"Applied Science, Faculty of"@en . "Civil Engineering, Department of"@en . "DSpace"@en . "UBCV"@en . "Srikongsri, Atitep"@en . "2010-05-25T17:36:43Z"@en . "2010"@en . "Doctor of Philosophy - PhD"@en . "University of British Columbia"@en . "In the absence of an extensive body of laboratory and field data, empirical criteria for soil retention in dynamic or cyclic flow are not yet well-defined with reference to a margin of safety. A performance-based approach is taken in this study: the method of investigation involves laboratory tests on a total of seven geotextiles (needle-punched nonwoven and woven materials) and a total of four uniformly-graded soils (non-plastic fine sand and coarse silt). Filtration compatibility in unidirectional and cyclic flow reversal is evaluated using two rigid-wall permeameters: a small bench-mounted device, and a large floor-mounted device. Analysis of the results addresses the effects of specimen size (small and large), sidewall friction and stress distribution, and examines the influence of filter ratio (AOS/Dn), hydraulic gradient (i) and confining stress (\u00CF\u0083\u00CA\u00B9) over a range of cyclic flow reversal times or wave period (T).\n\nA novel analytical framework is proposed from the permeameter test results, to unify AOS/Dn and a hydromechanical index that accounts for the combined effect of hydraulic gradient and confining stress. The framework provides a distinction between the benign actions of mass loss through the geotextile by washout, in contrast to the more problematic action of piping. A filter ratio AOS/D\u00E2\u0082\u0088\u00E2\u0082\u0085 appears better-suited to interpretation of the data than AOS/D\u00E2\u0082\u0085\u00E2\u0082\u0080. The framework is used to examine the margin of safety inherent in current design guidance. Independent verification of the framework through comparison with other laboratory studies, and a consideration of field observations reported by others, leads to a recommendation that AOS/D\u00E2\u0082\u0088\u00E2\u0082\u0085 \u00E2\u0089\u00A4 1 to address undue conservatism in design guidance."@en . "https://circle.library.ubc.ca/rest/handle/2429/25029?expand=metadata"@en . "A LABORATORY PERMEAMETER STUDY OF GEOTEXTILE-SOIL RETENTION IN CYCLIC FLOW by ATITEP SRIKONGSRI B.Eng. King Mongkut University of Technology Thonburi, 1996 M.Eng. Asian Institute of Technology, 1999 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in The Faculty of Graduate Studies (Civil Engineering) THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) May 2010 \u00C2\u00A9 Atitep Srikongsri, 2010 ii ABSTRACT In the absence of an extensive body of laboratory and field data, empirical criteria for soil retention in dynamic or cyclic flow are not yet well-defined with reference to a margin of safety. A performance-based approach is taken in this study: the method of investigation involves laboratory tests on a total of seven geotextiles (needle-punched nonwoven and woven materials) and a total of four uniformly-graded soils (non-plastic fine sand and coarse silt). Filtration compatibility in unidirectional and cyclic flow reversal is evaluated using two rigid-wall permeameters: a small bench-mounted device, and a large floor- mounted device. Analysis of the results addresses the effects of specimen size (small and large), sidewall friction and stress distribution, and examines the influence of filter ratio (AOS/Dn), hydraulic gradient (i) and confining stress (\u00CF\u0083\u00CA\u00B9) over a range of cyclic flow reversal times or wave period (T). A novel analytical framework is proposed from the permeameter test results, to unify AOS/Dn and a hydromechanical index that accounts for the combined effect of hydraulic gradient and confining stress. The framework provides a distinction between the benign actions of mass loss through the geotextile by washout, in contrast to the more problematic action of piping. A filter ratio AOS/D85 appears better-suited to interpretation of the data than AOS/D50. The framework is used to examine the margin of safety inherent in current design guidance. Independent verification of the framework through comparison with other laboratory studies, and a consideration of field observations reported by others, leads to a recommendation that AOS/D85 \u00E2\u0089\u00A4 1 to address undue conservatism in design guidance. iii TABLE OF CONTENTS Abstract ........................................................................................................................ ..ii Table of Contents ......................................................................................................... iii List of Tables ............................................................................................................. .. ix List of Figures ............................................................................................................... xi List of Symbols ......................................................................................................... ..xvi Acknowledgements ..................................................................................................... xix Dedication .................................................................................................................... xx Co-authorship Statement .......................................................................................... xxi 1 Introduction ............................................................................................................ 1 1.1 Geotextile filters................................................................................................ 1 1.2 Soil-geotextile filtration mechanism ................................................................. 3 1.3 Objectives and scope of the study ..................................................................... 6 1.4 Thesis organization ......................................................................................... 10 1.5 References ....................................................................................................... 12 2 Soil-Geotextile Compatibility Testing in Cyclic Flow ....................................... 16 2.1 Outline............................................................................................................. 16 2.2 Introduction ..................................................................................................... 17 2.3 Laboratory test program .................................................................................. 20 2.4 Results and discussion .................................................................................... 22 iv 2.5 Conclusions ..................................................................................................... 25 2.6 References ....................................................................................................... 32 3 Influence of Testing Methodology on Soil-Geotextile Compatibility in Cyclic Flow Using Rigid-Wall Permeameter ................................................................ 34 3.1 Outline............................................................................................................. 34 3.2 Introduction ..................................................................................................... 35 3.3 Test equipment ................................................................................................ 40 3.3.1 Small permeameter .................................................................................. 40 3.3.2 Large permeameter .................................................................................. 42 3.3.3 Control and measurement system ............................................................ 44 3.4 Test methodology............................................................................................ 46 3.4.1 Sample preparation .................................................................................. 46 3.4.2 Multi-stage test procedure ....................................................................... 48 3.4.3 Test materials ........................................................................................... 51 3.5 Results ............................................................................................................. 53 3.5.1 Hydraulic response in head-controlled system ........................................ 53 3.5.2 Large permeameter test data .................................................................... 55 3.5.3 Small permeameter test data .................................................................... 58 3.5.4 Reproducibility of findings ...................................................................... 60 3.5.5 Scale effect ............................................................................................... 61 3.6 Stress in a rigid-wall permeameter ................................................................. 62 3.6.1 Vertical stress distribution ....................................................................... 63 v 3.6.2 Influence of sidewall friction ................................................................... 65 3.6.3 Sidewall friction: large permeameter ....................................................... 67 3.6.3.1 Hydrostatic condition ................................................................... 68 3.6.3.2 Downward seepage flow .............................................................. 68 3.6.3.3 Upward seepage flow ................................................................... 69 3.6.3.4 Coefficient of sidewall friction .................................................... 69 3.6.4 Sidewall friction: small permeameter ...................................................... 70 3.6.4.1 Hydrostatic condition ................................................................... 71 3.6.4.2 Downward seepage flow .............................................................. 72 3.6.4.3 Upward seepage flow ................................................................... 73 3.7 Discussion ....................................................................................................... 74 3.7.1 Size of geotextile sample ......................................................................... 74 3.7.2 Influence of test procedure ...................................................................... 75 3.7.3 Significance of lateral stress .................................................................... 76 3.7.4 Influence of test variables ........................................................................ 77 3.8 Conclusions ..................................................................................................... 79 3.9 References ..................................................................................................... 104 4 Soil-Geotextile Retention in Cyclic Flow ......................................................... 108 4.1 Outline........................................................................................................... 108 4.2 Introduction ................................................................................................... 109 4.3 Experimental methodology ........................................................................... 113 4.3.1 Soils ....................................................................................................... 113 vi 4.3.2 Geotextiles ............................................................................................. 114 4.3.3 Test device and procedure ..................................................................... 115 4.4 Results ........................................................................................................... 118 4.4.1 Nonwoven geotextiles ............................................................................ 119 4.4.1.1 Filter ratio: 0.7 \u00E2\u0089\u00A4 AOS/D85 \u00E2\u0089\u00A4 1.0 ............................................... 119 4.4.1.2 Filter ratio: 1.0 < AOS/D85 \u00E2\u0089\u00A4 2.3 ............................................... 120 4.4.2 Woven geotextiles .................................................................................. 121 4.4.2.1 Filter ratio: 1.2 \u00E2\u0089\u00A4 AOS/D85 \u00E2\u0089\u00A4 2.0 ............................................... 121 4.4.2.2 Filter ratio: 2.0 < AOS/D85 \u00E2\u0089\u00A4 3.7 ............................................... 122 4.5 Analysis and discussion ................................................................................ 128 4.5.1 Soil washout ........................................................................................... 129 4.5.2 Soil retention: a hydromechanical approach .......................................... 130 4.5.2.1 Onset of piping for 2.6 \u00E2\u0089\u00A4 AOS/D85 \u00E2\u0089\u00A4 3.7: woven geotextiles .... 131 4.5.2.2 Hydromechanical influence: a unified plot ................................ 133 4.5.2.3 Unified plot for AOS/D85 ........................................................... 135 4.5.2.4 Unified plot for AOS/D50 ........................................................... 136 4.5.3 Soil retention in cyclic flow ................................................................... 137 4.6 Conclusions ................................................................................................... 138 4.7 References ..................................................................................................... 166 5 Retention Criteria for Geotextile Filter in Cyclic Flow .................................. 169 5.1 Outline........................................................................................................... 169 5.2 Introduction ................................................................................................... 170 vii 5.3 Select laboratory test data ............................................................................. 173 5.3.1 Hawley (2001) ....................................................................................... 175 5.3.2 Cazzuffi et al. (1999) ............................................................................. 177 5.3.3 Chew et al. (2000) .................................................................................. 179 5.4 Field data ....................................................................................................... 180 5.5 A recommended criterion for soil retention .................................................. 182 5.6 Conclusions ................................................................................................... 187 5.7 References ..................................................................................................... 205 6 Conclusions and Recommendations ................................................................. 208 6.1 Conclusions ................................................................................................... 208 6.1.1 Previous study ........................................................................................ 210 6.1.2 Influence of test method ........................................................................ 211 6.1.3 A hydromechanical framework (for onset of retention incompatibility) ............................................................................................................... 213 6.1.4 A recommended criterion for soil retention ........................................... 216 6.2 Recommendations for further study.............................................................. 218 6.3 References ..................................................................................................... 222 Appendices ................................................................................................................. 224 Appendix A Mobilization of sidewall friction ....................................................... 224 Appendix B Example stress calculation for section 3.6.4.2 ................................... 227 Appendix C Summary of key results ..................................................................... 228 Appendix D Select photographs: small permeameter ............................................ 242 viii Appendix E Select photographs: large permeameter ............................................. 250 ix LIST OF TABLES Table 2.1 Properties of the geotextiles ....................................................................... 27 Table 2.2 Multi-stage test procedure .......................................................................... 27 Table 2.3 Test results.................................................................................................. 28 Table 3.1 Properties of the woven geotextiles ........................................................... 82 Table 3.2 Test program .............................................................................................. 82 Table 3.3 Mass loss (g/m 2 ) ......................................................................................... 83 Table 3.3a Wave period T = 6s (900 cycles) ............................................................... 83 Table 3.3b Wave period T = 60s (90 cycles) and T = 120s (45 cycles) ....................... 83 Table 3.4 Soil-geotextile interface stress (small permeameter): parametric values ... 84 Table 4.1 Properties of soils ..................................................................................... 142 Table 4.2 Properties of geotextiles ........................................................................... 143 Table 4.3 Test combinations (and AOS/D85) ........................................................... 144 Table 4.4 Summary of modified values of Gradient Ratio (GR8) and qualitative mass loss observation ........................................................................................ 145 Table 4.5 Average mass loss (mav) in CYC stages (g/m 2 /100 cycles) ...................... 146 Table 5.1 Properties of soils ..................................................................................... 190 Table 5.2 Properties of geotextiles ........................................................................... 191 Table 5.3 Mass washout and mass piping (after Hawley, 2001) .............................. 192 Table 5.4 Mass loss (after Cazzuffi et al. 1999 and Chew et al. 2000) .................... 193 x Table 5.5 Field performance evaluation data (after G. Mannsbart & B.R. Christopher, 1997)......................................................................................................... 194 Table 5.5a Key summary of material properties and performence evaluation .......... 194 Table 5.5b Approximate hydromechanical loading regime ....................................... 195 xi LIST OF FIGURES Figure 2.1 Cyclic Gradient Ratio test device: (a) permeameter assembly; (b) arrangement of sidewall ports ............................................................... 29 Figure 2.2 Grain size distribution curves ................................................................ 30 Figure 2.3 Relation between GR8 (stage UNI4, see Table 2) and filter ratio AOS/D85 ................................................................................................ 31 Figure 2.4 Relation between mass loss and filter ratio AOS/D85 ........................... 31 Figure 3.1 Small permeameter: a) schematic drawing; b) test device; c) test specimen; d) mass collection ................................................................. 85 Figure 3.2 Large permeameter: a) schematic drawing; b) test device; c) test specimen ................................................................................................ 86 Figure 3.3 Schematic diagram of the cyclic head-controlled system ..................... 87 Figure 3.4 Flow chart of multistage test procedure ................................................ 88 Figure 3.5 Water head distribution in test W2-T6(S): a) starting CYC1-stage and b) ending CYC1-stage ............................................................................... 89 Figure 3.6 Water head distribution in tests W2-T6(S), W2-T60(S) and W2- T120(S): a) T = 6 s; b) T = 60 s; c) T = 120 s ....................................... 90 Figure 3.7 Measured stress at the soil-geotextile interface in the large permeameter: a) test W1-T60(L); b) test W2-T60(L) ........................... 92 xii Figure 3.8 Water head distribution in unidirectional flow: a) test W2-T60(S) and b) test W2-T60(L) ...................................................................................... 93 Figure 3.9 Gradient Ratio in tests W2-T60(S) and W2-T60(L): a) iav \u00E2\u0089\u0088 1; b) iav \u00E2\u0089\u0088 5; c) iav \u00E2\u0089\u0088 9 ................................................................................................. 94 Figure 3.10 Comparison of mass loss and volume change in the small and large permeameter .......................................................................................... 95 Figure 3.11 Schematic illustration of stress regime in the test specimen: a) hydrostatic; b) downward flow; c) upward flow ................................... 96 Figure 3.12 Relation of stress difference and average sidewall shear stress ............ 96 Figure 3.13 Stress analysis for large permeameter: a) test W1-T60(L); b) test W2- T60(L).................................................................................................... 97 Figure 3.14 Back-analyzed values of f: a) downward flow; b) upward flow ........... 99 Figure 3.15 Stress calculation procedure for small permeameter ........................... 100 Figure 3.16 Vertical effective stress at soil-geotextile interface ............................. 101 Figure 3.17 Variation of vertical effective stress (at specimen mid-height) in a rigid- wall permeameter: a) typical response to unloading (modified from Mayne and Kulhawy 1982); b) analyzed response based on results of the large permeameter ......................................................................... 102 Figure 3.18 Mean effective stress at soil-geotextile interface: small permeameter ...... ............................................................................................................. 103 Figure 4.1 Soils: a) photograph of Fraser River sand; b) photograph of Alouette River sand; c) grain size distribution curves ....................................... 147 xiii Figure 4.2 Cyclic Gradient Ratio device: a) permeameter; b) head-control system; c) schematic stress distribution; d) relation between top stress on the specimen and mean stress at the soil-geotextile interface (after Srikongsri and Fannin, see chapter 3) ................................................. 148 Figure 4.3 Multi-stage test procedure (test C-W2-T6) at iav \u00E2\u0089\u0088 9 ........................... 149 Figure 4.4 Relation between soil passing and filter ratio for stage CYC2: a) at iav = 1 and b) at iav \u00E2\u0089\u0088 5 ................................................................................. 150 Figure 4.5 SEM images of needle-punched geotextiles: a) new NW4; b) tested NW4 (from D-NW4-T6); c) new NW5; d) tested NW5 (from D-NW5- T6) ....................................................................................................... 151 Figure 4.6 Mass loss at AOS/D85 = 2.6 for T = 6 s: a) test D-W1 and b) repeated D- W1 ....................................................................................................... 152 Figure 4.7 Mass loss at AOS/D85 = 2.6 for T = 60 s: a) test D-W1-T60 and b) repeated D-W1-T60 ............................................................................. 153 Figure 4.8 Mass loss at AOS/D85 = 2.6 for test D-W1-T120 ................................ 155 Figure 4.9 Mass loss at AOS/D85 = 2.8 for T = 6 s: a) test C-W2-T6 and b) repeated C-W2-T6 ............................................................................... 156 Figure 4.10 Mass loss at AOS/D85 = 2.8 for T = 60 s: a) test C-W2-T60 and b) repeated C-W2-T60 ............................................................................. 157 Figure 4.11 Mass loss at AOS/D85 = 2.8 for test C-W2-T120 ................................ 158 Figure 4.12 results at AOS/D85 = 3.7 for test D-W2-T6: a) mass loss; b) end-of-test photograph ........................................................................................... 159 xiv Figure 4.13 SEM images of woven geotextiles: a) tested W1 (from D-W1-T6); b) tested W2 (from C-W2-T6) ................................................................. 160 Figure 4.14 Particle bridging: a) effect of particle shape on vibration-based stability (modified from Valdes and Santamarina, 2008); b) conceptual regime for mechanical instability of spherical particles .................................. 161 Figure 4.15 Inspection of retention compatibility (data from tests D-W1-T6, D-W1- T6-R, C-W2-T6 and C-W2-T6-R): a) piping; b) washout .................. 162 Figure 4.16 Onset of piping: a) soil D - geotextile W1; b) soil C \u00E2\u0080\u0093 geotextile W2; c) soil D \u00E2\u0080\u0093 geotextile W2; d) concept of hydromechanical stability ....... 163 Figure 4.17 Hydromechanical influences on soil retention in cyclic flow for woven geotextiles (data for T = 6 s from Fig. 4.16a, 4.16b, 4.16c and test C- W1-T6) ................................................................................................ 164 Figure 4.18 Retention compatibility for uniformly-graded soil for wave period T = 6 s: a) AOS/D85; b) AOS/D50; c) characteristic zone of soil retention ... 165 Figure 5.1 Geotextile-soil retention: a) AOS/D50; b) AOS/D85 ............................ 196 Figure 5.2 Grain size distribution curves .............................................................. 197 Figure 5.3 Comparison of Hawley (2001) test data for the FR and MT soil ........ 198 Figure 5.4 Comparison of Hawley (2001) test data for the PC soil ...................... 199 Figure 5.5 Original and post-test soil gradation curves: a) test PC-W43a; b) test PC-W43b ............................................................................................. 200 Figure 5.6 Comparison of Cazzuffi et al. (1999) and Chew et al. (2000) test data for the BS and RS soil ......................................................................... 201 xv Figure 5.7 Cross-section of Sungai Buntu ............................................................ 202 Figure 5.8 Comparison of Mannsbart and Christopher field observations .......... 203 Figure 5.9 Combined database for wave period in the range 2 s \u00E2\u0089\u00A4 T \u00E2\u0089\u00A4 20 s ........ 204 xvi LIST OF SYMBOLS AOS Apparent opening size of geotextile (m) c Side-wall interface cohesion (kPa) Cu Coefficient of uniformity which is equal to D60/D10 (dimensionless) CYC Cyclic flow D Diameter of the test specimen (m) Dn Indicative particle size of the base soil at n % passing by weight (m) DPT Differential pressure transducer f Coefficient of sidewall friction (dimensionless) Gs specific gravity of soil (dimensionless) Gr Specific gravity of rock (dimensionless) GRx Gradient Ratio of x mm thick of soil-geoetxtile composite (dimensionless) Hs Significant wave height or design value of wave height (m) Hxy Head loss across the ports x and y (m) iav Average hydraulic gradient across the test specimen (dimensionless) K0 Coefficient of lateral pressure at- rest (dimensionless) kg Permeability of the geotextile (m/s) ks Permeability of the base soil (m/s) LVDT Linear variable differential transformer mav Average mass loss (g/m 2 /100 cycles) xvii mp Mass loss per unit area (g/m 2 ) OCR Over consolidation ratio (dimensionless) OF Filtration opening size of geotextile (m) Ox Pore opening size of geotextile, at which x% of pores are finer (m) Pmax Maximum water pressure in the filter layer of rip-rap (kPa) pi Hydrodynamic mean effective stress at the soil-geotextile interface (kPa) pi(0) Hydrostatic mean effective stress at the soil-geotextile interface or initial confining stress (kPa) S Seepage pressure (kPa) UNI Unidirectional flow W50 Weight of median rock size (N) wr Density of rock mass (kg/m 3 ) Z Length of the test specimen (m) \u00CE\u0094\u00CF\u0083\u00E2\u0080\u00B2v Stress difference in hydrodynamic condition (kPa) \u00CE\u0094\u00CF\u0083\u00E2\u0080\u00B2v(0) Stress difference in hydrostatic condition (kPa) \u00CE\u00B4 Soil-wall interface friction angle (degree) \u00CE\u00B3\u00E2\u0080\u00B2 Submerged unit weight of soil (kN/m3) \u00CE\u00B3w Unit weight of water (kN/m 3 ) \u00CF\u0085 Internal friction angle (degree) \u00CF\u0083\u00E2\u0080\u00B2vb Hydrodynamic effective vertical stress at the base of the test specimen (kPa) xviii \u00CF\u0083\u00E2\u0080\u00B2vb(0) Hydrostatic effective vertical stress at the base of the test specimen (kPa) \u00CF\u0083\u00E2\u0080\u00B2vm Average effective vertical stress in hydrodynamic condition (kPa) \u00CF\u0083\u00E2\u0080\u00B2vm(0) Average effective vertical stress in hydrostatic condition (kPa) \u00CF\u0083\u00E2\u0080\u00B2vt Effective vertical stress applied on top of the test specimen (kPa) \u00CF\u0084av Average side-wall friction resistance (kPa) \u00CE\u00B8 Slope of the rip-rap (degree) xix ACKNOWLEDGEMENTS I wish to thank and express my sincere gratitude to my research supervisor, Dr. Jonathan Fannin, for his continued guidance and kind support throughout the course of this study. I also wish to thank the members of my supervisory committee, Dr. John Howie and Dr. Frank Ko, for their input and constructive comments on the manuscript. I am deeply grateful for funding in support of this research, which was provided by the Natural Sciences and Engineering Research Council (NSERC) of Canada and also Ten Cate Geosynthetics with assistance from Mr. Chris Lawson. I would like to thank the Civil Engineering Workshop staff (Bill Leung and Harald Shremp) and also the Electronics Workshop staff (Scott Jackson). My thanks are extended to all of the graduate students in the geotechnical laboratory for their support and valuable discussions. xx DEDICATION To my parents, who have been a constant source of love, support and encouragement to me. xxi CO-AUTHORSHIP STATEMENT The content of this thesis comprises four proposed manuscripts (chapters 2, 3, 4 and 5, respectively). For the first manuscript (chapter 2), I was responsible for novel revisions to data reduction and analysis, based on some of the laboratory tests performed by Ms. R. Hawley and reported in her MASc thesis in 2001. In recognition of her data collection, and my complementary revisions of the data analysis, she is included as a co-author. For the remaining manuscripts (chapters 3 to 5), I performed all the laboratory tests myself, including all data reduction and analysis. Preparation of all four manuscripts, including the contents, figures and tables, has been solely my responsibility. Dr. R. J. Fannin reviewed all the manuscripts, gave feedback on technical content, assisted with the formulation of my ideas and made suggestions for clarity of presentation. Accordingly, Dr. Fannin is a co-author on all four manuscripts. 1 1 Introduction The term \u00E2\u0080\u009Cfiltration\u00E2\u0080\u009D, as used with reference to civil engineering works, describes the restriction of particle migration from a soil (the \u00E2\u0080\u009Cbase\u00E2\u0080\u009D soil) into or through an adjacent medium (the \u00E2\u0080\u009Cfilter\u00E2\u0080\u009D material) as a consequence of groundwater seepage. The filtration process itself is predicated on the development, over time, of a stable interface between the base soil and the filter material. In construction practice, there is a considerable body of experience with the use of granular soils as a filter material and, in comparison, a growing body of experience with the use of geotextiles as a filter material. The primary objective of the filter is to protect against soil erosion in applications where groundwater flow has the potential to cause a seepage-induced movement of particles while, at the same time, providing for adequate discharge capacity and therefore an unimpeded drainage of the soil to be protected. Accordingly, properly designed filters are integral to the performance of construction works, both with respect to the economic concerns governing serviceability and also for safety concerns governing stability at the ultimate limit state. 1.1 Geotextile filters The manufacturing process yields several constructions or styles of geotextile, two of which, a nonwoven and a woven fabric, are typically used in filtration applications. 2 The styles are inherently different. A nonwoven geotextile comprises a layer of many randomly-oriented polymer strands that are bonded to obtain a planar fabric. The individual strands are usually a short fibre or a continuous filament, generally made of polypropylene and occasionally of polyester or polyethylene. The common methods of bonding are either physical entanglement of the strands, yielding a needle-punched nonwoven geotextile, or thermal fusing of contact points between the strands during a calendaring operation, which produces a heat-bonded nonwoven geotextile. In contrast, a woven geotextile is made from individual polymer strands that are aligned and orthogonally interlaced on a weaving loom, again yielding a planar fabric. The strand itself is usually a slit film, a monofilament, or a multifilament yarn. A fibrillated strand is one that has been intentionally split along portions of its length, as a part of the manufacturing process, to condition its properties. In contrast to a nonwoven geotextile, which has a wide range of pore opening sizes, a woven geotextile tends to have a narrower range of relatively larger openings. A characteristic opening size of the fabric is generally established through indirect means, by placement of a test gradation of either soil or glass ballotini on a specimen of the geotextile, and subsequent determination of the grain size distribution curve of the fraction of that gradation which passes through the fabric under a prescribed disturbance. The disturbing action typically involves either dry shaking or hydrodynamic flushing. A characteristic opening size, for example O95 (\u00C2\u00B5m), is taken to be the equivalent size of the fraction passing, in this case D95, with the implicit 3 understanding that 95% of the pore openings are less than or equal to this value (Fischer 1994; Fischer et al. 1996; Bhatia et al., 1996). Filtration compatibility is predicated on the geotextile satisfying a requirement for soil retention. Incompatibility may take the form of unacceptable piping or clogging. Piping refers to a particle migration through the geotextile, while clogging is a result of entrapment of particles on or within the geotextile. With reference to the permeability of the soil that is retained, piping yields a zone of relatively high permeability adjacent to the geotextile while, in contrast, clogging generates a zone of relatively low permeability. Compatibility may therefore be evaluated by placing soil and geotextile in a permeameter, imposing a prescribed seepage regime, and monitoring any change in the permeability of the soil-geotextile interface relative to that of the undisturbed soil. Interpretation of the results involves comparison of observed change against a threshold value of acceptable filtration compatibility. 1.2 Soil-geotextile filtration mechanism Filtration compatibility requires there be no unacceptable erosion as a consequence of soil loss through the geotextile while, at the same time, providing for unimpeded flow of water seeping from that soil into the drainage aggregate. The expectation, as with granular filters, is that retention of coarser particles in the soil then promotes 4 development of a stable interface or \u00E2\u0080\u009Ebridging zone\u00E2\u0080\u009F in a thin zone of soil adjacent to geotextile (Lawson, 1982). Given this expectation, the design approach is predicated on matching a characteristic pore size opening of the geotextile (for example, O95) to a characteristic particle size of the soil (for example, D85b or D50b and D15b) yielding: C1 x D85b > O95 > C2 x D15b (1) C3 x D50b > O95 (2) where C1, C2 and Cc are constants that depend on soil type and shape of the grain size distribution curve (Leuttich et al. 1992 and Holtz et al. 1997). The approach is very similar to that adopted in granular filters, where C1 addresses soil retention and C2 addresses clogging (Ogink, 1975; Schober and Teindl, 1979; Giroud 1982; Christopher and Holtz, 1985; Gourc and Faure, 1990; Lafleur et al. 1992; Palmeira and Fannin, 2002). With respect to the cross-plane permeability, filtration compatibility is contingent on the geotextile having a capacity for discharge flow significantly greater than that of the soil against which it is placed. The expectation, as for granular filters, is that if each successive layer in the direction of seepage flow exhibits a greater permeability, there is no potential to impede discharge flow through those layers. Geotextiles exhibit a relatively wide range of volumetric flow rate per unit area across the plane of the fabric, with discharge capacity again being largely determined by the 5 manufacturing process. To characterize discharge capacity, the geotextile is mounted in a permeameter and subject to flow under the influence of either a constant differential head or a falling head. A calculation is typically made of the normal permeability kn (cm/s), which may also be reported as a value of permittivity \u00CF\u0088 (s -1 ) if divided by the thickness of the fabric. In routine applications, the design approach is commonly based on matching an index value of cross-plane permeability for the geotextile (kn) to the permeability of the soil (ks). Where concern exists for entrapment of fine particles against and/or within the geotextile, which may result in blinding and/or clogging of the fabric (Koerner and Ko 1982; Lafleur et al. 1989), the ASTM Gradient-Ratio test (D5101) was developed as a performance-oriented test for evaluation of soil-geotextile compatibility (Calhoun 1972; Haliburton and Wood, 1982; Fannin et al., 1994; Fannin et al. 1996). In steady unidirectional flow, filtration compatibility is evaluated based on empirical acceptance criteria. No such performance-oriented criteria exist for cyclic, reversing or pulsating flow applications where flow regimes are not easily reproduced in a simple test device, and interpretation of the onset of retention incompatibility is generally found more complex than unidirectional flow (Giroud, 1996; Mlynarek, 2000; Fannin, 2007). Moreover, the absence of a standardized test method for cyclic flow, and very limited well-documented laboratory and field data, leave considerable uncertainty in design practice. Consequently, and perhaps with good reason, current design guidance adopts a conservative approach. The research of this thesis seeks to address that conservatism. 6 Onset of retention incompatibility in cyclic flow reversal is governed by several factors, including: (i) soil properties, (ii) geotextile properties (iii) effective stress at the soil-geotextile interface, (iv) hydraulic gradient across the soil-geotextile interface, and (v) period of flow reversal (as noted in several studies, including de Graauw et al., 1983; Cazzuffi et al., 1999; Chew et al., 2000; Hameiri, 2000; Hawley, 2001; Chen et al. 2008). The first and second factors represent a geometric constraint or capacity, termed \u00E2\u0080\u009Cfilter ratio\u00E2\u0080\u009D, that describes the ratio of characteristic filter opening size to indicative particle size of base soil (OF/Dn). Collectively, the third and fourth factors represent a hydromechanical demand on the pore size openings of the filter. At a certain period of flow reversal, it is postulated the relation between hydromechanical demand and geometric constraint determines the onset of retention incompatibility in a geotextile filter. 1.3 Objectives and scope of the study Fine sand, sand with some silt and sandy silt are commonly found in estuarine and coastal environments where reversing flow conditions prevail: they represent a challenging base soil for selection and use of geotextile products as a filter in erosion- control structures. The typical wave environment comprises a relatively fast reversing flow (e.g. wind-generated or gravity waves for which a wave period is typically in the 7 range 1 s \u00E2\u0089\u00A4 T \u00E2\u0089\u00A4 10 s), and a slower reversing flow resulting from energy conversion processes acting on gravity waves that filters out the higher frequency (e.g. infragravity waves for which a wave period is typically in the range 50 s \u00E2\u0089\u00A4 T \u00E2\u0089\u00A4 350 s), as described by Munk (1949). In the absence of an extensive body of laboratory and field experience, current design guidance takes an understandably conservative approach. For example, a criterion advocated by Canadian Foundation Engineering Manual for applications of cyclic flow (O95 or AOS \u00E2\u0089\u00A4 0.5D85, adopted from Holtz et al. 1997) yields a maximum value of 0.05 mm for the characteristic opening size of the fabric for a fine sand having D85 of 0.1 mm. Note that AOS is an Apparent Opening Size of a geotextile, according to ASTM D 4751: the O95 size obtained by dry sieving. This AOS or O95 value is much smaller than the typical range of opening size available in geotextile products, and thereby eliminates the options of using a geotextile in some situations where it may offer significant cost saving. Moreover, a smaller opening size tends to reduce the cross-plane permeability of the geotextile, which may have adverse consequences for filtration performance. As a result, selection of a suitable geotextile for cyclic flow applications is made with considerably less understanding than for unidirectional flow. Well-known criteria such as Holtz et al. (1997), and also Luettich et al. (1992) and Pilarczyk (2000), are wholly developed from a body of practical experience and judgment that is based on very limited field data. None of the design criteria for cyclic flow have been developed from systematic laboratory studies, and a consideration of hydromechanical analysis. Furthermore, they for soil retention have not yet been well-defined, and demonstrated valid with reference to a margin of safety. 8 The research of this thesis seeks to introduce a science-based explanation, based on a consideration of the five factors outlined above that are believed to govern retention incompatibility, and thereby enhance confidence in engineering practice. More specifically, the study has two main objectives. First, to develop a hydromechanics-based framework that accounts for the influence both of hydromechanical demand and geometric constraint on onset of retention incompatibility in cyclic flow. Second, to characterize the margin of safety between current design guidance, based on the work of Luettich et al. (1992), Holtz et al. (1997) and Pilarczyk (2000), if appropriate, to make recommendations for modifications to design guidance that address undue conservatism in current practice. The research takes a performance-based approach to the study of geotextile filtration compatibility. The method of investigation involves laboratory tests on a total of seven geotextiles (needle-punched nonwoven and woven materials) and a total of four uniformly-graded soils (non-plastic fine sands and coarse silt). Given the grain size distribution of the soils, and opening size of the geotextiles, the variety of soil- geotextile combinations yielded a range in filter ratio of 0.7 \u00E2\u0089\u00A4 AOS/D85 \u00E2\u0089\u00A4 3.7 (greater than the design guidance, AOS \u00E2\u0089\u00A4 0.5D85). Filtration compatibility is evaluated using two rigid-wall permeameters, a small bench-mounted device and a large floor-mounted device. The small permeameter is a modified version of the ASTM Gradient Ratio permeameter, and the large device was commissioned to examine seepage-induced 9 instability in the core of earth dams and was also modified for the purposes of the current study. Analysis of the results addresses the influence of filter ratio (AOS/D85), hydraulic gradient (i) and confining stress (\u00CF\u0083\u00CA\u00B9) over a range of cyclic flow reversal times or wave period (T). However, an explicit correlation of retention capability between nonwoven and woven geotextiles to account the influence of various fabric structures is not inclusive. The intended contribution of the research is to build confidence in implementing a simple geometric constraint, based on an empirical filter ratio AOS/Dn, through a unified plot against a measure of the hydromechanical demand that is quantified by normalized seepage pressure (S/pi(0)). The demand parameter is a ratio of seepage pressure (S) to initial mean effective stress at the soil-geotextile interface (pi(0)). The novel conceptual framework provides a science-based explanation for onset of soil piping through a geotextile filter and, furthermore, a means to identify the margin of safety inherent in current empirical design criteria. The conceptual framework is developed from testing in the small and large permeameters, and verified independently with reference to laboratory test data and a field study reported in the literature. 10 1.4 Thesis organization This manuscript-based thesis consists of six chapters, which are briefly outlined as follows: \u00E2\u0080\u00A2 Chapter 1 provides an introduction to the phenomenon of retention incompatibility in soil-geotextile filtration, and insight into the nature of conservatism in current design practice, as background context for the objectives and scope of study. \u00E2\u0080\u00A2 Chapter 2 presents an analysis and interpretation of data from a recently-completed study of soil-geotextile compatibility using the small permeameter, reviews selected design criterion for cyclic flow, addresses uncertainty in margins of safety; the findings identify the role and likely contribution of a systematic mechanics- based study to improving design practice. \u00E2\u0080\u00A2 Chapter 3 describes experimental data from the small and the large permeameter test device, with emphasis on matters of scale effect in the test equipment, test methodology and influence of the test device itself on the soil-geotextile interface, in order to ensure the appropriateness of using the small permeameter for a systematic study of variables governing cyclic flow. Additionally, the comparison of data leads to recommendations for the main experimental program. 11 \u00E2\u0080\u00A2 Chapter 4 presents the findings of the main experimental program. The utility of an empirical filter ratio, expressed as AOS/D50 and AOS/D85, is evaluated. The potential for a relation between filter ratio and normalized seepage pressure (S/pi(0)) to explain the onset of retention incompatibility in cyclic flow is explored, leading to a conceptual hydromechanical framework for interpretation of the test data. The framework is then used to consider the margin of safety inherent in current design guidance. \u00E2\u0080\u00A2 Chapter 5 provides an independent verification of the proposed framework, through a comparison with other laboratory studies and a consideration of field observations reported by others. Verification of the concept leads to a recommendation to address undue conservatism, by means of a revision to current design guidance. \u00E2\u0080\u00A2 Chapter 6 concludes the study. Findings of the experimental study are summarized, together with accompanying theoretical provisions for their analysis and interpretation. Given the nature of the experimental work, and its contribution, recommendations are then provided for future research. 12 1.5 References ASTM D 4751, Standard Test Method for Determining Apparent Opening Size of a Geotextile. American Society for Testing and Materials, West Conshohocken, PA, USA. ASTM D 5101, Standard Test Method for Measuring the Soil-Geotextile System Clogging Potential by the Gradient Ratio. American Society for Testing and Materials, West Conshohocken, PA, USA. Bathia, S. K., Smith, J. L. and Christopher, B. R. (1996). Characterization and pore-size distribution: part III: comparison of methods and application to design\u00E2\u0080\u009D, Geosynthetics International, 3, No. 3, 301-328. Calhoun, C.C., Jr. (1972). Development of design criteria and acceptance specifications for plastic filter cloths. Technical Report S-72-7, US Army Engineer Waterways Experiment Station, Vicksburg, Mississippi, USA, 82 p. Canadian Geotechnical Society (2006). Canadian Foundation Engineering Manual. 4th edition, BiTech Publishers Ltd. Cazzuffi, D. A., Mazzucato, A., Moraci, N., & Tondello, M. (1999). A new test apparatus for the study of geotextiles behaviour as filters in unsteady flow conditions: relevance and use. Geotextiles and Geomembranes 17, No. 5-6, 313 - 329. Chen, R.-H., Ho, C.-C. & Hsu, C.-Y. (2008). The effect of fine soil content on the filtration characteristics of geotextile under cyclic flows. Geosynthetics International, 15, No. 2, 95\u00E2\u0080\u0093 106. Christopher, B.R., and Holtz, R.D. (1985). Geotextile engineering manual, Report No. FHWA- TS-86/203, Federal Highway Administration, D.C., USA. 13 de Graauw, A., van der Meulen, T. & van der Does de Bye, M. (1983). Design criteria for granular filters. Publication no. 278, Delft hydraulics laboratory, Delft, The Netherlands. Fannin, R.J. (2007). Chapter 6: The use of geosynthetics as filters. Geosynthetics in Civil Engineering. Woodhead Publishing, Cambridge, UK, 295p. Fannin, R. J., Vaid, Y. P., Palmeira, E. M. & Shi, Y. (1996). A modified gradient ratio device. Recent Developments in Geotextile Filters and Prefabricated Drainage Geocomposites, ASTM STP 1281, Philadelphia, PA, USA, 100 \u00E2\u0080\u0093 112. Fannin, R. J., Vaid, Y. P. and Shi, Y. (1994). Filtration behaviour of nonwoven geotextiles. Canadian Geotechnical Journal, 31, No. 4, 555 \u00E2\u0080\u0093 563. Fischer, G. R. (1994). The influence of fabric pore structure on the behavior of geotextile filters. PhD Thesis, University of Washington, Washington, USA, 502 p. Fischer, G. R., Holtz, R. D., and Christopher, B.R. (1996). Evaluating geotextile pore structure. Recent Developments in Geotextile Filters and Prefabricated Drainage Geocomposites, ASTM STP 1281, Philadelphia, PA, USA, 100 \u00E2\u0080\u0093 112. Giroud, J.P. (1982). Filter criteria for geotextiles, Proceedings of the 2 nd International Conference on Geotextiles, Las Vegas, NV, Industrial Fabrics Association International, St. Paul, MN, USA, Vol. 1, pp. 103-108. Giroud, J. P. (1996). Granular filters and geotextile filters, Proceedings of Geofilters\u00E2\u0080\u009996 Conference, Montreal, Quebec, Canada, pp 565-680. Gourc, J. P. and Faure, Y. H. (1990). Filter criteria for geotextiles. Proceedings of the 4th International Conference on Geotextiles, Hague, the Netherlands, Vol. 4, pp. 949\u00E2\u0080\u0093971. 14 Haliburton, T.A. and Wood,. P.D. (1982). Evaluation of U.S. Army Corps of Engineers Gradient Ratio test for geotextile performance. Proceedings of the 2 nd International Conference on Geotextiles, Las Vegas, NV, Industrial Fabrics Association International, St. Paul, MN, USA, pp. 97-101 Hameiri, A. (2000). Soil geotextile filtration behavior under dynamic conditions of vibration and cyclic flow. PhD Thesis, University of British Columbia, BC, Canada, 270p. Hawley, R. (2001). Filtration performance of geotextiles in cyclic flow conditions. MASc Thesis, University of British Columbia, BC, Canada, 141p. Holtz, R.D, Christopher, B.R., and Berg, R.R. (1997). Geosynthetic Engineering. BiTech Publishers, Richmond, BC, Canada, 452 p. Koerner, R. M. and Ko, F. K. (1982). Laboratory studies on long-term drainage capability of geotextiles. Proceedings of the 2 nd International Conference on Geotextiles, Las Vegas, NV, Industrial Fabrics Association International, St. Paul, MN, USA, pp. 91 \u00E2\u0080\u0093 95. Lafleur J., Mlynarek J., and Rollin A. L. (1989). Filtration of broadly graded cohesionless soils. Journal of Geotechnical Engineering, ASCE, 115, No. 12, pp. 1747-1768. Lafleur J., Mlynarek J., and Rollin A. L. (1992). Filter criteria for well graded cohesionless soils. Filters in Geotechnical and Hydraulic Engineering, Balkema, Rotterdam, The Netherlands, pp. 97 \u00E2\u0080\u0093 106. Lawson, C.R. (1982). Filter criteria for geotextiles: relevance and use. ASCE Journal of Geotechnical Engineering, 108, No. 10, pp. 1301-1317 Luettich, S.M., Giroud, J.P., and Bachus, R.C. (1992). Geotextile filter design guide. Geotextiles and Geomembranes, 11, 355 - 370. 15 Mlynarek J. (2000). Geo drains and geo filter-retrospective and future trends. Filters and Drainage in Geotechnical and Geoenvironmental Engineering, Balkema, Rotterdam, The Netherlands, pp. 27 - 47. Munk, W. H. (1949). Surf beats. EOS, Transactions, American Geophysical Union, 30, 849 - 859. Ogink, H. J. M. (1975). Investigations on the hydraulic characteristics of synthetic fabrics. Publication no. 146, Delft Hydraulic Laboratory, Delft, The Netherlands. Palmeira, E.M. and Fannin, R.J. (2002). Soil-geotextile compatibility in filtration. Proceedings of the 7 th International Conference on Geosynthetics, Nice, France, pp. 853-872. Pilarczyk, K. W. (2000). Geosynthetics and geosystem in hydraulic and coastal engineering. A.A. Balkema, Rotterdam, The Netherlands, 913 p. Schober, W. and Teindl, H. (1979). Filter criteria for geotextiles, 7th European Conference on Soil Mechanics and Foundation Engineering, Brighton, UK, Vol. 1, pp. 121-129. 16 2 Soil-Geotextile Compatibility Testing in Cyclic Flow 1 2.1 Outline Influence of filter ratio on soil-geotextile interaction is examined under conditions of cyclic flow. Seven different geotextiles and three non-plastic soils are tested in a cyclic gradient ratio test device, using a multistage test procedure with variables of confining stress and wave period. Mass loss per unit area in stages of cyclic flow increases with larger AOS/D85. The empirical criterion of AOS/D85 \u00E2\u0089\u00A4 0.5 for soil retention in cyclic flow is found conservative, both for nonwoven and woven geotextiles. Wave period and confining stress influence soil-geotextile filter compatibility, and those parameters, in combination with hydraulic gradient, require systematic study in order to understand the margin of safety that governs a confident use of the empirical criterion. 1 A version of this chapter will be submitted for publication. Srikongsri, A., Fannin, R. J., Hawley, R. (2010). Soil-geotextile compatibility testing in cyclic flow. 17 2.2 Introduction In the current state-of-practice, it is reasonable to observe that the behavior of geotextile filters in earthworks subject to unidirectional flow of groundwater seepage is well-understood and, consequently, that companion design criteria may be used with confidence. The confidence is predicated on a longstanding appreciation of the fundamental physical processes that govern compatibility (Lawson, 1982; Hoare, 1982). Subsequent recommendations for design criteria are wholly empirical and, importantly, assume the base soil through which seepage flow occurs is internally stable (Palmeira and Fannin, 2002). In contrast to unidirectional seepage flow in routine filter applications, where a use of geotextiles is based on considerable field experience and many laboratory studies, the issue of bidirectional, reversing or cyclic flow is one for which our current understanding is more limited (Fannin, 2007). This may be attributed to several factors, including the relatively uncommon occurrence of reversing flow in routine engineering works, and corresponding lack of good documented field experience, coupled with a paucity of laboratory studies that address the specific nature of such flow regimes. Yet considerable challenges exist in the confident provision of filters for protection of civil infrastructure in estuarine and coastal environments, where a subtle distinction can be made between slow reversing flow, such as that of tidal 18 environments, and the relatively faster reversing flow that occurs in the presence of wave action. Fannin and Srikongsri (2007) provided a critical review of geotextile behavior in cyclic filtration experimental studies, with emphasis on governing factors and design criteria. Three factors, namely wave period, hydraulic gradient and confining stress were found to significantly influence compatibility of soil-geotextile filter. A limited number of studies report on development of a laboratory test device for cyclic flow (Cazzuffi et al., 1999; Chew et al., 2000; Hameiri, 2000; Fannin and Pishe, 2001; Hawley, 2001; Hameiri and Fannin, 2002). Cazzuffi et al. (1999) described the influence of hydraulic gradient and confining stress on the performance of geotextile filters in cyclic flow, from test data on four combinations of two sandy soils and a woven and a nonwoven geotextile. Although the laboratory permeameter test device did not conform to specifications of the ASTM Gradient Ratio device, its configuration permitted a similar characterization of filtration compatibility with reference to a ratio of hydraulic gradient across the soil-geotextile interface to that within the base soil. Considerable quantities of mass loss were observed in test combination for which the woven geotextile opening size was relatively large (AOS/D85 \u00E2\u0089\u0088 2.2). At constant AOS/D85, the general trend was characterized by greater mass loss with increasing hydraulic gradient or decreasing confining stress. 19 Effects of wave period in cyclic flow were examined by Chew et al. (2000), from tests on gravelly sand against a woven and a nonwoven geotextile. Once again, although their laboratory permeameter test device did not conform to specifications of the ASTM Gradient Ratio device, its configuration permitted measurement of hydraulic gradient across the soil-geotextile composite and within the soil. Mass loss was found to increase with the decrease of wave period, particularly from 15 s to 2 s. Recognizing the importance of wave period, they proposed an index value for rate of changing hydraulic gradient; Ri = 2\u00E2\u0088\u0086i/T, where \u00E2\u0088\u0086i = hydraulic gradient in each cycle and T is wave period. This index was recommended as a measure of hydraulic loading conditions. The ASTM Gradient Ratio test (ASTM D5101) was originally conceived to evaluate compatibility of base soil and geotextile filter in steady unidirectional flow. Modifications to enable tests with cyclic flow are described by Hameiri and Fannin (2002). The modified Gradient ratio test device allows testing of a soil-geotextile specimen that is subject to axial confining stress, with gradient-control of seepage flow at a constant period, and collection of soil loss through the geotexile during a test. Data obtained on soil-geotextile compatibility, using the modified Gradient Ratio test device with cyclic flow, are reported with the objective of evaluating its suitability as a performance test. Emphasis is placed on the utility of collecting soil that passes through the geotextile, where concern for filtration compatibility addresses a criterion 20 for soil retention. Evaluation of an empirical design criterion that employs the filter ratio AOS/D85 yields insight to the influence of effective stress on filtration compatibility. 2.3 Laboratory test program The modified Gradient Ratio test device used in testing is equipped with a loading frame to apply axial force to the top surface of the specimen, and a conical trough to collect soil washed though the geotextile (Fig. 2.1a). A series of multistage cyclic flow tests, on specimens of reconstituted glass beads, was used to commission the laboratory test device (Hameiri, 2000): the utility of a modified value of gradient ratio (GR8) was demonstrated, based on measurement of the hydraulic gradient across a gauge length of only 8 mm across the base soil and geotextile. The rationale for a relatively short gauge length is that any incompatibility of the filtration interface is most evident immediately upstream of the geotextile. An array of port locations on the side-wall of the permeameter (see Fig. 2.1b, with values in parentheses indicating height above the geotextile), define the ASTM Gradient Ratio (GR25) and modified value (GR8) respectively as: 35 57 25 i i GR \u00EF\u0080\u00BD (1) and 21 35 67 8 i i GR \u00EF\u0080\u00BD (2) where hydraulic gradient i35 is measured across the soil between ports P3 and P5, i57 is measured across the soil-geotextile zone between ports P5 and P7, and i67 across the same zone between ports P6 and P7. Details of the flow control and data acquisition have been summarized by Hameiri and Fannin (2002). Hawley (2001) used the device to test three soils (see Fig. 2.2). Soil FR, a Fraser River sand with trace silt, has a D85 = 0.33 mm and coefficient of uniformity Cu = 1.8. Soil PC is also a river-deposited sand with some silt, for which D85 = 0.215 mm and Cu = 5.8. Soil MT, a processed mine tailings with some silt, has a D85 = 0.29 mm and Cu = 3.3. The mine tailings exhibited an angular grain shape, while both alluvial soils had sub-rounded grains. The gradation curves are evaluated as internally stable according to the method of Kenney and Lau (1985; 1986). Testing in the permeameter established a typical permeability of 0.025, 0.0015 and 0.0001 cm/s respectively for reconstituted specimens of soil FR, MT and PC. Combinations of the three soils were tested against seven geotextiles, for which material properties are reported in Table 2.1. The two needle-punched nonwoven geotextiles have the same opening size of 0.212 mm, and the five woven geotextiles exhibit a range from 0.212 mm to 0.6 mm. Tests were conducted at an average 22 hydraulic gradient (i17) of 4. Test variables examined were (i) the flow regime (unidirectional or cyclic), (ii) the imposed normal stress (0 or 25 kPa), and (iii) the period of cyclic flow reversal (50 or 10 s). A summary of the multistage test sequence followed in testing is given in Table 2.2. In principle, a relatively long stage of cyclic flow at T = 50 s (CYC 1) was followed by a shorter stage at T = 10 s (CYC 2), whereupon the normal stress was reduced from 25 kPa to zero, and the shorter stage at T = 10 s then repeated (CYC 3). In order to characterize filtration compatibility, each cyclic stage was preceded and followed by a stage of unidirectional flow. The rationale for choosing these test conditions is based on simulation of bank or shore protection structures that experience a low confining stress, and may be subject to a wide range of wave periods. A wave period T= 10 s is deemed \u00E2\u0080\u009Cfast\u00E2\u0080\u009D reversing flow and recommended for simulation of wave action, while longer periods are deemed \u00E2\u0080\u009Cslow\u00E2\u0080\u009D reversing flow typical of tidal environments 2.4 Results and discussion Combinations of soil and geotextile examined in testing yield a filter ratio 0.6 \u00E2\u0089\u00A4 AOS/D85 \u00E2\u0089\u00A4 2.1 (see Table 2.3). For internally stable soil, a value of gradient ratio equal to one corresponds to a linear variation of head loss across the test specimen, believed indicative of excellent soil-geotextile compatibility; a value less than one implies a non-linear variation of head loss, in which the permeability of the soil-geotextile zone 23 (k57 or k67) is less than that of the soil (k35). Values of GR25 and GR8 obtained in the last stage of unidirectional flow (UNI 4) generally diminish with increasing AOS/D85 (see Table 2.3 and Fig. 2.3), to a GR8 \u00E2\u0089\u00A4 0.5 at AOS/D85 \u00E2\u0089\u00A5 1.5. The data suggest a relatively more permeable soil-geotextile interface develops with increasing pore size opening in the woven geotextiles. The needle-punched nonwoven geotextiles all have AOS/D85 \u00E2\u0089\u00A4 1, and a GR8 > 0.5 for the UNI 4 stage of testing. Mass loss is reported separately for each stage of cyclic flow (see Table 2.3). Data for the first and last cyclic stage are plotted against AOS/D85 (see Fig. 2.4). Mass loss generally increases with greater AOS/D85. The relation appears sensitive to opening size of the geotextile, given a loss smaller than 60 g/m 2 at AOS/D85 \u00E2\u0089\u00A4 1 during the first stage of cyclic loading performed at a confining stress of 25 kPa (Table 2.2). Indeed, no mass loss was observed in any stage of cyclic flow in the three tests with AOS/D85 = 0.6, and a minimal loss (mp \u00E2\u0089\u00A4 30 g/m 2 ) was observed in only the initial stage of cyclic flow (CYC 1) in the three tests with AOS/D85 = 0.7 (see Table 2.3). The results describe the response of both nonwoven and woven geotextiles to cyclic flow. For these six tests with AOS/D85 \u00E2\u0089\u00A4 0.7, inspection of gradient ratio values obtained in the last stage of unidirectional flow (UNI 4) establishes an average GR25 = 0.97 and GR8 = 0.93. The finding implies consistency between the gradient ratio index value of relative permeability and absence of seepage-induced mass loss through the geotextile. 24 In contrast, the three test combinations with AOS/D85 \u00E2\u0089\u00A5 2 yielded a cumulative mass loss mp \u00E2\u0089\u00A5 1250 g/m 2 , which is much greater than that observed in all other tests. It was sufficiently large to cause early termination of the two tests with AOS/D85 = 2, given mp \u00E2\u0089\u00A5 4000 g/m 2 . The nature of the response in these PC-G43aW and PC-G43bW tests, namely increased loss at reduced confining stress and wave period, appears consistent with the observations of Cazzuffi et al. (1999) and Chew et al. (2000). More generally, at greater values of AOS/D85, the results show gradient ratio values (Fig. 2.3) consistent with the observations of mass loss (Fig. 2.4). Accordingly, values of GR8 \u00E2\u0089\u00A4 0.5 are attributed to piping of soil through the woven geotextiles examined in testing. In filtration applications of dynamic, pulsating or reversing flow, Holtz et al. (1997) recommended an empirical design criterion of AOS/D85 \u00E2\u0089\u00A4 0.5 for purposes of soil retention by a woven or nonwoven geotextile. The criterion evolved from an earlier expression based on O50, rather O95, that was based on a general review of filter design recommendation in several countries (Christopher and Holtz, 1985). A comparison of the empirical criterion for soil retention and laboratory test data (of Fig. 2.4) suggests the recommendation is conservative to design practice. In making this observation it should be noted that three factors, namely wave period, the hydraulic gradient and the confining stress are postulated to influence compatibility of soil-geotextile filter (Fannin and Srikongsri, 2007). Yet, none of these factors can be confidently addressed using the data of Fig. 2.4, and should be considered systematically in future studies in 25 order to properly understand the margin of safety that is implied in use of such empirical criteria. 2.5 Conclusions The Gradient Ratio test was originally conceived to evaluate compatibility of base soil and geotextile filter in unidirectional flow. The test apparatus described has been configured to enable testing with unidirectional or cyclic flow, and to permit collection of soil that passes through the geotextile as a consequence of seepage flow. The following conclusions are made based on characterization of the soil (D85), geotextile (AOS), soil-geotextile compatibility (GR8) and mass of soil passing per unit area (mp): \u00EF\u0082\u00B7 test combinations of soil and geotextile that exhibit filtration compatibility in stages of cyclic flow, yield values of gradient ratio 0.5 \u00E2\u0089\u00A4 GR8 \u00E2\u0089\u00A4 2.0 (0.7 \u00E2\u0089\u00A4 GR25 \u00E2\u0089\u00A4 1.6) in a following stage of unidirectional flow; \u00EF\u0082\u00B7 gradient ratio values diminish with larger AOS/D85; \u00EF\u0082\u00B7 mass loss per unit area in stages of cyclic flow increases with larger AOS/D85; \u00EF\u0082\u00B7 values of GR8 \u00E2\u0089\u00A4 0.5 are attributed to piping of soil through the woven geotextiles examined in testing. \u00EF\u0082\u00B7 mass loss per unit area provides a very useful index of filtration compatibility for soil-geotextile combinations that exhibit piping (and a gradient ratio 26 significantly less than 1.0) rather than clogging (and a gradient ratio significantly greater than one), and should be reported to assist with test interpretation. A negligible mass loss (mp \u00E2\u0089\u00A4 30 g/m 2 ) in only the first stage of cyclic flow at AOS/D85 = 0.7, and no mass loss (mp = 0) in any stage of cyclic flow at AOS/D85 = 0.6, suggest: \u00EF\u0082\u00B7 the empirical criterion of AOS/D85 \u00E2\u0089\u00A4 0.5 for soil retention in cyclic flow is conservative, both for nonwoven and woven geotextiles. An unacceptable mass loss (mp \u00E2\u0089\u00A5 4000 g/m 2 ) in the third stage of cyclic flow at AOS/D85 = 2, preceded by a relatively modest loss in the first two stages of cyclic flow, suggests: \u00EF\u0082\u00B7 wave period and confining stress influence soil-geotextile filter compatibility, and those parameters, in combination with hydraulic gradient, require systematic study in order to understand the margin of safety that governs a confident use of the empirical criterion. 27 Table 2.1 Properties of the geotextiles Geotextile (code) Type Mass / Unit Area AOS (ASTM D4751) Permittivity (ASTM D4491) Permeability NW/W (g/m 2 ) (mm) (sec -1 ) (cm/s) G21Na NW 287 0.212 1.310 0.290 G21Nb NW 185 0.212 1.192 0.134 G21W W 218 0.212 0.511 0.021 G30W W 225 0.300 0.769 0.049 G43Wa W 282 0.425 0.881 0.080 G43Wb W 304 0.425 2.003 0.194 G60W W 453 0.600 0.366 0.061 Table 2.2 Multistage test procedure Stage UNI1 CYC1 UNI2 CYC2 UNI3 CYC3 UNI4 Normal Stress (kPa) 0 25 25 25 25 0 0 Wave Period (s) 0 50 0 10 0 10 0 Duration (min) 90 900 30 43 30 43 30 Number of cycles 1080 260 260 28 Table 2.3 Test results Test code AOS/D85 Mass loss (g/m 2 ) UNI4 CYC1 CYC2 CYC3 GR25 GR8 FR-G21aN 0.6 \u00E2\u0089\u0088 0 \u00E2\u0089\u0088 0 \u00E2\u0089\u0088 0 1.0 0.8 FR-G21bN 0.6 \u00E2\u0089\u0088 0 \u00E2\u0089\u0088 0 \u00E2\u0089\u0088 0 1.3 1.1 FR-G21W 0.6 \u00E2\u0089\u0088 0 \u00E2\u0089\u0088 0 \u00E2\u0089\u0088 0 0.9 0.8 FR-G30W 0.9 \u00E2\u0089\u0088 0 \u00E2\u0089\u0088 0 \u00E2\u0089\u0088 0 1.1 1.3 FR-G43aW 1.3 \u00E2\u0089\u0088 0 \u00E2\u0089\u0088 0 \u00E2\u0089\u0088 0 1.0 1.0 FR-G43bW 1.3 6 \u00E2\u0089\u0088 0 \u00E2\u0089\u0088 0 1.3 1.3 FR-G60W 1.8 40 \u00E2\u0089\u0088 0 \u00E2\u0089\u0088 0 0.9 0.9 MT-G21aN 0.7 26 \u00E2\u0089\u0088 0 \u00E2\u0089\u0088 0 0.9 0.7 MT-G21bN 0.7 13 \u00E2\u0089\u0088 0 \u00E2\u0089\u0088 0 1.6 1.2 MT-G21W 0.7 4 \u00E2\u0089\u0088 0 \u00E2\u0089\u0088 0 0.9 1.0 MT-G30W 1.0 60 \u00E2\u0089\u0088 0 \u00E2\u0089\u0088 0 1.2 0.7 MT-G43aW 1.5 313 115 128 0.3 0.1 MT-G43bW 1.5 162 17 53 0.6 0.4 MT-G60W 2.1 1246 \u00E2\u0089\u0088 0 88 0.4 0.2 PC-G21aN 1.0 34 \u00E2\u0089\u0088 0 39 1.0 0.8 PC-G21bN 1.0 56 \u00E2\u0089\u0088 0 1 1.0 0.7 PC-G21W 1.0 49 \u00E2\u0089\u0088 0 28 1.2 2.0 PC-G30W 1.4 55 \u00E2\u0089\u0088 0 51 1.4 1.9 PC-G43aW 2.0 95 \u00E2\u0089\u0088 0 4349 N/A N/A PC-G43bW 2.0 239 165 4953 N/A N/A 29 (a) (b) Figure 2.1 Cyclic Gradient Ratio test device: (a) permeameter assembly; (b) arrangement of sidewall ports 30 Figure 2.2 Grain size distribution curves 31 Figure 2.3 Relation between GR8 (stage UNI4, see Table 2) and filter ratio AOS/D85 Figure 2.4 Relation between mass loss and filter ratio AOS/D85 32 2.6 References ASTM D 5101, Standard Test Method for Measuring the Soil-Geotextile System Clogging Potential by the Gradient Ratio, American Society for Testing and Materials, West Conshohocken, PA, USA. Cazzuffi, D.A., Mazzucato, A., Moraci, N., & Tondello, M. (1999). A new test apparatus for the study of geotextiles behaviour as filters in unsteady flow conditions: relevance and use. Geotextiles and Geomembranes, 17, 313 \u00E2\u0080\u0093 329. Chew, S.H., Zhao, Z.K., Karunaratne, G.P., Tan, S.A, Delmas, P., and Loke, K.H. (2000). Revetment geotextile filter subjected to cyclic wave loading. Proceedings of Geo-Denver 2000, Denver, CO, USA, pp. 162 \u00E2\u0080\u0093 175. Christopher, B.R., and Holtz, R.D. (1985), Geotextile engineering manual, Report No. FHWA- TS-86/203, Federal Highway Administration, D.C., USA, pp. 1044 p. Fannin, R.J. (2007). Chapter 6: The use of geosynthetics as filters. Geosynthetics in Civil Engineering. Woodhead Publishing, Cambridge, UK, 295 p. Fannin, R.J. and Pishe, R. (2000). Testing and specifications for geotextile filters in cyclic flow applications. Proceedings of Geosynthetics 2001, Portland, OR, USA, pp. 423-435. Fannin, R.J., Vaid, Y.P., Palmeira, E.M. and Shi, Y.C., (1996). A modified gradient ratio device. Recent Developments in Geotextile Filters and Prefabricated Drainage Geocomposites. ASTM STP 1281, Philadelphia, PA, USA, pp. 100 \u00E2\u0080\u0093 112. Fannin, R.J. and Srikongsri, A. (2007). Geotextile filters in cyclic flow: test results and design criteria. Proceedings of Geosynthetics 2007, Washington, D.C., USA, pp. 170-185. 33 Hameiri, A. (2000). Soil geotextile filtration behavior under dynamic conditions of vibration and cyclic flow. PhD Thesis, University of British Columbia, British Columbia, Canada, 270 p. Hameiri, A. and Fannin, R.J. (2002). A cyclic gradient ratio test device. ASTM Geotechnical Testing Journal, 39, 266-276. Hawley, R. (2001). Filtration performance of geotextiles in cyclic flow conditions. MASc Thesis, University of British Columbia, British Columbia, Canada, 141 p. Hoare, D.J. (1982). Synthetic fabrics as soil filters: a review. ASCE Journal of Geotechnical Engineering, 108, 1230-1245. Holtz, R.D, Christopher, B.R., and Berg, R.R. (1997). Geosynthetic Engineering. BiTech Publishers, British Columbia, Canada, 452 p. Kenney, T. C. & Lau, D. (1985). Internal stability of granular filters. Canadian Geotechnical Journal, 22, No. 2, 215-225 Kenney, T. C. & Lau, D. (1986). Internal stability of granular filters: Reply. Canadian Geotechnical Journal, 23, No. 3, 420-423 Lawson, C.R. (1982). Filter criteria for geotextiles: relevance and use. ASCE Journal of Geotechnical Engineering, 108, 1301-1317. Palmeira, E.M. and Fannin, R.J. (2002). Soil-geotextile compatibility in filtration. Proceedings of the 7 th International Conference on Geosynthetics, Nice, France, pp. 853-872. 34 3 Influence of Test Methodology on Soil-Geotextile Compatibility in Cyclic Flow Using Rigid-Wall Permeameter 2 3.1 Outline Cyclic flow is reproduced in two rigid-wall permeameters, of different size, to examine soil-geotextile filtration compatibility. The influence of test methodology is reported, from data on uniform sand and a monofilament and a multifilament woven geotextiles. Cyclic flow regime is characterized with reference to hydraulic gradient and wave period. Vertical effective stress at the soil-geotextile interface is obtained directly by measurement in the large permeameter, and indirectly by computation in the small permeameter taking into account side-wall friction. A new approach is proposed for data interpretation that considers both seepage pressure and stress history, using a value of mean effective stress at the soil-geotextile interface. No scale effect is apparent in results from the two permeameters. Values of gradient ratio, mass loss and volume change indicate a concern for soil retention at a filter ratio (AOS/D85) between 2.0 and 2.8 that appears governed by effective stress and wave period. 2 A version of this chapter will be submitted for publication. Srikongsri, A.and Fannin, R. J. (2010). Influence of test methodology on soil-geotextile compatibility in cyclic flow using rigid-wall permeameter. 35 3.2 Introduction In filtration applications, reversing or pulsating seepage that yields a cyclic flow regime is encountered at river, lacustrine and marine protection works. The characteristics (i.e. wind wave, ship wave and tidal) of cyclic flow vary considerably with reference to spatial and temporal variations of hydraulic gradient, effective stress and period of flow reversal. Few systematic experimental studies of these variables are reported for granular filters (de Graauw et al., 1983 and de Graauw et al., 1984), and a very limited body of test data is reported for geotextile filters. No standard approach to testing is found in the literature, and design guidance for geotextile selection in applications of cyclic flow is, for the most part, unsupported by a firm understanding of the governing variables. Accordingly, the margin of safety inherent in use of empirical rules for such geotextile filtration design cannot be expressed with confidence. Over a considerable period of time, research efforts on filtration applications in cyclic flow have sought to examine the influence of wave period, hydraulic gradient and effective stress. Laboratory permeameter test devices may be generally categorized in two sizes, namely a large permeameter of approximately 300 mm diameter, and a small permeameter of approximately 100 mm diameter. To date, issues of any scale effect in laboratory testing have not received attention. In particular, the variation in 36 effective stress along the length of a rigid-wall permeameter, under hydrodynamic conditions, is not well-understood due to limitations of the test device regarding the characterization of sidewall friction. In a study of granular filters for storm surge barrier and coastal protection structures, de Graauw et al. (1984) reproduced a flow regime of in-plane cyclic flow (bi- directional flow parallel to the filter interface), and also cross-plane cyclic flow (bi- directional flow perpendicular to the filter interface). In contrast to in-plane cyclic flow, tests with cross-plane seepage resulted in movement of base soil into the filter layer. The later test series was conducted in a rigid-wall permeameter, 280 mm in diameter, employing the principle of head-control of seepage flow: a test performed at a constant differential water head across the soil sample to achieve the predetermined value of gradient. The specimen of base soil, and granular filter, were each approximately 350 mm in length. Importantly, a constant wave period (T) of 10 s, hydraulic gradient (i) from 1 to 8 and effective stress (\u00CF\u0083) from 0 to 130 kPa were all examined as test variables. A sensitivity of the base soil, a fine sand, to arching phenomena at the filter interface was noted at a large filter ratio D15f/D50b \u00E2\u0089\u0088 13. Importantly, critical gradient at which instability occurred was found to increase with increasing stress. No consideration was given to sidewall friction in analysis of the results, and repeatability of filtration behaviour was not explicitly addressed. 37 Cazzuffi et al. (1999) developed a cross-plane cyclic flow device, employing the principle of flow-control to generate wave actions at a range of wave period 2 s \u00E2\u0089\u00A4 T \u00E2\u0089\u00A4 20 s and hydraulic gradient 1.5 < i < 16. The flow-controlled system uses a cylindrical piston to push and pump the water seeping through the soil specimen in manner of a constant flow volume. Stress on the soil-geotextile interface was 0 \u00E2\u0089\u00A4 \u00CF\u0083 \u00E2\u0089\u00A4 150 kPa. The permeameter had a partially flexible wall that could accommodate axial compression of the test specimen, thereby reducing the influence of sidewall friction. The test specimen was 300 mm in diameter, and 400 mm in length. A ratio of hydraulic gradient across the soil-geotextile interface to that within the base soil was used to quantify the nature of filter compatibility, in an approach similar to that of the ASTM Gradient Ratio test. Retention capacity was characterized with reference to mass loss of base soil through the geotextile. The great loss (> 5000 g/m 2 ) that led to instability at very low stress (\u00E2\u0089\u0088 0 kPa), associated with the high gradients of values greater than 3, was reported for the combination of uniform sand and woven geotextile at a filter ratio O95/D85 \u00E2\u0089\u0088 2.2. The work of Cazzuffi et al. (1999) led to development of very similar cyclic flow test devices, using a rigid-wall permeameter with slightly different dimensions (Chew et al 2000; Chen et al. 2008). In a study of wave period in the range 2s < T < 15s, Chew et al. (2000) tested a uniform sand in combination with a woven monofilament and a needle-punched nonwoven geotextile. A constant value of top stress 110 kPa was imposed on the specimen, of 315 mm diameter and 300 mm length, with no 38 consideration of sidewall friction effects governing stress at the interface. A wave period shorter than 10 s, most notably a value of 2s, was found to induce the greatest quantity of soil passing. Chen et al. (2008) examined the effects of fines content in a uniformly-graded sand. The test specimen, 330 mm in diameter and 450 mm long, was subject to relatively slow cyclic flow at 300 s < T < 600 s, at a maximum hydraulic gradient of 10, and with a top vertical effective stress of 70 to 140 kPa. Grease was used to lubricate the soil- wall interface of the permeameter in order to reduce sidewall friction. However, an effectiveness of greasing on a reduction of sidewall friction was not addressed. Fines content, up to 10% of low plasticity silt, has a significant effect on the retention stability of a soil\u00E2\u0080\u0093geotextile system. Preferential piping channels are developed from washing out of the fines through the sand, and consequently the piping of sand particles. Chen et al. (2009) used a smaller permeameter, 100 mm in diameter, to examine spatial variations of water head distribution in a plastic silt sample, 100 mm long. The specimen dimensions are identical to those of the ASTM Gradient Ratio test device. In this test series, the filter medium was a wire mesh screen of 0.5 mm in opening size. A cross-plane cyclic flow was reproduced using the principle of head-control, yielding a hydraulic gradient of 0.8, for a range of wave period 30 s < T < 3750 s. No vertical stress was applied to the top of the specimen. For wave periods shorter than 150 s, a 39 temporal hydraulic gradient developed at the filter interface was observed as greater than that within soil, and found to be influenced by the permeability of the soil. Modifications to an ASTM Gradient Ratio test apparatus, to allow for testing with cyclic flow under head-control similar to that reported by de Graauw et al. (1984), are reported by Hameiri and Fannin (2002). The device was commissioned with a series of tests on glass beads (Hameiri, 2000) and subsequently used to examine the compatibility of various geotextiles and sandy soils in tests for wave periods of 10 s and 50 s, at a constant hydraulic gradient of 4, and a top axial stress of 25 kPa and 0 kPa (Hawley, 2001). A review of findings from this latter study is provided by Srikongsri et al. (2010: see chapter 2). The influence of wave period and confining stress was identified as important to geotextile filter compatibility, and found worthy of further systematic study in order to understand the safety margin of empirical design criteria. In this study, the compatibility of soil and geotextile in cross-plane cyclic flow is examined using the small permeameter of Hameiri (2000) and the large permemater of Moffat (2005). The latter device was originally developed for study of seepage- induced internal instability of cohesionless soil in steady unidirectional flow, and therefore it was modified for the purposes of this study. A multi-stage test procedure is described, with sequences of cross-plane uni-directional and bi-directional flow, which enables characterization of soil-geotextile compatibility in cyclic flow. Use of two 40 permeameters allows for the reporting of observations on scale effect of the test device. An innovative component of the large permeameter leads to a novel approach to account for sidewall friction testing with a rigid-wall permeameter, and hence the influence of effective stress at the soil-geotextile interface. Experimental data, and a companion theoretical analysis, are presented in support of the test method and approach to data interpretation. 3.3 Test equipment 3.3.1 Small permeameter The small permeameter is a Modified Gradient Ratio device that was designed and fabricated at UBC. It is a modified version of the ASTM device that allows for the application of unidirectional or cyclic flow, imposition of an axial stress to simulate in- situ confining pressures, and collection of particles passing through the geotextile. Additional measurements of water head along the sample length allow for a more comprehensive analysis of soil/geotextile compatibility. Details of the development and features of the design are reported by Fannin et al. (1996) and Hameiri (2000). A schematic diagram of the gradient ratio device (Fig. 3.1a) illustrates the various modifications. The rigid-wall permeameter mounts on the bench top. It is made of 8 41 mm thick Plexiglas, which facilitates visual observation during testing (Fig. 3.1b). It accommodates a soil specimen of 101 mm diameter that is typically reconstituted to a length of approximately 105 mm. The specimen rests on the geotextile sample to be tested (Fig. 3.1c). The geotextile is supported on a perforated base plate, made of anodized aluminum, with 6 mm holes at 15 mm centre-to-centre spacing. A coarse wire mesh (opening size \u00E2\u0089\u0088 1.5 mm) is inserted between the geotextile and the base plate to provide for unimpeded seepage across the geotextile. A collection trough, located below the base plate, collects soil particles that pass through the geotextile sample. The collection trough comprises an upper and lower section, where the upper section is made of a Plexiglas funnel with internal slope of 45\u00EF\u0082\u00B0 that directs particles passing through the geotextile into the lower section; the lower section is made of a flexible silicon tube with an internal diameter of 19 mm to facilitate the acquisition of discrete samples at any time during the test (Fig. 3.1d). Axial loading is applied by means of a piston and top plate made of anodized aluminum, to yield a value of normal effective stress on the top of the soil specimen (Fig. 3.1d). Measurements of water head are taken at five port locations on the sidewall of the permeameter (Fig. 3.1a). Note that two additional port locations reported by Fannin et al. (1996) are redundant, and are found not necessary for purposes of data analysis and interpretation in this study. Port 1 is located on the top plate to establish the value of water head at the inlet (top of sample). Ports 3, 5 and 6 are located at 75 mm, 25 mm 42 and 8 mm above the geotextile. Port 7 is located on the upper part of the collection trough, and establishes water head at the outlet (bottom of sample). With reference to Fig. 3.1a, the value of Gradient Ratio (GR25) and a companion modified value (GR8) are calculated as: 35 57 25 i i GR \u00EF\u0080\u00BD (1) 35 67 8 i i GR \u00EF\u0080\u00BD (2) where i35 = hydraulic gradient within soil between ports 3 and 5 i57 = hydraulic gradient across soil-geotextile between ports 5 and 7 i67 = hydraulic gradient across soil-geotextile between ports 6 and 7 3.3.2 Large permeameter The large permeameter was originally designed and fabricated at UBC for testing seepage-induced internal erosion of soil as a consequence of steady unidirectional flow, with application to earth dams (Moffat, 2005). Like the small permeameter, its design enables application of axial stress on the top of the specimen. In contrast to the small permeameter, axial load is measured at the top and also at the bottom of the specimen (Moffat and Fannin, 2006). For the purposes of the current study, a geotextile is located below the soil specimen. However, the configuration of the rigid- wall cell means that no special provision can be made to collect soil particles passing 43 through the geotextile during a test. Control of the permeameter was changed to enable application of a unidirectional or cyclic flow regime. Additional ports were added to the sidewall of the permeameter for measurements of water head along the specimen length at locations directly comparable to the small gradient ratio test device. A schematic diagram of the large device (Fig. 3.2a) provides for a direct comparison with the smaller device. The large permeameter is floor-mounted. The rigid-wall cell, which seats in a steel reaction frame, is made of 13 mm thick Plexiglas that facilitates visual observation during testing (Fig. 3.2b). It accommodates a soil specimen of 280 mm diameter that was reconstituted to a length of approximately 105 mm, which is the same length used in the smaller device. As noted above, the specimen rests on the geotextile sample to be tested (Fig. 3.2c). The geotextile is supported on a perforated base plate, made of anodized aluminum, with 30 mm holes at 45 mm centre-to-centre spacings. Once again, a layer of two wire meshes, which are coarse (opening size \u00E2\u0089\u0088 5 mm) and fine (opening size \u00E2\u0089\u0088 1.5 mm), is inserted between the geotextile and the base plate to provide for unimpeded seepage across the geotextile. The base plate is part of an integral base reaction frame that seats onto a submersible load cell that is used to measure vertical stress at soil-geotextile interface (see Fig. 3.2c). In the absence of any collection trough that facilitates discrete collection of soil passing during a test, the lower chamber of the device was carefully washed at the end of a test, thereby permitting a final value to be calculated. Axial loading is applied by means of a piston and top plate made of anodized aluminum. Measurements of water head are taken at 44 five port locations on the sidewall of the permeameter (Fig. 3.2a), the locations of which reproduce those of the companion small permeameter. Accordingly, measurement of gradient ratio value is identical to that in the small permeameter. 3.3.3 Control and measurement system In the small permeameter, vertical stress (\u00CF\u0083\u00E2\u0080\u00B2vt) is applied to the top of the specimen by means of a pneumatic piston. The magnitude is controlled manually by a pressure regulator on the laboratory air-supply. Seepage flow is imposed by a hydraulic supply system that employs a principle of head-control. Three constant-head cylinders, with adjustable elevation, are used to impose either unidirectional or cyclic flow (Fig. 3.3). De-aired water is continuously supplied from a reservoir, by a peristaltic pump, to both the upper cylinder and the middle cylinder. The lower cylinder receives downward flow from the permeameter. Overflow water from all three constant-head cylinders is drained without recirculation. In the large permeameter, vertical stress (\u00CF\u0083\u00E2\u0080\u00B2vt) is similarly applied to the top of the specimen by means of a pneumatic piston that is controlled manually by a pressure regulator. Additionally, vertical stress at the bottom (\u00CF\u0083\u00E2\u0080\u00B2vb) is established from readings by the submersible load cell. The hydraulic supply system is also head-controlled. However, in contrast to the small permeameter, the upper and middle cylinders (Fig. 3.3) are replaced by two floor mounted pressure-tanks with a capacity of 45 approximately 160 liters of de-aired water. In this case, water head across the specimen that is given by the differential head between ports 1 and 7 (H17), is controlled manually by a pressure regulator that controls pressure applied to the air- membrane-water interface in the tanks. The lower cylinder is replaced with a floor- mounted overflow tank that is subject to atmospheric pressure. The flow regime used in both the small and the large permeameter is reproduced using an identical flow operation. An equidistance system water head, namely +H and \u00E2\u0080\u0093H (Fig. 3.3) is adjusted to meet a pre-determined value of average hydraulic gradient across the specimen (iav), defined as iav = H17/Z, where Z is a total length of specimen (see Figs. 3.1a and 3.1b). To impose a stage of downward unidirectional flow, seepage is controlled by water head +H across the specimen, from the middle cylinder to the lower cylinder. To impose cyclic flow, a programmable solenoid valve (3-way valve) is used to switch the direction of seepage at a pre-determined wave period (T), wherein the downward seepage controlled by +H is followed by upward seepage controlled by -H, associated with flow from the upper to the middle cylinder. Differential pressure transducers (DPT) provide a record of differential water head between each port location. A linear variable differential transformer (LVDT) is used to measure change in specimen length during a test (Figs. 3.1a and 3.2a). A data acquisition system records the output of each transducer and writes the data to storage in real-time. Discharge flow is measured periodically from the overflow of the lower cylinder (tank for the large permeameter) and used to deduce a value of hydraulic conductivity or 46 permeability. Soil passing through the geotextile may be collected at discrete intervals during soil placement and testing in the small permeameter, and at the end-of-test in the large permeameter, whereupon it is oven-dried in order to calculate mass loss per unit area. 3.4 Test methodology The test methodology consists of preparing the geotextile and soil materials, reconstituting the soil specimen against the sample of geotextile, consolidation under the initial top vertical effective stress, performing the multi-stage seepage test, and collecting the soil, if any, that passes through the geotextile. The methodology is largely adapted from Hawley (2000), with some modifications to address the specific objectives of the current study. 3.4.1 Sample preparation It is necessary in any fundamental study of soil properties and filtration compatibility involving a reconstituted specimen that the specimen preparation technique replicates a homogeneous specimen. A water pluviation technique is used in this study to create homogenous, saturated specimens of the same density. The technique is well-suited to uniform soils (Vaid & Negussey, 1986). 47 In all tests, the sample of geotextile is soaked in a bath of de-aired water and squeezed manually until there is no visual observation of air bubbles. Thereafter it is left in the bath overnight to further ensure saturation. In preparing the soil, water is added to a predetermined quantity of dry soil and the resulting slurry then boiled in a flask to remove any entrapped air. The soil for each specimen is prepared as a series of batches. The saturated soil is then allowed to cool in the flask to room temperature, which in the laboratory is consistently 23 \u00E2\u0080\u0093 24\u00C2\u00BAC. The experimental routine involves assembly of the lower part of the permeameter, and back-filling the collection trough with de-aired water until the rising surface just reaches the perforated plate and wire mesh screen. The geotextile sample to be examined in testing is then transferred on to the wire mesh, and the permeameter cell mounted in position and immediately filled with de-aired water. After checking the saturation of each thin tube that connects a manometer port to its corresponding pressure transducer, soil is pluviated from the flask into the permeameter. During pluviation, the particles reach a terminal velocity at a very small drop height, and the resulting deposit is in a very loose state. Upon completing the process of reconstitution, the top surface is levelled to a targeted specimen length of approximately 100 - 105 mm. In order to prevent any significant filtration incompatibility at the top boundary, the surface was covered with a nonwoven geotextile (AOS = 0.08 mm) and a wire mesh (opening size \u00E2\u0089\u0088 3 mm) before seating the 48 top loading platen against it. The experimental set-up is completed by assembling the upper components and connecting the top inlet-outlet valve to the middle constant- head supply. 3.4.2 Multi-stage test procedure The test procedure is intended to recreate, in a general manner, conditions found at the soil-filter interface in bank or shore protection structures. The typical wave period of wind-generated waves recommended for purposes of design is in the range 2 to 10 s (Pilarczyk, 2000; USACoE, 2002). In contrast, infragravity waves typically generate a longer wave period in the range 50 to 350 s: the waves result from processes of energy conversion of gravity waves near the coastline. In addition, wave action creates a wide range of hydraulic gradient, which may attain values as large as 10 (Giroud, 1996). Accordingly, three wave periods (T) of 6, 60 and 120 s, and three average hydraulic gradients (iav) of approximately 1, 5 and 10 were selected as test variables in the current study. The multi-stage test procedure involves consolidation of the soil specimen to a targeted value of effective stress, whereupon it is subject to several stages of unidirectional and cyclic flow. Conditions at the filter interface in bank and shore protection structures are expected to invoke relatively low values of effective stress in the range 5 to 30 kPa (e.g. a stone armor layer of a typical thickness in the range 0.5 to 49 3 m, with a submerged unit weight of 10 kN/m 3 ). Accordingly, three values of vertical effective stress (\u00CF\u0083\u00CD\u00B4 vt) on the top surface of the soil were examined in testing; in the small permeameter, values of 66, 33 and 7 kPa and, in the large permeameter, values of 57, 28 and 7 kPa. The difference in values arises from the different aspect ratio (Z/D) of the specimen in the large permeameter (\u00E2\u0089\u0088 0.3) and the small permeameter (\u00E2\u0089\u00881): the influence of sidewall friction is expected less significant in the large permeameter and the selection of values was made with the objective of imposing an equal value of effective stress at the soil-geotextile interface, for purposes of data comparison. The selection was made based on reported experience from previous use of the large permeameter (Li, 2008) and the author\u00E2\u0080\u009Fs judgment based on use of the small permeameter. The combination of minimum top stress (\u00E2\u0089\u0088 7 kPa) and maximum hydraulic gradient (\u00E2\u0089\u0088 10) is anticipated to yield a very low value of vertical stress at the soil-geotextile interface. The multistage test procedure is summarized in Fig. 3.4. In any particular test, all stages of cyclic flow are imposed at a constant value of wave period (T), yielding two test variables of effective stress (\u00CF\u0083\u00CD\u00B4 vt) and hydraulic gradient (iav). A test commences with consolidation of the specimen to a value of \u00CF\u0083\u00CD\u00B4 vt = 66 kPa in the small permeameter, else 57 kPa in large permeameter. The first stage of cyclic flow (CYC1), at value of iav \u00E2\u0089\u0088 1, is preceded and followed by a stage of unidirectional flow (UNI-1a and UNI-1b) at the same gradient iav \u00E2\u0089\u0088 1: observations made during the stage 50 of unidirectional flow are used to characterize soil-geotextile filtration compatibility before, and after, the embedded stage of cyclic flow. Upon completion of the first stage of cyclic flow, the effective stress on the top of the specimen is reduced to a value of \u00CF\u0083\u00CD\u00B4 vt = 33 kPa in the small permeameter, else 28 kPa in large permeameter, and a second stage of cyclic flow (CYC2) then imposed at the same value iav \u00E2\u0089\u0088 1: the same sequence is repeated for a third stage of cyclic flow (CYC3) at a value of \u00CF\u0083\u00CD\u00B4 vt = 7 kPa for either small permeameter. At this point in the test, the top effective stress is re-established at 66 kPa in the small permeameter, else 57 kPa in large permeameter, the value of hydraulic gradient increased to iav \u00E2\u0089\u0088 5, and the imposed flow routine of CYC1 to CYC3 then repeated (Fig. 3.4). Upon completion, the top effective stress is re-established as before, the value of hydraulic gradient increased to iav \u00E2\u0089\u0088 9, and the imposed flow routine of CYC1 to CYC3 then repeated for a third and last time. All stages of unidirectional flow down through the specimen are imposed for a minimum of 30 mins, in order to observe any spatial and temporal variation in water head distribution. All stages of cyclic flow are imposed for exactly 90 mins, yielding 900 cycles per stage at T = 6 s (a total of 8100 cycles per test), 90 cycles per stage at T = 60 s (a total of 810 cycles per test), and 45 cycles per stage at T = 120 s (a total of 405 cycles per test). Typically, a test would take 3 to 4 days to complete in the small permeameter, and 7 to 10 days in the large permeameter. During testing, there was need for a periodic interruption that was usually timed to occur at the end of a 51 particular UNI stage, in order to recharge the supply reservoir of de-aired water. This interruption has no effect on the further stage of testing. At any point in time during the test, the lower collection trough may be clamped at discrete intervals to separate and collect the soil passing through the geotextile (small permeameter only). Upon completion of the test (small and large permeameter), the soil is removed and its mass is determined. This information provides the basis of evaluating soil retention criteria for that specific geotextile/soil combination. 3.4.3 Test materials The soil used in tests reported herein is a uniformly-graded alluvial sand, termed Alouette River sand. The soil is a very fine sand, which is believed as potentially problematic for soil erosion against selected geotextile filters. It has a D85 of 0.11 mm, a uniformity coefficient (Cu) of 1.8, and contains approximately 15% of non-plastic coarse silt. Microscopic inspection reveals the particles are sub-angular in shape. It is classified as silty sand (SM) according to the unified soil classification system, and found internally stable according to the method of Kenney and Lau (1985 and 1986). A water pluviation technique yields a saturated unit weight of 18 kN/m 3 . Direct shearbox tests were conducted in order to establish values for the angle of shearing resistance, to use in calculation of a lateral stress coefficient. The tests were 52 performed on reconstituted soil specimens in a standard shearbox, 100 mm x 100 mm in plan area, and 25 mm thick. The specimen was reconstituted by air pluviation with very small drop height, to yield a loose state believed representative of the permeameter test specimen. Specimens were consolidated at a normal effective stress in the range 5 to 100 kPa, and then sheared at a constant rate of 1 mm/minute. Results for the silty sand used in testing (termed soil C, in Appendix C) include a test that was repeated. The data exhibit a non-linear relation that is attributed to the influence of stress dependency at very low stress. Similar findings are reported in the literature for other soils at very low stress (e.g. Fannin et al. 2005). The maximum angle of shearing resistance (\u00CF\u0086 m\u00CC\u0081ax ) was 38\u00C2\u00BA, 40\u00C2\u00BA and 44\u00C2\u00BA for values of 40 \u00E2\u0089\u00A4 \u00CF\u0083 n\u00CC\u0081 \u00E2\u0089\u00A4 60 kPa, 10 \u00E2\u0089\u00A4 \u00CF\u0083 n\u00CC\u0081 \u00E2\u0089\u00A4 40 kPa and \u00CF\u0083 n\u00CC\u0081 \u00E2\u0089\u00A4 10 kPa, respectively. Two geotextiles, a monofilament woven geotextile (W1) and a multifilament woven geotextile (W2), were examined in testing. These two geotextiles were designed and recommended for filtration applications involving coastal or bank erosion protection such as riprap revetment by the manufacturer (Ten Cate Geosynthetics ). Properties of the geotextiles are given in Table 3.1: in combination with the sand they yield a filter ratio AOS/D85 of 2.0 and 2.8, respectively. Details of the test program are given in Table 3.2. 53 3.5 Results A total of nine tests were performed on the soil and two geotextiles (Table 3.2). In the small permeameter, two tests at T = 6 s, namely W1-T6(S) and W2-T6(S) for which AOS/D85 \u00E2\u0089\u0088 2 and 2.8 respectively, may be compared with findings for the W2 geotextile with larger filter ratio at T = 60 and 120 s, from tests W2-T60(S) and W2- T120(S). In order to address issues of reproducibility in the experimental findings, tests W2-T6(S) and W2-T60(S) were repeated. In the large permeameter, three tests were performed, namely W2-T6(L) at T = 6 s, and W1-T60(L) and W2-T60(L) at T = 60s. These latter tests allow issues of scale effect to be addressed in interpretation of the data. 3.5.1 Hydraulic response in head-controlled system The transition in test W2-T6(S), from stages UNI-1a to CYC1 (Fig. 3.5a) and CYC1 to UNI-1b (Fig. 3.5b) is typical of the hydraulic response observed in testing with head- control of seepage flow. The data are for a top stress of 66 kPa and iav \u00E2\u0089\u0088 9, with the average gradient determined knowing H17 (see ports P1 and P7 in Figs. 3.1 and 3.2) and the specimen length (Z). Inspection shows the control system to yield an excellent transition from unidirectional to cyclic flow, and vice versa. The response during cyclic flow alone for the same W2 geotextile at T = 6 s, 60 s and 120 s (Fig. 3.6) 54 shows the short wave period of 6 s does not permit the water head across the soil- geotextile interface (H57) and within the soil (H35) to reach equilibrium (Fig. 3.6a). In contrast, there is a relatively short duration of steady flow evident at T = 60 s (Fig. 3.6b) that becomes longer at T = 120 s (Fig. 3.6c). The companion plots of water head along the specimen length depict the temporal variation over one flow reversal. It takes about 5 to 7 s to reach the equilibrium state (Figs. 3.6b and 3.6c) given the hydraulic conductivity of this soil (0.007 cm/s), which explains the absence of a steady flow regime in the data at T = 6 s (Fig. 3.6a). The response is typical of the cyclic flow in both the small and large permeameter. The variation of time to achieve equilibrium during the transient phase in each cycle of reversing flow, in soils of different permeability, has been demonstrated in the work of Hameiri and Fannin (2002). Consider the general case of a soil element, which is subjected to upward seepage flow. The vertical effective stress at any depth x, below the top surface of the element is calculated by: \u00CF\u0083\u00CD\u00B4 v = \u00CE\u00B3\u00CD\u00B4 x \u00E2\u0080\u0093 i \u00CE\u00B3w x, where = \u00CE\u00B3\u00CD\u00B4 is a submerged unit weight of the soil (kN/m 3 ), i is the hydraulic gradient across the entire soil element and \u00CE\u00B3w is a unit weight of water (kN/m 3 ). If the applied gradient exceeds a critical value, seepage flow results in zero effective stress and the action of static liquefaction results in a condition of heave and boiling. This concept applies directly to the soil specimen in the permeameter, when flow reversal yields upward seepage flow component. The term i\u00CE\u00B3wx defines the pore water arising from the hydrodynamic component, which for 55 purposes of mechanical analysis (in further section), is defined herein as the \u00E2\u0080\u009Cseepage pressure\u00E2\u0080\u009D (S). Flow reversal produces a transient variation in water head distribution and hence hydraulic gradient, along the length of the test specimen. Inspection of the test data (Fig. 3.6a to 3.6c) yields a schematic representation of the transient hydraulic gradient (itr) at the lower part of the specimen, which exceeds the value of average hydraulic gradient (iav) (see Fig. 3.6d). Over time, it diminishes to the value of iav. The greater value of transient hydraulic gradient is likely to be the cause of seepage-induced instability. However, it acts over a length (Z\u00CD\u00B4 ) shorter than the total length of the specimen (x = Z) and, the magnitude of Z\u00CD\u00B4 cannot be defined accurately due to the limitation of port location on the permeameter. Accordingly, the term iav\u00CE\u00B3wZ is used as an index parameter, in order to calculate seepage pressure for purposes of data analysis. Recognizing that the transient hydraulic gradient is greater than the average gradient, the approach is believed conservative in defining a limit to filtration incompatibility with reference to the general principle of hydromechanics. 3.5.2 Large permeameter test data In the large permeameter, onset of seepage-induced soil loss is established from visual observations. During a test, compatibility of the soil-geotextile combination is characterized from volume change of the soil specimen and from Gradient Ratio 56 values in the soil-geotextile composite zone. Mass loss, obtained after completion of a test, is also reported. At AOS/D85 \u00E2\u0089\u0088 2 and T = 60 s for test W1-T60(L), visual observations indicated no loss of soil in any stage of unidirectional (UNI) flow and, similarly, no loss in any cyclic (CYC) stage. No volume change was observed during the test, and therefore the cumulative mass loss of 94 g/m 2 recorded at the end is attributed only to loss during reconstitution of the soil specimen against the geotextile. Values of GR25 \u00E2\u0089\u0088 1.1 and GR8 \u00E2\u0089\u0088 1.2, obtained in all UNI stages of the test, indicate a variation of water head distribution that is essentially linear across the length of the specimen. Accordingly, the test combination is deemed stable in both unidirectional and cyclic flow. At AOS/D85 \u00E2\u0089\u0088 2.8 and T = 60 s for test W2-T60(L), visual observations again established no soil loss during any of the UNI stages. In the CYC stages, no losses occurred at iav \u00E2\u0089\u0088 1, however small losses were noted for all CYC stages with iav \u00E2\u0089\u0088 5 and found to increase at iav \u00E2\u0089\u0088 9. The losses occurred as a pulsating action that was not restricted to any preferential location on the geotextile. The action was found transient, since each cycle of flow reversal yielding downward seepage was associated with the onset a soil loss that continued for approximately 7 s to 15 s, after which it diminished quickly to a negligible quantity. A volume change of 0.5% occurred over the duration of the test, associated with a cumulative mass loss of 959 g/m 2 that includes the component for specimen reconstitution. Values of 0.9 \u00E2\u0089\u00A4 GR25 \u00E2\u0089\u0088 GR8 \u00E2\u0089\u00A4 57 1.0 indicate a linear variation of head loss across the soil-geotextile composite zone. The results show the test combination to be stable in unidirectional flow, but provide evidence of a transition from stable to unstable response in cyclic flow. At AOS/D85 \u00E2\u0089\u0088 2.8 and T = 6 s for test W2-T6(L), there was no mass loss in any of the UNI stages, however it did occur in all of the CYC stages. Rather than pulsating and transient, the action was found continuous at this shorter wave period. As for test W2- T60(L), the losses were observed across the entire surface area of geotextile, with no evidence of any preferential location. Observations suggest the losses increased with application of larger hydraulic gradient, and increased with reduction in effective stress applied to the top surface of the specimen. No preferential flow channel developed within the body of soil specimen. A volume change of 3% occurred over the duration of the test, associated with a cumulative mass loss of 5487 g/m 2 , which again includes that for specimen reconstitution. Values of GR25 = GR8 = 1 confirm a linear head loss across the soil-geotextile composite zone that appears common to all three tests, and is attributed to the uniform gradation of the soil. The test combination is deemed stable in unidirectional flow, but unstable in cyclic flow. Axial load on the top and bottom surface of the test specimen are measured directly in the large permeameter. For the hydrostatic condition, a top vertical effective stress (\u00CF\u0083\u00E2\u0080\u00B2vt(0)) of 57 kPa, 28 kPa and 7 kPa yielded values at the base (\u00CF\u0083\u00E2\u0080\u00B2vb(0)) in the range 45 to 48 kPa, 22 to 24 kPa, and 4 to 6 kPa, respectively. In stages of unidirectional flow, 58 the downward seepage force resulted in a stress increase at the base of approximately 0.2 to 0.4 kPa, 2 to 3 kPa, and 4 to 5 kPa for values of iav \u00E2\u0089\u0088 1, 5 and 9, respectively. In stages of cyclic flow at T = 60 s, upward flow yielded a corresponding reduction of similar magnitude in \u00CF\u0083\u00E2\u0080\u00B2vb. Consequently, stress at the soil-geotextile interface is close to zero in the CYC3 stage when \u00CF\u0083\u00E2\u0080\u00B2vt(0) = 7 kPa at iav \u00E2\u0089\u0088 9 (see Fig. 3.7). Stress loss in the system that is influenced by sidewall friction is found to be about 20% to 45%. Stages of cyclic flow at T = 6s did not provide useful data on axial load at the base, because the reversing action is too fast. 3.5.3 Small permeameter test data In the small permeameter, onset of seepage-induced losses is based not only on visual observations, but also on values of mass loss for each individual stage of loading. The additional data enable a more precise characterization of filtration compatibility. At AOS/D85 \u00E2\u0089\u0088 2 and T = 6 s for test W1-T6(S), visual observations indicated no soil loss in any stage of unidirectional (UNI) or cyclic (CYC) flow. The finding is consistent with no volume change of the specimen. Only a mass loss of 65 g/m 2 was recorded during reconstitution of the soil specimen. Values of GR25 and GR8 in the range 1.2 to 1.5 indicate a nearly linear variation of water head throughout the test. Accordingly, the test combination is considered stable in both unidirectional and cyclic flow. 59 At AOS/D85 \u00E2\u0089\u0088 2.8 and T = 6 s for test W2-T6(S), mass loss occurred in none of the UNI stages, but all of the CYC stages. At iav \u00E2\u0089\u0088 1, the amount was reasonably similar (mav \u00E2\u0089\u0088 450 g/m 2) for each value of top stress (\u00CF\u0083\u00E2\u0080\u00B2vt = 66 kPa, 33 kPa and 7 kPa). Raising the gradient to iav \u00E2\u0089\u0088 5 led to increased loss (mav \u00E2\u0089\u0088 800 to 900 g/m 2 ). At the maximum gradient iav \u00E2\u0089\u0088 9, the loss exceeds 1000 g/m 2. Reducing \u00CF\u0083\u00E2\u0080\u00B2vt influences mass loss, particularly at 7 kPa where stress at the soil-geotextile interface is believed close to zero (see Table 3.3a). Although the losses were continuous, rather than pulsating, no preferential flow channels were evident in the soil. A volume change of 3.1 % during the test was associated with a cumulative mass loss of 7650 g/m 2 . Of note is the finding that soil piping in each CYC stage did not affect the gradient ratio in the following UNI stage, for which values of GR25 and GR8 in the range 0.9 to 1 were obtained. The test combination is deemed stable in unidirectional flow, but unstable in cyclic flow. At AOS/D85 \u00E2\u0089\u0088 2.8 and T = 60 s for test W2-T60(S), collection of soil passing through the geotextile confirmed visual observations of no loss during any of the UNI stages, but occurrence of loss in all the CYC stages. The loss was very small (mav \u00E2\u0089\u00A4 35 g/m 2 ) at iav \u00E2\u0089\u0088 1. Raising the gradient to iav \u00E2\u0089\u0088 5 nearly doubled the loss, and raising it again to iav \u00E2\u0089\u0088 9 produced another doubling of losses to a value of 115 g/m 2 (see Table 3.3b). Values of 0.9 \u00E2\u0089\u00A4 GR25 \u00E2\u0089\u0088 GR8 \u00E2\u0089\u00A4 1.0 are comparable to those of test W2-T6(S). At constant gradient, mass loss was found to increase with decreasing effective stress. As 60 for the large permeameter, losses occurred as a pulsating action that was not restricted to any preferential location on the geotextile. A volume change of 0.3 % during the test was associated with a cumulative mass loss of 473 g/m 2 . The results show the test combination to be stable in unidirectional flow, but provide further evidence of a susceptibility to instability in cyclic flow like that reported from the companion W2- T60(L) test. At AOS/D85 \u00E2\u0089\u0088 2.8 and T = 120 s for test W2-T120(S), no visual observation was made of soil loss through the geotextile in UNI or CYC stages of flow. In CYC flow, no loss was measured at iav \u00E2\u0089\u0088 1, and very little mass loss (mav \u00E2\u0089\u00A4 26 g/m 2 ) at iav \u00E2\u0089\u0088 5 and 9. In this regard, it appears the long wave period replicates behaviour found in unidirectional flow. A volume change less than 0.1% was associated with a very small cumulative mass loss of 83 g/m 2 (see Table 3.3b). Values of 1.1 \u00E2\u0089\u00A4 GR25 \u00E2\u0089\u0088 GR8 \u00E2\u0089\u00A4 1.3 imply a linear variation of head loss. Stability is again confirmed in unidirectional flow. In cyclic flow, it appears stable at relatively low hydraulic gradients, but may be susceptible at the highest gradient. 3.5.4 Reproducibility of findings Test results at a filter ratio AOS/D85 \u00E2\u0089\u0088 2.8 are used to examine issues of repeatability and scale effect in the laboratory data. Reproducibility of the findings is considered with reference to a qualitative assessment of general mass loss trends, and a 61 quantitative assessment of specific mass loss quantities and related volume change. Two tests were repeated in small permeameter, for T = 6 s and T = 60 s respectively. Tests W2-T6(S) and W2-T6(S)-R exhibit a similar trend of mass loss in response to change in gradient and vertical effective stress, and specific values of mass loss are also similar in comparable stages of testing (see Table 3.3a). For test W2-T6(s)-R (repeated test), a total volume change of 4.2 % during the test was associated with a cumulative mass loss of 6075 g/m 2 . Tests W2-T60(S) and W2-T60(S)-R also exhibit a similar trend of mass loss, however the variation in specific values of mass loss is up to 40 % (see Table 3.3b). The difference is attributed the spatial variation of pore size opening in the fabric of the geotextile, from which each sample is selected, which is sufficient to yield a difference in the \u00E2\u0080\u009Clocal\u00E2\u0080\u009D response, but insufficient to influence the \u00E2\u0080\u009Cgeneral\u00E2\u0080\u009D response. A total volume change of 0.3 % during the test was associated with a cumulative mass loss of 287 g/m 2 . 3.5.5 Scale effect Scale effect manifests itself as a difference in cross-sectional area of the test specimen, and also volume of soil, examined in the large and small permeameter. Water head distribution in unidirectional flow is first used for evaluation of reproducibility, and the excellent agreement (see Fig. 3.8 for example) is very encouraging. This is reflected directly in values of gradient ratio, which also show good agreement (Fig. 3.9). 62 In considering reproducibility with reference to mass loss and volume change, focus is placed on the data for tests at T = 6 s which experienced a relatively large number of flow cycles (8100 in total) and therefore provide a sufficiently large volume change. In contrast, the data for T = 60 s (810 cycles) and 120 s (405 cycles) provide a relatively small volume change. In all three tests, volume change increases linearly with increasing cumulative mass loss for this uniform sand (see Fig. 3.10). Data for the large permeameter test plot between those of the repeated test in the small permeameter. Recall the cross-sectional area of test specimen in the large permeameter is approximately eight times greater than that in the small device. It is reasonable to expect the larger specimen would exhibit a response indicative of the average found in the smaller specimens, given the influence of spatial variations in pore size opening. The data support this expectation and, it is therefore reasonable to conclude that the two permeameter devices exhibit no significant scale effect. Accordingly, it may be necessary to repeat tests in the small permeameter that demonstrate seepage-induced soil piping through the geotextile. Comparable quantity of mass loss from the repeated test is used to ensure that the geotextile samples accommodate spatial variations in pore size opening and hence represent a larger sample size. 63 3.6 Stress in a rigid-wall permeameter Stress exerts a controlling influence on filtration behaviour, especially in reversing or cyclic flow. Therefore, it is important to account for it in a mechanics-based interpretation of results in these filtration tests. Measured values of top and bottom stress in the large permeameter indicate a strong influence of sidewall friction, and a significant reduction in stress at the soil-geotextile interface. Accordingly, it is important to correct for this influence in the small device (for which the design yields a value of top stress only), because the ability of the small permeameter to provide multi-stage data on soil loss is instrumental to the main focus of the study. In this section, a simple force equilibrium approach is presented to account for the influence of sidewall friction, from interpretation of axial force measurements in both permeameters. 3.6.1 Vertical stress distribution Taking a stress equilibrium at the base of the specimen, the relation between top vertical effective stress (\u00CF\u0083\u00C2\u00B4vt) and basal stress (\u00CF\u0083\u00C2\u00B4vb(0)) on the specimen, for the hydrostatic condition (no seepage) (Fig. 3.11a) is given by: )0()0( vvtvb Z \u00EF\u0081\u00B3\u00EF\u0081\u00A7\u00EF\u0081\u00B3\u00EF\u0081\u00B3 \u00EF\u0082\u00A2\u00EF\u0081\u0084\u00EF\u0080\u00AD\u00EF\u0082\u00A2\u00EF\u0080\u00AB\u00EF\u0082\u00A2\u00EF\u0080\u00BD\u00EF\u0082\u00A2 (3) 64 where \u00CE\u00B3\u00C2\u00B4 is the buoyant unit weight of the soil (kN/m3), \u00CE\u0094\u00CF\u0083\u00E2\u0080\u00B2v(0) is the stress difference caused by sidewall friction (kPa) and Z is specimen length (m). Downward flow yields a stress increase at the base (Fig 3.11b) while, in contrast, upward seepage flow yields a decrease (Fig. 3.11c). Thus, vertical stress at the base for the hydrodynamic condition (seepage flow) \u00CF\u0083\u00C2\u00B4vb is given by: SZ vvtvb \u00EF\u0082\u00B1\u00EF\u0082\u00A2\u00EF\u0081\u0084\u00EF\u0080\u00AD\u00EF\u0082\u00A2\u00EF\u0080\u00AB\u00EF\u0082\u00A2\u00EF\u0080\u00BD\u00EF\u0082\u00A2 \u00EF\u0081\u00B3\u00EF\u0081\u00A7\u00EF\u0081\u00B3\u00EF\u0081\u00B3 (4) where \u00CE\u0094\u00CF\u0083\u00E2\u0080\u00B2v is the stress difference (kPa), S is seepage pressure applied at the base (kPa) taken equal to iav\u00CE\u00B3wZ, where iav is average hydraulic gradient across the specimen (negative for upward flow), and \u00CE\u00B3w is the unit weight of water (kN/m 3 ). Li (2008) simplified the solution (Eq. 3 and 4) by assuming a constant value of sidewall friction along the specimen, establishing the mean vertical effective stress at the mid-height of the specimen for hydrostatic (\u00CF\u0083\u00E2\u0080\u00B2vm(0)) and hydrodynamic conditions (\u00CF\u0083\u00E2\u0080\u00B2vm), respectively, as: )(5.0)0( vvtvm Z \u00EF\u0081\u00B3\u00EF\u0081\u00A7\u00EF\u0081\u00B3\u00EF\u0081\u00B3 \u00EF\u0082\u00A2\u00EF\u0081\u0084\u00EF\u0080\u00AD\u00EF\u0082\u00A2\u00EF\u0080\u00AB\u00EF\u0082\u00A2\u00EF\u0080\u00BD\u00EF\u0082\u00A2 (5a) )(5.0 SZ vvtvm \u00EF\u0082\u00B1\u00EF\u0082\u00A2\u00EF\u0081\u0084\u00EF\u0080\u00AD\u00EF\u0082\u00A2\u00EF\u0080\u00AB\u00EF\u0082\u00A2\u00EF\u0080\u00BD\u00EF\u0082\u00A2 \u00EF\u0081\u00B3\u00EF\u0081\u00A7\u00EF\u0081\u00B3\u00EF\u0081\u00B3 (5b) 65 The mean vertical effective stress was used by Li (2008) for mechanical analysis of internal soil erosion for the specimen tested in the large permeameter, and was also recommended for sidewall friction analysis. 3.6.2 Influence of sidewall friction Sidewall shear is governed by lateral stress and a value of interface friction. The \u00E2\u0080\u009Eat- rest\u00E2\u0080\u009F condition is commonly assumed for a soil element in a rigid-walled device that inhibits development of lateral strain, primarily from studies on sidewall resistance in consolidation test cells (for example, Shirato et al. 1968 and, more recently Sivrikaya and Togol, 2005). Lateral strain in the rigid-wall permeameter is believed negligible, and leads to a similar assumption of K0 for the lateral stress coefficient. Thus, average sidewall shear (\u00CF\u0084av) may be calculated with reference to mean vertical effective stress: \u00EF\u0080\u00A8 \u00EF\u0080\u00A9vmvmav orKfc \u00EF\u0081\u00B3\u00EF\u0081\u00B3\u00EF\u0081\u00B4 \u00EF\u0082\u00A2\u00EF\u0082\u00A2\u00EF\u0080\u00AB\u00EF\u0080\u00BD )0(0 (6) where c is a value of soil-wall adhesion (assumed zero in this study of cohesionless soil), and f is a coefficient of soil-wall interface friction. In theory, an upper-bound value of f \u00E2\u0089\u00A4 tan\u00CE\u00B4 is mobilized with development of large relative displacement between the soil specimen and inside wall of the permeameter. However, in practice, 66 this does not occur over the full length of the specimen because of the zero displacement boundary on which it rests. Considering force equilibrium (see Fig. 3.12) for the hydrostatic (\u00CE\u0094\u00CF\u0083\u00E2\u0080\u00B2v(0)) and the hydrodynamic condition (\u00CE\u0094\u00CF\u0083\u00E2\u0080\u00B2v) respectively, and introducing the aspect ratio (Z/D), yields: fK D Z D Z vmavv 0)0()0( 44 \u00EF\u0081\u00B3\u00EF\u0081\u00B4\u00EF\u0081\u00B3 \u00EF\u0082\u00A2\u00EF\u0080\u00BD\u00EF\u0080\u00BD\u00EF\u0082\u00A2\u00EF\u0081\u0084 (7a) fK D Z vmv 0 4 \u00EF\u0081\u00B3\u00EF\u0081\u00B3 \u00EF\u0082\u00A2\u00EF\u0080\u00BD\u00EF\u0082\u00A2\u00EF\u0081\u0084 (7b) The test procedure of applied stress and cyclic flow during a multistage test imposes load-unload-reload sequences (see Fig. 3.4). Unloading at zero lateral strain affects the magnitude of K0, which increases with value of over-consolidation ratio (OCR) in the specimen (Campanella and Vaid 1972; Mayne and Kulhawy 1982; Mayne and Kulhawy 1994). Mayne and Kulhawy (1982) suggested a lateral stress coefficient at-rest (K0) for unloading that is correlated with OCR and internal friction angle (\u00CF\u0085), based on analysis of test data from 81 clays and 90 sands, expressed as: 67 \u00EF\u0081\u00A6\u00EF\u0081\u00A6 sin0 )sin1( OCRK \u00EF\u0080\u00AD\u00EF\u0080\u00BD (8) In the rigid-wall permeameter, and with reference to mean vertical effective stress, OCR is defined as: vm vm OCR \u00EF\u0081\u00B3 \u00EF\u0081\u00B3 \u00EF\u0082\u00A2 \u00EF\u0082\u00A2 \u00EF\u0080\u00BD max, (9) where \u00CF\u0083\u00E2\u0080\u00B2vm,max is the maximum mean vertical effective stress experienced by the test specimen, and \u00CF\u0083\u00E2\u0080\u00B2vm is mean the vertical effective stress at the current stage of testing. 3.6.3 Sidewall friction: large permeameter Data from two tests in the large permeameter establish values of soil-wall interface friction coefficient that may be used in analysis and interpretation of data from the small permeameter. For the hydrostatic condition, stress difference (\u00CE\u0094\u00CF\u0083\u00E2\u0080\u00B2v(0)) is deduced from measurement of top stress, bottom stress and self weight of the specimen (see Eq. 3). For the hydrodynamic condition, and knowing the seepage pressure, the stress difference (\u00CE\u0094\u00CF\u0083\u00E2\u0080\u00B2v) is similarly deduced (see Eq. 4). In the absence of a condition of stress equilibrium fully developing in test W2-T6(L), as a consequence of the fast 68 cyclic flow, data from tests W1-T60(L) and W2-T60(L) only are used for purposes of stress analysis, and deduction of mobilized soil-wall interface friction. 3.6.3.1 Hydrostatic condition The calculation is made in four steps: (a) determine the mean vertical effective stress (\u00CF\u0083\u00E2\u0080\u00B2vm(0)) for \u00CF\u0083\u00C2\u00B4vt = 57 kPa, 28 kPa and 7 kPa using Eq. 5a; (b) from known values of \u00CE\u0094\u00CF\u0083\u00E2\u0080\u00B2v(0), calculate values of K0f using Eq. 7a (see Fig. 3.13a and 3.13b); (c) knowing OCR = 1 at \u00CF\u0083\u00C2\u00B4vt = 57 kPa, and recognizing that \u00CF\u0083\u00E2\u0080\u00B2vm(0) at \u00CF\u0083\u00C2\u00B4vt = 57 kPa is the value of \u00CF\u0083\u00E2\u0080\u00B2vm(0),max, determine OCR values (see Eq. 9) for unloading to \u00CF\u0083\u00C2\u00B4vt = 28 kPa (OCR = 2.1) and 7 kPa (OCR = 9.1); (d) calculate K0 (see Eq. 8) with \u00CF\u0086 = 38\u00CB\u009A, 40\u00CB\u009A and 44\u00CB\u009A for \u00CF\u0083\u00C2\u00B4vt = 57 kPa, 28 kPa and 7 kPa (see section 3.3.3), from which values of f may be deduced. The average value obtained for the tests without seepage flow is f = 0.32, 0.23 and 0.18, for \u00CF\u0083\u00C2\u00B4vt = 57 kPa, 28 kPa and 7 kPa, respectively (see Fig. 3.14, at S/\u00CE\u0094\u00CF\u0083\u00E2\u0080\u00B2v(0) = 0). 3.6.3.2 Downward seepage flow The calculation also consists of four steps: (a) knowing seepage pressure (S = iav\u00CE\u00B3wZ), calculate \u00CF\u0083\u00E2\u0080\u00B2vm (see Eq. 5b) in order to determine an OCR value for the nine combinations of \u00CF\u0083\u00C2\u00B4vt = 57 kPa, 28 kPa and 7 kPa, and iav \u00E2\u0089\u0088 1, 5 and 9; (b) from known values of \u00CE\u0094\u00CF\u0083\u00E2\u0080\u00B2v, calculate values of K0f using Eq. 7b (see Fig. 3.13a and 3.13b); (c) at a 69 given hydraulic gradient, and corresponding \u00CF\u0083\u00E2\u0080\u00B2vm, determine OCR (see Eq. 9) for CYC1 (note that \u00CF\u0083\u00E2\u0080\u00B2vm = \u00CF\u0083\u00E2\u0080\u00B2vm,max, and OCR = 1), CYC2 and CYC3; (d) knowing the value of OCR, calculate K0 with reference to the same values of friction angle (\u00CF\u0086 ) for the soil (see Eq. 8), allowing f to be deduced for different seepage conditions (see Fig. 3.14a). 3.6.3.3 Upward seepage flow The calculation steps to deduce a value of f are identical to those for the case of downward flow, except for the need to account for a seepage pressure (-S = -iav\u00CE\u00B3wZ) that is negative in the first step of the calculation. The remaining three steps again enable a deduction of f for different seepage conditions (see Fig. 3.14b). 3.6.3.4 Coefficient of sidewall friction (f) The relation between f and a normalized measure of seepage flow (see Fig. 3.14), from these large permeameter test data, depicts the combined influence of stress difference and hydraulic gradient. With downward flow (see in Fig. 3.14a), the value of f appears more sensitive to \u00CF\u0083\u00E2\u0080\u00B2vt than magnitude of seepage flow. In contrast, upward flow (see Fig. 3.14b) yields a significant reduction in the magnitude of sidewall friction: negative values indicate a reversal of the direction is which sidewall friction is 70 mobilized at relatively large seepage flow. The response is consistent with observations reported by Li (2008) for the same test device. The variety of seepage conditions examined in testing yield a range of -0.4 \u00E2\u0089\u00A4 f \u00E2\u0089\u00A4 0.4 for the coefficient of interface friction, where f \u00E2\u0089\u00A4 tan \u00CE\u00B4. Proportionally, this represents between \u00C2\u00BD and \u00E2\u0085\u0094 tan \u00CF\u0085 for the friction angle of 44\u00CB\u009A, 40\u00CB\u009A and 38\u00CB\u009A measured for the soil in direct shear box tests (section 3.3.3). The relation between f and S/\u00CE\u0094\u00CF\u0083\u00E2\u0080\u00B2v(0) established from analysis of tests in the large permeameter, Figure 3.14, is assumed appropriate for stress analysis of tests data in the small permeameter. The assumption is made based on the use of an acrylic cell in each device. 3.6.4 Sidewall friction: small permeameter In contrast to the large permeameter, effective stress at the soil-geotextile interface (\u00CF\u0083\u00E2\u0080\u00B2vb) must be deduced for the small permeameter, since there is no provision to measure axial force on the lower boundary of the test specimen. Accordingly, the objective of the following stress analysis is to deduce a value of \u00CF\u0083\u00E2\u0080\u00B2vb for the range of applied stress on the top boundary, namely 66 to 7 kPa. Note that the values of f deduced from the range 57 to 7 kPa in the large permeameter are assumed applicable to this range. 71 Since the value of stress difference arising from sidewall friction is unknown in the small permeameter, the mean effective stress for hydrostatic (\u00CF\u0083\u00E2\u0080\u00B2vm(0)) and hydrodynamic conditions (\u00CF\u0083\u00E2\u0080\u00B2vm) cannot be calculated directly by Eq. 5a or Eq. 5b. Thus, combining Eq. 7a with Eq. 5a and similarly Eq. 7b with Eq. 5b, yields the following new expressions: \u00EF\u0083\u00B7 \u00EF\u0083\u00B8 \u00EF\u0083\u00B6 \u00EF\u0083\u00A7 \u00EF\u0083\u00A8 \u00EF\u0083\u00A6 \u00EF\u0080\u00AB \u00EF\u0082\u00A2\u00EF\u0080\u00AB\u00EF\u0082\u00A2 \u00EF\u0080\u00BD\u00EF\u0082\u00A2 D ZfK Zvt vm 0 )0( 2 1 )(5.0 \u00EF\u0081\u00A7\u00EF\u0081\u00B3 \u00EF\u0081\u00B3 (10a) \u00EF\u0083\u00B7 \u00EF\u0083\u00B8 \u00EF\u0083\u00B6 \u00EF\u0083\u00A7 \u00EF\u0083\u00A8 \u00EF\u0083\u00A6 \u00EF\u0080\u00AB \u00EF\u0082\u00B1\u00EF\u0082\u00A2\u00EF\u0080\u00AB\u00EF\u0082\u00A2 \u00EF\u0080\u00BD\u00EF\u0082\u00A2 D ZfK SZvt vm 021 )(5.0 \u00EF\u0081\u00A7\u00EF\u0081\u00B3 \u00EF\u0081\u00B3 (10b) The stress analysis for the small permeameter is made in reverse order to that for the large permeameter (see Fig. 3.15): OCR is first obtained, and used to determine a value of K0 that, knowing f (from Fig. 3.14, else Table 3.4), enables a calculation of mean vertical stress (\u00CF\u0083\u00E2\u0080\u00B2vm(0), or \u00CF\u0083\u00E2\u0080\u00B2vm) and hence stress difference (\u00CE\u0094\u00CF\u0083\u00E2\u0080\u00B2v(0), or \u00CE\u0094\u00CF\u0083\u00E2\u0080\u00B2v), which then allows for a calculation of bottom stress (\u00CE\u0094\u00CF\u0083\u00E2\u0080\u00B2vb(0), or \u00CE\u0094\u00CF\u0083\u00E2\u0080\u00B2vb). 72 3.6.4.1 Hydrostatic condition Consider, for illustrative purposes, a calculation for the case of \u00CF\u0083\u00C2\u00B4vt = 33 kPa. The calculation is made in five steps: (a) for \u00CF\u0083\u00C2\u00B4vt = 66 kPa, calculate K0 from Eq. 8 (with \u00CF\u0086 = 38\u00C2\u00BA, and OCR = 1); (b) calculate \u00CF\u0083\u00E2\u0080\u00B2vm(0) from Eq. 10a (with f = 0.32, and noting \u00CF\u0083\u00E2\u0080\u00B2vm(0) = \u00CF\u0083\u00E2\u0080\u00B2vm(0),max); (c) for \u00CF\u0083\u00C2\u00B4vt = 33 kPa, assume a trial value of OCR (= 66/33 \u00E2\u0089\u0088 2) and calculate K0 (with \u00CF\u0086 = 40\u00CB\u009A, see section 3.3.3), and thereby determine \u00CF\u0083\u00E2\u0080\u00B2vm(0) from Eq. 10a (with f = 0.26); (d) calculate the OCR from Eq. 9 and compare with the assumed trial value (of step c), using iteration as a necessary to obtain agreement; (e) knowing \u00CF\u0083\u00E2\u0080\u00B2vm(0), calculate \u00CE\u0094\u00CF\u0083\u00E2\u0080\u00B2v(0) from Eq. 7a, and deduce the bottom stress from Eq. 3. 3.6.4.2 Downward seepage flow Like the hydrostatic condition, \u00CF\u0083\u00E2\u0080\u00B2vm is unknown for stages with OCR greater than 1, until a value of K0 is determined. Furthermore, OCR cannot be approximated from the top stress alone, or calculated from Eq. 9, rather a trial value must be established from: )5.0( )0( max, S OCR vm vm \u00EF\u0082\u00B1 \u00EF\u0082\u00BB \u00EF\u0081\u00B3 \u00EF\u0081\u00B3 (11) Consider, for illustrative purposes, a calculation for the CYC2 stage (\u00CF\u0083\u00C2\u00B4vt = 33 kPa) at iav = \u00C2\u00B1 9 (S \u00E2\u0089\u0088 8.8 kPa). In this case, use the value of \u00CE\u0094\u00CF\u0083\u00E2\u0080\u00B2v(0), calculated earlier, to define 73 an S/\u00CE\u0094\u00CF\u0083\u00E2\u0080\u00B2v(0) ratio that yields f from Fig. 3.14a, for conditions of downward flow. The calculation is made in five steps: (a) for \u00CF\u0083\u00C2\u00B4vt = 66 kPa and S = +8.8 kPa, calculate K0 from Eq. 8 (with \u00CF\u0086 = 38\u00C2\u00BA, and OCR = 1); (b) calculate \u00CF\u0083\u00E2\u0080\u00B2vm from Eq. 10b (noting \u00CF\u0083\u00E2\u0080\u00B2vm = \u00CF\u0083\u00E2\u0080\u00B2vm,max); (c) for \u00CF\u0083\u00C2\u00B4vt = 33 kPa, assume a trial value of OCR (Eq. 11) and calculate K0 (with \u00CF\u0086 = 40\u00CB\u009A), and thereby determine \u00CF\u0083\u00E2\u0080\u00B2vm from Eq. 10b; (d) calculate the OCR from Eq. 9 and compare with the assumed trial value (of step c), using iteration as a necessary to obtain agreement; (e) knowing \u00CF\u0083\u00E2\u0080\u00B2vm, calculate \u00CE\u0094\u00CF\u0083\u00E2\u0080\u00B2v from Eq. 7b, and deduce the bottom stress from Eq. 4. 3.6.4.3 Upward seepage flow The calculation steps to deduce a value of f are identical to those for the case of downward flow, except for the need once again to account for a seepage pressure (-S = -iav\u00CE\u00B3wZ) that is negative in the first step of the calculation: \u00CF\u0083\u00E2\u0080\u00B2vm,max has the same magnitude as that for downward flow. Importantly to note that Eq. 11 is invalid for a condition yielding a value 0.5S close to or in excess of \u00CF\u0083\u00E2\u0080\u00B2vm(0). There is only at the gradient \u00E2\u0089\u0088 9 (0.5S \u00E2\u0089\u0088 4.4 kPa) for the stage CYC3 (\u00CF\u0083\u00C2\u00B4vt = 7 kPa and \u00CF\u0083\u00E2\u0080\u00B2vm(0) = 4.6 kPa) falling into this case. Thus, a trial OCR value of 10 is firstly assumed for OCR in the step c (see 3.6.4.2). The agreement from iteration is obtained at the OCR value of 14. At this stage, the calculation yields a bottom stress of about 0 kPa. 74 In summary, a corresponding vertical stress at the base or at the soil-geotextile interface (\u00CF\u0083\u00C2\u00B4vb ) is calculated for each measured value of top applied stress (\u00CF\u0083\u00C2\u00B4vt ) (see Fig. 3.16). Inspection shows \u00CF\u0083\u00C2\u00B4vb to diminish with increasing iav (negative value depicts upward flow). The relation between \u00CF\u0083\u00C2\u00B4vt and \u00CF\u0083\u00C2\u00B4vb appears non-linear, as a result of the influence of unloading, and therefore OCR, on K0. For the same top stress, the calculations yield a value of interface stress for the small permeameter approximately 20 % lower than that for the large permeameter. The finding is attributed to the different aspect ratio of the respective test specimens. A summary of input parameters to the stress calculation, and resulting values deduced for f, is provided in Table 3.4. 3.7 Discussion 3.7.1 Size of geotextile sample Size of the geotextile sample is found to influence quantity of mass loss per unit area, which differs in the small and large permeameter, a behaviour that is attributed to spatial variation of material properties. In contrast, no companion variation of water head distribution was observed at the soil-geotextile interface, as evident from similar values of hydraulic gradient ratio. The response is attributed to the uniform gradation of the soil. More generally, mass loss per unit area and consequent volume change are comparable, a finding that implies the system response in the small permeameter 75 replicates that in the large permeameter. Furthermore, a limited comparison of data indicates the large permeameter likely defines an average or typical response of the small permeameter (see Fig. 3.10). For this reason, it is recommended to repeat a test in the small permeameter where it is believed the soil and geotextile exhibit filtration incompatibility, and report an average of the experimental findings for purposes of analysis. 3.7.2 Influence of test procedure The multi-stage test procedure is intended to challenge soil-geotextile compatibility, by means of reduced stress and increasing hydraulic gradient as the test progresses. As long as the specimen remains intact, it is believed the influence of stress and gradient can be examined in one test specimen. As shown by results at AOS/D85 = 2.8 in both the large and small permeameter, the GR25 and GR8 values obtained in unidirectional flow, after a cyclic stage in which some piping or mass loss occurred, are unchanged from the initial values. It indicates the readiness of the test specimen for testing in the next stage: the finding is expected for a uniformly-graded soil. This protocol is very important to multi-stage testing, because the influence of these variables is examined by continued testing of the same geotextile sample. In this manner, test variables are examined without any spatial variation in geotextile properties. Furthermore it is efficient, in a systematic study of a large number of test combinations, to employ a multi-stage test procedure. Accordingly, the test procedure is believed well-suited to 76 testing of uniformly-graded soil, and is recommended for a systematic study of soil- geotextile compatibility that addresses stress and hydraulic gradient as well as wave period in cyclic flow reversal. The loading routine, with load-unload-reload sequence, imposes a stress history on test specimen. Accordingly, the relation between vertical stress and lateral stress, which varies with stress history, has been addressed with reference to variation of mean vertical effective stress (see Fig. 3.17a) taking into account observations of Mayne and Kulhawy (1982). The point of \u00CF\u0083\u00CD\u00B4 vm,max occurs for downward flow during the CYC1 stage at maximum gradient, and the point of \u00CF\u0083\u00CD\u00B4 vm,min occurs for upward flow during the CYC3 stage at maximum gradient. Results for the small permeameter (see Fig. 3.17b) are provided for comparison: they underscore the difference between lateral stress and vertical stress in the rigid-wall permeameter. Inspection shows the values are comparable when vertical stress rebounds to a certain value, and shows that lateral stress exceeds vertical stress at relatively low values of vertical stress. The finding is meaningful because, if lateral stress influences a stress-based interpretation, then it must be addressed in data analysis and interpretation. 3.7.3 Significance of lateral stress Particle movement within a porous medium is believed governed by confining stress in a plane normal to the direction of movement (Indraratna and Vafai 1997; Indraratna 77 and Radampola 2002). The configuration of the laboratory permeameter imposes one- dimensional vertical flow in either a downward or upward direction. Thus, the confining stress normal to the direction of movement is lateral stress. Theory and analysis suggest that lateral stress in a rigid-wall permeameter may be much lower than vertical stress when downward seepage occurs under a condition of normal consolidation and, in contrast, it may be much greater when upward seepage pressure reduces the stress level to yield a relatively large OCR (see Fig. 3.17). Accordingly, vertical stress alone is not believed sufficiently representative of the actual confining stress, and could result in misleading interpretations. For this reason, lateral stress should be accounted for in any mechanics-based analysis, and is explicitly addressed in this study. More specifically, mean effective stress is proposed as an index value to account for the influence of vertical and lateral stress in the rigid-wall permeameter. At the soil- geotextile interface, it is taken as pi = \u00CF\u0083\u00E2\u0080\u00B2vb[1+2K0]/3, where calculated values of pi(0) for the hydrostatic condition, and values of pi for the hydrodynamic condition (upward flow) are illustrated in Fig. 3.18. 3.7.4 Influence of the test variables Mass loss describes the combined influence of confining stress and hydraulic gradient. In a relatively open filter (AOS/D85 = 2.8), a particle bridging network is expected to 78 develop over the pore openings of the fabric. Its inherent stability relies on the contact resistance between individual grains of soil. It is evident that upward seepage pressure reduces confining stress at the soil-geotextile interface (see Fig. 3.7), which diminishes the integrity of the bridging network. If the confining stress is insufficient, is it postulated that flow reversal and the reinstating of downward flow may result in localized collapse of the network, whereupon the quantity of mass loss (mass per unit area per flow cycle) is governed by seepage velocity. Therefore, hydraulic gradient exerts an influence in triggering the onset of soil piping, whereas confining stress opposes the role of hydraulic gradient. This combined effect is believed significant to soil retention phenomena in a geotextile filter, especially in the presence of cyclic flow reversal. Wave period appears to influence retention capacity of the woven geotextile. Consider, for example, results of all tests for geotextile W2 in the stage CYC3 at iav \u00E2\u0089\u0088 9 (see Table 3.3). Reporting mass loss as an average value per cycle yields 1.4 g/m 2 /cycle (average) for T = 6 s, 1.1 g/m 2 /cycle for T = 60s and 0.6 g/m 2 /cycle for T = 120 s. It appears the soil-geotextile interface stability is more sensitive to a shorter wave period than a longer wave period. This finding, from observation at T = 6 s, is tentatively attributed to a bridging network over the geotextile openings that cannot fully re-established itself before the onset of the next cycle of flow reversal. This apparent influence of wave period may explain why mass loss is greater and more continuous in the tests at the shortest wave period of T = 6 s. 79 From the test data reported for woven geotextiles, mass loss appears sensitive to a change in wave period, hydraulic gradient and stress. It is believed mass loss may be used to distinguish between a soil-combination that is compatible, versus incompatible in cyclic flow. In order to use this approach, more test data are required to characterize a greater range of AOS/D85. Confidence in the approach also requires data for nonwoven geotextiles. This will enable development of an empirical soil retention rule for geotextile filter compatibility in cyclic flow that is based on principle of mechanics. 3.8 Conclusions Wave period and confining stress influence soil-geotextile filter compatibility, and those two parameters, in combination with hydraulic gradient, require systematic study in order to understand the margin of safety that governs a confident use of empirical design criteria for applications of cyclic flow. Based on experimental results from the large and the small permeameter, for a uniformly-graded sand and two woven geotextiles, the following conclusions are drawn: \u00EF\u0082\u00B7 measurement of axial load in the large permeameter indicates a reduction of 20% to 40% in vertical effective stress along the specimen length that is 80 attributed to interface friction, a finding that implies any stress-based interpretation of soil-geotextile compatibility in a rigid-wall permeameter must address the phenomenon of sidewall friction; \u00EF\u0082\u00B7 mass loss-volume change relations in the large (280 mm diameter) and small (100 mm diameter) permeameter are attributed to a spatial variation of pore size opening in the geotextile specimen rather than a scale effect in the two permeameters, hence it is recommended to repeat a test in the small permeameter where it is believed the soil and geotextile exhibit filtration incompatibility, and report an average of the experimental findings for purposes of analysis; and, \u00EF\u0082\u00B7 therefore the small permeameter is considered sufficient and more practical for a systematic study of test variables. The experimental data and companion theoretical analysis show that: \u00EF\u0082\u00B7 for the multi-stage test method, and corresponding variation of lateral stress in the rigid wall permeameter, mean effective stress at the soil-geotextile interface (pi) is a better parameter for interpretation of performance than vertical stress; and 81 \u00EF\u0082\u00B7 soil retention is very sensitive to the upward component of cyclic flow that yields a reduction in mean effective stress and, it is postulated, thereby acts to destabilize arching in soil particles at the openings of the woven geotextile. For the range of variables examined in testing, mass loss is negligible in cyclic flow at a filter ratio AOS/D85 \u00E2\u0089\u0088 2, but very significant at AOS/D85 \u00E2\u0089\u0088 2.8, where soil-geotextile retention incompatibility is sensitive to loading conditions governed by a combination of wave period, hydraulic gradient and confining stress. The findings suggest that mass loss may be used to distinguish between a soil-combination that is compatible, versus incompatible, in cyclic flow. More test data are required, both for woven and nonwoven geotextiles, to characterize a greater range of AOS/D85 and thereby enable development of an empirical soil retention rule for geotextile filter compatibility in cyclic flow that is based on principle of mechanics. 82 Table 3.1 Properties of the woven geotextiles Geotextile AOS (\u00C2\u00B5m) Percent open area (%) Permittivity (sec -1 ) Filter Ratio (AOS/D85) W1 212 4 0.28 2.0 W2 300 4-6 0.51 2.8 Table 3.2 Test program Small permeameter Large permeameter Geotextile T = 6 s T = 60 s T = 120 s Geotextile T = 6 s T = 60 s W1 W1- T6(S) - W1 - W1- T60(L) W2 W2- T6(S) W2- T60(S) W2- T120(S) W2 W2- T6(L) W2- T60(L) 83 Table 3.3 Mass loss (g/m 2 ) Table 3.3a Wave period T = 6s (900 cycles) Test code W2(S)-T6 W2(S)-T6-R Stage CYC1 CYC2 CYC3 CYC1 CYC2 CYC3 iav 1.3 1.3 1.3 1.1 1.1 1.1 mass loss 465.4 460.0 430.8 322.8 551.3 643.6 iav 5.7 5.7 5.7 5.1 5.1 5.1 mass loss 876.9 838.5 898.7 638.5 661.5 664.1 iav 9.5 9.5 9.5 9.1 9.1 9.1 mass loss 1043.6 1159.0 1476.9 683.3 830.8 1079.5 Cum. loss 7648.7 6075.4 Table 3.3b Wave period T = 60s (90 cycles) and T = 120s (45 cycles) Test code W2(S)-T60 W2(S)-T60-R W2(S)-T120 Stage CYC1 CYC2 CYC3 CYC1 CYC2 CYC3 CYC1 CYC2 CYC3 iav 1.1 1.1 1.1 1.0 1.0 1.0 1.2 1.2 1.2 mass loss 19.2 21.8 34.6 5.6 12.8 17.1 0 0 5.1 iav 5.1 5.1 5.1 5.2 5.2 5.2 5.5 5.5 5.5 mass loss 38.5 37.2 64.1 21.8 34.2 40.6 6.4 10.3 11.5 iav 8.9 8.9 8.9 8.7 8.7 8.7 9.3 9.3 9.3 mass loss 75.6 67.9 114.3 41.9 40.6 76.1 8.9 15.4 25.6 Cum. loss 473.3 286.8 83.3 84 Table 3.4 Soil-geotextile interface stress (small permeameter): parametric values \u00C2\u00B1iav Hydrostatic 1 5 9 \u00CF\u0083\u00C2\u00B4vt (kPa) 66 33 7 66 33 7 66 33 7 66 33 7 \u00CE\u0094\u00CF\u0083\u00C2\u00B4v(0) (kPa) - - - 25.9 15.0 5.1 25.9 15.0 5.1 25.9 15.0 5.1 |\u00C2\u00B1S/\u00CE\u0094\u00CF\u0083\u00C2\u00B4v(0)| - - - 0.04 0.07 0.19 0.19 0.33 0.97 0.34 0.59 1.74 f 0.3 2 0.2 6 0.1 8 - - - - - - - - - Down f - - - 0.34 0.26 0.19 0.34 0.27 0.21 0.35 0.29 0.22 K0 - - - 0.38 0.56 1.50 0.38 0.54 1.23 0.38 0.53 1.06 Up f - - - 0.32 0.25 0.15 0.31 0.21 0.05 0.3 0.17 - 0.04 K0 - - - 0.38 0.57 1.73 0.38 0.59 1.80 0.38 0.62 1.89 85 (a) (b) (c) (d) Figure 3.1 Small permeameter: a) schematic drawing; b) test device; c) test specimen; d) mass collection P 1 Z Axial load LVDT Top inlet-outlet P 3 P 5 P 6 Silicone hose (discrete sample) Soil collection trough Bottom inlet-outlet P 7 Geotextile Soil 86 (a) (b) (c) Figure 3.2 Large permeameter: a) schematic drawing; b) test device; c) test specimen LVDT Soil Z Axial load Geotextile P 1 P 3 P 5 P 7 Bottom load cell Bottom inlet-outlet Top inlet-outlet P 6 87 Figure 3.3 Schematic diagram of head-controlled system for cyclic flow Upward flow Downward flow Permeameter 3-way valve Constant-head cylinder -H +H 88 Figure 3.4 Flow chart of multistage test procedure UNI-1a: downward flow, iav (\u00E2\u0089\u00A545 min) CYC1: cyclic flow, \u00C2\u00B1iav (90 min) \u00CF\u0083\u00CC\u0081 vt 1 (see Sec. 3.3 for values) UNI-1b: downward flow, iav (\u00E2\u0089\u00A530 min) UNI-2a: downward flow, iav (\u00E2\u0089\u00A530 min) CYC2: cyclic flow, \u00C2\u00B1iav (90 min) \u00CF\u0083\u00CC\u0081 vt 2 UNI-2b: downward flow, iav (\u00E2\u0089\u00A530 min) UNI-3a: downward flow, iav (\u00E2\u0089\u00A530 min) CYC3: cyclic flow, \u00C2\u00B1iav (90 min) \u00CF\u0083\u00CC\u0081 vt 3 UNI-3b: downward flow, iav (\u00E2\u0089\u00A530 min) Increase iav to a higher value? END TEST Yes No Select a constant T 89 (a) (b) Figure 3.5 Water head distribution in test W2-T6(S): a) starting CYC1-stage and b) ending CYC1-stage 90 (a) (b) (c) Figure 3.6 Water head distribution in tests W2-T6(S), W2-T60(S) and W2- T120(S): a) T = 6 s; b) T = 60 s; c) T = 120 s; d) schematic representation of transient hydraulic gradient 91 (d) Figure 3.6 (continued) 92 (a) (b) Figure 3.7 Measured stress at the soil-geotextile interface in the large permeameter: a) test W1-T60(L); b) test W2-T60(L) 93 (a) (b) Figure 3.8 Water head distribution in unidirectional flow: a) test W2-T60(S) and b) test W2-T60(L) 94 (a) (b) (c) Figure 3.9 Gradient Ratio in tests W2-T60(S) and W2-T60(L): a) iav \u00E2\u0089\u0088 1; b) iav \u00E2\u0089\u0088 5; c) iav \u00E2\u0089\u0088 9 95 Figure 3.10 Comparison of mass loss and volume change in the small and large permeameter 96 (a) (b) (c) Figure 3.11 Schematic illustration of stress regime in the test specimen: a) hydrostatic; b) downward flow; c) upward flow Figure 3.12 Relation of stress difference and average sidewall shear stress 97 (a) Figure 3.13 Stress analysis for large permeameter: a) test W1-T60(L); b) test W2- T60(L) 98 (b) Figure 3.13 (continued) 99 (a) (b) Figure 3.14 Back-analyzed values of f: a) downward flow; b) upward flow 100 Figure 3.15 Stress calculation procedure for small permeameter Calculate \u00CE\u0094\u00CF\u0083\u00CD\u00B4 v(0) (Eq. 7a) or \u00CE\u0094\u00CF\u0083\u00CD\u00B4 v (Eq. 7b) Calculate \u00CF\u0083\u00CD\u00B4 vm(0) (Eq. 10a), or \u00CF\u0083\u00CD\u00B4 vm (Eq. 10b) Calculate OCR (Eq. 9) to verify the assumed OCR value Calculate K0 (Eq. 8) HYDROSTATIC Given f = 0.32, 0.23 and 0.18 for \u00CF\u0083\u00CD\u00B4 vt = 57, 28 and 7 kPa HYDRODYNAMIC Given \u00CE\u0094\u00CF\u0083\u00CD\u00B4 v(0) and S =iav\u00CE\u00B3wZ Find f for S/\u00CE\u0094\u00CF\u0083\u00CD\u00B4 v(0) from Fig.3.14 Bottom stress: \u00CF\u0083\u00CD\u00B4 vb(0) (Eq. 3) or \u00CF\u0083\u00CD\u00B4 vb (Eq. 4) Define a value of OCR: a trial value for first loop approximated (Eq. 11) for hydrodynamic OCRas = OCRcal No Update with OCRcal 101 Figure 3.16 Vertical effective stress at soil-geotextile interface: small permeameter 102 (a) (b) Figure 3.17 Variation of vertical effective stress (at specimen mid-height) in a rigid- wall permeameter: a) typical response to unloading (modified from Mayne and Kulhawy 1982); b) analyzed response based on results of the large permeameter \u00CF\u0083\u00E2\u0080\u00B2v \u00CF\u0083\u00CD\u00B4 h \u00CF\u0083\u00E2\u0080\u00B2vm(max) \u00CF\u0083\u00E2\u0080\u00B2vm,min 103 Figure 3.18 Mean effective stress at soil-geotextile interface: small permeameter 104 3.9 References Campanella, R. G. & Vaid, Y. P. (1972). A simple K0 triaxial cell. Canadian Geotechnical Journal, 9, No.3, 249 \u00E2\u0080\u0093 260. Cazzuffi, D. A., Mazzucato, A., Moraci, N., & Tondello, M. (1999). A new test apparatus for the study of geotextiles behaviour as filters in unsteady flow conditions: relevance and use. Geotextiles and Geomembranes 17, No. 5-6, 313 - 329. Chen, L., Zhuang, Y-F., Wang, Z. & Xu, Q. (2009). Hydraulic behavior of filter protected silt under cyclic flow. Journal of Geotechnical and Geoenvironmental Engineering, 135, No. 8, 1161 \u00E2\u0080\u0093 1166. Chen, R.-H., Ho, C.-C. & Hsu, C.-Y. (2008). The effect of fine soil content on the filtration characteristics of geotextile under cyclic flows. Geosynthetics International, 15, No. 2, 95\u00E2\u0080\u0093 106. Chew, S. H., Zhao, Z. K., Karunaratne, G. P., Tan, S. A, Delmas, Ph., & Loke, K. H. (2000). Revetment geotextile filter subjected to cyclic wave loading. Proceedings of Geo-Denver 2000, Denver, CO, USA, pp. 162 \u00E2\u0080\u0093 175. de Graauw, A., van der Meulen, T. & van der Does de Bye, M. (1983). Design criteria for granular filters. Publication no. 278, Delft Hydraulics Laboratory, The Netherlands, 25 p. de Graauw, A., van der Meulen, T. & van der Does de Bye, M. (1984). Granular filters: design criteria. Journal of Waterway, Port, Coastal and Ocean Engineering, ASCE, 110, No. 1, 80- 96. Fannin, R. J. & Srikongsri, A. (2007). Geotextile filters in cyclic flow: test results and design criteria. Proceedings of Geosynthetics 2007, Washington, D.C., USA, pp 170-185. 105 Fannin, R. J., Eliadorani, A. & Wilkinson, J. M. T. (2005). Shear strength of cohesionless soils at low stress, Geotechnique, 55, No. 6, 467-478. Fannin, R. J., Vaid, Y. P., Palmeira, E. M. & Shi, Y. (1996). A modified gradient ratio device. Recent Developments in Geotextile Filters and Prefabricated Drainage Geocomposites, ASTM STP 1281, Philadelphia, PA, USA, pp. 100 \u00E2\u0080\u0093 112. Giroud, J. P. (1996). Granular filters and geotextile filters, Proceedings of Geofilters\u00E2\u0080\u009996 Conference, Montreal, Quebec, Canada, pp 565-680. Hameiri, A. (2000). Soil geotextile filtration behavior under dynamic conditions of vibration and cyclic flow. PhD Thesis, University of British Columbia, British Columbia, Canada, 270p. Hameiri, A. & Fannin, R. J. (2002). A cyclic gradient ratio test device. ASTM Geotechnical Testing Journal, 39, No.2, 266-276. Hawley, R. (2001). Filtration performance of geotextiles in cyclic flow conditions. MASc Thesis, University of British Columbia, Vancouver, B.C., Canada, 141p. Indraratna, B & Vafai, F (1997). Analytical model for particle migration within base soil-filter system, Journal of Geotechnical and Geoenvironmental Engineering, 123, No. 2, 100-109. Indraratna, B & Radampola, S. (2002). Analysis of Critical Hydraulic Gradient for Particle Movement in Filtration. Journal of Geotechnical and Geoenvironmental Engineering, 128, No. 4, 347-350. Kenney, T. C. & Lau, D. (1985). Internal stability of granular filters. Canadian Geotechnical Journal, 22, No. 2, 215-225. Kenney, T. C. & Lau, D. (1986). Internal stability of granular filters: Reply. Canadian Geotechnical Journal, 23, No. 3, 420-423. 106 Li, M. (2008). Seepage induced instability in widely graded soils. Ph.D. thesis, University of British Columbia, B.C., Canada, 297 p. Moffat, R. (2005). Experiments on the internal stability of widely graded cohesionless soils. Ph.D. thesis, University of British Columbia, B.C., Canada, 274 p. Mayne, P.W. and Kulhawy, F.H. (1982). \"Ko-OCR Relationships in Soil\", Journal of the Geotechnical Engineering Division, ASCE, 108, No. GT6, 851-872. Mayne, P.W. and Kulhawy, F.H. (1994). Discussion of \"The Coefficient of Earth Pressure At- Rest\", Canadian Geotechnical Journal, 31 No.5, pp. 788-790. Moffat, R.A. and Fannin, R.J. (2006). A large permeameter for study of internal stability in conhesionless soils. ASTM Geotechnical Testing Journal, 29, No. 4, pp. 273-279. Pilarczyk, K. W. (2000). Geosynthetics and Geosystem in Hydraulic and Coastal Engineering. A.A. Balkema, Rotterdam, The Netherlands, 913 p. Shirato, M., Aragaki, T., Mori, R. & Sawamoto, K. (1968). Predictions of constant pressure and constant rate filtrations based upon an approximate correction for side wall friction in compression permeability cell data. Journal of Chemical Engineering of Japan, 1, No. 1, 86- 90 Sivrikaya, O. & Togrol, E. (2005). Measurement of side friction between specimen and consolidation ring with newly designed oedometer cell. ASTM Geotechnical Testing Journal, 29, No. 1, 1-8. Srikongsri, A. and Fannin, R.J. (2010). Soil-geotextile compatibility testing in cyclic flow. Manuscript prepared for publication (chapter 2). 107 U.S. Army Corps of Engineers (2002). Coastal Engineering Manual, EM 1110-2-1100, Washington, DC., USA. Vaid, Y.P. and Negussey, D. (1988) Preparation of reconstituted sand specimens, Advanced Triaxial Testing of Soil and Rock, ASTM STP 977, ASTM, Philadelphia, pp. 405 \u00E2\u0080\u0093 417. 108 4 Geotextile-Soil Retention in Cyclic Flow 3 4.1 Outline Unidirectional and cyclic flow regimes are reproduced at laboratory scale using a cyclic Gradient Ratio device, in order to examine the influence of filter ratio AOS/Dn from combinations of four uniformly-graded soils and seven geotextiles. All combinations were found retention compatible for unidirectional flow. In cyclic flow, mass loss (g/m 2 /100 cycles) indicates a strong influence of wave period, hydraulic gradient and effective stress at 3 \u00E2\u0089\u00A4 AOS/D50 and 2.5 \u00E2\u0089\u00A4 AOS/D85. A novel analytical approach is proposed, to unify AOS/Dn and a normalized value of seepage pressure (S/pi(0)). The hydromechanical approach appears to distinguish between mass loss by washout, in contrast to the more problematic action of piping, in response to cyclic flow. Comparison of the laboratory data with empirical design criteria using AOS/D50 or AOS/D85 reveals a considerable margin of safety for applications to uniformly- graded soil. 3 A version of this chapter will be submitted for publication. Srikongsri, A.and Fannin, R. J. (2010). Soil-geotextile retention in cyclic flow. 109 4.2 Introduction Dynamic flow regimes are complex phenomena, yet cyclic flow (repeating cross-plane reversing flow) is often of major concern as it typically presents a severe challenge for filtration compatibility. In the absence of a standard test method for cyclic flow, a paucity of appropriate laboratory studies, and a lack of well-documented field performance data, the selection of a geotextile for protection against dynamic flow- induced erosion of a base soil is made with reference to empirical criteria that are believed very conservative (Srikongsri and Fannin, 2009). The criteria define a relation between characteristic value of pore size opening in the geotextile (AOS), and indicative particle size of the base soil (Dn), termed the filter ratio (AOS/Dn). The conservatism arises from recommendation of a small filter ratio, likely unnecessarily small, given uncertainty in thresholds for soil retention in dynamic flow. Retention criteria in cyclic flow are derived from judgement based on practical experience, with few reported data to support the selection process. Yet, over a considerable period of time, those criteria have not been evaluated in a systematic manner (Pilarczyk, 2000). Consider the relatively simple case of unidirectional flow through a base soil. Openings of the companion filter layer may be selected as large relative to the grain size of the base soil, and the selection made with reasonable confidence, because the margin of safety can be easily determined in a laboratory filtration test or indeed 110 verified through simulation. In contrast, examining the boundary for soil retention in cyclic flow is considerably more difficult, because dynamic flow regimes are not easily reproduced in a simple test device and the mechanism is governed by several factors (Fannin, 2007). Stress and hydraulic gradient have been demonstrated to govern the movement of soil particles in a granular filter (Indraratna and Vafai, 1997; Indraratna and Radampola, 2002). Similarly, soil passing through the geotextile is influenced by stress and hydraulic gradient (Cazzuffi et al. 1999), and has been found more sensitive to shorter wave periods (Chew et al. 2000). More specifically, Srikongsri et al. (2010a: see chapter 2) reviewed the experimental work of Hawley (2001), which determined the empirical criterion of AOS/D85 \u00E2\u0089\u00A4 0.5 for cyclic flow was overly conservative for the soils examined in testing, given that excessive mass loss through the geotextile was only first encountered at filter ratios AOS/D85 \u00E2\u0089\u0088 2. Consequently, a systematic study was recommended to explore more fully the threshold to seepage-induced instability. In undertaking the systematic study, the need was quickly established to understand better the spatial variation of stress in a rigid-wall permeameter. Furthermore, the need for a standardized multi-stage test procedure and interpretation also became evident. Variation of stress in a rigid-wall permeameter is believed a function of applied stress level, seepage pressure, stress history, soil properties and soil-wall interface shear. An approach to analysis and interpretation of test data is proposed that involves mean effective stress at the soil- geotextile interface (Srikongsri and Fannin, 2010b: see chapter 3). 111 Two empirical criteria used for soil retention in conditions of dynamic, pulsating or cyclic flow, namely O95 or AOS < D50 (Luettich et al., 1992), else O95 or AOS < 0.5D85 (Holtz et al., 1997), are often referenced in conference publications, book chapters and specialist design guidance (see for example, Koerner 1998; Pilarczyk 2000; Reddi 2003; Shukla and Yin 2006; Fannin 2006). Hence these two criteria are selected for evaluation in this systematic laboratory study, with the objective of establishing the margin of safety with regard to filter incompatibility. The Luettich et al. (1992) criterion, for severe hydraulic loading conditions, is based on the earlier proposal of Heerten (1982). In providing guidance on geotextile filter selection against clogging, namely a permeability criterion, Heerten (1982) suggested an approach that was developed from forensic evaluation of 12 needle-punched nonwoven and 4 woven geotextile samples, all of which were exhumed from in-land waterways and sea dikes constructed between the years 1970 - 1977. Comparison was made of clogging in exhumed nonwoven geotextiles, with emphasis on properties of the virgin geotextile and the exhumed geotextile (pore space clogging, reduced porosity and reduced permeability). From these field studies, and more specifically the issue of soil retention, the suggested approach was simply to adopt a conservative value for \u00E2\u0080\u009Csand-tightness\u00E2\u0080\u009D: it led to the soil retention criterion of O90 < D50 for non- plastic soils subject to dynamic loading conditions given by high turbulent flow, wave attack or pumping phenomenon. Accordingly, the empirical criterion is based on 112 experience that is both appropriate and valuable, but it was not developed from any fundamental study of soil retention and filtration compatibility. The Holtz et al. (1997) criterion is modified from earlier work reported by Christopher and Holtz (1985). It transpires the origin of the criterion takes the form of O50 < 0.5D85, and is based on a simple, and intentionally conservative, reduction of a companion filter ratio criterion for unidirectional flow. The objective was to provide for a greater, albeit unknown, margin of safety for more challenging applications in dynamic flow. At the time of the Christopher and Holtz (1985) contribution, few systematic studies were available to support design practice using geosynthetics, hence it is reasonable to find the criterion was not supported by any fundamental study of soil-geotextile behaviour in dynamic, pulsating or cyclic flow. The modified version of AOS < 0.5D85 has subsequently been adopted in the Canadian Foundation Engineering Manual (CFEM). The objective of the current study is to examine the inherent margin of safety in the empirical criteria for geotextile selection in applications of cyclic flow. A systematic experimental program is followed, based on testing of soil-geotextiles combinations in a gradient ratio device, in order to reproduce cyclic flow regimes at laboratory scale. A broad range of filter ratio (0.7 \u00E2\u0089\u00A4 AOS/D85 \u00E2\u0089\u00A4 3.7) or (0.9 \u00E2\u0089\u00A4 AOS/D85 \u00E2\u0089\u00A4 4.6) is evaluated, from combinations of four uniformly-graded soils against five needle- punched nonwoven geotextiles and two types of woven geotextile. The test results are 113 interpreted based on values of Gradient Ratio in the soil-geotextile filtration zone, mass loss through the geotextile, and volume change of the soil specimen. Soil retention is then characterized with reference to principles of hydromechanics, through which recognition is given to the influence of hydraulic gradient (i), stress (\u00CF\u0083) and wave period (T) on filtration compatibility. The findings address confidence in geotextile selection criteria, and margins of safety in use of empirical criteria for engineering design. 4.3 Experimental methodology 4.3.1 Soils Four uniformly-graded soils, termed A, B, C and D (see Fig. 4.1 and Table 4.1), were used to examine retention compatibility with seven different geotextiles. The gradations are derived from Fraser River sand (A and B) and Alouette River sand (C and D), taken from borrow sources located in the Vancouver Lower Mainland of British Columbia. Soils A and B are poorly graded sand (SP), soil C is a poorly graded silty sand (SM) and soil D is a non-plastic sandy silt (ML), according to the Unified Soil Classification System. The grain size distribution curves represent soils at the finer end of the coarse-grained size range, and are believed characteristic of problematic erodible soils in marine environments. 114 Permeability was determined from outflow measurements during test stages of unidirectional flow in the laboratory permeameter, yielding average values between 4 x 10 -3 cm/s for soil D and 2 x 10 -2 cm/s for soil A (see Table 4.1). Direct shearbox testing was conducted on dry specimens, reconstituted to a loose condition, at relatively low values of normal effective stress (\u00CF\u0083\u00E2\u0080\u00B2n \u00E2\u0089\u00A4 50 kPa). The objective was to determine the angle of shearing resistance at large displacement (\u00CF\u0086) at stress levels equal to those encountered in the permeameter test device, for purposes of stress analysis at the soil-geotextile interface. The values of \u00CF\u0086 are found to be very similar in magnitude, and exhibit an apparent stress-dependency (see Table 4.1). 4.3.2 Geotextiles Seven geotextiles, with Apparent Opening Size (AOS) ranging from 0.06 to 0.30 mm, were examined in the program of testing (see Table 4.2). Five needle-punched nonwoven geotextiles, termed NW1 to NW5, are manufactured from continuous filaments of polypropylene. They all exhibit a permeability, defined by the product of permittivity and thickness, greater than that of the four soils. The woven monofilament (W1) and woven multifilament (W2) geotextile are also made of polypropylene yarn. They have a percent open area (POA) greater than or equal to 4 %, and values of permeability comparable to those of soils A and B, and greater than 115 soils C and D. All seven geotextiles were manufactured and supplied by Ten Cate Geosynthetics. Altogether, 15 combinations of soil and geotextile were examined (see Table 4.3), in a total of 27 tests at a wave period of 6, 60 or 120 s. All test combinations were examined at T = 6 s, with additional testing at T = 60 s and 120 s for the case of soil loss or for occasional comparative checks on compatibility. In reporting the laboratory data, the test code employed (e.g. C-W2-T6) declares the sequence of soil gradation (C), geotextile type (W2) and wave period (T6). Given the grain size distribution of soils, and opening size of geotextiles, the variety of soil-geotextile combinations yielded a range in filter ratio of 0.7 \u00E2\u0089\u00A4 AOS/D85 \u00E2\u0089\u00A4 3.7 (0.9 \u00E2\u0089\u00A4 AOS/D50 \u00E2\u0089\u00A4 4.6). The smallest filter ratio is similar to that recommended in design guidance for applications of cyclic flow, while the largest is significantly greater than that recommended for unidirectional flow. 4.3.3 Test device and procedure A cyclic gradient ratio test device was used to characterize soil-geotextile compatibility (see Fig. 4.2a). The device is configured to apply stress on the top of the specimen, to record the distribution of water head along the test specimen, to measure change in specimen length, and to collection of soil loss through the geotextile in discreet quantities. Water head distribution is used to calculate a value of Gradient 116 Ratio: port 1 is located on the top plate; ports 3, 5 and 6 are located at 75 mm, 25 mm and 8 mm above the geotextile; port 7 is located beneath the geotextile sample (see Fig. 4.2a). Gradient Ratio, GR25 and a modified value GR8 are calculated as: 35 57 25 i i GR \u00EF\u0080\u00BD (1) 35 67 8 i i GR \u00EF\u0080\u00BD (2) where i35 = hydraulic gradient within the soil between ports 3 and 5 i57 = hydraulic gradient across the soil-geotextile interface between ports 5 and 7 i67 = hydraulic gradient across the soil-geotextile interface between ports 6 and 7 Cyclic flow is imposed by a head-controlled system (see Fig. 4.2b). Average hydraulic gradient (iav), in either downward or upward flow, is controlled by an equidistant spacing of constant head tanks (+H and -H). The gradient is defined by head loss between ports 1 and 7 (H17), expressed as iav = H17/Z, where Z is length of the specimen. Frequency of flow reversal is controlled using a 3-way solenoid valve programmed to switch at an interval t = T/2 sec, where T is the wave period. Further details of the test device are reported by Hameiri (2000) and Hameiri and Fannin (2002). 117 Preparation of the geotextile sample, reconstitution of the soil specimen, and imposition of seepage in stages of unidirectional (UNI) and cyclic flow (CYC) followed exactly the routine described by Srikongsri and Fannin (2010b: see chapter 3). A test was conducted with cyclic flow at constant wave period (T = 6, 60 or 120 s). Three values of gradient (iav \u00E2\u0089\u0088 1, 5 and 9) were imposed sequentially, with vertical effective stress on the top of the specimen changed to yield values of confining stress reducing from pi(0) = 23 kPa to 14 kPa and then 3 kPa at each value of iav. The nature of the loading routine for test C-W2-T6 is illustrated schematically for iav \u00E2\u0089\u0088 9 (see Fig. 4.3), in order to demonstrate how stages of unidirectional (UNI) flow were used to bound stages of cyclic (CYC) flow, yielding values of gradient ratio (GR8 and GR25). Consider the matter of stress distribution in the test specimen. Vertical effective stress at the soil-geotextile interface (\u00CF\u0083\u00E2\u0080\u00B2vb) is less than that applied at the top (\u00CF\u0083\u00E2\u0080\u00B2vt), as a consequence of sidewall friction (see Fig. 4.2c). The relation between \u00CF\u0083\u00E2\u0080\u00B2vt and \u00CF\u0083\u00E2\u0080\u00B2vb has been quantified (Srikongsri and Fannin, 2010b: see chapter 3) with respect to its variation in a multi-stage loading sequence identical to that of the program of testing reported herein. Recognizing the importance of overall confining stress on geotextile filtration compatibility, the relation established between \u00CF\u0083\u00E2\u0080\u00B2vt and initial mean effective stress (pi(0)) at the soil-geotextile interface is illustrated for soil C (see Fig. 4.2d). The relation was established for the initial hydrostatic condition: it is assumed to quantify the initial hydrostatic stress condition for tests on soils A, B and D given the similar properties of all four uniformly graded materials. 118 4.4 Results The five needle-punched nonwoven geotextiles were all tested at the relatively fast cyclic flow (T = 6 s), in a total of eight combinations of soil and geotextile that yield a filter ratio 0.7 \u00E2\u0089\u00A4 AOS/D85 \u00E2\u0089\u00A4 2.3. In presenting the data, results are examined over two distinct ranges of filter ratio, firstly for 0.7 \u00E2\u0089\u00A4 AOS/D85 \u00E2\u0089\u00A4 1, and secondly for 1 < AOS/D85 \u00E2\u0089\u00A4 2.3. The two woven geotextiles, which exhibit a relatively larger opening size than the nonwoven geotextiles (see Table 4.3), were also tested at the relatively fast cyclic flow of T = 6 s and, in several cases at T = 60 s and 120 s, with tests exhibiting significant soil loss then repeated. A total of seven combinations of soil and geotextile yield a filter ratio 1.2 \u00E2\u0089\u00A4 AOS/D85 \u00E2\u0089\u00A4 3.7. The data are again presented over two ranges of filter ratio, namely 1.2 \u00E2\u0089\u00A4 AOS/D85 \u00E2\u0089\u00A4 2 and 2 < AOS/D85 \u00E2\u0089\u00A4 3.7, in order to distinguish between subtle aspects of filtration compatibility. The generalized nature of filtration compatibility is summarized in Table 4.4, and examined herein with reference to type of geotextile and filter ratio. A general trend is evident in the experimental findings, with the nonwoven geotextiles yielding no mass loss, and the more open woven geotextiles yielding mass loss at larger filter ratio (see also Table 4.4). The general trend is best illustrated by the relation between mass loss and AOS/D85, for all test specimens, induced by cyclic flow in the CYC2 stage at the 119 same confining stress pi(0) = 14 kPa (\u00CF\u0083\u00CD\u00B4 vt = 33 kPa) at iav \u00E2\u0089\u0088 1 (Fig. 4.4a) and iav \u00E2\u0089\u0088 5 (Fig. 4.4b). Mass loss per unit area is reported for 100 cycles of flow reversal (g/m 2 /100 cycles): it is an average value for tests at T = 6 s (with 900 cycles/stage) and an extrapolated value for tests at T = 60 s (with 90 cycles/stage) and T = 120 s (with 45 cycles/stage). It appears the filter ratio 2 \u00E2\u0089\u00A4 AOS/D85 \u00E2\u0089\u00A4 3 exerts a governing influence on filtration compatibility, and deserves careful evaluation. 4.4.1 Nonwoven geotextiles 4.4.1.1 Filter ratio: 0.7 \u00E2\u0089\u00A4 AOS/D85 \u00E2\u0089\u00A4 1.0 Four soil-geotextile combinations with AOS/D85 \u00E2\u0089\u00A4 1 (tests A-NW4-T6, C-NW3-T6, D- NW1-T6 and D-NW2-T6, see Table 4.3), all exhibited a similar response in testing at T = 6 s. No soil was observed to pass through the geotextile during the process of soil reconstitution by water pluviation. Visual observations indicated no loss of soil in any stage of unidirectional (UNI) flow and, similarly, no loss in any cyclic (CYC) stage, and none was found in the silicone collection hose (see Fig. 4.2a). No volume change was observed in the specimen during the test. Gradient ratio values in the range 0.9 \u00E2\u0089\u00A4 GR25 \u00E2\u0089\u00A4 1.4 and 0.9 \u00E2\u0089\u00A4 GR8 \u00E2\u0089\u00A4 1.7, for all stages of unidirectional flow, confirm the homogeneity of the test specimen. Accordingly, these test combinations are deemed stable in both unidirectional and cyclic flow. 120 4.4.1.2 Filter ratio: 1.0 < AOS/D85 \u00E2\u0089\u00A4 2.3 Consider the two soil-geotextile combinations with AOS/D85 = 1.7 (tests C-NW4-T6 and C-NW5-T6), which again were tested at T = 6 s. No soil loss occurred during specimen reconstitution, and subsequent visual observations indicated no loss of soil in any stage of unidirectional (UNI) or cyclic (CYC) flow. The observations are consistent with no measurable accumulation of soil in the collection hose, and no volume change during the test. Values of GR25 and GR8 in the range 0.9 to 1.0 indicate an almost linear variation of water head distribution along the length of the specimen. Accordingly, both tests reveal a very similar response, and the combinations are deemed stable in both unidirectional and cyclic flow. The two soil-geotextile combinations with AOS/D85 = 2.3 (tests D-NW4-T6 and D- NW4-T60, and D-NW5-T6) include one test at T = 60 s, in addition to those at T = 6 s. Although some soil loss was observed at the beginning of specimen reconstitution, the amount was too small to measure. Visual observations indicated no loss of soil in any stage of unidirectional (UNI) flow and, similarly, no loss in any cyclic (CYC) stage, which is consistent with no recorded volume change in the specimens. Values of GR25 and GR8 ranging from 0.9 to 1.3 imply a water head distribution that is essentially linear along the specimen. Accordingly, these test combinations are also deemed stable in both unidirectional and cyclic flow. 121 Scanning Electron Microscope (SEM) images from D-NW4-T6 and D-NW5-T6, taken after testing (Figs. 4.5b and 4.5d), are compared to images of the geotextiles before testing (Figs. 4.5a and 4.5c). Note the geotextiles must be dried before imaging, and experience moderate disturbance as a consequence of preparation for the imaging process. In contrast to the specks of dust evident on the untested geotextile, the post- test specimens reveal particles of soil D embedded within the fabric of the geotextile. The particles, ranging in size up to 100 microns, illustrate clearly the manner in which soil retention is mobilized within the tortuous array of pore size openings and constrictions in the nonwoven material of AOS/D85 = 2.3. 4.4.2 Woven geotextiles 4.4.2.1 Filter ratio: 1.2 \u00E2\u0089\u00A4 AOS/D85 \u00E2\u0089\u00A4 2.0 Four soil-geotextile combinations were examined with 1.2 \u00E2\u0089\u00A4 AOS/D85 \u00E2\u0089\u00A4 2.0 (see Table 4.3). No soil was observed to pass through the geotextile during specimen reconstitution of tests A-W1-T6 and A-W1-T60 (AOS/D85 = 1.2), and B-W1-T6 and B-W1-T60 (AOS/D85 = 1.4). In contrast, tests B-W2-T6, B-W2-T60 and C-W1-T6 (AOS/D85 = 2) yielded a mass loss between 50 and 100 g/m 2 . Thereafter, visual observations indicated no loss of soil in any stage of unidirectional (UNI) flow and, similarly, no loss in any cyclic (CYC) stage. Additionally, no volume change was measured in the specimen length during these tests. Gradient ratio values in the range 122 1.0 \u00E2\u0089\u00A4 GR25 and GR8 \u00E2\u0089\u00A4 1.5 confirm the homogeneous nature of the test specimen. Accordingly, the test combinations for AOS/D85 \u00E2\u0089\u00A4 2.0 are deemed stable in both unidirectional and cyclic flow. 4.4.2.2 Filter ratio: 2.0 < AOS/D85 \u00E2\u0089\u00A4 3.7 In total, five tests were performed on soil D and geotextile W1 (AOS/D85 = 2.6) for wave periods of T = 6, 60 and 120 s and, similarly, five tests on soil C with geotextile W2 (AOS/D85 = 2.8) at the same values of wave period. One test was performed at the largest filter ratio, that for soil D and geotextile W2 (AOS/D85 = 3.7), at T = 6 s. A significant soil loss was found during specimen reconstitution, in the range 200 to 400 g/m 2 for 2.6 \u00E2\u0089\u00A4 AOS/D85 \u00E2\u0089\u00A4 2.8, and 450 g/m 2 for AOS/D85 = 3.7. In providing a detailed description of soil loss during cyclic flow (see Figs. 4.6 to 4.12), reference is made to imposed seepage pressure S = \u00C2\u00B1 iav\u00CE\u00B3wZ (kPa), where \u00CE\u00B3w is unit weight of water and Z is length of the test specimen, when discussing the influence of hydraulic gradient. For AOS/D85 = 2.6, at T = 6 s (test D-W1-T6), mass loss occurred in none of the UNI stages, but all of the CYC stages. The loss was not found particularly sensitive to S \u00E2\u0089\u0088 \u00C2\u00B1 1 kPa for each of the three values of confining stress (pi(0) = 23 kPa, 14 kPa and 3 kPa), but appears more sensitive at S \u00E2\u0089\u0088 \u00C2\u00B1 5 and \u00C2\u00B1 9 kPa (see Fig. 4.6). The response to seepage flow appears consistent, given repetition of the test on a different sample of 123 the same geotextile: average mass loss (mav) varies from 10 to 90 g/m 2 /100 cycles (see Fig. 4.6a), and from 10 to 150 g/m 2 /100 cycles (see Fig. 4.6b). The modest difference in values is believed a consequence of spatial variation in pore size distribution in the two samples tested. Visual observation indicates the losses were continuous, rather than pulsating, a response that is attributed to the relatively short wave period of 6 s. No preferential flow channels were found to develop in the soil specimen. Increasing the wave period to T = 60 s (test D-W1-T60) at the same AOS/D85 = 2.6, again yielded no mass loss during the UNI stages of flow, and a loss in all of the CYC stages. The general trend found in testing at T = 6 s is also apparent in testing at T = 60 s (compare Figs. 4.7 and 4.6). At the same value of confining stress, soil loss increases with greater seepage pressure (from S \u00E2\u0089\u0088 \u00C2\u00B1 1 kPa to S \u00E2\u0089\u0088 \u00C2\u00B1 5 kPa and S \u00E2\u0089\u0088 \u00C2\u00B1 9 kPa, see Fig. 4.7a); likewise, at the same condition of seepage pressure, mass loss increases with reduction in confining stress. In contrast to the experience at T = 6 s, the losses at T = 60 s occurred as a pulsating action that dimished to zero loss over the duration of each flow reversal. The loss was not restricted to any preferential location on the geotextile filter. The trend in mass loss is again confirmed by the repeated test (see Fig. 4.7b), where the same influence of spatial variation in geotextile opening size is believed evident (compare Figs. 7a and 7b). The losses at T = 60 s appear slightly smaller than those at T = 6 s. 124 Increasing the wave period to T = 120 s (test D-W1-T120) confirmed, once again, no soil loss through the geotextile in the UNI stages of flow. Furthermore, no loss was measured at S \u00E2\u0089\u0088 \u00C2\u00B1 1. Thereafter, some loss (25 g/m2 < mav \u00E2\u0089\u00A4 55 g/m 2 /100 cycles) was observed at S \u00E2\u0089\u0088 \u00C2\u00B1 5 and \u00C2\u00B1 9 kPa (see Fig. 4.8). In this regard, it appears the relatively long wave period does not replicate the responses at T = 6 and 60 s. All tests at this filter ratio gave values of GR25 and GR8 between 1.0 and 1.3, which confirms the homogeneity of the test specimen throughout the multi-stage test procedure. Hence, to summarize the findings for tests at AOS/D85 = 2.6, the soil- geotextile combination is deemed stable in unidirectional flow. However, given the greater value of soil loss, the soil retention and consequent nature of stability achieved in cyclic flow requires further analysis in order to characterize it more definitively. Consider now the data acquired for AOS/D85 = 2.8, which are re-plotted from Srikongsri and Fannin (2010a). No mass loss was observed during stages of unidirectional flow. In cyclic flow, test C-W2-T6 (see Fig. 4.9a) and the companion repeated test (see Fig. 4.9b) at T = 6 s both exhibit the general trend evident in data for the smaller filter ratio of 2.6 at the same relatively short wave period (see Fig. 4.6). Values of mav vary from 50 to 160 g/m 2 /100 cycles, which slightly exceeds those for the corresponding test at AOS/D85 = 2.6. The finding is attributed to the slightly larger filter ratio. 125 Increasing the wave period to T = 60 s (test C-W2-T60) at the same AOS/D85 = 2.8, again yielded no mass loss as a consequence of unidirectional flow. Losses during stages of cyclic flow in this test (see Fig. 4.10a) and the companion repeated test (see Fig. 4.10b) again increase with greater seepage pressure, and increase with reduction in confining stress. Values of mav recorded in the two tests do not appear to differ significantly, ranging from 25 to 125 g/m 2 /100 cycles and 10 to 80 g/m 2 /100 cycles, respectively. As was the case for AOS/D85 = 2.6, the losses at T = 60 s appear slightly smaller than those at T = 6 s. Increasing the wave period to T = 120 s (test C-W2-T120) yields a response (see Fig. 4.11) almost identical to that reported for D-W1-T120 (see Fig. 4.8). Comparison of the data shows the general pattern of soil loss with variation of seepage pressure and confining stress is very similar, with mav \u00E2\u0089\u00A4 55 g/m 2 /100 cycles in all stages of cyclic flow. All tests at this filter ratio gave values of GR25 and GR8 between 1.0 and 1.5, which like the previous filter ratio, confirms the homogeneity of the test specimen throughout the multi-stage test procedure. Hence, to summarize the findings for tests at AOS/D85 = 2.8, the soil-geotextile combination is deemed stable in unidirectional flow. However, given the greater value of soil loss, and its variation with wave period, soil retention and the consequent nature of stability achieved in cyclic flow requires further analysis in order to properly characterize it. 126 One test (D-W2-T6) was performed for the largest filter ratio AOS/D85 = 3.7, at T = 6 s. Unlike all of the preceding tests at relatively smaller filter ratios, it was terminated before completion of the multi-stage procedure. More specifically, the test was stopped after completion of the stage at iav \u00E2\u0089\u0088 5 and pi(0) \u00E2\u0089\u0088 14 kPa (CYC2), due to excessive mass loss through the geotextile and a corresponding loss of integrity in the test specimen itself. Again, in contrast to all of the preceding tests, mass loss was observed in every stage of unidirectional flow that was applied, and measured to be in the range 30 to 80 g/m 2 . Visual observations confirm the losses to occur over a relatively short time period after transition from cyclic to unidirectional flow, typically no longer than one minute, whereupon the downward seepage flow did not yield further loss and the system appeared \u00E2\u0080\u009Cmeta-stable\u00E2\u0080\u009D. Values of GR25 and GR8 in the range 0.7 to 0.8, which being less than 1.0, suggest a zone of slightly lower hydraulic conductivity at the soil-geotextile interface that is consistent with significant washout of soil particles. Accordingly, the soil-geotextile combination is believed close to a limiting value of filter ratio for soil retention in unidirectional flow. In the five stages of cyclic flow that were imposed (see Fig. 4.12a), mass loss was found to be much greater than in previous tests (Fig. 12a). More specifically, mav varied from 150 to nearly 320 g/m 2 /100 cycles, which resulted in development of significant voids in the body of soil specimen (see Fig. 4.12b). The AOS/D85 = 3.7 filter ratio is deemed unstable in cyclic flow, given the very clear evidence of retention incompatibility. 127 In summary, significant quantities of soil loss were only observed from tests on the two woven geotextiles, with soils C and soil D, for values of filter ratio in the range 2.6 \u00E2\u0089\u00A4 AOS/D85 \u00E2\u0089\u00A4 3.7. Loss occurs with cyclic flow alone at the lower end of the range, and with both unidirectional and cyclic flow at the higher end of the range. Variations in mass loss are evident at different values of confining stress, hydraulic gradient and wave period. Scanning Electron Microscope (SEM) images from D-W1-T6 and C-W2-T6, taken after testing (see Figs. 4.13a and 4.13b, respectively) illustrate the very different structure of the woven monofilament and multifilament fabrics. The AOS of geotextile W1 (210 microns) is smaller than geotextile W2 (300 microns), and both values are larger than those of the nonwoven geotextiles examined in testing (see Table 4.2). A common feature of both is the absence of a tortuous path, comprising many pore size openings and constrictions, across the plane of the fabric. It is in marked contrast to the nonwoven geotextile (see Fig. 4.5). The SEM images illustrate well the regular configuration of fibres and openings of a woven geotextile that does not allow particles to lodge within the fabric (see Fig. 4.13). It is believed that retention stability in these soil-geotextile combinations of filter ratio greater than 1 is a consequence of soil bridging over the filter opening (see Fig. 4.14). The range between 2.0 < AOS/D85 < 2.6 appears very important with regard to the onset of filtration incompatibility in cyclic flow. The work of Valdes and Santamarina 128 (2008) demonstrates, for spherical particles and a single circular orifice or opening, a series of conceptual regimes for particle bridging. The effect of particle shape on the formation of bridges across an orifice of diameter d0 is illustrated in Figure 4.14a. The conceptual regime for glass beads at 2 \u00E2\u0089\u00A4 d0/d \u00E2\u0089\u00A4 3 is further illustrated in Figure 4.14b: it suggests that for spherical particles and a single orifice, the onset of bridging instability may occur at d0/d \u00E2\u0089\u0088 3. The contribution of Valdes and Santamarina (2008) suggests the ratio d0/d increases with particle angularity. Likely, this fundamental study (with vibration, not cyclic flow) represents an upper threshold to the case for soil grains against a mesh or filter with many closely-spaced openings. Hence, the onset of filtration incompatibility in cyclic flow at 2.0 \u00E2\u0089\u00A4 AOS/D85 \u00E2\u0089\u00A4 2.6 observed in the current study appears consistent with the findings reported by these conceptual regimes. 4.5 Analysis and discussion The behaviour of soil-geotextile combinations in the range 2.6 \u00E2\u0089\u00A4 AOS/D85 \u00E2\u0089\u00A4 3.7 provides the basis for a hydromechanics-based analysis. More specifically, the measured soil loss and associated volume change may be used to quantify filtration compatibility with reference to soil retention. The analytical approach first involves defining an acceptable threshold for seepage-induced loss or washout through the geotextile, and then defining a retention criterion that accounts for the influence of wave period, hydraulic gradient and confining stress. Findings of the analysis are 129 reported for the complete set of laboratory data (0.7 \u00E2\u0089\u00A4 AOS/D85 \u00E2\u0089\u00A4 3.7, else the equivalent 0.9 \u00E2\u0089\u00A4 AOS/D50 \u00E2\u0089\u00A4 4.6). Discussion addresses the issue of a margin of safety in design guidance. 4.5.1 Soil washout In order to achieve stability in a relatively open filter, some amount of seepage- induced mass loss may be necessary, recognizing washout contributes to the formation of a stable bridging network at the filter openings. A threshold value for the associated losses may be established from consideration of volume change over time (or number of flow cycles): a constant or increasing rate of volume change is taken to denote a piping action. A preliminary consideration of volume change in the three soil-geotextile combinations that exhibit mass loss suggests a value of mav greater than about 60 g/m 2 /100 cycles is likely not a case of washout. More specifically, consider the data for tests D-W1-T6 and C-W2-T6. The response at T = 6 s is selected for analysis because volume loss at T = 60 s (90 cycles) and 120 s (45 cycles) is insufficient for systematic study, compared to the 900 cycles imposed at T = 6 s. The relation between volume change and number of flow cycles for these two tests (including test repetition), from the CYC1 stage at iav \u00E2\u0089\u0088 5, reveals three clustered responses (see Fig. 4.15). A nearly constant rate of loss occurs for 70 g/m 2 /100 cycles \u00E2\u0089\u00A4 mav \u00E2\u0089\u00A4 100 130 g/m 2 /100 cycles (see Fig. 4.15a), whereas the rate appears to diminish for 35 g/m 2 /100 cycles \u00E2\u0089\u00A4 mav \u00E2\u0089\u00A4 50 g/m 2 /100 cycles and clearly results in zero volume change after about 500 cycles for mav \u00E2\u0089\u00A4 30 g/m 2 /100 cycles (Fig. 4.15b). It is proposed the curves for 36, 38 and 52 g/m 2 /100 cycles represent a transition from washout to a piping action, which may be reasonably associated with a volume change less than 0.5 %. It therefore appears an upper threshold to washout, leading to stability in soil retention, is defined by the range 30 g/m 2 /100 cycles \u00E2\u0089\u00A4 mav \u00E2\u0089\u00A4 50 g/m 2 /100 cycles (or total mass loss of 270 to 450 g/m 2 , at 900 cycles). Comparison shows the range is similar to the mass loss of 200 g/m 2 to 400 g/m 2 through the geotextile that was observed during reconstitution of the soil specimen by water pluviation. 4.5.2 Soil retention: a hydromechanical approach A normalized seepage pressure (S/pi(0)), which accounts for both the initial stress condition and hydraulic gradient, is evaluated for its potential to explain the variation of mass loss during various stages of seepage flow in the multi-stage test procedure. The stress ratio serves as an index for intensity of the imposed hydromechanical loading condition. A low S/pi(0) value, significantly less than unity, represents relatively mild or moderate hydraulic loads. In contrast, if the ratio S/pi(0) exceeds a value of one, it implies that effective stress at the soil-geotextile filter interface is approaching zero. The range of variables examined in testing (see Table 4.5) yields a range of normalized seepage pressure 0.05 \u00E2\u0089\u00A4 S/pi(0) \u00E2\u0089\u00A4 3. 131 4.5.2.1 Onset of piping for 2.6 \u00E2\u0089\u00A4 AOS/D85 \u00E2\u0089\u00A4 3.7: woven geotextiles For AOS/D85 \u00E2\u0089\u0088 2.6, mass loss data obtained at T = 6 s (18 values from Fig. 6a and 6b) and at T = 60 s (18 values from Fig. 4.7a and 4.7b) is reported as an average value per 100 cycles (yielding 9 values for each wave period), and reproduced with reference to the normalized seepage pressure (see Fig. 4.16a). Data for T = 120 s (from Fig. 4.8) are also included, with the results from 45 cycles reported as an equivalent rate per 100 cycles. The influence of wave period is generally evident: data at T = 6 s represent the most sensitive response, characterized by greatest increase of mass loss with normalized seepage pressure, compared to data at T = 60 s and T = 120 s, with the latter yielding a relatively small increase in soil loss. It is reasonable to conclude the short wave period of T = 6 s governs retention stability in the range 6 s \u00E2\u0089\u00A4 T \u00E2\u0089\u00A4 120 s. Inspection of the T = 6 s data, at the threshold range of 30 \u00E2\u0080\u0093 50 g/m2/100 cycles for washout, suggests the onset of soil piping through the geotextile is triggered at values of S/pi(0) approximately 0.1 to 0.2. For AOS/D85 \u00E2\u0089\u0088 2.8, mass loss data are obtained in the same manner (see Fig. 4.16b), using data at T = 6 s in (Figs. 4.9a and 4.9b), T = 60 s (Figs. 4.10a and 4.10b) and T = 120 s (Fig. 4.11). Inspection of the plot confirms the data at T = 6 s again demonstrate the greatest increase of mass loss, and govern the retention stability. Indeed a subtle, yet clear difference, is evident in the relative plotting positions of the response at each 132 of three wave periods. Inspection of the T = 6 s data, at the threshold range for washout, suggests the onset of soil piping is triggered at values of S/pi(0) approximately 0.04 to 0.08. For AOS/D85 \u00E2\u0089\u0088 3.7, data for only five stages at T = 6 s (Fig. 12) are available (see Fig. 4.16c) given early termination of the test. They general trend of mass loss with S/pi(0) is consistent with that observed for the two smaller filter ratios (see Figs. 4.16a and 4.16b). The five data points locate significantly above the threshold range for washout, providing further evidence that the soil-geotextile combination experienced a severe piping action in response to seepage flow. The finding is consistent with the visual observation of voids within the body of the soil specimen (see Fig. 4.12b). Extrapolation of the data toward the origin of the plot, and therefore into the threshold range for washout, suggests the onset of soil piping is likely triggered at values of S/pi(0) approximately 0.01 to 0.02. The general relation between mass loss and normalized seepage pressure, in combination with a criterion for washout progressing to onset in piping, is found to be an appropriate technique for characterizing soil retention in cyclic flow. It addresses the most critical test variables and, although subjective in part, lends itself to quantification of the system response. Importantly it brings a hydromechanical framework to evaluation of soil-geotextile filtration compatibility, which has potential for general application (see Fig. 4.16d). 133 4.5.2.2 Hydromechanical influence: a unified plot The experimental data indicate the short wave period T = 6 s governs stability in soil retention, a response that is attributed to the more aggressive nature of this flow condition. Accordingly, it is reasonable to use the response at that wave period to unify the influence of hydromechanical constraint at the soil geotextile interface, characterized by S/pi(0), and the influence of geometric constraint from the filter medium, characterized either by AOS/D85 or AOS/D50 in various design guidance. The experimental findings on S/pi(0) values at the onset of piping are plotted against 2.6 \u00E2\u0089\u00A4 AOS/D85 \u00E2\u0089\u00A4 3.7 (see Fig. 4.17). Additionally the finding for AOS/D85 = 2.0, from test C-W1-T6 that exhibited essentially no loss for S/pi(0) \u00E2\u0089\u00A4 3, is included for purposes of trends analysis. The dashed curve represents the lower boundary (30 g/m 2 /100 cycles) to washout through the geotextile. The error bar represents the range between lower and upper bound values of S/pi(0) to washout. It should be noted that inspection of the error bars, derived from the postulated washout threshold in Fig. 4.16, implies a greater sensitivity to value of the threshold limits rather than value of effective stress at the soil-geotextile interface. At a particular value of S/pi(0), if a value of AOS/D85 plots above the curve, it implies there is susceptibility to soil piping. Consider now the data for a wave period longer than T = 6 s, namely 60 s and 120 s, which are available at AOS/D85 = 2.6 and 2.8: the threshold range is associated with 134 larger values of S/pi(0) for T = 60 s and 120 s (see Figs. 4.16a and 4.16b). The influence of wave period is evident in the unified plot (Fig. 4.17). The phenomenon may be explained through a consideration of the duration of steady flow condition between each cycle of flow reversal, in association with the rate of change of seepage pressure (i.e. \u00E2\u0088\u0086S, kPa \u00E2\u0080\u0093 s-1). For tests at T = 6 s, the soil-geotextile interface is continuously subjected to an unsteady flow condition yielding turbulent pressure fluctuations: the relatively short duration of each cycle is not sufficient to allow the value of \u00E2\u0088\u0086S, which is a function of the rate of change of gradient (\u00E2\u0088\u0086i), to diminish to zero before the next flow reversal (see chapter 3). In contrast, tests at T = 60 and 120 s experience a transient condition of steady flow between each flow reversal, and consequently the soil bridging network can stabilize between periods of instability. The plotting position of the data at T= 60 s and 120 s is attributed to this importance of wave period. Furthermore, the rate of change of seepage pressure is also found to influence the onset of incompatibility. Consider now the error bars for the data at AOS/D85 = 2.6 and 2.8, on the semi-log plot (Fig. 4.17), which depict values of S/pi(0) associated with a condition of washout (30 g/m 2 /100 cycles) and onset of retention incompatibility from piping action (50 g/m 2 /100 cycles). The range of S/pi(0), from a condition of washout through to onset of piping action, is found to be relatively small in tests at T = 6 s, nearly twice as large at T = 60, significantly larger again at T = 120 s (see also Fig. 4.16). In each case, \u00E2\u0088\u0086S/pi(0) increases by approximately 100%. The finding suggests that the rate of change 135 of seepage pressure is more important to soil retention for conditions of relatively fast cyclic flow. Hence, the threshold boundary at T = 6 s yields a conservative relation between S/pi(0) and AOS/D85 for the onset of incompatibility in soil retention. 4.5.2.3 Unified plot for AOS/D85 The hydromechanical influence examined in Fig. 4.17, for data on woven geotextiles only, is expanded to include data for all soil-geotextile combinations examined in the test program (see Fig. 4.18a). Recall that filter ratios of 1.2 \u00E2\u0089\u00A4 AOS/D85 \u00E2\u0089\u00A4 2.0 are for woven geotextiles, and 0.7 \u00E2\u0089\u00A4 AOS/D85 \u00E2\u0089\u00A4 2.3 are for needle-punched nonwoven geotextiles, yielding a variety of geotextile type at a filter ratio AOS/D85 \u00E2\u0089\u0088 2. The data for no soil loss are also plotted as a minimum and maximum range of S/pi(0), for any given AOS/D85: the minimum value is that for the first stage of cyclic loading (CYC1) at a lowest gradient (iav \u00E2\u0089\u0088 1), and the maximum value is obtained in the third stage (CYC3) at the highest gradient imposed in the test. A general inspection of Fig. 4.18a reveals three characteristic responses to cyclic flow. At relatively large filter ratio (AOS/D85 \u00E2\u0089\u00A5 3.5), the onset of piping that occurs at a very low value of S/pi(0) \u00E2\u0089\u00A4 0.05 means a gentle flow reversal is capable of triggering retention instability, and implies the filter ratio is close to a limit at which unidirectional flow alone could lead to filter incompatibility. In contrast, no soil loss is found with AOS/D85 \u00E2\u0089\u00A4 2.0 for a woven geotextile, and with AOS/D85 \u00E2\u0089\u00A4 2.2 for a 136 nonwoven geotextile. The interval between these two contrasting responses yields a third range of filter ratio wherein washout and piping are found very dependent on the influence of hydraulic gradient on the magnitude of effective stress. Within this range, a filter ratio 2.0 \u00E2\u0089\u00A4 AOS/D85 \u00E2\u0089\u00A4 2.5 is likely to result in washout, with increasing susceptibility to piping at higher seepage pressures; a filter ratio 2.5 \u00E2\u0089\u00A4 AOS/D85 \u00E2\u0089\u00A4 3.5 is likely to result in piping, with a small change in filter ratio being very sensitive to changes in the hydromechanical influences of stress and gradient. The design criterion O95/D85 or AOS/D85 \u00E2\u0089\u00A4 0.5 is suggested for applications of cyclic, pulsating or reversing flow (Holtz et al., 1997; CFEM, 2006). From inspection of the laboratory data (see Fig. 4.18a), it appears the criterion provides a substantial safety margin, given no soil loss in any stage of unidirectional or cyclic flow in a woven or nonwoven geotextile with AOS/D85 \u00E2\u0089\u00A4 2. 4.5.2.4 Unified plot for AOS/D50 The same database for all soil-geotextile combinations is also reported with reference to the alternate filter ratio of AOS/D50 (see Fig. 4.18b). Once again, the data for soil loss are plotted with the threshold range of S/pi(0) for onset of piping, and the data with no soil loss are plotted with a minimum and maximum range of S/pi(0) imposed in testing. Given the uniform gradation of the four soils examined in testing, the distribution 0.7 \u00E2\u0089\u00A4 AOS/D85 \u00E2\u0089\u00A4 3.7 is now expressed as 0.9 \u00E2\u0089\u00A4 AOS/D50 \u00E2\u0089\u00A4 4.6. 137 The same three characteristic responses to cyclic flow identified previously (in Fig. 4.18a) are again evident in Fig. 18b. At relatively large filter ratio (AOS/D50 \u00E2\u0089\u00A5 4.0), the onset of piping at a very low value of S/pi(0) \u00E2\u0089\u00A4 0.05 implies a filter ratio that is close to filter incompatibility in unidirectional flow alone. In contrast, no soil loss is found with AOS/D50 \u00E2\u0089\u00A4 2.4 for a woven geotextile, and with AOS/D85 \u00E2\u0089\u00A4 2.7 for a nonwoven geotextile. The third range of filter ratio now yields values of 2.5 \u00E2\u0089\u00A4 AOS/D50 \u00E2\u0089\u00A4 3.0 that are likely to result in washout and increasing susceptibility to piping at higher seepage pressures, and a filter ratio 3.0 \u00E2\u0089\u00A4 AOS/D50 \u00E2\u0089\u00A4 4.5 that is likely to yield piping, again with a small change in filter ratio being very sensitive to hydromechanical influences. The design criterion O95/D50 or AOS/D50 \u00E2\u0089\u00A4 0.5 is an alternate suggestion for applications of cyclic, pulsating or reversing flow (Luettich et al., 1992). From inspection of the laboratory data (see Fig. 4.18b), it appears this criterion also provides a substantial safety margin, given no soil loss in any stage of unidirectional or cyclic flow in a woven or nonwoven geotextile with AOS/D50 \u00E2\u0089\u00A4 2.5. As before, the finding is appropriate to filtration of uniformly-graded soils. 138 4.5.3 Soil retention in cyclic flow Analysis of the laboratory data identifies a general relation between normalized seepage pressure S/pi(0), and a normalized characteristic opening size of the geotextile AOS/Dn (Fig. 4.18c). It provides for a distinction between three domains of (i) no loss, (ii) washout and (iii) piping through the geotextile. The concept is predicated on a S/pi(0) characterizing the hydromechanical demand of the flow regime, and AOS/Dn describing the geometric resistance of the filter zone. The threshold boundary between the domains of no loss and washout appears governed mainly by AOS/Dn. In contrast, the threshold boundary between domains of washout and piping appears to be governed by both AOS/Dn and S/pi(0). In addition, transition from no loss to piping is very sensitive to a change of AOS/Dn when seepage pressure exceeds the initial mean effective stress (S/pi(0) > 1). The novel analytical framework is developed from results of a systematic laboratory study, involving head-control of one-dimensional seepage flow, with testing on reconstituted specimens of uniformly-graded soil. In this regard, it has been developed from a limited database of experience. Accordingly, it is recommended that confidence in the findings be established from applying the framework to explain the response of other soils and geotextiles reported in the literature. 139 4.6 Conclusions A systematic experimental study was conducted on a broad range of filter ratios in the cyclic Gradient Ratio device. Filtration compatibility was evaluated for conditions of unidirectional and cyclic flow under controlled conditions of hydraulic gradient and effective stress, in order to understand the fundamental mechanism of geotextile-soil retention incompatibility. The findings relate to applications of woven and needle- punched nonwoven geotextiles with uniformly-graded base soil (fine sands, silty sand and coarse sandy silt). With reference to a characteristic opening size AOS and an indicative particle size Dn, the following conclusions are drawn: \u00EF\u0082\u00B7 For unidirectional flow at iav < 10 , no unacceptable mass loss was observed for nonwoven geotextiles at 0.7 \u00E2\u0089\u00A4 AOS/D85 \u00E2\u0089\u00A4 2.3 (0.9 \u00E2\u0089\u00A4 AOS/D50 \u00E2\u0089\u00A4 2.7), and woven geotextiles at 1.2 \u00E2\u0089\u00A4 AOS/D85 \u00E2\u0089\u00A4 3.7 (1.3 \u00E2\u0089\u00A4 AOS/D50 \u00E2\u0089\u00A4 4.6). The finding implies retention compatibility; \u00EF\u0082\u00B7 For cyclic flow at iav < 10, with nonwoven geotextiles at 0.7 \u00E2\u0089\u00A4 AOS/D85 \u00E2\u0089\u00A4 2.3 (0.9 \u00E2\u0089\u00A4 AOS/D50 \u00E2\u0089\u00A4 2.7), and with woven geotextiles at 1.2 \u00E2\u0089\u00A4 AOS/D85 \u00E2\u0089\u00A4 2.0 (1.3 \u00E2\u0089\u00A4 AOS/D50 \u00E2\u0089\u00A4 2.5), there was again no unacceptable mass loss observed, which also implies retention compatibility. 140 However for woven geotextiles at AOS/D85 > 2.0, mass loss was observed in some cyclic loading conditions, which suggests: \u00EF\u0082\u00B7 From interpretation of specimen volume change, a value 30 \u00E2\u0089\u00A4 mav \u00E2\u0089\u00A4 50 g/m 2 /100 cycles defines a threshold for soil arching at openings of the geotextile that describes the boundary between washout and onset of incompatibility; \u00EF\u0082\u00B7 At 2.0 < AOS/D85 \u00E2\u0089\u00A4 2.5 (2.5 \u00E2\u0089\u00A4 AOS/D50 \u00E2\u0089\u00A4 3.0), a transition occurs from retention compatible to retention sensitivity; \u00EF\u0082\u00B7 At 2.5 < AOS/D85 \u00E2\u0089\u00A4 3.7 (3.0 \u00E2\u0089\u00A4 AOS/D50 \u00E2\u0089\u00A4 4.6), soil retention is found very sensitive to loading conditions that are governed by hydraulic gradient, stress and wave period; A hydromechanics-based interpretation is proposed to account for the combined influence of hydraulic gradient and effective stress. A normalized value of seepage pressure at the soil-geotextile interface (S/pi(0)) appears to explain the nature of mass loss through pore size openings of the geotextile. For woven geotextiles, data analysis implies that T = 6 s governs the filtration compatibility, for the range 6 s \u00E2\u0089\u00A4 T \u00E2\u0089\u00A4 120 s. The fundamental mechanism of soil retention, from interpretation based on the threshold range of mav (g/m 2 /100 cycles), for the test data at T = 6 s, suggests: 141 \u00EF\u0082\u00B7 A threshold boundary between no loss and washout is governed by AOS/Dn; \u00EF\u0082\u00B7 A threshold boundary between washout and piping is governed by AOS/Dn and S/pi(0); \u00EF\u0082\u00B7 Transition from no loss to piping is very sensitive to a change of AOS/Dn when seepage pressure exceeds the initial mean effective stress (S/pi(0) > 1). The findings of the laboratory experimental program on uniformly-graded soils indicate filtration compatibility in cyclic flow for a nonwoven geotextile at AOS/D50 \u00E2\u0089\u00A4 2.7 or AOS/D85 \u00E2\u0089\u00A4 2.3, and the onset of retention incompatibility for woven geotextile when AOS/D50 > 2.5 or AOS/D85 > 2. Accordingly, the design criterion AOS/D50 \u00E2\u0089\u00A4 1 (Luettich et al., 1992), or alternatively AOS/D85 \u00E2\u0089\u00A4 0.5 (Holtz et al., 1997; CFEM, 2006), appears to provide a very adequate margin of safety for applications involving uniformly graded soil. The novel analytical framework is developed from results of a systematic laboratory study, involving head-control of one-dimensional seepage flow, with testing on reconstituted specimens of uniformly-graded soil. In this regard, it has been developed from a limited database of experience. Accordingly, it is recommended that confidence in the findings be established from applying the framework to explain the response of other soils and geotextiles reported in the literature. 142 Table 4.1 Properties of soils Soil D85 (\u00CE\u00BCm) D50 (\u00CE\u00BCm) 44\u00CE\u00BCm < D < 75\u00CE\u00BCm (%) Cu \u00CE\u00B3t,avg (kN/m 3 ) kav (cm/s) \u00CF\u0085 (degree) 50 kPa 25 kPa < 10 kPa A 180 160 - 1.5 18 0.020 39 41 45 B 150 130 - 1.5 18 0.014 39 41 45 C 106 85 30 \u00E2\u0080\u00A0 1.8 18 0.007 38 40 44 D 80 65 80 \u00E2\u0080\u00A0 1.4 18 0.004 37 40 43 \u00E2\u0080\u00A0 Non-plastic coarse silt 143 Table 4.2 Properties of geotextiles Geotextile code Fabric structure AOS (O95) (microns) POA (%) ASTM Permittivity (s -1 ) Thickness (mm) Unit mass (g/m 2 ) NW1 NW-NP 60 N/A 2.5 \u00E2\u0080\u00A0 4.0 1200 NW2 NW-NP 75 N/A 1.0 4.0 500 NW3 NW-NP 80 N/A 8 \u00E2\u0080\u00A0 4.5 600 NW4 NW-NP 180 N/A 1.7 2.5 285 NW5 NW-NP 180 N/A 1.1 2.3 278 W1 W-M 210 4-6 0.3 N/A 190 W2 W-MT 300 4 0.5 N/A 170 Remarks: NW-NP = needle-punched nonwoven geotextile, W = woven, M = monofilament, MT = multifilament N/A = data not available, \u00E2\u0080\u00A0 = according to ISO standard 144 Table 4.3 Test combinations (and AOS/D85) Geotextile (AOS) Soil NW1 (60\u00C2\u00B5m) NW2 (75\u00C2\u00B5m) NW3 (80\u00C2\u00B5m) NW4 (180\u00C2\u00B5m) NW5 (180\u00C2\u00B5m) W1 (210\u00C2\u00B5m) W2 (300\u00C2\u00B5m) A T6 (1.0) T6, T60 (1.2) B T6, T60 (1.4) T6, T60 (2.0) C T6 (0.7) T6 (1.7) T6 (1.7) T6 (2.0) T6, T60, T120 (2.8) D T6 (0.8) T6 (0.9) T6,T60 (2.3) T6 (2.3) T6,T60, T120 T6 (3.7) (2.6) Remarks: 1) Test code: A-W1-T6 describes soil A and geotextile W1, at T = 6 s. 2) Bold letter denotes a repeated test 145 Table 4.4 Summary of modified values of Gradient Ratio (GR8) and qualitative observation of mass loss Geotextile (AOS) Soil NW1 (60\u00C2\u00B5m) NW2 (75\u00C2\u00B5m) NW3 (80\u00C2\u00B5m) NW4 (180\u00C2\u00B5m) NW5 (180\u00C2\u00B5m) W1 (210\u00C2\u00B5m) W2 (300\u00C2\u00B5m) A 0.9 No 1.2 No B 1.3 No 1.1 No C 1.7 No 0.9 No 0.9 No 1.5 No 0.9 \u00E2\u0080\u0093 1.1 T6 Yes T60 Yes T120 Yes D 1.4 No 0.9 No 1.0 No 1.3 No 1.0 - 1.3 0.7 T6 Yes T6 Yes T60 Yes T120 Yes Remarks: \u00E2\u0080\u009CNo\u00E2\u0080\u009D denotes for no loss or negligible cumulative quantity (< 0.2 g \u00E2\u0089\u0088 25 g/m2) Bold denotes test repeated 146 Table 4.5 Average mass loss (mav) in CYC stages (g/m 2 /100 cycles) Test code T=6s for confining stress, pi(0) (kPa) T=60s for confining stress, pi(0) (kPa) T=120s for confining stress, pi(0) (kPa) 23 14 3 23 14 3 23 14 3 C-W2 iav 1.3 1.3 1.3 1.1 1.1 1.1 1.2 1.2 1.2 mav 52 51 48 21 24 38 0 0 11 iav 5.7 5.7 5.7 5.1 5.1 5.1 5.5 5.5 5.5 mav 97 93 100 43 41 71 14 23 26 iav 9.5 9.5 9.5 8.9 8.9 8.9 9.3 9.3 9.3 mav 116 129 164 84 76 127 20 34 57 C-W2 (repeated) iav 1.1 1.1 1.1 1.0 1.0 1.0 mav 36 61 72 6 14 19 - - - iav 5.1 5.1 5.1 5.2 5.2 5.2 mav 71 74 74 24 38 41 - - - iav 9.1 9.1 9.1 8.7 8.7 8.7 mav 76 92 120 47 45 85 - - - D-W1 iav 1.1 1.1 1.1 1.2 1.2 1.2 1.1 1.1 1.1 mav 11 11 13 9 10 17 0 0 20 iav 5.2 5.2 5.2 5.3 5.3 5.3 5.4 5.4 5.4 mav 38 78 81 16 24 44 23 26 34 iav 9.5 9.5 9.5 9.4 9.4 9.4 9.6 9.6 9.6 mav 76 78 91 47 67 78 26 29 57 D-W1 (repeated) iav 1.1 1.1 1.1 1.0 1.0 1.0 mav 15 27 30 21 34 43 - - - iav 5.3 5.3 5.3 5.2 5.2 5.2 mav 67 69 90 46 77 85 - - - iav 9.7 9.7 9.7 9.7 9.7 9.7 mav 103 116 143 97 104 131 - - - D-W2 iav 1.3 1.3 1.3 mav 150 203 228 - - - - - - iav 5.4 5.4 mav 299 316 - - - - - - - iav mav - - - - - - - - - Remark: highlights denote data plotted in Figure 4.15 147 (a) (b) (c) Figure 4.1 Soils: a) photograph of Fraser River sand; b) photograph of Alouette River sand; c) grain size distribution curves 148 (a) (b) (c) (d) Figure 4.2 Cyclic Gradient Ratio device: a) permeameter; b) head-control system; c) schematic stress distribution; d) relation between top stress on the specimen and mean stress at the soil-geotextile interface (after Srikongsri and Fannin, see chapter 3) P 1 Z Axial load LVDT Top inlet-outlet P 3 P 5 P 6 Silicone hose (discrete sample) Soil collection trough Bottom inlet-outlet P 7 Geotextile Soil Upward flow Downward flow -H +H Soil 3-way valve Geotextile 149 Figure 4.3 Multi-stage test procedure (test C-W2-T6) at iav \u00E2\u0089\u0088 9 1b 2a b UNI 2b 3a b UNI 3b UNI UNI 1a Time CYC-1 CYC-2 CYC-3 Reduce stress Reduce stress pi(0) \u00E2\u0089\u008823 kPa pi(0) \u00E2\u0089\u00883 kPa pi(0) \u00E2\u0089\u008814 kPa 150 (a) (b) Figure 4.4 Relation between soil passing and filter ratio for stage CYC2: a) at iav = 1 and b) at iav \u00E2\u0089\u0088 5 151 (a) (b) Figure 4.5 SEM images of needle-punched geotextiles: a) new NW4; b) tested NW4 (from D-NW4-T6); c) new NW5; d) tested NW5 (from D-NW5-T6) 152 (c) (d) Figure 4.5 (continued) 153 (a) (b) Figure 4.6 Mass loss at AOS/D85 = 2.6 for T = 6 s: a) test D-W1 and b) repeated D-W1 154 (a) (b) Figure 4.7 Mass loss at AOS/D85 = 2.6 for T = 60 s: a) test D-W1-T60 and b) repeated D-W1-T60 155 Figure 4.8 Mass loss at AOS/D85 = 2.6 for test D-W1-T120 156 (a) (b) Figure 4.9 Mass loss at AOS/D85 = 2.8 for T = 6 s: a) test C-W2-T6 and b) repeated C-W2-T6 157 (a) (b) Figure 4.10 Mass loss at AOS/D85 = 2.8 for T = 60 s: a) test C-W2-T60 and b) repeated C-W2-T60 158 Figure 4.11 Mass loss at AOS/D85 = 2.8 for test C-W2-T120 159 (a) (b) Figure 4.12 results at AOS/D85 = 3.7 for test D-W2-T6: a) mass loss; b) end-of-test photograph 160 (a) (b) Figure 4.13 SEM images of woven geotextiles: a) tested W1 (from D-W1-T6); b) tested W2 (from C-W2-T6) 161 (a) (b) Figure 4.14 Particle bridging: a) effect of particle shape on vibration-based stability (modified from Valdes and Santamarina, 2008); b) conceptual regime for mechanical instability of spherical particles d0/d \u00E2\u0089\u0088 2 d0/d \u00E2\u0089\u0088 1.5 d0/d \u00E2\u0089\u0088 2.5 d0/d \u00E2\u0089\u0088 3 Stable bridge Mechanical instability Transition no bridge formation no bridge formation vibration-sensitive bridge vibration-sensitive bridge stable bridge stable bridge 162 (a) (b) Figure 4.15 Inspection of retention compatibility (data from tests D-W1-T6, D-W1- T6-R, C-W2-T6 and C-W2-T6-R): a) piping; b) washout 163 (a) (b) (c) (d) Figure 4.16 Onset of piping: a) soil D - geotextile W1; b) soil C \u00E2\u0080\u0093 geotextile W2; c) soil D \u00E2\u0080\u0093 geotextile W2; d) concept of hydromechanical stability 164 Figure 4.17 Hydromechanical influences on soil retention in cyclic flow for woven geotextiles (data for T = 6 s from Fig. 4.16a, 4.16b, 4.16c and test C-W1-T6) 165 (a) (b) (c) Figure 4.18 Retention compatibility for uniformly-graded soil and wave period T = 6 s: a) AOS/D85 and b) AOS/D50; c) characteristic zone of soil retention Washout Piping No loss AOS/Dn S/pi(0) 166 4.7 References Canadian Geotechnical Society (2006). Canadian Foundation Engineering Manual, 4th edition, BiTech Publishers Ltd. Cazzuffi, D. A., Mazzucato, A., Moraci, N., & Tondello, M. (1999). A new test apparatus for the study of geotextiles behaviour as filters in unsteady flow conditions: relevance and use. Geotextiles and Geomembranes 17, No. 5-6, 313 - 329. Chew, S. H., Zhao, Z. K., Karunaratne, G. P., Tan, S. A, Delmas, Ph., & Loke, K. H. (2000). Revetment geotextile filter subjected to cyclic wave loading. Proceedings of Geo-Denver 2000, Denver, CO, USA, pp. 162 \u00E2\u0080\u0093 175. Christopher, B.R., and Holtz, R.D. (1985), Geotextile engineering manual, Report No. FHWA- TS-86/203, Federal Highway Administration, D.C., USA, 1044 p. Fannin, R.J. (2007). Chapter 6: The use of geosynthetics as filters. Geosynthetics in Civil Engineering. Woodhead Publishing, Cambridge, UK, 295p. Hameiri, A. (2000). Soil geotextile filtration behavior under dynamic conditions of vibration and cyclic flow. PhD Thesis, University of British Columbia, British Columbia, Canada, 270p. 167 Hameiri, A. & Fannin, R. J. (2002). A cyclic gradient ratio test device. ASTM Geotechnical Testing Journal, 39, No.2, 266-276. Hawley, R. (2001). Filtration performance of geotextiles in cyclic flow conditions. MASc Thesis, University of British Columbia, Vancouver, B.C., Canada, 141p. Heerten, G. (1982). Dimensioning the filtration properties of geotextiles considering long- term conditions. Proceedings of 2 nd international conference on geotextiles, Las Vegas, NV, USA, pp. 115 \u00E2\u0080\u0093 20 Holtz, R.D, Christopher, B.R., and Berg, R.R. (1997). Geosynthetic Engineering. BiTech Publishers, Richmond, BC, Canada, 452 p. Indraratna, B & Vafai, F (1997). Analytical model for particle migration within base soil-filter system, Journal of Geotechnical and Geoenvironmental Engineering, 123, No. 2, 100-109. Indraratna, B & Radampola, S. (2002). Analysis of Critical Hydraulic Gradient for Particle Movement in Filtration. Journal of Geotechnical and Geoenvironmental Engineering, 128, No. 4, 347-350. Luettich, S.M., Giroud, J.P., and Bachus, R.C. (1992). Geotextile filter design guide. Geotextiles and Geomembranes, 11, 355 - 370. 168 Pilarczyk, K. W. (2000). Geosynthetics and geosystem in hydraulic and coastal engineering. A.A. Balkema, Rotterdam, The Netherlands, 913 p. Reddi, L. N. (2003). Seepage in soils principles and applications. John Wiley & Sons, NJ, USA, 402 p. Shukla, S. K., Yin, J-H. (2006). Fundamentals of geosynthetic engineering. Taylor & Francis/Balkema, Natherlands, 410 p. Srikongsri, A. and Fannin, R.J. (2009). Retention capacity of geotextile filters in cyclic flow. Proceedings of Geosynthetics 2009, Salt Lake City, UT, USA, pp. 498-508. Srikongsri, A. and Fannin, R.J. (2010a). Soil-geotextile compatibility testing in cyclic flow. Manuscript draft, see Chapter 2. Srikongsri, A. and Fannin, R.J. (2010b). Influences of testing methodology on soil-geotextile compatibility in cyclic flow using rigid-wall permeameters. Manuscript draft, see Chapter 3. Valdes, J. R. and Satamarina, J. C. (2008). Clogging: bridge formation and vibration-based destabilization. Canadian Geotechnical Journal. 45, No. 2, 177-184. 169 5 Retention Criteria for Geotextile Filter in Cyclic Flow 4 5.1 Outline A hydromechanical framework, for retention compatibility of soil-geotextile filters in cyclic flow, is evaluated through comparison with (i) laboratory test data from three independent studies and (ii) field performance data from four bank and coastal protection sites. The framework is found to distinguish between filter compatibility, and conditions leading to onset of a piping action through the geotextile. The AOS/D85 filter ratio appears more suitable than AOS/D50 as an indicative particle size for design. For a woven geotextile, the onset of piping occurs at 2 \u00E2\u0089\u00A4 AOS/D85 \u00E2\u0089\u00A4 2.5. In design for soil retention involving cyclic or dynamic flow, it is recommended that AOS \u00E2\u0089\u00A4 D85 in order to achieve an adequate margin of safety. The recommendation is empirical, and appropriate for a uniformly-graded non-plastic base soil that is internally stable. 4 A version of this chapter will be submitted for publication. Srikongsri, A., Fannin, R. J. (2010). Retention criteria for geotextile filter in cyclic flow. 170 5.2 Introduction The nature of dynamic flow, especially cyclic flow reversal, is believed to reduce the inherent margin of safety that is provided by retention criteria intended for applications of unidirectional flow. For example, if a criterion for unidirectional flow requires OF < BDi, where OF is a characteristic pore size opening of the geotextile, B is a constant (normally \u00E2\u0089\u0088 1 \u00E2\u0080\u0093 2) and Di is an indicative particle size of the base soil, then the criterion for cyclic flow may simply take the form of OF < 0.5BDi. Accordingly, Luettich et al. (1992) suggests (OF = O95)/(Di = D50) < 1 and Holtz et al. (1997) suggests (OF = O95)/(Di = D85) < 0.5. These criteria are currently used in the USA and the latter criterion is also adopted for recommendation in the current edition of the Canadian Foundation Engineering Manual. In applying this concept, the belief is that OF retains Di and, if Di retains the remainder of particles in the gradation curve, then filtration compatibility is achieved. Undoubtedly, the concept works for a certain range of soil gradations with relatively coarse grains. However, when the base soil to be protected has a relatively small Di, for example a fine sand or silt, then a geotextile may simply be disqualified as a candidate filter medium because the smallest available OF cannot satisfy the design filter ratio OF/Di (Klein Breteler and Verhey, 1990; Klein Breteler et al. 1994). In contrast, Dutch experience with geotextiles in filtration applications for coastal defense structures using an open filter (i.e. OF/Di > 1), yields a different approach for conditions of cyclic flow wherein (OF = O98)/(Di = D85) < 1 to 2, 171 for cohesionless soil with Cu < 6 to 10 (Pilarczyk, 2000). In addition, the current Dutch design practice also recommends the criterion (OF = O90)/(Di = D90) \u00E2\u0089\u00A4 1, for an internally stable base soil with D40 > 60 \u00CE\u00BCm (modified from Heerten, 1982 as reported by Pilarczyk, 2000). Note that the findings of Heerten (1982) have been briefly mentioned in Srikongsri and Fannin (2010b: see chapter 4). These two criteria are provided in a revetment design manual, which can be found at the website of the Expertise Network for Flood Protection (Expertise Netwerk Waterveiligheid, ENW) using the link: http://www.tawinfo.nl/engels/downloads/DesignRevetments.pdf. A hydromechanical framework has been proposed for retention compatibility. It relates filter ratio (AOS/D50 or AOS/D85) and normalized seepage pressure (S/pi(0)), based on laboratory test data from a systematic study of soil-geotextile compatibility in cyclic flow (Srikongsri and Fannin, (2010b: see chapter 4). Seepage pressure (S) is calculated from: ZiS wav\u00EF\u0081\u00A7\u00EF\u0080\u00BD (1) where iav is average hydraulic gradient across the soil specimen, \u00CE\u00B3w is unit weight of water and Z is specimen length. In a rigid-wall permeameter, the initial mean effective 172 stress (pi(0)) or \u00E2\u0080\u009Cconfining stress\u00E2\u0080\u009D is deduced from the vertical effective stress applied to the top of the specimen (Srikongsri and Fannin, (2010a: see chapter 3), which is calculated as: pi(0) = \u00CF\u0083\u00CD\u00B4 vb(0) (1+2K0)/3 (2) where \u00CF\u0083\u00CD\u00B4 vb(0) is vertical effective stress at the geotextile and K0 is an at-rest coefficient of lateral stress. A schematic illustration of the hydromechanical framework (see Fig. 5.1) depicts the general experimental finding that onset of retention instability in a woven geotextile occurs at 2.4 < AOS/D50 or 2 < AOS/D85. The fact that needle-punched nonwoven geotextiles were found retention compatible at AOS/D50 \u00E2\u0089\u00A4 2.7 or AOS/D85 \u00E2\u0089\u00A4 2.3 in the same experiments lends additional confidence to the finding. Consider now the criteria of Luettich et al. (1992) in Fig. 5.1a and Holtz et al. (1997) in Fig. 5.1b, which both locate in the lower part of the \u00E2\u0080\u009Cno loss\u00E2\u0080\u009D zone in the hydromechanical framework and are likely overly conservative. Consider also the more general Dutch approach represented in Fig. 5.1b, assuming O98 \u00E2\u0089\u0088 AOS. Again, the hydromechanical framework appears reasonably supportive of this approach at relatively low values of normalized seepage pressure. At higher values of S/pi(0), for example at S/pi(0) \u00E2\u0089\u00A5 1 (see Fig. 5.1b), the upper threshold of (O98 \u00E2\u0089\u0088 AOS)/D85 < 2 provides for a little or no margin of safety. Therefore, it seems prudent to undertake an independent evaluation 173 of the proposed hydromechanical framework for retention compatibility in cyclic flow through comparison with (i) other published laboratory test data and (ii) field performance data reported in the literature. Notable laboratory studies on the nature of soil-geotextile compatibility in cyclic flow are reported by Cazzuffi et al. (1999), Chew et al. (2000) and Hawley (2001). Given the limited test data used in support of the findings of Srikongsri and Fannin (2010b: chapter 4), a comparison with test data from three additional laboratory studies is of tremendous benefit. In addition, field performance data reported by Mannsbart and Christopher (1997) from four bank and coastal protection sites enable further evaluation of the hydromechanical framework, and address the confidence with which it can be considered for use in engineering practice. 5.3 Select laboratory test data Grain size distribution curves of five soils from three experimental studies are shown in Figure 5.2. Three soils, namely Fraser River Sand (FS), Mine Waste Tailings (MT) and Port Coquitlam Sand (PC) are examined by Hawley (2001): they were used in laboratory testing at UBC, in work to evaluate filtration compatibility of two needle- punched nonwoven geotextiles and five woven geotextiles, in an investigation that preceded the current study. One soil, Beach sand (BS), was tested against a nonwoven 174 and a woven geotextile, in work to examine the influence of hydraulic gradient and effective stress in an investigation by Cazzuffi et al. (1999). Finally, tests on sand used for reclamation purposes (RS) are reported from an investigation by Chew et al. (2000), involving one nonwoven geotextile and one woven geotextile. Notwithstanding the fact that all three test programs invoked different test methods, the results are still useful for purposes of evaluating the proposed hydromechanical framework. All five gradation curves from the three separate investigations are narrowly graded (Cu \u00E2\u0089\u00A4 6), and deemed internally stable soil according to the method of Kenney and Lau (1985 and 1986) and Li and Fannin (2008). Properties of the soils are summarized in Table 5.1. The characteristic pore size opening of a geotextile may be established by a dry sieving technique (Apparent Opening Size (AOS), according to ASTM D-4751). However, there are also a number of different standardized methods used in various countries that involve techniques of wet, or alternatively hydrodynamic sieving, that define a similar, but not identical, value of O95 (see Table 5.2). For a woven geotextile, the index values of opening size are believed reasonably comparable. In contrast, they can be significantly different for a nonwoven geotextile. Typically, the AOS value is found to be larger than those obtained from wet or hydrodynamic sieving (Bathia et al. 1996). 175 5.3.1 Hawley (2001) The study was undertaken at UBC in the same cyclic gradient ratio test device used by Srikongsri and Fannin (2010a: see chapter 3), to examine the filtration compatibility of three sands and seven geotextiles. The resulting database comprises twenty one test combinations examining a filter ratio in the range 0.8 < AOS/D50 < 3.4 and 0.6 < AOS/D85 < 2.8, for two needle-punched nonwoven and five woven geotextiles. Tests were conducted at only one average hydraulic gradient (iav) of approximately 4. Test variables examined comprise vertical effective stress on the top surface of the soil specimen (\u00CF\u0083vt \u00CC\u0081= 25 and 0 kPa, namely unloaded), and the wave period of cyclic flow reversal (T = 50 or 10 s). The test sequence involved a relatively long stage of cyclic flow at T = 50 s (1080 cycles) that was followed by a shorter stage at T = 10 s (260 cycles), whereupon the normal stress was reduced from 25 kPa to zero, and the shorter stage at T = 10 s then repeated (260 cycles). Each cyclic stage was preceded and followed by a stage of unidirectional flow. Interpreting data in the hydromechanical framework requires the applied top stress and the imposed hydraulic gradient to be expressed as a value of S/pi(0). Srikongsri and Fannin (2010b: see chapter 4) established a relation between \u00CF\u0083vt \u00CC\u0081 and pi(0) for a uniformly-graded fine sand: this relation is assumed, for purposes of analysis, to apply for soils FR, MT and PC. Accordingly, the combinations of iav \u00E2\u0089\u0088 4 at \u00CF\u0083vt \u00CC\u0081= 25 kPa and 0 kPa yield values of S/pi(0) = 0.37 and 4.2, respectively. 176 The combination of soil type FS and seven geotextiles in the range 1 < AOS/D50 < 2.2 or 0.6 < AOS/D85 < 1.8 was reported as retention compatible with essentially no soil loss through the geotextile (Hawley, 2001). Soil MT is also reported compatible with no loss in the range AOS/D50 < 2.4 or AOS/D85 < 1.5, but evidence of some mass loss was reported in the range 2.4 \u00E2\u0089\u00A4 AOS/D50 \u00E2\u0089\u00A4 3.4 or 1.5 \u00E2\u0089\u00A4 AOS/D85 \u00E2\u0089\u00A4 2.1 (see Table 5.3). In the latter case, the losses were deemed necessary for development of filter stability and are not believed to represent retention incompatibility of the soil-geotextile combination: it was concluded that all of these test combinations were also stable. A companion analysis of the data is made using AOS/D50 (see Fig. 5.3a) and AOS/D85 (see Fig. 5.3b), which enables comparison with the hydromechanical framework developed in this study. Given the observed filtration compatibility that was common to all test combinations, the trend appears to be better explained by the AOS/D85 value rather than the AOS/D50 value. The filter ratio AOS/D85 = 2.1 (test MT-G60W, Table 5.3) at which washout occurred appears to agree very well with that anticipated by the hydromechanical framework (see Fig. 5.3b and Fig. 5.1b). Soil PC was reported stable for a filter ratio AOS/D50 \u00E2\u0089\u00A4 1.7 or AOS/D85 \u00E2\u0089\u00A4 1.4. However, results of tests on two woven geotextiles at AOS/D50 = 2.4 or AOS/D85 = 2.0 (tests PC-G43aW and PC-G43bW) exhibited washout at the stage with \u00CF\u0083vt \u00CC\u0081= 25 kPa at T = 10 s and subsequently exhibited a piping failure that involved excessive amounts of mass loss (see Table 5.3) in the test stage with \u00CF\u0083vt \u00CC\u0081 = 0 kPa, for which 177 S/pi(0) = 4.2 (see Fig. 5.4). Additionally, one planned test involving a woven geotextile at AOS/D50 = 3.3 or AOS/D85 = 2.8 exhibited an excessive soil loss through the geotextile during reconstitution of the soil specimen by water pluviation against the geotextile, which it is clearly a case of retention incompatible. For plotting purposes, the data point for this latter soil-geotextile combination is nominally located at S/pi(0) = 0.01, which places it in the washout zone of the framework (see Fig. 5.4). Generally, the piping response with soil PC is found at AOS/D50 and AOS/D85 values that plot at or below the curve-linear relation postulated from the current study, which is based on testing soil of uniform gradation. The response is attributed to the relatively broad gradation curve of the PC soil compared to the others (see Table 1) yielding a different bridging structure at the soil-geotextile interface, and one that may be more susceptible to disturbance by cyclic flow. The grain size distribution reported for soil passing through the G43aW and G43bW geotextile (AOS = 425 \u00CE\u00BCm, see Table 5.2) was predominantly of fines fraction less than 70 \u00CE\u00BCm (see Fig. 5.5). 5.3.2 Cazzuffi et al. (1999) The data of Cazzuffi et al. (1999) described results from cyclic flow tests on one soil, a uniform fine sand (D85 = 0.2 mm and D50 = 0.14 mm, Cu = 1.4 and no presence of fines, see Fig. 5.2), in combination with two types of geotextile, a nonwoven (O95 = 0.16 mm) or a woven (O95 = 0.44 mm) geotextile. The laboratory tests were performed in a bi-directional flow apparatus, with a permeameter that accommodates a 178 cylindrical specimen 300 mm in diameter and 400 mm long. In testing, the influence of vertical effective stress at the soil-geotextile interface was examined the range 4 to 154 kPa, for a hydraulic gradient in the range 3 to 16, and an unspecified wave period in the range 2 to 20 s, for a test duration of typically 1500 cycles. At a filter ratio O95/D50 = 1.1 or O95/D85 = 0.8, for the nonwoven geotextile, a negligible washout less than 10 g/m 2 /100 cycles was reported and the combination was deemed retention compatible (see Table 5.4). At a filter ratio O95/D50 = 3.1 or O95/D85 = 2.2, for the woven geotextile, the response was more subtle: mass loss for this BS- G44W test was negligible at \u00CF\u0083vt \u00CC\u0081\u00E2\u0089\u0088 50 kPa, but was reported to increase dramatically when the top stress was reduced to zero (only self weight applies to the geeotextile) yielding an approximate pi(0 = 2.4 kPa (see Table 5.4, for an assumed K0 = 0.42). More specifically, consider the mass loss of less than 30 - 50 g/m 2 /100 cycles at S = 11.7 kPa and pi(0) = 2.4 kPa, that increased to a mass loss of 323 g/m 2 /100 cycles when S was increased to 19.6 kPa. The finding suggests the onset of retention incompatibility was triggered at 4.9 < S/pi(0) < 8.2 (see Table 5.4). A separate presentation of the data is made using AOS/D50 (see Fig. 5.6a) and AOS/D85 (see Fig. 5.6b), which enables comparison with the hydromechanical framework developed in the current study. Given the filtration compatibility that was common to all tests except one, once again the trend appears to be moderately better explained by the AOS/D85 value rather than the AOS/D50 value. 179 5.3.3 Chew et al. (2000) The testing of Chew et al. (2000) was performed using a slightly modified version of the apparatus developed by Cazzuffi et al. (1999). The primary objective of the work was to observe the influence of wave period in the range T = 2 to 15 s. The tests were performed at \u00CF\u0083vt \u00CC\u0081= 113 kPa and a maximum hydraulic gradient of 26, for a typical test duration of 2000 to 2500 flow cycles. The data describe results from cyclic flow tests on one soil, a coarse sand (D85 = 2 mm and D50 = 0.9 mm, Cu = 3.3 and fines < 2%, see Fig. 5.2), in combination with two types of geotextile, a nonwoven (O95 = 0.1 mm) or a woven (O95 = 0.47 mm) geotextile. At a filter ratio O95/D50 = 0.1 or O95/D85 = 0.05, for the nonwoven geotextile (see Fig. 5.6), a negligible washout was reported and the combination was deemed retention compatible. Likewise, the same conclusion was drawn for the woven geotextile (see Fig. 5.6b), at filter ratio O95/D50 = 0.52 or O95/D85 = 0.24. The values of filter ratio AOS/D50 and AOS/D85 examined by Chew et al. (2000) are the lowest of all test data compared to evaluate the proposed hydromechanical framework, and are found to be in good agreement with it. 180 5.4 Field data Mannsbart and Christopher (1997) report on the long-term field performance of geotextiles used as a filter in coastal and bank protection applications. They provide schematic drawings, together with information on the grain size distribution curve of each base soil, material properties of each geotextile and describe, albeit in very general terms, the severity of the hydraulic action at each site location. Filtration compatibility was considered satisfied at all sites, based on the fact that no evidence was found to indicate concern either for clogging or piping activity over the service life of the installation at that time. Taken collectively, it yields a sufficiently comprehensive record to allow for comparison with the proposed hydromechanical framework. All of the geotextiles were needle-punched nonwoven fabrics, samples of which were exhumed for forensic analysis. Pore size of the fabric was reported with reference to values of O90, based on a wet sieving technique, and therefore values of AOS were sourced from the manufacturer\u00E2\u0080\u009Fs technical database for the purposes of this study (see Table 5.5a), enabling determination of the corresponding filter ratio AOS/D50 and AOS/D85. From the drawings provided for each site (see, for example, Fig. 5.7), an approximate value of initial confining stress (pi(0)) at the soil-geotextile interface was calculated based on thickness of the overlying armor layer (assuming a unit weight of 15 kN/m 3 , 181 and K0 = 0.42). An approximate value of seepage pressure at the soil-geotextile interface was established indirectly. Wave-generated pressure on the armor layer is a function of wave characteristic and geometry of the slope. It is a stochastic phenomenon. This value may be assumed equal to a seepage pressure (S) for the purpose of defining S/pi(0) at particular field location. Pilarczyk (2000) proposed a simplified relation between the maximum value (Pmax) of wave-generated pressure and a significant wave height (Hs): swHAPS \u00EF\u0081\u00A70max \u00EF\u0080\u00BD\u00EF\u0080\u00BD (3) where A0 is an empirical factor which may be obtained by experiments, \u00CE\u00B3w is the unit weight of water, and Hs is the significant wave height or design value of wave height. The value of A0 is a function of wave characteristic that accounts for the influence of both hydrostatic and hydrodynamic components of wave energy. Pilarczyk (2000) suggested, for calculation purposes, that the value of A0 can be assumed equal to 2 as an approximation for a value of Pmax within the armor stone; in a filter layer (the soil- geotextile interface), the wave impact will be partly damped by the armor stone, in which case a value of A0/2 is suggested. 182 A value of Hs is not reported for each site in the work of Mannsbart and Christopher (1997). However, it is possible to obtain an approximate value of Hs by indirect means, knowing the characteristic size (or mass) of the armor stone. The Hudson formula (USACoE, 1984) relates Hs to mass of the median rock size (W50): \u00EF\u0081\u00B1cot)1( 3 50 \u00EF\u0080\u00AD \u00EF\u0080\u00BD rD sr GK Hw W (4) Where KD is the stability coefficient, wr is the density of rock mass, Gr is the specific gravity of rock, and \u00CE\u00B8 is the slope of the rip-rap. The value of KD varies significantly with type and shape of armor materials, whether the wave is non-breaking or breaking, and method of armor placement. Breaking waves may result in large pressure variation. At all sites reported herein, inspection of the drawings indicates the stones are angular in shape, and randomly placed. Assuming a condition of breaking wave leads to KD of approximately 2.2 (USACoE, 1984). Values of wr = 2650 kg/m 3 and Gr = 2.65 were assumed for all four sites. Establishing W50, again from inspection of the drawings, allows for a back-calculation of Hs (Eq. 4). Knowing Hs, a value of Pmax is then back-calculated (Eq. 3), and used to determined the ratio S/pi(0) (see Table 5.5b). 183 Each site is reported with reference to the ratio AOS/D50 and AOS/D85, and for purposes of comparison, O90/D50 and O90/D85 (see Fig. 5.8). Therefore, each site yields a total of four data points. The range of S/pi(0) depicted for each point (horizontal error bar) results from the constant of A0 having a lower bound A0 =1 and upper bound of A0 = 2, which is believed appropriate given the uncertainty in determination of normalized seepage pressure. It is not feasible to provide a companion error bar for the AOS/Dn values. By definition 95% of pore size openings are smaller than the AOS value, but the range is unknown. Furthermore, the spatial variations in grain size of the base soil are unknown. Accordingly, it is not possible to assign a range of AOS/Dn, as was done for S/pi(0). All four sites yields a value of S/pi(0) \u00E2\u0089\u0088 1 (see Table 5.5b), which implies a seepage pressure that likely results in very low effective stress in the base soil adjacent to the geotextile. Three of the four sites exhibit a relatively low AOS/D85 \u00E2\u0089\u00A4 0.5 (and AOS/D50 \u00E2\u0089\u00A4 1) and one, at Sungai Buntu, a relatively large value of AOS/D85 \u00E2\u0089\u0088 1.5 (and AOS/D50 \u00E2\u0089\u0088 5). Recall all four sites are deemed filtration compatible, based on field observations (Mannsbart and Christopher, 1997). From comparison of the AOS/D50 (see Fig. 5.8a) and AOS/D85 (see Fig. 5.8b) filter ratio, and the postulated curve-linear threshold to onset of piping (shown dashed) it appears, once again, that the AOS/D85 values provide a better agreement to the proposed hydromechanical framework. 184 5.5 A recommended criterion for soil retention Verification of the hydromechanical framework with the selected additional laboratory and field studies implies that AOS/D85 is more suitable than AOS/D50 as an indicative filter ratio for design. The finding is consistent with the observation of Watson and John (1999) who established, from statistical analysis, that particles in the size range D70 to D90 govern the formation of a soil bridge at the openings of a filter. Therefore, all laboratory test data, including test data of the current study by Srikongsri and Fannin (2010b: see chapter 4), and the field data (section 5.4) are reproduced in Figure 5.9, with reference to an AOS/D85 filter ratio. Recall the Holtz et al. (1997) criterion of (O95 or AOS)/D85 \u00E2\u0089\u00A4 0.5 that was modified from earlier work reported by Christopher and Holtz (1985): its development was not supported by any systematic mechanics-based study of soil-geotextile behaviour in dynamic, pulsating or cyclic flow. The unified plot of AOS/D85 versus S/pi(0) confirms that the current design guidance used in North America is conservative for these uniformly-graded soils. Comparison of the database and the empirical criteria suggests there is sufficient conservatism to warrant a revision to the criterion. 185 Recall the Dutch approach that a filter ratio, O98/D85 < 1 to 2 be used for cohesionless soil with Cu < 6 to 10. Although, the upper bound value of 2 appears to provide a broader choice for selecting a candidate geotextile, it is deemed too close to the zone of soil piping for S/pi(0) \u00E2\u0089\u00A5 1. Over the service life of a rip-rap revetment, the filter is expected to occasionally experience a very large S/pi(0). Hence, the lower bound O98/D85 < 1 is considered more appropriate as it provides an adequate margin of safety for the uncertainty in hydraulic loads (see Fig. 5.9). In summary, the newly proposed hydromechanical framework of AOS/D85 (a geometric index of capacity) versus S/pi(0) (a hydromechanical index of demand) both demonstrates and explains the nature of conservatism in empirical criteria for soil retention in cyclic flow, with reference to a qualitative margin of safety against onset of piping. In addition, with the assumption that AOS \u00E2\u0089\u0088 O98, it enables a comparison with experience reported from practice in the Netherlands, which likewise has not been developed with reference to any systematic mechanics-based study of soil-geotextile behaviour in dynamic, pulsating or cyclic flow. Accordingly, and with recognition of the Holtz et al. (1997) and Pilarczyk (2000) guidance, it is proposed that the criterion O95 or AOS/D85 \u00E2\u0089\u00A4 0.5 be relaxed to yield a recommended criterion of AOS \u00E2\u0089\u00A4 D85. The recommended criterion is appropriate for a 186 uniformly-graded (Cu \u00E2\u0089\u00A4 6) non-plastic base soil that is internally stable. It provides a significant margin of safety against piping failure, given the difference between AOS/D85 = 1 and the filter ratio AOS/D85 \u00E2\u0089\u0088 2 at which retention incompatibility is found to initiate for a woven geotextile (see Fig. 5.9). The relative margin of safety is believed greater for a nonwoven geotextile, and attributed to the more tortuous nature of the pore size distributions and related constrictions. In all likelihood, entrapment of soil particles may occur with time over the service life of the geotextile filter, particularly as a result of wave loading and episodic fluidization of the soil immediately adjacent to the fabric, where the contact is not intimate. It is believed that any such entrapment of soil particles makes the retention criterion yet more conservative. Use of the recommended criterion, and indeed any criterion for soil retention, is made on the basis that adequate site supervision during handling and installation of the geotextile eliminates the possibility of mechanical damage yielding any significant increase in the characteristic pore size opening. 187 5.6 Conclusions Analysis of laboratory test data from three studies, and select field data from four project sites, is made with reference to attributes of the base soil (D50 and D85), the pore size opening of the geotextile (AOS), an estimate of likely seepage pressure (S) from cyclic flow, and initial mean effective stress (pi(0)) in the base soil. The three laboratory studies involve both woven and needle-punched nonwoven geotextiles, in filtration applications that include both compatible and incompatible soil-geotextile combinations. The four field applications involve needle-punched nonwoven geotextiles that are compatible soil-geotextile combination. Accordingly, the database provides opportunity to critically evaluate the merits of a proposed hydromechanical framework, for retention compatibility in cyclic flow, based on a relation between filter ratio (AOS/D85, and AOS/D50) and normalized seepage pressure (S/pi(0)). Generally good agreement is found between the proposed framework and the database of independent laboratory and field experience. The framework is found to distinguish between filter compatibility with some washout, and filter incompatibility in the form of a piping action through the geotextile. Inspection of the data suggests that, AOS/D85 is generally more suitable than AOS/D50 as an indicative filter ratio for design. 188 More specifically, and with reference to AOS/D85 filter ratio, analysis of the laboratory test data suggests: \u00EF\u0082\u00B7 For narrowly-graded non-plastic soils (Cu \u00E2\u0089\u00A4 6) in combination with a woven geotextile, a transition occurs from filter compatibility to onset of piping action at a filter ratio 2 \u00E2\u0089\u00A4 AOS/D85 \u00E2\u0089\u00A4 2.5. Analysis of the laboratory and field data also suggests: \u00EF\u0082\u00B7 For narrowly-graded non-plastic soils (Cu \u00E2\u0089\u00A4 6) in combination with a needle- punched nonwoven geotextile, retention compatibility is satisfied at AOS/D85 \u00E2\u0089\u0088 1.5. The findings are consistent with Srikongsri and Fannin (2010b: see chapter 4) reporting onset of piping for woven geotextiles at 2.0 \u00E2\u0089\u00A4 AOS/D85, and retention compatibility for needle-punched nonwoven geotextiles at 2.3 \u00E2\u0089\u00A4 AOS/D85. The concept of a unified plot of AOS/D85 versus S/pi(0) yields a hydromechanical framework that describes the relation between geometric index of capacity and hydromechanical index of demand. 189 This novel framework is used to illustrate the unspecified margin of safety associated with criteria for soil retention, and thereby improve confidence in their application to engineering practice. Accordingly, the following empirical criterion is recommended for cyclic or dynamic flow conditions: \u00EF\u0082\u00B7 O95 or AOS \u00E2\u0089\u00A4 D85 The criterion is verified from inspection of a database comprising 50 values of filter ratio (AOS/D85) and normalized seepage pressure (S/pi(0)), 46 of which are established from laboratory studies and 4 of which are derived from field observations. Given the nature of the soils examined in the tests and in the field, it is appropriate for a uniformly-graded (Cu \u00E2\u0089\u00A4 6) non-plastic base soil that is internally stable. The form of the empirical relation is consistent with the lower bound suggested by Pilarczyk (2000), which was developed with reference to practical experience but not supported by any systematic laboratory study. It provides a significant margin of safety against piping failure for woven geotextile. The relative margin of safety is believed greater for a nonwoven geotextile, and attributed to the more tortuous nature of the pore size distributions and related constrictions. 190 Table 5.1 Properties of soils Author Soil code D85 (mm) D50 (mm) D < 75\u00CE\u00BC (%) Cu Internal stability (Kenney & Lau) Hawley FR 0.33 0.26 < 3 1.8 Stable MT 0.29 0.18 10 3.3 Stable PC 0.21 0.18 15 5.8 Stable Cazzuffi et al. BS 0.20 0.14 < 2 1.5 Stable Chew et al. RS 2.20 0.90 < 2 3.3 Stable 191 Table 5.2 Properties of geotextiles Author Geotextile code Fabric structure AOS or O95 (\u00C2\u00B5m) POA (%) Thickness (mm) Unit mass (g/m 2 ) Hawley G21aN NW-NP 212 N/A 1.0 163 G21bN NW-NP 212 N/A 1.7 220 G21W W-M 212 4-6 N/A 190 G30W W-M 300 4 N/A 170 G43aW W-M 425 <5 N/A 282 G43bW W-M 425 10 N/A 304 G60W W-MT 600 N/A N/A 490 Cazzuffi et al. G16N NW-NP 160 \u00E2\u0080\u00A0 N/A 3.0 300 G44W W 440 \u00E2\u0080\u00A0 N/A 0.8 250 Chew et al. G10N NW 100 N/A 3.5 400 G47W W 470 \u00E2\u0080\u00A0\u00E2\u0080\u00A0 N/A 1.6 400 NW - NP denotes needle-punched nonwoven W denotes woven (M = Monofilament and MT = Multifilament) POA is percent open area N/A denotes no data available \u00E2\u0080\u00A0 = hydrodynamic sieving, and \u00E2\u0080\u00A0\u00E2\u0080\u00A0 = wet sieving 192 Table 5.3 Mass washout and mass piping (after Hawley 2001) Test combination Filter ratio mav (g/m 2 /100 cyc) for iav \u00E2\u0089\u0088 4 (S = 3.9 kPa) AOS/D85 AOS/D50 T = 50 s pi(0) \u00E2\u0089\u0088 10 kPa T = 10 s pi(0) \u00E2\u0089\u0088 10 kPa T = 10 s pi(0) \u00E2\u0089\u0088 1.5 kPa FR-G21aN 0.6 0.8 \u00E2\u0089\u0088 0 \u00E2\u0089\u0088 0 \u00E2\u0089\u0088 0 FR-G21bN 0.6 0.8 \u00E2\u0089\u0088 0 \u00E2\u0089\u0088 0 \u00E2\u0089\u0088 0 FR-G21W 0.6 0.8 \u00E2\u0089\u0088 0 \u00E2\u0089\u0088 0 \u00E2\u0089\u0088 0 FR-G30W 0.9 1.2 \u00E2\u0089\u0088 0 \u00E2\u0089\u0088 0 \u00E2\u0089\u0088 0 FR-G43aW 1.3 1.6 \u00E2\u0089\u0088 0 \u00E2\u0089\u0088 0 \u00E2\u0089\u0088 0 FR-G43bW 1.3 1.6 1 \u00E2\u0089\u0088 0 \u00E2\u0089\u0088 0 FR-G60W 1.8 2.3 4 \u00E2\u0089\u0088 0 \u00E2\u0089\u0088 0 MT-G21aN 0.7 1.2 2 \u00E2\u0089\u0088 0 \u00E2\u0089\u0088 0 MT-G21bN 0.7 1.2 1 \u00E2\u0089\u0088 0 \u00E2\u0089\u0088 0 MT-G21W 0.7 1.2 \u00E2\u0089\u0088 0 \u00E2\u0089\u0088 0 \u00E2\u0089\u0088 0 MT-G30W 1.0 1.7 6 \u00E2\u0089\u0088 0 \u00E2\u0089\u0088 0 MT-G43aW 1.5 2.4 29 44 49 MT-G43bW 1.5 2.4 15 7 20 MT-G60W 2.1 3.4 115 \u00E2\u0089\u0088 0 34 PC-G21aN 1.0 1.2 3 \u00E2\u0089\u0088 0 15 PC-G21bN 1.0 1.2 4 \u00E2\u0089\u0088 0 \u00E2\u0089\u0088 0 PC-G21W 1.0 1.2 5 \u00E2\u0089\u0088 0 11 PC-G30W 1.4 1.7 5 \u00E2\u0089\u0088 0 20 PC-G43aW 2.0 2.4 9 \u00E2\u0089\u0088 0 1673\u00E2\u0080\u00A0 PC-G43bW 2.0 2.4 22 63 1905\u00E2\u0080\u00A0 \u00E2\u0080\u00A0 denotes mass piping 193 Table 5.4 Mass loss (after Cazzuffi et al. 1999 and Chew et al. 2000) Test combination iav and [S in kPa] Filter ratio mav (g/m 2 /100 cyc), for \u00CF\u0083\u00E2\u0080\u00B2vt and [approximate pi(0)value] AOS/D85 AOS/D50 154 [93] (kPa) 113 [67] (kPa) 54 [33] (kPa) 4 [2.4] (kPa) BS-G16N (a) 12 [47 kPa] 0.8 1.1 1 - 2 8 BS-G44W (a) 3 [11.7 kPa] 2.2 3.1 3 - 4 21.9 5 [19.6 kPa] 2.2 3.1 5 - 8 323 (c) 16 [62.7 kPa] 2.2 3.1 7 - 10 426 (c) RS-G-10N (b) 26 [76.4 kPa] 0.05 0.1 - 20 - - RS-G47W (b) 26 [76.4 kPa] 0.12 0.3 - 30 - - a) T varies from 2 s \u00E2\u0080\u0093 20 s and mav values deduced from a duration of 1500 cycles b) T = 2 s and mav values deduced from a duration of 2000 cycles c) Unacceptable mass loss 194 Table 5.5 Field performance evaluation data (after G. Mannsbart & B.R. Christopher, 1997) Table 5.5a Key summary of material properties and performence evaluation Project title and year of construction Type of protection NP-NW Geotextile opening size\u00E2\u0080\u00A0 (mm) Soil indicative particle size (mm) Evaluation AOS O90w D85 D50 Cu Piping Clogging Lacanau, France (1984) Rip-rap Coastal 0.15 0.10 0.8 0.35 2 No No Pantai Murni, Malaysia (1986) Rip-rap Coastal 0.21 0.11 2.5 0.4 15 No No Greifenstein, Austria (1981) Rip-rap River bank 0.21 0.10 7.5 \u00E2\u0080\u00A0 0.2 4 No No Sungai Buntu, Malaysia (1986) Rip-rap Coastal 0.25 0.15 0.16 0.05 5 No No \u00E2\u0080\u00A0 Gap-graded gravelly sand 195 Table 5.5b Approximate hydromechanical loading regime Project title Armor layer thickness (m) slope K0 pi(0) (kPa) W50 (kg) Significant wave height, Hs (m) S (kPa) S/pi(0) Lacanau 3 1:2.0 0.42 27.6 1500 2.24 22.0 0.8 Pantai Murni 1.6 1:3.0 0.42 14.7 430 1.69 16.6 1.13 Greifenstein 1 1:1.5 0.42 9.2 125 0.89 8.7 0.95 Sungai, Buntu 1.6 3.00 0.42 14.7 200 1.31 12.8 0.87 196 (a) (b) Figure 5.1 Geotextile-soil retention: a) AOS/D50; b) AOS/D85 197 Figure 5.2 Grain size distribution curves 198 (a) (b) Figure 5.3 Comparison of Hawley (2000) test data for the FR and MT soil 199 (a) (b) Figure 5.4 Comparison of Hawley (2000) test data for the PC soil 200 (a) (b) Figure 5.5 Original and post-test soil gradation curves: a) test PC-W43a; b) test PC-W43b 201 (a) (b) Figure 5.6 Comparison of Cazzuffi et al. (1999) and Chew et al. (2000) test data for the BS and RS soil 202 Figure 5.7 Cross-section of Sungai Buntu Rip-rap 100 \u00E2\u0080\u0093 300 kg 0.60 Nonwoven geotextile Rip-rap 5 \u00E2\u0080\u0093 20 kg 1.00 (Not to scale) 203 (a) (b) Figure 5.8 Comparison of Mannsbart and Christopher field observations 204 Figure 5.9 Combined database for wave period in the range 2 s \u00E2\u0089\u00A4 T \u00E2\u0089\u00A4 20 s 205 5.7 References ASTM D 4751, Standard Test Method for Determining Apparent Opening Size of a Geotextile. American Society foe Testing and Materials, West Conshohocken, PA, USA. Bathia, S. K., Smith, J. L. and Christopher, B. R. (1996). Characterization and pore-Size distribution: part III: comparison of methods and application to design\u00E2\u0080\u009D, Geosynthetics International, 3, No. 3, 301-328. Cazzuffi, D. A., Mazzucato, A., Moraci, N., & Tondello, M. (1999). A new test apparatus for the study of geotextiles behaviour as filters in unsteady flow conditions: relevance and use. Geotextiles and Geomembranes, 17, No. 5-6, 313 - 329. Chew, S. H., Zhao, Z. K., Karunaratne, G. P., Tan, S. A, Delmas, Ph., & Loke, K. H. (2000). Revetment geotextile filter subjected to cyclic wave loading. Proceedings of Geo-Denver 2000, Denver, CO, USA, pp 162 \u00E2\u0080\u0093 175. Hawley, R. (2001). Filtration performance of geotextiles in cyclic flow conditions. MASc Thesis, University of British Columbia, Vancouver, B.C., Canada, 141p. Kenney, T. C. and Lau, D. (1985). Internal stability of granular filters. Canadian Geotechnical Journal, 22, No. 2, 215-225 206 Kenney, T. C. and Lau, D. (1986). Internal stability of granular filters: Reply. Canadian Geotechnical Journal, 23, No. 3, 420-423 Klein Breteler, M., Pilarczyk, K. W. and Smith, G. M. (1995). Geotextiles in bed and bank protection structures. Publication No. 488, Delft hydraulics laboratory, Delft, The Netherlands. Klein Breteler, M. and Verhey, H. J. (1990). Erosion control by hydrodynamically sandtight geotextiles. Proceedings of 4 th International conference on geotextiles, geomembranes and related products, Hague, The Natherlands, Vol. 1, pp. 385 \u00E2\u0080\u0093 390 Li, M. and Fannin, R.J. (2008). A comparison of two criteria for internal stability of granular soil. Canadian Geotechnical Journal, 45, No. 9, 1303-1309. Mannsbart, G. and Christopher, B.R. (1997). Long-term performance of nonwoven geotextile filters in five coastal and bank protection projects, Geotextiles and Geomembranes, 15, 207 \u00E2\u0080\u0093 221. Pilarczyk, K. W. (2000). Geosynthetics and geosystem in hydraulic and coastal engineering. A.A. Balkema, Rotterdam, The Netherlands, 913 p. Srikongsri, A. and Fannin, R.J. (2010a). Influences of testing methodology on soil-geotextile compatibility in cyclic flow using rigid-wall permeameters. Manuscript draft, see Chapter 3. 207 Srikongsri, A. and Fannin, R.J. (2010b). Geotextile-soil retention in cyclic flow. Manuscript draft, see Chapter 4. U.S. Army Corps of Engineers (1984). Shore Protection Manual. Coastal Engineering Research Center, Washington, DC, USA. Watson, P. D. J., and John, N. W. M. (1999). Geotextile filter design and simulated bridge formation at the soil-geotextile interface. Geotextiles and Geomembranes, 17, 265 - 280 208 6 Conclusions and Recommendations The research study has two main objectives. First, to develop the concept of a hydromechanics-based framework that accounts for (i) capacity of the soil-geotextile filter, by means of a filter ratio or geometric constraint to the onset of retention incompatibility in cyclic flow, and also accounts for (ii) the transient demand on the soil-geotextile filter, by means of seepage-induced change in effective stress. Second, to characterize the unspecified margin of safety that exists in current design guidance for a geotextile filter in applications of cyclic flow, using the empirical rules of Luettich et al. (1992), Holtz et al. (1997) and Pilarczyk (2000), and if appropriate, to recommend changes to address conservatism in those rules. 6.1 Conclusions The filtration compatibility of a geotextile, for soil retention in cyclic flow, has been systematically investigated in a program of laboratory permeameter testing. Onset of retention incompatibility is governed by: (i) grain size distribution of the base soil, (ii) opening size of the geotextile, (iii) a combination of effective stress and hydraulic gradient, and (iv) wave period of flow reversal. A conceptual hydromechanical framework is proposed that unifies these governing factors. The framework relates 209 filter ratio (AOS/Dn) to a value of normalized seepage pressure (S/pi(0)) associated with quantities of soil loss by \u00E2\u0080\u009Cwashout\u00E2\u0080\u009D that do not exceed a threshold value (g/m2/100 cycles) considered representative of piping through the geotextile: the threshold value of soil loss is defined empirically by the rate of loss-induced volume change. A soil- geotextile combination is deemed retention compatible if the rate diminishes with number of cycles, and hence time; in contrast, a rate that was constant or increased during the test was indicative of a piping action. Using the proposed concept, conservatism of select current design guidance is examined with regard to the filter ratios AOS/D50 and AOS/D85. Verification of the concept through comparison with other laboratory studies, and also with field observations, then leads to a recommendation for current design guidance. The following summary of findings from the study addresses (i) insights to important factors governing soil-geotextile compatibility and the need for a systematic study of cyclic flow conditions, (ii) the influence of test method, scale effect in a small and large permeameter, and stress distribution in a rigid-wall permeameter, (iii) a novel hydromechanical-based approach to interpretation of retention incompatibility and, (iv) evidence of undue conservatism in current design guidance, and a proposed revision to an empirical criterion for soil retention in cyclic flow. 210 6.1.1 Previous study Analysis of data on soil-geotextile compatibility using the small permeameter, from a study (Hawley, 2001) that preceded the current research, examines filtration combinations of three cohesionless soils and seven geotextiles (5 woven geotextiles, and 2 needle-punched nonwoven geotextiles). Uncertainty surrounding inherent margins of safety in design guidance is addressed through re-interpretation of the data. Specifically, based on characterization of the soil (D85), geotextile (AOS), soil- geotextile compatibility (GR8) and mass of soil passing per unit area (mp), the findings suggest: \u00EF\u0082\u00B7 mass loss per unit area provides a very useful index of filtration compatibility for soil-geotextile combinations that exhibit piping, and should be reported to assist with test interpretation; \u00EF\u0082\u00B7 mass loss per unit area increases with larger AOS/D85; \u00EF\u0082\u00B7 the empirical design criterion of AOS/D85 \u00E2\u0089\u00A4 0.5 for soil retention in cyclic flow is unduly conservative; \u00EF\u0082\u00B7 wave period and confining stress influence the filtration compatibility of a soil- geotextile combination, and those two parameters, in combination with hydraulic gradient, require systematic study in order to properly understand the margin of safety in empirical design criteria. 211 6.1.2 Influence of test method Newly-acquired data from the small (100 mm diameter) permeameter, and a large permeameter (280 mm diameter) test device, for a uniformly-graded sand and two woven geotextiles were used to evaluate the influence of test method, with emphasis on matters of scale effect in the test equipment, test procedure and the test device itself on hydromechanical conditions at the soil-geotextile interface. In the absence of a standard test device and procedure, the objective was to develop and validate a suitable laboratory technique for systematic study of soil-geotextile compatibility in cyclic flow. The following conclusions are drawn: \u00EF\u0082\u00B7 measurement of axial load in the large permeameter indicates a reduction of 20% to 40% in effective stress along the specimen length that is attributed to interface friction, a finding that implies any stress-based interpretation of soil- geotextile compatibility in a rigid-wall permeameter must address the phenomenon of sidewall friction; \u00EF\u0082\u00B7 inspection of mass loss-volume change in the large and small permeameter indicates no scale effect in the two permeameters. The difference of results (mass loss) is attributed to spatial variation of the pore opening size of the geotextile, hence it is recommended to repeat a test in the small permeameter 212 where it is believed the soil and geotextile exhibit filtration incompatibility, and report an average of the experimental findings for purposes of analysis. Findings in the small permeameter appropriate for a systematic study of test variables, the experimental data and a companion theoretical analysis show that: \u00EF\u0082\u00B7 a multi-stage test method involving reduction of axial stress, and the corresponding variation of lateral stress in the rigid wall permeameter, suggest mean effective stress at the soil-geotextile interface (pi) is a better parameter for interpretation of test performance than vertical stress; and \u00EF\u0082\u00B7 soil retention is very sensitive to the upward component of cyclic flow that yields a reduction in mean effective stress and, it is postulated, thereby acts to destabilize arching in soil particles at the openings of the woven geotextile. Finally, for the range of variables examined in testing, mass loss is found negligible in cyclic flow at a filter ratio AOS/D85 \u00E2\u0089\u0088 2, but very significant at AOS/D85 \u00E2\u0089\u0088 2.8, where soil-geotextile retention incompatibility is triggered by loading conditions governed by a combination of wave period, hydraulic gradient and confining stress. The findings suggest that mass loss may be used to distinguish between a soil-combination that is compatible, versus incompatible in cyclic flow. The results are sufficiently encouraging that more data, also including tests on nonwoven geotextiles, are required 213 to characterize a greater range of AOS/D85 to enable development of an empirical rule for soil-geotextile retention compatibility in cyclic flow based on principles of mechanics. 6.1.3 A hydromechanical framework (for onset of retention incompatibility) The main experimental program of the current study was performed on combinations of four uniformly-graded base soils (two fine sands, a silty sand and a coarse sandy silt), five needle-punched nonwoven geotextiles and two woven geotextiles. The utility of an empirical filter ratio, expressed as AOS/D50 and AOS/D85, is evaluated. The potential for a relation between filter ratio and normalized seepage pressure (S/pi(0)) to explain the onset of retention incompatibility in cyclic flow is explored, leading to a conceptual hydromechanical framework for interpretation of the test data. With reference to a characteristic opening size AOS and an indicative particle size Dn, the following conclusions are drawn: \u00EF\u0082\u00B7 For unidirectional flow at iav < 10 , no unacceptable mass loss was observed for nonwoven geotextiles at 0.7 \u00E2\u0089\u00A4 AOS/D85 \u00E2\u0089\u00A4 2.3 (0.9 \u00E2\u0089\u00A4 AOS/D50 \u00E2\u0089\u00A4 2.7), and woven geotextiles at 1.2 \u00E2\u0089\u00A4 AOS/D85 \u00E2\u0089\u00A4 3.7 (1.3 \u00E2\u0089\u00A4 AOS/D50 \u00E2\u0089\u00A4 4.6). The finding implies retention compatibility; \u00EF\u0082\u00B7 For cyclic flow at iav < 10, with nonwoven geotextiles at 0.7 \u00E2\u0089\u00A4 AOS/D85 \u00E2\u0089\u00A4 2.3 (0.9 \u00E2\u0089\u00A4 AOS/D50 \u00E2\u0089\u00A4 2.7), and with woven geotextiles examined at 1.2 \u00E2\u0089\u00A4 AOS/D85 214 \u00E2\u0089\u00A4 2.0 (1.3 \u00E2\u0089\u00A4 AOS/D50 \u00E2\u0089\u00A4 2.5), there was again no unacceptable mass loss observed, which also implies retention compatibility. However for woven geotextiles at AOS/D85 > 2.0, mass loss (mav, g/m 2 /100 cycles) was observed under cyclic loading conditions. From interpretation of specimen volume change, the following conclusions are made: \u00EF\u0082\u00B7 a value 30 \u00E2\u0089\u00A4 mav \u00E2\u0089\u00A4 50 g/m 2 /100 cycles defines a threshold for soil arching at openings of the geotextile that describes the boundary between washout and onset of retention incompatibility; \u00EF\u0082\u00B7 At 2.0 < AOS/D85 \u00E2\u0089\u00A4 2.5 (2.5 \u00E2\u0089\u00A4 AOS/D50 \u00E2\u0089\u00A4 3.0), a transition occurs from retention compatible to retention sensitivity; \u00EF\u0082\u00B7 At 2.5 < AOS/D85 \u00E2\u0089\u00A4 3.7 (3.0 \u00E2\u0089\u00A4 AOS/D50 \u00E2\u0089\u00A4 4.6), soil retention is found very sensitive to loading conditions that are governed by hydraulic gradient, stress and wave period. A hydromechanics-based interpretation is proposed to account for the combined influence of hydraulic gradient and effective stress. A normalized value of seepage pressure at the soil-geotextile interface (S/pi(0)) appears to explain the nature of mass loss through pore size openings of the geotextile. For woven geotextiles, the finding implies that T = 6 s governs the filtration compatibility, for the range 6 s \u00E2\u0089\u00A4 T \u00E2\u0089\u00A4 120 s. 215 The fundamental mechanism of soil retention, from interpretation based on the range of mav and analysis of the test data at T = 6 s, suggests: \u00EF\u0082\u00B7 A threshold boundary between no loss and washout is governed by AOS/Dn; \u00EF\u0082\u00B7 A threshold boundary between washout and piping is governed by AOS/Dn and S/pi(0); \u00EF\u0082\u00B7 Transition from no loss to piping is very sensitive to a change of AOS/Dn when seepage pressure exceeds the initial mean effective stress (S/pi(0) > 1). Overall, the findings of the laboratory experimental program on uniformly-graded soils indicate filtration compatibility in cyclic flow reversal for a nonwoven geotextile at AOS/D50 \u00E2\u0089\u00A4 2.7 or AOS/D85 \u00E2\u0089\u00A4 2.3, and the onset of retention incompatibility for a woven geotextile at 2.5 \u00CA\u00B9 AOS/D50 or 2 \u00CA\u00B9 AOS/D85. Accordingly, the design criterion AOS/D50 \u00E2\u0089\u00A4 1 (Luettich et al., 1992), or alternatively AOS/D85 \u00E2\u0089\u00A4 0.5 (Holtz et al., 1997; CFEM, 2006), appear to provide a generous margin of safety for applications involving filtration of a uniformly-graded base soil. Indeed, it is considered unduly conservative, because it has potential to eliminate the use of a geotextile with base soil of fine sand - yet results of the current laboratory test program, supported by insights from the proposed hydromechanical framework, demonstrate the combinations to yield a retention compatible filter. 216 6.1.4 A recommended criterion for soil retention In order to more thoroughly examine this concern for undue conservatism in design guidance for cyclic flow, arising from interpretation of the newly acquired laboratory permeameter data, the study evolved to re-analyze test data from three other laboratory studies published in the literature, and select field data reported from four project sites. The re-analysis is undertaken with reference to attributes of the base soil (D50 and D85), the pore size opening of the geotextile (AOS), an estimate of likely seepage pressure (S) from cyclic flow, and an estimate of initial mean effective stress (pi(0)) in the base soil. The laboratory and field applications involve both woven and needle-punched nonwoven geotextiles, in filtration applications that include both compatible and incompatible soil-geotextile combinations. Accordingly, the database provides opportunity to critically evaluate the merits of a proposed hydromechanical framework, for retention compatibility in cyclic flow, based on a relation between filter ratio (AOS/D85, and AOS/D50) and normalized seepage pressure (S/pi(0)). Generally good agreement is found between the proposed framework and the database of independent laboratory and field experience. The framework is found to distinguish between filter compatibility, and incompatibility in the form of a piping action through the geotextile. It appears that AOS/D85 is more suitable than AOS/D50 as an indicative filter ratio for design. The finding is consistent with the observation of Watson and 217 John (1999) who established, from statistical analysis, that the particles in the size range D70 to D90 govern the formation of a soil bridge at the openings of a filter. More specifically, analysis of the laboratory test data suggests: \u00EF\u0082\u00B7 For narrowly-graded non-plastic soils (Cu \u00E2\u0089\u00A4 6) in combination with a woven geotextile, a transition occurs from filter compatibility to onset of piping action at a filter ratio 2 \u00E2\u0089\u00A4 AOS/D85 \u00E2\u0089\u00A4 2.5. Analysis of the laboratory and field data also suggests: \u00EF\u0082\u00B7 For narrowly-graded non-plastic soils (Cu \u00E2\u0089\u00A4 6) in combination with a needle- punched nonwoven geotextile, retention compatibility is satisfied at AOS/D85 \u00E2\u0089\u0088 1.5. Recall the experimental findings, reported in section 6.1.3, that onset of piping for woven geotextiles is found at 2.0 \u00E2\u0089\u00A4 AOS/D85, and retention compatibility for needle- punched nonwoven geotextiles is found at 2.3 \u00E2\u0089\u00A4 AOS/D85. The proposed novel hydromechanical framework of AOS/D85 versus S/Pi(0) proves useful to illustrate the unspecified margin of safety associated with criteria for soil retention, and thereby improve confidence in their application to engineering practice. Accordingly, and in 218 order to address undue conservatism, it is proposed the current empirical criterion of AOS \u00E2\u0089\u00A4 0.5 D85 for soil retention cyclic or dynamic flow conditions be revised to: \u00EF\u0082\u00B7 O95 or AOS \u00E2\u0089\u00A4 D85 The recommended criterion is appropriate for a uniformly-graded (Cu \u00E2\u0089\u00A4 6) non-plastic base soil that is internally stable, and in agreement with the lower bound (but not the upper bound) to design guidance in the Netherlands. It provides a significant margin of safety against piping failure, given the difference between AOS/D85 = 1 and the filter ratio AOS/D85 \u00E2\u0089\u0088 2 at which retention incompatibility is found to initiate for a woven geotextile. The relative margin of safety is believed greater for a nonwoven geotextile, and attributed to the more tortuous nature of the pore size distributions and related inter-fiber constrictions. 6.2 Recommendations for further study It has been postulated, and then proven, that a relation between hydromechanical demand and geometric constraint determines the onset of retention incompatibility in a geotextile in combination with a uniformly-graded non-plastic base soil in cyclic flow. Many existing retention criteria for unidirectional flow involve a value of Cu (= D60/D10) to address the influence of shape of the grain size distribution curve (see, for 219 example, Geotechnical Engineering Office, 1993; Pilarczyk, 2000; Fannin, 2007). In general, the recommended filter ratio is reduced for a broader soil gradation. It is also reduced if the base soil is internally unstable and therefore susceptible to seepage- induced migration of the finer fraction (Lafleur 1999; Mylnarek 2000). This gives recognition to the fact that, in broadly-graded and internally unstable soil, a stable bridging network of particles may not easily develop at the pore size openings of the geotextile (Bhatia and Huang 1995). Furthermore, fines content also appears to influence retention compatibility (Hawley, 2001). Clogging potential or long-term permeability in dynamic or cyclic flow is not yet supported by an extensive body of experimental data and supporting theoretical concepts, and therefore not yet properly addressed in design criteria. Accordingly, the following recommendations are made for further study: \u00EF\u0082\u00B7 a systematic study of broadly-graded soils and internally unstable soils; and \u00EF\u0082\u00B7 a systematic study of fines content; both of which have potential to challenge empirical criteria not only for soil retention, but also physical clogging. In addition, influence of pore size distribution should be investigated in order to provide a better recommendation for design guidance that may distinguish between a woven and a nonwoven style of geotextile. 220 The study has been undertaken in the absence of a standard test method for cyclic flow, and very limited well-documented laboratory and field data. A novel concept is proposed that links geometric constraint (O95/Dn, in the form of AOS/D85) to normalized seepage pressure (S/pi(0)). A comparison of experience between different research studies would benefit greatly from a standardized method of testing, and it is recommended that efforts be placed in developing such a method. This will require a standard form of permeameter, for which the choice rests between: (i) rigid-wall without sidewall lubrication (used in the current study); (ii) rigid-wall with sidewall lubrication (for example, Chen et al. 2008) and (iii) a flexible-wall permeameter (for example, Harney and Holtz, 2001). The first option requires a correction for the influence of sidewall friction in order to obtain a value of S/pi(0), and the third option eliminates sidewall friction. It is recommended that a comparative study be made to establish the merits of each approach. It is also recommended that greater emphasis be placed in characterizing effective stress at the soil-geotextile interface, in fundamental studies involving cyclic flow. However, a proper interpretation of retention incompatibility requires a confident distinction between acceptable washout and unacceptable piping actions. This distinction is likely more important to address than greater precision of soil-geotextile interface stress. Finally, any confidence derived from analysis and interpretation of laboratory studies will be greatly enhanced by corroboration through comparison with field performance 221 data. Construction applications in dynamic flow conditions may involve flow in many directions: issues of cyclic flow parallel to the plane of filter, as well as normal to it, may have implications for the proposed hydromechanical interpretation of soil retention. The careful reporting of such field performance data, and long-term monitoring of field test sites, will only serve to enhance confidence in the applicability of design guidance to engineering practice. 222 6.3 References Bhatia, S.K. and Huang, Q. (1995). Geotextile Filters for Internally Stable/Unstable Soils. Geosynthetics International, 2, No. 3, 537-565. Canadian Geotechnical Society. Canadian Foundation Engineering Manual, 4th edition, BiTech Publishers Ltd. Chen, R.-H., Ho, C.-C. & Hsu, C.-Y. (2008). The effect of fine soil content on the filtration characteristics of geotextile under cyclic flows. Geosynthetics International, 15, No. 2, 95\u00E2\u0080\u0093 106. Fannin, R.J. (2007). Chapter 6: The use of geosynthetics as filters. Geosynthetics in Civil Engineering. Woodhead Publishing, Cambridge, UK, 295p. Geotechnical Engineering Office Civil Engineering Department, Hong Kong (1993), Review of Granular and Geotextile Filters, Geo Publication No. 1/93, The Government Printer, Hong Kong Harney, D. H. and Holtz, R. D. (2001). A flexible gradient ratio test. Proceedings of geosynthetics conference 2001, Portland, Oregon, USA, pp. 409 \u00E2\u0080\u0093 422. Hawley, R. (2001). Filtration performance of geotextiles in cyclic flow conditions. MASc Thesis, University of British Columbia, Vancouver, B.C., Canada, 141p. Holtz, R.D, Christopher, B.R., and Berg, R.R. (1997). Geosynthetic Engineering. BiTech Publishers, Richmond, BC, Canada, 452 p. Lafleur, J. (1999). Selection of geotextiles to filter broadly graded cohesionless soils. Geotextiles and Geomembranes, 17, 299 - 312 223 Luettich, S.M., Giroud, J.P., and Bachus, R.C. (1992). Geotextile filter design guide. Geotextiles and Geomembranes, 11, 355 - 370. Mlynarek J. (2000). Geo drains and geo filter-retrospective and future trends. Filters and Drainage in Geotechnical and Geoenvironmental Engineering, Balkema, Rotterdam, The Netherlands, pp. 27 - 47. Pilarczyk, K. W. (2000). Geosynthetics and geosystem in hydraulic and coastal engineering. A.A. Balkema, Rotterdam, The Netherlands, 913 p. Watson, P. D. J., and John, N. W. M. (1999). Geotextile filter design and simulated bridge formation at the soil-geotextile interface. Geotextiles and Geomembranes, 17, 265 \u00E2\u0080\u0093 280 224 Appendix A Mobilization of sidewall friction Consider a rigid-wall permeameter in which a test specimen rests on a rigid, zero displacement (ds = 0) at lower boundary. It is postulated that application of an axial load on the top surface of the test specimen develops a maximum value of soil-wall relative displacement at the top of the specimen (dst), which decreases to zero at the base of the specimen. Accordingly, there is a constraint to full mobilization of sidewall friction along the entire specimen length (see Fig. A1a). Hence, the coefficient of side-wall friction for a cohesionless soil may be expressed as: \u00EF\u0081\u00A4tan\u00EF\u0080\u00BCf (A1) where f is a coefficient of sidewall friction and \u00CE\u00B4 is the soil-wall interface friction angle. For purposes of analysis, a simplified linear elastic-perfectly plastic is adopted (Fig. A1b). The value of f is influenced by stress magnitude and compressibility of the soil. Hence, the magnitude of f may be categorized according to two simple cases. First, the relative displacement at the top of the specimen is less than or equal to the value of relative displacement at mobilization of full interface friction; (dst \u00E2\u0089\u00A4 dsp, see Fig. A1c and Fig. A1e), for which: 225 \u00EF\u0081\u00A4\u00EF\u0081\u00A4 tan5.0tan \u00EF\u0082\u00A3\u00EF\u0083\u00B7\u00EF\u0083\u00B7 \u00EF\u0083\u00B8 \u00EF\u0083\u00B6 \u00EF\u0083\u00A7\u00EF\u0083\u00A7 \u00EF\u0083\u00A8 \u00EF\u0083\u00A6 \u00EF\u0080\u00BD OACEarea OABarea f (A2) Second, the relative displacement at the top of the specimen exceeds the value of relative displacement at mobilization of full friction; dst > dsp, see Fig. A1d and Fig. A1e). \u00EF\u0081\u00A4\u00EF\u0081\u00A4\u00EF\u0081\u00A4 tantantan5.0 \u00EF\u0080\u00BC\u00EF\u0083\u00B7\u00EF\u0083\u00B7 \u00EF\u0083\u00B8 \u00EF\u0083\u00B6 \u00EF\u0083\u00A7\u00EF\u0083\u00A7 \u00EF\u0083\u00A8 \u00EF\u0083\u00A6 \u00EF\u0080\u00BD\u00EF\u0080\u00BC OACEarea OACDarea f (A3) 226 (a) (b) (c) (d) (e) Figure A1 Mobilization of sidewall friction in the rigid-wall permeameter 227 Appendix B Example stress calculation for section 3.6.4.2 For stage CYC1 during downward flow: \u00CF\u0083\u00C2\u00B4vt = 66 kPa, \u00CE\u0094\u00CF\u0083\u00C2\u00B4v(0) = 25.9 kPa, S = 8.8 kPa, f = 0.35 (from Fig. 3.14a), \u00CE\u00B3\u00CD\u00B4 = 8 kN/m3, Z = D = 0.1 m, OCR = 1 and \u00CF\u0086 = 38\u00CB\u009A K0 = (1-sin38\u00CB\u009A) = 0.384 \u00EF\u0083\u00B7 \u00EF\u0083\u00B8 \u00EF\u0083\u00B6 \u00EF\u0083\u00A7 \u00EF\u0083\u00A8 \u00EF\u0083\u00A6 \u00EF\u0080\u00AB \u00EF\u0082\u00B1\u00EF\u0082\u00A2\u00EF\u0080\u00AB\u00EF\u0082\u00A2 \u00EF\u0080\u00BD\u00EF\u0082\u00A2 D ZfK SZvt vm 021 )(5.0 \u00EF\u0081\u00A7\u00EF\u0081\u00B3 \u00EF\u0081\u00B3 = \u00EF\u0083\u00B7\u00EF\u0083\u00B7 \u00EF\u0083\u00B8 \u00EF\u0083\u00B6 \u00EF\u0083\u00A7\u00EF\u0083\u00A7 \u00EF\u0083\u00A8 \u00EF\u0083\u00A6 \u00EF\u0080\u00AB \u00EF\u0080\u00AB\u00EF\u0080\u00AB \u00EF\u0080\u00BD\u00EF\u0082\u00A2 )1.0( )1.0)(35.0)(384.0(2 1 )8.8)1.0(8(5.066 max,vm\u00EF\u0081\u00B3 = 55.8 kPa For stage CYC3 during downward flow: \u00CF\u0083\u00C2\u00B4vt = 33 kPa, \u00CF\u0083\u00C2\u00B4vm(0) = 26 kPa, \u00CE\u0094\u00CF\u0083\u00C2\u00B4v(0) = 15.0 kPa, S = 8.8 kPa, f = 0.29 (from Fig. 3.14a), \u00CE\u00B3\u00CD\u00B4 = 8 kN/m3, Z = D = 0.1 m and \u00CF\u0086 = 40\u00CB\u009A Trial )5.0( )0( max, S OCR vm vm \u00EF\u0082\u00B1 \u00EF\u0082\u00BB \u00EF\u0081\u00B3 \u00EF\u0081\u00B3 = ))8.8(5.026( 8.55 \u00EF\u0082\u00B1 = 1.83 40sin 0 83.1)40sin1( \u00EF\u0080\u00AD\u00EF\u0080\u00BDK = 0.53 \u00EF\u0083\u00B7 \u00EF\u0083\u00B8 \u00EF\u0083\u00B6 \u00EF\u0083\u00A7 \u00EF\u0083\u00A8 \u00EF\u0083\u00A6 \u00EF\u0080\u00AB \u00EF\u0082\u00B1\u00EF\u0082\u00A2\u00EF\u0080\u00AB\u00EF\u0082\u00A2 \u00EF\u0080\u00BD\u00EF\u0082\u00A2 D ZfK SZvt vm 021 )(5.0 \u00EF\u0081\u00A7\u00EF\u0081\u00B3 \u00EF\u0081\u00B3 = \u00EF\u0083\u00B7\u00EF\u0083\u00B7 \u00EF\u0083\u00B8 \u00EF\u0083\u00B6 \u00EF\u0083\u00A7\u00EF\u0083\u00A7 \u00EF\u0083\u00A8 \u00EF\u0083\u00A6 \u00EF\u0080\u00AB \u00EF\u0080\u00AB\u00EF\u0080\u00AB )1.0( )1.0)(29.0)(53.0(2 1 )8.8)1.0(8(5.033 = 29.0 kPa Check vm vm OCR \u00EF\u0081\u00B3 \u00EF\u0081\u00B3 \u00EF\u0082\u00A2 \u00EF\u0082\u00A2 \u00EF\u0080\u00BD max, = 29 8.55 = 1.93 \u00E2\u0089\u0088 the trial value \u00E2\u0080\u0093 OK fK D Z vmv 0 4 \u00EF\u0081\u00B3\u00EF\u0081\u00B3 \u00EF\u0082\u00A2\u00EF\u0080\u00BD\u00EF\u0082\u00A2\u00EF\u0081\u0084 = 4(29)(0.53)(0.29) = 17.7 kPa SZ vvtvb \u00EF\u0082\u00B1\u00EF\u0082\u00A2\u00EF\u0081\u0084\u00EF\u0080\u00AD\u00EF\u0082\u00A2\u00EF\u0080\u00AB\u00EF\u0082\u00A2\u00EF\u0080\u00BD\u00EF\u0082\u00A2 \u00EF\u0081\u00B3\u00EF\u0081\u00A7\u00EF\u0081\u00B3\u00EF\u0081\u00B3 = 33 + (8)(0.1) \u00E2\u0080\u0093 17.7 + 8.8 = 24.9 kPa 228 Appendix C Summary of key results Small permeameter: test A-NW4-T6 iav UNI 1a UNI 1b UNI 2a UNI 2b UNI 3a UNI 3b kav (cm/s) GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 1 1.0 0.9 1.0 0.9 1.0 0.9 1.0 0.9 0.9 0.9 1.0 0.9 0.026 5 1.0 1.0 1.0 0.9 1.0 0.9 1.0 0.9 1.0 0.9 1.0 0.9 0.022 9 1.0 0.9 0.9 0.9 0.9 0.9 0.9 0.9 1.0 0.9 0.9 0.9 0.018 Remarks: (a) NO MASS LOSS was observed in all UNI and CYC stages. (b) ZERO volume change. Small permeameter: test A-W1-T6 iav UNI 1a UNI 1b UNI 2a UNI 2b UNI 3a UNI 3b kav (cm/s) GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 1 1.1 1.2 1.1 1.2 1.0 1.1 1.1 1.1 1.1 1.2 1.1 1.1 0.019 5 1.1 1.3 1.0 1.2 1.0 1.2 1.0 1.1 1.0 1.1 1.1 1.2 0.021 9 1.1 1.3 1.1 1.2 1.2 1.3 1.2 1.3 1.2 1.3 1.2 1.2 0.016 Remarks: (a) NO MASS LOSS was observed in all UNI and CYC stages. (b) ZERO volume change. Small permeameter: test A-W1-T60 iav UNI 1a UNI 1b UNI 2a UNI 2b UNI 3a UNI 3b kav (cm/s) GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 1 1.2 1.4 1.2 1.4 1.2 1.3 1.2 1.3 1.2 1.4 1.2 1.4 0.023 5 1.1 1.3 1.0 1.2 1.0 1.2 1.0 1.1 1.0 1.1 1.1 1.2 0.016 9 1.1 1.3 1.1 1.2 1.2 1.3 1.2 1.3 1.2 1.3 1.2 1.2 0.018 Remarks: (a) NO MASS LOSS was observed in all UNI and CYC stages. (b) ZERO volume change. 229 Small permeameter: test B-W1-T6 iav UNI 1a UNI 1b UNI 2a UNI 2b UNI 3a UNI 3b kav (cm/s) GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 1 1.0 1.2 1.0 1.2 1.0 1.2 1.0 1.2 1.0 1.2 1.0 1.2 0.015 5 1.1 1.3 1.1 1.3 1.0 1.3 1.0 1.2 1.0 1.3 1.1 1.2 0.012 9 1.1 1.3 1.1 1.1 1.1 1.1 1.1 1.1 1.0 1.1 1.0 1.1 0.015 Remarks: (a) NO MASS LOSS was observed in all UNI and CYC stages. (b) ZERO volume change. Small permeameter: test B-W1-T60 iav UNI 1a UNI 1b UNI 2a UNI 2b UNI 3a UNI 3b kav (cm/s) GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 1 1.2 1.4 1.2 1.4 1.2 1.3 1.2 1.3 1.2 1.4 1.2 1.4 0.018 5 1.1 1.3 1.0 1.2 1.0 1.3 1.0 1.2 1.0 1.2 1.1 1.3 0.016 9 1.1 1.4 1.1 1.2 1.2 1.3 1.1 1.3 1.1 1.3 1.1 1.3 0.016 Remarks: (a) NO MASS LOSS was observed in all UNI and CYC stages. (b) ZERO volume change. Small permeameter: test B-W2-T6 iav UNI 1a UNI 1b UNI 2a UNI 2b UNI 3a UNI 3b kav (cm/s) GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 1 1.1 1.2 1.1 1.2 1.0 1.1 1.1 1.1 1.1 1.2 1.0 1.1 0.019 5 1.1 1.2 1.0 1.2 1.0 1.2 1.0 1.1 1.0 1.1 1.1 1.1 0.011 9 1.1 1.2 1.1 1.2 1.1 1.2 1.1 1.2 1.2 1.2 1.0 1.1 0.012 Remarks: (a) NO MASS LOSS was observed in all UNI and CYC stages. (b) ZERO volume change. 230 Small permeameter: test B-W2-T60 iav UNI 1a UNI 1b UNI 2a UNI 2b UNI 3a UNI 3b kav (cm/s) GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 1 1.0 1.0 1.0 1.0 1.0 1.1 1.0 1.1 1.0 1.1 1.0 1.1 0.018 5 1.0 1.0 1.0 1.1 1.0 1.1 1.0 1.1 1.0 1.1 1.0 1.0 0.015 9 1.0 1.1 0.9 1.1 0.9 1.1 0.9 1.0 1.0 1.0 1.0 1.1 0.015 Remarks: (a) NO MASS LOSS was observed in all UNI and CYC stages. (b) ZERO volume change. Small permeameter: test C-NW3-T6 iav UNI 1a UNI 1b UNI 2a UNI 2b UNI 3a UNI 3b kav (cm/s) GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 1 1.3 1.6 1.3 1.6 1.2 1.5 1.2 1.7 1.3 1.8 1.4 1.8 0.008 5 1.4 1.8 1.4 1.8 1.4 1.8 1.4 1.7 1.4 1.6 1.4 1.6 0.007 9 1.4 1.7 1.4 1.7 1.4 1.7 1.4 1.6 1.4 1.6 1.4 1.6 0.007 Remarks: (a) NO MASS LOSS was observed in all UNI and CYC stages. (b) ZERO volume change. Small permeameter: test C-NW4-T6 iav UNI 1a UNI 1b UNI 2a UNI 2b UNI 3a UNI 3b kav (cm/s) GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 1 1.0 0.9 1.0 0.9 1.0 0.9 1.0 0.9 1.0 0.9 0.9 0.9 0.009 5 1.0 0.9 1.0 0.9 1.0 0.9 0.9 0.9 0.9 0.9 0.9 0.9 0.008 9 1.0 1.0 1.0 0.9 0.9 0.9 0.9 0.9 0.9 0.9 0.9 0.9 0.008 Remarks: (a) NO MASS LOSS was observed in all UNI and CYC stages. (b) ZERO volume change. 231 Small permeameter: test C-NW5-T6 iav UNI 1a UNI 1b UNI 2a UNI 2b UNI 3a UNI 3b kav (cm/s) GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 1 1.0 0.9 1.0 0.9 1.0 0.9 1.0 1.0 0.9 1.0 1.0 1.0 0.010 5 1.0 1.0 1.0 0.9 1.0 0.9 1.0 0.9 1.0 0.9 1.0 0.9 0.008 9 1.0 0.9 1.0 1.0 1.0 0.9 1.0 0.9 1.0 0.9 1.0 0.9 0.008 Remarks: (a) NO MASS LOSS was observed in all UNI and CYC stages. (b) ZERO volume change. Small permeameter: test C-W1-T6 iav UNI 1a UNI 1b UNI 2a UNI 2b UNI 3a UNI 3b kav (cm/s) GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 1 1.2 1.3 1.1 1.3 1.2 1.3 1.1 1.3 1.2 1.4 1.2 1.3 0.007 5 1.2 1.3 1.1 1.2 1.1 1.2 1.0 1.1 1.0 1.1 1.1 1.2 0.006 9 1.1 1.2 1.1 1.2 1.2 1.3 1.2 1.3 1.1 1.2 1.1 1.2 0.007 Remarks: (a) NO MASS LOSS was observed in all UNI and CYC stages. (b) ZERO volume change. 232 Small permeameter: test C-W2-T6 iav UNI 1a UNI 1b UNI 2a UNI 2b UNI 3a UNI 3b kav (cm/s) GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 1.3 1.1 1.2 1.1 1.2 1.0 1.1 1.1 1.1 1.1 1.2 1.1 1.1 0.006 5.7 1.1 1.1 1.0 1.2 1.0 1.2 1.0 1.1 1.0 1.1 1.0 1.1 0.009 9.5 1.1 1.1 1.1 1.2 1.0 1.2 1.0 1.0 1.0 1.0 1.0 1.1 0.009 Remarks: MASS LOSS (chapter 3 & 4) was observed in all CYC stages, and associated with volume change. Small permeameter: test C-W2-T6: Change in specimen height (mm) iav UNI1a CYC1 UNI1b UNI2a CYC2 UNI2b UNI3a CYC3 UNI3b 1.3 0 0.22 0 0 0.09 0 0 0.05 0 5.7 0 0.41 0 0 0.28 0 0 0.45 0 9.5 0 0.41 0 0 0.43 0 0 0.81 0 Small permeameter: test C-W2-T6-R (repeated) iav UNI 1a UNI 1b UNI 2a UNI 2b UNI 3a UNI 3b kav (cm/s) GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 1.1 1.0 1.1 1.0 1.1 1.0 1.1 1.0 1.1 1.0 1.1 1.0 1.1 0.007 5.1 1.0 1.1 1.0 1.1 1.0 1.1 1.0 1.1 1.0 1.1 1.0 1.0 0.009 9.1 1.0 1.0 1.0 1.0 1.0 0.9 0.9 0.9 0.9 0.9 0.9 0.9 0.010 Remarks: MASS LOSS (chapter 3 & 4) was observed in all CYC stages, and associated with volume change. Small permeameter: test C-W2-T6-R (repeated): change in specimen height (mm) iav UNI1a CYC1 UNI1b UNI2a CYC2 UNI2b UNI3a CYC3 UNI3b 1.1 0 0.28 0 0 0.26 0 0 0.32 0 5.1 0 0.48 0 0 0.44 0 0 0.68 0 9.1 0 0.34 0 0 0.52 0 0 0.94 0 233 Small permeameter: test C-W2-T60 iav UNI 1a UNI 1b UNI 2a UNI 2b UNI 3a UNI 3b kav (cm/s) GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 1.1 1.0 1.0 1.0 1.0 1.0 1.0 1.0 0.9 1.0 0.9 1.0 0.9 0.006 5.1 1.0 1.0 1.0 0.9 1.0 0.9 1.0 0.9 1.0 0.9 1.0 0.9 0.008 8.9 0.9 0.9 0.9 0.9 0.9 0.9 1.0 0.9 1.0 0.9 0.9 0.9 0.008 Remarks: MASS LOSS (chapter 3 & 4) was observed in all CYC stages, and associated with volume change. Small permeameter: test C-W2-T60: change in specimen height (mm) iav UNI1a CYC1 UNI1b UNI2a CYC2 UNI2b UNI3a CYC3 UNI3b 1.1 0 0.07 0 0 0.01 0 0 0.02 0 5.1 0 0.08 0 0 0.02 0 0 0.16 0 8.9 0 0.02 0 0 0.02 0 0 0.09 0 Small permeameter: test C-W2-T60-R iav UNI 1a UNI 1b UNI 2a UNI 2b UNI 3a UNI 3b kav (cm/s) GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 1.0 1.0 1.0 0.9 1.0 1.0 1.0 0.9 1.0 1.0 1.0 1.0 1.0 0.007 5.2 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 0.006 8.7 0.9 1.0 0.9 1.0 0.9 1.0 0.9 0.9 0.9 0.9 0.9 0.9 0.007 Remarks: MASS LOSS (chapter 3 & 4) was observed in all CYC stages, and associated with volume change. Small permeameter: test C-W2-T60-R: change in specimen height (mm) iav UNI1a CYC1 UNI1b UNI2a CYC2 UNI2b UNI3a CYC3 UNI3b 1.0 0 0 0 0 0.03 0 0 0.05 0 5.2 0 0.01 0 0 0.08 0 0 0.03 0 8.7 0 0.02 0 0 0.03 0 0 0.06 0 234 Small permeameter: test C-W2-T120 iav UNI 1a UNI 1b UNI 2a UNI 2b UNI 3a UNI 3b kav (cm/s) GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 1.2 1.2 1.4 1.3 1.3 1.3 1.3 1.3 1.4 1.3 1.4 1.3 1.3 0.008 5.5 1.3 1.3 1.3 1.2 1.3 1.2 1.2 1.2 1.2 1.2 1.2 1.2 0.007 9.3 1.2 1.2 1.1 1.1 1.1 1.1 1.1 1.1 1.1 1.1 1.1 1.1 0.006 Remarks: MASS LOSS (chapter 3 & 4) was observed in some CYC stages, and associated with very small volume change. Small permeameter: test C-W2-T120: change in specimen height (mm) iav UNI1a CYC1 UNI1b UNI2a CYC2 UNI2b UNI3a CYC3 UNI3b 1.2 0 0 0 0 0 0 0 0 0 5.5 0 0 0 0 0.01 0 0 0.02 0 9.3 0 0.02 0 0 0.02 0 0 0.04 0 Small permeameter: test D-NW1-T6 iav UNI 1a UNI 1b UNI 2a UNI 2b UNI 3a UNI 3b kav (cm/s) GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 1 1.2 1.4 1.2 1.4 1.2 1.3 1.2 1.3 1.2 1.4 1.2 1.4 0.004 5 1.1 1.3 1.0 1.2 1.0 1.3 1.0 1.2 1.0 1.2 1.1 1.3 0.003 9 1.1 1.4 1.1 1.2 1.2 1.3 1.1 1.3 1.1 1.3 1.1 1.4 0.003 Remarks: (a) NO MASS LOSS was observed in all UNI and CYC stages. (b) ZERO volume change. 235 Small permeameter: test D-NW2-T6 iav UNI 1a UNI 1b UNI 2a UNI 2b UNI 3a UNI 3b kav (cm/s) GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 1 0.9 0.9 0.9 0.8 1.0 0.9 1.0 0.9 0.9 0.8 1.0 0.9 0.005 5 1.0 1.0 1.0 0.9 1.0 0.8 1.0 0.9 1.0 0.9 0.9 0.8 0.005 9 1.0 0.9 0.9 0.8 0.9 0.8 0.9 0.8 1.0 0.9 0.9 0.9 0.004 Remarks: (a) NO MASS LOSS was observed in all UNI and CYC stages. (b) ZERO volume change. Small permeameter: test D-NW4-T6 iav UNI 1a UNI 1b UNI 2a UNI 2b UNI 3a UNI 3b kav (cm/s) GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 1 1.0 1.0 1.0 0.9 1.0 0.9 1.0 1.0 1.0 0.9 1.0 1.0 0.005 5 1.0 1.0 1.0 1.0 1.0 0.9 1.0 1.0 1.0 0.9 1.0 0.9 0.004 9 1.0 0.9 0.9 0.9 1.0 0.9 1.0 0.9 1.0 1.0 1.0 1.0 0.004 Remarks: (a) NO MASS LOSS was observed in all UNI and CYC stages. (b) ZERO volume change. Small permeameter: test D-NW4-T60 iav UNI 1a UNI 1b UNI 2a UNI 2b UNI 3a UNI 3b kav (cm/s) GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 1 1.0 1.0 1.0 0.9 1.0 0.9 1.0 1.0 1.0 0.9 1.0 1.0 0.005 5 1.0 1.0 1.0 1.0 1.0 0.9 1.0 1.0 1.0 0.9 1.0 0.9 0.005 9 1.0 0.9 0.9 0.9 1.0 0.9 1.0 0.9 1.0 1.0 1.0 1.0 0.003 Remarks: (a) NO MASS LOSS was observed in all UNI and CYC stages. (b) ZERO volume change. 236 Small permeameter: test D-NW5-T6 iav UNI 1a UNI 1b UNI 2a UNI 2b UNI 3a UNI 3b kav (cm/s) GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 1 1.2 1.3 1.2 1.3 1.2 1.3 1.2 1.4 1.2 1.4 1.2 1.3 0.006 5 1.2 1.3 1.1 1.3 1.1 1.3 1.2 1.3 1.2 1.3 1.2 1.3 0.005 9 1.2 1.3 1.1 1.3 1.1 1.3 1.1 1.3 1.1 1.3 1.1 1.3 0.005 Remarks: (a) NO MASS LOSS was observed in all UNI and CYC stages. (b) ZERO volume change. Small permeameter: test D-W1-T6 iav UNI 1a UNI 1b UNI 2a UNI 2b UNI 3a UNI 3b kav (cm/s) GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 1.1 1.1 1.2 1.1 1.2 1.0 1.2 1.0 1.2 1.0 1.2 1.1 1.3 0.004 5.2 1.1 1.3 1.1 1.3 1.0 1.3 1.0 1.2 1.0 1.3 1.1 1.2 0.006 9.5 1.1 1.2 1.1 1.1 1.1 1.1 1.1 1.1 1.0 1.1 1.0 1.1 0.007 Remarks: MASS LOSS (chapter 4) was observed in all CYC stages, and associated with volume change. Small permeameter: test D-W1-T6: change in specimen height (mm) iav UNI1a CYC1 UNI1b UNI2a CYC2 UNI2b UNI3a CYC3 UNI3b 1.1 0 0.12 0 0 0.04 0 0 0.15 0 5.2 0 0.25 0 0 0.44 0 0 0.83 0 9.5 0 0.37 0 0 0.42 0 0 0.90 0 237 Small permeameter: test D-W1-T6-R (repeated) iav UNI 1a UNI 1b UNI 2a UNI 2b UNI 3a UNI 3b kav (cm/s) GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 1.1 1.1 1.2 1.1 1.2 1.0 1.1 1.1 1.1 1.1 1.2 1.1 1.1 0.005 5.3 1.1 1.1 1.0 1.2 1.0 1.2 1.0 1.1 1.0 1.1 1.0 1.1 0.006 9.7 1.1 1.1 1.1 1.2 1.0 1.2 1.0 1.0 1.0 1.0 1.0 1.1 0.006 Remarks: MASS LOSS (chapter 4) was observed in all CYC stages, and associated with volume change. Small permeameter: test D-W1-T6-R (repeated): change in specimen height (mm) iav UNI1a CYC1 UNI1b UNI2a CYC2 UNI2b UNI3a CYC3 UNI3b 1.1 0 0.15 0 0 0.15 0 0 0.18 0 5.3 0 0.26 0 0 0.49 0 0 0.74 0 9.7 0 0.56 0 0 0.61 0 0 1.01 0 Small permeameter: test D-W1-T60 iav UNI 1a UNI 1b UNI 2a UNI 2b UNI 3a UNI 3b kav (cm/s) GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 1.2 1.2 1.5 1.2 1.5 1.2 1.5 1.2 1.4 1.2 1.4 1.2 1.4 0.004 5.3 1.2 1.4 1.1 1.4 1.1 1.4 1.0 1.3 1.0 1.3 1.1 1.3 0.004 9.4 1.1 1.3 1.1 1.2 1.1 1.3 1.1 1.3 1.1 1.2 1.1 1.2 0.005 Remarks: MASS LOSS (chapter 4) was observed in all CYC stages, and associated with volume change. Small permeameter: test D-W1-T60: change in specimen height (mm) iav UNI1a CYC1 UNI1b UNI2a CYC2 UNI2b UNI3a CYC3 UNI3b 1.2 0 0 0 0 0.03 0 0 0.04 0 5.3 0 0 0 0 0.04 0 0 0.04 0 9.4 0 0.02 0 0 0.05 0 0 0.06 0 238 Small permeameter: test D-W1-T60-R (repeated) iav UNI 1a UNI 1b UNI 2a UNI 2b UNI 3a UNI 3b kav (cm/s) GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 1.0 1.3 1.3 1.3 1.4 1.3 1.3 1.3 1.3 1.3 1.4 1.3 1.4 0.005 5.2 1.3 1.3 1.3 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.3 0.004 9.7 1.2 1.3 1.2 1.2 1.2 1.3 1.2 1.3 1.3 1.3 1.3 1.3 0.006 Remarks: MASS LOSS (chapter 4) was observed in all CYC stages, and associated with volume change. Small permeameter: test D-W1-T60-R (repeated): change in specimen height (mm) iav UNI1a CYC1 UNI1b UNI2a CYC2 UNI2b UNI3a CYC3 UNI3b 1.0 0 0.02 0 0 0.03 0 0 0.05 0 5.2 0 0.04 0 0 0.04 0 0 0.05 0 9.7 0 0.03 0 0 0.06 0 0 0.09 0 Small permeameter: test D-W1-T120 iav UNI 1a UNI 1b UNI 2a UNI 2b UNI 3a UNI 3b kav (cm/s) GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 1 1.1 1.1 1.0 1.0 1.0 1.1 1.0 1.1 1.0 1.1 1.0 1.0 0.005 5 1.0 1.0 1.0 1.1 1.0 1.1 1.0 1.1 1.0 1.1 1.0 1.1 0.005 9 1.0 1.1 1.0 1.1 0.9 1.1 1.0 1.0 1.0 1.0 1.0 1.1 0.004 Remarks: MASS LOSS (chapter 4) was observed in some CYC stages, and associated with volume change. Small permeameter: test D-W1-T120: change in specimen height (mm) iav UNI1a CYC1 UNI1b UNI2a CYC2 UNI2b UNI3a CYC3 UNI3b 1 0 0 0 0 0 0 0 0 0 5 0 0 0 0 0 0 0 0.03 0 9 0 0.02 0 0 0.02 0 0 0.03 0 239 Small permeameter: test D-W2-T6 iav UNI 1a UNI 1b UNI 2a UNI 2b UNI 3a UNI 3b kav (cm/s) GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 1.3 1.1 1.2 1.1 1.0 1.1 1.0 1.0 0.9 1.0 0.9 0.9 0.9 0.004 5.4 0.9 0.8 0.9 0.8 0.9 0.8 - - - - - - 0.005 Remarks: MASS LOSS (chapter 4) was observed in all CYC stages, and associated with volume change. Test stopped at the end of stage CYC2 at iav of 5.4. Small permeameter: test D-W2-T6: change in specimen height (mm) iav UNI1a CYC1 UNI1b UNI2a CYC2 UNI2b UNI3a CYC3 UNI3b 1.3 0 0.73 0 0 0.96 0 0 1.22 0 5.4 0 1.73 0 0 1.50 - - - - 240 Large permeameter: test C-W1-T6 (as W1-T6(L) in chapter 3) iav UNI 1a UNI 1b UNI 2a UNI 2b UNI 3a UNI 3b kav (cm/s) GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 1.3 1.2 1.0 1.1 1.0 1.1 1.0 1.1 1.0 1.1 1.0 1.1 1.0 0.008 4.8 1.1 1.0 1.1 0.9 1.1 0.9 1.1 0.9 1.1 0.9 1.1 0.9 0.007 8.9 1.0 0.9 1.0 0.9 1.0 0.9 1.0 0.9 1.0 0.9 1.0 0.9 0.007 Remarks: (a) NO MASS LOSS was observed in all UNI and CYC stages. (b) ZERO volume change. Large permeameter: test C-W2-T6 (as W2-T6(L) in chapter 3) iav UNI 1a UNI 1b UNI 2a UNI 2b UNI 3a UNI 3b kav (cm/s) GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 1.3 0.9 0.9 0.9 0.9 0.9 0.9 0.9 0.8 0.9 0.8 0.9 0.8 0.007 4.9 0.9 0.9 0.9 0.9 0.9 0.8 0.9 0.8 0.8 0.8 0.8 0.8 0.009 8.8 0.9 0.8 0.8 0.8 0.8 0.8 0.8 0.7 0.8 0.7 0.8 0.7 0.010 Remarks: (a) MASS LOSS was observed in all CYC stages (visual observations): total loss = 5487 g/m 2 . (b) Total volume change = 3.1%. Large permeameter: test C-W2-T60 (as W2-T60(L) in chapter 3) iav UNI 1a UNI 1b UNI 2a UNI 2b UNI 3a UNI 3b kav (cm/s) GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 GR 25 GR 8 1.4 0.9 1.3 0.9 1.3 0.9 1.2 0.9 1.2 0.9 1.2 0.9 1.2 0.006 4.7 0.9 1.2 0.9 1.0 0.9 1.0 0.9 1.0 0.9 1.1 0.9 0.9 0.007 9.0 0.9 1.0 0.8 1.0 0.8 1.0 0.8 1.0 0.8 1.0 0.9 1.0 0.006 Remarks: (a) NO MASS LOSS was observed in some CYC stages (visual observations): total loss = 959 g/m 2 . (b) Total volume change = 0.47%. 241 Figure C1 Direct shear test results 242 Appendix D Select photographs: small permeameter Figure D1 Placement of geotextile W2 sample on permeameter base 243 Figure D2 Reconstituted test specimen (soil C) 244 Figure D3 Head-controlled in cyclic flow: 3-way solenoid valve 245 Figure D4 Assembly of the small permeameter: axial loading device 246 Figure D5 Flexible hose discrete sampling of mass loss through the geotextile 247 Figure D6 Geotextile NW4 after test D-NW4-T6 (AOS/D85 \u00E2\u0089\u0088 2.3) 248 Figure D7 Geotextile W2 after test D-W2-T6 (AOS/D85 \u00E2\u0089\u0088 3.7) 249 Figure D8 Soil specimen after test D-W2-T6 (AOS/D85 \u00E2\u0089\u0088 3.7) 250 Appendix E Select photographs: large permeameter Figure E1 Assembly of the large permeameter 251 (a) (b) Figure E2 Test C-W2-T = 6 s (AOS/D85 = 2.8): a) post-test specimen; b) deposition of soil loss through the geotextile 252 (a) (b) Figure E3 Test C-W2-T = 60 s (AOS/D85 = 2.8): a) post-test specimen; b) deposition of soil loss through the geotextile"@en . "Thesis/Dissertation"@en . "2010-11"@en . "10.14288/1.0062955"@en . "eng"@en . "Civil Engineering"@en . "Vancouver : University of British Columbia Library"@en . "University of British Columbia"@en . "Attribution-NonCommercial-NoDerivatives 4.0 International"@en . "http://creativecommons.org/licenses/by-nc-nd/4.0/"@en . "Graduate"@en . "A laboratory permeameter study of geotextile-soil retention in cyclic flow"@en . "Text"@en . "http://hdl.handle.net/2429/25029"@en .