ABSTRACT. Work in this dissertation consists of three numerical studies undertaken to investigate the

Transcription

1 ABSTRACT BAHADOR, MAHDI. Thermal, Hydraulic, and Mechanical Response of Unbound Pavement Layers with Geosynthetics at the Subgrade-Base Course Interface. (Under the direction of Dr. Mohammed A. Gabr and T. Matthew Evans.) Work in this dissertation consists of three numerical studies undertaken to investigate the thermal, hydraulic, and mechanical behavior of unbound pavement layers with embedded geotextile used as drainage/moisture barrier layer under unsaturated conditions. Thermal gradient can change the capillary barrier and drainage properties of the geotextile and to fully understand the hydraulic behavior of embedded geotextile and soil layers, thermo-hydraulic response of the system should be considered. Geotextiles emplaced in a pavement section can change the moisture distribution and mechanical response of the pavement layers. The mechanical response of the pavement layers is affected by the change in moisture distribution of the section and reinforcement effect of geotextile. The effect of thermal gradient on the capillary barrier and drainage properties of the geotextile is investigated using 1-D thermo-hydraulic numerical simulations. A series of coupled thermo-hydraulic simulations were performed on a soil-geotextile column using the computer program UNSAT-H to understand the effect of temperature gradient on suction distribution throughout the soil column and on the hydraulic performance of the geotextile as a drainage/capillary barrier layer. Unsaturated seepage analyses are performed using the computer program SIGMA/W to study the effect of the hydraulic properties of geotextiles used as drainage/moisture barrier layers in a pavement section on the moisture distribution within the section during infiltration and drainage. A sensitivity analysis was performed to study the effect of the air entry value and saturated hydraulic conductivity of the drainage/moisture barrier layer on drainage system performance. The Computer program

2 FLAC is used to perform stress-deformation analyses on the paved and unpaved road sections and study the effect of geotextile layers on mechanical response and plastic deformation of the pavement sections. Two types of geotextile were modeled as the transport layer: woven fiberglass and nonwoven polypropylene. The coupled thermo-hydraulic simulation results showed that during drainage, the temperature gradient and lower temperature at the top of the column increased suction in the geotextile and its ability to function as a capillary barrier. During capillary rise, the temperature gradient and lower temperature at the top of the column decreased the suction in the geotextile and its ability to function as a capillary barrier. Changing the direction of the thermal gradient reversed the water vapor flow direction and its effect on the suction in the geotextile. The unsaturated seepage analyses indicated that a geocomposite consisting of a geotextile with a high air entry value overlying a geonet decreased the degree of saturation in the underlying subgrade soil while maintaining suction in the overlying aggregate base course during infiltration. A higher air entry value for the geotextile resulted in higher suction in the underlying layer so long as this air entry value did not exceed the air entry value of the overlying soil layer. Lastly, stress-deformation analyses showed that the geosynthetic layers decreased the plastic deformation in both paved and unpaved road sections through reinforcement and hydraulic effects. Increasing the thickness of the asphalt and aggregate base course decreased the reinforcement effect of the geotextile while increasing its hydraulic effect. In low volume road sections with thinner asphalt layer, using the woven fiberglass as the transport layer decreased the plastic deformation of the profile by up to 20% more than the profile with the nonwoven polypropylene geotextile and increasing the thickness of the asphalt layer reduced this difference to approximately 4%. In unpaved

3 road sections, the woven fiberglass decreased the plastic deformation of the profile by approximately 24% more than the profile with nonwoven polypropylene geotextile regardless of the aggregate base course thickness.

4 Copyright 2012 by Mahdi Bahador All Rights Reserved

5 Thermal, hydraulic, and mechanical Response of Unbound Pavement Layers with Geosynthetics at the Subgrade-Base Course Interface by Mahdi Bahador A dissertation submitted to the Graduate Faculty of North Carolina State University in partial fulfillment of the requirements for the degree of Doctor of Philosophy Civil Engineering Raleigh, North Carolina 2012 APPROVED BY: Dr. Mohammed A. Gabr Committee Chair Dr. T. Matthew Evans Co-Chair Dr. Shamim Rahman Dr. Paul Khosla

6 DEDICATION This dissertation is dedicated to all those who have sacrificed their lives for human rights and justice. ii

7 BIOGRAPHY Mahdi Bahador was born in Tehran, Iran. He graduated in 2001 from high school and entered college that autumn as a Civil Engineering major. He graduated from college in 2005 and entered graduate school as a Civil Engineer, with an emphasis in Geotechnical Engineering. It took him two years to finish his master s degree. After obtaining his master s degree, he worked as a researcher in International Institute of Earthquake Engineering and seismology, before moving to North Carolina to pursue a PhD in Civil Engineering. After he graduates in May 2012, Mahdi plans to move to Charlotte, NC, where he will work as a senior staff engineer for GeoSyntec Consultants. iii

8 ACKNOWLEDGMENTS I would like to first acknowledge my advisors Dr. Matthew Evans and Dr. Mohammed Gabr, for their precious guidance and support throughout my graduate life at North Carolina State University. Their research style and enthusiasm was a model for me to improve my research skills. Their experience in unsaturated soil mechanics and geosynthetic design helped me a lot to complete this research successfully. I would also like to extend my sincere thanks to my research committee members Dr. Paul Khosla and Dr. Shamim Rahman for their valuable advice and support. I am also grateful to Dr. Joshua Heitman from Soil Science department for sharing his knowledge to perform themo-hydraulic simulations successfully. I would also like to give my recognition to all civil engineering department staff at North Carolina State University, especially to Ms. Renee Howard for her valuable advice and help. I would also like to give heartfelt thanks to my family and my beloved girl friend for their love and moral support. I am also indebted to all my friends, especially to Mahdi Khalilzad who was part of my graduate life at North Carolina State University. iv

9 TABLE OF CONTENTS LIST OF TABLES... vii LIST OF FIGURES... viii CHAPTER 1- Overview of Dissertation Background Scope of Research Dissertation Organization... 5 CHAPTER 2- Numerical studies of the effect of temperature on the unsaturated hydraulic response of geotextiles Introduction Overview of the Numerical Model Material Properties Simulation Results Geotextile at the bottom of the column drainage condition Geotextile at the top of the column capillary rise condition Geotextile in the middle of the column drainage condition Geotextile in the middle of the column capillary rise Discussion of Results Conclusion Acknowledgements References CHAPTER 3- Moisture distribution in a partially saturated pavement section with drainage/moisture barrier layer Introduction Material Properties Model Verification Numerical Simulations: Single-layer Geotextile Geotextile Air Entry Value Numerical Simulations: Single-layer Geotextile, Varying AEVs Numerical Simulations: Geotextile Over Geomembrane Numerical Simulations: Geotextile Over Geonet v

10 3.8.1 Air Entry Value Saturated Hydraulic Conductivity Summary and Conclusions Acknowledgement References CHAPTER 4- Effect of Geocomposite Drainage Layers on Moisture Distribution and Plastic Deformation of Paved and Unpaved Road Sections Introduction Effect of Suction on Soil Strength Modeling Plastic Deformations Material Properties Hydraulic properties Mechanical properties Modeling Approach Model Verification Seepage Analysis Paved roads Unpaved roads Stress-deformation Analysis Paved roads Unpaved roads Conclusion References CHAPTER 5- Conclusions and Future Work Summary and Conclusions Thermo-Hydraulic Response of Geotextile Unsaturated Seepage Analysis Hydro-Mechanical Response of Geotextile Suggestions for Future Work vi

11 LIST OF TABLES Table 2-1. Hydraulic properties of the materials used in the simulations Table 3-1. van Genuchten parameters and saturated hydraulic conductivity for materials used in the modeling. (SEEP/W, 2002; EICM, 2006; Ramos, 2001) Table 3-2. Hydraulic properties of fine sand and geotextile used in verification simulations (after Krisdani et al. 2008) Table 4-1. van Genuchten parameters and saturated hydraulic conductivity for materials used in the modeling Table 4-2. Elastic strength properties of materials used in the model for various layers (Brunton et al. 1992; Ramos, 2001; Bergado et al. 2001; EICM, 2006; Pease, 2010) Table 4-3. Mohr-Coulomb and cam-clay material properties for ABC and subgrade (After Desai and Siriwardane, 1984) Table 4-4. Hydraulic properties of fine sand and geotextile used in verification simulations (after Krisdani et al. 2008) Table 4-5. Plastic deformations calculated in MEPDG and FLAC Table 4-6. ABC and asphalt thickness used in paved road profile simulations vii

12 LIST OF FIGURES Figure 2-1. Schematic diagram of the discritized profile Figure 2-2. a) Moisture characteristic curves b) hydraulic conductivity curves Figure 2-3. Suction head distribution along the silty clay-geotextile culomn a) constant temperature of 0 C b) constant temperature of 38 C Figure 2-4. Suction head distribution along the silty clay-geotextile column a) constant temperature of 0 C at the top and constant temperature of 4 C at the bottom b) constant temperature of 4 C at the top and constant temperature of 0 C at the bottom Figure 2-5. Difference in suction head between the silty clay-geotextile column with a constant temperature of 0 C and the silty clay-geotextile column with the temperature gradient of a) 0 C at the top and 4 C at the bottom b) 4 C at the top and 0 C at the bottom ( the bottom figures are the magnified view of suction head profiles in the geotextile layer) Figure 2-6. Suction head distribution along the sand-geotextile culomn a) constant temperature of 0 C b) constant temperature of 38 C Figure 2-7. Suction head distribution along the silty clay-geotextile culomn a) constant temperature of 0 C b) constant temperature of 38 C (the top figures are the magnified view of suction head profiles in the geotextile layer) Figure 2-8. Difference in suction head between the column with a constant temperature of 0 C and the silty clay-geotextile column with the temperature gradient of a) 0 C at the top and 4 C at the bottom b) 4 C at the top and 0 C at the bottom (the top figures are the magnified view of suction head profiles in the geotextile layer) Figure 2-9. Suction head distribution under constant temperature of 0 C in a) silty claygeotextile column b) sand-geotextile column Figure Difference in suction head between the column with a constant temperature of 0 C and the silty clay-geotextile column with the temperature gradient of 0 C at the top and 4 C at the bottom a) in soil b) in geotextile Figure Difference in suction head between the column with a constant temperature of 0 C and the silty clay-geotextile column with the temperature gradient of 4 C at the top and 0 C at the bottom a) in soil b) in geotextile Figure Difference in suction head between the column with a constant temperature of 0 C and the sand-geotextile column with the temperature gradient of a) 0 C at the top and 4 C at the bottom b) 4 C at the top and 0 C at the bottom viii

13 Figure Suction head under constant temperature of 0 C in a) silty clay-geotextile column b) sand-geotextile column Figure Suction head difference between the silty clay-geotextile column with constant temperature of 0 C and the silty clay-geotextile column with temperature gradient of a) 0 C at the top and 4 C at the bottom b) 4 C at the top and 0 C at the bottom (the top figures are the magnified view of suction head profiles in the geotextile layer) Figure Suction head difference between the sand-geotextile column with constant temperature of 0 C and the sand-geotextile column with temperature gradient of a) 0 C at the top and 4 C at the bottom b) 4 C at the top and 0 C at the bottom (the top figures are the magnified view of suction head profiles in the geotextile layer) Figure 3-1. (a) Moisture characteristic curves (b) unsaturated hydraulic conductivity curves Figure 3-2. Comparison of the measured experimental pore pressures by Krisdani et al. (2008) and the computed pore pressures (numerical) in this paper along the column during drainage Figure 3-3. Schematic diagram of the profile used for model verification (after Krisdani et al., 2008) Figure 3-4. Schematic diagram of the profile used in one layer geotextile simulations Figure 3-5. Pore pressure contours after 10 hrs infiltration (a) profile with geotextile (b)profile without geotextile Figure 3-6. a) Pore pressure distribution along the centerline of the profile: stage 1 b) Magnified view of pore water pressure profiles in crushed stone layer Figure 3-7. a) Pore pressure distribution along the centerline of the profile: stage 2 b) Magnified view of pore water pressure profiles in crushed stone layer Figure 3-8. MCC and hydraulic conductivity curves for DMBLs with different AEVs Figure 3-9. a) Pore pressure distributions for DMBLs with different AEVs b) Magnified view of pore pressure profiles in crushed stone layer Figure Pore pressure contours (kpa) after (a) 3 hr; and (b) 6 hr rainfall Figure Pore pressure contours (kpa) after 6 hr rainfall, geotextile with saturated hydraulic conductivity increased by a factor of two Figure Schematic diagram of narrow profile ix

14 Figure Pore pressure distribution after 6, 12, and 24 hrs of rainfall Figure Pore pressure distribution for different AEVs in the two-layer DMBL system 81 Figure (a) MCCs and (b) hydraulic conductivity curves of geotextiles with different AEVs Figure Pore pressure distributions for different saturated hydraulic conductivity of the upper DMBL Figure Hydraulic conductivity curves for (a) different saturated hydraulic conductivity values; and (b) different AEVs Figure 4-1. Unloading reloading stress-deformation curves of the simulated displacementcontrolled triaxial test Figure 4-2. Properties of study materials: (a) Moisture characteristic curves (b) unsaturated hydraulic conductivity curves Figure 4-3. (a) Woven fiberglass geotextile specimens (b) extensometer (c) Woven fiberglass geotextile at failure Figure 4-4. Stress strain curves of woven fiberglass geotextile Figure 4-5. Comparison of the measured experimental pore pressures by Krisdani et al. (2008) and the computed pore pressures (numerical) in this paper along the column during drainage Figure 4-6. Schematic diagram of the profile used for model verification (after Krisdani et al., 2008) Figure 4-7. Schematic diagram of the profile used for seepage analyses Figure 4-8. Pore pressure contours in the profile : (a) without geocomposite (b) with WF geotextile transport layer (c) with NWP transport layer Figure 4-9. Pore pressure distribution in ABC and subgrade of the paved road section with ABC thickness of (a) 15.2 cm (b) 25.4 cm (c) 45.7 cm Figure Pore pressure distribution in ABC and subgrade of the unpaved road section with ABC thickness of (a) 50.8 cm (b) 63.5 cm and (c) 68.6 cm Figure Total plastic deformations under 1 million cycle x

15 Figure Percent decrease in total plastic deformation of profile with geocomposite compared to the profile without geocomposite for ABC thicknesses of a) 15.2 cm b) 25.4 cm c) 45.7 cm Figure Percent decrease in total plastic deformation due to the total effect of the geocomposite for ABC thicknesses of (a) 15.2 cm (b) 25.4 cm (c) 45.7 cm Figure The ratio of the plastic deformation of the profile with geocomposite (d pg ) to the one for the profile without geocomposite and with 45.7 cm ABC (d p45.7 ) Figure Percent decrease in total plastic deformation of profile with geocomposite compared to the profile without geocomposite in unpaved road section Figure Percent decrease in total plastic deformation due to the total effect of the geocomposite in unpaved road section Figure Ratio of the plastic deformation of the profile with geocomposite (d pg ) to the one for the profile without geocomposite and with 68.6 cm ABC (d p68.6 ) xi

16 CHAPTER 1 Overview of Dissertation 1.1 Background Geotextiles are used in many aspects of earth construction, including as cushion, separation, or reinforcement layers. However, recent studies have shown that geotextiles (alone or in combination with other synthetics) may also act as effective drainage and/or moisture barrier layers when in contact with partially saturated soils (e.g. Henry, 1988; Stormont et al., 1998; Henry and Holtz, 2001; Stockton, 2001; Krisdani et al., 2006). One application of geotextiles as a drainage/moisture barrier layer is in pavement sections. Conventional drainage systems in pavement sections consist of a permeable aggregate base course under the asphalt layer and an edge drainage to divert the water toward drainage pipes. This drainage system, however, assumes saturated flow and is not always effective in maintaining the water content of the unbound pavement layers (Henkel, 1997). In order to design an appropriate drainage system, water flow under unsaturated conditions and the unsaturated properties of the geomaterials should be considered. In addition to varying degrees of saturation, geotextiles will typically be subjected to thermal gradients, such as those caused by solar heating and to fully understand the behavior of geotextiles emplaced in unsaturated soils, the effect of temperature and thermal gradients must also be considered. Drainage/moisture barrier layers that utilize geosynthetics have been reported in the literature, and among such systems is a patent by Henry and Stormont (2000) for a 1

17 configuration termed the geocomposite capillary barrier drain (GCBD). GCBD consists of a geonet sandwiched between a transport geotextile at the top and a separation geotextile at the bottom. In general, capillary barriers are composed of a layer of a relatively low permeability material over a more highly permeable material (e.g., Stormont and Anderson, 1999). In a pavement system, a capillary barrier can be formed by placing a high conductivity material at the interface of the aggregate base course (ABC) and subgrade to divert infiltrated water laterally toward edge drains and minimize moisture variations and shear strength reduction in the profile (Christopher et al., 2000; Henry and Stormont, 2000; Henry et al., 2002; Stormont et al., 2009). The laboratory proof-of-concept test for GCBDs was undertaken by Henry and Stormont (2002). The system used in their tests simulated a base course material over a GCBD and separate outflow volumes were recorded for the two layers. They showed that a significant amount of water was drained by the transport layer in the GCBD demonstrating the ability of the GCBD to drain water from unsaturated soils subject to infiltration. Henry et al. (2002) demonstrated the use of GCBDs at laboratory scale. Nine tests were performed using control sections and GCBD sections and assuming infiltration from ten-year and onehour design storms. The results indicate that the GCBD drains when a suction of 100 mm or higher is applied. During testing a breakthrough to the subgrade material was observed. However, in subsequent tests the GCBD was observed to recover from the breakthrough and protect the subgrade. A few field tests were also performed to investigate potential road improvement using GCBDs (Christopher et al. 2000; Stormont et al. 2009). 2

18 Previous experimental studies on the effect of moisture content on mechanical behavior of aggregate base course and subgrade materials have been reported in the literature. In general, increasing water content decreases the shear strength of geomaterials (e.g., Fredlund and Rahardjo 1993; Lu and Likos 2004) and such a reduction in strength can be considered in pavement design. Rada and Witczal (1981) concluded that the resilient modulus of granular materials under saturated conditions can be decreased to one third one of the value measured at lower water contents. Tian et al. (1998) measured the resilient modulus of a coarse granular material under dry of optimum and wet of optimum water contents and showed that the specimen compacted wet of optimum had a 20% lower resilient modulus compared to that at optimum water content. Thus, in order to maintain the strength of geomaterials in a pavement section, the drainage system should function to not only prevent the development of positive pore pressure, but also to mitigate moisture increases in unsaturated aggregate base course and subgrade layers. In this dissertation, thermal, hydraulic, and mechanical response of geotextile emplaced within unsaturated soil as a drainage/moisture barrier layer is numerically studied. The effect of hydraulic properties of geotextile and thermal gradients on suction head distribution in the soil above and below the geotextile and the efficacy of the geotextile as a drainage/moisture barrier layer are investigated. The mechanical response of the geotextile embedded in elastoplastic geomaterials is then studied considering the moisture distribution in the profile. 3

19 1.2 Scope of Research Geotextile materials have been widely used as drainage layers in many geotechnical engineering projects. One of the recent applications of geotextiles as drainage/moisture barrier layers is in pavement sections to minimize significant moisture accumulation in the aggregate base course while minimizing the moisture change in subgrade. Since geomaterials are rarely saturated in pavement sections, the unsaturated properties of both geomaterials and geosynthetic layers should be considered in the design. Temperature and thermal gradients, also, affect the hydraulic behavior of geotextiles and should be considered in the design. The research presented in this dissertation has been conducted to understand thermal, hydraulic, and mechanical response of geotextiles embedded in a partially saturated geomaterial and to study its efficacy in reducing moisture content and plastic deformation of the section. The objective of this research is to assess the impact of hydraulic properties of the geotextile and the thermal condition of soil-geotextile system on the functionality of geotextile as a drainage/moisture barrier in reducing the moisture content and plastic deformation of the system. Thermo-hydraulic analysis was performed to investigate the effects of matric suction and thermal gradients on the infiltration and drainage properties of a vertical soil-geotextile column through numerical modeling. An unsaturated seepage analysis was performed on a three layer pavement section consisting of aggregate base course, geotextile, and subgrade layers to assess the effect of the hydraulic properties of geotextile used as a drainage/moisture barrier layer on the moisture distribution within the section during infiltration and drainage. Hydro-mechanical analyses were performed on numerically 4

20 simulated paved and unpaved road sections with different asphalt and aggregate base course thicknesses to study the effect of a three-layer drainage system consisting of a geonet sandwiched between two geotextiles on plastic deformation of the section and to study the effect of individual pavement layer thicknesses on the hydraulic and mechanical behavior of the drainage system. 1.3 Dissertation Organization There are 4 additional chapters in this dissertation. Chapters 2 through 4 are papers that will be submitted for publication. The author was the lead author in all three papers and responsible for organizing and writing the papers. References are presented at the end of each chapter. Chapter 5 is the summary of the main findings of the research and suggestions for future work. Chapter 2 presents thermo-hydraulic numerical analysis on a vertical soil-geotextile column. Two soil types and four temperature conditions were considered, and the behavior of the geotextile under drainage and capillary rise were evaluated. Matric suction profiles along the column under different temperature conditions are presented and the performance of the geotextile as a drainage/capillary barrier is evaluated in terms of variations in suction head and associated change in the hydraulic conductivity. This paper will be submitted to Geosynthetics International. Chapter 3 presents the unsaturated seepage analyses on a three layer pavement section consisting of aggregate base course, geosynthetics, and subgrade layers. Parametric analyses are performed on the effects of air entry value and saturated hydraulic transmissivity of the 5

21 geotextile on moisture variation in the pavement section and corresponding accumulation in the aggregate base course, and the efficacy of the geotextile as drainage/moisture barrier layer is assessed. The drainage/moisture barrier system was modeled as one geotextile layer and a geotextile layer in combination with other geosynthetic materials to assess the most efficient configuration for the drainage system to minimize moisture accumulation in the aggregate base course and moisture change in subgrade. This paper has been submitted to Geotextiles and Geomembranes and is currently under review. Chapter 4 presents the hydro-mechanical analysis on paved and unpaved road sections including geosynthetic layers at the interface of aggregate base course and subgrade as drainage/moisture barrier layers. The drainage/moisture barrier system consists of a geonet sandwiched between a nonwoven geotextile at the bottom and a transport geotextile at the top. Two geotextile types are considered for the transport layer in the drainage system: nonwoven polypropylene geotextile and woven fiberglass. The moisture distribution in the pavement section with different aggregate base course thicknesses are studied during rainfall using SIGMA/W. The moisture distributions are then used in FLAC to investigate the effect of asphalt and aggregate base course thicknesses on the functionality of the drainage system in reducing the total plastic deformation of the pavement section. The reinforcement and hydraulic effect of the nonwoven polypropylene and woven fiberglass geotextiles on section deformation response is separated and the effects of asphalt and aggregate base course thicknesses on each of these two components are studied. This paper will be submitted to International Journal of Geomechanics. 6

22 CHAPTER 2 Numerical studies on the effect of temperature on the unsaturated hydraulic response of geotextiles This chapter will be submitted for publication in the Geosynthetics International. The authors are: Mahdi Bahador, T. Matthew Evans, and Mohammed A. Gabr ABSTRACT: A series of coupled thermo-hydraulic simulations were performed on a soilgeotextile column to understand the effect of temperature on suction distribution throughout the soil column and on the hydraulic performance of the geotextile as a drainage/capillary barrier layer. Two different constant temperatures of 0 C and 38 C and a temperature gradient of 4 C along the column were modeled. Changing the temperature from 0 C to 38 C did not have a significant effect on the suction head distribution in the soil-geotextile column. The temperature gradient resulted in appreciable thermal vapor flow and changes in suction head and hydraulic conductivity of the geotextile. During drainage, the temperature gradient and lower temperature at the top of the column increased suction in the geotextile and its ability to function as a capillary barrier. During capillary rise, the temperature gradient and lower temperature at the top of the column decreased the suction in the geotextile and its ability to function as a capillary barrier. Changing the direction of the thermal gradient reversed the water vapor flow direction and its effect on the suction in the geotextile. A 7

23 temperature gradient did not have a noticeable effect on the suction head of the geotextile when positive pore pressure was developed in the geotextile and adjacent soil during drainage. KEYWORDS: Geosynthetics, unsaturated flow, thermohydraulic behavior, capillary barrier 2.1 Introduction Nonwoven geotextiles are often used for drainage, filtration, separation, and/or reinforcement in the design of geotechnical systems. During construction, geotextiles are emplaced in soils that are not fully saturated. The soils with which a given geotextile is in contact are expected remain unsaturated for much of the design life of the system and periods of saturation will likely be brief and intermittent. In addition to varying degrees of saturation, geotextiles will typically be subjected to thermal gradients, such as those caused by solar heating or exothermic waste decomposition. In this work, we will focus on the hydraulic behavior of geotextiles in contact with unsaturated soils and how that behavior varies with applied thermal gradients. Due to their specific hydraulic properties (Stormont et al. 1997), nonwoven geotextiles can function both as drainage layers and capillary barriers (Stormont et al. 2001). When saturated (or nearly saturated) geotextiles have a high hydraulic transmissivity and will act as a drainage layer. However, hydraulic transmissivity decreases rapidly with increasing suction and geotextiles will begin to function as a capillary barrier due to the discontinuity in conductivity across the soil-geotextile interface. Consequently, geotextiles have been proposed for use both as drainage and capillary barrier layers in landfills and pavement 8

24 sections (Dwyer 1998; Dwyer 2001; Henry and Holtz 2001; Henry et al. 2002; Stormont et al. 2009). Clough and French (1982) performed an early study on geotextiles as capillary barriers in pavement sections. They showed that geotextiles can reduce capillary rise in soil. Henry (1996) proposed the use of geotextiles as capillary barriers in pavement sections to reduce frost heave. McCartney et al. (2005) compared the performance of the geotextile capillary barriers with that of soil-only capillary barriers in landfill covers. They showed that a geotextile capillary barrier provides higher water storage in the overlying fine soil compared to a soil-only capillary barrier. In addition to causing a capillary break, geotextiles have been used to drain water from unsaturated soil in pavement sections (Christopher et al. 2000; Stormont et al. 2001). The hydraulic properties of geotextiles constitute a key design parameter. The specific properties of interest are the moisture characteristic curve (MCC) and the saturated transmissivity. These properties have been the subject of several studies found in the literature (Stormont et al. 1997; Stormont and Morris 2000; Ramos 2001; Bouazza et al. 2006). Soil column (McCartney et al. 2005; Stormont et al. 2008; McCartney and Zornberg 2010a) and capillary rise (Henry and Holtz, 2001; Henry, et al. 2002) studies have been conducted to evaluate the behavior of geotextiles as a capillary barrier. These studies have shown that the geotextile increased water storage in the overlying soil during infiltration (see Stormont and Anderson (1999) for a thorough discussion of water storage in capillary barriers). 9

25 To fully understand the behavior of geotextiles emplaced in unsaturated soils, the effect of temperature and thermal gradients must also be considered. In unsaturated porous materials (particularly those that are quite dry), water vapor flow may be the predominant mechanism for moisture movement. Philip and De Vries (1957) studied the thermal effects in unsaturated soils. They developed an approximate analytical analysis to predict the general behavior and to describe moisture and heat transfer in porous media under suction and temperature gradients. Later, Milly (1982) extended that work to a heterogeneous and hysteretic medium. These approaches assumed that air flow can be neglected, which is adequate in many situations. Faust and Mercer (1979) developed a basic formulation for modeling geothermal reservoirs and handled the problem of phase change induced by heating. However, in this approach, capillary effects and the possible presence of air are neglected. Philip and De Vries s (1957) approach is based on suction head and temperature as state variables. Temperature gradients result in thermal water vapor flow and consequently, changing suction in the medium (Yong and Mohamed 1992; Yong et al. 1997; Shooshpasha et al. 2000). Mohamed and Shooshpasha (2004) performed a series of one dimensional coupled heat and moisture tests on a capillary barrier consisting of three soil layers. The capillary barrier was subjected to thermal and suction gradients in opposite directions to simulate arid lands with sub-irrigation systems. They showed that both thermal and suction gradients resulted in moisture movement. Suction gradient caused upward moisture flow from the water source to the mid-part of the barrier and temperature gradient resulted in moisture flow from higher temperature at the top toward the lower temperature at mid-part of the barrier. 10

26 The effect of temperature and thermal gradients on matric suction distributions in soilgeotextile systems has not previously been considered. In the current work, we study numerically the effects of matric suction and thermal gradients on the infiltration and drainage properties of a vertical soil-geotextile column through numerical modeling using the computer program UNSAT-H. Two soil types and four temperature conditions were considered, and the behavior of the geotextile under drainage and capillary rise were evaluated. Matric suction profiles along the column under different temperature conditions are presented and the performance of the geotextile as a drainage/capillary barrier is evaluated in terms of variations in suction head and associated change in the hydraulic conductivity. 2.2 Overview of the Numerical Model UNSAT-H (Fayer and Jones, 1990; Fayer, 2000) is a finite difference program for simulating air, water, and heat flow in one dimension. It has been previously shown to reasonably reproduce field measurements for infiltration, storage, and drainage in soil profiles (Fayer et al. 1992; Khire et al. 1997; Khire et al. 1999; Wilson et al. 1999). UNSAT-H simulates water flow using a modified form of Richards equation (Richards 1931): where θ is volumetric water content, h is suction head, t is time, K L is hydraulic conductivity, z is depth below the surface, H is total hydraulic head, and S is a sink term used to capture transpiration (zero in the current work). Note that suction is equivalent to positive-valued negative pore water pressure. (2-1) 11

27 Richards equation does not include the vapor flow, so Fick s law is used to capture thermal and isothermal vapor flow: (2-2) where q v is flux density of water vapor, D is vapor diffusivity in soil, is density of water, is saturated vapor density, M is molecular weight of water, g is gravitational constant, R is the gas constant, T is temperature, H R is relative humidity, h is suction head, z is depth below the surface, and is an enhancement factor to consider increased cross section area and decreased path length for vapor diffusion. The first term in the right hand side of the Fick s law accounts for isothermal vapor flow, and the second term accounts for thermal vapor flow. Heat flow is calculated using Fourier s law of heat conduction: where q h is heat flux density, k h is thermal conductivity, T is temperature, and z is depth below the surface. (2-3) Thermal conductivity may be approximated as a function of volumetric water content as follows (Cass et al. 1984): where θ is water content, θ s is saturated water content, and a, b, c, d, and e are constants. (2-4) 12

28 The heat flux, vapor flow, and liquid flow equations are solved using a modified Picard iteration scheme (Celia et al. 1990; Fayer and Jones 1990). The user specifies the spatial discretization of the profile (an example profile with the geotextile on the bottom is presented in Figure 2-1) and the time step is adjusted to a value necessary to provide a stable solution. UNSAT-H does not simulate the freeze/thaw process. 2.3 Material Properties Two types of soil were considered in the simulations: sand and silty clay. A layer of geotextile was embedded inside the soil profile as a capillary barrier. Hydraulic properties for the sand, silty clay, and geotextile are presented in Table 2-1. The van Genuchten equation (van Genuchten 1980) was used to model the moisture characteristics curves, and the van Genuchten-Mualem model (Mualem 1976; van Genuchten 1980) was used to predict the unsaturated hydraulic conductivity of the materials. Soil properties were selected from the probabilistic recommendations of Carsel and Parrish (1988). Hydraulic properties of gravel were chosen for the geotextile as it has been shown by Stormont et al. (1997) and Stormont and Morris (1998) that geotextiles behave similarly to gravels when unsaturated (Stormont et al., 1997; Stormont and Morris, 1998). The moisture characteristic and hydraulic conductivity curves for the soils and geotextile are shown in Figure 2-2. The constants in thermal conductivity equation (equation 2-4) were selected the same for both types of soil and geotextile (a = 0.6, b = 0.7, c = 8.0, d = 0.26, e = 3.0). These values were taken from Cass et al. (1981). The constants in thermal conductivity equation affect the temperature distribution in the profile with temperature gradient. Given the small temperature gradient 13

29 used in this study (4 C), the effects of thermal conductivity constants on the temperature distribution and, consequently, thermal vapor flow was not substantial. 2.4 Simulation Results Four cases were considered in the current work: (1) the geotextile placed at the bottom of the column with an initial pressure head of zero along the column to simulate drainage; (2) the geotextile placed at the top of the column, and the water table initially placed at the bottom to simulate capillary rise; (3) the geotextile placed in the middle with an initial pressure head of zero along the column; and (4) the geotextile placed in the middle and the water table at the bottom. The conditions for Cases 3 and 4 describe a scenario in which the geotextile layer may be embedded within the soil. For all of the simulations, the height of the soil column was 30 cm, and the thickness of the geotextile layer was 1 cm. As mentioned previously, two types of soil were simulated: sand and silty clay. The silty clay is considered to be relatively susceptible to frost heave due to capillary and electrochemical retention of water in the soil pores. The sandy soil has a higher permeability than the silty clay and allows a downward flow of water into the subsurface during precipitation. Four different temperature conditions were considered: a constant temperature of 0 C or 38 C along the column, and two temperature gradients: one of 0 C at the top and 4 C at the bottom of the profile and one of 4 C at the top and 0 C at the bottom of the column. Temperatures of 0 C and 38 C were selected to represent cold and hot regions in the United States, respectively. The temperature gradient of 4 C was selected for the 30 cm column to be consistent with a typical temperature gradient range for pavements of 0 C/m to 12 C/m (Henry 1988). 14

30 2.4.1 Geotextile at the bottom of the column drainage condition The boundary conditions in the first case were a no flux boundary at the top of the profile and a unit gradient at the bottom. The initial condition was zero pressure head along the length of the column. The geotextile was placed at the bottom of the profile. The soil-geotextile column was allowed to drain under gravitational force for 50 days. Figure 2-3 shows suction head distributions along the silty clay-geotextile column for two constant temperatures of 0 C and 38 C. According to equation 2-2, when temperature is constant along the length of the model, there is no thermally induced vapor flow and changing the constant temperature from 0 C to 38 C does not have a significant effect on the suction head distribution. A temperature gradient, however, results in thermally-driven vapor flow and, consequently, changes the suction head distribution. Flux density of water vapor can be obtained as follows: where q v is flux density of water vapor, D is vapor diffusivity, ρ w is density of water, ρ v is vapor density, z is depth below the surface. (2-5) Vapor density is a function of relative humidity through the following equation: where ρ v is vapor density, ρ vs is saturated vapor density, and H R is relative humidity. (2-6) Relative humidity is related to temperature as follows: 15

31 where H R is relative humidity, h is suction, M is molecular weight of water, g is the gravitational constant, R is the gas constant, and T is temperature. (2-7) According to equation 2-6 and 2-7, relative humidity and consequently vapor density have a direct relationship with temperature. Based on equation 2-5, water vapor flows from a point with higher vapor density to a point with lower vapor density and thus, a temperature gradient induces thermal vapor flow from the higher temperature toward the lower temperature. Figure 2-4 shows the suction head distribution along the silty clay-geotextile column for both temperature gradients. To facilitate comparison, the difference in suction head between the column with a constant temperature of 0 C and the columns with the temperature gradient is presented in Figure 2-5. Negative values indicate an increase in suction head compared to the column with constant temperature of 0 C. Thus, a temperature gradient with 0 C at the top results in thermal vapor flow from the bottom of the profile to the top of the profile, as shown in Figure 2-5(a), and suction head decreases in the soil. However, suction head increases at the bottom of the geotextile from approximately 8 cm to 10 cm after 50 days. As can be seen in Figure 2-2(b), the geotextile has a very steep hydraulic conductivity curve in this suction range. Thus, the temperature gradient of 0 C at the top and 4 C at the bottom decreased the hydraulic conductivity of the geotextile by a factor of two. Reversing the temperature gradient reversed the thermal vapor flow direction and suction head increased in soil and decreased at the bottom of the geotextile. A temperature gradient of 4 C at the top and 0 C at the bottom decreased the suction head in 16

32 the geotextile from 8 cm to 6 cm and increased its hydraulic conductivity by approximately a factor of four after 50 days. Suction head distributions in sand-geotextile columns for constant temperatures of 0 C and 38 C are shown in Figure 2-6. Similar to the silty clay-geotextile column, changing the constant temperature from 0 C to 38 C did not have a significant effect on suction head distributions. However, a temperature gradient resulted in changes to the suction head profile and the hydraulic conductivity of the geotextile. A temperature gradient from 0 C at the top to 4 C at the bottom increased the suction head in the geotextile from 4 cm to 5 cm and decreased the hydraulic conductivity of the geotextile by half. The inverse temperature gradient decreased the suction head of the geotextile from 4 cm to 3 cm and increased the hydraulic conductivity of the geotextile by a factor of three Geotextile at the top of the column capillary rise condition In the second case of simulations, the geotextile was placed at the top of the profile to investigate the temperature effects on a soil-geotextile column under capillary rise conditions (upward water flow). The boundary and initial conditions were no flux boundary at the top and along the side walls, initial pressure head of zero at the bottom and hydrostatic pressure distribution along the column, and heat and vapor flow were allowed. Then, the suction head at the bottom of the column was decreased to -5 cm (5 cm positive pore pressure) to cause upward water flow due to capillary forces for 50 days. Figure 2-7 shows suction head distributions along the silty clay-geotextile columns with constant temperature of 0 C and 38 C. Similar to Case 1, changing the constant temperature from 0 C to 38 C had a slight 17

33 effect on the suction head profile throughout the silty clay and sand-geotextile columns. Temperature gradient had a small effect on the suction head of both the silty clay and the sand (<0.05 cm); however, it had a significant effect on the suction head in the geotextile in both the silty clay and sand columns. Figure 2-8 shows the difference in suction head between the silty clay-geotextile column with a constant temperature of 0 C and the silty clay-geotextile columns with the temperature gradient. As it can be seen from Figure 2-8, temperature gradient with 0 C at the top decreased suction in the geotextile by 28 cm after 50 days due to upward thermal vapor flow into the geotextile. Consequently, the hydraulic conductivity of the geotextile became close to its saturated value and increased by eight orders of magnitude. On the other hand, the temperature gradient with 4 C at the top increased the suction in the geotextile by cm due to downward thermal vapor flow from the geotextile. The initial suction head in the geotextile is 30 cm and as it can be seen from Figure 2-2b, at this suction level, decreasing a small amount of moisture content results in a large increase in suction head. Consequently, the downward thermal vapor flow increased the suction head significantly in the geotextile and the turned the geotextile into an impermeable layer. In sand-geotextile columns, temperature gradient with 0 C at the top decreased the suction by 27 cm in the geotextile and the reverse temperature gradient increased the suction by cm in the geotextile Geotextile in the middle of the column drainage condition In Case 3, the initial and boundary conditions are the same as for Case 1, but the geotextile layer is placed in the middle of the simulated column to study the effect of location on suction head distribution. During the drainage, water flows into the geotextile from the 18

34 overlying soil and from geotextile to the underlying soil under gravitational force. Figure 2-9 shows suction head distributions for both silty clay and sand geotextile columns under constant temperature of 0 C along the column. In the silty clay-geotextile column (Figure 2-9a), suction head increases with time along the column. When suction head increases, hydraulic conductivity of the geotextile decreases rapidly and the geotextile functions as a barrier and reduces the drainage from the overlying soil to the underlying soil. However, water can be drained freely from the underlying soil to the bottom boundary (unit gradient). As a result, higher suction head is developed in the underlying soil compared to the overlying soil during the drainage. In sand-geotextile column (Figure 2-9b), a negative suction head (positive pore pressure) is developed in the geotextile and the adjacent soil above it during drainage. This can be attributed to the developed suction head at the bottom of the geotextile due to drainage of the underlying soil and high hydraulic conductivity of the overlying soil under low suction heads (<10 cm). Accordingly, the geotextile cannot drain all the water coming from the overlying layer and water accumulates above it. This is consistent with observations made by Iryo and Rowe (2004), Bathurst et. al (2007), and Krisdani et. al (2008) for sand-geotextile columns where thermal effects were not considered. The drainage simulations were performed for the constant temperature of 38 C and the two temperature gradients as well. Similar to case 1, changing the constant temperature from 0 C to 38 C did not have a significant effect on the suction head distribution. Figure 2-10 a and b show the suction head difference between the silty clay-geotextile column with constant temperature of 0 C and the silty clay-geotextile column with temperature gradient of 0 C at 19

35 the top and 4 C at the bottom in the soil and geotextile, respectively. As it can be seen from Figure 2-10, temperature gradient increased the suction head by 125 cm in the geotextile after 50 days. In the underlying soil, however, suction head first decreased due to upward vapor flow and then increased by approximately 5 cm after 50 days. The reason for increasing suction in the underlying soil can be attributed to the increased suction in the geotextile which brings about lower hydraulic conductivity and preventing water from breaking into the underlying soil. Reversing the temperature gradient decreased suction head in the geotextile and consequently increased its hydraulic conductivity. This trend is similar to observations made for the previous cases. Figure 2-11 shows the suction head difference between the silty claygeotextile column with constant temperature of 0 C and the silty clay-geotextile column with temperature gradient of 4 C at the top and 0 C at the bottom. Temperature gradient decreased the suction head in the geotextile and consequently, increased its hydraulic conductivity. During the initial period of the drainage, and when suction head is small, the geotextile has higher hydraulic conductivity and the change in suction head has higher effect on its hydraulic conductivity. For example, after the first day, suction head decreased by 10 cm (from 45 cm to 35 cm) at the bottom of the geotextile and doubled the hydraulic conductivity of the geotextile. Consequently, more water flowed into the underlying soil during the drainage and suction head decreased in the underlying soil. In the sand-geotextile column, temperature gradient did not have a significant effect on the suction head in the geotextile due to the developed positive pore pressure in the geotextile 20

36 and the adjacent soil above it. The suction head difference between the sand-geotextile column with constant temperature of 0 C and the sand-geotextile column with temperature gradients are shown in Figure As it can be observed from Figure 2-12a, temperature gradient with 0 C at the top, decreased the suction head in the soil below the geotextile due to upward thermal vapor flow. Because geotextile gets saturated in sand-geotextile columns and temperature gradient does not affect its hydraulic conductivity, contrary to silty claygeotextile columns (Figure 2-10a) suction head did not increase in the soil below the geotextile with passing time. Similarly, the temperature gradient with 4 C at the top did not change the hydraulic conductivity of the geotextile, and contrary to silty clay-geotextile column (Figure 2-11a) the suction head in the soil below the geotextile increased due to downward thermal vapor flow (Figure 2-12b) Geotextile in the middle of the column capillary rise In Case 4, the geotextile layer is placed in the middle of the simulated column and the initial and boundary conditions are the same as for Case 2. Figure 2-13 shows suction head along the silty clay and sand-geotextile columns with constant temperature of 0 C. Similar to previous cases, changing the constant temperature from 0 C to 38 C did not have a significant effect on the suction head in both columns. Figure 2-13 shows the suction head difference between the silty clay-geotextile column with constant temperature of 0 C and the silty clay-geotextile column with temperature gradients. As it can be seen from Figure 2-14, similar to what was observed in Figure 2-8a, temperature gradient with 0 C at the top decreased suction and increased hydraulic conductivity in the geotextile by a factor of three and consequently, more water flows into the overlying soil leading to 2.2 cm decrease in 21

37 suction in the soil above the geotextile after 50 days. Conversely, the temperature gradient with 4 C at the top turns the geotextile into a moisture barrier and increases the suction head by 2 cm in the soil above the geotextile. The same trend was observed for sand-geotextile columns as it is shown in Figure After 50 days, temperature gradient with 0 C at the top decreased the suction in the geotextile and the overlying soil by 2.8 and 0.4 cm, respectively. Similarly to the silty clay-geotextile column, a reverse temperature gradient turned the geotextile into a barrier and increased the suction in the overlying soil by 0.4 cm after 50 days. 2.5 Discussion of Results The hydraulic behavior of geotextiles emplaced in unsaturated soils depends on the suction head and thermal gradient. Thermal gradients cause thermal vapor flow and moisture movement in unsaturated porous materials and changes the suction head profile. Geotextiles have relatively steep moisture retention characteristics and hydraulic transmissivity curves and a small change in the suction can cause significant changes in their hydraulic behavior. Accordingly, thermal gradients can significantly change the hydraulic transmissivity of a geotextile. Thus, the thermal and hydraulic conditions under which a geotextile will function should be considered in the design. According to the results, changing the constant temperature did not affect the hydraulic behavior of the geotextile. The temperature along an unsaturated soil profile, however, is not constant and changes with depth resulting in vapor flow. The change in temperature decreases with depth and accordingly, thermal gradients and thermal vapor flow are higher closer to the soil surface. 22

38 In hot regions, the temperature of the soil surface is higher which leads to downward water vapor flow toward the lower temperature. This downward water vapor flow increased the transmissivity of the geotextile emplaced in the unsaturated soil by a factor of four during drainage. This increase in hydraulic transmissivity eased the downward moisture flow and decreased the suction in the underlying soil. Thus, a drainage geotextile placed closer to the soil surface will have higher transmissivity and drainage capacity and, if sloped, can divert the moisture from the overlying soil toward the edge. The temperature gradient, however, did not affect the hydraulic transmissivity of the geotextile when positive pore water pressure developed in the geotextile and the adjacent soil during drainage. During capillary rise, the downward vapor flow decreased the hydraulic transmissivity of the embedded geotextile and reduced the upward moisture flow into the overlying layer. Accordingly, in hot regions, capillary barrier geotextiles placed closer to the surface have lower transmissivity and mitigate the upward water flow into the upper layer. In cold regions, the lower temperature on the surface results in upward water vapor flow toward the surface. According to the results, the upward water vapor flow decreased the hydraulic transmissivity of the geotextile by a factor of two during drainage and reduced the drainage capacity of the geotextile. Thus, in cold regions, the geotextile placed deeper in the soil profile has higher drainage capacity during drainage. Conversely, the upward water vapor flow increased the hydraulic transmissivity of the geotextile close to its saturation value during capillary rise and increased the upward water flow into the overlying soil. Thus, in cold regions, the geotextile emplaced deeper in the soil profile has lower hydraulic transmissivity during capillary rise and can mitigate the upward water flow into the overlying soil. 23

39 2.6 Conclusion The effect of temperature on the hydraulic transmissivity of the geotextile as a drainage/capillary barrier layer and on the suction head distribution in a soil-geotextile column is studied in this paper. A series of 1-D thermo-hydraulic simulations were performed using UNSAT-H and the geotextile functionality in terms of drainage/moisture barrier under different temperature and hydraulic gradient conditions was investigated. Changing the constant temperature from 0 C to 38 C had only minor effect on suction head distributions in the soil and geotextile. Temperature gradient changed the suction head and hydraulic conductivity of the geotextile and accordingly, the suction head in the soil below and above it. During drainage condition, temperature gradient with 0 C at the top and 4 C at the bottom of the silty clay-geotextile column increased the suction in the geotextile by up to 125 cm. Consequently, the decreased hydraulic transmissivity of the geotextile reduced the downward water flow into the underlying soil and increased the suction head in the soil below the geotextile by 5 cm after 50 days. Temperature gradient with 4 C at the top and 0 C at the bottom of the column increased the hydraulic transmissivity of the geotextile by a factor of two and led into more water flow and 5 cm suction decrease in the underlying soil. In sand-geotextile column and when positive pore pressure was developed in the geotextile, temperature gradient did not change the suction and hydraulic conductivity of the geotextile. During the capillary rise condition, temperature gradient with 0 C at the top and 4 C at the bottom of the column decreased the suction head in the geotextile by 4 and 3 cm in silty clay and sand columns, respectively. Consequently, the hydraulic conductivity of the geotextile 24

40 increased and more water flew into the overlying soil and decreased its suction by 2.2 cm and 0.4 cm in silty clay and sand columns, respectively. Temperature gradient with 4 C at the top and 0 C at the bottom of the column decreased the hydraulic transmissivity of the geotextile and reduced the upward water flow into the soil above the geotextile. Accordingly, suction head increased by 2 cm and 0.4 in the soil above the geotextile in silty clay and sand columns, respectively after 50 days. 2.7 Acknowledgements The work described in this paper was supported by Federal Highway Administration, Central Federal Lands Highway Division with Marilyn Dodson serving as the project committee chair and Roger Surdahl serving as the project administrator. The authors acknowledge the project committee (Daniel Alzamora, Rich Barrows, Jason Dietz; and Khalid Mohamed) for their input and guidance. 2.8 References Bathurst, R.J, Ho, A. F. & Siemens, G. (2007). A column apparatus for investigation of 1-D unsaturated saturated response of sand-geotextile systems. Geotechnical Testing Journal, 30, No. 6, Bouazza, A., Fruend, M. & Nahlawi, H. (2006). Water retention of nonwoven polyester geotextiles. Polymer Testing, 25, No. 8, Carsel, R. & Parrish, R.S. (1988). Developing joint probability distribution of soil water retention characteristics. Water Resources Research, 24, No. 5, Cass, GS Campbell & Jones, T.L. (1981). Hydraulic and thermal properties of soil samples from the Buried Waste Test Facility. Report No. PNL-4015, Pacific Northwest Laboratory, Richland, Washington. Cass, A., Campbell, G.S. & Jones, T. L. (1984). Enhancement of thermal water vapor diffusion in soil. Soil Science Society of America, 48,

41 Christopher, B. R., Hayden, S. A. & Zhao, A. (2000). Roadway Base and Subgrade Geocomposite Drainage Layers. Testing and Performance of Geosynthetics in Subsurface Drainage, ASTM STP Clough, I. R. & French, W. J. (1982). Laboratory and field work relating to the use of geotextiles in arid regions. Proceedings of the 2 nd International Conference on Geotextiles, IFAI, Las Vegas, Nevada. Dwyer, S. F. (1998). Alternative landfill covers pass the test. Civil Engineering, 68, No. 9, Dwyer, S. F. (2001). Finding a better cover. Civil Engineering, 71, No. 1, Faust, C. R. & Mercer, J. W. (1979). Geothermal reservoir simulation: 1. Mathematical models for liquid- and vapour- dominated hydrothermal systems. Water Resources Research, 15, No. 1, Fayer M. (2000). UNSAT-H Version 3.0: unsaturated soil water and heat flow model, theory, user manual, and examples. Rep Battelle Pacific Northwest Laboratory, Hanford, Washington. Fayer, M., and Jones, T. (1990). Unsaturated soil-water and heat flow model, ver:2.0, pacific Northwest Laboratory, Richland, Wash. Fayer, M., Rockhold, M. & Campbell, M. (1992). Hydrologic modeling of protective barriers: comparison of field data and simulation results. Soil Science Society of America Journal, 56, Henry, K. (1996). Geotextiles to mitigate frost effects in soils: A critical review. Transportation Research Record, 1534, pp Henry, K. & Holtz, R. (2001). Geocomposite capillary barriers to reduce frost heave in soils. Canadian Geotechnical Journal, 28, No. 4, Henry, K. S., Stormont, J. C., Barna, L. A. & Ramos, R. D. (2002). Geocomposite capillary barrier drain for unsaturated drainage of pavements. Proceedings of the Seventh International Conference on Geosynthetics, Nice, France, pp Iryo, T. & Rowe, R. K. (2005). Hydraulic behavior of soil-geocomposite layers in slopes. Geosynthetics International, 12, No. 3, Khire, M., Benson, C., & Bosscher, P. (1997). Water balance modeling of earthen final covers. Journal of Geotechnical & Geoenvironmental Engineering, ASCE, 123, No.8,

42 Khire, M., Benson, C. & Bosscher, P. (1999). Field data from a capillary barrier and model predictions with UNSAT-H. Journal of Geotechnical & Geoenvironmental Engineering, ASCE, 125, No. 6, Milly, P. C. D. (1982). Moisture and heat transport in hysteretic, inhomogeneous porous media: a matric head-based formulation and a numerical model. Water Resources Research, 18, No. 3, Krisdani, H., Rahardjo, H. & Leong, E. C. (2008). Measurement of geotextile-water characteristic curve using capillary rise principle. Geosynthetics International, 15, No. 2, McCartney, J. S., Kuhn, J. A. & Zornberg, J. G. (2005). Geosynthetic drainage layers in contact with unsaturated soils. Proceedings of the 16th International Conference on Soil Mechanics and Geotechnical Engineering (ISSMGE), Osaka, Japan, September, pp McCartney, J. S. & Zornberg, J. G. (2010a). Effect of infiltration and evaporation on geosynthetic capillary barrier performance. Canadian Geotechnical Journal, 47, Mohamed, A.M.O. & Shooshpasha, I. (2004). Hydro-thermal performance of multilayer capillary barriers in arid lands. Geotechnical and geological engineering, 22, Mualem, Y. (1976). A new model for predicting the hydraulic conductivity of unsaturated porous media. Water Resources Research, 12, No. 3, Philip, J. R. and de Vries, D. A. (1957). Moisture movement in porous materials under temperature gradients, Eos, Transactions, American Geophysical Union, 38, No. 2, Ramos, R. D. (2001). Performance of a Fiberglass Based Geocomposite Capillary Barrier Drain. M.Sc. thesis, University of New Mexico, Albuquerque. Richards LA. (1931). Capillary conduction of liquids through porous mediums. Physics, 1, Shooshpasha, I., Mohamed, A. M. O. & Yong, R. N. (2000). Geoengineering in Arid Lands. Coupled heat and moisture flow in unsaturated soils under opposing thermal and hydraulic gradients, Mohamed and Al Hosani Editors, Balkema, Rotterdam, pp Stormont, J.C. and Anderson, C.E. (1999). Capillary barrier effect from underlying coarser soil layer. Journal of Geotechnical and Geoenvironmental Engineering, ASCE, 125, No. 8, Stormont, J. C., Henry, K. S. &Evans, T. M. (1997). Water retention functions of four nonwoven polypropylene geotextiles. Geosynthetics International, 4, No. 6,

43 Stormont, J. C., Henry, K. S. & Roberson, R. (2009). Geocomposite Capillary Barrier Drain for Limiting Moisture Changes in Pavements: Product Application. Contract No. NCHRP- 113, Highway IDEA Project, Washington, D.C. Stormont, J. C., Hines, J. S., Pease, R. E., Kelsey, J. A & Dowd, D. (2008). The effectiveness of a geotextile as a capillary barrier. GeoAmericas 2008, Cancun, Mexico (CD-ROM). Stormont, J. C.& Morris, C.e. (1998). Method to estimate water storage capacity of capillary barriers. Journal of Geotechnical and Geoenvironmental Engineering. 124, No. 4, Stormont, J. C. & Morris, C. E. (2000). Characterization of unsaturated nonwoven geotextiles. Advances in Unsaturated Geotechnics Proceedings, Geotechnical Special Publication No. 99, C. D. Shackelford, S. L. Houston, and N. Y. Chang, eds., Geo-Institute of the ASCE, Denver, Colorado, Aug. 2000, pp Stormont, J. C., Ramos, R. & Henry, K. S. (2007). Geocomposite capillary barrier drain system with fiberglass transport layer. Transportation Research Record, 1772, pp Van Genuchten, M.Th. (1980). A closed-form equation for predicting the hydraulic conductivity of unsaturated soils. Soil Science Society of America Journal, 44, No. 5, Wilson, G., Albright, W., Gee, G., Fayer, M. & Ogan, B. (1999). Alternative cover assessment project phase I report. Control No. 68-C5-0036, WA#14, U.S. EPA, U.S. Environmental protection Agency, Washington, D.C. Yong, R. N. & Mohamed, A. M. O. (1992). A study of particle interaction energies in wetting of unsaturated expansive clays. Canadian Geotechnical Journal, 29, No. 6, Yong, R. N., Mohamed, A. M. O., Shooshpasha, I. & Onofrei, C. (1997). Hydro-thermal performance of unsaturated bentonite-sand buffer material. Engineering Geology, 47,

44 Table 2-1. Hydraulic properties of the materials used in the simulations θ s θ r α (1/cm) n K s (cm/hr) Sand Silty clay Geotextile

45 Figure 2-1. Schematic diagram of the discritized profile 30

46 Figure 2-2. a) Moisture characteristic curves b) hydraulic conductivity curves 31

47 Figure 2-3. Suction head distribution along the silty clay-geotextile culomn a) constant temperature of 0 C b) constant temperature of 38 C 32

48 Figure 2-4. Suction head distribution along the silty clay-geotextile column a) constant temperature of 0 C at the top and constant temperature of 4 C at the bottom b) constant temperature of 4 C at the top and constant temperature of 0 C at the bottom 33

49 Figure 2-5. Difference in suction head between the silty clay-geotextile column with a constant temperature of 0 C and the silty clay-geotextile column with the temperature gradient of a) 0 C at the top and 4 C at the bottom b) 4 C at the top and 0 C at the bottom ( the bottom figures are the magnified view of suction head profiles in the geotextile layer) 34

50 Figure 2-6. Suction head distribution along the sand-geotextile culomn a) constant temperature of 0 C b) constant temperature of 38 C 35

51 Figure 2-7. Suction head distribution along the silty clay-geotextile culomn a) constant temperature of 0 C b) constant temperature of 38 C (the top figures are the magnified view of suction head profiles in the geotextile layer) 36

52 Figure 2-8. Difference in suction head between the column with a constant temperature of 0 C and the silty clay-geotextile column with the temperature gradient of a) 0 C at the top and 4 C at the bottom b) 4 C at the top and 0 C at the bottom (the top figures are the magnified view of suction head profiles in the geotextile layer) 37

53 Figure 2-9. Suction head distribution under constant temperature of 0 C in a) silty clay-geotextile column b) sand-geotextile column 38

54 Figure Difference in suction head between the column with a constant temperature of 0 C and the silty clay-geotextile column with the temperature gradient of 0 C at the top and 4 C at the bottom a) in soil b) in geotextile 39

55 Figure Difference in suction head between the column with a constant temperature of 0 C and the silty clay-geotextile column with the temperature gradient of 4 C at the top and 0 C at the bottom a) in soil b) in geotextile 40

56 Figure Difference in suction head between the column with a constant temperature of 0 C and the sand-geotextile column with the temperature gradient of a) 0 C at the top and 4 C at the bottom b) 4 C at the top and 0 C at the bottom 41

57 Figure Suction head under constant temperature of 0 C in a) silty clay-geotextile column b) sandgeotextile column 42

58 Figure Suction head difference between the silty clay-geotextile column with constant temperature of 0 C and the silty clay-geotextile column with temperature gradient of a) 0 C at the top and 4 C at the bottom b) 4 C at the top and 0 C at the bottom (the top figures are the magnified view of suction head profiles in the geotextile layer) 43

59 Figure Suction head difference between the sand-geotextile column with constant temperature of 0 C and the sand-geotextile column with temperature gradient of a) 0 C at the top and 4 C at the bottom b) 4 C at the top and 0 C at the bottom (the top figures are the magnified view of suction head profiles in the geotextile layer) 44

60 CHAPTER 3 Moisture distribution in a partially saturated pavement section with drainage/moisture barrier layer This chapter is submitted for publication in Geotextiles and Geomembranes and is currently under review. The authors are: Mahdi Bahador, Mohammed A. Gabr, and T. Matthew Evans ABSTRACT: The effect of geosynthetic layers on moisture distribution in a partially saturated pavement profile was studied. The FEM program SIGMA/W was used to simulate infiltration into an unsaturated three-layer unpaved road consisting of, from top to bottom, crushed stone, a geosynthetic layer, and silty sand subgrade. One and two layers of geosynthetic materials were placed at the crushed stone-silty sand interface as drainage/moisture barrier layers to divert water toward the edge drains under partially saturated conditions. A sensitivity analysis was performed to study the effect of the air entry value and saturated hydraulic conductivity of the drainage/moisture barrier layer on drainage system performance. Simulation results indicated that the drainage/moisture barrier layer with either only a geotextile layer or a geomembrane underlying a geotextile layer increased the degree of saturation in the partially saturated profile during infiltration. A geocomposite consisting of a geotextile with a high air entry value overlying a geonet decreased the degree of saturation in the underlying subgrade soil while maintaining suction in the overlying 45

61 crushed stone during infiltration. A higher air entry value for the geotextile resulted in higher suction in the underlying layer so long as this air entry value did not exceed the air entry value of the overlying soil layer. KEYWORDS: geocomposite, moisture barrier, drainage, unsaturated flow, pavement 3.1 Introduction Geotextiles are widely used in several aspects of earth construction including as cushion, separation, or reinforcement layers. However, recent studies have shown that geotextiles (alone or in combination with other synthetics) may also act as effective drainage and/or moisture barrier layers when in contact with partially saturated soils (e.g. Henry, 1988; Stormont et al., 1998; Henry and Holtz, 2001; Stockton, 2001; Krisdani et al., 2006). One application of geotextiles as a drainage/moisture barrier layer (DMBL) is in pavement sections. Such a system has been patented by Henry and Stormont (2000) for a configuration termed the geocomposite capillary barrier drain (GCBD). In general, capillary barriers are composed of a layer of a relatively low permeability material over a more highly permeable material (e.g., Stormont and Anderson, 1999). In a pavement system, a capillary barrier can be formed by placing a high conductivity material at the interface of the aggregate base course (ABC) and subgrade to divert infiltrated water laterally toward edge drains and minimize moisture variations and shear strength reduction in the profile (Christopher et al., 2000; Henry and Stormont, 2000; Henry et al., 2002; Stormont et al., 2009). The geomaterials in a typical road cross-section are rarely saturated, so the unsaturated hydraulic properties of the geosynthetics used as a DMBL constitute a key design issue. Specifically, 46

62 these properties are the moisture characteristic curve (MCC) and in-plane transmissivity. The MCC represents the relationship between the water content of a porous medium and the potential energy of that pore water and can be measured experimentally for soils (ASTM D6836, Standard Test Methods for Determination of the Soil Water Characteristic Curve for Desorption Using Hanging Column, Pressure Extractor, Chilled Mirror Hydrometer, or Centrifuge) and geotextiles (Stormont et al., 1997; Lafleuer, 2000; Knight and Kotha, 2001; Stormont et al., 2001; Kuhn et al., 2005; Garcia et al., 2007; Nahlaw et al., 2007). Two important parameters for the design of porous geosystems for partially saturated conditions are the air entry value (AEV) and the water entry value (WEV). These two parameters can be determined from the MCC. The WEV is the suction level at which water begins to enter an initially dry matrix during the wetting process (related to the point at which the matrix becomes conductive for liquid water). The AEV is the suction at which an initially saturated porous matrix begins to desaturate during the drying process (related to the largest pore size in the matrix). Below the AEV, the matrix is saturated and exhibits high relative hydraulic conductivity (close to its saturated hydraulic conductivity). MCC and hydraulic conductivity curve of a porous matrix can be obtained from van Genuchten-Mualem equations, (Mualem, 1976; van Genuchten, 1980) as follows: (3-1) (3-2) 47

63 where: θ is volumetric water content, k is hydraulic conductivity, h is suction, is residual water content, is saturated water content, and α and n are fitting parameters. It has been shown that the MCCs of geotextiles are steeper than those for most types of fine grained soils, and are similar to coarse-grained soils such as gravel or uniform coarse sand (Stormont et al., 1997; Stormont et al., 2001; Iryo and Rowe, 2003). It was also found that the WEV of geotextiles is smaller than for most soils (Stormont et al., 1997; Stormont et al., 2001). These properties allow geotextiles to function as both drainage and moisture barrier layers (Zornberg and Mitchell, 1994). Under high suctions, water cannot enter the geotextile and it functions as a moisture barrier. However, as suction in the adjacent soil decreases to the WEV of the geotextile, water may enter the geotextile (the capillary barrier effect, see e.g., Stormont and Anderson, 1999), its hydraulic conductivity increases rapidly, and the geotextile functions as a drainage layer (Stormont et al., 2001). To quantify the one dimensional unsaturated-saturated hydraulic behavior of geotextiles embedded in soil layers, laboratory soil-geotextile column tests have been performed under drainage and/or surface infiltration (Ho, 2000; Iryo and Rowe, 2004; Bathurst et al., 2007; Krisdani et al., 2008; Stormont et al. 2008). In these studies, a geotextile was embedded in a soil column and the moisture content along the column axis was measured during drainage and/or infiltration. These studies showed that at suctions higher than the WEV of the geotextile, the low hydraulic conductivity of the unsaturated geotextile may lead to an accumulation of infiltrated water at the soil-geotextile interface. Such an accumulation leads to an increase in the degree of saturation of the soil above the geotextile and the consequent 48

64 decrease in shear strength. The accumulated water above the geotextile will drain when the soil suction reaches the WEV of the geotextile and the geotextile then acts as a drainage layer. In pavement sections, the accumulation of moisture above the geotextile can decrease the shear strength of the ABC and increase rutting. To mitigate the generation of positive (or neutral) pore pressure above the geotextile, a layer of geotextile with an appropriate WEV should be used and the geotextile should be inclined to convey water toward edge drains. Bahador et al. (2011) showed that if an inappropriate DMBL configuration is used, moisture content increases above the DMBL leading to a decrease in the shear strength of the ABC material and an increase in deformation by 68% compared to a pavement section not containing a geosynthetic. Previous studies on inclined capillary barriers showed that capillary barriers are effective over an effective length called diversion length (Ross, 1990; Morris and Stormont, 1999; Iryo and Rowe, 2005). Beyond this diversion length the capillary barrier cannot divert any additional water and water breaks into the underlying layer and the capillary barrier fails. The objective of this paper is to study the effect of the hydraulic properties of geotextile used as a DMBL on the moisture distribution within the section and its efficacy in minimizing significant moisture accumulation in the ABC while minimizing the moisture change in subgrade. An unsaturated seepage analysis is performed on a three layer pavement section consisting of ABC, a DMBL, and subgrade layers to assess moisture distribution throughout the profile during infiltration and drainage. A parametric analysis is performed on the effect 49

65 of AEV and saturated hydraulic transmissivity of the DMBL on moisture variation in the pavement section and corresponding accumulation in the ABC. The results of the simulations are discussed in view of proposed configuration and hydraulic properties of DMBL. 3.2 Material Properties The simulated profile consisted of three layers representing a typical pavement section: a crushed stone ABC, a woven fiberglass geotextile as the DMBL, and a silty sand subgrade. The hydraulic properties of these three layers are presented in Table 3-1, and MCCs and hydraulic conductivity curves are shown in Figure 3-1. In this paper, MCCs and hydraulic conductivity curves are presented using the van Genuchten-Mualem formulation and wetting and drying hysteresis is not considered. Hydraulic properties of the crushed stone were obtained from the Enhanced Integrated Climate Model database (EICM, 2006) and hydraulic properties for the silty sand were selected from the SEEP/W database (GEO-SLOPE International, 2002). The hydraulic properties for the fiberglass geotextile considered for the DMBL were measured by Ramos (2001). Fiberglass is a hydrophilic (low contact angle), organic, cross-linked, polymeric material which becomes transmissive at higher suctions relative to other types of polymeric geotextiles (Stormont and Ramos, 2004). 3.3 Model Verification The finite element software SIGMA/W was used to simulate infiltration and drainage into a three layer profile. The model was first verified against published results and then used for further simulations. The pore pressure data measured by Krisdani et al. (2008) in a soilgeotextile column were used for model verification. Krisdani et al. (2008) performed soil- 50

66 geotextile column tests on a one meter fine sand column in which a Polyfelt Megadrain 2040 geosynthetic was embedded 0.51 m from the top of the column. Polyfelt Megadrain 2040 consisted of a geonet sandwiched between two filter geotextiles. Drainage tests were performed by lowering the water table from the top of the column to the bottom. The pressure head at the bottom and along the column were measured during the test. The measured pore water pressures along the column are shown as data points in Figure 3-2 (Krisdani et al., 2008). The schematic diagram of the profile used in SIGMA/W to simulate the test configuration is presented in Figure 3-3. Boundary conditions were a no-flow boundary on the sides and the top, and a time dependent pressure head at the bottom based on the measured data. The hydraulic properties of the fine sand and geotextile layer used in the simulation were provided in Krisdani et al. (2008) and are shown in Table 3-2. The geonet was modeled as a layer with low air entry value (0.2 kpa). Additional test details can be found in Krisdani et al. (2008). The results of the numerical simulations are shown in Figure 3-2 along with the measured experimental data. Comparative data shown in Figure 3-2 indicate good agreement between numerical computations and experimental measurements, and serve to validate the applicability of the developed model 3.4 Numerical Simulations: Single-layer Geotextile Two dimensional seepage analyses were performed on a three layer profile simulating an unpaved roadway under infiltration and drainage conditions. The profile consisted of crushed stone, a fiberglass geotextile, and a silty sand subgrade. The surface of the profile and the fiberglass geotextile were inclined at a slope of 4.4%. Figure 3-4 shows a schematic diagram 51

67 of the model profile. In this case, a very fine mesh was necessary to represent the geotextile layer due to its relative thinness. Some of the elements were as small as 0.9 mm, such that four nodes were located vertically within the geotextile layer. Transition elements were used to increase the size of the mesh for the remaining model area. The hydraulic properties of the layers are presented in Table 3-1. The boundary conditions were a no-flow boundary on the left side, zero pressure head at the bottom, and a potential seepage boundary condition on the right side of the crushed stone and geotextile. The potential seepage boundary condition allows drainage of moisture from the boundary under saturated conditions. This simulates drainage of moisture from the geotextile into the edge pipe which is assumed saturated for simplicity. The top of the profile was a no-flow boundary during drainage and a constant flux of m/s (0.84 in/hr) during infiltration. This flux corresponds to a 6-hr rainfall with a 50 year return period for Raleigh, NC (North Carolina Erosion and Sediment Control Planning and Design, 2002). The initial conditions were a hydrostatic pore water pressure distribution throughout the silty sand and a constant pore pressure of kpa in crushed stone (corresponding to the optimum gravimetric water content of ~7.4%). Numerical simulations were performed in two stages. In the first stage, a constant hydraulic flux of m/s was applied to the top of the profile for 24 hrs to simulate infiltration. In the second stage, pore pressures at the end of the first stage were used as the initial condition and the profile was drained under gravity for 24 hrs. Figure 3-5 shows pore pressure contours after 10 hrs of simulated rainfall. Figure 3-5 shows that pore pressure contours are approximately horizontal, which also holds for all other simulation hours. To 52

68 allow for a better comparison, the pore pressure distributions along the centerline of the profile for the two cases (with and without geotextile) are plotted in Figure 3-6, which shows that after approximately six hours of infiltration and before the suction reaches the WEV of the geotextile (~10 kpa), the inclusion of the geotextile has a small effect on the pore pressure profile. However, as the moisture front advances in the crushed stone, the presence of the geotextile changes the pore pressure distribution in both the crushed stone and the silty sand. During the first 16 hours of infiltration and under unsaturated conditions, the geotextile exhibits very low hydraulic conductivity and functions as moisture barrier and prevents water from breaking into the silty sand. Consequently, the suction decreases in the crushed stone and increases in the silty sand compared to the case without the geotextile. Over time, and with the increase in infiltration volume, suction decreases, the hydraulic conductivity of the geotextile increases, and water breaks into the silty sand and eventually exits the profile from the right edge of the geotextile. Consequently, water is drained from the crushed stone and the suction in the crushed stone increases. Figure 3-7 shows pore pressure distributions for Stage 2 (24 hour drainage after 24 rainfall). When water is drained under gravity, the geotextile becomes unsaturated and its hydraulic conductivity decreases. Consequently, suction decreases in the crushed stone, whereas it remains higher in the silty sand as compared to the case without the geotextile. Thus, in order to mitigate water accumulation in the crushed stone, the geotextile needs to remain transmissive under unsaturated conditions and drain moisture from the crushed stone to the edge drain. 53

69 3.5 Geotextile Air Entry Value The AEV of a porous matrix is the suction at which moisture begins to leave the matrix. Thus, increasing the AEV increases the range of the suction under which the matrix is saturated and its hydraulic conductivity is close to the saturated value. In the van Genuchten- Mualem formulation, a decreasing α value corresponds to an increasing AEV. Keeping all other parameters constant, increasing the AEV implies an increase in the WEV of the matrix as well. This implies that water can enter the partially saturated porous matrix which can then function as a drainage layer at higher suction levels. Thus, a porous matrix with a higher AEV becomes conductive under higher suctions and may prevent water accumulation above it during infiltration and drainage. Figure 3-8 shows the MCC and hydraulic conductivity curves for the geotextiles with three different AEVs (0.19, 0.23, and 0.30 kpa), along with those for silty sand and crushed stone (the inverse of α value was selected as the AEV of geotextile). The selected AEVs for the geotextile are lower than that of Fiberglass geotextile (~0.39 kpa) and consistent with AEVs of non-woven polypropylene/polyester geotextiles (Ramos, 2001; Stormont and Morris, 2000). Figure 3-8 shows that increasing AEV increases the unsaturated hydraulic conductivity of DMBLs at constant suction and, consequently, conveys more water, both toward the edge drain and into the subgrade layer. 3.6 Numerical Simulations: Single-layer Geotextile, Varying AEVs In order to study the effect of AEV on moisture distribution during infiltration, three different AEVs were simulated for the DMBL: 0.19, 0.23, and 0.30 kpa. MCC and hydraulic conductivity curves for the DMBLs with the different AEVs are shown in Figure 3-8. The 54

70 same boundary and initial conditions as was used in previous analyses were applied and simulations were performed for 24 hr infiltration. Figure 3-9 shows pore pressure distributions along the centerline of the profiles with the different AEVs. During the first hours of rainfall and before the suction in the DMBL drops to its WEV, the AEV does not affect the pore pressure distribution (~ 6 hours). When suction drops to the WEV of the DMBL and water starts to enter the DMBL, the higher AEV results in lower suction in silty sand and a slight higher suction in crushed stone. The reason is that higher AEV increases unsaturated hydraulic conductivity of DMBL and infiltrating moisture moves into the silty sand. Thus, although a DMBL with a relatively higher AEV can drain more water from the overlying layer, it conveys much of this moisture into the underlying layer because the crosssectional area of flow is larger in the vertical direction than in the horizontal direction. Over time and with continued infiltration, the DMBL becomes saturated and conveys most of the infiltrated water from the crushed stone into the silty sand, and pore pressure distribution becomes almost identical for all AEVs. Accordingly, increasing the AEV of the DMBL increases the unsaturated hydraulic conductivity of the DMBL at a given suction which implies that if downward water flow into the silty sand is eliminated or limited, water can be drained from crushed stone while suction in the subgrade remains relatively unchanged. 3.7 Numerical Simulations: Geotextile Over Geomembrane A geocomposite DMBL consisting of a fiberglass geotextile with an AEV of 0.30 kpa overlying a geomembrane was modeled to eliminate downward water flow into the silty sand. In order to simulate the geomembrane, a no flow boundary was selected at the bottom 55

71 of the geotextile. Crushed stone was modeled overlying the geocomposite and the boundary and initial conditions were the same as the previous analyses and a flux of m/s was applied at the top of crushed stone to simulate rainfall. Figure 3-10 shows pore pressure contours after three and six hours of rainfall. Figure 3-10 shows that using a geomembrane underneath the geotextile can cause water accumulation and positive pore pressure in the crushed stone. The reason is that there is no downward water flow due to the presence of the geomembrane and the geotextile in this case does not have enough drainage capacity to convey all the water coming from crushed stone toward the edge drain. In order to increase the drainage capacity of the geotextile, its saturated hydraulic conductivity was increased by a factor of two and the simulation was repeated. Figure 3-11 presents pore pressure contours after six hours of rainfall ( m/s). As Figure 3-11 shows, doubling the saturated hydraulic conductivity of the geotextile decreased the developed positive pore pressure in the crushed stone, placing a geomembrane underneath the geotextile still generated positive pore pressure in the overlying layer. 3.8 Numerical Simulations: Geotextile Over Geonet One possible approach to mitigate water accumulation in the crushed stone is to place a second geosynthetic layer with a very low AEV below the upper geotextile to create a capillary break, as was proposed by Henry and Stormont (2000). A geocomposite DMBL consisting of a geotextile overlying a geonet was modeled to study the use of two synthetic layers at the interface of the crushed stone and silty sand. A very low AEV of 0.05 kpa was selected for the geonet. This low AEV is consistent with data reported in previous studies 56

72 (Ramos, 2001; Krisdani et al., 2008). The boundary and initial conditions were the same as what was used in the previous analyses and m/s infiltration rate was simulated for 24 hr. In order to model the two-layered DMBL, the finite element mesh is further refined to overcome an observed mesh and time step dependency. Adding the second layer nearly doubled the number of elements in the profile, leading to computational difficulties due to the very large stiffness matrices. Such difficulties in modeling thin layers of geotextile in a two dimensions domain have been reported in previous studies (Park and Fleming, 2006; Stormont, et al. 2009) as well. In order to model the two-layer DMBL, a narrow profile with 15 cm width was considered. A schematic diagram of the narrow profile is shown in Figure Pore pressure distributions along the centerline of the profile after 24 hrs rainfall are shown in Figure 3-13 for three different cases: without DMBL, a single-layer (geotextile only) DMBL, and a two-layer DMBL. The use of a two-layer DMBL increased suction level in the silty sand, as compared to the one-layer DMBL system and the profile without the DMBL, and led to maintaining suction level in the crushed stone. The geonet functions as a moisture barrier when it is unsaturated and increases the suction in the silty sand. Over time, suction decreases in the geonet until capillary condensation occurs and it begins to function as a drainage layer, thereby increasing the drainage capacity of the DMBL system. Thus, contrary to the geomembrane, using a geonet underneath a geotextile as a DMBL not only increased suction in subgrade, but also it drained water from the crushed stone. This finding is 57

73 consistent with previous experimental and field-scale studies (Henry et al., 2002; Stormont et al., 2009) Air Entry Value To study the effect of changing hydraulic properties of the DMBL on its function as a drainage/moisture barrier, four different AEVs of 0.19, 0.30, 0.76, and 6.65 kpa were selected for the geotextile overlying the geonet. The first two AEVs are lower than the AEV of the fiberglass geotextile (~0.39 kpa). The AEV of 0.76 kpa corresponds to that of sandy soils that exhibit high hydraulic conductivity and high AEVs (UNSODA, 1999). The last AEV is higher than AEV of crushed stone (5.41 kpa) and was chosen to study ways that high AEVs affect the functionality of the DMBL (the inverse of α value of the crushed stone was selected as its AEV). Figure 3-14 shows the effect of the AEV of the geotextile on pore pressure distributions in the profile. In this case, increasing the AEV of the geotextile increased the suction in the silty sand except for the case with the highest AEV during the first 6 hrs of rainfall and before the geotextile gets saturated. Changes in the AEV of the geotextile had a small effect on the suction in the crushed stone compared to that of the silty sand. With continued rainfall and saturating the geotextile layer, suction in the silty sand remained relatively constant regardless of AEVs used for the geotextile. Thus, increasing the AEV of the geotextile from 0.19 to 0.76 kpa increased suction in subgrade while maintaining the suction in crushed stone during the first six hours of rainfall and before saturating the geotextile. However, increasing the AEV of the geotextile to higher than AEV of the crushed stone decreased suction in the silty sand compared to other cases with lower AEVs. Figure 3-15 shows the MCCs and hydraulic conductivity curves for all the geotextiles 58

74 considered along with those for silty sand and crushed stone. Figure 3-15 shows that the WEV of the geotextile with an AEV of 6.65 kpa is higher than the AEV of the crushed stone. Thus, once the water can leave the crushed stone it can enter the geotextile (at a suction of about 5 kpa). At a suction equal to the AEV of crushed stone, the hydraulic conductivity of the geotextile is higher than the hydraulic conductivity of the crushed stone and is close to its saturated hydraulic conductivity. Consequently, the geotextile remains unsaturated and water cannot be drained into the saturated drain pipe. Over time, and once the suction decreases to the WEV of the geonet, water breaks into the geonet and then toward the edge and into the silty sand. As a result, suction in the silty sand decreases Saturated Hydraulic Conductivity In order to study the effect of the saturated hydraulic conductivity (k s ) of the geotextile on drainage capacity of the two layer DMBL system (geotextile over geonet), three different saturated hydraulic conductivities were used for the upper DMBL with AEV of 0.30 kpa: , , and m/s. The middle value is the saturated hydraulic conductivity of the fiberglass geotextile. Figure 3-16 shows pore pressure distributions along the centerline of the profile for these three saturated hydraulic conductivity values. Changing the saturated hydraulic conductivity of the upper layer had only a small effect on pore pressure distributions compared to the effect of different AEVs on pore pressures. Recall that increasing the AEV of the upper DMBL from 0.30 kpa to 0.76 kpa increased the suction in the silty sand by approximately 45% during the first 6 hours of rainfall and before the geotextile gets saturated (Figure 3-14) due to the shape of the hydraulic conductivity curves for geotextiles with different AEVs. Figure 3-17(a) shows hydraulic conductivity curves of a 59

75 geotextile with an AEV of 0.30 kpa and three different k s values, and Figure 3-17(b) shows the hydraulic conductivity curves of a geotextile with k s = and for two AEVs of 0.30 and 0.76 kpa. To enhance clarity, the hydraulic conductivity curves are plotted in both arithmetic and logarithmic scales. Figure 3-17 shows that changing k s changes the hydraulic conductivity at low suction values (<0.1 kpa), whereas high AEVs cause greater hydraulic conductivity under higher suctions. Based on the simulation results, during the infiltration and when water infiltrated the geonet, the geotextile became unsaturated and remained under suction (~0.25 to 3 kpa) for most of the analysis. Accordingly, changing k s and, consequently, the hydraulic conductivity of the geotextile under low suctions, did not have a significant effect on the pore pressure distribution. 3.9 Summary and Conclusions Moisture distributions in a three-layer profile simulating an unpaved road section with drainage/moisture barrier layers are presented in this paper. A series of seepage analyses were performed using the software SIGMA/W, and the effects of inclusion of drainage/moisture barrier layers on pore pressure distributions were examined. DMBLs with one and two layers were modeled and the pore pressure distribution along the centerline of the profile for each case is presented. The results showed that using only one layer of geotextile as the drainage/moisture barrier first decreased suction in the crushed stone during infiltration. With continuation of rainfall, water was drained from the ABC toward the edge drain and into the silty sand through the saturated geotextile which resulted in suction increase in the crushed stone and suction decrease in the silty sand. However, a geocomposite 60

76 consisting of a geotextile overlying a geonet increased suction in the silty sand during infiltration while the crushed stone remained unsaturated. Placing a geomembrane underneath the geotextile resulted in the development of positive pore pressure in the crushed stone (up to 1 kpa). Increasing the drainage capacity of the geotextile by doubling its saturated hydraulic conductivity decreased the developed positive pore pressure in the crushed stone but could not eliminate it. The air entry value of the geotextile was found to have a significant effect on the hydraulic behavior of the drainage/moisture barrier. In a one layer DMBL, increasing the air entry value of the geotextile from 0.19 to 0.30 kpa decreased the suction in the silty sand by up to 2 kpa because increasing the air entry value increased both the lateral and downward water flow. A geocomposite consisting of a geotextile with a high air entry value (0.76 kpa) overlying a geonet showed the highest suction in the silty sand during the initial stages of infiltration, whereas the suction in the crushed stone did not change. Increasing the AEV from 0.19 to 0.76 kpa increased the suction in the silty sand by about 45% during the first 6 hours of rainfall infiltration and before saturating the geotextile. However, the geocomposite consisting of a geonet underlying a geotextile with an air entry value higher than the air entry value of the crushed stone (6.65 kpa) decreased the suction in the silty sand by approximately 60% compared to the case with the AEV of 0.76 kpa during the first 6 hours of rainfall and before saturating the geotextile. Increasing the saturated hydraulic conductivity of the geotextile from to m/s did not have a significant effect on pore pressure distribution as long as the geotextile remained at a suction greater than 0.1 kpa. 61

77 3.10 Acknowledgement The work described in this paper was supported by Federal Highway Administration, Central Federal Lands Highway Division with Marilyn Dodson serving as the project committee chair and Roger Surdahl serving as the project administrator. The authors acknowledge the project committee (Daniel Alzamora, Rich Barrows, Jason Dietz; and Khalid Mohamed) for their input and guidance. The first author acknowledges the financial support of North Carolina State University References American Society for Testing and Materials (ASTM D )., Standard Test Methods for Determination of the Soil Water Characteristic Curve for Desorption Using a Hanging Column, Pressure Extractor, Chilled Mirror Hygrometer, and/or Centrifuge. Vol Bahador, M., Evans, T.M., and Gabr, M.A., Geotextile drains in road sections subjected to unsaturated conditions. In: Proceeding of Geo Frontiers 2011 Conference, Dallas, Texas. Bathurst, R., Ho, A. F., and Siemens, G., A column apparatus for investigation of 1-D unsaturated saturated response of sand-geotextile systems. Geotechnical Testing Journal, 30 (6), Christopher, B. R., Hayden, S. A., and Zhao, A., Roadway Base and Subgrade Geocomposite Drainage Layers. In: Testing and Performance of Geosynthetics in Subsurface Drainage, Seattle, Washington, Enhanced Integrated Climate Model (EICM)) version 3.2 Beta, Texas Transportation Institute, Texas A&M university. Garcia, E. F., Function of permeable geosynthetics in unsaturated embankments subjected to rainfall infiltration. Geosynthetics International, 14 (2), Henry, K. S., Use of geotextiles to mitigate frost heave in soils. In: Proceedings of the 5th International Conference on Permafrost, Trondheim, Norway,

78 Henry, K. S. and Holtz, R. D., Capillary rise of water in geotextiles. In: Proceedings of the 1997 International Symposium on Ground Freezing and Frost Action in Soils, Lulea, Sweden, Henry, K. S. and Stormont, J. C., Geocomposite Capillary Barriers Drain. US Patent No. 6,162,653.Henry, K. S. and Stormont, J. C., Geocomposite Capillary Barrier Drain for Limiting Moisture Changes in Pavement Subgrades and Base Courses. NCHRP-IDEA 68 Final Report, Transportation Research Board. Henry, K. S., Stormont, J. C., Barna, L. A., and Ramos, R. D., Geocomposite capillary barrier drain for unsaturated drainage of pavements. In: Proceedings of the Seventh International Conference on Geosynthetics, Nice, France, 3, Ho, A. F., Experimental and Numerical Investigation of Infiltration Ponding in Onedimensional Sand geotextile Columns. MSc thesis, Queen s University, Kingston, Ontario, Canada. Iryo, T. and Rowe, R. K., On the hydraulic behavior of unsaturated nonwoven geotextiles. Geotextiles and Geomembranes, 21 (6), Iryo, T. and Rowe, R. K., Numerical study of infiltration into a soil-geotextile column. Geosynthetics International, 11 (5), Iryo, T. and Rowe, R. K., Hydraulic behavior of soil-geocomposite layers in slopes. Geosynthetics International, 12 (3), Knight, M. A. and Kotha, S.M., Measurement of geotextile-water characteristic curves using a controlled outflow capillary pressure cell. Geosynthetics International, 8 (3), Krisdani, H., Rahardjo, H. & Leong, E. C., Experimental study of 1-D capillary barrier model using geosynthetic material as the coarse-grained layer. In: Proceeding of the 4 th International Conference on Unsaturated Soils, Carefree, Arizona, Krisdani, H., Rahardjo, H., and Leong, E. C., Measurement of geotextile-water characteristic curve using capillary rise principle. Geosynthetics International, 15 (2), Kuhn, J. A., McCartney, J. S., and Zornberg, J. G., Impinging flow over drainage layers including a geocomposite. In: Proceedings of the Sessions of the Geo-Frontiers 2005 Congress, Austin, Texas. Lafleur, J., Lebeau, M., Faure, Y. H., Savard, Y., and Kehila, Y Influence of Matric Suction on the Drainage Performance of Polyester Geotextiles. In: Proceedings of the 3rd International Conference, Geofilters 2000, Warsaw, Poland,

79 Morris, C. M. and Stormont, J. C., Parametric study of unsaturated drainage layers in a capillary barrier. Journal of Geotechnical and Geoenvironmental Engineering, 125 (12), Mualem, Y., A new model for predicting the hydraulic conductivity of unsaturated porous media. Water Resources Research, 12 (3), Nahlawi, H., Bouazza, A., and Kodikara, J., Characterisation of geotextiles water retention using a modified capillary pressure cell. Geotextiles and Geomembranes, 25 (3), Park, K. D. and Fleming, I. R., Evaluation of geosynthetic capillary barrier. Geotextiles and Geomembranes, 24 (1), Ramos, R. D., Performance of a Fiberglass Based Geocomposite Capillary Barrier Drain. M.Sc. thesis, University of New Mexico, Albuquerque, USA. Ross, B., The diversion capacity of capillary barriers. Water Resources Research, 26, GEO-SLOPE International, SEEP/W (Version 5.20) Manual, Calgary, Alberta, Canada. Stockton, T. B., Performance Experimental Testing of Diversion Lengths for a Geocomposite Unsaturated Drainage System. M.Sc. thesis, University of New Mexico, Albuquerque, New Mexico. North Carolina Erosion and Sediment Control Planning and Design., City of Raleigh's Storm Drainage Design Manual, Chapter 2, P40. Stormont, J. C., and Anderson, C. E., Capillary barrier effect from underlying coarser soil layer. Journal of Geotechnical and Geoenvironmental Engineering, 125(8), Stormont, J. C., Evans, T. M., Stockton, T., and Ray, C., Unsaturated hydraulic properties of a nonwoven polypropylene geotextile. In: Proceedings of the 98 Joint Conference on the Environment, Waste-Management Education and Research Consortium, Albuquerque, New Mexico. Stormont, J. C., Henry, K. S., and Evans, T. M., Water retention functions of four nonwoven polypropylene geotextiles. Geosynthetics International, 4 (6), Stormont, J. C., Henry, K. S., and Roberson, R., Geocomposite Capillary Barrier Drain for Limiting Moisture Changes in Pavements: Product Application. Final Report, Contract No. NCHRP

80 Stormont, J. C., Hines, J. S., Pease, R. E., Kelsey, J. A, and Dowd, D., The effectiveness of a geotextile as a capillary barrier. In: Proceedings of GeoAmericas 2008, Cancun, Mexico. Stormont, J. C., and Morris, C. E., Characterization of unsaturated nonwoven geotextiles. In: Proceeding of Advances in Unsaturated Geotechnics, Geotechnical Special Publication, ASCE, Denver, CO, Stormont, J. C., Ray, C., and Evans, T. M., Transmissivity of a nonwoven polypropylene geotextile under suction. ASTM Geotechnical Testing Journal, 24 (2), Stormont, J. C., and Ramos, R. D., Characterization of a fiberglass geotextile for unsaturated in-plane water transport. ASTM Geotechnical Testing Journal, 27 (2), Unsaturated soil hydraulic database (UNSODA), United States Department of Agriculture, v EPA/600/R-96/095. Van Genuchten, M.Th., A closed-form equation for predicting the hydraulic conductivity of unsaturated soils. Soil Science Society of America Journal, 44 (5): Zornberg, J. G. and Mitchell, J. K., Reinforced soil structures with poorly draining backfills. Part I: Reinforcement interactions and functions. Geosynthetics International, 1(2),

81 Table 3-1. van Genuchten parameters and saturated hydraulic conductivity for materials used in the modeling. (SEEP/W, 2002; EICM, 2006; Ramos, 2001) Saturated Water Content θ s Residual Water Content θ r α (1/kPa) n Saturated Hydraulic Conductivity (m/s) Silty sand Crushed stone Fiberglass geotextile

82 Table 3-2. Hydraulic properties of fine sand and geotextile used in verification simulations (after Krisdani et al. 2008) Saturated Water Content θ s Residual Water Content θ r α (1/kPa) n Saturated Hydraulic Conductivity (m/s) Fine sand Filter geotextile Geonet

83 (a) Figure 3-1. (a) Moisture characteristic curves (b) unsaturated hydraulic conductivity curves (b) 68

84 Figure 3-2. Comparison of the measured experimental pore pressures by Krisdani et al. (2008) and the computed pore pressures (numerical) in this paper along the column during drainage 69

85 Figure 3-3. Schematic diagram of the profile used for model verification (after Krisdani et al., 2008) 70

86 Figure 3-4. Schematic diagram of the profile used in one layer geotextile simulations 71

87 (a) (b) Figure 3-5. Pore pressure contours after 10 hrs infiltration (a) profile with geotextile (b)profile without geotextile 72

88 Figure 3-6. a) Pore pressure distribution along the centerline of the profile: stage 1 b) Magnified view of pore water pressure profiles in crushed stone layer 73

89 Figure 3-7. a) Pore pressure distribution along the centerline of the profile: stage 2 b) Magnified view of pore water pressure profiles in crushed stone layer 74

90 Figure 3-8. MCC and hydraulic conductivity curves for DMBLs with different AEVs 75

91 Height Above Base of Profile (cm) Initial condition 4hr - AEV = 0.3 kpa 8hr - AEV = 0.3 kpa 12hr - AEV = 0.3 kpa 24hr - AEV = 0.3 kpa 4hr - AEV = 0.23 kpa 8hr - AEV = 0.23 kpa 12hr - AEV = 0.23 kpa 24hr - AEV = 0.23 kpa 4hr- AEV =0.19 kpa 8hr- AEV =0.19 kpa 12hr- AEV =0.19 kpa 24hr- AEV =0.19 kpa Pore Water Pressure (kpa) a b Figure 3-9. a) Pore pressure distributions for DMBLs with different AEVs b) Magnified view of pore pressure profiles in crushed stone layer 76

92 Figure Pore pressure contours (kpa) after (a) 3 hr; and (b) 6 hr rainfall 77

93 Figure Pore pressure contours (kpa) after 6 hr rainfall, geotextile with saturated hydraulic conductivity increased by a factor of two. 78

94 Figure Schematic diagram of narrow profile 79

95 Height Above Base of Profile (cm) Height Above Base of Profile (cm) Height Above Base of Profile (cm) 200 3hrs hrs Initial condition Two DMBL One DMBL Without DM BL Pore Water Pressure (kpa) 50 Initial condition Two DMBL One DMBL Without DM BL Pore Water Pressure (kpa) hrs Figure Pore pressure distribution after 6, 12, and 24 hrs of rainfall 50 Initial condition Two DMBL One DMBL Without DM BL Pore Water Pressure (kpa) 80

96 Height Above base of Profile (cm) Height Above base of Profile (cm) Height Above base of Profile (cm) 200 6hrs hrs Initial condition AEV = 6.65 kpa AEV = 0.76 kpa AEV = 0.30 kpa AEV = 0.19 kpa Pore Water Pressure (kpa) Initial condition AEV = 6.65 kpa AEV = 0.76 kpa AEV = 0.30 kpa AEV = 0.19 kpa Pore Water Pressure (kpa) hrs Initial condition AEV = 6.65 kpa AEV = 0.76 kpa AEV = 0.30 kpa AEV = 0.19 kpa Pore Water Pressure (kpa) Figure Pore pressure distribution for different AEVs in the two-layer DMBL system 81

97 (a) Figure (a) MCCs and (b) hydraulic conductivity curves of geotextiles with different AEVs (b) 82

98 Figure Pore pressure distributions for different saturated hydraulic conductivity of the upper DMBL 83

99 (a) Figure Hydraulic conductivity curves for (a) different saturated hydraulic conductivity values; and (b) different AEVs (b) 84

100 CHAPTER 4 Effect of Geocomposite Drainage Layers on Moisture Distribution and Plastic Deformation of Paved and Unpaved Road Sections This chapter will be submitted for publication in the International Journal of Geomechanics. The authors are: Mahdi Bahador, Mohammed A. Gabr, and T. Matthew Evans ABSTRACT: The effect of geosynthetic layers on moisture distribution and plastic deformation of paved and unpaved road sections is studied using computer programs SIGMA/W and FLAC. The geosynthetic layers consisted of a geonet sandwiched between a nonwoven geotextile at the bottom and a transport layer at the top. Two types of geotextile were modeled as the transport layer: woven fiberglass and nonwoven polypropylene geotextile. Inclusion of the geosynthetic layers at the interface of aggregate base course and subgrade increased suction in subgrade and decreased the suction in aggregate base course during rainfall. Woven fiberglass geotextile caused higher suction in aggregate base course compared to nonwoven polypropylene geotextile. The geosynthetic layers decreased the plastic deformation in both paved and unpaved road sections through reinforcement and hydraulic effects. Increasing the thickness of the asphalt and aggregate base course decreased the reinforcement effect of the geotextile while increased its hydraulic effect. In low volume 85

101 road sections with thinner asphalt layer, using the woven fiberglass as the transport layer decreased the plastic deformation of the profile by up to 20% more than the profile with the nonwoven polypropylene geotextile and increasing the thickness of the asphalt layer reduced this difference to approximately 4%. In unpaved road sections, the woven fiberglass decreased the plastic deformation of the profile by approximately 24% more than the profile with nonwoven polypropylene geotextile regardless of the aggregate base course thickness. Author keywords: geocomposite, pavement, reinforcement, suction 4.1 Introduction Water is the main cause of deterioration in pavement sections (Cedergren, 1994). Excess moisture in pavement sections can decrease the pavement life by more than half (Christopher and McGuffey, 1997). Excess moisture in a pavement section can cause issues such as breaking the cementation bond in stabilized aggregate base course (ABC), separating asphalt from aggregates (stripping), water bleeding and pumping, shrinking, swelling, and frost heave of subgrade. Water can enter the pavement section through rainfall or upward water flow due to capillary forces. An interlayer drainage system can be used to address moisture changes of soils within pavement sections due to these two sources (Henry 1988; Stormont and Zhou, 2001; Stormont et al., 2009). Conventional drainage systems consist of a permeable aggregate base course under the asphalt layer and an edge drainage to divert the water toward drainage pipes. This drainage system, however, assumes saturated flow and is not always effective in maintaining the water content of the unbound pavement layers (Henkel, 1997). In order to design an appropriate drainage system, water flow under 86

102 unsaturated conditions and the unsaturated properties of the geomaterials should be considered. The state-of-the art drainage design in pavement sections is using geosynthetic layers at the interface of the aggregate base course (ABC) and subgrade and considering unsaturated conditions (Christopher et al., 2000; Henry and Stormont, 2000; Henry et al., 2002; Stormont et al., 2009). Since geomaterials are rarely saturated in pavement sections, the unsaturated properties of both geomaterials and geosynthetic layers should be considered in the drainage design. Specifically, these properties are the moisture characteristic curve (MCC) and inplane transmissivity. The MCC represents the relationship between the water content of a porous medium and the potential energy of that pore water. The MCC can be measured experimentally for soils (ASTM D6836, Standard Test Methods for Determination of the Soil Water Characteristic Curve for Desorption Using Hanging Column, Pressure Extractor, Chilled Mirror Hydrometer, or Centrifuge) and geotextiles (Stormont et al., 1997; Lafleuer, 2000; Knight and Kotha, 2001; Stormont et al., 2001; Kuhn et al., 2005; Garcia et al., 2007; Nahlaw et al., 2007). The MCC of the geotextiles is steeper than most of the geomaterials and they have lower air entry value (AEV) and water entry value (WEV) compared to finegrained soils (Stormont et al., 1997; Stormont et al., 2001; Iryo and Rowe, 2003). These properties allow geotextiles to function as drainage/moisture barrier layers when in contact with partially saturated soils in pavement sections (e.g. Henry, 1988; Henry and Holtz, 2001; Stockton, 2001; Krisdani et al., 2006; Stormont et al., 2009). Henry and Stormont (2000) proposed a drainage/moisture barrier system in pavement sections using geosynthetic layers 87

103 termed the geocomposite capillary barrier drain (GCBD). In their design, the drainage system consisted of a geonet as a capillary barrier sandwiched between two geotextiles. The top geotextile functioned as a transport layer. Fiberglass geotextiles generally have a higher AEV and become transmissive at higher suctions relative to other types of polymeric nonwoven geotextiles, and are recommended for use as the transport layer in the GCBD (Stormont and Ramos, 2004). In the GCBD system, the fiberglass layer diverts water from the ABC toward the edge drain and the geonet functions as a capillary barrier for downward water flow into the subgrade under unsaturated conditions. Stormont and Zhou (2001) modeled a pavement section with a GCBD at the interface of ABC and subgrade using computer program VS2DHI and showed that the GCBD prevents downward water flow into the subgrade during rainfall. Previous experimental studies on the effect of water content on mechanical behavior of ABC and subgrade materials have been reported in the literature. In general, increasing water content decreases the shear strength of geomaterials (e.g., Fredlund and Rahardjo 1993; Lu and Likos 2004) and such a reduction in strength should be considered in pavement design. The Mechanistic-Empirical Pavement Design Guide (MEPDG) uses the resilient modulus of the subgrade as the fundamental parameter for pavement design. It has been reported that increasing water content results in a considerable reduction in the resilient modulus of the ABC and subgrade (Sweere 1990; Dawson et al. 1996; Drumm 1997; Gehling 1998; Witczak 2000; Kolisoja et al. 2002). Rada and Witczal (1981) concluded that the resilient modulus of granular materials under saturated conditions can be decreased to one third one of the value 88

104 measured at lower water contents. Tian et al. (1998) measured the resilient modulus of a coarse granular material under dry of optimum and wet of optimum water contents and showed that the specimen compacted wet of optimum had a 20% lower resilient modulus compared to that at optimum water content. Thus, in order to maintain the strength of geomaterials in a pavement section, the drainage system should function to not only prevent the development of positive pore pressure, but also to mitigate moisture increases in unsaturated ABC and subgrade layers. Conversely, the effect of drainage geosynthetics on plastic deformation in ABC and subgrade layers has not been extensively studied. In this paper, paved and unpaved road sections with different asphalt and ABC thicknesses are simulated using SIGMA/W (Krahn, 2004) and FLAC (ITASCA consulting group, 2008) to study the effect of a three layers drainage system on hydraulic and mechanical behavior of the pavement section. The drainage system consists of a geonet sandwiched between a nonwoven geotextile at the bottom and a transport geotextile at the top. Two geotextile types are considered for the transport layer in the drainage system: nonwoven polypropylene (NWP) geotextile and woven fiberglass (WF) geotextile. The moisture distribution in the pavement section with different ABC thicknesses are studied during rainfall using SIGMA/W. The moisture distributions are then used in FLAC to investigate the effect of asphalt and ABC thicknesses on the functionality of the drainage system in reducing the total plastic deformation of the pavement section. The reinforcement and hydraulic effect of the NWP and WF geotextiles on section deformation 89

105 response is separated and the effects of asphalt and ABC thicknesses on each of these two components are studied. 4.2 Effect of Suction on Soil Strength In the mechanistic-empirical pavement design guide (MEPDG), the resilient moduli of geomaterials are a function of stress state and degree of saturation (NCHRP, 2000). In this model, resilient modulus increases with increasing the stress state and decreases with increasing the degree of saturation. In this paper, FLAC was used to perform the stressdeformation analysis. Subgrade was modeled using Cam-Clay constitutive model since it can capture the effect of soil moisture content and stress state on soil strength. Furthermore, Cam-Clay considers elastoplastic behavior, making it an appropriate model for simulating geomaterial behavior. In the Cam-Clay model, bulk modulus is a function of mean effective stress through the following equation: (4-1) where K is bulk modulus, is specific volume defined as the ratio of total volume to volume of solids, P is mean effective stress, and κ is the slope of the overconsolidation line in space. Effective stress is calculated as follows: ` (4-2) where is effective stress, S is degree of saturation, u a is pore air pressure, and u w is pore water pressure. 90

106 Degree of saturation may be calculated at any suction using (e.g.) the van Genuchten equation (van Genuchten, 1980): (4-3) where: S is degree of saturation, h is suction head, is residual water saturation, and α and n are fitting parameters. Accordingly, in the Cam-Clay model the effects of stress state and suction on soil strength are considered using equations 4-1 to 4-3. Cam-Clay model parameters are rarely reported in the literature for coarse-grained geomaterials due to the required high isotropic pressures in the tests (Atkinson, and Bransby, 1978). In this paper, the ABC was modeled as a Mohr-Coulomb material with suction and stress state effects on strength. In the Mohr-Coulomb model, decreasing the elastic modulus decreases plastic deformations and misrepresents the effect of moisture content on soil strength. Figure 4-1 shows the unloading-reloading stress-deformation curves for a numerically modeled displacement-controlled triaxial test on a 20 cm tall specimen with 5 cm radius and three different elastic moduli of 10.0, 9.0, and 8.0 MPa. The specimen was first isotropically confined under a cell pressure of 0.1 MPa and then a vertical displacement of 5 mm was specified at the top of the specimen. As is evident in Figure 4-1, decreasing the elastic modulus of the specimen decreases the slope of the linear elastic portion of the curve and consequently, increases the elastic deformation and decreases the plastic deformation. In order to avoid this misrepresentation, elastic modulus of the ABC material is not changed 91

107 during the stress-deformation analysis herein. Instead, the effective stresses are computed using equation 4-2. Considering suction in effective stress calculations results in increase in effective principal stresses and moves the Mohr s circles away from the origin and the failure envelope and consequently considers the effects of suction and stress state on soil strength. 4.3 Modeling Plastic Deformations Plastic deformations are of particular interest when modeling pavement sections because they are permanent. To calculate the plastic deformation in MEPDG, elastic strains are first calculated using a linear elastic analysis and then plastic strains are calculated as a function of the calculated elastic strain, water content, and number of load cycles. In the numerical model, the vertical elastic strain increments at each step in all the elements below the centerline of the load were calculated as follows: (4-4) where is the vertical elastic strain increment,,, and are stress increments, G is shear modulus, and K is bulk modulus. The elastic deformation in each element below the centerline of the load was calculated by multiplying the thickness of each element by its total elastic strain. The total elastic deformation in the ABC was computed by summing the calculated elastic deformations in each of these elements. The plastic deformation in the ABC was then calculated by subtracting the elastic deformation from the total deformation of the ABC as computed from FLAC. 92

108 In the Cam-Clay model, bulk modulus is a function of mean effective stress and changes at each step. The total and plastic volume change is computed in FLAC at each step. To compute the plastic deformations in the subgrade, the elastic volumetric strain in each element was calculated by subtracting the plastic volumetric strain from total volumetric strain, and the vertical elastic strain increment was calculated as follows in each step using the elasticity theory and axisymmetric loading condition: (4-5) where is vertical elastic strain increment, is vertical stress increment, is elastic volumetric strain increment, is bulk modulus at each step, and G is shear modulus. The same approach used in plastic deformation calculations in the ABC was used to find the plastic deformation in the subgrade by subtracting the total elastic deformation from the total deformation in subgrade. To verify the formulation described above, results from a simulated displacement-controlled triaxial test were used. The triaxial test specimen was modeled first as a Mohr-Coulomb material and then as a Cam-Clay material, and the plastic deformation in each case was calculated using the unloading-reloading curves. Then, the same triaxial test was simulated and the aforementioned approach was used to calculate the plastic deformations. The difference between the plastic deformations obtained from each of the two methods was less than 0.05% for both Cam-Clay and Mohr-Coulomb models. 93

109 4.4 Material Properties Hydraulic properties The paved road section was modeled as a four-layer profile: asphalt, ABC, geocomposite, and subgrade. The hydraulic properties of these materials are shown in Table 4-1. Permeability of asphalt depends on the initial in-place air void content of the mixture. It has been shown that the initial in-place air void content and consequently, the permeability of asphalt is a function of nominal maximum aggregate size (NMAS), density, and lift thickness (Choubane et al., 1998; Musselman et al., 1998; Mallick et al, 1999; Cooley and Brown, 2000; Maupin, 2000; Cooley and Brown, 2001). Musselman et al. (1998) suggested that the ratio of lift thickness to NMAS should be 4.0. The permeability of asphalt becomes an issue especially in coarse-graded superpave mixes due to water permeation into the void structure of the pavement. Cooley Jr, et al. (2002) conducted a series of experimental and field permeability tests on 23 superpave pavement construction projects with different NMAS and lift thicknesses. The permeability of the superpave asphalt with NMAS of 25 mm and lift thickness to NMAS ratio of 4.0, as was measured by Cooley Jr et al (2002), is used herein in the seepage simulations. The geocomposite drainage layer was modeled as a three-layer system: a geonet sandwiched between a nonwoven geotextile at the bottom and a transport geotextile at the top. Two types of transport geotextile were considered: nonwoven polypropylene (NWP) and woven fiberglass (WF) geotextile. The hydraulic properties of the WF geotextile and geonet considered for the geocomposite drainage system were measured by Ramos (2001) and the 94

110 hydraulic properties of the NWP geotextile were measured by Stormont and Morris (2000). Fiberglass is a hydrophilic (low contact angle), polymeric material which becomes transmissive at higher suctions relative to other types of polymeric geotextiles (Stormont and Ramos, 2004). Hydraulic properties of crushed stone and silty sand were used for ABC and subgrade, respectively (Henry et al, 2001; EICM, 2006). The MCCs and hydraulic conductivity curves for all materials used in this study are shown in Figure Mechanical properties The elastic strength properties of the materials used in this study are presented in Table 4-2.The asphalt was modeled as a linear elastic material with properties presented by Brunton et al. (1992) and Pease (2010). Subgrade and ABC were modeled using Cam-clay and Mohr- Coulomb constitutive models, respectively and their model properties are presented in Table 4-3 (Desai and Siriwardane, 1984). The elastic modulus and Poisson ratio of crushed stone and silty sand along with friction and dilation angle of crushed stone were obtained using model calibration with MEPDG. The model calibration is presented next. NWP and WF geotextiles were modeled as linear elastic materials. The typical elastic modulus for nonwoven needle-punched geotextiles was reported by Bergado et al. (2001). They measured the elastic modulus of three different nonwoven geotextiles under axisymmetric loading and showed that when the tensile load (N/m) is divided by the thickness of the geotextile, the stress-strain curves of all three geotextiles fall into a narrow band and thus, have similar elastic moduli. 95

111 The elastic modulus of the WF geotextile was measured as a part of this study since no such information was found in the literature. A special clamping adapter for a standard load frame was necessary for the WF geotextile due to its high strength and gripping problems (ASTM D ). The two ends of specimens 2 cm wide and 13 cm long were prepared and glued to a metal wedge clamp using a special resin to prevent stress concentration and slipping in the clamp (Figure 4-3(a)). Strains were measured using an extensometer with inch accuracy (Figure 4-3(b)). Figure 4-3(c) shows that the WF geotextile failed in the middle of the specimen with no slippage in the clamps. The stress-strain curves from two tests are shown in Figure 4-4.The slopes of the initial linear portion of the curves for the two specimens were averaged and used as the elastic modulus for the material (Table 4-2). 4.5 Modeling Approach The seepage analysis was performed using SIGMA/W. Then, the pore water pressure distribution at steady state was then used in FLAC as the initial hydraulic condition to perform the stress-deformation analysis. SIGMA/W was not used for stress-deformation analysis because it computes the effective stress as the subtraction of pore water pressure from total stress, which overestimates the effective stress in high negative pore water pressures. Indeed, the balance between suction and wetted area within the matrix must be considered when calculating effective stress in unsaturated soils (Vanapalli, et al. 1996; Lu and Likos, 2004). FLAC, however, uses equation 4-2 to capture suction stress in a more robust manner. 96

112 FLAC was not used for a seepage (or coupled hydromechanical) analysis due to unreasonably the long simulation times. The time step in transient seepage analysis is constant in FLAC and is estimated as follows: (4-6) where is time step, L z is the smallest zone size in the simulation, K w is bulk modulus of water, K g is bulk modulus of air, k w is water saturated conductivity, and k a is air saturated conductivity. A high saturated hydraulic conductivity ( m/s) coupled with the thinness of the geotextile layer (3.2 mm) resulted in a small time step (~10-10 sec) and long simulation times (on the order of weeks). However, in SIGMA/W, the backward difference scheme is used for time integration. As a result, the solution is stable for any time step size (Smith, 1971). To avoid the numerical oscillation, the minimum time step is estimated as follows: (4-7) where t min is minimum time step, l is element size, E is modulus of elasticity, ν is poison s ratio, k is hydraulic conductivity, and γ w is unit weight of water. The time step is then automatically varied from s to 1 s based on the results of the previous step using an adaptive time-stepping routine, thus significantly reducing overall simulation time. 97

113 4.6 Model Verification The models were first verified against published results and then used for further simulations. The pore pressure data measured by Krisdani et al. (2008) in a soil-geotextile column were used for SIGMA/W model verification. Krisdani et al. (2008) performed soil-geotextile column tests on a one meter fine sand column in which a Polyfelt Megadrain 2040 geosynthetic was embedded 0.51 m from the top of the column. Polyfelt Megadrain 2040 consisted of a geonet sandwiched between two filter geotextiles. Drainage tests were performed by lowering the water table from the top of the column to the bottom. The pressure head at the bottom and along the column were measured during the test. The measured pore water pressures along the column are shown as symbols in Figure 4-5 (Krisdani et al., 2008). The schematic diagram of the profile used in SIGMA/W to simulate the test configuration is presented in Figure 4-6. Boundary conditions were a no-flow boundary on the sides and the top of the profile, and a time dependent pressure head at the bottom based on the measured data. The hydraulic properties of the fine sand and geotextile layer used in the simulation were provided in Krisdani et al. (2008) and are shown in Table 4-4. The geonet was modeled as a layer with low air entry value (0.2 kpa). Additional test details can be found in Krisdani et al. (2008). The results of the numerical simulations are shown in Figure 4-5 along with the measured experimental data. Comparative results shown in Figure 4-5 indicate good agreement between numerical computations and experimental measurements, and serve to validate the applicability of the seepage model. 98

114 In order to calibrate the deformation model used herein, MEPDG software was used as base line for comparison (MEPDG, 2009). A profile 100 cm wide and consisting of 12.7 cm asphalt, 25.4 cm ABC, and 100 cm subgrade was modeled in FLAC under axisymmetric loading conditions. Hydraulic and mechanical properties presented in Table 4-1,Table 4-2Table 4-2, and Table 4-3 for asphalt, crushed stone, and silty sand were used and the water table was placed at the top of the profile. The elastic modulus and Poisson s ratio of the crushed stone and silty sand and the friction and dilation angles of the crushed stone were obtained through the calibration process. A 690 kpa stress was applied over 13.6 cm of the top left corner of the profile, simulating a tire load of 40 kn under axisymmetric conditions. The same calculation was performed with the MEPDG software using one pass of vehicle class number 5 (one single axle with a single tire) to represent the loading condition used in FLAC. After performing each calculation in MEPDG, the calculated elastic modulus for asphalt, ABC, and subgrade were used in the FLAC simulations. Multiple simulations were performed using both approaches with varying elastic moduli, Poisson s ratios, friction angles and dilation angles for the crushed stone to calibrate the plastic deformations calculated in FLAC. The appropriate material parameters are presented in Table 4-2 andtable 4-3Table 4-3. These parameters are consistent with the published values for silty sand and crushed stone (Brunton et al., 1992; Evdorides and Snaith, 1996; Budhu, 1999). Using the parameters in Table 4-2 and Table 4-3, the difference between plastic deformations computed in MEPDG and FLAC was less than 3%. The plastic deformations computed in FLAC and MEPDG for ABC and subgrade are presented in Table

115 4.7 Seepage Analysis Paved roads The schematic diagram of the profile used for seepage modeling is shown in Figure 4-7. A very fine mesh was necessary to represent the geotextile layer due to its relative thinness, leading to computational difficulties due to the very large conductivity matrices. Such difficulties in modeling thin layers of geotextile in a two dimensions domain have been reported in previous studies (Park and Fleming, 2006; Stormont, et al. 2009) as well. In order to be able to model the profile with the three thin layers of the geocomposite in two dimensions, a profile with 100 cm width was considered (Figure 4-7). Some of the elements in the geotextile layer were as small as 1.5 mm, such that three nodes were located vertically within the geotextile layer. Transition elements were used to increase the size of the mesh for the remaining model area. The geocomposite was placed at the interface of the ABC and 100 cm thick subgrade. The surface of the profile and the geocomposite layer were sloped 3%. Two types of transport geotextile were considered: woven fiberglass (WF) and nonwoven polypropylene (NWP) geotextile. The thicknesses of the geonet, WF, and NWP geotextile were selected as 5.9, 3.2, and 5.9 mm, respectively (Stormont and Morris, 2000; Ramos, 2001; Stormont et al., 2001). Various thicknesses were considered for the ABC and asphalt, representing low, medium, and high volume road sections, as presented in Table 4-6. Repetitions of 10 7, 10 6, and 10 5 E18KSAL loads were considered for high, medium, and low traffic volume, respectively (NCHRP, 2004). The minimum asphalt required for the ABC thicknesses were then determined using the traffic volume and the Asphalt Institute design 100

116 method (AI, 1982) for a mean annual air temperature of 15.5 o C, which covers a major part of the United States. These values are presented in Table 4-6. The hydraulic properties of the materials used in the simulations are presented in Table 4-1. The boundary conditions were a no-flow boundary on the left side, zero pressure head at the base, and a potential seepage boundary condition on the right side. The potential seepage boundary condition allows drainage of moisture from the boundary under saturated conditions. This simulates drainage of moisture from the geocomposite into the edge pipe which is assumed saturated for simplicity. A constant flux of m/s (19 mm/hr) was applied on the top of the profile simulating rainfall condition. This flux corresponds to a 6-hr rainfall with a 50 year return period for Greensboro, NC (McDowell and Borchers 2007). The initial conditions were a hydrostatic pore water pressure distribution throughout the profile. A volumetric water content equal to the air-void content of the superpave asphalt and a constant permeability corresponding to the field measurements were used for the asphalt layer and thus, the thickness of the asphalt layer did not affect the pore pressure distributions in ABC and subgrade. This condition simulates a saturated asphalt layer. Figure 4-8(a) through Figure 4-8(c) show pore pressure contours after 6 hours of simulated rainfall for a paved road profile without a geocomposite layer, and profiles with NWP and WF geotextile as the transport layer, respectively. The thickness of the ABC layer in Figure 4-8 is 25.4 cm. In all cases, steady state was reached during the 6 hours rainfall and the amount of water flux entering the system was equal to the water leaving the system (~0.019 m 3 /hr). As can be seen from Figure 4-8, the inclusion of the geocomposite at the interface of 101

117 the subgrade and ABC increased the suction in subgrade by up to 8.0 kpa. The hydraulic conductivity of the geotextile transport layer is very low under suction levels higher than its water entry value (WEV). When the wetting front reaches the geotextile transport layer, water accumulates above the geotextile and decreases the suction in the ABC as shown in Figure 4-8(b) and Figure 4-8(c). When suction decreases to the WEV of the geotextile, water enters the geotextile and is diverted toward the edge drain. The WF geotextile has a higher WEV and becomes transmissive at higher suctions relative to other types of polymeric geotextiles (Stormont and Ramos, 2004). Consequently, the WF geotextile drains more water from the ABC, resulting in a higher suction in the ABC compared to the NWP geotextile. This is consistent with findings from previous studies (Stormont and Zhou, 2001; Stormont et al, 2009) Unpaved roads The unpaved road section was modeled similarly to the paved road, but excluding the asphalt layer. Three different ABC thicknesses of 50.8 cm, 63.5 cm, and 68.6 cm were selected, corresponding to E18KSAL repetitions of 10 4, , and 10 3 cycles. These were computed using the standard equation for unpaved roads (Giroud and Noiray, 1981): (4-8) where h 0 is ABC thickness and N is the number of E18KSAL repetitions. Similar to the paved road section, the seepage simulations were performed on the unpaved road profiles for the three ABC thicknesses with and without a geocomposite layer. To facilitate a better comparison, the pore pressure distributions along the left side of the profile 102

118 for the three cases (without geocomposite, with WF transport layer, and with NWP geotextile transport layer) and the three ABC thicknesses are plotted in Figure 4-9 and Figure 4-10 for paved and unpaved road sections, respectively. In all the cases the geocomposite layer increases the suction in the subgrade and decreases the suction in the ABC compared to the profile without a geocomposite layer. Also, the geocomposite with the WF geotextile results in higher suctions in the ABC compared to the NWP geotextile. 4.8 Stress-deformation Analysis The geocomposite layer affects the pore pressure distributions in the ABC and subgrade (Figure 4-9 and Figure 4-10) and consequently changes the effective stress and shear strength (equations 4-1 and 4-2). The geocomposite layer also functions as a reinforcement layer and affects the stress distribution and deformations of the ABC and subgrade. In this paper, the latter effect is termed hydraulic effect and the former one is called mechanical effect of the geocomposite layer. In order to study the effect of asphalt and ABC thicknesses on the contribution of each of these two components to reduce the plastic deformation, stressdeformation analyses were performed Paved roads A schematic diagram of the profile used in stress-deformation analysis is shown in Figure 4-7. The left and right boundaries of the profile were fixed in the lateral direction and the bottom boundary was fixed in both directions to simulate axisymmetric loading. The profile was allowed to equilibrate under gravity to simulate initial in-situ stresses and then a stress of 103

119 690 kpa was applied over 13.6 cm of the top left corner of the profile, simulating an axisymmetric tire load of 40 kn. The computed deformations were relatively small (~0.01 mm) since only one load cycle was modeled. In order to investigate whether the displacement at the first load cycle can be used as an indication of the deformation trend for higher number of load cycles, a saturated profile and a dry profile were modeled using MEPDG formulations. The thickness of the ABC and asphalt layers were 25.4 and 12.7 cm, respectively. The total plastic deformations under one load cycle in the saturated and dry profiles were and mm, respectively. The simulations were performed for 10 6 load cycles and the total deformations are shown in Figure Figure 4-11 indicates that the initial deformation difference of mm resulted in 7.4 mm difference in the total plastic deformation after 10 6 load cycles. In this paper, deformations were computed under one load cycle and the percent change in deformation was used rather than the absolute values to compare different cases and study the effect of asphalt and ABC thicknesses on the efficacy of the geocomposite in reducing the plastic deformation. A profile without geocomposite but with the pore pressure distribution of the profile with the geocomposite was modeled for each case as well, to separate the hydraulic and mechanical effects and to find the hydraulic effect of the geocomposite on plastic deformation. The mechanical effect was then calculated as the total effect minus the hydraulic effect. Figure 4-12Figure 4-12 shows the percent decrease in the total (ABC and subgrade) plastic deformation due to mechanical and hydraulic effect of the geocomposite layer compared to 104

120 the profile without a geocomposite for the three different ABC and asphalt thicknesses. Increasing the thickness of the asphalt layer decreases the stress level in the geocomposite layer and thus reduces its reinforcement effect. As is shown in Figure 4-9, the geocomposite layer decreases the suction in the ABC and increases the suction in the subgrade. A thicker asphalt layer decreases the stress level in the ABC and thus the negative hydraulic effect of the geocomposite in the ABC is mitigated. Consequently, a thicker asphalt layer results in an increased hydraulic and decreased mechanical effect from the geocomposite. Figure 4-12 also shows that the mechanical effect of the WF is more significant than that of the NWP geotextile for lower asphalt thicknesses due to the higher elastic modulus of the WF geotextile. With increasing asphalt thickness, the mechanical effect of the WF decreases and becomes similar to the NWP geotextile. Figure 4-13 shows the relative decrease in the total plastic deformation considering both the mechanical and hydraulic contributions of the geocomposite. Increasing the asphalt thickness decreases the total effect of the geocomposite on plastic deformations in the profile with the WF geotextile and increases the total effect in the profile with NWP geotextile. In the profile with NWP geotextile, if the asphalt thickness increases, the increase in the hydraulic effect is more than the decrease in mechanical effect and thus, the total effect of the NWP geotextile on plastic deformations increases. However, the decrease in the mechanical effect of the WF is higher than its hydraulic effect increase and the total effect of the geocomposite decreases with increasing the asphalt thickness. As a result, the effect of the geocomposite layers on the total plastic deformation for the NWP and WF geotextiles become similar with increasing asphalt thickness (Figure 4-13). 105

121 It is also apparent from Figure 4-13 that the geocomposite can decrease the total plastic deformation up to 55%. To investigate how much the ABC thickness can be reduced by inclusion of the geocomposite layer, plastic deformations in a profile without a geocomposite and with 45.7 cm ABC was compared to the profiles with the geocomposite and 15.2, 25.4, and 45.7 cm ABC. This comparison was performed on the profiles with the same asphalt thickness. Three asphalt thicknesses of 19.1, 26.9 and 38.1 cm were used. The plastic deformations of the profiles with geocomposite (d pg ) were divided by the one for the profile without geocomposite and with 45.7 cm ABC (d p45.7 ) and the results are shown in Figure For each of the asphalt thicknesses, the plastic deformation of the profiles with the geocomposite was lower than without the geocomposite (between 0.5 to 0.7). In the profile with the thinnest asphalt layer, WF had higher effect on plastic deformations and increasing the asphalt layer reduced this effect. However, NWP geotextile has higher effect on the plastic deformations in the profile with thicker asphalt Unpaved roads To simulate the unpaved road, a profile similar to Figure 4-7 and with excluding the asphalt layer was modeled. Three ABC thicknesses of 50.8 cm, 63.5 cm, and 68.6 cm corresponding to 10 3, , and 10 4 E18KSAL repetition were used (Giroud and Noiray, 1981). In unpaved road simulations, kpa (70 psi) stress was applied and a small cohesion of 3.5 kpa was chosen for ABC to overcome a convergence problem. The rest of the modeling details were similar to the paved road section. Figure 4-15 shows the hydraulic and mechanical effect of the geocomposite on total plastic deformations for all the three ABC thicknesses. Increasing the ABC thickness increases the hydraulic effect and decreases the 106

122 mechanical effect of the geocomposite. Figure 4-12 and Figure 4-15 also show that the geocomposite has higher mechanical effect and lower hydraulic effect in unpaved roads. The reason is that the asphalt layer in paved roads decreases the stress level in the underlying layers and thus, decreases the reinforcement effect and the negative hydraulic effect of the geocomposite. The elastic modulus of geosynthetics usually decreases with increasing the tensile strain. Bergado et al. (2001) showed that the elastic modulus of the nonwoven geotextiles decreases to 13.3% of its initial value (from MPa to MPa) in tensile strains higher than 2.5%. In unpaved road sections, the induced tensile strain at the top of the subgrade may reduce the elastic modulus and mechanical effect of the geocomposite layer. To consider this decrease in elastic modulus, the elastic modulus of the WF and NWP geotextiles was reduced to 13.3% of their initial value and and MPa, respectively. Figure 4-16 shows the total effect of the geocomposite on the plastic deformation with the initial and reduced elastic modulus of the geotextiles. Figure 4-15 shows that in unpaved road sections, WF geotextile maintains its mechanical effect in the profiles with thicker ABC and thus, its effect on the total plastic deformation does not decrease with increasing ABC thickness (Figure 4-16). The WF geotextile decreases the total plastic deformations by approximately 24% more than the profile with NWP geotextile regardless of the ABC thickness in the unpaved road sections. Figure 4-16 shows that decreasing the elastic modulus of the WF and NWP geotextiles decreases their effect on the total plastic deformations by 20 and 3.4%, respectively. Figure 4-15 shows that with total elimination of the mechanical effect, WF and 107

123 NWP geotextiles still decrease the total plastic deformations by the average of 31 and 28%, respectively. In order to determine how much of the ABC thickness can be reduced with the inclusion of the geocomposite, the ratio of the plastic deformation of the profile with geocomposite and different ABC thicknesses (d pg ) was divided by the one for the profile without geocomposite and with 68.6 cm ABC thickness (d p68.6 ). The results for both the initial and reduced elastic modulus of the WF and NWP geotextiles are shown in Figure Figure 4-17 shows that when the initial elastic modulus is used, the plastic deformation of all the profiles with WF and NWP geotextiles is lower than the one without geocomposite. However, when the elastic modulus is reduced, the plastic deformation of the profile with NWP geotextile and 50.8 cm ABC becomes more than the one without geocomposite and with 68.6 cm ABC. Simulations for a NWP geotextile with reduced elastic modulus and 58.4 cm (23 inch) ABC was also performed and the results are shown in Figure According to Figure 4-17, WF and NWP geotextiles with reduced elastic modulus can decrease the required ABC thickness from 68.9 cm to 50.8 cm and 58.4 cm, respectively. 4.9 Conclusion A comprehensive study on the effect of inclusion of geocomposite on the moisture distribution and plastic deformations of the profiles simulating unpaved and paved road sections are presented. In order to overcome the numerical difficulties in modeling the thin geocomposite layers in two dimensions domain, computer program FLAC was used for stress-deformation and SIGMA/W was used for seepage analysis. Simulations were 108

124 performed for different ABC and asphalt thicknesses and the built in programming language in FLAC was used to compute the plastic deformations in the profiles with and without geocomposite. The following conclusions can be drawn based on the results: 1- To study the effect of the suction on soil strength and plastic deformations using Mohr- Coulomb model, failure envelope should be changed instead of the elastic modulus. 2- Seepage analysis showed that during rainfall infiltration, geocomposite increases the suction in subgrade by up to 8 kpa and decreases the suction in ABC by up to 3.6 kpa. WF geotextile causes about 2 kpa less reduction in suction in ABC compared to NWP geotextile due to its higher WEV. 3- Stress deformation analysis showed that increasing the pavement thickness increases the hydraulic effect and decreases the mechanical effect of the geocomposite on the total plastic deformations in both the paved and unpaved road sections. 4- WF and NWP geotextiles can decrease the total plastic deformation by up to 55% and 35% in paved road and by up to 60% and 30% in unpaved road sections, respectively depending on the ABC and asphalt thicknesses. In low volume paved road sections, WF geotextile decreased the total plastic deformation by up to 20% more than NWP geotextile. Increasing the thickness of the asphalt layer decreased the difference between the effect of the WF and NWP geotextile on plastic deformation to less than 5%. In unpaved road sections, WF geotextile decreases the total plastic deformation by the average of 24% more than the NWP geotextile regardless of the ABC thickness. 109

125 5- In unpaved road sections, diminishing the elastic modulus of the WF and NWP geotextile by about one order of magnitude decreases their effect on the total plastic deformations by approximately 20% and 3.4%, respectively. 6- In unpaved road sections, geocomposite with reduced elastic modulus of WF and NWP geotextile can decreases the required thickness of the ABC by 26% and 15%, respectively References American Society for Testing and Materials (ASTM D ). (2008). Standard test methods for determination of the soil water characteristic curve for desorption using a hanging column, pressure extractor, chilled mirror hygrometer, and/or centrifuge. Vol American Society for Testing and Materials (ASTM D ). (2008). Standard test methods for tensile properties of geotextiles by the wide-width strip method. Asphalt institute (AI). (1982). Research and development of the asphalt institute s thickness design manual (MS-1), 9 th ed., Research report Atkinson, J.H and Bransby, P.L. (1978). The mechanics of soils: an introduction to critical state soil mechanics. The McGraw-Hill Companies, Inc. NY. USA. Bergado, D.T., Youwai, S., Hai, C.N., and Voottiipruex, P. (2001). Interaction of nonwoven needle-punched geotextiles under axisymmetric loading conditions. Geotextile and Geomembrane. 19, Brunton, J.M., Armitage, R.J., and Brown, S.F. (1992). Seven years experience of pavement evaluation. Proc., 7 th International Conference on Asphalt Pavements, 3, Budhu, M. (1999). Soil mechanics and foundations. Publisher. John Wiley and sons, Inc. New York, USA. Cedergren, H.R. (1994). America s pavements: world s longest bathtubs civil engineerin., American Society of Civil Engineering, 64(9), Christopher, B. R., Hayden, S. A., and Zhao, A. (2000.) Roadway base and subgrade geocomposite drainage layers. Proc., Testing and Performance of Geosynthetics in Subsurface Drainage, Seattle, Washington,

126 Christopher, B.R. and McGuffey, V.C. (1997). NCHRP Synthesis of Highway Practice 239: Pavement Subsurface Drainage Systems. TRB, National Research Council, Washington, D.C. Choubane, B., Page, G.C., and Musselman, J.A., (1998). Investigation of water permeability of coarse graded superpave pavements. Association of asphalt paving technologists, 67, P Coolet Jr, L.A., Prowell, B.D., and Brown, E.R. (2002). Issues pertaining to the permeability characteristics on coarse-graded superpave mixes. Final report. NCAT report Cooley Jr, L.A. and Brown, E.R., (2000). Selection and evaluation of a field permeability device for asphalt pavements. Transportation research record. 1723, Cooley Jr, L.A., Brown, E.R., and Maghsoodloo, S., (2001). Development of critical field permeability and pavement density values for coarse graded superpave pavements. Journal of the transportation research board, 1761, Dawson, A.R., Paute, J.L., and Thom, N.H., (1996). Mechanical characteristic of unbound granular materials as a function of condition. Flexible pavements., A. Gomes Correia editor Desai, C.S. and Siriwardane, H.J. (1984). Constitutive laws for engineering materials with emphasis on geologic materials. Prentice-hall, Inc., Englewood Cliffs, NJ. Drumm E.C., Reeves J.S., Madgett M.R. and Trolinger W.D. (1997). Subgrade resilient modulus correction for saturation effects. ASCE Journal of Geotechnical and Geoenvironmental Engineering, 123(7), Enhanced Integrated Climate Model (EICM)) version 3.2 Beta, (2006). Texas Transportation Institute, Texas A&M university. Evdorides, H.T. and Snaith, M.S. (1996). A knowledge based analysis process for road pavement condition assessment. Proc., The institute of civil engineers, 117(3), Garcia, E. F., (2007). Function of permeable geosynthetics in unsaturated embankments subjected to rainfall infiltration. Geosynthetics International, 14 (2), Gehling W.Y.Y., Ceratti J.A., Nunez W.P. and Rodrigues M.R. (1998). A study on the influence of suction on the resilient behavior of soils from southern Brazil. Proc of the Second Int. Conf. On Unsaturated Soils, Beijing, China, 1, Giroud, J. P., and Noiray, L. (1981). Geotextile-reinforced unpaved road design. Journal of the Geotechnical Engineering Division.,ACSE., 107(9),

127 Heckel, L. (1997). Performance problems of open-graded drainage layers under continuously reinforced concrete pavement in Illinois. In Transportation Research Record, 1596, TRB, National Research Council, Washington, D.C., Henry, K. S., (1988). Use of geotextiles to mitigate frost heave in soils. Proc., The 5th International Conference on Permafrost, Trondheim, Norway, Henry, K. S. and Holtz, R. D., (1997). Capillary rise of water in geotextiles. Proc., The 1997 International Symposium on Ground Freezing and Frost Action in Soils, Lulea, Sweden, Henry, K. S. and Stormont, J. C., (2000). Geocomposite capillary barriers drain. US Patent No. 6,162,653. Henry, K. S. and Stormont, J. C., (2002). Geocomposite capillary barrier drain for limiting moisture changes in pavement subgrades and base courses. NCHRP-IDEA 68 Final Report, Transportation Research Board. Henry, K.S., Stormont, J.C., Ramos, R.D. and Barna, L. (2001). Geocomposite capillary barrier drain for limiting moisture changes in pavement subgrades and bases. Report prepared for the IDEA program, Transportation research board, National research Council, Washington, D.C. Henry, K. S., Stormont, J. C., Barna, L. A., and Ramos, R. D., (2002). Geocomposite capillary barrier drain for unsaturated drainage of pavements. Proc., The Seventh International Conference on Geosynthetics, Nice, France, 3, Iryo, T. and Rowe, R. K., (2003). On the hydraulic behavior of unsaturated nonwoven geotextiles. Geotextiles and Geomembranes, 21 (6), ITASCA CONSULTING GROUP, INC., Minneapolis, MN. (2008). FLAC (Fast Lagrangian Analysis of Continua), Version 6.0. Knight, M. A. and Kotha, S.M., (2001). Measurement of geotextile-water characteristic curves using a controlled outflow capillary pressure cell. Geosynthetics International, 8 (3), Krahn, J. (2004). Stress and deformation modeling with SIGMA/W, an engineering methodology. GEO-SLOPE International Ltd., Calgary, Alberta. Krisdani, H., Rahardjo, H. & Leong, E. C., (2006). Experimental study of 1-D capillary barrier model using geosynthetic material as the coarse-grained layer. Proc., The 4 th International Conference on Unsaturated Soils, Carefree, Arizona,

128 Krisdani, H., Rahardjo, H., and Leong, E. C., (2008). Measurement of geotextile-water characteristic curve using capillary rise principle. Geosynthetics International, 15 (2), Kolisoja, P., Saarenketo, T., Peltoniemi, H., and Vuorimies, N. (2002). Laboratory testing of suction and deformation properties of base course aggregates. Transportation research record. 1787: Kuhn, J. A., McCartney, J. S., and Zornberg, J. G., (2005). Impinging flow over drainage layers including a geocomposite. Proc., The Sessions of the Geo-Frontiers 2005 Congress, Austin, Texas. Lafleur, J., Lebeau, M., Faure, Y. H., Savard, Y., and Kehila, Y. (2000) "Influence of Matric Suction on the Drainage Performance of Polyester Geotextiles." Proc., The 53 rd annual conference of Canadian geotechnical society, Montreal, Canada, 2, Lu, N and Likos, W.J. (2004). Unsaturated soil mechanics. John-Wiley, Hoboken, N.J. Miller, D.J., Nelson, J.D., (1993). Osmotic Suction as a Valid Stress State Variable in Unsaturated Soil Mechanics. Unsaturated Soils, Geotechnical Special Publication, 39, ASCE. pp Mallick, R.B., Cooley, L.A., and Teto, M., (1999). Evaluation of permeability of superpave mixes in Maine. Final report. Technical report ME-001 Maine department of transportation, Augusta, Maine. Maupin, G.W. and Jr., (2000). Asphalt permeability testing in Virginia. Transportation research record, McDowell, H. K and Borchers, M. (2007). Storm drainage design manual. North Carolina Erosion and Sediment Control Planning and Design. Storm Water Management Division. Mechanistic-Empirical Pavement design guide (MEPDG) Version (2009). Applied research associates transportation. Arizona State University. Musselman, J.A., Choubane, B., Page, G.C., and Upshaw, P.B., (1998). Superpave field implementation: Florida s early experience. Transportation research record. 1609, Nahlawi, H., Bouazza, A., and Kodikara, J., (2007). Characterisation of geotextiles water retention using a modified capillary pressure cell. Geotextiles and Geomembranes, 25 (3), National Cooperative Highway Research Program (NCHRP). (2000). Guide for mechanistic-empirical design of new and rehabilitated pavement structures. Appendix DD- 113

129 1: Resilient modulus as function of soil moisture- summary of predictive models. Submitted by ARA, Inc., ERES division. Champaign, Illinois. National Cooperative Highway Research Program (NCHRP). (2004). Guide for mechanistic-empirical design of new and rehabilitated pavement structures. Appendix GG- 1: Calibration of permanent deformation models for flexible pavements. Submitted by ARA, Inc., ERES division. Champaign, Illinois. Park, K. D. and Fleming, I. R., (2006). Evaluation of geosynthetic capillary barrier. Geotextiles and Geomembranes, 24 (1), Pease, R.E. (2010). Hydraulic properties of asphalt concrete. PhD. Thesis. The University of New Mexico. Albuquerque, New Mexico. Rada, G. and Witczak, M.W. (1981). Comprehensive evaluation of laboratory resilient moduli for granular material. Transportation research record. 810: Ramos, R. D., (2001). Performance of a fiberglass based geocomposite capillary barrier drain. M.Sc. thesis, University of New Mexico, Albuquerque, USA. Smith, G.D., (1971). Numerical solution of partial differential equations. Oxford University Press, Oxford. Stockton, T. B., (2001). Performance experimental testing of diversion lengths for a geocomposite unsaturated drainage system. M.Sc. thesis, University of New Mexico, Albuquerque, New Mexico. Stormont, J. C., Henry, K. S., and Evans, T. M., (1997). Water retention functions of four nonwoven polypropylene geotextiles. Geosynthetics International, 4 (6), Stormont, J. C., Henry, K. S., and Roberson, R., (2009). Geocomposite capillary barrier drain for limiting moisture changes in pavements: Product Application. Final Report, Contract No. NCHRP-113. Transportation Research Board. Stormont, J. C., and Morris, C. E., (2000). Characterization of unsaturated nonwoven geotextiles. Proc., Advances in Unsaturated Geotechnics, Geotechnical Special Publication, ASCE, Denver, CO, Stormont, J. C., and Ramos, R. D., (2004). Characterization of a fiberglass geotextile for unsaturated in-plane water transport. ASTM Geotechnical Testing Journal, 27 (2), Stormont, J. C., Ramos, R. D., and Henry, K. S. (2001). "Geocomposite capillary barrier drain system with fiberglass transport layer." Transportation Research Record,1772,

130 Stormont, J. C., Ray, C., and Evans, T. M., (2001). Transmissivity of a nonwoven polypropylene geotextile under suction. ASTM Geotechnical Testing Journal, 24 (2), Stormont, J., and Zhou, S., (2001), Improving pavement sub-surface drainage systems by considering unsaturated water flow, Final Report DTFH61, Federal Highway Administration Sweere, G.T.H. (1990). Unbound granular bases for roads. PhD thesis, University of Delft, Netherland. Tian, P., Zaman, M.M., and Laguros, J.G. (1998). Variation of resilient modulus of aggregate base and its influence on pavement performance. Journal of testing and evaluation. 26(4): Van Genuchten, M.Th., (1980). A closed-form equation for predicting the hydraulic conductivity of unsaturated soils. Soil Science Society of America Journal, 44 (5): Vanapalli, S.K., Fredlund, D.G., Pufahi, D.E., and Clifton, A.W. (1996). Model for the prediction of shear strength with respect to soil suction. Canadian geotexhnical journal Witczak M.W., Houston W.N. and Andrei D. (2000). Resilient modulus as function of soil moisture a study of the expected changes in resilient modulus of the unbound layers with changes in moisture for 10 LTPP sites. Development of the 2002 Guide for the Development of New and Rehabilitated Pavement Structures, NCHRP 1-37 A, Inter Team Technical Report (Seasonal 2). 115

131 Table 4-1. van Genuchten parameters and saturated hydraulic conductivity for materials used in the modeling. Saturated Residual Saturated References Water Water α Hydraulic n Content Content (1/kPa) Conductivity θ s θ r (m/s) (EICM, 2006) Silty sand (subgrade) Crushed (Henry et al. stone ) (ABC) WF (Stormont and geotextile Ramos, 2004) NWP (Stormont and geotextile Morris, 2000) Geonet (Ramos, 2001) Asphalt (Cooley Jr, et al. (2002) 116

132 Table 4-2. Elastic strength properties of materials used in the model for various layers (Brunton et al. 1992; Ramos, 2001; Bergado et al. 2001; EICM, 2006; Pease, 2010) Materials Elastic modulus (MPa) Poisson s ratio Unit weight (kn/m 3 ) Asphalt Crushed stone (ABC) WF geotextile NWP geotextile Silty sand (subgrade)

133 Table 4-3. Mohr-Coulomb and cam-clay material properties for ABC and subgrade (After Desai and Siriwardane, 1984) Materials Friction Dilation Slope Slope M Reference Specific Silty sand (subgrade) Crushed stone (ABC) angle (ϕ) * Calculated from M angle (d) of NCL (λ) of OCL (κ) pressure (P 0 ) (kpa) volume at p * (ν 0 ) 118

134 Table 4-4. Hydraulic properties of fine sand and geotextile used in verification simulations (after Krisdani et al. 2008) Materials Saturated Saturated Residual Water α Hydraulic Water Content Content n (1/kPa) Conductivity θ s θ r (m/s) Fine sand Filter geotextile Geonet

135 Table 4-5. Plastic deformations calculated in MEPDG and FLAC Plastic deformation (mm) Material MEPDG FLAC Subgrade ABC

136 Table 4-6. ABC and asphalt thickness used in paved road profile simulations Asphalt thickness (cm) E18KSAL repetitions ABC thickness (cm)

137 Figure 4-1. Unloading reloading stress-deformation curves of the simulated displacement-controlled triaxial test 122

138 Figure 4-2. Properties of study materials: (a) Moisture characteristic curves (b) unsaturated hydraulic conductivity curves 123

139 Figure 4-3. (a) Woven fiberglass geotextile specimens (b) extensometer (c) Woven fiberglass geotextile at failure 124

140 Figure 4-4. Stress strain curves of woven fiberglass geotextile 125

141 Figure 4-5. Comparison of the measured experimental pore pressures by Krisdani et al. (2008) and the computed pore pressures (numerical) in this paper along the column during drainage 126

142 Figure 4-6. Schematic diagram of the profile used for model verification (after Krisdani et al., 2008) 127

143 Figure 4-7. Schematic diagram of the profile used for seepage analyses 128

144 Figure 4-8. Pore pressure contours in the profile : (a) without geocomposite (b) with WF geotextile transport layer (c) with NWP transport layer 129