Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Soil/geotextile filtration behavior under dynamic conditions of cyclic flow and vibration Hameiri, Avikam 2000

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

Item Metadata

Download

Media
831-ubc_2000-486427.pdf [ 11.57MB ]
Metadata
JSON: 831-1.0063513.json
JSON-LD: 831-1.0063513-ld.json
RDF/XML (Pretty): 831-1.0063513-rdf.xml
RDF/JSON: 831-1.0063513-rdf.json
Turtle: 831-1.0063513-turtle.txt
N-Triples: 831-1.0063513-rdf-ntriples.txt
Original Record: 831-1.0063513-source.json
Full Text
831-1.0063513-fulltext.txt
Citation
831-1.0063513.ris

Full Text

SOJL/GEOTEXTHE FILTRATION BEHAVIOR UNDER DYNAMIC CONDITIONS OF CYCLIC FLOW AND VIBRATION by AVIKAM HAMEIRI B.Sc, Technion - Israel Institute of Technology, 1994 M.Sc, Technion - Israel Institute of Technology, 1996  A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PFflLOSOPHY in THE FACULTY OF GRADUATE STUDIES Department of Civil Engineering We accept this thesis as conforming to the required standard  THE UNIVERSITY OF BRITISH COLUMBIA April 2000 © Avikam Hameiri, 2000  In presenting this thesis in partial fulfillment of the requirement for an advanced degree at the University of British Columbia, I agree that the library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis forfinancialgain shall not be allowed without my written permission.  Department of Civil Engineering The University of British Columbia Vancouver, Canada  Date  MM I  K)  K  i ooo  ABSTRACT  An evaluation of soil internal stability and soil/geotextile filtration compatibility for different types of loading conditions is important for the proper use and extension of empirically based design procedures. Static conditions like those found in earth dams and vertical drains, are characterized by steady flow. Dynamic conditions can, however, be present in some situations, for example under revetments due to hydraulic disturbance and under railways due to physical mechanical disturbance. To better understand the filtration performance of nonwoven geotextiles under dynamic conditions, this experimental study was undertaken. An existing permeation (Gradient Ratio) device was modified to perform tests with vibration. In addition, a new automatic cyclic Gradient Ratio device with a computerized control system was designed and commissioned to perform tests with cyclic flow. Forty-one combinations of four nonwoven geotextiles with narrow, wide, or gap-graded model soils were examined in testing. The reconstituted model soils were composed of glass bead fractions in the range between coarse silt andfinesand. The test program was conducted under a hydraulic gradient of four, and involved multi-stage tests. In the vibration tests, an initial stage of unidirectional flow was imposed and followed by a dynamic stage that involved physical disturbance using controlled energy blows at a frequency of 3 FIz. In cyclic flow tests the initial unidirectional flow was followed by reversing the direction of flow at frequencies of 0.2 Hz and 0.02 Hz, under both confined (a =25 kPa) and unconfined (ov=0 kPa) conditions. v  ii  The results of each test were interpreted from the measurements of flow rate, water head distribution along the sample length, visual observations both during and after testing, and the weight and gradation of the particles that passed through the geotextile. Reflecting on existing design criteria and previous research work, the analysis and conclusions address issues of internal stability and soil geotextile compatibility under both static and dynamic conditions.  iii  TABLE OF CONTENTS  ABSTRACT  ii  LIST OF TABLES  xiii  LIST OF FIGURES  xvii  LIST OF SYMBOLS  xxv  ACKNOWLEDGEMENTS  xxx  1. INTRODUCTION  1  1.1 Geotextile Properties for Design  1  1.2 Design Approach  1  1.3 Objectives  2  1.4 Thesis Organization  3  2. LITERATURE REVTEW  4  2.1 The Basic Principles of Filters and the Related Conditions  4  2.2 Internal Stability ofNoncohesive Soils  6  2.3 Factors Affecting the Internal Stability ofNoncohesive Soils  7  2.3.1 Particle Shape and Size  7  2.3.2 Density  8  2.3.3 Confining Stress  8  2.3.4 Mechanical Loading  8 iv  2.3.5 Hydraulic Loading  9  2.4 Geotextiles  10  2.4.1 Properties of Geotextiles  11  2.4.2 The Geotextile Opening Size  12  2.5 The Principles of Geotextile Filter Performance  13  2.6 Factors Affecting the Performance of the Soil/Geotextile Combination  15  2.7 Design Criteria for Geotextile Filters Under Unidirectional and Cyclic Flow Conditions  19  2.7.1 Soil Retention  19  2.7.2 Permeability  22  2.7.3 Clogging Resistance  23  2.8 Selected Previous Laboratory Studies  23  2.8.1 Static Conditions (The Gradient Ratio Device - a State of Practice)  24  2.8.2 Dynamic Conditions (State of the Art)  26  2.8.2.1 Railroads and Highways (Pulsating Load)  26  2.8.2.2 Sloping Bank  27  2.8.2.3 Pulsating Flow Condition  28  2.8.2.4 Hydrodynamic Sieving (Cyclic Flow)  28  2.8.2.5 Cyclic Flow Control (Using Flow Pump)  28  2.9 Research Needs and Objectives  29  3. A P P A R A T U S  37  3.1 Introduction  37  v  3.2 The Vibration GR Device  37  3.2.1 Permeameter  38  3.2.2 Collector Trough  38  3.2.3 Hydraulic Supply System  39  3.2.4 Automatic Controlled Vibration System  39  3.3 The Cyclic GR Device  40  3.3.1 Permeameter  40  3.3.2 Collector Trough  41  3.3.3 Vertical Pressure System  42  3.3.4 Top and Bottom Boundaries  42  3.3.5 Hydraulic Control System  43  3.3.6 Water Supply and Flow Rate Measurement System  44  3.3.7 Data Acquisition and Control System  45  3.4 Sedigraph Particle Size Analyzer  46  4. MATERIAL PROPERTIES  61  4.1 Introduction  61  4.2 Model Soils  61  4.2.1 Vibration Tests  62  4.2.2 Cyclic Flow Tests  63  4.3 Geotextile Samples  63  4.4 Test Program  64  5. TEST PROCEDURE  70  vi  5.1 Introduction  70  5.2 Preparation of the Geotextiles and Model Soil  70  5.3 Vibration Tests  71  5.3.1 Preparation of the Test Apparatus  71  5.3.2 Geotextile Placement  72  5.3.3 Soil Placement and Test Initiation  72  5.3.4 Test Program  73  5.3.5 On The Severity of Vibration (Vibration Amplitude)  74  5.3.6 Control and Data Acquisition  75  5.3.7 Post-Test Procedure  76  5.4 Cyclic Flow Tests  76  5.4.1 Preparation of The Test Apparatus  77  5.4.2 Placement of the Soil and Geotextile  77  5.4.3 Application of the Confining Stress and the Test Initiation  79  5.4.4 Control and Data Acquisition  80  5.4.5 Post-Test Procedure  82  5.4.6 On The Severity of Cyclic Flow  83  5.5 Summary  83  5.5.1 Vibration  83  5.5.2 Cyclic Flow  84  6. TEST RESULTS AND PRELIMINARY OBSERVATIONS 6.1 Introduction  87 87  vii  6.1.1 Treatment of Top Blinding  87  6.2 Vibration Tests  88  6.2.1 Static Stage  89  6.2.1.1 Water Head Distribution  89  6.2.1.2 Sample Conditions  91  6.2.1.2.1 Permeability  91  6.2.1.2.2 Void Ratio  92  6.2.1.2.3 Mass of Passing Through Particles  94  6.2.2 Dynamic Stage  96  6.2.2.1 Water Head Distribution  96  6.2.2.2 Sample Conditions  97  6.2.2.2.1 Permeability  97  6.2.2.2.2 Void Ratio  98  6.2.2.2.3 Mass of Passing Through Particles  99  6.3 Cyclic Flow Test  100  6.3.1 Stage 1: Static Stage  100  6.3.1.1 Water Head Distribution  100  6.3.1.2 Sample Conditions  102  6.3.1.2.1 Permeability  102  6.3.1.2.2 Void Ratio  103  6.3.1.2.3 Mass of Passing Through Particles  103  6.3.2 Stages 2-4: Dynamic Stages  104  6.3.2.1 Water Head Distribution  104 viii  6.3.2.1.1 Stage 2 (f=0.2 Hz, o =25 kPa)  105  6.3.2.1.2 Stage 3 (f= 0.02 Hz, a =25 kPa)  105  v  v  6.3.2.1.3 Stage 4 (£=0.2 Hz, o =0 kPa)  107  v  6.3.2.2 Sample Condition  107  6.3.2.2.1 Permeability and Void Ratio  107  6.3.2.2.2 Mass of Passing Through Particles  108  6.3.3 Stage 5: Static Stage  109  6.4 Gradation of the Passing Through Particles  109  6.5 Repeatability  Ill  6.6 Summary  112  7. A N A L Y S I S O F T E S T R E S U L T S  122  7.1 Introduction  122  7.2 The Phenomenon of an External Blinding Layer  122  7.2.1 Static Conditions  123  7.2.1.1 Circulated Water  124  7.2.1.2 Non-Circulated Water  126  7.2.1.3 Synthesis  127  7.2.2 Dynamic Conditions  128  7.2.2.1 Vibration  :  7.2.2.2 Cyclic Flow  128 128  7.3 Base Soil Internal Stability under Static Conditions  129  7.4 Base Soil Behavior under Dynamic Conditions  131  ix  7.4.1 Generalized State  131  7.4.2 Quick Conditions  133  7.4.3 Cyclic Flow Permeability  134  7.4.3.1 Finer Fraction Mobility  135  7.4.3.2 Electrical analogy  137  7.4.3.3 Apparent anisotropy  139  7.4.3.4 Implications for Practice  141  7.4.4 Base Soil Internal Stability Under Dynamic Conditions 7.4.4.1 Vibration  142 143  7.4.4.1.1 Step 1: Migration of Particles Out of the Soil Sample  144  7.4.4.1.2 Step 2: Migration of Particles Within the Sample  146  7.4.4.2 Cyclic Flow  149  7.4.4.2.1 Step 1: Migration of Particles Out of the Soil Sample  149  7.4.4.2.2 Step 2: Migration of Particles Within the Soil Sample  151  7.4.4.3 Criteria Development  152  7.4.4.3.1 The "Concave Degree" (CD) for Narrow and Wide Gradations  154  7.4.4.3.2 Gap-Graded Samples  156  7.5 Gradation of the Passing Through Particles  157  7.6 Soil/Geotextile Compatibility  159  7.6.1 Static Conditions  160  7.6.1.1 Piping  160  7.6.1.2 Blinding  162  7.6.2 Vibration Disturbance  163  7.6.2.1 Piping  163  7.6.2.2 Blinding  164  7.6.3 Cyclic Flow Disturbance  165  7.6.3.1 Piping  165  7.6.3.1.1 Confined Conditions  166  7.6.3.1.1 Unconfined Conditions  166  7.6.3.2 Blinding  167  7.6.5 Summary of the Suggested Criteria: Narrow and Wide Gradations  168  7.6.5.1 Static and Confined Dynamic  168  7.6.5.2 Unconfined Dynamic  168  8. S U M M A R Y A N D C O N C L U S I O N S  197  8.1 Objectives  197  8.2 Apparatus  198  8.3 Materials  198  8.4 Test Procedures  199  8.5 Analysis and Design Implications  199  8.5.1 General Behavior  199  8.5.1.1 Blinding Layer and the Gradient Ratio Value (Static Unidirectional Flow)  199  8.5.1.2 Vibration  200  8.5.1.3 Cyclic Flow  200  8.5.1.3.1 Head Distribution  200  xi  8.5.1.3.2 Permeability  201  8.5.1.4 The Passing Through Particles 8.5.2 Internal Stability.  202 202  8.5.2.1 Static Conditions  202  8.5.2.2 Dynamic Conditions  203  8.5.3 Soil-Geotextile Compatibility  204  8.5.4 Implications for Research and Practice  205  8.6 Recommendations for Further Studies  206  BIBLIOGRAPHY  209  APPENDIX A. TESTS RESULTS  225  A l : Vibration tests  225  A2: Cyclic flow test  250  APPENDIX B. BASIC CALCULATIONS  268  B.l Void ratio calculation  268  B.2 Darcy's law  268  B.3 Validity of Darcy's law  268  B.4 Gradation correction  269  xii  LIST OF TABLES  Table 2.1: The severity of the physical conditions (after Holtz et al., 1997)  5  Table 2.2: Typical hydraulic gradients (after Giroud, 1996)  9  Table 2.3: Possible hydraulic loads based on the external applied conditions  17  Table 2.4: Cyclic load and related period (after Mouw et. al., 1986)  18  Table 2.5: Existing geotextile retention criteria  20  Table 2.6: Filtration criteria that include cyclic flow conditions  21  Table 2.7: Permeability criteria that can be relevant to cyclic flow conditions  22  Table 3.1: Vibration test device: port locations (mm)  38  Table 3.2: Cyclic flow test device: port locations (mm)  41  Table 4.1: Physical properties of the glass beads (from the manufacturer's technical  literature)  ;  61  Table 4.2: Description of narrow and wide gradations (vibration tests)  62  Table 4.3: Description of gradations that show a gap (vibration tests)  62  Table 4.4: Description of narrow and wide gradations (cyclic flow tests)  63  Table 4.5: Description of gap graded gradations (cyclic flow tests)  63  xiii  Table 4.6: Manufacturers of the geotextiles used  64  Table 4.7: Geotextile material properties  64  Table 4.8: Vibration test program  65  Table 4.9: Cyclic flow test program  65  Table 5.1: Cumulative mass per unit area (m , g/m ) of the passing through particles  75  Table 6.1: The GR values at the end static and dynamic stages  90  2  A  Table 6.2: The permeability (ky, 1E-3 cm/sec.) and flow rate (Q, cm3/sec.) at the end of static and dynamic stages  92  Table 6.3: The void ratio (e) at the end of static and dynamic stages  93  Table 6.4: Cumulative mass per unit area of the passing through particles before and  during vibration (m , g/m )  94  2  A  Table 6.5: The GR values at the end of the static stages (stages 1 and 5)  101  Table 6.6: The void ratio (e), permeability (kn, 1E-3 cm/sec) and flow rate (Q,  cm /sec.) at the end of static and dynamic stages 3  102  Table 6.7: Mass per unit area (m , g/m ) of passing through particles  104  Table 6.8: The GR values at the end of half a cycle (after 10 cycles)  106  2  A  xiv  Table 6.9: D95p of the passing through particles from the vibration GR device (1E3mm)  :  110  Table 6.10: D95p of the passing through particles from the cyclic GR device  (lE-3mm)  110  Table 7.1: The time for preliminary stabilization (tp, sec.) of GR within half a cycle (after 10 cycles): stage 3  133  Table 7.2: Equivalencies in seepage and flow of electrical current (after Bardet 1997).. 137  Table 7.3: A comparison of D95p and that of the original sample: vibration tests  146  Table 7.4: A comparison of D95p and that of the original sample: stage 2  150  Table 7.5: A comparison of D95p and that of the original sample: stage 4  151  Table 7.6: The soil samples stability: vibration tests  152  Table 7.7: The soil samples stability: continuous cyclic flow  153  Table 7.8: A comparison between the predicted performance of narrowly and widely graded soil samples based on the Kenney and Lau (1985,1986) technique and the observed performances  154  Table 7.9: The Interaction ratio (IR) values for static unidirectional conditions: piping.. 161  Table 7.10: The Interaction ratio values (IR) for static unidirectional conditions: blinding  163 xv  Table 7 . 1 1 : The interaction ratio (IR) for confined cyclic flow conditions  165  Table 7 . 1 2 : The interaction ratio (IR) for unconfined cyclic flow conditions  166  Table 8 . 1 : Suggested retention criteria for narrow and wide gradations  205  xvi  LIST OF FIGURES Figure 2.1: A graphical procedure to evaluate the internal stability of granular materials, (Based Kenney, andLau 1985, 1986)  31  Figure 2.2: The three classes of gradation profiles, (from Lafleur et al., 1989)  31  Figure 2.3 idealizedfiltercake (after Ingold 1994)  32  Figure 2.4: Flow direction: a.) Parallel to the plane (in plane), b.) Perpendicular to the plane, (cross plane)  32  Figure 2.5: The geotextile local displacement as a result of cyclic flow: a.) Flapping b.) Uplift (After Cazzuffi, 1996)  33  Figure 2.6: The GR permeameter test arrangement, (after Fannin et al., 1994b)  33  Figure 2.7: Three zones development in filtration testing  34  Figure 2.8: A schematic principle of the tests that were carried out in order to simulate railroads or highway conditions  34  Figure 2.9: Sloping bank test equipment for non unidirectional flow study (after Wewerka, 1986)  35  Figure 2.10: Schematic diagram of the pulsatingflowfiltrationtest set up (After Narejo and Koerner 1992)  35  Figure 2.11: Hydrodynamic sieving apparatus and testing cylinders setup (After Batia et al., 1996)  36  Figure 2.12:Cross plane cyclicflowcontrol test device (After Cazzuffi., et al., 1996)  36  Figure 3.1: The unidirectional GR device with the automatic hammer (picture)  48  Figure 3.2: The vibration GR device (schematic drawing)  49  xvii  Figure 3.3: The automatic vibration system (schematic drawings): a.) Double acting drop-hammer (picture), b.) Supporting system  50  Figure 3.4: The cyclic GR device (picture)  51  Figure 3.5: Schematic view of the cyclic test device  52  Figure 3.5: Schematic view of the cyclic test device  53  Figure 3.6: Cyclic GR device: location of ports and the confining boundaries  54  Figure 3.7: Pressure transducers/manometer interface  55  Figure 3.8: Schematic view of the Ports connections: a.) The ports/pressure transducers/manometers interface; b.) Distribution of the Pressure transducers-port connections  56  Figure 3.9: The discrete samples collection in the lower collector through (lc)  57  Figure 3.10: Modeling the reversing flow regime in the permeameter  58  Figure 3.11: The flow rate measuring assembly  59  Figure 3.12: Schematic view of the open loop reading control program  60  Figure 4.1: Reference name for narrow and wide gradations  66  Figure 4.2: Reference name for gap graded gradations  66  Figure 4.3: Vibration test program particles size distribution curves of : a) narrow gradations, b) wide gradations, and c) gap graded gradation  67  Figure 4.4: Cyclic flow test program particles size distribution curves of: a.) narrow and wide gradation, b.) gap graded gradations  68  Figure 4.5: Reference name for the combinations tested  69  Figure 5.1: The vibration test process  85  Figure 5.2: The cyclic flow test process  86 xviii  Figure 6.1: The water head distribution before and after top blinding formation (s.(20-1.5.gl49)  114  ;  Figure 6.2: Head loss with elapsed time due to top blinding layer  114  Figure 6.3: The influence of the water head distribution on the GR values  115  Figure 6.4: The water head distribution as a result of vibration (N =3780 blows)  116  Figure 6.5: The influences of blow count on the gradient ratio  117  Figure 6.6: The permeability values response due to vibration  117  b  Figure 6.7: Stage 1 - The water head distribution at the end of the static stage  118  Figure 6.8: Stage 2 - The measured head difference with time (f=0.2 Hz, after 10 cycles)  118  Figure 6.9: Stage 3 - The measured head difference with time (f=0.02 Hz, after 10 cycles)  119  Figure 6.10: Stage 3 - The water head response to cyclic flow  119  Figure 6.11: Stage 4 - The measured head difference with time (f=0.2 Hz, after 10 cycles)  120  Figure 6.12: The influence of the flow rate on the total head difference along the sample  120  Figure 6.13: Grain size distribution of flushing through particles (cyclic flow test)  121  Figure 7.1: Using recycled water, the change in the water head distribution along s(4.397). g 103 as a result of the external blinding layer  170  Figure 7.2: Schematic influence of the external blinding layer on the GR: a.) top external blinding layer, and b.) bottom external blinding layer  xix  170  Figure 7.3: The variation of the GR(Mod.) with time (redrawn after Nishigata et al., 2000)  171  Figure 7.4: The variation of GR with blow count (initially non- uniform samples)  171  Figure 7.5: Using Kenney and Lau (1985,1986) technique to evaluate the internal stability of the gradations tested  172  Figure 7.6: The transient dynamic water hammer effect in the sample (after 10 cycles). 172 Figure 7.7: The influence of the change in the flow direction on the GR values (after 10 cycles and 4 seconds, f=0.02 Hz)  173  Figure 7.8: The quick conditions along the sample (s.(5.1-144).gl22; f=0.2 Hz; crv=0) 174 Figure 7.9: The influence of flow rate and permeability on the measured head difference  175  Figure 7.10: The influence of the cycling flow on the arrangement of the finer fraction within the voids: a.) steady flow, b.) unsteady flow  176  Figure 7.11: The influence of the finer fraction size on changes in the permeability, as the cyclic flow frequency increases (observed)  176  Figure 7.12: The influence of thefinerfraction size on the changes in the permeability as the cycling flow frequency increase (hypothesized)  177  Figure 7.13: Applying the fictional alternating emfs to the circuit  177  Figure 7.14: The head difference changes across the soil layer (h35) during 25sec: a.) f=0.02HZ, b.) f=0.2HZ  178  Figure 7.15: The potential change on a capacitor with time: a.) charging, b.) discharging  178  xx  Figure 7.16: Modeling the overall cycling flow behavior, through resistors, capacitor and an alternatingfictionalemfs circuit  179  Figure 7.17: The measured current (i ) on the ammeter (A) during charging or A  discharging periods  179  Figure 7.18: The water head distributions at different time steps within half a cycle f=0.2Hz)  180  Figure 7.19: The process of evaluating the internal stability of soil samples that were tested: a.) generic, b.) vibration tests, c.) cyclic flow tests  183  Figure 7.20: The m -N curves  184  Figure 7.21: Piping susceptibility shapes: a.) Instantaneous b.) Continuous  184  Figure 7.22: The m -SPR relationship  185  D  D  Figure 7.23: The influence of vibration on the void ratio, gradient ratio, and permeability  186  Figure 7.24: The influence of the GR value on SBR  186  Figure 7.25: The influence of the concavity of the gradation profile on the analyzed performance: for soil samples with Cu>4 Figure 7.26: The concave degree (CD) procedure: soil sample s(6.4-3.6)  187 188  Figure 7.27: The boundary between stable and unstable gradations (Cu>4): CD procedure  188  Figure 7.28: Internal stability: gap graded samples  189  Figure 7.29: The influence of the pore channel constriction on passing through particles  189  Figure 7.30: The gradation of the passing through particles: soil sample s(3.2-68) xxi  190  Figure 7.31: The conditions applied here compared to one of Tondello (1998) limit states  190  Figure 7.32: The interaction ratio (I ) based on different piping criteria with the current R  test resultsfromstatic unidirectional conditions  191  Figure 7.33: Applying Lafleur (1998) criteria for the onset of piping (2500 g/m ) to the 2  results obtained by Fannin et al., (1994a) using soils with 3<Cu<6  191  Figure 7.34: Filtration behavior of soils with Cu<2 using the results obtained by Fannin et al., (1994a) and in this work  192  Figure 7.35: The interaction ratio (IR) based on the different blinding criteria with the current test resultsfromstatic unidirectional conditions  192  Figure 7.36: The interaction ratio (I ) based on the different piping criteria with the R  current test results from vibration tests  193  Figure 7.37: The influence of the FOS/D30 on the SBR values  193  Figure 7.38: Using FCGG (1986) criterion for dense and confined soils on the collected data  194  Figure 7.39: Using CGS (1992) criterion on the collected data  194  Figure 7.40: Using Mlynarek et al. (1999) criterion for non-severe hydraulic conditions on the collected data  195  Figure 7.41: Using FCGG (1986) criteria for loose or unconfined soils on the collected data  195  Figure 7.42: Using Mlynarek et al. (1999) for Severe hydraulic conditions (wave attack or pumping) on the collected data Figure 7.43: The retention criteria using the D85 as an indicative grain size xxii  196 196  Figure A l . l : s(1.9-39).gl03i  226  Figure A1.2: s(1.9-39).gl03  227  Figure A1.3: s(2.2-52).gl03  228  Figure A1.4: s(2.2-78).gl03  229  Figure A1.5: s(4.3-97).gl03  230  Figure A1.6: s(6.4-213).gl03  231  Figure A1.7: s(6.4-306).gl03  232  Figure A1.8: s(15-2.7).gl03  233  Figure A1.9: s(20-1.5).gl03  234  Figure ALIO: s(20-3.1).gl03  235  Figure Al.11: s(45-3.1).gl03  236  Figure A1.12: s(70-1.6).gl03  237  Figure A1.13: s(2.2-52).gl22  238  Figure A1.14: s(2.2-78).gl22  239  Figure A1.15: s(6.4-213).gl22  240  Figure A1.16: s(15-2.7).gl22  241  Figure A1.17: s(70-1.6).gl22  242  Figure A1.18: s(4.3-97).gl49  243  Figure A1.19: s(6.4-306).gl49  244  Figure A1.20: s(20-1.5).gl49  245  Figure A1.21: s(4.3-97).g290  246  Figure A1.22: s(6.4-213).g290  247  2  xxiii  Figure A1.23: s(6.4-306).g290  248  Figure A1.24: s(15-4.5).g290  249  Figure A2.1: Symbols used  251  Figure A2.2: s(3.2-68).gl03  252  Figure A2.3: s(20-1.9).gl03  253  Figure A2.4: s(20-3.1).gl03  254  Figure A2.5: s(45-3.1).gl03  254  Figure A2.6: s(70-2.2).gl03  255  Figure A2.7: s(1.3-134).gl22  256  Figure A2.8: s(3.2-68).gl22  257  Figure A2.9: s(4.8-333).gl22  258  Figure A2.10: s(5.1-144).gl22  259  Figure A2.11: s(1.3-134).gl49  260  Figure A2.12: s(4.8-333).gl49  261  Figure A2.13: s(5.1-144).gl49  262  Figure A2.14: s(7.2-250).gl49  263  Figure A2.15: s(7.2-250).gl49  264  Figure A2.16: s(20-1.9).gl49  265  Figure A2.17: s(4.8-333).g290  266  Figure A2.18: s(7.2-250).g290  268  xxiv  LIST OF SYMBOLS  A  1.) Ammeter; 2.) Cross section o f the permeameter  AOS  Apparent Opening Size  b  Reaction bar  C  Capacity o f a capacitor (electrical)  CTJ  Coefficient of uniformity  CD  Concave degree  c  Collection trough  DF  The maximum diameter of the soil sample finer fraction.  Dgap  Diameter at the beginning o f the gap  Di  Indicative grain size diameter  Dn  Grain size corresponding to n% finer  D95p  The D95 of particles passing through the geotextile  e  1.) Water energy dissipater or 2.) V o i d ratio  ep  V o i d ratio o f the primary skeleton fabric ep  F  Mass fraction smaller than D  FOS  Filtration Opening Size  f  Frequency  /]  Fraction o f the loose particles within the primary skeleton fabric  GR  Gradient ratio.  Gs  Specific gravity  xxv  g  Geotextile sample  H  Constant head difference, or mass fraction between D and 4D  HT  Total head  h  Automatic double acting drop hammer Water head Head difference between ports i and j  I(P)  The preliminary slope of the cumulated M D - N or the G R - N curves  IR  Interaction ratio  I(S)  The secondary slope of the cumulated M D - N or GR-N curves  i  1.) Constant head inlet tank, or 2.) Hydraulic gradient, or 3.) Electrical current  ii  Measured steady unidirectional flow  12  Measured unsteady flow component  i  A  Ammeter current  i  cr  Critical gradient to cause quick conditions  i-o  Constant head inlet outlet tank  [  Hydraulic gradient across the soil  i  Hydraulic gradient across the soil geotextile composite  sg  k k  Geotextile permeability  s  k-.  Soil permeability Coefficient of permeability between port i and j.  xxvi  k sg  Soil geotextile interface permeability  L  Soil specimen length  lc  lower collection trough  1  Cell base legs  j.  distance from the bottom of the geotextile  j.j  distance between ports i and j  LVDT Linear variable differential transformer m  Flow rate measurement tank  mA  Cumulative mass per unit area of the passing through particles  m  Cumulative mass per unit area of the passing through particles during the  D  dynamic phase N  Cumulative number of blow count (b) or flow cycles (c).  n  Porosity  nj  Porosity of the loose particles within the pore spaces of the primary skeleton fabric  On  Geotextile opening size in which n% of the openings are smaller.  Ow90  Geotextile opening size (90% finer) based on wet sieving  o  Constant head outlet tank  p  Ports  p  a  The middle point of the concave section in a grain size distribution curve  p  h  The highest point of the concave section in a grain size distribution curve  xxvii  Pj  The lowest point of the concave section in a grain size distribution curve  POS  Performance opening size  PS1  Flow rate measurement pressure transducer  Q  Volumetric flow rate of water  R  Gap width ratio  Re  Reynolds number  r  Reservoir  S  Electrical on/off switch  SBR  Secant blinding ratio  SPR  Secant piping ratio  s  Soil sample  si  Flow regime solenoid valve  s2  Flow rate measurement solenoid valve  t  1.) Cell top cover plate or 2.) elapsed time  tL  Time needed for the water to travel from the top to the bottom of the specimen  tp  Time to reach preliminary stabilization  U  Fluid velocity  uc  Upper collection trough  V  Voltage (electrical)  Vc  Voltage across the capacitor  v  Test initiation valve  w  Cell wall  y  Dry unit weight  d  xxviii  y  w  Unit weight of water  r]  Fluid viscosity  p  Fluid unit mass  \\i  Pemittivity  xxix  ACKNOWLEDGEMENTS  I am deeply indebted to my thesis supervisor Dr. Jonathan Fannin for his support, advice, and engineering judgement throughout the study. Without his advice and understanding this research project would not have been possible.  I would like to express my sincere gratitude for Mr. Harald Schrempp a technician in the civil engineering workshop, for his help in fabrication of the cyclic and vibration Gradient Ratio devices. I would also like to thank Mr. Scott Jackson and Mr. John Wong, electrical technicians, for their help in commissioning of the data acquisition and control system.  This research is funded by a Research Grant from Natural Science and Engineering Research Council of Canada. I wish to express my thanks for the financial support provided.  Finally, I wish to thank my family overseas for their love and support.  xxx  1. INTRODUCTION 1.1 Geotextile Properties for Design The process of comparison and selection of a geotextile for filtration is not easy and depends on the specific application and function it is expected to provide. In the design process, the designer usually refers to the geotextile based on its index and performance properties. The index properties (mass per unit area, permeability, and others) are usually provided by the manufacturer and can be used for specifications, and quality control evaluations. The performance properties, however, are determined by testing a specific soil/geotextile combination using a standard test method such as the gradient ratio test (ASTM 5101). Generally, the performance tests should model the conditions in the field. These conditions include hydraulic regime, mechanical vibration and confining stresses.  1.2 Design Approach The geotextile should provide retention of the upstream soil particles without inducing an impediment to the flow. Based on a large database accumulated through tests and field observations, the retention demand is well established for static conditions. For dynamic conditions of vibration and cyclic flow the design practice is lacking as described below: •  Vibration - No research was found in the literature which specifically investigated the influence of vibration. This has been recognized as an area for further research (Kenney and Lau, 1985). Therefore, none of the criteria that have been developed address the condition of vibration.  •  Cyclic flow - Very little was done in studying the soil/geotextile compatibility in the laboratory under the conditions of cyclic flow. Therefore, some of the existing criteria 1  were established to be very conservative and can be difficult to fulfill (for example, CGS, 1992). In addition, the existing criteria do not correlate confining stresses to maximum hydraulic gradient and do not explicitly address gap-graded gradations and unstable soils.  1.3 Objectives The existing design criteria are largely limited to conditions of static unidirectional flow on soils that are internally stable. The challenges infiltrationcompatibility include: 1.) Soils that are internally unstable, 2.) Conditions of mechanical disturbance and, 3.) Hydraulic conditions that vary from those of unidirectional flow. Specifically the objectives of this study were to: •  Add an automatic vibration component and associated mechanical controls to an existing gradient ratio device. This component should allow the performance of filtration tests under energy and frequency control vibration;  •  Develop, design and commission a new cyclic gradient ratio device, and associated computerized control system, to performfiltrationtests under a head-controlled cyclic flow regime;  •  Taking into account the existing test standards, conduct new test procedures for vibration loading in the vibration device and cyclic flow loading in the cyclic flow device;  •  Extend methods of interpretation based on the measurements taken from both static and dynamic conditions;  •  Comprehensively investigate thefiltration-relatedphenomena under both static and dynamic conditions from the instrumentation on the test devices and related theoretical considerations;  2  •  Examine existing filtration criteria for soil-geotextile compatibility and independently validate new ones;  •  Compare and contrast the results of this work with design practice. It should be noted that the primary objectives of this research were related to the  development and commissioning of the cyclic flow device, and the analysis of the results using it. To achieve these objectives, the new device and procedures were used to perform head control filtration tests with different soil/geotextile combinations under static, vibration and cyclic flow disturbances.  1.4 Thesis Organization The state of the art, design practice and further research needed are reviewed in Chapter 2 . Chapter 3 describes the design and fabrications of the vibration and cyclic flow GR devices. Properties of materials used and the test program are reported in Chapter 4 . Chapter 5 describes the new experimental procedures. Test results are reported in Chapter 6. Chapter 7 includes analysis and discussion of the results. Chapter 8 summarizes the original contributions of this work, and in addition includes the conclusions and recommendations for further study.  3  2. LITERATURE REVIEW 2.1 The Basic Principles of Filters and the Related Conditions Filters in civil engineering applications are used to assure retention of upstream soil particles, without creating a barrier to free flow of water. The use offiltersinvolves aspects of applications, filter properties, soil properties, flowing fluid, hydraulic conditions, stresses, quality insurance and quality control. Thefirststep in designing a geotextilefilteris to identify the conditions under which it will be required to perform (Christopher, 1998; Holtz et al., 1997). Carroll, (1983) developed this concept using the following indices to distinguish between a critical and non-critical condition: 1. ) The influence of drain failure on either a decrease in structure life or significant structural damage, 2. ) Clogging potential and the ability to observe evidence of drain clogging in advance of failure, 3. ) Repair costs compared to installation costs for drain, 4. ) Level of hydraulic gradients through the soil  This concept was extended later by Holtz, et al., (1995, 1997). They suggested that the critical nature of the project should help determine the level of design effort, and the physical conditions should establish the geotextile requirements. The physical conditions are imposed by a hydraulic regime and stress which together dictate the potential for the soil to move. Table 2.1 provides a review of the severity of these conditions.  4  Table 2.1: The severity of the physical conditions (after Holtz et al., 1997) Less severe Severe Well-graded or uniform Gap-graded, pipable, or dispersable High Low Hydraulic gradient Dynamic, cyclic, or Steady flow Flow conditions pulsating 'Fine grained natural soils that deflocculate in the presence of water and therefore are highly susceptible to erosion and piping (Sherard et al., 1972). Item Soil to be drained  8  The physical conditions imposed in this work were generically divided into static and dynamic where: •  Static conditions are characterized by steady state flow without any external disturbance. Examples for potential development of such conditions can be in earth dams; in foundations of all types of hydraulic structures ponding water; in vertical drains; in trench drains adjacent to roads and parking lots; and behind retaining walls.  •  Dynamic conditions are induced by hydraulic or mechanical disturbances. Hydraulic i  disturbance can be imposed due to wave action under revetments or other applications for coastal and canals. Mechanical disturbance can be imposed during earthquake or can be present in railways and in pavement edge drains due to traffic loading and construction works.  The review presented here begins by addressing the state of the art in evaluation of the ability of soils to withstand migration of their loose fraction. The review continues by presenting some physical characteristics of geotextiles that are relevant to design. This is followed by details related to soil/geotextile interaction characteristics, and concludes with a  5  description of some state of the art and state of practice equipment that has been used to investigate soil/geotextile compatibility.  2 . 2 Internal Stability of Noncohesive Soils Noncohesive soils can be categorized based on their internal stability: in internally stable soils, the finer particles are entrapped in the soil fabric; and in internally unstable soils, the finer particles can migrate within the soil fabric. Investigations of the finer particle motion in porous media have been carried out since the 1930's, (Wolski, 1987). Kenney and Lau (1985, 1986) proposed an elegant graphical procedure to evaluate the instability of compacted granular material. This procedure was verified later by Skempton and Brogan (1994) and can be represented in the following way, (Figure 2.1): •  For different points on the gradation curve until F=20% finer (where F is the percent finer) in case of granular widely graded soils, and F=30% finer in case of granular poorly graded soils, find a grain size D.  •  If the location of some of the (4D,2F) points is below the gradation curve or in other words if H>F, where H is the mass fraction between D and 4D, then the gradation is internally stable, otherwise it is internally unstable.  For broadly graded cohesionless soils Lafleur et. al., (1989) used three classes of gradation profiles (Figure 2.2): •  Linearly graded: that can be categorized as internally stable and includes soils with all particles uniformly distributed, or soils with very few coarse particles (<40% according to Lafleur et. al., 1993) distributed in a matrix of fine particles;  6  •  Gap graded: which were described by Kenney and Lau, (1985) as possessing a "deficiency in a number of particles of a certain size range". Honjo et al., (1996) termed gap ratio as the ratio between the particle sizes at the beginning and the end of the gap. Honjo et a l , observed that base soils having a gap ratio less than 2.8 are stable and as the gap ratio increases to 4, the migration of fines through the pores of the coarser skeleton become significant. However, the conclusions of Honjo et al. are based on  settlement  considerations rather than just the mobility of the finer fraction. •  Concave upward: that were described by Kenney and Lau, (1985), as "soils that have a gentle inclined section in the lower part of their grading curve," and in cases where CTJ 20 >  showed (Kenney and Lau 1985; Lafluer et al. 1989) to be internally unstable. 2.3 Factors Affecting the Internal Stability of Noncohesive Soils Internal stability of granular soils is affected by many factors including particle shape, density, confining stress, mechanical loading, and hydraulic loading. These factors will now be discussed in more detail.  2.3.1 Particle Shape and Size By testing crushed angular quartz versus sub-rounded sands, Bertram (1940) suggested that the grain shape was not a factor in potential for erosion. Kenney at al., (1985) stated that as long as the materials are composed of particles coarser than silt size coupled with large seepage velocity and mild vibration, the absolute sizes of the particles are of little importance in comparison with the shape of the grading curve.  7  2.3.2 Density Density can have a significant effect on the stability of the base soil. Sherard and Dunningan, (1989) showed that for well graded soils, especially for silty soils, a decrease in density causes an increase in the movement of particles. As uniform gradations are stable it is reasonable that they are less sensitive to the density. Using uniform clean sand, Bertram (1940) showed that the density had little influence on the filtration behavior of the base soil.  2.3.3 Confining Stress One of the mechanisms that affect the internal stability of the base soil is friction between particles. The friction depends on the magnitude of the compressive stresses. If the compressive stresses become equal to zero, the base soil and the filter (if it is cohesionless) are in critical conditions and particles may become free to move. These conditions can occur in the following cases (Giroud, 1996): cyclic flow (de Graauw, 1984); earthquake; and repeated traffic loads (e.g. as in railroad tracks) which cause "pumping" of the soil. Although the confining pressure is important to stability of granular soils, high confining pressure can cause extrusion of fine grained soils into the filter. This can reduce the permeability of the filter (Luettich 1993).  2.3.4 Mechanical Loading Kenney et. al., (1985), Kenney and Lau, (1985, 1986), and Lafleur et. al., (1989), performed their tests applying light vibration by manually tapping the samples. These methods were considered to induce conditions more severe than would generally be expected in practice. Kenney and Lau (1985) reported that the "Vibration had a profound influence on the behavior of some of the tested materials for which even a mild vibration increased the loss 8  of fines". Kenney and Lau were motivated by being conservative as compared to the standard unidirectional flow. Therefore they do not apply their conclusions to conditions in which the filter is performing under vibration.  2.3.5 Hydraulic Loading •  Unidirectional flow: Bertram (1940) suggested thatfiltrationtests under upward and downward flow should behave similarly.  Lafleur, (1984), however, suggested that  because of viscous drag of water acting in the same direction as gravity, downward flow is worst for migration of particles. •  Cyclic flow: de Graauw et al., (1984) showed that the minimum amplitude of the hydraulic gradient to cause movement of particles under cyclic flow conditions are substantially lower than that needed under steady flow conditions and suggested that the soil arches (bridges) are less stable under cyclic flow conditions. Table 2.2 gives typical hydraulic gradients for some phenomena and applications  Table 2.2: Typical hydraulic gradients (after Giroud, 1996) Drainage application Typical hydraulic gradient 1.0 Standard dewatering trench 1.5 Vertical wall drain 1 Pavement edge drain 1.5 Landfill leachate collection/detection removal system 1.5 Landfill leachate collection removal system Landfill closure surface water collection removal system 1.5 Dam toe drain 2 3 to >10 Dam clay cores 1 Inland channel protection 10 Shoreline protection >10 Liquid impoundment with clay liners Critical application may require designing for higher gradients than those given  a  9  •  Magnitude: Bertram (1940) justified using large gradients as a compensating factor for the short time scale of the laboratory experiments relative to a life over which a filter is expected to perform. However, Bertram did not observe any difference in the stability using gradients as high as 18-20 as opposed to 6-8. Sherard and Dunnigan (1986) observed that 20 times more sand migrated into the filter at a gradient of 45 than at a gradient of 3. Fischer and Holtz (1996) stated that a dynamic application of gradient, such as in pulsating flow, would collapse any soil bridge that was developing after initial stabilization. Tomlinson (1997) concluded that imposing the gradient rapidly prevents a proper formation of the filtration structure.  2.4 Geotextiles As noted by Christopher and Holtz, (1989) compare to natural filters geotextile filters are cost effective. The first recorded use of synthetic fabrics as geotextiles was in Florida (USA) in the late 195 0's. This application involved placement of a woven geotextile beneath coastal erosion controlrip-raprevetment in place of graded granular filters (Christopher and Valero, 1999). Geotextile filters are usually woven or nonwoven fabrics.  Woven geotextiles are  manufactured generally by interlacing two perpendicular sets of monofilaments, yarns or tapes. Nonwoven geotextiles are manufactured byfirstforming a loose "web", (i.e. a layer of short fibers or continuousfilaments),arranged in an oriented or random pattern, and then subjecting this to some form of bonding in order to achieve a cohesive planar structure. Nonwoven geotextiles can be divided based on the bonding method. For example in this work, needle punched nonwoven geotextiles were used. In this method of bonding  10  thousands of barbed needles, are punched through the formed web. As the needle travels through the web thickness, it carries with it fibers which are caught on the barbs. The needle is then withdrawn, leaving thefibersphysically entangled within the flexible structure.  2.4.1 Properties of Geotextiles The geotextiles' functional properties are determined by index or performance tests. Index tests are used to measure properties such as pore size openings, cross plane permeability, thickness, mass per unit area, and strength. Performance tests such as the Gradient Ratio (GR) test are used to model thefieldperformance. The major index properties of the geotextiles, which are relevant to cross plane filtration applications are: pore size openings; coefficient of cross plane permeability (or permittivity); thickness; mass per unit area; and tear strength. •  Thickness: is measured as a distance between the upper and the lower surface of the fabric at a specific pressure (ASTMD5199).  •  Mass per unit area: is determined by cutting specimens to known minimum dimensions and weighing them (ASTM D5261 or CGSB-148.1 No. 2-M85).  •  Permittivity (crossplane permeability): is defined as the volumetric flow rate of water per unit cross section area, per unit head, under laminar flow conditions, in the normal direction (ASTM D4491-85). The permittivity \y is related to the coefficient of permeability (k) and the geotextile thickness (t) by:  \|/=k/t  (2.1)  11  •  Porosity: is defined as the ratio of void volume to total volume. It is usually calculated from other properties of the geotextile  •  Tensile Strength (Grab tensile): is determined by stretching the geotextile in tension until failure occurs (ASTM-D4632).  •  Tear Tests (Trapezoidal tear, Grab tear): is the force required to break individual yarns of the fabric (ASTM-D4533).  2.4.2 The Geotextile Opening Size Most design criteria for geotextile filters, especially for soil retention are based on relationships that were developed between the near largest pore size of the geotextile, and some indicative grain size of the soil. Usually the near largest pore size of the geotextile is determined by sieving. There are three basic sieving methods: •  Dry sieving (US standard): shaking glass beads or natural soil of progressive larger size (fractions) on the top of the geotextile until the bead size is found, where 95% by mass of the beads pass through the geotextile. This bead size is termed apparent opening size (AOS).  •  Wet sieving (Swiss and German standard): water is sprinkled on the geotextile and by doing so facilitate passage of a glass beads mixture through the geotextile. The pore size is usually designated as O90w.  •  Hydrodynamic sieving (Canadian, French and Belgium standard): the geotextile and a graded soil or beads mixture of a known weight are systematically immersed and retrieved from a water bath. Upon completion, the gradation of the retained soil or bead is  12  deterrnined. The equivalent pore size is defined as Filtration Opening Size (FOS) and usually designated as 095.  Where the pore channel is constricted at its minimum opening the term "pore channel constriction" is defined (Kenney et al., 1985). The pore sizes obtained from sieving methods depend on the pore channel constrictions and the probability of various particles to reach and pass a specific channel. Fischer et al., (1996) explained that the chances of obtaining a close size to the larger pore constriction is higher than the chance of obtaining a close size to smaller constrictions. Bhatia et al., (1994) performed a comparison between the close to the largest geotextiles opening size (larger than 95% of all the openings - 095) obtained from different sieving methods. Bhatia et al., found that wet and hydrodynamic sieving techniques produce the same results. For geotextiles with AOS less than 0.1 mm a linear relationship was observed between the dry, wet, and hydrodynamic sieving results. For geotextiles with AOS larger than 0.1 mm, 095 obtained from hydrodynamic and wet sieving were observed to be 60-75% of 095 obtained from dry sieving.  2.5 The Principles of Geotextile Filter Performance The use of geotextiles infiltrationapplications requires selection of a fabric that will provide retention of the upstream soil particles, without development of an unacceptable cross plan permeability (Christopher and Holtz, 1985). Based on these principles, design criteria were established. These criteria should fulfill the following demands: retention; permeability;  13  lower bound retention (anti-clogging,  anti-blocking, anti-blinding); survivability; and  durability. •  Retention criteria (upper bound retention): ensures that the geotextile openings are small enough to prevent migration of the soil particles through it. It is accepted that there will be an initial small migration of base soil fines through the geotextile filter (Lafleur, 1984; Lafleur et al, 1989). This initial migration allows the formation of a bridging network at the soil geotextile interface and a graded soil filter upstream of the geotextile filter (Figure 2.3)  •  Permeability criteria: ensures that the geotextile is permeable enough to allow liquid to pass through it without inducing an impediment to the flow.  •  Lower bound retention criteria (anti-blinding, anti-clogging and anti-blocking): ensures that the geotextile and its vicinity will adequately meet the permeability throughout the life of the structure. As described by Giroud (1996), blinding occurs when soil particles form a thin layer or "cake" at the surface of the geotextile. Blocking occurs when soil particles obstruct the geotextile filter opening, and clogging occurs when soil particles get trapped within the geotextile. In this work, however, for the purpose of discussion the general term, "blinding" will be used to describe the combined influence of clogging, blocking and specifically blinding. The clogging mechanisms can be divided into two main categories: particulate clogging that formed as a result of suspended particles; and non particulate clogging that formed as a result of biological, chemical or biochemical processes (Rohde and Gribb, 1990;Rollin, 1996; Rowe, 1998; Mackey and Koerner, 1999).  •  Survivability criteria: ensures that the geotextile is strong enough to survive its installation. 14  •  Durability criteria: ensures that the geotextile is resistant enough to withstand adverse chemical and ultraviolet exposure for the design life of the project.  2.6 Factors Affecting the Performance of the Soil/Geotextile Combination The performance of the geotextile depends on the following factors: •  Geotextile itself, as a function of the geotextile physical properties and pore structure, which in turn depends on thefibertype, and the manufacture process (Bhatia and Smith, 1996).  •  Base soil permeability: that may restrict the amount of discharged water through the geotextile.  •  Base soil internal stability: that dictates the behavior of the soil/geotextile system. In some cases, like very uniform gradations, retention of all the base soil should be dictated. In other cases like gap graded soils a limited amount of soil migration through the geotextile should take place to allow a better performance (Fluet and Luettich, 1993, Giroud, 1996).  •  Particle shape: It was noted above, that compared to the gradation profile, the particle shape does not have a significant influence on the internal stability of the base soil. However, Bhatia and Huang (1995) noted that the irregular shape of the particles can increase their resistance to movement and therefore can yield a less conservative geotextile retention criteria.  •  Geotextile extension: The influence of tensile strain on the geotextile opening size was alluded to by Giroud (1980). Most researchers believe that the applied load conditions in the field do not affect the apparent opening size (Adel et al., 1996; Young and Ochola,  15  1999). However Fourie and Addis (1999) found that the opening size of a thicker woven geotextile decreased with increasing biaxial load, whereas the opposite occurred for a thinner woven geotextile, and therefore suggested that this effect should not be ignored in applications where there are in plane tensile stresses. Top-load: The importance of the confining stresses in filtration tests was pointed out by Giroud (1982). The confining stresses have the potential to influence the stability of the base soil alone (see above), the soil/geotextile interface stability (as will be explained later) and to modify the geotextile performance. Palmeira and Fannin (1998) found that the 095 for relatively thin geotextiles, with characteristically larger pore openings, is sensitive to the value of the imposed pressure. However, all geotextiles converged to a similar 095 at higher confining pressure and little significant dependence was observed for any of the geotextiles at confining pressures between 25 and 200 kPa. Mechanical disturbing force: that can be divided into two categories: pumping load and vibration. Pumping load: describes changes in the applied external load such as underneath highways and railways as a result of traffic (Bell et al., 1982, Hoare, 1982, Mcmorrow, 1990). The Canadian Foundation Engineering manual (CGS 1992) does not give any criteria for pumping loads. The CGS even emphasizes not to use its criteria for dynamic, pulsating and cyclic flow in the case of pulsating (pumping) loads such as in highways and railways where the "flows are small and the cyclic loads are large". Vibration: can occur as a result of traffic load, construction works or an earthquake. Vibration has the potential to disturb the internal stability of the base soil and/or of the soil geotextile interface.  No research was found in the literature, which specifically 16  investigated the influence of vibration. This has been recognized as an area for further research (Kenney and Lau, 1985). •  Hydraulic load: According to Heerten (1982), dynamic flow conditions are given by high turbulent, wave attack or pumping phenomenon. Heerten suggests that changes of flow direction under laminar flow regime can be categorized as static flow conditions. As illustrated in Figure 2.4 the flow direction can be perpendicular to the plane or parallel to the plane. This research is within the spectrum of flow perpendicular to the plane or in other words cross plane flow. The terminology shown in Table 2.3 is suggested here: If the absolute value of the external applied gradient is constant (for example i=| +10| =| -10| ) than the flow is either unidirectional (i=+10) or cyclic (i changes from i=+10 to i=-10 cyclically). If the absolute value of the external applied gradient is not constant than the flow is termed pulsating unidirectional or pulsating cyclic. If steady conditions can be developed before the next pulse or half a cycle (like tides) than the flow regime is termed steady dynamic. If steady flow can not be developed at all (like highfrequencywaves) than the flow regime is termed unsteady dynamic. Some different types of hydraulic loads and typical cyclic flow periods are summarized in Table 2.4.  Table 2.3: Possible hydraulic loads based on the external applied conditions Flow direction One direction two direction 'Dynamic conditions.  Constant absolute gradient Unidirectional Cyclic 8  17  Variable absolute gradient Pulsating unidirectional Pulsating cyclic  3  3  T a b l e 2 . 4 : C y c l i c l o a d a n d related p e r i o d (after M o u w et. a l . , 1 9 8 6 )  Phenomenon  Storm surges; Tidal waves. Seiches. Translation waves; Swell. Wind waves. Ship waves Turbulence Dynamic impact  Remark  F r e q u e n c y (Hz)  10 -6 A  to be considered stationary for geotextiles  10 -3 A  Transition area 0.1 0.5  Cyclic  100  Dynamic flow conditions have the potential to disturb the internal stability of the base soil and/or the soil geotextile interface. However, under unidirectional flow conditions small changes in the applied gradient are not expected to influence the filtration behavior (Fannin et al., 1994a, Chin et al., 1994). Bhatia and Huang (1995) found that one directional pulsated hydraulic load has just a short term influence that is more apparent in the behavior of coarser and internally unstable soils than finer and internally stable soils. Cyclic flow regime can displace the geotextile locally and consequently decrease the potential of the geotextile to retain the base soil. As illustrated in Figure 2.5, the local displacement can occur due to flapping (free movement of the fabric - Mouw et. al., 1986; Kohler, 1993; Cazzuffi, et al., 1996), or due to high gradient that uplifts the base soil particles (Wewerka, 1986; Cazzuffi, et al., 1996; Tondello, 1998). Tonldello (1998) suggested a relationship between eroded soil mass and hydraulic gradient or effective stresses. Tonldello observed that a stable soil geotextile interface could reach instability for both an increase in the gradient and a reduction of stress. •  Blocking, clogging, survivability and durability: see section 2.5.  18  2.7 Design Criteria for Geotextile Filters Under Unidirectional and Cyclic Flow Conditions Using existing criteria in design practice, the designer is trying to address different factors that have the potential to influence the performance of the geotextile. These criteria address retention, permeability, lower bound retention, survivability, and durability. Some of these criteria are described here.  2.7.1 Soil Retention Defining soil retention is somewhat subjective. However, it is generally agreed that in widely graded soils minor losses of base soil at the filter interface are necessary to develop a state of equilibrium. Lafleur et al., (1989) reported that for broadly graded cohesionless soils, up to 0.25g/cm of the soil could be lost before the soil structure became internally unstable, 2  which in some instances, may be too large an amount of base soil migrating into a downstream drain (Lafleur, 1999). Bhatia and Huang (1995) found that the mass of the fine particles passing through the geotextilefilterswas relatively insignificant under ~0.3g/cm . Lawson 2  (1998) used the rate of soil passing through the geotextile as an index for piping suggesting that over time with continuous soil passing, loss of serviceability may arise and potentially lead to collapse. Common geotextile retention criteria that have been proposed for unidirectional flow using the Filtration Opening Size (FOS) values are summarized in Table 2.5. Some additional retention criteria using other characteristic opening sizes are summarized in Fischer et al. (1990).  19  Table 2.5: E x i s t i n g geotextile retention criteria Source FCGG, 1986  Criterion FOS/D85<0.381.25 FOS/D15>4  CGS (1992)  FOS/D85<1.5 F0S/D85O F0S/D85<1 and FOS>0.5D85 or FOS> 0.040 mm FOS/D85<1.5 FOS/D50<1.8 F0S/D85O.2 FOS/D50<2 FOS/D50r<2.5 FOS/D15<4 FOS/DK1  OMT (1992)  UBC (Fannin et al., 1994a)  Lafleur (1998, 1999)  1<FOS/D30<5  Dependent on soil type, compaction, hydraulic and application conditions For soils from whichfinescan easily be put in suspension Uniform soils Broadly jgraded soils  1<CTJ<2  3<CTJ<7, and where D50i is the mean particle size of the soil fraction smaller than the FOS of the geotextile For internally stable soils, where the indicative grain size diameter (Di) is a function of CTJ and the gradation shape profile For internally unstable soils  A significant body of test data exists to describe soil/geotextile interaction for unidirectional (steady state) flow, and validate design criteria for soil retention. In contrast, few data exist to describe behavior under reversing flow. However to satisfy the demands of design practice, several different criteria have been suggested. A comparative review of these criteria is given in Table 2.6. Although most of these criteria are proposed for use in design practice, they have not been explicitly validated from laboratory studies. In this research work some of these criteria (FCGG, 1986; CGS, 19921; U.S. FHWA, 1995; Mlynarek et al., 1999) are compared to the new laboratory data produced here.  20  Table 2.6: Filtration criteria that include cyclic flow conditions Source  Schober and Teindl, 1979.  Lawson, 1982.  Heerten & Wittmann, 1985  Criterion  The grain diameter that must be retained is set equal to the opening size. This diameter is dependent on the intensity of flow and the allowable amount of soil loss. D50>O90>D15  Characteristic pore size  Remarks  Dry sieving with sand (acc. to Orgink, 1975).  For nonwoven. For sand with 0.01 mm<D50<0.3 mm and  Dry sieving with glass beads.  Have been used in design and limited number of erosion control structures and appears to validate the criterion Based on field investigations. In the case of dispersive soil it may be necessary to employ a more stringent design, (Heerten and Wittmamm, (1985). In case of silt it can be very hard to meet. This criteria formed the base to Luettich et al., 1992, criteria, which in turn is the source to Koerner 1998 criteria.  Cyclic Loading: Wet sieving D50>O90w* with sand "This Criterion is based on Heerten (1982) criterion for dynamic flow conditions. Static load conditions: 5^ Cu O90<10D50 and O90SD90 CTJ<5 O90<2.5D50 and O90<D90  Fine grained cohesive material: 10D50>O90w and D90>O90w and 0.1 mm>O90w FCGG, 1986.  Ingold, 1985  Christopher and Holtz, 1985. PIANC 1990.  Hydrodynamic CD85S095 C= a function of grain size sieving distribution, soil density, hydraulic flow, and geotextile function. O90w/D50<l Very similar to the wet sieving technique using sand, developed by Heerten, (1982). <50% passing #200 D15S095 (if Dry sieving with glass the soil can move beneath the beads fabric) or 0.5D85a£)50 >50% passing #200 0.5D85>O50 0.7D90>O90>0.05 mm for 5>CTJ Suspected Wet sieving. D90>O90>0.05 mm for CTJ>5  CGS 1992.  < 50% passing #200 Hydrodynamic 095(FOS)<D15 sieving >50% passing #200 the lesser of the following criteria should be taken O95(FOS)<0.5D15 O95(FOS)<0.3 mm. 21  1.5<CTJ<5  Semi-empirical  For dynamic, pulsating and cyclic flow. Proposed as a general guideline for the design of flexible revetments should not be used for pulsing loads such as in highways and railways where the flow are small and cyclic loads are large  Source  Criterion  Klein Breteler, 1994.  U.S. D.O.T FHWA - 1995. Mlynarek et al., 1999.  Characteristic pore size  Published preliminary design criteria for geometrically open geotextile constructions on sand and cohesive soils for geotextiles with 0.1 mm<O90<0.3 mm  Remarks  reported results shows scattering (Berensen & Smith, 1996). Not clear whether this criteria can be used for cross section cyclic flow  dry sieving with A revised Christopher and glass beads Holtz, (1985) criterion (see also Holtz et al. 1997). 095<A*Di or 095<B mm and Hydrodynamic Semi - empirical 095>C*Di or 095>D mm sieving A, B, C and D are determined according to the nature of the retained soil and the severity of the hydraulic conditions 0.5D85^O95  2.7.2 Permeability Many research works and design standards address the geotextile permeability under static conditions (Schober and Teindl, 1979; Carroll, 1983; FCGG, 1986; CGS, 1992; Lafleur, et al., 1993; Giroud, 1996; Holtz, et al., 1997 etc.). However, not many research works and design standards address permeability criteria specifically under dynamic conditions. Three permeability criteria that can be related to cyclicflowconditions are described in Table 2.7 : Table 2.7: Permeability criteria that can be relevant to cyclic flow conditions Source PIANC, 1987.  Criteria  k <J3k B - for NW geotextiles = 50 B - for woven fabric = within the range of 1 to 10000. g  Giroud, 1996.  8  s  k>10ik i - the imposed gradient s  Holtz et al. 1997; U.S. D.O.T FHWA - 1995.  k >10k g  s  'kg, k - geotextile and soil permeability respectively s  22  Remarks Was adopted from Heerten (1982). The B value increases as the D10 of the base soil decreases. Not addressing explicitly dynamic (cyclic) flow conditions. For critical applications and severe conditions  2.7.3 Clogging Resistance Carroll (1983) showed that fine grained soils that satisfy the geotextile permeability criteria could cause clogging of the geotextile. A summary of existing criteria for static flow conditions have been presented by Fischer et al. (1990), and by Christopher and Fischer (1992). The existing clogging criteria for geotextiles are based on either (Smith et al., 1999): relationships between the pore opening of the geotextile and the grain size of the surrounding soil; the porosity or percent open area of the geotextile; or the soil/geotextile performance tests. For cyclic flow conditions no clogging resistance criteria has been suggested yet. 2.8 Selected Previous Laboratory Studies Wewerka (1986), Bezuyen et al., (1987) Breteler and Verheij (1990); Breteler et al., (1994) and Berendsen and Smith (1996) ran state of the art performance tests to model flow conditions in dikes, banks or bed protection. However, the induced hydraulic conditions in their tests did not have a pure cross plane component and therefore can not specifically reflect the influence of the cross plane gradient. This research is intended to run a filtration performance investigation under cross plane flow. Many devices have been used to test the soil/geotextile compatibility in cross plane filtration applications. As shown below, some of these devices are based on applying static conditions like unidirectional flow, and some are based on applying dynamic conditions like pulsating unidirectional, or cyclic flow.  23  2.8.1 Static Conditions (The Gradient Ratio Device - a State of Practice) Many devices have been used to study the soil/geotextile compatibility under static conditions including the Long Term Flow Test (Koerner and Ko, 1982; Siva and Bhatia, 1993), Hydraulic Conductivity Ratio test (Williams and Abouzakhm 1989) and Fine Fraction Filtration test (Hoover, 1982; Legge, 1990; Sansone and Koerner, 1992). However the most common performance test to model the soil geotextile behavior is the Gradient Ratio (GR) test. In this test method a rigid wall permeameter accommodates a cylindrical soil sample and geotextile specimen (Figure 2.6) that seats on a wire mesh. Water passes through the system by applying various total differential heads. Measurements of hydraulic heads are taken at several locations on the apparatus and used to establish the variations of hydraulic gradients along the soil and the geotextile. Flow rate through the system is determined and can be used to calculate permeability values. The Gradient Ratio (GR) is defined as the ratio of hydraulic gradient in the soil geotextile composite (i ) to that of the soil (i ). sg  s  GR(ASTM)=i /i sg  s  (2.2)  Nominally, in order to calculate the GR(ASTM), the head loss in the soil/geotextile interface is measured across the layer of geotextile and 25mm of soil upstream the geotextile (between ports 7 and 5, see Figure 2.6), and the head loss across the soil is measured across  24  the 50mm layer of soil upstream to the soil/geotextile interface layer (between ports 5 and 3, see Figure 2.6). Under laminar and steady flow the permeability of the soil (k ) is related to the s  permeability of the soil-geotextile interface (k ) by: sg  GR=i /i =k /k sg  s  s  sg  (2.3)  Scott (1980) used a modified gradient ratio device. This device accommodated a longer sample (250mm) and installation of manometers every 25 mm. Scott (1980) observed three distinct zones within the sample (see Figure 2.7) •  Top blinding zone: considered as an experimental problem. The top blinding zone caused larger head loss along the top of the sample. This phenomenon was also reported later by Chin et al., (1994) and Fannin et al., (1994b).  •  Undisturbed zone: Where the permeability remained unchanged.  •  Filter zone: In which movement of soil particles occurred. In this zone depending on the ability of thefinesto pass through the geotextile the permeability could either decrease or increase.  A linear distribution of water heads gives a value of GR=1 and suggests that the geotextile has not influenced the flow through the system. A value of GR>1 implies that some impediment to flow exists near the geotextile, and furthermore a value GR(ASTM)>3 is interpreted as a poor compatibility of the two materials (U.S. Army Corps of Engineers, 1977; 25  Haliburton and Wood, 1982). In contrast a value of GR<1 implies, that some soil particles adjacent to the geotextile have migrated through the geotextile. To increase the sensitivity of i  s g  to the changes in the vicinity of the geotextile Fannin et  al. (1994a,b;1996) used an additional port closer to the geotextile (port 6, see Figure 2.6). Based on this port that is located 8mm above the geotextile the modified GR is defined as:  GR(Mod.)=i(67)/i(35)  (2.4)  Where i(67) is the gradient between port 6 and 7 and i(35) is the gradient between ports 3 and 5. Austin et al. (1997) has further decreased the distance between the geotextile and port 6 by 2mm.  Advantages of the GR device: 1.) Water head distribution can be measured; 2.) Field behavior under a flexible gradient control can be modeled. Disadvantages of the GR device: No means to control the stress conditions are provided.  2.8.2 Dynamic Conditions (State of the Art)  2.8.2.1 Railroads and Highways (Pulsating Load) The stability of a railroad or a highway sub-base can be impaired if it becomes contaminated by clay size particles migrating vertically (upward) from a cohesive subgrade  26  (Bell, 1982). To prevent this contamination it is possible to use geotextiles as separators. Under such conditions the geotextile will have to serve as a cover to avoid punching through the low bearing capacity soil, and as a filter to the finer consolidating subgrade particles. Under the applied pumping loads upward and downward flows take place along the subgrade geotextile interface. Different laboratory research works have been carried out to model pumping loads acting on a sub-base of coarse aggregates that are separated from a fine subgrade soil by a geotextile. The principle of these tests is shown qualitatively in the Figure 2.8 (Bell et al., 1982, Hoare, 1982, Mcmorrow, 1990): Advantages: Simple set up Disadvantages:  1.) No control on the flow regime 2.) No accurate  measurements. 2.8.2.2 Sloping Bank Wewerka (1986) tried to simulate hydraulic loads on a nonwoven fabric filter covering sloping banks (Figure 2.9). The tested soil was placed in a square container with a depth of 5 cm. The geotextile was placed on top of the tested soil andfixedat the sides. To simulate the covering material, an additional load was applied in the form of steel screen. To simulate different bank slopes the inclination of the surface can be changed. The container was dipped into the water tank and withdrawnfromit at a velocity of 0.2m/sec to a depth of 30cm, at a predetermined frequency of 10 seconds interval. Disadvantages: 1.) Gives just general perspective of the phenomena, 2.) can not provide valuable data to quantify the behavior  27  2.8.2.3 Pulsating Flow Condition Using the device shown in Figure 2.10, Narejo and Koerner (1992) performed flow rate tests after conducting incremental periods of dynamic pulses. Between each pulsing cycle, selected increments of slurry are added upstream of the geotextile. Advantages: 1.) Simple set up, 2.) Can model situations where the geotextile is not in contact with the base soil Disadvantages: Can not model situations where the geotextile is in contact with the base soil. 2.8.2.4 Hydrodynamic Sieving (Cyclic Flow) Bhatia et al., (1996) performed a series of performance tests using the hydrodynamic sieving apparatus (see Figure 2.11). In this apparatus, four tests can be performed simultaneously. The immersion and retrieval cycles of the four cylinders were adjusted so that 10cm of waterfilledand drained during each cycle. Advantages: Simple set up Disadvantages: The hydraulic condition can not be quantified.  2.8.2.5 Cyclic Flow Control (Using Flow Pump) Cazzuffi., et al. (1996), Tondello (1998) and Cazzuffi., et al. (1999) reported on a development of a new test apparatus (Figure 2.12). This apparatus was designed in order to investigate the influence of cyclic flow infiltrationapplications and the related effect of some boundary conditions. Advantages: 1.) Field behavior can be modeled, 2.) Head distributions can be measured, 3.) stresses can be controlled. 28  Disadvantages: 1.) Can not provide head control flow regime 2.) The flow duration in one direction is limited to the dimension of the pump, 3.) Allows performing tests within a limited range of frequencies. 2.9 Research Needs and Objectives From the literature review it is apparent that geotextiles have been used as filters under dynamic conditions. These conditions include cyclic flow and vibration. The cyclic flow regime can be induced in revetments, dikes, bank protections or other coastal and canals applications. Vibration can be induced in railways and in pavement edge drains due to traffic loading and construction works. No research has been carried out to date to investigate the influence of vibration on the filtration performance. Under cyclic flow very little data was published to describe filtration related phenomena. The review of literature reveals that there is a need for a better understanding of the soil internal stability and the geotextile filter performance under dynamic conditions. The following specific research needs are: •  Develop experimental tools that allow performance of flexible investigations into the influence of dynamic conditions on thefiltrationperformance.  •  Taking into account the current test procedures for static conditions (ASTM 5101), develop compatible procedures for dynamic conditions.  •  Investigate the influence of different factors on thefiltrationperformance under dynamic conditions. These factors include: applied gradient or flow rate; confining pressure; soil gradation profile and cohesion; disturbance frequency; geotextile structure and others  29  •  Compare laboratory derived criteria or test data with criteria that were proposed and used for design.  Using performance tests, the challenge in this research is to investigate the influence of vibration and cyclic flow on different combinations of geotextile filters with stable and unstable soils. To achieve these objectives new experimental tools were developed and compatible test procedures are suggested. Based on the program of testing, theoretical considerations and new methods of interpretations, implications for design practice are discussed.  30  Finer Unstable  S  / H 2F  Jr.  4D  D  Grain size on a log scale, m m  Figure 2.1: A graphical procedure to evaluate the internal stability of granula materials, (Based Kenney, and Lau 1985, 1986)  100  log' d  too  Figure 2.2: The three classes of gradation profiles, (from Lafleur et al., 1989)  31  Well distributed particles  Filter formation  Bridging network Geotextile Aggregate  Figure 2.3:Idealized filter cake (after In gold 1994)  <  —  a  ::::::':  : : : i • •:: i: :  Base soil  1  b c  ::::::::::  Geotextile  c  This research interest  Figure 2.4: Flow direction: a.) Parallel to the plane (in plane), b.) Perpendicular to the plane, (cross plane).  32  a.)  b.)  Figure 2.5: The geotextile local displacement as a result of cyclic flow: a.) Flapping b.) Uplift (After Cazzuffi, 1996)  Figure 2.6: The GR permeameter test arrangement, (after Fannin et al., 1994b)  33  A  •Top blinding zone (lower permeability)  Soil Specimen  -Undisturbed zone (Same permeability)  -Filter zone (lower or higher permeability)  Geotextile sample  Figure 2.7: Three zones development infiltrationtesting  Dynamic Load  i  k  Load plate Sub-base aggregate <:  geotextile  Water level  Subgrade  Figure 2.8: A schematic principle of the tests that were carried out in order to simulate railroad or highway conditions  34  wolf  ttnlvstn  Figure 2.9: Sloping bank test equipment for non unidirectional flow study (after Wewerka, 1986)  pratsur* Gaug*  Slurry In  Row Raw H M W I t l M M  Wlra Mcslt Support  Figure 2.10: Schematic diagram of the pulsating flow filtration test set up (After Narejo and Koerner 1992)  35  Figure 2.11: Hydrodynamic sieving apparatus and testing cylinders setup (After Batia et al., 1996).  Figure 2.12:Cross plane cyclic flow control test device (After Cazzuffi., et al., 1996)  36  3. APPARATUS 3.1 Introduction An existing Gradient Ratio (GR) device (Fannin et al., 1994a) was modified to perform vibration tests and facilitate the collection of passing through particles at any stage of the test (section 3.2). The modifications include the addition of a vibration component and associated mechanical controls. In addition, a new cyclic GR device and associated computerized control system were developed and designed (section 3.3). Both devices can be used to evaluate the compatibility of different soil/geotextile combinations under static and dynamic conditions of vibration and cyclic flow. Valuable insight can be gained in any study of soil retention criteria by conducting a grain size analysis on particles that pass through the geotextile. A Sedigraph X-ray system was used for the grain size analysis. It is described shortly in section 3.4. 3.2 The Vibration GR Device A photograph of the apparatus is given in Figure 3.1. A schematic figure of the apparatus and its components are illustrated in Figure 3.2. Generally the apparatus comprises: a rigid cell wall permeameter (w) that contains the soil (s) and the geotextile (g); a collector trough (c) for the passing through particles; a unidirectional hydraulic supply system (tanks i and o) to control the applied unidirectional gradient; and an automatic double acting drop hammer (h) that allows the performance of filtration tests under energy controlled and frequency controlled vibration.  37  3.2.1 Permeameter The transparent permeameter is made of a cell wall (w) that accommodates a soil sample (s) of diameter 102mm and length of approximately 125 mm. A gradation tape is glued on the permeameter to facilitate reading the sample height. Manometer ports (p) are used to monitor the water head distribution at seven locations.  The connection of the  manometer ports to the permeameter pipe is covered with a fine mesh to prevent soil particles from being discharged from the permeameter. The location of the ports along the permeameter is given in Table 3.1. Table 3.1: Vibration test device: port locations (mm) Location Port 88.3 2 75.0 3 50.0 4 25.0 5 8.0 6 "From the geotextile  8  Using flexible tubes, the ports are connected to the manometers. The manometers seat on a mounting board. Below each manometer, a valve is used for connection or disconnection between the manometer and a flexible tube. A water energy dissipater (e) is mounted below the inlet. Experience has shown that it prevents disturbance of the top surface when flow rates are high (Fannin et al., 1994a). 3.2.2 Collector Trough A collector trough (c) is located below the permeameter cell.  It comprises two  removable pans mounted on aframethat rotates into position between the cell base legs (1). This arrangement allows for one pan to catch any soil particles that pass through the  38  geotextile, while the other is removed. Consequently, the migration of the soil particles passing through the geotextile can be monitored periodically with time. 3.2.3 Hydraulic Supply System The head difference is imposed by two constant head tanks (Palmeira et al., 1996) termed the constant head inlet (i) and outlet (o) tanks. For a given sample length (L) the imposed hydraulic gradient is controlled by the vertical distance between the water level in the two tanks (H, constant head difference). Overflow water from the constant head outlet tank is routed to the reservoirfromwhich the water can be recycled using a peristaltic pump. This pump continuously recharges the constant head inlet tank with de-aired water.  3.2.4 Automatic Controlled Vibration System The vibration is transmitted (see Figure 3.3) using an American model 1062DVS1.50-2 automatic double acting drop hammer (h), with a total length of 139.7 mm, diameter of 28.4 mm stroke of 38.1 mm and bore of 27.0 mm. The hammer is located on the cell top cover plate (t) above the center of the device (see also Figure 3.1 and Figure 3.2). Two flexible 6.3 mm diameter tubes connect the hammer to Numeric model L23BA4520 solenoid valve. The valve transfers air pressure (400 kPa) alternately to the tubes at a constant frequency. Air pressure is applied and monitored using an air pressure Fairchild model 10 regulator with an output range of 13.7-1034.2kPa, and a March gauge model B270-1 with a range of 0-689.4kPa.  39  The frequency is applied using a Wavetek model F62A function generator that generates square wave signals to the UBC made switching box, that in turn amplifies and transmits the frequency switch signals to the solenoid valve. 3.3 The Cyclic GR Device The Cyclic UBC GR device (Figure 3.4) was developed, designed, and built as part of the current study. The apparatus is used to evaluate the soil geotextile compatibility under both static and/or cyclic flow regimes. Instrumentation and a computerized system are used to apply a cyclic flow regime under confining pressure while measuring the sample length, flow rate and the water pressure along the sample. The automatic data acquisition system is described in section 3.3.7. Several components comprise the apparatus (Figure 3.4 and 3.5): a rigid wall permeameter which contains the soil sample and the geotextile sample; collector trough that collects the downward passing through particles; vertical pressure system that allows the application and control of confining pressures; sample top and bottom boundaries that bound the sample with minimum interference; head control hydraulic supply system that allows application of cyclic flow under any desirable frequency; and the flow rate measuring system that allows derivation of the average flow rate. All of these components are described in detail below.  3.3.1 Permeameter The transparent permeameter was built from a smooth Plexiglas pipe with a wall thickness of 6.4 mm. The inside diameter is 101 mm according to the ASTM 5101-96, and the length is 160 mm. The sample length (L) can be visually read using a gradation tape that  40  was glued on the permeameter walls.  Manometer ports are made of a 6.35mm brass  Swagelock tube fitting with a 0.125mm Polyethelyn tube filter. As illustrated in Figure 3.6 the ports are located on the cell top cover plate (port 1), at four locations along the permeameter cell wall (ports 2, 3, 5 and 6, see Table 3.2) and on the upper collector trough (port 7). T a b l e 3 . 2 : C y c l i c flow t e s t d e v i c e : p o r t l o c a t i o n s ( m m )  Port Location 2 101.0 3 75.0 5 25.0 6 8.0 From the geotextile  81  a  Imperial Eastman nylon 6.4mm diameter tubing connects the ports to a pore water pressure measurement system. This system includes differential Setra wet/wet pressure transducers (PS2) with a range of+/- 6.9kPa and manometers (see Figure 3.7). The interface between the pore water measurement system and the ports is illustrated schematically in Figure 3.8. According to the manufacturer, the response time of the pressure transducers is 30 to 50 milliseconds and their accuracy after calibration in the laboratory was found to be ±0.5 mm of water.  In addition the transducers operate with a negligible flow requirement and  therefore are able to detect transient pore water pressures.  3.3.2 Collector Trough The collector trough was made of an upper part and a lower part. The purpose of the upper part (uc) is to seal the bottom of the reaction frame (see Figure 3.5) and to allow visual observation of the passing through particles. For these purposes it was made of a Plexiglas  41  funnel with an internal slope of 45° that directs the passing through particles into the lower collector trough (lc). A test initiation valve (v) is mounted on the upper collector trough (Figure 3.5) and connects the bottom of the device to the hydraulic system. The lower collector trough (Figure 3.9) is made of a flexible silicon tube with a diameter of 19.1 mm. A series of discrete samples can be captured at any time during sample preparation and testing by clamping across the flexible tube to seal it in individual sections. After testing, the tube is removed and discrete samples of the passing through particles can be removed from it section by section. 3.3.3 Vertical Pressure System Using air pressure, a constant value of vertical stress is applied to the top of the soil sample (Figure 3.5). The applied air pressure is monitored using an air pressure Fairchild model 30 regulator with a range of 0-69.0 kPa and a Marchal town low pressure diaphragm gauge with a range of 0-69.0 kPa. This pressure is transmitted to the soil sample through a loading piston and a top loading plate (see Figure 3.6). The clearance between the top loading plate and the permeameter walls is 0.1 mm. The vertical displacement of the loading piston can be measured using a gradation tape glued on the permeameter cell wall and/or a Trans-Tel DC-DC LVDT model 244-000. This displacement reflects the movement of the top loading plate and the changes in the length of the soil sample.  3.3.4 Top and Bottom Boundaries Geotextile samples are placed above and below the soil sample (see Figure 3.6). The geotextile below the sample (lower geotextile) seats on a coarse wire mesh, which in turn is placed on the bottom plate. The bottom plate is made of stainless steel and is approximately 8  42  mm thick. It is perforated with many holes of 2 mm diameter and a triangular spacing of 3 mm. The objective of the boundary between the loading plate and the soil sample is not to replicate or examinefiltrationcompatibility (see Figure 3.6), but to retain the soil sample during upward flow and to provide forfreepassage of water. Two geotextiles, each having FOS=0.103mm (CAN/CGSB-148.1 No. 10), were used as part of the loading plate (see Figure 3.6). Using six flat head screws and a wire mesh below these geotextiles is connected to the top loading plate. This is done to attach the geotextiles to the top loading plate and to pre-compress them. Following this attachment, in order to seal the clearance between the top loading plate and the permeameter walls, the geotextiles diameter should be 0.2 mm larger than the top loading plate. A plastic circular insert to the top loading plate is 6 mm thick, and perforated with many holes of 5 mm diameter and a triangle spacing of 12 mm. Water flow passes across large openings of the top loading plate, the perforated insert, the geotextiles and the wire mesh. The top of the permeameter is sealed with a hard anodized aluminum cover plate. Through this plate, the sample is connected to the constant inlet/outlet tank (i-o) and to port number 1. To further ensure saturation in the pore water pressure measurement system of port number 1, this port is extended through a 10mm length and a 3 mm diameter brass pipe into the permeameter.  3 . 3 . 5 Hydraulic Control System The hydraulic system was designed to model the behavior of a horizontal soil column where the geotextile is placed on a revetment face (see Figure 3.10). In this structure, the  43  revetment face is subjected to an alternating head wave action (+/-hj) while the opposite side (into the revetment) is subjected to a relatively stable hydraulic head (hi). The configuration of the new cyclic head-control system comprises three constant head tanks (Figure 3.5), termed the constant head inlet tank (i); the constant head inlet-outlet tank (i-o); and constant head outlet tank (o). The constant head inlet-outlet tank serves two functions: an inlet tank to the sample during downward flow; and an outlet tank to the sample during upward flow. The switching operator employs a Goyen three way Solenoid water service valve (si) controlled by the computer. A downward hydraulic gradient in the test sample has the inletoutlet tank routed through the cell top, while discharging from the cell base routed to the outlet tank. By switching the valve (si), an upward gradient is imposed by routing the inlet tank through the cell bottom and taking discharge flow from the cell top to the inlet-outlet tank. As illustrated in Figure 3.5 for a given soil sample length (L), the imposed hydraulic gradient is controlled by the equidistant head differences (FT) between the three tanks. Assuming no energy losses, the equidistant head difference (H) between the three tanks can be measured using port number 1 that is located above the sample and port number 7 that is located below the sample (see Figure 3.6 and 3.10).  3.3.6 W a t e r S u p p l y a n d F l o w R a t e M e a s u r e m e n t  System  As illustrated in Figure 3.5, a peristaltic pump recharges the inlet constant head tank (i) with de-aired water from a 10 liter capacity reservoir (r). Overflow water from the inlet  44  constant head tank is routed to the inlet outlet constant head tank (i-o) and overflow water from the inlet outlet constant head tank is routed back to the reservoir. The downward flow rate is determined by the discharging rate of the measuring tank. This is done in the following process: •  During downward flow the outlet constant head tank is discharged with water from the permeameter;  •  Overflow from the outlet constant head tank is routed to the flow rate measuring tank (m);  •  The discharge rate of the flow rate measuring tank is automatically recorded using a pressure transducer (PS1) located below this tank (see Figure 3.5). The flow rate measuring tank (m) dimensions are 41.9mm length; 44.5mm diameter; and  6.35mm thickness. A Solenoid valve (s2) which is located below the flow rate measuring tank and above the pressure transducer (see Figure 3.11) is used to empty periodically the flow rate measuring tank to the reservoir. 3.3.7 Data Acquisition and Control System The data acquisition system consists of a Metrabyte DAS-16 board, a 386-SX desktop computer, a signal conditioning unit, and a data acquisition program.  The DAS-16 is a  multifunction board with a 12 bit high speed A/D (Analog to Digital), digital counters, and digital I/O (Input/Output). The analog input can be configured for 16 signal ended channels or 8 double ended (differential) channels.  45  The pressure transducers use output of 4-20 mA signal which is converted to a voltage by passing through a resistor. This voltage is then amplified by a factor of 10 in the UBC signal conditioning box, and then transferred to the DAS-16 that converts the signals to digital values. The signal conditioner also supplies an excitation voltage to the LVDT, and conditions to two digital control signals from the DAS-16 board for switching the 110 volts to the solenoid valves. Data from the channels is collected at a frequency of 20 Hz, and an average of the last 10 readings is written to an output file. The open loop reading control program is shown schematically in Figure 3.12.  3.4 Sedigraph Particle Size Analyzer The Sedigraph 5100 X-ray system, manufactured by Micrometrics Instrument Corporation, consists of a particle size analyzer and a PC compatible computer.  The  Sedigraph 5100 determines particle size by measuring the gravity-induced travel rates of different particles in a liquid with known properties. The rate at which particles fall through a liquid is described by Stokes' Law. The largest particles fall fastest, while the smallest particles fall slowest, until all have settled and the liquid is clear. Vitton et al., (1997) concluded that the X-ray absorption instruments produce particle size distributions very close to the standard hydrometer method, with the exception of soils with high mica concentrations. Since different particles rarely exhibit a uniform shape, each particle size is reported as an "Equivalent Spherical Diameter ", the diameter of a sphere of the same material with the same gravitational speed. The sedimentation rate is measured using a finely collimated beam of low energy X-rays that passes through the sample cell to a detector. As the particles  46  in the cell absorb X-rays, only a percentage of the original X-ray beam reaches the detector. This raw data is used by the PC to determine the distribution of particle sizes in the cell.  47  48  c - Collector trough  1 - Cell base legs  e - Water energy dissipater  o - Constant head outlet tank  g - Geotextile sample (on a supported screen)  p - Manometer ports 1 to 7  h - Automatic double acting drop hammer  s - Soil sample  H - Constant head difference  t - Cell top cover plate  i - Constant head inlet tank  v - Test initiation valve  L - Sample length  w - Cell wall  Figure 3.2: The vibration GR device (schematic drawing)  49  Double acting drop hammer (h)  Alternately air pressure supply  Cell top cover plate (t)  (a.)  Air pressure regulator Gauge  Solenoid valve  Air pressure supply  Switching box • Power supply  ->Q>  Double acting drop hammer (h) p 3  Function generator  (b.) Figure 3.3: The automatic vibration system (schematic drawings): a.) Double acting drop-hammer (picture), b.) Supporting system  50  Figure 3.4: The cyclic GR device (picture) 51  Vertical force LVDT  H  H  V See Fig. 3.11  V See Fig. 3.6 V See Fig. 3.9  F i g u r e 3.5: Schematic view of the cyclic test device  52  b  Reaction bars  H  Constant head difference  i  Constant head inlet tank  i-o  Constant head inlet outlet tank  L  Soil sample length  lc  Lower collector trough  LVDT Linear variable differential transformer m  Flow rate measurement tank  o  Outlet constant head tank  PS1  Flow rate measurement pressure transducer  r  Reservoir  si  Flow regime solenoid valve  s2  Flow rate measurement solenoid valve  uc  Upper collector trough  v  Test initiation valve  w  Cell wall  F i g u r e 3.5: Schematic view of the cyclic test device  53  Loading piston Top  cover plate,  S2L  1  Rigid wall permeameter  1-0  The  loading plate  2 Perforated insert  II  \ i m  I  Upper collector trough  ^/ •'  Top  | j | 1111111111111111  loading plate \  11  2 layers of geotextile  \Wire mesh  Lower geotextile (  Bottom plate  Permeameter walls ^ 1  Geotextile above a wired mesh Bottom plate  ^ ring Reaction frame base  Figure 3.6:  * \  Cyclic GR  ^  device: location of ports and the confining boundaries  54  \  Figure 3.7: Pressure transducers/manometer interface  55  Ports on the permeameter  Whitey ball valves (a) PS2  Port  (b) Figure 3.8: Schematic view of the Ports connections: a.) The ports/pressure transducers/manometers  interface;  transducers-port connections  56  b.)  Distribution of the Pressure  Flow of Flushing through particles  Figure 3.9: The discrete samples collection in the lower collector trough (lc)  57  Figure 3.10: Modeling the reversingflowregime in the permeameter  58  Figure 3.11: The flow rate measuring assembly  59  Data acquisition Pressure transducers After statistical and preliminary calculations write on the screen and on output files.  Read LVDT  Control  Data file l datafile2  Control simultaneously the valves that switch the flow direction and empty the flow rate measuring tank.  a  1 1  Read  ^Inputfilethat help controlling the frequency of SI (see Figure 3.5) Input file that help controlling the frequency of S2 (see Figure 3.5)  Figure 3.12: Schematic view of the open loop reading control program.  60  4. MATERIAL PROPERTIES 4.1 Introduction Materials used in the program offiltrationtesting are model soils made of glass bead, and geotextiles. Based on laboratory testing and the manufacturers' technical literature, some material properties that are relevant to the present study are reported in the following sections. 4.2 Model Soils The small spherical glass beads used in the present study came from Rotair Industries Ltd. Some physical properties of the glass beads are reported in Table 4.1. Table 4.1: Physical properties of the glass beads (from the manufacturer's technical literature) Property Color Compressive strength Specific gravity Density  Average value Clear 248 Mpa 2.45-2.5 2.5 g/cm 3  Tests were performed on reconstituted saturated soil samples. The soil samples were made by mixing different fractions of glass beads. The original gradations of the samples were corrected based on the mass and gradation of the passing through particles during preparation (see Appendix B). The soil samples are referenced according to their gradation characteristics.  As illustrated in Figure 4.1, for narrow and wide gradations, the soil is  labeled according to the coefficient of uniformity (Cu) and the median diameter (D50). For gap graded gradations the soil is labeled using the gap location and the gap width ratio. As illustrated in Figure 4.2, the gap location is the percent finer that fits the plateau of the 61  gradation. The gap width ratio is the ratio between the diameter at the end of the gap (at the beginning of the coarser fraction) to the diameter at the beginning of the gap (at the end of the finer fraction). 4.2.1 Vibration Tests The soil samples include four narrowly graded (defined by 1<CTJ<6); two widely graded; and 6 gap graded model soils (see Figure 4.3). Two of the gap graded soils are "narrowly graded" and four are "widely graded". Tables 4.2 and 4.3 show some of the significant characteristics of the gradations that were used. Table 4.2: Description of narrow and wide gradations (vibration tests) Code  D85 (mm)  D50 (mm)  D30 (mm)  Cu  s(1.9-39) s(2.2-52) s(2.2-78)  0.054 0.101 0.121  0.039 0.052 0.078  0.032 0.039 0.063  1.9 2.2 2.2  s(4.3-97)  0.270  0.097  0.052  4.3  s(6.4-213)  0.461  0.213  0.097  6.4  s(6.4-306)  0.520  0.306  0.216  6.4  Description (BS - From Bardet 1997) Coarse silt Sandy silt Very silty sand Very silty sand Very silty sand Silty Sand  Table 4.3: Description of gradations that show a gap (vibration tests) Code  D85 (mm)  Did (mm)  Cu  Gap location [ %  s(15-2.7) s(15-4.5) s(20-1.5) s(20-3.1) s(45-3.1) s(70-1.6)  0.580 0.786 0.237 0.505 0.396 0.211  0.062 0.062 0.037 0.039 0.030 0.026  8.5 11.9 5.7 10.5 12.1 2.1  62  Gap width ratio  finer]  15 15 20 20 45 70  2.7 4.5 1.5 3.1 3.1 1.6  4.2.2 Cyclic Flow Tests The soil samples includes four narrowly graded; one widely graded; and four gap graded gradations (see Figure 4.4). One of the gap graded gradations is "narrowly graded" and three are "widely graded". Tables 4.4 and 4.5 are showing some of the significant characteristics of the gradations that were used. Table 4.4: Description of narrow and wide gradations (cyclic flow tests) Code  D85 (mm)  D50 (mm)  D30 (mm)  Cu  s(1.3-134) s(3.2-68) s(4.8-333) s(5.1-144) s(7.2-250)  0.174 0.152 0.576 0.321 0.564  0.134 0.068 0.333 0.144 0.250  0.119 0.042 0.252 0.075 0.132  1.3 3.2 4.8 5.1 7.2  Description (BS - from Bardet 1997) Fine sand Sandy silt Very silty sand Very silty sand Very silty sand  Table 4.5: Description of gap graded gradations (cyclic flow tests) Code  D85 (mm)  D10 (mm)  Cu  s(20-1.9) s(20-3.1) s(45-3.1) s(70-2.2)  0.282 0.404 0.398 0.254  0.037 0.039 0.030 0.026  6.8 9.8 12.1 2.1  Gap location [% finerl 20 20 45 70  Gap width  1.9 3.1 3.1 2.2  4.3 Geotextile Samples Four different non-woven needle-punched polypropylene geotextile samples were used.  The index tests that are relevant to this research were carried out by SAGEOS  (Quebec, Canada). The geotextiles manufacturers, their relevant index properties and related test methods are given in Tables 4.6 and 4.7.  63  Table 4.6: Manufacturers of the geotextiles used Manufacturer Amco Polyfelt Texel SAGEOS  Internal code gl03 gl22 gl49 g290  Table 4.7: Geotextile material properties Internal code  Thickness [mm]  8  Mass per unit area [g/m 2] 286 294 126 216  FOS (mm)  Permeability [cm/s]  Porosity [%]  0.103 0.122 0.149 0.290  0.64 0.59 0.20 0.99  90 92 85 82  c  b  A  a b  3.02 gl03 2.50 gl22 0.92 gl49 1.33 g290 CAN/CGSB-148.1 No. 3 - M85 CAN/CGSB-148.1 No. 2 - M85 GAN/CGSB-148.1 No. 10 - 94 ASTMD4491-85  C d  4.4 T e s t  P r o g r a m  The combinations of soil samples and geotextiles that were tested are shown in tables 4.8 and 4.9. As illustrated in Figure 4.5 reference names were given to the each combination based on the geotextile and the gradation tested. For example s(l.9-39). 103 is the reference name for test that includes soil sample s( 1.9-3 9) and geotextile gl03.  64  Table 4.8: Vibration test program Code s(1.9-39) s(2.2-52) s(2.2-78) s(4.3-97) s(6.4-213) s(6.4-306) s(15-2.7) s(15-4.5) s(20-1.5) s(20-3.1) s(45-3.1) s(70-1.6)  gl03  gl49  gl22  SS S S S S S S  S S S  G290  S  •  S  S S  S  S S  S s s s  s  Table 4.9: Cyclicflowtest program Code s(1.3-134) s(3.2-68) s(4.8-333) s(5.1-144) 8(7.2-250) s(20-1.9) s(20-3.1) s(45-3.1) s(70-2.2)  gl03  s  gl49  gl22  S s s s  g290  •  S S SS S  s s s s  65  S S  Grain size (mm)  Figure 4.1: Reference name for narrow and wide gradations  Grain size (mm)  Figure 4.2: Reference name for gap graded gradations  66  0.01  0.10 Grain size (mm)  1.00  0.01  0.10 Grain size (mm)  a.)  1.00  b.) >—  s(15-2.7) s(15-4.5)  0.01  0.10 Grain size (mm)  1.00  c) Figure 4.3: Vibration test program particles size distribution curves of : a) narrow gradations, b) wide gradations, and c) gap graded gradation.  67  0.01  0.10 Grain size (mm)  1.00  b.) Figure 4.4: Cyclic flow test program particles size distribution curves of: a.) narrow and wide gradation, b.) gap graded gradations.  68  — Geotextile  soil gradation  ure 4.5: Reference name for the combinations tested  69  geotextile.soil gradaion  5. TEST PROCEDURE 5.1 Introduction Dynamic conditions are induced by hydraulic or mechanical disturbances. Hydraulic disturbance occurs due to wave action under revetments or other applications for coastal works and canals. Mechanical disturbance occurs during earthquakes or can be present in railways and in pavement edge drains, due to traffic loading and construction works. The devices described in Chapter 3 were designed to allow the performance of filtration tests with either hydraulic or mechanical disturbance. In order to use these new features in an efficient and optimum way new multi-stage programs of testing were conducted. The process of performing dynamic filtration tests includes: preparation of the device, the geotextile samples and model soil sample; performing the test according to the test procedures; and post-test observations. This chapter begins with the preliminary preparation that is common to both devices (section 5.2), and then for each device it separately addresses further preparation, the test procedure and post-test activities (sections 5.3 and 5.4).  5.2 Preparation of the Geotextiles and Model Soil The preparation of uniform and saturated samples is important for a comparison between different tests. Geotextile sample: In order to test geotextile samples with similar opening sizes, a sample of a square geotextile (112mmx 112mm in the vibration test program, and 1 lOmmxl 10mm in the cyclic flow test program) was cut so that it was within 5% of the mass per unit area shown in Table 4.7. This sample was then further trimmed to a diameter of 111 mm for the vibration test and 109 mm for the cyclic flow tests, put in a small bath of de-aired  70  water and squeezed manually. This was done until the geotextile appeared completely wet and no bubbles were observed in the geotextile. The submerged geotextile was then left in the small bath of de-aired water for 24 hours. ASTM 5101 specifies soaking the geotextile in de-aired water and an additional overnight soaking of the entire system. However, using heatbonded and needlepunched nonwoven geotextiles, Fischer et. al., (1999) suggested that soaking the geotextile overnight is probably sufficient for saturation. Soil sample: A known mass (1800g) of glass beads (the model soil) was boiled with de-aired water in flasks for about 30 minutes to remove any entrapped air. It was then allowed to cool at room temperature (23°C-24°C).  5.3 Vibration Tests The objectives of the vibration tests were to investigate the influence of vibration on the internal stability of different soil gradations and therefore the soil/geotextile compatibility in filtration. Using Figure 3.2 the following sections describe the procedures that were carried out to achieve these objectives.  5.3.1 Preparation of the Test Apparatus Prior to each test, the permeameter sections were thoroughly cleaned and dried. Orings gaskets and the junctions between the permeameter and the collector trough were slightly lubricated. Ensuring that bubbles were not present in the system, half of the outlet tank (Figure 3.2) was filled with de aired water. Then the permeameter, the cell base legs section, and the collector trough, were seated separately in the outlet tank. observation permitted any bubbles to be removed from the submerged system.  71  Visual  5.3.2 Geotextile Placement The support screen (Figure 3.2) was placed on the top of the cell base legs section. After submerging the small bath of de-aired water within which the geotextile seats in the larger outlet constant head tank, the submerged geotextile was transferred to the top of the support screen. The permeameter was then slipped on the permeameter legs section and pressed down until the geotextile sample and the support screen were tightly secured. These sections were then clamped together.  5.3.3 Soil Placement and Test Initiation As described in chapter 3, the apparatus set up and instrumentation used in this research work are different from those used by ASTM D 5101. Therefore, as described below, the process of sample preparation that was performed in this work is different from the ASTM D 5101 method that specifies pumping CO2 through the soil and backfilling with water. Like the ASTM D 5101 method, the process of sample preparation used here allows preparation of saturated and uniform samples. This process is based on the slurry deposition technique (Kuerbis and Vaid, 1988) that has been successfully performed for filtration tests (Fannin et al., 1994a). Water from the outlet head tank (see Figure 3.2) was taken until the water level was about 2 cm above the geotextile. A spoon was used to transfer small portions of the saturated sample to the permeameter. The soil sample was reconstituted in layers of 2 cm. To achieve homogeneity, after every two layers the sample was mixed with a spoon. Using the slurry deposition technique the sample was built to a height of about 130 mm while filling the constant head outlet tank with de-aired water. The sample was then top leveled, by siphoning 72  off the soil at the top. The siphoning process is very common in reconstitution of soil samples. This technique (Shi, 1993) is based on sucking a thin layer at the top of the sample through a glass pipe by using a small gradient in a closed system. The length of the sample (L ) after leveling was maintained at approximately 125mm. The permeameter and the outlet constant head tank were filled with de-aired water. The particles that passed through the geotextile during preparation were transferred from the collector trough to an empty tin. The inlet constant head tank was connected to the top cover plate, and the top cover plate with the energy dissipater was attached to the permeameter. The test initiation valve was used to make sure that no bubbles were entrapped below. The test initiation valve was left partially opened and flexible tubes were used to connect the ports to the manometers. Then the test initiation valve was closed. A small amount of red dye was added to the water in the manometer tubes, to facilitate reading the transparent gradation tape. The test was then ready to begin by opening the test initiation valves and the manometer valves.  5.3.4 T e s t  P r o g r a m  A new multi-stage program of testing was conducted to investigate the behavior of different soil/geotextile combinations under vibration (see Figure 5.1). All stages were performed under an externally imposed gradient of four (H/L=4, Figure 3.1). This gradient was chosen to be larger than that expected in most field applications (see section 2.3.5). As described below the test was divided into static and dynamic stages: During the static stage a downward unidirectional flow was applied in order to examine the behavior of the sample under unidirectional flow and its homogeneity. The sample was left to run overnight. 73  The objective behind the dynamic stage was to examine the sample performance under unidirectional flow conditions while applying vibration. The degree of disturbance during vibration was chosen in an attempt to impose conditions above which the behavior will be insensitive to any further increase in the disturbance severity. During this level of disturbance, the particles of the primary soil fabric are not fully restrained and the instability of the constriction sizes (where the pore channel is constricted at its minimum opening) causes particle migration. The disturbance was applied with the automatic hammer (see section 3.2.4) at series of 180, 720, 1440, and 1440 blows (see Figure 5.1) and observations were taken arbitrary using these numbers. However, the first series of blows was chosen in an attempt to examine thefiltrationbehavior during a short period of vibration, for example, a 30 sec. long earthquake with 3 cycles per second. The rate of blows was applied at 6 blows per second (f=3 Hz). However, it is assumed that the filtration behavior is frequency independent. Nevertheless, as discussed later (section 5.3.5), the energy of blows (vibration amplitude) was chosen in an attempt to cause particle migration even under dense conditions. If continuous piping (will be discussed in section 7.4.4) was observed during the application of blows the test was ended. After each series of blows the sample was left to stabilize under unidirectional flow. Stabilization was characterized by no observable changes in the water head (hd) and flow rate (q) within a 90min period. After stabilization the next series of blows was applied. This looped process was continued until the last series of blows (N4) after which the test was ended.  5.3.5 On The Severity of Vibration (Vibration Amplitude) The energy of blows in this work is chosen in an attempt to impose conditions above which the behavior will be energy independent. At this level of energy, particles in a loose or 74  a dense state are not fully restrained and the instability of the constriction sizes (where the pore channel is constricted at its minimum opening) may cause an increase in particle migration. Two tests were performed using s(1.9-39).gl03 while changing the applied energy by a factor of three. It was observed that the increase in the applied energy did not cause any influence on the amount of the passing through particles (see Table 5.1 below). Furthermore no densification was observed (see Figures Al.lc and A1.2c). Table 5.1: Cumulative mass per unit area (mA, g/m ) of the passing through particles 2  Test  Applied  Preparation  Static stage  Code  Energy (Joules) 11 32  End 7283 7200  End 9679 9100  Dynamic stage (blows)  8  s(1.9-39).gl03i s(1.9-39).gl03  2  180 11728 11200  720 17543 17062  2880 34209 33924  see section 3.2.4  a  Based on these observations it was concluded that any further increase in the energy of blows would not affect the behavior. Furthermore to ensure that the tests conducted in this work were independent of energy level, the higher value of energy (32 J) was chosen for testing. 5.3.6 Control and Data Acquisition The externally imposed hydraulic gradient (H/L=4), the frequency (f=3 Hz) and energy of vibration (blows) were controlled. During each blow of the automatic double acting drop hammer 32 J were applied. The number of blows (Nb) was determined by knowing the applied frequency of vibration and measuring the time during which the vibration was applied. The mass of the passing through particles (m ) was collected after preparation and p  75  the stabilization periods. Other measurements taken during the stabilization periods of both static and dynamic stages are: •  Volumetric flow rate (Q) by measuring the overflow from the outlet tank with time;  •  Water head (hd) along the sample by reading the manometers;  •  Sample length (L) by visual readings of the gradation tape on the permeameter; and,  •  Visual observations of sample homogeneity through the transparent device.  5.3.7 Post-Test Procedure Immediately after the test, the test initiation valve was closed. All the tubes that were connected to the permeameter were removed and the outlet constant head tank was emptied. The cell top cover plate was removed.  The cell base legs section together with the  permeameter were then disconnected from the collector trough and the soil from the permeameter was taken out by inverting the permeameter with the collector trough. When back in the upright position the geotextile was released by disconnecting the cell base legs section from the permeameter. The passing through particles extracted from the collector trough were then taken to the Sedigraph X ray system for gradation analysis.  5.4 Cyclic Flow Tests The objectives of the cyclic flow tests were to investigate the influence of cyclic flow on the internal stability of different soil gradations and on the performance of the soil/geotextile combinations. This was conducted under two extreme conditions: a.) under vertical confining stresses that will ensure no boiling conditions; and, b.) under vertically unconfined conditions that will allow boiling of the soil sample during the upward flow (half a cycle).  76  The applied head difference (Ff), vertical pressure ( c ) , and reversing flow frequency v  (f), were controlled parameters during the tests. The results were interpreted from the measurements of flow rate (q), water head along the sample (ha), visual observations through the transparent device, and the gradations of the passing through particles that were collected during the test as discrete samples. 5.4.1 Preparation of The Test Apparatus Prior to each test, the apparatus was thoroughly cleaned, the ports were sealed with a swagelock fitting, and the O rings were lubricated and placed in position. After sealing the bottom of the lower collector trough with clamps, it was connected to the base of the reaction frame. Then the collector trough and the base of the reactionframewere filled with de-aired water.  Using the test initiation valve, the upper collector trough was connected to the  solenoid valve (si, see Figure 3.5). The perforated bottom plate was then placed in the base of the reactionframe,and the coarse wire mesh was placed on the perforated bottom plate. Independent of this process, the loading plate was boiled in water for about 30 minutes to remove any air entrapped and then was cooled to room temperature.  5.4.2 Placement of the Soil and Geotextile After making sure that the system is fully saturated up to and above the top of the wire mesh, the saturated and de-aired geotextile was placed on it. Therigidwall permeameter was then placed on top of the geotextile and secured with wing nuts. About 3/4 of the permeameter was filled with de-aired water, and again the geotextile was squeezed manually and suspended under these conditions for 24 hours.  77  In general the connection of the ports to the pressure transducers and manometer system was carried out in the following manner (see Figure 3.8a): •  At the beginning the manometers werefilledwith water up to 60 mm of their height, valves 1 and 3 were opened and valve 2 was closed.  •  The water level in the permeameter was above the port to be connected.  •  Partially opening valve 2 to allow a small amount of water to flow through the tube, the connection between the tube and the relevant port was made.  A slurry deposition technique was used to reconstitute the soil samples. To increase the homogeneity of the soil samples they were placed in several layers of 1 cm followed by a manual spoon stirring after each layer. The layer thickness was chosen to fit the process of the ports connection (as described below). The manual stirring caused an increase in the density of the samples. Even though the ASTM standard is using a conservative approach by calling for loose placement of soil in the permeameter, in most field applications the soil is densified during construction. Therefore thefiltrationperformance of densified samples in the laboratory will not be under conservative. During sample reconstitution the ports were connected to the manometer pressure transducers interface system. This process was carried out in the following way, (see Figure 3.8b): •  port 7 and 6 were connected;  •  after filling 2 cm of the sample port 5 was connected;  •  after filling 7 cm of the sample port 3 was connected;  •  after filling 10 cm of the sample port 2 was connected.  78  After filling another 1.5-2 cm of the sample, the reconstituted sample was leveled using a siphon (Fannin et al., 1994a) and the permeameter wasfilledwith de-aired water. The length of the sample after top leveling was maintained at about 11 cm. After submerging the bath that contain the de-aired loading plate in the upper part of the permeameter, the loading plate was transferred to cover the top of the soil sample. Then the permeameter was sealed with the top cover plate, which was connected to the constant head inlet outlet tank in advance (see Figure 3.6). The sample was then sealed and seated under the same head as the constant head inlet outlet tank and Port 1 was connected to the pressure tranducers/manometer interface. 5.4.3 Application of the Confining Stress and the Test Initiation The length of the sample was measured using the gradation tape that was glued on the permeameter wall (visual reading). To restrain the particles of the primary soil fabric during upward flow a vertical pressure (c%) of 25kPa was applied to the top of the sample (Figure 5.2) through the confining vertical pressure system (Figure 3.4).  Then to establish  connection between the manometers and the permeameter, all valves number 1 (see Figure 3.8) were opened on all the port connections, and the sample was left under the applied pressure for 2 hours to assure equilibrium. Since the samples were characterized as silty sand to fine sand, settlements were expected to take place immediately. After this period, valve 2 was closed on all the port connections. Using the manometers and the gradation tape on the permeameter, the pressure transducers and LVDT offsets were respectively entered into the computer. Then valve 1 was closed and valve 2 was opened again. After reading the sample length, the test was initiated by opening the test initiation valve and simultaneously executing the computer program. 79  5.4.4  C o n t r o l a n dD a t a  A c q u i s i t i o n  A new multiple stage program of testing was conducted to investigate the behavior of the soil geotextile combinations under different applied cyclic flow conditions that can be present in the field (see Figure 5.2). The input data was collected using the automatic data acquisition system that was described in Chapter 3. Furthermore as mentioned in Chapter 3 the vertical displacement of the sample can be measured using a gradation tape glued on the permeameter cell wall and/or LVDT. However, as the sample length during testing was insensitive to the imposed conditions (see section 6.3.1.2.2), its measurements were relied on the gradation tape. All stages were performed under an externally imposed hydraulic gradient (H/L)  of four. This gradient was controlled by the distance (H) between the constant head  tanks. In progressing from unidirectional flow to confined cyclic flow and unconfined cyclic flow the severity of the hydraulic disturbance increase. At the end of the test program or in some cases if failure was observed at an earlier stage, the test was ended. Failure was defined by observation of continuous piping through the collector trough (will be further discussed in Chapter 6). The test was ended by closing the test initiation valve and stopping the computer program. After 30 min.  of letting all the passing through particles settle down  in the collector trough, a discrete sample was taken by clamping above it. Then after setting the computer the next stage was initiated by opening the test initiation valve and executing the relevant computer program. The motive behind each stage (Figure 5.2) and its duration (when continuous piping was not observed) is described below.  80  Stage 1: downward unidirectional flow: this stage allowed study of sample behavior under unidirectional flow and assessment of the initial homogeneity. This stage was run until no changes in the water heads along the sample were observed within 90 minutes. Stage 2: confined cyclic flow at a frequency of 0.2 Hz: this stage was run to check the influence of cyclic flow under confined conditions. The applied frequency was chosen to simulate high frequency gravity waves (Shore protection manual, 1984). These waves are of primary concern in coastal engineering problems. This stage was run for 12 hours to allow accumulation of the passing through particles. Stage 3: confined cyclic flow at a frequency of 0.02 Hz: This stage was run to be compared to the previous stage when the frequency was 10 times higher. This stage was run for 15 cycles, a duration that was chosen to reflect on the water head distribution and on the flow rate. Stage 4: unconfined cyclic flow at a frequency of 0.2 Hz: this stage was run after releasing the vertical pressure to zero. This was confirmed by making sure that the cover plate could move freely. The objective of this stage that was to check the influence of cyclic flow under unconfined conditions. This stage was run for 20 hours. Stage 5: downward unidirectional flow: this stage was run to examine the sample homogeneity and behavior after the cyclic flow disturbances, that were imposed during previous stages. This stage was run until no changes in the water heads along the sample were observed within 90 minutes.  Data were automatically written to output files (Figure 3.12) at the following frequencies:  81  1. ) Channel 1 that was connected to the flow rate measuring tank: every 50sec. 2. ) Channel 2 to 6 that were connected to the manometer ports: a. ) during stages 1 and 5: every 100 sec; b. ) during stages 2 and 4: every 0.5 sees, during the first 40 cycles, and then every 200 cycles, at 0.5 sec. before the direction of the flow was switched upward; c. ) during stage 3: every 1 sec.  5.4.5 P o s t - T e s t  P r o c e d u r e  The procedure of ending any of the stages and ending the test was the same. After the test, valve 1 (see Figure 3.8) was opened on all the port connections. This was done to let water in the manometers reach the same level and then valve 2 was closed. The pressure transducers were then checked again to confirm that no changes in their calibration occurred during the test program. By simply opening the clamps from the bottom to the top, discrete samples from the lower collector trough were taken to different containers. After emptying the water from the permeameter through the lower collector trough, the vertical pressure system with the loading plate, the tubes that are connected to the ports, and the lower collector trough were removed. Releasing the bottom plate and the permeameter together, the soil sample is then emptied by inverting the permeameter together with the bottom plate. Once back in the upright position, the geotextile is released by disconnecting the bottom plate from the permeameter When enough material was collected in the collector trough (>250g/m =2g), discrete 2  samples of the passing through particles were taken to the X ray absorption system for gradation analysis. 82  5.4.6 On The Severity of Cyclic Flow Generally, the severity of the cyclic flow conditions applied in this work, increased as the test proceeded. •  i=4 and <%= 25kPa (stage 2 and 3): These loads were imposed not to induce quick conditions of the soil fabric but to cause quick conditions of the loose particles within the pores of the skeleton.  •  i=4 and o- =0 kPa (stage 4): These loads were imposed to induce quick conditions of the v  soil fabric, and therefore to influence the internal stability of the soil and the stability of the soil geotextile interface. 5.5 Summary Taking into account the existing test standards, new multi-stage programs of testing were conducted. The programs of testing were performed under an externally imposed gradient of four and were divided into static and dynamic stages. The results of each test were interpretedfromthe measurements of flow rate, water head distribution along the sample, visual observations through the transparent device, and the weight and gradations of the passing through particles.  5.5.1 Vibration In the static stage downward unidirectional flow was imposed. In the dynamic stage vibration disturbance was added by four series of blows (180, 720, 1440, and 1440 blows). The blows were applied at afrequencyof 3 Hz. Between each series of blows the sample was left to stabilize under unidirectional flow.  83  5.5.2 Cyclic Flow Similarly to the vibration test program, during the static stage a constant unidirectional gradient of four was externally imposed. In the dynamic stage disturbance was applied by cyclically changing the direction of flow under both confined and unconfined conditions. The confining stresses during confined conditions (a =25 kPa) were chosen to simulate v  conditions in which the primary soil fabric is restrained. Conversely the unconfined conditions (a =0 kPa) were chosen to simulate conditions during which the primary soil v  fabric is not fully restrained during upward flow. Under confined conditions cyclic flow frequencies of 0.02 Hz; and 0.2 Hz were imposed and under unconfined conditions cyclic flow with a frequency of 0.2 Hz was imposed. The applied frequency of 0.2 Hz was chosen to simulate high frequency gravity waves. These waves are of primary concern in coastal engineering problems. The applied frequency of 0.02 Hz was chosen to be compared to the 0.2 Hz.  84  Sample preparation  Static stage  Applying downward unidirectional flow  Stabilization*  Piping Yes  End  Dynamic stage  1  Series l : N i = l 80 blows Series 2: N2=720 blows Series 3: N3=1440 blows Series 4: N4=1440 blows End  Stabilization**  Piping ? No  Yes  >End  Apply next series of blows  * Overnight stabilization (t=10hrs) ** No continuous piping or changes in the water heads and in the flow rate was observed during t=90min.  Figure 5.1: The vibration test process  85  Test drocess bath!  Begin  A I  •.•.•.•••••Begih-E nd..  V^ • •  • Stage i 2,3. v.-. •.Stage .4,5.  Preparation  T  Applying vertical pressure (o\ =25kPa)  End  A Stage 1:  Applying downward unidirectional flow  A No End  Stage 2:  ^  "Ves i NoX K  Static  stage  Piping? YesK, "End A }  A  Applying cyclic flow at afrequencyof 0.2Hz  End  : Stage 5  I  Stage 4  IK  Piping ?  ^Yes ^No  Stage 3:  Applying cyclic flow at a frequency of 0.02 Hz  I End  Piping ?  es n  XNo  Releasing the vertical pressure, a =0. v  D y n a m i c  Figure 5.2: The cyclic flow test process  86  stages  6. TEST RESULTS AND PRELIMINARY OBSERVATIONS 6.1 Introduction The vibration and cyclic Gradient Ratio (GR) devices, described in Chapter 3, were developed to allow the performance of filtration tests with different dynamic components of vibration and cyclic flow. Using different soil/geotextile combinations (Chapter 4) a multiple stage program of testing (Chapter 5) was carried out. The test results are reported in this chapter. Additional graphs are provided in Appendix A. The results include visual observations, and measurements of the following variables: 1.) water heads along the sample; 2.) flow rate; 3.) sample length; 4.) mass of passing through particles; and 5.) gradations of the passing through particles. The repeatability of these measurements is addressed at the end of this chapter. Results obtained using the vibration device are reportedfirst,followed by those using the cyclic device. For each test program the results of the static and dynamic stages are discussed separately. They are presented in terms of water head distribution, Gradient Ratio value, permeability, void ratio, and mass of passing through particles.  6.1.1 Treatment of Top Blinding As discussed previously Scott (1980), Fannin et al., (1994b) and Chin et al., (1994) observed a blinding layer that developed on the top of the soil sample during unidirectional flow. This top blinding layer was considered by these authors to be a problem during experiments. As described below, in this work the top blinding layer was also observed using both the vibration and the cyclic Gradient Ratio (GR) devices.  87  Vibration Gradient Ratio device: All homogeneous samples (21 tests) that were run in the static stage of the vibration test program developed, overnight, a greenish top blinding layer less than 1 mm thick. The influence of this lower permeability layer was to reduce flow rate, therefore this blinding layer was siphoned at the end of the static stage. In order to check whether it was a result of recycling the water through the sample, one test (s(20-1.5).gl49) was performed using non-recycled water. However, as illustrated in Figure 6.1, the top blinding layer developed similarly to all tests using recycled water. Cyclic Gradient Ratio device: Since the duration of unidirectional flow in the static stage, in the cyclic flow device, was relatively short (~90min.), no top blinding layer was observed. One test was performed to check if it would form during a longer period of unidirectional flow. It was run for approximately 140 hours, using distilled and deaired water with 0.5% concentration of bleach. The sample comprised a lower uniform layer (CTJ=1.3) with D50=0.135 mm and an upper uniform layer (CTJ=1.9) with D50=0.039 mm. The results, Figure 6.2, show development of head loss across the top layer (hi2) with time indicating once again the onset of top blinding layer. The onset occurs after 12 hours. Its influence increases at a rapid and constant rate until complete blinding occurs, defined by a head loss across hi2 equal to that applied across the sample.  6.2 Vibration Tests As described before, the vibration test program comprised two stages: a static stage in which unidirectional flow was imposed; and a dynamic stage in which both unidirectional flow and mechanical vibration were imposed simultaneously.  88  The results below are presented separately for each stage. A total of 24 tests were performed. As described in Chapter 5, two tests were performed using the same soil/geotextile combination (s(1.9-39).103i and s(1.9-39).103 ). 2  6.2.1 Static Stage  6.2.1.1 Water Head Distribution Based on the resolution of readings, the Gradient Ratio (GR) value is calculated to ±0.04. In tabulating the results they are reported to ±0.1. Typically tests for which GR>1 exhibited a GR(Mod.)>GR(ASTM),  and conversely  tests with GR<1 exhibited a  GR(Mod.)<GR(ASTM). This observation confirms the GR(Mod.) to be a more sensitive index than the GR(ASTM), as suggested by Fannin et al., (1994b). Visual observations taken immediately after preparation indicated that 21 tests were homogenous and 3 tests were not. Those that were homogenous are termed "initially homogeneous samples", and the remainder are termed "initially non-homogeneous samples". The three initially non-homogeneous samples possessed the lowest gap location (at 15% finer). In these tests (s(15-2.7).gl03, s(15-2.7).gl22 and s(15-4.5).g290) a thin layer of fines (1-2 cm) was observed at the bottom of the sample. The presentation of the results below distinguishes between the behavior of the initially homogeneous and the initially nonhomogeneous samples. •  The initially homogeneous samples showed a linear water head distribution, both at the beginning (after preparation) and at the end (after removal of the top blinding layer) of the static stage (see for example Figure 6.3a). Consequently all tests exhibited a constant  89  GR value (see for example Figure 6.3b). The GR values were within a range bounded by tests s(4.3-97).gl03 and s(6.4-306).gl49, of 0.6<GR<1.4 (see Table 6.1). Table 6.1: The GR values at the end static and dynamic stages Test End of static stage Code GR(Mod.) GR(ASTM) N s(1.9-39).gl03, 0.8 0.9 N s(1.9-39).gl03 0.9 0.9 N s(2.2-52).gl03 0.9 0.9 N s(2.2-78).gl03 1.0 1.0 N s(4.3-97).gl03 0.6 0.7 W s(6.4-213).gl03 0.6 0.8 W s(6.4-306).gl03 0.9 1.1 G s(15-2.7).gl03 827 264 G s(20-1.5).gl03 0.7 0.8 G s(20-3.1).gl03 0.9 1.1 G s(45-3.1).gl03 0.7 1.0 G s(70-1.6).gl03 0.8 0.9 N 1.0 s(2.2-52).gl22 1.0 N s(2.2-78).gl22 0.9 1.0 W s(6.4-213).gl22 0.8 0.9 G s(15-2.7).gl22 662 210 G s(70-1.6).gl22 0.8 1.0 N s(4.3-97).gl49 1.2 1.0 W s(6.4-306).gl49 1.4 1.1 G s(20-1.5).gl49 1.0 1.0 N s(4.3-97).g.290 0.9 0.9 W s(6.4-213).g290 0.9 0.9 w s(6.4-306).g290 1.1 0.9 G s(15-4.5).g290 13.7 6.3 'using third of the energy applied in the rest of the tests * N , W, G narrow wide and gap-graded respectively a  2  End of dynamic stage GR(Mod.) GR(ASTM) 0.8 0.9 0.9 0.9 1.1 1.0 1.2 1.1 1.1 1.0 1.0 0.9 1.8 1.4 283 132 0.8 0.9 1.1 1.2 0.1 0.3 1.0 1.0 1.4 1.1 1.1 1.1 1.7 1.2 131 149 1.0 1.1 1.6 1.2 3.1 1.7 1.2 1.1 1.4 1.1 1.6 1.4 1.9 1.2 1.9 1.2  W ; G  •  The thin, interface basal fines layer that was observed visually after preparation of the initially non-homogeneous samples suggests a certain internal instability of these samples. It is attributed to particle migration during the preparation process. As illustrated in Figure 6.3c and Figure 6.3e, its presence is confirmed by the concave upward shape of the water head distribution. Two of the samples on geotextiles with relatively small opening size 90  (s(15-2.7).gl03 and s(15-2.7).gl22) yielded, after sample preparation, a GR(Mod.)>650. The effect of the blinding layer increased during the static stage (Figure 6.3 d). As for the homogeneous samples this latter behavior was related to the appearance of a thin (less than 1mm) greenish blinding layer that was observed on top of the basal fines layer. In contrast, test s(15-4.5).g290 yielded a much smaller gradient ratio (GR(Mod.)=13.7), and a water head distribution that did not vary much with time (Figure 6.3e and Figure 6.3f)  6.2.1.2 Sample Conditions  6.2.1.2.1 Permeability The coefficients of permeability were calculated using Darcy's law. Based on the resolution of measurement it was found that, under a flow rate of Q=3 cm /sec, 3  k g 7 , IC57,  and  k/35 are reported to ±0.6E-3, ±0.2E-3 and ±0.1E-3 cm/sec respectively (Table 6.2). •  In all initially homogeneous samples, the flow rate at the end of the static stage (see Table 6.2 below) was similar to the initial flow rate after sample preparation. The deduced values of permeability are typical of coarse silts to silty sands (Craig, 1995).  •  All the initially non-homogeneous samples yielded IC35 values that were at least an order of magnitude larger than those of the homogeneous samples. Consider test s(154.5).g290, which had the lowest gap location (at 15% finer), the largest gap width ratio (R=4.5), and a geotextile with the largest opening size (FOS=0.290 mm). A significantly higher flow rate resulted in non-laminar flow conditions (see Table 6.2).  91  Table 6.2: The permeability (kjj, 1E-3 cm/sec.) and flow rate (Q, cm3/sec.) at the end of static and dynamic stages Test Code  End of static stage  s(1.9-39).gl03i s(1.9-39).gl03 s(2.2-52).gl03 s(2.2-78).gl03 s(4.3-97).gl03 s(6.4-213).gl03 s(6.4-306).gl03 s(15-2.7).gl03 s(20-1.5).gl03 s(20-3.1).gl03 s(45-3.1).gl03 s(70-1.6).gl03 s(2.2-78).gl22 s(2.2-52).gl22 s(6.4-213).gl22 s(15-2.7).gl22 s(70-1.6).gl22 s(4.3-97).gl49 s(6.4-306).gl49 s(20-1.5).gl49 s(4.3-97).g290 s(6.4-213).g290 s(6.4-306).g290 s(15-4.5).g290  N N  2  N N N W W G G G G G N N W G G N W G N W W G  Q  K35  0.4 0.4 0.5 2.1 0.8 1.0 2.4 0.9 1.0 3.5 0.3 0.4 2.3 0.6 1.0 0.6 0.5 0.7 2.9 1.4 0.7 1.0 2.8 37  1.27 1.30 1.61 6.11 2.37 3.01 7.21 134.70 3.01 11.54 1.12 1.27 7.40 1.72 2.88 74.09 1.41 2.03 8.68 4.57 2.35 2.90 7.97 203.98  k 1.40 1.44 1.83 6.20 3.22 3.80 6.63 0.51 3.58 12.48 1.08 1.36 7.07 1.71 3.06 0.35 1.43 2.08 7.87 4.72 2.72 3.30 8.81 32.29 57  a  a  End of dynamic stage k67 1.59 1.44 1.77 6.15 3.70 5.32 7.80 0.16 4.41 10.57 1.63 1.58 7.95 1.78 3.46 0.11 1.74 1.70 6.23 4.46 2.71 3.15 7.11 14.79"  Q  k5  0.4 0.4 0.4 1.7 0.5 0.8 2.7 2.0 0.7 18.7 0.3 0.5 1.9 0.3 0.7 1.0 0.5 0.5 2.9 1.0 0.5 0.6 3.5 64  1.31 1.30 1.21 5.11 1.42 2.28 7.92 155.43 2.02 63.07 0.66 1.16 5.89 0.95 1.92 88.82 1.36 1.36 9.21 3.02 1.39 1.77 9.58 357.35  k7 k67 1.48 1.60 1.44 1.44 1.25 1.15 4.77 4.24 1.42 1.27 2.44 2.18 5.66 4.35 1.18 0.55 2.39 2.57 53.73 58.95 2.21 6.89 1.13 1.17 5.38 5.28 0.90 0.70 1.56 1.13 0.59 0.68 1.23 1.31 1.14 0.85 5.40 2.96 2.74 2.44 1.31 1.02 1.30 1.12 7.72 5.12 292.61" 192.70 5  3  a  'Non laminar flow (see appendix B) and therefore apparent kjj. * N , W , G narrow, wide and gap-graded respectively W ; G  6.2.1.2.2 Void Ratio The void ratio is reported to ±0.01. In all tests no changes in the void ratio were observed during the static stage (Table 6.3). The range of values obtained after preparation was between 0.40 and 0.56, with the exception of one sample for which e = 0.33. No change occurred with unidirectional flow during the static stage. The lowest value of e = 0.33 was 92  obtained for a gap-graded sample in which the gap was located at 45% finer and the gap width was relatively large (s(45-3.1).gl03). It is attributed to fine particles filling the voids created by the significant coarser fraction. Table 6.3: The void ratio (e) at the end of static and dynamic stages Test End of static End of dynamic code stage stages s(1.9-39).gl03i N 0.53 0.53 N s(1.9-39).gl03 0.55 0.55 N s(2.2-52).gl03 0.49 0.42 N s(2.2-78).gl03 0.56 0.50 N s(4.3-97).gl03 0.47 0.39 W s(6.4-213).gl03 0.40 0.33 s(6.4-306).gl03 W 0.43 0.38 G s(15-2.7).gl03 0.46 0.40 G s(20-1.5).gl03 0.43 0.39 G s(20-3.1).gl03 0.47 0.50 G s(45-3.1).gl03 0.33 0.31 G s(70-1.6).gl03 0.49 0.44 N s(2.2-78).gl22 0.56 0.48 N s(2.2-52).gl22 0.52 0.41 s(6.4-213).gl22 W 0.40 0.32 G s(15-2.7).gl22 0.47 0.37 G s(70-1.6).gl22 0.49 0.43 N s(4.3-97).gl49 0.43 0.33 s(6.4-306).gl49 W 0.43 0.39 G s(20-1.5).gl49 0.50 0.43 s(4.3-97).g.290 N 0.40 0.34 W s(6.4-213).g290 0.40 0.31 s(6.4-306).g290 W 0.42 0.40 G s(15-4.5).g290 0.54 0.52 'using third of the energy applied in the rest of the tests * N , W, G narrow, wide and gap-graded respectively a  2  W ; G  Samples of wider particle size gradation had lower values of e, for example s(6.4213).gl03 with e=0.4. In contrast narrower gradations had higher values of e, like s(2.278).gl03 and s(2.2-78).gl22 with e=0.56. In gap-graded samples, the void ratio is a function of the gap location: 93  •  Samples in which the coarser fraction "floats" in the finer matrix behaved like homogeneous samples yielding higher values of void ratio (for example s(70-1.6).gl03 and s(70-1.6).gl22 with e=0.49).  •  Samples in which a smaller fines fraction had the potential to fill voids created by a dominant coarser fraction yielded a wide range of void ratios (see for example s(201.5).gl03 and s(20-1.5).g230 in which just 20% of the sample was located below the gap and yielded e=0.43 and e=0.5 respectively). The range is attributed to manual stirring that was used to increase the homogeneity of the samples and could vary from sample to sample.  •  Samples that were categorized as initially non-homogeneous (all samples in which the gap location is at 15% finer) showed a blinding layer above the goetextile. The void ratio was a function of the blinding layer thickness, and thefractionof coarse and fine particles: it is not considered truly representative of the sample.  6.2.1.2.3 Mass of Passing Through Particles As indicated in Table 6.4, excluding one test (s(15-4.5).g290), during preparation the mass of the passing through particles was in the range <250 to -12300 g/m . For reference 2  250 g/m represents just ~2g of particles, and is considered the threshold for detection and 2  collection; ~12300g/m represents ~100g which is about 7% of the prepared sample. For 2  samples that lost more than 1000g/m during preparation, the original particle size distribution 2  curve was corrected based on the mass and gradation of the passing through particles (see Appendix B). In most tests the greatest quantity of particles passed through the geotextile as a  94  result of preparation rather than unidirectional flow in the static stage. This is attributed to the manual stirring technique used in preparation. Table 6.4: Cumulative mass per unit area of the passing through particles before and during vibration (m , g/m ). 2  A  Test Preparation Static stage, Code End end N 7283 9679 s(1.9-39).gl03i N 7200 9100 s(1.9-39).gl03 N 444 481 s(2.2-52).gl03 N 37 61 s(2.2-78).gl03 N 740 s(4.3-97).gl03 753 W 666 728 s(6.4-213).gl03 W 283 s(6.4-306).gl03 370 G 3740 3864 s(15-2.7).gl03 G 61 86 s(20-1.5).gl03 G 12308 12962 s(20-3.1).gl03 G 4395 4555 s(45-3.1).gl03 N 1543 1716 s(70-1.6).gl03 N 469 567 s(2.2-52).gl22 G 111 135 s(2.8-78).gl22 W 1864 1987 s(6.4-213).gl22 G 5864 s(15-2.7).gl22 5962 G 2061 s(70-1.6).gl22 2123 N 641 666 s(4.3-97).gl49 W 1567 s(6.4-306).gl49 1629 G 469 518 s(20-1.5).gl49 N 4444 s(4.3-97).g.290 4506 W 2827 s(6.4-213).g290 2851 W 4148 s(6.4-306).g290 4185 G 22839 s(15-4.5).g290 23432 'using third of the energy applied in the rest of the tests ^> -yV, G narrow, wide and gap-graded respectively a  2  Dynamic stage (blows) 180 720 1440 1440 11728 17543 34209 -> 11200 17062 33924 -> 604 777 888 -> 148 —> -» -» 790 1012 1086 -> 802 864 888 925 444 567 728 888 4037 4185 4259 4296 -> 197 -» -> 15172 22246 43728 ->• 5419 8098 12185 15506 1740 -» —> -> 666 716 740 765 172 197 222 246 2098 2283 2345 2395 6222 6333 6419 6444 2370 2617 2753 2777 703 753 790 814 1765 1925 2135 2283 567 604 629 629 4654 4790 4876 4925 3037 3172 3432 3518 4222 4444 4839 5790 26938 32197 33444 34160  w  Test s(15-4.5).g290 is characterized by the most sensitive gap-graded soil sample and geotextile of largest opening size. It lost significantly more particles (~23000g/m ) than any 2  other test. Thereafter tests s(20-3.1).gl03 and s(1.9-39).gl03 experienced the greatest loss of 95  particles. The former, of all initially homogeneous samples, had the lowest gap location and the highest gap width ratio. The latter, of all narrowly graded samples, had the highest ratio of FOS/D85. During unidirectional flow in the static stage, the incremental mass of the passing through particles was from <250 to -2400 g/m . Again tests s(1.9-39).gl03i, s(20-3.1).gl03 2  and s(15-4.5).g290 lost the greatest amount of particles.  6.2.2 Dynamic Stage  6.2.2.1 Water Head Distribution Measurements of water heads were taken with time beginning immediately at the end of vibration. In the initially homogeneous samples no change was observed. In the initially non-homogeneous samples, due to dissipation of excess pore water pressure from the basal fines layer, a decrease in the water head was observed visually during the first few seconds after vibration. The samples are again categorized by their homogeneity after preparation. After the generally rapid dissipation of any excess pore water pressure, the following observations of the modified Gradient Ratio values were taken (Table 6.1). •  Regarding the initially homogeneous samples, 13 tests showed no change or a change of less than 0.5 (see for example Figure 6.4a), 7 tests showed an increase of 0.5 or more (see for example Figure 6.4b) and 1 test showed a decrease of 0.6 (see for example Figure 6.4c). Based on these observations, and for purposes of description, a change of 0.5 or more is considered significant. Those tests showing no change include all gap-graded samples where the gap is at 20% or 70% finer, and all samples with CTJ<4.3. With the 96  exception of two tests that could be due to the resolution of readings (s(4.3-97).gl49 and s(6.4-213).gl03, see Appendix A), all tests with CTJ^4.3 showed an increase in GR(Mod.). The sample that showed a decrease in GR(Mod.) had a large gap close to the middle of the gradation. •  Due to an increase in k  sg  (k57 and k^j), all Initially non-homogeneous samples showed a  decrease in GR(Mod.). As can be seen for example in Figure 6.4d and Figure 6.5 (see also Appendix A) the changes took place mainly during thefirsttwo series of vibration (900 blows).  6.2.2.2 Sample Conditions  6.2.2.2.1 Permeability During vibration the samples may density (see Table 6.3) and/or lose thefinerfraction. While densification tends to decrease the permeability, loss of the finer fraction tends to increase the permeability. The permeability observations that were taken as a result of vibration (see Figure 6.6) can be distinguished based on the gradation of the soil samples used. •  Narrow and wide gradations (15 tests): With the exception of tests including gradation s(6.4-306), in all tests (12 tests) the permeability along the sample was observed to stay constant or to slightly decrease (see Table 6.2 above, and for example Figures 6.6a and 6.6b). Tests including soil sample s(6.4-306) that could be characterized as having the largest CTJ and D50 (3 tests) showed a slight increase in k (k35) and a decrease in k s  sg  (see  for example Figure 6.6c). A behavior that can explain the increase in GR(Mod.) values during vibration. 97  •  Gap gradations: The behavior of these gradations can be divided into two groups based on the gap location: a. ) Gap location above 15% finer and initially homogeneous samples (6 tests): In gapgraded sample gradations with a smaller gap width ratio (R<1.6; 4 tests), the permeability values along the sample decreased. Samples with a larger gap width ratio (R=3.1; 2 samples) had experienced an increase in k  sg  (see for example Figure 6.6d).  Test s(20-3.1).gl03 that had a gap at 20%finershowed an increase also in k and s  therefore a stable GR(Mod.) value. However test s(45-3.1).gl03 that possessed a gap at 45%finershowed a slight decrease in k and therefore a decrease in GR(Mod.). s  b. ) Gap location at 15%finerand initially non-homogenous samples (3 tests): Generally the initially non-homogenous samples showed an increase in the permeability along the sample. Due to higher influence of the increase in k , all these tests showed a sg  decrease in GR(Mod.).  6.2.2.2.2 Void Ratio Changes in void ratio due to vibration, are a function of densification and loss of the finer loose fraction of the gradation. The void ratio observations that were taken during vibration (see Table 6.3 above) can be divided based on the gradation of the soil samples, as follows: •  Narrow and wide gradations (15 tests): With the exception of tests including gradation s(6.4-306) and s(1.9-39) in all others (10 tests) the void ratio showed a decrease in the range of 0.06 to 0.11. Tests including gradation s(6.4-306), have the largest Cu and D50 98  (3 tests): these tests, that were previously noted to have an increase in k , exhibited a sg  decrease in the void ratio of less than 0.06. Tests s(1.9-39).gl03, and s(1.9-39).gl03 had 2  the lowest coefficient of uniformity. These tests, which did not show significant changes in GR and permeability, experienced no change in the void ratio. •  Gap gradation: The void ratio of the initially non-homogenous samples is not considered a representative value. Therefore these gradations are excluded from the void ratio observations below, which consider only the 6 initially homogenous samples. Those with a smaller gap width ratio (R<1.6; 4 tests), showed a decrease in void ratio between 0.04 and 0.07. Samples with a larger gap width ratio (R=3.1; 2 tests) had a decrease in the void ratio of less than 0.04 or an increase of 0.03. The increase of 0.03 was observed in test s(20-3.1).gl03, which had the lowest gap location and the largest gap width ratio. In addition this test showed during vibration an increase in the permeability along the sample (see Table 6.2).  6.2.2.2.3 Mass of Passing Through Particles In the initially non-homogenous samples the thickness of the blinding layer at the vicinity of the geotextile is a function of sample preparation. Therefore these samples are excludedfromthe observations below. Observing the cumulative mass of the passing through particles (Table 6.4 above) as a result of vibration it is possible to categorize the results in two groups. Thefirstgroup lost more than 10,000g/m of particles and the second group less than 2  1700g/m . Inspection of Table 6.4 indicates: 2  99  •  Samples that lost more than 10,000 g/m (4 tests): This group includes samples with the 2  narrowest gradation (CTJ=1.9, 2 tests) and gap-graded gradations with a larger gap width ratio (R=3.1, 2 tests). •  Samples that lost less than 1700 g/m (17 tests): This group includes all the samples in 2  which the void ratio showed a decrease. 6.3 Cyclic Flow Test Each test comprised five stages: two static stages in which unidirectional flow was imposed and three dynamic stages in which there was cyclic flow. In unidirectional flow, water flows down from the top of the sample (hd=0) to the bottom of the sample (hd<0). Under cyclic flow the gradient is imposed by maintaining a constant head on the top of the sample (hd=0) and changing cyclically, the head on the bottom of the sample between constant positive (hd>0) and negative (hd<0) values (see Fig. 3.10).  6.3.1 Stage 1: Static Stage  6.3.1.1 Water Head Distribution During this stage, in all 17 tests the GR(Mod.) values did not change at all. As indicated in Table 6.5 below, 4 tests yielded 0.3<GR(Mod.)<0.5 and the remainder 0.7<GR(Mod.)<1.3 (see for example Figure 6.7). The low GR(Mod.) values are in agreement with visual observations through the transparent permeameter that confirmed no blinding layer at the bottom of the sample after preparation and during the static stage. Tests that combined gap-graded samples in which the gap was lower (at 20% finer) with a smaller opening size geotextile (FOS=0.103mm), yielded higher GR(Mod.) values (see 100  tests s(20-3.1).gl03 and s(20-1.9).gl03, Table 6.5). This phenomenon can be explained by the increased tendency o f the finer fraction to migrate within these samples. Test s(453.1).gl03 in which the gap was located around the middle o f the gradation (at 4 5 % finer) yielded GR(Mod.)=0.3. In this test the spoon stirring during preparation caused a loss o f particles from the finer fraction at the very vicinity o f the geotextile (Table 6.7), which in turn caused a lower G R ( M o d . ) value.  Table 6.5: The GR values at the end of the static stages (stages 1 and 5)  Test Code s(3.2-68).gl03 s(20-1.9).gl03 s(20-3.1).gl03 s(45-3.1).gl03 s(70-2.2).gl03 s(1.3-134).gl22 s(3.2-68).gl22 s(4.8-333).gl22 s(5.1-144).gl22 s(1.3-134).gl49 s(4.8-333).gl49 s(5.1-144).gl49 s(7.2-250).gl49, s(7.2-250).gl49 s(20-1.9).gl49 s(4.8-333).g290 s(7.2-250).g290  GR(ASTM)  GR(Mod.) N G G G G N N N N N N N W w  2  G N W  stage 1 0.7 1.3 1.2 0.3 0.5 1.0 1.1 0.9 0.3 1.3 1.1 0.9 1.1 0.7 0.9 0.8 0.5  Stage 5 0.9 1.0' a  L  D  u  D  i_ D  1.1' 0.4 a  C  0.7 1.4  a  c  0.6 c c  0.6  a  D  0.4  a  Stage 1 0.9 0.9 1.0 0.6 0.9 1.1 1.0 0.8 0.6 1.0 0.9 0.9 0.9 0.8 0.9 0.7 0.3  Stage 5 0.9 1.0' a  t, u  u  t,  0.9 0.3  a a  C  0.8 1.2  a  C  0.7 C c  0.6  a  uD  0.3  a  "Continuous piping observed in previous stage This stage were not performed and continuous piping was observed during previous stage. This stage was not performed. Movement o f particles to the top o f the sample was observed during stage 4 ° N , W , G narrow wide and gap-graded respectively b  c  N ; W ;  101  6.3.1.2 Sample Conditions  6.3.1.2.1 Permeability No changes in the permeability (k\j) were observed during the static stage. The permeability values were a function of the density and gradation profile. These values were in the range of 0.96E-3 to 11.42E-3 cm/sec (see Table 6.6) and were in agreement with expected values for the gradations of sandy silt tofinesand size (Craig, 1995). Table 6.6: The void ratio (e), permeability (k.17,1E-3 cm/sec) and flow rate ( Q , cm /sec.) 3  at the end of static and dynamic stages Test  e  k  n  / Q  Stage 4 Stage 5 Stage 3 Stage 2 Stage 1 code Stage 1 N 1.29/0.4 6.34/2.2 3.0/0.9 3.56/1.3 0.96/0.3 0.42 s(3.2-68).gl03 G 5.30/1.8 4.0/1.3 6.46/2.0 4.49/1.3 3.23/1.0 0.44 s(20-1.9).gl03 G a a a 8.77/2.5 6.89/1.9 0.43 s(20-3.1).gl03 G a a a 0.98/0.3 3.18/1.2 0.35 s(45-3.1).gl03 G a 3.31/1.2 1.42/0.5 4.52/0.7 1.07/0.3 0.49 s(70-2.2).gl03 N 12.14/2.8 11.46/2.5 18.65/3.8 11.50/2.5 10.7/2.4 0.63 s(1.3-134).gl22 N 1.31/0.4 8.32/2.7 1.28/0.4 3.45/1.3 1.05/0.3 0.42 s(3.2-68).gl22 N c 10.41/2.8 7.95/2.4 7.02/1.9 8.48/2.1 0.47 s(4.8-333).gl22 N 3.79/1.4 1.43/0.5 3.82/1.3 1.41/0.5 1.34/0.4 0.42 s(5.1-144).gl22 N 21.89/3.8 11.832.5 12.09/2.6 12.79/2.8 11.42/2.5 0.63 s(1.3-134).gl49 N c 6.01/1.7 9.84/2.6 7.67/2.2 7.89/2.0 0.46 s(4.8-333).gl49 N 3.68/1.4 1.45/0.5 4.80/1.7 1.56/0.5 1.30/0.4 0.38 s(5.1-144).gl49 W c 4.14/1.5 2.15/0.7 1.93/0.6 4.05/1.5 0.41 s(7.2-250).gl49! W c 2.20/0.7 4.45/1.5 4.08/1.5 1.82/0.5 0.39 s(7.2-250).gl49 G 5.30/1.8 3.99/1.2 6.46/2.0 4.49/1.3 3.23/1.0 0.44 s(20-1.9).gl49 N a 5.64/1.6 8.95/2.5 6.66//1.7 5.89/2.1 0.45 s(4.8-333).g290 W 2.15/0.7 5.64/2.0 3.93/1.2 1.96/0.6 3.84/1.5 0.39 s(7.2-250).g290 "Stage was not performed Continuous loss of particles was observed and therefore the permeability was calculated at the beginning of the stage This stage was not performed. Movement of particles to the top of the sample was observed during stage 4 N wGj^ Q narrow, wide and gap-graded respectively b  b  b  b  b  b  b  b  2  b  b  b  b  0  ;  ;  102  Test s(3.2-68).gl03 that among the non-gap graded gradations possessed the lowest D50 (D50=0.068 mm) showed a lower value of permeability (ki7=0.96E-3cm/sec). Test s(l.3-134).gl49 that among the non-gap graded gradations showed the lowest Cu (Cu=13) also showed the highest value of permeability (ki7=l 1.42E-3 cm/sec).  6.3.1.2.2 Void Ratio In all tests the void ratio did not change as a result of the vertical stresses applied before the test program or as a result of flow during the static stage. The void ratio values were in the range of 0.35 to 0.49 (see Table 6.6) with the exception of two tests. The lower value (e=0.35) is attributed to the potential of test s(45-3.1).gl03 to form a denser fabric because of its larger gap at 45% finer. The upper value (e=0.49) for test s(70-2.2).gl03 is likely a consequence of the smaller fraction above the gap floating in the finer but very uniform fraction. The two exceptions had the lowest coefficient of uniformity (s(1.3134).gl22 and s(1.3-134).gl49), which is consistent with the largest values of void ratio (e=0.63).  6.3.1.2.3 Mass of Passing Through Particles It was impractical to collect and therefore report accurately on quantities of less than 250g/m (~2g). This small amount constitutes less than 0.1% of the mass of the soil sample, 2  and was considered the threshold of measurement. The mass per unit area (m ) of the passing A  through particles collected after preparation, and after each stage of the test program are reported in Table 6.7. Most samples lost more than 250g/m as a result of preparation. In no 2  test was a continuous flow of passing through particles visually observed during unidirectional 103  flow in stage 1. Indeed the total mass of passing through particles during this stage was always less than 250g/m . The greatest initial losses were in those samples that exhibited a 2  large gap width ratio (s(20-3.1).gl03 and s(45-3.1).gl03) or a large portion of finer fraction (s(70-2.2).gl03). Table 6.7: Mass per unit area (m , g/m ) of passing through particles 2  A  Stage 3 Stage 4 Stage 2 Stage 1,5 >2500 <250 <250 <250 G <250 >2500 <250 <250 G <250 >2500 test stopped G >2500 test stopped <250 G <250 <250 <250 >2500 N <250 <250 <250 >2500 N <250 <250 >2500 <250 N <250 <250 <250 <250 N <250 <250 <250 <250 N <250 <250 <250 >2500 N <250 <250 <250 <250 N <250 <250 <250 <250 W <250 <250 <250 <250 W <250 <250 <250 <250 s(7.2-250).gl49 G <250 <250 >2500 <250 618 s(20-1.9).gl49 N <250 <250 >2500 495 <250 s(4.8-333).g290 W >2500 <250 <250 <250 983 s(7.2-250).g290 Continuous flow of passing though particles was observed. Stage 5 was not performed This stage was not performed. Movement of particles to the top of the sample was observed during stage 4 N w Gj^ ^ Q id ± gap-graded respectively Sampe code s(3.2-68).gl03 s(20-1.9).gl03 s(20-3.1).gl03 s(45-3.1).gl03 s(70-2.2).gl03 s(1.3-134).gl22 s(3.2-68).gl22 s(4.8-333).gl22 s(5.1-144).gl22 s(1.3-134).gl49 s(4.8-333).gl49 s(5.1-144).gl49 s(7.2-250).gl49i  N  2  Preparation 624 978 1793 2237 2122 621 492 331 <250 328 <250 324 <250 <250  a  a  b  a  b  a  b  a  a  a  c  a  c  c  c  a  b  a  a  a  b  c  ;  ;  n  a  r  r  o  w  w  e  m <  6.3.2 Stages 2-4: Dynamic Stages  6.3.2.1 Water Head Distribution Observations of the water head distribution in the dynamic stages are presented separately for each stage.  104  6.3.2.1.1 Stage 2 (f=0.2 Hz, rj =25 kPa) v  The externally applied demand was a step wave. Though, as shown in Figure 6.8, the overall head difference (hi7) that was actually applied is that approximately of a sine wave. This variation of I117 corresponds well to the shape and frequency of gravity waves (SPM, 1984). Changes in the direction of flow are indicated by the vertical dashed grid lines in Figure 6.8. Inspection of Figure 6.8 suggests the peak value of head difference (hi 7) occurs at approximately half way between the changes in flow direction. The other head differences along the sample ( h i 6 , hi5, h i 3 , h i 2 ) show a time lag in reaching their peak. The governing unsteady flow conditions at f=0.2 Hz precluded the calculation of Gradient Ratio (GR) values. 6.3.2.1.2 Stage 3 (f= 0.02 Hz, a =25 kPa) v  At the slower frequency of f=0.02HZ, variation of the overall head difference was that of a step wave (see for example Figure 6.9). The constant value represents steady unidirectional flow conditions. The switching operation, shown again by the vertical grid lines, is accompanied by a minor transient disturbance that is attributed to a water hammer effect. This peak was observed to be almost an integral part of the sine wave at the higher frequency applied in Stage 2 (f=0.2Hz, see Figure 6.8). The GR value near the end (lsec. before the change in flow direction) of each successive half cycle was deduced. The results, see for example Figure 6.10a, suggest that one cycle was significant to achieve consistency in the GR(Mod.) and in the GR(ASTM) in both directions. 105  As indicated in Table 6.8 below the consistency between upward and downward GR values at the end of half a cycle during the tenth cycle is apparent. Comparing the GR obtained at the last second of the tenth cycle (Table 6.8) to the values obtained at the end of stage 1 (Table 6.5 above), it was observed that in 9 tests the GR values did not change significantly (see for example Figure 6.10b); four tests showed an increase of more than 0.5 in the GR(Mod.); and one test showed an increase of more than 0.5 in the GR(ASTM). It should be emphasized, that even though steady flow conditions were observed during half a cycle, the duration of less than half a period (25sec. for f=0.02Hz) can not be considered as long enough to achieve filtration stability. Since in all tests the GR(Mod.) and GR(ASTM) values at the end of half a cycle were below 5.2 and 1.7 respectively, it is concluded that the samples were homogeneous. This conclusion was supported by visual observations through the transparent permeameter. Table 6.8: The G R values at the end of half a cycle (after 10 cycles) GR(Mod) GR(ASTM) Test Upward Downward Upward Downward code Flow Flow Flow flow N 2.4 2.7 s(3.2-68).gl03 1 1 G s(70-2.2).gl03 0.1 0.1 1.5 1.6 N s(1.3-134).gl22 •1.0 1.0 1.1 1.1 N s(3.2-68).gl22 3.2 3.6 1.0 1.0 N s(4.8-333).gl22 0.8 0.8 0.5 0.5 N s(5.1-144).gl22 0.6 0.6 0.8 0.8 N s(1.3-134).gl49 5.4 •5.1 1.1 1.1 N s(4.8-333).gl49 0.7 0.7 1.1 1.1 N s(5.1-144).gl49 1.3 1.3 0.9 0.9 W s(7.2-250).gl49 0.9 0.9 1.0 0.9 W s(7.2-250).gl49 0.9 0.9 1.0 0.9 G s(20-1.9).gl49 0.4 1.1 1.0 1.1 N s(4.8-333).g290 0.5 0.5 0.6 0.6 W s(7.2-250).g290 1.5 1.6 0.5 0.5 N,W, G narrow wide and gap-graded respectively ,;G  106  6.3.2.1.3 Stage 4 (f=0.2 Hz, a =0 kPa) v  As illustrated in Figure 6.11 similarly to stage 2 unsteady flow conditions were applied along the sample. During the upward flow of stage 4 quick conditions were present in the sample (will be discussed later in Chapter 7, section 7.4.2).  6.3.2.2 Sample Condition  6.3.2.2.1 Permeability and Void Ratio Based on the flow rate measurements (Table 6.6) and the median particle diameter of the tested soils (see Chapter 4), it was found that laminar flow conditions occurred in all tests (for explanation see section B.3 in appendix B). However, as shown above (section 6.3.2.1.1), unsteady flow conditions were observed under cyclic flow with f=0.2 Hz. Therefore for the latter case Darcy's law is not applicable as an integral relationship. However, in order to interpret the measured flow rate (Q) and the measured total head difference (hi7), Darcy's law was used as an integral relationship to derive the coefficients of permeability reported in Table 6.6. Further comments on this analysis is given in section 7.4.3. The measurements of flow rate (Q) and total head difference (hi 7), under confined conditions (stages 1, 2 and 3) are summarized in Figure 6.12. From inspection of Figure 6.12 it can be seen that the measured total head difference (hi 7) varies between different tests and different stages. Although the distance between the constant head tanks (42-43cm) that imposed the external hydraulic load was kept the same during the test program, for each frequency the measured total head difference (hi 7) decreased as the flow rate increased. This  107  phenomenon increased as the flow rate increased. However, the measured total head difference of tests with lower flow rates at f=0.2 Hz was greater than the external applied head difference of 42-43cm. Furthermore an increasefromf=0.02 Hz to f=0.2 Hz caused an increase in the flow rate. This phenomenon was observed to be more significant in tests with lower flow rate. The changes in the measured total head difference (hi7) caused small variations in the maximum applied gradient. Though, such small variations in the imposed hydraulic gradient are not significant to thefiltrationperformance (Bertram ,1940; Fannin et al, 1994a, Chin et al, 1995). During stages 2 and 3 the void ratios of the retained samples did not change. From Table 6.6 it can also be seen that the flow rates and the permeability values during stage 4 were higher than during stages 2, and 3. This is attributed to the looser state of the soil during the quick conditions that were applied in this stage.  6.3.2.2.2 Mass of Passing Through Particles The behavior during cyclicflowis divided to two groups (see Table 6.7): •  Samples that lost less than 250g/m (~2g) ofparticles: The portion of less than 250g/m is 2  2  considered here as a noise (see explanation in section 6.3.1.2.3). •  Samples that continuously lost more than 2500g/m of particles: A Continuous loss of 2  particles was observed through the transparent collector trough. In all samples that showed a continuous loss of particles, the quantity in the collector trough was found to be above 2500g/m The value of 2500g/m was established by Lafleur, et. al, 1989 as a 2  2  boundary for initiation of piping.  108  6 . 3 . 3 Stage 5 : Static Stage In this stage nine tests were performed. Although the soil samples were severely disturbed during stage 4, the additional observations that can help further investigation are presented below. •  Changes in GR: The GR values were in range of 0.4 to 1.5 (see Table 6.5 above) and did not change during this stage. Therefore it appears that none of the tests showed blinding.  •  Permeability and void ratio: Changes in permeability as a result of the quick conditions applied during stage 4 were a function of the passing through particles, and changes in the void ratio. Two tests did not show a continuous loss of particles during previous stages (s(5.1-144).gl22 and s(5.1-144).gl49). These tests had the same permeability values during stages 1 and 5 (unidirectional stages).  •  Mass ofpassing though particles: Similarly to stage 1, there was no continuous migration of particles and all samples lost less than 250g/m . 2  6.4 Gradation of the Passing Through Particles Selected samples of passing through particles were taken to the Sedigraph 5100 for particle size analysis. An example of such analysis is given in Figure 6.13, which compares a gradation after preparation with that after stage 4. It was observed that the Cu of the passing through particles was in the range of 1.2 to 1.7. The behavior is essentially independent of the model soil and geotextile sample. To be consistent with the characteristic opening size (Apparent Opening Size or Filtration Opening Size), the D95 of the passing through particles was chosen to represent the largest geotextile opening size, defined as D95p.  109  Table 6.9: D95p of the passing through particles from the vibration GR device (1E3mm) M o d e l soil sample  Geotextile  FOS  s(1.9  s(2.2  s(4.3  s(6.4  s(6.4  s(15  s(15  s(20  s(20  s(70  (lE-3mm)  -39)  -52)  -97)  -213)  -306)  -2.7)  -4.5)  -1.5)  -3.1)  -1.6)  103  P V  122  P V  60 60  -  65 65 80 75  -  60  70  -  -  60 65  75 80  -  85 90 105 110 290 p 105 110 V p - after preparation; v - after vibration. 149  P V  -  90 90 105 105  70 70 85 80 90  -  65 65  -  -  -  80 80  -  -  90 90  -  -  105  -  -  -  -  Table 6.10: D95p of the passing through particles from the cyclic GR device (lE-3mm) Geotextile  M o d e l soil sample  FOS 103  s(1.3-  s(3.2  s(4.8  s(5.1-  s(7.2  s(20  s(20  s(45  s(70  134)  -68)  -333)  144)  -250)  -1.9)  -3.1)  -3.1)  -2.2)  60 60  65 65  60 60  65 65  P c  122  P c  149  p c  145 150 150 165  65 65 80 80  a  a  80 b  90 90  b  105 105  105 p 105 c p - after preparation; c - after cyclic flow. "From the confined stage ''Not available 290  A summary of the measured D95p values is given in Tables 6.9 and 6.10. From inspection of these tables it can be seen that: •  D95p obtained after preparation, vibration, or cyclic flow was essentially the same.  •  generally, D95p increased from 0.060 mm to 0.105 mm as the Filtration Opening Size (FOS) increased from 0.103 mm to 0.290 mm; and  110  •  two tests (s(1.3-134).gl22 and s(1.3-134).gl49) gave higher D95p values (D95p=0.1450.165 mm). This behavior is attributed to their coefficients of uniformity and is discussed in Chapter 7 (section 7.5).  6.5 R e p e a t a b i l i t y  To assess experimental repeatability, four sets of coupled tests were examined: •  Vibration test program - two tests in the vibration device using s( 1.9-3 9). g 103;  •  Cyclic test program - two tests in the cyclic flow device using s(7.2-250).gl49;  •  Static stage (unidirectionalflow)- one test in the vibration device and one in the cyclic flow device using tests s(20-3.1).gl03 and s(45-3.1).gl03  The repeatability of the performance tests is discussed below, with reference to the GR values, permeability, void ratio, and mass of the passing through particles. •  GR values: As indicated in Table 6.1 and Table 6.5, in tests s(1.3-39).gl03, s(7.2250).gl49 and s(20-3.1).gl03 the GR values were close to 1. Due to sample preparation, test s(45-3.1).gl03 in the cyclic flow device gave lower GR values (GR(Mod.) = 0.3 and GR(ASTM)=0.6). These values were smaller than the GR values obtained in the vibration GR device by less than 0.5. The data indicate homogenous samples (no blinding layer). This was confirmed by visual observations through the transparent permeameter.  •  Permeability and void ratio: In all tests the water temperature was measured to be 23 °C. Good repeatability was observed in test s(1.9-39).gl03 in the vibration device and in test s(7.2-250).gl49 in the cyclicflowdevice (see Appendix A and Table 6.6) .  Ill  Test s(20-3.1).gl03 showed after preparation in the vibration device e=0.47 and k=HE-3 cm/sec. In the cyclic flow device this test showed a lower void ratio (e=0.43) and as expected a slightly lower permeability value (k=7E-3 cm/sec). Test s(45-3.1).gl03 showed in the vibration device e=0.33 and in the cyclic flow device e=0.35. However, taking into account the accuracy of 0.01 in the void ratio derivation, these differences were not significant. In addition the permeability values of these samples in both devices were essentially the same (kslE-3 cm/sec). Furthermore, it is interesting to.note, that in both devices the void ratio and the permeability of test s(453.1).gl03 showed to be the smallest among all tests. •  Mass ofpassing through particles: Based on Tables 6.4 and 6.7, taking into account that 125g/m is equivalent to just lg of soil and that the mass of the retained samples is more 2  than 1500g, it is apparent that good repeatability was demonstrated in test s(1.9-39).gl03 in the vibration device and test s(7.2-250).gl49 in the cyclic flow device. As a result of preparation an apparent poor repeatability is observed in tests s(203.1).gl03 and s(45-3.1).gl03 (see Tables 6.4 and 6.7). However, it should be noted that the gradation and void ratio after sample preparation were corrected using the mass and gradation of the passing through particles (see Appendix B). Therefore the mass of the passing through particles during preparation did not influence following stages.  6.6 Summary •  Twenty-four tests were performed under the vibration test program. After preparation, 21 samples were homogeneous and due to particles segregation 3 samples were nonhomogeneous. 112  As a result of the environmental conditions during the static stage, top blinding layer developed on top of the homogeneous samples. The top blinding layer was siphoned at the end of the static stage. A significant increase in the Gradient Ratio (GR) values was observed just in two of the non-homogeneous samples. During vibration two samples with the narrowest gradation and two samples with a larger gap width significantly lost more particles than all other samples. Seventeen tests were performed in the cyclic flow test program. A l l samples were homogeneous  after  sample preparation. No changes  were observed  during the  unidirectional flow (static stage). During stages of cyclic flow 11 samples continuously lost their particles. Fifty-seven gradation tests were performed on selected samples of particles that passed through the geotextiles during the test program. The geotextile characteristic opening sizes obtained using these tests were generally smaller than those obtained by the standard Filtration Opening Size tests (CAN/CGSB-148.1 No. 10 - 94). Inspection of four coupled tests showed good repeatability in each of the devices and between the two devices.  113  After preparation After 1170 minutes  o E o  Water head h 4 , (cm)  Figure 6.1: The water head distribution before and after top blinding formation (s.(201.5.gl49)  Elapsed time t, (hours)  Figure 6.2: Head loss with elapsed time due to top blinding layer  114  s(4.3-97).g103 (end of static stage)  -B-  s(6.4-306).g149 (end of static stage) Initial theoretical reference line  O  of initial uniform sample  #  ASTM Mod.  2-^ s(6.4-306).g149  te-  #=8  s(4.3-97).g103  i ' i 400 800 Elapsed time t, (min.)  20 40 Water head h , (cm) d  b.)  a.) After preparation s(15-2.7).g122  -e  1000 - ,  After preparation s(15-2.7).g103 after 950 minutes s(15-2.7).g122 800 •  After 900 minutes s(15.2.7).g103  ASTM s(15-2.7).g122  #—  Mod. s(15-2.7).g122  A—  ASTM s(15-2.7).g103  ± —  Mod. s(15-2.7).g103  600 - \  400 - \  200  0  -20  H ~i—  20 40 Water head h j (cm)  200  s(15-4.5).g290  -Q—  i  r  1  800  1000  d.)  C.)  —  1  400 600 Elapsed time t, (min.)  s(15-4.5).g290  After preparation  -©—  After 980 minutes  •  a.  ASTM Mod.  12-  CD  o" 2  •s 5!  s O  8  -G  I ' 20 40 Water head h <i (cm)  —I 200  60  1  1 1 ' 400 600 Elapsed time t, (min.)  ©  1  800  1000  e.) Figure 6.3: The influence of the water head distribution on the G R values  115  116  800 - ,  s(15-2.7).g122 -©—  ASTM  -#—  Mod.  1000  4000  2000 3000 Number of blows, N b  Figure 6.5: The influences of blow count on the gradient ratio 0.004 -  0.005  s(1.9-39). g 103 1  K  35  K  57  K  67  0.003  0.004 H  0.003  0.002 —j  0.001 - \  0.000  0.002 H  1 1000  '  1  '  1  '  2000 3000 Number of blows, N ^  0.001  1  1000  4000  2000 3000 Number of blows, N ^  b.)  a.) 0.008 - i  0.008 - i  0.007  0.006  f  0.006  0.004 H  to CD  e O-  0.005 H  0.004  -I  1000  1  1  1  1  2000 3000 Number of blows, N ^  0.002 H  0.000 I  I  ~l  1000  4000  '  1  1  1  2000 3000 Number of blows, N b  c.) d.) Figure 6.6: The permeability values response due to vibration  117  1  1  s(5.1-144).g122 120  -e  Beginning of static stage 1 End of static stage 1 Initial theoretical reference line of homogeneous sample  80  .Q  E o  40  T -50  -40  i  1  r  -30 -20 Water head h ^ , [cm]  Figure 6.7: Stage 1 - The water head distribution at the end of the static stage  s(5.1-144).g122 Port Location (mm)  50  55  60 65 Elapsed time t, (sec.)  Figure 6.8: Stage 2 - The measured head difference with time (f=0.2 Hz, after 10 cycles)  118  s(5.1-144).g122 Port Location (mm) 0 8  -B-e-  25 75 101  500  525  550 575 Elapsed time t, (sec.)  600  "17 16  h  "15 13  h  h  1  2  625  Figure 6.9: Stage 3 - The measured head difference with time (f=0.02 Hz, after 10 cycles).  5-1  - 3H  120 - ,  s(5.1-144).g122 -f-  Mod. downward flow  X  A S T M downward flow  O  Mod. upward flow  |  |  s(5.1-144).g122 End of static stage 1 End of half a cycle in stage 3, N =10 c  A S T M upward flow  I  2  40  -| 4  ,  1  ,  8 Number of cycles, N  1 12  1  1 16  Water head h , , ( c m )  c  a.) b.) Figure 6.10: Stage 3 - The water head response to cyclic flow.  119  s(5.1-144).g122 Port location mm] "17  •2  h 16  0  -40 - \  50  55  60 65 Elapsed time t, (sec.)  F i g u r e 6.11: Stage 4 - T h e measured head difference w i t h time (f=0.2 H z , after 10 cycles).  120  0.01  0.1  1  Grain size, (mm)  Figure 6.13: Grain size distribution of flushing through particles (cyclic flow test).  121  7. ANALYSIS OF TEST RESULTS 7.1 Introduction As described previously, in both the vibration and cyclic flow test programs the sample was subjected to a combination of static and dynamic conditions. Static conditions were applied by imposing unidirectional flow under an applied external gradient of approximately 4. Dynamic conditions were subsequently imposed by a vibration component in the vibration test program and a cyclic flow component in the cyclic flow test program. In this chapter the test results presented in chapter 6 are analyzed and discussed. Interpretations are made with reference to previous related studies and the possible implications for practice are discussed. Treating the static and dynamic conditions separately, the analysis addressesfirstthe overall performance and the internal stability of the model soil and then the soil/geotextile interaction. This interaction includes the phenomena of piping and blinding.  7.2 The Phenomenon of an External Blinding Layer In order to evaluate soil/geotextile compatibility using the Gradient Ratio (GR) device it is necessary to accommodate responses that are related to the conditions applied in the laboratory. The ASTM D 5101-96 (section 8.1.2) suggests the use of an algae inhibitor to control biological clogging, but does not specify the type or how to introduce it. Fannin et al. (1994b) concluded that biological clogging can be eliminated by treatment of the water with algaecide, but that some physical clogging of the soil may result due to fine particles in the water. Fischer et al. (1999) supports this observation and suggested the use of both algae inhibitor and a micro-screen. 122  The phenomena of clogging, blocking or specifically blinding, that cause a decrease in the system permeability (kn), will be termed here for the purposes of discussion "blinding". In addition the terms internal and external blinding layer will be used. The external blinding layer is related to the environmental conditions that are induced by the supplied water. This can be due to suspended solids in the water supply (Bertram 1940) and/or chemical, biological, or biochemical processes (Rollin, 1996; Mackey and Koerner, 1999). The internal blinding layer is related to the potential mobility of thefinerloose fraction within the primary soil fabric and therefore is a function of the internal stability. Such a blinding layer was observed in the vicinity of the geotextile in soil samples that were non-homogeneous after preparation.  7.2.1 S t a t i c  C o n d i t i o n s  Assuming a gradient, i, is applied to a soil sample of permeability, k, porosity, n, and length, L, the average time for water to travel from the top to the bottom of the sample, tL, is given by: t =(Lxn)/(kxi) L  (7.1)  Therefore, assuming one pore volume exchange is sufficient to cause blinding, the time for an internal blinding layer to develop should not exceed the time needed for the water to travel from the top to the bottom of the soil samples. Assuming a gradient of i=4 is applied to a soil sample with a permeability of coarse silt to very fine sand (k=lE-3cm/sec), a porosity of n=0.45 and a length of L=10 cm, it can be shown (Equation 7.1) that the time needed for the migrating particles to travel from the top to the bottom of the sample is less than 20 min. Based on this approximation and taking into 123  account that the internal blinding layer is made of the migrating particles, it can be concluded that in all tests performed here, the potential development of an internal blinding layer should be observed in less than 20 min. The static stage (stage 1- unidirectional flow) in the cyclic flow test program was not long enough (90 minutes) for the imposed environmental conditions to cause an external blinding layer. However during the static stage of the vibration test program (1000 minutes) the external blinding layer developed on top of the sample or on top on the internal blinding layer and was visually observed to have a greenish color. In cases where the external blinding layer developed on top of the internal one the large head losses across the external blinding layer caused an increase in the Gradient Ratio (GR) values. The investigation that was carried out here was performed while changing the source of the supplied water. The tests can be divided to those using circulated water (25 tests) and one test without circulated water.  7.2.1.1 Circulated Water These tests include: 1.) twenty three tests that were performed using de-aired water, 2.) one test that was reported by Nishigata et al., (2000) using de-aired water, and 3.) one test that was performed using distilled water. 1.) Twenty three tests in the static stage of the vibration test program - the observations that were taken in these tests can be divided based on the homogeneity of the soil samples after sample preparation: •  In all soil samples that were initially homogeneous (20 tests) the external blinding layer was visually observed on the top of the soil samples after -800 minutes of  124  unidirectional flow. Head losses across the external blinding layer reduced the differential water head across the remaining sample length and as a result the seepage force on the potential migrating particles. Therefore the Gradient Ratio (GR) values did not change (see for example Figure 7.1 and schematically Figure 7.2a). •  In the three initially non - homogeneous soil samples, larger values of permeability were observed in the soil above the internal blinding layer  (IC35,  see Table 6.2).  Interestingly, a blinding layer did not develop on top of these samples. However, it was visually observed that in tests s(15-2.7).g203 and s(15-2.7).gl22 the external blinding layer developed on top of the internal one. As illustrated in Figure 6.3c and 6.3d this was also observed by changes in the water head distributions and the GR values. Conversely, in sample s(15-4.5).g290 that possessed larger permeability values at the vicinity of the geotextile (IC57 and k^, see Table 6.2) no external blinding layer was visually observed to develop. As illustrated in Figure 6.3e and 6.3f this was also observed by a relative stability in the water head distribution and the GR values with time. 2.) One static test that was reported by Nishigata et al, (2000) - Nishigata et al. (2000) performed a long-term filtration test, at i= 5 for a duration of 14 days. The model soil was a gap graded (gap location -20% finer) glass bead sample, where the fraction below the gap was a coarse silt (D10=0.036mm) and the fraction above the gap a medium sand (D85=0.45mm). The geotextile was a nonwoven needle-punched sample with an apparent opening size of 0.07 mm. Based on the results obtained by Nishigata et al. (2000), Figure 7.3 was redrawn and re-analyzed in this work. Two stages that are characterized by an  125  increase in the GR values (Figure 7.3a) can be observed: stage 1 that took place within the first 20 minutes; and stage 2 that began 1500 minutes after the beginning of the test and ended after 7000 min. Based on the explanation that was given here using Eg. 7.1 it can be concluded, that the first stage (for which ti<20 minutes) is related to the internal blinding layer and the second stage to the external blinding layer. It can be further observed (see Figure 7.3b and schematically 7.2b) that once the process of an external blinding layer began to take place it continued essentially at a constant rate until reaching a value after which no further changes can be observed. 3.) Distilled water - one test was performed using distilled (purified) and de-aired water with 0.5% concentration of bleach. As illustrated in Figure 6.2 in this test head difference readings along the top of the sample were taken (hi2)- In agreement with the observations above, from thisfigureit is apparent that once the process of an external blinding layer begins to take place it continues at a constant rate until reaching a value after which no further changes can be observed.  7.2.1.2 Non-Circulated Water To examine the possibility that the phenomenon of external blinding is a consequence of recirculating the de-aired water supply, a test was performed without recirculation (using one reservoir as an inlet and another reservoir as an outlet). As illustrated in Figure 6.1, the external blinding layer was again observed to develop, implying that it was not a result of recirculation and hence could not be attributed to the influence of suspended very fine particles of the model soil.  126  7.2.1.3 Synthesis Analysis of the observations suggests the following: 1.) Duration of the Gradient Ratio (GR) test: As noted above the ASTM D 5101-96 (section 8.1.2) suggests the use of algae inhibitor to control biological clogging. Assuming that the algae treatment in the laboratory eliminates the influence of the environmental conditions the ASTM (see note 3) standard implies that the GR test may be terminated at or before 24 h if there is no tendency toward stabilization. In order to be able to approximate the time needed for the internal blinding layer to develop, a simple equation was suggested here (Equation 7.1). Furthermore based on this equation it was demonstrated theoretically that using coarse silt, stabilization should be observed in less than 20 min, which implies that the test can be terminated after 20 min. However this same equation implies that for a fine silt the test should be run for approximately 20 h. 2.) Implication for laboratory testing practice: Based on the work done by Haliburton and Wood, (1982) and taking into account just the internal blinding layer, a laboratory performance of GR(ASTM)>3 is interpreted as poor soil/geotextile compatibility (U.S. Army Corps of Engineers, 1977; USFHWA, 1995). However, it was shown here, that in the laboratory the external blinding layer has the potential to develop on top of the internal one. Hence, there is a growing body of evidence that a major component of the clogging is microbiological related (Rowe, 1998). Therefore, in modeling the field performance the potential development of an external blinding layer above the internal one should be considered.  127  7.2.2 Dynamic Conditions  7.2.2.1 Vibration In the initially homogeneous samples where the top blinding layer developed during the static stage it was removed by siphoning prior to the vibration stage. Three samples (s(152.7).gl03, s(15-2.7).gl22 and s(15-4.5).g290)) experienced a segregated behavior after sample preparation. At the onset of unidirectional flow they had GR(Mod.) values of 283, 131 and 2 respectively, increasing to 827, 662 and 14 respectively due to the external blinding layer. Following the application of vibration these values decreased to 104, 34, and 16 after 180 blows and changed again to 289, 131 and 2 after 1440 blows (see Figure 7.4). The changes in the GR(Mod.) values following vibration are believed to be due to limited migration of particles in the sample, limited movement of particles that formed the internal blinding layer through the geotextile, and limited release of the external blinding layer.  7.2.2.2 Cyclic Flow As shown above, under unidirectional flow the top external blinding layer developed in less than 15 hours (see Figure 6.2). In most cyclic flow tests, the dynamic stages (stages 2, 3, and 4) exceeded duration of more than 30 hours. However, visual observations and the water head distributions (see Figure 6.10b) suggest that under these flow conditions the external blinding layer did not develop, implying that this phenomenon is dependent on unidirectional flow.  128  7.3 Base Soil Internal Stability under Static Conditions - The loosefraction:The term internal stability is related to the potential mobility of the finer loose fraction within the primary skeleton fabric of the soil. As discussed further, using granular soil, Kenney and Lau (1985) showed that thefinerloose particles (/i) within a primary skeleton fabric, can be expressed in relation to the void ratio of the primary skeleton fabric (ep) and the porosity of the loose particles (ni) within the pore spaces of the primary skeleton fabric, where: /i<l-(l+e x(l-m))"  1  p  (Equation 7.2)  The values of ep and ni are a function of compaction and the gradation width of the primary skeleton fabric and the loose fraction. Estimating extreme values of ep=0.7 and ni=0.4 for narrowly graded soil samples, Kenney and Lau (1985) found that /i<30%, and estimating values of ep=0.4 and ni=0.4 for widely graded samples, they found that /i<20%. In this study the widest gradation gave CTJ=7.2 (see Table 4.4). Therefore, for purposes of analysis, it is conservatively assumed that in all narrowly and widely graded soil samples tested here the maximum potential amount of thefinerloose fraction is 30%. As an extreme case, consider that in the gap graded soils the fraction above the gap composed of nearly uniform size particles at a loose state for which ^=0.9. Furthermore, consider that the pores contain very widely graded particles for which ni=0.1. From Equation 7.2 the value of /i<45% is obtained. This suggests that the fraction below the gap will be considered as the maximum potential finer loose fraction just in soil samples that showed a gap at 45%fineror less.  129  - The stability of the soil samples: A sample was considered here as internally unstable if migration of the loose fraction yielded an increase in the Gradient Ratio (GR) value or a continuous loss of particles was observed. However, considering that all samples lost less than 1% of their original gradations (see Tables 6.4 and 6.7) during the static stage, which has been noted to be less than the piping boundary of 2500 g/m established by Lafleur et al., (1989), 2  the stability evaluation focused essentially on the GR. Examining the 19 gradations (10 in the vibration, 7 in the cyclic and 2 in both vibration and cyclic devices) used in this test program with the Kenney and Lau (1985,1986) technique, suggested that 5 samples were unstable. The results are illustrated in Figure 7.5, modified for purposes of clarity of presentation A total of forty-one GR tests were performed, for which the analysis shows the following (see Table 6.1): •  Fourteen soil samples (32 tests) that were predicted to be stable by Kenney and Lau (1985,1986) were observed to be homogeneous after sample preparation and showed no internal instability as a result of unidirectional seepage flow, as indicated by the stability of the Gradient Ratio (GR) values and by their magnitude (GR(Mod.)<1.5).  •  Three of the soil samples (6 tests) that were predicted to be unstable turned out to be stable (GR(Mod.)<1.5). This disagreement might be due to a higher severity of disturbance imposed by Kenney and Lau (1985) that included a mild vibration by manually tapping the permeameter, and a flow rate that was higher by at least an order of magnitude compared to that imposed in this study  •  The remaining two soil samples (3 tests) that were predicted to be unstable were found to segregate during sample preparation (GR(Mod)>ll). Therefore the internal instability of 130  these samples is attributed to the stirring action of the preparation method rather than the true behavior under seepage flow. From these observations it is apparent that the Kenney and Lau (1985,1986) technique developed using narrowly and widely soil samples successfully predicted the behavior of the narrowly and widely graded glass beads tested here. The attempt to extend it to gap-graded samples was successful in some cases (2 samples) and proved conservative in the others (3 samples).  7.4 Base Soil Behavior under Dynamic Conditions  7.4.1 Generalized State The dynamic conditions of vibration or cyclic flow have the potential to cause unsteady pore water pressures and/or particle migration. The manometers of the vibration device are not suited to the measurement of excess pore water pressure, hence they were used simply to characterize the sample before and after vibration. Inspection of the data reported in Appendix A l (Figures b) shows that the GR values did not change with time after each series of vibration. This and the fact that losses of particles occurred only during vibration lead to the conclusion that vibration disturbance did not cause any subsequent effects. Similarly in the cyclic flow test program seven samples that showed continuous piping during stage 4 were also tested in the static stage 5 (see Table 6.5) and found to show no change in GR during this stage and no piping (see Table 6.7). Therefore, it is concluded that like the case of vibration disturbance, the cyclic flow did not cause any subsequent effects. Cyclic flow was applied using head control (see Figure 5.2). The water head distributions characterize the behavior of each sample, and allow a better understanding of the related 131  filtration mechanisms. As illustrated for example in Figure 6.8 (see also Appendix A2 Figures a), in stage 2, performed under confinement at a frequency of 0.2 Hz, a sinusoidal variation of water head was imposed with time such that unsteady flow conditions developed in the sample (see also Appendix A2 - Figures b). This behavior reflects well the shape and frequency of a gravity wave. In contrast, as illustrated for example in Figure 6.9 (see also Appendix A2 - Figures c), in stage 3, performed under confinement at a frequency of 0.02 Hz, the appliedflowregime imposed an external head difference with the shape of a step wave. In this stage the water head distribution reached a steady state that is seen by a plateau during half a cycle. As illustrated for example in Figure 7.6, under these conditions a transient dynamic water hammer effect was observed immediately after the switching operation (every 25 sec.) by a temporary peak. This peak occurred at the bottom of the sample and was damped mainly along the lower 25 mm of the sample. The time for preliminary stabilization (tp) is defined here as the elapsed time during half a cycle until a less than 10% change in successive GR readings can be observed over a period of 1 sec. (see for example Figure 7.7 that begins at 4 seconds after the direction of flow was switched). The tp values that were derived based on stage 3 (f=0.02 Hz) are given in Table 7.1 below. It is reasonable to expect that samples of higher permeability will stabilize faster. Inspection confirms that sample s(1.3-gl34).gl49 with the highest value of permeability showed the smallest tp value and samples s(3.2-68). 103 and (70-2.2).gl03 that showed the lowest permeability showed the highest value of tp based on GR(Mod.). The tp using the GR(ASTM) value tends to be shorter than that using the GR(Mod.) value. This is attributed to the previous observation that the influence of the switching  132  operation at the bottom of the sample was mainly damped along the lower 25 mm of the sample. Table 7.1: The time for preliminary stabilization (tp, sec.) of GR within half a cycle (after 10 cycles): stage 3 GR(Mod.)  Test s(3.2-68).gl03 s(70-2.2).gl03 s(1.3-134).gl22 s(3.2-68).gl22 s(4.8-333).gl22 s(5.1-144).gl22 s(1.3-134).gl49 s(4.8-333).gl49 s(5.1-144).gl49 s(7.2-250).gl49 s(7.2-250).gl49 s(20-1.9).gl49 s(4.8-333).g290 s(7.2-250).g290  GR(ASTM)  KIT  9 15 4 10 4 8 5 4 8 3 6 6 4 3  1 E-3 (cm/sec.) 0.96 1.07 10.7 1.05 8.48 1.34 11.42 7.89 1.30 1.93 1.82 3.23 6.66 1.96  13 16 6 11 6 10 4 5 13 5 7 10 6 4  7.4.2 Quick Conditions As mentioned before a positive gradient implies downward flow and a negative gradient implies upward flow. While under confined conditions the critical gradient (to cause quick conditions) is ic =-25, unconfined conditions yield ic =-l. Theoretically, the overall gradient r  r  applied under the unconfined stage was high enough (in=4) to cause quick conditions (in>icr) under steady upward flow. Figures 7.8a and 7.8b show the variation of hydraulic gradient with time under unconfined conditions (stage 4) at different locations within the sample. Ports 2,3,5 and 6 were located along the sample and ports 1 and 7 were located above and below it (see Figure 3.6). Inspection of the trend with time shows that at t=50 sec, when flow was  133  upward, all the hydraulic gradients were consistent with quick conditions. As a result of the unsteady conditions the gradients are changing with time.  7.4.3 Cyclic Flow Permeability Figure 7.9a illustrates the differences between the externally applied and the measured head differences between ports 1 and 7 (hu). Inspection of this figure shows that: •  For each frequency (0 Hz, 0.02 Hz, 0.2 Hz), as the flow rate increased the head losses in the system increased. Based on the generalized Bernouli or Euler equation the linear relationship between the head losses and Q suggest that the head losses are due to energy 2  losses in the system. •  Under the dynamic conditions of f=0.2 Hz, in samples that yielded lower flow rates the total head difference in the system (hi7) is above the external applied one. This was attributed to the water hammer effect that under a frequency of f=0.2 Hz and lower flow rates could overcome the head losses in the system and therefore caused a net increase in the total applied head (hi7<0).  Figure 7.9b illustrates the influence of the permeability coefficients that were derived under different stages of the test (stages 1,2,3) as a function of frequency, on the measured total head in the system. Inspection of thisfigureshows that:  134  •  For a specific total head difference, changes in the conditions from steady flow (f=0 Ffz) or mainly steady flow (f=0.02 Ffz) to unsteady flow (f=0.2 Hz) could cause an increase in the permeability coefficient.  •  As the permeability increases these difference lessens.  To fully understand these phenomena, and their implications for compatibility, three qualitative models were developed. These models are presented in the following sections. The first is based on the finer fraction mobility within the soil skeleton fabric, the second on an extension of the well - recognized analogy between hydraulic and electrical trends, and the third on the water head distribution within the sample (apparent anisotropy).  7.4.3.1 Finer Fraction Mobility Thefinerloose fraction under steady flow might come to rest as shown in Figure 7.10a or under unsteady flow, remain in suspension as shown in Figure 7.10b. Therefore, compared to the unsteady flow, the structure formed by the loose fraction under steady flow may manifest itself in the measured permeability. Furthermore, theoretical calculation showed that the upward seepage force induced by the overall system hydraulic gradient of in=4 is capable of lifting a particle of diameter 0.3 mm. This diameter is larger than the maximum diameter of the finer loose fraction in all soils tested here (see Figure 4.4). Therefore, based on this model it can be expected that: •  An increase in frequency has the potential to increase the suspension period of the finer loose particles within the voids, and increase the resultant permeability (a frequency dependent behavior); and 135  •  The frequency dependent behavior will increase as the size of the finer loose fraction, which impedes flow under steady conditions, decreases.  As noted in section 7.3, based on the work of Kenney and Lau (1985), in this study the finer loose fraction of the non-gap graded soils was taken to be that below 30% finer. Taking into account the common approach in filtration criteria to treat samples with CTJ<2 as almost one fraction ( C G S , 1992; U . S . FHWA, 1995), Figure 7.11 includes all tests on samples with CTJ^2 that did not show migration of particles during stages 1 (f=0 FIz), 2 (f=0.02 Ffz) and 3 (f=0.2 Hz). It shows permeability ratios against D30 that provide two observations in support of the above hypothesis: •  A frequency dependent behavior at the higher frequency of f=0.2Hz compared to a frequency independent behavior at the lower frequency cyclic flow (f=0.02Hz); and,  •  As the D30 increases the influence of the frequency decreases.  Based on the above conceptual model and the observed behavior, Figure 7.12 illustrates few hypothesized phenomena within a certain range of D30 that can be summarized as follows: •  At a lower bound frequency of cyclic flow the behavior is frequency independent.  •  As the frequency increases above the lower bound value the relative influence increases until an upper bound value, above which an increase in the frequency does not cause further changes in the calculated permeability.  •  The ratio k(f>0 Hz)/k(f=0 Hz) changes linearly as a function of D30.  136  7.4.3.2 E l e c t r i c a l  a n a l o g y  Using resistive paper the analogy between unidirectional seepage and conduction theories is often used to interpret seepage problems under steady flow conditions. The main points of comparison are tabulated in Table 7.2. Table 7.2: Equivalencies in seepage and flow of electrical current (after Bardet 1997) flow of electrical current  Flow of water  V, voltage Conductivity i, current Ohm's law V V =0 Equipotential lines: V=constant Insulated bounbdary  Ff , total head k, coefficient of permeability Discharge velocity Darcy Law V h =0 Equipotential lines: FfT=constant Impervious boundary T  2  2  The challenge in this work is to capture the phenomenon of unsteady flow. Consequently the electrical analogy has to account conceptually for the way in which flow is imposed and the sample response. The flow regime is imposed by keeping the top of the sample under a constant head (hi), and alternating the head at the bottom (from +b.7 to -hj). As shown in Fig. 7.13, using a switch (S), this was modeled by alternating between two emfs, a fictional configuration in which therightside is constant and the left side alternates between +V and - V . Figure 7.14 illustrates the observed response of the head difference with time across the soil layer (1135). While under f=0.02 Ffz the soil column reached a steady state, under f=0.2 Ffz a steady state was not observed, and therefore the maximum value of 1135 was lower. This  137  behavior is modeled here using a capacitor. The response of a capacitor with time in charging and discharging is shown in Figure 7.15. Applying the fictional alternating emfs through a circuit including a capacitor and resistors, as illustrated in Figure 7.16, it is possible to model the overall behavior observed in the cyclic flow tests. When flow is steady the capacitor (C ) is fully charged and therefore the current is passing just through R l . During unsteady flow the capacitor is in the process of charging or discharging and therefore current is passing through R l and R2, where R2 represents the unsteady component. The measured current in the ammeter (A) represents the overall discharge velocity (or volumetric flow rate, Q ) in the system, and the measured potential difference in the voltmeter (Vc) represents the measured overall head difference across the soil layer (I135). Figure 7.17 illustrates current variations in the ammeter HA) as a function of time. In this figure, ii represents the steady (Figure 7.17a) and i the unsteady (Figure 7.17b) flow 2  component respectively. It can be seen that during the unstable periods i >0. Therefore during 2  the unstable periods the measured current in the ammeter (Figure 7.17c), is higher than ii (iA > ii). Furthermore if the frequency of the alternating potential difference (cyclic flow) is high enough to increase the effect of the unsteady component than the measured current (flow) will be higher. Conversely, if the effect of the alternating potential is smaller, than the influence of the unsteady period is smaller and as a result the measured current will be close to that observed under steady conditions (i =ii). A  In the operation of this model two physical phenomena are captured (see Figure 7.9b):  138  1. ) An increase in the frequency that causes transformation from steady to unsteady flow causes an increase in the flow rate, and therefore in the derived permeability; and 2. ) As the true Dacry's coefficient of permeability (under f=0 Hz) increases the time to reach steady conditions (tp) decreases and therefore the influence of an increase in the frequency diminishes. 7.4.3.3 A p p a r e n t  a n i s o t r o p y  For purposes of analysis, as illustrated in Figure 7.18a, the samples are divided to two layers: a top layer with head difference of hi6 between ports 1 and 6 and a bottom layer with head difference of h^j between ports 6 and 7. Flow continuity exist across cross-sectional area A, where: ki6xi].6xA=k67xi67xA  (7.3)  and ky is the permeability and ijj is the gradient between ports i and j. Observing the sample response in Figure 7.18a suggests an anisotropic behavior, given by 16 67- However, as noted in Chapter 6, observations suggest that the sample is  1  <1  homogeneous. Thus it can be concluded that the anisotropic behavior is apparent. Applying Darcy's law to each layer in turn the overall permeability (kij) is given by: kl7=Cl6/kl6+l67/k67)/0l6+l67)  (7.4)  or ki7=Constantx[(l xi )+(l xi )]' 16  16  67  67  1  (7.5)  However, the head difference was applied at the bottom of the sample (see Figure 3.10), therefore based on Figure 7.18 it can be inferred that: 139  [k67]unsteady [ 17]steady [ 16]unsteady < k  < k  ( ) 7 6  Since li6Sl0xi67, following Equation 7.4 it is apparent that ki6 has more influence on the overall permeability (k^) than k67 and therefore:  [kl7]unsteady [ 17]steady  C - )  > k  7  7  From inspection of Figure 7.18 it appears that under the applied flow regime a large portion of the head losses occur across the very bottom of the sample (layer 67). Hence it appears that: •  An increase in the sample length will lower the gradient across layer 16 and therefore increase the sensitivity to the applied unsteady flow conditions.  •  An increase in the applied gradient across the sample (between port 1 and 7) will manifest itself by a higher anisotropic behavior and therefore increase the frequency dependent behavior.  In  contrast to the above analysis of sample s(5.1-144).gl22 which yielded  ki7(0Hz)slE-5 m/sec. a similar plot (Figure 7.18b) is provided for sample s(l.3-134).gl22 which yielded ki7(0 Hz) of about an order of magnitude larger (see Table 6.6). Inspection of the water head distributions with time indicates that the apparent anisotropy in sample s(1.3134).gl22 is much less significant. These differences in turn caused changes in the derived permeability under f=0.2 Hz: compared to s(5.1-144).gl22 in which ki7(f=0.2Hz)=3ki7(0 Hz),  in sample s(1.3-134).gl22 ki7(f=0.2 Hz)=k]7(0 Hz). Based on these two selected  140  observations (see also Figures b in Appendix A2), it is concluded that samples with lower permeability have a higher damping potential and are therefore more likely to show a pronounced apparent anisotropy (frequency dependent) behavior.  7.4.3.4 Implications for Practice The observations above (Section 7.4.3) suggest that under unsteady flow conditions, soils with a lower coefficient of permeability are going to exhibit an apparent anisotropy and therefore an apparent higher permeability. This may have implication for design practice predicated on ensuring the geotextile permeability meets or exceeds that of the adjacent soil. However, the pore water pressure distribution generated by cyclic flow has not previously been recognized as a factor that should be taken into account in assessing geotextile permeability (FCGG, 1986; PIANC, 1987; CGS, 1992; U.S. FHWA - 1995). Currently, depending on the project specific conditions, in choosing the appropriate geotextile permeability a factor of safety for one or more of the following is taken into account: long term clogging and blocking; pollution during layering and while in service; compression under load; hydraulic gradient; allowable head loss; and general safety coefficient. It should be acknowledged that in many cases in practice under cyclic flow conditions it is very difficult to evaluate the exact pore water pressure distribution in the soil at the geotextile (like revetments for example). This is due to the complex mechanisms involved: the compressibility of the aerated water, the Darcy's permeability coefficient of the soil, the frequency and maximum head difference of the cyclic flow, solubility, surface tension, gradation of base soil,finerfraction mobility (as shown above), density and others.  141  In this test program soils that exhibited a true Darcy's coefficient of permeability of 1E-5 m/sec. showed at a frequency of f=0.2 Hz an apparent permeability of-6.5 times higher, which is attributed to the phenomena of an increase in the volumetric flux. Therefore, in the design process of structures that are subjected to wave action it is recommended to add a factor of safety of 10 to the existing geotextile permeability criteria. These criteria use the true Darcy's coefficient of permeability (see Table 2.7).  7.4.4  B a s e Soil  Internal Stability  U n d e r D y n a m i c  C o n d i t i o n s  As noted in Chapter 5 the extent of the applied vibration and the nature of the applied unconfined cyclic flow can be considered to be severe. This was done in an attempt to impose a maximum degree of disturbance above which the behavior will be insensitive to any further increase in the disturbance and therefore ensure that the observed phenomena will allow for conservative observations compared to less severe conditions. Kenney and Lau (1985,1986) developed a technique intended to assess internal stability under unidirectional flow but not to address explicitly issues of vibration or cyclic flow. As noted before (section 7.3) this technique provided an excellent assessment under unidirectional flow of the narrowly and widely graded samples tested but had mixed success with the gap graded samples. In this work a new technique is being developed for the conditions of vibration and cyclic flow. Followingfromthe work of Kenney and Lau (1985, 1986) it is assumed that the sample may comprise twofractions:the primary soil fabric and the finer loosefraction.The maximumfinerloosefractionis taken to be thefractionbelow 30%finerfor narrow and wide soils and thefractionbelow the gap for gap graded soils where the gap is at 45% orfiner(see section 7.3).  142  A new framework is suggested for the analysis of the test results. It is illustrated in Figure 7.19a and was applied to both the vibration and cyclic flow results. The common approach, for which the outcomes are independent of the number of cycles and sample length, involves two steps: Step 1, migration  of particles  out of the soil sample: In this step, the size of the  particles that pass through the geotextile is compared to the original gradation of the original sample. •  If the sample exhibits a continuous loss of particles that are considered part of the primary fabric then the sample is categorized as piping. Conversely, if these particles are not part of the primary fabric then the sample is considered to exhibit internal instability.  •  If the sample is not showing a continuous loss of particles, then if instantaneous loss of the primary fabric is observed the sample is considered to show piping otherwise the sample is examined under step 2. Step 2, migration  of particles within the soil samples: If the finer loose fraction  migrates within the sample then the sample is internally unstable; otherwise the sample is internally stable.  7.4.4.1  V i b r a t i o n  As shown in Chapter 6, three tests including soil samples s(15-2.7) and s(15-4.5) showed segregation during sample preparation and were termed initially non-homogeneous. Sample preparation included manual stirring and therefore it is believed that even if these soil samples were initially homogeneous, they would eventually have become to be unstable as a result of the applied dynamic conditions. Therefore soil samples s(15-2.7) and s(15-4.5) were  143  classified as unstable under dynamic conditions. The internal stability of the rest of the samples (21 tests) that were homogeneous and found to be internally stable following unidirectional flow is examined here (see Figure 7.19a and 7.19b) under dynamic conditions.  7.4.4.1.1 Step 1: Migration of Particles Out of the Soil Sample In thefirstpart of this two part analysis the shape of the m -N curves was considered, D  b  where m is the cumulative mass per unit area of the passing through particles during the D  dynamic stage and N is the cumulative number of blows. To recognize that m <250g/m  2  b  D  constitutes an insignificant loss (see section 6.2.1.2.3), 8 tests that gave m <250g/m were 2  D  categorized as not showing continuous or instantaneous loss of particles and will be examined later in step 2. The rest of the tests (13 tests for which in which m >250g/m ) are examined in 2  D  Figure 7.20 (Note: sample s(6.4-306).g290 is plotted in Figure 7.20a and in Figure 7.20b to emphasize the scale differences between these two graphs). As illustrated in Figure 7.21, inspection of the data shows two characteristic responses. In one response there is a rapid initial loss of particles followed by a moderate to insignificant loss with further blow counts (Figure 7.21a). The second response (Figure 7.21b) indicates a continuous loss with blows that may increase or remain comparable to that which occurred initially. Therefore the response illustrated in Figure 7.21a is not considered to exhibit a continuous loss of particles, and in contrast, the response shown in Figure 7.21b is termed a "continuous loss of particles".  In order to quantify these responses a Secant Piping Ratio, (SPR) is introduced: US) SPR% = j^xl00 where: 144  (Eg. 7.8)  I(S) - The secondary slope of the m -Nb curve (arbitrarily taken here as the slope between D  900 and 3780 blows). I(P) - The preliminary slope of the m -Nb curve (arbitrarily taken here as the slope between 0 D  and 180 blows).  For the 13 tests (including one repeated test) shown in Figure 7.20, the Secant Piping Ratio is plotted against m in Figure 7.22. Inspection of the data suggests that there are three trends, D  characterized by: •  low Secant Piping Ratio (SPR) and m values (8 tests) suggesting that there was an D  insignificant loss of particles that occurred just at the first series of vibration, •  60%>SPR>50% and m >10,000 g/m (4 tests). This group would appear to have a 2  D  potential to lose a significant quantity of material, but exhibit a declining loss with number of blows (SPR< 100%); and •  one test that exhibited a clear increase in loss of particles (SPR>100%), though the low value of m suggests that the magnitude of this loss is relatively small. D  Based on these observations only a Secant Piping Ratio (SPR) greater than 30% is considered a response indicative of a continuous loss of particles. Applying this criterion tests s(1.939).gl03i , s(20-3.1).gl03, s(45-3.1).gl03 and s(6.4-306).g290 are categorized as showing >2  continuous loss of particles.  To address the issue of whether the migrating particles constitute the primary fabric of the soil, the close to the largest particle diameter (D95) of the passing through particles, which is termed here D95p, is compared to the original gradation of the sample. This comparison for 145  the five samples that exhibited a continuous loss of particles shows the following (see Table 7.3): •  As the D95p of sample s(6.4-306) is smaller than D30, it is considered to have lost a portion of thefinerloosefractionand therefore to be unstable.  •  D95p of soil samples s(20-3.1) and s(45-3.1) is approximately equal to Dgap (the diameter at the beginning of the gap). These samples lost their finer loose fraction (located below the gap) and are unstable.. •  •  D95p of soil sample s(1.9-39) is essentially equal to D95 of the original sample, suggesting that this soil sample is susceptible to piping.  Table 7.3: A comparison of D95p and that of the original sample: vibration tests Test  D95p (1E-3 mm) s(1.9-39).gl03 60 s(20-3.1).gl03 65 s(45-3.1).gl03 65 s(6.4-306).g290 105 "Piping (see explanation above) 1>2  Original sample ~D95 -Dgap -Dgap <D30 a  Tests that exhibited a low m and SPR values (8 tests) and therefore did not show loss of D  the primary soil fabric or continuous loss of thefinerloosefractionare further examined in step 2.  7.4.4.1.2 Step 2: Migration of Particles Within the Sample In order to decide whether the increase in the Gradient Ratio (GR) was due to non homogeneous densification or due to the migration of particles within the soil sample the observations in this step accounted for changes in the following: void ratio; permeability along the sample; and GR (Mod.). Inspection of the collective data (see chapter 6 and 146  Appendix A l ) shows an increase in the GR(Mod.) just in 6 samples. Their behavior with vibration is divided into two main categories based on their characteristics when densification essentially did not take place anymore (see for example Figure 7.23 for Nb>900 blows): 1. Four tests (s(4.3-97).gl03; s(6.4-213).gl03; s(4.3-97).g290; and s(6.4-213).g290) in which no further changes were observed. 2. Two tests (s(6.4-306).gl03 and s(6.4-306).149) in which, k  sg  (k^ and k67) decreased  while k (k35) remained constant. s  To characterize the changes in Gradient Ratio (GR) with vibration, a Secant Blinding Ratio (SBR) is introduced where: KS) SBR% = -7-7 x 100  KP)  (7.9)  where: I(S) - The secondary slope of the GR-Nb curve, (taken here as the slope between 900 and 3780 blows). I(P) - The preliminary slope of the GR-Nb curve, (taken here as the slope between 0 and 900 blows). Note: The GR was found to be rather resistive to the initial vibration, therefore the initial slope during the first 900 blows was taken rather than just during thefirst180 blows. Figure 7.24 is intended to check the relationship between GR(Mod.) and SBR. Inspection of thisfiguresuggests that: •  As GR increases SBR increases, 147  •  Short duration of vibration (NI = 900 blows) caused just minor changes of less than 0.5 in GR(Mod.).  Assuming no densification, internal migration of the finer loose fraction toward the geotextile with vibration involves a decrease in the permeability at the vicinity of the geotextile (IC57,  k67) and therefore an increase in GR. In association, the permeability of the upstream  soil layer (k35) is expected to increase. However, layer 35 is twice as long as layer 57 which in turn is about three times longer than layer 67. Therefore compared to layer 57, and 67, the permeability of layer 35 is expected to be less sensitive. Inspection of the GR(Mod.), void ratio (e), and permeability (kjj) values, in the two categories of the 6 samples that exhibited an increase in the GR(Mod.) with vibration (see Appendix A) suggest the following: •  Samples in the first category exhibited lower GR and therefore (as shown above) lower SBR values (GR(Mod.)<1.7). Once the densities of these samples essentially stop changing no further changes in the permeability along the sample and thus in the GR were observed. Therefore it is concluded that the insignificant changes in the GR were a result of densification. Based on these observations soil samples s(4.3-97) and s(6.4-213) are classified as stable.  •  Samples in the second category include tests using the same soil sample with two different geotextiles. These samples exhibited higher GR and SBR values and a decrease in kgg even when densification did not take place anymore, which suggests that these samples are  148  susceptible to blinding. This outcome that classifies soil sample s(6.4-306) as unstable is in agreement with the outcome of step 1.  7.4.4.2 Cyclic Flow Based on the same framework explained in section 7.4.4 (see Figure 7.19a), the process of evaluating the internal stability of the soil samples tested in the cyclic flow device was comprised of two steps: migration of particles out of the sample; and migration of particles within the sample. These steps are described below (see Figure 7.19c).  7.4.4.2.1 Step 1: Migration of Particles Out of the Soil Sample As mentioned in Chapter 6, based on visual observations through the collector trough and the mass of the passing through particles the behavior of the samples in cyclic flow can be divided into two groups: 1) samples that did not show continuous loss of particles and that lost less than 250g/m (~2g) of particles, and 2) samples that showed continuous loss of 2  particles and lost more than 2500g/m (~20g) of particles. These characterizations are applied 2  at each stage of the cyclicflowtests. Stage 2, (cr =25 kPa, f=0.2 Hz): From the seventeen tests that were investigated in this v  stage, two that possessed the largest gap width ratio (R=3.1), showed continuous loss of particles (s(20-3.1).gl03 and s(45-3.1).gl03). In addition the D95p of these samples was approximately equal to the Dgap of the original soil sample (see Table 7.4). These observations suggest that these samples were losing theirfinerloose fraction and therefore soil samples s(20-3.1) and s(45-3.1) were classified as unstable under these conditions.  149  Table 7.4: A comparison of D95p and that of the original sample: stage 2 D95p (1E-3 mm) 65 65  Test s(20-3.1).gl03 s(45-3.1).gl03  Original sample -Dgap -Dgap  Stage 3, (cr =25 kPa,f=0.02 Hz): In all the 15 tests that were performed under this stage v  no loss of particles was observed. This suggests that the range of the frequencies applied in stages 2 and 3 (f=0.2Hz to f=0.02Hz) did not play an important role in the stability of the soil sample. Stage 4, (<J =0 kPa, f=0.2 Hz): From the 15 soil samples investigated, 9 visually showed v  continuous loss of particles. A comparison of the D95p and the gradation of the original soil is given in Table 7.5, and leads to the following: Narrowly and widely graded soils: •  Tests using soil samples s(l.3-134) and s(3.2-68) showed D95p>D30, and in fact even D95p>D50. Therefore it is concluded that they lost the primary soil fabric (skeleton) and can be classified as showing piping.  •  Tests using soil samples s(4.8-333) and s(7.2-250) showed D95p<D30. Therefore these soil samples lost theirfinerfraction and were classified as unstable  Gap graded soils: In tests using soil samples s(20-1.9) and s(70-2.2) it was observed that D95pEDgap. However, the gap location in soil sample s(70-2.2) is at 70% finer which indicates that the coarser fraction is floating in thefinerfraction. Therefore the fraction that passed through the geotextile is the primary soil fabric and this combination is classified as showing piping. Conversely, in soil sample s(20-1.9) thefinerloose fraction which is at 20%  150  finer isfloatingin the voids of the coarsefraction.Therefore thefractionthat passed through the geotextile is thefinerloosefractionand this soil sample is considered to be unstable. Table 7.5: A comparison of D95p and that of the original sample: stage 4 Test s(3.2-68).gl03 s(20-1.9).gl03 s(70-2.2).gl03 s(1.3-134).gl22 s(3.2-68).gl22 s(1.3-134).gl49 s(20-1.9).gl49 s(4.8-333).g290 s(7.2-250).g290 "Piping ^ot available  D95p (1E-3 mm) 65 65 65 150 80 165  Original sample >D30 -Dgap -Dgap >D30 >D30 >D30 a  3  a  a  a  b  105 105  <D30 <D30  Tests that as a result of the analysis performed in step 1 did not show piping or internal instability will be investigated under step 2. 7.4.4.2.2 Step 2: Migration of Particles Within the Soil Sample From the tests investigated under this step just four were observed visually to experience migration of particles within the soil sample. These tests were s(4.8-333).gl22, s(4.8-333).gl49, s(7.2-250).gl49, and s(7.2-250).gl49 showed during stage 4 migration of 2  finer particles toward the top of the sample. The migration of the finer particles toward the top of the sample in stage 4 rather than toward the bottom could be explained by the boiling conditions during upward flow. These conditions increased the ability of finer particles to move upward rather than downward. Based on these behavior soil samples s(4.8-333) and s(7.2-250) were classified as unstable.  151  7.4.4.3 Criteria Development As indicated in Tables 7.6 and 7.7, five soil samples were internally unstable under vibration, two under confined cyclic flow, and three under unconfined cyclic flow. However, no instantaneous loss of the primary soil fabric (see Figure 7.19a) or instantaneous development of blinding layer (see Section 7.4.4.1.2) was observed. Therefore, it can be concluded that for filtration design a short duration of dynamic disturbance like an earthquake should be treated as static conditions. Table 7.6: The soil samples stability: vibration tests Dynamic stage Soil sample Test code N P s(1.9-39) s(1.9-39).gl03i N s(1.9-39) P s(1.9-39).gl03 N s(2.2-52) S s(2.2-52).gl03 N s(2.2-78) S s(2.2-78).gl03 N s(4.3-97) S s(4.3-97).gl03 W s(6.4-213) S s(6.4-213).gl03 W s(6.4-306) s(6.4-306).gl03 uba G s(15-2.7) s(15-2.7).gl03 u G s(20-1.5) S s(20-1.5).gl03 G s(20-3.1).gl03 s(20-3.1) u G s(45-3.1) s(45-3.1).gl03 u G s(70-1.6) S s(70-1.6).gl03 N s(2.2-52) S s(2.2-52).gl22 N s(2.2-78) S s(2.2-78).gl22 W s(6.4-213) S s(6.4-213).gl22 b G s(15-2.7) s(15-2.7).gl22 u G s(70-1.6) S s(70-1.6).gl22 N s(4.3-97) S s(4.3-97).gl49 W s(6.4-306) s(6.4-306).gl49 ua G s(20-1.5) S s(20-1.5).gl49 N s(4.3-97) s(4.3-97).g.290 S W s(6.4-213) S s(6.4-213).g290 s(6.4-306) w s(6.4-306).g290 u G s(15-4.5) s(15-4.5).g290 ub "Blinding susceptibility initially non-homogeneous S - Stable; U - unstable; P - piping. N w ^ Q ^jjg j gap-graded respectively 2  ;  ;  n a r T 0 W  a n (  152  T a b l e 7.7: T h e soil samples stability: continuous cyclic flow Soil sample  Test code  N  Stage 2 S  s(3.2-68) s(3.2-68).gl03 G s(20-1.9) s(20-1.9).gl03 S G s(20-3.1) s(20-3.1).gl03 u G s(45-3.1) s(45-3.1).gl03 u G S s(70-2.2).gl03 s(70-2.2) N S s(1.3-134) s(1.3-134).gl22 N s(3.2-68) s(3.2-68).gl22 s N s(4.8-333) S s(4.8-333).gl22 N S s(5.1-144) s(5.1-144).gl22 N s(1.3-134) s(1.3-134).gl49 s N S s(4.8-333).gl49 s(4.8-333) N S s(5.1-144) s(5.1-144).gl49 W S s(7.2-250) s(7.2-250).gl49i W s(7.2-250) S s(7.2-250).gl49 G s(20-1.9) S s(20-1.9).gl49 N S s(4.8-333).g290 s(4.8-333) W s(7.2-250) s(7.2-250).g290 s S - Stable; U - unstable; P - piping. "Stage was not performed. Not available N wG^ ^ Q i ( j j gap-graded respectively 2  Stage 3 S  Stage 4 P  s  U  a  a  a  a  s s s s s s s s s s s s s  P P P  u s p  u s u u b  u u  b  ;  ;  n a r r o w  w  e anc  Under severe dynamic conditions the constriction sizes are changing, causing a maximum increase in the potential mobility of the finer loose fraction. Hence, it is assumed here that the internal stability of the soil samples under these conditions is independent of the disturbance type (vibration or cyclic flow). A new technique for predicting the internal stability of soil samples, is developed here. As described below, this is done separately for the narrowly and widely graded soil samples and for the gap graded soil samples.  153  7.4.4.3.1 The "Concave Degree" (CD) for Narrow and Wide Gradations The influence of the shape of the grain size distribution curve on the stability of narrowly and widely graded soils under conditions of unidirectional flow is widely recognized in existing criteria (Kenney and Lau, 1985; Lafleur et al., 1989). However, as noted previously, the conditions for which those criteria were developed do not include those of severe vibration or cyclic flow. Indeed, as can be observed from Table 7.8, The Kenney and Lau (1985, 1986) criteria do not address the performance of those samples found to be internally unstable under severe dynamic conditions. Therefore the new data accumulated here are used to examine the characteristic differences between the internally stable and unstable samples. T a b l e 7.8: A comparison between the predicted performance of n a r r o w l y a n d widely graded soil samples based on the K e n n e y a n d L a u (1985,1986) technique and the observed performances Soil sample s(1.3-134) s(1.9-39) s(2.2-52) s(2.2-78) s(3.2-68) s(4.3-97) s(4.8-333) s(5.1-144) s(6.4-213) s(6.4-306) s(7.2-250)  N N N N N N N N W W  Cu  Prediction based on K & L  1.3 1.9 2.2 2.2 3.2 4.3 4.8 5.1 6.4 6.4 7.2  S S S S S S S S S S S  Analyzed Performance P P  S S P  S  u  S S  w "For both vibration and cyclic flow S - Stable; U - unstable; P - piping. N w ^ Q j gap-graded respectively ;  ;  n a r r o w  anc  154  u u  8  Inspection of Table 7.8 suggest that the coefficient of uniformity has a role to play: those samples with CTJ<4 were always stable, and some of those samples with CTJ>4 were unstable. Coupling that with inspection of the shape of the grain size distribution curves (Figure 7.25) it would appear that the performance is also influenced by the concavity of the gradation profile. In order to quantify the relation between the sample performance and the gradation profile, a Concave Degree (CD) procedure is suggested here for further investigation (see Figure 7.26): Step J: Draw a secant line between the lowest point of the concave (Pi %) and the highest point on the concave (Ph %). Step 2: Find the middle of this line, (P % = Pi+(P -Pi)/2) a  h  Step 3: Determine the corresponding D (mm) to P (%) a  a  Steps 4: Based on the gradation profilefindthe corresponding Pi (%) to D (%) a  Step 5: CD=P /Pi a  According to this technique CD<1 for concave downward profile, CD^l for linear profile, and CD>1 for concave upward profile. Using the "CD" procedure for soil samples with CTJ>4, Figure 7.27 illustrates the boundary between stable and unstable soil samples. From inspection of Figure 7.27 it can be observed that, independent of the disturbance type, internally stable soils lie below the CD=2 line and internally unstable soils above it. Furthermore, this observation is in agreement with the assumption made here (Section 7.4.4.3) that the internal stability under severe dynamic conditions is independent of the 155  disturbance type (vibration or cyclic flow). Based on these observations the following criterion is suggested here: CTJ<  4 or CTJ>4 and CD<2  Cu>4andCD>2  =>  =>  internal stability internal instability  (7.10) (7.11)  7.4.4.3.2 Gap-Graded Samples Under cyclic flow the unconfined conditions are considered to be more severe than confined conditions, therefore soil samples that were unstable under the confined conditions (stage 2) are also considered for the purposes of analysis to be unstable under unconfined conditions (stage 4). Inspection of the results in Tables 7.6 and 7.7 leads to the following observations: •  sample s(20-1.5) was stable in the vibration test program.  •  samples s(15-2.7) and s(15-4.5) were unstable in the vibration test program  •  sample s(20-l .9) was unstable in the cyclic flow test program.  •  samples s(20-3.1) and s(45-3.1) were unstable in both the vibration and cyclic flow test programs.  Furthermore, as explained above (section 7.3), it is assumed that the fraction below the gap is the maximum finer loose fraction just in soil samples where the gap is at 45% finer or less. Inspection of Figure 7.28 which summarizes the above observations, suggests the following:  156  •  samples with a gap location between 15% and 45% finer and with R>1.5 (R - gap width ratio) were unstable, thus it can be expected that all soil samples with gap location smaller than 15%finerand with R>1.5 will be unstable, implying that: R>1.5 and 45%finer>Gaplocation => unstable  •  (7.12)  a sample with a gap location at 20%finerand R=1.5 is stable, thus it can be expected that all soil samples with a higher gap location and a narrower gap width ratio will be stable, implying that: R< 1.5 and Gap location>20%finer=> stable  (7.13)  Note: no data exist yet for R<1.5 and Gap location<20% finer.  Thefindingthat the gap width ratio has a major influence on the stability of the soil sample is in agreement with the work of Honjo et al. (1996). They observed that soils having a gap width ratio less than 2.8 are stable and as the gap width ratio increases to 4, the migration of fines through the pores of the coarser skeleton become significant. However, while in this work the conclusions are based on mobility of the finer loose fraction, Honjo et al. (1996) based their criterion on settlement considerations.  7.5 Gradation of the Passing Through Particles The mass and gradation of the passing through particles are a very useful index of filtration compatibility. Specifically in this section the gradation is addressed. As noted by Fischer et al. (1996), during sieving, larger particles may become trapped in channels with smaller constrictions (Figure 7.29a) and smaller particles may pass through channels with larger constrictions (Figure 7.29b). Therefore, they concluded that the largest particle can 157  provide an index of the largest geotextile opening size, but the grain size distribution is not a good index of the geotextile pore size distribution. Indeed, in routine characterization of opening size, either in the Apparent Opening Size (AOS) or the Filtration Opening Size (FOS) tests, the D95 of the passing through particles (D95p) is chosen as the characteristic geotextile opening size 095 (see Figure 7.30). For the purpose of analysis here the same approach is adopted. However, the D95p is defined as the Performance Opening Size (POS) since it is derived based on energy levels applied using thefiltrationperformance test. Inspection of Tables 6.9 and 6.10 suggest that the POS (or D95p) after preparation were very similar to that measured after applying dynamic conditions. The consistency in the POS values as a result of these stages is attributed to the high severity of disturbance. With the exception of tests including soil sample s(l.3-134) it was also observed that the POS was smaller than the Filtration Opening Size (FOS) values. This observation can be explained by differences in the mobility of the particles: in sieving-like methods the particles have some flexibility to move across the geotextile until reaching a channel with a larger openings. Conversely, in the GR performance test their mobility is more restricted. This in turn decreases the chances of the particles tofinda larger channel and pass through it.  Sample s(l.3-134) yielded the largest POS values (Table 6.10). This is attributed to its very low CTJ. The significant influence of CTJ on the observed behavior is supported by the following: •  Bhatia et al. (1994) found that 095 (close to the largest opening of the geotextile) from Filtration Opening Size (FOS) tests can be 60-75% of 095 obtained from Apparent Opening Size (AOS) tests. This is attributed to fact that while the FOS method is 158  performed using a mixture of fractions (higher Cu) the AOS is performed using separate fractions (lower Cu)•  The state of practice in North America (CGS, 1992; U.S. FHWA, 1995) treats soil samples with Cu<2 as almost one fraction suggesting a more stringent filtration criterion for retention.  •  Watson and John (1999) found that for Cu>2 the larger particles act as catalysts for bridge formation, but at Cu<2 any bridge formation is dependent upon particle interlocking. It can be postulated that the dynamic disturbance applied in this work caused movements of particles that in turn released the interlocking system.  7.6 Soil/Geotextile Compatibility For the conditions of cyclic flow, Tondello (1998) and Cazzuffi et al., (1999) suggested erosion limit states with reference to the mass of soil passing through the filter and with respect to both effective stresses and maximum applied gradient (see Figure 7.31). However as noted previously (see Table 7.7) under the cyclic flow conditions imposed here (see Figure 7.31) two of the tests were internally unstable under stage 2, seven were internally unstable under stage 4, and five piped during stage 4. This observation implies that design criteria for cyclic flow should consider in addition to the applied effective stresses and cyclic gradients also the gradation of the base soil and the geotextile characteristics. In design, the relationship between the geotextile Filtration Opening Size (FOS) and the base soil is evaluated using an interaction ratio (I ): R  I =FOS/(CxDi)  (Eg. 7.14)  R  159  Where, C - Constant D i - The indicative grain size (usually D85, D 5 0 or D30) Using the interaction ratio, in the following sections the soil/geotextile compatibility is examined for narrow and wide gradations under static, vibration, cyclic-confined and cyclicunconfined conditions. In addition an interpretation is performed to assess the response o f gap graded soils for unconfined conditions.  7.6.1 Static Conditions  7.6.1.1 Piping During the static stage in both the cyclic and vibration test programs no continuous loss o f particles was observed. The total mass o f passing through particles was found to be less than 1% o f the original gradation (see Tables 6.4 and 6.7), which had been noted to be less than the piping boundary o f 2500 g/m established by Lafleur et a l , (1989). Therefore it is 2  concluded that none o f the samples showed piping. A comparison is performed between the experimental observations made here and some common Filtration Opening Size (FOS) criteria (see Table 7.9). This was done for each criterion using the IR value and presented graphically in Figure 7.32.  160  Table 7.9: The Interaction ratio (I ) values for static unidirectional conditions: piping R  Source IR French Committee of Geotextiles and FOS/(0.64D85) Geomembranes (FCGG, 1986) FOS/(0.8D85) Canadian Geotechnical Society (CGS, FOS/(1.5D85) 1992) FOS/(3D85) Ontario Ministry (OMT, 1992)  of  Uniform (CTJ<6)  Broadly graded (CTJ^6) Uniform (CTJ<6)  Broadly graded (CTJ^6)  Transportation FOS/D85  University of British Columbia (UBC, FOS/(1.5D85) Fannin et al., 1994a) FOS/(0.2D85) Lafleur (1998, 1999)  Comments  FOS/(D85) FOS/(D50) FOS/(D30) FOS/(5D30)  1<CTJ<2 3<CTJ<7 CU<6  6<CTJ linearly graded 6<CTJ concave upward stable 6<CTJ concave upward unstable  The slopes of the curves in Figure 7.32 are related to the Constant (C) given in equation 7.14 (as shown for example using UBC criteria for 3<CTJ<7). In order to prevent piping the Interaction ratio (IR, Table 7.9) should be smaller that 1. From inspection of Figure 7.32 it is apparent that the CGS (1992) predicts well the behavior of almost all samples. It can also be seen that the UBC criteria for CTJ>2 is conservative: This is because a much smaller mass of soil  (=500  g/m2) was selected as a threshold value for initiation of piping. The OMT  (1992) and Lafleur (1998, 1999) criteria are also a bit conservative as the FOS/D85 increase. Interestingly, applying Lafleur (1989) criterion for the onset of piping (2500 g/m ) to 2  those results reported by Fannin et al., (1994a) for soils with 6>CTJ>3 yields a criterion in very good agreement (see Figure 7.33) with that of OMT (1992) and Lafleur (1998, 1999).  As discussed above, in filtration criteria using AOS values it is common to treat soil samples with CTJ<2 as almost one fraction (CGS, 1992; U.S. FHWA, 1995). This approach suggests a more stringent criterion for these soil samples. Inspection of the data collected by Fannin et al., (1994a) and in this work (see Figure 7.34) suggests, that in agreement with the CGS (1992) and Fannin et al., (1994a), the actual onset of piping occurs at FOS/D85=1.5. However, due to the potential for catastrophic piping associated with these soil samples (CTJ<2) for design purposes the OMT (1992) and Lafleur (1998, 1999) criteria (FOS<D85) are recommended for both uniformly and broadly graded soils.  7.6.1.2 Blinding As mentioned previously the general term, "blinding," is used here collectively to describe the general phenomenon of a decrease in soil/geotextile permeability (k ) as a result sg  of clogging, blocking or specifically blinding. The tendency toward blinding in the static stage was monitored through visual observations and the GR values. As noted in section 6.2.1.1 in all tests the GR values were observed to be less than 1.4 and essentially constant with time, therefore it was concluded that none of the tests showed a susceptibility to blinding. As noted in Table 7.10 and plotted in Figure 7.35, to compare the results obtained in this study to different common Filtration Opening Size (FOS) criteria, the Interaction ratio (I ) was again R  used.  162  T a b l e 7.10: T h e Interaction ratio values (I ) f o r static unidirectional conditions: R  blinding Source  IR  FOS/(4D15)  French Committee of Geotextiles and Geomembranes (FCGG, 1986) F6S/(D30) Lafleur (1998) "For soils from which fines can easily be put in suspension For broadly graded soils with gradation curves that are markedly concave upward or that show a gap below 30% passing a  6  b  In order to prevent blinding the retention ratio should be larger than one. Inspection of the data points indicates that: 1.) both criteria are essentially conservative, 2.) Lafleur (1998) criterion is less conservative than FCGG (1986). 7.6.2 Vibration Disturbance  7.6.2.1 Piping  As mentioned above (see Table 7.6), under vibration two tests (see Figure 7.19a) that included one soil geotextile combination s(1.9-39).gl03i and s(1.9-39).gl032 were categorized as susceptible to piping. Otherwise the remaining tests did not show a continuous loss of the primary soil fabric and were categorized as not susceptible to piping. As noted in the literature review there are no design criteria for filtration compatibility under the application of mechanical vibration. In the absence of such criteria the test data from the vibration stage are compared to the same unidirectional flow criteria in Figure 7.36. As before an interaction ratio (Table 7.9) less than one suggests compatibility for piping. Inspection of Figure 7.36 suggests that piping is well described by the existing criteria. It is apparent that some criteria were more conservative than others and that the OMT (1992), 163  CGS (1992) and Lafleur (1998, 1999) criteria describe better the onset of piping. However, due to the potential for catastrophic piping associated with uniform soils  (CTJ^),  for design  purposes the OMT (1992) criteria (FOS<D85) is recommended for both uniformly and broadly graded soils.  7.6.2.2 Blinding The tendency toward blinding during vibration was observed through visual observations, the Gradient Ratio (GR) values, and an unsatisfactory stabilization of the GR(Mod.)-Nb curve. As noted previously, two of the tests that include narrow and widely graded soil samples were found to be susceptible to blinding during vibration (see Table 7.6). The results have been summarized Figure 7.37 with respect to the Secant Blinding Ratio. D30 has been selected as an index to be consistent with the observations of Kenney and Lau, (1985), (see section 7.3) and the analysis performed here. Inspection of Figure 7.37 suggests that the blinding tests fall below the threshold value given by Equation 7.15.  F0S/D3O1  (7.15)  This criterion for vibration is identical to that of Lafleur (1998) for blinding under unidirectional flow without vibration. However, while the Lafleur criterion was suggested for broadly graded soils that qualitatively possess a concave upward shape, in this work also narrowly graded soils have been found to be unstable. Furthermore a quantification of the concave upward degree was introduced here (see section 7.4.4.3.1) and it was shown that  164  soils with CTJ>4 and CD>2 are likely to be internally unstable. Therefore, for the purposes of design it is suggested that Equation 7.15 be applied for soils with Cu>4 and CD>2.  7.6.3 Cyclic Flow Disturbance  7.6.3.1 Piping From experimental observations of the mass of the passing through particles, gradation of the retained soil, gradation of the passing through particles, and whether a continuous loss of particles was observed, a retention criterion for cyclic flow conditions is developed. As reported in the literature review there are some existing FOS criteria for conditions of cyclic flow. Figures 7.38 to 7.42 illustrates them graphically for confined (see Table 7.11) and unconfined conditions (see Table 7.12). T a b l e 7.11:  T h e interaction ratio (I ) for confined cyclic flow R  Source French Committee of Geotextiles and Geomembranes (FCGG, 1986)  IR  FOS/(0.6D85)  conditions  Comments Uniform (CTJ<6)  a  FOS/(0.75D85) Canadian Geotechnicai Society (CGS, i992) Geo synthetics Technology Center (Mlynarek et al., 1999)°  b  Broadly graded (Cu>6)  FOS/(D15) FOS/(D85)  Uniform (CTJ<6)  FOS/(D90)  Broadly graded (CTJ>6)  Dense and confined soils Where pulsating flow is large, the geotextile should be sufficiently open to prevent blow up and should be weighted down. This criterion should not be used for pulsating load such as in highways and railways where the flow is small and cyclic loads are large. Non severe hydraulic conditions (Mlynarek, 1999 - personal communication) a  b  c  165  Table 7.12: The interaction ratio (I ) for unconfined cyclic flow conditions R  Source French Committee of Geotextiles and Geomembranes (FCGG, 1986)  IR FOS/(0.38D85)  Comments Uniform (CTJ<6)  FOS/(0.48D85)  Broadly graded  a  (Cu>6) Geosynthetics Technology Center (Mlynarek et al., 1999) Loose or unconfined soils  FOS/D50  b  a  b  Severe hydraulic conditions (wave attack or pumping)  7.6.3.1.1 Confined Conditions The tests performed in this work under confined conditions (ov=25 kPa) did not show piping. Using these tests Figures 7.38, 7.39, and 7.40 are prepared to illustrate the behavior under confined conditions. It is apparent that the FCGG (1986) criterion trends to be conservative as FOS/D85 increases and that the CGS (1992) criterion is too conservative. However the Mlynarek et al. (1999) criterion for non-severe hydraulic conditions predicts well the response of the tests. This criterion is essentially the same criterion as the one suggested by the OMT (1992) for static unidirectional flow. Furthermore, as during both static unidirectionalflow and the confined cyclic flow the particles that make the primary soil fabric are not mobile, it is suggested to use the criterion that was recommended here for unidirectional flow (OMT, 1992, see Table 7.9) also for confined cyclic flow.  7.6.3.1.1 Unconfined Conditions Inspection of Figures 7.41 and 7.42 suggest that both the FCGG (1986) and Mlynarek et al. (1999) criteria predict well the response of the tests. In an attempt to further to improve upon these criteria the data are reported again in Figure 7.43. It can be observed that in order to fulfill the retention demand Equation 7.16 should be fulfilled. 166  FOS/D85<0.5  (7.16)  Bhatia et al. (1994) found that for geotextiles with Apparent Opening Size (AOS) larger than 0.1mm, the FOS values are smaller by 25% to 40% than the AOS values (AOS=0.600.75AOS). To compare the data obtained here to the well recognized criterion suggested by Holtz et al. (1997), (0.5D85>AOS) an approximation was made to transform the AOS values to equivalent FOS values. Inspection of Figure 7.43 suggests that the FOS approximation of Holtz et al., (1997) is reasonably consistent with the data obtained here. 7.6.3.2 Blinding No blinding layer was visually observed. This observation could be expected as under a very high frequency of cyclic flow if the finer loose fraction migrate cyclically, it clogs and opens pores. Though in practicefiltersare often subjected to different frequencies of cyclic flow, where during the lower frequency, the steady component can promote a blinding layer. Inspection of differences between piping criteria that were suggested here for vibration and unconfined cyclic flow suggests, that during cyclic flow particles that migrate toward the geotextile are much more likely to pass through it. Hence, it can be concluded that for the conditions of cyclic flow it will be conservative to use the blinding criterion that was recommended here for the conditions of vibration. Therefore in order to prevent blinding it is recommended to apply the Lafleur (1998) criterion that was recommended for steady flow also for unstable soils under cyclic flow. For practicality purposes the instability of the soils under confined and unconfined cyclic flow can be examined using the CD procedure (see section 7.4.4.3.1) that was developed here. Note: soils that are classified as unstable under unconfined cyclic flow can conservatively be classified as unstable under confined cyclic flow.  167  7.6.5 Summary of the Suggested Criteria: Narrow and Wide Gradations The design criteria recommended here uncouple between the potential mobility of the primary soil fabric and the migration of the loose fraction. Furthermore these criteria take into account the gradation of the base soil and the geotextile characteristic opening size. 7.6.5.1 Static and Confined Dynamic These criteria should be applied in cases where the particles that form the primary soil fabric are fully restrained. •  Apply Equation 7.19 against piping. FOS<D85  •  (7.19)  Apply Equation 7.20 against blinding if the base soil is internally unstable: to evaluate the internal stability of the base soil under steady flow apply Kenney and Lau, (1985, 1986) procedure, and to evaluate the internal stability of the base soils under confined cyclic flow apply the CD procedure developed here (see section 7.4.4.3.1) FOS>D30  (7.20)  7.6.5.2 Unconfined Dynamic These criteria should be applied in cases where the particles that form the primary soil fabric are not fully restrained. •  Vibration disturbance: Apply Equation 7.19 against piping.  •  Cyclic flow disturbance: Apply Equation 7.21 against piping. F0SO.5D85  168  (7.21)  •  Apply Equation 7.20 against blinding if the base soil is internally unstable: To evaluate the internal stability of the base soils apply the CD procedure developed here (see section 7.4.4.3.1).  Note: In cases where it is hard to fulfill both piping and blinding criteria, Gradient Ratio performance tests that examine the conditions in the field are recommended.  169  16  Before external blinding layer  O —  After external blinding layer  Flow direction  III  .,,,.,,•,,,<  E  External blinding layer (less than 1mm thickness)  Q  20 40 Water head h ^ (cm)  Figure 7.1: Using recycled water, the change in the water head distribution along s(4.397).gl03 as a result of the external blinding layer  GR a.) time  GR b.) time Figure 7.2: Schematic influence of the external blinding layer on the GR: a.) top external blinding layer, and b.) bottom external blinding layer  170  10 •  8H TJ O  o  s  rZ" O  c  s  6  • • • — • • •  of  4  n  c5  i i — ' — i — ' — i 10000 15000 20000 25000 Elapsed time t, (min.)  1  I  1  I I Nlllj  10  1  I TTTTTTj  1 I J I IJ111  1  I I  100 1000 Elapsed time t, (min.)  1  10000  I I M||||  5000  100000  1  b.)  a  Figure 7.3: The variation of the GR(Mod.) with time (redrawn after Nishigata et al.,  2000) 1000 —i  —  0  — | — 800  ~  01 O  600  H  400  H  "  After preparation After stabilization  - © •  After 180 blows  —#-  After 1440 blows  200  0.1  0.2  0.3  F O S , (mm)  Figure 7.4: The variation of G R with blow count (initially non- uniform samples)  171  40.00  s(20-3.1)  •O-  s(45-3.1) s(15-2.7)  Q  D c c CO  o £  20.00 - \  Ii  -O-  s(15-4.5)  X  s(20-1.9)  Mr Stable  •5  £ t> 1/J  t  0.00-  1  Unstable  -20.00 • 0.00  10.00 20.00 Mass fraction smaller than D, F (%)  30.00  Figure 7.5: Using Kenney and Lau (1985,1986) technique to evaluate the internal stability of the gradations tested  s(5.1-144).g122 h67  0—  h56 h35  A—  500  505  510 515 Elapsed time t, (sec.)  520  h23  525  Figure 7.6: The transient dynamic water hammer effect in the sample (after 10 cycles)  172  8 s(5.1-144).g122  500  505  510 515 Elapsed time t, (sec.)  520  Figure 7.7: The influence of the change in theflowdirection on the GR values (after 10 cycles and 4 seconds, f=0.02 Hz).  173  174  -10-,  Permeability: k yj 10 -3, (cm/sec.) A  b.) Figure 7.9: The influence offlowrate and permeability on the measured head difference  175  a  b  Figure 7 . 1 0 : The influence of the cycling flow on the arrangement of the finer fraction within the voids: a.) steady flow, b.) unsteady flow  4  -(-  a=k (f=0.2Hz);b=k (f=0Hz)  O  a=k 17 (f=0.02 Hz); b=k  17  17  1  7  (f=0 Hz)  - i  +  000  0.10  0.20  0.30  D30 (mm)  Figure 7 . 1 1 : The influence of the finer fraction size on changes in the permeability, as the cyclicflowfrequency increases (observed) 176  a=k 1 (f>0 Hz); b=k 7  1 7  (f=0 Hz)  D30 (mm)  Figure 7.12: The influence of the finer fraction size on the changes in the permeability as the cyclingflowfrequency increases (hypothesized)  -V  Left (Alternated)  J  Right (Constant)  +v  Figure 7.13: Applying the fictional alternating emfs to the circuit  177  525  530  535 540 Elapsed time t, (sec.)  545  550  50  55  a.)  60 65 Elapsed time t, (sec.)  70  75  b.)  Figure 7.14: The head difference changes across the soil layer (h35) during 25sec: a.) f=0.02HZ, b.) f=0.2HZ  Elapsed time, t  Elapsed time, t  a.)  b.)  Figure 7.15: The potential change on a capacitor with time: a.) charging, b.) discharging  178  rWV\A C  R2  W  W  —  1  I  r  Figure 7.16: Modeling the overall cycling flow behavior, through resistors, capacitor and an alternating fictional emfs circuit  +  a.)  b.)  c.)  Figure 7.17: The measured current (i ) on the ammeter (A) during charging or A  discharging periods 179  a.  K16  E  a CO  b r  j " - - '  -40  K 7 6  0 40 Water head h j , (cm)  b. Figure 7.18: The water head distributions at different time steps within half a cycle f=0.2 Hz): a.) s(5.1-144).gl22, b.) s(1.3-134).gl22 180  21 vibration tests 16 cyclic flow tests Step 1: Continuous loss of particles ? No  Yes  Instantaneous loss of the primary soil fabric ?  Potential loss of the primary soil fabric ? No  Yes  Yes  1r  Piping 0 tests  Step 2:  r Piping 2 vibration tests 5 cyclic flow tests  Particles migrating ? No  Internal stability 14 vibration tests 2 cyclic flow tests  Yes  Internal instability 2 vibration tests 4 cyclic flow tests a.)  181  No  r Internal instability 3 vibration tests 5 cyclic flow tests  Step 1: MD>250 g/m and SPR>30% ? 2  No  Yes  M >2500 g/m and {D95p>D30 or D95p>Dgap}? 2  D  D95p>D30 or D95p>Dgap ? No  Yes  Yes  1r  r  r  Piping  Step 2:  No  Piping  Internal instability  Inspection of the GR(Mod.), void ratio (e), and permeability (k ) values suggest particles migration? y  No  Internal stability  Yes  Internal instability  b.)  182  Step 1: M >2500g/m ? 2  D  No  Yes  D95p>D30 or D95p>Dgap ?  D95p>D30 or D95p>Dgap ? No  Yes  Yes  r  r  Piping  Step 2:  No  Piping  r  Internal instability  Visual observation suggest particles migration ? No  Yes  Internal stability  Internal instability  c.)  Figure 7.19: The process of evaluating the internal stability of soil samples that were tested: a.) generic, b.) vibration tests, c.) cyclic flow tests 183  250  +  • 3780 blows  200  or o.  co  150  ~  100  100%  +  +  30%  r~  — 1  0  i  n  1  —i—'—i  10000 20000 30000 40000 50000 Mass per unit area of passing through particls  m D • (9  ure 7.22: The m -SPR relationship D  185  /m2  )  0.50 W—  s(6.4-213).g103  f—  s(6.4-306).g103  4-,  s(6.4-213).g103 s(6.4-306).g103  01 o TJ  e o  i  1 r 1000 2000 3000 Number of blows, N ^ 1  1  4000 1000 2000 3000 Number of blows, N ^  a.)  4000  b.)  0.006 s(6.4-213).g103  0.005  0.004  0.006  0.003  0.005 H  I rl  0.002  1  1  1  1000 2000 3000 Number of blows, N  1  1 4000  0.004 Number of blows, N i  d.)  c)  Figure 7.23: The influence of vibration on the void ratio, gradient ratio, and permeability 186  80 - ,  £  O  60-  co w  +  N =0  O  N  b  = 900  b  2 o> 40 H  + + ° o + o + + o o  20 - \  + +  I  0.0  I  +  Q  °  oo + o  t-  I  0.5  1  ,  1.0  1— 1.5  ~ l 2.0  Gradient ratio, GR(Mod.)  Figure 7.24: The influence of the GR value on SBR 100 - ,  80  2  H  40  Grain size, (mm)  Figure 7.25: The influence of the concavity of the gradation profile on the analyzed performance: for soil samples with Cu>4  187  Grain size (mm)  Figure 7.26: The concave degree ( C D ) procedure: soil sample s(6.4-3.6)  VIBRATION T E S T S  +  Stable  •  CYCLIC FLOW T E S T S  •  5-1  o o  Unstable  Stable  •  Unstable  •  o o> o TJ 3 '  •  n  1  1  '  r  5 6 7 Coefficient of uniformity, C u  Figure 7.27: The boundary between stable and unstable gradations (CTJ>4): C D procedure  188  VIBRATION TESTS + A  Stable Unstable CYCLIC FLOW TESTS  •  0  10  Unstable  20 30 Gap location [% finer]  Figure 7.28: Internal stability: gap graded samples.  189  40  100  -  90  -  80  -  70 -  Mass finer  60 50 40 30 20 10 n  1  y  U — I 0.0  .  . J  .  D  9  5  P =  P  O  S  r*=—i—i—i  i i i i'[ 1 0.1 Grain size (mm)  1  1—  i 1.0  Figure 7.30: The gradation of the passing through particles: soil sample s(3.2-68)  15 —  i=4, 0 < a  v  < 25 kPa  10 Unstable  /  / / / / / /  •  0  I 0  Stable  I  '  50 Vertical stress a  v  100 , (KPa)  '  I 150  Figure 7.31: The conditions applied here compared to one of Tondello (1998) limit states  190  o  •  o + A  0.0  0.4  F C G G 1986 CGS 1992 OMT 1992 U B C 1994 Lafleur 1998  0.8 1.2 FOS/D85  1.6  2.0  Figure 7.32: The interaction ratio (I ) based on different piping criteria with the R  current test results from static unidirectional conditions 4500 - ,  0.5  1.0 1.5 FOS/D85  2.0  2.5  Figure 7.33: Applying Lafleur (1998) criteria for the onset of piping (2500 g/m ) to the 2  results obtained by Fannin et al., (1994a) using soils with 3<CTJ<6  191  •0—  0.5  Fannin et al. (1994a)  1.0 1.5 FOS/D85  2.0  2.5  Figure 7.34: Filtration behavior of soils with CTJ<2 using the results obtained by Fannin et al., (1994a) and in this work  +  Lafleur (1998)  o or  2  FCGG (1986)  H  o  5  o  c o  •8 2  o  + p  o  FOS/D  1  5  Figure 7.35: The interaction ratio (I ) based on the different blinding criteria with the R  current test results from static unidirectional conditions 192  o  F C G G 1986  •  CGS1992  o +  8-i  OMT 1992 U B C 1994  A  Lafleur 1998  +  at  a  o  1  "co  I  4-  V  CO  fe  0.0  0.4  0.8 1.2 FOS/D85  1.6  2.0  Figure 7.36: The interaction ratio (I ) based on the different piping criteria with the R  current test results from vibration tests  +  Non blinding susceptibilty  %  Blinding susceptibility  80  iE.  60—|  of m co  c  40  TJ  o  +  +  20  "  i 2  1  i  1  3 FOS/D30  r 4  Figure 7.37: The influence of the FOS/D30 on the SBR values  193  Non piping susceptibility  -f2.0  1.6 -  +  DC  •S 2 c .2 t> S <u  "  1 2  0.8 •  +  +  + +  0.4  0.0  -" 0.0  '  1  1 '  0.2  1  I  1  0.4 0.6 FOS/D85  1  I  0.8  1.0  Figure 7.38: Using F C G G (1986) criterion for dense and confined soils on the collected data -f5.0  Non piping susceptibility  -i  +  4.0 •  + +  ce o"  3.0  2 c o  '« 2  +  + 2.0 •  1.0  H 0.0 -i  0.0  + +  +  1 0.2  +  J-  1  1  1  1  0.4 0.6 FOS/D85  1  1  1  0.8  Figure 7.39: Using CGS (1992) criterion on the collected data  194  I  1.0  -\-  Non piping susceptibility  2.0-,  1.6 •  .2 2 c o  1-2-1  2  0.8 -  ts -f+  +  0.4 •  ++ 0.0  —  0.0  ++  1  i  1  -  0.4 0.6 FOS/D85  0.2  0.8  1.0  Figure 7.40: Using Mlynarek et al. (1999) criterion for non-severe hydraulic conditions on the collected data  2.5 -  o  Piping susceptibility  +  Non piping susceptibility  o  2.0  (9  or o  •8 2 o c  o  1.5-1  +  o  1.0 - f -  0.5  +  H  0.0  ++ "i  0.0  1  0.2  r-  1  I  0.4 0.6 FOS/D85  1  I 0.8  '  I 1.0  Figure 7.41: Using F C G G (1986) criteria for loose or unconfined soils on the collected data  195  o +  Piping susceptibility Non piping susceptibility  2.0 - ,  O 1.6  H  o  or o  1.2 H  2  0.8  o  + +  o o  + +  0.4  0.0 0.0  - , — I — , — I — I — I — I — I 0.2 0.4 0.6 0.8 FOS/D85  I  I 1.0  Figure 7.42: Using Mlynarek et al. (1999) for Severe hydraulic conditions (wave or pumping) on the collected data  o +  Piping susceptibility Non piping susceptibility  I.O-i  O O  0.8 H  in 0.6 co O W  o  oo  H  +  -4=-  0.4 H  +  0.2  0.0 0.0  ,  1 , 1 1 0.1 0.2 Filtration Opening Size, FOS (mm)  1 0.3  Figure 7.43: The retention criteria using the D85 as an indicative grain size. 196  8. SUMMARY AND CONCLUSIONS  8.1 Objectives The assessment of filtration compatibility is based on empirical design guidance. There are extensive criteria available for design. These criteria are largely limited to conditions of static unidirectional flow. The challenges infiltrationcompatibility include: 1.) soils that are internally unstable, 2.) conditions of mechanical disturbance and, 3.) hydraulic conditions that vary from those of unidirectional flow. Therefore, in this work using the Gradient Ratio test as a performance test, a path of investigation has been taken to assess filtration compatibility by contrasting dynamic conditions that include mechanical vibration and cyclic flow against those of routine static unidirectional flow. Specifically the objectives of this research were as follows: •  Adding an automatic vibration component and associated mechanical controls to an existing gradient ratio device. This component should be able to allow the performance of filtration tests under energy and frequency controlled vibration;  •  Develop, design and commission a new cyclic gradient ratio device, and associated computerized control system, to performfiltrationtests under a head-controlled cyclic flow regime;  •  Taking into account the existing test standards, conduct new test procedures for vibration loading in the vibration device and cyclic flow loading in the cyclic flow device;  •  Extend methods of interpretation based on the measurements taken from both static and dynamic conditions;  197  •  Comprehensively investigate the filtration-related phenomena under both static and dynamic conditions from the instrumentation on the test devices and related theoretical considerations;  •  Examine existing filtration criteria for soil-geotextile compatibility and independently validating new ones;  •  Compare and contrast the results of this work with design practice. It should be noted that the primary objectives of this research were related to the  development and commissioning of the cyclic flow device, and the analysis of the results using it. To achieve these objectives, the new device and procedures were used to perform head control filtration tests with different soil/geotextile combinations under static, vibration and cyclic flow disturbances. 8.2 Apparatus An existing Gradient Ratio (GR) device was modified to allow performance of vibration tests and collection of the passing through particles at any stage. Furthermore a new cyclic Gradient Ratio device and associated computerized control system was developed and designed. Both devices can be used to evaluate the compatibility of different soil/geotextile combinations under static and dynamic conditions of vibration and cyclic flow.  8.3 Materials Forty-one different combinations of model soils (glass beads) and nonwoven geotextiles were used. Twenty-four combinations were tested in the vibration device and seventeen in the cyclic device. The model soils were of narrow, wide and gap-graded gradations in the range of coarse silt to fine sand. 198  8.4 Test Procedures Taking into account the existing test standards, new multi-stage programs of testing were conducted. The programs of testing were performed under an externally imposed gradient of four and were divided into static and dynamic stages. In the static stage unidirectional flow was imposed. In the dynamic stage vibration was imposed in the vibration device and cyclic flow was imposed in the cyclic flow device. The results of each test were interpreted from the measurements of flow rate, water head distribution along the sample, visual observations through the transparent device, and the weight and gradation of the passing through particles.  8.5 Analysis and Design Implications  8.5.1 General Behavior  8.5.1.1 Blinding Layer and the Gradient Ratio Value (Static Unidirectional Flow) For the purpose of analysis the general term, "blinding," was used here to describe the overall phenomenon of a decrease in the sample combination permeability (k^) as a result of clogging, blocking or specifically blinding. In addition the terms internal and external blinding layer were introduced. The internal blinding layer is related to the mobility of thefinerloose fraction within the pores of the primary soil fabric and the external blinding layer is related to the environmental conditions imposed in the laboratory. The external blinding layer developed just under static unidirectional flow.  199  The investigation that was carried out under static unidirectional flow yielded the following: •  It was demonstrated that the external blinding layer can have a significant influence on the soil/geotextile performance if it can develop on top of the internal blinding layer. Therefore in laboratory modeling of field performance the potential development of an external blinding layer should be considered  •  Observations confirmed the modified Gradient Ratio to be a more sensitive index than the ASTM Gradient Ratio.  •  To evaluate the time needed for the Gradient Ratio test to be performed a simple equation is suggested here. This equation is based on the assumption that the permeation time needed is the time for one pore volume exchange.  8.5.1.2 Vibration Inspection of the results showed that changes in the Gradient Ratio values and passage of particles through the geotextile occurred just during vibration. These observations suggest that vibration disturbance has an effect only during its application.  8.5.1.3 Cyclic Flow  8.5.1.3.1  HEAD  DISTRIBUTION  Observing the water head distribution under the cyclic flow characterizes the behavior of the samples: •  At a frequency of 0.2 Hz a sinusoidal distribution of water head with time was imposed during which unsteady flow conditions developed in the sample. This behavior reflects 200  well the shape and frequency of gravity waves. In contrast under a frequency of 0.02 Hz, a step wave was imposed during which the water head distribution reached, periodically, a steady state. •  Immediately after the flow direction was changed, a temporary peak was observed reflecting a water hammer effect. This effect occurred at the bottom of the sample and was damped with time along the bottom 25 mm of the sample.  8.5.1.3.2 P E R M E A B I L I T Y  A comparative analysis of the permeability coefficients, the flow rates, and the water head distributions, yields the following: •  Head losses in the system increased with increasing flow rate, due to energy losses associated with flow.  •  For a specific total head difference, changes in the conditions from laminar steady flow or mainly steady flow (f=0.02 Hz) to unsteady flow (f=0.2 Hz) tended to cause an increase in the permeability. However this phenomenon diminished as the permeability increased. To fully understand this behavior and its implications, three qualitative models were developed. Thefirstis based on thefinerfraction mobility within the soil skeleton fabric, the second on extension of the well - recognized analogy between hydraulic and electrical characteristics, and the third on trends in the water head distribution within the sample (apparent anisotropy).  •  To account for the pore water pressure redistribution, in designing for a wave action loading, it is recommended here to add a factor of safety of 10 to the existing geotextile permeability criteria (FCGG, 1986; PIANC, 1987; CGS, 1992; U.S. FHWA - 1995).  201  8.5.1.4 The Passing Through Particles To be consistent with the standardized methods to determine the characteristic opening size, the close to the largest diameter (D95) of the passing through particles was chosen to be the characteristic opening of the geotextile. Using this value, termed here the Performance Opening Size (POS), the following observations are made: •  The Performance Opening Sizes were essentially smaller than the Filtration Opening Sizes that were derived based on hydrodynamic sieving. This was attributed to the restricted mobility of the migrating particles in the Gradient Ratio Device to move through the sample.  •  Tests that included soil samples with the lowest coefficient of uniformity yielded the largest Performance Opening Size values. This is in agreement with design practice approach (CGS, 1992; U.S. D.O.T FHWA, 1995) and previous research works (Bhatia et al., 1994; and Watson and John 1999).  8.5.2 Internal Stability The term internal stability is related to the potential mobility of thefinerloose fraction within the primary fabric. In internally stable soils thefinerloose particles are trapped in the primary fabric; and in internally unstable soils thefinerloose particles can migrate within the primary fabric.  8.5.2.1 Static Conditions The Kenney and Lau (1985,1986) technique for assessing internal stability was developed using soil samples with particle sizes in the range between sand and gravel. However, this technique successfully predicted the behavior of the narrowly and widely 202  graded soil samples tested here, which were constituted from particle sizes in the range between coarse silt andfinesand.  8.5.2.2 Dynamic Conditions The degree of disturbance under vibration and unconfined cyclic flow was chosen in an attempt to impose conditions above which the behavior is insensitive to any further increase in the disturbance severity. This level of disturbance increases the potential mobility of the finer loose fraction within the primary soil fabric. •  Independent of the number of cycles or blows and sample length a new framework of analysis for both vibration and cyclic flow was suggested here to distinguish between internal instability and piping.  •  It was concluded that infiltrationdesign a short duration of dynamic disturbance like an earthquake should be treated as static conditions.  •  A new technique to predict the internal stability of soil samples based on their gradation profile, was developed here for dynamic conditions. This technique can also be used for static conditions. As described below this was done separately for the narrow and wide graded soil samples and for the gap graded soil samples: •  Narrow and wide gradations: Under static conditions internal instability is associated just with very wide gradations. However in this study it was found that under severe dynamic conditions, narrow gradations also have the potential to be internally unstable. Therefore a "Concave Degree" (CD) procedure that accounts for the concavity of the grain size distribution curve was developed here.  203  •  Gap graded soils: No results or criterion exist yet to classify the internal stability of gap graded soils under severe and continuous dynamic conditions. Based on the results obtained in this study, boundaries for internal stability states are suggested. It appears that the ratio between the diameter at the coarser end of the gap to the diameter at the finer end the gap, has a major influence on the internal stability.  8.5.3 Soil-Geotextile Compatibility The geotextile should provide retention to the upstream soil particles (piping criteria) without inducing an impediment to the flow (blinding criteria). Based on a large database of accumulated test information, the retention demand is well established for static conditions. However for dynamic conditions the design practice is lacking as is evident in the provisions for vibration and cyclic flow. •  Vibration - No research was found in the literature, which specifically investigated the influence of vibration and therefore, none of the criteria that have been developed address the condition of vibration.  •  Cyclic flow - Very little was done in studying the soil/geotextile compatibility in the laboratory under the conditions of cyclic flow. Therefore, some of the existing criteria were established to be very conservative and can be difficult to fulfill (for example, CGS, 1992). In addition the existing criteria do not couple between confining stresses and maximum hydraulic gradient and do not explicitly address unstable soils.  204  The Filtration Opening Size (FOS) design criteria suggested here for dynamic conditions (Table 8.1) couple the combined influence of effective stresses and maximum applied gradient. Furthermore these criteria address gap-graded gradations and unstable soils. Table 8.1: Suggested retention criteria for narrow and wide gradations Criteria Conditions FOS<D85 Static, vibration, confined cyclic F0SO.5D85 Unconfined cyclic FOS>D30 All conditions "the particles that form the primary soil fabric are fully restrained. the particles that form the primary soil fabric are not fully restrained. for internally unstable soils. To evaluate the internal stability of the base soils under static conditions of steady flow apply Kenney and Lau, (1985, 1986) procedure, and under cyclic flow or vibration apply the Concave Degree procedure developed here. 3  b  0  b  c  8.5.4 Implications for Research and Practice The contributions of this work to current research and design practice are as follows: The major contributions include: •  The development and commissioning of a new cyclic flow gradient ratio test device to model the filtration conditions imposed by wave actions on offshore structures. The cyclic head control hydraulic regime is imposed using three constant head tanks. One middle tank routed to the top of the sample and the other upper and lower tanks are routed through a switching valve to the bottom of the sample.  •  A proposed new framework of analysis to distinguish between internal instability and piping based on the gradation of the retained soil sample, gradation and mass of particles that passed through the geotextile, and migration of particles within the retained soil sample. This framework was validated by the experimental work.  205  •  A proposed new technique to evaluate the internal stability of soils under dynamic conditions based on the curvature of the grain size distribution curve, that is validated by results from 3 internally stable and 3 internally unstable soils.  •  An evaluation of 6 existing soil retention and blinding criteria suggested for design practice that showed limited success.  •  A proposed soil retention and blinding criteria for static and dynamic conditions that was validated by the experimental results.  •  An assessment of the influence of pore water pressure distribution, induced by cyclic flow, on the volumetric flux that shows a greater flux under unsteady cyclic flow compared to unidirectional flow. An additional factor of safety for geotextile permeability is considered appropriate for such conditions.  The minor contributions include: •  The demonstration of external blinding as a potentially significant influence on filter performance, if it develops on top of the internal blinding layer.  •  The observation that the D95 of particles passing through the geotextile in the Gradient Ratio Test was essentially smaller than that reported from hydrodynamic sieving. This was attributed to the restricted mobility of the migrating particles in the Gradient Ratio device to move within the sample. Such conditions might also govern in manyfieldapplications.  8.6 Recommendations for Further Studies This study has involved design and development of equipment, the use of model soil/geotextile combinations and investigation offiltrationphenomena under both static and  206  dynamic conditions to make recommendations for design practice. The following recommendations are made for further research: •  It was found that the environmental conditions could have an effect on the Gradient Ratio value. Therefore it is recommended that in addition to the algaecide treatment, future investigations will include removable water filters.  •  The characteristic opening sizes of the geotextiles that were derived based on the Gradient Ratio tests were different than those obtained from standard procedures. The geotextile characteristic opening sizes duringfieldperformance should be further addressed under different vertical stresses, and flow regimes.  •  It was concluded that the imposed hydraulic conditions under unsteady flow should be considered in dictating the geotextile permeability. However, to improve the factor of safety suggested here for design, the influence of different flow regimes on the volumetric flux should be further addressed.  •  To reflect on the homogeneity of the sample it is recommended to add an unidirectional (steady) flow stage after each stage of cyclic flow.  •  For narrowly and widely graded soils made of coarse silt to fine sand a procedure was developed to measure the Concave Degree" (CD) which quantify the influence of the gradation profile on the internal stability of the soil. However the validity of this procedure should be further investigated for gradation profiles with larger uniformity coefficients. Furthermore using this procedure and the pattern suggested for gap graded soils, research is needed to evaluate boundary states of internal stability under different dynamic conditions that include for example pulsating loads.  207  The soil/geotextile filtration criteria suggested here were derived using nonwoven geotextiles. It is recommended that these criteria be extended for use with woven geotextiles. This can be done according to the research methodology performed here. To further investigate the influence of other dynamic conditions in filtration applications, more data should be accumulated as a function of: 1.) gradient and confining stresses, 2.) gradation profile, and 3.) disturbance frequency.  208  BIBLIOGRAPHY  Adel, H.D, De Bruin, P., and Hofrnan, G. 1996, "Influence of Stresses on the Characteristic Opening Size," Proceedings of Geofilters'96, Montreal, Quebec, Canada, May 29-31, 1996, pp. 157-166.  ASTM D4491-85 Standard Test method for water permeability of geotextiles by permittivity, Annual Book of ASTM standards, Vol. 04.09, ASTM Philadelphia.  ASTM D4533-85 Standard Test Method for trapezoid tearing strength of geotextiles, Annual Book of ASTM standards, Vol. 04.09, ASTM Philadelpia.  ASTM D4632-91 Standard Test Method for Grab Breaking Load and Elongation of Geotextiles, Annual Book of ASTM standards, Vol. 04.09, ASTM Philadelpia.  ASTM D5101-96. Standard test method for measuring the soil-geotextile clogging potential by the gradient ratio, Annual Book of ASTM standards, Vol. 04.09, ASTM Philadelpia.  ASTM D5199-98 Standard Test Method for Measuring Nominal Thickness of Geotextiles and Geomembranes, Annual Book of ASTM standards, Vol. 04.09, ASTM Philadelphia.  ASTM D5261-92 Standard Test Method for Measuring Mass per Unit Area of Geotextiles, Annual Book of ASTM standards, Vol. 04.09, ASTM Philadelphia.  Austin, D.N., Mlynarek, J., and Blond, E. 1997, "Expanded Anti-Clogging Criteria For Woven Filtration Geotextiles," Proceedings of Geosynthetics' 97 Conference, Long Beach California, USA, March 11-13, 1997, Vol. 2. pp. 1123-1144.  209  Bardet, J.P. 1997, "Experimental soil mechanics, " Prentice-Hall Inc., 583 p.  Bear, J. 1972, "Dynamic of Fluids in Porous Media," American Elsevier publishing company, Inc. New York, London, Amsterdam, 764 p.  Bell, A.L., Mccullough, L.M., and Snaith, M.S., 1982, "An Experimental Investigation of Sub-Base Protection using Geotextiles," Proceedings of Second International Conference on Geotextiles. Las Vegas, U.S.A., August 1-6, 1982, Vol. 2, pp. 435-440.  Berendsen, E., and Smith, G.M., 1996, "Design Criteria For Geotextile on Sand and Cohesive Soils," Proceedings of Geofilters'96 Conference, Montreal, Quebec, Canada, May 29-31, 1996, pp.439-448.  Bertram, G.E. 1940, "An Experimental Investigation of Protective Filters," Harvard Soil Mechanics Series No. 7, Harvard University, No. 267, 21 p.  Bezuijen, A., Breteler, M. K. and Bakker, K J . 1987, "Design Criteria for Placed Block Revetment and Granular Filters," Proceedings of the Second International Conference on Coastal and Port Engineering in Developing Countries, Beijing, Chiana, September 1987, pp. 1852-1866  Bhatia S.K., Smith J.L., and Christopher, B.R. 1994, "Interrelationship Between Pore Opening of Geotextiles and Methods of Evaluation," Proceedings of the Fifth International Conference on Geotextiles and Related Products, Singapore, 5-9 Sept, Vol. 2, pp. 705-710.  Bhatia, S.K. and Huang, Q. 1995, "Geotextile Filters for Internally Stable/Unstable Soils," Geosynthetics International, Vol. 2, No. 2, pp. 537-565.  210  Bhatia, S.K., Moraille, J., Keith, K., and Smith, J.L. 1996, "Comparative Performance of Granular Vs. Geotextile Filters In the Laboratory," Proceedings of Geofilters'96 Conference, Montreal, Quebec, Canada, May 29-31, 1996, pp.449-460.  Bhatia, S.K., and Smith J.L. 1996, "Geotextile Characterization and Pore-Size Distribution: Part 1, A Review of Manufacturing Processes," Geosynthetics International, Vol. 3, No. 1, pp. 85-105.  Breteler, K. M., and Verhey, H. 1990, "Erosion Control by Hydrodynamically Sandtight Geotextiles," Proceedings of the Fourth International Conference on Geotextile, Geomembranes and Related Products, Hague, Netherlands, May 28-June 1, 1990, pp. 385-389.  Breteler, K.M., Smith M. G., and Pilarczyk, K.W. 1994, "Performance of Geotextile on Clay, Silt and Fine Sand in Bed and Bank Protection Structures," Proceedings of the Fifth International Conference on Geotextiles Geomembranes and Related Products, Singapore, September 5-9, 1994, Vol. 2, pp. 771-774.  Carroll, R.G. 1983, "Geotextile Filter Criteria," Transportation Research Record, No. 916, pp. 46-53.  Cazzuffi, D.A,, Mazzucato, A., Moraci, N., and Tondello, M. 1996, "A New Test Apparatus For The study of Geotextile Behavior As Filters In Unsteady Flow Conditions," Proceedings of Geofilters'96 Conference, Montreal, Quebec, Canada, May 29-31, 1996, pp. 183-191.  Cazzuffi, D.A., Mazzucato, A., Moraci, N., and Tondello, M . 1999, "A New Test Apparatus For The study of Geotextile Behavior As Filters In Unsteady Flow Conditions: Relevance and Use," Geotextiles and Geomembranes, Vol. 17, No. 56, pp. 313-329.  211  CGS, Canadian Geotechnical Society 1992, "Canadian Foundation Engineering Manual," 3 edition. BiTech Publisher Ltd., Richmond, B.C, p. 512. rd  Canadian General Standard Board, CGSB-148.1 No. 2-M85 Method of testing Geotextiles and Geomembranes-Mass per unit area, National Standards of Canada, CGSB, Sixth Draft, July 1992.  Canadian General Standard Board, CGSB-148.1 No. 3-M85 Methods of Testing Geotextiles and Geomembranes-Thickness of geotextiles, National Standards of Canada, CGSB, Sixth Draft, July 1992.  Canadian General Standard Board, CGSB-148.1 No. 10-M85 Method of Testing Geotextiles and Geomembranes-Filtration Opening Size of Geotextiles, Proposed National Standard of Canada, CGSB, Sixth Draft, July 1992.  Chin, Y.M., Nikraz, H. R., and Press, M. 1994, "Laboratory Filtration Performance of Nonwoven Geotextiles," Proceedings of the Fifth International Conference on Geotextiles Geomembranes and Related Products, Singapore, September 5-9, 1994, Vol. 2, pp. 635-638.  Christopher, B.R., and Holtz, R.D. 1985, "Geotextile Engineering Manual," Course text, prepared for Federal Highway Administration National Highway Institute, Under Contract No. DTFH61-80-C-00094, Washington, D.C.  Christopher, B.R. and Holtz, R.D. 1989, "Geotextile Engineering Manual," Report No. FHWA-TS-86/203, U.S. Federal Highway Administration, Washington, DC.  Christopher, B.R. 1998, "The First Step in Geotextile Filter Design," Geotechnical Fabrics Report, Vol. 16, No. 6, pp. 36-38.  212  Christopher, B.R., and Fisher, GR. 1992, "Geotextile Filtration Principles, Practices and Problems," Geotextiles and Geomembranes, Vol. 11, No. 4-6, pp. 337-353.  Christopher, B.R., and Valero, S.N. 1999, "Thirty Years of Evaluation of a Geotextile Filter," Proceeding of Geosynthetics '99 Conference, Boston, Massachusetts, USA, April 28-30, Vol. 2, pp. 977-990.  Craig, R.F. 1995, "Soil Mechanics," Published by Chapman and Hall, London, UK, p 427.  Fannin, R.J., Vaid, Y.P., and Shi, Y.C. 1994a, "Filtration Behavior of Nonwoven Geotextiles," Canadian Geotechnical Journal, Vol. 31, No. 4, pp. 555-563.  Fannin, R.J., Vaid, Y.P., and Shi, Y.C. 1994b, "A Critical Evaluation of the Gradient Ratio Test," Geotechnical Testing Journal, Vol. 17, No.l, pp. 35-42.  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 128, Editors Shobha K. Bhatia and L David Suit, pp. 100-112.  Fischer, G.R., Christopher, B.R., and Holtz, R.D. 1990, "Filter Criteria Based on Pore Size Distribution," Proceedings of the Fourth International Conference on Geotextile, Geomembranes and Related Products, Hague, Netherlands, May 28June 1, 1990, Vol. 1, pp. 289-294.  Fischer, G.R. 1994, "The Influence of Fabric Pore Structure on the Behavior of Geotextile Filters," Doctoral Dissertation, University of Washington, 502 p.  213  Fischer, G.R, and Holtz, R.D. 1996, "A Critical Review of Granular Soil Filter Retention Criteria," Proceedings of Geofilters'96 Conference, Montreal, Quebec, Canada, May 29-31, 1996, pp. 409-418.  Fischer, G. R., Holtz, R. D., and Christopher, B. R. 1996, "Evaluation of Geotextile Pore Structure, " Recent Developments in Geotextile Filters and Prefabricated Drainage Geocomposites," ASTM STP 128, Editors Shobha K. Bhatia and L David Suits. ASTM., pp. 3-18  Fischer, G. R., Mare, A. D., and Holtz R. D. 1999, "Influence of Procedural Variables on the Gradient Ratio Test," Geotechnical Testing Journal, Volume 22, No. 1, pp. 22-31.  Fluet Jr. J. E., and Luettich, S. M., 1993, "Geotextile Filter Criteria for Gap Graded Silty Sands," Proceedings of Geosynthetics '93 Conference, Vancouver, Canada, March 30-Apr. 1, 1993, pp. 469-480.  Fourie A.B., and Addis, P. 1999, "Changes in Filtration Opening Size of Woven Geotextiles Subjected to Tensile loads," Geotextiles and Geomembranes, Vol. 17, No. 5-6, pp. 331-340.  FCGG, French Committee of Geotextiles and Geomembranes, 1986, "Recommendations for the Use of Geotextiles in Drainage and Filtration System," Geotextile Manual, Institute Texile de France, Boulogne-Billancourt, France, pp. 43-54.  Giroud, J.P. 1980, "Introduction to Geotextiles and their Applications," Proceedings of the First Canadian Symposium on Geotextiles, Calgery, Alberta, Canada, Sept. 23, 1980, pp. 3-31.  214  Giroud, J.P. 1982, "Filter Criteria for Geotextiles," Proceedings of the Second International Conference on Geotextiles, Las Vegas, USA, August 1-6, 1982, Vol. 1, pp. 103-108.  Giroud, J.P. 1996,  "Granular Filters And Geotextiles  Filters," Proceedings  of  Geofilters'96 Conference, Montreal, Quebec, Canada, May 39-31, 1996, pp.565680.  de Graauw, A.F., van der Meulen, T., and van der Does de Bye, M . 1984, "Granular Filter: Design Criteria," Journal of Waterway, Port, Coastal and Ocean Engineering, ASCE, Vol. 110, No. 1, pp. 80-96.  Haliburton, T . A , and Wood, P.D., 1982, "Evaluation of the U.S. Army Corps of Engineers Gradient Ratio Test for Geotextile Performnce," Proceedings of the Second International Conference on Geotextiles, Las Vegas, USA, August 1-6, 1982, Vol. 1, pp. 97-101.  Heerten, 1982, "Dimensioning the Filtration Properties of Geotextiles Considering Long Term Conditions," Proceedings of the Second International Conference on Geotextiles, Las Vegas, USA, August 1-6, 1982, Vol. 1, pp. 115-120  Heerten, G. and Wittmann, L. 1985, "Filtration Properties of Geotextiles and Minerals Filters Related to River and Canal Bank Protection," Geotextiles and Geomembranes, Vol. 2, No. 1, pp. 47-63.  Heerten, G. 1986. "Functional Design of Filters using Geotextiles," Proceedings of the Third International Conference on Geotextiles, Vienna, Austria, April 7-11, 1986, Vol. 4, pp. 1191-1196.  215  Hoare, D.J. 1982, "A Laboratory Study into Clay through Geotextiles under Dynamic Loading," Proceedings of the Second International Conference on Geotextiles, Las Vegas, USA., August 1-6, 1982, Vol. 2, pp. 423-428  Holtz, R.D., Christopher, B.R., and Berg, R.R. 1995, "Geosynthetics Design and Construction Guidelines," Publication No. FHWAHI-95-038, Washington, D . C , Federal Highway Administration, March, 1995, p. 396.  Holtz, R.D., Christopher, B. R., and Berg, R.R. 1997, "Geosynthetic Engineering," BiTech Publisher Ltd., Richmond, B.C., p. 452.  Honjo, Y., Haque, M.A., and Tsai, K.A. 1996, "Self-Filtration Behavior of Broadly and Gap Graded Cohesionless Soils," Proceedings of Geofilters'96  Conference,  Montreal, Quebec, Canada, May 29-31, 1996, pp. 227-236.  Hoover, T.P. 1982, "Laboratory Testing of Geotextile Fabric Filters," Proceedings of the Second International Conference on Geotextiles, Las Vegas, Nevada, Aug. 1-6, Vol. 3, 1982, pp. 839-843.  Ingold, T.S. 1985, "A Theoretical and Laboratory Investigation of Alternating Flow Filtration Criteria for Woven Structures," Geotextiles and Geomembranes, Vol. 2, No. 1, pp. 31-45.  Ingold, T.S. 1994, "The Geotextiles and Geomembranes Manual", Elsevier Advanced Technology, Oxfordshire, England, 610 p.  Kenney, T.C., Chahal, R., Chiu, E., Ofeogbu, G.I., Omanga, G.N, and Ume, C A . 1985, "Controlling Constriction Sizes of Granular Filters," Canadian Geotechnical Journal, Vol. 22, No. 1, pp. 32-43.  216  Kenney, T.C., and Lau, D. 1985, "Internal Stability of Granular Filters," Canadian Geotechnical Journal, Vol. 22, No. 2, pp. 215-222.  Kenney, T.C., and Lau, D. 1986, "Reply: Internal Stability of Granular Filters," Canadian Geotechnical Journal, Vol. 23, No. 3, pp. 420-423.  Knaap, F.C.M., Klein Bretler, M., and Meulen, T. 1986. "Design Criteria for Geotextiles Beyond the Sandtightness Requirement," Proceedings of the Third International Conference on Geotextiles, Vienna, Austria, April 7-11, 1986, Vol. 2, pp. 495500  Koerner, R.M., and Ko, F.K. 1982, "Laboratory Studies on Long Term Drainage Capacity of Geotextiles," Proceedings of the Second International Conference on Geotextiles, Las Vegas, Nevada, Aug. 1-6, Vol. 1, pp. 91-95.  Koerner, R.M. 1998, "Designing with Geotextiles, " Fourth edition, Prentice Hall, New Jersey, U.S.A, 761 p.  Kohler, H. J. 1993, "The Influence of Hydraulic Head and Hydraulic Gradient on the Filtration Process,"  Proceedings  of the First international 'Geo-Filters'  Conference, Karlsruhe , Germany, Oct. 20-22, 1992, pp. 225-240.  Kuerbis, R.H., and Vaid, Y.P. 1988, "Sand Sample Preparation - the Slurry Deposition Method," Soils and Foundations, Vol. 28, No. 4, pp. 107-118.  Lafleur, J. 1984, "Filter Testing of Broadly Graded Cohesionless tills," Canadian Geotechnical Journal, Vol. 21, No. 4, pp. 634-643.  217  Lafleur J., Mlynarek J., and Rollin A. L. 1989, "Filtration of Broadly Graded Cohesionless Soils", Journal of Geotechnical Engineering, ASCE, Vol. 115, No. 2, pp. 17471768.  Lafleur, J., Mlynarek, J., and Rollin A. L. 1993, "Filter Criteria for Well Graded Cohesionless  Soils," Proceedings  of the First international 'Geo-Filters'  Conference, Karlsruhe, Germany, Oct. 20-22, 1992, pp. 97-106.  Lafleur, J. 1998, "Particles Washout Associated with the Retention of Broadly Graded Soils by Geotextiles," Proceedings of the Sixth International Conference on Geosynthetics, Atlanta, Georgia, USA, March 25-29, 1998, Vol. 2, pp. 10011004.  Lafleur, J. 1999, "Selection of Geotextiles to Filter Broadly Graded Cohesionless Soils," Geotextiles and Geomembranes, Vol. 17, No. 5-6, pp. 299-312.  Lawson, C R . 1982,  "Geotextile Requirements for Erosion Control Structures,"  Proceedings of the International Symposium on Recent Developments in Ground Improvement Techniques, AIT, Bankok, November 29-December 3, 1982, pp. 177-192.  Lawson,  C. R.  1998,  "Retention  Criteria and Geotextile-Filter Performance,"  Geotechnical Fabrics Report, Vol. 16, No. 2, pp. 26-29.  Legge, K.R. 1990, "A New Approach to Geotextiles Selection," Proceedings of the Fourth International Conference on Geotextile, Geomembranes and Related products, Hague, Netherlands, May 28-June 1, 1990, Vol. 1, pp. 269-272.  Luettich, S.M., Giroud, J.P., and Bachus, R.C. 1992, "Geotextile Filter Design Guide", Geosynthetics and Geomembranes, Vol. 11, No. 4-6, pp. 355-370.  218  Luettich, S.M. 1993, "Geotextile Filter: Designing for the application," Geotechnical Fabrics Report, Vol. 11, No. 1, pp. 70-78.  Mackey, R., and Koerner, G.R. 1999, "Biological Clogging of Geotextiles Filters, a Five Year  Study",  Proceedings  of  Geosynthetics  '99  Conference,  Boston,  Massachusetts, USA, 28-30, April, 1999, Vol. 2, pp. 783-798.  Mcmorrow, J. 1990, "Filtration Action of Non-Woven Geotextiles under Dynamic Loading," Proceedings of the Fourth International Conference on Geotextile, Geomembranes and Related Products, Hague, Netherlands, May 28-June 1, 1990, Vol. 1, pp. 233-238.  Mlynarek, J., Blond, E., Bouthot, M., and Marcil R. 1999, "Geosynthetic Prefabricated Drains Behavior During a Thunder Rainfall in Experimental Lasalle Municipality Roadway," Proceeding of Geosynthetics '99 Conference, Boston, Massachusetts USA, April. 28-30, Vol. 1, pp. 291-305.  Mouw, K. A., Nederlof, K. D. C , Stuip, J., and Veldhuijzen, V. Z. R. 1986, "Geotextiles in Shore and Bed Protection Works," Proceedings of the Third International Conference on Geotextiles, Vienna, Austria, April 7-11, Vol. 2, pp.349-354.  Narejo, D.B., and Koerner, R.M. 1992, "A Dynamic Filtration Test for geotextile Filters," Geosynthetics and Geomembranes, V. 11, pp. 395-400.  Nishigata, T., Fannin, R.J., and Vaid, Y.P. 2000, "Blinding and Clogging of Nonwoven Geotextiles," Journal of Soil and Foundations, in press. , 35 pp.  219  OMT, Ontario Ministry of Transportation, 1992, Draft Engineering Materials Report, Guidelines for Design and Quality Control of Geotextiles, Engineering Materials Office, Soils and Aggregates Section, Ministry of Transportation Ontario, Downsview.  Palmeira, E.M., Fannin, RJ., and Vaid, Y.P. 1996, "A Study on the Behavior of SoilGeotextile Systems in Filtration Tests", Canadian Geotechnical Journal, Vol. 33, No. 6, pp. 899-912.  Palmeira, E.M., and Fannin R. J. 1998, "A Methodology for the Evaluation of Geotextile Pore Opening Sizes Under Confining Pressure," Geosynthetics International, Vol. 5, No. 3, pp. 347-357.  Permanent International Association of Navigation Congresses - PIANC, "Guidelines  for the  Design  1990.  and Construction of Flexible Revetments  Incorporating Geotextiles for Inland Waterways," Supplement to Bulletin no. 57, 1987, 27 Congress, May 20-26, 1990, 133 p. th  Poskitt, M. F.F., (Chairman), Carlyle, W.J., Earp D.N.W, Sadgrove. B.M., and Vaughan, P.E. 1986, "Geotextiles As Filter And Transition in Fill Dams", Report prepared on behalf of BNCOLD for Committee on Material for Fill Dams, Bulletin 55 Paris.  Rohde, J. R., and Gribb, M. M. 1990, "Biological and Particulate Clogging of Geotextile/Soil Filter Systems," Geosynthetic Testing for Waste Containment Application," ASTM STP 1081, Robert M. Koerner, editor, ASTM, Philadelpia.  Rollin, A.L. 1996, "Bacterial Clogging of Geotextiles," Proceedings of Geofilters'96 Conference, Montreal, Quebec, Canada, May 29-31, 1996, pp. 125-134.  220  Rowe, R.K. 1998, "Geosynthetics and the Minimization of Contaminant Migration through Barrier System Beneath Solid Waste," Proceedings of the Sixth International Conference on Geosynthetics, Atlanta, Georgia, USA, March 25-29, 1998, Vol. 1, pp. 27-102.  Van Santvoot, G.P.T.M. 1994, "Geotextiles and Geomembranes in Civil Engineering," A.A. Balkema, Rotterdam, Brookfield, Revised Edition, 595 p.  Sansone, L.J., and Koerner, R.M. 1992, "Fine Fraction Filtration Test to Asses Geotextile Filter Performance," Geotextile and Geomembranes, Vol. 11, No. 4-6, pp. 371393.  Schober, W. and Teindl, Ff. 1979, "Filter-Criteria for Geotextiles," Seventh European Conference on Soil Mechanics and Foundation Engineeing, Brighton, UK, September 1979, Vol. 1, pp. 121-129.  Scott, J.D. 1980, "The Filtration - Permeability Test," Proceedings of the First Canadian Symposium on Geotextiles, Calgary, Alberta, Canada, September 23, Vol. 2, pp. 121-129.  Sherard, J.L., Decker, R.S. and Ryker, N.L. 1972, "Piping of Earth Dams of Dispersive Clay," Proceedings of The ASCE Special Conference on Performance of Earth and Earth Support Structures, Purdue University, New York, Vol. 1, pp. 589626.  Sherard, J.L., Dunningan, L.P., and Talbot, J.R. 1984, "Basic Properties of Sand and Gravel Filters," Journal of Geotechnical Engineering, Vol. 110, No. 6, pp. 684700.  221  Sherard, J.L., and Dunningan, L.P. 1986, "Internal Stability of Granular Filters: Discussion," Canadian Geotechnical Journal Vol. 23, No. 2, pp. 418-420.  Sherard, J.L. and Dunningan, L.P. 1989, "Critical Filter for Imprevious Soils," Journal of Geotechnical Engineering, ASCE, Vol. 115, No. 7, pp. 927-947.  Shi, T. 1993, "Filtration behaviour of Non-Woven Geotextiles in the Gradient Ratio Test," MASc thesis, The University of British Columbia, Vancouver, Canada, 127 p.  Siva, U. and Bhatia, S.K. 1993, "Filtration Performance of Geotextiles with Fine-Grained Soils," Proceedings of Geosynthetics' 93 Conference, Vancouver, Canada, March 30-Aprl. Vol. 1, pp. 483-499.  SPM, "Shore protection manual," 1984, Department of Army Waterway Experiment Station, Corps of Engineers, Coastal Engineering Research Center, Vol. 1, Chapter 2, Washington, DC, U.S.A.  Skempton, A.W., and Brogan, J.M. 1994, "Experiments on Piping in Sandy Gravels," Geotechnique 44, No. 3, pp. 449-460.  Smith, J. L., Bathia, S. K., Ridgway, J., and Hawkins, W. 1999, "Are Porosity and 095 Important in The Retention and Particulate Clogging Behavior of Geotextiles," Proceeding of Geosynthetics '99 Conference, Boston, Massachusetts USA, April 28-30. Vol. 2, pp. 813-832.  Tomlinson, S.S. 1997, "Seepage Forces and Confining Pressure Effects On Piping Erosion", M.Sc. Thesis, UBC, Department of Civil Engineering, 62 p.  Tondello, M . 1998,  "Geotextile Filters in Unsteady Flow Conditions, "Italian  Geotechnical Journal, ANNO XXXII, No. 4, Oct.-Dec. pp. 18-29.  222  United States Army Corps of Engineers, 1977, "Plastic Filter Cloth," Civil Works Construction Guide Specification No. CE-02215, Office Chief of Engineers, Washington, D . C , 16 p.  USFHWA, United States Department of Transportation Federal Flighway Administration, 1995, "Geosynthetic Design and Construction Guidelines, "NFfl Course No. 13213, Mclean, Virginia, USA, May 1995.  Veldhuijzen van Zanten, R and Thabet, R.A.H. 1982, "Investigation on Long Term Behavior of Geotextile in Bank Protection Works," Proceedings of the Second International Conference on Geotextiles, Las Vegas, USA, August 1-6, 1982. Vol. 1, pp. 259-264.  Vitton, S. J., and Salder L.Y. 1997, "Particle-size Analysis of Soils Using Laser Light Scattering and X-Ray Absorption Technology, " Geotechnical Testing Journal, Vol. 20, No.l, pp. 63-73.  Watson, P.D.J., and John, N.W.M. 1999, "Geotextile Filter Design and Simulated Bridge Formation at Soil-Geotextile Interface," Geotextiles and Geomembranes, Vol. 17, No. 5-6, pp. 265-280.  Williams, N.D., and Abouszakhm, M.A. 1989, "Evaluation of Geotextile/Soil Filtration Characteristics Using the Hydraulic Conductivity Ratio Analysis," Journal of Geotextiles and Geomembranes, Vol. 8, No. 1, pp. 1-26.  Wewerka, M . 1986, "Filtration Capability of Geotextiles - Testing And Practical Experience", Proceedings of the Third International Conference On Geotextiles, Vienna, Austria, April 7-11, Vol. 4, pp. 1203-1206.  223  Wolski, W. 1987, "General Report", Groundwater Effect in Geotechnical Engineering," Proceedings  of the Ninth European Conference on Soil Mechanics and  Foundation Engineering, Dublin, Ireland, August - September 1987, Vol. 3, pp. 1351-1366.  Young, H. M . , and Ochola C. 1999, "Strain Effects on the Filtration Properties of Geotextiles,"  Proceeding  of  Geosynthetics  '99  Massachusetts USA, April 28-30. Vol. 2, pp. 757-768.  224  Conference,  Boston,  APPENDIX A. TESTS RESULTS Although the results and the sample behavior during testing are presented in Chapter 6, in the form of tables and figures, some additional figures that may be of interest are presented here. The vibration test results are presented in section A l and the cyclic flow test results are presented in section A2. A 1 : Vibration tests Static stage: a.) The variations of Gradient ratio with time Dynamic stage: Variations with respect to cumulative number of blows of: b.) Gradient ratio, c.) Void ratio, and d.) Permeability along the sample.  225  -eO  ASTM  #  ASTM Mod.  Mod.  2A  2H  I 200  — |  400  1  | —  600 t (min.)  —r~  2000  800  b.)  a.)  0.004 •  0.60 •  s(1.9-39).g103  I 0.56  H  0.52  H  K  35  0.003 •  0.48 0.001  0.40 1000  2000  3000  1000  4000  2000 3000 Number of blows, N i  d.)  c)  Figure A l . l : s(1.9-39).gl03!  226  4000  -e-0-  ASTM Mod.  ASTM Mod.  a.  o  1  4—•  1^=§=§= T  200  400  600  800  1000  1000  2000 N  t (min.)  a.)  3000  4000  h  b.)  K  35  K  57  K  67  0.003 •  >  0.002 - A  0.48  0.001 0.44  0.000  0.40 0  1000  2000 N  3000  T 1000  4000  2000  h  d.)  c) Figure A1.2: s(1.9-39).gl03  2  227  3000  4000  Figure A1.3: s(2.2-52).gl03  228  229  230  231  232  233  234  c.)  d.)  Figure ALIO: s(20-3.1).gl03  235  236  237  238  239  240  241  242  243  244  245  246  247  248  Figure A1.24: s(15-4.5).g290  249  A 2 : Cyclic flow test  Static stages 1 and 5: No change of water heads was observed with time, and therefore all the results are presented in Chapter 6 in a form of tables and figures. Cyclic flow stages 2-4: Stage 2: Figure a.) water heads with time during four cycles using the symbols presented in Figure A2.1a. Figure b.) water head distribution with time during half a cycle using symbols presented in Fig. A2. lb. Stage 3: Figure c.) water heads with time during two and a half cycles using the symbols presented in Fig. A2.1a. Figure d.) water head distribution with time during half a cycle. This is done in a similar way to Figure b. Stage 4: the behavior is similar to stage 2 and was addressed in Chapters 6 and 7.  250  h  d  hi7 hi6  his hi3  •e-  hi2 a.)  T i m e step  1 —  2  —  5  Steps: 2 3 4 5 6  Downward flow  b.) Figure A2.1: Symbols used  251  Upward flow  Steps: 7 8 9 10 1  252  253  254  255  256  257  258  259  260  261  262  263  264  265  266  267  APPENDIX B. BASIC CALCULATIONS  B.1 Void ratio calculation e=y Gs/y -l w  d  B.l.l  e - Void ratio Gs - Specific graviity Yd - Dry unit weight y - Unit weight of water w  Note: In the vibration test program the dry unit weight was corrected accounting for the of the passing through particles. B.2 Darcy's law q=kiA  B.2.2  q - Volumetric flow rate of water k - Coefficient of permeability i - Hydraulic gradient A - Cross section of the permeameter  B.3 Validity of Darcy's law The Reynolds number can be used to check whether flow is laminar or turbulent Re=(pUD)/ri Where: p - Fluid unit mass (=lg/cm for water at ~20°C) 3  268  B.3.1  TJ - Fluid velocity D - Median particle diameter rj - Fluid viscosity (=0.01 g/cm*sec. for water at ~20°C) Darcy's law is valid where the flow is laminar and viscous forces are predominant. The upper limit of this range is at a value of Re between 1 and 10 (Bear, 1972) B.4 Gradation correction In all samples that lost more than 1000 g/m , the original gradation of the soil sample was 2  corrected based on the mass and gradation of the passing through particles according to the following steps: 1. ) Find D95 of the passing though particles (D95p) 2. ) Find the potential maximum diameter of the original soil sample loose fraction, (D ): this F  diameter is assumed to be equal to D30 in case of narrow and wide gradations or in gap graded gradations where the gap is above 45%finer.In gap graded gradations where the gap is below 45% finer DF=Dgap (where Dgap is the diameter at the beginning of the gap)3. ) If  D95P<DF  then subtract the mass of the passing through particles from the  correspondingfractionin the original soil sample.  Note: In this correction two assumptions are made: 1.) If D 9 5 P > D F then the sample is losing its primary fabric and no correction is necessary; and  269  2.) The gradation of thefinerfraction in the soil samples was essentially narrow. Therefore if D95p<D then the gradation of the passing through particles is approximately equal to the F  gradation of the corresponding fraction in the original soil sample.  270  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0063513/manifest

Comment

Related Items