Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Experimental study for evaluating the internal stability of gap-graded soils Khan, Abdul Sattar 2003

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

Item Metadata

Download

Media
831-ubc_2003-0531.pdf [ 37.12MB ]
Metadata
JSON: 831-1.0063846.json
JSON-LD: 831-1.0063846-ld.json
RDF/XML (Pretty): 831-1.0063846-rdf.xml
RDF/JSON: 831-1.0063846-rdf.json
Turtle: 831-1.0063846-turtle.txt
N-Triples: 831-1.0063846-rdf-ntriples.txt
Original Record: 831-1.0063846-source.json
Full Text
831-1.0063846-fulltext.txt
Citation
831-1.0063846.ris

Full Text

E X P E R I M E N T A L S T U D Y F O R E V A L U A T I N G T H E I N T E R N A L STABILITY O F G A P - G R A D E D SOILS By ABDUL SATTAR KHAN B.Sc. (Civil Engineering) - The University of Engineering and Technology, Lahore, Pakistan, 1998 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF APPLIED SCIENCE 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 October 2003 © Abdul Sattar Khan, 2003 UBC Rare Books and Special Collections - Thesis Authorisation Form Page 1 of 1 In p r e s e n t i n g t h i s t h e s i s i n p a r t i a l f u l f i l m e n t of the requirements for an advanced degree at the U n i v e r s i t y of B r i t i s h Columbia, I agree tha t the L i b r a r y s h a l l make i t f r e e l y a v a i l a b l e fo r reference and s tudy . I fu r the r agree tha t pe rmis s ion for ex tens ive copying of t h i s t h e s i s fo r s c h o l a r l y purposes may be granted by the head of my department or by h i s or her r e p r e s e n t a t i v e s . I t i s understood tha t copying or p u b l i c a t i o n of t h i s t h e s i s fo r f i n a n c i a l ga in s h a l l not be a l lowed without my w r i t t e n p e r m i s s i o n . Department of f iy"L F_K61 NEEfitNG The U n i v e r s i t y of B r i t i s h Columbia Vancouver, Canada Date OC-roB£ft {O 2CD3 7 http://vvww.library.ubc.ca/spcoll/thesauth.html 10/9/03 ABSTRACT Experience has shown that certain soils may be susceptible to internal erosion. These soils, for example gap-graded and broadly graded (concave-up) soils are vulnerable to migration of the fine fraction during permeation. Factors governing this phenomenon of internal stability may be categorized as geometric constraints (pore size constrictions, porosity and shape of gradation curve) and hydrodynamic actions (seepage force and direction of flow). The objective of this study was to evaluate the success of empirical rules developed for assessing the internal stability of granular media subjected to seepage flow. Specifically, it seeks to examine the influence of hydraulic gradient, vertical effective stress on the susceptibility of reconstituted gradation of glass beads to internal instability with reference to the boundary conditions of the rigid-wall permeameter. A total of twelve multi-stage permeability tests were executed on glass beads specimens of five different gradations. The specimens were reconstituted using a method of slurry preparation and discrete deposition that resulted in repeatable homogeneous specimens. Analysis of results reveals that the Kezdi (1979) proposed for a split gradation analysis could be used, with reasonable confidence, for assessing the internal stability of gap graded soils. In contrast the criterion of Kenny and Lau (1985, 1986), failed to adequately detect and properly differentiate between the gradations, characterised by a completely horizontal gap in the gradation. At a more fundamental level, it appears that potential for internal stability, and particle migration, diminishes with an increase in vertical effective stress. With regard to the apparatus, the influence of the boundary wall condition is very subtle. For specimens that are either prone to segregation or, on the other hand, stable (which are two extreme stability conditions), the boundary wall condition does not exert any influence. Whereas, for the intermediate gradations, it effects the material behaviour. Abdul Sattar Khan ii Abstract TABLE OF CONTENTS Page ABSTRACT ii LIST OF TABLES vi LIST OF FIGURES vii LIST OF SYMBOLS iX ACKNOWLEDGEMENT X 1.0 INTORODUCTION 1 1.1 Objectives Of The Study 1 1.2 Scope Of The Study 2 2.0 LITERATURE REVIEW 3 2.1 Principle of Internal Stability in Gap-Graded Soils 3 2.1.1 Geometric Constraints 4 2.1.2 Hydrodynamic Actions 5 2.1.3 Critical Hydraulic Gradient 5 2.2 Filter Design Criterion of Particle Retention 5 2.3 Review of Filtration Testing 10 3.0 APPARATUS, MATERIALS AND PROCEDURES 23a. 3.1 Modified Gradient Ratio Device 23a, 3.1.1 Soil Specimen and Filter Mesh 24 3.1.2 Water Supply and Control System 24 3.1.3 Instrumentation 25 3.1.4 Data Acquisition System 26 3.2 Materials 26 3.3 Procedures 29 3.3.1 Specimen Reconstitution 29 Abdul Sattar Khan iii Table of Contents 3.3.2 Test Set-Up 3.3.3 Multi-Stage Test Procedure 3.3.4 Post Test Measurements And Observations 29 30 30 3.4 Program of Investigation 30 TEST RESULTS 34 4.1 Head Losses in the Permeameter 34 4.2 Loss of Fines from the Specimen 36 4.3 Pre and Post Test Gradations 40 4.4 Water Head Distribution 43 4.5 Permeability 46 4.6 Axial Strains attributed to the Seepage Forces 47 4.7 Repeatability 48 ANALYSIS AND DISCUSSION 50 5.1 Critical Hydraulic Gradient 50 5.1.1 Influence of the Boundary Conditions 5.1.2 Influence of Elevated Vertical Effective Stress 53 54 5.2 Loss of Fines from the Test Specimen 54 5.2.1 Influence of the Boundary Conditions 5.2.2 Influence of Vertical Effective Stress 55 56 5.3 Comparison with other Studies 57 5.3.1 Evaluation of the Critical Hydraulic Gradient 5.3.2 Observed total loss of Fine Fraction 5.3.3 Effect of Vertical Effective Stress 57 59 60 5.4 Validation of the Kezdi' s (1979) Criterion 60 5.5 Rationale of the Kenny and Lau (1985, 1986) Criterion 61 6.0 CONCLUSSION AND RECOMMENDATIONS 62 Abdul Sattar Khan iv Table of Contents LIST OF REFERENCES 65 APPENDIX - A Assessment of Internal Stability by Kenney and Lau Criterion 67 (1985, 1986) APPENDIX - B Pre and Post-Test Gradations 71 APPENDIX-C Water Head Distributions 78 APPENDIX - D Photographs 84 Abdul Sattar Khan v Table of Contents L I S T O F T A B L E S Page Table 2.1 Published design rules for granular filters (Modified after 7 Geotechnical Engg. Office, Government of Hong Kong, 1993) Table 2.2 Design criteria for granular filters by US Bureau of Reclamation 8 (after Moffat, 2002) Table 2.3 Table 2.3 Design criteria for granular filters (after Geotechnical 9 Engg. Office, Government of Hong Kong, 1993) Table 3.1 Characteristics of the selected gradation curves 28 Table 3.2 Test code developed for the study 31 Table 3.3 Summary of the observed system hydraulic gradients, in and duration 31 of the test stages Table 4.1 Summary of fine fraction passing through the filter mesh 36 Table 4.2 Information of the observed system hydraulic gradients, in, for 39 every test stage Table 4.3 Summary of the mass of fine fraction loss during each test stage 39 Table 4.4 Average permeabilities (x 10"3 cm/sec) for different test stages 47 Table 4.5 Summary of specimen lengths after each test stage 48 Table 4.6 Summary of tests for repeatability 49 Table 5.1 Summary of the influence of the hydraulic gradients 51 Table 5.2 Summary of the normalized fines loss in % of the total fines 55 present in the test specimen at the start of the test Abdul Sattar Khan vi List of Tables L I S T O F F I G U R E S P a g e Fig. 2.1 Primary fabric and fine particles of a glass beads sample 3 Fig. 2.2 Microscopic view of the particle arrangement resulting in small 4 constriction sizes Fig. 2.3 Close up view of the particle arrangement leading to the larger 4 constriction sizes Fig. 2.4 Constant head apparatus (after Bertram, 1940) 11 Fig. 2.5 Filter criterion (after Terzaghi, 1922) 12 Fig. 2.6 Permeameter used by Karpoff (1955) 13 Fig. 2.7 Methodology to split the gradation (after Kezdi, 1979) 14 Fig. 2.8 Permeameter test assembly (after Lafleur, 1984) 15 Fig. 2.9 Test set-up (after Sherard etal. 1984) 16 Fig. 2.10 Permeameter arrangement (after Kenny et al. 1985) 17 Fig. 2.11 Stability criterion (after Kenny et al. 1985) 18 Fig. 2.12 Test specimen (after Honjo et al. 1996) 19 Fig. 2.13 Typical test arrangement (after Tomlinson et al. 2000) 20 Fig. 2.14 Specimen dimension (after Lafleur, 2000) 20 Fig. 2.15 Flow control system (after Moffat, 2002) 22 Fig. 3.1 Details of the permeameter and flow control 23o-Fig. 3.2 Different port locations along the length of permeameter 25 Fig. 3.3 View of a filter mesh used for the test 3.7S25 27 Fig. 3.4 Glass beads used in specimen reconstitution 27 Fig. 3.5 Particle size distribution curves of the test gradations 28 Fig. 3.6 Schematic diagram of the textured wall of the rigid-wall permeameter 32 Fig. 3.7 Top view of the textured wall permeameter 32 Fig. 4.1 Variation in the applied and observed system gradient with 35 volumetric flow rate for test 4.4S25 Abdul Sattar Khan vii List of Figures Fig. 4.2 Variation in the applied and observed system gradients, in, with 35 volumetric flow rate for test 5.9S25 Fig. 4.3 Total fine fraction loss for different gradations 37 Fig. 4.4 Development of pipe leads to the localized fine loss through the 38 specimen (Test: 5.9S25) Fig. 4.5 Universal fines loss throughout the specimen in test 7.4T25 38 Fig. 4.6 Grain size distribution curves for specimen used in test 3.7S25 40 Fig 4.7 Gradation curves for specimen used in test 4.4S25 41 Fig. 4.8 Grain size distribution curve for specimen used in test 5.9S25 42 Fig. 4.9 Gradation curve for specimen used in test 7.4S25 42 Fig. 4.10 Three exclusive behaviours deduced from the water head distributions 43 Fig. 4.11 Water head distributions measured during test 4.4S25 44 Fig. 4.12 Water head distributions for the test 5.9S50 45 Fig. 4.13 Water head distributions measured for the test 5.9S25 45 Fig. 4.14 Water head distribution for test 7.4S25 46 Fig. 4.15 Water head distribution curves for tests 5.9T25and 5.9T25R 49 Fig. 5.1 Development of pipe through the specimen in test 5.9S25 52 Fig. 5.2 Universal fines loss through the specimen (suffosion) in test 8.9T25 52 Fig. 5.3 Influence of the boundary conditions and elevated vertical effective 53 stress over critical hydraulic gradient Fig. 5.4 Normalized fines loss under different boundary conditions of the 56 rigid-wall permeameter Fig. 5.5 Test results reported by Tomlinson et al. (2000) 58 Fig. 5.6 Total loss of fines observed during various studies 59 Fig. 5.7 Evaluation of gradations for internal stability using Kenney and Lau 61 criterion (1985, 1986) Fig. 6.1 Design rules and observations reported by various researchers 63 Abdul Sattar Khan viii List of Figures LIST OF SYMBOLS A Cross sectional area of the permeameter C, Coefficient of uniformity Coefficient of curvature D B n Grain size of the base soil corresponding to n % finer by weight D n F Grain size of filter medium corresponding to n % finer by weight D, ns Grain size of base soil corresponding to n % finer by weight D n- F Grain size of filter medium corresponding to n % finer by weight (Kezdi, 1979) D n' S Grain size of base soil corresponding to n % finer by weight (Kezdi, 1979) F Mass fraction smaller than D in % (Kenny et al., 1985 and 1986) Gs Specific gravity H Mass fraction measured between D and 4d (Kenny et al., 1985 and 1986) h Water head hjj Water head measured across the length of specimen between ports i and j i Hydraulic gradient i c r Critical hydraulic gradient iy Hydraulic gradient between across the length of specimen between ports i and j k Hydraulic conductivity kjj Hydraulic conductivity for the length of specimen between port i and j L Length of the specimen LVDT Linear variable differential transformer Abdul Sattar Khan ix List of Symbols ACKNOWLEDGEMENT I wish to express my sincere gratitude to Dr. Jonathan Fannin, my academic research supervisor, for his thorough and persistent professional advice along with undaunting moral support that was instrumental for the timely completion of my research studies and subsequent submission of this thesis. I would also like to extend a word of thanks to Dr. Dharma Wijewickreme and Dr. John Howie for their additional suggestions during my experimental work. I am most grateful to Dr. Dawn Shuttle for her reading of this thesis and additional comments. I would like to thank Mr. Harald Schrempp and Bill Leung of the UBC Civil Engineering machine shop, for their support in maintaining the test device. My thanks also to Mr Scott Jackson for his technical support during modifications to the instrumentation and data acquisition system. Additional thanks are extended to my colleagues Ali Amini, Chris Anderson, Sriskandra Kumar Somasundarm, Tanya van Halderen, Megan Sheffer and especially Ricardo Moffat for initiating healthy debates. Financial assistance for this study was obtained through a University Graduate Fellowship (UGF) of the University of British Columbia, and the Natural Science and Engineering Research Council of Canada (NSERC). Contribution of each organization is much appreciated. I also wish to express my deepest love for my parents in Pakistan, who have always been a beacon of hope and supported me at every point of my life. Last but not the least; I am deeply touched by my wife, Bushra, and daughter, Aleeza, for their love and support. Abdul Sattar Khan x Acknowledgement 1.0 INTRODUCTION Gap graded soils are abundant in different parts of the world, especially in higher latitudes of the northern and southern hemispheres. Glacial tills and moraines, which are intermingled materials of heterogeneous size and shape, are examples of gap-graded soils. These soils of deficient intermediate grain sizes are the result of denudation process associated with alpine glaciers and transported along the riverbeds. Many earth dams had been either constructed over or with these soils. However, gap-graded soils are susceptible to particle migration when subject to seepage flow and so potentially unstable. Exposure to seepage can not only change the gradation of parent soil but also trigger the structural instability in the soil matrix. Gap-graded soils are also prone to material segregation during placement The greater appreciation of potential problems with the gap-graded soils has ushered in a series of studies to understand their grain structure and associated material behaviour. During evolution of knowledge, the concept of internal stability was established. Internal stability is associated with the size of finer fractions and constriction openings formed at the contact points of coarser particles. Factors affecting the internal stability can be categorized as: i. A geometric constraint. For example, pore size, porosity, shape of gradation curve; and, ii. A hydrodynamic state. For example seepage force, direction of flow and exsolution of dissolved air In order to better understand the internal stability, a series of multi-stage permeameter tests were conducted. The variables examined included the grain sizes and seepage force. Observations of the variation in head loss along the length of specimen, and any of soil loss during each stage, were used to characterize the potential for instability. 1.1 Objectives of the Study Ralph B. Peck, at the fifth Laurits Bjerrum memorial lecture, shared his disenchanted observation that "Designers and regulatory bodies tend to place increasing reliance on analytical procedures of growing complexities and to discount judgement as a non-quantitative, undependable contributor to design". Good judgment develops from both Abdul Sattar Khan 1 Chapter No. 1 experience and a confident understanding based on careful observation. This study was planned to not only be a contribution towards the advancement of knowledge, but also an effort to understand the complexities involved in the interparticle behaviour of soils under seepage flow. Objectives set forth at the beginning of this study were: 1. To expand the database acquired using the UBC modified gradient ratio test device on gap-gradations (both soils and glass beads) 2. To further validate the threshold value of D i 5 - F / dg5'S ratio, Kezdi's criterion (1979), marking the shift from an internally stable to an unstable gradation 3. To comment upon the influence (if any) of boundary conditions in the rigid-walled permeameter on the onset of instability in the test specimen 4. To examine the hydrodynamic conditions necessary to trigger instability in the test specimen 5. To observe the effects of elevated effective stress level over the response to seepage flow 1.2 Scope of the Study This thesis is based on the interpretation of twelve multi stage permeameter tests conducted on reconstituted glass beads samples and five different D i 5 - F / d^ys ratios were examined. Tests variables were seepage force, level of vertical effective stress and boundary condition of the rigid-walled permeameter. A unidirectional flow regime was imposed using constant head tanks, to achieve the target hydraulic gradients. The level of vertical effective stress was also controlled. Interpretation of the results is based on the water head distribution along the length of specimen, volumetric flow rate, measurement of specimen height, mass of loss of fines through the sample and visual observations during the series of experiments. A synthesis of the results provides us an empirical rule to assess the internal stability of gap-graded soils with greater confidence. Abdul Sattar Khan 2 Chapter No. 1 2.0 LITERATURE REVIEW Over the last fifty years, many earth fill dams have been constructed using new and evolving criteria for filter design and provision of drainage. Through this process of knowledge advancement, innovative concepts evolved to explain material performance under the influence of seepage forces. Growth of a self-healing layer at the soil filter interface, self-filtering filters, development of filter cakes and the use of widely graded filters are few such examples. However, field monitoring suggests that some dams have not always performed well, with problematic behaviour attributed to the use of internally unstable core and filter materials during construction. It is concluded that the core and filter were not able to successfully block the migration of particles from the core-filter interface and to retain their finer fractions from being drained out of the system (Ripley 1985). The focus of this review is the principle of internal stability in gap-graded soils, with consideration given to influencing factors and a concise presentation of available design criteria. As filtration testing is a key tool used to assess the internal stability of granular materials, a brief commentary on previous studies is also presented. 2.1 Principle of Internal Stability in Gap-Graded Soils Gap-graded soils have a skeleton of coarser particles or primary fabric that is responsible for the transfer of load or stress (see Fig. 2.1). Pores or openings form between the contact points of these coarser particles, termed constrictions. Within the pores of the primary fabric are mobile fine particles that are not fixed in position. They do not transfer stress and may potentially move to neighbouring pores or out of the system, if sufficiently small (Kenney et al. 1985). Fig. 2.1 Primary fabric and fine particles of a glass beads sample Abdul Sattar Khan 3 Chapter No. 2 When soil is subjected to seepage flow, the flowing water exerts a drag force on these potentially mobile fine particles. Movement of fines within the primary fabric and hence the onset of internal instability depends upon the constraints that may be categorized as geometric and hydrodynamic in origin (Kezdi. 1979). The terms piping, suffusion and suffosion have been used to describe several forms of internal erosion. Piping is usually applied to a process that starts at the exit point of seepage and a continuous fine deficient passage or pipe developed in the soil by backward erosion (Charles, 2001). Suffusion describes the redistribution of fines within the soil layer causing no change in the overall porosity and permeability of the soil layer (Kovacs 1981). Suffosion refers to a universal fines loss in a soil layer and changes the porosity and permeability values of the original soil matrix (Charles 2001) 2.1.1 Geometric Constraints Geometric constraints address the comparative sizes of fine particles and constriction sizes. If the constriction size is small enough (see Fig. 2.2) to control the movement of a finer fraction even under the application of severe seepage forces, then the gradation is termed "self-filtering". In a self-filtering or internally stable gradation, there is a well knitted primary fabric with dominant arching effect by the coarser particles giving small constriction sizes that will efficiently prevent the fines movement. Whereas, in internally unstable gradations, the constriction size is large enough to allow an unimpeded movement of fines (see Fig. 2.3) Fig. 2.2 Microscopic view of the particle Fig. 2.3 Close up view of the particle arrangement arrangement resulting in small constriction sizes leading to the larger constriction sizes Factors that define the constriction size include shape of the gradation curve, porosity, shape of the particle and magnitude of effective confining stress. Abdul Sattar Khan 4 Chapter No. 2 2.1.2 Hydrodynamic Actions A threshold gradation is postulated that, upon severe to mild level of hydrodynamic disturbance, yields the internal instability. These hydrodynamic actions include seepage force, flow direction and vibration. As internally unstable gradations approached, even a mild hydrodynamic condition has potential to initiate a movement of fines through the primary fabric. These hydrodynamic actions exert a specific influence on the soil behaviour hence, it is necessary to examine the response under prototype conditions in any laboratory-testing programme. 2.1.3 Critical Hydraulic Gradient It has been postulated that a relatively mild hydrodynamic force might cause internal instability to occur in a soil with large constrictions and fine particles of small diameter. In order to categorize such behaviour, the term critical hydraulic gradient (icr) has widely been used in the literature. For the case of a two layer system, base soil and filter medium, the critical hydraulic gradient may cause fine particles of the base layer to clog the pores of the filter material (Bertram 1940) or induce a free movement of base soil particles through the filter medium leading to the development of an open channel in the soil (Tomlinson et al. 2000). In the current study, critical hydraulic gradient is similarly defined as the gradient to cause an appreciable movement of fines out of the specimen. This fines movement either leads to the formation of a fine deficient channel along length of the specimen or results in a suffosion through entire or part of the cross sectional area and length of the specimen. Either mechanism is associated with a marked drop in the hydraulic resistance of the system. 2.2 Filter Design Criterion of Particle Retention In principle, since the measure of internal stability is the capacity for fines retention of a particular gradation, the filter design criterion of particle retention could be used to assess the internal stability of soils. Considerable attention has been afforded to the development of empirical rules for particle retention, recognizing the variable nature of field conditions. Many regulatory and monitoring agencies, such as U.S. Army Corps of Engineers, the U.S. Bureau of Reclamation and the Geotechnical Engineering Office of the Government of Hong Kong, have both general and specific design guidelines on filtration and drainage works. All of these empirical formulations are based upon three major assumptions. First, both filter and Abdul Sattar Khan 5 Chapter No. 2 base soil are cohesionless, and the particles are free to move unless constrained by other particles. Second, some self-filtering occurs at the interface with large base soil particles moving to fill constrictions in the filter, and progressively smaller particles then filling the resulting constrictions until stability is achieved. Third, both base soil and filter are internally stable (Vaughan 2000). This concept of particle retention raises the issue of effective particle diameters to represent both the base soil and filter material. Conservatively, the largest filter particle size, DIOOF, would result in the largest constriction size whereas the smallest soil particle, Dos, is the size most likely to travel through the filter. However, this DIOOF / Dos would eliminate the formation of a self-filtering layer and resultant bridging zone. These processes allow the small size particles to be progressively retained by relatively larger diameter particles. A more realistic grain size ratio to serve as an index of whether or not erosion will occur was proposed as D15F / D85s (Terzaghi 1939). A concise summary of different particle retention criteria suggested by individual researchers and regulatory organizations is given in Table 2.1 to 2.3. Abdul Sattar Khan 6 Chapter No. 2 Table 2.1 Published design rules for granular filters (modified after Geotechnical Eng. Office, Government of Hong Kong, 1993) Reference Coefficient of Uniformity C u Filter criteria for soil retention Remarks Base soil Filter D1 5 F/D 85S D50F/D50S Terzaghi(1939) None None 4 None Criterion based on experience Bertram (1940) 1.2 1.2 <6.5 None Hydraulic gradient 8 to 20 Newton and Hurley (1940) 3.1 -4.8 Uniform None None Failed sample has D1 5 F/D85s < 4 USCE(1941) Uniform 2.3-8 <5 None Hydraulic gradient 2 Lund (1949) 1.1-7.0 Uniform None None Hydraulic gradient 10; Lund opined that the Terzaghi (1922) criteria had a safety factor of 2 USCE(1953) 1.2-6.1 2-23 <5 None Hydraulic gradient 1 to 26 Leatherwood & Peterson (1954) 1.3-2.9 1.2-1.3 <4.1 <5.3 Criteria based on maximum head loss at interface USBR(1955) 1.4-7.0 1.2-1.4 None 5 to 10 Uniform filters, DIOOF < 75 mm 7-25 5-30 None 12 to 58 Graded filters (grading of base soil and filters of similar shape) None None None 9 to 30 Crushed materials (base soil for particles finer than 4.76 mm only) Kolbuszewski (1957) 1.15 1.05-1.08 < 10 None All filters tested were stable max D I 5 F / D 8 5 S was 12.5 Zweck and Davidenkoff (1957) 1.2 1.2 None 5 to 10 Medium sand (max. hydraulic gradient of 2.2) 5 to 15 Fine sand 5 to 25 Graded sand Thanikachalam and Sakthivadivel (1974a; 1974b) None None None None They proposed the following criterion: D 6 0 F / D 1 0 S = 0.941 D I 0 F / D 1 0 S - 5.65 Log (D,0F / D 1 0 S - 3) = 1.55/ Log (1000D10S -1) Lafleur(1984) 6.9 1.8-23 <9 None Coarser particle (> 4.76 mm) omitted in calculating the D85S Sherard et al. (1984a) Uniform 1.1-4.4 <5 None No significant amount of fine particles (< 75 um) in filter. Filter and base soil grading need not to be of similar shape Well-graded 1.1-4.4 <9 None Kenney et al. (1985) 3-14.3 1.2-12 None None Additionally, D 5 F / D 5 0 S < 4 and D 1 5 F / D 5 O S < 5 US Army Corps of Engineers (1986) None None <5 <25 Tomlinson et al. (2000) Poorly graded Poorly graded <8 None A soil filter will not fail at Di5 F/D 8 5s <8 and at 12 filter will not been able to retain the base soil Moffat (2002) Gap graded None <5 None Seepage forces and vibrations with were applied Abdul Sattar Khan 7 Chapter No. 2 Table 2.2 Design criteria for granular filters by US Bureau of Reclamation (after US Bureau of Reclamation 1987) Rule number Filter design rule Requirement 1 D,5F<4*d 8 5B (1) D 1 5 F < 0.7 mm + (40-A) (4*d85B-0.7 mm) / 25 (2) STABILITY (1) Sand and gravels: less than 15% fines (2) Silty and clayey sands and gravels: 15 to 39 finer 2 For gap-graded and unstable, broadly graded base soils, the filter should be designed to protect the fine matrix of the base soil 3 The permeability of the filter should be at least 25 times that of the base material. Generally: D i 5 F > 5 D 1 5 B PERMEABILITY D15F not less than 0.1 mm 4 The percent fines finer than No. 200 sieve must not exceed 5% by weight after compaction 5 The ratio D 9 0 F / D 1 0 F should decrease rapidly with increasing D ] 0 F SEGREGATION Generally, a filter should be uniformly graded to prevent segregation during processing, hauling and placing. 6 Filter should have relatively uniform grain-size distribution curves, without "gap-grading" Notes: D 1 0 F and D 9 0 F limits for preventing segregation Minimum DIOF (mm) Maximum D90F (mm) < 0.5 20 0.5-1.0 25 1.0-2.0 30 2.0-5.0 40 5.0-10.0 50 10.0-50.0 60 Abdul Sattar Khan 8 Chapter No. 2 Table 2.3 Design criteria for granular filters (after Geotechnical Eng. Office, Government of Hong Kong, 1993) Rule number Filter design rule Requirement 1 D 1 5 F C < 5 * D 8 5 S F STABILITY (i.e. the pores in the filter must be small enough to prevent infiltration of the material being drained) 2 Should not be gap-graded (i.e. having two or more distinct sections of the grading curves separated by sub-horizontal portions) 3 D 1 5 F f > 5 * D 1 S S C PERMEABILITY (i.e. the filter must be much mire permeable than the material being drained 4 Not more than 5% to pass 63 um sieve and that fraction to be cohesion less 5 Uniformity Coefficient 4 < D 6 0 F / D 1 0 F < 20 SEGREGATION (i.e. the filter must not become segregated or contaminated prior to, during, and after installation). 6 Maximum size of particles should not be greater than 50 mm Notes: In this table, D] 5F is the size of sieve (in mm) that allows 15% by weight of the filter ^ material to pass through. Similarly, D 8 5 S is the size of sieve (in mm) that allows 85% by weight of the base soil to pass through. The subscript c denotes the coarser side of the envelope, and subscript f denotes the fine side. For a widely graded base soil, the original D 9 0 S > 2 mm and D] 0S < 0.06 mm, the above 2. criteria should be applied to the "revised" base soil grading curve consisting of the particles smaller than 5 mm only. ^ The thickness of a filter should not be less than 300 mm for a hand-placed layer, or 450 mm for a machine-placed layer. ^ Rule 5 should be used to check individual filter grading curves rather than to design the limits of the grading envelope. ^ The determination of particle size distributions of the base soil and the filter should be carried out without using dispersants. Abdul Sattar Khan 9 Chapter No. 2 2.3 Review of Filtration Testing Karl Terzaghi (1939) first proposed empirical rules for the protection of errodable cohesionless base soil by a protective filter material, when subjected to excessive seepage flow. Bertram (1940), under the supervision of Dr. Casagrande at the Harvard Graduate School of Engineering, later validated those rules through a systematic laboratory study of filtration phenomenon. Sample preparation with de-aired water was found to be necessary, to prevent tiny bubbles of dissolved air in the water supply coming out of solution, resulting in a reduction of the degree of saturation, and hence permeability. Accumulation of suspended particles in the regular water supply was also found to be problematic, leading to the recommended use of distilled water. Constant head permeability tests were performed (see Figure 2.4) on samples with 50 mm in diameter and 60 mm in length of, the base soil and filter medium respectively. The base soils were uniformly graded fine sands and the filter materials were uniformly graded coarser sands. Some of the tests were executed with well-graded fine to medium gravel filter samples that were 100 mm in diameter. No surcharge load was applied, and downward flow, with few exceptions, was imposed for a test duration of 2 hours (at i = 18 to 20) or 4 hours (at i = 6 to 8). Interpretation of results established that the minimum size ratio (Di 5 F /D 8 5 s) at the limit of stability is approximately 6 and 8 for soil retention. This important observation lent greater confidence to the filtration criteria proposed by Terzaghi and Peck (1948) (see Fig. 2.5) based on the earlier recommendation of Terzaghi (1922). Specifically the soil ratio, D I5F/D 8 5 S , should be less than or equal to 4 to provide an adequate level of safety against particle migration. Abdul Sattar Khan 10 Chapter No. 2 Fig. 2.4 Constant head apparatus (after Bertram, 1940) Abdul Sattar Khan 11 Chapter No. 2 005 0.1 02 03 i 2 5 Grain Size in mm (Log Sca/eJ Fig. 2.5 Filter Criteria (after Terzaghi, 1939) Karpoff (1955) studied the issue of filter design criteria, based on earlier filtration work done by the US Bureau of Reclamation (1947 and 1955). The permeameter used in his study is shown in Fig. 2.6. He used well-graded silt, uniformly graded fine and medium sands, and well graded gravelly sand as the base soils, in combination with a uniform gradation of medium or coarse sand and fine to medium gravel as the filter medium. Unidirectional flow of a regular water supply was imposed under no surcharge load in downward direction. From a synthesis of these experiments, he proposed the following guidelines that later became integral parts of many design specifications: • The filter material should pass the 75 mm (3") screen for minimizing particle segregation and bridging during placement. Also filters must not have more than 5% minus 0.075 mm (Sieve No. 200) particles, to prevent excessive movement of fines in the filter and into drainage pipes, causing clogging • The gradation curves of the filter and the base material should be approximately parallel in the range of the finer sizes, because the stability and proper function of protective filters depends upon skewness of the gradation curve of the filter toward the fines, giving support to the fines in the base Abdul Sattar Khan 12 Chapter No. 2 In designing of filters for base soils containing particles larger than 4.75 mm (Sieve No. 4) size, the base soil should be analyzed on the basis of the gradation smaller than the No. 4 sieve. Fig. 2.6 Permeameter used by Karpoff (1955) Abdul Sattar Khan 13 Chapter No. 2 Kezdi (1979) reported tests on the behaviour of internally unstable soils. These soils were considered to be made up of two major components, termed as the coarser fraction, which acts as a filter, and the finer fraction, considered similar to the base soil. In order to define these fractions on the grain size distribution curve, a split gradation technique was developed (see Fig . 2.7). It was then suggested that i f these two components satisfy Terzaghi's filtration rule (1922), where Dis-f/dgs-s < 4, then the composite gradation w i l l be self-filtering and therefore internally stable. cu cn o c fJu u C U CL cn if) o _> 100 80 60 40 20 0 das 8E ) % \ \ \ •——: 1 2 * * ^ r - rl— J mm, - mmmm — - i ^ 4ds —o— Ad 15 0.5 0.2 0.1 0.05 0.02 0.01 0.005 0.002 0.001 Grain size d, mm Fig. 2.7 Methodology to split the gradation (after Kezdi, 1979) Lafleur (1984) performed tests on base soils that were well-graded gravelly silt sands, representing the glacial tills used in the construction of various earth dams in the James Bay Project, Quebec. The filter materials used for this study were uniform gravels and well-graded gravel-sands. The test set-up is shown in Fig . 2.8: a water head was imposed using calibrated springs to support water tanks, thereby maintaining a constant differential head across the test sample. The sample had a diameter of 150 mm, and a total length of 350 mm, comprising 150 mm of base soil and 200 mm of filter medium. A n effective cell pressure of 100 kPa was applied, over 800 kPa of backpressure to ensure saturation of the sample. A unidirectional flow was imposed in the downward direction, to achieve a hydraulic gradient value of 8 for duration of 50 to 880 hours. Analysis of the experimental data confirms the design guidelines suggested by Karpoff (1955) and Bertram (1940) Abdul Sattar Khan 14 Chapter No. 2 /1 •/•••/ tit ((t ft f •/./ / /, /'//•/ Rubber Membrane Gall Pressure (-BGOkPa) C O U N T E R WCI6HT SWITCHES TO COMMAND 50LINOIO v*uyts OJAMETCfl • I W M M Pressure transducer Synthetic Filter Filter paper Fig. 2.8 Permeameter test assembly (after Lafleur, 1984) Sherard et al. (1984) carried out an extensive series of experiments to better understand the behaviour of filter materials. The base soil was 100 mm in diameter, with a length ranging from 50 to 100mm, and was composed of uniform fine, medium or coarse sand. The filter material was a uniformly graded coarser sand, uniform gravel, or well-graded sandy gravel with a length in the range 125 to 175 mm. No surcharge load is applied, and a unidirectional downward flow regime was imposed for 5 to 10 minutes with a direct supply of tap water Abdul Sattar Khan 15 Chapter No. 2 without any control over the hydraulic gradient (see Fig. 2.9). After this brief exposure to seepage forces, the specimen was then placed on a shake table for one minute if little or no base soil had migrated through the filter medium. Pressure gauge 4" I.D. Water source under pressure Pea gravel (# 4 - 3/8") End Plate • Window screen n Clear plastic cylinder r '~~ (1/4" wall thickness Side material. Uniform sand coarser than base sand and finer than filter Uniform sand or gravel small enough to prevent filter movement 2 - 14" discharge pipes Fig. 2.9 Test set-up (after Sherard et al. 1984) From the experimental results it was concluded that Di5F/d85s 5 is conservative for filters with Di5 greater than 1.00 mm. It was also argued that the particle size distribution curve of the filter is not required to be similar in shape to the base soil. To better understand the concept of internal stability and the influence of seepage force and vibrations, Kenny and Lau (1985) conducted a series of experiments. Constant head tests were performed, with unidirectional flow of tap water in a downward direction, on specimens of approximate diameter of 245 mm and 580 mm. The smaller specimens were 450 mm (base Abdul Sattar Khan 16 Chapter No. 2 soil) in length whereas the larger specimens were 860 mm long (base soil) (see Fig. 2.10). The base soils were well-graded sandy gravels and the filter materials were a uniform medium or coarser gravel and cobbles. These experiments were run for test duration between 30 and 100 hours, with mild vibration applied to replicate the most severe field condition. Interpretation of test results concluded that the potential for grading instability is based on the shape of the particle size distribution curve. Furthermore, instability is more likely to occur in soils that have gently inclined sections (a wide range of particle sizes) in the lower part of their grading curves. Upper reservoir (Oil drum 580 mm 0 x 860 mm) inlet tube (variable length; Seepage cell (245 mm <p x 450 mm) Test specimen Drainage layer Discharge pipe Sedimentation tank with overflow Lower reservoir with pump return Upper reservoir (oil drum) Drums are held together by 4 rods bolted through 2 crossover bars top and bottom , Centring and sealing ring Seepage cell (580 mm 0 x 860 mm) Test specimen Drainage layer Bottom plate contains 25 mm holes Sedimentation tank with overflow Lower reservoir with pump return Fig. 2.10 Permeameter arrangement (after Kenny et al. 1985) Synthesis of test results led to a stability criterion between stable and unstable gradations as defined in Fig. 2.11. As illustrated, a discrete envelope of points (H) is established for selected intervals on the grading curve (F). If the grading curve lies below this envelope of points then the gradation is considered to be unstable. The threshold boundary line is defined by a line H/F = 1.3, but was later on modified as H/F = 1.0 by the authors due to constructive discussions by Ripley (1986), Milligan (1986), and additional work by Sherard et al. (1986). Abdul Sattar Khan 17 Chapter No. 2 SAND GRAVEL Fine IMediurnl Coarse Fine 1 Medium I Coarse Grain size D, mm 0.02 0.06 0.1 0.2 0.61.0 2 6 10 20 60 100 200 Fig. 2.11 Stability criterion (after Kenny et al. 1985) Honjo et al. (1996) performed a series of multi- and single-stage permeability tests to examine the internal stability of broadly and gap-graded soils. For broadly graded samples, gravelly sands were used while the gap-graded samples were sands without the medium fraction. In their test arrangement, base soils samples, 300 mm diameter and 100 mm length (see Fig. 2.12), were placed against a filter mesh of known opening size. A light surcharge of 0.85 kPa was applied to the top of the sample. Tap water was used to apply the required seepage force in downward direction, either by the use of constant head tanks or by direct pumping pressure. Light vibrations, in addition to seepage forces, were also applied to duplicate the severe conditions. Abdul Sattar Khan 18 Chapter No. 2 The duration of these tests was 2 h. From the interpretation of these experiments, they concluded that the diameter of the coarser particles of a broadly graded soil close to the filter opening size (DR7O to DB95) is very important for self filtration process. For successful self-filtration, it is necessary that the base soil's coarser particles start accumulating at the base soils-filter interface followed by the fines agglomerations in constrictions formed by the coarser particles. For gap-graded soils, a gap ratio, ratio between the particle sizes at the location of a gap, of 4 is considered to be the upper bound limit for stability of gap-graded soils irrespective of fines percentage. Tomlinson et al. (2000) examined the effects of confining stress, filter thickness, hydraulic gradient and the rate of gradient increase on a series of base soil-filter combinations. Glass beads were used instead of soils. The base medium (fine sand size equivalent) was 100 mm in diameter and approximately 30 mm long. The filter medium (coarse sand equivalent) had a thickness of 25 to 37 mm. A confining stress was imposed with a magnitude ranging from 50 to 400 kPa. Tap water was circulated in downward direction to attain the required differential head (max. of 100 cm). The test arrangement is shown in Fig. 2.13. From the tests they deduced that combinations with Di5F/Dg 5 B < 8 were immune to piping. Fig. 2.12 Test specimen (after Honjo et al. 1996) Abdul Sattar Khan 19 Chapter No. 2 Additionally a higher confining stress, lesser thickness of filter medium and higher rate of gradient increase were found to have a detrimental effect on the performance of a base soil-filter medium. Computed Loading Frame Loading Ram Regulated Air Supply Pneumatic Piston Load Cell * Displacement Transducer-Differential Pressure '"' Transducer Plexiglas Bath Graduated Cylinder Fig. 2.13 Typical test arrangement (after Tomlinson et al. 2000) Lafleur et al. (2000) established a laboratory-testing program to gain insight to high head losses observed in the core immediately upstream of the filter in the LG-4 main dam of the James Bay project. A series of tests were conducted with or without filter medium (see Fig. 2.14). The direction of flow, of distilled de-aired water, was from top to the bottom. Materials used as the base soil were broadly graded non-plastic moraines, representative of the core material, with a fines ranging from 28% to 6%. Filter materials were gravelly sand or sandy gravels, having a D15 at the lower and upper bound of acceptable filter materials for the Abdul Sattar Khan 20 Chapter No. 2 Water inlet Rame used to appfy bad Fbrous stone Rezo meter 84:6 mm 84,6 mm 84.6 mm| 84.6 mm v_cji moraine sample nt if tn\ y (compacted) I54mfirch- Water outlet grain size anaVs*! . Her paper Boundaries for | DEAD WBGHT | Fig. 2.14 Specimen dimension (after Lafleur, 2000) James Bay project. A surcharge load of 10 kPa was applied in some of the tests and an average hydraulic gradient of 10 was imposed to replicate the typical conditions of the James Bay structures. The duration of these tests varied between 12 hours and 51 days. Analysis of test data showed that blinding or formation of "clay cakes" was observed in specimens with less than 11% of fines. This led to a conclusion that excessive head loss resulted from the compression of gases, and degasification inside the unsaturated compacted moraine cores, with approximately 25 % of fines. Garner et al. (2002) further describes the response of the gap-graded soils, of glacial origin, to seepage flow. A large permeameter test was performed to observe the influence both of hydraulic gradient and gas exsolution. The base soil sample has a length of 660 mm whereas the filter medium has a thickness of 300 mm. The soils were compacted to 96% of their Standard Procter density. De-aired water was circulated from the bottom to saturate the specimen under a seating load of 312 kPa and was elevated to 600 kPa once the water appeared at the top of the specimen. During the first phase of 65 days, the system was subjected to a max hydraulic gradient of 37. During the second phase, gassy water was introduced with the intent of triggering suffosion. They concluded that: 1. The process of suffosion can be triggered by the introduction of gassy water. 2. The concept of self-healing in widely graded cores and filters may not be applicable to gap-graded soils vulnerable to suffosion. Abdul Sattar Khan 21 Chapter No. 2 3. Kenny and Lau internal stability criterion of H/F = 1 could possibly extended to beyond F=20% Moffat (2002) examined selected gradations of Kenny et al. (1985) and Honjo et al. (1996) to assess their internal stability with reference to the empirical rule of Kezdi (1979). The soils used in testing were broadly graded sandy gravels and gap-graded sands with a deficiency of medium sand fraction. They were placed against a wire mesh of known opening size. The test specimen was 100 mm in diameter and approximately 100 mm in length. In multistage tests, a light surcharge pressure of 25 kPa was applied, and distilled de-aired water was circulated. Downward flow was maintained with the constant head tanks, to apply hydraulic gradients in the range of 0.1 to 18.5 (see Fig. 2.15). A pneumatic hammer was used to impose vibration to duplicate severe conditions. He concluded a threshold value of Di5F'/dg5s' = 4 marks the onset of internal instability, which is in agreement to Kezdi's criterion (1979). Fig. 2.15 Flow system (after Moffat, 2002) Careful evaluation of these studies reported in the literature shows there is a large diversity in equipment and material specifications. Therefore a cautious approach is required to develop a Inflow tank Normal stress (25 kPa) L Abdul Sattar Khan 22 Chapter No. 2 broad-based judgment using such a diverse database. A separate issue is that of how to apply judgment developed in studies of base soil against filter medium to predict the internal stability of gap-graded soils. Regarding internal stability, Kezdi (1979), Kenny et al. (1985), and Honjo et al. (1996) all proposed techniques to examine widely graded and gap-graded soils for their potential to exhibit internal instability. In developing these criteria, different hydrodynamic constraints were used in testing hence doubts exists about the interconnectivity of the findings. The current study has been designed to better develop an understanding of gap-graded soils building upon the observations of Moffat (2002). More specifically, it seeks to examine the influence of hydraulic gradient, vertical effective stress and boundary conditions of the permeameter wall on the susceptibility of reconstituted gradation of glass beads to internal instability. Abdul Sattar Khan 23 Chapter No. 2 3.0 APPARATUS, MATERIALS AND PROCEDURES 3.1 Modified Gradient Ratio Device The apparatus used in this study is the Modified Gradient Ratio test device that was designed at UBC (Hameiri, 2000). It is a modified version of the ASTM device (ASTM D 5101-96) that allows for the application of a vertical effective stress to the test specimen, collection of particles passing through the test specimen, and has additional ports along the permeameter wall to provide a more detailed analysis of head loss in the specimen. In addition, the system is completely automated using a process control and data acquisition system via personal computer and custom software. Fig. 3.1 shows a schematic diagram of the UBC Modified Gradient Ratio test device. Axial Force Inlet Constant Head Water Tank LVDT Permeameter Specimen Outlet Constant HeadWater Tank \ J i Length of Specimen, L Applied Constant Head, H Peristaltic Pump Fig. 3.1 Details of the permeameter and flow control Abdul Sattar Khan 23 OL Chapter No. 3 3.1.1 Soil Specimen and Filter Mesh The rigid-wall permeameter (Plexiglas pipe) holds a specimen 100 mm in diameter with a length of approximately 120 mm. The top plate is made up of anodized aluminium with a provision for applying the axial force. A loading piston, with perforations, is placed at the top of the specimen to apply the axial force. Perforations in the loading piston are provided for an unimpeded water circulation through the specimen and, to maintain a constant head across the test specimen. The test specimen rests on a wire mesh, of known opening size, supported by rigid base plate (see Fig. 3.2). This base plate has systematic circular perforations, of 10 mm in diameter on 10 mm center-to-center spacing. The wire mesh is used to provide for soil retention during specimen reconstitution and during subsequent testing. Additionally it provides for an uninterrupted migration of fines under the influence of seepage forces. Wire meshes with opening sizes of 0.31, 0.86 and 1.19 mm were used during this study (see Fig. 3.3). Underneath, is a collection trough, which collects the fines passing through the filter mesh. It comprises two parts. The upper part is a Plexiglas funnel, with an internal slope of 45°, that directs the particles to the lower section. The lower part is a flexible silicon tube that can be clamped at discrete intervals to isolate quantities of migrating particles at any stage during the test. 3.1.2 Water Supply and Control System The water supply system is composed of two constant heads tanks, used to impose a hydraulic gradient that is controlled by length of the specimen L, and the constant differential head H. A peristaltic pump, Model 7529-20 manufactured by Masterflex, was used to pump, distilled, de-aired water from the reservoir, to the inlet constant head water tank. The overflow in the inlet tank is returned to the reservoir, via a tube that maintains the constant head. During this study, the inlet constant head water tank was fixed and the position of outlet constant head water tank was changed to impose the desired hydraulic gradient in any stage of the test. Selection of distilled de-aired water was made to eliminate the effects of dissolved air and suspended particles (Bertram 1940) Abdul Sattar Khan 24 Chapter No. 3 3.1.3 Instrumentation Measurements of the water head, along the length of specimen, were made using the port locations shown in Fig. 3.2. Port no. 1 is located on the top plate to establish the water head at the top of the specimen. Port no. 3, 5 and 6 are located at a distance of 75 mm, 25 mm and 8 mm above the filter mesh. Port no. 7, which establishes the water head below the specimen, is located in the upper part of the collection trough. Axial Force Top Plate From Inlet Constant Head Water Tank Rigid Base Plate (Perforated) To Outlet Constant Head Water Tank Fig 3.2 Different port locations along the length of permeameter The setup consists of four ± 7 kPa (in, i 3 5 , i 5 6 , i67) and one ± 17 kPa (in) differential wet/wet pressure transducers. These differential pressure transducers, measure difference in pressure between two ports, are Model C230 units manufactured by Setra and recorded the water heads to a resolution of ± 1 mm of water. Abdul Sattar Khan 25 Chapter No. 3 The hydraulic conductivity of the reconstituted specimen was deduced form measurement of volumetric flow rate at different time intervals. Volumetric flow rate was measured using the hose connecting the outlet constant head water tank with the reservoir. It is important to note that the most common term used for hydraulic conductivity is permeability and is more prominently acquainted. Therefore the term permeability was used during this study to refer hydraulic conductivity. Axial force was applied to the specimen through the top plate, targeting to impose a chosen vertical effective stress. It was measured using compression load cells having full scale capacity of either 50 or 400 kg and had a resolution of ± 0.01 and 0.1 kg respectively A Linear Vertical Differential Transformer (LVDT) was used to measure displacement of the loading piston in contact with the top of the specimen. The LVDT recorded displacement to a resolution of ± 0.1 mm 3.1.4 Data Acquisition System An electric data logger was used to record all output voltages from the differential pressure transducers, LVDT and the load cell. The system comprised a power supply, a signal conditioning unit that amplifies the output signals and a Metrabyte DAS-16 board connected to a personal computer. The DAS-16 board was a multifunction board with a 12-bit resolution and digital input and output. Software, Labtech Notebook by Laboratory Technologies Corporation, was used to collect the data (seven channels) at a rate of 3 Hz and write it to an output file. 3.2 Materials During this study, test specimens were reconstituted from glass beads of different size ranges (see Fig. 3.4). These glass beads are manufactured by Rotair Industries and Potters Industries Inc. Their chemical composition is soda-lime silica glass. They have a specific gravity of 2.45 to 2.50 with a static coefficient of friction 0.9 to 1.0. These particles are almost perfect spheres. Due to their translucent nature, they have proved very beneficial in detecting anomalies during any experimental phase. Abdul Sattar Khan 2 6 Chapter No . 3 In total nine different size ranges of beads were used: 0.10 to 0.15 mm, 0.15 to 0.246 mm, 0.60 to 0.875 mm, 0.78 to 1.26 mm, 1.20 to 1.70 mm, 1.76 to 2.00 mm, 1.40 to 2.00 mm, 2.00 to 2.80 mm, and 2.80 to 3.35 mm . They were mixed to achieve a particular gradation curve. In order to retain the coarser fractions, a metallic wire mesh of known opening size was used as a filter screen at the bottom of the specimen. The opening size was matched to the gradation curve of the test specimen, to ensure free passage of any migrating fine fraction and to retain the primary fabric or coarser particles in the gap-graded gradation test specimen. The selection of test gradations was governed by the geometric constraints for internal stability. Therefore an effort was made to have a fair range of gradations with small to significantly larger constriction sizes. The use of uniform cohesion less sand or gravely sand as a filter layer was reported as an excellent option (Ripley 1985 and Milligan 1986). On the other hand, little information is available about the filtration capabilities of the gap graded uniform or gravely sands. Based on these observations, five different particle size gradations were selected for testing. Three of them are the selected gradations of Honjo et al. (1996) i.e. modified G-ld, G-3c and G-4c. To fill in the gap between the gradations modified G-3c and G-4c, gradation no.4 was selected whereas to extend the observational range, gradation no. 2 was introduced. It is important to note that all these gradations have a similar particle size range before the gap, at 40% of fines content, resulting in a constant dgs-s value. On the other hand, changes in the coarser particle sizes were made to get the required soil ratios, DIS-F / d Abdul Sattar Khan 2 7 Chapter No. 3 85-s. The gradation characteristics and particle size distribution curves of the five selected gradations are tabulated and shown in Table 3.1 and Fig 3.5 respectively. Table 3.1 Characteristics of the selected gradation curves Gradation D 1 0 (mm) D 1 5 (mm) D 3 0 (mm) D 5 0 (mm) D 6 0 (mm) D 8 5 (mm) Cu C c Soil ratio Dl5F'/d85's Gap ratio Filter Mesh @ (mm) G-ld 0.129 0.144 0.199 0.774 0.848 1.024 6.6 0.36 3.7 2.9 0.31 G-3c 0.129 0.144 0.199 0.967 1.132 1.565 8.8 0.27 4.4 3.9 0.86 No. 4 0.129 0.144 0.199 1.174 1.419 1.725 11.0 0.22 5.9 4.8 1.19 G-4c 0.129 0.144 0.199 1.621 1.897 2.622 14.7 0.16 7.4 5.7 1.19 No. 2 0.129 0.144 0.199 1.908 2.156 2.770 16.7 0.14 8.7 6.9 1.19 Fig. 3.5 Particle size distribution curves of the test gradations Abdul Sattar Khan 28 Chapter No. 3 3.3 Procedures The test procedure consists of reconstituting the glass beads specimen, setting-up the apparatus, running the multi-stage test, and finally acquiring a couple of post-test measurements. 3.3.1 Specimen Reconstitution The objective of the reconstitution technique is to replicate a homogeneous test specimen of similar density and degree of saturation. It was found challenging to reconstitute identical specimens with gradations exhibiting a relatively larger D ^ - F / dg5>s ratio: they were found to be prone to segregation. The method used in this study was a modified slurry deposition technique (Kuerbis, 1989). The glass beads were prepared as slurry and the specimen itself then reconstituted directly in the permeameter using a discrete deposition technique. In this method around 1700 g of glass beads were boiled in a stainless steal bowl with distilled de-aired water for at least 30 minutes. After leaving it to cool down to room temperature (20°C), it was placed under a vacuum for 12 hours. The beads were thoroughly mixed to attain a uniform consistency. Thereafter a small quantity was taken out in a beaker and thoroughly remixed. The specimen was reconstituted under a thin layer of standing water in the permeameter, no deeper than 1 to 1.5 cm, to ensure saturation and yet also achieve minimal segregation of coarse and finer fraction in the gap gradation. This placement technique yields a loose specimen with a dry density in the range of 1.80 to 2.51 Mg/m3. Homogeneity of the specimen was assessed by visual observations, and from the water head distribution recorded along the specimen at the beginning of the experiment. Minor to significant amount of fines loss was observed during the reconstitution process. 3.3.2 TestSet-Up The next stage was to set up the apparatus. First, it was ensured that all the instrumentation and data acquisition system was ready to start taking and recording the measurements. Then the collection trough was filled with distilled de-air water and the filter mesh was placed over the base plate. The permeameter is then lowered down and attached to the base plate. The permeameter was filled to a depth of 1.0 to 1.5 cm above the filter mesh and specimen reconstitution was started by discrete deposition. All the ports were connected with their respective differential pressure transducers as they were submerged under the rising water Abdul Sattar Khan 29 Chapter No. 3 surface, as the length of the specimen increased. Care was taken to ensure all of the port connections were air-free. Once the specimen was reconstituted, a siphon was used to level its top surface to an approximate length of 10 cm. Thereafter the top cover plate assembly, including loading plate and LVDT, was attached and the connection made to the inlet constant head water tank (again ensuring an air-free connection). Finally, the top port no. 1 was connected to the respective differential pressure transducer. 3.3.3 Multi-Stage Test Procedure All the tests were executed in a multi-stage procedure. The multi-stage procedure allowed for an assessment of specimen response to different test variables. Test variable examined were hydraulic gradient across the specimen (typically 1 < in < 20), vertical effective stress on the specimen (25 and 50 kPa) and boundary wall condition of the rigid-wall permeameter (smooth and textured). Specific details are reported in Section 3.4 below. Unidirectional flow of water was imposed from the top to the bottom of the specimen, with the hydraulic gradient was controlled by the difference in height between the inlet and outlet constant head water tanks. All particles of the fine fraction of the specimen passing through the filter mesh, including that during specimen reconstitution, were collected using clamps to separate the amount accumulated during each test stage. 3.3.4 Post-test Observations and Measurements Upon completion of the test, the nature of the particle migration was quantified. The fines fraction passing during each test stage was measured. The dry mass was collected, compared with the initial weight of the specimen, and then reported as percentage of the total mass of the specimen. In addition sieve analyses were done both for all the fractions collected during the test stages along and for different layers of the post-test itself. A series of photographs were taken, while systematically excavating the specimen, after the test. These images were used to compare and visualize the generalized nature of fines migration. 3.4 Program of Investigation A total of twelve tests were performed on five selected gradations (see Table 3.2). A test code is used for reporting purposes, with reference to the test variables that were examined. Details of the observed system hydraulic gradient, in and duration of every test stage are reported in Abdul Sattar Khan 30 Chapter No. 3 Table 3.3. It is important to note that the reported in values in Table 3.3 are at the start of the test stage. Also one test was repeated (5.9T25R), to allow for the issues of operator dependence to be addressed. Table 3.2 Test code developed for the study Test Soil Boundary Vertical code ratio condition effective D|5'F / d 85's stress (kPa) 3.7S25 3.7 Smooth (S) 25 4.4S25 4.4 Smooth (S) 25 5.9S25 5.9 Smooth (S) 25 7.4S25 7.4 Smooth (S) 25 8.7S25 8.7 Smooth (S) 25 4.4T25 4.4 Textured (T) 25 5.9T25 5.9 Textured (T) 25 5.9T25R 5.9 Textured (T) 25 7.4T25 7.4 Textured (T) 25 8.7T25 8.7 Textured (T) 25 5.9S50 5.9 Smooth (S) 50 7.4S50 7.4 Smooth (S) 50 Table 3.3 Summary of the observed system hydraulic gradients, i , 7 and duration of the test stages Test code Dry Stage No. 1 Stage No. 2 Stage No. 3 Stage No. 4 Stage No. 5 density ' ii7 Period in Period in Period in Period in Period Mg/m3 (min) (min) (min) (min) (min) 3.7S25 1.88 5.0 90 - - - - - - - -4.4S25 1.98 1.0 60 4.8 60 10.1 60 18.1 60 - -5.9S25 1.94 1.0 30 8.7 60 11.8 60 14.2 30 17.5 5 7.4S25 2.11 0.6 30 - - - - - - - -8.7S25 2.51 Unable to reconstitute the specimen 4.4T25 1.80 1.0 30 5.4 30 9.9 30 17.1 30 - -5.9T25 1.84 1.0 30 5.3 12 - - - - - -5.9T25R 1.85 1.0 30 9.2 3.5 - - - - - -7.4T25 2.08 0.4 45 - - - - - - - -8.7T25 2.21 Unable to reconstitute the specimen 5.9S50 1.99 1.0 30 9.6 30 12.0 30 13.7 30 18.2 4 7.4S50 2.12 0.4 30 0.6 30 1.0 30 1.5 30 - -Abdul Sattar Khan 31 Chapter No. 3 In order to observe the boundary wall effect of the rigid-wall permeameter, glass beads of 2.0 to 2.8 mm diameter were glued with silicone glue to the walls of permeameter. A schematic view of this modification and the top view of the permeameter with the textured wall are shown in Fig. 3.6 and Fig. 3.7. Wall of the permeameter Layer of silicone glue Fig. 3.6 Schematic diagram of the textured wall of the rigid-wall permeameter Fig. 3.7 Top view of the textured wall permeameter Abdul Sattar Khan 32 Chapter No. 3 Selection of test stage duration was based primarily on observations of previous researchers like Bertram (1940) who reported that "It was observed that the movement of the base layer and the rearrangement of the grains occurred only at the very beginning of a test and that visible movement ceased within three to five minutes. On the basis of this investigation, the filter tests were operated for approximately 4 hours for low hydraulic gradients and approximately for 2 hours for higher gradients". Therefore, in this research, each test stage was lasted for 30 minutes on average resulting in total test durations between 30 minutes to 4 hours. Some of the test stages were longer than 30 minutes to reconfirm the time dependency of the fines migration. Additionally to accelerate the onset of material instability, and to address the shorter test duration compared with the actual field conditions, vertical effective stress and the seepage forces were applied at a higher rate (always less than 1 minute). Abdul Sattar Khan 33 Chapter No. 3 4.0 TEST RESULTS To address the objectives set forth for this study, test results are presented that show the response of the test specimen to seepage flow. These results include measurements of mass of fine fraction passing, water head distribution, volumetric flow rate and the specimen length. Post-test measurements comprise a gradation analysis of the specimen by discrete layers, and of the fine fraction passing. Visual observations and video images were used to complement the test measurements. The mass of fine fraction passing through the filter mesh is first summarized. Based on these values, selected pre and post-test gradations, and water head distributions, are reported. Data on permeability and axial strain in the test specimen are described. Finally results are reported for two tests that were performed to demonstrate the repeatability, of the test procedures. A complete set of pre and post-test gradations, and water head distributions, is provided in Appendix B and C. Also a comprehensive set of pictures for every experiment, during and after the tests, is given in Appendix D. 4.1 Head Losses in the Permeameter During the series of experiments, observation showed that an increase in volumetric flow rate led to greater system losses (hi 7) through the hoses and connections, which in turn diminishes the system hydraulic gradients in (=hn/L). This phenomenon is more pronounced in soils exhibiting a higher permeability: from the inspection of Fig. 4.1 and Fig. 4.2, it is evident that the difference between the applied and measured system hydraulic gradient is smaller in test 4.4S25. This observation confirms the need for measuring the hydraulic gradient across the test specimen. Pishe (2001) and Moffat (2002) also reported this observation while working on the same permeameter. Additionally from Fig. 4.1 and 4.2, it appears that the volumetric flow rate varies linearly with the hydraulic gradient hence proving the validity of Darcy's law. This validation appears reasonable, assuming that the fine fraction is controlling the permeability. From Table 4.4, for test 5.9T25, permeability values remain constant providing the specimen is homogeneous. But, at the onset of material instability or migration of the fine fraction, permeability values increased rapidly and resulted in the considerable loss in applied system gradient. Abdul Sattar Khan 34 Chapter No. 4 25 i 0.0 5.0 10.0 15.0 20.0 25.0 30.0 Volumetric flow rate (cm3/sec) Fig. 4.1 Variation in the applied and observed system gradient with volumetric flow rate for test 4.4S25 Fig. 4.2 Variation in the applied and observed system gradients, i 1 7 , with volumetric flow rate for test 5.9S25 Abdul Sattar Khan 35 Chapter No. 4 4.2 Loss of Fine Fraction from the Specimen The loss of fine fraction of the gap-graded specimen, measured during the test comprised two components. The first portion is that collected during the process of specimen reconstitution, whereas the second is that washed out of the specimen under the influence of seepage forces. These fractions were carefully collected, and weighed, for every stage of each test. The measured values are given in Table 4.1. Observations are reported as a mass per unit cross-sectional area of the specimen, and a percentage of the initial mass of the specimen. For reference, Lafleur (1989) proposed a value of 2500 g/m2 as a threshold boundary for the onset of an unacceptable piping condition in Geotextile filters. Table 4.1 Summary of fine fraction passing through the filter mesh Tes t code D u r i n g spec imen reconstitut ion W i t h seepage forces T o t a l loss (g/m 2) (%) (g/m 2) (%) (g/m 2) (%) 3.7S25 1103 0.6 83 0.1 1186 0.7 4 .4S25 8330 4.3 884 0.5 9214 4.7 5.9S25 8797 4.5 20013 10.0 28810 14.5 7.4S25 4 2 0 9 9 17.3 30541 12.5 72640 29.8 8.7S25 72010 29.3 - - 72010 29.3 4 . 4 T 2 5 6053 3.3 2451 1.3 8504 4.7 5 .9T25 8813 4.7 21178 11.6 29991 16.3 5 . 9 T 2 5 R 8657 4.6 22647 12.1 31304 16.7 7 .4T25 42166 18.1 25807 11.1 67972 29.2 8 .7T25 62065 25.8 - - 62065 25.8 5.9S50 10175 5.1 8631 4.3 18806 9.4 7.4S50 46698 20.7 15678 6.9 62376 27.6 Inspection of Table 4.1 indicates an orderly variation of fine fraction loss. Increasing loss occurs with an increase in the soil ratio DJS-F / dg5> s, which ranges from 3.7 to 8.7. The results are classified in four major categories. The first is that for which negligible fine fraction loss occurs (1500 g/m ) both during specimen reconstitution and permeation was observed. Test 3.7S25, lies in this category. Abdul Sattar Khan 36 Chapter No. 4 10 c 0= o r-50 i 45 40 35 30 25 20 15 10 5 0 • Smooth wall (25 kPa) O Textured wall (25 kPa) X Smooth wall (50 kPa) X • 0 • • 10 0 1 2 3 4 5 6 7 8 9 Dl5T/d85"s Fig. 4.3 Total fine fraction loss for different gradations The second category is that for which a significant level of fine fraction (~10,000g/m ) was lost during specimen reconstitution, followed by a negligible loss under seepage flow. It was observed that the relatively large loss during specimen reconstitution was a result of two factors: 1. Material segregation during the discrete depositional process, and 2. Application of the vertical effective stress. Typically the movement of finer particles was initiated by a faster application of the vertical effective stress. These fine fractions were lost from the lower part of the specimen. Test 4.4S25 falls in this category. The third category defines significant amount of fine fraction loss both during specimen reconstitution and under the influence of seepage flow. Applied hydraulic gradients and resulting fines loss are reported for each test stage in Table 4.2 and 4.3. For example in test 5.9S25, minor losses were observed in first the four test stages, during which the gradient has raised from in = 1 to in to 14.2. Later, when it was subjected to a system hydraulic gradient in = 17.5, a comprehensive but localised fines loss commenced at the side and centre of the Abdu l Sattar K h a n 37 Chapter No. 4 specimen. This was followed by the formation of a fine fraction deficient zone or pipe along full length of the specimen. Typically any fine fraction migration was either localized to a certain part of the specimen (piping) (see Fig. 4.4) or occurred universally through out the entire cross section and length of the specimen (suffosion) (see Fig. 4.4). During Test 5.9S25, a localized pipe was developed where as for Tests 7.4S25, 5.9T25, 5.9T25R, 7.4T25, 5.9S50 and 7.4S50 universal fine fraction migration or suffosion was observed. Fig. 4.4 Development of pipe leads to the localized fine loss through the specimen (Test: 5.9S25) Fig. 4.5 Universal fines loss throughout the specimen in test 7.4T25 Abdul Sattar Khan 38 Chapter No. 4 Table 4.2 Information of the observed system hydraulic gradients, in, for every test stage Test in at the inat the i 1 7 at the in at the in at the code start of start of start of start of start of stage no. 1 stage no. 2 stage no. 3 stage no. 4 stage no. 5 3.7S25 5.0 - - - -4.4S25 1.0 4.8 10.1 18.1 -5.9S25 1.0 8.7 11.8 14.2 17.5 7.4S25 0.6 - - - -8.7S25 - - - - -4.4T25 1.0 5.4 9.9 17.1 -5.9T25 1.0 5.3 - - -5.9T25R 1.0 9.2 - - -7.4T25 0.4 - - - -8.7T25 - - - - -5.9S50 1.0 9.6 12.0 13.7 18.2 7.4S50 0.4 0.6 1.0 1.5 -Table 4.3 Summary of the mass of fine fraction loss during each test stage Test Stage Stage Stage Stage Stage code no. 1 no. 2 no. 3 no. 4 no. 5 g/m2 (%) g/m2 (%) g/m2 (%) g/m2 (%) g/m2 (%) 3.7S25 83 (0.1) - - - -4.4S25 176 (0.1) 188 (0.1) 208 (0.1) 300 (0.1) -5.9S25 151 (0.1) 3479(1.7) 2145 (1.1) 930 (0.5) 13045 (6.6) 7.4S25 30541 (12.5) - - - -8.7S25 - - - - -4.4T25 235 (0.1) 530 (0.3) 627 (0.3) 1060 (0.6) -5.9T25 621 (0.3) 21115 (11.3) - - -5.9T25R 762 (0.4) 21885 (11.7) - - -7.4T25 25807(11.1) - - - -8.7T25 - - - - -5.9S50 949 (0.5) 2134(1.1) 517(0.3) 1523 (0.8) 3508 (1.8) 7.4S50 6853 (3.0) 3523 (1.6) 2050 (0.9) 3252(1.4) -The fourth category describes a continuous fine fraction loss during specimen reconstitution. In these tests, using extremely internally unstable gradation, the fine fraction was falling through the large pore size constrictions under its own weight. The specimens shed almost 70 Abdul Sattar Khan 39 Chapter No. 4 % of their fines within few minutes in a process of segregation that occurred in the absence of seepage flow (see Fig. 4.5). Tests 8.7S25 and 8.7T25 are of this type. 4.3 Pre and Post Test Gradations The particle size gradation curve before and after testing provides valuable information on the size of fine fraction passing through the filter mesh, and hence changes in the gap-gradation of the specimen. This could efficiently be used to interpret the integrity of a reconstituted specimen and subsequent water head distributions during different test stages. All pre and post test gradations are provided in the Appendix B. Evaluation of these gradation changes suggests that four distinct scenarios of particle loss were observed during this study. In the first scenario, a negligible amount of fine fraction was lost during specimen reconstitution and washed out under the action of seepage forces. Only a slight change was observe in the gradation of the bottom third part of the specimen, where the fine fraction was reduced (see Fig. 4.6). Test 3.7S25 is an example of this scenario where the fine fraction was reduced from an original value of 40% to 39.3%, which represents a negligible change. O) I n i_ a> c c g 0. 100 90 80 70 60 50 40 30 20 10 • Driginal gradation Jpper third part of the specimen vliddle third part of the specimen Bottom third part of the specimen - • • fian Rat in = ' ! . 9 i — D m = 0.78 m . _ d, s., = 0.21 m _ I W d , 5 . . = 3 Set of Sieve #16 (1.18 r — #25 (0.71 n #60 (0.246 — #140 (0.106 n 1 m Wire Mesh _ @ 0.31 mm i l . 7 1 m -1 f I « s: • 4 rm) m) / mm) mm) # m 0.01 0.10 Opening Size (mm) 1.00 Fig. 4.6 Grain size distribution curves for specimen used in test 3.7S25 10.00 In the second scenario, a significant amount of the fine fraction was lost during specimen reconstitution and a negligible amount then collected during permeation. Visual observations suggest these particles originated from the lower third part of the specimen. This was later Abdul Sattar Khan 40 Chapter No. 4 confirmed by the post-test gradation analysis in the case of test 4.4S25 (see Fig. 4.7). In the lower third of the test specimen the fine fraction was reduced to 29%, while the remaining two thirds of the specimen was unchanged. , 0.10 Opening size (mm) 1.00 10.00 Fig 4.7 Gradation curves for specimen used in test 4.4S25 The third scenario involves a significant loss of the fine fraction during specimen reconstitution, primarily from the lower 3 cm of the specimen, and further losses as a result of seepage flow. Tests 5.9S25, 5.9T25, 5.9T25R, 5.9S50 and 7.4S50 represent this scenario. For test 5.9S25, the lower part of the specimen had a 22% fine fraction whereas the upper part region had a 30% fraction (compared to the original value of 40%). The central part exhibit an average fine fraction of 35% (see Fig. 4.8). The fourth scenario involves a very significant loss of the fine fraction during specimen reconstitution and, in some cases, seepage flow. In the case of two specimens (8.7S25 and 8.7T25) it was impossible to subject them to permeation. The particle loss during reconstitution represented complete segregation. Those specimens subjected to the seepage flows and lost their fines very rapidly. Tests 7.4S25, 7.4T25 and 7.4S50 represent this scenario. For test 7.4S25, particle loss throughout the specimen left a variable fine fraction between 0.5 % and 8.0 % (see Fig. 4.9) Abdul Sattar Khan 41 Chapter No. 4 0 0 1 0.10 Opening size (mm) 1.00 Fig. 4.8 Grain size distribution curve for specimen used in test 5.9S25 10.00 0.01 0.10 Opening size (mm) 1.00 Fig. 4.9 Gradation curve for specimen used in test 7.4S25 10.00 Abdul Sattar Khan 42 Chapter No. 4 4.4 Water Head Distributions One of the primary objectives of this study was to examine the hydrodynamic constraints governing internal instability in gap-graded test specimens. This issue is now examined with reference to the variation of water head distribution with time. A complete set of water head distribution curves for all the experiments is given in Appendix C. From inspection, the water head distribution curves can be considered in one of three categories (see Fig. 4.10). A homogeneous test specimen, yields a uniform head loss along the Water head (cm) Fig. 4.10 Three exclusive behaviours deduced from the water head distributions length of the specimen, resulting in an "ideal" linear relation. Material segregation caused either during specimen reconstitution or under permeation, results in a non linear relation. If the fine fraction is mobile and is moved out from the lower part of the specimen, the head loss in that region diminishes and the curve lies above the linear response. On the other hand, during internal migration, any accumulation of the fine fraction will increase the head loss in the lower part of the specimen and the water head distribution curve will be below the "ideal" curve of the homogeneous specimen. For purposes of demonstration, four tests are selected, each of which illustrates a characteristic response. The water head distribution curves for test 4.4S25, shown in Fig. 4.11, confirm that the lower 3 cm of the specimen had a diminished concentration of the fine fraction. This loss of particles yielded a reduction in head loss: the remaining portion of the specimen showed a reasonably linear distribution of water head. That Linear relation was independent of hydraulic gradient. This interpretation is well-matched with the observations based on pre and post test gradations that showed a reduction in the level of fines in bottom third portion. This specimen remained stable under hydraulic gradients up to 18.1, with a negligible fine fraction loss. The same kind of behaviour was observed in test 5.9S50, (see Fig. 4.12), which lost more fine fractions when subjected to permeation. 0 20 40 60 80 100 120 140 160 180 Water head (cm) Fig. 4.11 Water head distributions measured during test 4.4S25 For test 5.9S25, the water head distribution curves validate the reduction in fine fraction in the upper and lower third of the specimen (see Fig. 4.13). This particle loss resulted in the reduction of head losses. Whereas, the middle third of the specimen retained most of the original gradation: represented by fairly linear water head distributions. This observation is compatible with the conclusion drawn from the pre and post test gradations suggesting fine fraction deficiency in the upper and lower third of the specimen. The test specimen failed under a hydraulic gradient of 17.5 resulting in significant loss of fines and hence significant reduction in the head loss along the length of the specimen. Abdul Sattar Khan 44 Chapter No. 4 Water Head (cm) Fig. 4.12 Water head distributions for the test 5.9S50 £ 5 E — p ' • """"" m A \ * \ r — / — / — / / — / — / — * * / / /—• / * / / L-g I * T ? * I • . J _ • o iaye I N O . I , n/ = i.u Stage No. 2, i17 = 8.7 —A—Stage No. 3, i17 =11.8 —X - Stage No. 4, i17 = 14.2 —Hi - Stage No. 5, i17 = 17.5 1 I i .' 0 20 40 60 80 100 120 140 Water hnaH trjm\ Fig. 4.13 Water head distributions measured for the test 5.9S25 For experiment 7.4S25, water head distribution curves (see Fig. 4.14) reveal that the lower most and upper third of the specimen had significantly diminished fine fraction concentration at the start of first test stage. This observation complements the significant particle loss exhibited during specimen reconstitution. Under permeation, universal fine fraction loss through the entire cross section and length of the specimen (suffosion) was observed. Deficiency of the fine particles resulted in the reduction in head loss along the length of the Abdul Sattar Khan 45 Chapter No. 4 specimen. This mass movement of fine fraction is compatible with the observations based on pre and post-test gradations. s 2 i l o £ E o •b 8 c 3 w 5 10 9 8 7 6 5 4 3 2 1 0 • 01 Minute, i17= 0.6 — Time = 05 Minutes, i17 = 0.6 0 1 2 3 4 Water head (cm) Fig. 4.14 Water head distribution for test 7.4S25 4.5 Permeability Based on the measurements of water head and volumetric flow rate, coefficients of permeability were calculated for discrete intervals of the specimen. Specifically, the values of interest are kjs and kn which describe permeability of the central (likely undisturbed) portion and the average permeability of the specimen respectively. These values are summarized in Table 4.4 Recognizing that, for most of the tests, there was no significant fine particle loss during Stage no. 1, k 3 5 is assumed representative of the initial permeability of the specimen. Permeability values were from 11.9 to 18.8 x 10"3 cm/sec in those tests. This narrow range is attributed to the common fine fraction in all test gradations (see Fig. 3.5). It also implies a uniformity between specimens and hence a repeatability of the reconstitution procedure. Nearly constant values of permeability during the respective stages of imposed hydraulic gradient are interpreted as evidence of little migration of fine fraction particles. Internal instability is evident in the large concurrent increase in permeability. Abdul Sattar Khan 46 Chapter No. 4 Table 4.4 Average permeabilities (x 10"3 cm/sec) for different test stages Test After stage no. After stage After stage After stage After stage no. code 1 no. 2 no. 3 no. 4 5 k35 k17 k35 k17 k35 k,7 k35 k17 k35 k17 3.7S25 14.8 13.8 - - -• - - -. - -4.4S25 11.9 14.4 12.8 15.8 12.9 16.2 12.9 16.2 - -(Specimen 5.9S25 12.8 15.5 11.7 18.4 13.4 18.1 12.6 17.4 failed during stage no. 5) 7.4S25 510.5 891.1 (Specimen failed during stage no. 1) 8.7S25 Unable to reconstitute the specimen 4.4T25 16.4 19.3 3.4 3.3 14.7 15.1 15.1 15.6 - -5.9T25 15.1 18.8 179.4 175.5 (Specimen failed during stage no. 2) 5.9T25R 18.8 21.4 (Specimen failed during stage no. 2) 7.4T25 1007.8 816.9 (Specimen failed during stage no. 1) 8.7T25 Unable to reconstitute the specimen 5.9S50 13.7 15.8 13.8 17.4 14.1 15.7 14.2 16.1 - -7.4S50 301.7 405.1 417.1 419.8 (Specimen failed during stage no. 2) 4.6 Axial Strains attributed to the Seepage Forces To correlate fines loss, attributed to the influence of seepage forces, with change in specimen length, a summary of axial strains during testing is presented in Table 4.5. The specimen is composed of coarse and fine grained particles. Axial strains in the absence of fine fraction loss are suggestive of specimen consolidation, and are always smaller in magnitude. On the other hand, axial strains associated with significant loss of particles imply restructuring of the particle and are larger in magnitude. Interpretation of the data suggests the results can be differentiated in three categories. First, specimens that experienced little fine fraction loss during permeation yielded a relatively small axial strain (< 2 %). Tests 3.7S25, 4.4S25 and 4.4T25 fall in this category. Second, specimens with a significant level of fine fraction loss (between 1,500 g/m2and 10,000 g/m2) exhibit an intermediate level of axial strains (< 10%). Test 5.9S50 is of this type. For the third category, significant fine fraction loss under permeation (10,000 g/m2), the result was a Abdul Sattar Khan 47 Chapter No . 4 relatively large axial strain greater than 10%. Tests 5.9S25, 7.4S25, 5.9T25, 5.9T25R, 7.4T25, and 7.4S50 fall in this category. Table 4.5 Summary of specimen lengths after each test stage Test code After At the After After After After After Axial specimen start of stage stage stage stage stage strain recons. stage No. 1 no. 1 no. 2 no. 3 no. 4 no. 5 (attributed to seepage only) (cm) (cm) (cm) (cm) (cm) (cm) (cm) (%) 3.7S25 9.80 9.70 9.70 - - - - 0.00 4.4S25 9.94 9.94 9.86 9.84 9.82 9.81 - 1.30 5.9S25 10.29 10.09 10.05 9.68 9.47 9.40 9.11 9.71 7.4S25 11.69 10.99 9.43 - - - - 14.19 8.7S25 Unable to reconstitute the specimen 4.4T25 10.25 9.95 9.94 9.92 9.87 9.77 - 1.80 5.9T25 10.29 9.99 9.96 8.89 - - - 11.04 5.9T25R 10.29 10.09 10.04 8.60 - - - 14.76 7.4T25 11.29 10.79 9.23 - - - - 14.45 8.7T25 unable to reconstitute the specimen 5.9S50 10.19 9.99 9.99 9.85 9.65 9.63 9.40 5.91 7.4S50 10.79 9.69 9.34 9.09 9.08 9.07 - 6.39 4.7 Repeatability It is important to demonstrate the test results are repeatable and therefore provide a reliable basis of interpretation. With respect to specimen reconstitution, two tests up to stage 1 of seepage flow were performed with identical test variables. The results of these tests, 5.9T25 and 5.9T25R, are summarised in Table 4.6. During specimen reconstitution, a consistent loss of fine fraction was observed, which that confirms the replication of the specimen reconstitution technique. Similarly, particle migration out of the specimen during test Stage no. 1 (in = 1 in both cases) and compatible axial strains of 0.5% and 0.3% respectively further confirm the reproductively of the data. Water head distribution curves, for both tests, are shown in Fig. 4.15. As previously noted, in 5.9T25, fine fraction loss during specimen reconstitution was primarily from the lower third of the specimen, whereas the remaining portion of the specimen is homogeneous. Based upon these observations and measurements, Abdul Sattar Khan 48 Chapter No. 4 comparable values of dry densities, similar loss of fine fraction during specimen reconstitution and under seepage forces, along with similar water head distribution curves, it is concluded that the test procedure is repeatable and the test results are reliable. Table 4.6 Summary of tests for repeatability Test code Dry density (Mg/m3) Loss of fine fraction during specimen reconstitution g/m2 (%) Loss of fines during stage no. 1 g/m2 (%) Length of specimen before stage no. 1 (cm) Length of specimen after stage no. 1 (cm) Axial strain (%) 5.9T25 5.9T25R 1.84 1.85 8813 (4.7) 8657 (4.6) 621 (0.3) 762 (0.4) 10.09 9.99 10.04 9.96 0.5 0.3 2 3 4 5 6 7 8 9 10 11 Water head (cm) Fig. 4.15 Water head distribution curves for tests 5.9T25and 5.9T25R Abdul Sattar Khan 49 Chapter No. 4 5.0 ANALYSIS AND DISCUSSION Test data, from the series of experiments were analysed to address the objectives set forth in Chapter 1. The analysis examines the concept of a critical hydraulic gradient, and therefore, the loss of fine fraction through the specimen. The influence of boundary conditions in the rigid-wall permeameter on the onset of instability is considered, as is the influence of vertical effective stress on the response to seepage flow. In a discussion of the findings, results from the current study are compared with data reported from other studies. In closing, a brief commentary is made on the validation of criteria for internal stability proposed by Kezdi (1979) and Kenny and Lau (1985, 1986). 5.1 Critical Hydraulic Gradient It is generally understood that if the gradation curve of soil exceeds a threshold Di5>F/d85*s value, a geometric constraint to the migration of the finer fraction is lost and internally instability becomes governed by a hydrodynamics. With regard to the influence of hydraulic gradient, the term critical hydraulic gradient (icr), has extensively been used in the literature. There appears to be some variation in its definition. For purpose of the current study i c r is defined as the hydraulic gradient that causes an appreciable movement of the fine fraction out of the specimen, leading to the formation of a fines deficient channel or pipe within the specimen (piping) or resulting in a universal fines loss (suffosion) throughout the specimen. Development of the preferential flow path(s) is associated with a noticeable reduction in the value of applied hydraulic head. Table 5.1 provides a summary of the hydraulic gradient (i)7) at the start of each test stage, duration of the stage and associated loss of fine fraction (% of the total mass of the specimen at the start of the test). Additionally, hydraulic head measured across the specimen before and after the test stage is also reported. It has been observed that excessive fine fraction loss (piping and suffosion) resulted in a rapid increase in the average permeability of the specimen that, in turn, led to an increase in volumetric flow rate. This greater flow rate caused higher losses in the hoses and connections, and resulted in a significant reduction in the applied hydraulic head across the specimen. Abdul Sattar Khan 50 Chapter No. 5 Table 5.1 Summary of the influence of the hydraulic gradients Test Stage in at the Duration Loss of h1 7 H, 7 % red. code no. start of fines (at the start (at the end in h ] 7 the test during of test of test stage permeation stage) stage) (min) (%) (cm) (cm) (%) 3.7S25 1 5.0 90 0.05 48.8 48.8 0.0 4.4S25 1 1.0 60 0.09 10.5 10.5 0.0 2 4.8 60 0.1 47.1 46.9 0.5 3 10.1 60 0.1 98.9 98.9 0.0 4 18.1 60 0.12 178.2 177.5 0.4 5.9S25 1 1 30 0.08 11.0 10.9 1.1 2 8.7 60 1.8 95.1 93.4 1.8 3 11.8 60 1.1 114.3 114.1 0.2 4 14.2 30 0.5 137.1 132.8 3.1 5 17.5 5 6.6 165.1 90.6 45.1 7.4S25 1 0.6 30. 12.5 6.9 1.0 85.2 8.7S25 Unable to reconstitute the specimen 4.4T25 1 1.0 30 0.1 10.4 10.4 0.4 2 5.4 30 0.3 55.8 55.3 0.9 3 9.9 30 0.3 100.8 99.3 1.5 4 17.1 30 0.6 168.7 168.5 0.1 5.9T25 1 1.0 30 0.3 10.1 10.1 0.4 2 5.3 12 11.3 53.5 13.3 75.2 5.9T25R 1 1.0 30 0.4 10.6 10.4 2.4 2 9.2 3.5 11.7 91.4 40.7 55.5 7.4T25 1 0.4 45 11.0 4.6 1.6 65.1 8.7T25 Unable to reconstitute the specimen 5.9S50 1 1.0 30 0.5 10.2 10.1 1.0 2 9.6 30 1.1 95.2 93.9 1.4 3 12.0 30 0.3 119.1 119.0 0.1 4 13.7 30 0.8 134.6 133.9 0.5 5 18.2 4 1.8 175.5 175.1 0.2 7.4S50 1 0.4 30 . 3.0 2.8 2.6 7.2 2 0.6 30 1.6 4.6 4.6 1.3 3 1.0 30 0.9 9.3 7.5 19.2 4 1.5 30 1.4 13.4 8.5 37.1 Abdul Sattar Khan 51 Chapter No. 5 It is evident from the inspection of Table 5.1 that, during tests 5.9S25, 7.4S25, 5.9T25, 5.9T25R, 7.4T25 and 7.4S50, a critical hydraulic gradient was encountered in those test stages in which significant fine fraction loss is associated with a marked reduction in the value of applied hydraulic head. For test 7.4S50, small particle migration was observed at the critical hydraulic conditions. This apparent departure form the definition of the critical hydraulic Fig. 5.1 Development of pipe through the specimen in test 5.9S25 Fig. 5.2 Universal fines loss through the specimen (suffosion) in test 8.9T25 Abdul Sattar Khan 52 Chapter No. 5 gradient is because of higher fine fraction loss during specimen reconstitution (see Table 5.2). During test 5.9S25, a fines deficient conduit was observed resulting in the piping at in = 17.5 (see Fig. 5.1). Whereas, for test 7.4T25, a universal fines loss (or suffosion), occurred at in of 0.4 (see Fig. 5.2) 5.1.1 Influence of the Boundary Conditions in the Rigid-Wall Permeameter From Table 5.1, it is evident that the critical hydraulic gradient was not encountered during the tests on more benign or internally stable gradations. Tests 3.7S25, 4.4S25 and 4.4T25 may be considered in this category. Hence, a threshold value of Di5-F/dg5'S that represents a shift from the influence of a geometric constraint to hydrodynamic control could be deduced as between 4.4 and 5.9 (see Fig. 5.3). The observation must be qualified by the operating range of the test device, which permits a maximum system hydraulic gradient in = 20. 20 -I 18 -16 -C 14 -Oi 2 12 CO o 3 10 -2 "D >* 8 --C 6 -O 4 -2 -0 1 • Smooth wall (25kRa) A Textured wall (25 kFa • Smooth wall (50 kFa) J 14.2 to 17.5 5.9-r25 1.0 to 9.2 b.yi25K 1.0 to 5.3 • 1 8 10 Fig. 5 D 1 5 F ' / d 85's 3 Influence of the boundary conditions and elevated vertical effective stress over critical hydraulic gradient For the gradations having a DIS'FAUJS'S, greater or equal to 5.9, critical hydraulic gradient was typically encountered. A ratio of Dis-rVdss's = 5.9 yields a complex relation. The critical hydraulic gradient is between 14.2 and 17.5 for the smooth wall, and 1.0 to 9.2 for the textured wall. A ratio of Dis /^dss's equals 7.4 demonstrates no influence of boundary conditions on the critical hydraulic gradient, which is between 0.4 and 1.5. Additionally it was noticed that, Abdul Sattar Khan 53 Chapter No. 5 during tests 5.9T25 and 5.9T25R, the specimens took 10 minutes to fail at the critical hydraulic gradients whereas higher hydraulic gradients in excess of the critical value failed the specimens in less than 5 minutes. 5.1.2 Influence of the Elevated Vertical Effective Stress A comparison of four tests (5.9S25, 5.9S50, 7.4S25 and 7.4S50) examines the effect of vertical effective stress. From Table 5.1 it is evident that the increase in vertical effective stress led to an elevated critical hydraulic gradient. For the gradations with Dis^ /dgs-s = 7.4, the critical hydraulic gradient appears slightly greater at higher stress (see Fig. 5.3). The gradation, with D '^p/dgs's = 5.9, yielded a moderate fine loss (6.6%, at ip = 17.5, over 5 minutes) associated significant reduction of hi7 (45.1%) at a 25 kPa confining stress. This represents a failure condition that was not observed in the companion test at higher effective stress (5.9S50). 5.2 Loss of fines from the Test Specimen Loss of the finer fraction from the test specimen has been considered in two categories. First, the loss associated with specimen reconstitution. This depends upon the care and precision of the operator, size and shape of the particles and the opening size of the wire mesh screen over which the specimen rests. From Table 4.1, inspection has shown the fine fraction loss during specimen reconstitution is independent of the boundary conditions in the rigid-wall permeameter. This observation confirmed the replication of reconstitution technique used during this study. The second category is the loss occurring during the permeation process. In all previous tables, fines loss is expressed either as mass per unit cross sectional area of the specimen (g/m ) or % of the total mass of the specimen at the start of the test. However, to obtain a better understanding of fine fraction migration, loss in Table 5.2 is reported in % of the total mass of the fine fraction present at the start of the test. Abdul Sattar Khan 54 Chapter No . 5 Table 5.2 Summary of the normalized fines loss in % of the total fines present in the test specimen at the start of the test Test Mass of fines in . Mass of Normalized During During Code the specimen total fines fines specimen Permeation at the start of test loss loss recons. (g) (g) (%) (%). (%) 3.7S25 591.4 9.6 1.6 1.5 0.1 4.4S25 631.5 74.6 11.8 10.7 1.1 5.9S25 640.0 231.4 36.2 11.1 25.1 7.4S25 791.4 588.9 74.4 43.1 31.3 8.7S25 796.3 583.7 73.3 73.3 -4.4T25 589.9 68.9 11.7 8.3 3.4 5.9T25 606.9 247.6 40.8 11.8 29.0 5.9T25R 608.3 253.8 41.7 11.5 30.2 7.4T25 754.3 550.8 . 73.0 45.3 27.7 8.7T25 768.9 436.5 64.6 64.6 -5.9S50 643.3 152.4 23.5 12.7 10.8 7.4S50 733.1 505.6 68.4 51.6 16.8 5.2.1 Influence of the Boundary Conditions in the Rigid-Wall Permeameter From Table 5.2, a small percentage of fines loss was observed, under permeation, for gradations having Dis-F/dss-s ratios ranging from 3.7 to 4.4. This observation reiterates the finding that the fine fraction loss is independent of the boundary conditions for internally stable gradations. For Di5'F/d85'S = 5.9, it appears that the boundary conditions in the rigid-wall permeameter exert some influence on the magnitude of fine fraction loss. Direct comparisons require careful consideration because the test stage durations and the applied hydraulic gradients vary from test to test (see Table 5.1). To address this problem, the rate of fine fraction loss (% of the mass of total fine particles present at the start of the test) is taken into consideration which is a ratio between fine fractions lost during permeation to the total test duration. For test 5.9S25, the fine fraction loss was at a rate of 8.5 % per hour (1.0 < in <17.5) whereas for tests 5.9T25 and 5.9T25R, it was found out to be 41.5 % (1.0 < i>7 < 9.2) and 51.0 % (1.0 < i n < 5.3) per hour respectively (see Fig. 5.4). This increased rate of fines loss with the textured boundary is attributed to: Abdul Sattar Khan 55 Chapter No. 5 1. A localized change in Dis-F/dgs-s ratio near the boundary walls: relatively coarse beads (D = 2.0 to 2.8 mm) were used to textured the walls of the permeameter, resulting in larger pore size constrictions adjacent to the wall; and, 2. The textured wall has a higher interface friction value, which will mobilise greater friction losses. Hence the applied vertical effective stress of 25 kPa is considerably diminished along the length of the test specimen. Lower stresses yield more fine fraction losses. in o •a o> N 35 30 25 20 15 10 • Smooth wall (25 kPa) A Textured w all (25 kPa) Test Code Duration Fines loss 5.9S25 5.9T25 5.9T25R 7.4S25 7.4T25 3 h, 5 min. 25.1% 42 min. 29.0 % 35.5 min. 30.2 % 30 min. 45 min. 31.3 % 27.7 % 10 )l5'F / das ' s Fig. 5.4 Normalized fines loss under different boundary conditions of the rigid-wall permeameter For test gradations with a Di5-p/d85-sratio of 7.4, higher rates of fine fraction loss were observed, and which appear independent of the boundary wall condition. Again, these observations based on the rate of fines loss further imply the boundary wall condition does not significantly affect the response of gradations that are either clearly internally stable or, in contrast, clearly unstable. It appears only to influence gradations that are in a "intermediate" range of D^-pAWs. 5.2.2 Influence of Vertical Effective Stress Inspection of Table 5.2 reveals the higher vertical effective stress slightly increased the fine fraction loss during specimen reconstitution, more specifically during its application. During Abdul Sattar Khan 56 Chapter No. 5 permeation, however, it provides confinement to the specimen and results in a lesser fine fraction loss. During permeation for tests 5.9S25 and 5.9S50, considerably less loss of fine particles (25.1% vs. 10.8%), under reasonably comparable test duration and applied system hydraulic gradient, was observed. Similar change in the material response was noticed for test 7.4S50.This observation implies the elevated vertical effective stress imparts stability to the particles. 5.3 Comparison with other Studies In the preceding paragraphs, an effort was made to critically analyze the test data obtained during this study and deduce conclusions based on careful observations. Results from current study are compared with similar studies to revalidate the observations made during the experiments and elevate the level of confidence in understanding of internal stability for gap-graded soils 5.3.1 Evaluation of the Critical Hydraulic Gradient During previous studies, little information was reported about the influence of the hydraulic gradients in general and the critical hydraulic gradients in particular. Moffat (2002), Bertram (1940) and Tomlinson et al. (2000) and are amongst those few researchers who had studied the effects caused by the magnitude of the seepage forces. Their work is now compared to the data generated during this study. Moffat (2002) conducted series of tests on a single layer system over broadly and gap-graded soils. The tests specimens were reconstituted using materials of a broadly graded glacial till. He used the same UBC modified Gradient Ratio Test device with smooth boundary condition, specimen reconstitution technique and vertical effective stress (25 kPa) during his study. Comparison of Moffat's results with those of the present study allows estimation of particle shape effects. For test gradation with Di5>F/dg5'S, = 7.4, Moffat (2002) report a critical hydraulic gradient between 1.0 and 9.. During current study a critical hydraulic gradient, on the companion gradation, between 0.5 and 1.0 was observed. This difference is attributed to the use of glass bead specimens during current study. It had been observed that permeability is several orders of magnitude lower for rough particles as compared to the rounded particles of smooth surfaces (after US Corps of Engineers, 1986). For test gradation with Di5-F/d85's, = 5.3, Moffat (2002) reported that the seepage forces did not induce any failure in the range of Abdul Sattar Khan 57 Chapter No. 5 applicable system hydraulic gradients (0.1 < in < 18.5). Whereas during current study, for gradation with Dis-p/dgs-s, = 5.9, seepage induced failure was found out at in - 14.2 and 17.5. For a two layer system, base soil against a filter medium, Bertram (1940) reported "The size ratio, D15F / dg5 S , at the limit of stability decreased to approximately 8 and 6 for the high gradient (18 to 20) tests". It is important to note that Bertram had performed these experiments in a rigid-wall permeameter with smooth boundary conditions. From Table 5.1, it is found out that the test 5.9S25 (Dis-p/dgs-s, = 5.9) did fail at a system hydraulic gradient between 14.2 and 17.5, which appears to reconfirm the conclusion made by Bertram (1940) during his research. Tomlinson et al. (2000) readdressed the filtration characteristics of the uniformly graded materials in a two-layer system. They had reported very high values of the critical hydraulic gradients (see Fig. 5.5). This observation, surprisingly, differs considerably from the results reported by the Bertram (1940) on the same type of materials in a two layer system. Inspection of the Fig. 5.5 reveals that for Di5p/dg5S between 8 and 10, critical hydraulic gradient value varies between 100 and 70. Whereas comparison with current study suggests that the range of reported critical hydraulic gradient is somewhat higher. It is important to note Abdul Sattar Khan 58 Chapter No. 5 TO" Fig. 5.5 Test results reported by Tomlinson et al. (2000) that a direct comparison cannot be made because of many differences, like the size of test specimen, different aspect ratios and the use of uniformly graded materials. This discrepancy deserves further experimental study. 5.3.2 Observed Total Loss of Fines One of the main objectives of the current study was to enhance a data base acquired using the UBC modified gradient ratio device on gap gradations (both soils and glass beads). To compare the test data obtained during this study (on glass beads) with the existing data base (on soils) and to include selected work of other researchers, a plot of DIS-FA^'S versus total loss of fines (% in total mass of the specimen at the start of the test) developed by Moffat (2002) is modified and reproduced in Fig. 5.6 V) o c CO V) o s o 100 90 80 70 60 50 40 30 20 10 0 \ O Kenny etal. (1985) • Honjoetal. (1996) # Kenny-Moffat (2002) Honjo-Moffat (2002) • Khan (2003) n — t j ~ — n •_• L_J—— • • n • u • • • FJ 1 • I 0 u 1 ee- 0 ^ 0 # = T ° 1 1 0 1 2 3 4 5 6 7 8 9 10 D 1 5 T / d 8 5 . s Fig. 5.6 Total loss of fines observed during various studies Inspection of the Fig. 5.6 suggests that the particle loss observed during current study is consistent with the values reported by previous researchers. Following comments are made in recognition of the factors influencing fine fraction loss including magnitude of the applied system hydraulic gradient, test duration, level of vertical effective stress and vibrations. Not withstanding the influence of these factors, there are certain trends in these data. For example data reported at Di5>F/d85's = 7.4 by Honjo et al. (1996), Moffat (2002) and during the current Abdul Sattar Khan 59 Chapter No. 5 study appear reasonably consistent. However, although Honjo et al. (1996) reported a higher fine fraction loss for gradation with same Di5- F/dg5' S values, this difference appears to be associated with the increased fine fraction present in their test gradation. The loss of fine fraction noticed during this work at Di5>F/dg5-s = 3.7 and 4.4 appears in excellent agreement with the broad data base. This confirms that for stable gradations, geometric constraints do not allow the hydrodynamics to influence the material response. On the other hand, triggering of the material instability at Di 5'F/d 85'S = 7.4 under a very mild seepage force is compatible to the range defined by Bertram (1940) for material instability. 5.3.3 Effect of Vertical Effective Stress Fewer studies exists that report on the relationship between the vertical effective stress and the critical hydraulic gradient. Current work by Moffat (2003) using a large size permeameter to test gap graded soils (gravely sand) suggests that the critical hydraulic gradient appears to increase with the increase in vertical effective stress. This emerging finding is consistent with the observation from this study, although on a limited number of tests. Interestingly the observations of Tomlinson et al. (2000) are in contrast. They had used small uniformly graded two layer system as compared to a gap graded single layer system during this study. This deserves further work to address the nature of apparent inconsistency. 5.4 Validation of the Kezdi's (1979) Criterion Kezdi (1979) proposed a technique to split the gradation curve of a broadly or gap-graded soil into two components namely coarser and finer fractions. It was then suggested that, if the particular gradation split satisfied the Terzaghi's Rule of filtration (1939), where Di5>F /dg5-s = 4 and 5 then the composite gradation would be internally stable. Based upon these empirical guidelines, modified gradation G-ld (Di5-F /dg5>s = 3.7) and modified G-3c (Di5-F /dg5-s = 4.4) are deemed to be internally stable. In contrast, gradations No. 4, modified G-4c and No. 2 (Di5'F /dg5-s Of 5.9, 7.4 and 8.7 respectively) are deemed to be internally unstable. Inspection of Fig. 5.6 inspection shows that a Di 5 - F /d 8 5' S between 4 and 5 appears threshold ratio for the onset of a significant fine fraction loss that in turn is evidence of an internal instability. Hence it appears that the Kezdi's (1979) criterion can be used, with reasonable confidence, in predicting the internal stability of the gap-graded soils. Abdul Sattar Khan 60 Chapter No . 5 5.5 Rationale of the Kenny and Lau (1985,1986) criterion Test gradations used in the current study are evaluated using the Kenney and Lau (1985,1986) criterion for internal stability, which is based on the shape of the gradation curve (see Appendix - A). Fig. 5.7 shows that all the gradations used in testing lie on or above the H/F = 1 line for a value F 20 %, the boundary defined for well graded materials, and hence would be deemed internally stable. Test results indicate that the gradations modified G-ld and modified G-3c were stable but gradations No.2, modified G-4c and No. 2 showed an 10 15 20 25 30 35 F, mass fraction smaller than Fig. 5.7 Evaluation of gradations for internal stability using Kenney and Lau (1985, 1986) criterion intermediate to severe extent of internal instability (see Table 5.1). Hence, this empirical rule failed to adequately detect and properly differentiate between the gradations, characterised by a completely horizontal gap in the gradation. This observation from analysis of results raises the question of revaluating or qualifying the use of the criterion for markedly gap-graded soils. Abdul Sattar Khan 61 Chapter No. 5 6.0 CONCLUSIONS AND RECOMMENDATIONS FOR FUTURE WORK A modified Gradient Ratio test device was used to examine the internal stability of five sand gradations characterised by a gap at 40% of mass passing. Glass beads of various sizes were reconstituted to achieve a desired gradation of fine to coarser sands. Test specimens were subjected to a range of hydraulic gradients (in = 0.4 to 18.2) to observe the effect of seepage flow on the triggering instability in the test specimen. The influence of different boundary conditions in the rigid-wall permeameter, and vertical effective stress, were examined. Internal stability is governed by a geometric constraint and hydrodynamic activation. Those specimens with a larger Dis-pA s^'s ratio, allow for the examination of critical hydraulic gradient. The critical hydraulic gradient is reported with consideration of energy losses occurring throughout the test device. From measurement of water head distributions, axial deformation, volumetric flow rate and mass of fine fraction lost during various test stages, the following conclusions are drawn. 1. Internal instability is assured in gradations with D -^pA '^s <3.7 whereas material segregation is expected with gradation having Di5>FAl85's> 8.7. Intermediate gradations yield an increase in susceptibility for material instability with an increase in DIS-FAISS'S ratio. 2. For specimens with same gradation that are subjected to a similar magnitude and duration of hydraulic gradient, an increase in vertical effective stress results in a decrease in loss of fine fraction. 3. For potentially unstable gradations (Dis-F/dgs-s > 3.7), the critical hydraulic gradient appears to vary significantly with Di 5 ' F/dg 5 ' S ratio. At Di5-F/dg5'S = 8.7, no gradient is required as material experienced excessive segregation. For Di5'F/dg5'S = 7.4, the critical hydraulic gradient is very low. At Di5'F/d85'S = 5.9 a relatively high critical hydraulic gradient was observed whereas for Di5>FAlg5'S = 4.4 essentially nothing was found. 4. The influence of the boundary wall condition is very subtle. For specimens that are either prone to segregation or, on the other hand, stable (which are two extreme Abdul Sattar Khan 62 Chapter No. 6 stability conditions), the boundary wall condition does not exert any influence. Whereas for the intermediate ones, it effects the material behaviour. For a comparison of the findings from this study with the work of others, additional conclusions are drown, which are as follows: 5. The critical hydraulic gradient to trigger the onset of material instability appears higher for real soils in comparison to the glass beads of this study. However, following the onset of instability, identical gradation of soils or glass beads will likely yields similar amount of fine fraction loss. 6. Bertram (1940) observed that the material instability encountered at Disp/dgss approximately 6 for a uniform base soil against uniform filter. This finding appears compatible with glass bead specimen having gap gradation 7. The Kezdi's (1979) empirical rule based on the split gradation and a threshold Di5'F/d85'S between 4 and 5 for the onset of material instability can be used with confidence in assessing the internal stability (see Fig. 6.1). 100 T 90 \ 80 \ g 7 0 ; 0) c 60; «*-o 5 0 ; tn lit o 40 : ra o H 30 • 20 10 0 :-O Kenny et al. (1985) • Honjo e ta l . (1996) * Kenny-Moffat (2002) • Honjo-Moffat (2002) • Khan (2003) Transitional zone for the onset of internal instability Kezdi's rule (1979) for internal stability Terzaghi's rule (1939) for soil retentioin CTJ OO 0 B • • Onset of instability, Bertram (1940) 1 • Fig. 0 1 2 3 4 6.1 Design rules and observations reported by 5 Dl5'F/d»5's Segregation Per = 0) 7 8 9 various researchers 10 8. Kenney and Lau (1985, 1986) criterion needs some modifications to capture the material instability for gap-gradations having a completely horizontal gap Abdul Sattar Khan 63 Chapter No. 6 9. In general Dis-p/dgs's is the measure of potential for material instability. Once a threshold DIS-FA^'S is exceeded, the onset of instability is governed by a critical value of a hydraulic gradient. If the critical hydraulic gradient is exceeded, nearly all the available fines are then removed under the influence of seepage flow. Two pattern of fine fraction loss were observed. Either the development of a preferential channel (piping) or universal fine fraction loss (suffosion). With regard to further work, it is suggested that: 1. Any future study should examine thoroughly the influence of vertical effective stress. 2. Detailed investigation is required in the zone of soil ratios 5 < Dis-p/dgs's < 7 to establish better understanding of material response to seepage flow 3. It would be advantageous to ensure any test device is capable of imposing the system hydraulic gradients is excess of 20 to examine the gradations having Dis-p/dgs-s ratio in the range of 4 and 6. 4. It would be helpful to further examine the effect of boundary toughness by changing the size of the attached glass beads. Abdul Sattar Khan 64 Chapter No. 6 LIST OF REFERENCES ASTM. (1996) Standard Test Method for Measuring ht Soil-Geotextile Clogging Potential by the Gradient Ratio (D5101-96), in the Annual Book of ASTM Standards, Vol. 04.09, ASTM Philadelphia, USA. Bertram, G.E. (1940) An Experimental Investigation of Protective Filters, Soil Mechanics Series No. 7, Graduate school of Engineering, Harvard University, Cambridge, MA, USA. Charles, J.A. (2001) Internal Erosion in European Embankments Dams, Proceedings of Hydropower 01, ICOLD European Symposium, June 25-27, Geiranger, Norway. Garner, S.J. and Sobkowicz, J.C. (2002) Internal Stability in Gap-Graded Cores and Filters, Proceeding of Canadian Dam Association Annual Conference, October 6-10, Victoria, BC. Canada. Geotechnical Engineering Office, (1993) Review of Granular and Geotextile Filters, Geo publication No. 1/93, Civil Engineering Department, Hong Kong, pp. 88-92. Hameiri, A. (2000) Soil Geotextile Filtration Behaviour under Dynamic Conditions of Vibrations a d Cyclic Flow, PhD Thesis, University of British Columbia, Vancouver, BC, Canada. Hawley, R. (2001) Filtration Performance of Geotextile in Cyclic Flow Conditions: A Laboratory Study, M.A.Sc Thesis, University of British Columbia, Vancouver, BC, Canada. Honjo, Y., Haque, M.A. and Tsai, K.A. (1996) Self-Filtration Behaviour of Broadly and Gap-Graded Cohesionless Soils, Proceedings of Geofilters' 96 Conference, Montreal, Quebec, Canada, May 29-31, pp. 227-236. Karpoff, K.P. (1955) The use of Laboratory Tests to Develop Design Criteria for Protective Filters, Proceedings of American Society for Testing Materials 5, pp. 1183-1198. • Kenney, T.C. and Lau, D. (1985) Internal Stability of Granular Filters, Canadian Geotechnical Journal, Vol. 22, No. 2, pp. 215-225. Kenney, T.C. and Lau, D. (1986) Reply: Internal Stability of Granular Filters, Canadian Geotechnical Journal, Vol. 23, No. 3, pp. 420-423. Kezdi, A. (1979) Soil Physics, Elsevier Scientific Publishing Company, Amsterdam, Netherlands. Kovacs, G. (1981) Seepage Hydraulics, Elsevier Scientific Publishing Company, Amsterdam, Netherlands. Kuerbis, R.H. Vaid, Y.P (1988) Sand Sample Preparation - The Slurry Deposition Method, Soils and Foundations, Vol. 28, No. 4, pp. 107-118. Abdul Sattar Khan 65 List of References Lafleur, J. (1984) Filter Testing of Broadly Graded Cohesionless Tills, Canadian Geotechnical Journal, Vol. 21, No. 4, pp. 634-643. Lafleur, J., Montes, P., Alicescu. V., and Phuong, N. (2000) Laboratory Simulations of Filtration through Dam Cores made of Broadly Graded Moraines, Proceedings of Geofilters 2000 Conference, Warsaw, Poland, June 5-7, pp. 135-144. Moffat, R. (2002) A Laboratory Study of Particle Migration in Cohesionless Soils, M.A.Sc Thesis, University of British Columbia, Vancouver, BC, Canada. Moffat, R. (2003) Ongoing Research Work on Large Size Permeameter, University of British Columbia, Vancouver, BC, Canada. Milligan, V. (1986) Internal Stability of Granular Filters: Discussion, Canadian Geotechnical Journal, Vol. 23, pp. 414-418. Peck, R.B. (1980) Where has all the Judgment gone?, Fifth Laurits Bjerrum Memorial Lecture, Canadian Geotechnical Journal, Vol. 17, pp. 584-590. Ripley, C.F. (1986) Internal Stability of Granular Filters: Discussion, Canadian Geotechnical Journal, Vol. 23, pp. 255-258. Sherard, J.L., Dunnigan, L.P., Talbot, J.R. (1984) Basic Properties of Sand and Gravel Filters, Journal of Geotechnical Engineering, Vol. 110, No. 6, pp. 684-699. Sherard, J.L. and Dunnigan, L.P. (1986) Internal Stability of Granular Filters: Discussion, Canadian Geotechnical Journal, Vol. 23, pp. 418-420. Terzaghi, K., Peck, R.B. and Mesri, G. (1996) Soil Mechanics in Engineering Practice, Third Edition, John Wiley and Sons, New York, NY, USA. Tomlinson, S.S. and Vaid, Y.P. (2000) Seepage Forces and Confining Pressure Effects on Piping Erosion, Canadian Geotechnical Journal, Vol. Vol. 37, pp. 1-13. U.S. Department of the Interior, Bureau of Reclamation (1987) Design Standards No. 13-Embankment Dams, Chapter No. 5, Protective Filters. U.S. Corps of Engineer, (1986) Engineering Manual EM 1110-2-1901: Seepage Analysis and Control for Dams, Chapter No. 2, Determination of Permeability of Soil and Chemical Composition of Water. Vaughan, P.R. (2000) Filter Design for Dam Cores of Clay, A Retrospect, Proceedings of Geofilters 2000 Conference, Warsaw, Poland, June 5-7, pp. 189-196. Abdul Sattar Khan 66 List of References APPENDIX - A Assessment of Internal Stability by Kenney and Lau Criterion (1985 and 1986) Abdul Sattar Khan 67 APPENDIX - A GRADATION: MODIFIED G-1d 0.01 0.10 Opening size (mm) 100 10.00 GRADATION: MODIFIED G-3C 0.10 Opening size (mm) 100 10.00 Abdul Sattar Khan 68 APPENDIX - A GRADATION: No. 4 GRADATION: MODIFIED G - 4 C 0.10 Opening size (mm) 100 10.00 Abdul Sattar Khan 69 APPENDIX - A GRADATION: No. 2 0.10 Opening size (mm) 100 10.00 Abdul Sattar Khan 70 APPENDIX - A A P P E N D I X - B Pre and Post-Test Gradations Abdul Sattar Khan 71 APPENDIX - B TEST 3.7S25 1 0 0 90 80 O) s >. .Q k. 0) c <1> 0. 70 j— Gap Ratio = 2.9 60 50 <g 40-1— 8 30 20 10 D „ , / d M . . = 3 . 7 Set of Sieves: - • — O r i g i n a l gradation • • Upper third part of the spec imen • Middle third part of the spec imen - * - Bottom third part of the spec imen #16 (1.18 mm) #25 (0.71 rrm) #60 (0.246 mm) #140 (0.106 mm) 0.01 Wire Mesh @ 0.31 rrm TO r 0.10 Opening size (mm) 1.00 10.00 TEST 4.4S25 - , , — i — i — i — i — i — i 0.10 Opening size (mm) 1.00 10.00 Abdul Sattar Khan 72 APPENDIX - B TEST 7.4S25 Abdul Sattar Khan 73 APPENDIX - B TEST 8.7S25 Abdu l Sattar Khan 74 A P P E N D I X - B T E S T 5 . 9 T 2 5 0.01 0.10 O p e n i n g s i z e ( m m ) 1.00 10.00 T E S T 5 . 9 T 2 5 R 0.01 0.10 O p e n i n g s i z e ( m m ) 1.00 10.00 Abdul Sattar Khan 7 5 APPENDIX - B T E S T 7.4T25 T E S T 8.7T25 0.01 0.10 Opening size (mm) 1 0 0 10.00 Abdul Sattar Khan 76 APPENDIX - B TEST 5.9S50 0.10 Opening size (mm) 1.00 10.00 TEST 7.4S50 Abdul Sattar Khan 77 APPENDIX - B A P P E N D I X - C Water Head Distributions Abdul Sattar Khan 78 APPENDIX - C TEST 3.7S25 TEST 4.4S25 40 60 80 100 120 140 160 Water head (cm) Abdul Sattar Khan 79 APPENDIX - C TEST 5.9S25 Abdul Sattar Khan 80 APPENDIX - C TEST 4.4T25 Water head (cm) Abdul Sattar Khan 81 APPENDIX - C T E S T 5 .9T25R 4) 5 0 » A. / y A / m / < • A / * / * / 1 m /.* — * — S t a g e No. 1, i17 = 1.00 Stage No. 2, i17 = 9.2 at 1.0 minute — — S t a g e No. 2, i17 = 9.2 at 2.0 minutes - Stage No.2, i17 = 9.2 at 3.0 minutes —U - S t a g e No. 2, i17 = 9.2 at 3.5 minutes L- 1 1 ifiii. ' I I I 1 1 ] 1 1 0 10 20 30 40 50 60 70 80 90 Water head (cm) T E S T 7.4T25 0-0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 Water head (cm) Abdul Sattar Khan 82 A P P E N D I X - C Abdu l Sattar Khan 83 A P P E N D I X - C A P P E N D I X - D Photographs Abdul Sattar Khan 84 A P P E N D I X - D TEST 3.7S25 Abdul Sattar Khan 85 APPENDIX - D TEST 4.4S25 Test set-up at the start o f test stage no. 1 Fine fractions in the lower collection trough Sample at 2.00 cm above the wire mesh Abdul Sattar Khan 8 6 APPENDIX-D TEST 5.9S25 Top view of the sample after test Sample at 8.0 cm above the wire mesh Abdul Sattar Khan 87 A P P E N D I X - D Sample at 6.0 cm above the wire mesh Sample at 4.0 cm above the wire mesh Sample after test stage no. 5 Abdul Sattar Khan 88 APPENDIX - D TEST 7.4S25 Abdul Sattar Khan 89 APPENDIX - D Top view of the sample after test Sample at 8.0 cm above the wire mesh Sample at 2.0 cm above the wire mesh Abdul Sattar Khan 90 APPENDIX TEST 8.7S25 Top view of the specimen after reconstitution Specimen after reconstitution Sample at 8.0 cm above the wire mesh Universal fine fraction loss resulted in the filling of collection trough Abdul Sattar Khan 9 1 APPENDIX - D Sample at 6.0 cm above the wire mesh Sample at 4.0 cm above the wire mesh Sample at 2.0 cm above the wire mesh Abdul Sattar Khan 92 APPENDIX - D T E S T 4 . 4 T 2 5 Sample after test stage no. 2 Sample after test stage no. 3 Abdu l Sattar Khan 93 A P P E N D I X - D Sample at 8.0 cm above the wire mesh Sample at 6.0 cm above the wire mesh Sample at 4.0 cm above the wire mesh Sample at 2.0 cm above the wire mesh Abdul Sattar Khan 94 A P P E N D I X - D TEST 5.9T25 Abdu l Sattar Khan 95 A P P E N D I X - D Top view of the sample after test Sample at 8.0 cm above the wire mesh Abdul Sattar Khan 96 A P P E N D I X - D TEST 5.9T25R Specimen after reconstitution Top view of the specimen after reconstitution Localized fines deficient region during test Development of the fines deficient channel stage no. 1 along the length of specimen Abdul Sattar Khan 97 APPENDIX - D TEST 7.4T25 Sample during test stage no. 1 Loss o f fine fraction is in progress Sample at 8.0 cm above the wire mesh Sample at 6.0 cm above the wire mesh Abdul Sattar Khan 99 APPENDIX - D TEST 8.7T25 Abdu l Sattar Khan 100 A P P E N D I X - D Abdul Sattar Khan 101 APPENDIX - D TEST 5.9S50 Specimen after reconstitution Sample after test stage no. 1 Sample after test stage no. 2 Sample after test stage no. 3 Abdul Sattar Khan 102 APPENDIX - D Abdu l Sattar Khan 103 A P P E N D I X - D Sample at 6.0 cm above the wire mesh Sample at 4.0 cm above the wire mesh Sample at 2.0 cm above the wire mesh Localized fine fraction loss during test stage no. 5 Abdu l Sattar Khan 104 A P P E N D I X - D TEST 7.4S50 Pockets of the fine fraction depleted zones Sample after test stage no. 1 Abdul Sattar Khan 105 APPENDIX - D Sample after test stage no. 2 Sample after test stage no. 3 Abdu l Sattar Khan 106 A P P E N D I X - D Abdul Sattar Khan 107 APPENDIX - D 

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:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0063846/manifest

Comment

Related Items