertical ground deformation), vertical uplift or vertical bearing load will also be applied to the pipes; these also can result in bending moment and shear stress on the pipe in a vertical plane. In essence, a pipe subjected to landslide might experience one or a combination of soil loading conditions presented above. Another example of ground movement is shown in Figure 1-3, where a pipe crosses a strike slip fault with normal displacement component. The direction of the pipe and the angle in which pipe crosses the fault are key parameters in identifying pipe restraints. For a pipe crosses the fault in a normal angle, it is mainly subjected to lateral horizontal load. On the other hand, if the pipe crosses the fault in a low angle, the axial load on pipe is dominant. The direction of the fault movement decides whether the pipe section is in tension or compression. In both cases, if the fault movement has a normal component, pipe will also be subjected to the vertical uplift and bearing soil loads. In Figure 1-3, the pipe is subjected to all four restraints and the magnitude of each soil load component is a function of fault movement direction and crossing angle. 3 Figure 1-2: Soil loads on pipeline(s) passing through an area subject to landsliding 4 It is believed that the deformations and stresses on pipelines as a result of permanent ground movement are higher than those associated with displacement from transient ground shaking (Trautmann and O'Rourke 1983); hence, in most instances, designing a pipeline against permanent ground deformation would automatically provide resistance against transient ground shaking. Due to the large number of variables, full-scale testing provides a meaningful approach to understanding the pipe-soil interaction problem and obtaining reliable information. Tests in smaller scales are subjected to errors associated with scaling. Moreover, although numerical models and analytical approaches can help in understanding pipe/soil interface behaviour during ground movement in the field, physical models of the field situations are required to calibrate the results of computer models and validate analytical approaches and numerical models. Physical model tests have been mainly performed (mostly by private companies) for specific uses and the results are either not published and documented, or cannot be generalized to other cases. The other reason for the scarcity of information is the large costs associated with full-scale testing. Cornell University is one of the pioneers in buried pipeline research and physical testing. Research testing, specially lateral and uplift pullout, conducted during the past decades at Cornell University has contributed to the development of A S C E Guidelines for the Seismic Design of Oil and Gas Pipeline Systems (1984). In the past 10 years, some full-scale tests on buried pipelines focusing more on axial and lateral pullout tests has also been performed at the Center for Cold Oceans Resources Engineering (C-CORE) at the Memorial University of Newfoundland, St. John's, Newfoundland , Canada. In spite of these contributions, as discussed in Chapter 2 - Literature Review, there are a number of areas that require extensive research to advance our fundamental understanding of the pipe-soil interaction problem. These required areas of research, including evaluation of axial soil loads on pipe buried in sand backfill with various states and clarifying the significant differences in the suggested lateral soil loads on buried pipes in the literature provided the stimulus for the current research work. 5 1.2 Soil load reduction methods Often, substantial reduction in pipeline strains can be achieved if the soil load generated on buried pipelines from ground displacement can be reduced. Past approaches to this problem have largely focused on four general strategies applied individually or in combination: • Modify the pipeline alignment to provide a more beneficial orientation of the pipeline with respect to the direction of permanent ground displacement. The viability of this strategy is often limited by local topography or right-of-way restrictions. • Use a low friction pipeline coating, such as fusion bonded epoxy, to reduce axial soil loads. This strategy is rarely practical for assessing existing pipelines and is primarily an advantage when cohesionless backfill soils are present. • Modify the pipeline trench and backfill to achieve reduced soil resistance. Implementation of this strategy typically includes reducing the depth of soil cover and using a trapezoidal trench with selected backfill material. • Increase the ability of the pipeline to resist soil loads by specifying greater wall thickness or a higher grade of pipe steel. This strategy is typically limited to new design unless other factors (e.g., corrosion, class change) necessitate replacing or relocating an existing pipeline. Geosynthetic materials are increasingly used in current practice to reduce soil loads on pipelines: (a) wrapping of pipeline with two layers of geosynthetic material as shown in Figure 1-4 to reduce axial soil loads; and (b) dual-geotextile-lined trench as shown in Figure 1-5 to reduce lateral soil loads. This is particularly true for existing pipelines 6 where other alternatives including pipeline replacement or realignment are not practical. The design of above geosynthetic-fabric-based solutions are largely based upon the interface frictional properties of the geosynthetic fabrics and assumptions regarding the soil failure mechanisms at the geosynthetic fabric interface. Except discrete field tests for special purposes, no controlled test has been undertaken to quantify the efficiency of methods and to understand the fundamental failure mechanism. In the absence of a set of recommended practices regarding the application of these materials, implementation of design concepts is largely left to the construction contractor. 1 S T layer of Geosynthetic 2 n d layer of Geosynthetic Figure 1-4: Proposed method for reduction of axial soil loads on buried pipe 7 Figure 1-5: Introduced method for reduction of transverse soil loads on buried pipes 1.3 Objectives of the thesis The main goal of this study included fundamental understanding of the soil loads on relatively rigid steel pipelines subjected to horizontal ground movements (i.e., resulting in transverse and longitudinal soil loads on pipe). With this in mind a series of full-scale testing accompanied with numerical modeling was conducted at the University of British Columbia, to investigate the above pipe-soil interaction response, and this work is described in this thesis. In particular, the specific objectives described herein are as follows: i) to examine the response of rigid steel pipes subjected to lateral and axial permanent ground displacement using physical modeling, including direct measurement of the distribution of normal stress on pipe surface under "at-rest" conditions and during lateral movement; 8 ii) to study the effect of soil parameters and especially stress-dilatancy effects on the response of buried pipes during axial soil movement using data from physical modeling supported by numerical analysis; iii) to provide information for all problems where soil/structure interface is an issue. Study on variation of normal stress due to the soil dilation provides first hand information that is applicable to all buried structures. Finally, measurement of the stress on the pipe surface, leads to a new picture of stress distribution around a curved surface. iv) to observe and, where possible, quantify the effect of key parameters that govern axial and lateral soil loads; v) to review the applicability of the current approaches for pipeline design in light of the test results of the current study; vi) to examine several currently utilized, or proposed, methods for reduction of soil loads on pipelines during horizontal-lateral and longitudinal-axial differential displacements. 1.4 Scope of the thesis To accomplish the research objectives as mentioned above, the scope of work conducted throughout the research program and discussed in this thesis is as follows: • Modification of an existing soil chamber for pipe-soil testing of small-diameter polyethylene pipes to accommodate testing of large diameter steel pipes in both axial and transverse directions. This also included the development of a new loading mechanism and control system. 9 Development of an instrumentation array to observe pullout resistance, pipe and geosynthetic layers displacement, and normal soil stresses on pipe surface in real time, and to monitor sand particle displacement and surface deformation after the tests. Conducting a series of axial pullout tests on bare pipes buried in sand with different density levels to provide first hand data on the effect of soil density and dilatancy on axial soil loads on pipe. The testing program also included pullout tests on geosynthetic-wrapped pipes to investigate the efficiency of soil load reduction method. Performing transverse pulling tests on pipes buried in native soil and also in trapezoidal trench configuration, using different backfill materials to evaluate the effect of trench backfill and native soil conditions on lateral soil loads on the pipe. The program also included tests on pipes buried in geotextile-lined trenches to observe the efficiency of proposed method to reduce lateral soil loads on the pipe. Development of numerical models to capture full-scale testing results during both axial and lateral pullout tests. This included calibration and validation of the model, as well as generalization of the findings from the physical modelling to other conditions. Comparison of the results from the physical testing and numerical modeling with the currently used methods for prediction of soil loads on buried pipes subjected to ground movements and providing suggestions to improve current approaches of buried pipeline design. 10 1.5 Organization This thesis is organized into 6 chapters and 3 appendices. Chapter 1: It is this chapter, which provides the introduction to this thesis. Chapter 2: In this chapter, a review of the literature on the soil-pipeline interaction problem, with particular emphasis on the performance of buried pipelines subject to axial and lateral movements is presented. The chapter presents past research work and findings in terms of both physical and numerical modelling. Chapter 3: This chapter describes the development of a full-scale physical model testing facility (2.5 m x 5.0 m pipe-soil testing chamber) to conduct pullout tests on relatively large diameter steel pipelines. The equipment used, setup, methodology, measurements, and limitations of the experimental setup are presented in detail. The results from laboratory element testing to characterize material properties for numerical modeling are also provided. Chapter 4: The results of axial pullout tests on bare pipes as well as those from cases with soil load reduction methods are presented. Common approaches to estimate soil loads on pipes are compared with test results. Numerical modeling of axial pullout tests and a review of the effective parameters on axial soil loads on pipe are included. Recommendations are made on prediction of soil loads on pipe during axial ground deformation. Chapter 5: In this chapter, the results from lateral pullout tests on pipe: in native soil, in different trench configurations, and with proposed soil load reduction methods implemented are presented. The results are compared with commonly used approaches and previous studies. The efficiency of the proposed load reduction method is also discussed. A study on the parameters affecting soil loads is undertaken using a numerical 11 model calibrated using data from physical modeling. Some modifications to the available methods for prediction of lateral soil loads are recommended. Chapter 6: This chapter presents a summary and conclusion of the current research. Recommendations for design of buried pipelines are also provided. Suggestions for future work in this area are made. Appendix A : The results from element testing performed to characterize the soil and interface parameters are presented. Appendix B: A l l the results from full-scale axial and lateral pullout tests are presented. 12 CHAPTER 2 LITERATURE REVIEW In this chapter, previous research on pipelines subject to ground movement with particular reference to lateral and axial force-displacement response is presented. In addition to presenting the findings from experimental and analytical studies on this subject, some relevant information from topics (e.g., vertical anchors and pile foundations) that have contributed to understanding the soil-pipe interaction problem has also been presented. Under real-life conditions, pipelines are placed in trenches excavated in native soils; in a general context, the trench backfill may or may not necessarily be the same as the native soil. Most of the previous research has been focused on the fundamental behaviour of bare pipes buried in native soil while only a few studies on the response of pipelines buried with trench backfill different to native soil can be found. This chapter also includes a summary of previous studies on soil load response to permanent ground displacement (PGD) on pipes buried in trenches. Some details related to limited research on the currently used methods for reduction of soil loads are also presented. The guidelines presently used in the current design practice for the computation of soil loads and pipe-soil interaction effects are also summarized as a part of the chapter. 13 2.1 Response of buried pipes subject to axial loading 2.1.1 Longitudinal soil loads on pipes Current approaches to evaluate axial soil loads on pipelines during ground movement are essentially based on interface parameters and initial average normal effective soil stress levels on the pipe. An average normal soil load on the pipe in at rest conditions is often estimated using Equation [2-1] below. K L = ( ^ ) x r ^ P-i] Where: (o"'n)flV = average normal soil loads on the pipe in at rest conditions H = height of the soil over pipe springline Ko = coefficient of lateral earth pressure at rest And y = soil density (NOTE: select y so that yH would give rise to vertical effective soil stress at pipe springline level) Then based on the assumption that the failure mechanism would be formed at the pipe-soil interface, and using average normal stress from Equation [2-1], axial soil loads on pipes per unit length of the pipe can be estimated from Equation [2-2] presented here. 1 + K Fa = nx Dx(a'n )av x tan5 = nx Dx (—^—9-)xyxHx tan5 [2-2] Where: D = pipe nominal diameter and 8 = interface friction angle 14 Data for interface friction angle between soil and pipe surface , 5, for different materials have been collected and summarized by Brumunds and Leonards (1973) and Kulhawy and Peterson (1979). Kulhawy et al. (1983) later suggested that the friction angle between sand and smooth steel and sand and rough steel varies in the range of 0.5(p to 0.7cp and 0.7(p to 1 -Ocp, respectively, where cp = internal friction angle of sand. This formula has been in wide use in the design of pipeline systems. Newmark and Hall (1975) used this equation to calculate axial soil loads on pipes subjected to strike slip fault movement. To approximate the average radial pressure on a pipe subjected to large axial strains, Kennedy et al. (1977) used this formula assuming mobilization of frictional resistance at the pipe surface. Moreover, this equation has been recommended for the computation of axial soil loads in ASCE (1984) "Guidelines for the Seismic Design of Oil and Gas Pipeline Systems", American Lifeline Alliance (2001) "Guidelines for the Design of Buried Steel Pipe", and PRCI (2004) "Guidelines for the Seismic Design and Assessment of Natural Gas and Liquid Hydrocarbon Pipelines". There are a few other formulae that have been proposed for the estimation of axial soil loads. McAllister (2001) in his handbook has suggested using Equation [2-3] for estimating axial soil loads on pipes. Fa = {2.D.y.(H -®) + Wp}xtan5 [2-3] where W p is the weight of pipe and content per unit length. Danish Submarine Pipeline Guidelines (1985) suggested using of Equation [2-4] to estimate maximum possible axial loads on pipe. This equation was derived by integrating shear stresses at interface around the pipe. 15 z, = v n D A W v n ^tan<p.(H + ^)Ml + K0) + -.^X2 + K0)-±—(2 + K0) 2 2 3 K 5 .tan cp [2-4] Paulin et al. (1998) conducted full-scale testing on a 324 mm steel pipe buried in loose and dense sand. Comparison of results between tests indicates that the axial load on pipe in loose sand is lower, and in dense sand is much higher, than those predicted by Equation [2-2]. The axial pullout force on buried pipe in loose sand is about 30% to 50% of that in dense sand. The test results, however, are not published in detail, and, therefore, cannot be compared with other studies in a meaningful manner. Cappelletto and co-workers (1998) performed a series of field tests on pipes, having diameters of 200 mm (8") and 600 mm (24"), mainly buried in cohesive material. After examining their results, they indicated that for pipes subject to slow rate of longitudinal ground deformation, use of material cohesion (Su) to predict axial soil loads on pipe, overestimates the load. Knowing that a very small area around the pipe is sheared during axial displacement, they suggested that in the mentioned situation, using an effective stress model provides more reasonable values of soil loads than a total stress model. In other word assuming undrained behaviour of clay for these conditions is not rational; instead it is more appropriate to use Equation [2-2] which is suggested for axial loads on pipes buried in granular material. After Northridge earthquake, Honegger (1999) performed a series of field tests on pipeline buried in desiccated clayey soil. He reported that repeated loading yielded lower axial force than virgin loading. Anderson (2003, 2004) conducted a series of lab testing on buried HDPE pipes in sand. The result of these tests also indicated that the simplified formula suggested by ASCE (1984), A L A (2001), and PRCI (2004) did not adequately predict the axial soil loads on pipe. His results were in agreement with those reported by Paulin et al. (1998), indicating axial soil loads on pipes buried in dense sand were much higher than those predicted by suggested formula, yet no explanation was presented. 16 Singhal (1980) after a series of tests on buried pipelines concluded that the maximum load on pipe during axial ground deformation is mobilized after 0.1 to 0.2 inches of differential displacement. Holloway (1976) suggested that the load-displacement relationship can be approximated by a hyperbolic relationship or a simplified bilinear relation. 2.1.1.1 Tests on wrapped pipes Kulhawy et al. (1983) indicated that wrapping the pipe (with one layer) may result in increase in resistance and especially when the wrapping material is soft; interface friction can be as high as the internal angle of friction of soils (cp) since soil particles could become embedded in the wrapping material causing shear to take place within the soil mass. Uncoated steel pipes which have been buried for a long time are subjected to oxidization and during that process sand particles become cemented around the pipe. In this condition, interface friction angle can also increase up to internal friction of sand. O'Rourke (1994) proposed a method for reducing soil loads on pipelines during differential axial displacement. The method consisted of wrapping pipe with 2 layers of geosynthetic materials to take advantage of lower interface friction angle of layers. Nova Gas Transmission Ltd. (1994) performed a series of field tests on a 40 cm diameter steel pipe. The intent was to examine the efficiency of the method proposed by O'Rourke (1994). The tests were performed using different geosynthetic materials as pipe wrapping. Overburden ratio (H/D) of all the field tests conducted by Nova was approximately 3, and instrumentation was limited to the measurement of only the load and displacement of the pipe. Test results indicated that using proper wrapping material might decrease the load on pipe to 40% of that on bare pipe. Current pipeline projects in the United States and overseas are being designed using geosynthetic fabrics as a means to reduce soil loads based mainly upon the interface frictional properties of the geosynthetic fabrics and assumptions regarding the soil failure 17 mechanisms at the geosynthetic fabric interface. In the absence of a set of recommended practices regarding the application of these materials, implementation of design concepts is largely left to the construction contractor. 2.1.2 Relevance to shaft friction in axially loaded piles As noted from the equations presented in the previous section, interface friction angle and normal stress on the pipe are the key parameters that govern the axial soil loads during ground movement. This subject is not well-addressed in the field of pipelines, yet has been largely investigated in research on pile foundations. Clear differences that exist between buried pipes and piles (i.e., in relation to configurations, methods of installation, etc.) would not allow direct use of the approaches developed for axially loaded piles to be directly adopted for pipeline problems; however, already understood effects of soil density and dilation in the development of shaft friction in axially loaded piles becomes useful in assessing the mobilization of axial soil loads on pipes. With this background, it was judged reasonable to provide the brief overview below on relevant research findings from axially loaded piles. Nauroy et al. (1983) performed tests on piles driven into sand to measure the axial capacity of piles, for different types of sand with differing compressibility, and different friction angles. They observed that frictional force contributed to axial loads pn piles in medium dense siliceous sand were more than that of piles in loose calcareous sand, although the loose sand had a higher friction angle than the dense siliceous sand. These results of the tests were later used by Kraft (1990, 1991) in a study on performance of axially loaded piles in sand. He presented a series of methods for calculating axial soil loads on piles in sand. He suggested that the sand shows more dilatancy at lower depths, hence variations in lateral soil pressures at areas close to the surface due to the axial loads on piles were more noticeable. Since pipes are usually buried at relatively shallow depths, it can be inferred from the above results that the effect of dilatancy on pipes during axial displacement could also be significant. 18 Lehane et al. (1993) conducted a series of field tests on an instrumented pile. Again, the variation of radial effective stress on piles during axial loading was measured and observed to be more noticeable at shallow depths, in agreement with the findings of Leland and Kraft (1991). Lehane et al. related the increase in radial soil stress into dilation at interface and rotation of principal stress in the sand mass. Also, after examining the test results and comparing with analytical models, they found that the operational interface friction between pile and sand can be assumed to be the constant volume friction angle. These findings are of relevance to the assessment of buried pipes. Foray et al. (1993) performed a series of pile driving tests in a large sand chamber to investigate the effect of sand density on the skin friction, effect of direction of loading on skin friction, and effect of confining stress. They also investigated the effect of overconsolidation ratio on end bearing and skin friction. The results indicated that lateral earth pressure in overconsolidated sand can be much higher compared to normally consolidated sand. They observed that the skin friction force on the pile can be doubled and coefficient of lateral earth pressure can be as high as 1 in overconsolidated sand, Lehane and Jardine (1993, 1994, 1996) performed a thorough study on the behaviour of pipe piles driven into sands and considered the effect of dilation in increasing soil stresses on the shaft. They suggested that the normal stress is a combination of initial radial stress and increased stress due to dilatancy. Using the result of field tests, cavity expansion theory and laboratory test data, they presented a relation to calculate the variation in radial stress after axial displacement. Based on the suggested relation, increase in radial stress is proportional to soil relative density and square root of overburden stress. Also, it is proportional to 1/R, where R is radius of the shaft (i.e. this effect is more noticeable for smaller shafts). They also reported that the direction of the loading can affect radial stress due to the Poisson's effect. The suggested relation was in agreement with the field test results, and indicated that previous approaches underpredict radial stresses on piles. It is rational to expect this effect in axial loads on pipes especially for flexible pipes such as polyethylene and P V C pipelines. Randolph (1994) developed an analytical model and 19 presented a closed form solution to consider the phenomenon of increase in radial soil stress. The proposed relation by Jardine and Lehane (1996) was in agreement with this closed form solution as well. 2.1.3 Summary of key observations The following can be noted from the above presented review of previous research on axial soil loads on pipes and also on axially loaded piles: • Research on pipes buried in dense sand (compacted to different density levels) has shown that measured axial soil resistance in physical model tests is significantly higher than that predicted using simple formulae given in current design guidelines. • Research work to understand axial loads in piles have shown that dilation of soil around the immediate vicinity of pile during shear is one of the key reasons for increase in axial loads [could cause changes (increase) in normal stress distribution on the pile]. However, so far, no focused research in the way of experimental work or numerical modeling have been undertaken to explicitly understand high axial soil loads on pipes buried in dense sand. • The wrapping of pipelines using geosynthetic layer as a method of soil load reduction during axial ground deformation is used in practice, especially as a measure of safeguarding existing pipelines. Although discrete lab and field tests have been performed (mostly by private companies) for evaluating the efficiency of soil load reduction methods, no published data on controlled tests are available to confirm the effectiveness of these methods in a fundamental manner. • There is a strong need for further research to understand the development of axial soil loads on pipelines in a fundamental manner and then evaluate the currently used methods for axial soil load calculations. This would require carefully 20 conducted physical model tests under controlled conditions along with numerical modeling as appropriate. The redistribution of soil stress around the pipe, as a result of soil dilatancy, with axial pipe displacement seems to be one of the keys to developing this understanding. 2.2 Response of buried pipes subject to lateral loading Early approaches for evaluation of soil-pipe interaction during horizontal-lateral ground movement were mainly based on the experiments conducted to understand the response of retaining walls, laterally loaded piles, and vertical anchors. Initial analytical approaches to calculate soil loads on pipes were developed based on the active and passive soil resistance and simplified failure surfaces. The areas of research on horizontal capacity of vertical anchors and lateral soil loads on buried pipes are closely related to each other. Therefore, some of the methods currently used to evaluate soil loads on pipes have been derived from studies on anchors. For example, due to the similarity of failure surfaces, the experimental results on anchors were used directly, or after correction for aspect ratio and shape factor to estimate peak lateral soil loads on pipes. To the best of the author's knowledge, the first well-documented tests on pipes subject to ground movements were performed by Audibert and Nyman (1977), and understanding of pipe-soil interaction prior to that was limited to the tests on vertical anchors. In this section, the results from previous experimental and analytical approaches are initially presented (Sections of 2.2.1 and 2.2.2). This is followed by a summary (Section 2.2.3) leading to identification of the need to conduct the proposed research in this study. As may be noted, considering the relevance to initial understanding of response of buried pipelines, past work on anchors is also presented in Section 2.2.1, and included in the discussion. 21 2.2.1 E x p e r i m e n t a l research Hansen (1953) made early contributions to understanding of the soil-structure interaction problem by conducting experiments to model effects of passive lateral movements of retaining walls with loose sand backfill. With the main objective of recording patterns of failure and identifying failure mechanism, he tested a small-scale wall in a box with transparent windows and took photographs during wall displacement, but he did not measure the passive soil force on the wall. Later, he continued his studies by measuring passive soil force on laterally loaded piles in sand (Hansen, 1961) In continuation of the studies by Hansen, Ovesen (1964) performed a series of tests on 15 cm high plate anchors in loose and dense sand. The width of the anchor was selected long enough to represent a plane strain condition. He conducted tests for a range of overburden ratios from 1 to 10 and observed that this range of overburden ratios covers both shallow and deep failure mechanisms. The test results were used to develop an analytical model which was used to estimate passive soil loads on anchors. The results were summarized and presented later by Ovesen and Stromann (1972) along with an analytical model (which is presented in Section 2.2.2). Another series of tests on vertical anchors were performed by Kosteyukov (1967) to study the behaviour of deadman anchors on port structures. During the tests, he monitored changes in soil density in front of anchors buried in loose and dense sands using a gamma ray absorption technique. The results showed different pattern of density variation for different density levels. Unlike the case of dense sand in which the density of soil in front of the plate remained almost constant during pullout, the tests in loose sand indicated that the density of sand in a triangle in front of the anchor increased significantly during movement of the anchor. He also reported that the density of sand above the tip of vertical plate up to surface and along the failure surface remains almost constant for loose sand and decreases slightly for dense sand. The tests were conducted on both dry and saturated materials. Measurement of soil loads indicated that for similar 22 soil unit weights and overburden ratio equal to 2, anchor resistance in saturated sand in about 30% less than those in dry sand. Neely and co-workers (1973) performed a series of laboratory tests on vertical anchors buried in loose sand, with a reported friction angle of 35° for the sand. They compared their test results with those from tests performed by Smith (1962) on deadman anchors in loose sand, and also with available theoretical approaches. The experimental results showed good agreement with an analytical model based on Sokolovskii (1965) method of stress characterization. They also observed that for tests with high aspect ratios (ratio of the width to the height of the plate) of five, which can be assumed as plane strain conditions, displacement at failure varied in the range of 0.1 D to more than 0.2D, where D is the height of the anchor, for overburden ratios (depth of the bottom of the plate to the height of the plate) of one to five respectively. Das and Seeley (1975) investigated the effect of aspect ratio (ratio of the width to the height of the plate) on vertical anchor resistance against horizontal movement. They concluded that the dimensionless force-displacement curves can be approximated by a single rectangular hyperbola; this idea has been largely used by researchers in the development of "soil-springs" used in pipe/pile-soil interaction modeling. According to the authors, Equation [2-5] describes rectangular hyperbola relation between anchor resistance and displacement. j . L _ 0.15 + 0.857 [2-5] P = P/PU (P<P U ) and Y = Y/YU (Y <YU) where: P = lateral soil load on anchor Pu = lateral soil load on anchor at failure Y = horizontal displacement of anchor 23 Yu = horizontal displacement of anchor at failure. Moreover, through this work Das and Seeley (1975) observed that the load per unit width of the anchor decreased with increasing aspect ratio. This finding would imply that if the effect of aspect ratio is not taken into account, using the results of vertical anchors for buried pipes would lead to overprediction of lateral soil loads on pipes; in other words, this suggests the importance of ensuring appropriate geometric conditions in physical modeling that suitably represent plane strain conditions. The first series of tests on lateral soil loads on buried pipes was performed by Audibert and Nyman (1975, 1977) on steel pipes with outside diameters of 25 mm, 60 mm, and 114 mm. Soil behaviour during lateral displacement of pipes buried in sand under a wide range of overburden ratios from 1.5 to 24.5 was observed. The tests were performed in both loose and dense sand surroundings. Besides the laboratory testing program, one field test on a steel pipe with outside diameter of 230 mm was also performed to confirm the results of small scale tests and validate the physical model. Comparing their laboratory model test results, Audibert and Nyman (1977) observed good agreement between experimental results with recommendations by Hansen (1961) based on analytical works. Moreover, independent of the results of Das and Seely (1975), they derived a rectangular hyperbola relationship as shown in Equation [2-6] to fit the dimensionless load-displacement curves obtained from experimental results. It can be noted that their relation is in agreement with the relation reported by Das and Seeley. Audibert and Nyman (1977) also observed an increase in normalized displacement at failure with decrease in sand density and pipe size. It should be noted that the friction angle they were using for comparison of test results with Hansen (1961) method and in-situ test performed by themselves, were empirically based on the density of sand and not on specific laboratory element testing. P = = [2-6] 0.145 + 0.8557 24 After a series of tests on anchor plates buried in different depths to investigate the effect of overburden ratio on failure mechanism and effect of shape of anchor on capacity, Akinmusura (1978) noticed a transition from shallow failure mechanism to deep failure mechanism for loose sand at an overburden ratio of 6.5. His test results were questioned by Dickin and Leung (1979) who repeated the tests and reported soil loads that were 50% more than those measured by Akinmusuru (1978). Dickin and Leung (1979) also compared the test results with the tests performed by Das and Seeley (1975) and noted that their (Dickin and Leung, 1979) test results were more than double of that reported by Das and Seely (1975) for the same range of overburden ratios. On the closure, Akinmusuru (1979) concluded that the variation in soil loads by different researchers can be explained by the effect of anchor plate rotation as a result of using single cable for pulling the plate. After examining different shapes of anchors, he also made the conclusion that the wider anchor plates have less horizontal capacity per unit length. This finding was in agreement with observations made by Das and Seeley (1975). Trautmann and O'Rourke (1983, 1985) performed a series of tests on pipes with outside diameters of 102 mm and 324 mm buried in dry sand. The main goal of their research was to measure the soil loads on buried pipes as a function of pipe depth and soil density; hence, the test program was designed to test pipes buried in sand with different density levels and with different overburden ratios ranging from 1.5 to 22.0. Three density levels were tested with corresponding peak friction angles of 44°, 36°, and 31° (friction angles were based on the results of direct shear tests). After a comprehensive comparison with the previous research, they observed that the experimental results from tests in medium-dense and dense sand are in agreement with the analytical model developed by Rowe and Davis (1982); in the case of loose sand, the soil loads on pipe was more than those estimated from analytical model for loose sand - but, closer to the computed loads for pipe buried in medium sand from the analytical model. This observation was in agreement with the pattern of density change in front of a pulling vertical anchor as observed by Kosteyukov (1967) and described earlier. 25 The results of tests performed by Trautmann and O'Rourke (1983) were also in agreement with suggestions made by Das and Seely (1975) in terms of the shape of load-displacement response. Accepting rectangular hyperbola model for load displacement relation, Trautmann and O'Rourke (1983) estimated pipe displacement at failure (Y u) as equal to 0.13H, 0.08H, and 0.03H for loose, medium, and dense sand, respectively. A series of laboratory tests were performed by Murray and Geddes (1989) on anchor plates buried in sand with different overburden ratios and aspect ratio ranging from 1 to 5. The main objective of the research was to study soil loads on inclined anchors and a method was suggested to approximate the soil loads on inclined anchors. The results of their tests (and calculation method) for case of vertical anchor was compared with previous studies and found to be in agreement with the tests on vertical anchors performed by Neely and Stewart (1973). Hsu (1994) performed a series of tests on pipes buried in dry sand. A 1.8mx 1.8mx 1.2 m sand chamber was employed to conduct 120 tests on different pipes with outside diameters ranging from 38 mm to 229 mm. Tests with different sand densities, overburden ratios and loading rates were performed. The results of maximum dimensionless soil loads on pipe for different overburden ratios and for different pipe diameters were lying between the results of tests by Audibert and Nymann (1977) and Trautmann and O'Rourke (1985). Hsu (1994) studied the effect of pullout rate and presented a power law relation between strain rate and maximum soil loads on pipe. According to this relationship, the peak soil load on pipe would increase with increase in strain rate; however, this increase is less than 5% for 10 times increase in the pipe displacement rate from 0.01 D/sec to 0.1 D/sec in which D is pipe diameter. He also proposed a series of rectangular hyperbola relationships between lateral soil loads and pullout displacements for different strain rates. These relations were in accord with previous research by Audibert and Nyman (1977), Das and Seeley (1975), and Trautmann and O'Rourke (1983). 26 To investigate lateral soil loads on pipelines buried oblique to the direction of ground movement, Hsu et al. (2001, 2005) performed a series of full scale tests on pipes buried in dense and loose sand. The same sand chamber as described above and steel pipes with outside diameters of 15 cm, 23 cm, and 30 cm were used. The internal friction angles of the sand for their tests were determined, based on direct shear tests results, as 33° and 42° for loose and dense sand, respectively. When the pullout direction was normal to the pipe axis (i.e., conventional lateral pullout configuration), the result of the tests showed slightly smaller values of soil loads compared to the tests performed by Trautmann and O'Rourke (1983). After comparing with the analytical models, they concluded that for shallow embedded pipes, the lateral soil load can be well estimated in loose sand by using the limit equilibrium model with the assumption of a planar rupture surface. They also concluded that for the same embedment conditions, but in dense sand, the results of tests can be best approximated if the modified Meyerhof approach of the foundation bearing capacity theory is applied assuming a logarithmic spiral failure surface. A research program on buried pipelines was also commenced at the Centre for Cold Oceans Resources Engineering (C-CORE), Memorial University of Newfoundland, Canada in the mid-1990s. The full-scale testing facility at C-CORE is described in Paulin et al. (1996), and the work reported includes initial results from a series of full-scale tests on a pipe with 324 mm outside diameter buried in sand (Paulin et al. 1997, and 1998). Popescu et al. (1999) and Konuk et al. (1999) have also reported other series of tests on pipeline response using the same testing facility. The pullout load results were presented in terms of percentage of the maximum load and no absolute value for the results was presented. The test results were used in calibration of numerical models developed at C-CORE. Yoshizaki et al. (2001) conducted research with the objective of simulating full-scale permanent ground deformation (PGD) effects on buried pipelines with elbows. During this test, they had the opportunity to observe soil reaction on buried steel pipes. In a 9 m L x 5 m W x 1.5mH box, steel pipe having outside diameter of 100 mm was subjected to differential ground deformations. The results of this test were also used to calibrate 27 and validate a numerical model developed to capture the effect of PGD on buried pipes with elbows. However, the flexibility of pipe was a key consideration in this test and associated numerical modeling and the test results cannot be used to estimate lateral soil loads on pipe under plane strain conditions. Calvetti et al. (2004), conducted a series of small scale tests to investigate lateral soil loads on pipe during ground movement. Pipe diameters ranging from 20 mm to 50 mm were used in their tests. The results showed values of dimensionless soil loads on pipe that are higher than those from Trautmann and O'Rourke (1983) and Hsu (1994). The test results, were accompanied with a Distinct Element Method model developed with Particle Flow Code (PFC2©) software. Both the test results and the developed model showed good agreement with the Hansen (1961) method for predicting soil loads on pipes. In continuation of the work by Trautmann and O'Rourke (1983) at Cornell University, Turner (2005) investigated the effect of moisture content on the lateral soil loads on pipe during ground movement, throughout an experimental program on partially saturated sand with different moisture content and different densities. He used a steel pipe having outside diameter of 119 mm, buried in sand with overburden ratios varying from 6 to 20. Based on his tests on dry sand, he modified the charts presented by Trautmann and O'Rourke (1983) to predict soil loads on pipes buried in dry sand. The charts yield slightly higher values than those suggested by Trautmann and O'Rourke (1983). Turner (2005) also presented a series of curves to predict soil loads on pipes buried in partially saturated soil and concluded that soil loads on pipe buried in moist sand can be as high as double of those in dry sand. 2.2.2 Interpretation of experimental findings and numerical modeling The analytical method for calculation of lateral soil loads on shallow buried anchors is based on the assumption of developing active soil pressure at the back of the anchor and passive soil pressure in front of the anchor. Hansen (1953) presented a model in which 28 failure surfaces were assumed to be straight and extended to the surface. He mentioned that this method yielded good results for anchors with the buried overburden ratios of up to 2. Since the method did not account for the vertical load equilibrium, for deep-buried anchors, the predictions yielded relatively lower values and were not in agreement with experimental results. It was, however, noted that the analytical methods that were developed for buried anchors, obviously do not account for the curvature of pipe, thus leading to potential overprediction of soil loads on pipelines. The first analytical model that was adopted to predict the soil loads on pipelines subjected to lateral ground movements was developed by Hansen (1961). This model initially was suggested by Hansen to predict the lateral capacity of rigid piles. He assumed that a laterally loaded pile at shallow depths behaves like a retaining wall, and at great depths the behaviour is expected to be similar to a deep strip footing. In order to compute soil loads on piles at intermediate depths, he interpolated soil loads for shallow and deep failure mechanisms. Dimensions of passive failure wedge on a pile (or anchor) with intermediate buried depth was calculated using Rankine passive earth pressure theory. Knowing the dimension of the failure surface, and assuming mobilization of frictional force along both sides of failure wedge, an interpolation function was approximated. This model however was found to have some limitations when it was used to predict soil loads on pipelines. Later, it was shown that this model can in fact result in overprediction of soil loads on buried pipes (Ovesen 1964). Moreover, the Hansen (1961) model does not allow for the vertical movement of the buried structure. It is known that a buried structure when subjected to lateral ground movement may move upward; hence, the additional restraint of fixing vertical direction of movement in the model would result in predicting higher loads on pipe than those expected in real life problems. Hansen's (1961) model was adopted by Audibert and Nyman (1977), and they reported good agreement between results of lateral pullout loads on pipes and their new analytical model. In this model, the predicted lateral loads on the pipelines are represented by a series of curves as shown in Figure 2-1. The curves would yield dimensionless soil loads on pipes buried in sand with different friction angles and different overburden ratios. 29 Dimensionless lateral soil loads , or N q h values as noted in Figure 2-1, are essentially the ratio of total lateral load on pipe normalized with respect to the overburden stress at the pipe springline level (Nqh = F L /y.D.H.L) Figure 2-1: Prediction of horizontal bearing capacity on buried pipes; developed by Audibert and Nyman (1977) from method suggested by Hansen (1961) By modifying Hansen's (1961) model to account for curvature in failure surface, Ovesen (1964) suggested a method to predict soil loads on buried anchors with low overburden ratios subjected to lateral ground displacement. The model showed a reduction in soil loads compared to Hansen's model. To extend the result of this model for buried anchors to intermediate burial depths, Ovesen (1964) used the results of an experimental program performed by himself and suggested a method to predict soil loads on vertical anchors 30 with buried depth ratio up to 10. As noted by Ovesen and Stromann (1972), the results from these experiments further supported the method presented by Ovesen (1964). In his new model, Ovesen allowed for vertical displacement of the vertical anchor which led to mobilization of lower frictional force on the anchor and consequently lower lateral resistance (see Figure 2-2). Displacement constrained to horizontal direction Displacement not constrained Figure 2-2: Comparison between assumption of Hansen (1961) method (a) and Ovesen (1964) method (b); (Figure adopted from Trautmann and O'Rourke, 1983) Using Sokolovskii's (1965) method for soil stress characterization and ignoring active soil loads at the back of the anchor, Neely et al. (1973) investigated two different analytical methods and compared them with the experimental results. Focusing on shallow overburden ratios of up to 5, Neely et al. showed that application of the Meyerhof free surface model unacceptably overpredicts the soil loads while a simpler model, in which soil above the anchor is represented as a surcharge load with no shear strength, yields results closer to the experimental results. Using upper bound limit analysis along with the results of tests on anchor plates, Wang and Wu (1980) calculated the capacity of anchor plates with different orientation angles. 31 Applying this method to estimate soil loads on vertical anchors yielded loads 30% to 50% higher than those predicted by Hansen (1961) method. Trautmann and O'Rourke (1983) adopted the analytical model presented by Ovesen (1964) and Ovesen and Stromann (1972) which was used in prediction of soil loads on vertical anchors. They showed that the results of this model when used to estimate soil loads on pipes, is in agreement with full-scale test results. They developed the series of graphs shown in Figure 2-3 based on their test results in a similar manner to those shown in Figure 2-1. 0 1 2 3 4 5 6 7 8 9 10 DIMENSIONLESS DEPTH, H/D Figure 2-3: Prediction of horizontal bearing capacity on buried pipes; developed by Trautmann and O'Rourke (1983) from method suggested by Ovesen (1964) 32 In 1980's and with the advent of computers and their applications in engineering, numerical models started to develop. Rowe and Davis (1982), through a finite element model, investigated the behaviour of anchor plates buried in sand. A parametric study was performed with consideration of burial depth, friction angle, dilatancy effect, initial soil stress conditions, and surface roughness of the plate response. They observed that soil dilatancy significantly affects anchors' capacity especially for deep embedment. Based on their observations, the effect of anchor roughness is more noticeable under shallow embedment conditions. They also indicated that the effect of initial lateral soil stress on anchor capacity is insignificant. After comparing the results with physical model tests by authors as well as a comparison with previous experiments, they presented a series of charts that could be used to estimates capacity of vertical anchors. The presented graphs consisted of a series of charts, yielding anchor capacity for different embedment ratios. Correction factors presented by authors allow for modification of anchor capacity for different surface roughness, soil dilatancy, and initial stress conditions. Popescu and co-workers (1997) developed a 2D finite element model in ABAQUS and simulated the full-scale tests performed at C-CORE (as described in Section 2.2.1). After calibration and validation of the numerical model, they provided a class " C " prediction of their physical model (i.e. calibrated the numerical model with a series of test results which yields reasonable values for other cases). With the use of Mohr-Coulomb model as soil constitutive model, the numerical model showed a good agreement with test results in terms of peak load and mobilization of load on the pipe. Since their results have been presented only as normalized values detailed assessment of these data is difficult. In practice, pipelines are buried in trenches and the stiffness of native soil is usually higher than that of backfill material. Using a two dimensional finite element model in ABAQUS, Phillips et al. (2004) investigated the effect of ground movement on pipelines buried in trenches. Both backfill and native soil materials were modelled as cohesive 33 materials. The results of numerical modeling were supported by physical centrifuge tests. The effects of rate of ground movement and trench wall inclination were investigated. It was shown that a decrease in trench surface slope from vertical position to 60° does not noticeably affect the soil load on the pipe while trench wall inclination of 45° and less can result in reduction of soil loads on pipes significantly (depending on relative density of native soil and backfill material and distance of the pipe from the trench surface). Using the experimental results of Trautmann and O'Rourke for tests in sand with overburden ratios of 2 to 11, Yimsiri et al. (2004) calibrated a finite element numerical model to investigate soil behaviour in deep embedment. They used two different constitutive models: Mohr-Coulomb model and Nor-Sand model. They extended the result of numerical model to overburden ratios of 100 and suggested limiting values for dimensionless load for different friction angles. As shown in Figure 2-4, a series of curves were also provided to estimate the soil loads on pipes buried in sand with different friction angle and at high overburden ratios. Guo and Stolle (2005) collected the result of previous experiments to investigate the effect of different parameters on lateral soil loads on pipe subjected to ground movement, including soil dilatancy and scale effect. They used a finite element model to simulate buried rigid pipe displacement in sand material. ABAQUS software was used for modelling and material properties were obtained from triaxial tests. They used a Mohr-Coulomb model with constant dilation angle and constant friction angle as initial constitutive model. The main goals of this study were capturing the effect of geometrical factors such as burial depth and overburden ratio, investigating the scale-effect on soil loads, and sensitivity analysis of soil parameters. The model was calibrated and validated with the full-scale tests performed by Popescu (2002) on pipe with outside diameter of 203 mm and the results were compared with the published results of previous experiments. 34 0 5 10 15 20 25 30 35 40 49 SO Erab«d!*i«rrt ratio, Htfft Figure 2-4: Prediction of horizontal bearing capacity on buried pipes in deep embedment; developed by Yimsiri et al (2004). Guo and Stolle (2005) suggested a series of relations to account for the effect of pipe diameter, soil dilatancy, and burial depth. Application of suggested relations indicates that the dimensionless soil loads on a pipe with outside diameter of 30 mm are about 80% more than that of a pipe with outside diameter of 300 mm, buried in the same material and with the same overburden ratio. This finding can explain the scatter in dimensionless loads on pipes reported by different researchers without expressly recognizing the difference in pipe diameter. They also concluded that the effect of dilation angle increases with the increase in overburden ratio, and the effect of stress dependency of friction angle decrease with the increase in overburden ratio. As explained in Section 2.2.1, Turner (2004) also investigated soil loads on pipes buried in dry and wet sand through an experimental program. He suggested a correction on 35 curves presented by Trautmann and O'Rourke (1983) and added the curves for moist sand with 10% moisture content. Figure 2-5 shows the curves presented by him. 0 2 4 6 8 10 12 Dimensionless Depth, H ^ D Figure 2-5: Prediction of horizontal bearing capacity of buried pipes is dry and moist sand; developed by Turner (2004) 36 2.2.3 Summary of key observations The above review of the results of physical model tests and a comparison between different analytical and experimental models lead into the following notes with regard to the response of buried pipelines to lateral soil movement: • The shapes of the curves describing the lateral load vs. displacement response proposed by different researchers (Das and Seeley, 1975; Audibert and Nyman, 1977; Trautmann and O'Rourke, 1983; and Hsu, 1994) were in general agreement. The rectangular hyperbola relation originally proposed by Das and Seeley (1975) was commonly adopted to describe the shape of the curve. • Unlike the shape of the load-displacement curve, there was less agreement in the maximum lateral soil load for a given case estimated on the basis of both experimental studies and those assessed using empirical and analytical methods. • Dimensionless lateral soil loads on pipes have been treated mainly as a function of soil friction angle, soil density, and overburden ratio. Several geometric parameters (e.g. pipe diameter), soil parameters (e.g. soil dilatancy), and pipe and interface parameters (e.g. surface roughness and pipe and content weight) that are likely to affect the lateral loading response, seemed to have been ignored. • Although under real-life conditions, pipelines are placed in trenches excavated in native soils there are only few studies that have been performed to capture the effect of the trench in pipeline response. Developed numerical models in this area were also limited to the cohesive material as backfill and native soil. 37 2.3 Pipeline design guidelines The research works described in the above sections have also led to several guidelines for the design of buried pipelines (ASCE 1984; A L A 2001; PRCI 2005). 2.3.1 Estimation of axial soil loads The simple equation proposed for the estimation of axial soil loads on pipes based on interface friction angle and lateral earth pressure coefficient normal soil stresses on pipes (see Equation [2-2]), has been essentially adopted by all the design guidelines. The usage of this equation, however, becomes complex since the selection of a representative lateral earth pressure coefficient varies based on the normal soil stress distribution around the pipe surface. As such, for reliable estimation of axial soil load, there is a need for further research to understand the normal soil stress distribution around the pipe surface under "at-rest" condition as well as its variation during pipe pullout. 2.3.2 Estimation of lateral soil loads To estimate lateral soil loads on buried pipes, A L A (2001) and PRCI (2004) suggests the use of charts based on works by Hansen (1961). ASCE (1984) guideline presents two methods to predict soil loads based on the works by Hansen (1961) and Trautmann and O'Rourke (1983), without specific details as to which methods to be used. As previously indicated in Section 2.2.2, Hansen method predicts soil loads more than 2 times of that predicted by Trautmann and O'Rourke (see Figure 2-1 and Figure 2-3). This further corroborates the need to understand the response of pipelines subject to lateral ground movements. Although previous research clearly improved understanding of pipeline response during transverse and longitudinal ground deformation, scatter in recommendations and test 38 results indicates that more tests, analytical studies, and numerical models are required in this field for better understanding of failure mechanism and soil behaviour. 2.4 Closure The survey of literature, combined with the summary observations made in Sections 2.1 and 2.2, clearly suggests that while significant research work has already been undertaken, the current fundamental understanding of the performance of buried pipelines when subjected to ground movements is still limited. In recognition of this need to further this understanding, a systematic experimental and analytical research program was undertaken at the Department of Civi l Engineering, University of British Columbia, Vancouver, Canada, and this thesis describes in detail the research work and its findings. The research work focused on the performance of buried steel piping subjected to axial and lateral ground movements. In addition to increasing the fundamental understanding, the research is also aimed at generating information for potential enhancement / modification of current design guidelines and criteria used by the pipeline industry. 39 CHAPTER 3 PHYSICAL MODELING ASPECTS 3.1 Experimental apparatus A testing apparatus that was designed and constructed at the University of British Columbia to conduct full-scale physical modeling research on pipe-soil interaction problems (Anderson et al., 2005) was modified to undertake the testing work of the present program. The apparatus allows investigating the force-displacement behaviour of buried pipeline configurations subjected to axial and horizontal lateral loadings, essentially simulating the typical field soil loading conditions encountered by the components of buried natural gas, oil and water supply systems. The details of the testing chamber, loading mechanisms, and data acquisition system are described in the following sections. Figure 3-1 and Figure 3-2 show scaled layout of testing facilities for two different testing configurations: axial pullout tests and lateral pullout tests respectively. Figure 3-1 and Figure 3-2 are accompanied by Table 3-1 in which the facilities are named and numbered. 40 Table 3-1: Identification of testing equipment No. Description No. Description 1 Long Soil Box (5 m x 2.5 m) 13 Shackles 2 Shortened Soil Box (3.78 m x 2.5 m) 14 >/2" Steel Cable 3 Steel Column Supports 15 String Potentiometers 4 Steel Pipe (6 or 5 m long) 16 Hydraulic Actuator: System 2 5 Hydraulic Actuator: System 1 17 LVDT: System 2 6 Pedestal: System 1 18 Pedestal: System 2 7 LVDT: System 1 19 Servo Controller: System 2 8 Load Cell 20 Steel Pipe (2.4 m long) 9 Servo Controller: System 1 21 3 bolts end clamps 10 To Hydraulic power 22 1 1/8" Steel Cables 11 Control System: System 1 23 Control System: System 2 12 Data acquisition system & Computer 3.1.1 Testing chamber The testing chamber (soil box) had been constructed to make use of available space and equipment in the University of British Columbia structural engineering laboratory. As described in Anderson (2002), the apparatus is capable of performing axial and lateral pullout tests on buried pipe specimens, and the design of the soil box was based on several considerations pertaining to both pipeline and soil deformation mechanisms. A perspective view of the soil box and the layout of the testing chamber for axial pullout and transverse pullout are shown in Figure 3-3 and Figure 3-4, respectively. 42 Figure 3-3: Testing chamber Some of the key criteria considered in the design (and sizing) of the box are: 1) full development of active and passive soil wedges in horizontal lateral pullout tests for pipe sizes up to 500 mm; 2) minimal effects from the end walls and side walls during axial pipe pullout tests; 3) promoting essentially plain strain conditions in horizontal lateral pullout tests, including rigid boundary wall conditions and minimum side friction; 4) flexible dimensions and configuration to allow easy modifications to meet alternative test configuration requirements. This modular construction of the chamber provides the ability to change the configuration of the chamber to best accommodate multiple testing configurations and applications. 43 Figure 3-4: Testing configuration for lateral pullout (left) and axial pullout (right) tests The published experience of the work performed at Cornell University (Trautmann and O'Rourke, 1983), and C-CORE (Paulin et al., 1997) was considered in the design process. The plan dimensions of the internal chamber were kept at 2.5 m x 5.0 m for most of the axial pullout tests, except for a few tests where a plan dimension of 2.5 x 3.8 m was used. In axial tests, the pipe axis was aligned parallel to the longer direction of the box. As may be noted later in the discussion of results in Section 4.4.1, axial pullout tests conducted with buried pipe lengths of 3.8 m and 5.0 m, on otherwise identical test conditions, yielded essentially similar axial force per unit length vs. displacement characteristics confirming that the end wall effects in the selected configuration are negligible. As such, the minimum pipe length (i.e., length of the box) of 3.8 m employed in the axial pullout tests is judged suitable for axial pullout tests conducted on pipe diameters up to 0.5 m. 44 Al l horizontal lateral pullout tests were performed with the box set up to have a plan dimension 2.5 x 3.8 m; in horizontal lateral pullout tests, the pipe was aligned parallel to the shorter direction of the box. The influence zones arising from passive and active wedges for typical pipe sizes were examined based on classical soil mechanics solutions and numerical modeling (see Section 3.4.2.2). Based on this work, it was determined that the inferred extents of passive and active wedges (in front and back of the pipe) during horizontal lateral pullout tests, were well within the space available in a soil box having a length of 3.8 m. The design of the chamber permits providing up to 2 m of soil cover above pipeline configurations. It is, however, noted that the failure zones with deeper soil covers are wider; as such, the length of the box required for the formation of failure zones inside the box may constrict the feasible burial depth in lateral pullout tests. 3.1.2 Backfill material - handling, placement, and removal Pipelines are installed in the field using either trenchless technology or in excavated trenches. By nature of the method, when trenchless technology is used, the pipe generally ends up in direct contact with the native soil. For pipes buried in trenches, the surrounding soil may vary depending on the site and design related factors. While in urban areas granular backfill may be used, in other areas the trench is usually backfilled with native soil. Due to these variables and also natural variability of the soil, a physical modeller is faced with some difficulty in selecting the material type to be used in a given testing process. When the objective is to understand the pipe-soil interaction process in a fundamental manner (e.g., this study), it is prudent to use a material that is relatively convenient to handle and place as well as whose material properties are reasonably well understood. Based on the above considerations, a decision was made to use, locally available, uniformly graded, Fraser River sand as backfill material for all the tests conducted herein. This sand has been extensively used in research programmes in the past 20 years at the 45 University of British Columbia. The description of the material and characterization of its properties are given in Section 3.3. For the tests performed during the research herein, depending on the test, a sand volume ranging approximately between 12 and 23 m 3 was required to achieve the required configuration. Sand was stored outside of the box in large bulk-storage bags, each of which contained approximately 0.9 m 3 of soil, and the bags were moved out using the overhead crane in the structures laboratory. Once suspended, a "draw-string" chute in the base of the bulk-storage bag would allow the sand to be placed at a chosen location inside the box. With the chute open, it was possible to spread the sand throughout the chamber by using the movement of the crane. The drop height of the sand from the base of the bags was limited to less than about 30 cm to meet safety concerns, minimize the generation of dust, and achieve a uniform "as-placed" density across the chamber. Upon completion of a given test, the backfill sand was removed via two access-holes provided near the base of the test chamber, one at either end. Each access-hole measured approximately 450 mm by 300 mm, and was covered by a plywood panel during the tests. The sand exiting from the box was transferred back into the bulk bags using a conveyor belt mechanism that was set right beside the access-hole. 3.1.3 Loading mechanism In all the test configurations, the pipe was loaded in a displacement-controlled manner. The displacement rate chosen for the testing varied between 2 mm/s to 50 mm/s. These different displacement rates were selected to evaluate the effect of loading rate. The tests results indicated that the loading rates selected has no noticeable effect on the results; this observation was not unexpected since the material used for the tests was dry sand. The relatively fast loading rate is selected to simulate the natural situation of rapid ground displacement. Other considerations that controlled the loading rates included the capacity of the hydraulic pumps and costs related to the actuators. 46 The test configurations were modified during the test program. Initially and for the six axial puljout tests, an available system during construction of new earthquake engineering research facility at U B C was used. The system was a double-acting hydraulic actuator with an analog hydraulic control system (System 1, in Table 3-1). The actuator comprises a single 225-kN (50-kip) hydraulic actuator with an integrated linear variable displacement transducer (LVDT). The hydraulic actuator, manufactured by Royal Westcoast Cylinders Inc., New Westminster, BC , Canada had a 175-mm (7-in.) bore diameter, with a full stroke of 900 mm (36 in). The actuator was trunnion-mounted to a loading pedestal that was bolted to the concrete strong-floor of the laboratory at a location separate from the soil box. The mounting arrangement places the loading axis of the actuator at a height of about 450 mm above the concrete floor (See Figure 3-1 for layout details). The in-house hydraulic system in the structures laboratory that supplied the actuator has a capacity of 75 litres per minute at 21 MPa (3000 psi). The hydraulic pump was manufactured by MTS Inc., Eden Prairie, M N USA. During the testing, a hydraulic fluid flow of 13.3 litres per minute was required to achieve a displacement rate of 10 mm/sec. The control system for the hydraulic actuator consisted of the following components: • Analog signal generator, model 340A, manufactured by Exact Electronics Inc., Tillamook, Oregon, USA; (item 11 in Table 3-1) • Temposonic integrated linear variable displacement transducer (LVDT), model LP, manufactured by MTS Inc., Eden Prairie, M N , USA; (item 7 in Table 3-1) • Servo Control Signal Interface, manufactured by the U B C Department of Civil Engineering Electronics shop; (item 11 in Table 3-1) • Servo controller, manufactured by MTS Inc., Eden Prairie, M N USA Servo control valves, manufactured by Moog, East Aurora, N Y , USA. (item 9 in Table 3-1) 47 The analog signal generator has the ability to provide a uniform "ramp" loading signal representing the desired displacement rate. A feedback signal, representing the actual actuator position, is provided from the integrated LVDT on the actuator. The servo controller compares the two signals, and when necessary, sends a command signal to the control valves to adjust the flow of hydraulic oil into and out of the actuator. The volume of the hydraulic fluid pumped into the actuator could also be controlled manually using a knob on the control box. This manual control allowed movement of the actuator during initial positioning as well as loading. A stop-button on the control box was available for emergency shutdown as needed. The loading system was upgraded during the course of the test program, primarily because the horizontal tests required the use of 2 parallel actuators along with a control system to synchronize the displacement of the actuator loading rams (only one of the actuators was needed to conduct the axial pullout tests). The remaining axial pullout tests and all horizontal pullout tests were undertaken using the new upgraded system (System 2 in Table 3-1). The new system consisted of two double-acting hydraulic actuators with a digital hydraulic control system. The capacity of the actuators is 418 kN (93 kips) at 21 MPa (3000 psi) working pressure, and 5.6 MPa (800 psi) pressure drop over valve with an externally mounted Temposonic SSI feedback. The resolution of the SSI probe is 2 microns. The hydraulic actuators, manufactured by Royal Cylinders Inc., has a 200-mm (8-in) bore diameter, with a full stroke of ±305 mm (±12 in) and a 90-mm (3.5-in) rod diameter. Again, the actuator is trunnion-mounted to a loading pedestal attached to the reinforced concrete floor, separate from the soil box. The mounting arrangement provided a connection height of 700 mm above the concrete floor (See Figure 3-2 for layout details). While two actuators of system II were being used at the same time, a hydraulic fluid flow of 37.7 litres per minute is required to achieve a displacement rate of 10 mm/sec for both actuators. 48 The control system for the hydraulic cylinder consisted of the following components: • Delta R M C controller, model RMC100-S2-ENET, with SSI interface to the probe and analog command output to the valves, manufactured by Delta Computer Systems Inc., Vancouver, WA, USA; (item 23 in Table 3-1) • Ethernet communication link to PC for data acquisition; • 4 controlled axes; • Analog input card for reading of pressure transducer for pressure control; • SSI feedback probe, Temposonic RP with captive sliding magnet; (item 17 in Table 3-1) • Servo proportional valve, 10 GMP, for speeds up to 25 mm/sec; • Servo life filters manufactured by PQ Systems ltd; (item 19 in Table 3-1) RMCWin software by Delta Computer Systems Inc., Vancouver, WA, USA, was employed to interface with the R M C controller. The SSI probe has the ability to check the position of each cylinder so that actual and target positions are being compared continuously. The controller sends command signals to the servo valve to adjust the valve opening, and consequently the actual position of the loading ram to match the target. The controller can provide synchronized or proportional movements of up to four control axes. The movement of one of the actuators was geared to the 2 n d axis to provide uniform and steady loading at each end of the pipe during horizontal pullout tests. A schematic of the control system is shown in Figure 3-5. 49 Hydraulic Power 4 • Servo Valve 1 4—• Actuator 1 Controller 4 » Computer (RMCWin) Servo Valve 2 Actuator 2 Figure 3-5: Schematic figure of control system 3.1.4 Pipe specimen placement and actuator connections In axial pullout tests, the load cell was attached to the actuator and a shackle for pulling on the pipe. The system for attaching the pipe to the load cell consisted of two shackles attached to the pipe and a 1-metre length of 12.7-mm (0.5-in) cable with hooks at both ends. As shown in Figure 3-6, the hooks on the cable were attached to the shackles on the pipe and the cable passed through the shackle attached to the load cell. The cable system provided a loose connection during test preparation and prevented damage in the event of movement of the actuator toward the soil box. During reverse loading, the load from the actuator ram was transferred to the pipe using an approximately 200 mm thick wooden block so that the load was symmetrically imposed on the pipe circumference. 50 Figure 3-6: Coupling system for axial pullout tests The length of the pipeline test specimen was longer than the length of the box so that the pipe extended through both ends of the soil box. This ensured a constant soil-pipe test length and also avoided soil disturbance at the back of the box during pullout. Two 54-cm circles were cut out of both ends of the box to provide approximately 4 cm of clearance for the 457-mm pipe diameter. To prevent soil loss at the end-openings of the soil box during axial pullout, a gasket made using " M L C foam", a closed cell foam material, was secured at the end-openings. Circular holes, 40-cm in diameter, were cut out of 18 mm x 600 mm x 600 mm (0.75 in x 24 in x 24 in) sheets of M L C foam. The foam sheets were attached on the inside face of the soil box at each opening. The smaller diameter of the hole in foam, compared to the pipe, provided compression in the foam when pipe passed through the openings and 51 prevented sand from flowing out. The foam material exhibited minimal resistance to pipe movement. The 600 mm x 600 mm size of the foam sheets was also expected to provide a reasonable width of flexible "cushion" material to allow free deformation of the sand next to the end wall and reduce end effects. Figure 3-7: Coupling system in lateral pullout tests The coupling system for horizontal pullout tests consisted of end clamps at the each end of the pipe, combined with 29-mm (1-1/8-in) steel cables and shackles (see Figure 3-2 and Figure 3-7). The 1-1/8-in cables were directly attached to the clamps and passed through vertical slots in the end wall of the box. The vertical slots were provided to promote vertical movement of the cables in the event of vertical pipe uplift during the tests. The cell foam used to seal pipe openings in the axial tests was used to seal the vertical slots. The cables extending out of the box were attached to the load cells using shackles, similar to the approach used to set up axial pullout test configurations. 52 3.1.5 Instrumentation The primary measurements made during the tests included the forces acting on the pipeline specimens and the displacement of the pipelines relative to the soil box. Numerous other measurements were added to specific tests during the course of this research to attempt to provide information on the fundamental mechanics of pipe-soil interaction and to better understand differences between observed behaviour and the expected behaviour. These measurements included pipeline interface pressure, backfill soil pressure, surface soil displacement for horizontal tests, and width of interface shear zone for axial tests. The movement of geosynthetic layers was also measured as appropriate. A l l instrumentation and measurement methods employed during the research are described in this section. A matrix identifying the types of measurements made during specific tests is provided in Section 3.2. 3.1.5.1 Force measurement The applied load on the pipeline segments in axial pullout tests was measured using a load cell mounted in-line to the actuator loading ram (item 8 in Table 3-1). The load cell was a MTS model 661.22, with a maximum load capacity of 225 kN (50,000 lbs). The load cell was operated at an excitation voltage of 10 V , and was calibrated over the range of expected axial loads, which was less than 150 kN. For measuring load on the pipe during transverse pullout tests, load cells were mounted on each of the loading rams of the two actuators. Both of the load cells were essentially identical to the one used in axial pullout tests. Total load on the pipe in lateral load tests was taken as the sum of the load measured from each load cell. 3.1.5.2 Actuator displacements As described in Section 3.1.3, two types of actuators were used during the tests. In the first system, the displacement of the actuator was measured using the integrated LVDT, 53 which is a Temposonic type LP analog magnetostrictive linear displacement transducer; in the second system, a Temposonic type RP with captive sliding magnet LVDT was used. 3.1.5.3 Pipe displacements Pipe displacements relative to the concrete floor of the laboratory were measured using string potentiometers. For axial tests, measurements were made at both ends of the pipe with the string for the potentiometers passed through the inside of the pipe. For horizontal tests, 1.6 mm (1/16") diameter steel cables were attached to the both ends of the pipe. The cables were passed through the soil and outside the box and attached to the string potentiometers mounted at the back of the box. 3.1.5.4 Displacement of geosynthetic String potentiometers were used to observe displacements of geosynthetic layers (as described in Section 1.2) in order to assess the slippage behaviour of the layers. In axial pullout tests, very thin extension cables attached to the mid-length of each geosynthetic layer (passed through the geosynthetic layers and then to the outside of the soil box) were attached to the string potentiometers mounted on the back of the box. In horizontal pullout tests, string potentiometers were attached to the top of the geosynthetic layers. 3.1.5.5 Pipe interface pressure measurements In tests where the soil pressures imparted on the pipe during testing were monitored, the measurements of normal stress on the pipe were accomplished using five total pressure transducers (TPT). Data Instruments brand, bonded semiconductor strain gauge pressure transducers, Type AB/HP manufactured by Honeywell, Freeport, PL, USA were used in this regard. Three of the transducers had a range 0 kPa to 100 kPa while the other two 54 had a range 0 kPa to 50 kPa. Figure 3-8 shows a typical transducer mounting arrangement on the pipe. Figure 3-8: Pressure Transducers mounted on pipe After cutting holes in the pipe, the transducers were mounted to be flush with the surface of the pipe so that there is minimum opportunity for soil arching or localized disturbance at the pipe-soil interface in the vicinity of the transducer. The holes in the pipe were threaded and transducers were mounted in hollow threaded shafts, which allowed convenient screw-in type attachment. The wires of the pressure transducers were passed through the pipe and connected to the data acquisition system. The diameter of pressure transducers was 19.1 mm (0.75 in) and the diameter of transducer mounted on hollow shaft was 28.6 mm (1-1/8 in). Considering the diameters of the pipe used in testing (i.e., 457 mm or 324 mm), the angular discontinuity in the pipe circumference associated with the pressure transducer installation was approximately ±3.6° for a pipe diameter of 457 mm and ±5.4° for pipe diameter of 324 mm. To simulate the roughness of the pipe surface, graded 0.4-mm size (1/64") uniform sand, which was comparable to the size of the asperities on the pipe surface resulting from sand blasting, was glued on top of 55 transducers. Considering the efforts to reduce pipe surface discontinuities and match pipe roughness, it is judged that the introduction of the pressure transducer did not affect the local soil-pipe interaction at the transducer location. The five transducers were located along half of the pipe circumference, at five equidistant locations (45° radial spacing) between the crown and invert of the pipe. Although this arrangement only covered one half of the pipe, it was judged suitable since the soil pressures around the pipe during axial tests could be considered symmetric around the pipe cross section and the most significant pressures for horizontal tests occur on the side of the pipe facing the actuators. 3.1.5.6 Backfill soil pressure Total stress in the soil mass at selected locations was measured using a 150-mm (6-in) diameter, 13-mm (0.50-in) thick, total earth pressure transducer fabricated by RST Instruments, Coquitlam, BC , Canada. The 150-mm diameter sensing steel plate of the pressure transducer is mounted on a chamber filled with oil. The range of the transducer is 0 kPa to 200 kPa with an excitation voltage of 9 V to 30 V. The relatively large diameter provided a sensing surface area wide enough to minimize arching effects and improve the accuracy in stress measurement. 3.1.5.7 Backfill density measurements The density and moisture content of the as-placed backfill sand was measured using a calibrated nuclear densimeter with a sensitivity of ±1 kg/m . The densimeter readings were further checked with independent density measurements using sand bowls placed during compaction. 56 3.1.5.8 Data Acquisition Al l measurements from the instrumentation array monitoring the pipe specimens were recorded at 10 sps (10 samples per second). The high rate of sample acquisition was chosen to match with the relatively fast rate of pipe pullout that ranged from 2 mm to 25 mm per second. Signals from the instrumentation array were collected via a 16-channel National Instruments, Austin, TX , USA, signal conditioning boards model NI SCXI-1001 using the commercially available software package Daisylab, manufactured by Superlogics Inc, Waltham, M A , USA, for real-time plotting of all data. The system was controlled using a dedicated Pentium III computer system running MS Windows XP Professional. 3.2 Testing Program A total of 26 tests were performed during 23-month time period between September 2003 and July 2005. Two trial tests (i.e., two axial pullout tests and one horizontal lateral pullout test) were performed to calibrate the instruments and to check the efficiency of the test preparation procedure; the results of these trial tests are not presented in this report. Sixteen different configurations were tested and seven tests were performed for configurations that were previously examined to confirm the repeatability and reliability of the test data. A matrix of all 26 tests is provided in Table 3-2 that identifies the test configuration and test measurements employed during the tests. The following discussion addresses the system used to identify individual tests and the behaviour that was the focus of the test. Each test was given an identification code to distinguish between different tests with relative ease. Test specimens are identified using the following labelling convention: 57 W X Y Z - N where W = A for axial tests = L for horizontal tests X = B for axial tests on bare pipe and N for horizontal tests in native soil = G in axial test where first layer of wrapping is geotextile fabric = R in axial tests where first layer of wrapping is Tensar bi-directional geonet = T in horizontal test i f rigid trench boundary used Y = blank for bare pipe = G in axial test i f second layer of geotextile fabric is used = G in horizontal test i f trench wall lined with two layers of geotextile fabric Z = C i f two layers of wrapping applied with a cigar-wrapped pattern = S i f two layers of wrapping applied with a spiral-wrapped pattern = blank for axial tests on bare pipe and horizontal tests N = test number for test configuration For example, Test No . AGGS-1 represents the first axial pullout tests on a spiral-wrapped pipe with two layers of geotextile and Test No . L T G - 2 is the second horizontal pullout test with a hard trench boundary that covered with two layers of geotextile. 58 Table 3-2: Summary of performed tests and different measurements in each test Test ID Pipe OD (mm) Test Length (m) Target Density (kg/m3) Soil Cover (m) Soil Moisture (%) Measurements Performed During Test Notes Soil Density (nuclear) Soil Density (bowls) Geosynthetic Displacement Interface Pressure Backfill Pressure Loading Svstem No. AB-0 457 5.0 1,600 0.9 1% 1 1,2 AB-1 457 5.0 1,600 0.9 1% • 1 1,2 A B - 2 457 5.0 1,600 0.9 1% • - • 1 2,4 AB-3 457 5.0 1,600 0.9 1% 1 2 A B - 4 457 3.8 1,430 1.05 1% • • 2 2 AB-5 457 3.8 1,600 0.9 1% • • 2 3 AB-6 457 3.8 1,600 0.9 1% • • 2 2 AGGC-1 457 5.0 1,600 0.9 1% • • 1 2,4 A G G C - 2 457 5.0 1,600 0.9 1% • • • 1 2,4 A G G S - I 457 5.0 1,600 0.9 1% • 1 2 A R G C - I 457 5.0 1,600 0.9 1% • • 1 2 LN-0 457 1.8 1,600 0.65 1% 2 1,2 LN-1 457 2.4 1,600 0.65 0% • 2 2 LN-2 457 2.4 1,600 0.65 0% • 2 2 LN-3 324 2.4 1,600 0.46 0% • 2 2 LN-4 324 2.4 1,600 0.46 0% • 2 2 LN-5 324 2.4 1,600 0.16 0% 2 2 LN-6 324 2.4 1,600 0.73 0% • 2 2 -LN-7 324 2.4 1,600 0.46 0% 2 2 LNG-1 457 2.4 1,600 0.65 0% • • 2 2 LNG-2 457 2.4 1,600 0.65 0% • • 2 2 LT-1 457 2.4 1,600 0.65 0% • 2 2 LT-2 457 2.4 1,600 0.65 10% • 2 2 LT-3 457 2.4 1,600 0.65 10% • 2 2 LTG-I 457 2.4 1,600 0.65 0% • • 2 2 LTG-2 457 2.4 1,600 0.65 10% • • • 2 2 N O T E S : 1 . Preliminary test, not presented in this report 2. Performed within 24 hours of preparation 3. Performed 45 days after preparation 4. Reverse loading test was performed 3.2.1 Comments on the implementation of axial pullout test Al l tests were performed on a sand-blasted steel pipe specimen with an outside diameter of 457 mm (18 in) and a 12.7-mm (0.500-in) wall thickness. Two different configurations of axial pullout tests were performed: bare pipe tests and wrapped pipe tests. Before each test, the box was emptied to the bottom level of the pipe to remove any possible residual stress in the soil. The axial pullout testing program is summarized in Table 3-3. Table 3-3: List of performed axial pullout tests Tes t No. Pipe Size (mm) H/D Rati 0 Soil Density Loading time Pipe Condition Box Length (m) Test ID 1 18" 2.5 Dense Immediate* Loading Bare Pipe 4.98 AB-2 2 18" 2.5 Dense Immediate Loading Bare Pipe 4.98 AB-3 3 18" 2.5 Dense Immediate Loading Bare Pipe 3.76 AB-4 4 18" 2.7 Medium -loose Immediate Loading Bare Pipe 3.76 AB-5 5 18" 2.5 Dense Loading After 45 Days Bare Pipe 3.76 AB-6 6 18" 2.5 Dense Immediate Loading 2-Layers Geotextile -** CW 4.98 AGGC-1 7 18" 2.5 Dense Immediate Loading 2-Layers Geotextile -** CW 4.98 AGGC-2 8 18" 2.5 Dense Immediate Loading 2-Layers Geotextile - SW** 4.98 AGGS-1 9 18" 2.5 Dense Immediate Loading 2Layers: Geotextile, Geonet - CW 4.98 ARGC-1 The pullout test performed within 24 hours after preparation of the specimen. **CW and SW stand for "cigar wrapping" method and "spiral wrapping" method respectively. 60 3.2.1.1 Loading rate Figure 3-9 and Figure 3-10 show plots of axial displacement (8) versus time (8-t) for the Tests No. AB-2, AB-3 and AB-4 through AB-6. Displacement rates in Tests No. AB-2 and AB-3 were manually controlled, while the other three tests, Tests No. AB-4 through AB-6, were displacement controlled. The following points are noted with respect to Figure 3-9 and Figure 3-10: • In Test No. AB-2, the displacement rate was about 25 mm/sec until 334 mm (first loading) and after unloading and reloading, displacement rate of about 10 mm/sec was applied to the pipe. The 8-t for the reverse loading of Test No. AB-2 had the same rate as initial loading. • Initial displacement rate for Test No. AB-3 was about 20 mm/sec which was decreased to about 10 mm/sec for subsequent reloading. • The 8-t for the other three tests consists of an initial loading with the rate of 2 mm/sec and repeated loadings (after unloading and reloading) with a rate of 4 mm/sec. 900 Time (sec) Figure 3-9: Displacement rate for Tests No. AB-2 and AB-3 61 500 0 100 200 300 400 500 Time (Sec) Figure 3-10: Displacement rate for Tests No. AB-4, AB-5 and AB-6 The rate of loading within the range used exhibited no effect on the pullout load as presented in Section 4.2.1. Figure 3-11 presents typical axial displacement (5) vs. time profiles (8-t profiles) for the tests conducted on steel pipe wrapped with geosynthetics. The Test No. AGGC-1 which was displacement-controlled with a constant rate of 50 mm (2") per second, gave rise to an approximate sinusoidal oscillatory response in the axial load. It was judged that the impact from abrupt commencement of pulling (i.e., abrupt change from 0 mm/s, specimen at rest, to 50 mm/s) likely had imparted instability in the feedback loading system causing this oscillatory motion. As such, the loading system was manually controlled so that rate the displacement rate would increase gradually from zero to displacement rates of 10 mm/s to 35 mm/sec. 62 900 0 10 20 30 40 50 60 70 80 Time (Sec) Figure 3-11: Displacement rate of wrapped pipe tests 3.2.1.2 Quality and loss of measurement data Instability in the hydraulic system and feed back system resulted in some sinusoidal noise in the measured pullout force of one test (Test No. AGGC-1) . This configuration was retested (Test No. AGGC-2) to remove ambiguity in the test results. The result of the repeated test showed that an average value of the sinusoidal force measurement could represent the load on the pipe. A voltage cut-off occurred during running axial pullout Test No. AB-2, resulting in no record of the maximum load. The test data is still presented as can be used for comparing reloading and reverse loading behaviour. In addition, the peak loads for Test No. AB-2 can be interpreted from two other tests performed with the same configuration (Tests No. AB-3 and AB - 4 ) . 63 3.2.1.3 Preparation and variations in axial tests on bare pipe (AB Tests) Five axial tests were performed on bare pipe. In four of the tests, dry sand was compacted to reach average density of 1600 kg/m3. The sand backfill was compacted using two different methods. In Tests No. AB-2 and AB-3, a model AP2000B plate tamper, manufactured by M-B-W Slinger, WI, USA, with an equivalent dynamic force of 11 kN applied to a 53 cm x 48 cm tamping plate, was used to compact the soils. The backfill for Tests No. AB-4 and AB-6 was compacted using a static roller. As noted in Section 3.4.4, acceptably uniform density conditions were achieved regardless of the compaction method. Test No. AB-5 was performed using "as-placed" sand to observe the effect of reduced soil density on the pipe response. For this test, no additional compaction was applied to sand after emptying from the bulk-storage bags, resulting in a medium loose condition with an average density of about 1430 kg/m . The depth of burial in a given test was defined with respect to an H/D ratio, in which H is the height of the sand overburden above the centre line of the pipe, and D is outside diameter of the pipe. H/D ratio was kept constant at 2.5 for all the tests performed with dense sand. Test No. AB-5 (medium loose sand) was performed with an H/D ratio of 2.7 to provide nearly the same vertical effective stress at the pipe level for all tests. A l l the pullout tests but Test No. AB-6 were conducted within 24 hours after filling the soil box. Test No. AB-6 was performed 45 days after preparation in order evaluate any effects of aging or relaxation of the box on pipe response. The size and positions of the box, was required to be modified after Tests No. AB-1, A B -2 and tests on wrapped pipe to free up lab floor space for other experiments. The loading system (Loading System 1, see Section 3.1.3) used for the initial tests was also needed to be moved for use in a new earthquake engineering research facility. As such, after repositioning the box and changing the dimensions from 2.5m x 5m to 2.5 m x 3.8m, the Loading System 2 was set up for use in the subsequent tests as described in Section 3.1.3. 64 There were some differences in the instrumentation between the tests. In Tests No. AB-2 and AB-3, only the axial load and axial pipe displacement were monitored; whereas, in the other tests, soil stress on the pipe was measured using total pressure transducers mounted on the pipe. The pressure transducer readings provided an opportunity to observe the changes on the normal stress on the pipe while filling the box and compacting the sand. In Test No. AB-4, the sand surface was monitored using surveying to provide information on surface deformations during axial pullout. Twenty-four points, arranged in a grid pattern on the backfill surface, were surveyed before and after the test to a resolution of 1 mm. In Tests No. AB-4 and AB-5, some predefined zones in the immediate vicinity of the pipe were backfilled with colored sand. Following these tests, the backfill was carefully removed and changes in the location of colored sand were used to identify active shear zones and particle displacements in the proximity of the pipe during pullout. In Test No. AB-2, the pipe was loaded in the reverse direction after completing the axial pullout to observe the variation in pullout force on a pipe in pre-sheared backfill. 3.2.1.4 Preparation and variation of axial tests on wrapped pipe Four tests (Tests No. AGGC-1 , AGGC-2, AGGS-1, and ARGC-1) were performed on the pipe wrapped with two layers of geosynthetic fabrics. Two geosynthetic fabric products were used in the tests: • Mirafi Filterweave (No. FW700) woven geotextile produced by Mirafi Construction Products, Georgia, USA • Tensar DC4105 Bidirectional Geonet produced by The Tensar Corporation, Georgia, USA 65 The objective was to assess the effectiveness of geosynthetics in reducing the axial soil loads. In essence, an inner layer of geotextile or geonet (termed first layer) that is in immediate contact with the pipe constituted the first wrapping; this layer was overlain by an outer geotextile layer (or second layer) that would be sandwiched between the surrounding soil and the first layer. Two types of geosynthetic-wrapping methods were used: (a) "cigar-wrapping"; and (b) "spiral-wrapping" as illustrated in Figure 3-12. In the case of cigar-wrapped pipe, each layer of the geotextiles were tightly wrapped around the pipe (with about 100 mm of overlap along the longitudinal seams) and secured using common "duct tape". In spiral-wrapping, each layer of geotextile was wrapped with an overlap of about 200 mm with the ends secured using "duct tape". Figure 3-12: Schematic figure showing cigar-wrapping (top) and spiral-wrapping (bottom) methods for geotextile-wrapped pipes. 66 Three tests were performed with the pipe wrapped in two layers of Mirafi geotextile fabric. For one test (Test No. ARGC-1), bi-directional geonet was used for the first wrapping layer. The wrapping of pipe was undertaken outside the box, and then placed in the box using a crane and forklift. A l l of the wrapped pipes were buried in sand compacted to the target density of 1600 kg/m using the model AP2000B plate tamper described in the previous section. The density of the backfill was checked in Tests No. AGGC-1 , AGGC-2 and ARGC-1. The H/D ratio for all tests was kept at 2.5 and the tests were conducted within 24 hours after completion of filling of the box. Reverse loading tests were also performed for Tests No. AGGC-1 and AGGC-2. 3.2.2 Comments on the implementation of transverse pullout tests Fourteen horizontal pullout tests were conducted using a dense sand backfill medium with bare steel pipe. Although different methods were employed to compact the sand, as described in previous section, regular density measurements indicated an average density of about 1600 kg/m and a standard deviation of ±19.2 kg/m for all tests. Horizontal tests were conducted using three pipe backfill configurations: I. Sand-sand: Pipeline buried in a uniform dense sand backfill (i.e., Fraser River sand). This configuration is representative of pipe burial in well-compacted cohesionless soil. II. Sand-lining-sand: Pipeline buried in a geotextile-lined trapezoidal "trench" in Fraser River sand fill, as shown in Figure 3-13. The trench is lined with two layers of Mirafi Filterweave (No. FW700) woven geotextile. After placement of the pipe, the trench is backfilled with the same sand that was used to form the soil base below the geotextile. III. Sand-lining-hard boundary: Pipeline buried in a "rigid trench", backfilled with Fraser River sand. This configuration is similar to the configuration described in 67 Item II above except that it is designed to represent a case where the pipe is buried in a trench excavated in a hard "native" soil. This configuration includes cases with and without two layers of Mirafi geotextile fabric lining of the trench wall. 3.8m 2.5m. Figure 3-13: Scaled schematic drawing of the pipe and trench position 3.2.2.1 Loading rate The imposed displacement vs. time (8-t) profile on the pipe consists of an initial part with a constant velocity until displacement rate of 5 mm/sec was obtained, which essentially occurred within the first 10 mm of pipe pullout. After the initial part, in most tests, the pipe was laterally pulled at a constant displacement rate 5 mm/s. In some of the tests, the rate of displacement was doubled at the middle of the test to assess any effects from the rate of loading on lateral pullout load. As can be noted from the discussions in Section 5.3.1 and Appendix B, changes to the rate of loading seemed to have no noticeable effect on the measured load during unloading and reloading. 68 3.2.2.2 Preparation and variation of horizontal tests Sand blasted steel pipe in two different pipe sizes: 324-mm (12.75-in) diameter with 6.4-mm (0.25-in) wall thickness and 457-mm (18-in) diameter with 12.7-mm (0.50-in) wall thickness. The pipe used in all of the tests had a length of 2.4 m, which is 16 cm shorter than the width of the box to provide some nominal clearance to protect against potential jamming of the pipe in the box during pipe pullout in the event of differential movement of actuators. In Tests No. LT-2, LT-3 and LTG-3, the model AP2000B plate tamper, described previously, was used to compact the sand. Other tests used a static roller for compaction. Table 3-4: list of performed lateral pullout tests Pipe Size (mm) H/D Ratio Soil Density Box Configuration Geotextile Liner Separation Backfill Material Test ID 18" 1.92 Dense Normal No Dry Sand LN-1 18" 1.92 Dense Normal No Dry Sand LN-2 12.75" 1.92 Dense Normal No Dry Sand LN-3 12.75" 1.92 Dense Normal No Dry Sand LN-4 12.75" 1 Dense Normal No Dry Sand LN-5 12.75" 2.75 Dense Normal No Dry Sand LN-6 12.75" 1.92 Dense Normal No Dry Sand LN-7 18" 1.92 Dense Normal Yes -45° Dry Sand LNG-1 18" 1.92 Dense Normal Yes - 35° Dry Sand LNG-2 18" 1.92 Dense Trench Type, 35° Hard Boundary No Dry Sand LT-1 18" 1.92 Dense Trench Type, 35° Hard Boundary No Moist Sand LT-2 18" 1.92 Dense Trench Type, 35° Hard Boundary No Moist Sand LT-3 18" 1.92 Dense Trench Type, 35° Hard Boundary Yes - 35° Dry Sand LTG-1 18" 1.92 Dense Trench Type, 35° Hard Boundary Yes - 35° Moist Sand LTG-2 69 3.2.2.3 Comments on configuration I tests (sand-sand) Seven horizontal pullout tests were performed on buried pipeline representing Configuration I. As summarized in Table 2.4, the bare pipe tests were performed for H/D ratios of 1, 1.92 and 2.75. Tests LN-1 and LN-2, with 457-mm (18-in) diameter pipe, were performed at an H/D ratio of 1.92. The smaller pipe size, with a diameter of 324 mm (12.75 in), was used for the remaining Configuration I tests. Deformations of the sand surface were captured by measuring the soil surface before and after each test. Following each test, the soil box was emptied to the bottom level of the pipe to examine vertical and horizontal displacement of the pipe during the test and also remove any potential residual stress build up in the sand mass prior to the next tests. Interface soil pressures on the pipe during filling of the box with sand and during pullout were measured in Test No. LN-7 using pressure transducers mounted on the pipe (as described in Section 3.1.5). 3.2.2.4 Comments on configuration II tests (sand-lining-sand) Configuration II tests included Tests No. LNG-1 and LNG-2 with a two-layer geotextile trench lining inclined at 45° and 35°, respectively (see Figure 3-13). Under typical field conditions, a symmetric double-sided trapezoidal trench is cut in the native soil; after lining with two geosynthetics, the trench is then filled with backfill material. In the physical model as displacement on the pipe is applied in only one direction and in order to save space, only half of the trench was modeled. In order to form the trench and place the geotextile layers in desired angles, two 76.2-mm x 76.2-mm x 6-mm (3-inch x 3-inch x 1/4-inch) angles inclined at the desired trench angle were attached to opposite side walls and used as guide rails (see Figure 3-14). A 2.50 m x 0.6 m steel plate was laid across the guide rails to serve as a temporary trench-surface former. Two geotextile layers, sized to cover the entire trench surface, were 70 placed on top of the trench-forming steel plate. Once the two-layers of geotextile had been positioned, the pipe was placed inside the box at the base of the trench in preparation for sand placement. Fraser River sand was placed to a depth close to the top of, and on either side of, the trench-forming plate. The sand on both sides of the formed trench was carefully compacted before the steel plate was gently raised up (by dragging along the angle guide rails), leaving the two layers of geotextile as the separator between the "backfill" and "native soil" zones. This process was repeated until the geotextile-lined trench was fully constructed. For the test LNG-1, with a trench angle of 45°, the horizontal distance from centre of the pipe to the trench face was 46 cm. This distance was 55 cm for Test No. LNG-2, which had a trench angle of 35°. Figure 3-14: procedure of constructing a trench in native soil 71 3.2.2.5 Comments of configuration III tests (sand-lining-hard boundary) Assessment of test results from Tests No. LNG-1 and LNG-2 led to a modification of the test configuration to represent conditions that would better represent those where geotextile trench lining would be considered. One such condition is where very stiff to hard native soils are present. Since logistics and time constraints did not allow for identifying and using alternate "native" soils in the soil box, it was decided to conduct a number of lateral pipe pullout tests in an artificial trench with a "hard" boundary. A l l tests with hard boundary were performed with a 35° trench angle. The hard boundary was constructed as follows: • Six lengths of 50 mm x 300 mm (2"xl2") rough Douglas Fir timber boards were spanned across the angle guide rails (see previous section) set at an angle of 35°. • The timber boards spanning the angle guide rails were braced against the end wall using sixteen, evenly spaced, 100 mm x 200 mm (4"x8") timber boards. • The space between the timber board bracing and the end walls was backfilled with sand and compacted. • A layer of Fraser River sand was glued to the surface of the timber boards spanning the angle guide rails to simulate the roughness of an excavated soil trench wall. Configuration III tests were conducted with and without two layers of geotextile lining on the trench wall and two sand backfill conditions, dry and moist. Dry backfill was placed in a similar manner to tests for Configurations I and II, except that there was no need to use the steel plate to form a temporary trench wall. For moist sand backfill tests, the target moisture content was 10% and was achieved by sprinkling water over each lift of sand placed in the soil box and mixing it with rake and shovel prior to compacting with the model AP2000B plate tamper (NOTE: In order to have uniformity of water content, measured quantities of water was sprinkled so that approximately 140 litres of water per 72 m3 was maintained). Since the negative pore water pressures in compacted moist sand typically keeps the sand particles together, this backfill was used to simulate a field situation in which cohesive material is used as backfill. All Configuration III tests included measurements of pipe interface pressure at the ± 45° to horizon and springline of the pipe. 3.3 Characterization of material properties Understanding the fundamental mechanical behaviour of soil as a continuum and at the interface between soil and pipe surface is critical to the interpretation of data derived from the pipe-soil interaction tests. This section provides information about the materials used in the tests and summarizes the findings from laboratory tests undertaken to define important soil strength parameters including internal friction angle of sand backfill, and interface friction behaviour for sand-pipe, sand-geosynthetic, geosynthetic-geosynthetic interfaces. 3.3.1 Materials used in physical model testing Fraser River Sand Dredged sand from the Fraser River in the Lower Mainland of British Columbia, Canada, has been extensively used in laboratory research at UBC over the past 10 years. Based on the available literature, the composition of Fraser River sand is 40% quartzite and chert, 11% feldspar, 45% unaltered rock fragments, and 4% other minerals (Garrison et al. 1969). Fraser River sand can be characterized as a fine to medium sand with sand grains that are generally angular to sub-rounded in shape. Constant volume internal friction angles reported in the literature for this sand range from 32° to 34° (Uthayakumar, 1996; Sivathayalan, 2000). Previous tests of the Fraser River sand used in this study indicate an average particle size D50 = 0.23 mm, minimum particle size of 0.074 mm, a coefficient of uniformity, C u of 73 1.5, and a specific gravity (Gs) of 2.70. The minimum and maximum void ratios, determined in accordance with American Society for Testing and Materials Standards ASTM-4254 and ASTM-4253, are reported to be 0.62 and 0.94, respectively (Anderson, 2004). Grain size distribution tests of the Fraser River sand used in the current test program were performed at the start and end of the test program. The similarity in test results, shown in Figure 3-15, suggests that repeated moving and compaction of the sand has not significantly altered the sand particles. 10 0.1 G r a i n S i z e ( m m ) 0.01 0.001 Figure 3-15: Grain size distribution for Fraser River sand, before and after the testing program. Pipeline Material The pipes used in the tests, were Grade A524 steel pipe. The surface of the pipes was prepared by sand blasting (using coarse sand). The diameter of "pits" due to sand blasting on the pipe surface had a maximum size of approximately 0.8 mm (0.030 in) and an average size of approximately 0.4 mm (0.015 in). 74 TC Mirafi Filterweave 700 woven-Reotextile TC Mirafi Filterweave 700 woven-geotextile, manufactured by Mirafi Construction Products, Georgia, USA was used as the geotextile material (for wrapping of pipe) in the testing program. As described by the manufacturer, the material is composed of high-tenacity monofilament polypropylene yarns, which are woven into a stable network such that the yarns retain their relative position. The wide-width tensile strength is 26 kN/m to 40 kN/m and the apparent opening size and percentage of open area are 0.212 mm and 4% to 6% respectively (http://www.tcmirafi.com/PDF/TDS/FW/FW700.pdf). The peak and residual interface friction angle between two layers of geotextile material have been reported to be 21.1° and 19.7° for the normal stresses between 27 kPa and 40 kPa (Texas Research International Company, 2000). 3.3.2 Laboratory element testing of materials: direct shear tests The results of laboratory tests conducted to characterize stress-strain-strength characteristics of Fraser River sand and interface friction parameters are presented below. 3.3.2.1 Direct shear tests on Fraser River sand Numerical models and contact pressure measurements during transverse pullout tests indicated that the mean stress imparted on the leading part of the pipe circumference in most instances did not exceed 100 kPa. Knowing this, conventional direct shear tests were performed to cover an effective normal stress from 20 kPa to 100 kPa, with the lower value corresponding to the normal stress at the maximum pipe burial depth of 1.3 m. Detailed procedures for direct shear testing can be found in Lambe (1991). In the preparation of direct shear test specimens, Fraser River sand was placed in layers of about 5 to 10 mm in thickness, and each layer was compacted using a square-shaped 75 wooden tamper essentially covering the footprint of the specimen and vibration imparted on the direct shear box using a small vibrator. This technique resulted in a maximum average density of 1575 kg/m (D r = 68%); this is comparable to the average target density of 1600 kg/m 3 (D r = 75%) used in most tests during full scale testing. The measured shear stress at peak and large-strains vs. normal stress values from these tests are shown in Figure 3-16. The peak friction angle of Fraser River sand between individual tests varied between 44° and 41°, with an average peak friction angle of about 42° as shown in Figure 3-16. The friction angle at larger strains also seemed to be in good agreement for all tests, and has an average of about 36° (see Figure 3-16). While the direct shear apparatus provides a relatively convenient means of conducting a shear test on sand, one of the main disadvantages is its inability to deliver uniform shear strain across the plane of shear at the commencement of a test, leading to an underestimate of the peak friction angle. At the initiation of shear, the shear strains 76 mobilized at the ends of the specimen would reach very large values (i.e., bringing the mobilized shear stress at the ends to the ultimate shear strength), whereas the middle of the specimen would still be experiencing much lower shear strains. Suppression of the peak friction angle due to this strain non-uniformity is well accepted. Since predicting peak loads on buried pipelines is highly dependent upon the peak friction angle, making proper characterization of the friction angle is important for the interpretation of data from physical modeling tests. It is generally accepted (Rowe 1969; Lee 1970) that the peak friction angle calculated from the results of direct shear tests is lower than the peak friction angle calculated from triaxial test results. Independent triaxial tests results as shown in Section 3.3.3 demonstrates that direct shear tests underestimate the peak friction angle by about 2°. The result of triaxial tests on different soil densities also showed that every 3% to 4% increase in relative density (or 10 kg/m3 to 15 kg/m 3 increase in dry soil density), leads to a 1° increase in peak friction angle. The result is generalized to estimate peak friction angle at target density of 1600 kg/m3 from direct shear test results. The normal displacements were also monitored during the direct shear tests. It was observed that the effect of dilation caused up to 0.8 mm increase in the height of specimen during testing. 3.3.2.2 Direct shear tests on Fraser River sand/sand-blasted steel interface Direct shear tests were performed to determine the interface friction angle between sand-blasted steel and Fraser River sand. A 75 mm x 75 mm steel coupon (sized to fit the bottom part of the direct shear box) was prepared using the same sand-blasting procedure followed in the preparation of pipe specimens. The coupon was mounted so that its top surface was flush with the top level of the bottom part of the direct shear box. The upper part of the direct shear box was filled with compacted Fraser River sand, following the same procedure used for the direct shear tests on sand. Three tests were conducted with an initial sand density of 1600 kg/m3 (D r = 75%), and two tests were conducted with an initial sand density of 1450 kg/m3 (D r = 20%). Detailed test results are presented in 77 Appendix A. The average peak interface friction angle for the sand-blasted steel and sand interface for dense sand and loose sand were determined to be 36° and 33°, respectively. The average interface friction angle at large strains was determined to be 31°, for both dense and loose sand. The interface friction factor (5) both at peak and at large strains is about 0.85. Similar to tests on Fraser River sand, the dilation was measured in these direct shear tests via measurement of normal displacement of top cap. About 0.2 mm and 0.05 mm of dilative vertical displacements were noted for the tests with dense and loose sand, respectively. 3.3.2.3 Direct shear tests on geotextile/geotextile interface For the evaluation of interface friction characteristics between two layers of TC Mirafi Filterweave 700 woven geotextile, 2 pieces of wood were cut to fit in the bottom and top halves of the direct shear box. The wooden blocks were wrapped with geotextile fabrics with some fine sand sandwiched between the wooden block and geotextile to simulate the anticipated contact conditions between geotextile layers during physical modeling of pipelines. Five tests were performed at a normal stress between 10 kPa to 20 kPa. The average peak and large-strain interface friction angles were determined to be 21° and 20° respectively. These results are in line with those reported by Texas Research International Company (2000) for the range of normal stress between 27 kPa and 40 kPa (i.e., 21° for the peak and 20° for residual interface friction angle). The friction angles measured at the interface between Fraser River sand and other materials are compared with the measured internal friction angle of Fraser River sand alone in Table 3-5. 78 Table 3-5: Summary of friction angles from laboratory direct shear testing of this study Slippage Surface Peak Large Strain Normal Displacementa Average density Dense Sand 440.410 36° ~ 0.72 mm 1575 kg/m j (Dr = 68%) 6 Loose Sand-Sand Blasted Steel 33° 31° ~ 0.05 mm 1450 kg/m j (Dr = 20%) Dense Sand-Sand Blasted Steel 36° 31° ~ 0.20 mm 1600 kg/m J (Dr = 75%) Geotextile-Geotextile 21° 20° a. Maximum Normal Displacement of top platen of Direct shear box b. 2° should be added to get friction at 1600 kg/m3 (suggested based on triaxial testing as per Section 3.3.3) 3.3.3 Laboratory testing: Triaxial tests As mentioned in section 3.3.2, the estimated friction angle from direct shear tests are likely to be smaller than the actual value due to the non-uniform strain distribution along the length of the box. With the need to obtain reliable and realistic stress-strain characteristics for numerical modeling of full-scale test configurations, it was decided to conduct a series of triaxial tests on Fraser River sand. In particular, it was noted that the knowledge of the variation of friction angle, and deformation moduli, with stress level and relative density would provide key parameters for numerical modeling; triaxial element testing is able to provide such parameters. To simulate the same condition as full-scale soil box tests, specimens for triaxial element testing were prepared using dry Fraser River sand. The specimens were 76 mm (2.5") in diameter and about 152 mm (5") in height. As shown in Figure 3-17, the confining stress to the specimens was applied via vacuum lines attached to the top and bottom of the specimens. The vacuum (or confining stress) was kept constant throughout the tests using a vacuum regulator. The axial displacement and change in the diameter at the centre of the specimens were measured via LVDTs (Figure 3-17). This allowed the estimation of both axial and volumetric strain during testing. The computation of stresses 79 was undertaken with appropriate corrections for variation in the cross-sectional area of the specimen. The eight triaxial tests were conducted using this approach. The sand specimens were compacted via tamping to reach target densities of 1575 and 1665 corresponding to relative densities of 69% and 100% respectively. At each density level, four tests was performed at effective confining stress levels of 15 kPa, 25 kPa, 35 kPa, and 50 kPa by applying different vacuum levels as appropriate. This range of confining stresses were selected after observation of preliminary numerical model results indicating that the minor principal stress in the soil elements during both axial pullout tests and lateral pullout tests would be in the order of 50 kPa. Due to the relatively low confining stress levels, the effect of membrane stiffness was judged to be significant on the results; hence, the results were corrected for the membrane effect as appropriate. Figure 3-17: Triaxial rest setup for testing dry Fraser River sand under relatively low confining stress levels (< 50 kPa) 80 Friction angle The peak friction angles computed from the data obtained from triaxial testing are shown in Figure 3-18. The results show that, as expected, the peak friction angle varies with both density level and effective confining stress. As may be noted, the peak friction angle drops by about 2° to 3° when the confining stress increases from 15 kPa to 50 kPa. At each density level, the relation between peak friction angle and confining stress in the range of 15 kPa to 50 kPa can be approximated by a quadratic equation. In order to assist the numerical modeling (as presented in later chapters), based on these results, it was assumed that the peak friction angle would be constant for minor principal stresses greater than 50 kPa. The variation of peak friction angle with density level for the tested relative density ranging from 68% to 100% was assumed to be linear. This approach allowed the peak friction angle at density of 1600 kg/m (Dr = 75%), which is the target density of backfill material in full-scale soil box testing, to be interpolated from triaxial test results for D r = 69% and 100%. 4J "5b e a c o QJ 50 49 48 47 46 45 44 43 42 ity. 1665 kg/ Dens 3 m ——~ILZ_~ 1 Density. 160( Intrapolated 1 t o / m 3 -V . s _K>fcfcc --J K g / 111 . Density. 1575 kg/m 3 ~~ "* — 1 1 10 20 30 40 Confining Stress (kPa) 50 60 Figure 3-18: Peak friction angle calculated from triaxial test results 8 1 Initial Elastic (Young's') Modulus The results of triaxial tests were also used to calculate initial elastic modulus (Ej), and shear modulus (Gj) to obtain constitutive model parameters for numerical analysis. As noted by Duncan (1970), plotting si / ad vs Sj , where 81 = axial strains, and o"d = deviatoric stress in triaxial testing can be approximated by a single line for the purpose of obtaining parameters for the hyperbolic stress-strain model. As an example, this plot for test with 1665 kg/m3 density and 50 kPa confining stress is shown in Figure 3-19. As shown in Equation [3-1] the intercept of this line is equal to (1/Ei) in which Ej is initial elastic modulus. £•, /(cr, -<r3) = Asx + B £ , - > ( ) = > £ , /(CT, -a3) = \/E, =>B = l/E, [3-1] 0.016 0.014 0.012 - 0.01 £ 0.008 " 0.006 0.004 0.002 0 _JI. = 0.0033x + 0 y 0011 s <f - - -1/Ei • el/(ol-o3) = Forel -> 0 : ... l /Ei = B , Ael +B £l / (al-o-3)-Ei= 1/B= 100 + 1/Ei / 0.0011 =900 00 kN/m 2 0 Figure 3-19: Calculation of initial elastic modulus using triaxial test results (test with D r = 75% and a ' 3 = 50 kPa) 82 This approach was repeated for all tests and the initial elastic modulus was calculated for tests with different relative densities and at different confining stresses. The calculated initial elastic moduli for different tests are shown in Figure 3-20. The variation of elastic modulus with confining stress at each relative density can be approximated by a power equation as shown on the graphs. The results indicate that when confining stress varies in the range of 15kPa to 50kPa, the elastic modulus increases by more than two-fold. Again, assuming linear variation of elastic modulus with relative density in the range of 69% to 100%, elastic modulus at target density of 1600 kg/m 3 can be interpolated. 0 1 0 2 0 3 0 4 0 5 0 6 0 Confining Stress (kPa) Figure 3-20: Initial elastic modulus calculated from triaxial tests results 3 . 4 Experimental limitations and associated errors As with any experimental testing work, the interpretation of data from physical model testing should be undertaken with due consideration of possible test errors. Some of the key errors and uncertainties associated with the pipe-soil interaction testing of this study are judged to arise from: chamber-boundary effects, control of backfill density (uniformity of test specimens), friction in pulling system during lateral pullout tests, and 83 methods of measurements. With particular reference to chamber-boundary, pullout of the pipe results in shearing of the sand around the pipe which in turn causes formation of elastic and plastic strains and redistribution of stresses in the backfill soil. During axial pullout tests, native soils at sides and below the pipe are being replaced by side walls and the laboratory strong-floor. The following sections present an evaluation of the effect of these considerations on the test results. The intent of this evaluation is to, if possible, modify the test configurations to minimize/eliminate the cause(s) of errors. Alternatively, the process allows obtaining an appreciation of the significance of these errors on the assessments made from test measurements. 3.4.1 Boundary constraints perpendicular to the direction of pipe movement (Sidewall friction) The lateral pullout tests of the current study (see Chapter 5) are expected to simulate a pipeline subjected to relative lateral horizontal soil movement in a 2-dimensional plane strain manner. The results from the test then could be used to assess maximum soil loads and pipe-soil interaction characteristics ("soil springs") on a per-unit-length-of-pipe basis. However, mobilized frictional force between soil and vertical side-walls of the box during lateral pipe pullout would add another resistance to the movement of the pipe (i.e., in addition to the resistance of the soil mass round the pipe). Clearly, the effect of sidewall friction would increase with the decreasing width of the testing chamber. If the width of the box is large, then the frictional force can be negligible. This sidewall friction effect can also be reduced by using appropriate material for the inside of the sidewall so that the interface friction angle between side wall and the soil backfill is less. Several methods that have been used in the past research work to reduce the sidewall friction can be found. For example, Audibert and Nyman (1977) used 2 layers of polyethylene at the side walls. The one in immediate contact with sand was cut into strips to move freely with the soil, while the second layer was attached to the box. This 84 method was intended to reduce the sidewall friction to friction between polyethylene layers. Trautmann and O'Rourke (1983) used a glass window on one side of the box and Formica on the other side to reduce the effect of sidewall friction. Paulin and co-workers (1998) at C-CORE used a 3 m wide steel box, and they did not use any additional material to reduce the side wall friction. Figure 3-21: Side walls covered with stainless steel sheets during lateral pullout tests In the current study, the vertical side walls that are 2.5-m apart were covered with 20Ga 304 stainless steel sheets as seen in Figure 3-21. The friction angle between the sheets and the sand was measured to be 20°. The frictional force at side walls can be calculated applying some simplifying assumptions. It is fair to assume that the frictional force would be largest for those tests having higher burial depths of pipe. Considering this, sidewall friction was calculated here for overburden ratio of 1.92 and pipe diameter of 457mm (representing Test No. LN-1 and LN-2). As schematically shown in Figure 3-22, failure surface is assumed to be triangular with failure angle of 30° to the horizon. The normal stress on the sidewalls was calculated from o' n = Ko .a ' v where a ' n , Ko, and a ' v are normal stress on the walls, coefficient of lateral earth pressure at rest, and effective overburden 85 stress respectively. It is judged that this simplified approach should provide a reasonable order-of-magnitude estimate of side friction. If Ko assumed to be 0.5, the frictional force is calculated to be about 0.9 kN per meter of the pipe (NOTE: the width of the box = 2.5 m). The low interface friction angle between sand and stainless steel sheets reduced the shearing resistance of sidewalls to half of that compared to a situation in which the backfill material is in direct contact with plywood panels. This computed sidewall friction force of 0.9 kN is less than 2% of the soil loads on the pipe that ranged was in the order of 50 kN/m (see Figure 5-48 in Chapter 5). Clearly, an error of less than 2% suggests that the effect of sidewalls is negligible for the tested configurations of the present study. Fn=2.JKQ.avxlA = 5.9\iN Ff = F„. tan(tf) = 2.14 kN (or 0.9 kN/m) Figure 3-22: The schematic effect of side wall friction during lateral pullout tests (H/D=2.5 and D=457 mm, Tests No. LN-1 and LN-2) The observation of sand particle movements as presented in Section 4.2.3.2 indicated that only particles in the immediate vicinity of the pipe are subjected to movement during axial pullout tests. This fact indicates that the presence of sidewall at distances in the order of 10 mm from the pipe surface is not considered to impact the performance during axial pullout tests. 86 3.4.2 Boundary constraints in the direction of pipe movement (front and rear wall effects) Ideally, the physical modeling of the real-life problem would be best achieved if there are no artificial boundaries in the front or rear sides during pipe pullout. Since such boundaries are needed to keep the scale of testing to manageable levels, efforts should be made to place these boundaries so that the impacts of their presence on the pipe-soil interaction mechanisms are minimal. In this section, the effects of boundaries in the direction of pipe movement are reviewed briefly on both axial and lateral pullout configurations. 3.4.2.1 Effect of boundary constraints in axial pullout tests Because of the steel-frame-design, the boundaries of the soil box can be assumed as rigid compared to the soil; hence, i f the width of the box or height of the pipe from the ground is not chosen properly, the measured soil loads on the pipe may be significantly different from those expected in the field. The effect of front and end walls on the test results during axial pullout were evaluated using: (a) direct measurement of stresses on the walls using a total earth pressure transducer as described in Section 3.1.5; and (b) using numerical modeling. The results from stress measurements indicate no noticeable change during pipe pullout. As may be noted from Section 4.5.3.3, the results from numerical modeling indicated that doubling the distance of front and rear walls and the laboratory strong-floor ground from the pipe did not change the computed stress distribution around the pipe. This indirectly suggested that side walls have no effect on pullout loads on pipe. In addition to this, as mentioned in Section 3.2.1, data from two axial pullout tests performed on pipes with different length, but otherwise identical configurations, were 87 available for this assessment. The computed loads per unit length of pipe were within 2% of each other in the tests; this, again, indirectly confirmed that the end walls had no significant effect on the soil loads during axial pullout tests of pipe. 3.4.2.2 Effect of boundary constraints in lateral pullout test Three distinct zones of sand displacement, as shown in Figure 3-23, during lateral pullout tests have been already identified (Audibert and Nymann 1977; Trautmann and O'Rourke 1983). One of the key parameters in selecting the length of the box is to allow free formation of displacement zones in the sand chamber. The length of the box during lateral pullout tests of the current study was 3.8 m which is 25% longer than the box used by Paulin et al. and Popescu et al. (1997), 70% longer than that used by Trautmann and O'Rourke (1983), more than 2 times of that used by Hsu (1992) and 4 times of the box used by Audibert and Nyman (1977). The use of a longer box gives the opportunity to test larger pipe sizes and higher overburden ratios. Figure 3-23: Zones of displacements during lateral horizontal pipe pullout as identified by Audibert and Nyman (1977) and Trautmann and O'Rourke (1983) 88 The shape and size of the failure surface, if purely based on Rankine theory, depends on the buried depth and the peak friction angle. For a peak friction angle of 45°, passive and active wedge in the front and the back of the pipe would have failure surface angles of 22.5° and 67.5°, respectively, in relation to the horizontal; however, Audibert and Nyman (1977) reported a higher angle for passive wedge based on their test results and Trautmann and O'Rourke (1983) reported the angle as high as 40° for dense sand. The axis of the pipe during horizontal lateral pullout tests is about 2.5 m from the front wall and 1.3 m from the back wall which are more than the lengths of influence of the failure surfaces based on above-estimated angles. Observations of the deformation on the surface of the sand backfill after each test also showed that both active wedge and passive wedge formed freely inside the box (see Section 5.3.3). In addition, the results of numerical model of the lateral pullout tests using different box lengths (see Section 5.7.3 for details) also suggested that the computed soil load on pipe was not affected when the length of the box in the numerical model was changed from 3.8 m to 7.6 m. 3.4.3 Pulling system friction As shown in Figure 3-7, steel cables having 50 mm diameter formed a key apt of the pulling system. Since these cable pass through the soil mass, a part of the load applied by the actuators to the pulling system is required to overcome friction between the pulling cable and soil mass. The length of the cables in horizontal pullout tests was about 2 m. To estimate the frictional force on the cables, in an independent test, a 2 m cable with free end was pulled with the soil cover of about 112 cm. The measured pullout load on the cable was about 3.5 kN. In tests with various cable length and buried depth, pullout load on the cables can be modified using commonly used approaches as indicated in ASCE (1984) and PRCI (2004). Based on conventional formulae, the axial soil load on the pipe (or cable in here) is proportional to the cable length and soil cover. The measured loads applied by actuators during lateral pipe pullout tests conducted as a part of the study were modified to consider the effect of soil/cable friction. 89 3.4.4 Control of backfill density A good control of the density (and uniformity) of the sand backfill is an important consideration in the preparation of test specimens for physical modeling. The changes in density wi l l reflect in the changes to friction angle, dilatancy, and deformation moduli, in turn, affecting the stress distribution in the soil mass during testing. A nuclear densimeter as explained in Section 3.1.5.7 was employed to measure the local density and moisture content of the backfill during 4 of the axial pullout and 8 of the lateral pullout tests. The sand was poured into the box using sand bags shown in Figure 3-24 (note: each sand bag contains about 0.9 m 3 of sand). The density measurements were performed at 4 to 5 random points after emptying 3 to 4 sand bags and compaction o f the layer. The average soil density in each axial pullout varied in the range o f 1594 kg/m to 1607 kg/m with average of 1603 kg/m . The average moisture content was also measured as 0.9% indicating essentially dry soil. Independent tests using sand retrieved from bowls buried in the backfill were performed to check validity of the nuclear densimeter measurements in Tests No . A B - 2 and A G G C - 2 . The measured density at two different points was 1590 and 1586 kg/m3 in Test No . A B - 2 and 1592 kg/m3 and 1612 kg/m3 in Test N o . A G G C - 1 , which was in line with density measurement via nuclear densitometer. The average moisture content in lateral pullout tests was 0.3% while the density varied in the range of 1592 to 1608 kg/m with average of 1600 kg/m . The low measured moisture content during the tests with dry sand (< 1%) indicates that the soil can be treated as cohesionless material. In the axial test with loose sand (Test No. A B - 5 ) , density was not measured during filling the box as it is believed that the sand density around the pipe would have changed by emptying sand bags and by inevitable walking on sand layers; hence to estimate the average density in this test, all the sand used for the tests with dense backfill was poured into the box. Knowing the height of the sand in this test and the total weight of poured 90 sand into the box from previous tests density and height measurements, the average density can be calculated. Figure 3-24: Emptying the sand bags into the box For Test No LTG-1 in which moist sand was used, the nuclear densimeter was employed to measure the sand density and moisture content. The average measured moisture content was in complete harmony with the volume of water added to the sand. A summary of density and moisture content measurement is presented in Table 3-6. The result of density measurements in axial and pullout tests are also shown graphically in Figure 3-25 and Figure 3-26 respectively. 91 Table 3-6: Summary of density and moisture content measurements Test ID Number of Measurements Average Density (kg/m 3 ) Standard Deviation of Density Average Moisture Content Standard Deviation of Moisture LN-1 19 1592 15.8 0.32% 0.24% LN-2 9 1594 22.2 0.33% 0.19% LN-3 16 1608 27.5 0.34% 0.25% LN-4 9 1607 28.0 0.33% 0.23% LN-6 17 1595 13.3 0.34% 0.16% LNG-1 17 1606 13.3 0.36% 0.22% LNG-2 12 1602 20.1 0.38% 0.28% LTG-1 10 1721 23.3 10.24% 2.50% Average Lateral 1 99 1600 19.2 0.34% 0.23% AB-2 22 1594 22.2 1.09% 0.39% AGGC-1 20 1606 14.0 1.07% 0.27% AGGC-2 20 1608 18.9 0.81% 0.28% ABGC-1 20 1608 22.2 0.71% 0.17% Average A x i a l 2 82 1603 20.1 0.92% 0.34% 1: The average based on measurements in tests LN-1, LN-2, LN-3, LN-4, LN-6, LNG-1 and LNG-2. LTG-1 is not included in the averaging as the material is different 2: The average based on measurements in tests AB-2, AGGC-1, AGGC-2 and ABGC-1 92 1545 1555 1565 1575 1585 1595 1605 1615 1625 1635 1645 1655 Density (kg/m3) Figure 3-25: Density measurement distribution in axial pullout tests 1545 1555 1565 1575 1585 1595 1605 1615 1625 1635 1645 1655 Density (kg/m3) Figure 3-26: Density measurement distribution in lateral pullout tests 93 3.5 Summary of the chapter The full-scale testing system was developed to study the response of buried pipes to ground movements. The system was specifically designed to accommodate testing of buried (bare) pipes, geosynthetic-wrapped pipes, and pipes buried in trapezoidal trenches with different trench backfill and native soil conditions. Development of this system included: • Large soil chamber: a soil box with constant width and height of 2.5 m and variable length of 3.8 m to 5 m was designed. The dimensions of the box were selected considering boundary effects during axial and lateral pullout tests. • Instrumentation: string potentiometers and load cells were used to monitor pulling loads and displacement of pipe and geosynthetic layers. Also a series of pressure transducers was mounted on the pipe surface to record normal stresses on the pipe prior, during, and after pullout. • Characterization of backfill material and interfaces: Uniformly graded, Fraser River sand was used as backfill material. The characteristic of this material was studied through laboratory testing including direct shear tests and triaxial tests. Moreover, to obtain reliable data, direct shear tests were performed on interfaces between soil and sand blasted steel and between geotextile layers. • Testing program: A total of 9 axial pullout tests on both bare pipes and geosynthetic-wrapped pipes were performed. Also 14 pulling tests in transverse direction were conducted on pipes buried in native soil and trench configuration, backfilled with different materials. • Experimental limitations: These limitations including the effect of side wall friction, pulling system friction, and boundary conditions and associated errors were investigated in this chapter. Where the effects found to impact the recorded data, modifications were made on row data to consider those effects. 94 CHAPTER 4 BURIED PIPES SUBJECT TO LONGITUDINAL GROUND MOVEMENTS 4.1 Introduction This chapter describes the results from axial pullout tests on buried steel pipe as well as findings from numerical modeling undertaken to capture the pipe response during axial pipe pullout. As described in Chapter 3, the test program consisted of nine tests using 457-mm (18-in) diameter steel pipes with about 91 cm of soil cover above the crown of pipes. The tests were conducted on buried pipe with and without various types of geosynthetic wrapping. Some of the key parameters related to the tests are summarized in Table 4-1 for the readers' convenience. Initially the experimental results obtained from axial pullout tests conducted on bare pipe are presented. This is followed by results from tests on pipes wrapped with geosynthetic layers. The experimental results are discussed and also compared with the currently recommended methods for prediction of axial loads on buried pipelines during ground movement. The detailed load-displacement response and other physical measurements for the axial pullout tests are presented in Appendix B. In order to facilitate comparison of different tests, the axial soil resistance is presented in the form of a normalized axial soil resistance as defined below: 95 (F A ' ) = F A / Y . H . 7 i . D . L [4-1] where: F A = measured axial load on pipe y = dry density of soil H = depth to centerline of pipe D = diameter of pipe L = length of pipe test section Table 4-1: Summary of parameters in axial pullout testing Test Configuration Axial Longitudinal Pullout Soil Type Fraser River Sand Average Density 1600 and 1430 kg/m3 for tests on dense and loose sand Average moisture less than 1% Internal Friction Angle Peak: 46°~43° for a ' 3 of 10 to 50 kPa, respectively for 1600 kg/m 3, Constant volume friction angle: 33° Box Size 5m L x 2.5m W x 2.5m H & 3.8 m L x 2.5m W x 2.5m H Pipe Size 457 mm x 12.7 mm (18 in x 0.50 in) - Pipe Length: 6m and 5m Pipe Grade & Surface Steel Grade A524, Sand Blasted Surface Soil - Pipe Friction Angle Dense Sand: peak = 36°, Constant volume friction = 31° Loose Sand: peak = 33°, Constant volume friction = 31° Geosynthetic Material TC Mirafi Filterweave 700 geotextile Rockshield bidirectional grid geonet Interface Friction Angle of Geotextile Layers Peak friction angle: 21° Post peak friction angle: 20° Maximum Pipe Displacement 400 mm to 800 mm Loading Rate 2 mm/sec - 60mm/sec 96 The value of F A ' represents the average shear force around the pipe normalized with respect to the vertical effective stress from the soil overburden at the centerline of the pipe. Similar normalization approach has been used by previous researchers to describe lateral soil forces. 4.2 Axial pullout tests on bare pipe buried in dry sand In this section, the results of different measurements during axial pullout tests on bare pipe are presented. The observed load displacement curves and pressure measurements on pipe surface during pullout where applicable are presented. Visual observations made in relation to movement of sand particles in the vicinity of pipe-soil interface during the tests are also presented. 4.2.1 Axial load vs. displacement response The observed normalized axial load resistance vs. displacement response during first loading for Tests N o . A B - 3 , A B - 4 and A B - 6 that were conducted on dense sand are presented in Figure 4 - 1 . Results for Test A B - 2 are not shown because, as mentioned in chapter 2 , a voltage overflow during first loading of Test No . A B - 2 caused a loss of data. Test No . A B - 4 was a repeat of Test N o . A B - 3 and the almost identical response displayed in Figure 4-1 demonstrates very good repeatability in the test preparation and pipeline response. Test No. A B - 6 , which was performed after 4 5 days from the date of specimen preparation, shows a slightly higher peak normalized axial soil resistance ( F A ' ) of about 1 0 % reflecting some likely effects due to aging of backfill. The value of peak F A ' observed for Tests N o . A B - 3 and A B - 4 is about 1 .02 , and for Test No . A B - 6 , it is about 1.13. In all three cases, the peak load was achieved at an axial displacement of about 7 mm. The post-peak F A ' values for all tests approached a constant value between 0 . 7 5 and 0 . 8 after reaching axial displacements in excess of about 2 5 0 mm. 9 7 Upon completion of the initial axial loading to displacements of about 200 mm to 250 mm, in which axial load reached a constant value, all the test specimens were subjected to unloading and subsequent reloading at least two times. Figure 4-2 shows the results obtained from these unloading and subsequent reloading sequences for Tests No. AB-2, AB-3, AB-4 and AB-6. The axial load reached during these tests was significantly lower than the initial loading tests and did not exhibit a prominent peak load. The average normalized axial soil resistance ( F A ' ) during the second loading was about 0.65 for all of the tests, compared to 0.75 to 0.80 post-peak F A ' observed during initial loading. The value of F A ' was approximately 10% lower in the third loading compared to the second loading. Little change in F A ' was observed between the fourth loading and third loading. The load versus displacement response for Test No. AB-5, which was performed with a loose sand backfill, is presented in Figure 4-3. The peak normalized axial soil resistance ( F A ' ) is about 0.42, and this value drops to about 0.37 when sheared to larger strain levels. Unlike for the tests on dense sand the value of F A ' during second loading is similar to that noted during first loading. The observed peak F A ' is less than half of the 98 peak load observed for Tests No. AB-2 and AB-3 conducted with the same depth of pipe burial, but in dense sand. 2 0 4 0 60 80 100 120 Displacement (mm) Figure 4-2: Load displacement response, Tests No. AB-2, AB-3, AB-4 and AB-6, during subsequent loadings after first loading/unloading 0.60 T 0.50 4 < Ex. * 0.40 o 0.20 0 . 1 0 0.00 50 100 150 200 Displacement (mm) — AB-5 — AB-2, Reverse loading L -1 - • 1 — 250 300 Figure 4-3: Load displacement response, Test No. AB-5 and reverse loading part of Test No. AB2 99 The results from the reverse axial loading part of Test No. AB-2 are also superimposed in Figure 4-3 for comparison. Reverse axial loading of Test No. AB-2 was performed after completing the initial and pulling and subsequent unloading and reloading of the pipe for about 800 mm (i.e., full stroke of the actuator) in one direction. As may be noted, the large strain F A ' observed for the loose sand Test No. AB-5 is comparable to the corresponding reverse loading value from reverse test on dense sand. 4.2.2 Soil pressures on pipe during specimen preparation and axial pullout As indicated in Chapter 3, the soil pressure on the pipe surface was monitored using five pressure transducers (PT1 through PT5) mounted on one side of the pipe circumference (see Figure 4-4). The soil pressures were monitored both during specimen preparation and axial pullout during the Tests No. AB-4, AB-5 and AB-6. Figure 4-4: Radial position of pressure transducers 100 4.2.2.1 Soil pressures during specimen preparation The normal stress on the pipe recorded at the transducers PT5 and PT3 are plotted in Figure 4-5 and Figure 4-6 with respect to the computed vertical effective overburden stress (a'vo) (equal to total stress, as dry sand was used) at the level of the transducer during backfilling of the box in selected tests. As shown in Figure 4-5, the measured vertical stress on the pipe at PT5 (crown of the pipe) is in good agreement with the computed overburden effective stress based upon the average soil density of 1600 kg/m . This agreement provides assurance of the suitability and sensitivity of pressure transducers and the adopted mounting approach to measure the imparted soil pressures even at relatively low stress levels. CC "a E i_ o Z •a y.< A • • k „ r < • A • Dense Sand (AB-4, AB-6) A Loose Sand (AB-5) 4 6 8 10 Computed Vertical Stress (kPa) 12 14 16 Figure 4-5: Measured soil stress normal to the pipe at transducer PT5 101 03 I. 73 E o V s 9 8 7 6 5 4 3 2 1 0 / / / • / ' K 0 = 1.0 • • / • • / < / • * .. KO = 0.4 • / / / / --/ A ' • Dense Sand (AB-4, AB-6) A Loose Sand (AB-5) 0 10 15 20 Computed Vertical Stress (kPa) Figure 4-6: Measured soil stress normal to the pipe at transducer PT3 The measured lateral stress at the location of transducer PT3 is shown in Figure 4-6 for Tests No. AB-5 through AB-6. The measured lateral stress in the case of dense sand specimens (Tests No. AB-4 and AB-6), is higher than that obtained for the loosely prepared specimen (Test No. AB-5). In Figure 4-6, it can be observed that the ratio between lateral and vertical stress decreases with increasing vertical stress for both dense and loose soil conditions. However, the rate of decrease in the ratio between lateral and vertical stress is greater for the dense soil conditions, resulting in a decrease in the difference in measured lateral stresses between the loose and dense specimens with increasing overburden stress. Because of the high structural stiffness of the steel pipe and also transducer's diaphragm, it can be argued that the measured stress at PT3 is an indication of the lateral earth pressure "at rest" (a'ho)- The measurements can be interpreted using the concept of coefficient of lateral earth pressure "at rest" (Ko): K. [4-2] 102 where: <j'ho = effective lateral stress a'm = effective vertical overburden stress As may be noted from Figure 4.8, a value of Ko in the order of 1.0 can be inferred from the stress measurements in the denser specimens at lower stress levels (say < 5 kPa), and the inferred Ko decreases with increasing overburden effective stress. These observations are in accord with the findings reported by Duncan (1986) and Carder (1976) from their tests on retaining walls with regard to lateral earth pressures. Figure 4-7 presents the measured pressures at transducer locations P T l and PT2. Clearly, the transducer readings show significantly lower stresses than the estimated vertical effective overburden stress at the PT l position, which is believed to be due to local arching. In addition, although a trend can be observed in the pressure measurements at PT2 for loose sand backfill, there is considerable scatter for dense sand, which can be contributed to the same effect (i.e. local arching). 12 • PTl - Dense Sand A PTl - Loose Sand o PT2- Dense Sand A PT2 - Loose Sand O O o o O A •<>" o ._o O A A • A * 0 m * 1 i • A ^  A • • A O • 1 1 • £10 t 8 t» S3 E u o Z cu L. 9 a 0> 6 4 4 4 § 2 10 15 Computed Vertical Stress (kPa) 20 25 Figure 4-7: Measured soil stress normal to the pipe at transducers P T l and PT2 103 4.2.2.2 Soil pressures during axial pullout testing Soil pressure transducers were monitored during pullout Tests No. AB-4, AB-5, and A B -6. The measured soil stress on the pressure transducers had high frequency "noise" during pullout possibly due to the localized particle movements and "abrasive" action at the soil/pipe interface. As such, moving averages of the measurements were computed to present the data; the "window" used for computations was varied using judgment to avoid visually unfair estimations that would have occurred in the event of using a constant "window". These average soil pressure readings were judged sufficient for cross- comparisons with axial load measurements and in identifying relative trends in the variation of soil pressures during axial pullout tests. In order to facilitate and ease of comparison, the measured pressures at a given transducer location (a'„) were normalized with respect to the computed vertical overburden effective stress (a ' v ) at the depth of the pipe springline to obtain the dimensionless normal stress,(o'N). [4-3] where: cr'n = averaged pressures at a given transducer location <j'v = computed effective overburden stress at the pipe springline The computed dimensionless normal stress on different locations of the pipe in Tests No. AB-4, AB-5, and AB-6 are shown in Figure 4-8, Figure 4-9, and Figure 4-10 respectively. 104 0.0 -f 1 i i 1 1 i 0 50 100 150 200 250 300 Pipe Displacement (mm) Figure 4-8: Dimensionless normal stress on the pipe during Test No. AB-4 Figure 4-9: Dimensionless normal stress on the pipe during Test No. AB-5 105 0.0 i i i i i ' i 0 50 100 150 200 250 300 Pipe Displacement (mm) Figure 4-10: Dimensionless normal stress on the pipe during Test No. AB-6 4.2.3 Other visual observations In addition to the above measurements, surface deformations and sand particle movement in the vicinity of the pipe/soil interface were also monitored, during some of the tests conducted on bare pipe to provide additional information with regard to understanding the basic failure mechanisms during axial pullout testing. 4.2.3.1 Surface deformations After completion of backfilling of the specimen for pullout Test No. AB-4, a rectangular grid pattern of twenty-four survey locations was established to capture possible surface deformations (uplift or subsidence) resulting from the axial pullout tests. The locations were surveyed before and after the tests with the position of each point measured with an accuracy of ±0.5 mm. Surveying was conducted using a Pentax transit, model GT4 obtained from the U B C surveying laboratory. 106 The depth of soil cover above the crown of the pipe in Test No. AB-4 was about 92 cm. The measurements indicated no detectable deformations of survey points before and after pipe pullout. In other words, any volume changes due to shearing within the annular soil mass in the vicinity of the pipe during pullout did not manifest as surface expressions. 4.2.3.2 De fo rmat ions at the Soi l /p ipe Inter face The movement of soil particles in the vicinity of the pipe was investigated by using colored sand backfill in Tests No. AB-4 and AB-5. Discrete patterns of colored sand were placed in the backfill on a horizontal plane at the level of the pipe springline. The colored sand, made by mixing Fraser Paver sand with red or black concrete mix-in color, provided well-defined contrasting zones in the backfill for convenient visual inspection before and after testing. After raising the backfill to pipe center line level in the specimen preparation for a given test, the colored sand was placed to form regular geometric patterns. Straight colored strips were constructed perpendicular to the pipeline and extending outward from the pipe surface towards the side walls "forming strips" (see Figure 4-11). Similarly, colored triangles were constructed immediately adjacent to the pipe as shown in Figure 4-12. The initial positions of the patterns were located by measurements and markings on the pipe and the walls of the box. Subsequent lifts of sand were gently placed to minimize the potential for disturbance to the colored patterns below. Removal of sand upon completion of a pullout test also had to be conducted with extreme care to avoid disturbance of the colored pattern. In this regard, a vacuum was employed to lift and remove the sand backfill above the colored areas. 107 Figure 4-11: Forming strips of colored to observe particle movements Figure 4-12: Colored sand patterns at the pipe centreline level prior to placement of remaining backfill during specimen preparation 108 After exposing the colored sand upon completion of axial testing changes in the colored geometric patterns (i.e., colored sand strips and triangles) were carefully measured. Figure 4-13 shows a typical post-test condition of a colored strip. The measurements of the deformations in the colored sand were made to an accuracy of about ±1 mm. Measurements from the three to four identical patterns were averaged to for use in interpretations. NI About 2 mm Pipe Surface O 0 .6 ' Figure 4-13: Sand particles displacement after pipe pullout Figure 4-14 was constructed using the average measurements made on the displacement of colored patterns (strips and triangles) for Test No. AB-4, although the observations made during Tests No. AB-4 and AB-5 were consistent. The observed soil movements (Figure 4-13 and Figure 4-14) indicate that nearly all shearing deformation occurs within about a 2-mm zone from the surface of the pipe. The soil particles located about 1 cm away from the pipe surface moved less than 2 mm in the direction of the pipe displacement while those located about 4 cm away from the pipe surface moved by negligible amounts (less than 0.5 mm). The details on measurements made on the colored patterns from the Tests No. AB-4 and AB-5 are presented in Appendix B. 109 Uncolored Sand into colored sand = 1.2-2 mm Y : Normal to pipe axis j o i j g co \o o II II II II No observation was made X — 37 mm Y « 0.4 mm X ~ 12.5 mm Y= 1.2-1 6 mm X ** 6.5 mm Y= 2.4-3.2 mm ;ure 4-14: Measurement of sand particles movement in Test No. AB-4 110 4.3 Axial pullout tests on steel pipe wrapped with geosynthetics buried in dry sand Three different test configurations for steel pipe wrapped with geosynthetics were tested: 1. Pipe wrapped with two layers of woven-geotextile in a "cigar-wrapped" configuration (tests labelled AGGC) ; 2. Pipe wrapped with two layers of woven-geotextile in a "spiral-wrapped" configuration (tests labelled AGGS); 3. Pipe wrapped with a layers of woven-geotextile over bi-directional geonet (tests labelled ARGC) ; A l l tests on wrapped pipe were performed in dense sand (target density of 1600 kg/m ) on a 457 mm diameter steel pipe with an H/D ratio of 2.5. The tests were performed within 24 hours after preparing the specimen. In all tests, the initial loading was followed by an unloading/reloading cycle. In Tests No. AGGC-1 and AGGC-2 , reverse loading was also conducted. 4.3.1 Axial Load vs. Displacement Response The measured axial loads were normalized to obtain the dimensionless axial resistance F A ' in a manner identical to that described in Section 4.2.1 for the bare pipe tests. 4.3.1.1 Tests with two layers of woven-geotextile Figure 4-15 shows the normalized axial soil resistance (FA ' ) versus displacement for the tests with two layers of geotextile. Tests No. AGGC-1 and AGGC-2 were identical, except for the changes in the rates of applied loading as discussed in the previous section. As may be noted, the peak normalized axial soil resistance (FA ' ) generated in the Tests No. AGGC-1 and AGGC-2 can be considered similar if the "mean" values of data are 111 compared so that transient peaks arising due to oscillations in Test AGGC-2 are excluded (see Section 3.2.1.2 for the reasoning for oscillation). Based on these two tests, an average F A ' of about 0.27 can be obtained for the "cigar-wrapped" configuration. On the other hand, the Test No. AGGS-1 with the "spiral-wrapped" configuration yielded an average F A ' of about 0.32. These resistance values are about one third of that of bare pipe in dense sand and about 25% less than that of bare pipe in loose sand. Figure 4-15 also indicate that no significant change takes place in the pullout resistance after unloading and reloading. 0.35 n 0.30 a J 0.20 — .2 0.15 0.10 t 0.05 0.00 AGGS-1 1 . L . VGGC-2 i f \ AGGC-1 AGGC-2 AGGS-1 1 —r- 1 100 200 300 Displacment (mm) 400 500 600 Figure 4-15: Load displacement response of Tests No. AGGC-1 , AGGC-2 , and AGGS-1 The observed response during reverse loading in Tests No. AGGC-1 and AGGC-2 are compared with initial loading in Figure 4-16. The average normalized axial soil resistance (F A ' ) is about 0.18 which is about 25% less than first pullout force. This might be a result of weakened annular sand zone around the pipe and reduction of normal stress on the pipe. 112 0.35 0.00 50 100 150 200 250 300 Displacment (mm) 350 400 450 500 Figure 4-16: Load displacement response: Reverse loading of Tests No. AGGC-1 and AGGC-2 4.3.1.2 Test with one layer of woven-geotextile and one layer of bi-directional geonet Figure 4-17 presents the axial loading response during the Test No. ARGC-1 , where the pipe was wrapped with a layer of woven-geotextile over a bi-directional geonet layer. Clearly the peak F A ' developed in this test is higher than that observed for those tests conducted using pipe wrapped only with geotextile. 113 Figure 4-17: Load displacement response of Test No. ARGC-1 4.3.2 Displacement of geosynthetics during axial pullout testing The displacement of inner and outer geotextile layers for Tests No. AGGC-1 and A G G C -2 are shown in Figure 4-18. As may be noted, the inner first layer of geotextile appears to move almost in harmony with the pipe. The second outer layer appears to remain stationary with the surrounding soil, indicating the establishment of full slippage between the two layers of geotextile from the very beginning of the pullout process. The measured displacements of the geotextile layers in the Test No. AGGS-1 are shown in Figure 4-19. For the spirally-wrapped configuration, there is some minor slippage between the pipe and the first inner layer, and significant slippage between the first and second layers. The results plotted in Figure 4-19 also indicate slippage between the pipe and the first layer increases at larger displacements. This behaviour is believed to be related to deformations occurring in the shape of spiral wrapping from its original configuration. It is clear from Figure 4-19 that the second outer layer did not move during the test. 114 800 0 100 200 300 400 500 600 700 800 Pipe Displacement (mm) Figure 4-18: Geotextile layers displacement; Tests No. AGGC-1 and AGGC-2 The measured displacements of the geotextile (outer) and geonet (inner) layers for the Test No. ARGC-1 are shown in Figure 4-20. As the width of the geonet was limited, four 115 pieces of geonet were required to cover the test length of pipe. The geonet pieces were attached to each other using duct tape with a small overlap between the separate pieces, as shown in Figure 4-21. The displacement recording wire was attached to the middle piece of geonet. It is possible that the discontinuity of the geonet pieces could have permitted non-uniform movement, leading to interference between pieces during pullout. Since the outer layer was continuous, its recorded displacement can be argued to represent the displacement of the whole layer in unison. In the first 100-mm of axial pipe movement, the inner layer of geonet appears to move with the pipe while the second outer layer of geotextile remains stationary with the surrounding soil. Slippage between the geonet and the pipe occurs during 100 mm to 200 mm of axial pipe displacement. 0 100 200 300 400 500 600 700 800 Pipe Displacement (mm) Figure 4-20: Geosynthetic layers displacement; Test No. AGGS-1 116 Figure 4-21: Geonet material wrapped around the pipe in Test No. ARGC-1 4.4 Discussion of test results 4.4.1 Axial pullout tests on bare pipe buried in dry sand 4.4.1.1 Comparison of measured versus predicted axial soil loads for bare pipe The variation of dimensionless force F A ' with axial displacement obtained from the current study is compared herein with those computed using the formula given in Equation [4-4] as per A S C E (1984) and PRCI seismic design guidelines (2004). = y-H.([-^-).tm(5)in.D.L) [4-4] 117 Where: FA= axial load on pipe (soil resistance) D = external pipeline diameter L = buried pipeline length Y = unit weight of soil H = depth from ground surface to centreline of the pipe Ko = coefficient of lateral earth pressure "at rest" 5 = interface angle of friction between soil and pipeline The calculated axial pullout loads using Equation [4-4] for loose and dense sand are shown in Figure 4-22 and Figure 4-23, respectively, along with the corresponding test results from the axial pullout tests. The values of Ko from pressure transducer readings taken prior to pipe pullout as 0.37 and 0.42 for loose sand and dense sand, respectively (see Figure 4-6; pressure measurements at vertical stress level of 15 to 20 kPa) and the interface friction angle (5) from direct shear tests (peak values of (5) of 33° and 36° for loose and dense sand, respectively, and a large strain value of 31 0 for both loose and dense sand) were used to compute F A -The results of direct shear tests on soil-sand blasted steel interface indicated that the peak friction angle mobilizes within 1 mm of differential displacement at the interface and then friction angle gradually drops to a constant value after about 1 to 2 mm of movement. The results of axial pullout tests also indicated that the peak soil loads on pipe are mobilized in the range of 5 to 10mm of pipe pullout (see Figure 4-1, Figure 4-2, and Figure 4-3). Complexity of strain contours around the pipe and lack of data makes it impossible to identify differential displacement at interface at peak soil loads during pullout tests. This argument indicates that the mobilized friction angle at interface and at peak soil loads on pipe is between peak and residual interface friction angle as measured during direct shear tests. From Figure 4-22, it is clear that the dense sand test results (Tests No. AB-3, AB-4, and AB-6) exhibit much higher axial resistance than that predicted using Equation [4-4]. This 118 contrasts with the test results for loose sand (Tests No. AB-2 and AB-5) as shown in Figure 4-23 that are in good agreement with Equation [4-4]. 0 50 100 150 200 250 300 350 400 Displacement (mm) Figure 4-22: Comparison of test results with predictions using A S C E (1984) formula -pipe in dense sand The results of reverse loading in Test No. AB-2 are also shown in Figure 4-24. Comparison of the axial soil loads on the pipe during reverse loading with computed soil loads from Equation [4-4] for loose sand indicated that they are in good agreement. The results of test in loose sand are superimposed in the same figure. It can be observed that the axial soil loads on the pipe during reverse loading in dense sand exhibit values close to those observed on pipe buried in loose sand. This is believed to be related to a formation of a loose soil structure around the pipe as a result of shear deformations at the pipe interface during initial and repeated loading. Since the reverse loading test is conducted on pre-sheared sand, it is likely that the large strain friction angle of sand would mobilize from the very beginning of the reverse loading test and produce results that are more comparable with loose sand. 119 0.60 0.50 < u. •6 M o -J 0.40 -h jS 0.30 a c E 0.20 0.10 0.00 K0=0.37, 8=33° 50 Prediction of Soil Load: K0=0.37, 5=31 100 150 200 Displacement (mm) A B - 5 - - A S C E , A L A 250 300 Figure 4-23: Comparison of test results with predictions using A S C E (1984) formula pipe in loose sand 0.60 0.50 < K0=0.37, 8=33° A B - 5 - - A S C E , A L A — - A B - 2 , Reverse load ing Prediction of Soil Load: K0=0.37, 8=3 T 50 100 150 200 Displacement (mm) 250 300 Figure 4-24: Comparison of test results on pipe in dense sand during reverse loading with those on pipe in loose sand and predictions using A S C E (1984) formula 120 4.4.1.2 Role of K 0 in observed differences between tests and predictions Instead of using a Ko value as a known quantity, it was decided to back-calculate the values of K needed in Equation [4-4] to obtain a match with the results from tests with dense sand backfill (Note: Since Ko is the lateral earth pressure coefficient at rest, the use of symbol " K " , indicating a lateral stress coefficient, is proposed herein to represent the back-calculated values of "Ko" using Equation [4-4]). This approach is considered rational since there is a good understanding of the interface friction angle (5) from direct shear tests. As evident from the Figure 4-25, taking the steel-sand interface friction angle (8) to be between 31° and 36°, a K value of between 2.5 and 1.8 is estimated to obtain correspondence with peak axial resistance from Tests No. AB-3, and AB-4. This value would be about 2.0 immediately after drop of peak load. At large displacements and after unloading and reloading, the corresponding K value computed assuming 8 = 31° drops to about 1.5 and 1.0 respectively. The back-calculated coefficient of lateral earth pressure at peak F A ' for Test No. AB-6 shows slightly higher values that can be associated with aging effect, as the test specimen was left for 45 days prior to testing. The significant difference between back-calculated values of K and those based upon Ko measurements just prior to the tests suggests that there is a substantial increase in the value of the normal soil stresses on the pipe during pullout. The soil pressure measurements on the pipe undertaken during some of the pullout tests provided an opportunity directly to investigate the validity of this thinking. 121 1.20 150 200 250 Displacement (mm) 300 350 Figure 4-25: Comparison of measured axial pullout loads and predicted loads using different K values The normalized pressure (CJ'N) at each transducer location vs. axial displacement during Tests No. AB-4 and AB-6 with dense sand backfill (see Figure 4-8 and Figure 4-10) were available to assess the stress changes on the pipe in dense sand during pullout. The averaged values from the two tests were computed and plotted in Figure 4-26 for the first 100 mm of the pipe displacement. For each transducer location, the O ' N value corresponding to displacement at which the peak F A ' value occurred (about 10 mm of displacement) were extracted from Figure 4-26 and presented as a radial plot Figure 4-27 (right side). The O ' N values under "at-rest" conditions computed from the transducer measurements prior to pipe pullout are also plotted in Figure 4-27 (left side). 122 0.0 H i i i i i 0 20 40 60 80 100 Pipe Displacement (mm) Figure 4-26: Normal stress on the pipe: average of Tests No. AB-4 and AB-6 (tests on dense sand) "at-rest" conditions Corresponding stresses at peak pullout load * Assumed values based on calculations Figure 4-27: Normalized soil stress O ' N on the pipe just prior to and during axial pullout in Tests No. AB-4 and AB-6 (Computed using stress measurements) 123 As pressure measurements of PT1 during specimen preparation showed, it appears that potential arching of soils around the pipe is the leading cause of reduction of soil pressures on the bottom of the pipe. The scatter of the results in the same pressure transducer during sample preparation and pullout is likely a result of non-uniformity in the initial stress at contact points of pipe and soil in different tests, the variability of the soil densities around the pipe and the resulting arching effects. As such, the pressure transducer mounted at the bottom of the pipe is unlikely to represent the actual average stress along the pipe (at PT1). For this reason, normal stress at the bottom of the pipe used in constructing Figure 4-27 was assumed to be the calculated overburden stress. As may be noted from Figure 4-27 the normalized stress value at PT1 of 1.2 has been assumed to remain constant throughout the test. In an overall sense, it is clear that the normal stresses on the pipe during pipe pullout of Tests No. AB-4 and AB-6 increased significantly in comparison to the "at-rest" values. As may be noted from Figure 4-27, the largest stress increase occurred at the springline. This change in normal stress during pipe pullout can be explained in terms of dilation (volumetric expansion) of the soil. Dense sand would exhibit a tendency to dilate as it undergoes shear deformations at the interface during pullout. However, the tendency to dilate in the horizontal direction is significantly constrained by the surrounding soil, leading to the observed increase in lateral soil stress (up to 5 times the "at-rest" values) at the springline of the pipe. The normalized pressure (O'N) at each transducer location vs. axial displacement during the Test No. AB-5 with loose sand (see Figure 4-1) is re-plotted in Figure 4-28 to display the response for the first 100 mm of pipe displacement. In a similar manner to Figure 4-27, the O ' N values corresponding to the occurrence of the peak F A ' value in Test AB-5 were extracted from Figure 4-28 to develop the radial distribution of O ' N shown in Figure 4-29 (right side). The O ' N values from transducer measurements prior to pipe pullout are plotted on the left side of Figure 4.31. In an overall sense, the O ' N on the pipe during pullout in Test No. AB-5 did not change significantly from those observed under "at-rest" 124 conditions. This result is believed to be a result of a reduced tendency of the loose sand to dilate during shearing deformations. The variation of normal stress around the pipe showed that the average normal stress on the pipe surface is not necessarily equal to the average of vertical stress and normal stress on the surface (horizontal stress) at the pipe springline. On the other hand the average O ' N value over the perimeter of the pipe by definition would be equivalent to the term (l+K)/2 in Equation [4-4]. Hence it can be concluded that K value is not necessarily equal to the ratio of horizontal stress to vertical stress at pipe axis level. Based upon this, the values of K for the tests in dense and loose sand were computed by averaging distributions of O ' N presented in Figure 4-27 and Figure 4-29; these K values are compared in Table 4-2 with those values back-calculated from the measured axial loads and using Equation [4-4] as per Section 4.4.1.1. The value of lateral earth pressure at-rest or Ko value, estimated from the pipe stress measurements shown in Figure 4-6 is also shown in the same table. "b </ tsi u Vi E e e c 4) E 5 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 / PT5 / PT4 \ - - PT5 (top) PT4 (+45°) PT3 (Horizontal) —-PT2 (-45°) - - PTl (Bottom) A / \ J  N . s ""[ PT3CT \ PT2 / V PTl J> <— -^/—PT5 s N. .>«»_.. \^PT4 PTl 1 PT3 25 50 75 Pipe Displacement (mm) 100 Figure 4-28: Normalized soil stress O ' N on the pipe: test AB-5 loose sand 125 * Assumed values based on calculations Figure 4-29: Normalized normal stress on pipe during test AB-5 There is a general correlation between the K values back-calculated from axial load measurements and values obtained from soil pressure measurements. This is taken as support for a conclusion that relatively high K values developed during pullout as a result of constrained dilative behaviour of the soils during shear. Table 4-2 also shows that the back-calculated K values from axial loads on buried pipes in dense sand and those computed from pressure transducers measurements are higher in several folds than "at-rest" lateral soil pressure. This increase in normal stress is believed to be associated with constrained dilation of the dense sand during shear deformations. Using Ko value in Equation [4-4] to predict axial soil loads on pipes as suggested in ASCE (1984) and PRCI (2004) greatly underestimates the peak load when pipe is buried in dense material. 126 Table 4-2: Comparison of K values back-calculated from axial load vs. soil pressure measurements Test backfill density Back-calc. from axial load measurements on pipe (computed using Equation [4-41) From soil pressure measurements on pipe (computed from O ' N ) Lateral soil pressure on pipe springline prior to pullout (Ko) Dense sand 2.5 to 1.8 (Figure 4-25) 2.5 (Figure 4-27) 0.42 (Figure 4-6) Loose sand 0.37 (Figure 4-23) 0.23 (Figure 4-29) 0.36 (Figure 4-6) Similar trends have been noted in pile research (Kraft 1990, Randolph et al. 1994, Jardine et al. 1996, Foray et al. 1998, Lehane et al 2001). 4.4.1.3 B e h a v i o u r at in ter face shear zone As can be observed from Figure 4-14, and previously mentioned in Section 4.2.3.2, the sand in vicinity of the pipe is being sheared during pullout. Dilation happens not only at the interface of sand and pipe, but also in the sand in vicinity of the pipe. The dilation in the sand and interface results in increase in normal stress on the pipe. Dilation angle is high at the start of pullout decreases with increase in strain until reaches zero value (it happens at the strain level in which friction angle decreases to a constant value). It can explain the fact that the pullout load remains constant after third reloading and unloading. It is of interest to note that the above observations on a limited annular shear zone around the pipe surface is in accord with the inferences obtained by surface measurements during pullout tests. 127 4.4.2 Observations and discussion on axial pullout tests on pipe wrapped with geosynthetic layers 4.4.2.1 Effectiveness of geotextile-wrapped pipe in reducing soil loads The results of axial pullout tests conducted with and without wrapping are compared in Figure 4-30. It is clear that the use of geotextile-wrapping provides an effective means of reducing axial soil loads in comparison to those mobilized for the case of bare pipe. The suggested formula in A S C E (1984) and PRCI seismic design guideline (2004) can also be used to predict the loads on wrapped pipe tests with the interface friction angle taken from geotextile interface friction tests. The predicted load using an interface friction angle of 20° (from direct shear test results) and K 0 value of 0.42 (from soil pressure measurements made in bare pipe buried in dense sand) compare well with the test data as shown in Figure 4-30. In addition to having a lower interface friction angle, it is also possible that the two-layer geotextile-wrapping contributes to a reduction in the normal earth pressures on the pipe during pullout. The presence of two layers of wrapped geotextile introduces a somewhat "flexible and irregular" composite interface between the pipe and dense soil, as opposed to the almost rigid boundary encountered by soil when compacted against a bare pipe. The flexibility of this geotextile composite interface could prevent the development of high K values for sand in case of potential sand disturbance. More importantly, as displacement of geosynthetic layers during pullout indicates, slippage surface is promoted to occur between geotextile layers. Therefore the sand adjacent to the pipe is less likely to shear and as a result, no dilation is expected. 128 100 200 300 400 Displacment (mm) 500 600 Figure 4-30: Comparison of load displacement curve of Tests No. A G G C - 1 , AGGC-2, and AGGS-1 with predictions using ASCE (1984) formula and tests on bare pipe Unlike for the "cigar-wrapped" pipe, the relative direction of the weaving between the geotextile layers is not aligned in the "spiral-wrapped" pipe (see Figure 3-9). The alignment of overlaps between 1.8-m wide spiral wrap of the inner first layer are also at oblique angles to those of the outer second layer above. These factors may have resulted in an increase in the interface friction characteristics for the spiral-wrapped pipe specimen. More importantly, the spiral-wrapped configuration allows non-uniform movement of the inner or outer layer as one section of the spiral can slide relative an adjacent spiral section. Differential sliding between different spiral sections of the same layer of wrap will lead to expansion or contraction of the spiral wrapping. A l l of these factors are suspected of playing some role in the increased large-displacement axial resistance for the spiral-wrapped pipe (Test No. AGGS-1) in comparison to the cigar-wrapped pipe (Tests No. AGGC-1 and AGGC-2) shown in Figure 4-30. 129 4.4.2.2 Effectiveness of pipes wrapped with geotextile/geonet in reducing soil loads In the case where the pipe was wrapped with a layer of woven-geotextile over geonet (Test No. ARGC-1 , see Figure 4-21), the test results (Figure 4-17) indicate that jamming of the geonet at the seams and also separation of geonet segments could lead to increases in axial soil load. This suggests that the use of "geotextile-over-geonet" wrapping configuration may not be effective in reducing axial soil loads; however other configurations of "geotextile-over-geonet" with continuous geonet wrapping should be tested to address this issue. 4.4.2.3 General comments Comparing the wrapped pipe test results in Figure 4-17 and Figure 4-3 with the bare pipe test results in Figure 4-1, it can be seen that the initial part of the load displacement curve is steeper than what was observed in bare pipe tests. In bare pipe tests, peak axial load is mobilized after 5 mm to 10 mm of pipe displacement compared to almost immediate mobilization of the axial load (within first 2 mm) for the wrapped pipe tests. Furthermore, the noticeable peak in the bare pipe tests is absent in the wrapped pipe tests. The close values of peak and post peak interface friction angle between geotextile layers can explain the absence of a peak in Tests No. AGGC-1 , AGGC-2 , and ARGC-1. As mentioned in Chapter 3, because of almost immediate slippage between geotextile layers in direct shear tests, the peak interface friction is reached at much lower displacements compared to that noted for sand-sand interfaces. This behaviour is consistent with the observation of a steeper initial load-displacement curve for pipe specimens wrapped with two layers of geotextile. 130 4.5 Numerical modeling The results from full-scale pullout tests performed on 457 mm (18") sand-blasted steel pipe provides an opportunity to understand the development of axial soil loads on pipe during ground movement. The results can also be used to assess existing analytical approaches and validate numerical models developed to simulate this problem. The validated model then can be used to examine the effects of different soil and geometric parameters affecting pipe pullout load. The analysis was conducted using F L A C 2D® version 4.0. The program uses a two-dimensional explicit finite difference method based code. Material is represented by elements within an adjustable grid to fit the shape of the modeled object. The elements are quadrilateral with no intermediate node. This software is based on a "Lagrangian" calculation which is well suited for large deformations and material collapse. F L A C allows for yield and flow of material and while large strain mode is used, the grid can deform and position of nodes are updated after each step. Additional information regarding this software can be found in F L A C 4.0 manual. Because of the relatively shallow burial depth, a complete numerical modeling of the axial pullout condition would involve a 3-D simulation. Considering the extensive time, cost, and effort associated with 3-D modeling, it was decided to simulate the problem in a pseudo manner using plane strain 2-D analysis. The selection of parameters and constitutive models for the representation of soil behaviour was completed as a part of numerical analysis undertaken to simulate lateral soil loads on pipes. The selected soil constitutive model is a modified hyperbolic model to account for the effect of density change on material properties. Selection of model parameters is mainly based on the element testing and the developed model is validated with the results of full-scale testing. Unbonded interface elements with Coulomb shear-strength criterion were used to model the interface between pipe and soil. The reader is referred to Sections 5.7.2.1 and 5.7.2.4 for more details which are not repeated herein for brevity. 131 4.5.1 Numerical modeling of axial pullout considering a vertical plane normal to the pipe axis 4.5.1.1 Consideration for the development of model Observation of particle movements around the pipe using colored sand as described in Section 4.2.3.2 indicated that only a small annular zone in the immediate vicinity of the pipe is being actively sheared (1.2 to 2.8 mm as shown in Figure 4-14). It is also known that the axial soil load on the pipe is a function of the normal stress on the pipe. The potential for dilation could cause sand in the sheared area around the pipe (Figure 4-31) to increase in volume. Since this increase in volume in the sheared zone is constrained by the surrounding soil mass, it will likely result in an increase in the radial stress in the annular zone. Particularly considering that the sheared annular zone is thin (<3 mm), this should also cause an increase in the normal stress on the pipe. Shear zone boundary (before dilation) Shear zone boundary (after dilation) b Pipe surface (after expansion) Pipe surface (before expansion) Figure 4-31: Modeling the effect of dilation on soil loads on pipe 132 The fundamentals of the axial pullout problem and proposed approach above are similar to the classical cavity expansion problem to some extent. In most solutions, those problems with an axisymmetric model are analysed (Chadwick, P., 1959; Vesic, A. S., 1972; and etc). In the buried pipeline problem herein, due to the direction of gravity, and the ground surface nearby (see Figure 4-36 and Figure 4-38 for more clarification), the axisymmetric solutions are not applicable, although symmetric cavity expansion solutions may be extended for deep buried pipes. The solution of ground subsidence during tunnelling as described by Robert Mair (2006 Rankine Lecture: Tunnelling and Geotechnics-New Horizons) is an example where simplified approaches have been used to address a problem similar to the buried pipelines. In spite of the possible limitations, at the beginning of the present research, symmetric cavity expansion solution was attempted to examine the stress increases around pipe during axial pullout without much success. With the above background, it was considered more appropriate to use a 2-D numerical model, considering a vertical plane perpendicular to the pipe axis as shown in Figure 4-31-b. Because of the relatively small thickness of the shear zone, it was considered acceptable to assume that the perimeter of pipe and the outside perimeter of the shear zone are coincident. It was also assumed that the effect of volume change that takes place in the shear zone can be simulated by expanding the pipe "numerically" by the same amount as the estimated dilation effect. Herein the objective is to obtain a numerical prediction of the increase in radial stress of the shear zone in the real life case. Clearly, the axial pullout that takes place normal to the plane of the paper is not modelled in this analysis. The underlying premise is that it is more critical to have a model that is capable of modeling the shear zone and capturing the changes in radial (normal) stress distribution around the pipe than a model that simulates the longitudinal direction. 133 4.5.1.2 Developing of numerical model The numerical model configuration developed on the above basis is shown in Figure 4-32. The size of the model, position of the pipe and boundary conditions were set similar to the experiment. The model consisted of 1073 soil elements and 38 beam elements that formed the pipe. The wall thickness and rigidity of the pipe were chosen similar to the pipe that was used in the full-scale testing. The locations of pressure transducers mounted on the pipe in the real life test are also shown in the same figure for completeness. A hyperbolic elastic model with Mohr-Coulomb failure criterion was used as the constitutive model for the soil similar to the one used in modeling of lateral pipe pullout tests and described in Section 5.7.2.4. Free surface Displacements in both directions are restrained Figure 4-32: model geometry and mesh size for modeling of axial pullout tests 134 4.5.2 Thickness of shear zone and the degree of dilation at interface In order to model the pipe as per the proposed numerical model configuration (see Figure 4-32), it is critical to provide a realistic value for change in the thickness of sand around the pipe. With this objective in mind, the thickness of the shear zone and the associated expansion of the shear zone during pullout in dense sand were estimated considering the information available from the following sources: (a) Measured deformation of colored sand zones during axial pullout tests of this study; (b) Previous research conducted by others (Roscoe 1970, Bridgewater 1981, Scarpelli and Wood 1982, Palmeira and Milligan 1989) (c) Measured vertical strains in direct shear box tests conducted as a part of this study. As suggested by Roscoe (1970) and Bridgewater (1981), the thickness of actively sheared zone is about 1 Odso; this was supported later by observations via radiography method by Scarpelli and Woods (1980). Based on that, the thickness of shear zone can be estimated 2.3 mm for Fraser River sand. (i.e. 0.23 mm x 10). On the other hand, measurements in colored sand zone (Figure 4-13 and Figure 4-14) indicated that the shear zone thickness is about 1.2 to 2.8 mm. Since the shear zone during axial pullout tests can be assumed to be fully sheared and constant volume condition is reached, the estimate of lOdso for shear zone thickness is comparable with those measured during the tests using colored sand. The comparison exhibits a good agreement. The observations during direct shear tests of Fraser River sand was available to assess the level of dilation when sheared up to constant volume phase (see Appendix A for detail results of direct shear tests). In this regards, the tests on compacted sand with an initial average relative density of 75% indicated that the average vertical deformation is about 135 0.72 mm, which is almost 30% of the thickness of the active shear zone (2.3 mm) estimated based on guidelines by Roscoe (1970). With this information in mind, and considering the fact that relative density in full-scale tests are slightly higher than those in direct shear tests, it was considered reasonable to assume a dilation of 0.7 to 1 mm for the purpose of numerical modeling. The particle movements around the pipe were also observed in an independent axial pullout test on a polyethylene pipe buried in dense Fraser River sand. The pipe diameter and soil cover were 114 mm and 600 mm respectively. The interface friction between PE pipe and the dense Sand is in the range of 0.5cp and 0.6cp. The deformation of sand particles around the pipe after 60cm of pipe pullout is shown In Figure 4-33. The particle movements were observed at 5 different points and the average thickness of the sheared area was measured in the range of 1.6mm to 2.0mm. The thickness of active shear zone as observed in three different configurations of pipe and backfill material (i.e. (i) steel pipe and compacted sand, Tests No. AB-4 and AB-6; (ii) steel pipe and loose sand, Test No. AB-5; and (iii) polyethylene pipe and compacted sand) exhibited close values, indicating that the thickness of shear zone in axial pullout tests can be assumed independent from the test condition and calculated form t = lO.dso as suggested by Roscoe (1970). 136 Figure 4-33: Particle movement at pipe springline level in axial pullout test on a PE pipe 137 4.5.3 Validation of the model and discussion of the results As described earlier, normal stresses on the pipe at five positions were measured during filling of the box and pipe pullout, and they were available for comparison with those computed from numerical modeling. Calculated average normal stress on the pipe from the numerical model could also be compared with average stress back-calculated from pullout loads in experimental model. 4.5.3.1 Stresses under static conditions prior to pullout To simulate the condition around the pipe and before "expanding" the pipe, the soil weight was applied in two steps during numerical modeling. At first, the soil up to the level of pipe invert and the pipe on top of it were modeled, and iteration was continued to reach equilibrium and then the weight of soil from pipe invert level to the level corresponding to desired overburden ratio (2.5) was applied to the model. At this stage, the iteration was continued to achieve full equilibrium before expansion of the pipe. The contours of computed horizontal and vertical stresses prior to simulating "expansion" of the pipe are shown in Figure 4-34. Normal stress on the top (PT5) and bottom (PT1) of the pipe was directly available from the computed vertical stresses at those levels, and at springline level (PT3) it is equal to the computed horizontal stress. Also considering Mohr's circle, normal stress at position of PT2 and PT4 (-45° and +45° to the horizon respectively) could be computed using Equation [4-5]. + <7 o-±4r = yy + T xy [4-5] 138 Effec. SXX-Stress Contours M - I 50E+04 m -1 20E+04 • -9.00E+03 • -6.00E+03 • -3.00E+03 • O.OOE+00 Contour interval 1.50E+03 (a) Effec. SYY-Stress Contours -3.O0E+O4 -2.40E+04 -1.80E+Q4 m -1.20E+04 -6.O3E+03 • O.GDE+GO Contour interval 3.00E+03 (b) * Units: Pa Figure 4-34: (a) Horizontal and (b) vertical stresses contours prior to expanding the pipe The computed normal stress (a'n) at locations PT l through PT5 were non-dimensionalized with respect to the vertical effective stress at springline; these results are compared with the normalized measured stresses at the same locations in Figure 4-35. [Note: the values in Figure 4-35 are average from two or three neighbour elements in numerical model] 139 Numerical model results * Assumed values based on calculations Measurement of pressure transducers Figure 4-35: "at-rest" stress distribution around the pipe; comparison of numerical model results and pressure measurements via transducers As may be noted, the normal stresses at position of PT3, PT4, and PT5 are comparable with those measured during physical tests. During filling the box with sand, the areas in level of PT2 are hard to access, therefore it is expected that the sand in that level is looser than the average sand density and it can explain measurement of relatively lower normal stresses at PT2 compared to calculated normal stress from numerical modeling results. Also the result of numerical modeling indicated that the assumption of normal stress at position of PT1 to be equal to overburden stresses at that level seems reasonable. 4.5.3.2 Normal stresses on pipe during pullout The horizontal and vertical stress contours after 1 mm of expansion of the pipe which represents dilation at interface during pipe pullout are shown in Figure 4-36. A significant increase in the horizontal stress in the soil mass around the pipe in comparison 140 to the static condition can be seen from the stress contours. This horizontal stress increase seems to reduce rapidly with increasing radial distance from the pipe surface. For example at the end walls (AA' and B B ' in Figure 4-36), horizontal stress increase is negligible compared to that at the pipe surface. In the same figure, a noticeable increase in the vertical stress at top of the pipe (position of PT5) occurs after expansion of the pipe (compare Figure 4-34 and Figure 4-36); this is in agreement with the measured increase in stress at PT5 as shown in Figure 4-27. The calculated normalized stresses from numerical model results (after expansion of the pipe for 0.7 mm to 1.0 mm) are compared with those measured from pressure transducers in Figure 4-37. [Note that due to discontinuity of beam elements, for a better estimation of normal stress, the values shown in this figure is average normal stress at 2 or 3 neighbour elements]. The stress distribution in "at-rest" conditions as shown in Figure 4-35 are superimposed in the same figure for comparison purposes. Significant amount of increase is observed due to dilation of sand at shear zone. The dashed curves in Figure 4-37 represents the suggested stress distribution around the pipe by ASCE (1984), considering linear variation of stress between crown, springline and invert of the pipe. Examining this figure indicates that suggested stress distribution around the pipe exhibits comparable values with those prior to pullout; however during pullout, both measured and computed normal stresses on the pipe are in agreement that normal stresses interpreted from ASCE (1984) are greatly underestimated. 141 Effec. SXX-Siress Contours -4.00E-04 -3.2DE*D4 -2.4QE^04 -1.60E-04 -8.0QE+O3 0.00E+0D Contour interval= 4.00E+Q3 A ' Effec. SW-Stess Cantons -4.00E^04 -3.20E^04 -14DE-04 -1.60E-D4 O.OOE+OD V ' Contour irrterval= 4.DOE+03 * Units: Pa Figure 4-36: (a) Horizontal and (b) vertical stresses contours after 1 mm expansion of the pipe As mentioned previously, the term (l+K)/2 in Equation [4-4] represents the average O ' N value over the perimeter of the pipe. The back-calculated K value from computed normal stresses on the pipe perimeter (as shown in Figure 4-37) is between 1.9 to 2.2. Similarly, the K value back-calculated from axial pullout loads using Equation [4-4] is about 2.15 (considering interface friction angle 8 = 33° which is between peak and residual, see Section 4.4.1.2). Clearly the K value from numerical modeling has captured the normal stresses more effectively than using Ko value as suggested by A S C E (1984), A L A (2001), and PRCI (2004). 142 1.55-1.8 Stress distribution interpreted from ASCE (19841 1.4-1.55 Prior to expansion of the pipe Computed normal stresses from numerical model results After expansion of the pipe 1.2-1.35 Stress distribution interpreted from ASCE (19841 During pullout Measured normal stresses from pressure transducers Prior to pullout * Assumed values based on calculations ure 4-37: Stress distribution around the pipe during pullout; comparison of numerical model results and pressure measurements 143 4.5.3.3 The effect of box size The numerical modeling approach was also used to assess the effect of side wall and ground constraints. In this regards, another mesh configuration simulating a larger soil box with 6 m width and 2.3 m height was modeled. In this model, the distances of boundaries from pipe surface are more than two times of that in experiments. After expansion of pipe for 1 mm (similar to that in the previous model) the horizontal and vertical stress contours are shown in Figure 4-38. The computed normal stresses at position of pressure transducers are plotted in Figure 4-39 along with those from the previous model for comparison. E H K . S¥¥-StHS {tonsure H -4.KJE+04 -3.23E+EA 11 -2..CE+W B -1.i3&#f • fumm Caftor lifef¥*> 4JOE+GB * Units: Pa Figure 4-38: Horizontal and Vertical stresses contours after 0.92mm expansion of the pipe (box size: 6m W x 2.3m H) 144 Computed normal stress, 2.5 m x 1.6 m box Computed normal stress, 6 m x 2.3 m box Figure 4-39: Computed normal stress on the pipe for different model configurations The stress distribution in the soil mass as observed in Figure 4-38 did not show a noticeable change from previous model (see Figure 4-36 for comparison). Examining Figure 4-39 also indicated that the normal stresses on the pipe are almost similar for both configurations. This suggests that any boundary effects associated with the soil box boundaries are either not present or insignificant. 4.6 Variation of K in relation to the level of dilation As noted from the work presented in previous sections, the current approach of estimating axial soil loads using Equation [4-4] is significantly affected by the value of K representing the normal soil stress on the pipe surface during pipe pullout. Considering that the normal stress appears to increase due to dilation of soils, and the numerical model proposed in Section 4.5.1 seems to generally capture this effect, it was considered worthwhile to investigate the influence of the degree of dilation on the K value for compacted soils. 145 While the value of interface friction angle (8) also contributes to the computed axial load, it is of importance to note that the impact of the variations in 8 on the F A ' is relatively small compared to the effects arising due to changes in K (as observed in the present study). With this rationale, the numerical model representing pipe diameter D = 457 mm and depth of H = 115 cm (see Figure 4-32) was used to back-calculate K value as a function of the level of dilation (represented by expansion at interface). The computed results are presented in Figure 4-40. The results of the numerical model clearly show the sensitivity of the K value to dilation and consequently the effect on the magnitude of axial soil loads on the pipe. In other words, evaluation of K value on pipe becomes an important issue while dilative soil surrounds the pipe. 0 0.5 0.7 1 1.5 2 Effect of dilation; Expansion at pipe surface (mm) Figure 4-40: Calculation of equivalent K value based on the numerical model results The use of the numerical modeling results in Figure 4-40 to real-life involves estimation of the dilative displacement (amount of expansion at interface) corresponding to the axial pullout: (i) under laboratory pipe testing or (ii) in field pipe installation problems. The 146 amount of dilation at interface is primarily a function of soil parameters such as density, stress level, and grain size distribution. In the case of laboratory testing, it can be argued that (as also explained in Section 4.5.1.2), the amount of dilation at the interface can be estimated by the normal displacement observed during direct shear testing of backfill soil specimens tested at the same conditions as the physical modeling test. The estimation of dilative displacement at shear zone in the filed condition is going to be dependent on in-situ soil parameters which are more difficult to quantify. As such, the use of these results to the field conditions will involve more investigation, particularly addressing different soil types and states. This work was not included as part of the current study. Since selecting a proper K value for pipeline buried in dense soils is critical for axial soil loads, it is of interest to predict K value for cases with different burial depths, pipe diameters, and soil properties (e.g. friction angle, and elastic properties). The configuration of the validated numerical model was appropriately adjusted to model the effect of some of these parameters on the K value. 4.6.1 Ef fect o f b u r i a l depth on K va lue The numerical model was executed with pipe buried in material representing properties of Fraser River sand at D r = 75%) at different burial depths (H) between 0.93 m to 2.8 m resulting overburden stresses at the pipe springline level between 15 kPa and 45 kPa. Again, the constitutive model and the soil parameters adopted for modeling are described in Section 5.7.2.4. The pipe diameter for the model for all the cases was kept constant at 457 mm. The back-calculated K values using the results from analysis are shown in Figure 4-41. To reach convergence in the numerical model, the results are shown for expansion greater than 0.5 mm. Observations of vertical deformation in direct shear box as discussed in Section 4.5.2 are also in agreement that a range of 0.5 mm to 2.0 mm is a reasonable range for dilative soils. The variation of K value with soil dilation at the interface is higher for shallow buried pipes; however it is worth noting that when the expansion of the dilation zone is more 147 than about 0.5 mm, the computed values of K for pipes buried at different depths fall within a narrow band. An envelope was drawn to provide an upper bound for K for all cases with overburden stress greater than 15 kPa at the pipe centreline. Ej = 94 (a 3/P a) u ° Mpa H = 0.93 m, ov'= 15kPa H= 1.86 m, av' = 30kPa H = 2.8 m, oV = 45 kPa Upper-bound 0 0.5 1 1.5 Effect of dilation; Expansion at pipe surface (mm) Figure 4-41: Variation of K value with burial depth, D = 0.46 m 4.6.2 Effect of pipe diameter on K value The model was re-analyzed to simulate pipes with outside diameters of 0.23 m and 0.92 m at burial depths corresponding to overburden stresses at pipe axis level of 15kPa, 30kPa, and 45kPa. The computed K values versus expansion of shear zone are shown in Figure 4-42 and Figure 4-43. The envelopes that represent upper bound K value for overburden stresses ranging from 15 kPa to 45 kPa are also shown. The results for pipe with outside diameters of 0.23 m and 0.92 m are comparable to those presented in Section 4.6.1; again the value of K for dilations greater than 0.5 mm seems to be insensitive to overburden stress at pipe level. 148 0.5 1 1.5 Effect of dilation; Expansion at pipe surface (mm) Figure 4-42: Variation of K value with burial depth, D = 0.23 m The upper bound values of K obtained from the numerical model results on different pipe diameters are plotted together in Figure 4-44. Since the upper bound is plotted, the curves can be considered independent of the overburden stress. As may be noted in the next section, this figure will form the backbone for prediction of K values in soils with different material parameters. It is however important to note that this figure corresponds to modeling conducted for pipes buried in Fraser River sand at a relative density of D r = 75%. The curves indicate that the effect of dilation is higher for smaller pipes. 149 0.5 q> = 45° Ej = 94 (oyp a) 0 6 Mpa H = 0.93 m,oV = 15kPa — H = 1.86 m, av' = 30kPa — H = 2.8 m, oV = 45 kPa — • Upper-bound 0 0.5 1 1.5 Effect of dilation; Expansion at pipe surface (mm) Figure 4-43: Variation of K value with burial depth, D = 0.92 m 4 3.5 3 2.5 -< ; 2 1.5 1 0.5 0 D = 23 cm D = 46 cm D = 92 cm cp = 4J O Ej = 94 (a 3/P a) u 6 Mpa 0 0.5 1 1.5 Effect of dilation; Expansion at pipe surface (mm) Figure 4-44: K value as a function of dilation effect and pipe size 150 4.6.3 Effect of soil parameters on K value The sensitivity of the computed value K to the soil parameters were also investigated. It was found that the value of K is sensitive to both elastic modulus (or shear modulus) and friction angle, whereas other parameters such as interface friction angle and soil dilation angle did not cause any significant effect. Therefore only the effects of these parameters (friction angle and elastic modulus) were examined. Specific cases with different elastic modulus (E = Eo, 2 x Eo and 4 x Eo where Eo is elastic modulus of Fraser River sand at 75% relative density) and friction angles (cp = 40°, 45° and 50° at G3'=15kPa) were modeled in this regard. Ei / E i 0 > E i 0 = 94 (rj3/Pa)0-6 MPa Figure 4-45: Variation of K value with elastic modulus The values of K for different elastic modulus at two different pipe diameters are shown in Figure 5-45 after 0.5 mm, 1mm, and 1.5 mm of expansion of pipe. For dilative soils (i.e. more than 0.5 mm increase in the thickness of sheared area), the K value varies significantly with the change in shear modulus. For example, doubling the shear modulus of soil results in increase in K value of about 25%. Plotted values of K vs. 151 elastic modulus on a logarithmic scale exhibited a linear variation regardless of the pipe diameter and amount of expansion. These outcomes can be represented using a formula as shown in Equation [4-6]. K(D,AV) = K'(D,AV)x(Ei/EJ 0.3 [4-6] where: E j : initial elastic modulus of the material EJO : initial elastic modulus of Fraser River sand at D r = 75% (= 94 ( 0 3 ' / P a ) 0 6 ) K*(D,AV) : Computed K value for certain geometry and soil parameters as found from Figure 4-44 The same approach was undertaken to observe and quantify the effect of friction angle on K value. As presented in Figure 4-46, the results indicated a linear relationship between friction angle K value, independent from pipe diameter or effect of dilation. As may be noted, the increase in friction angle of 5° resulted in about 10% increase in K value. Based on this, the K values can be expressed as a function of friction angle as given in Equation [4-7] for a given geometry. where: 9: Peak friction angle of soil material K*(D,AV) : Computed K value for certain geometry and,soil parameters as found from Figure 4-44 Based on this study it is proposed that Equation [4-6] and Equation [4-7] be combined to predict the K value for different pipeline geometry and soil properties. It should be noted that this procedure provides an upper bound for K value which results in a conservative design of pipelines for most real-life situations. K(D,AV) = K\D,AV)x(<p/45) 0.3 [4-7] 152 0 -I i i i ' 35 40 45 50 55 Friction angle (deg) Figure 4-46: Variation of K value with friction angle 4.7 Summary of the chapter A series of axial pullout tests were conducted in sand that included experiments with bare pipe and pipe wrapped with geosynthetics. In addition to the measurement of axial loads and axial pipe displacements during pullout, the soil pressure on the pipe surface was monitored using pressure transducers mounted at several circumferential locations on the pipe. These soil pressure measurements were critical in developing a theory to explain differences in the load deformation responses observed during the tests. Monitoring of soil surface deformations and sand particle movements in the vicinity of the pipe/soil interface provided additional information with regard to understanding the basic failure mechanisms during axial pullout testing. Some of the key findings/assessments are summarized below: 153 1. The axial soil loads predicted using the ASCE (1984) and PRCI (2004) equation were in good agreement with the results from axial pipe pullout tests conducted for bare pipe buried in loose sand backfill. 2. The results from pipe pullout tests in dense sand exhibited axial resistance approximately three times greater than that predicted using A S C E (1984) and PRCI (2004) equation (see Figure 4-22). The soil pressure measurements undertaken during pullout tests in dense sand indicated that the overall normal soil stresses on the pipe during pullout increased substantially in comparison to the initial values. This increase in normal stress is believed to be associated with constrained dilation of the dense sand during shear deformations. Therefore, the recommended use of "at-rest" earth pressure coefficient (Ko) in the ASCE (1984) equation is inappropriate for estimating initial axial resistance where dilative soil behaviour is expected. 3. The maximum soil resistance decreases with increasing displacement of the pipe through the soil and repeated displacement cycles. Subsequent reverse loading tests achieved maximum axial soil resistance values comparable to those predicted by the A S C E (1984) and PRCI (2004) equation (see Figure 4-24). This is believed to be a result of substantial disturbance at the pipe-soil interface before reverse loading (pipe was pulled for more than 800 mm prior to reverse loading). 4. The observed soil particle movements using colored sand zones indicated that predominant shearing action in axial pullout occurs within about a 2-mm zone from the surface of the pipe. Following 500 mm of pipe displacement, the soil particles that are located about 1 cm away from the pipe surface appear to have moved less than 2 mm in the direction of the pipe displacement; those located about 4 cm away from the pipe surface seem to have moved by negligible amounts (less than 0.5 mm) 154 5. The use of two-layers of geotextile wrapping provides an effective means of reducing axial loads on buried pipelines subject to ground movements (see Figure 4.33). Wrapping the pipe with geosynthetic layers not only decreases the load by reducing the interface friction angle at the slippage surface, but also may minimize development of high normal stress by preventing shear deformation in the dense sand (thereby reducing dilatent behaviour) and by providing some flexibility to accommodate potential dilation strains from any shear deformation within dense sand. 6. The "cigar-wrapped" two-layer geotextile system is more effective in reducing the soil loads than the "spiral-wrapped" system. 7. In the case where the pipe was wrapped with a layer of woven-geotextile over geonet, the test results indicate that jamming of the geonet at the seams could lead to increases in axial soil load. This suggests that the use of an alternate "geotextile-over-geonet" wrapping configuration is needed to effectively reduce axial soil load. 30 25 OJ) s - 15 u a . •o « o J 5 1 Dense sarid, Bare pipe i ^ Dense s and, Bare pipe, Reverse loading Medium sand, Bare pipe / X . Ir 1 x Dense sand Wrapped pipe, Reverse loading X Dense sand, Wrapped pipe 100 200 300 Pipe Displacement (mm) 400 500 gure 4-47: Summary of axial pullout tests; Moving average and averaging of different tests are applied. Starting points of unloading / reloading parts are shifted. 155 Since the commonly used methods for prediction of axial soil loads on pipe buried in dense sand seemed inappropriate, numerical modeling was undertaken using finite difference method based software, F L A C 4.0 , to explain high soil loads on pipe during axial pullout. The constitutive model and soil material properties were initially selected based on the element testing on sand and calibrated numerical model for lateral pullout tests as can be found in Section 5.7.2.4. The key findings from numerical modelling and are summarized here: 1. A plane strain model that models a vertical plane normal to the box axis can acceptably show variations in the soil stress and strain due to dilation of sand at interface as a result of shearing. In this model, dilation of sand at interface was simulated by numerically "expanding" of the pipe surface. The model predictions were in reasonable agreement with the measured pressures in transducers mounted on the pipe surface both before and during pipe pullout. Also the model was able to predict a value of K that is in accord with those from axial pullout load measurements. 2. Previous research on shear band and shear zone in granular material indicated that the width of active shear zone can be approximated by lO.dso in which dso is average grain size. Observation of particle movement around the pipe showed that the thickness of the actively sheared sand zone around the pipe is in the order of 1.2 to 2.8 mm. This confirms that assuming lO.dso as the thickness of shear zone, provides reasonable values (for Fraser River sand: dso = 0.23, results in shear zone thickness of 2.3 mm) 3. Examining the results of the validated model with the test data indicated that the K value is highly dependent on the amount of dilation, especially at the onset of dilation where 0.5 mm to 1.0 mm of increase in the thickness of sheared sand area, results an increase of more than 200% to 300% in K value (compared to Ko). Dilative displacement at shear zone can be approximated by measurement of 156 normal displacement during direct shear testing on the backfill soil tested at the same (density and normal stress) conditions as backfill material at the pipe springline level. 4. Models with different pipe diameters and different buried depths were developed using the same soil properties and modelling approach. The results of the models indicated that the variation of K value with burial depth is in a range that an envelope can cover all burial depths. However the results showed that the K value is different for pipes with different diameters and the effect of dilation is more significant on the smaller pipes. A series of charts were developed to estimate K value for pipes with different diameters and different dilation effect (see Figure 4-44). 5. Sensitivity analysis on the effect of different soil and interface parameters on the value of K indicated that this value is sensitive to the variation of soil elastic modulus and friction angle. Increase in both friction angle and elastic modulus resulted in increase in the value of K. Equation [4-6] and Equation [4-7] are presented to modify this value for different soil material properties. The effects of other parameters (e.g. interface friction angle, soil dilation) on the value of K were found to be negligible. 157 CHAPTER 5 BURIED PIPES SUBJECT TO TRANSVERSE GROUND MOVEMENTS 5.1 Introduction This chapter describes the results from horizontal lateral pulling tests on buried steel pipe as well as results of numerical modelling of the test configuration. A s described in Chapter 3, the lateral pulling test program consists of fourteen tests. Some of the parameters related to the tests are also summarized in Table 5-1 herein prior to the presentation of the results. The tests results are presented in three sub-sections: (i) tests on pipe in dense sand, (ii) tests on buried pipe in trench excavated in dense cohesionless "native" soil, and (iii) tests simulating buried pipe in trench excavated in hard "native" soil (see Figure 3-10 in Chapter 3 for the definition of native soil and backfill zones). Following the presentation of the test data, the results are discussed and compared with those from previous experiments and the recommended methods for prediction of horizontal lateral loads on buried pipelines during ground movement ( A S C E 1984). Recommendations are also made with regard to the selection of material parameters. The effect o f soil load reduction methods as described in Chapter 1 is evaluated in combination with available analytical methods. Recommendations are made for the applicability of the method in the field. 158 5.2 Summary of test parameters Tests on buried pipelines have been performed on a variety of different pipe sizes and model scales. To establish a bench mark for presenting the results and comparing with other studies, the concept of dimensionless load and normalized displacement is used in this report as previously suggested by Hansen (1964) and used by Audibert and Nyman (1975), Rowe and Davis (1982), Trautmann and O'Rourke (1983) and Paulin et al. (1998). The dimensionless load F L ' = F L / (y.H.D.L) versus normalized displacement Y ' = Y / D where F L = pulling load, y = soil density, L= length of pipe, H = height of soil over pipe centre line, D = pipe diameter, and Y = transverse pipe displacement are presented instead of load versus displacement. The details related to the testing program and test parameters were shown in Tables 3-2 and 3-4 of Chapter 3, and important test characteristics are summarized in Table 5-1. Table 5-1: Summary of parameters in lateral pulling tests Test Configuration Horizontal Lateral Pulling Soil Type Fraser River Sand Average Density 1600kg/mJ Average moisture > 1% for Dry Sand Tests & ~ 10% for Moist Sand Tests Internal Peak Friction Angle 46°~42° for 1600 kg/m J and o 3 of 10 to 50 kPa Box Size 4 m L x 2 . 5 m W x 2 . 5 m H Pipe Size D. 12.75" & VA" Thickness - D. 18" & lAn Thickness Pipe Length: 2.4 m Pipe Grade & Surface Steel Grade 524A, Sand Blasted Surface Soil - Pipe Friction Angle Peak of 36° and Constant Volume of 30.5° Geosynthetic Material TC Mirafi Filterweave 700 Interface Friction Angle of Geotextile Layers Peak Friction Angle of 21° Residual Friction Angle of 19.7° Maximum Pipe Displacement 500 mm to 600 mm Loading Rate 5 mm/sec - 10 mm/sec 159 Dimensionless load F L ' can be generally interpreted as a net average of lateral soil stress on the pipe for a given overburden soil stress during lateral displacement of the pipe. For example an F L ' value of 10 can be considered to imply the soil stress in a plane perpendicular to the direction of pipe displacement is 10 times of overburden stress at the centreline level of the pipe. Normalized displacement of Y ' also represents pipe displacement over the pipe diameter. A series of lateral displacement-controlled experiments were performed to investigate the load-displacement behaviour of buried pipes. The test apparatus and experimental issues were discussed in detail in Chapter 3 and Chapter 4. 5.3 Lateral loading response of pipeline buried in cohesionless soil Detailed of load-displacement curves and other physical measurements are presented in Appendix B. 5.3.1 Lateral load-displacement response Tests No. LN-1 and LN-2 were identical, and they were performed using a 457-mm diameter pipe, with overburden ratio (H/D) of 1.92 and dry sand having an average soil density of 16 kN/m J. Tests No. LN-3, LN-4 and LN-7 were performed using a 324-mm diameter pipe and the same overburden ratio and soil density. The remaining two tests, Tests No. LN-5 and LN-6, were performed again using a 324-mm diameter pipe, but with overburden ratios of 1 and 2.75 respectively. The results of Tests No. LN-1 and LN-2 are shown in Figure 5-1. Maximum load per unit length of the pipe in both cases were about 50 kN/m, resulting in a dimensionless force F L ' of about 8. The maximum lateral load is mobilized at pipe displacements in the order of 45 mm to 70 mm, or 0.1 -0.15 of normalized displacement (Y'). Since the two 160 tests are identical, the results indirectly confirm very good repeatability of specimen preparation, test control, and instrumentation. 10.0 9.0 8.0 7.0 6.0 5.0 4.0 3.0 2.0 1.0 0.0 — LN-1 — LN-2 0.00 0.05 0.10 0.15 0.20 0.25 0.30 Normalized Displacement Figure 5-1: Load-displacement response - Tests No. LN-1 and LN-2 (H/D=1.92; D=457mm) Figure 5-2 shows the result of the Tests No. LN-3, LN-4 and LN-7. The peak load was reached after 25 to 30 mm of the pipe displacement which is about 0.1 of normalized displacement. The average peak mobilized dimensionless lateral load force F L ' is about 7 for these tests with an overburden ratio of 1.92 and pipe diameter of 324 mm. The dimensionless load values in tests with either 457 mm pipe diameter of 324 mm pipe diameter are almost the same, due to the similar overburden ratios. During test LN-7, a pressure film with 61 cm width and a very smooth surface was wrapped around the pipe to capture the stress variation in front of the pipe during lateral pulling of the pipe. The decrease in peak F L ' by about 10% for LN-7 is not observed in any other tests and might be related to the non-uniformity of the pipe surface roughness due to introduction of pressure film. 161 0.00 0.05 0.10 0.15 0.20 0.25 0.30 Normalized Displacement Figure 5-2: Load-displacement response, Tests No. LN-3, LN-4 and LN-7 (H/D=1.92; D=324mm) In Figure 5-3, the result of Tests No. LN-5 and LN-6 are presented as well as an average of the Tests No. LN-1, LN-2 and also LN-4, LN-5 and LN-7. The peak loads in Tests No. LN-5 and LN-6 in terms of F L ' are 6 and 8.5, respectively (i.e., in terms of peak load 10 kN and 38 kN, respectively). The peak load was mobilized at about 15 mm or 0.05 of normalized displacement for LN-5, and 35 mm or 0.1 of Y ' for LN-6. As expected, the dimensionless load increases with increasing overburden ratio (increasing depth of pipe burial) and the peak load is mobilized at larger displacements. These effects are further discussed in Section 5.6. 162 10.0 9.0 8.0 7.0 6.0 5.0 4.0 3.0 2 0 F 1.0 ; L N - 6 . i X X l N - i j L N - i " >LN^37LN^471 .N-7 jfj " ^ L N - 5 m - -] Al [ l 0.0 LN-5 (H/D=l, D=324mm) LN-6 (H/D=2.75, D=324mm) ~ - LN-1-2 (H/D=l .92, D=456mm) - -LN-3-4-7 (H/D=1.92, D=324mm) 0.00 0.05 0.10 0.15 0.20 Normalized Displacement 0.25 0.30 Figure 5-3: Load-displacement response, Tests No. LN-1 to LN-7 5.3.2 Soil contact pressures on the pipe during lateral loading In test LN-7, five pressure transducers were mounted on the pipe (see Section 3.1.5.5) to observe the variation of stress in front of the pipe. The transducer configuration was oriented in the soil box so that the transducer No. PT3 is facing the pulling direction (see inset of Figure 5-4 for transducer identification numbers). The transducers recorded the normal stress from the soil at the contact point. The overall soil stresses in the lateral direction could be calculated by appropriately resolving normal stresses on the pipe, as discussed in Section 5.6.1.2. As in the case of axial tests (see Section 4.2.2), examination of the moving averages of the soil pressure measurements were considered reasonable since it allowed removing the high-frequency "noise" during lateral pulling. Significant judgement was used to vary the window used for computing the averages in order to obtain a smoothed result that fairly tracked the raw data. These average soil pressure readings were judged sufficient for identifying relative trends in the variation of soil pressures during lateral pulling tests. 163 In order to be compatible with the lateral load-displacement response, which was presented using a dimensionless load, the measured normal stresses (a' n) were also normalized with respect to the overburden stress (a' v) at the centre line of the pipe; hence the value of dimensionless normal stress can be defined as O ' N = o"'n /o ' v . Figure 5-4 shows the moving average of dimensionless normal stress (G 'N ) at the pressure transducers versus normalized displacement of the pipe. The details of raw measurements are presented in Appendix B for this test; however for other tests with pressure transducers measurements, only results after applying moving average are presented. PT5 PT4~\ PT4 PT3q PT3 PT2 PT2/ — P T l 0.00 0.05 0.10 0.15 0.20 Normalized Displacement 0.25 0.30 Figure 5-4: Dimensionless stresses around the pipe during Test No. LN-7 (moving average is applied) It can be observed that the O ' N from transducer at top of the pipe (PT5) and that located at 45° to the horizontal (PT4) show a small increase at the beginning of the pipe displacement; the value then drops after about 0.1 of Y \ The increase is negligible compared to the increase in normal stress in front of the pipe (at PT3) and in normal stress on 45° below the horizontal level (at PT2). The stress increase in PT5 was not 164 observed in the other tests (Tests No . LT-1 to LT-4 , L T G - 1 and L T G - 2 which w i l l be presented in Section 5.5.2). The initial reading of the pressure transducer mounted at the invert of the pipe ( P T l ) is slightly higher than vertical stress at that point (leading to a dimensionless normal stress of 1). The measured pressures at transducer location P T l are difficult to interpret because of several factors such as potential arching of soils around the pipe, potentially large stress non-uniformities due to varying contact in supporting the weight of pipe, and the effects arising due to upward displacement of the pipe during lateral pulling. 5.3.3 Surface deformations and pipe movement The deformed surface of the sand was observed after completion of a number of tests. Figure 5-5 presents the deformed surface after Test No . L N - 4 to illustrate these observations. The "active" and "passive" zones/fields of soil deformation are clearly distinguishable. The active zone is at the trailing side with respect to the pipe movement wedge, and a passive field is formed in front of the pipe. The observed surface deformation profiles obtained from Tests No . L N - 5 and L N - 6 are shown in Figure 5-6. This pattern of surface movement was observed to be consistent for all tests. 165 Figure 5-5: Surface deformation after Test No. LN-4 Upon completion of testing and after removal of the sand in some of the tests, the final position of the pipe was surveyed for comparison with respect to the initial position, a movement pattern can be estimated. The "before-test" and "after-test" positions of the pipe for Tests No. LN-4 and LN-6 are shown in Figure 5-6. Surface is slightly less deformed in Test No. LN-6 compared as can be observe in Figure 5.6 to Tests No. LN-4 and LN-5 most likely due to the deeper embedment in the first one. After emptying the box, in both the tests, it was observed that the pipe had moved upward about 200 mm during lateral pulling of 500 to 600 mm. Upward movement of the pipe seems to be insensitive to the overburden ratios (H/D) for the range of tested H/D (1 to 2.75). The path of the pipe displacement of about 20° to the horizon was also captured in the numerical modeling. The observed uplift of the pipe during movement is in accord with the characteristic failure mechanisms observed by others for shallow buried pipe. 166 Figure 5-6: Surface deformation and pipe position after Tests No. LN-4 and LN-6 5.4 Lateral load response of pipeline buried in trench-configurations constructed in "cohesionless native soil" Several tests were conducted on pipes buried in "trench-configurations" to study the effectiveness of some selected methods used for reducing soil loads on pipes due to ground movement. The details related to the test configurations, the rationale for testing and associated terminology used for description, are given in Chapters 1 and 2, and they are not repeated herein for brevity. This section presents the results from two tests (Tests No. LNG-1 and LNG-2) on pipeline buried in trench-configurations lined with two geotextiles, constructed in cohesionless "native" soil. Compacted Fraser River sand was used for the "trench backfill" as well as "native soil" in these tests. The two tests were performed with trench slope angles (9) of 45° and 35°, respectively, as is illustrated in Figure 5-7. 167 e Dual Layer of Geosynthet ic Figure 5-7: Typical dual-geotextile-lined trench-configuration for reduction of soil loads due to transverse ground movements 5.4.1 Load-displacement response The load-deformation response was examined using the dimensionless load ( F L ' ) and normalized displacement ( Y ' ) , as previously defined in Section 5.2. Figure 5-8 shows the load-displacement curves obtained from the Tests No . L N G - 1 and L N G - 2 . The lateral load gradually increases until a F L ' value of about 8.0 is reach at about at Y ' = 0.1. Between Y'=0.1 and 0.4, F L ' remains essentially unchanged. This response is very much similar to the response observed in Tests No. LN-1 and L N - 2 which were conducted on pipe without geotextile trench. However, under displacements beyond Y'=0.4 (about 200 mm of the pipe displacement), a gradual increase in the lateral load could be noted (Note: This increase was not noted in Test No. LN-1 and LN-2) . Observations made after removal of soil, upon completion of tests, indicated that the geotextile layers on trench slope in front o f the pipe had got pushed into the native soil, likely preventing the formation of a passive wedge and occurrence of slippage between the two geotextile layers. It appears that, with increased lateral movements, the additional constraints from the geotextile layers are mobilized 168 thus increasing the F L ' to levels above those observed for tests in pipes buried in sand without geotextile layers. 9 8 •O 7 M O J 6 09 09 s s 5 4 3 E - i f e V k" ( — LNG-1,9=45° LNG-2, 8=35° 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 Normalized Displacement 0.80 0.90 1.00 Figure 5-8: Load displacement response; Tests No. LNG-1 and LNG -2 5.4.2 Movement of geotextile layers during lateral pulling Displacement of individual geotextile layers during the Tests No. LNG-1 and LNG -2 measured using string potentiometers, provided an opportunity to examine the potential slippage between the two layers of geotextile, or between the inner layer of geotextile and native backfill. Figure 5-9 presents the measured displacements of the geotextiles in tests No. LNG-1 and LNG -2 . In both of the tests, the two geotextile layers seemed to have moved in harmony, indicating that no significant slippage has occurred at the geotextile/geotextile interface. Clearly, the objective of forcing the slippage surface to occur between two geotextile layers does not seem to have materialized; it appears that the presence of the geotextile is 169 not recognized by the system likely due to the identical deformation stiffness of "trench backfill" and "native soil". E «V w J& 5. 5 •a CV = m E u o Z CV -1 O 1.00 0.90 0.80 0.70 0.60 -f 0.50 0.40 4 0.30 0.20 0.10 4 0.00 Outer Layer, LNG-1 Inner Layer, LNG-1 Outer Layer, LNG-2 Inner Layer, LNG-2 0.00 0.20 0.40 0.60 0.80 Pipe Normalized Displacement 1.00 .20 Figure 5-9: Movement of geotextile layers during Tests No. LNG-1 and LNG - 2 In Test No. LNG - 2 , the differential displacement between geotextile layers is slightly more than that in Test No. LNG-1. The angle of trench slope in this test (0 = 35°) is similar to the angle of anticipated slope failure determined using classical soil mechanics approaches for a case without geotextiles. It is possible that this may be the reason for somewhat increased slippage and no prominent increase of F L ' (see Figure 5-9) after Y'=0.5in Test No. LNG-1 . 5.4.3 Surface deformations and pipe movement Figure 5-10 presents the deformed surface movements after Test No. L N G - 2 . Figure 5-11 also shows a photograph taken from the sand surface after completion of the same test. The "active" and "passive" fields of soil deformation are clearly apparent and distinguishable. The active zone is at the trailing side with respect to the pipe movement 170 wedge, and a passive zone is formed in front of the pipe. Deformed sand surface after completion of Test No. LN-2 which has the same test parameters as LNG-2 (except for geotextile liners) is also shown in Figure 5-10 for comparison purpose. Examining this figure shows that both tests follow a similar pattern for surface deformation. Movement of a higher volume of sand during Test No. LNG-2 compared to Test No. LN-2 is in harmony with slightly higher lateral soil loads in test where geotextile layers are present. Uplift of pipe during pulling the pipe was measured as about 200 mm, similar to what was observed in tests in native soil. Distance from Pipe Centre (mm) Figure 5-10: Surface deformation and pipe position after Tests No. LNG-2 and LN-2 171 Figure 5-11: Surface deformation after Test No. LN-4 5.5 Lateral load response of pipeline buried in trench-configurations constructed in "hard native soil" As discussed in Section 3.2.2, lateral pulling tests were conducted to simulate the response under lateral loading of pipes buried in trench-configurations with "hard native soil" conditions (e.g., pipe buried in glacial till-like material or bedrock). In this investigation, five full scale tests were performed on buried pipelines in trenches made from significantly stiff material with a trench slope of 35°. Tests were conducted with and without geotextile-lining on the trench surface (see Section 3.2.2 and Figure 5-10 for details). In two of the tests, Tests No. LT-1 and LTG-1, compacted dry Fraser River sand was used as backfill to capture the effect of cohesionless material in trench backfill. In the other three tests, Tests No. LT-2, LT-3 and LTG-2, moist sand (with a moisture content of about 10%) was used to simulate the use of a cohesive material as backfill. 172 The load-displacement behaviour and the relative displacement of geotextile layers (in the geotextile-lined configuration), and soil pressure measurements on the pipe (in some of the tests) during lateral pulling tests are presented in this section. 5.5.1 Load displacement response Figure 5-12 presents the variation of the dimensionless load ( F L ' ) and normalized displacement Y ' obtained from Tests No. LT-1 and LTG-1. Except for the geotextile-lined trench in LTG-1, the other conditions are identical for both the tests. After a relatively rapid initial build-up (from Y ' = 0 to -0.05), a significant drop in the rate of increase of dimensionless load F L ' is observed for both the Tests No. LT-1 and LTG-1. This rate of increase in F L ' beyond Y'~0.05 is still steeper than, and different from, the response observed for the trench configurations constructed in "cohesionless native soil" (see Figure 5-1 to Figure 5-3) in which the value of F L ' reached a plateau or a relatively mild increase with increasing Y ' . o E 5 < 4 • \ . -< . . — L T - 1 , 3 5 ° . — L T G -1 , 35° > I 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 Normalized Displacement 0.80 0.90 1.00 Figure 5-12: Load displacement response; Tests No. LT-1 and LTG-1 173 Comparison between Tests No. LT-1 and LTG-1 suggests that the introduction of geotextile lining would contribute to reducing the dimensionless load (FL ' ) by about 15% to 20%. However, considering that the interface friction between two geotextiles (20°) is significantly lower than interface friction angle between trench-backfill and trench surface (-34° to 45°), a reduction of the dimensionless load ( F L ' ) between the above tests would be estimated to be in the order of 50% from an analytical point of view. It appears that the increasing proximity to the hard boundary during pulling of the pipe have contributed to increasing the lateral load in the tests, as discussed later in Section 5.6.2. The load displacement curve of Tests No. LT-2, LT-3, and LTG-2, where compacted moist sand was used as backfill material, are presented in Figure 5-13. The sand-glued trench wall was covered with two layers of geotextile in Test No. LTG-2 to examine the efficiency of soil load reduction method. Tests No. LT-2 and LT-3 are identical and were conducted to assess the repeatability. T3 es o - J E 5 LT-3 i LT-2 t_~--LTG-2 f ! LT-2, 35° LT-3, 35° LTG-2, 35° I 1 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 Normalized Displacement 0.80 0.90 .00 Figure 5-13: Load displacement response; Tests No. LT-2, LT-3 and LTG-2 174 The shapes of the load-displacement response obtained for these tests with moist sand backfill are generally similar to that observed for the tests with dry sand. However, the dimensionless load F L ' starts to build up at Y ' = 0.4 (or pipe displacement of 180 mm). This is higher than those observed during tests in dry sand. The initial part of the curve is steeper here and indicates that the load is mobilized faster and at smaller pipe displacements. Comparison of results from the identical Tests No. LT-2 and LT-3, again, confirms very good repeatability of specimen preparation and testing techniques. The dimensionless load ( F L ' ) obtained with the trench lined with geotextiles (Tests No. LT-2 and LT-3) is about 15% to 20% lower than that obtained for the unlined trench (Test No. LTG-2). This trend is similar to the observations made for the tests conducted with dry sand backfill. In an overall sense, the magnitudes of F L ' obtained for both the cases (i.e., with and without geotextile-lining) are slightly lower than the counterpart values of F L ' obtained from tests conducted with dry sand backfill. As presented in detail in Section 5.5.4, clear formation of a coherent passive block (wedge) in front of the pipe was observed during the tests with moist sand; the block appeared to have slipped along the trench backfill interface essentially as a unit. This observation suggest that the local shearing of sand in the immediate vicinity of the pipe is likely less with moist sand backfill in comparison to that observed under test configurations with dry sand. Such block movements, with less energy dissipation required for local shearing, might explain the lower lateral load on the pipe in tests with moist sand. 5.5.2 Soil contact pressures on the pipe during lateral loading The contact soil pressures at transducers PT2, PT3, and PT4 were monitored during Tests No. LT-1, LT-2, LT-3, LTG-1 and LTG-2. The radial location of the transducers are given in the inset of Figure 5-14. As mentioned earlier, the positioning was so that the transducer No. PT3 would face the pulling direction. 175 Figure 5-14 shows the typical variation of the moving average of the dimensionless normal stress (O 'N) at the transducers PT2, PT3, and PT4. The definition of O ' N and the reasoning for calculating the moving average are described in Section 5.3.2. The interpretation of the pressure measurements and associated comparisons with the variations on the horizontal load are discussed in Section 5.6.2.2. 12 10 H a. c o in 09 «U "e .o '« c n E 64 I P T 4 \ \ \ p \ P T 1 w PT2 PT4 1 PT4 PT3 PT2 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 Normalized Displacement 0.40 0.45 0.50 Figure 5-14: Dimensionless stress around the pipe during Test No. LT-1 (Moving averaging is applied) 5.5.3 Movement of geotextile layers during lateral pulling Figure 5-15 shows the displacement of two geotextile layers lining the trench in Tests No. LTG-1 and LTG-2. In this figure, inner geotextile layer describes the layer in contact with backfill material and the outer geotextile layer describes the layer that is sandwiched between inner layer and native soil material. The results clearly indicate that the geotextile layer that was in contact with the native soil zone did not experience any movement in both of the tests. The slippage appears to have occurred in one or 176 combination of the following interfaces: (i) between the two geotextile layers; (ii) outer geotextile/sand backfill zone; or (iii) within the sand mass. 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 Pipe Normalized Displacement Figure 5-15: Geotextile layers movement in tests LTG-1 and LTG-2 5.5.4 Surface deformation and pipe movement The surface deformation patterns observed in Tests No. LT-1 and LTG-1 (with dry sand backfill) are shown in Figure 5-16. Similar patterns observed for Tests No. LT-2 and LTG-2 conducted with moist sand backfill material are presented in Figure 5-17. The "before-test" and "after-test" positions of the pipe for these tests are also shown in the figures. This information is also complemented by the photograph in Figure 5-18 that presents the deformed backfill surface taken after Test No. LTG-2. The observed deformation patterns in the tests conducted with dry sand backfill suggests the development of active and passive fields of deformation in the soil mass during pipe movement. On the other hand, in tests with moist sand, the passive zone in front of the 177 pipe moved as a solid block with no evidence of formation of an active deformation zone (see Figure 5-17 and Figure 5-18). It is important to note that in all the tests conducted simulating trench-configurations in "hard native soil", passive zone was forced to form inside the trench. This is clearly in contrast to the tests with no hard boundary (such as Tests No. LN-1 and LN-2), where the trace of passive wedge in front of the pipe could be extended for distances in the order of 250 cm in front of the pipe. 250 Displacement from Pipe Centre (mm) Figure 5-16: Surface deformation and pipe position after pulling of the pipe: tests LT-1 and LTG-1 178 40 -Displacement From Pipe Centre (mm) Figure 5-17: Surface deformation and pipe position after pulling of the pipe: tests LT-2 and LTG-2 Figure 5-18: Photo of surface deformation after pulling of the pipe: tests LTG-2 179 In all "hard boundary" tests, when the pipe location was examined after removal of backfill upon completion of testing, revealing the pipe and emptying backfill, the closest distance between pipe and the boundary was about 3 to 5 cm (see Figure 5-16 and Figure 5-17 for pipe locations). It was noted that this proximity was essentially the same regardless of the pipe displacement during testing. The pipe movement in the initial part of "hard-boundary" tests may be similar to those conducted without trenches. However, once the presence of the hard boundary is "felt" by the pipe (i.e., once the leading pipe surface reaches within about 3 to 5 cm from the trench surface), it appears to preferably move upwards parallel to the trench surface while maintaining a 3 to 5 cm proximity from the trench surface. In Test No. LTG-2 with moist sand backfill, the maximum pipe displacement was about 400 mm and the passive wedge in front of the pipe also slid about 400 mm along the trench wall (see Figure 5-17). For the case in which there is no local shearing in sand and pipe directly pushes the sand in front of it to move as a solid block, 400 mm of horizontal displacement of pipe, should result in about 500 mm of soil block movement along the trench (400 mm / cos (35) = 490 mm). It can be inferred that although observation of "after-test" surface deformation implies that the soil in front of the pipe moves as a block, there is some penetration of pipe in backfill sand. Figure 5-15 also shows that the differential displacement between geotextile layers is almost the same as the above "soil block" displacements. This would imply that the soil block essentially slipped at the interface between the two geotextile layers. In Test No. LTG-1 with dry sand backfill, the differential displacement between geotextile layers is less than half of the pipe displacement (see Figure 5-15). The comparison of these displacements with the surface deformations (see Figure 5-16) suggests that, in the test with dry sand backfill, the soil in front of the pipe would not have moved as a solid block. Instead, it appears that more complex mechanism would have taken place with combined involvement of deformations: at the interface between the two geotextiles, at the interface between backfill/outside layer, and within the sand 180 mass itself. This argument would also indirectly support the relatively higher values of peak load observed in Test No. LTG-1 (with dry sand backfill) compared to LTG-2 (with moist sand backfill). 5.6 Discussion of test results 5.6.1 Pipeline buried in cohesionless soil The test results from the current study provided an opportunity to compare with published data from other lateral pipe load testing work and assess the different methods available for the estimation of loads on pipelines subjected to lateral soil loadings. The data from soil pressure measurements on pullout testing were also helpful in obtaining more insight into the pullout mechanisms involved, further evaluating analytical approaches, and indirectly assessing the measured pullout loads. 5.6.1.1 Comparisons with previous studies and current approaches Laboratory testing and field experiments on buried pipes, anchors, and foundations have led to the current state-of-practice guidelines for pipeline design such as A S C E (1984), A L A (2001), and PRCI (2004). In this section, an attempt is made to compare the results from previous leading studies with the observations from this study. The differences in the comparisons are identified and the reasoning for the variations is discussed. Figure 5-19 presents the dimensionless peak load versus overburden ratio relations obtained from the test results reported by Audibert and Nymann (1977), Trautmann and O'Rourke (1983), Hsu (1992), and Calvetti et al. (2004). The pipe diameter in each test is also indicated in Figure 5-19. As may be noted, these previous testing have been performed on pipes smaller in diameter than those used in the present study, and they have been mainly limited to tests in dense sand backfill, with relatively minor variation in the reported friction angle for each case. For example: (i) in tests performed by Audibert 181 and Nyman (1977), a friction angle of 40° has been reported; (ii) Trautmann and O'Rourke (1983) performed direct shear test on dense sand and reported a friction angle of 46° for the sand used in their experiments; (iii) friction angle in the tests performed by Hsu (1992) was 42°, again, from direct shear test results; (iv) Calvetti (2004) estimated friction angle of 40° from their triaxial testing. He extrapolated the results for dense sand from the tests using loose and medium sand with lower friction angles. The results from different tests shown in Figure 5-19 indicate a wide variation in the lateral soil loads. This can be partly attributed to the differences in soil parameters including (i.e., friction angle and density) combined with lack of information on the testing details, and random errors. In spite of these possibilities, it is of relevance to observe that the results from the current study plot in lower part of the chart compared to the data from most previous studies on pipe buried in dense sand backfill. In particular, the data from Audibert and Nymann (1977), Calvetti (2004), and Hsu (on D.76mm pipes) show much higher values for peak loads even though the friction angle in these tests are lower than the current study, and experiments performed by Trautmann and O'Rourke (1983). Scale effects have observed to result in higher predicted loads in small scale tests than prototype (Yamaguchi and Kimura (1976), Vesic (1970), Guo (2005)). The observed higher dimensionless peak loads from tests on relatively smaller pipes (20, 25, 50, 62 and 76mm diameter) as per Figure 5.19 seem to be in accord with this view point. The results of Turner (2004) is close to those of Hsu (1992) on 76 mm diameter pipes and Trautmann and O'Rourke (1983). 182 -•-Audibert and Nymann, D=25mm, <p=40° (1977) -^-Audibert and Nymann, D=62mm, (p=40° (1977) -a-Trautmann & O'Rourke, D=T02mm, (p=46° (1983) Hsu, D=76mm, (p=42° (1992) • Hsu, D=152 and 228mm, 9=42° (1992) —1— Calvetti, D=20 and 50mm, (p=40° (2004) -•—Current study, D=324mm, cp=43 to 46° O Current study, D=457mm, cp=43 to 46° A Turner, D=l 19mm, cp=41 to 44° (2004) A Turner, D=l 19mm, q>=47° (2004) L ' k k \ 0 1 2 3 4 5 6 7 Overburden Ration (H/D) Figure 5-19: Comparison between the dimensionless peak lateral load ( F L ' ) obtained from the current study and published test results 183 The tests load-displacement response curves can also be compared to the A S C E (1984) and A L A (2001) recommendations for the designs of buried pipelines in cohesionless material. The load displacement characteristic suggested by the both guidelines is based on the assumption of hyperbolic stress strain response for soil. Two charts are presented in the guidelines are based on: (i) studies conducted by Hansen (1961) on buried piles which was later adopted by Audibert and Nymann and reported good agreements in pipeline problems and ii) studies by Ovesen (1964) and Ovesen and Stromann (1972) on vertical anchors. Trautmann and O'Rourke (1983) reported good agreements between the loads predicted by charts presented by Ovesen with the test results on pipelines. The relation between load per unit length of the pipe and pipe displacement can be found from Equation [5-1] Y P = [5-1] (0.\5xYuIPu) + (0.Z5IPu)xYu Where Pu = Ultimate soil resistance = Y.H.N q n.D Nqh~ Horizontal bearing capacity (see Figures 2-1 and 2-3) D = external pipeline diameter Y = Pipe displacement Yu = Pipe displacement at which maximum soil resistance is expected H = depth from ground surface to centreline of the pipe 184 180 T Displacement (mm) Figure 5-20: Comparison of test results with ASCE (1984) / A L A (2001) Guidelines (H/D = 1.92; D =457 mm) Figure 5-20 through Figure 5-23 present a comparison between the test results and the predictions made using the above methods for Tests No. LN-1 through LN-7. These tests simulate lateral loading response of pipeline buried in cohesionless "native soil" (see Section 5.3.1 for test details). The following soil parameters (derived from laboratory testing that was undertaken as a part of this study) were used for predictions using the Hansen (1964) and O'Rourke (1983) methods. Average peak friction angle of soil for normal stress range of 20 kPa to 100 kPa (in direct shear) (j) = 44° (see Table 5-1) and density of soil y = 1600 kg/m [Note: The peak friction angle from direct shear testing was used in the computations to be compatible with the test methods used by Hansen (1964) and O'Rourke (1983)]. It can be seen that the O'Rourke (1983) method slightly over-predicts peak loads while Hansen (1964) method over-predicts the load by several-fold. The American Lifeline Alliance (ALA 2001) suggests the application of Hansen (1964) chart to predict the soil loads; the above tests results from the current study suggest that this recommendation may lead to significant over-prediction of soil loads. 185 90 -• ~ 80 -0 10 20 30 40 50 60 70 80 90 100 Displacement (mm) Figure 5-21: Comparison of test results with ASCE (1984) / A L A (2001) Guidelines (H/D = 1.92; D =324 mm) Nqh from the guidelines has the same definition as F L ' , and the values for friction angle of 44° (as described above) are plotted in Figure 5-24 along with the test results. Clearly, there is correspondence between the prediction using O'Rourke (1983) method and the soil loads from pipe testing in this study. On the other hand, the over-prediction of the lateral loads using the Hansen (1964) method is again visible. Guo and Stolle (2005) suggested modifications on predicted load from ASCE (1984) with consideration given to scale effect and effect of dilation. Since the tests of this study were on relatively large diameter pipes buried with moderate overburden ratios, the correction proposed by Guo and Stolle (2005) for effect of size and dilation on Nqh values are less than 1% and 6% respectively and the modified values of Nqh are almost the same as original values. 186 40 ^ 35 E " 0 30 ja +-M S 25 M 1 20 u-u a 0 J 2 1 0 u «3 N J 5 ____ _„ •-? / / 7 / / ; £v LN-5 - - O'Rourke (1984) - - Hansen (1964) 1 1 i i 20 40 60 Displacement (mm) 80 100 Figure 5-22: Comparison of test results with ASCE (1984) / A L A (2001) Guidelines ( H / D = 1.0; D = 324) 140 g 120 -" 100 -bSo a u J u CL •a a o J / / / / / / / ^ LN-6 O'Rourke (1984) f 20 40 60 Displacement (mm) 80 100 Figure 5-23: Comparison of test results with ASCE (1984) / ALA(2001) Guidelines (H/D = 2.75; D = 324 mm) 187 O 5 - -- ;• r -j 0 ] ; ; j 1 j 1.0 1.5 2.0 2.5 3.0 3.5 4.0 Overburden Ration (H/D) Figure 5-24: Comparison of N Q N from guidelines with F L ' from test results 5.6.1.2 Interpretation based on measured soil stress on pipe It was considered reasonable to independently assess the horizontal loads on the basis of the data from pressure transducers that directly measured soil stresses on the pipe during lateral pulling. As mentioned in Section 5.3.2, the pressure transducer readings were averaged (using moving averages) and normalized to obtain the O ' N ( = o ' n /o\) stress variation around the pipe and with increasing lateral displacement. The transducers used in the pressure measurements were installed to be flush with the circumferential surface of the test pipe (see Figure 3-8 and inset of Figure 5-26 for the location of pressure transducers around the circumference). As may be noted, the transducer PT3 would face the horizontal direction and it allowed direct measurement of the lateral soil stress at the pipe centreline level. However, the transducers PT2 and PT4 are inclined to the horizontal by 45 degrees. Since the objective is to examine the correspondence between the pressure transducer readings and the measured lateral loads 188 during lateral pulling, the non-dimensional normal stress O ' N on the pipe from the transducers PT2 and PT4 was "resolved" to compute the stress components O 'NH on the pipe in the horizontal direction (with depth). This "resolving" of stresses was achieved using the concept of Mohr circle. It was assumed that, at transducer locations PT2 and PT4, the direction of maximum shear is tangent to the pipe surface and that the vertical stress at these two transducer locations are equal to the direct stress from the overburden. Using these assumptions, a Mohr circle was constructed as shown in Figure 5-25, in turn, allowing the interpretation of O 'NH at the positions PT2 and PT4. Figure 5-25: Concept of calculating normalized horizontal soil stress component CJ'NH at the locations PT2 and PT4 using stresses measured perpendicular to the pipe circumference The a ' N H components calculated at the three transducer locations as per above were normalized with respect to the vertical overburden effective stress at the center line level of the pipe; Figure 5-26 shows the variation of these values with respect to the normalized displacement up to 0.10 during lateral pulling. The peak stress components (derived from the pressure transducers) in Figure 5-26 seems to occur at a normalized 189 displacement Y ' (=Y/D) of about 0.07. It is interesting to note that the dimensionless load F L ' observed during the same test (See Figure 5-3) using load cell measurements also occurs around the same Y ' value, clearly indicating the correspondence between the lateral loads and the induced soil pressures on the pipe. (Note: Due to the almost constant lateral load observed on the pipe after reaching the peak, for presentation purposes, the post-peak dimensionless horizontal stress component C J ' N H depicted in Figure 5-26 was computed by averaging the pressure transducer readings observed between Y ' = 0.07 to Y'=0.1) . 14.0 0.00 0.02 0.04 0.06 0.08 0.10 N o r m a l i z e d D i s p l a c e m e n t Figure 5-26: Average dimensionless horizontal stress component C J ' N H at the locations of pressure transducers P T l , PT2, and PT3, during Test No. LN-7 The correspondence between the pressure transducer readings and the measured lateral loads can be further examined, as shown in Figure 5-27, by plotting the peak C J ' N H component values extracted from Figure 5-26 with pipe depth. This interpretation indicates that the horizontal component of stress along the leading perimeter zone increases with increasing pipe depth. This irregular C J ' N H distribution can be replaced by 190 an equivalent uniform distribution, as shown by the shaded area, and it would result in an equivalent normalized horizontal stress component (o 'NH)eq value of 7.3. If the lateral stresses on the trailing side of the pipe is assumed to be relatively small in magnitude (considering the "active" soil conditions prevalent on that side), the above parameter (o'Ni-Oeq is directly comparable with the peak dimensionless lateral load F L ' , which is the net average lateral soil stress on the pipe for a given overburden soil stress during lateral displacement as per Section 5.2. It is of interest to note that the above (c'NhOeq of 7.3 interpreted for Test No. LN-7 is in fact in good agreement with the peak F L ' of 7.2 estimated from the pulling loads for the same test. (0"'NH)eq -7.3 Dimensionless Load = 6.8-7.2 Test LN-7 Figure 5-27: Lateral soil stress distribution in front of the pipe in Test No. LN-7 The numerical similarity between (o'NH)eq and F L ' values also indirectly suggests that the soil chamber used in the current study has demonstrated success in physical modeling of the lateral pipe loading problem in a 2-D plane-strain manner (i.e., the correspondence between lateral loads estimated from pressure transducers and pulling load cells 191 indirectly suggests that the sidewall friction and pulling system friction in the lateral load testing operation). To the best of the author's knowledge, no previous attempts have been made to measure the soil pressures on pipelines and use these measurements for the interpretation/estimation of lateral loads on pipes subject to relative ground movement. The soil pressure measurements on pipe, in addition to conventionally measured lateral load vs. displacement response, also provides an independent set of data that would assist effective validation of numerical models simulating this problem. 5.6.2 Pipeline buried in trench-configurations 5.6.2.1 Evaluation of simplified mechanisms to estimate the sources of soil loads on pipe As discussed in Section 3.2.2, it was originally perceived that pipe embedded in a dual-geosynthetic fabric-lined trench would experience preferred slippage at the low-friction geosynthetic fabric-geosynthetic fabric interface when subjected to ground movements. Comparing the plotted results for tests without geosynthetic fabric lining and tests with geosynthetic fabric lining, it can be concluded that the geosynthetic fabric lining reduced F L ' by about 15% to 20%. However, considering that the 20° interface friction between two geosynthetic fabrics is significantly lower than interface friction angle between trench-backfill and trench surface (36° to 45°), a reduction of F L ' would be estimated to be at least 50% based upon simple sliding models. Given below is an attempt to understand the potential mechanisms that would have led to the experimental observations. Since the formation of passive wedge and sliding movement along the interface was more evident with moist sand backfill, the tests conducted using moist sand are used to assess the validity of mechanical models proposed to represent pipeline response. 192 The ideal scenario for effective load reduction would be to have a passive soil wedge in front of the pipe move as a rigid block with slippage between the geosynthetic fabric layers, as shown schematically in Figure 5-28. The expected forces, considering the passive wedge as a free body, are given by Equation [5-2]. [5-2] Figure 5-28 Applied load on the pipe and passive wedge Assuming a soil density of 1600 kg/cm and a mobilized friction angle 20° and 35° at the sliding interface with and without geosynthetic fabric-lining, respectively, the required horizontal pipe load (F L) to push the passive wedge along the trench can be calculated to be 20 kN/m (equivalent F L ' of 3.1) and 38 kN/m (equivalent F L ' of 6.0), with and without geosynthetic fabric-lining, respectively. As shown in Figure 5-12 and Figure 5-13, the average experimental horizontal load to induce significant movement of the pipe for these 193 two cases is about 45 kN/m (F L ' =7.0) and 52 kN/m (F L ' =8.1), respectively. Furthermore, these loads tend to increase as the pipe approaches the trench wall. These significant differences between the experimental and computed values (both in terms of the magnitude of F L and the percentage reduction of F L observed between the cases with and with geosynthetic fabric liner) suggest that the mechanism in Figure 5-28 does not represent the test condition. An alternate mechanism, as shown in Figure 5-29, is based upon an assumption of the soil behaviour between the pipe and the hard boundary. Observations following the tests clearly indicated that the pipe moved through the backfill and approached the hard boundary and the pipe experienced some vertical displacement. However, the pipe was observed to not contact the hard boundary in any of the tests. Instead, the pipe reached approximately the same distance to the hard boundary. This implies that the soil between the pipe and the hard boundary becomes "locked" in place and the soil between the pipe and the hard boundary is forced to squeeze through the space between the pipe and the hard boundary as the pipe moves horizontally and vertically. If it is assumed that this squeezing effect results in shearing of the soil between the pipe and the hard boundary, the component of the horizontal pipe force acting normal to the hard boundary creates a frictional resistance that is more dependent upon the internal friction of the soil instead of the geotextile boundary. This leads to the modified equation for estimating F L given by Equation [5-3]. _ cos(a). tan(c>) + sin(a) 1 cos(a)-sin(a).tan(^) [5-3] Assuming the behaviour described above leads to values of F L ' of 7.2 and 9.4 for the case with and without geosynthetic fabric-lining, respectively. These values are a much better match to the average experimental values of 7.0 and 8.1 for the case with and without geosynthetic fabric-lining, respectively. 194 Figure 5-29 Free body diagram of the passive wedge and pipe 5.6.2.2 Evaluation of measured soil pressures on pipe during lateral pulling tests Data from pressure transducers, that directly measured soil stresses on the pipe during lateral pulling tests on trenched configurations, enabled further examining the credibility of the above postulated mechanisms with regard to the effects of the proximity to the "interface boundary" on the lateral load on pipes. In a similar manner to the approach for the lateral pulling tests in "native soils" described in Section 5.6.1.2, the pressure transducer readings were initially averaged (using moving averages) and normalized to 195 obtain the O 'N ( - o-'n /o' v) stress variation around the pipe and with increasing lateral displacement. Then the stress components O'NH on the pipe in the horizontal direction (with depth) were computed using the Mohr circle concept (see Figure 5-25). In the tests conducted with trenched configurations, both the load-displacement response and pressure transducers measurements were examined considering three different phases of pipe movement: (a) initial stage where the load is rapidly mobilized on the pipe (Dimensionless pipe displacement of Y'= Y/D = 0 to 0.02 ~ 0.1); (b) the second stage of displacement where the pipe is yet to "recognize" the presence of the trench boundary (Y' = 0.02-0.1 to Y ' = 0.25-0.3), and (c) the final stage where the pipe moves along the trench after recognizing the presence of the trench boundary (Y' = 0.3 to the end of the test). Three tests (Tests No. LT-1, LT-2, and LT-3) were conducted with hard boundary, at constant H/D ratios and pipe diameter but without geotextile lining, were initially examined. (Note: The Test No. LT-1 was conducted with a dry sand backfill; the Tests No. LT-2 and LT-3 were conducted with moist backfill, and they were identical). The variation of normalized horizontal stress components (O'NH) from Test No. LT-1 and from Tests No. LT-2 and LT-3 (average response from the two tests) are shown in Figure 5-30 and Figure 5-31, respectively, for the first and second stages of pipe movement. If these two stages occur before the pipe has "recognized" the presence of trench boundary, then it can be argued that the response in these stages should be comparable to the result of tests in native soil. In order to investigate this aspect, the representative CJ'NH values were extracted from information presented in Figure 5-30 and Figure 5-31 and re-plotted in Figure 5-32 using a format similar to Figure 5-27. Again, similar to the format in Figure 5-27, the irregular O 'NH distribution was noted to be replaced by a uniformly distributed equivalent normalized horizontal stress component value (a'NH)eq, as shown by the shaded area in Figure 5-32 (Note: Figure 5-27 was constructed based on pressure transducer measurements for the Test No. LN-7 with native soil backfill). 196 Figure 5-30: Lateral soil stress based on pressure transducers reading in Test No. LT-1 • x 1 ! PT2 PT3 V PT3 7 P T 2 j / . , PT4 PT3 — PT2 f 0.00 0.05 0.10 0.15 0.20 0.25 0.30 Normalized Displacement Figure 5-31: Lateral soil stress based on pressure transducer reading in Tests No. LT-2 & LT-3 197 (tf'NH)eq-7.2 (o 'NH)eq-5.9 Dimensionless Load = 7.2-7.5 Dimensionless Load = 6.0-7.0 Test No. LT-1 Tests No. LT-2 and LT-3 Figure 5-32: Lateral soil stress distribution in front of the pipe in Tests No. LT-1 to LT-3 A comparison between Figure 5-32 and Figure 5-27 would reveal the following: (i) the horizontal component of stress distribution around the pipe (O'NH), a n d therefore the equivalent normalized horizontal stress component value (o'NH)eq, obtained from Test No. LT-1 (approx. 7.2) is similar to what was observed during the Test No. LN-7 (approx. 7.3); (ii) the value of (o'NH)eq on the pipe in Tests No. LT-2 and LT-3 (approx. 5.9) are lower than those for Tests No. LT-1 and LN-7. While this difference is in accord with the observed -15% lower lateral load on the pipe in Tests No. LT-2 and LT-3, The shape of the (O'NH) distribution is in agreement with the Tests No. LN-7 and LT-1. and shows that using moist sand as backfill material, which results in movement of the passive wedge as a block, does not significantly influence the stress distribution around the pipe. These observations support the postulate that the pipe has not "recognized" the presence of the trench boundary during first and second stages of pipe movement (i.e., dimensionless pipe displacement = Y'= Y/D < 0.25 - 0.3). With this background, the averaged normalized horizontal stress components (O'NH) for all tests with the hard boundary are presented in Figure 5-33. Clearly, a significant increase in lateral soil stress can be noted at the pressure transducer PT2 (i.e., pressure 198 transducer located at -45° radial location to horizon. The other two transducers either do not show an increase in the stress or the increase is negligible compared to PT2). 18 T 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00 Normalized Displacement Figure 5-33: variation of average lateral soil stress in tests with trench configuration with pipe displacement Figure 5-34 is prepared (in a similar manner to Figure 5-32) with the results of average dimensionless lateral soil stress (O'NH) values extracted from Figure 5-33 based on tests conducted with trenches simulating hard native soil. In this figure, the variation of average dimensionless lateral soil stress (O'NH) around the pipe in the first and second stages of pipe displacement (i.e., first -200 mm of pipe displacement) are compared with those observed in the third stage of pipe displacement (i.e., from 200 mm to 400 mm of pipe displacement). The measured soil pressures on the pipe are also examined in Figure 5-34 in terms of the uniformly distributed equivalent normalized horizontal stress component value (o 'NH)eq, as defined earlier. The observed (a 'NH)eq and F L ' (computed from measured loads) correspond well providing mutual confirmation of the validity of the measured values. 199 The comparisons in Figure 5-34 clearly demonstrate the significant increase in the soil pressures in the lower transducer PT2 located at -45° radial location to horizon when the pipe moves towards the hard boundary (Note: transducer No. PT2 is the one nearest to the hard boundary). This suggests that it would be prudent to provide an increased clearance between the pipe and the trench surface, especially in the case of trenches in hard native soil, or perhaps use a softer material between the pipe and trench. (O" )Average 5.7 (o )Average 7.7 Dimensionless Load ~ 6.5 Dimensionless Load ~ 8.0 Before reaching the trench After reaching the trench Figure 5-34: Lateral soil stress distribution in front of the pipe in tests with trench configuration 5.7 Numerical modeling The soil-pipe interaction during horizontal lateral pipe displacement was numerically modeled using F L A C 2D® version 4.0, finite difference method based software. Soil behaviour was represented using parameters obtained from laboratory element testing. The results of the numerical model were calibrated and validated with the results of performed tests. In addition to the validations, the model could also be used to predict 200 the soil loads on pipe in conditions with different configurations and to investigate the effect of different parameters on pipeline response to the ground movement. 5.7.1 Development of mesh configurations The problem was modeled considering a plane strain condition with structural elements representing the pipe. The size of the model and geometric conditions including boundary conditions were defined similar to the conditions of full-scale tests. As described in Chapter 3, length and height of the box were 3.78 m and 2.5 m respectively and the pipe centre was located 70 cm from the bottom of the box and 120 cm from the back wall. The finite difference (F-D) mesh was defined to provide finer discretization around the pipe and areas subjected to high strains. A coarser discretization was considered acceptable in the areas where no significant deformations were expected. Some key parameters such as analysis time, geometry, accuracy of the results, and convergence were considered in designing the mesh. While a larger mesh size is more time efficient, it might yield inaccurate results or might not converge to a unique value. On the other hand, using a finer mesh guarantees more accurate results, but increases the analysis time. Using a very fine mesh in areas that are subjected to relatively high strains may result in instability of the model and distorted elements. After examining several mesh configurations for convergence and minimum analysis time, those shown in Figure 5-35 through Figure 5-38 were selected to simulate tests conditions of: (I) H/D = 1.92 and D= 457mm (LN-1 and LN-2), (II) H/D = 1.92 and D = 324 mm (LN-3, LN-4, and LN-7), (III) H/D = 1 and D = 324 (LN5), and (IV) H/D = 2.75 and D = 324 mm (LN-6), respectively. The number of elements in the model is equal to 1176 for configurations (I) and (IV), 1029 for configuration (II) and 833 for configuration (III). The mesh size allowed more than 100 mm of pipe displacement in all conditions. The use of mesh dimensions equal to half of the above mesh sizes was found to limit the pipe displacement to less than 60 mm and increased the analysis time by four times; while the load-displacement curve for this finer mesh was within 1% of that computed 201 with the selected meshes shown in Figure 5-35 through Figure 5-38. Boundary conditions are identical in all models as shown in Figure 5-35. The pipe was modeled with a series of beam elements to form a circular shape with weight and stiffness of the beam elements matching the actual pipe test specimen. When compared to the stiffness of the soil, the pipe behaves like a rigid body. 1.58 m 0.88 m 3.78 m Free surface Horizontal displacement is restrained Displacement are restrained in both directions 1.20 m Figure 5-35: Mesh configuration for modeling Tests No. LN-1 and LN-2 3.78 m 1.32 m « M hi fij '.1 Cl nm r i •••» rt§ '.1 M rM M M >:t l » « n mam mmmm EE 1.20 m Figure 5-36: Mesh configuration for modeling Tests No. LN-3, LN-4, and LN-7 202 3.78 m 1.02 id r c t* L | r J 1 fc" sll 1 L~ ISIC'J" >isifl C" r K B HB B H ti ti ti ti ti t H y H h I fe r • H H i H B H H H U H H H B B « — > 1.20 m Figure 5-37: Mesh configuration for modeling Tests No. LN-5 Figure 5-38: Mesh configuration for modeling Test No. LN-6 5.7.2 Parameters and constitutive models for numerical analysis Both the shape of the load-displacement curve and the magnitude of the maximum load are sensitive to the selected soil constitutive model and the representation of the soil-pipe interface. In consideration of this, numerical modeling was undertaken using a number of constitutive models for comparison with the test results as described in Sections 5.7.2.2 to 5.7.2.4. 203 Since the soil-pipe interface was modeled using only one approach (i.e. same interface models with all cases and soil constitutive models), the approach for modeling the interface is first described below in Section 5.7.2.1. The overall objective is to select a model that represents the soil-pipe system within acceptable error, yet is simple to use, with parameters that are easily obtained from simple laboratory tests. To simulate horizontal displacement of the pipe, a steady rate of displacement equal to lxlO" 8 to 2xl0" 8 m per step was applied to the structural elements, based on the rate of convergence. 5.7.2.1 Modeling of pipe soil interface Unbonded interface elements were used to model the interface between pipe and soil. These interface elements with Coulomb shear-strength criterion allows for slippage when shear force between soil and pipe exceeds Fsmax as per Equation [5-4]. Fsmaii =c.L + Fn.\an(d) [5-4] Where: c = cohesion (in stress units) along the interface L = effective contact length along pipe d = interface friction angle and F n = normal force per unit length on the interface In the current model, since cohesionless material is considered, Fsmax is only a function of interface friction and normal stress. Interface friction angle between dense sand and sand-blasted steel was obtained using the results of direct shear tests. As presented in Chapter 3, peak and constant volume friction angle at pipe-sand interface was estimated to be 36° and 31° respectively. A constant value for friction angle of 31° was selected as input parameter for numerical model in this study. Sensitivity analysis on the effect of interface friction angle on the soil loads on pipe indicated that such variations (e.g. from 204 36° to 31°) have negligible effect. The selection of this value for interface friction angle was also in line with direct shear tests results, in which it was observed that the interface friction reaches the constant volume stage relatively quickly after 1 to 2mm. In real life situations, the effect of dilatancy is high at the onset of shear and decreases with increase in strain when the shear strength changes from peak to constant volume. In the current model, dilation at the interface is defined by a constant specified dilation angle, y/. The interface dilates if the shear displacement increment is in the same direction as the total shear displacement and contracts when they move in opposite directions. Based on this, a dilation angle equal to a constant value of 3° was used; however, sensitivity analysis on interface dilation angle showed that variation of this value from 0° to 10° results in less than 1% increase in soil loads on pipe. 5.7.2.2 Linear Elastic-Perfectly Plastic Mohr-Coulomb model Soil elements used in the analytical model exhibited linear elastic-plastic behaviour based upon a Mohr-Coulomb failure criterion. Elastic properties parameters are required as well as parameters describing failure surface and sand density (y). As for elastic properties, each two parameters from shear modulus (G), Poisson's ratio (u), elastic modulus (E), and bulk modulus (B) can describe elastic behaviour of the material. Internal friction angle (cp) and dilation angle (\|/) are also required to determine failure surface and volume change of the material. Although studies performed by Byrne et al. (1987) showed that Poisson's ratio (u) varies in the range of 0.1 to 0.5 for small strain levels and strains at failure respectively, a constant value between 0.3 to 0.35 is usually selected in practice. In the current study, it is assumed that Poisson's ratio is constant and has a value of 0.3. Other elastic parameters such as elastic modulus or shear modulus were derived from the results of triaxial testing as described in Chapter 2. 205 Initial elastic modulus, Ej, calculated from triaxial test results as shown in Figure 3-15 and are repeated here in Figure 5-39. The predicted initial elastic modulus for target soil density of 1600 kg/m 3 and various confining stresses using curve fitting is also shown in Figure 5-39. The results indicated that Ej at this target density and for confining stress level of 20 kPa, which is almost equal to the overburden stress at the pipe springline during the test configurations (I) and (IV), can be estimated as 36 MPa. While using a linear elastic - perfectly plastic model, a secant elastic modulus should be selected to represent the behaviour of material. As accepted in previous analytical works (Byrne et al. 1987), a secant elastic modulus (E s e c) equal to one third of initial modulus was selected as reasonable. The elastic properties that were used in bilinear model as per above are given in Equations [5-5] and [5-6]. E s e c = Ej / 3 = 12 MPa [5-5] G s e c = E s e c / 2(1+D) = 12 / 2(1+0.3) = 4.6 MPa [5-6] E Z S 10 20 30 40 Confining Stress (kPa) 50 60 Figure 5-39: Initial elastic modulus calculated from triaxial tests results 206 Figure 5-40 showing the variation of peak friction angle with stress level and soil density based on the results of triaxial testing is also repeated here from Chapter 2. Similar to Figure 5-39, the interpolated results show peak friction angle at a target soil density of 1600 kg/m are shown in this figure as well. It can be observed that the peak friction angle at the target density and confining stress of 20 kPa is about 45°. Dilation was modeled based upon the relationship proposed by Bolton (1986) as shown in Equation [5-7]. Based on Bolton's relationship and considering a constant volume friction angle ((pcv) of 33° (Uthayakumar, 1996, Syvathayalan 2000), dilation angle of dense sand is calculated to be about 15°. y = (9max-(Pcv)/0.8 [5-7] Figure 5-40: Peak friction angle calculated from triaxial tests results Figure 5-41 shows the load-displacement response predicted using the numerical model results, with soil behaviour represented by linear elastic-perfectly plastic Mohr-Coulomb constitutive model as described above for test configuration (I). Also the result of tests LN-1 and LN-2 are plotted in the same figure for comparison purposes. The bilinear 207 model over-predicted the soil loads on pipe and did not represent the development of soil loads with displacement. The predicted peak load was almost 17% higher that that recorded in experiments. 70 60 50 4 M An a 40 a 30 M O 20 10 / a " * " Mohr-Coulomb Model — LN-1 — LN-2 1 j ' ' • v 20 40 60 80 Displacement (mm) 100 120 Figure 5-41: Simulation of Tests No. LN-1 and LN-2, using bilinear Mohr-Coulomb compared with test results 5.7.2.3 Hyperbolic Mohr Coulomb Model Predictions were also made using hyperbolic model as described by Duncan and Chang (1970) and Byrne et al. (1987). The Mohr-Coulomb criterion controls failure of elements. The constitutive model offered in FLAC 4.0, accounts for both elastic and plastic deformation of soil elements. Principal stresses and directions are evaluated from the stress tensor and corresponding strain increments are decomposed in the form of plastic and elastic strains as presented in Equation [5-8]. Asi = As- + As? i = l,2,3 [5-8] 208 where the subscript " i " represents principal directions and superscripts "e" and "p" refer to the elastic and plastic part of the strain increment respectively. In this model, the plastic components have values other than zero only during plastic flow (i.e. after failure of elements). The relationship between elastic strains and principal stresses is described by Hooke's law while flow rule controls the relation between plastic strains and principal stresses. The Hooke's law in incremental expression is as appear in Equation [5-9]: where: ai = B + 4G / 3 a 2 = B - 2G /3 and B and G are bulk and shear modulus of the material respectively. The failure envelope consists of a non associated shear flow rule (Mohr-Coulomb) and an associated tensile flow rule (tension cut-off). The failure envelope as defined by Mohr-Coulomb yield function is presented in Equation [5-10]: l - sin cp c = cohesion and <p = internal friction angle The shear potential function, corresponding to a non-associated flow rule where dilation angle is not zero, is in the form of Equation [5-11]. ACT', = a, Ae\ + a2 (Ae^ + Ae\ ) [5-9] [5-10] where: major principal stress minor principal stress l + sin^» 209 gs =cr\-cr\.N, [5-11] where: l + s inT N„ = * l - s i n T and \|/ = dilation angle The flow rule can be written as Equation [5-12]: KeP-X'^- [5-12] da' in which Is is a parameter of magnitude and a function of friction angle, dilation angle and stress condition. Implementing g s as expressed in Equation [5-11] in Equation [5-12] led into the Equation [5-13] as shown here: Aef = Xs Asp2=Q [5-13] As shown in Equation [5-8], elastic strain can be expressed as total strain minus plastic strain. The flow rule as shown above in Equation [5-13] can be used to rewrite Equation [5-9] in terms of total strain increments: ACT ' , = a,.Ae, + a2(Ae2 + A<?3)-Xs{ax -a2N^) A<j'2 = al.Ae2+a2(Ael + Ae3)-Axa2(l-N^) [5-14] Acr'j = avAe3 +a2(Ael + Ae2) - Xs {-axN^, +a2) For more details of formulation and the modeling procedure of Mohr-Coulomb constitutive model refer to Section 2.4.2 of F L A C 4.0 manual. 210 The tangent shear modulus at each strain level in the hyperbolic model, which provides the stiffness for Hooke's law, is defined by Equation [5-15]. G, tan = G , x ( l - max [5-15] max where: Gtan = Tangent shear modulus q = o'i - a '3 = Deviative stress qf = Deviative stress at failure Rf = Failure ratio The shear stress at failure (tf) can be calculated from Equation [5-16]. where a '3 = effective confining stress and cp = peak friction angle. In this model, stress dependency of material properties is also considered. As shown in Figure 5-39, curve fitting provides the best estimation of initial elastic modulus. The generally accepted relationship between stress and elastic modulus presented by Duncan and Chang (1970) is provided in Equation [5-17]. 9/ = 2xcr' 3xsin(^) [5-16] (1 - sin(p)) £ , = 2 W — ) [5-17] 211 where: Ej = initial elastic modulus Eiref = initial elastic modulus at reference confining stress a '3 = effective confining stress o r e f ~ reference confining stress (usually equal to 100 kPa) Interpolation of the triaxial tests results, with reasonable assumption of linear variation of elastic modulus with soil density in the range of 1575 kg/m3 to 1665 kg/m , and considering the relationship between elastic modulus and shear modulus (i.e. G = E / 2(l+u)) has been performed. Equation [5-18] can best describe the variation of initial shear modulus with confining stress level for dry Fraser River sand with density of 1600 kg/m3 (D r = 75%). G , = 3 6 0 x P f l x ( ^ l ) 0 6 [5-18] In which P a is atmospheric pressure. Substituting if as presented in Equation [5-16] and Gj as presented in Equation [5-18] in Equation [5-15], led into Equation [5-19]. This equation was directly used in numerical modeling of lateral horizontal pipe pulling. G „ . ^ x ( V x ( l - J , i ^ ) ' [5-19] 100 zxcr 3 x s i n ^ R f is an experimental factor which depends on the material properties and varies in the range of 0.6 to 1.0. Various researchers have been studied to estimate R f value. Duncan (1970) suggested 0.9 and Byrne (1987) suggested 0.7 as Rf. The value of R f was back-calculated from triaxial test results. The back-calculated values of R f are presented in Table 5-2 for different tests. 212 Table 5-2: Rf values based on Triaxial tests results Confining stress (kPa) R f Value Density of 1575 kg / m 3 Density of 1665 kg/m 15 0.9 0.83 25 0.91 0.91 35 0.88 0.9 50 0.9 0.88 As can be observed from in Figure 5-40 peak friction angle was found to be dependent on the confining stress level. For a better representation of soil behaviour, stress dependency of friction angle was also considered in this model. Curve fitting showed that a quadratic polynomial reasonably represents the variation of friction angle with confining stress. Assuming that variation of peak friction angle with relative density is linear for soil density ranging from 1575 kg/m3 to 1665 kg/m 3, interpolation of triaxial tests results led to Equation [5-20] which represents variation of peak friction angle with confining stress for Fraser River sand at density of 1600 kg/m3 (D r = 75%). <P = 0.002CT'32 -0.2<T'3 +48 [5-20] The trend of variation in friction angle indicates that it is rational to assume that friction angle remains almost constant for minor stresses higher than 50 kPa. The peak friction angle for 75% density is calculated to be 43.3 at minor stress of 50 kPa and assumed to be constant in the model for stresses greater than that. Results of the numerical model indicated that for just a few elements confining stresses over 50 kPa was calculated during horizontal pullout. Dilation angle also changes with the variation of friction angle. Bolton's relationship as indicated in Equation [5-6] was used to update dilation angle for different confining stress levels. The computed lateral load versus displacement responses using the hyperbolic stress-strain model of test configuration (I) for different failure ratios (Rf) are shown in Figure 213 5-42. The results of Tests No. LN-1 and LN-2 are also overlain in the same figure. A comparison between Figure 5-41 and Figure 5-42 indicates that using a hyperbolic model with an appropriate failure ratio yields results that are in better agreement with results from full-scale tests that those obtained using a bilinear elastic model. As may be noted, the failure ratio (Rf) not only controls the slope of the load vs. displacement profile, but also controls the peak load as well. A value of Rf equal to 0.9 was found to produce the best fit between numerical results and the physical tests.. This value is in agreement with approximately calculated Rf value from triaxial tests results and also with suggested Rf value by Duncan and Chang (1970) for silica sand. | * 40 JS bt B V J 30 ' B S3 -—  ~™ —* — -—* —  § JSLL r — — Hyperbolic Model, Rf = 0.7 —- Hyperbolic Model, Rf = 0.8 -•- Hyperbolic Model, Rf = 0.9 — LN-1 — LN-2 i 1 1 1 ' i l 0 20 40 60 80 100 120 Displacement (mm) Figure 5-42: Numerical model result using hyperbolic Mohr-Coulomb model (with different failure ratios) versus test results (H/D=2.5, D=457 mm) 214 5.7.2.4 Modified hyperbolic model to account for density change during pulling of the pipe During pipe movement, it is reasonable to assume that there would be effects from density change in the soil mass. These variations in the soil density may be due to compression of sand in front of the pipe and dilation at the back of the pipe and heavily strained zones. Such variation of soil density could affect deformation moduli and the failure criterion (in terms of friction angle). It was judged that incorporating these variations into the hyperbolic model would further improve the ability of the analytical model to match the test results. A relationship that accounts for the dependency of friction angle to the soil density could be developed using the results of triaxial tests as shown in Figure 5-40. Having the friction angle of sand at two relative density levels and knowing that at loosest condition with 0% relative density the internal friction angle would be equal to cpcv, the relation in Equation [5-21] was developed to obtain friction angle at a given confining stress level. Combining this equation with Equation [5-20] results in Equation [5-22] which allows the friction angle to be determined as a function of relative density and minor principal stress. Accepting that the soil shear and elastic modulus have power law relations with soil relative density, the same approach was undertaken to characterize elastic properties as a function of soil density. Based on the triaxial test data, Equation [5-23] which account for the dependency of initial elastic modulus on the relative density and confining stress level could be obtained. <p(Dr, CT\ = 15 kPa) = <pcv + 0.16Dr [5-21] (p(Dr,cr\ ) = 0.002CT'3 2-0.2a'3 +36.6 + 0.16Dr [5-22] 215 Ei(Dr,cj3) = 2.9x(Dry,3xPax(^.) a [5-23] The computer model in Section 5.7.2.3 was updated to reflect the above relation of cp and Ej with density. The program was then executed for D r = 75%, representing the condition of physical modeling, and the results are presented in Figure 5-43. The results from numerical modeling without considering the effect of change in density (as per Section 5.7.2.3), and those from physical model testing are also superimposed. The computed soil loads are similar between the two models and in good agreement with physical measurements. In other words, accounting for the effect of changing density during pulling of the pipe did not seem to impact the measured lateral pullout performance in dense sand (D r = 75%). This was confirmed by executing a model for higher relative densities as well. 3 0 2 0 M C 0> " 15 B S Im 01 Cm "O M o -1 10 1 — Ignoring effect of density change Considering effect of density change Physical test results; Test No. LN-3 i i i 2 0 4 0 6 0 Displacement (mm) 8 0 100 Figure 5-43: Numerical model result using modified hyperbolic Mohr-Coulomb model to account for densification; D = 324 mm and H/D = 1.92, Density = 1600 kg/m 3 (D r= 75%) 216 Although no experiment was conducted on loose sands, the numerical models would provide an opportunity to investigate the effect of loose soil conditions. As such, the two models (i.e. with and without considering the effect of density change during pulling of the pipe) were reanalyzed considering a loose backfill with relative density of 20% (corresponding friction angle of 38°), and the results are given in Figure 5-44. Displacement (mm) Figure 5-44: Numerical model result using modified hyperbolic Mohr-Coulomb model to account for densification; D = 324 mm and H/D = 1.92, Density = 1450 kg/m 3 (D r= 20%) The model that considered the effect of soil density changes (during pipe displacement) yielded a higher lateral pulling load for D r = 20% than that of the model that did not consider this effect. The reasoning for this difference also becomes evident when contours of computed density and friction angles presented in Figure 5-45 are examined. As may be noted, the density and consequently friction angle in front of the pipe have increased significantly due to pulling of the pipe in loose sand; while at heavily sheared zones the density and friction angle seemed to be unaffected. These observations from the numerical model are in general accord with the observations made by Trautmann and 217 O'Rourke (1983) and Koskyukov (1967) regarding the effect of change in the soil density in front of the pipe. In this context, the second model that accounts for the effect of density variation is considered more appropriate to predict the response of the pipe soil interaction problem in general. Direction of pipe movement friction 3.40E+01 3.60E+01 180E+O1 : 4.00E+01 4.20E+01 IBS;' 4.40E+01 Density Direction of pipe movement • 38E+G3 4QE+03 42E+G3 44E+03 46E+03 48E+03 50E+03 52E+03 54E+03 56E+Q3 Figure 5-45: Density and friction angle contour for test in loose sand with initial density of 1450 kg/m3 (D r = 20%) after 30 mm of pipe displacement (H/D=1.92, D=324 mm) 5.7.3 Effect of size of the box Numerical modeling was also used to assess the impact of boundary conditions in the soil box during physical modeling of the horizontal pipe pulling. It should be noted that this section contributes directly to the development of the soil-pipe interaction physical modeling system presented in Chapter 3. Because of the intimate connection with numerical modeling, it was decided to present this section here. In order to assess the effect of front and rear wall boundaries as well as ground boundary of the soil box during pulling of the pipe, a numerical mesh configuration was developed to represent a box size of 7.6 m x 2.3 m with pipe outside diameter of 324 mm and overburden ratio of 2.75. This configuration represents a box with pipe distances from boundaries that are double of those dimensions in the real-life experiments. The results from numerical analysis of this configuration could then be directly compared with those 218 obtained from the numerical model that accurately represent the actual model. The predicted soil loads versus displacement curves for the two models analysed with identical soil properties, interface properties, etc., are presented in Figure 5-46. In addition to soil loads on pipe, the contours of minor principal stresses for the two models are also compared in Figure 5-47. As may be noted, the results for the small box size representing the test set up are in very good agreement with the model simulating a larger box size. It can be concluded that, for burial depths tested, the end and front walls in the physical model are located at an acceptable distance from the pipe location and that the results can be extended to field conditions. 50 40 S tt> 30 B cu J _ — _ it &r M i l • 11 - • - M o d e l o f 3.8 m x 1.6 m box -v- M o d e l o f 7.6 m x 2.3 m box i i i c -cu a. •a S3 o 20 10 20 40 60 80 100 D i s p l a c e m e n t ( m m ) 120 140 Figure 5-46: Soil loads on pipe in numerical modeling of box with different sizes 219 Figure 5-47: Minor principal stress contours after 30 mm of pulling of the pipe, effect of box size 5.8 Comparison between numerical model results and experimental results As pointed out earlier, it was decided to use the model described in Section 5.7.2.4 to represent soil behaviour. Besides accuracy and simplicity, the model also has the advantage of using soil parameters that are obtainable from common laboratory testing such as triaxial tests as done in this research. The selected model was used to predict soil loads for configuration of physical modeling. The calculated soil loads on the pipe from the numerical model are compared with measured loads on the pipe for different tests as shown in Figure 5-48 through Figure 5-51. It is observed that the numerical model slightly overpredicts soil loads on the pipe. However considering variability in soil parameters, errors associated with testing conditions, and limitations in numerical modeling, the results look reasonable. 220 60 50 » 40 JS EJD B 1> J 30 3 S-01 T3 O —1 20 10 1 f — * 1 1 , Numerical model results — LN-1 — LN-2 1 . 20 40 60 80 D i s p l a c e m e n t ( m m ) 100 120 Figure 5-48: Comparison between numerical and experimental results; Configuration with H/D = 1.92 and D = 457 mm (Tests No. LN-1 and LN-2) 30.0 • Numerical model results •LN-3 LN-4 •LN-7 0 10 20 30 40 50 60 70 80 90 100 D i s p l a c e m e n t ( m m ) Figure 5-49: Comparison between numerical and experimental results; Configuration with H/D = 1.92 and D = 324 mm (Tests No. LN3, LN-4, and LN-7) 221 12 10 Z M c « 6 3 L. u Q. •D a o — 1 1 1 Numerical model results LN-5 10 20 30 40 50 60 Displacement (mm) 70 80 90 100 Figure 5-50: Comparison between numerical and experimental results; Configuration with H/D = 1 and D = 324 mm (Test No. LN-5) 50 40 ID 30 a -1 20 4 <u a TJ O -1 10 --*- Numerical model results — LN-6 20 40 60 80 100 Displacement (mm) 120 140 Figure 5-51: Comparison between numerical and experimental results; Configuration with H/D = 2.75 and D = 324 mm (Test No. LN-6) 222 5.9 Effect of different variables on N q h value Full-scale physical modeling of buried pipelines provides reliable results for prediction of soil loads on pipe; however the modeling is costly and time consuming. Considering the large number of variables affecting soil behaviour and various geometrical configurations of buried pipes with regard to the burial depth and pipe size, it is not practical to conduct extensive full-scale physical modeling to explore this multiplicity of cases. Numerical models provide an alternate opportunity to investigate these aspects at relatively low cost, time, and effort requirements. With this background, it was decided to use the numerical models that have already been calibrated and validated in section 5.7.2 to examine horizontal bearing capacity (N qh), with respect to soil parameters, pipe-soil interface characteristics, and geometric variables such as pipe diameter. [Note: Dimensionless horizontal soil loads on the pipe (FL ' ) as described in Section 5.2, is also called horizontal bearing capacity (Nqh) by previous researchers (Hansen, 1961; O'Rourke, 1983; and Turner, 2004)] Considering the need to relate to the field conditions, and also limitations in the calibrated model, some restrictions were applied to the range of parameters investigated herein. For example, common oil and gas transmission buried pipelines have a diameter ranging from 20 cm to 100 cm and overburden depth of approximately 1 m. Previous studies (Audibert and Nyman, 1977; Trautmann and O'Rourke, 1983) have shown that the mechanism of failure is gradually changed from shallow to deep with increasing overburden ratio. Trautmann and O'Rourke (1983) have noted that deep failure mechanism will not take place for overburden ratios smaller than 5. Based on these considerations, in the current study, the overburden ratio was limited to 5 so that the investigation would represent shallow failure mechanisms. This ratio also represents common oil and gas buried pipeline conditions. 223 A s indicate in chapter 2, a family of curves have been proposed by Hansen (1961), Trautmann and O'Rourke (1983), and Turner (2004) for estimation of horizontal bearing capacity, N q h (see Figures 2-1, 2-3, and 2-5 respectively). These charts have been based on data from a limited number of tests pertaining to specific soil types and pipe configurations. In particular, these charts do not specifically account for the effect o f pipe diameter, pipe-soil interface friction, weight of pipeline contents, and dilatancy of surrounding soil. With this background, in this study, a series of numerical models were executed to examine the effects of the above mentioned parameters on N q h . The following approach was undertaken in this regard: (i) Using the numerical modeling configuration that was calibrated with experimental data is Section 5.7.2.4, a series of curves was developed to obtain the relations between N qh* vs. overburden ratio (H/D) and peak friction angle (cp) [Note: the computed horizontal bearing capacity in this study for a certain geometrical configuration and soil and interface properties is denoted by symbol N q h* for clarity]. This item is described in Section 5.9.1 (ii) The effect of pipe diameter (D), interface friction angle (8), pipe content weight (W), and soil dilatancy (\|/) were then investigated accordingly by variation o f these parameters in the above numerical model. The results for this part is described is Sections 5.9.2, 5.9.3, 5.9.4, and 5.9.5 respectively. (iii) The outcome of items (i) and (ii) would enable obtaining lateral bearing capacity for various combination of geometrical configuration, soil, pipe and interface parameters. 5.9.1 Variation of horizontal bearing capacity factor (Nqh) with overburden ratio and friction angle A family of curves were developed to show the effect of different burial depths and friction angle (cp) on N qh*. A s mentioned earlier, horizontal bearing capacity factor for a 224 specific geometrical, soil, pipe, and interface parameters is called Nqh* for clarity. The curves are developed for steel pipe with outside diameter of 324 mm, and wall thickness of 10 mm. Soil and interface parameters used in developing of presented curves are similar to those described in Sections 5.7.2.4 and 5.7.2.1 respectively. The numerical model was executed for 5 different overburden ratios of 1, 1.92, 2.75, 3.5, and 5 and at 3 different density levels which resulted in three different friction angles of 45°, 40°, and 35°. The computed values of N qh* are presented in Figure 5-52 in term of dimensionless peak load versus overburden ratio for different soil friction angles 9 = 45° cp = 40° 9 = 35° 9 = 45° — 9 = 40° 9 = 35° H/D Ratio Figure 5-52: Prediction of dimensionless soil loads for different friction angle and overburden ratios It is also of interest to compare the results obtained from numerical model results on (Nqh*) with previous studies. The results of N qh* are plotted in Figure 5-53 along with curves after Hansen (1964) (and as suggested in A S C E 1984, A L A 2001 and PRCI 2004) and those suggested by Turner (2004) [Note: presented curves by Trautmann and O'Rourke, 1983, exhibited close values to those suggested by Turner 2004; hence are not 225 presented here]. To avoid confusion, the graphs for only two friction angles are presented. Examining this figure indicates that the predicted loads by Hansen and Turner are about 200% and 20% more than those predicted from the results of current study. Slightly higher soil loads predicted from curves presented by Turner (2004) (and Trautmann and O'Rourke, 1983) can result in a conservative design; however the soil loads predicted by Hansen (1964) obviously overestimate soil loads and wil l result in an inefficient design. 40 35 30 25 Z ' 2 0 15 10 5 0 cp = 45°^ Hansen f (1964) - - " cp =4Q^ -_- • •r . _TT —' /(p -45^ -/(p = 40! —-~-0 = 45a J Turner ' (2004) L Current > Study (N^*) y — 3 H/D Figure 5-53: Variation of N q n value with overburden ratio; Comparison of the results of current study with previous research 5.9.2 Effect of pipe diameter Guo and Stolle (2005), numerically investigated the effect of pipe diameter on the soil resistance during transverse ground movement. Their numerical models (calibrated using data from a flexible 5.8 m long pipe with outside diameter of 203 mm) simulated pipe diameters ranging from 30 mm to 3300 mm and it was concluded that the effect of scaling on normalized soil loads is dependent on pipe diameter, but not overburden ratio. 226 They suggested Equation [5-24] to modify the horizontal soil loads on pipes with different diameters while other geometrical and soil parameters such as overburden ratio, friction angle, and soil density are constant. = 0.91(1 + 10S m ) [5-24] Where: FL ' - Dimensionless horizontal soil load on pipe FLO ' - Dimensionless horizontal soil load on pipe with reference pipe diameter D = external pipeline diameter Do - external reference pipeline diameter: 0.33 m L = buried pipeline length H ~ height of the soil on top of the springline of the pipe Ho - reference height of the soil on top of the springline of the pipe Since the numerical model in Section 5.7.2.4 has already been calibrated with respect to experimental data from full-scale testing of large diameter rigid pipes (D = 324 mm and 457 mm), it was considered appropriate to evaluate the suitability of the modification equation proposed by Guo and Stolle (2005). Such an analysis for large pipe sizes is prudent, knowing that the diameters of commonly used oil and gas transmission pipelines are generally greater than 200 mm. The soil loads on pipe for three different pipe sizes of 200mm, 324 mm and 1000 mm and for two overburden ratios of 1 and 5 was numerically modeled. The results of this analytical work were used to capture ( F L ' / F LO ' ) as per Equation [5-24] and the outcome is directly compared with those by Guo and Stolle (2005) in Figure 5-54. As may be noted, the results for the present study are in good agreement with those presented by Guo and Stolle (2005) for pipe diameter in the range of 200 mm to 1000 mm. 227 3 — Guo and Stolle (2005) * Current study (H / D = 1) * Current study (H / D = 5) i 0.1 1 10 D / D 0 Figure 5-54: The effect of pipe diameter; Comparison of numerical model results with proposed equation The adjustment in Equation [5-24] provides the opportunity to examine findings from the past research (for other pipe diameter) as shown in Figure 5-55. Several notable observations can be made from this figure: (i) The wide scatter observed without the correction for pipe diameter (see Figure 5-19) is now significantly reduced, thus making all experimental data for Nqh vs. H/D to fall within a relatively "narrow" band. (ii) The data from experimental work of this study is well within this narrow band of data (iii) The numerical model proposed in the recent study seems to be in reasonable agreement with most experimental results 228 A O • Audibert and Nymann, D=25mm, cp=40° (1977) •O • Audibert and Nymann, D=62mm, cp=40° (1977) •e— Trautmann & O'Rourke, D= 102mm, <p=46° (1983) Hsu, D=76mm, cp=42° (1992) o Hsu, D= 152 and 228mm, cp=42° (1992) H— Calvetti, D=20 and 50mm, cp=40° (2004) A Turner, D= 119mm, cp=41 to 47° (2004) • Current study, D=324mm, cp=43 to 46° • Current study, D=457mm, cp=43 to 46° Nqh*, (p=45°, Current study Nqh*, cp=40°, Current study Overburden Ration (H/D) Figure 5-55: Comparison of the results of current study (numerical and experimental) with previous research after correction for the effect of pipe diameter 229 5.9.3 Ef fec t o f sur face roughness The interface friction angle between pipe and dense sand was selected in the current study as 31° (see Section 5.7.2.1). This interface friction angle is about 0.7(p in which (p is internal friction angle of sand. Interface friction angle between sand and steel with different surface roughness varies in the range of 0.5cp to l.Ocp (ASCE 1984, Kalhawy et al., 1983). To investigate the effect of interface friction angle on horizontal soil loads on pipe, the numerical model was executed with identical material properties and different interface friction angles of 22.5° (0.5(p), 30° (0.7(p) and 45° (l.Ocp). The results are presented for configuration with overburden ratio of 1.92 and pipe diameter of 324 mm in Figure 5-56. The results indicated that doubling the interface friction angle (from 22.5° to 45.0°) results in less than 10% increase in lateral soil loads on the pipe. 30.0 25.0 4 5 20.0 ex c o Z 15.0 10.0 o —< 5.0 0.0 J ... * — -r 1 i -•- Interface friction angle (5) - 45.0° ~*~ Interface friction angle (6) = 30.0° -•- Interface friction angle (8) = 22.5° 10 20 30 40 50 60 Dimensionless Displacement 70 80 90 100 Figure 5-56: Numerical model results of test configuration (II) (H/D = 1.92 and D = 324 mm); Comparison of different surface roughness The same procedure was repeated for test configurations with H/D of 1 and 2.75 and the results are shown in Figure 5-57 and Figure 5-58 respectively. As may be noted, variation in the pipe surface roughness has a greater effect on lateral soil loads for lower 230 overburden ratios. Increase in surface roughness from 22.5° (0.5(p) to 45° (l.Ocp) results in increase in the soil loads on pipe of about 18% for overburden ratio of 1, while the same increase in surface roughness results in less than 5% increase in soil loads for H/D ratio of 2.75. 14 12 10 o ->< £\ ' — H ' ' -•- Interface friction angle (5) = 22.5° Interface friction angle (5) = 30.0° Interface friction angle (5) = 45.0° t 10 20 30 40 50 60 Displacement (mm) 70 80 90 10(1 Figure 5-57: Numerical model results of test configuration (III) (H/D = 1 and D = 324 mm); Comparison of different surface roughness To account for the effect of surface roughness, Equation [5-25] is presented to modify the Nqh* values. The values of F L ' / FLO ' as per Equation [5-25] is shown graphically vs. overburden ratio in Figure 5-59. The results of the numerical model were also overlain in the same figure to show suitability of the presented equation. Examining the figures confirms the suitability of the proposed equation. The trend of variation from Figure 5-59 also indicates that for overburden ratios higher than 5, the effect of surface roughness on the lateral soil loads on the pipe is negligible. 231 50 i Displacement (mm) Figure 5-58: Numerical model results of test configuration (IV) (H/D = 2.75 and D = mm); Comparison of different surface roughness —^- = l + ( — - l ) x [5-Where: FL ' = dimensionless horizontal load on pipe Fxo' = dimensionless horizontal load on pipe with reference interface friction angle d/<p = interface friction angle ratio between soil and pipe 8(/(p - reference interface friction angle ratio between soil and pipe (= 0.7) H/D = overburden ratio 232 1.2 1.1 §S 1 0.8 0.9 — Proposed Equation (5 = 22.5°) — Proposed Equation (8 = 45°) A Result of numerical model (8 = 22.5°) • Result of numerical model (8 = 45°) 0 1 2 3 4 5 H / D ratio Figure 5-59: Suitability of the proposed equation for modification the effect of surface roughness; Comparison between proposed equation and numerical model results 5.9.4 Effect of pipe and content weight on the soil loads The effect of pipe and content weight on the lateral soil loads on pipes was also investigated using the calibrated numerical. A number of cases with and without pipe content was examined for different pipe diameters (d) and overburden ratios (H/D). The results of maximum load and maximum dimensionless load for the different cases analysed are shown in Table 5-3. Examining this table indicates that the value of F L is also dependent on the weight of pipe and additional horizontal force is different for different pipe diameters. It can be seen that horizontal soil loads on the pipe may be affected by content weight as much as 10% in comparison with the case with no content and up to 20% in comparison with the case of weightless pipe with no content. The results indicate that the effect of pipe and content weight is more significant in low overburden ratios in term of percentage to the 233 total soil load on pipe. Examining this table also suggests that the additional force on the pipe due to the weight of pipe and content can be assumed independent from overburden ratio. Table 5-3: Variation of soil loads due to pipe and content weight D (cm) H/D Condition Change in weight of pipe and content (kg) A F L (kN) Calc. from numerical model A F L ( % ) Calc. from numerical model A F L (kN) Estimated from eq. [5-20] 32 1 Weightless pipe, no content -80 -0.9 - 9 % -1.0 32 1 1 cm steel wall thickness, no content 0 — — — 32 1 1 cm steel wall thickness, water filled 80 0.9 9% 1.0 32 5 Weightless pipe, no content -80 -1.1 1% -1.0 32 5 1 cm steel wall thickness, no content 0 0 — 32 5 1 cm steel wall thickness, water filled 80 1.1 1% 1.0 100 1 1 cm steel wall thickness, no content 0 0 — — 100 1 1 cm steel wall thickness, water filled 800 6.4 7% 7.1 100 5 1 cm steel wall thickness, no content 0 0 — — 100 5 1 cm steel wall thickness, water filled 800 8 1% 7.1 Based on this investigation, the empirical relation as in Equation [5-26] is suggested to calculate the effect of pipe and content weight on the lateral soil loads. The value of A F L from this equation could then be used to modify the soil loads on pipe obtained from values of N q n * as presented in Figure 5-52 to consider the effect of pipe and content weight (note that the curves in Figure 5-52 were developed for steel pipe with 1 cm wall thickness and no content). As shown in Table 5-3, the estimated change in the load on pipe is in good agreement with the numerical modelling results in the range of overburden ratio of 1 to 5. 234 AFL = 2 . 5 x ^ x t a n ( £ ) x ( l - 0 . 1 2 ^ ) Do p-26J Z ) 0 = 32cm Where AF= variation of soil loads from reference condition (steel pipe with 1cm wall thickness) W = variation of weight of the pipe and content from reference condition S = interface friction angle D = Pipe diameter To examine the validity of the proposed relationship, another model was developed with a 20 cm diameter pipe and overburden ratio of 3.0. The model was executed for three different pipe and content weight conditions. The first condition represented a weightless pipe with no content. In the second and third conditions, the model represented a steel pipe with 1 cm wall thickness. In the second case the pipe was empty, while for the last case a pipe filled with water was modelled. Peak load on pipe with 1 cm steel wall thickness and no content was calculated to be 19.7 kN/m. For weightless pipe, the calculated soil load was 19.1 k N / m and for the last case, in which the model represented a water filled pipe the soil load was 20.2 kN/m. Knowing the soil loads on steel pipe, with 1 cm wall thickness, Equation [5-26] suggests 20.1 kN/m and 19.1 k N / m lateral soil loads on pipe for weightless and water filled conditions respectively which are in agreement with the results from numerical modelling. 5.9.5 Effect of dilation angle The calibrated numerical model was also used to investigate the effect of soil dilatancy on horizontal soil loads on pipe. In the calibrated model, dilation angle was defined as a function of friction angle as suggested by Bolton (1986) (see Equation [5-7]). It is reasonable to assume that dilation angle decreases with decrease in friction angle. 235 Dilation angle in sand can be assumed to change in the range of 25° to 0° for highly compacted sand and loose sand respectively. The numerical model was employed to calculate the soil loads on pipe using sand with different dilation angles and friction angles. Two different models were developed to represent different overburden ratios (H/D) of 2 and 4. The calculated soil loads per unit length (FL ) for different cases are shown in Table 5-4. Table 5-4: Effect of constant volume friction angle on the soil loads on pipe Friction angle (cp) Dilation angle (i|/) H / D = 2 H / D = 4 F L (kN) A F L (%) (from case with lower dilation) F L (kN) A F L ( % ) (from case with lower dilation) 45° 15° 24.4 — 52.3 — 45° 25° 25.3 3% 53.5 2% 35° 5° 15 — 41.4 — 35° 15° 15.6 4% 42.3 2% Examining the results indicates that the effect of dilation for the range of overburden ratio as referred to in this study is negligible. This is in accord with suggestions by Rowe and Davis (1982) that the effect of soil dilatancy decreases with decrease in overburden ratio. The results of current study can also be compared with those suggested by Guo and Stolle (2005). As a part of their study on effect of different parameters on soil loads on pipe during transverse ground movement, they suggested Equation [5-27] to modify dimensionless soil loads for dilatancy effect. The equation is approximated from the results of numerical modeling, repeated for soils with different dilation angles. In their modeling, dilation angle in soil mass remained constant during pulling of the pipe. 236 = 1 + 0.23(1 + .24—). sin y/ [5-27] F ' D Where: FL ' - Dimensionless horizontal load on pipe FLo'= Dimensionless horizontal load on pipe buried in soil with zero dilation angle D = external diameter of the pipe H = height of the soil on top of the springline of the pipe yj - Dilation angle Based on this equation, variation in dilation angle of 10° results in approximately 5%, 6%, and 8% increase in soil loads for H/D of 1, 2, and 3 respectively. The higher effect from dilation on soil loads as observed by Guo and Stolle (2005) compared to current study can be associated with the assumption of constant dilation angle throughout pulling of the pipe and ignoring the effect on shear strain and density change in dilation angle; however the trend of variation of soil loads with the overburden ratio confirms that this effect is less for shallow buried pipes. 5.10 Some observations regarding modelling of pipe in trench The performance of pipe buried in trench configuration with hard native soil zones (see Figure 3-13 for definition of "native soil" and "backfill" zones) was also simulated using numerical modeling. 5.10.1 Modeling of pipe buried in dry sand "backfill" The case of pipe in trench with sand backfill as shown in Figure 5-60 was initially modelled. This essentially models the Test No. LT-1 of Section 5.5, where the test was conducted with dry Fraser River sand as backfill. The already available material 237 parameters and constitutive model used for modeling in Section 5.7.2.4 was employed to represent sand. Backfill-Trench Figure 5-60: Numerical model of pipe buried in trench configuration The interface parameters of pipe and soil were also kept same as in Section 5.7.2.1 while new interface parameters were used to describe the interface between backfill material and trench slope interface. Since the trench surface in the box in Test No. LT-1 was glued with the same sand as backfill material, it was judged reasonable to assume that interface friction is equal to the constant volume friction angle of sand. The results of load displacement response of numerical model are plotted in Figure 5-61 along with the results of Test No. LT-1 and average of Tests No. LN-1 and LN-2. The effect of trench surface on pipeline response observed in the physical model, as pipe approaches the trench, was not captured in the numerical model. This seems to be partly due to the limitation of mesh geometry. However, the initial part of the load displacement response is in good agreement with the physical model. 238 70 T 60 -Displacement (mm) Figure 5-61: Comparison of numerical model results and test results for pipe in trench 5.10.2 Modeling of pipe with moist sand backfill Tests No. LT-2 and LT-3 were performed using sand with 10% moisture content as backfill, and the results of these tests as presented in Figure 5-13 indicates that the load on the pipe is slightly lower than that when dry sand is being used as backfill. This finding initially seemed to be in contrast with what Turner (2004) observed during his studies. After a series of tests on buried pipes with different overburden ratios, he concluded that (a) the soil loads on pipes buried in moist sand with moisture content in the range of 4% to 8% is almost identical and (b) soil loads on pipes in moist sand with the range of moisture content mentioned above are almost twice as those observed when native soil is dry sand. Trench configuration in the current study might be the reason of getting lower soil loads on pipes than those predicted by Turner (2004). Turner also performed a series of direct shear tests on sand with different moisture content. He reported that for the same dry unit weight of the sand, friction angle of moist sand with 4 to 8% moisture content is about 3 to 5 degrees higher than that of dry sand. 239 Limited numerical modeling was attempted to obtain a preliminary understanding of soil response to pipeline movement when the backfill material is moist sand (Tests No. LT-2 and LT-3). The moist sand was modeled using material with apparent cohesion. A value for apparent cohesion can be calculated based on the measurements made on the height of free standing soil at the back of the pipe and friction angle of the soil. Considering classical earth pressure theory where the horizontal stress is zero, Equation [5-28] yields apparent cohesion of the material. Where h = height of free standing soil at the back of the pipe c' = apparent cohesion cp' = internal friction angle and y = total unit weight of the soil With height of cracks observed to be at least 1.1 m (see Figure 5-17 and Figure 5-18), internal friction angle estimated to be about 47° (based on the direct shear tests on dry sand and observations by Turner, 2004) and total density of about 1700 kg/m for backfill material, the apparent cohesion was estimated to be at least 4 kPa. The interface between trench surface and backfill material was still assumed to be frictional, with interface friction angle equal to constant friction angle of the sand. The peak load on the pipe for the first 100 mm of the pipe movement was calculated to be 49 kN/m which is in agreement with experimental results of Tests No. LT-2 and LT-3. Typical displacement vectors, obtained from this model are shown in Figure 5-62. The soil wedge in front of the pipe appears to move almost as a rigid block. Some local failure seems to occur around the pipe. N,= tan2 (45 + -<p') [5-28] Y 240 Figure 5-62: Displacement contours after 100 mm of pulling of the pipe, trench configuration with cohesive backfill material For comparison purposes, another numerical configuration was developed to model a hypothetical case with moist sand material in both "native soil" and 'backfill" zones. The geometric configuration was kept identical to that shown in Figure 5-60. The computed peak soil load on the pipe for this case, with the same material properties as explained above, was calculated to be 78 kN/m. The results show significantly higher values of soil loads on pipe where moist sand is used as backfill material. These results are in accord with those observed by Turner (2005) during his tests on moist sand. It is of interest to overlay the results for the latter cases as shown in Figure 5-63. From the comparison of the effect of apparent cohesion of backfill material on soil loads on buried pipes in configurations with and without a hard boundary, it can be concluded that in configurations where a trench exists, slippage occurs at trench surfaces and hence, the failure mechanism at the trench surface dominates soil loads on the pipe. In other words, if the failure mechanism at the interface is frictional, adding apparent cohesion to the backfill material does not increase soil loads on the pipe significantly. Alternatively, in configurations with no trench, backfill material with apparent cohesion exhibited dramatically higher resistance to the pipe displacement 241 8 0 7 0 / i 6 0 - / }.... 0* <p'=44° c = 0kPa O * <P' = 47° c = 4kPa / ! 7 .rf Mpdel of tests in native soil Model of tests in trench configuration 0 » / ^ 4 7 ° c = 0kPa c = 4kPa 2 0 4 0 6 0 Displacement (mm) 80 100 Figure 5-63: The effect of apparent cohesion on soil loads on pipe, comparison between trench configuration and case with no trench 5.11 Summary of the chapter The performance of buried steel pipes due to lateral soil loading was investigated through full-scale horizontal pulling tests, conducted using a large soil chamber at UBC. The pipeline configurations simulated in this research program included: pipe placed in dense dry sand and pipes buried in a trench excavated in dense cohesionless "native" soil, backfilled with dry or moist compacted sand. Application of a dual-layer geosynthetic fabric boundary to provide a preferential failure surface was also tested in both test configurations. In addition to the measurement of horizontal loads and pipe displacements, the soil pressure on the pipe surface was monitored using pressure transducers mounted at several circumferential locations on the pipe test specimen. These soil pressure measurements 242 were critical in improving the understanding of the load deformation response observed during horizontal pulling tests. Monitoring of the movement of geosynthetic fabrics lining trench surface, in combination with the observations on surface deformations and pipe position after pulling, provided useful information to understand the basic failure mechanisms and provide benchmarks for comparison with numerical modeling that was performed as part of the work. Some of the key findings and observations from the horizontal tests are summarized below: 1. The maximum horizontal load developed in the tests in uniform dry dense sand indicates that horizontal bearing capacity factors recommended by Trautmann and O'Rourke (1983) slightly overpredict soil loads the on pipe. The loads observed in the tests were far lower than what was computed using the horizontal bearing capacity factors from Hansen (1961). 2. The shape of the load displacement response observed during the tests showed a good agreement with the rectangular hyperbolic relation shown in Equation [5-1] which was suggested by Das and Seeley (1975) and adopted by Trautmann and O'Rourke (1983) 3. To the best of the author's knowledge, no previous attempts have been made to understand the soil pressure distribution on pipelines subject to relative ground movement. The soil pressure measurements on the pipe, in addition to conventionally measured horizontal load versus displacement response, also provides an independent set of data that would assist effective validation oL numerical models simulating this problem. 4. The concept of using dual layers of geosynthetic fabric to reduce soil loads due to horizontal ground deformation relies on a mechanism where slippage would preferentially occur at the geosynthetic fabric interface separating the trench backfill and native soil. The test results suggest that in addition to the interface frictional characteristics between the two geosynthetic fabric materials, two other 243 factors play a key role in governing the resistance pipe pullout: (i) the relative stiffness of the native soil (in which the trench is excavated) to the backfill material; and (ii) the "cohesive ability" of the trench backfill to promote soil movement in front of the pipe as a "wedge" or a block. 5. The results suggest that i f the native soil (in which the trench is excavated) is relatively stiff/hard, then the presence of the hard trench surface would be realized as the pipe moves towards this trench surface during horizontal ground movement. The resulting increase in soil pressures on the pipe due to this realization would leads to a significant increase in the resistance imparted by the soil on the pipe. However, this increase is still much less than the load that would occur if the backfill material and in situ material were the same. 6. For the trench conditions tested, the presence of geosynthetic fabric was helpful in reducing lateral soil loads on the pipe, although it did not provide as much benefit as would be expected based upon ideal slippage conditions at the trench interface. 7. Tests conducted using moist soil backfills suggested that backfills having some "cohesive ability" would cause soil in front of the pipe to move as a "wedge", in turn promoting the interface sliding along the geotextile interfaces in dual-geotextile-lined trenches. However, the reduction of soils loads due to this aspect could be significantly masked by the increase in soil loads due to the "hard boundary effect" explained in item 5. A numerical model was also developed as a part of this research to study the effect of different parameters on soil loads on pipe during transverse ground movement. Material properties were obtained directly from element tests on the soil and interfaces including direct shear tests and triaxial tests. A modified hyperbolic model was selected after examining a series of different constitutive models. The model was calibrated by comparing with the results of full scale testing. The model then was used to observe the effect of different parameters on the pipeline response to transverse ground movement. 244 In addition to modeling the configuration of pipe buried in native soil, a preliminary model was also developed to model the pipe buried in trench configuration and to capture the effect of stiff native soil on pipeline response. Some of the key findings and observations from numerical modeling are summarized below: 1. A modified hyperbolic model employed in a finite difference computer code (FLAC 4.0) was found to reasonably predict soil loads on pipe subjected to lateral movements in full-scale tests. 2. Using the validated model, a family of curves was developed, to predict horizontal bearing capacity (N q n) as a function of overburden ratio and peak friction angle similar to those previously suggested by other researchers (see Figure 5-52). In comparison with curves suggested by Hansen (1964) and Turner (2004), the curves presented here predict lower soil loads on buried pipes. 3. The effect of pipe diameter (D), interface friction angle (8), pipe and content weight (W), and dilation of soil (y) on the value of N q h was investigated, it was found: (a) The variation of Nqh with pipe diameter is similar to those observed by Guo and Stolle (2005). This effect results in higher dimensionless soil loads on pipe with smaller diameter. When applied to the previous tests on pipes with different diameters, the effect can explain the wide scatter in the results. (b) The weight of the pipe and content was determined to have a noticeable effect on lateral soil loads. A formula was introduced to provide a correction for this effect. (c) Surface roughness was also found to be a key parameter in identifying soil loads on pipe. The effect of surface roughness is a function of overburden ration and decreases with increase of burial 245 depth. For overburden ratios higher than 5, the effect can be neglected. (d) Executing numerical model using material properties with different dilation angles indicated that the effect of soil dilatancy is negligible for shallow buried pipes (overburden ratio lower than 5). This is in agreement with observations by Rowe and Davis (1982). The responses of pipe buried in hard "native soil" with both dry and moist sand as "backfill" material were adequately captured in numerical modeling. 246 CHAPTER 6 SUMMARY AND CONCLUSIONS The main objective of this thesis was to study the response of relatively large diameter buried steel pipes subject to lateral ground movements and the effectiveness of currently used methods to reduce soil loads on buried pipes. The work included the physical model testing and numerical analysis for the investigation, evaluation, and prediction of the response of buried pipelines under relative permanent ground movements. While significant research work has already been undertaken at a number of research institutions, the current fundamental understanding of the performance of buried pipelines when subjected to ground movements is still limited. In particular, most of the available experimental data have been derived using tests conducted on relatively small diameter pipelines. There is a need to account for the complexities in the mechanical behaviour of soils, and better understand the distribution of stresses around buried pipes in the presence of ground movements. With the above considerations in mind, a large soil testing chamber that is capable of subjecting relatively large diameter pipes to relative ground movements was developed, with provisions for measurement of pipe displacements, loads, and soil pressures imparted on pipe, localized soil shear zones, and ground surface movements. The testing was undertaken to model the effects of horizontal ground movements by subjecting the buried pipes to transverse (lateral) and/or longitudinal (axial) movements. Numerical models were developed to capture the results of full-scale testing during both axial and lateral pullout tests. A finite difference based software, F L A C 2D ®, version 247 4.0, was used in this study. Parameters for characterizing soil and soil-pipe interface response used in the numerical model were obtained from element testing of specific soil used in the physical modeling. After calibration with full-scale test results, the numerical models were used to investigate the effect of different parameters. In addition to providing fundamental insight into the pipe-soil interaction problem, the work also contributes to the design of buried pipelines located in landslide and earthquake-prone areas, and potentially improving some of the engineering guidelines used by the oil and gas industry. The key contribution can be summarized as follows: • Development of a laboratory facility, capable of testing large diameter pipes subjected to lateral and axial ground movements. This includes pipes buried in trench configuration with various native soil and backfill material conditions • Observation of soil normal stresses on the pipe during specimen preparation, in "at-rest" condition and during pullout test using pressure transducers mounted on the pipe surface • Development of numerical model analysis to assess the response of buried pipes to lateral and axial ground movements with respect to the pipe direction • Understanding the mechanism of axial pullout from basic soil-structure interaction perspectives with special consideration given to soil density and dilatancy effect as well as evaluation of practice guidelines • Evaluation of the current approaches for prediction of lateral soil loads on buried pipes using physical and numerical model results with special attention to the effect of pipe diameter • Evaluating the effectiveness of proposed methods for reduction of soil loads on buried pipes subjected to ground movements using geosynthetic layers 248 For the purpose of clarity, summary and conclusions (including contributions) have been presented as separate sections for each of the following main areas of work: (a) development of full-scale testing facility; (b) response to axial pullout; (c) response to lateral pulling. 6.1 Full-scale testing facility The testing facility included a 2.5 m height, 2.5 m width sand box with variable length of either 3.8 m or 5 m. The test chamber and the loading system were developed to enable testing of steel pipes up to 457 mm (18") in diameter. Two 418 kN (93 kips) actuators were employed to pull the buried pipe in a controlled rate and up to displacements in the order of 1 m. The position and alignment of actuators and soil box provided the opportunity to conduct both lateral and axial pullout tests. Steel pipes having outside diameters of 457 mm (18") and 324 mm (12.75") were used in the testing. The wall thickness of the two pipe sizes was 12.7 mm and 6.4 mm, respectively. The measurements during testing program included of load on each actuator, displacement of each end of the pipe, and displacement of geosynthetic layers, where applicable. Also for selected tests, soil pressures on the pipe surface were measured via pressure transducers. Observation of soil particle movements at soil-pipe interface during axial tests was also investigated by using colored sand in certain pre-defined zones at the interface. Locally available uniform dredged sand from the Fraser River (i.e., Fraser River sand) was used as the backfill material for pipe burial. Density of the material was controlled throughout the tests using a nuclear densitometer. The range of overburden ratios (H/D) for all tests were between 1 and 2.75 (where H = burial depth to pipe springline, and D = pipe diameter) to represent typical burial depths for oil and gas pipelines. The tests with trenches were conducted with trench slopes at 35° based on both applicability in the field and shape of failure surface. 249 With this effort, a world-class facility was developed at the University of BC for pipe-soil interaction testing particularly contributing to the design of oil and gas pipelines. The tests were conducted to simulate specific cases, as appropriate, with buried: (a) bare pipes in sand; (b) pipes wrapped with geosynthetics for soil load reduction; and (c) pipe in trench configurations with simulating different "backfill" and "native soil" zones, with and without geosynthetics for soil load reduction. Overall, the testing program consisted of 9 axial pullout tests and 14 lateral pulling tests conducted using the physical model testing facility. 6.2 Findings from pipeline response during axial pullout 6.2.1 Response of bare pipes The measured axial soil loads from full-scale tests performed on pipes buried in loose dry sand are comparable to those predicted using the equation (see Equation [2-2], Section 2.1.1) given in commonly used guidelines (ASCE 1984; A L A 2001; PRCI 2004). On the other hand, the peak axial resistance observed in pullout tests conducted on pipes buried in dense sand can be several fold (in excess of 2 times) higher than those predicted from the commonly used guidelines. The soil pressure measurements, made using pressure transducers mounted on the circumference of the pipe buried in dense sand, indicated that overall normal soil stresses on the pipe increased substantially during pullout tests in comparison to the initial stresses on the pipe prior to pullout. This increase in normal stress is believed to be associated with dilation of sand in the shear zone at the pipe-soil interface constrained by the surrounding "un-sheared" compacted soil. Observation of soil particle movements at the soil-pipe interface using colored sand zones indicated that the visible active shear zone is limited to a distance of about 2-mm from the surface of the pipe. This is in accord with previous research that suggests that the thickness of the active shear zone is about 10 times the mean particle size (dso), based on studies on direct shear tests. 250 A 2-D plane strain model was developed to capture the effect of dilation in the shear zone at the pipe soil interface, on the soil stress distribution around the pipe. In numerical model, the dilation of shear zone was simulated by expansion at the interface. The required amount of expansion at the interface in numerical model (which represents the amount of dilation at shear zone) is a function of various soil parameters such as density, stress level, and grain size distribution. For laboratory applications, the dilative displacement can be estimated by the normal displacement observed during direct shear tests. The computed normal stresses on the pipe after applying such expansion to the interface were in agreement with normal stress measurements during axial pullout of the pipe. This further supported the notion that constrained dilation of sand in the shear zone is responsible for the increase in overall stress on the pipe; it appears that the use of the coefficient of lateral earth pressure "at rest" (Ko) in Equation [2-2] (Section 2.1.1) as per ASCE (1984) is inappropriate for soils that are likely to dilate during interface shear and may underestimate the actual soil loads by several fold. The above numerical model, after calibrating with respect to the experimental normal stress measurements, was employed to compute the stress distribution on the pipe under different geometric conditions and soil parameters while accounting for dilation. It was determined that the use of a new parameter K (called equivalent lateral earth pressure coefficient) that is derived to represent an average of the normal stress distribution on the pipe is a more appropriate parameter than the "at rest" lateral earth pressure coefficient (Ko) as proposed by A S C E (1984). The model computations indicated that the effect of dilation is more significant on the smaller pipes, causing the K value to be dependent on the pipe diameters. Moreover, it was shown that the value of K is sensitive to the variation of soil elastic modulus and internal friction angle. Increase in both friction angle and elastic modulus resulted in increase in the value of K. The effects of other parameters (e.g. interface friction angle, soil dilation) were found to be negligible. Based on the above outcomes of numerical modeling of this study, a series of charts and formulas were developed to obtain the value of K under different geometrical conditions and soil material properties for potential consideration by practitioners for use in design. 251 6.2.2 Response of geosynthetic-wrapped pipes Al l three of the tested geosynthetic-wrapping schemes (spiral wrapping of two layers of geotextile, cigar wrapping of two layers of geotextile, and cigar wrapping of a layer of geotextile over a layer of geonet) were found to be effective in reducing the axial soil loads. Wrapping the pipe with geosynthetic layers would not only decrease the load by reducing the interface friction angle at the slippage surface, but also may minimize development of high normal stress by preventing the opportunity for dilation (as a result of avoiding shear strains in soil surrounding) at the geosynthetic-interface during shear deformations. The cigar-wrapped geotextile fabric configuration proved to be the most effective while the geosynthetic fabric over geonet was the least effective. 6.3 Findings from Pipeline response during lateral pulling 6.3.1 Response of pipe buried in dry sand The results of full-scale tests showed that the shape of load displacement response during lateral loading is in good agreement with rectangular hyperbola relation suggested by Das and Seeley (1975), and adopted by Trautmann and O'Rourke (1983). The maximum horizontal load mobilized on the pipe during pulling tests in uniform dry sand indicated that horizontal bearing capacity factors (N q n) recommended by Trautmann and O'Rourke (1983) and Turner (2005) slightly overpredict soil loads on pipe. On the other hand, the use of horizontal bearing capacity factors from Hansen (1961) significantly overpredicted (by about 3 fold) the measured horizontal peak loads (or horizontal bearing capacity factor, N q n ) from physical model testing. A numerical model using a modified hyperbolic model was developed to study the effect of different parameters on soil loads on the pipe during transverse ground movement. 252 The material parameters for modeling were obtained directly from element tests on the soil and interfaces. Both direct shear tests and triaxial element tests were performed to achieve soil and interface parameters. The model was calibrated with the results of full-scale testing. The calibrated model then was used in various configurations, using different material properties to capture the effect of geometric, material, and interface parameters on the lateral soil loads on buried pipes. Instead of the currently used single set of curves to obtain N q h as a function of H/D and cp (ASCE 1985, A L A 2001, PRCI 2005), a series of curves and/or formulae to reflect the effects of pipe diameter, surface roughness, and pipe and content weight were developed using the calibrated numerical model. It is also noted that due to the significant effect of pipe diameter on dimensionless soil loads on the pipe, it is not prudent to generalize the results of small scale tests to field conditions, unless the effect of pipe diameter is taken into account. The approach suggested by Guo and Stolle (2005) was found to work well for the data obtained in this study. 6.3.2 Response of pipe buried in trench configuration with hard "native soil" A series of tests was conducted to simulate a trench excavated in relatively stiff native soil, and bare pipe buried in the trench backfilled with sand. Such use of a sand-backfilled trapezoidal trench as a means to reduce horizontal soil resistance has been a standard approach in practice for some 30 years. The physical modeling of this study showed that the presence of the hard boundary with dry backfill material would result in an increase in horizontal resistance as the pipeline moved within the backfill and approached the hard boundary. The resulting increase in soil pressures on the pipe due to this realization of hard native soil (as the pipe approaches the hard boundary) is the reason for the observed significant increase in the lateral soil resistance on the pipe. In spite of this increase, the developed loads are still much lower than those that would have occurred if the backfill trench material was also as stiff as native soil. In an overall sense, the results from tests with a hard boundary confirm that if the pipe is buried in sand in a 253 suitably wide trapezoidal trench, so that it would maintain a reasonable horizontal distance from the hard boundary, it could provide an effective means to reduce the horizontal soil resistance against permanent ground movements. The results of tests using moist sand backfill, in trench excavated in hard "native soil", exhibited lateral soil loads comparable with those when dry sand was used. The numerical modeling undertaken indicated that the existence of a trench promoted a "rigid-block failure" which led to lower soil loads than those observed when no trench exists. In other words, when pipe is buried in trench configuration and backfilled with moist sand material, failure criterion at trench surface dominates the soil loads on the pipe and hence these loads are almost independent of the material cohesion. In configurations where no trench exists, cohesion of the material significantly increases soil resistance to the lateral movement of the pipe. It should be noted that only limited experimental work has been completed so far to study this configuration with moist sand backfill; therefore, further work will be required before more definitive conclusions can be derived. 6.3.3 Response of pipe buried in trench configuration with surface covered with dual-geotextile layers The use of dual-geotextile lined trench configurations to reduce soil loads from horizontal ground deformations relies on the formation of a preferential slippage surface at the geosynthetic fabric interface. The results of tests on pipes buried with a number of native soils conditions indicated that, in addition to the interface frictional characteristics between the two geosynthetic layers, the relative stiffness of the "native soil" zone in comparison to the backfill material and the ability of the trench backfill to move as a "cohesive block" become critical in promoting the desired slippage mechanisms (in turn, achieving the desired efficiency of reducing the lateral soil loads). For example, the use of dual-geotextile liners on the trench surface had no effect when the tests were conducted in configurations having both trench backfill and native soil 254 zones constructed of dense sand. The presence of geosynthetic fabric appeared to be helpful in reducing the lateral soil loads when the native soil is significantly harder than the trench backfill; in spite of this reduction, the loads are still higher than those estimated, assuming ideal slippage conditions at the trench interface. Using a simplified analytical model, it was possible to hypothesize that these relatively higher observed soil loads (than those expected under ideal slippage) can be attributed to local shearing around the pipe. Based on the present study, it is suggested that the maximum horizontal soil resistance be estimated: (a) using Equation [5-3] if there is a potential for the pipe to experience relative movement within the backfill; and (b) using Equation [5-2] if the pipe is expected to move in unison with the backfill and backfill moves almost as a rigid body along the trench surface. 6.4 Recommendations for future research It is recommended that the following additional work on this topic be undertaken to obtain a better fundamental understanding of the pipeline response to ground movement, development of more refined guidelines for pipeline design, and reduction of soil loads on pipelines: • Conduct more axial pullout tests to obtain more data on different configurations (i.e. different pipe size, overburden ratio, interface friction, and backfill material). In addition to the use of approximate 2-D numerical models developed in this study for further validation, consideration should be given to use of 3-D numerical modeling (e.g., F L A C 3D) to more accurately represent the axial pullout problem. • As mentioned in Section 6.2.1, evaluation of axial soil loads on the pipe requires obtaining a meaningful " K " value that reflects dilation in the shear zone around the pipe. The current study showed that this parameter, K, can be estimated knowing the amount of dilation at interface. In modeling the laboratory tests, the 255 dilation observed from direct shear tests can be reasonably used to estimate K value. However the estimation of dilation for in-situ conditions is more complex; more study, perhaps using direct shear tests involving different soil materials and states would be required to extend the numerically developed charts to general field problems. In the present study, the numerical model to predict lateral soil loads on pipes was calibrated with results of full-scale tests on various pipe diameters and overburden ratios, but only with pipes buried in backfills having a relative soil density (D r) in the order of 75%. Conducting lateral pulling tests on pipe buried in loose material (i.e., friction angle, different dilatancy etc.) would provide additional support to increase the current understanding. For pipes buried in a trench configuration, the angle of the trench surface and distance of pipe from trench are believed to be key parameters in determining mobilized soil loads on pipes during transverse' ground movement. A series of tests on pipes buried in different trench configurations is recommended for a better understanding of pipeline response. The hard trench boundary resulted in increased horizontal lateral soil resistance once the pipeline displaced a sufficient amount to be in close proximity to the boundary. Presumably, the pipe displacement at which the hard boundary effects begin to occur can be reduced by increasing the initial separation distance between the pipe and the boundary. Tests are necessary to confirm this behavior and determine how the required separation distance varies with the amount of relative displacement imposed on the pipeline. Tests performed by Turner (2004) on pipes buried in dense sand raise questions regarding the reasonableness of recommendations on horizontal soil resistance based upon tests in dry sand. The results from tests performed in this study on pipes buried in moist sand in trench configurations with hard "native soil" zone 256 were not in accord with findings by Turner (2005). This disagreement in the results may be due to: (a) the difference in the range of tested H/D ratio; (b) inconsistencies in the characterization of backfill material; or (c) the presence of a hard trench boundary. The developed numerical model suggested that item (c) is the reason of this disagreement; however, to completely address this concern, tests should be performed on moist sand, but without the hard trench boundary, along with appropriate testing conducted to obtain the mechanical characteristics of the test sand. The present investigations indicated that a simple mechanical analog is not able to estimate the horizontal soil resistance for the case of a hard trench boundary lined with two layers of geosynthetic fabric. It is hypothesized that the relative movement of the pipe within the backfill, is the main reason for this inability. This hypothesis can be evaluated with tests that employ alternate backfill materials. At one extreme, the backfill could consist of relatively high strength material that would prevent any relative pipe movement; in other words, use a backfill that promotes movement of backfill within the trench as a "block" in unison with the pipe. If this extreme condition leads to a confirmation that the resistance is estimated by the simple mechanical analog, additional tests with backfill materials representing various degrees of strength and cohesion, would lead to more specific guidance on backfill requirements. The numerical model in this study allowed for large strains and pipe displacement up to 100 mm. This displacement of the pipe seemed to be sufficient to obtain peak loads during lateral pulling of the pipe buried in native soil. However, the model was unable to capture the increase in soil loads due to realization of trench surface while pipe moves forward. To capture this effect through numerical models, substitute computer programs are required. This is especially important if alternate trench configurations were to be tested with the pipe at greater distances from rigid trench wall. 257 REFERENCES Akinmusuru, J. O., (1978). "Horizontally loaded vertical plate anchors in sand." Journal of Geotechnical Engineering Division, 104(GT2), 283-286. Akinmusuru, J. O., (1979). Closure to: "Horizontally loaded vertical plate anchors in sand." Journal of Geotechnical Engineering Division, 105(GT11), 1370-1371. American Lifeline Alliance, (2001). "Guidelines for the design of buried steel pipe." <http://vv^vw.americanlifelinesalliance.org/pdf/Update061305.pdf>. Anderson, C. (2004) "Response of buried polyethylene natural gas pipelines subjected to lateral ground displacement." M.A.Sc. Thesis, University of British Columbia, Vancouver, BC, Canada. Anderson, C , Wijewickreme, D., and Mitchell, A . (2003) "Development of a Full-Scale th Laboratory Testing Facility for Soil-Pipeline Interaction Research." Proceedings, 56 Canadian Geotechnical Conference, Winnipeg, Sept. 2003. Anderson, C , Wijewickreme, D., Ventura, C , and Mitchell, A . (2004) "Full-scale laboratory testing of soil-pipe interaction in branched polyethylene pipelines." Experimental Techniques, 29(2), 33-37. ASCE, (1984) "Guidelines for the seismic design of oil and gas pipeline systems." Committee onGas and Liquid Fuel Lifelines, ASCE, New York. 258 Audibert, J. M. E., and Nyman, K. J. (1975) "Coefficient of subgrade reaction for the design of buried piping." Proceedings, 2 n d Specialty Conference on Structural Design of Nuclear Plant Facilities, A S C E , Vol . IA, 109-141. Audibert, J. M. E., and Nyman, K. J. (1977) "Soil restraint against horizontal motion of pipes." Journal of the Geotechnical Engineering Division, A S C E , 103(GT10), 1119-1142. Baumgard, A. J. (2000) "Monotonic and cyclic responces to upheaval buckling offshore buried pipelines." Ph.D. Thesis, University of Cambridge, Churchill College, Cambridge, U.K. Bishop, A. W. (1966) "Sixth Rankine lecture: Stength of soils as engineering materials." Geotechnique, 16, 89-130. Bolton, M. D. (1986) "The strength and dilatancy of sands." Geotechnique, 36(1), 65-78. Brachman, R. W. I., Moore, I. D., and Rowe, R. W. (2000) "The design of a laboratory facility for evaluation the structural response of small-diameter buried pipes." Canadian Geotechnical Journal, 37, 281-295. Bridgewater, J. (1980) "On the width of failure zones." Geotechnique, 30(4), 533-536. Brooker, E. W., and Ireland, H. O. (1965) "Earth pressure at rest related to stress history." Canadian Geotechnical Journal, 2, 1-15. Brumund, W. F., and Leonards, G. A. (1973) "Experimental study of static and dynamic friction between sand and typical construction materials." Journal of Testing and Evaluation, 1(2), 162-165. 259 Byrne, P. M., Cheung, H., and Yan, L. (1987) "Soil parameters for deformation analysis of sand masses." Canadian Geotechnical Journal, 24(3), 366-376. Byrne, M. R., Roy, D., Campanella, G., and Hughes J. (1995) "Predicting liquefaction response of granular soils from pressuremeter tests." Geotechnical Special Publication, 56, 122-135. Calvetti, F., Prisco, C , and Nova, R. (2004) "Experimental and numerical analysis of soil-pipe interaction." Journal of Geotechnical and Geoenvironmental Engineering, ASCE, 130(2), 1292-1299. Capelletto, A., Tagliaferri, R., Giuriani, G., Andrei, G., Furiani, G., and Scarpelli, G. (1998) "Field full scale tests on longitudinal pipeline-soil interaction." Proceedings, International Pipeline Conference, Calgary, A B , Canada, 2, 771-778. Carder, D. R., Pocock, R. J., and Murray, R. T. (1977) "Experimental retaining wall facility-Lateral stress measurements with sand backfill." Laboratory Report 766. Transport and Road Research Laboratory, Crowthorne, Berkshire, U.K. C-CORE and Honegger, D. (2003) "Extended model of pipe soil interaction." PRCI Report, Catalogue No. L51990, PRCI, 340-3801 Kirby Dr., Houston, Texas, USA. Chadwick, P. (1952) "The quasi-static expansion of a spherical cavity in metals and ideal soil." Journal of Mechanics and Applied Mathematics, Part 1, XII, 52-71. Clayton, C. R. I., and Symons, I. F. (1992) "Pressure of compacted fill on retaining walls." Geotechnique, 42(1), 127-130. Danish Hydraulic Institute (DHI) and Ramboll and Hannemann (1985) "Danish Submarine Guidelines." Danish Energy Agency. 260 Das, B. M., Seeley, G. R. (1975) "Load displacement relationship for vertical anchor plates." Journal of Geotechnical Engineering Division, 101(GT7), 711-715. De Beer, E.E. (1970) "Experimental determination of shape factors and the bearing capacity of sands." Geotechnique, 20(4), pp 387-411. De Nicola, A., and Randolph, M. F. (1999) "Centrifuge modeling of pipe piles in sand under axial loads." Geotechnique, 49(3), 295-318. Dickin, E. A., and Leung, C. F. (1979). Discussion on: "Horizontally loaded vertical plate anchors in sand." Journal of Geotechnical Engineering Division, 105(GT3), 442-443. Duncan, J. M., and Chang, C. Y. (1970) "Nonlinear analysis of stress and strain in soils." 96(SM5), 1629-1653. Duncan, J. M., and Seed, R. B. (1986) "Compaction-induced earth pressures under Ko-Conditions." Journal of Geotechnical Engineering, 112(1), 1-22. Duncan, J. M., Williams, G. W., Sehn, A. L., and Seed, R. B. (1991) "Estimation earth pressures due to compaction." Journal of Geotechnical Engineering, 117(12), 1833-1847. Edil, T. B., and Dhowian, A. W., (1981) "At-rest lateral pressure of peat soils." Journal of Geotechnical Engineering Division, 107(GT2), 201-217. Fannin, R. J., Eliadorani, A., and Wilkinson, M. T. (2005) "Shear strength of cohesionless soils at low stress." Geotechnique, 55(6), 467-478. Feda, J. , Bohac, J . , and Herle, I. (1995) "Ko-compression of reconstituted loess and sand with stress perturbation." Soils and Foundations, 35(3), 97-104. 261 Filz, G. M., and Duncan, M. J. (1996) "Earth pressures due to compaction: comparison of theory with laboratory and field behaviour." Transportation Research Records, 1526, 28-37. Foray, P., Balachowski L., and Colliat, J. L. (1998) "Bearing capacity of model piles driven into dense overconsolidated sand" Canadian Geotechnical Journal, 35(2), pp 374-385. Foray, P. Y. , Colliat, J. L., and Nauroy, J. F. (1993) "Bearing capacity of driven model piles in dense sands from calibration chamber tests." Proceedings, Offshore Technology Conference, Houston, Texas, USA, Vol . 2, 655-666. Fukagawa, R., and Ohta, H. (1988) "Effect of some factors on Ko-value of a sand." Soils and Foundations, 28(4), 93-106. Garrison, R. E., Luternauer, J. L., Gril l, E. V., MacDonald, R. D., and Murray, J. W. (1969) "Early diagenetic cementation of recent sands, Fraser River Delta, British Columbia." Sedimentology, 12, 27-46. Ghionna, V., Jamiolkowski, M., and Pasqualini, E. (1981). Discussion on: "On estimating Ko for overconsolidated granular soils." Geotechnique, 31(4), 474-577. Guo, P. J. (2005). "Numerical modeling of pipe-soil interaction under oblique loading." Journal of Geotechnical and Geoenvironmental Engineering, 131(2), 260-268. Guo, P. J., and Stolle, D. F. E. (2005) "Lateral pipe-soil interaction in sand with reference to scale effect." Journal of Geotechnical and Geoenvironmental Engineering, 131(3), 338-349. Hendron, A. J . , Jr. (1963) "The behavior of sand in one-dimensional compression." Ph.D. Thesis, University of Illinois, atUrbana, Champaign. Illinois, USA. 262 Hansen J. B. (1953) "Earth pressure calculation." Danish Technical Press, Copenhagen., Denmark. Hansen, B.J. (1961) "The ultimate resistance of rigid piles against transversal forces." Bulletin 12, Danish Geotechnical Institute, Copenhagen, Denmark. Holloway, D. M. (1976) "Mechanics of pipe-soil interaction in cohesionless soil." Ph.D. Thesis, Duke University, Durham, North Carolina, USA. Honegger, D. G. (1999) "Field measurements of axial soil friction forces on buried pipelines." Technical Council on Lifeline Earthquake Engineering Monograph, 16, 703-710. Honegger, D. G., and Nyman D. J. (2004) "Guidelines for the Seismic Design and Assessment of Natural Gas and Liquid Hydrocarbon Pipelines." Pipeline Research Council International, Inc. (PRCI), Catalogue No. L51927. Hsu, T. W. (1994) "Rate effect on lateral soil restraint of pipeline." Soils and Foundations, 33(4), 159-169. Hsu, T. W., Chen, Y . J., and Hung, W. C. (1996) "Soil resistant to oblique movement of buried pipes in dense sand." Journal of Transportation Engineering, 132(2), 175-181. Hsu, T. W., Chen, Y . J., and Wy, C. Y. (2001) "Soil resistant to oblique movement of buried pipes in dense sand." Journal of Transportation Engineering, 127(1), 82-87. Jaky J. (1944) "The coefficient of earth pressure at rest." Journal of Society of Hungarian Engineers and Architectures, 355-358. 263 Jardine, R. J., and Overy, R. F. (1996) "Axial capacity of offshore piles driven in dense sand." Proceedings, Offshore International Conference, Houston, Texas, USA., 161-180. Kennedy, R. P., Chow, A. W., and Williamson, R. A. (1977) "Fault movement effects on buried oil pipelines." Transportation Engineering Journal, 103(TE5), 617-633. Konuk, I., Fredj, A., Yu, S. (2005) "3-Dimensional bifurcations of pipe-in-pipe structures." Proceedings, 24 t h International Conference on Offshore Mechanics and Arctic Engineering, O M A E , 3, 747-753. Kostyukov, V. D. (1967) "Distribution of the density of sand in the sliding wedge ion front of anchor plates." Soil Mechanics and Foundation Engineering, 1, 12-13 Kraft, L. M. (1990) "Computing axial pipe capacity in sands for offshore conditions." Geotechnique, 9, 61-92. Kraft, L. M. (1991). "Performance of axially loaded pipe piles in sand." Journal of Geotechnical Engineering, 107(2), 272-296. Kulhawy, F. H., and Peterson, M. S. (1979) "Behaviour of sand-concrete interface." Proceedings, 6 t h Pan-American Conference on Soil Mechanics and Engineering, Lima, Peru, 2, 225-236. Kulhawy, F. H., Trautmann, C. H., Beech, J. F., O'Rourke, T. D., and McGuire, W. (1983) "Transmission line structure foundations for uplift-compression loading." Report No. EL-2870, Electric Power Research Institute. Ladd, R. S. (1974) "Specimen preparation and liquefaction of sands." Proceedings, ASCE Geotechnical Division, 100(10), 1180-1184. 264 Lambe, T. W. (1991) "Soil testing for engineers." Bitech Publishers Ltd., Vancouver, BC, Canada. Leathers, F. D., and Ladd, C. C. (1987) "Behaviour of an embankment on New York Varved clay." Canadian Geotechnical Journal, 15(2), 250-268 Lee, K. L. (1970) "Comparison between plane strain and triaxial tests on sand." Journal of the Soil Mechanics and Foundation Division, 1, 12-13. Lehane, B. M., and Jardine, R. J. (1994) "Displacement-pile behaviour in a soft marine clay." Canadian Geotechnical Journal, 31(1), 79-90. Lehane, B. M., Jardine, R. J. , Bond, A. J., and Frank, R. (1993) "Mechanisms of shaft friction in sand from instrumented pile tests." Journal of Geotechnical Engineering, 119(1), 19-35. Leroueil, S., and Flight, D. W. (2003) "Behaviour and properties of natural soils and soft rocks." In Characterization and Engineering Properties of Natural Soils. Edited by T. S. Tan, K. K. Phoon, D. W. Hight, and S. Leroueil., A. A. Balkema, Rotterdam, The Netherlands., 1,29-254. Liu, B., Moffitt, K., Nixon, J. F., Zhou, J., and Xiao Y. (2004) "Numerical studies of pipeline uplift resistance in frozen ground." Proceedings, International Pipeline Conference, Calgary, A B , Canada. Liu, J. X . (2004) "Design guide developed for buried pipelines crossing active faults." Oil and Gas Journal, 102(26), 58-65. Liu, S. H. (2006) "Simulating a direct shear box test by D E M . " Canadian Geotechnical Journal, 43, 155-168. 265 Mayne, P. W., and Kulhawy, F. H. (1982) " K 0 - O C R relationships in soil." Proceedings, ASCE Geotechnical Division, 108(6), 851-872. McAllister, E. W. (2001) "Pipeline rules of Thumb handbook." Gulf Professional Publishing, 648 p. Mesri, G., and Hayat, T. M. (1993) "The coefficient of earth pressure at rest." Canadian Geotechnical Journal, 30(4), 647-666. Meyehof, G. G. (1951) "The ultimate bearing capacity of foundations." Geotechnique, 2(4), 301-332. Michalowski, R. L. (2005) "Coefficient of earth pressure at rest." Journal of Geotechnical and Geoenvironmental Engineering, 131(11), 1429-1433. Motan, E. S., and Jacot, B. R. (1987) "Lateral response and earth pressure parameters of cohesionless soils related to flat dilatometer data: a laboratory study." Transportation Research Records, 1119, 98-104. Murray, E. J., and Geddes, J. D. (1989). "Resistance of passive inclined anchors in cohesionless medium." Geotechnique, 39(3), 417-431. Nauroy, J. F., and Le Tirant, P. (1983) "Model tests of piles in calcareous sands." Proceedings, Conference on Geotechnical Practice in Offshore Engineering, Austin, Texas, 356-369 Neely, W. J., and Stuart, J. G., and Graham, J. (1973) " Failure loads on vertical anchor plates in sand." Journal of the Soil Mechanics and Foundation Division., 99(SM9), 669-685. 266 Newmark, M., Hall, W. J., (1975) "Pipeline design to resist large fault displacement." VDI Forschungsheft, 1975, p 416-425. Nobahar, A., and Popescu, R. (2001) "Paramter calibration of strain hardening/softening of sand from direct shear tests." Proceedings, 10 t h International Conference on Computer Methods and Advances in Geomechanics, Tucson, Arizona, USA. Nova Gas Transmission (1995) "Frictionless pipe wrap test results-ASC pipe/soil interaction test centre." Technical Report, July 18 1995, Nova Gas Transmission Ltd., Calgary, A B , Canada. Oda, M. (1972) "Initial fabric and their relations to mechanical properties of granular material." Soils and Foundations, 12(1), 17-36. O'Rourke, M. J., Liu, X . , and Flores-Berrones, R. (1995) "Steel pipe wrinkling due to longitudinal permanent ground deformation." Journal of Transportation Engineering, 121(5) O'Rourke, M., Liu, J., (1998) "Seismic loading and behaviour of buried pipelines." Pressure Vessel and Piping Codes and Standards, 360, 513-519. O'Rourke, M., Nordberg, C. (1992) "Behavior of buried pipelines subject to permanent ground deformation." Proceedings of the World Conference on Earthquake Engineering. Madrid, Spain. O'Rourke, T. D. (1988) "Critical aspects of soil-pipeline interaction for large ground deformation." Proceedings, 1 s t Japan-US Workshop on Liquefaction, Large Ground Deformation and Their Effects on Lifelines, Association for Development of Earthquake Prediction, Japan, 118-126. 267 O'Rourke, T. D., and O'Rourke, M. J. (1995) "Pipeline response to permanent ground deformation: A benchmark case." Proceedings, 4tg US Conference on Lifeline Earthquake Engineering, TCLEF, ASCE, San Francisco, CA , Aug 1995, 288-295. O'Rourke, T. D., and Palmer, M. C. (1996) "Earthquake performance of gas transmission pipelines." Earthquake Spectra, 20(3), 493-527. O'Rourke, T. D., and Tawfik, M. (1983) "Effects of lateral spreading on buried pipelines during the 1971 San Fernando Earthquake." Earthquake Behaviour and Safety of Oil and Gas Storage Facilities, Buried Pipelines and Equipments, PVP, 77, A S M E , N.Y., 124-132. Ovesen N. K. (1964) "Anchor slabs, calculation method and model tests." Bulletin 16, Danish Geotechnical Institute, Copenhagen. Ovesen, N. K., and Stromann, H. (1972) "Design method for vertical anchor slabs in sand." Proceedings, Specialty Conference on Performance of Earth and Earth-Supported Structures, Purdue University, Indiana,Vol. I, Part 2, 1481-1500. Palmeira, E. M., and Milligan, G. W. E. (1989) "Scale effects in direct shear tests on sand." Proceedings, 12 t h international conference on soil mechanics and foundation engineering. 1(1) 739-742. Paulin, M. J., Phillips, R., Clark, J. I., Hurley, S., Trigg, A. (1997) "Establishment of a ful-scale pipeline/soil interaction test facility and results from lateral and axial investigations in sand." Proceedings, 16 t h International Conference of Offshore Mechanics and Arctic Engineering, OMAE, Vol. 5, Pipeline Technology, 139-146. Paulin, M.J., Phillips, R., Clark, J.L, Trigg, A., and Konuk, I. (1998) " A full-scale investigation into pipeline/soil interaction" Proceedings, International Pipeline Conference, Calgary, A B , A S M E , 779-788. , 268 Petroff, L., (1990) "Review of the relationship between internal shear resistance and arching in plastic pipe installations." A S T M Special Technical Publication, 1093, 266-280. Phillips, R., Nobahar, A., and Zhou, J. (2004) "Trench effects on pipe-soil interaction." Proceedings, International Pipeline Conference, Calgary, A B , Canada. Phillips, R., Nobahar, A., and Zhou, J. (2004) "Combined Axial and lateral pipe-soil interaction relationships." Proceedings, International Pipeline Conference, Calgary, A B , Canada. Popescu, R., Phillips, R., Konuk, I., and Deacu D. (1999) "Numerical and physical modelling of pipe-soil interaction." Proceedings, 52 n d Canadian Geotechnical Conference, Regina, SK, pp. 437-444. Popescu, R., Phillips, R., Konuk, I., Guo, P., and Nonahar, A. (2002) "Pipe-soil interaction: Large scale tests and numerical modeling." Proceedings, International Conference on Physical Modeling in Geotechnics, St. John's, NF, Canada. Poulos H. G. (1978) "Normalized deformation parameters for kaolin." A S T M Geotechnical Testing Journal, 1(2), 102-106. Randolph, M. F., Dolwin, J., and Beck, R. (1994) "Design of driven piles in sand." Geotechnique, 44(3), 427-448. Roscoe, K. H. (1970). "10 t h Rankine lecture: The influence of strains in soil mechanics." Geotechnique, 20(2), 129-170. Rowe, P. W. (1969) "The relation between the shear strength of sands in triaxial compression, plane strain and direct shear." Geotechnique, 19(1), 75-86. 269 Rowe, R. K., and Davis, E.H. (1982) "The behaviour of anchor plates in sand." Geotechnique, 32(1), 25-41. Saglamer, A. (1975) "Soil parameters affecting coefficient of earth pressure at rest of cohesionless soil." Proceedings, Conf. on Soil Mechanics and Foundation Engineering, Istanbul, Turkey, 9-16. Scarnio, J. (2002). Discussion on: "Soil resistant of oblique pipelines in loose sand." Journal of Transportation Engineering, 128(2), 198-200. Scarpelli, G., and Wood, D. M. (1982) "Experimental observations of shear band patterns in direct shear tests." I U T A M Conference on Deformation and Failure of Granular Materials. Schmidt, B. (1966). Discussion on: "Earth pressure at rest related to stress history." Canadian Geotechnical Journal, 3(4), 239-242. Sexena, S., Hedberg, J., and Ladd, C. C. (1978) "Geotechnical properties of Hackensack Valley Varved clays of N. J . " A S T M Geotechnical Testing Journal, 1(3), 148-161. Singhal, A. C. (1980) "Experiments of pipeline joints." American Society of Mechanical Engineers, n 80-C2/PVP-70, 5p. Sivathayalan, S. (2000) "Fabric, initial state and stress path effects on liquefaction susceptibility of sand," Ph.D Thesis, University of British Columbia, Vancouver, B.C. Sherif, M. A., and Koch, D. E. (1970) "Ko as related to soil recompression ratio." Highway Research Record, National Academy of Science, Washington, D . C , 323, 39-48. 270 Smith J. E. (1962) "Deadman anchorages in sand." Technical Report R199, U.S. Naval Civil Engineering Laboratory, Port Hueneme, CA. Sokolovskii, V. V. (1965) "Statics of granular material." Pergamon Presee, Oxford, 270 P-Texas Research International Company (2000). "Test results for interface friction testing-pipeline project." TRI/Environmental, Inc., June 2000, TRI log #: E2128-22-01. Trautmann C.H., and O'Rourke, T.D. (1983) "Behaviour of pipe in dry sand under lateral and uplift loading." Geotechnical Engineering Report 83-7, Cornell University, Ithaca, N.Y. Trautmann, C.H., and O'Rourke, T.D. (1985) "Lateral force displacement response of buried pipe," Journal of Geotechnical Engineering, ASCE, 111(9), 1077-1092. Tuncer, B. E., and Dhowian, A. W. (1981). "At rest lateral pressure of peat soils." Proceedings, A S C E Geotechnical Division, 107(2), 201-217. Turner, J. E. (2004) "Lateral force displacement behaviour of pipes in partially saturated sand." M.A.Sc. Thesis, Cornell University, Ithaca, N.Y. Uthayakumar, M. (1996) "Liquefaction of sands under multiaxial loading." Ph.D Thesis, University of British Columbia, Vancouver, B.C. Vaid, Y. P., Byrne, P. M., and Hughes, J. M. O. (1980) "Dilation angle and liquefaction potential." Proceedings, International conference on Recent Advances in Geotechnical Earthquake Engineering and Soil Dynamics, St. Louis, MO., USA. Vesic, A. S. (1972) "Expansion of cavities in infinite soil mass." Journal of Soil Mechanics, Foundation Division, A S C E 98, SM3, 265-290. 271 Vesic, A. S., and Claugh, G. W., "Behaviour of granular materials under high stress." J. Soil Mech. Found. Div., A S C E , 94 (SM3), 661-688. Wang, M. C , and Wu, A. H. (1980) "Yielding load of anchor in sand." Proceedings, Symposium of Application of Plasticity and Generalized Stress Strain in Geotechnical Engineering, ASCE, N Y , 291-307. Wijewickreme, D., Sriskandakumar S., and Byrne P. M. (2005) "Cyclic loading response of loose air-pluviated Fraser River sand for validation of Numerical Models Simulating Centrifuge Tests." Canadian Geotechnical Journal, 42(2), 550-561. Wijewickreme, D., Karimian, H., Honegger, D. G. (2005) "Effectiveness of some methods for reducing axial soil loads on buried pipelines subjected to ground movements." Proceedings, 58 t h Canadian Geotechnical Conference, September 2005, Saskatoon, Canada. Yamaguchi, H., Kimura, T., and Fuji-I, N. (1976) "On the influence of progressive failure on the bearing capacity of shallow foundations in dense sand." Soils and Foundations, 16(4), 11-22. Yimsiri, S., Soga, K., Yoshizaki, K., Dasari, G. R., and O'Rourke, T. D. (2004). "Lateral and uplift soil-pipeline interactions in sand for deep embedment conditions." Journal of Geotechnical and Geoenvironmental Engineering, 130(8), 830-842. Yoshizaki, K., O'Rourke, T. D., and Hamada, M. (2001) "Large scale experiments of buried steel pipelines with elbows subjected to permanent ground deformation." Structural Eng./ Earthquake Eng., 18(1), 41-52. Youd, T. L., and Craven, T. N. (1975) "Lateral stress in sands during cyclic loading." Proceedings, A S C E Geotechnical Division, 101(2), 217-221. 272 Youd, T. L., Idriss, I.M., Andrus, R.D., Arango, I., Castro, G., Christian, J.T., Dobry, R., Finn, W.D.L., Harder Jr., L.F., Hynes, M.E., Ishihara, K., Koester, J.P., Liao, S.S.C., Marcuson III, W.F., Martin, G.R., Mitchell, J.K., Moriwaki, Y. , Power, M.S., Robertson, P.K.,Seed, R.B., and Stokoe II, K.H. (2001). "Liquefaction resistance of Soils: summary report from the 1996 N C E E R and 1998 NCEER/NSF workshops on evaluation of liquefaction resistance of soils." Journal of Geotechnical and Geoenvironmental Engineering, 127 (10), 817-833. Zhu, F., and Clark, J. I. (1994) "The effect of dynamic loading on lateral stress in sand." Canadian Geotechnical Journal, 31, 308-311. Zhu, F., Clark, J. I., and Paulin, M. J. (1995) "Factors affecting at-rest lateral stress in artificially cemented sands." Canadian Geotechnical Journal, 32, 195-203. 273 APPENDIX A (Result of laboratory element testing) 274 Direct shear tests on medium dense Fraser River sand 275 1.0 • -4 t A 4 b * 4 r • 1 f * 1 i> - • - Normal stress = 20 kPa i 1 r— 1 1 0 1 2 3 4 5 6 7 8 Displacement (mm) Figure A - l : Variation of shear stress with displacement in direct shear test; Density=l 564 kg/m 3 (D r = 65%) and normal stress of 20 kPa 0.8 Displacement (mm) Figure A-2: Vertical displacement of top cap vs. displacement during direct shear testing; Density=1564 kg/m3 (D r = 65%) and normal stress of 20 kPa 276 1.0 Disp lacement (mm) Figure A-4: Vertical displacement of top cap vs. displacement during direct shear testing; Density=1584 kg/m3 (D r = 71%) and normal stress of 35 kPa 277 1.0 0.8 0.6 0.4 4 0.2 0.0 *• 0 > — ' <• I 1 -•-Normal stress = 53 kPa 1 1 1 1 1 8 2 3 4 5 6 7 Displacement (mm) Figure A-5: Variation of shear stress with displacement in direct shear test; Density=1579 kg/m3 (D r = 70%) and normal stress of 53 kPa 0.8 £ S, 0.6 e <u E <u u « 0.4 a. •3 u r 0.2 > 0.0 * — * — * - • - Normal stress = 53 kPa i 0 2 4 6 8 Displacement (mm) Figure A-6: Vertical displacement of top cap vs. displacement during direct shear testing; Density=1579 kg/m3 (D r = 70%) and normal stress of 53 kPa 278 1.0 (3 • — • - " <> - • - Normal stress = 96 kPa i 1 1 1 1 0 1 2 3 4 5 6 7 8 Displacement (mm) Figure A-7: Variation of shear stress with displacement in direct shear test; Density=1569 kg/m3 (D r = 66%) and normal stress of 96 kPa 0.8 Displacement (mm) Figure A-8: Vertical displacement of top cap vs. displacement during direct shear testing; Density=l 569 kg/m3 (D r = 66%) and normal stress of 96 kPa 279 Direct shear tests on interface of medium dense Fraser River sand and sand-blasted steel 280 1.0 J 3 0 0.5 1 1.5 2 Displacement (mm) 2.5' Figure A-9: Variation of shear stress with displacement in direct shear test of interface of sand and steel; Density=1600 kg/m3 (D r = 75%) and normal stress of 20 kPa E E £ a 0.20 0.15 0.10 -} 0.05 g 0.00 > • < ( i \ y o 5 1 5 : » 2 5 :i - • Normal stress = 20 kPa •0.05 1 -0.10 Displacement (mm) Figure A-10: Vertical displacement of top cap vs. displacement during direct shear test on interface of sand and steel; Density=1600 kg/m3 (D r = 75%) and normal stress of 20 kPa 281 1.0 0.8 4 b 0.4 0.2 ^ — • — . • ---»--» -•-Normal stress = 37 kPa i i 0 0.5 1 1.5 2 2.5 3 Displacement (mm) Figure A - l 1: Variation of shear stress with displacement in direct shear test of interface of sand and steel; Density=1600 kg/m3 (D r = 75%) and normal stress of 37 kPa 0.25 -0.05 J 1 1 J ; : 1 Displacement (mm) Figure A-12: Vertical displacement of top cap vs. displacement during direct shear test on interface of sand and steel; Density=1600 kg/m3 (D r = 75%) and normal stress of 37 kPa 282 1.0 0.8 0.6 0.4 0.2 0.0 • 0 : — • A — — i > A A -— • ' A> - • - Normal stress = 20 kPa i i 0.5 2.5 1 1.5 2 Displacement (mm) Figure A - l 3: Variation of shear stress with displacement in direct shear test of interface of sand and steel; Density=1450 kg/m3 (D r = 20%) and normal stress of 20 kPa £ E S <u u et a •3 "« '43 u a > 0.08 0.06 0.04 0.02 -0.00 <• ( -0.02 --0.04 Displacement (mm) Figure A-14: Vertical displacement of top cap vs. displacement during direct shear test on interface of sand and steel; Density=1450 kg/m3 (D r = 20%) and normal stress of 20 kPa A \ -A 0 \ 0 5 I 1 5 : \ 2 5 3 - • - Normal stress = 20 kPa 283 b « • • • W w w 41 Normal stress = 37 kPa i i 0 0.5 1 1.5 2 2.5 3 Displacement (mm) Figure A - l 5: Variation of shear stress with displacement in direct shear test of interface of sand and steel; Density=1438 kg/m3 (D r = 16%) and normal stress of 37 kPa 0.06 •S 0.04 4 « 0.02 0.00 -0.02 -0.04 Normal stress = 37 kPa Displacement (mm) Figure A-16: Vertical displacement of top cap vs. displacement during direct shear test on interface of sand and steel; Density=1438 kg/m3 (D r = 16%) and normal stress of 37 kPa 284 A.3 Direct shear tests on interface of 2 layers of Filterweave Mirafi 700 woven geotextile 285 0.5 0.4 0.3 4 0.2 0.1 0.0 • 0 — A . d m. - J 1 " <> - • - Normal stress =10 kPa i i i 0.5 2.5 1 1.5 2 Displacement (mm) Figure A-17: Variation of shear stress with displacement in direct shear test of interface of 2 layers of Filterweave Mirafi 700 geotextile; Normal stress of 10 kPa 0.5 0.4 0.3 0.2 o.i 4 0.0 • 0 4m 4 h It V V V • f f <> - • - Normal stress = 15 kPa i i 0.5 2.5 1 1.5 2 Displacement (mm) Figure A - l 8: Variation of shear stress with displacement in direct shear test of interface of 2 layers of Filterweave Mirafi 700 geotextile; Normal stress of 15 kPa 286 b k i > <> .—A ^ 1 <> - • - Normal stress = 20 kPa i i 0 0.5 1 1.5 2 2.5 3 Displacement (mm) Figure A-19: Variation of shear stress with displacement in direct shear test of interface of 2 layers of Filterweave Mirafi 700 geotextile; Normal stress of 15 kPa 287 A.4 Triaxial tests on Fraser River sand 288 8 a 3 -X T i ::: I : : 0 -I i i i i 0 3 6 9 12 15 Axial strain, 8] (%) Figure A-20: Stress ratio versus axial strain, results of triaxial test on dry sand; Density=1662 kg/m3 (Dr=97%) and confining stress of 15 kPa 0.05 T ^ 0.04 -es 5 0.03 -o I 0.02 -" 0.01 -0 --0 1 2 3 4 Axial strain, 8 j (%) Figure A-21: Variation of axial strain over deviatory stress vs. axial strain in triaxial test (to calculate Ej value); Density=1662 kg/m3 (Dr=97%) and confining stress of 15 kPa 289 2 -1 -0 -I i i i i 0 3 6 9 12 15 Axial strain, E t ( % ) Figure A-22: Stress ratio versus axial strain, results of triaxial test on dry sand; Density=1642 kg/m3 (Dr=91%) and confining stress of 25 kPa 0.03 T - r - - T - — - - - i 0 i i i i 1 0 1 2 3 4 Axial strain, 8j (%) Figure A-23: Variation of axial strain over deviatory stress vs. axial strain in triaxial test (to calculate Ej value); Density=1642 kg/m3 (Dr=91%) and confining stress of 25 kPa 290 8 T 7 -6 .2 5 -at u x 4 -la ^ 3 -2 -1 -' 0 -0 3 6 9 12 15 Axial strain, et (%) Figure A-24: Stress ratio versus axial strain, results of triaxial test on dry sand; Density=1670 kg/m3 (Dr=99%) and confining stress of 35 kPa 0.025 i ; - : i £ 0.015 -b • 0.01 -b > » • 1—, » 0.005 -0 --0 1 2 3 4 5 Axial strain, 6j (%) Figure A-25: Variation of axial strain over deviatory stress vs. axial strain in triaxial test (to calculate Ej value); Density=1670 kg/m3 (Dr=99%) and confining stress of 35 kPa 291 6 9 Axial strain, E , ( % ) 12 15 Figure A-26: Stress ratio versus axial strain, results of triaxial test on dry sand; Density=1658 kg/m3 (Dr=95%) and confining stress of 50 kPa 03 0.016 0.014 -\ 0.012 0.01 0.008 0.006 0.004 0.002 0 0 1 y = 0.0033x +0.0011 2 3 Axial strain, Ej (%) Figure A-27: Variation of axial strain over deviatory stress vs. axial strain in triaxial test (to calculate Ej value); Density=1658 kg/m3 (Dr=95%) and confining stress of 50 kPa 292 0 3 6 9 12 15 Axial strain, ej (%) Figure A-28: Stress ratio versus axial strain, results of triaxial test on dry sand; Density=1578 kg/m3 (Dr=70%) and confining stress of 15 kPa 0.06 -~ 0.04 ^ 0.03 •a i 3, 0.02 0.01 -----y= tT.Ol3lx +0.0032 i • •-0 1 2 3 4 Axial strain, zx (%) Figure A-29: Variation of axial strain over deviatory stress vs. axial strain in triaxial test (to calculate Ej value); Density=1578 kg/m3 (Dr=70%) and confining stress of 15 kPa 293 6 1 -0 -I i i i i 1 0 3 6 9 12 15 Axial strain (%) Figure A-30: Stress ratio versus axial strain, results of triaxial test on dry sand; Density=1576 kg/m3 (Dr=69%) and confining stress of 25 kPa 0.05 " 0.01 -0 -I i i 1 1 0 1 2 3 4 5 Axial strain, e, (%) Figure A-31: Variation of axial strain over deviatory stress vs. axial strain in triaxial test (to calculate Ej value); Density=1576 kg/m3 (Dr=69%) and confining stress of 25 kPa 294 1 - -0 -I 1 i i i 1 0 3 6 9 12 15 Axial strian, 6 ! ( % ) Figure A-3 2: Stress ratio versus axial strain, results of triaxial test on dry sand; Density=1570 kg/m3 (Dr=67%) and confining stress of 35 kPa 0.03 0 -I i i i i i 0 1 2 3 4 5 Axial strain, Ej (%) Figure A-33: Variation of axial strain over deviatory stress vs. axial strain in triaxial test (to calculate Ej value); Density=1570 kg/m3 (Dr=67%) and confining stress of 35 kPa 295 6 9 Axial strain, E, (%) 12 15 Figure A-34: Stress ratio versus axial strain, results of triaxial test on dry sand; Density=1575 kg/m3 (Dr=69%) and confining stress of 50 kPa cs 2a 0.03 0.025 0.02 -I ^ 0.015 <*) 3 0.01 u 0.005 0 y = 0.( i044x + 0.0 319 > ^ 1 , 0 1 2 3 4 5 6 Axial strain, Ej (%) Figure A-35: Variation of axial strain over deviatory stress vs. axial strain in triaxial test (to calculate Ej value); Density=1575 kg/m3 (Dr=69%) and confining stress of 50 kPa 296 APPENDIX B (Result of full-scale testing) 297 Results of axial pullout tests 298 B.l. l Axial load vs. displacement response, tests on bare pipe 25 2 0 Z 5 15 c o> 3 10 i -n. •o A O J 5 0 — A B - 2 . ~\ | - — i -1 0 100 2 0 0 3 0 0 4 0 0 5 0 0 6 0 0 7 0 0 8 0 0 9 0 0 Displacment (mm) Figure B - l : Axial pullout load vs. displacement, Test No. AB-2, Average Dr=75%, D=457 mm, H/D=2.5 3 0 25 Hi S z 2 0 -JS it lengl 15 -c s per 10 -•o C3 J 0 — A B - 3 0 100 2 0 0 3 0 0 4 0 0 5 0 0 6 0 0 7 0 0 8 0 0 9 0 0 Displacment (mm) Figure B-2: Axial pullout load vs. displacement, Test No. AB-3, Average Dr=75%, D=457 mm, H/D=2.5 299 3 0 25 § 2 0 61 C •a 15 C L •a o mi 10 0 0 — A B - 4 > 1 • 5 0 100 2 5 0 3 0 0 3 5 0 150 2 0 0 D i s p l a c m e n t ( m m ) Figure B-3: Axial pullout load vs. displacement, Test No. AB-4, Average 0^75%, D=457 mm, H/D=2.5 Z DH B ii C L "O 93 O -J 0 — A B - 5 J i 5 0 2 5 0 3 0 0 100 150 2 0 0 D i s p l a c m e n t ( m m ) Figure B-4: Axial pullout load vs. displacement, Test No. AB-5, Average Dr=20%, D=457 mm, H/D=2.5 300 cs o 5 -o 4 i i *> 4— i * -M 0 100 2 0 0 3 0 0 4 0 0 5 0 0 Displacment (mm) Figure B-5: Axial pullout load vs. displacement, Test No. AB-6, Average Dr=75%, D=457 mm, H/D=2.5 (tested after 45 days of specimen preparation) 301 B.1.2 Pressure transducer measurements on bare pipe tests .z 2.5 Vi "w E i_ o Z •2 1.0 f O 0.5 0 50 300 350 100 150 200 250 Pipe Displacement (mm) Figure B-6: Variation of dimensionless normal stress on the pipe vs. axial displacement; Test No. AB-4, Average Dr=75%, D=457 mm, H/D=2.5 (moving average is applied) 50 250 300 100 150 200 Pipe Displacement (mm) Figure B-7: Variation of dimensionless normal stress on the pipe vs. axial displacement; Test No. AB-5, Average Dr=20%, D=457 mm, H/D=2.5 (moving average is applied) 302 3.5 0.0 A fr 5T5 PT4 \ PT3 6 PT2 / - - PT5 (top) PT4 (+45°) PT3 (Horizontal) PT2 (-45°) PTl (Bottom) t ^ PT3 - i—p ii • hc< A \ [ -——' PT4 < * \ PTl PT5 \ j 1 _ 0 50 100 150 200 250 300 350 400 450 500 Pipe Displacement (mm) Figure B-8: Variation of dimensionless normal stress on the pipe vs. axial displacement; Test No. AB-6, Average Dr=75%, D=457 mm, H/D=2.5 (moving average is applied) 0 50 100 150 200 250 300 350 400 Pipe Displacement (mm) Figure B-9: sample of measurements of pressure transducers, raw data (prior to applying moving average and normalization), Test No. AB-4 303 B.1.3 Pattern of sand particles movements around the pipe after pullout Uncolored Sand into colored sand = I .2-2 mm Y : Normal lo pipe axis i s CO No observation was made Xs*'37 mm Y = 0.4 mm • X = 12:5.mm Y = i:2—l .6 mm X - «s > mm Y - 2 -1-3 2 mm Figure B-10: Sand particles movement pattern, Test No. AB-4; Average Dr=75%, D=457 mm, H/D=2.5 304 U n c b l d r e d s a n d i n t o c o l o r e d s a n d ™ 1 . 6 - 2 . 4 m m Y : Normal to pipe axis Figure B-l 1: Sand particles movement pattern, Test No. AB-4; Average Dr=20%, D=457 mm, H/D=2. 305 .1.4 Axial load vs. displacement, tests on wrapped pipe o 100 2 0 0 6 0 0 7 0 0 800 Figure 3 0 0 4 0 0 5 0 0 D i s p l a c m e n t ( m m ) B-12: Axial pullout load vs. displacement, Test No. AGGC-1 (2 layers of geotextile, cigar wrap); Average Dr=75%, D=457 mm, H/D=2.5 10 Q. •o 03 O - J 9 4 Z 7 •—-•5 6 a * 5 4 3 2 1 0 0 — A G G C - 2 -1 — — ~ ~ 100 2 0 0 6 0 0 7 0 0 8 0 0 Figure 3 0 0 4 0 0 5 0 0 D i s p l a c m e n t ( m m ) B-13: Axial pullout load vs. displacement, Test No. AGGC-2 (2 layers of geotextile, cigar wrap); Average Dr=75%, D=457 mm, H/D=2.5 306 10 9 Z 7 •a 5 'S = 4 >-v o. . O J 2 0 —AGGS-1 f 1 1 —' i 100 200 600 700 800 Figure 300 400 500 Displacment (mm) B-14: Axial pullout load vs. displacement, Test No. AGGS-1 (2 layers of geotextile, spiral wrap); Average Dr=75%, D=457 mm, H/D=2.5 20 18 16 Z 14 5 12 WO B •a io ft o J 4 0 •ARGC-1 100 200 600 700 800 300 400 500 Displacment (mm) gure B-15: Axial pullout load vs. displacement, Test No. ARGC-1 (1 layer of geotextile and 1 layer of geonet, cigar wrap); Average Dr=75%, D=457 mm, H/D=2.5 307 B.1.5 Displacement of geosynthetic layers wrapped around the pipe 0 100 200 300 400 500 600 700 800 Pipe Displacement (mm) Figure B-16: Movement of geosynthetic layers vs. axial pipe displacement, Test No. AGGC-1 (2 layers of geotextile, cigar wrap); Average Dr=75%, D=457 mm, H/D=2.5 — AGGC-2, Inner Layer 0 100 200 300 400 500 600 700 800 Pipe Displacement (mm) Figure B-17: Movement of geosynthetic layers vs. axial pipe displacement, Test No. AGGC-2 (2 layers of geotextile, cigar wrap); Average Dr=75%, D=457 mm, H/D=2.5 308 Pipe Displacement (mm) Figure B-18: Movement of geosynthetic layers vs. axial pipe displacement, Test No. AGGS-1 (2 layers of geotextile, spiral wrap); Average Dr=75%, D=457 mm, H/D=2.5 0 100 200 300 400 500 600 700 800 Pipe Displacement (mm) Figure B-19: Movement of geosynthetic layers vs. axial pipe displacement, Test No. ARGC-1 (1 layer of geotextile and 1 layer of geonet, spiral wrap); Average Dr=75%, D=457 mm, H/D=2.5 309 Results of lateral pulling tests 310 B.2.1 Tests on pipe buried in cohesionless soil: lateral load vs. displacement response, 60 50 4 e 40 ee a <u Z 30 '3 s k. tu a. 20 tj cs o -1 10 0 -f — LN-1 i 0 100 400 500 200 300 Displacement (mm) Figure B-20: Lateral pulling load vs. displacement, Test No. LN-1, Average Dr=75%, D=457mm,H/D=1.92 60 50 £ 40 6D C Z 30 c •a cs o -1 10 0 — LN-2 r 0 100 500 600 200 300 400 Displacement (mm) Figure B-21: Lateral pulling load vs. displacement, Test No. LN-2, Average Dr=75%, D=457mm,H/D=1.92 311 25 20 z 5 15 10 •a cs o 0 • ' H i m : — L N - 3 / 1— I -I i 1 1 i 1 1 0 100 5 0 0 600 2 0 0 300 4 0 0 D i s p l a c e m e n t ( m m ) Figure B-22: Lateral pulling load vs. displacement, Test No. LN-3, Average Dr=75%, D=324 mm,H/D=1.92 25 20 Z ** 15 Z io a. •o cs o J 5 4 0 0 100 — L N - 4 - i t 500 600 2 0 0 300 4 0 0 D i s p l a c e m e n t ( m m ) Figure B-23: Lateral pulling load vs. displacement, Test No. LN-4, Average Dr=75%, D=324 mm, H/D=1.92 312 a Z 6 -o. 4 r r—- i — T r •a i : ei : : : : o : ; ; ; : J 2 - -- -j \ -j ~ 4 j - -0 4 i 1 i i i 0 100 2 0 0 300 4 0 0 500 600 Displacement (mm) Figure B-24: Lateral pulling load vs. displacement, Test No. LN-5, Average Dr=75%, D=324 mm,H/D=1.0 0 4 1 : j 1 i 1 1 1 0 100 2 0 0 300 4 0 0 500 600 Displacement (mm) Figure B-25: Lateral pulling load vs. displacement, Test No. LN-6, Average Dr=75%, D=324 mm, H/D=2.75 313 25 20 ** I5 10 a. •n cd o -I 0 0 100 1 — L N - 7 1 5 0 0 600 2 0 0 300 4 0 0 Displacement (mm) Figure B-26: Lateral pulling load vs. displacement, Test No. LN-7, Average Dr=75%, D=324 mm,H/D=1.92 314 B.2.2 Tests on pipe buried in cohesionless soil: pressure transducers measurements (normal stress) 9.0 s-o 6.0 u <S 5.0 co J 4.0 c o 1.0 0.0 Y \ 5 \ — PT5 PT4 PT3 r n \ PT3 P PTzZ \ PT 3 — r i z T V T " 1 \ P T 1 _y PT1 PT-2 PT-1 I PT-4 - ^ "*•» «>»* 1 'III MH —-—1 20 40 160 180 200 60 80 100 120 140 D i m e n s i o n l e s s D i s p l a c e m e n t Figure B-27: Dimensionless normal stress on the pipe (normalized with overburden stress at springline level) vs. lateral displacement; Test No. LN-7, Average Dr=75%, D=324 mm, H/D=1.92 (moving average is applied) 315 B.2.3 Tests on pipe buried in cohesionless soil: observations after the test (surface deformation and pipe position) S - 1 0 0 u 3 Vi E o u a 03 -40 =+20-100 150 200 250 300 350 Pipe position, before and after pullout (test LN-2) — LN-2 Distance from Pipe Centre (cm) Figure B-28: Surface deformation and pipe position after pulling the pipe, Test No LN-2; Average Dr=75%, D=457 mm, H/D=1.92 § -lpo C D U .3 s. 3 Vi -100 4 -125 200 250 Pipe position, before and after pullout (LN-4) — LN-4 Distance from Pipe Centre (cm) Figure B-29: Surface deformation and pipe position after pulling the pipe, Test No LN-4; Average Dr=75%, D=324 mm, H/D=1.92 316 Figure B-31: Surface deformation and pipe position after pulling the pipe, Test No LN-6; Average Dr=75%, D=324 mm, H/D=2.75 317 B.2.4 Tests on pipe buried in trench configuration constructed in cohesionless native soil: lateral load vs. displacement response CS o 10 | i \ ; i 0 -I i • i i • 1 i 0 100 200 300 4 0 0 500 600 Displacement (mm) Figure B-32: Lateral pulling load vs. displacement, Test No. LNG-1 (45° trench);,Average Dr=75%, D=457 mm, H/D=1.92 a. 2 0 -•n cs o - J 10 -0 1 i i i r— i 1 0 100 2 0 0 300 4 0 0 5 0 0 600 Displacement (mm) Figure B-33: Lateral pulling load vs. displacement, Test No. LNG-1 (35° trench); Average Dr=75%, D=457 mm, H/D=1.92 318 B.2.5 Tests on pipe buried in trench configuration constructed in cohesionless native soil: Geotextile layers displacement 450 400 350 A § 300 | 150 X cu 1 100 50 4 0 0 / — Inner layer, LNG-1 Y — Outer layer, LNG-1 ( I 1 / V_X Axis mmmsP^. : 1 100 200 300 400 P ipe Disp lacement (mm) 500 600 Figure B-34: Movement of geotextile layers vs. lateral pipe displacement, Test No. LNG-1 (45° trench); Average Dr=75%, D=457 mm, H/D=2.5 250 200 300 400 P ipe Displacement (mm) Figure B-35: Movement of geotextile layers vs. lateral pipe displacement, Test No. L N G -2 (35° trench); Average Dr=75%, D=457 mm, H/D=2.5 319 B.2.6 Tests on pipe buried in trench configuration constructed in cohesionless native soil: surface deformation and pipe position 350 Distance from Pipe Centre (cm) Figure B-36: Surface deformation and pipe position after pulling the pipe, Test No L N G -2 (35° trench); Average Dr=75%, D=457 mm, H/D=2.75 320 B.2.7 Tests on pipe buried in trench configuration constructed in hard native soil: load vs. displacement response E 0 50 100 150 2 0 0 250 300 3 5 0 4 0 0 4 5 0 Displacement (mm) Figure B-37: Lateral pulling load vs. displacement, Test No. LT-1 (35° trench); Average Dr=75%, D=324 mm, H/D=1.92 0 + ; 1—i ; 1 ; i ' 1 0 100 2 0 0 3 0 0 4 0 0 5 0 0 6 0 0 Displacement (mm) Figure B-38: Lateral pulling load vs. displacement, Test No. LT-2 (35° trench); Average dry density Dr=75%, moisture content=10% D=324 mm, H/D=l .92 321 80 70 ? 60 fl 50 - 40 '3 a S 30 o. •a S 2 0 —1 10 0 — L T - 3 0 100 500 600 2 0 0 300 4 0 0 D i s p l a c e m e n t ( m m ) Figure B-39: Lateral pulling load vs. displacement, Test No. LT-3 (35° trench); Average dry density Dr=75%, moisture content=10%, D=324 mm, H/D=1.92 60 50 4 E % 40 •a 6X1 a tu Z 30 "3 a tu a- 20 •o cs o -J 10 0 0 100 4 0 0 — L T G - 1 < 1 500 Figure 2 0 0 3 0 0 D i s p l a c e m e n t ( m m ) B-40: Lateral pulling load vs. displacement, Test No. LTG-1 (35° trench); Average Dr=75%, D=324 mm, H/D=1.92 322 0 100 2 0 0 3 0 0 4 0 0 500 Displacement (mm) Figure B-41: Lateral pulling load vs. displacement, Test No. LTG-1 (35° trench); Average dry density Dr=75%, moisture content=10%, D=324 mm, H/D=1.92 323 B.2.8 Tests on pipe buried in trench configuration constructed in hard native soil: measurements of pressure transducers c o CO cu a o a CW E 5 f PT5 >i / PT4\ PT3 q A YXTJ / P T I ^ X y X — -x / LT-1 — PT4 — PT3 PT2 50 200 250 100 150 Dimensionless Displacement Figure B-42: Dimensionless normal stress on the pipe (normalized with overburden stress at springline level) vs. lateral displacement; Test No. LT-1 (35° trench), Average Dr=75%, D=457 mm, H/D=1.92 (moving average is applied) § 6 a 5 CO £ 4 a o '8 a 3 cu E 5 2 1 0 f PTS >i / PT4 v ( PT3 9 f \ > — ^ 1 = ^ • \ PT2/ X P T l ^ X \/ / LT-2 —PT4 — PT3 PI 2 i 50 200 250 100 150 Dimensionless Displacement Figure B-43: Dimensionless normal stress on the pipe (normalized with overburden stress at springline level) vs. lateral displacement; Test No. LT-2 (35° trench), Avg. dry density Dr=75%, moisture content=10%, D=457 mm, H/D=1.92 (moving average is applied) 324 0 50 100 150 200 250 Dimensionless Displacement Figure B-44: Dimensionless normal stress on the pipe (normalized with overburden stress at springline level) vs. lateral displacement; Test No. LT-3 (35° trench), Avg. dry density Dr=75%, moisture content=10%, D=457 mm, H/D=1.92 (moving average is applied) 0 50 100 150 200 250 Dimensionless Displacement Figure B-45: Dimensionless normal stress on the pipe (normalized with overburden stress at springline level) vs. lateral displacement; Test No. LTG-1 (35° trench), Average Dr=75%, D=457 mm, H/D=1.92 (moving average is applied) 325 0 50 100 150 200 250 Dimensionless Displacement Figure B-46: Dimensionless normal stress on the pipe (normalized with overburden stress at springline level) vs. lateral displacement; Test No. LTG-2 (35° trench), Avg. dry density Dr=75%, moisture content=10%, D=457 mm, H/D=1.92 (moving average is applied) 326 B.2.9 Tests on pipe buried in trench configuration constructed in hard native soil: movement of geotextile layers 0.50 B 0.45 tu I 0.40 .2 1" 0.35 5 8 0.30 "a I 0.20 tu 2 o.io 4 H O 0.05 0.00 — Outer Layer, LTG-1 — Inner Layer, LTG-1 Y st? V^XAxis Layers, in .c - 4 • mtactwithT lard fl,mf'' 0.00 0.10 0.60 0.70 0.20 0.30 0.40 0.50 Pipe Dimensionless Displacement Figure B-47: Movement of geotextile layers vs. lateral pipe displacement, Test No. LTG-1 (35° trench); Average Dr=75%, D=457 mm, H/D=2.5 0.70 g 0.60 c v u CS o. 0.50 5 J 0.40 cs .o . 'vi § 0.30 E s g 0.20 >> es J H 0.10 0.00 4 — Outer Layer, LTG-2 — Inner Layer, LTG-2 Y y/ O • / V_X Axis Layers in contact with f)ard ! 1 ! 0.00 0.10 0.60 0.70 0.20 0.30 0.40 0.50 Pipe Dimensionless Displacement Figure B-48: Movement of geotextile layers vs. lateral displacement, Test No. LTG-2 (35° trench); Average dry density Dr=75%, moisture content=10%, D=457 mm, H/D=2.5 327 B.2.10 Tests on pipe buried in trench configuration constructed in hard native soil: surface deformation and pipe position 250 Displacement from Pipe Centre (cm) Figure B-49: Surface deformation and pipe position after pulling the pipe, Test No LT-1 (35° trench); Average Dr=75%, D=457 mm, H/D=2.75 u C M s Vi 2 5 0 Displacement from Pipe Centre (cm) Figure B-50: Surface deformation and pipe position after pulling the pipe, Test No LTG-1 (35° trench); Average Dr=75%, D=457 mm, H/D=2.75 328 a Vi "3 -100 250 Displacement From Pipe Centre (cm) Figure B-51: Surface deformation and pipe position after pulling the pipe, Test No LT-2 (35° trench); Average dry density Dr=75%, moisture content=10%, D=457 mm, H/D=2.75 u u a _ a Vi B O 4) U a 250 Displacement From Pipe Centre (cm) Figure B-52: Surface deformation and pipe position after pulling the pipe, Test No LTG-2 (35° trench); Average dry density Dr=75%, moisture content=10%, D=457 mm, H/D=2.75 329 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0063294/manifest

Comment

Related Items