10.1.1.137

A Review of Water Movement in the Highway Environment:

Implications for Recycled Materials Use

By Defne S. Apul

Dr. Kevin Gardner Dr. Taylor Eighmy

Dr. Jean Benoit Dr. Larry Brannaka

Recycled Materials Resource Center Environmental Technology Building

University of New Hampshire Durham, NH 03824

February 2002

ii

Abstract Significant amounts of recycled materials are being used in structural components of highways

such as base, subbase, Portland cement and asphalt concrete, and embankments. Past experience with recycled materials in highways has not shown a risk with respect to environmental impact and human health. However, environmental regulatory agencies frequently must assess risk to human health and the environment when conducting beneficial use determinations, particularly for new recycled materials. Contaminant leaching and advective transport are believed to be the primary transport pathways and both of them depend on in situ moisture conditions. The purpose of this study was to describe the state of the knowledge about water movement in the highway environment so that this knowledge may be incorporated into (1) practical leaching and preliminary impact assessments and (2) fate/transport models for use in risk assessment.

Measurement techniques for moisture content, pore water pressure, and rainfall were discussed.

For measuring in situ water content, time domain reflectometry (TDR) is a growing and promising technique. However, there is not a universal agreement on a TDR calibration equation. Laboratory techniques for measuring saturated hydraulic conductivity center on constant and falling head permeameter tests that should be conducted at low hydraulic gradients, and in addition, laminar and horizontal flow conditions. Hydraulic conductivity data for asphalt concrete, Portland cement concrete (PCC), and base/subbase layers displayed wide scatter due to high variability in mixture designs and in measurement techniques. Factors effecting hydraulic conductivity are shape, size, and interconnectivity of air voids, and in PCC, the curing temperature and the type and extent of chemical reactions during hardening. Most of the evidence suggested that addition of recycled materials to PCC decreases the hydraulic conductivity after curing.

Water enters pavements despite efforts to prevent it, but the extent of pavement deterioration can be reduced by proper drainage and maintenance. The major water ingress routes are infiltration through the pavement surfaces (through cracks and joints) and shoulders, melting of ice during the freezing/thawing cycles, capillary action, and seasonal changes in the water table. Most important water pathways that are discussed in greater detail in the literature are infiltration through cracks, joints and shoulders and drainage through edge drains. Groundwater level affects moisture conditions in the pavement and the subgrade if it is within approximately six meters from the surface. There is also literature on water pumping, temperature variations, geotechnical properties of base/subbase and subgrade layers as they relate to the water movement in the highway environment.

Simplistic and more comprehensive approaches to modeling water movement in the highway

environment were reviewed. Among all models presented, none closely matched the purpose of the present study. The IMPACT model, HWIR model, and the Integrated Model of the Climatic Effects on Pavements may be used as supplemental programs once a more sophisticated model is selected. Commercially available finite element models capable of simulating unsaturated water flow and contaminant transport in heterogeneous media in at least two dimensions is required.

iii

Table of Contents

1. Introduction .................................................................................................................................... 1 1.1 Recycled Materials Use in the Highway Environment .................................................. 1 1.2 Pavement Structure......................................................................................................... 4 1.3 Incentives for Studying Water Movement in Pavements............................................... 6 1.2 Objectives and Scope of the Report ............................................................................... 8 Chapter 1 Synopsis..................................................................................................................... 8

2. Drainage Systems ........................................................................................................................... 9 2.1 Routes of Water Ingress and Egress............................................................................... 9 2.2 Significance of Drainage .............................................................................................. 11 2.3 Engineered Systems for Drainage ................................................................................ 12 2.4 Drainage Efficiency...................................................................................................... 16 Chapter 2 Synopsis................................................................................................................... 19

3. Water Flow Theory ...................................................................................................................... 20 3.1 Darcy’s Law for Saturated Water Movement .............................................................. 20 3.2 Soil Moisture Retention Function ................................................................................ 22 3.3 Unsaturated Hydraulic Conductivity............................................................................ 23 3.4 Infiltration..................................................................................................................... 25

4. Measurement Techniques for Water Movement in Pavements.................................................... 27 4.1 Moisture Content Measurements................................................................................. 27 4.2 Hydraulic Conductivity Measurements........................................................................ 32 4.3 Pore Water Pressure ..................................................................................................... 35 4.4 Rainfall ......................................................................................................................... 37 Chapter 4 Synopsis................................................................................................................... 37

5. Moisture Content in Pavements ................................................................................................... 38 5.1 Groundwater Table Effect ............................................................................................ 38 5.2 Temporal and Spatial Variability ................................................................................. 40 5.3 Field Observations........................................................................................................ 42 Chapter 5 Synopsis................................................................................................................... 44

6. Hydraulic Conductivity of Asphalt Concrete............................................................................... 45 Chapter 6 Synopsis................................................................................................................... 50

7.Hydraulic Conductivity of PCC .................................................................................................... 51 Chapter 7 Synopsis................................................................................................................... 52

8.Hydraulic conductivity of Bases / Subbases / Embankments ....................................................... 53 Chapter 8 Synopsis................................................................................................................... 59

9. Factors Affecting Water Flow in the Highway Environment ...................................................... 60 9.1 Pumping........................................................................................................................ 60 9.2 Infiltration Through Cracks.......................................................................................... 60 9.3 Temperature.................................................................................................................. 67 9.4 Soil Mechanics ............................................................................................................. 69 Chapter 9 Synopsis................................................................................................................... 69

10.Computer Models ........................................................................................................................ 70 10.1 Modeling Approaches .................................................................................................. 70 10.2 Examples of Simplified Pavement Water Movement Models ..................................... 70

iv

Simulation of Subsurface Drainage of Pavements ........................................................... 70 Subbase Moisture Conditions........................................................................................... 72 Water Movement in PCC ................................................................................................. 73

10.3 Examples of Comprehensive Pavement Water Movement Models............................. 73 Water Flow Modeling in Unsaturated Base and Subbase Materials ............................... 73 Base and Subgrade Moisture Regimes............................................................................. 75 Water Migration Model in Cracked Concrete.................................................................. 77 Integrated Model of the Climatic Effects on Pavements.................................................. 78 Relevant Modules of Hazardous Waste Identification Rule (HWIR).............................. 79 Recycled Materials Fate and Transport Model (IMPACT).............................................. 80 PURDRAIN Model .......................................................................................................... 83

10.4 Need for Commercially Available Models and Model Selection .................................... 83 Chapter 10 Synopsis................................................................................................................. 85

11. Summary and Conclusions......................................................................................................... 87 12. Research Needs .......................................................................................................................... 89

Appendix ...................................................................................................................................... 90 A. Recommended Hydraulic Conductivity Values ...................................................................... 90 B. ASTM and AASHTO Standards Cited in This Report............................................................ 90 C. Standard Specifications for Materials for Soil-Aggregate Subbase, Base, and Surface Courses (ASTM D1241-68) ....................................................................................................................... 91 D. Terminology ............................................................................................................................ 91 References .................................................................................................................................... 93

v

Index of Tables Table 1.1 Annual production and use of recycled materials in roads in the United States (adapted from Collins and Ciesielski, 1994; Schroeder, 1994; Chesner et al., 1998). ......................................... 2 Table 1.2 Recycled materials used in various highway applications (adapted from Chesner et al., 1998)....................................................................................................................................................... 3 Table 1.3 Pollutant constituents in runoff from conventional road surfaces (adapted from Ball et al., 1998)....................................................................................................................................................... 7 Table 2.1 Routes of ingress and egress (adapted from Van Sambeek, 1989 and Dawson, 1998) ....... 10 Table 2.2 Pavement drainage system components............................................................................... 14 Table 2.3 Drainage performance of Indiana pavements (Feng et al., 1999) ........................................ 17 Table 2.4 Drainage data of pavements for the first two hours (Hagen and Cochran, 1996)................ 19 Table 2.5 Drainage data of pavements in Indiana (Ahmed et al., 1997).............................................. 19 Table 3.1 Empirical equations simulating the soil-water characteristic curve..................................... 24 Table 4.1 Comparison of moisture content measurement techniques (Stephens, 1996; Klute, 1986; ASTM, 2000, AASHTO, 1993). .......................................................................................................... 29 Table 4.2 TDR Equations..................................................................................................................... 31 Table 4.3 Methods for measuring pore water pressure. ....................................................................... 36 Table 5.1 Moisture content data in the field......................................................................................... 43 Table 6.1 Hydraulic conductivity values of asphalt concrete ............................................................. 50 Table 7.1 Hydraulic conductivity values of PCC................................................................................. 52 Table 8.1 Hydraulic conductivity of base, subbase, and subgrade layers ............................................ 55 Table 8.2 Hydraulic conductivity correlations for base layers (Elsayed and Lindly, 1996; Lindly and Elsayed, 1995). ..................................................................................................................................... 58 Table 8.3 Hydraulic conductivity of embankments ............................................................................. 59 Table 9.1 Cracking types and causes (Roberson, 2001)....................................................................... 61 Table 9.2 Percent of runoff entering surface cracks............................................................................. 61 Table 9.3 Hydraulic conductivity of seals as measured using a falling head test (Button, 1996)........ 62 Table 9.4 Infiltration rates. ................................................................................................................... 65 Table 10.1 Properties and parameter for flow, temperature and deformation analysis (adapted from Alonso, 1998))...................................................................................................................................... 76 Table 10.2 Rainfall factors for design of highway subbase drainage recommended by Cedergren (1974) ................................................................................................................................................... 82 Table 10.3 Comparison of commercially available unsaturated zone groundwater flow and contaminant transport models .............................................................................................................. 86 Table C.1 Gradation requirement for soil-aggregate materials............................................................ 91

vi

Index of Figures Figure 1.1 The highway environment .................................................................................................... 4 Figure 1.2 Structural layers of a typical flexible and rigid pavement .................................................... 5 Figure 2.1 Routes of water ingress and egress. .................................................................................... 10 Figure 2.2 Typical permeable base pavement sections: (a) PCC pavement and asphalt concrete shoulder section, (b) PCC pavement and PCC shoulder section, (c) asphalt concrete pavement and asphalt concrete shoulder section (adapted from Mathis, 1990) .......................................................... 13 Figure 2.3 Downhill sloping, crowned road cross section showing longitudinal and transverse drains and water flowing perpendicular to the equal elevation plane (adapted from Crovetti and Dempsey, 1993)..................................................................................................................................................... 13 Figure 2.4 Cross-sectional view of PET............................................................................................... 16 Figure 2.5 Drainage criteria for granular layers (adapted from Darter and Carpenter, 1987) ............. 17 Figure 2.6 Precipitation-outflow relationship for concrete pavement with pipe drain: outflow/precipitation = 69.8%, section 2 in Table 2.5 (adapted from Ahmed et al., 1997)................. 17 Figure 2.7 Time lag between precipitation and outflow for concrete pavement with pipe drain: outflow/precipitation = 5.5%, section 1 in Table 2.5 (adapted from Ahmed et al., 1997)................... 18 Figure 3.1 Experimental apparatus illustrating Darcy’s Law............................................................... 21 Figure 3.2 Typical soil retention curve (θsat= saturated water content = porosity, θres= residual water content)................................................................................................................................................. 23 Figure 3.3 Infiltration curve for an unsaturated soil (K = hydraulic conductivity).............................. 25 Figure 4.1 Schematic curve showing deviation from Darcy’s Law at turbulent flow (Re=Reynolds number; adapted from Bear, 1979)....................................................................................................... 33 Figure 5.1 Water level effect on base and subgrade moisture, Route 62, Florida (adapted from Ksaibati et al, 2000).............................................................................................................................. 38 Figure 5.2 Variations in moisture content with fluctuations in groundwater table (adapted from Chu et al., 1972)............................................................................................................................................... 39 Figure 5.3 Moisture content and capillary rise in granite that is a typical subbase material rich in fines (adapted from Jessep, 1998)................................................................................................................. 39 Figure 5.4 A sketch of initial and seasonal moisture changes for construction starting compacted dry or wet of equilibrium (adapted from Look et al., 1994 and Alonso, 1998) ......................................... 40 Figure 6.1 Hydraulic conductivity-air void content relationship for coarse-graded Superpave mixes. Hydraulic conductivity is essentially zero below 6 percent air void ratio (Choubane et al., 1998)..... 46 Figure 6.2 Hydraulic conductivity-air void content relationship for dense-graded asphalt mixes. Note that the range of hydraulic conductivities is lower than that presented in figure 6.1 (Terrel and Al-Swailmi, 1993). .................................................................................................................................... 47 Figure 6.3 Air-void percentage of asphalt concrete material ............................................................... 47 Figure 6.4 Hydraulic conductivity in-place air void content relationship for two different hot mix asphalt pavements (adapted from Cooley and Brown, 2000) .............................................................. 48 Figure 6.5 Hydraulic conductivity air-void content relationship for porous asphalt mixes (Fwa et al., 1999)..................................................................................................................................................... 48 Figure 6.6 Hydraulic conductivity as a function of shape and size of voids, presence of asphalt and mineral fillers and interconnectivity..................................................................................................... 49 Figure 6.7 Hydraulic conductivity versus effective porosity relationship for open and dense-graded asphalt mixes (Huang et al., 1999). ...................................................................................................... 49

vii

Figure 8.1 Hydraulic conductivity and gradation of base and filter materials (Cedergren, 1974). The Cedergren chart cannot be used for hydraulic conductivity estimation of gradations other than those shown.................................................................................................................................................... 56 Figure 8.2 Moulton nomograph is valid only for materials that have a specific gravity of 2.7(Elsayed and Lindly 1996). ................................................................................................................................. 57 Figure 9.1 Pavement cross-section showing path of draining water (adapted from Crovetti and Dempsey, 1993).................................................................................................................................... 65 Figure 9.2 Infiltration rate versus crack width for the Searsmont sample (adapted from Koch and Sandford, 1998). ................................................................................................................................... 67 Figure 10.1 Non-linear increase in hydraulic conductivity with increase in crack width (adapted from Oshita and Tanabe, 2000c)................................................................................................................... 77 Figure 10.2 Schematic interpretation of interaction of modules (adapted from Pufahl et al., 1990)... 78 Figure 10.3 Pavement segments where component models are used (adapted from Pufahl et al. 1990).............................................................................................................................................................. 79 Figure 10.4 Highway reference environments for fate and transport model application..................... 81

viii

1. Introduction

1.1 Recycled Materials Use in the Highway Environment

Around the world, the growth of the industrial sector results in increasing amounts and types of generated waste. Many of these wastes are non-decaying and some may be hazardous. The management of these wastes has become an issue for environmentally conscious societies. The issue is intensified by decreasing landfill space. Interest has grown in reuse of wastes as an alternative, which will ease landfill disposal, and may reduce raw material demand.

Both waste materials and by-products can be reused. If a material has served its purpose and is

no longer usable it is classified as a waste material. If a material is generated from production of another material, it is a by-product. Secondary products may be re-used in the same application as before, or they may be recycled to be used in a similar or different application. In this document, these materials will be collectively referred to as recycled materials.

The U.S. Highway agencies have been using recycled materials with varying degrees of success

for the past 20 years (Schimmoller et al., 2000). Several components of agricultural, domestic and industrial wastes have the potential to be used in the construction of highways (see table 1.1). At least 22 states have approved the use of coal fly ash and coal bottom ash in road construction (Callahan, 2000). Other popular recycled materials are asphalt pavement, reclaimed concrete pavement, and blast furnace slag (Schimmoller et al., 2000). State-specific details on beneficial use of recycled materials by highway agencies is provided by Callahan (2000).

A schematic of typical highway construction and materials used is shown in figure 1.1. Recycled materials may be used in almost all components of the highway environment. However, some applications such as in the manufacturing of appurtenances (e.g. fences, signs, sound barricades, and drain pipes) and in vegetative cover are relatively minor compared to the volume of recycled materials used in structural components of pavements. Recycled materials research often concentrates on their incorporation into asphalt and Portland cement concrete (PCC), granular or stabilized bases and subbases, embankments, and flowable fill because these applications are needed in high volumes. A detailed list of recycled materials used in these sections is shown in table 1.2.

The use of recycled materials raises concerns including (1) increased costs, (2) environmental impacts, (3) human health, and (4) long-term performance of pavements containing recycled materials (Decker, 1994). However, more recent studies indicate that some of these barriers have been overcome (Schimoller et al., 2000; Eighmy and Magee, 2001). Research is ongoing to develop cost effective and well-engineered applications of recycled materials. Existing design protocols for concrete, base course or other pavement components may not be applicable to waste modified pavement materials. New protocols are being developed to incorporate recycled materials in highway construction (Terrel et al., 1994; Anderson, 1996; Eighmy and Chesner, 2001).

1

Table 1.1 Annual production and use of recycled materials in roads in the United States (adapted from Collins and Ciesielski, 1994(a); Schroeder, 1994(b); Chesner et al., 1998(c)).

Production Recycled Uses in Highways Waste Material

(million metric tons) Asphalt Pavement Concrete Pavement Base Course Embankment Other

Animal manure 1,450a U1 Fertilizer, compost, oil production by thermal processing

Crop wastes 362a U Cellulosic waste as asphalt extender Rice husks as supplementary cementing material Animal feed

Agr

icul

tura

l w

aste

s

Lumber and wood wastes 64a U Mulch, fill for

embankments

Incinerator ash 7.8a, 7.3b, 8c <0.7b, 0c

Cement-stabilized base Past research indicated good performance, environmental questions

Asphalt paving aggregate Cement stabilized base Vitrified aggregate Palletized aggregate

Soil stabilization, fill, embankments Reef blocks, masonry block

Sewage sludge ash 0.5-0.9a,c U Asphalt mineral filler and aggregate Concrete course aggregate Flowable fill aggregate

Scrap tires 2.2a, 2.3b U

Use accepted, extensive research being conducted Asphalt-rubber binder Asphalt fine aggregate

Experimental stages Used as an insulator Used with some success – research continuing

Tire-derived fuel, stress-absorbing membranes, rubberized crack sealant, lightweight fill material Being marketed for use as noise or retaining wall, molded posts

Compost 2.3a Mulching material

Glass and ceramics 11.3a , 12b,c 2.4b, 3.2c Asphalt fine aggregate, accepted use, long term performance research under way

Past research indicated performance problems

Unbound base course

Some research projects under way

Glass cullet Pipe bedding material

Plastic waste 13.1a, 14.7b 0.3b Asphalt-cement modifier Experimental stage No known use No known use Fence and sign posts, delineators, geotextile manufacture, composite pipe pilings, sound barriers

Dom

estic

was

tes

Used motor oil 1.8a Fuel in asphalt plants Fuel in cement production Recycled as lubricant

Coal ash – Fly ash 43.5a, 45b 11b Past use as a mineral filler, research ongoing

Cement replacement, accepted use, research ongoing Stabilized base Flowable fill and grout

Embankments

Coal ash - Bottom ash 12.7a , 16b, 14.5c 5.0b, 4.3c Combined ash as a fine aggregate, performance

data limited Unbound and stabilized base

Embankments, backfill, flowable fill

Anti-skid material Lightweight concrete, abrasives

Coal ash – Boiler slag 3.6a , 2.3c 2.1c Asphalt paving Unbound and stabilized base Blasting grit

Roofing granules

Advanced SO2 control by products

4.5a, 18.0b, 21.4c

At least 1.0c Stabilized base Used in embankments in

Pennsylvania Soil stabilization

Construction and demolition debris 22.7a U Unbound base course Embankment borrow Wood as mulch

Blast furnace slag 14.1a,c 14.1b,c Accepted use as an aggregate in base and surface (friction) course, research indicates good performance

Concrete aggregate Accepted use as a cement additive in granulated form, research ongoing

Unbound base course Accepted use, good, hard, durable aggregate

Limited but accepted use Leaching of sulfurous compounds reported

Accepted use as ice control abrasive

Steel making slag 7.2a, 7.5b 7-7.5c Asphalt concrete aggregate Past research indicated good performance

Extensive research, poor performance

Limited use as granular base Accepted use Anti-skid material, ice control, railroad ballast

Non ferrous slags 9.1a,7.6-8.1c U Asphalt concrete aggregate Concrete aggregate Unbound base course Embankment or fill Blasting grit

Railroad ballast Cement and lime kiln dusts >8.3 Mineral filler, aggregate and cement modifier in

asphalt concrete Cementitious material in stabilized base

Cementitious material in flowable fill

Recycled into clinker, waste stabilization, agricultural lime

Bag house fines 5.4-7.2c U Mineral filler in asphalt Reclaimed asphalt and concrete pavements

45a,c, 94b 33c Asphalt concrete aggregate and asphalt cement supplement

Pavement recycling Coarse aggregate in concrete

Unbound base course Stabilized base Embankment or fill

Foundry sand 9.1a, 9.0-13.6c U Asphalt paving Fill material, flowable fill, pipe bedding

Roofing shingle waste 9.1a, 8.1b, 10c U Asphalt cement modifier, aggregate substitute

and mineral filler

Sulfate waste U Cement production Stabilized base Embankment, fill Wallboard manufacture Lime waste 1.8 U Mineral filler in asphalt Soil stabilization Paper mill sludge U U Cement replacement Dust pallative, fly ash-bark ash blend Petroleum contaminated soils U U Asphalt paving Stabilized base

Indu

stria

l was

tes

Mineral processing wastes 1,600c U Waste rock and mill tailings as asphalt

aggregates Waste rock, mill tailings and coal refuse as aggregates in granular base

Waste rock, mill tailings and coal refuse in embankments and fill

1 U = unavailable

2

Table 1.2 Recycled materials used in various highway applications (adapted from Chesner et al., 1998)

Application Material

Asphalt concrete – Aggregate (Hot mix asphalt)

Blast furnace slag Coal bottom ash Coal boiler slag Foundry sand Mineral processing wastes Municipal solid waste combustor ash Nonferrous slags Reclaimed asphalt pavement Roofing shingle scrap Scrap tires Steel slag Waste glass

Asphalt concrete – Aggregate (Cold mix asphalt) Coal bottom ash Reclaimed asphalt pavement

Asphalt concrete – Aggregate (Seal coat or surface treatment) Blast furnace slag Coal boiler slag Steel slag

Asphalt concrete – Mineral filler

Bag house dust Sewage sludge ash Cement kiln dust Lime kiln dust Coal fly ash

Asphalt concrete – Asphalt cement modifier Roofing shingle scrap Scrap tires

PCC – Aggregate Reclaimed concrete

PCC – Supplementary cementitious materials Coal fly ash Blast furnace slag

Granular base

Blast furnace slag Coal bottom ash Coal boiler slag Mineral processing wastes Municipal solid waste combustor ash Nonferrous slags Reclaimed asphalt pavement Reclaimed concrete Steel slag Waste glass

Embankment or Fill

Coal fly ash Mineral processing wastes Nonferrous slags Reclaimed asphalt pavement Reclaimed concrete Scrap tires

Stabilized base – Aggregate Coal bottom ash Coal boiler slag

Stabilized base – Cementitious materials (Pozzolan, pozzolan activator, or self-cementing material)

Coal fly ash Cement kiln dust Lime kiln dust Sulfate wastes

Flowable fill – Aggregate Coal fly ash Foundry sand Quarry fines

Flowable fill – Cementitious material (Pozzolan, pozzolan activator, or self-cementing material)

Coal fly ash Cement kiln dust Lime kiln dust

3

Figure 1.1 The highway environment

6Embankment

1Guard

2Sign

9Bridge

7Culvert

8Vegetative cover and landscape material

Pavement structure

Structure Material 1,2 Metal, plastic, wood 3,4 Binder, aggregate 5,6 Aggregate, soil 7 Concrete 8 Topsoil 9 Concrete, steel, plastic

3 Wearing surface 4Base 5Subbase

1.2 Pavement Structure Structural components of pavements are relatively well defined although there are many

variations to pavement systems. The two major types of pavements are flexible pavement (asphalt pavement) and rigid pavement (PCC pavement). Component layers for each of these sections are similar with minor differences in types of material used and number of layers (see figure 1.2). The purpose of the surface layer is to provide a safe and smooth riding surface with maximal skid resistance, and minimal load- and non-load associated fractures and deformation. The surface layer could be asphaltic or PCC. In PCC pavements, the surface layer is composed of slabs that are connected with dowel bars. Those constructed of asphaltic concrete are referred to as the wearing surface layer. Below the concrete layer is the base course, which may sometimes be underlain by a subbase layer. Base and subbase layers may be designed to (1) prevent water pumping, (2) protect against frost action, (3) drain excess water, (4) prevent volume change of the subgrade, (5) increase structural or load-supporting capacity, and (6) expedite construction (Yoder and Witczak, 1975). Shoulders of a pavement may have asphalt concrete or PCC surface layer underlain by base or subbase materials. Base and subbase layers are constructed above the subgrade, which is the native soil. Desirable properties of subgrades include strength, ease of drainage, ease of compaction, permanency of compaction, and permanency of strength (Yoder and Witczak, 1975).

Asphalt concrete pavements have an asphalt concrete surface and an asphalt treated base if not

a granular base. Asphalt concrete consists of asphalt aggregate (95 percent by mass) and asphalt binder. If the asphalt concrete mix is prepared hot instead of cold, then mineral filler consisting of very fine and inert material can be added to the hot mix asphalt (3-6 percent by mass) to improve the density and strength of the mixture. Recently, polymers or other antistripping agents may be used to

4

reduce rutting, fatigue distress, low temperature performance or moisture susceptibility (Crossley and Hesp, 2000). Typically the asphalt concrete mixture properties that are adjusted for optimal performance are flow, air voids, stripping resistance, resilient modulus, and compacted density. To some extent, each of these properties affects the water movement in the asphalt concrete.

Base

Wearing

Base

Wearing Asphalt Concrete Layer, 5-10 cm Aggregate or stabilized layer, 10-20 cm

Aggregate layer, 10-20 cm

Native soil

(a) Flexible pave

Subgrade

Contraction joint

Portland Cement Concrete slab, 20-30 cm Base, aggregate or stabilized layer, 10-20 cm

(b) Rigid pavem

Figure 1.2 Structural layers of a typical

Components of a PCC pavement include a granulathe surface, and occasionally an overlay of asphalt concPortland cement (15 percent by mass), water, coarse aggaggregate (usually sand). Sometimes, supplementary ceadmixtures may be used to modify the properties of fresand hardens when its raw materials (containing lime, irosupplementary cementitious materials commonly used aash, blast furnace slag, and silica fume.

The base and the subbase may be granular or stabi

to requirements for type I mixtures (gradations A,B,C, omay be used (see Appendix E for explanation of gradatiof a mixture of aggregate, cementitious materials (e.g. cgranulated blast furnace slag), and water that is compactGranular base and subbase layers do not have any reactiaggregates. The aggregates may be sand, gravel, crushedurable material of mineral origin. Typically, crushed scoarse aggregate particles (Chesner et al., 1998). Similagranular base and subbase provide bearing strength but astructure. For better drainage and reduced frost susceptiunbound layers is limited. Granular subbase is coarser t

5

Subbase Subgrade

ment

Dowel bar for load transfer

ent

flexible and rigid pavement

r or stabilized base, a subbase, a PCC slab on rete. The basic materials used in PCC are regate (crushed stone or gravel), and fine mentitious materials and chemical h or hardened concrete. Portland cement sets n, silica, and alumina) react with water. The re also reactive materials, and include coal fly

lized. Coarse or fine aggregates conforming r D) or type II mixtures (gradations E or F) ons). The stabilized base or subbase consists oal fly ash, silica fume, and ground ed to form a dense and strong layer. ng materials but instead consist of only d stone or quarry rock, slag or other hard, tone consists more than 50 percent of the r to the stabilized base and subbase, the lso significant drainage for the pavement bility, the amount of fines used in these two hen granular base (Chesner et al., 1998).

Angular, nearly equidimensional aggregates with rough surface texture are preferred over rounded, smooth aggregate particles because they are stronger, have better interlock, and do not break and form fines.

Soils and oversize materials are used for embankments. Different types of soils ranging from granular (sand and gravel) to more finely sized (silt and clay) soils may be used for construction. Saturated clays, silty clay, and presence of organic matter are not desirable. In granular base and subbase layers, aggregates ranging from 0.075 mm to 5-10 cm in diameter are used whereas in embankments larger materials such as rocks, large stones, and reclaimed paving materials may be included. The materials used in granular base and subbase layers as well as embankments are first spread in thin lifts of 150 mm to 200 mm and then compacted.

Flowable fill can be used in lieu of compacted earth to fill in voids in irregular excavations and

hard to reach places (such as under and around pipes) (Chesner et al., 1998). Bhat and Lovell (1996) noted that if properly designed and used, flowable fill can be an economic alternative to conventional compacted fills. Flowable fill is a self-compacting, cementitious slurry consisting of a mixture of fine aggregate or filler, water and cementitious materials (coal fly ash or Portland cement). For fine aggregate or filler, sand or recycled materials such as bottom ash, fly ash, spent foundry sand, quarry fines, and baghouse dust may be used (Chesner et al., 1998). Permeability of flowable fill is very low reducing the opportunity of any toxic component being leached out (Bhat and Lovell 1996 and 1997).

1.3 Incentives for Studying Water Movement in Pavements Compilation of literature data on water movement in pavements was necessary from a

pavement engineering perspective to better establish the relationship between moisture in pavements and pavement integrity. It is well accepted that moisture shortens pavement life. Water pumping and freeze-thaw phenomena are two examples causing pavement damage in the form of cracking, rutting, and stripping. To examine damaging effects of moisture, water regimes in pavements need to be known. However, a complete understanding of water regimes in pavements has been difficult given the wide range of pavement designs and wide range of natural and recycled materials. Water regimes depend on moisture ingress and egress routes and material hydraulic properties. What are the ingress and egress routes? How would these routes be affected by various pavement designs? What hydraulic properties of materials dictate water regimes? Is there a significant difference between the hydraulic properties of materials used in different structural layers of pavements? Compilation of literature data was necessary to answer these questions.

In the U.S., environmental compatibility of the recycled materials has been one of the primary

concerns in light of stringent solid waste regulations. Decker (1993) pointed out that the hot mix asphalt industry wants to ensure that ‘linear landfills’ are not being built. Similarly, Callahan (2000) reported that among 40 surveyed state environmental protection agencies (EPAs), all confirmed that foremost concern was placed on protecting human health and the environment while evaluating beneficial use. At present, there is not a standard, validated approach for evaluating the risk posed by use of these recycled materials (Callahan, 2000). From a recycled materials perspective, characterization of water movement in pavements was also essential because possible risks associated with natural and recycled materials used in highways can only be determined if the water regimes in pavements are known.

6

Hazardous constituents may be released into the environment from highway components by (1) dispersion of fugitive dust, and particulate and volatile emissions into the ambient air, (2) dissolution and transport in surface runoff, and (3) leaching of soluble components in percolating groundwater (Chesner et al., 1998). Highways may release pollutants to the environment even when recycled materials are not used. Ball et al. (1998) listed the primary sources of pollutants found in runoff from road surfaces (see table 1.3). Many of the pollutants released as part of surface runoff may also be found leaching into the groundwater. In the pavement, possible leachable pollutants include trace metals (e.g. As, Cd, Cu, Cr, Hg, Pb, Zn) and trace organics (e.g. benzenes, phenols, vinyl chloride) (Chesner et al., 1998). Air quality is compromised by volatile constituents such as volatile organics as well as fine particulate matter that contains trace metals and organics (Chesner et al., 1998). In order to evaluate environmental risks related to the highway environment, each potential release mechanism must be fully investigated.

Table 1.3 Pollutant constituents in runoff from conventional road surfaces (adapted from Ball et al., 1998)

Constituents Primary Sources Particulate Pavement wear, vehicles, atmosphere, maintenance Nitrogen, phosphorus Atmosphere, roadside fertilizer application

Lead Auto exhaust, tire wear, lubricating oil and grease, bearing wear Zinc Tire wear, motor oil, grease Iron Auto rust, steel highway structures (e.g. guard rails, moving engine parts)

Copper Metal plating, bearing and brushing wear, moving engine parts, brake lining wear, fungicides and insecticides

Cadmium Tire wear, insecticide application Chromium Metal plating, moving parts, break lining wear

Nickel Diesel fuel and petrol exhaust, lubricating oil, metal plating, bushing wear, brake lining wear, asphalt paving

Manganese Moving engine parts Cyanide Deicing compounds Sodium/calcium chloride Deicing salts

Sulfate Roadway beds, fuels, deicing salts

Petroleum Spills, leaks or blow-by of motor lubricants, antifreeze and hydraulic fluids, asphalt surface leachate, dust suppressants and roadbed stabilizers (Kimball, 1997)

PCB Background atmospheric deposition, PCB catalyst in synthetic tires, spraying of rights-of-way Significant research is underway to evaluate the risk of using recycled materials in the

construction of highways. Efforts have focused on understanding the leaching behavior of metals from unstabilized wastes (Mizutani et al., 1996; Kida et al., 1996; Johnson et al., 1996; Herck et al., 2000) and stabilized wastes (Schreurs et al., 2000; Kim and Batchelor, 2001). Leaching of metals is relatively well understood and is regulated whereas more research is needed in to evaluate the risk of volatile and non-volatile trace organics found in wastes (Eikelbloom et al., 2000). The use of recycled materials in highways will become a more attractive alternative if it can be shown that the risk associated with the use of recycled materials in highways is not significant compared to the risk arising from the release of and exposure to contaminants described in table 1.3. Vashith et al. (1998) reported that recycled materials use in highways may not be detrimental to the environment and human health. Similarly, Humphrey and Katz (2000) indicated that shredded tire fill used above the groundwater table did not increase the concentration of metals (except for manganese and iron) above primary and secondary drinking water standards after five years. More research is needed to evaluate all possible scenarios with all candidate waste materials established.

7

1.2 Objectives and Scope of the Report Past experience with recycled materials in highways has not shown risk with respect to

environmental impact and human health. Environmental regulatory agencies are tasked with assessing the potential for environmental impact or risk to human health. Leaching and advective transport of possible contaminants in recycled materials is believed to be the primary transport pathway. Both of these mechanisms depend on water content of the material. Unfortunately, little is known about water movement in engineered highway systems, particularly ones containing recycled materials. Limited literature on water movement in highways also poses a problem for pavement engineers who need to understand the hydraulic regimes to study the deteriorating effect of water in pavement structures. The purpose of this literature review was to describe the state of the knowledge about water movement in the roadway environment with and without recycled materials so that this knowledge may be incorporated into fate/transport models for use in risk assessment.

The scope of this paper is relatively broad. First, importance of drainage is discussed by examining paths of water ingress and egress, and highway components engineered to provide drainage (Chapter 2). Then, mathematical equations modeling water flow and parameters important for describing water flow are introduced (Chapter 3). Measurement techniques of parameters describing water flow are discussed in Chapter 4. Chapter 5 is devoted to a discussion of water content in pavements, and the effect of ground water table as well as the spatial and temporal variability. Hydraulic conductivity of asphalt concrete, PCC, and base/subbase/ embankments are discussed in chapters 6,7 and 8, respectively. Base, subbase, subgrade and embankments are lumped into one category because materials used for these layers are similar and information learned from one layer is applicable to another layer under the same category. After presenting water content and hydraulic conductivity data, the next step is to introduce factors that affect water flow in the highway environment (Chapter 9). The physical phenomena considered are water pumping, infiltration through fissures, and the effect of temperature and soil structure on water flow. In chapter 10 computer modeling of water movement in the highway environment is introduced. Simplified and comprehensive approaches that integrate the concepts presented in earlier chapters are presented. Finally, summary and conclusions (Chapter 11) and research needs (Chapter 12) are presented. A glossary is provided in the appendix.

Chapter 1 Synopsis In chapter 1 the motivation behind this study was introduced. High production rates of agricultural, domestic, and industrial wastes and their possible re-use in highway applications were summarized in tables 1.1 and 1.2. Having shown the high potential for beneficial use of these wastes, it was stated that an educated decision for recycling should incorporate environmental concerns determined by risk assessment. Addressing the risk involves determination of the extent of moisture and subsequent chemical leaching from recycled materials in the highway environment. It was stated that the objective of the present report was to determine the state of the knowledge about water movement in the highway environment so that this knowledge may be used in later stages of the research, in developing fate/transport models that will aid in assessment of the risk. The report should also serve as a useful resource for pavement engineers interested in improving drainage to prolong pavement life. In chapter 1, components of rigid and flexible pavements were introduced since a significant portion of the report requires knowledge of pavement layers.

8

2. Drainage Systems The importance of drainage for durability of pavements was recognized as early as when the

ancient Romans built extensive road networks with enhanced pavement drainage (Forsyth et al., 1987). Until the 1940s many designers adhered to the belief that drainage was essential for good performance. However, starting in the late 1930s, many modern testing procedures were developed to measure the strength of soils and emphasis was placed on strength in road design. Interest in increasing the strength of roads continued until the late 1960s, when it was realized that durability could be achieved only if water was properly drained out of the pavement. However, accepted designs with a focus on strength of pavement structures persisted for many more years. Cedergren (1974, 1988, and 1994) emphasized the importance of drainage throughout his career and the importance of drainage gradually gained acceptance by designers. Currently, drainage is of interest to transportation engineers because many failures in pavements are attributable to elevated moisture conditions. For the present study, water movement in pavements and subgrades were fully examined considering routes of water ingress and egress, and engineered drainage systems.

2.1 Routes of Water Ingress and Egress Current engineering practice is predicated on the fact that water enters the pavement despite

efforts to prevent it. The presence of water in the pavement is mainly due to infiltration through the pavement surfaces and shoulders, melting of ice during freezing/thawing cycles, capillary action, and seasonal changes in the water table. The significance of the respective routes depends on the materials, climate, and topography.

Elsayed and Lindly (1996) note that until the study by Ridgeway (1982), high water table and

capillary water were thought to be the primary causes of excess water in pavements. Recently, crack and shoulder infiltration, and to some extent subgrade capillary action, were considered to be the major routes of water entry to the pavement (Elsayed and Lindly, 1996; Dawson and Hill, 1998). The significance of infiltration was shown by an immediate increase in edge drain outflow following a precipitation event (Ahmed et al., 1993). Van Sambeek (1989) reported that surface water infiltration can account for as much as 90 to 95 percent of the total moisture in a pavement system. Van Sambeek (1989) also identified transverse and longitudinal joints as major routes of ingress. Similarly, field studies by Ahmed et al. (1997) showed that pavement-shoulder joints were a major source of surface infiltration. For routes of egress, Dawson (1998) noted that the lateral or median drain is the most significant route except when a highly conductive underdrain (subgrade unsaturated hydraulic conductivity >0.1 cm/s) is provided. Thus, infiltration through cracks and joints is thought to be the major ingress route and engineered drainage is believed to be the major egress route. Many studies on drainage efficiency report a ratio of precipitation to drainage outflow because of this conception.

A simplified schematic and a list for routes of ingress and egress are provided in table 2.1 and

figure 2.1. No study was found in the literature that examined evaporation, reverse gradient of permeable layers above formation level, or direct rainfall on pavement during construction. Some studies were found that examined the water table effect on moisture content of base and subgrade (section 5.1). In addition, many studies focused on the phenomenon of pumping (section 9.1), and melting of frost lenses (sections 9.3), however, the emphasis in these studies was on pavement

9

performance as a function of these processes rather than quantifying water movement under these conditions.

Table 2.1 Routes of ingress and egress (adapted from Van Sambeek, 1989 and Dawson, 1998) Directio From Route

Construction joints Cracks resulting from shrinkage during/after construction Cracks resulting from distress due to loading

Pavement Surface

Diffusion through intact materials Artesian flow Pumping action under traffic loading Capillary action of lower pavement layer(s)

Subgrade

Water vapor rising through subgrade soils Reverse gradient of permeable layers above formation level Lateral or median drain surcharging Road Margins

Capillary action of pavement layers Pavement or ground run-off via unsealed shoulder Direct rainfall on pavement during construction

Ingress

Other Sources

Frost lenses melting during spring thaw Pumping through cracks/joints existing Capillary rise and evaporation through cracks

Pavement Surface

Diffusion/evaporation through intact material Infiltration to permeable, low water-table subgrade Subgrade Capillary action of subgrade Gravitational flow in aggregate to lateral or median drain

Egress

Road Margins Vertical flow in aggregate to open-graded drainage layer below

Edge

SurfaceEntry

Infiltration and Drainage

BBBound Layyyereerroouunndd LLaa

High

Figure 2.1 Routes of water ingress and egress.

VaporMovementCapillary

SuctionWaterTableRise

GroundDrainage

10

2.2 Significance of Drainage

Once in the pavement, water can stay for days or weeks after each saturating rainfall (Cedergren, 1988). To sustain the strength of the subgrade soil, it was necessary to remove the water from the pavement structure before it reached the subgrade soil. Cedergren (1974) noted that drainage systems were required for all pavements. According to Cedergren (1974), exceptions for the need of drainage systems include areas in which there is no groundwater or spring inflow, or where the annual rainfall is less than 20 or 25 cm, and no significant amount of snow or ice can enter structural sections. Areas in which subgrades are very permeable and are not subject to freezing, and the natural water table is very deep do not require drainage systems, nor do pavements that will be subjected to very limited numbers of heavy wheel loads over their design life, for example, on highways with less than 150 or 200 axle loads (8200 kg each) per day. Since drainage of pavements is required in most cases, the need for removing infiltrated water from the pavement increased interest and research into pavement subdrainage systems.

In the past decade, pavement designers realized that the pavement life could be extended three-fold by proper installation and maintenance of subsurface pavement drainage systems (Christopher and McGuffey, 1997). The accepted practice was to remove water to minimize moisture-related problems such as rutting, stripping, cracking and pumping. To remove the water in the pavement, many states adopted the use of drainage systems such as permeable bases and edge drains. Experience showed that this costly practice extended the pavement life only if the subsurface drainage systems were well maintained. Maintenance included cleaning outlets, replacing rodent screens, flushing or replacing outlet pipes, repairing damages, and deepening ditches. Current belief supports installation of drainage systems only if routine maintenance can be provided. In the absence of maintenance, pavements become flooded and susceptible to increased water damage. Unfortu-nately, in a national survey, several of the states admitted that proper maintenance was not practiced (Christopher and McGuffey, 1997).

Installation of drainage systems may have varying degrees of importance for pavement performance. A nation-wide survey and review of the literature by Christopher (1998) and Van Sambeek (1989) compared the extent of drainage requirements. It is suggested that because of the spacing between the slabs, and the slabs and shoulders, subsurface drainage systems for jointed concrete pavements are more critical compared to asphalt pavements. Similarly, subsurface drainage systems are highly important for freeze-thaw areas. Christopher (1998) and Van Sambeek (1989) stated similar criteria to those suggested by Cedergren (1974). Subsurface drainage systems may not be needed if annual rain does not exceed 40 cm, the subgrade has more than 300 cm/day hydraulic conductivity1, the pavement is structurally inadequate for drainage, lateral and vertical drainage in the pavement section exceeds infiltration, or predicted equivalent standard axle loads per day is less than 250 for a rigid pavement. More sophisticated decision criteria for evaluating the need for drainage are described by AASHTO (1998) and further modified by Mallela et al. (2000).

1 Contrasting Christopher’s findings, Dempsey (1988) and Forsyth et al. (1987) note that drainage provides no additional benefit if average annual rainfall is less than 25 cm (not 40 cm) and hydraulic conductivity of the subgrade exceeds 1,500 cm/day (not 300cm/day).

11

Institutions differ in design decision approaches on the use of drainage systems. The Federal Highway Administration (FHWA) requires permeable bases for all interstate pavements. Similarly, the World Road Association (PIARC) requires permeable bases for all PCC pavements. On the other hand, U.S. Army Corps of Engineers (ACOE) requires permeable bases only for pavements over 200 mm thick and makes it optional for any thinner pavements. The Ministry of Transportation of Ontario (MTO) requires that a 100 mm layer of open-graded drainage layer be placed beneath the concrete slab in all new rigid pavement designs (Kazmierowski et al., 1994).

2.3 Engineered Systems for Drainage Several components of pavements aid drainage. A permeable base and a collector pipe are the

two major components that enable drainage. A longitudinal collector and an outlet pipe are referred to as edge drains. In this section, permeable base and edge drains will be briefly described and other moisture-related components of pavements such as a filter layer, partial exfiltration trench (PET), and vertical moisture barrier will be introduced. Typical permeable base pavement sections, shown in figures 2.2 and 2.3, are provided to visualize pavement drainage components and water flow directions. Table 2.2 is a summary of drainage component functions that are described in this section.

A common design approach in subsurface drainage systems is the installation of a permeable

base, which serves to remove infiltration water. The highly permeable base drainage layer is at least seven to ten centimeters thick and extends under the full width of the roadway exposed to traffic loads. Permeable bases are used in both PCC and asphalt concrete pavements (see figure 2.2). The permeable base may be located just above the subgrade or above the base (Van Sambeek, 1989). When placed above the subgrade, the entire pavement structure can be drained. Placement above the base layer allows faster drainage of overlying layers (especially water from frost melt) and prevents water from reaching the subgrade. On the other hand, drainage of base and subgrade may be limited in the latter design. The permeable layer may also be used without another base.

A properly designed and constructed permeable base layer may function as a conventional dense-graded base, supporting the pavement by distributing the loads. In addition, a permeable base layer provides improved drainage and minimizes frost action. Inclusion of asphalt-treated permeable bases may also add more strength to the pavement and increase pavement fatigue life compared to pavements that have aggregate bases and not asphalt treated permeable base (Long et al., 1996). McEnroe and Zou (1993) noted that the amount of damage per load application is roughly 10-20 times less on the same pavement if an unsaturated, permeable base is used instead of a saturated impervious base. After years of experience, many states agree that permeable bases prolong life although philosophies differ with respect to the degree of permeability (Kozeliski, 1992; Mathis, 1990).

12

Figure 2.2 Typical permeable base pavement sections: (a) PCC pavement and asphalt concrete shoulder section, (b) PCC pavement and PCC shoulder section, (c) asphalt concrete pavement and asphalt concrete shoulder section

(adapted from Mathis, 1990)

Figure 2.3 Downhill sloping, crowned road cross section showing longitudflowing perpendicular to the equal elevation plane (adapted from C

PCC Pavement Asphalt Concrete Shoulder

Base and/or Subbase

Permeable Base

Permeable Base

Filter

(a)

Collector Pipe

PCC Pavement PCC Shoulder (b)

Subgrade

Filter

Subgrade

Asphalt Concrete Shoulder

Asphalt Concrete Pavement

Permeable Base

Base and/or Subbase

Collector Pipe

Subgrade

Filter

Permeable Base

PCC Concrete

Subgrade Transverse Drains and Outlet Pipes

Longitudinal Grade

Cross Slope Equal Elevation

Collector Pipe

Base

(c)

13

Longitudinal Drains

inal and transverse drains and water rovetti and Dempsey, 1993).

Table 2.2 Pavement drainage system components

Drainage Component Function Permeable base Collects infiltrating water and moves it to the edgedrains while providing adequate support to

the pavement May be used with or without another base May be stabilized with asphalt or cement

Collector pipe Slotted or perforated pipe Receives water from drainage layer and conveys it to an outlet pipe Can be longitudinal or transverse

Outlet pipe Conveys water from collector to a drainage facility outside the pavement

Known as edge drains, longitudinal edgedrains, retrofit edgedrains

Filter layer and/or separator

Can be geotextile, dense-graded base layer, subbase layer or cement stabilized subgrade Maintains separation of permeable base and subgrade and prevent them from intermixing Should reduce penetration of subbase particles into subgrade soil Should prevent intrusion and clogging

PET Controls quantity and quality of water Immobilizes suspended solids and dissolved metals

Vertical moisture barrier Required only for pavements constructed on expansive soils Prevents seasonal lateral migration of moisture to and from the subgrade

In the last few years, in the Netherlands and in France, the asphalt surface, which is typically

impervious asphalt, is being substituted for a pervious type of asphalt, also known as porous asphalt (Berbee et al, 1999; Pagotto et al., 2000). Porous asphalt is also used in Sweden as a straining layer for the surficial trapping of the larger size suspended solids (Sansalone, 1999). Berbee et al. (1999) compared impervious and pervious (porous) asphalt and concluded that some of the advantages of switching to pervious asphalt in the Netherlands include a decrease in concentration of pollutants, noise abatement, and improved skid resistance in wet weather. Pagotto et al. (2000) noted that a porous asphalt (>20% voids) allows a gradual evacuation of water onto the outlet (peak flow is limited and time of discharge is longer) and improves runoff water quality. The disadvantages of porous asphalt are shorter pavement life, a risk of clogging of the voids, the need for higher salt dosages during snow and frost, and somewhat higher construction costs (Berbee et al., 1999).

In the late 1970s, filter layers were adapted into drainage design. The need for a filter layer was

widely agreed upon; however, its application was not economically feasible until relatively inexpensive materials such as geotextiles were introduced to drainage applications (Forsyth et al., 1986). Currently, a separator/filter reinforcement layer is typically placed between the permeable base and natural soils to prevent infiltration of fines into the subbase and the migration of subbase into the permeable base. To function properly, the filter layer should not be clogged with suspended particles and it should also be strong enough to carry and/or distribute the applied loads (Van Sambeek, 1989).

In the field, a filter layer may consist of a dense-graded subbase or geotextile may be used as the filter layer (Christopher, 1998). Mathis (1990) noted that those states using an untreated permeable base use a dense-graded base or subbase layer as a filter layer whereas those states that use a treated permeable base use a geotextile as a filter layer. Field studies by Alobaidi and Hoare (1996) show that geotextiles do reduce the penetration of subbase particles into the subgrade soil. However, at the same time, geotextiles may allow for quick dissipation of the cyclic pore water pressure and therefore cause erosion of the subgrade surface and migration of fines with water across the geotextiles into the subbase layer (e.g. high permeability geotextiles; Alobaidi and Hoare, 1996).

14

If the permeable base is not daylighted, then the design should include an appropriate collector

and outlet pipe. The collector is a slotted or perforated pipe or conduit that removes water from a drainage layer and/or aggregate trench surrounding the collector and quickly conveys it to suitable outlets along the roadway (Van Sambeek, 1989). A system of transverse or longitudinal collectors may be used. Pipe diameters should be greater than or equal to 100mm to allow for video camera inspections and pipe maintenance (Mallela et al., 2000). Outlet pipes receive water from collectors and drain it to a ditch, storm sewer, catch basin, or other surface drainage facility outside the pavement structure. Spacing of outlet pipes should not exceed 75m to allow for effective cleaning (Mallela et al., 2000).

One of the most commonly used drainage systems is the longitudinal edgedrain that is also known simply as edgedrain or retrofit edgedrain if it is an addition to an existing pavement. Existing pavements can be retrofitted with prefabricated edge drains (geotextile fin drains) that are easy to place and are cost effective (Ahmed et al., 1997). Edgedrains usually consist of longitudinal collectors, outlet pipes and possibly transverse collectors (Van Sambeek, 1989). Edgedrains may be installed in new or existing pavements because they do not require a drainage layer and a protective filter. When installed, edge drains receive water from the base/subbase layers and discharge it outside of the pavement through outlet pipes. Edge drains may reduce the subgrade moisture by as much as 28 percent (Fleckenstein and Allen, 1996). Both the permeable base and the edge drain have to be installed properly and regularly maintained to provide a cost effective drainage solution. Recently, improved technological methods for monitoring highway edge drainpipe systems have provided more efficient tools to determine the interior conditions of edge drains (Daleiden and Peirce, 1997).

Another design element related to water movement in road structures is vertical moisture

barriers. Vertical moisture barriers are not part of a drainage system but they prevent seasonal lateral migration of moisture to and from the subgrade beneath the pavement. They are required for areas that have expansive soils that expand and shrink during wet and dry periods, respectively. Vertical moisture barriers are used successfully in many cases across the U.S. to control roughness and cracking generated from expansive soil subgrades. At present, the construction methods of these barriers are still crude and relatively expensive. A more detailed discussion on the construction and use of vertical moisture barriers is provided by Evans and McManus (1999) and Evans et al. (1996).

An innovative approach for controlling rainwater is the PET, which is designed to control runoff that does not infiltrate into the pavement. The hybrid design of a PET consists of a porous pavement cap, a filter sand column, and a wrapped underdrain (see figure 3.4). The primary differences between current underdrain designs and the PET are a perforated underdrain surrounded by an engineered porous media and capped with porous pavement directly above the trench to promote infiltration, and selection of the engineered porous media to enhance sorptive capacity (e.g. preference of oxide coated sand over regular sand; Sansalone, 1999). Not only storm water quantity, but also water quality is controlled by PETs. Suspended solids and dissolved metals transported with the flow are immobilized in the PET by filtration and sorption, respectively (Sansalone, 1999; Li et al., 1999). These relatively new designs can be easily installed along paved surfaces, such as urban highways and parking lots.

15

w

10cm diameter Underdrain

Exfiltration (to subgrade)

n

Figure 2.4 C

2.4 Drainage Efficiency Subgrade and pavement materials may

completely full of water, then the medium is addition to water, and because water does noas an unsaturated medium. The degree of sacontains water. For saturated media, saturatipercent. For any saturation level in between,

One approach used for evaluating drainagrequired for a certain percentage of free wateand Hicks (1977) suggested a time of 2-6 hoairport pavements. Darter and Carpenter (19an 85 percent saturation level (see figure 2.540-60 percent of the cumulative outflow volu2.6). Similarly, Feng et al. (1999) also repor(see table 2.3). If there is a longer time lag, aprecipitation intensity and relatively dry baseopposed the time criteria for drainage approadrain fast. He considered that not the time, budrainage to hydraulic conductivity of materiaconductivity less than 0.017 cm/s do not draipercent saturated and that a hydraulic conducpercent drainage. At 0.017 cm/s, only 20 perfor drainage is also not suitable considering tis detrimental to pavement performance. Noof saturation in the pavement to pavement pehigh enough to decrease the pavement life si

Infiltratio

Highway pavement

m

Filter sand

ross-se

be insaturat comturatioon is the m

e desr to durs fo87) p). Fieme ta

ted ths sho condch, stt the ls andn at ativitycent shat it literarformgnific

1

30 cm

ctional view of P

an unsaturateted. Sometimpletely saturatn refers to the

100 percent, wedium is refe

ign and perforain from a sar removing 50roposed a timeld experimentkes place withat the drainagewn in figure 2itions prior toating that if thextent of drain noted that gr

ll, that at 0.03 value of 0.07aturation remis not clearly ture was founance. Saturatantly.

6

40-90 cm

dee h

r

rmtu p

s

. re

a84addio

10 c

Lateral flo

ET.

or saturated condition. If pores are s the pores are filled with air in the soil, this condition is referred to percentage of the void space that ereas for dry media, saturation is 0

red to as unsaturated.

ance is to consider the time rated base or subbase. Barksdale ercent of the drainable water from

of five hours as acceptable to reach by Ahmed et al. (1997) showed that in the first four hours (see figure period varied between 4 and 7 hours 7, the reason may be low ain event. McEnroe (1994) pavement will drain at all, it will age is important. McEnroe related nular materials with a hydraulic cm/s material will remain 85 cm/s is required to achieve 50 ins after drainage. The time criteria efined what percentage of saturation that specifically related the degree n levels of 50 or 85 percent may be

Table 2.3 Drainage performance of Indiana pavements (Feng et al., 1999)

Pavement % rainfall drained Average drainage time (hrs)

Open graded asphalt drainage layer over a dense asphalt base filter/separator layer 7.26 4.0

Open-graded asphalt drainage layer over a dense aggregate filter/separator layer 7.57 6.45

Open-graded asphalt drainage layer over a dense aggregate filter/separator layer 8.4 7.0

02468

1012141618

80 85 90 95 100

% Saturation

Tim

e (h

rs) Unacceptable

Marginal

Satisfactory

Figure 2.5 Drainage criteria for granular layers (adapted from Darter and Carpenter, 1987)

0

0.5

1

1.5

2

2.5

1 5 9 13 17 21 25 29

Time (hrs)

Pre

cipi

tatio

n (m

m)

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8O

utflo

w (m

3 )

PrecipitationFlow

Figure 2.6 Precipitation-outflow relationship for concrete pavement with pipe drain: outflow/precipitation = 69.8%, section 2 in table 2.5 (adapted from Ahmed et al., 1997)

17

0

0.5

1

1.5

2

2.5

3

1 5 9 13 17 21 25 29 33 37 41 45 49 53 57 61 65

Time (hrs)

Prec

ipita

tion

(mm

)

0

0.02

0.04

0.06

0.08

0.1

Out

flow

(m3 )

PrecipitationFlow

Figure 2.7 Time lag between precipitation and outflow for concrete pavement with pipe drain: outflow/precipitation = 5.5%, section 1 in table 2.5 (adapted from Ahmed et al., 1997)

Another approach for evaluating the drainage efficiency is to compare precipitation to

edgedrain outflow rates. Hagen and Cochran (1996) compared four different pavement drainage systems to determine if transverse drains may be an efficient alternative to drainable base layers. The pavements were tested at three rain events (14, 21, and 41 mm). The four pavements had a PCC pavement surface and a dense-graded aggregate base (see table 2.4; A). In addition, three of the pavements had the following drainage systems: (B) permeable asphalt stabilized base, (C) transverse pipe drain under joints, and (D) geocomposite transverse drain under joints. The results, suggest that the permeable asphalt-stabilized base drains the most water within two hours after the rainfall ends and provides the driest pavement foundation.

Ahmed et al. (1997) compared drainage efficiency of PCC, asphaltic pavements, and asphalt

overlain concrete pavements with pipe or fin drains (see table 2.5) and concluded that overlaid pavements drained only negligible amounts of precipitation; concrete pavements exhibited the highest flow volumes, possibly due to the presence of joints; the type of drainage system did not affect drainage efficiency considerably; between two overlain pavement systems both with fin drains, the one with less permeable base and lower intensity of cracks had lower outflow volume; and at most sections, a lower intensity of precipitation yielded similar outflow volumes suggesting that precipitation duration or intensity are not the sole factors affecting drainage. Ahmed et al. (1997) recorded precipitation in cumulative volume since a single value for precipitation is often not descriptive considering the variations of the rate of precipitation. Finally, Feng et al. (1999) reported that the infiltration rate into the pavement increased after the first winter and then decreased in the next three years, possibly because cracks did not occur during the first three years and instead air voids were densified by traffic load and also filled with sand.

18

Table 2.4 Drainage data of pavements for the first two hours (Hagen and Cochran, 1996)

Pavement or drainage description % rainfall drained at 14, 21, and 41 mm rainfall A PCC pavement surface and a dense-graded aggregate base 32, 25, and 43 B (A) + permeable asphalt stabilized base 20, 13, and 46 C (A) + transverse pipe drain under joints 26, 25, and 59 D (A) + geocomposite transverse drain under joints 34, 35, and 46

Table 2.5 Drainage data of pavements in Indiana (Ahmed et al., 1997)

Section Route and county

Pavement and drain

type

Subgrade soil

Base/ subbase material

Max. ppt.

(mm/hr)

Cumulative ppt. (m3)

Outflow/ ppt. (%)

Moisture related

distress or joint

condition 2.7 18.82 5.5 NR 79.66 40.4 1 US-31

Hamilton Concrete, pipe drain

Sandy loam

Bituminous stabilized No. 5D NR 57.79 26.5

Edge cracks/joint seal damage

2.2 7.10 69.8 NR 14.21 32.1 2 US-36

Hendricks Concrete, pipe drain Loam

Bituminous stabilized No. 5D NR 10.67 33.8

Edge cracks

2.7 9.82 59.9 3 US-41

Sullivan Concrete, fin drain Silty clay

Bituminous base No.

53B NR 5.07 34.6

Edge cracks/shoul

d damage NR 1.95 50.6

4 SR-63 Vermillion

Asphalt, pipe drain

Gravelly sand

Crushed aggregate

No. 53 NR 3.40 41.7

All major distress types

5 SR-9 Noble

Asphalt, pipe drain

Gravelly sand

Crushed aggregate

No. 53 23.0

41.86 26.3 Edge cracks

NR 4.25 1.3 NR 43.02 2.4 30 64.81 2.8 NR 2.12 2.2

6 US-30 Laporte

Asphalt overlay on concrete, fin drain

Fine sand Sand No. 24

NR 29.15 2.8

Edge cracks/reflection cracks

NR 52.21 0.2 NR 21.73 0.5 7 US-31 St.

Joseph

Asphalt overlay on concrete, fin drain

Fine sand Crushed

aggregate No. 53 NR 27.56 0.1

Edge cracks/reflection cracks

NR= Not Reported, ppt = precipitation

Chapter 2 Synopsis In chapter 2 routes of water ingress and egress for pavement structures were discussed and it was pointed out that crack and joint infiltration and engineering drainage are the two major ingress and egress routes, respectively. Drainage systems can remove excess water if they are well maintained. For drainage, a permeable surface and a base/subbase are required. Permeable base collects infiltrating water and moves it to collectors and outlet pipes (edge drains). An edge drain design that improves water quality is the PET, which can treat the suspended solids and dissolved metals of the water being drained. Porous asphalt can also be used to reduce suspended solids. In presence of swelling soils, vertical moisture barriers can be used to prevent seasonal lateral moisture migration. Drainage efficiency can be quantified by the time required to remove a certain fraction of drainable water and by comparing precipitation amount to volume drained. Drainage appears to occur within 4-7 hours after precipitation. Ratio of precipitation to outflow volume from edge drains varies significantly (6-70 percent) depending possibly on pavement and edgedrain type, geometry and condition as well as intensity and duration of precipitation.

19

3. Water Flow Theory

3.1 Darcy’s Law for Saturated Water Movement Water movement in saturated conditions can be described by the well-known Darcy’s Law, an

empirical equation. Although Darcy’s Law is valid only under low flow rates where flow is laminar, the relationship is applicable for most hydrogeological regimes. Darcy’s Law states that the flow rate in one dimension through a porous medium is proportional to the cross-sectional area and the head loss along the length of the medium (see figure 3.1). Stated differently, the hydraulic gradient in porous materials is proportional to the volumetric flow rate. Darcy’s Law can be expressed as,

dldhKA

∆l∆hKAQ −=−= (Eq. 3.1)

where: Q = volumetric discharge [L3T-1], K = proportionality constant [LT-1], A = cross-sectional area [L2], and dh/dl = ∆h/∆l = gradient of hydraulic gradient [LL-1].

The proportionality constant (K) is referred to as the hydraulic conductivity and is a function of

the properties of the fluid and the medium in which the fluid flows. A related term is the intrinsic permeability, which is a function of the properties of the medium only. Hydraulic conductivity, in qualitative terms, is the ease with which fluid can move through a porous material (Domenico and Schwartz 1990). Hydraulic conductivity and intrinsic permeability are related by the following equation:

µgkρK w= (Eq. 3.2)

where: K = hydraulic conductivity [LT-1] ρw = fluid density [M L-3], g = acceleration due to gravity[LT-2], k = intrinsic permeability [L2], and µ = viscosity of fluid [ML-1T-1].

20

∆l

∆h

Cross section A

Q

Q

Figure 3.1 Experimental apparatus illustrating Darcy’s Law.

A discussion on testing concerns and valid approaches for non-laminar flows is presented by

Huang et al. (1999) and Fwa et al. (1999). Darcy’s equation shown in equation 3.1 can be rewritten as:

v = K i (Eq. 3.3) where:

v = q = discharge velocity = Q/A [LT-1] , K= hydraulic conductivity [LT-1], and i = dh/dl = hydraulic gradient [LL-1].

In turbulent or non-laminar conditions, a polynomial or a potential form may be used to relate

hydraulic gradient and discharge velocity. A binomial form is more common: i = av + bv2 (Eq. 3.4)

where: a = curve fitting parameter [TL-1], and b = curve fitting parameter [T2L-2].

The potential form is represented as v =K´in (Eq. 3.5)

where:

K´ = pseudo-hydraulic conductivity.

21

Fwa et al. (1999) noted that the experimental coefficient, n, has a value of 0.5 for turbulent

conditions and 0.7 for porous asphalt mixture. However, there is no unified literature for parameters of both the potential and the polynomial equations.

3.2 Soil Moisture Retention Function With slight modification, Darcy’s Law remains valid for unsaturated flow. While hydraulic

conductivity is constant for saturated conditions, in unsaturated media, hydraulic conductivity is a function of moisture content, which, in turn depends upon the soil suction or matric potential. Thus, to expand Darcy’s Law for unsaturated conditions, determination of hydraulic conductivity as a function of water content or matric potential is necessary. In defining the water content of porous media, volumetric water content (θ) is used in soil science and gravimetric water content (w) is used in geotechnical engineering practice (Fredlund and Xing 1994). The definition of water content and calculation of it using the gravimetric method is shown in equation 3.6.

il solidsMass of so

terMass of waDry weight

eightht - Dry wMoist weig ent ==Water cont (Eq. 3.6)

Due to surface tension, the pore water in unsaturated conditions is under a negative pressure,

which is referred to as capillary potential or matric potential. Unsaturated conditions where negative pressures are observed can be found above the ground water table. Above the water table, pores are increasingly less saturated with distance from the water table. Closer to the water table, negative pressures are small; with distance, negative pressures (suction) increase as the medium becomes drier.

An interesting behavior of matric potential is its relation to the water content. Water content and matric potential are positively related (see figure 3.2). Lowering the water content makes the matric potential more negative. However, the relation between matric potential and water content shows a slightly different curve during drying compared to wetting. Thus, the equilibrium water content and matric potential depend on whether the soil is letting water in or out. Occurrence of a distinct drying and wetting curve is referred to as hysteresis. Hysteresis implies that to characterize the state of a soil, the water content and the water potential as well as the drying and wetting history needs to be known. Hysteresis is caused by variations of the pore diameter (ink bottle effect), differences of radii in the advancing and receding meniscus (contact angle), entrapped air, and swelling/shrinking processes in the soil grains (Lebeau et al., 1998; Tindall et al., 1999).

22

θ (water content)

wetting

drying

θres θsat

Ψ (m

atric

pot

entia

l)

Figure 3.2 Typical soil retention curve (θsat= saturated water content = porosity, θres= residual water content).

3.3 Unsaturated Hydraulic Conductivity Estimation of unsaturated hydraulic conductivity for various volumetric water contents is still a

challenge. A range of hydraulic conductivity values (hydraulic conductivity functions) is of interest because hydraulic conductivity changes with water content (or matric potential). The unsaturated hydraulic conductivity function can be estimated using empirical equations as well as macroscopic models and statistical models that are summarized by Fredlund et al. (1994) and Leong and Rahardjo (1997). However, there are no universal relations available for unsaturated hydraulic conductivity versus soil suction or water content (Tindall et al., 1999).

The relation between water content and suction for the soil is referred to as the soil-water characteristic curve and can be used to estimate the hydraulic conductivity (see figure 3.2). In the past decade, there has been significant interest in understanding the soil-water characteristic curve (Fredlund and Xing, 1994; Fredlund et al., 1994). As the name implies, the soil-water characteristic curve is specific for different materials. Thus, the empirical equations for the soil-water characteristic curve are limited to a narrow range of conditions. Fredlund and Xing (1994) discussed the limitations of the commonly used equations such as those proposed by Brooks and Corey (1964), Williams et al. (1983), McKee and Bumb (1984), McKee and Bumb (1987), Bumb (1987), and van Genuchten (1980) (see table 3.1). One of the newer empirical equations is introduced by Fredlund and Xing (1994). It is a more general empirical equation for soil-water characteristic curve and it is based on the pore size distribution of the medium. Once a general model for soil-water characteristic curve is established, estimation of unsaturated hydraulic conductivity is made easier.

23

Table 3.1 Empirical equations simulating the soil-water characteristic curve

Author Equations Strengths Limitations DefinitionsBrooks and Corey (1964)

λ

ψbψ

=Θ

Verified by many studies Not valid near maximum desaturation or under fully saturated conditions

Williams et al. (1983)

ln ψ = a1+b1 ln θ Many times used for soils in Australia

McKee and Bumb (1984) 2)/b2a(ψ−−

=Θ e Not valid near maximum

desaturation or under fully saturated conditions

McKee and Bumb (1987) Bumb (1987) 3)/b3a(ψ

1

1−

+=Θ

e

Better approximation in the low-suction range

Not suitable in the high suction range

van Genuchten (1980)

m

nψ1

1

+

=Θq

Frequently used, Can be turned into a closed form expression of hydraulic conductivity if m = 1 – 1/n

Gardner (1958)

+

=Θ nψ1

1

q

Special case of the van Genuchten equation

Fredlund (1994) m

e

+

= nψ/a)(ln[

1sθθ

Theoretical equation uniquely determines the soil-water characteristic curve once the pore-size distribution of a soil is known Fits experimental data reasonably well within the suction range of 0 to 106 kPa

rs

rθθ

θθ

−

−==Θ content water normalized , dimensionless,

θr = residual volumetric water content, dimensionless, θs= saturated volumetric water content, dimensionless, Ψ = suction [ML-1T-2] Ψb = air entry value [ML-1T-2] λ = pore size distribution index, dimensionless, a1, b1, a2, b2 = curve fitting parameters , a,m,n,p = different soil parameters.

24

3.4 Infiltration Infiltration is the process of vertical movement of water into a soil from rainfall, snowmelt, or

irrigation (Bedient et al., 1999). Tindall et al. (1999) provided a brief summary of some of the well recognized models for infiltration: the Green-Ampt approach, the Horton and Kostiakov equations, the Holtan model, the Morel-Seytoux and Khanji model, and the Smith-Parlange model.

The Green-Ampt approach is one of the earliest physically based approaches to infiltration. It

assumes that the water moves into dry soil as a sharp wetting front that separates the wetted and unwetted zones. This model suggests that as long as there is sufficient water at the surface for infiltration, the infiltration rate is a value greater than the saturated hydraulic conductivity. Initially, the infiltration rate is high, then, it decreases and asymptotically approaches the saturated hydraulic conductivity (see igure 3.3).

K

Time

Infil

tratio

n ra

te

(L/T

)

Figure 3.3 Infiltration curve for an unsaturated soil (K = hydraulic conductivity).

Generally water flow in the unsaturated zone is solved using a form of the Richards Equation.

Richards Equation is derived from substituting Darcy’s Law into the unsaturated continuity equation. The unsaturated continuity equation is an expression for conservation of mass and states that the mass entering a specific volume of interest less the mass leaving the volume is equal to the change in mass storage with time. It is expressed as:

ρθtz

)ρq(y

)ρq(x

)ρq( zyx

∂∂

=

∂

∂+

∂

∂+

∂∂

− (Eq. 3.7)

where: θ = volumetric water content, dimensionless, t = time [T], x,y,z = dimensional coordinates [L], ρ = density of fluid [ML–3], and qx,qy,qz = Darcy’s velocity in three dimensions [LT-1].

25

For constant density, the ρ term cancels out on both sides:

tθ

z)q(

y)q(

x)q( zyx

∂∂

=

∂

∂+

∂

∂+

∂∂

− (Eq. 3.8)

A shortened form of the above equation can be written using vector calculus as follows:

tθqq ∂∂

=∇=− div (Eq. 3.9)

Stated in words, divergence of q (Darcy’s velocity) as expressed on the left side, is equal to

mass storage with time. Equations 3.8 and 3.9 are different forms of the continuity equation and are used for unsaturated water movement modeling.

Richards Equation is obtained by substituting for Darcy’s velocity (q) in the continuity equation for one dimensional flow. The following equation is the most common representation of the Richards Equation:

−∂∂

∂∂

=∂∂

=∂∂ 1

zhK(h)

zthC(h)

tθ (Eq. 3.10)

where:

C = hθ

dd = soil water capacity [L-1],

z = vertical coordinate taken positive downward [L], h = pressure head = pore water pressure per unit weight of water, and K = hydraulic conductivity [LT-1].

Richards Equation can also be written for two or three dimensions. For one dimension, the

solution, θ(z,t) describes the pressure head at any vertical point in flow field at any time. The solution requires knowledge of the characteristic curves for K as a function of pressure head. Another form of Richards Equation is the nonlinear diffusion convection equation for moisture content θ:

−= )()( θθθ K

dzdhD

dzd

dtd (Eq. 3.13)

where: D(θ) = moisture diffusivity [T-1].

26

4. Measurement Techniques for Water Movement in Pavements Measurement techniques and details on sample preparation and testing equipment for pavement

studies are described by Ahmed (1993) as part of a comprehensive field and laboratory study on pavement drainage. The following discussion is a summary of measurement approaches for estimating moisture content, hydraulic conductivity, and pore water pressure in pavements. Where available, precision, accuracy and procedures for measurement techniques are briefly described and appropriateness for pavement studies and observed difficulties of methods are noted.

4.1 Moisture Content Measurements One of the major variables required to describe water movement in pavements is moisture

content. The moisture content provides an indication of whether the media is saturated or unsaturated. The saturation, in turn, dictates the mathematical model used for computing water flow. Understanding subgrade water content variability and its affect on pavement response is crucial for finding solutions that will extend pavement life. However, because of the lack of accurate, precise and continuous measurements, the moisture content in pavements and its relationship to pavement integrity is still not well understood (Rainwater and Yoder, 1999).

Several methods are available for measuring moisture content. Some methods, such as the gravimetric method, measure water content directly whereas neutron scattering, gamma attenuation, x-ray attenuation, nuclear magnetic resonance, frequency domain reflectometry (FDR), time domain reflectometry (TDR), electrical resistance block, and thermal conductivity measure a sample property that can be related to water content. Van der Aa and Boer (1997) note that gravimetric and nuclear surface probe measurements are the two most widely used methods to determine moisture content in the Netherlands. Literature summary by Rainwater et al. (1999) reported that early researchers used time consuming and destructive methods such as extensive coring to determine water content and density values of subgrades which were gradually replaced by a variety of methods, including gypsum blocks, tensiometers, and neutron probes to measure soil water. Ahmed et al (1993) concluded that due to deterioration in constant wet or salty conditions, the gypsum block method is not appropriate for pavement moisture studies. Moisture content can also indirectly be determined if soil suction is measured and the characteristic soil moisture retention curve is known. Soil suction methods are described in section 4.3. Among all methods, only those most commonly used will be discussed here. A comparison of moisture content techniques is provided in table 4.1.

The gravimetric method is based on weighing a moist soil sample and drying it in an oven (105oC) until a constant weight is achieved. The moisture content is calculated from the ratio of the weight loss during drying to the dry weight of the sample. If the volume of the sample and density of the water is known, volumetric water content can also be calculated. One disadvantage of the gravimetric method is that it requires discrete field sampling. The sample needs to be removed from its location for measurement. The other disadvantage is that only point samples can be measured and at one instant in time. Spatial and temporal continuous measurements using the gravimetric method is nearly impossible. For precise measurements, errors arising from the oven as well as the balance should be minimized and accounted for as described by Gardner (1987). The gravimetric method is the most common method for measuring water content and is often used to calibrate other measurement techniques.

27

In the neutron scattering or neutron thermalization method a probe acts as a radioactive source and detector. High energy (fast) neutrons are emitted into the soil, which bounce off of soil and soil moisture reducing the energy level of the neutrons. Hydrogen atoms are much more effective at reducing the energy level of neutrons making this technique more sensitive to moisture content. The proportion of neutrons returning to the probe and reduction in neutron energy is related to the water content. This method is nondestructive and can do profiling as well as continuous measurements. However, the neutron scattering method is based on radioactive decay and thus other radioactive elements may interfere if they are present in the medium. Effective absorbers of neutrons such as boron, lithium, cadmium, iron and chlorine may also interfere by slowing the neutrons.

Moisture content determination with TDR is based on the measurement of the velocity of an

electromagnetic signal through a probe inserted in the soil. The measured velocity of electromagnetic energy is used to calculate the apparent dielectric constant, which in turn is used to calculate the moisture content. Most soil minerals have a dielectric constant of less than five, whereas water has a dielectric constant of about 78 (Stephens, 1996). Various empirical equations are reported in the literature for calibration of TDR probes because of differences in soil media. Examples of calibration equations that relate TDR output to moisture content are shown in table 4.2.

In the U.S., TDR seems to be the most common technique for measuring water content in

pavement structures and subgrades. Possibly the major reason for its increasing popularity is that depth profiles and continuous measurements of water content can be rapidly obtained using TDR. Several studies measuring water content values in subgrades, base and subbase layers, and concrete (PCC) and mortar have shown promising results (Rainwater and Yoder, 1999; Janoo et al. 1999; Janoo et al. 1994; Look et al. 1994). These studies suggested that to improve reliable data acquisition using TDR more research is needed on cumulative effects of vibrations caused by heavy traffic on the measurements and the equipment; influence of dissolved salts on the measurements; calibration equations for pavement materials and subgrade; and probe failures during installation.

28

Table 4.1 Comparison of moisture content measurement techniques (Stephens, 1996(a); Klute, 1986(b); ASTM, 2000(c), AASHTO, 1993(d)).

Method Standard Direct/ Indirect

Measured variable Requires calibration to Challenges Precision (%) Comments

Gravimetric (direct heating)

ASTM D 4959-89, AASHTO T265-01

Direct Mass of water - Uneven temperature distribution in oven 0.1-1b

Most basic method. Values from other methods are typically compared to gravimetric

measurements

Gravimetric (microwave)

ASTM D 4643 -93

AASHTO T 255-92

Direct Mass of water - Uneven heating within sample

± <0.96d

Values are within 0.61% of those measured by

convective heating ovens

Faster than the direct heating method.

Neutron scattering

(thermalization)

ASTM D 3017-96, ASTM D 5220-92

AASHTO T 239-91 Indirect

Electrical pulse count

ratio Soil

Requires uniform contact with soil, cannot be used

within 30cm deep of surface soil. Other neutron absorbers such as Fe, Bo, Cl

may interfere, casing material,

borehole diameter effects calibration

± 0.7a

± 0.2d

(+) Repeatable; temporal data (-) Does not detect discontinuities in water

content, averages over an area (-) Radioactive: license is required

(-) Can not be used within 30cm of land surface

Gamma attenuation - Indirect

Photon count from gamma

ray detector → Bulk density

Casing, source intensity, mud density, detector sensitivity

(+) Nondestructive (+) High resolution (1cm) depth profile

(-) Radioactive; license is required (-) Uncommon in the field because drilling, casing and equal-distancing of two holes are

required

FDR - Indirect Dielectric constant

Soil (non-linear calibration curve) Close fit required ± 0.2a (by volume)

(+) Nondestructive (-) Not sensitive at high saturation

(-) Not recommended for saline environments

TDR - Indirect Apparent dielectric constant

Zero frequency conductivity, texture structure Close fit required ±1-10a depending on

probe

(+) Continuous monitoring (+) Nondestructive (+) Rapid response (+) Depth profile

Electrical resistance block - Indirect

Electrical resistivity or conductivity

Soil Sensitive to electrolytic solutions >± 2%

Thermal conductivity - Indirect (-) Insensitive at high saturation

(-) No longer commonly used

29

The use of TDR to measure water content in pavements has become widespread since 1989. However, proper application procedures for pavements have been the subject of debate. Janoo et al. (1999) noted that earlier correlation equations relating the dielectric constant and moisture content developed for soils by Topp et al. (1980) produce erroneous results for concrete and mortar2. Janoo et al. (1999) reported two improved correlations for concrete and mortar and also note that the amount (and possibly size) of aggregate and cement used in the pavement affects the correlation. Diefenderfer et al. (2000) tested two commercially available TDR probes, CS610 and CS615, and noted that neither Topp’s equation (1980) nor the equations provided by the manufacturer matched the subbase material calibration model developed by them in the laboratory (see table 2.2). Furthermore, Diefenderfer et al. (2000) suggested that the composition of the pavement structure may have an affect on the moisture measured in the subbase layer and note that the two probes tested may yield different data in some circumstances. Rainwater and Yoder (1999) noted that some of the failures and discrepancies of the TDR method may be overcome if probes are installed during construction instead of after construction. Finally, there is evidence that TDR yields more reliable results in partially saturated conditions than in fully saturated media (van der Aa and Boer, 1997). Water content measurements in pavements using TDR seem to be a promising method; however, more experience and research is required to ensure its applicability for extended periods (probe survival), in saline environments (salting in winter), and different seasonal conditions (freezing and thawing).

A technology gaining wide acceptance in pavement site characterization studies is the cone penetrometer test (CPT), which can provide a relatively quick method to obtain properties of the subgrade throughout the construction project. In the CPT, the cone remoulds soils and pushes it aside as it penetrates the ground and slightly modifies the in situ conditions near the probe. Procedures for advancing an electric friction cone penetrometer attached to a truck-mounted drill rig unit are described in ASTM D3441-86. There are promising attempts to relate cone resistance to water content in field studies of subgrades below the wheel paths of lanes and shoulders (Houston et al., 1995). A wide range of sensors can be attached to the cone penetrometer (laser induced fluorescence for fuel fluorescence detection, electrical resistivity, soil moisture, temperature, pH, oxidation-reduction potential, Raman spectroscopy, seismic wave speeds and damping, soil gas, pore water pressure sensor for groundwater flow detection, ground penetrating radar, electrical resistivity tomography and ground penetrating radar tomography). The moisture probe is one of the most promising applications. Mitchell and Shinn’s (1998) probe design greatly reduces disturbances (water relaxation effects due to polarization) that prevented earlier researchers from accurately measuring the dielectric constant. The probe consists of four concentric rings placed along the penetration rod with insulators in between. The outer rings determine soil resistivity; the inner rings measure the capacitance with a transistor oscillator operating at 100MHz. Once the dielectric constant is measured, an empirical equation, similar to those used in TDR, can be used to estimate soil volumetric water content or the probe can be calibrated by other measurements of the soil water content using other techniques. Mitchell and Shinn (1998) reported that their CPT-moisture probe can provide moisture data within ± 5 percent of the real soil moisture. Ease of operation and continuous data are the major advantages of the new technique. Currently, this method is limited to testing subgrade soils only.

2 Mortar contains no coarse aggregate and more paste than concrete.

30

Table 4.2 TDR Equations

Source Equation Special Conditions Comments

Topp et al. (1980) θv=-5.3 10-2 + 2.92 10-2 εa-5.5 10-4 εa

2+4.3 10-6 εa3 General, not stated Seems to work for crushed concrete (van der Aa and Boer, 1997).

Accuracy is 1.3% for mineral soils, <10% for clay subgrades. θv(t)=-18.7 10-2 + 3.07 10-2 t- 3350 10-4 t2 Electrical conductivity <1.0dS/m

θv(t)=-20.7 10-2 + 9.7 10-2 t- 2280 10-4 t2 Electrical conductivity <1.8dS/m

Diefender et al. (2000) Probe CS615 Manufacturer’s calibration equations

θv(t)=-29.8 10-2 + 36.1 10-2 t- 96 10-4 t2 Electrical conductivity <3.0dS/m

Do not work for base materials (Diefenderfer et al., 2000).

Diefender et al. (2000) Probe CS615 Laboratory calibration model

θv(t)= 515.72 – 1474.6 10-2 t – 1389.6 t2 +444.04 t3, R2 = 0.9953

Averaged for all conductivities Developed for base materials.

Ledieu et al. (1986) θv =0.1138 ε1/2 – 0.1758 Not stated. Birgisson and Roberson (2000) note that this calibration is suitable for granular base materials

Janoo et al. (1999) θv =-4.425 + 1.146 ε + 0.0001928 ε2 Not stated Works for the particular PCC used in the study

θv= volumetric water content, εa= apparent dielectric constant, t = probe output expressed as time in milliseconds.

31

4.2 Hydraulic Conductivity Measurements Hydraulic conductivity is a key property that affects water flow through porous media. Water

flow rates are directly proportional to hydraulic conductivity of the medium as stated in Darcy’s Law. A significant portion of this literature review (chapters 6, 7, and 8) is devoted to hydraulic conductivity of materials used in highways. Hydraulic conductivity measurement techniques discussed in this section point out some of the difficulties in estimating in situ hydraulic conductivities of highway materials.

Hydraulic conductivity measurements of pavement materials are crucial to estimating pavement drainage capacity and thus water movement. A common laboratory technique to measure saturated hydraulic conductivity is the constant head permeameter method. A permeameter allows measurement of hydraulic gradient and specific discharge, both of which can be related to K, using Darcy’s Law. In the constant head test, a constant head drop is applied to the soil sample, and the resulting seepage quantity is measured. The constant head test is used primarily for coarse-grained soils (clean sands and gravels) with hydraulic conductivities equal to or greater than 10-3cm/s (Bardet, 1997). In soil physics and groundwater fields, both constant and falling head methods are practiced (Fredlund and Rahardjo, 1993). The ASTM standards for these tests are D5084-90 (reapproved 1997), D 5856-95, and D 5084. In the falling head test, the head is not fixed. The head falls too rapidly to be measured if the soil has a hydraulic conductivity greater than10-3cm/s. Thus the falling head test is generally used for less permeable soils with hydraulic conductivity values less than 10-

3cm/s (Bardet, 1997). Both the constant and the falling head tests assume presence of laminar flow conditions. If tests are conducted under turbulent conditions, then modifications of Darcy’s Law (see section 3.1) as described by Huang et al. (1999) and Fwa et al. (1999), can be used.

Another method for measuring hydraulic conductivity in the laboratory is the pulsed hydraulic

conductivity test, which can be used for samples with low hydraulic conductivities. In this method, a jacketed sample is confined between two pressurized reservoirs, which contain the penetrating fluid. The pores of the sample are also filled with the fluid. When the experiment begins, the pressure in one of the reservoirs is suddenly changed to a higher or lower value and the resultant pressure changes on the high or low pressure side are measured as a function of time. Hydraulic conductivity is calculated using a non-linear partial differential equation with time and distance variables as described in Roy et al. (1992). The advantage of the pulsed hydraulic conductivity test is that it is based on a pressure model instead of Darcy’s Law, which is valid only under laminar flow conditions. Roy et al. (1992) demonstrated the applicability of the pulsed hydraulic conductivity test to concrete specimens. This test is not currently widely used.

Jones and Jones (1989) showed that the hydraulic conductivity values indicative of normally

operating (i.e. horizontally draining) pavement drainage layers should be determined by testing at low hydraulic gradients duplicating those found in the field. Thus, when using a permeameter, the flow should be adjusted to low specific discharges and low hydraulic gradients where flow is laminar. Significant deviation from laminar flow and Darcy’s Law occurs at a Reynolds number (Re) = 10 (see figure 3.1). AASHTO recommends (test procedure T-215) that the hydraulic gradient be between 0.2 and 0.5 to obtain a laminar flow. The hydraulic gradient value suggested by Jones and Jones (1989) is less than 0.075 for a pavement with a cross fall of 1:40 (vertical: horizontal).

32

Similarly, laboratory results by Tandon and Picornell (1997) suggest that the flow is turbulent at hydraulic gradients greater than 0.05.

gradient Hydraulic1 dh/dl

Re=10

Re=1

Hyd

raul

ic g

radi

ent

Discharge

Figure 4.1 Schematic curve show

Standard methods for meahead test are described in ASTMstandard laboratory procedures afor evaluating field hydraulic cohave been made to better simulaboundary conditions, minimizin(Randolph et al., 1996a). Furtheconductivity instead of vertical hsubject to horizontal flow ratherhydraulic conductivity of porousfor 65 pairs of samples. Moynah(2000) presented a more detailedinnovative approaches for measu

To compute pavement draimportant than estimating the hyhydraulic conductivity determinfacilities. In unsaturated conditiconductivity may be a more descpercent saturation remaining undthe pavement are filled with watHowever, once the rain event is with water under negative (capilresidual moisture. If the soil sucthe capillary reservoir in the baspresence of potential gradients b

y velocitDischargeqK==

velocity

ing deviation from Darcy’s Law at turbulent flow (Re=Reynolds number; adapted from Bear, 1979).

suring hydraulic conductivity of granular materials by the constant D2434-68 (2000) and AASHTO T215-70 (1993). However, the nd equipment for measuring base and subbase material are not suited nductivities. Improvements in laboratory measurement techniques te field conditions, including improvements on simulating field g sidewall leakage, and reducing the effect of sample variation rmore, more recent approaches measure horizontal hydraulic ydraulic conductivity since bases and subbases are predominantly

than vertical flow in the field. The ratio of horizontal to vertical media may range from 1 to 42 according to Muskat’s (1937) results an and Sternberg (1974), Randolph et al. (1996), and Li and Chau summary on horizontal hydraulic conductivity measurements and rement of hydraulic conductivity of concrete and base layers.

inage, evaluating the water retention capacity may be equally or more draulic conductivity (Tandon and Picornell, 1997). The saturated es the speed of seepage through the base toward side drainage ons, the water retention capacity in lieu of unsaturated hydraulic riptive parameter. Water retention capacity may be expressed as er a certain negative pressure. In a rainfall event, when the pores of

er, the water moves downward under a unit (gravity) gradient. over, many of the pores in the base layer begin to desaturate (drain), lary) pressure. This process results in an unsaturated medium with tion capacity of the subgrade layer is greater than the base layer, then e layer may provide water to the subgrade layer. Thus, in the etween base and subbase layers and subgrade soils, the drainage

33

capacity may depend not only on hydraulic conductivity, but also on water retention capacity. Tandon and Picornell (1997) provided a detailed discussion of water retention capacity measurements in pavements.

It is well accepted that unsaturated conditions exist in pavements. Yet many laboratory measurements are conducted under a more permeable, saturated condition (falling or constant head techniques), which is rarely achieved in the field. Elsayed and Lindly (1996) noted that designers often account for this inconsistency by using laboratory hydraulic conductivity values greater than the design range. Measurements of unsaturated hydraulic conductivity in pavements have also been studied, although not as extensively. The unsaturated hydraulic conductivity can be measured by observing recovery of pore water to hydrostatic equilibrium in a permeameter (El Tani, 1991) or by using moisture content-matric suction relationship of soils using ‘Tempe’ cells (Elzeftway and Dempsey, 1976). A Tempe cell is a short soil column used for determining soil water retention curves. Once the soil moisture retention curve is determined, unsaturated hydraulic conductivity can be estimated using an empirical equation or a statistical model. Empirical equations relating water content to suction (see table 3.1) can be rewritten to express hydraulic conductivity in terms of water content or suction. Another approach is to use a statistical model based on the fact that both the permeability function and the soil-water characteristic curve are primarily determined by the pore-size distribution of the soil. Both of these methods are mathematically involved as described by Fredlund and Rahardjo (1993).

There are no established methods for determining the in situ hydraulic conductivity of layers in pavement structures. However, there are promising techniques. Constant head or falling head tests may be used for measuring saturated hydraulic conductivities near the middle of the PCC slab or near the edge of the pavement. Kozeliski (1992) suggested a qualitative and a simple quantitative method for testing the hydraulic conductivity of the base layer in the field. In the qualitative method, a gallon of water is poured through a 5-cm-diameter pipe held 1.2 cm above a base surface and the spread of water is observed as it penetrates the base. Bases with high hydraulic conductivities do not allow water to spread much before percolating. The quantitative method entails placing a 5-cm-diameter pipe directly on the base surface and filling it with water. Percolation of the water through the base layer is computed by the rate of the decreasing water level within the pipe. Koch and Sandford (1998) used a modified version of this approach for measuring infiltration through cracks in asphalt concrete. Koch and Sandford (1998) connected a 1 cm diameter pipe to a graduated cylinder and placed it in a bigger pipe (30 cm diameter) full of water to minimize edge effects. Use of a smaller diameter pipe (1 cm) minimized loss of water through evaporation and allowed measurement of infiltration into the cracked asphalt concrete in less than an hour. More recently, Cooley and Brown (2000) compared four different falling head field permeameters and a laboratory falling head permeameter on multiple hot mix asphalt pavements at saturated conditions. They documented that the hydraulic conductivity measurements from all five devices were within one order of magnitude of each other and most displayed similar repeatability. Laboratory measurements did not necessarily yield higher permeabilities all the time. Cooley and Brown (2000) noted that field hydraulic conductivity was likely to yield higher values since water is not confined to one dimensional flow (flow in horizontal, vertical or both directions is possible) in field measurements.

34

4.3 Pore Water Pressure

Suction or pore water pressure is a stress variable that expresses the attraction of soil for capillary water. Many different approaches exist for measuring pore water pressure or soil suction. For example, pressure plate apparatus, tensiometers, heat dissipation sensors, pressure membrane apparatus, centrifuges, gypsum blocks, and fiberglass moisture cells are used to measure matric suction. Matric potential is just one component of the total potential of the soil moisture. Other major components of total potential are osmotic suction and gravitational potential. To measure total suction, filter paper, thermocouple psychrometers, and vacuum desiccators may be used. A comparison of measurement techniques and more detailed explanations can be found in studies by Tsai and Petry (1995), Fredlund and Rahardjo (1993), and Van der Raadt (1987). A summary table including ranges for measurement capabilities is provided in table 4.3. Because there is no accepted reference value for measurement of pore water pressure, bias of each method is not included (ASTM, 2000). Of the aforementioned techniques, only tensiometers and thermal conductivity sensors have been used for pavements studies.

One of the most common devices used for measuring matric potential in the field is a

tensiometer. A tensiometer consists of a fine porous ceramic cup connected by a tube to a vacuum gage. The entire device is filled with deaired water. The porous tip is placed in intimate contact with the soil and the water flows through the porous cup (in or out) until the pressure inside the tensiometer cup is in equilibrium with the pore water in the soil. The reading on the pressure measuring device, once corrected for the water column in the device, is the matric suction. The water pressure that can be measured by this method is limited to approximately –90 kPa (Fredlund et al., 1991), otherwise water will begin to boil inside the tensiometer.

For pavement studies, thermal conductivity sensors are applicable. Loi et al. (1992) showed

that long-term stable and reliable matric suction readings in highway subgrades can be obtained using thermal conductivity sensors, as long as the sensor is not subjected to prolonged positive pore water pressures. A thermal conductivity sensor consists of a porous ceramic block containing temperature-sensing elements and a miniature heater. Once heat dissipating from the heater element causes the temperature to rise at the temperature-sensing element, the rise depends on the water content in the porous block. The water content can be correlated with matric suction using a pressure plate technique. This is an indirect method since the soil moisture retention curve is required. Fredlund and Rahardjo (1987) and Fredlund et al. (1991) noted that thermal conductivity sensors are suitable for field use including freezing conditions.

35

Table 4.3 Methods for measuring pore water pressure.

Technique Standard Suctioncomponent measured

Precision Range (kPa) Comments

Psychrometers AASHTO T 273-86 (1993) Total 100- ~8000* ϒ

50-3000∞ Constant temperature environment required

Filter paper ASTM D 5298-94 Total Entire range* May measure matric suction when in good contact with moist soil. Compared to psychrometers and pressure plates exhibits more scatter of data (Tsai and Petry, 1995) Simplest and least expensive of all indirect methodsς

Labor intensive, can not be automatedς

Tensiometers ASTM D3404-91 (Reapproved 1998)

Matric 0.10-0.19kPa 0-90* Difficulties with cavitation and air diffusion through ceramic cup Can be effected by soil gas pressure, temperature and overburden pressureς

Pressure plate ASTM D2325-68 (Reapproved 1994)

Matric 0-1500* Range of measurement is a function of the air entry value of the ceramic disk

Heat dissipation sensor

Matric ± 10kPaς 0 - ~ 400* >100ϒ ∏

10-1500∞

Indirect measurement using a variable pore size ceramic sensor Maximum sensitivity of response occurs 0-1000kPaς

Pore fluid squeezer

Osmotic Entire range Used in conjunction with a psychrometers or electrical conductivity measurement

*

Electrical resistance blocks

Nylon and fiberglass blocks most sensitive in the range of 0 to -100kPaς

Gypsum blocks sensitive at potentials below –-30 kPaς

*Fredlund and Rahardjo (1993) ϒChandra et al. (1989) ∏Picornell et al. (1983) ∞ Tindall et al. (1999) ς Stephens (1996)

36

4.4 Rainfall The amount of water in a pavement system is dictated by meteorological conditions such as

rainfall and snow. As mentioned earlier, the major moisture ingress route is from the pavement surface after a rainfall event. Thus, it is necessary to be able to measure and account for rainfall events.

There are a few different techniques for measuring rainfall. The most common device is a recording or manual rainfall gauge that collects falling rain on a standard area. Rainfall can be measured at a precision of 0.25 mm using a standard rain gauge. A recording rainfall gauge is a tipping bucket that has two reservoirs. When one reservoir is full, it tips and the other reservoir begins to fill. Each tip is recorded electronically and the data can be downloaded. Tipping buckets are also used to measure edge drain outflow (Ahmed et al., 1997; Birgisson and Roberson, 2000). Other emerging technologies for rainfall measurement are ground-based radar and satellite imagery. In pavement studies, the most common method for estimating rainfall intensity is either the rainfall gauge or using meteorological data provided by the National Oceanic and Atmospheric Administration.

Chapter 4 Synopsis For purpose of this study, the water related parameters of concern are moisture content, hydraulic conductivity, suction and rainfall. Field and measurement techniques for these variables and their applicability to pavement studies were discussed. For moisture content, the technique that is gaining popularity in pavement studies is the TDR method, which can be automated and is not extremely destructive. However, there is not a universal agreement on a TDR calibration equation. Many studies use the common Topp et al. (1980) equation even though it may not be accurate for base and subbase materials. To ensure quality data, TDRs may be installed before construction and may be calibrated for specific composition of the tested material. The laboratory techniques for measuring hydraulic conductivity are constant head and falling head methods, both of which measure saturated hydraulic conductivity. The concern about these techniques is the presence of turbulent conditions under which Darcy’s Law no longer applies. Ideally, tests should be conducted at low hydraulic gradients and flows that simulate field conditions. However, if turbulence is observed, pseudo-hydraulic conductivity can be measured by models discussed by Huang et al. (1999) and Fwa et al. (1999). To simulate horizontal flow towards edge drains, horizontal hydraulic conductivity, in addition to vertical hydraulic conductivity should be measured. For pavement structures, there are no established field techniques for measuring saturated hydraulic conductivity, and discussion of measurement techniques for laboratory and field unsaturated hydraulic conductivity was found to be minimal in the literature. Information on most common techniques for measuring pore water pressure in pavement structures was also sparse. Most of the studies suffice by measuring moisture content only. Loi et al. (1992) discussed applicability of thermal conductivity sensors for pavement studies. Precipitation also needs to be measured to quantify drainage or specify climate conditions of pavements. Mostly, a tipping bucket is used for measuring rainfall or edge drain outflow. The common aspect of all measurement techniques for all variables mentioned is the absence of standardization. Sometimes States have their own standards, however, a nationwide standardization is required to ensure data quality; most importantly consistency.

37

5. Moisture Content in Pavements

5.1 Groundwater Table Effect Groundwater conditions may affect the moisture in pavement systems and may be the major

factor influencing subgrade water content if the ground water table is within approximately six meters from the surface (Yoder and Witczak, 1975). Capillary water and water vapor may migrate towards ground surface and increase the moisture content especially in subgrades. Development of a perched water table may also increase head buildup in subbase layers (Ahmed et al. 1993). Ksaibati et al. (2000) reported that lower groundwater table depth results in lower moisture content in base and subbase layers as shown in figure 5.1. Similarly, Chu et al. (1972) found a somewhat positive correlation between subgrade moisture content and high groundwater table for pavement systems in South Carolina as illustrated in figure 5.2.

The affect of the ground water table on the moisture content of the overlying compacted aggregate material is further supported by laboratory experiments. Jessep (1998) compacted coarse and fine subbase materials in a tube. Figure 5.3 shows that after partial immersion of the tube in water for seven days, the moisture content of the subbase material within the first 25 cm above the water table increased. Jessep (1998) estimated that at a height of 25 cm above the water table, the moisture content rose to a value of 50 percent of the saturation value for a subbase rich in fines, and to a value of 12 percent for coarser subbase mixtures. These laboratory results supported the argument that a granular material with fines may more readily act as a sponge (due to capillary action) than as a drain (Dawson, 1985).

00.20.40.60.8

11.21.41.61.8

5 6 7 8 9 10 11 12 13 14 15 16 17 18 19

Moisture (%)

Wat

er T

able

Dep

th (m

) BaseSubgrade

Figure 5.1 Water level effect on base and subgrade moisture, Route 62, Florida (adapted from Ksaibati et al, 2000).

38

0

2

4

6

8

10

12

M-68

J-68

A-68

O-68

D-68

F-69

A-69

J-69

A-69

O-69

D-69

F-70

A-70

J-70

A-70

O-70

D-70

F-71

A-71

J-71

Subg

rade

moi

stur

e co

nten

t (%

) 0

5

10

15

20

25

30

35

40

45 Dep

th o

f gro

undw

ater

tabl

e (m

)

5

7

9

11

13

15

17

19

21

23

Subg

rade

moi

stur

e co

nten

t (%

)

0

5

10

15

20

25

30

Dept

h of

gro

undw

ater

tabl

e (m

)

Subgrade moisture content

Depth to groundwater table

Figure 5.2 Variations in moisture content with fluctuations in groundwater table (adapted from Chu et al., 1972).

0

25

50

75

100

3 5 7 9 11 13 1

Moisture content (%)

Capi

llary

ris

e (c

m)

5

Figure 5.3 Moisture content and capillary rise in granite that is a typical subbase material rich in fines (adapted from Jessep, 1998).

39

5.2 Temporal and Spatial Variability

Moisture varies temporarily and spatially as a response to environmental factors. During construction, compacting to 95 to 100 percent of dry unit weight and at optimum moisture content as determined by AASHTO T99 is standard practice (Thadkamalla and George, 1995). After construction, moisture either decreases or increases until it reaches an equilibrium water content (Look et al., 1994; Alonso, 1998). Field studies suggest that the finer the soil, the greater the difference between the equilibrium and optimum moisture content (Chu and Humphries, 1972). Pore water pressures usually remain negative after they attain an equilibrium with the local system of groundwater flow (Lebeau et al., 1998). The period from after construction to the time of equilibrium is a transient phase. After reaching equilibrium, during the stable period, the water content shows a seasonally cyclic behavior controlled by climate. Seasonal changes of temperature with maximum changes at the surface (a variation of 22oC between months of August and January in Rhode Island) were also observed (Jin et al., 1994). Regardless of climatic conditions, a stable moisture period with slight seasonal variations was always achieved (see figure 5.4).

Stab

le

Zone

Seasonal Changes

Wetting

Drying

Moi

stur

e C

onte

nt

Time (Years) After Construction

Figure 5.4 A sketch of initial and seasonal moisture changes for construction starting compacted dry or wet of equilibrium (adapted from Look et al., 1994 and Alonso, 1998)

Moisture related pavement problems are often amplified in cold regions where frost penetration or freeze-thaw cycles occur. During winter months, water trapped in pavements forms ice crystals that melt when temperatures rise. Because pavements are not homogenous, ice crystals are distributed randomly and result in uneven, differential heave and damage (Lafleur and Savard, 1996). During thawing, moisture content increases and the pavement is weakened. In some regions, subgrade thawing occurs only once whereas in relatively milder areas, continuous freezing and thawing may occur depending on the climate and amount of deicing salts used on the pavement. In cold regions, the thaw period may spread over several weeks if the frost has penetrated more than 3.0 m below the surface (Macmaster et al., 1982). To reduce the frost effect, existing material may be replaced by frost-durable material or an insulating layer within the pavement structure may be preferred due to lower cost (Kestler and Berg, 1995). Geotextiles may be used as capillary barriers to

40

reduce the unsaturated flow of water toward the surface caused by freezing or evaporation (Henry, 1996). Laboratory results suggest that addition of lime to clay soil increases frost resistance and may be an alternative solution for frost susceptible subgrades (Arabi et al., 1989). More information on common and alternative materials for road and airfield insulation and their performances can be found in studies by Dore et al. (1995) and Esch (1995). Xu et al. (1991) presented mathematical models of water flow and heat transport in frozen soil.

Moisture conditions in pavements can be spatially variable. Roads are three-dimensional structures. Longitudinally, major differences arise from changing subgrade properties and water table levels. When these two variables remain constant, longitudinal variability is likely to be minimized. Variability of moisture conditions among horizontal components of pavements such as surface layer, base, subbase and subgrade depends on material type, crack formation, and proximity to the groundwater table. The transverse variability is often summarized by the term ‘edge effects’. Typically, close to edges, moisture contents are higher due to greater infiltration. More frequent occurrence of cracks along and closer to shoulders typically indicate edge effects. A study by Gordon and Waters (1984) emphasized laterally heterogeneous moisture conditions and their consequences by stating that the performance of the pavements is dictated by edge effects, regardless of the pavement thickness and shoulder details. In the presence of a paved shoulder, edge effects can be delayed but not stopped.

Spatial variability of moisture content has been observed in the field. Houston et al. (1995)

measured subgrade water contents at 18 different Arizona sites, at three different locations on each site. Two of the locations were 10-m apart, along the right-wheel path of the right lane and the third location was along the adjacent shoulder. The water content data showed significant lateral and vertical variability. On one site, the moisture content varied from 5.5 percent to 15.1 percent on the same location within four vertical meters. At the same depth, the moisture content varied from 15.1 percent to 11.1 percent between a shoulder location and right-wheel path location. Similar variations were observed for many other sites suggesting that within a few meters of depth or horizontal distance there may be significant material type changes, negative pore-water pressure differences, or both.

Janssen (1987) studied depth-related moisture gradients in PCC pavements. Moisture contents

were measured throughout summer and fall by installing psychrometers in three sections (5, 14, and 17 cm deep) of a PCC pavement. Psychrometers measure relative humidity, which can be converted to degree of saturation by means of laboratory calibrations (Janssen, 1985). The data showed that the degree of saturation varied little among different depths with lowest at 82 percent and highest at 96 percent. For mathematical modeling, a finite-difference computer program, developed for moisture movement in soil, was used. An initial degree of saturation of 90 percent, observed in field measurements, was used in the model to simulate water movement for three months. The output of the model showed that surface drying does not extend far into the concrete, possibly due to the low hydraulic conductivity of PCC. Measurements taken deeper than approximately three cm did not reflect surface drying.

41

5.3 Field Observations The equilibrium moisture content depends on several variables; subgrade soil type, level of water table in the vicinity, and condition of the surface layer are the major controlling variables. Thadkamalla and George (1995) note that the equilibrium moisture content was not affected by annual rainfall, and that the degree of saturation corresponding to the equilibrium moisture content of plastic soils exceeds 85 percent. Equilibrium moisture content may not be correlated with climatic differences such as seasonal precipitation (Andrew et al., 1998). An empirical method for estimating the equilibrium moisture conditions is to use the Thornthwhite index, a climatic classification developed in 1948, that relates moisture conditions to climatic variables such as precipitation, evapotranspiration, mean monthly air temperature, and number of hours of daylight per day (Thadkamalla and George, 1995).

Examples of moisture content data are presented in table 5.1. The wide variation of moisture content data suggest that both saturated and unsaturated conditions occur in bases and subgrades. Loi et al. (1992) also noted that subgrades and embankments are almost always unsaturated.

Hall and Rao (1999) investigated factors affecting subgrade moisture in Arkansas and used a monthly (sometimes bi- or tri-monthly) data set from 14 sites collected by the Arkansas Highway and Transportation Department. This approach provided poor time resolution of data. However, by studying numerous field sites, Hall and Rao (1999) were able to correlate some environmental factors with subgrade moisture content. Their observations suggested that precipitation and subgrade moisture content were positively correlated; temperature and subgrade moisture content were negatively correlated; when close to the surface, the groundwater table played a major role in determining subgrade moisture content, not only as a source of water, but also by aiding the establishment of an equilibrium condition between soil moisture and the water table; and soil type may be the single most important factor that affects long-term subgrade moisture content.

42

Table 5.1 Moisture content data in the field

Percent moisture content (v/v)

Degree of saturation Vertical location Geographical location /Material Specification

Comments Author

15-33 Mostly unsaturated 0.15 and 0.75 m Vermont, base (silty sand with gravel), subgrade (sandy silt)

Greater fluctuations at the base layer than the subgrade

Simonsen et al. (1997) Janoo et al. (1994)

18-35 Mostly unsaturated 0.6-1.5 m Tennessee, various subgrade soils

Changes in θm in asphalt stabilized bases (increseased 0.05 units over a precipitation event) and asphalt concrete (showed no change when it rained) were also measured.

Rainwater et al. (1999)

3-28 Mostly unsaturated 0-10 m Arizona, various subgrade soils

No obvious vertical trends. Houston et al. (1995)

37-43 - 0.6-1.8 Australia, expansive clay Embankments Look et al. (1994) 42 85% Subgrade at 0.7 Subgrade at higher

moisture content than base 18-28 Unsaturated Bottom of aggregate base

at 0.5m 15-17 Unsaturated Top of aggregate base at

0.36m

Tennessee

The range is due to seasonal variation with lowest values observed in October

Andrew et al. (1998)

13-30 Mostly unsaturated 0.4-0.8m from surface The Netherlands Crushed concrete base layer

Van der Aa and Boer (1997)

6-18 Mostly unsaturated 3.5-5.2m from surface Virginia Smart Road Test facility, sections B,E,F,G, and H during May-November / 21B aggregate subbase layer

Two different probes yield different temporal results on section B.

Diefenderfer et al. (2000)

22-32 Mostly unsaturated 3-4 m from surface MnRoad test cells 10 and 12

Dense-graded base layer Birgisson and Roberson (2000)

43

Pavement failure may be an indication of presence of water in the pavements. A study conducted in Texas PCC and asphalt pavements analyzed the failure condition of pavements when the annual rainfall varied between 25 and 125 cm (Saraf et al., 1987). A strong positive correlation was found between rate of failure of pavement per mile and annual rainfall. Increasing rainfall resulted in more frequent failures. The age of pavements and the type of clay in the subgrade influenced the extent of the affect of the rainfall on pavement failure. Older pavements (6+ years) were more affected by rainfall than younger (0-6 years) pavements. Pavements constructed over high swelling clays had a higher number of failures per mile than pavements that had low swelling clay as the subgrade.

Chapter 5 Synopsis Chapter 5 focused on the moisture content in pavement structures, its spatial and temporal variability, and the effect of the groundwater table on the moisture content. A wide variation of moisture content (5-30% by volume) was observed while many studies documented both saturated and unsaturated conditions in pavements. As expected, moisture contents in the base or subgrade layers are higher if the groundwater table is shallower. Capillary water overlying the water table not only increases the moisture content of granular materials, but also allows subbase or subgrade layers to act as a sponge rather than as a drain. Variation of groundwater table depth and subgrade material type results from longitudinal variability along a highway environment. Vertically, pavement variability is a consequence of the use of different layers, extent of cracking and proximity to the groundwater table. The transverse variability is also reported in the literature. Shoulders may be a different material than the asphalt concrete or PCC surface layers. Often shoulders drain most of the water because (1) the road surface is crowned, (2) the joints and cracks in shoulders allow easy passage of water, and (3) the base/subbase materials used in shoulders may be free draining. Because of these factors, higher infiltration occurs along the shoulders (edge effects). Temporal variability of moisture content in the pavement becomes seasonally cyclic once the pavement reaches its equilibrium moisture content after construction. Freeze-thaw cycles during winter and early spring may significantly affect pavement performance.

44

6. Hydraulic Conductivity of Asphalt Concrete Moisture damage documented in the form of cracks, stripping, and rutting has prompted

researchers to investigate water movement and retention in pavements. Moisture damage appears to be one of the major concerns for asphalt concrete material, especially in regions that have high rainfall and high water tables. Infiltration of water and other liquids into the asphalt concrete causes softening and stripping of asphalt pavement (Kennedy, 1985). Softening is where there is reduced cohesion, which causes a reduction in strength and stiffness of the asphalt mixture. Stripping is characterized by the loss of adhesion and the physical separation of the asphalt cement and aggregate. The exact mechanisms of the loss of adhesion are not well understood, yet a combination of several properties of the asphalt concrete determines the strength of adhesion and thus indirectly, the hydraulic conductivity of the asphalt concrete. Terrel (1993) listed these properties as surface tension and chemical composition of the asphalt cement and aggregate, asphalt viscosity, surface texture of the aggregate, aggregate porosity and cleanliness, aggregate moisture content, and temperature at the time of mixing with asphalt cement. A combination of these variables determines the susceptibility of asphalt concrete to moisture damage.

Moisture damage is a result of the continued presence of water in the asphalt concrete material. If asphalt concrete cannot be designed to be impervious, then it should allow for drainage. To be considered an impervious asphalt concrete, the air-void content should be less than 6 or 7 percent3 (Kennedy, 1985; Choubane et al., 1998)(see figures 6.1 and 6.2). If a relative density of at least 93 percent of the theoretical maximum density is achieved during compaction, then the penetration of moisture into the mixture can be minimized. If the air void ratio is above 15 percent, then the mixture is considered free draining. In the U.S., the conventional dense-graded asphalt concrete has a medium air void percentage (~ 7-12 percent). However, laboratory studies summarized in figure 6.3, suggest that asphalt concrete is more resistant to water damage at either higher or lower than this medium void range (Terrel and Al-Swailmi, 1993). Thus, both free draining and impervious mixtures retain higher strength and perform better in the presence of water than the medium range commonly used in the U.S.

European researchers have shown that asphalt concrete with both low and high voids may have significant advantages over the medium void range (Terrel and Al-Swailmi, 1993). Porous asphalt pavements with better skid resistance in wet weather are also in use in the United States. For these types of pavements, susceptibility of porous asphalt mixtures to clogging may be of concern if the road is used by vehicles that have dirty wheels or carry earth (Fwa et al., 1999).

Even though linearity is generally assumed, the relation between the percentage of air voids and hydraulic conductivity in asphalt concrete is not linear (Terrel and Al-Swailmi, 1993; Choubane et al., 1998; Cooley and Brown, 2000)(see figure 6.1, 6.2 and 6.4). Experiments conducted under turbulent conditions that measure pseudo-hydraulic conductivity4 also show considerable spread of points (see figure 6.5). The structure and interconnection of air voids may affect hydraulic

3 Choubane et al. (1998) note that the fine-graded mixes are relatively impermeable, even at air voids significantly higher than seven percent, possibly because voids in a fine-graded mix are not as interconnected as compared to a coarse-graded mix. 4 See section 3.1 for pseudo-hydraulic conductivity

45

conductivity noticeably. Possibly, the water flow through larger and interconnected voids (so called macropores; Beven and Germann, 1982) is easier than if the mixture contains small voids that are not connected (see figure 6.6). The shape of the voids may be another variable affecting hydraulic conductivity. Spatial variability of voids within a sample may also contribute to the nonlinearity of the relation between voids and hydraulic conductivity. Vertical variations within the asphalt concrete are observed in the field. For example, Masad et al. (1999) note that more air voids were concentrated at the top and the bottom portions of an asphalt concrete specimen compacted by superpave gyratory compactor, whereas the specimen compacted by the linear kneading compactor had most of the voids concentrated at the bottom. Furthermore, in asphalt mixes, a portion of the air voids is trapped by asphalt and mineral fillers and therefore is not water-permeable (Huang et al., 1999). The effective porosity, which is the difference between the total air voids and the undrainable air voids, may be more meaningful than porosity by itself. Though with fewer data, Huang et al. (1999) present a more linear relation between effective porosity and pseudo-hydraulic conductivity (see figure 6.7) than the relationships shown in figures 6.1, 6.2, and 6.4.

Air

Voi

ds, (

%)

Figure 6.

16.0

14.0

12.0

10.0

8.0

6.0

4.0

2.0

0.0

Hydraulic conductivity (K), x 10-6 cm/sec

1 Hydraulic conductivity-air void content relationship for coarse-graded Superpave mixes. Hydraulic conductivity is essentially zero below 6 percent air void ratio (Choubane et al., 1998).

46

A

ir V

oids

, (%

) 121110

9876543210

0 20 40 60 80 100 120 140 160 180 200 Hydraulic conductivity (K), x 10-6 cm/sec

Figure 6.2 Hydraulic conductivity-air void content relationship for dense-graded asphalt mixes. Note that the range of hydraulic conductivities is lower than that presented in figure 6.1 (Terrel and Al-Swailmi, 1993).

Impervious

More resistant to

water damage

6-7% 12% 15%

Conventional

dense-graded

asphalt concrete

Free draining

More resistant to

water damage

Ret

aine

d M

ix S

treng

th

Air-void percentage

Figure 6.3 Air-void percentage of asphalt concrete material

47

0

50

100

150

200

250

300

0 1 2 3 4 5 6 7 8 9 10

Air void content (%)

Hyd

raul

ic c

ondu

ctiv

ity x

10-5

cm/s

Figure 6.4 Hydraulic conductivity in-place air void content relationship for two different hot mix asphalt pavements (adapted from Cooley and Brown, 2000)

0

100000

200000

300000

400000

500000

600000

700000

0 5 10 15 20 25

Air-void content (%)

Pse

udo-

hydr

aulic

con

duct

ivity

(10-5

cm/s

)

Figure 6.5 Hydraulic conductivity air-void content relationship for porous asphalt mixes (Fwa et al., 1999)

48

Asphalt and mineral fillers: Rock, dust, slag hydrated lime, hydraulic cement, fly ash and loess

Shape and Size of Voids

Interconnectivity

Hydraulic conductivity

Figure 6.6 Hydraulic conductivity as a function of shape and size of voids, presence of asphalt and mineral fillers and interconnectivity.

4

3

3

2

2

1

1

Pseu

do-h

ydra

ulic

con

duct

ivity

(mm

/s)

15 25 35 Effective porosity (%)

Figure 6.7 Hydraulic

Hydraulic co

The data compiled was typically less thydraulic conductiv

0

5

0

5

0

5

0

5

0 5

conductivity versus effective po(Huang

nductivity of asphalt concreteshowed that the hydraulic cohan the hydraulic conductivitity of Superpave asphalt ran

rosity relationship for open and dense-graded asphalt mixes et al., 1999).

varies six orders of magnitude as shown in table 6.1. nductivity of dense graded asphalt (10-2-10-4cm/s) y of porous asphalt concrete (10-2-101cm/s). The ged four orders of magnitude (10-5-10-1cm/s).

49

Table 6.1 Hydraulic conductivity values of asphalt concrete

Sample Hydraulic

conductivity (10-5cm/s)

Air Voids (%) Author

Coarse-graded asphalt Superpave Samples from I-75, Columbia 0-976 4.0-12.1 Coarse-graded asphalt Superpave Samples from I-10, Columbia 2-638 1.9-8.7 Coarse-graded asphalt Superpave Samples from I-10, Escambia 13-927 6.3-11.8 Fine-graded Marshall Mix Field samples 14-52 5.9-8.3 Coarse-graded asphalt Superpave Lab Fabricated Samples 22-145 6.5-8.3

Choubane et al. (1998)

Laboratory prepared asphalt Superpave mixtures 10- 10,000 3.5-15.0 Lynn et al. (1999) Porous asphalt mix specimens 62,900-535,900* 15.6-23.9 Porous asphalt mix specimens clogged with soil 1,300-337,600* Fwa et al. (1999)

Open-graded coarse asphalt mixture 27,000-148,000 Dense-graded asphalt Superpave wearing course specimens 10,200 – 11,600* Dense-graded mix specimens from I-10 176 – 529* Dense-graded mix specimens from I-12 35*

Huang et al. (1999)

Asphalt concrete mixture specimens 0-20 4-11 Terrel and Al-Swailmi (1993) Open-graded friction coarse 2,069 16.7 Open-graded friction coarse with 16% crumb rubber 5,502 15.8 Open-graded friction coarse with mineral fillers 3,203 19.9 Open-graded friction coarse with cellulose fibers 8,552 16.2 Open-graded friction coarse with styrene-butadiene (SB) polymer 1,801 13.9 Open-graded friction coarse with SB and cellulose fibers 8,121 19.2

Cooley et al. (2000)δ

Various hot mix asphalt pavements (Field and lab measurements) 100-16,000α Cooley and Brown (2000) *Psuedo-hydraulic conductivity αVariability within 24m longitudinal to pavement may vary up to one order of magnitude δ Most of the samples had rutting and cracking. Samples are from I-75 Atlanta, Georgia

Chapter 6 Synopsis In chapter 6, the hydraulic conductivity of the wearing course, the asphalt concrete layer, on flexible pavements were discussed and it was noted that: (1) asphalt deterioration such as softening (caused by reduced cohesion) and stripping (caused by loss of adhesion and physical separation of the asphalt cement and aggregate) has prompted researchers to investigate moisture in asphalt concrete and hydraulic conductivity, (2) asphalt concrete becomes impermeable below void contents of 6-7 percent, (3) asphalt concrete is free draining at void contents above 15 percent, (4) both impermeable and free draining asphalt concrete display higher strength and perform better in high moisture conditions than the middle range of void ratios typically used in the U.S., (5) exceptions to points (2) and (3) may occur if fine-graded mixes are used, (6) compaction techniques may affect void content distribution and thus hydraulic conductivity of asphalt concrete, however, field or laboratory measurements ignore its effect, and (7) hydraulic conductivity is not linearly dependent on void content and is a function of interconnectivity of air voids, shape and size of the voids, and the extent of presence of asphalt and mineral fillers.

50

7.Hydraulic Conductivity of PCC The hydraulic conductivity of PCC depends on having a continuous network of pores in the

material. The pores exist in the cementitious matrix of concrete and in the interfacial regions between aggregates and pastes (Roy et al., 1992). The pore structure of PCC is modified during the hardening of fresh concrete. Bakker (1983) described the fresh concrete as a granular structure with continuous capillary pores. During the hardening period the hydration products glue the particles together and block the capillary pores. These processes increase the strength of the material and decrease the hydraulic conductivity. The hydraulic conductivity of hardened concrete also depends on the temperature during hydration. Higher curing temperatures increase the hydraulic conductivity of PCC, but decrease it if by-product blended cement is used (Bakker, 1983). Thus, the type of raw materials (cementitious materials and chemical admixtures) used, the type and extent of chemical reactions during hardening, and curing temperature affect pore size distribution and hydraulic conductivity of PCC.

Roy et al. (1992) summarized several concrete porosity studies and the percolation theory that is often used in describing flow in sedimentary rocks. As explained by Roy et al. (1992), concrete porosity studies show that size distributions in the cementitious component of the concrete can be described in terms of a mixture of log normal distributions. In a mercury porosimetry experiment, the inflection point of a mixture of lognormal distributions corresponds with the inflection point of the cumulative pore size distribution. The inflection point determined by experiment indicates the minimum diameter of pores that are continuous through all regions of the material.

Significance of pore continuity may be better understood once it is related to the percolation theory and the preliminary percolation-hydraulic conductivity model developed by Roy et al. (1992). Percolation theory assumes that in a lattice structure, a point can be vacant or randomly occupied independent of the states of the neighboring sites. Randomly occupied points can be though of as the open pores in a material. Occupied sites may be isolated from each other or connected with neighboring sites to form a cluster. As the fraction of occupied points increase and form bigger clusters the probability of a route (infinite-path) that starts on one side and extends to the other side of the lattice becomes greater than zero. In other words, continuous paths begin to form. Below a percolation threshold, which can be calculated statistically, the possibility of an infinite path is zero. The percolation threshold is called the critical porosity and can be related to the fraction of connected pores (which is proportional to hydraulic conductivity) by a function. Roy et al. (1992) uses such a relation to estimate hydraulic conductivity from porosity and concludes that verification of the model with more data is required to test its validity.

Recycled materials possess properties differing from those of well-known binders. Compared to Portland cement paste, introduction of mineral by-products results in different hydraulic conductivity and pore structure in the concrete. Before and during hardening, the hydraulic conductivity of concretes containing slag or fly ash is greater than that of Portland cement concrete. However, once the reactions are complete the reverse is observed (Berry and Malhotra, 1978). Similarly, laboratory studies by Feldman (1983), Ozyildirim (1998), and Bakker (1983), respectively showed that (1) hydrated blended cement has lower hydraulic conductivity than hydrated Portland cement and that (2) concretes containing a pozzolan or slag have lower long-term hydraulic conductivity than the control, and (3) that blast furnace cement and cement containing fly-ash has

51

lower hydraulic conductivity than Portland cement. Virtanen (1983) and Pigeon and Regourd (1983) noted that the air content of concrete containing slag, fly ash or silica fume is smaller relative to the air content of pure cement mixes. On the contrary, laboratory experiments by Nakamoto (1998) suggest (1) that higher slag content in PCC may result in increased porosity and hydraulic conductivity, and (2) that the water tightness may be improved by utilizing more fine slag. With the exception of Nakamoto’s (1998) results, these studies suggest that addition of mineral by-products to Portland cement mix decreases the hydraulic conductivity of the concrete. Fineness of the by-product may also affect hydraulic conductivity.

Similar to the asphaltic concrete, the hydraulic conductivity of PCC is also more complex than

a simple function of porosity. Nakamoto et al. (1998) noted that hydraulic conductivity of concrete depends on the size, distribution and continuity of pores as well as total porosity. Laboratory studies by Nakamoto et al. (1998) suggest that the hydraulic conductivity of concrete may be more closely related to the pore volume over a certain threshold value of diameter (e.g. 500 or 1000nm) rather than to the mean diameter of pores or total porosity. Table 7.1 presents a compilation of hydraulic conductivity and air void content data for PCC. Typically, the hydraulic conductivity of PCC (<10-

7cm/s) is less than the hydraulic conductivity of asphalt concrete. Table 7.1 Hydraulic conductivity values of PCC

Sample Hydraulic

conductivity (10-5cm/s)

Air Voids (%) Author

Cements containing 0, 28.3, and 66% slag 4.2-20 Pigeon and Regourd (1983) Concrete mix (cement, slag, fly ash, silica fume, no aggregate) 1-7 Virtanen (1983)

Cement concrete after curing for 56daysβ 0.000001-0.0000026 3.9-6.1 Cramer and Carpenter (1999)

Fly ash blended cement ~35 Feldman (1993) No cracks, 25mm thick sample 0.00005 No cracks, 50mm thick sample 0.0002 50µm crack, 25mm thick sample 0.0003 50µm crack, 50mm thick sample 0.0006

High strength concrete

250µm crack, 25 and 50mm thick samples

Aldea et al. (1999)

0.003 βResults may not be representative since such low hydraulic conductivities are difficult to measure

Chapter 7 Synopsis Chapter 7 extended the discussion on hydraulic conductivity of asphalt concrete to hydraulic conductivity of PCC and presented hydraulic conductivity data for both types of concrete. Air void content of asphalt concrete and PCC varies from 3 to 35 whereas the range of hydraulic conductivity was five orders of magnitude with the exception of data from Cramer and Carpenter (1999) that would increase the span of the range to eleven orders of magnitude. The hydraulic conductivity of asphalt concrete is a function of size, shape and interconnectivity of voids in addition to raw materials used (see chapter 6). In PCC, the hydraulic conductivity also depends on the type of raw materials (chemical materials and chemical admixtures) used and, in addition, the type and extent of chemical reactions during hardening and the curing temperature. Noting that porosity may not relate closely to hydraulic conductivity, Nakamoto et al. (1998) suggested that the use of the pore volume over a certain threshold value of diameter to estimate hydraulic conductivity. Finally, the hydraulic conductivity of PCC when recycled materials are used was discussed and some contrary findings were noted. Before and during hardening, hydraulic conductivity of PCC containing slag or fly ash seems to be greater than PCC without recycled materials. Once the reactions are over, the air content and hydraulic conductivity of PCC containing recycled materials is lower. On the contrary, Nakamoto et al. (1998) reported that higher slag content in PCC might result in increased hydraulic conductivity and porosity. Nakamoto’s results suggest that authors should specify the fineness of recycled material used since it affects hydraulic conductivity.

52

8.Hydraulic conductivity of Bases / Subbases / Embankments Hydraulic conductivity of base layers is an important consideration for water movement, drainage and pavement service life. Early pavements were constructed with dense-graded base materials that had poor drainage. In the past, dense-graded aggregate bases, cement-treated bases, and asphalt-treated bases were used because they were strong and non-erodible, and at that time roads were built with emphasis on strength and not on drainage for performance (Mathis, 1990). During the late 1960s, the ACOE discovered that the subsurface pavement layers of most pavements remained near or in a saturated condition causing premature failing of pavements (Grogan, 1994). By the early 1970s, it was well recognized that in the conventional base layer; a dense-graded, well compacted base course material with more than five percent passing the number 200 sieve, would not meet the Corps’ criteria for drainage (Moynahan and Sternberg, 1974). Thus, interest was directed to designing more permeable bases.

Recently, the need to drain excess water in pavements has been better appreciated. It was recognized that the amount of damage per load application is roughly 10 to 20 times greater for a pavement with a saturated base than for the same pavement with an unsaturated base (McEnroe and Zou, 1993). Recent pavement design guidelines include a permeable base layer that serves to provide a medium to remove the water that enters the pavement. Higher hydraulic conductivity of base materials compared to subgrade materials allows water to flow by gravity to a collection system. Monitoring of untreated permeable bases constructed by the ACOE shows that the permeable drainage layers drain the excess water adequately and rapidly (Grogan, 1994).

A permeable base is characterized by an open-graded, crushed, angular aggregate with

essentially no fines (Mathis, 1990) (also see section 2.3). Depending on the required drainage capacity of the unbound permeable layer, open-graded or rapid-draining materials can be used (Freeman and Anderton, 1994). Comparison of field performances and hydraulic conductivity measurements of various permeable bases was investigated by Kazmierowski et al. (1994)(see table 8.1). Field studies by Ahmed et al (1993) show that for identical pavement geometry, sections with more permeable base layers do exhibit higher edge drain outflow volumes (Ahmed et al., 1993).

A permeable base pavement is not necessarily unstabilized. To provide a more stable

construction work area, permeable bases may be stabilized with a cement or asphalt binder with only a slight decrease in hydraulic conductivity (<10 percent of the initial material hydraulic conductivity)(Christopher, 1998). Stabilized bases have lower values of hydraulic conductivity (0.08-1.14 cm/s) than those of untreated permeable bases (1.14-7.6 cm/s or higher)(Mathis, 1990; Zhou et al., 1993). Tandon and Picornell (1997) suggested that the best alternative material for base layers is gravel stabilized with five percent cement, which provides the required stiffness, strength, and drainage.

A wide range of values has been reported for hydraulic conductivity for materials used in bases (see table 8.1). For adequate drainage, Baumgardner (1992) and the FHWA suggested the use of 0.34 cm/s for a minimum hydraulic conductivity, whereas Lindly and Elsayed (1995) note that a range of 0.18 to 0.36 cm/s is typically used for drainage design. The ACOE differentiates between open-graded and rapid-draining materials that have hydraulic conductivity of greater than 1.8 cm/s and 0.35-1.8 cm/sec respectively (Freeman and Anderton, 1994). Recommendation by the National Stone

53

Association for unstabilized permeable base is on the order of 0.49 cm/s. State Departments of Transportation often differ in their specifications ranging from greater than 0.35 cm/s for New Jersey Department of Transportation to a range of 0.35 to 6.4 for the Pennsylvania Department of Transportation (Freeman and Anderton, 1994). On the other hand, European researchers note that for mean annual rain intensities greater than about 400mm, the adequate draining capacity is greater than 0.5 cm/s (Alonso, 1998). Excluding the values cited by Zhou et al. (1993) for untreated base materials and the European recommendation by Alonso (1998), a minimum value common to sources cited above seems to be the hydraulic conductivity noted by Baumgardner (1992), which is 0.34 cm/s. Similarly, Elsayed and Lindly (1996) also note that a minimum laboratory hydraulic conductivity of 0.36 cm/s is often preferred by designers.

The measurement of hydraulic conductivity was discussed in section 4.2 and it was noted that

laboratory measurements might not be sufficiently accurate (Jones and Jones, 1989). Another approach to determine hydraulic conductivity is to estimate it as a function of physical properties of the material. The hydraulic conductivity of unbound materials used in bases depends on the gradation of the aggregates and the density and porosity of the compacted materials. The gradation is important partially because the extent of fines in the material affect the hydraulic conductivity considerably (Lytton et al., 1993). The relatively more well known Cedergren chart and the Moulton nomograph use the knowledge of physical properties to estimate the hydraulic conductivity of unbound materials (Lindly and Elsayed, 1995) (see figures 8.1 and 8.2). Finally, a more recent correlation for estimating hydraulic conductivity of both dense-graded and open-graded untreated permeable base layers is provided by Elsayed and Lindly (1996).

The shape and texture of particles, the degree of saturation of the sample during compaction, and temperature of permeating water are not considered by the Moulton nomograph or the Cedergren charts. For the latter two variables, the reader may find the discussion and references by Moynahan and Sternberg (1974) helpful. The former two variables are relatively well known. Gravels have relatively round shapes and smooth surfaces whereas slags are angular and have rough surfaces. The surface smoothness of limestones falls between gravels and slags but limestones are also angular. If hydraulic conductivity of base materials strongly depended on shape and texture, gravel would be expected to have the highest hydraulic conductivity followed by limestone and slags. However, laboratory results by Randolph et al. (1996a) show a reverse order supporting Moulton and Cedergren’s approach that other factors such as void ratio, gradation and density may be more important for hydraulic conductivity. Estimating the hydraulic conductivity for treated bases may be more complex. Lindly and Elsayed (1995) provide a regression correlation for asphalt-treated bases (see table 8.2). However, the correlation is for open-graded materials and may not be useful for dense-graded asphalt-treated bases. Yet, since the addition of two to three percent asphalt cement has markedly less effect on hydraulic conductivity than the aggregate gradation, approaches used for untreated bases may closely approximate the hydraulic conductivity for treated bases (Zhou et al., 19 93).

54

Table 8.1 Hydraulic conductivity of base, subbase, and subgrade layers

Hydraulic conductivity (cm/s) Source S,Uα Type of base

Field Lab

Recommended for drainage (see Appendix for more values) 0.34

Baumgardner (1992) Elsayed and Lindly (1996)

Mallela et al. (2000) FHWA (1990)

S MnDOT Permeable Asphalt stabilized base 0.35-0.71 U MnDOT Class 5 Dense-graded base 0.00016

Hagen and Cochran (1996)

U No. 24 sand with 3% passing No. 200 sieve 0.001 U No 24 sand with 6% passing No. 200 sieve 0.00043 U No. 53 stone with 5% passing No. 200 sieve 0.000038 U No. 53 stone with 10% passing No. 200 sieve 0.000043 U No. 73 stone with 7.5% passing No. 200 sieve 0.068 U No. 73 stone with 10% passing No. 200 sieve 0.038 U No. 53B base with 2.5 % passing No. 200 sieve 0.025 U No. 53B base with 5% passing No. 200 sieve 0.009 S

Base or Subbase materials used by

Indiana DOT

No. 5D hot asphalt concrete base 0.00022

Ahmed et al. (1997)

S Asphalt stabilized gravel/slag/limestone 8.1-13.0

S Portland cement stabilized gravel/slag/limestone 7.5-11.0 U Untreated gravel/slag/limestone 9.5-20.0

Randolph et al. (1996a)

U Untreated aggregate base (IDOT specification 41-21) 0.046 0.4-3.35, 0.044x

U Untreated aggregate base (ODOT specification No. 304) 0.00503 0.036-0.4, 0.018x

U Untreated aggregate base (ODOT specification No. 310) 0.0154 0.007-4.2, 0.0077x

U New Jersey untreated drainable base 0.611 2.6, 0.252x

S Portland cement stabilized free draining base (AASHTO No.57) 5.85 11.9 S Asphalt stabilized free draining base (AASHTO No. 57) 3.93 10.1-13.2

Randolph et al. (1996b)

S Asphalt treated open-graded base 0.17-1.45 Zhou et al. (1993) U Permeable base 0.35 – 1.7* Grogan (1992) U Dense-graded base material - Limestone 0.05-0.75 U Dense-graded base material – Iron-Ore 0.004 U Dense-graded base material – Sand and Gravel 0.04

Tandon and Picornell (1997)

S Cement Stabilized limestone 70.3 S 11 Open-graded large stone asphalt base mix samples 0.06-1.47 ϒ S 2 Asphalt treated drainable base mix 2.4-3.6ϒ

Huang et al. (1999)

U Fine to coarse-graded aggregate 0.0003 – 0.5 Biczysko (1985) U Open graded untreated base layers from roadway samples 0.075 S Asphalt treated permeable base 0.086 S Cement treated permeable base 0.059 U Aggregate from stockpile for treated permeable base 0.063

Kazmierowski et al. (1994)

U Untreated permeable bases (range) ≥ 1.14-7.6 S Treated permeable bases (range) 0.08-1.14

Mathis (1990) Zhou et al. (1993)

S Open-graded asphalt treated base 0.05 – 0.8 Lindly and Elsayed (1995) U Dense and open-graded aggregate bases 0-1.0 Elsayed and Lindly (1996) Geotextile layer (used to reduce pumping) 0.15 Alobaidi and Hoare (1996) Geotextile used in drainage trenches 0.28-1.66 Lafleur and Savard (1996) Till subgrade north of Montreal 0.000005-0.0001 Lafleur and Savard (1996)

S Permeable asphalt stabilized base 0.35 U Dense-graded base (Class 4 special) 0.000135 U Dense-graded base (Class 5 special) 0.000220 Pea gravel used in edge drains 0.35

Birgisson and Roberson (2000)

U Dense-graded (Class 4 special) 0.000038 S Permeable asphalt stabilized base 0.35

Roberson and Birgisson (1998)

U Subbase in Searsmont, Maine 0.0004-0.01 U Subbase in Cyr Plantation, Maine 0.000006-0.001 U Subbase in Passadumkeag, Maine 0.0001-0.007 U Subbase in Lebanon, Maine 0.0005-0.001

Manion et al. (1995)

U 2% fines subbase 0.000068 or 0.000190 U 12% fines subbase 0.000014 or 0.000053 U Free draining 0.031 or 0.31

Koch and Sandford (1998) Two values are from two different testing apparatus

U Clayey-sand 0.00013-0.00058 U Quartzitic crushed rock 0.000005-0.00024 U

Base course materials used in Western Australia. Also used in shoulders.

Clayey-gravel 0.00008-0.00023 McInnes, 1972

U Compacted natural gravel used in low volume bituminous surfaced roads in Kenya and Botswana 0.00000025-

0.000000045 Toll, 1991 αU = unstabilized, S = stabilized; *Based on design calculations; xMoulton estimation; ϒPseudo- hydraulic conductivity

55

Figure 8.1 Hydraulic conductivity and gradation of base and filter materials (Cedergren, 1974). The Cedergren chart cannot be used for hydraulic conductivity estimation of gradations other than those shown.

56

Figure 8.2 Moulton nomograph is valid only for materials that have a specific gravity of 2.7(Elsayed and Lindly 1996).

57

Table 8.2 Hydraulic conductivity correlations for base layers (Elsayed and Lindly, 1996; Lindly and Elsayed, 1995).

Equation Valid Hydraulic conductivity Range (cm/s)

Hydraulic conductivity of reduced accuracy (cm/s) Medium Reference

K =300.63- 87.71 × PrcntAC + 34.39 × PrcntAIR – 33.69 × pass8, R2 = 0.87 0.2-0.7 cm/s <0.2 Open-graded Asphalt Treated Bases Lindly and Elsayed (1995)

K =-0.251 + 0.92 × Voidratio + 2.68 / Pass30 - 0.005 × Pass200, R2 = 0.78 0.2-0.7 cm/s 0.004 –0.2 Untreated Roadway Bases

K = C × D10 Clean Filter Sand

Elsayed and Lindly (1996)

K = (0.5-1) × D5 Compacted sand, gravel Kenney et al. (1984)

K = hydraulic conductivity (cm/s), Pass8 = percent by weight passing 2.36 mm (No. 8) sieve, Pass30 = percent by weight passing 0.6 mm (No. 30) sieve, Pass200 = percent by weight passing 0.075 mm (No. 200) sieve in aggregate specimen, PrcntAC = percent asphalt cement by total weight of sample, PrcntAIR = percent air voids by total volume of sample, Voidratio = ratio of volume of voids to volume of solids in aggregate specimen, C = coefficient that varied from 90 to 120. Often a value of 100 is used, D10 = effective grain size in centimeters, D5 = diameter at which there is a 5 percent mass of smaller particles in the sample.

58

As mentioned in the section on measurement techniques, drainage capacity of bases may not be well studied by estimating hydraulic conductivity of base layers only (Tandon and Picornell, 1997). Water retention capacity may be more representative of drainage capacity of pavements than the hydraulic conductivity. The base layer retains capillary water after rainfall and until another rainfall event the water movement in base layers is governed by unsaturated flow conditions. Furthermore, there is a large range of hydraulic conductivity values of the base layer in the literature suggesting that the hydraulic conductivity measurement techniques may not be sufficiently accurate. Considering that the test for water retention capacity is relatively more accurate and easier to perform and that for significant periods of time the base layer remains unsaturated, in the future, studies of water retention capacity may gain equal importance in estimating hydraulic conductivity. Hydraulic conductivity and water regimes in embankments are not as widely discussed in the literature as asphaltic concrete, PCC, base, or subbase layers. There is some literature on the use of recycled materials in embankments, however most of these focus on strength and workability of the material rather than its hydraulic conductivity. Kim et al. (1992) presented a knowledge-based expert system for utilization of solid flue gas desulfurization by-product (a coal combustion by-product) in highway embankments and note that the hydraulic conductivity of solid flue gas desulfurization by-product may range from 3.1×10-9 to 1.6×10-4 cm/s at 28-day curing while its hydraulic conductivity in place may gradually decrease with aging (see table 8.3). Partridge et al. (1999) noted that compacted waste foundry sand used in embankments is not a free draining material. Its laboratory and field hydraulic conductivity ranges from 0.1×10-5 to 7.1×10-5 cm/s. Bhat and Lovell (1996) examined the design of flowable fill by using waste foundry sand as a fine aggregate. They note that the hydraulic conductivity of flowable fill is low and that the hydraulic conductivity does not necessarily decrease with increasing contents of fly ash possibly because the advantage from the fine particle size of fly ash is outweighed by the uniform spherical shape of these particles.

Table 8.3 Hydraulic conductivity of embankments

Embankment material Hydraulic conductivity (cm/s)

Source

Flue gas desulfurization by-product 3.1×10-9 – 1.6×10-4 Kim et al. (1992) Compacted waste foundry sand 0.1×10-5 – 7.1×10-5 Partridge et al. (1999) Flowable fill 2.6×10-6 – 1.2×10-5 Bhat and Lovell (1997)

Chapter 8 Synopsis Hydraulic conductivity measurements of base/subbase layers are more common in the literature than hydraulic conductivity discussions for concrete layers because recent notion centering on free draining bases/subbases has gradually replaced impervious but strong base/subbase design approaches and the transition required more research on hydraulic conductivity of base and subbase layers. Chapter 8 discussed the suggested hydraulic conductivity values (0.34 cm/s) for design purposes and presented the field and laboratory hydraulic conductivity data for base/subbases, subgrades and geotextiles. The hydraulic conductivity of base layers (0.00004-590 cm/s) varies significantly because both free draining and impermeable layers are currently used. The hydraulic conductivity of untreated base layers may be approximated by the Cedergren chart, the Moulton nomograph, the Elsayed and Lindly (1996) correlation, or the approach presented by Kenney et al. (1984). To estimate hydraulic conductivity of open-graded asphalt treated base layers, the correlation developed by Lindly and Elsayed (1995) may be used. Hydraulic conductivity of embankments is not widely discussed in the literature. The few values noted are quite low, suggesting that embankments are not free draining structures.

59

9. Factors Affecting Water Flow in the Highway Environment

9.1 Pumping Horizontal and vertical movement of water and particles beneath concrete pavement is referred

to as water pumping. Water pumping occurs when high pore pressures in the base/subbase layers of the pavement system are produced due to trapped water and moving wheel loads. If pore pressures are not dissipated within a reasonable time frame, pumping of material from the base begins and the pavement eventually fails. Pumping is related to water movement studies because understanding pore water pressures and consequent faulting also provides information on water flow.

Water pumping may cause faulting, which is the vertical movement of pavement slabs relative to one another. A literature summary by Hansen et al. (1992) notes that (1) faulting can occur in newly laid pavement in as little as three months; (2) sealing the lower edges of the pavement (in contrast to enhanced edge drainage) reduces faulting in comparison to pavement with no edge drains or edge seals; (3) there may be a positive correlation between faulting and rainfall. These notes suggest that the proper moisture content of the pavement and the subbase is significant for keeping the concrete pavement and the subbase intact.

Numerous interstate highway pavements experience faulting caused by pumping. Thus, there has been a motivation to prevent pumping of fines and faulting. One approach has been to use geotextiles in between the pavement and the subgrade soil. Research shows that in the presence of cyclic loading, geotextiles may be effective in reducing pumping if they are thick (lower hydraulic gradients observed in 0.5-mm and 1.0-mm thick geotextile than 2-mm thick geotextile), whereas they increase pumping if they are highly permeable and compressible (Alobaidi and Hoare, 1996). Another approach has been to study pumping characteristics of highly permeable (open-graded) bases. Crovetti and Dempsey (1991) noted that when open-graded materials were used, pore pressure build-up was reduced and possibility of pumping of fines was eliminated.

A comprehensive laboratory and mathematical study on pumping was conducted by Hansen (1991) who measured flow from water pumping beneath concrete pavement slabs. This study showed that water pumping was an unsteady phenomenon that could not be modeled using the steady energy flow equation. Peak water velocity and the Reynolds number of the flow were found to be 0.9 m/s and 560, respectively. The water movement beneath the pavement slab first moved slowly (0.2m/s) in the vehicle direction, then moved rapidly (0.9 m/s) in the reverse direction. Another study by Hansen et al. (1991) reported similar water velocities (0.15 m/s for a car to 0.9 m/s for a 79-kN truck axle).

9.2 Infiltration Through Cracks Formation of cracks is a widespread phenomenon in both asphalt concrete and PCC pavements.

In PCC, before application of an external load, bond cracks may occur in concrete at the mortar-aggregate interface, with negligible cracking in either the mortar or aggregate phases (Tawfiq et al., 1996). Cyclic loading propagates cracking, and with extension and widening of microcracks, a network of cracks may form. Slate and Hover (1984) reported that the increase in bond cracking is negligible at applied loads up to 30 percent of ultimate load. Environmental conditions and especially freeze-thaw cycles enhance crack development. Expansive soils underlying pavements also

60

contribute significantly to cracking (Wiseman et al., 1985). Causes of cracking and other deterioration resulting in surface roughness can be determined by profilometers that provide accurate and reproducible longitudinal profile data (Novak and DeFrain, 1992). Often cracks are referred to as transverse and longitudinal cracks. Roberson (2001) related crack formation and type to material problems (see table 9.1). Frabissio and Buch (1999) reported that pavements containing slag or recycled concrete coarse aggregate appear to have more transverse cracks than those using natural gravel or carbonate aggregates. They suggest that slag and recycled pavement have a greater tendency for shrinkage cracking when proper curing considerations are neglected. With so many factors promoting crack formation and growth, many pavement structures are subjected to undesired conditions arising from cracking.

Table 9.1 Cracking types and causes (Roberson, 2001)

Pavement type Cracking Material problem Corner Follows pumping Diagonal Transverse Longitudinal

Follows moisture build up PCC

Punch out Deformation following cracking Longitudinal Strength Alligator Drainage Transverse Freeze thaw cycling Asphalt concrete

Shrinkage Suction (i.e. moisture loss)

Table 9.2 Percent of runoff entering surface cracks

Crack width (mm) Pavement Slope (%) Percent of runoff entering crack Author

0.89 1.25 70 0.89 2.50 76 0.89 2.75 79 1.27 2.50 89 1.27 3.75 87 3.17 2.50 97 3.17 3.75 95

Cedergren (1974)

0.90 70 3.1 97 Barksdale and Hicks (1977)

The presence of cracks as well as joints significantly increases hydraulic conductivity of the

surface layer. Studies as early as 1952 have pointed to the enhancement of water permeability due to crack formation. The effect of cracks in water permeability of pavements is so significant that some authors believe infiltration through the surface layer depends on the extent of cracking rather than the hydraulic conductivity of the material itself (Barber and Sawyer, 1952). Cedergren and Godfrey (1974) noted that up to 70 percent of surface runoff can enter a crack no wider than 0.8 mm if there is no obstruction at the bottom of the crack (see table 9.2). Similarly, Barksdale and Hicks (1977) noted that it is possible for as much as 70 to 97 percent of rainfall to enter open joints with opening of 0.9 mm to 3.1 mm when dry conditions exist beneath the pavements.

There is continued interest to find methods to minimize the undesirable effects from cracking.

In asphalt concrete, mineral fillers such as rock, dust, slag dust, hydrated lime, hydraulic cement, fly ash and loess may be used to increase density and strength of asphalt concrete mixtures. A laboratory

61

study by Tawfiq et al. (1996) suggests that addition of mineral filler such as silica fume and fly ash reduces the extent of crack formation, and thus the hydraulic conductivity of concrete. Yet, more of the studies focus on sealing of cracks rather than preventing them. In asphalt pavements, the cracks may be sealed not only to minimize water penetration, but also to renew skid resistance, fill ruts, retard raveling, restore ride quality, reduce stresses due to traffic and reduce effect of thermal variations. Sealing appears to be an effective method for blocking water ingress since laboratory studies show that hydraulic conductivity of seals is quite small (see table 9.3). On the other hand, there is data suggesting that for PCC pavements, the presence of crack/joint sealants does not play a significant role in altering surface infiltration rates over dense-graded materials (Ridgeway, 1976). Christopher and McGuffey (1997) attributed the failure of sealing to the short life span of sealants (2-3 weeks). Similarly, Hagen and Cochran (1995) noted reduced or no flow through the sealed joints during the first two weeks and observed significant infiltration on the third week after a big rain event. Sealants were ineffective although they seemed to be in good condition.

Table 9.3 Hydraulic conductivity of seals as measured using a falling head test (Button, 1996)

Material tested Water head (cm) Measured hydraulic conductivity (cm/s)

Average hydraulic conductivity (cm/s)

5.2×10-5

5.2×10-5 20 1.2×10-5

3.9×10-5

1.1×10-7 3.0×10-6

Slurry seal

5 2.8×10-6

2.0×10-6

5.2×10-6 2.8×10-5 20

0.0 1.1×10-5

0.0 0.0

Micro-surfacing

5 0.0

0

20 0

Seal coat

5

0.0*

0

* Seal coat specimens exhibited no permeability to water with a head of 20cm for a measurement period of 72 hr. In the U.S., one approach taken to prevent moisture penetration has been to seal the top and

bottom surface of the asphalt mixture. At present, it is believed that sealing does not prevent water penetration (Christopher, 1998), although sealing of shoulder joints may significantly reduce infiltration (Roberson and Olson, 2000). Sealing can slow infiltration and may prevent particles from entering the pavement. On the other hand, several studies have suggested that surface sealing did not hinder rutting and stripping partially because routes of evaporation were blocked, causing moisture accumulation within the asphalt concrete mixture (Kennedy, 1985).

Other than sealing, several variables affect the water movement in cracked surface layers. For example, the type of base may influence the water flow through cracks in PCC pavements. The water

62

flow in PCC pavements with open bases may be greater than the water flow in PCC pavements with densely-graded bases (Ridgeway, 1976). Koch and Sandford (1998) documented that in presence of free draining subbase, infiltration through cracks is limited by crack width, whereas in other subbases, it is the subbase that limits infiltration rate. Duration of the rainfall is also relevant. Rainfall of higher duration may result in higher water entrance and flow in pavements than rainfall of short duration, but high intensity (Ridgeway, 1976). Related to crack conditions, the variables affecting water hydraulic conductivity through cracks are the length, width and spacing of cracks as well as whether the cracks are filled by debris or not (Ridgeway, 1976; Oshita and Tanabe, 2000c). Laboratory tests by Koch and Sandford (1998) confirmed that infiltration rates through infilled cracks are significantly higher than infiltration rates through unfilled cracks. Since cracks are free of debris only during their initiation (fresh cracks) and possibly after they are washed clean by pumping or rain events, infiltration rates through debris filled cracks are more representative of field conditions. Finally, on retrofitted pavements that have an additional asphalt layer on top of old pavement, crack width as observed on the new surface layer may be misleading if the old pavement was already cracked at some other width (Koch and Sandford, 1998).

Cedergren (1974) emphasized the significance of surface infiltration from rain events and reports that asphalt concrete pavement and PCC pavement could allow infiltration of 33-50 percent and 50-67 percent of precipitation water to the road base. Modeling the water movement in the surface layer (cracked or not-cracked) is essential for determining the water conditions in the base course. Modeling infiltration through the surface layer and through cracks relates the amount of water or degree of saturation in the base course to an equivalent amount of rainfall required to achieve that condition (Lytton et al., 1993). Thus, in order to understand the water regimes in the base and subbase, the movement of water in the surface layer first must be determined.

Water movement in cracked pavement is a markedly complicated phenomenon, and thus rather difficult to model. Oshita and Tanabe (2000a, b, c) have developed a detailed model that is described in section 9.3. Recently, hydrodynamic models of flow in unsaturated fractured porous media are being developed that may eventually be useful for modeling cracked pavements (Or and Tuller, 2000). At present, for calculating surface infiltration rate, an equation described by Ridgeway (1976) seems to have formed the basis for current highway subdrainage design calculations (Moulton, 1980; Crovetti and Dempsey, 1993). The approach calculates the potential surface infiltration rate in bituminous concrete or PCC pavement using the following equation:

)WCW

W1N(Iq

s

cci +

+= (Eq. 9.1)

where:

(L). jointsor cracks e transversof spacingCand (L),layer material permeable graded-open theofwidth W

(L), jointsand/or cracks e transversoflength averageWlanes, trafficofnumber N

/TL),(L rateon infiltraticrack I

),/TL(L rateon infiltrati surfaceq

s

c

3c

23i

=====

=

63

Equation 9.1 may also be rewritten as:

)WY(Iq ci = (Eq. 9.2)

where: Y = average length of cracks/joints per unit meter of pavement.

From equation 9.2, it may be easier to notice that this formulation expresses surface infiltration

rate on the basis of the total length of cracks/joints per unit area of pavement. The assumptions for these two equations are that the pavement surface layer is impermeable in uncracked locations, continuous longitudinal joints separate at least two individual driving lanes and separate outer driving lanes and shoulders, and transverse joints or cracks are regularly spaced.

Crovetti and Dempsey (1991) presented a modified form of the Ridgeway equation:

ps

cci k)

WCW

WNc(Iq ++= (Eq. 9.3)

where: Nc = number of contributing longitudinal cracks, kp = infiltration rate of uncracked pavement (L3/TL2). Drainage path of infiltrated water is also described by Crovetti and Dempsey (1991)(see figure 9.1). As shown in the following equation, the drainage path is a function of the cross slope, the width of drainage layer and the longitudinal gradient.

2

1 WL

+=

cSg (Eq. 9.4)

where: L = length of the drainage path in [L], g = longitudinal gradient [LL-1], W = width of drainage layer [L], and Sc = cross slope [LL-1]. All the previously mentioned variables affect the duration of drainage and thus the hydraulic conductivity requirement of the pavement. For example, a material with lower hydraulic conductivity can be used if the cross slope is increased from 1 to 2 percent.

64

W

Equal Elevation

Permeable Base

PCC Concrete

Subgrade

Longitudinal Grade,g

Cross Slope, Sc

Drainage path length, L

Figure 9.1 Pavement cross-section showing path of draining water (adapted from Crovetti and Dempsey, 1993).

Ridgeway (1976) presented a wide range of data for crack infiltration rates (Ic)(see table 9.4). In bituminous concrete pavements underlain by open-graded materials, the measured infiltration rate ranges from 0.005 to 1.521 m3d-1m-1, whereas in PCC pavements with dense-graded base materials the infiltration rate varies between zero and 0.181 m3d-1m-1. For design purposes, a conservative value of Ic = 0.223 m3d-1m-1 is suggested for PCC pavements with open-graded bases.

Table 9.4 Infiltration rates.

Comments Infiltration Rate (m3d-1m-1) Author Concrete pavement underlain by open-graded material 0.005-1.521 Concrete pavement underlain by dense-graded base material

0-0.181 Ridgeway (1976)

Asphalt, cracks unfilled with debris, measured in lab 1700-65000 Asphalt, cracks infilled with debris, measured in lab 130-1700 Asphalt underlain by 2 or 12% fines subbase, measured in lab

10-54

Asphalt pavement, field measurement 0.9-82 Asphalt pavement saw cut to 3.4 width, no debris, field measurement

>35000

Koch and Sandford (1998) (measurements performed at varying crack widths and water heads)

In a different context, in a study on autogenous healing of cracks in reinforced concrete, Edvardsen (1999) suggested the use of Poiseuille Law to model water flow through a smooth parallel-sided crack. The equation used in this approach is derived from the parallel-plate theory described in fluid mechanics textbooks. The following equation is suggested for use with incompressible fluids under laminar flow:

dwbpqo ⋅⋅⋅⋅∆

=η12

3

(Eq. 9.5)

65

where: qo = water flow of idealized smooth cracks [L3T-1], ∆p = differential water pressure between inlet and outlet of the crack [ML-1S2], b = length of the crack [L], w = crack width [L], d = flow path length of a crack (thickness of concrete layer) [L], and η = absolute viscosity [MT-1L-1].

The water flow estimated from equation 9.5 is an overestimation for several reasons. First, crack surfaces are not smooth and roughness resulting from course aggregate and surface roughness of aggregates and paste slow water flow. Second, crack width along the flow path and the visible crack length are variable, and third, cracks branch out. Finally, physical factors such as the effects of adhesion and cohesion are ignored. To account for these considerations, an overall reduction coefficient may be added to equation 9.5. Edvardsen (1999) cited a range of data for the reduction coefficient varying between 0.02 and 0.53. Edvardsen’s approach may be modifiable to study hydraulic conductivity in and infiltration through cracks in pavements instead of autogenous healing of cracks. However, its applicability is doubtful considering that in a study of crack infiltration, Koch and Sandford (1998) found that the rate of infiltration through cracks is hardly linear with crack width and not even close to having a cubic relationship as suggested in equation 9.5 (see figure 9.2). Koch and Sandford’s results also contradict the model and laboratory results of Oshita and Tanabe (2000c) (see chapter section 9.1). Finally, it should be noted that, Poiseuille’s Law applies for lateral flow through cracks, not drainage due to gravity.

Significance of crack infiltration as a major ingress route for the pavement system was

emphasized throughout this section. Infiltration rate data based on crack length was presented assuming crack infiltration was the major factor limiting the amount of volume that enters pavements. It should be noted that crack infiltration as a function of crack width, depth, and amount of infilled debris may not relate closely to how water enters and drains from the pavement system. Emphasizing this point, Ahmed et al. (1997) note that infiltration criteria based solely on length of crack (Ridgeway, 1976) is not adequate. Once the pavement cracks and pores of the subbase become saturated, infiltration in to the pavement layers depend on the rate at which water flow laterally in the subbase layer towards the drain (Ahmed et al., 1997). Thus, the rate of flow depends on pavement geometry, hydraulic properties of the pavement layers, and conditions of the edge drains in addition to duration of precipitation and length of cracks (Ahmed et al, 1997).

66

0

5000

10000

15000

20000

0 0.5 1 1.5 2 2.5

Crack width (mm)

Infil

trat

ion

rate

(cc/

min

-30c

m)

Figure 9.2 Infiltration rate versus crack width for the Searsmont sample (adapted from Koch and Sandford, 1998)5.

9.3 Temperature Water flow in porous media under variable temperatures is a relatively new discipline (Arnold

and Nishigaki, 1998). Although modeling of mass transfer in porous media under iso-thermal conditions is less complicated and better understood, many environmental studies have to address thermal affects. Water movement in pavements is one such case where the temperature of the pavement affects the pavement water regime. The extent of freezing, condensation, and evaporation depends on the temperature of the water in the pavement. Thus, estimation of pavement temperature becomes an important issue for modeling water movement in roads. This section presents the literature on estimation of temperature in the pavement. No work was found that related the temperature to condensation or evaporation.

Two approaches for estimating pavement surface temperature are described by Lytton et al. (1993). A simplistic way is to use a surface correction factor to relate the air temperature to the pavement surface temperature. The ACOE have worked on correction factors for different surfaces for temperatures above and below freezing. However, Lytton et al. (1993) noted that the correction factor approach has many shortcomings and proposes an energy balance to estimate the surface temperature. The energy balance consists of a vectoral sum of all heat fluxes between the pavement surface and air on a sunny day. Thus, the sum of incoming and reflected short and long wave

5 Note that 10 000cc/min-30cm = 43 200 m3/d-m and that measured infiltration rates by Ridgeway (1976) (0.005 to 1.521 m3d-1m-1) were much smaller. Measurements by Koch and Sandford (1998) where cracks are filled with debris are on the order of 43-260 m3d-1m-1.

67

radiation, convective heat transfer, energy absorbed by the ground, effects of transpiration, condensation, evaporation and sublimation are set equal to zero. Net short and long wave radiation are estimated by equations that consider cloud cover, absorptivity and emissivity of pavement surface, air temperature, vapor pressure of the air, latitude of the site, and declination of the sun. Lytton et al. (1993) noted that magnitude of the effects of phase changes are relatively small and can be neglected. Finally, the estimation of convective heat transfer is simplified by using an empirical equation.

In the literature, numerous approaches exist for estimating pavement temperatures. For example, Armaghani et al. (1986) present field data that show that pavement temperatures are always higher than ambient air and that minimum and maximum daily temperatures are reached between 6:00 a.m. and 8:00 a.m. and 12:00 noon and 2:00 p.m. respectively. Thompson et al. (1986) described the Climatic-Materials-Structural (CMS) heat transfer computer model (later improved by Lytton et al., 1993) that takes into account thermal properties of materials and soils, air temperature data, solar radiation data, and wind velocity data as input, processes the data with Fourier heat-transfer equation for transient heat flow, and outputs pavement temperature information. Similarly, Hermansson (2000) also presented a pavement temperature simulation model that takes hourly values for solar radiation, air temperature, and wind velocity as input and yields pavement temperature data.

Another approach is provided by Ovik et al (1998). They propose using the following empirical equation for pavement temperature estimation:

Tsurf = 0.859 Tair+1.7 (Eq. 9.6) where: Tair = one day minimum air temperature (oC), Tsurf = Tmean = mean surface temperature surface temperature (oC).

This empirical equation may underestimate the actual temperature, however, adjustment of Tsurf provides close approximations to field conditions over a typical year at a given geographical site when the equation 9.7 is used to estimate temperature with depth and time. Ovik et al. (1999) noted that equation 9.7 gave reasonable predicted values compared to the field temperatures at the MnDOT research site Mn/ROAD.

−+=

−

αππα

π

Pxt

PAeTtxT P

x

mean2)(2sin),(

2

(Eq. 9.7)

where: T(x,t) = soil temperature as a function of depth and time, x = depth [L], α = thermal diffusivity, [L2/T], P = period or recurrence cycle (365 days for one year) [L], T = time measure from when the temperature passes through Tsurf [L]. For surface of the pavement the equation reduces to:

T(t)=Tmean+Asinωt (Eq. 9.8)

68

where: w =2π/P = 2π/365 [L-1].

9.4 Soil Mechanics A significant portion of pavement design elements is based on the mechanical properties of

pavements. These elements include resilient modulus, bearing capacity, or other variables related to shear strength. Many authors have studied strength properties of pavements and some have investigated their relationship to water content and temperature in pavements. For example, an increase in subgrade water content decreases the resilient modulus, resulting in greater deflections in the pavement subgrade, which decreases the pavement design life (Rainwater et al. 1999; Tien et al. 1998; Ksaibati et al., 2000). Oloo et al. (1997) noted that the thickness of the base layer required to support a given tire pressure decreases as the matric suction increases. Matric suction pulls soil particles together and increases shear strength due to the forces between the soil grains arising from this pull. Other relationships between strength properties and temperature and moisture are described by Ali and Lopez (1996), Thadkamalla and George (1995), Tian et al. (1998), and Mohammad et al. (1999). Thus, in absence of information on water conditions in pavements, strength properties and their correlation to saturation or soil suction can be used to estimate saturation levels in pavements. Use of strength properties to understand water movement may be convenient since considerable literature exists on pavement strength and its dependence on moisture. Many of the pavement water movement models are developed from a strength perspective rather than a hydrology perspective.

Chapter 9 Synopsis Chapter 9 grouped water pumping, infiltration through cracks, temperature, and mechanical properties of soils as factors affecting water movement in the highway environment. Water pumping occurs due to differential pore water pressure as a consequence of moving wheel loads. A positive relation has been found between precipitation and faulting caused by pumping. During pumping, water may reach a velocity of 0.9 m/s and fines are exchanged between the base/subbase and subgrade layers. To prevent pumping, geotextiles may be used. Chapter 9 emphasized infiltration through cracks because by current knowledge, it is the major ingress route for pavement structures. Causes of cracking (freeze-thaw cycles, cyclic loading, expansive soils) and methods to prevent it (addition of silica fume, fly ash) are commonly discussed in the literature. Sealents are used to prevent infiltration, however they are ineffective because of their short life span, and they also block evaporation routes causing moisture to stay within the pavement. Variables affecting crack infiltration are hydraulic conductivity of base; crack length, width, and spacing (contradicting opinions exist); amount of infilling material; duration and intensity of precipitation (contradicting opinions exist, see section 2.4); pavement geometry; hydraulic properties of the pavement layers; and conditions of the edge drains. The crack infiltration equation by Ridgeway (1982) is still used in the literature. Temperature of the pavement structure is important for studying evaporation and freeze-thaw phenomena. In the literature, freeze-thaw estimations are widely discussed because differential phase changes of water cause pavement deterioration. Temperature of the pavement may be estimated by (1) a correction factor relating air temperature to pavement surface temperature, (2) an energy balance that includes incoming and reflected short and long wave radiation, convective heat transfer, energy absorbed by the ground, effects of transpiration, condensation, evaporation and sublimation, (3) the empirical equation provided by Ovik et al. (1999). Chapter 9 included a brief discussion on the effect of mechanical properties of pavement layers on moisture content in the pavement because matric suction increases shear strength and this phenomena has lead to development of relationships between strength properties and saturation. Many studies modeling water movement in pavements are developed from a soil mechanics perspective and some of the data and approaches as such may be useful for the present study also.

69

10.Computer Models

10.1 Modeling Approaches There are several modeling approaches. Deterministic models do not consider random

variations in variables and one combination of input data will always give the same output. Stochastic models incorporate randomness that may arise from measurement errors or the nature of data such as changing environmental conditions. In a stochastic approach, a model element may also be considered random if there is not enough information about it or if it is too complicated to be modeled. Uncertainty or the error in estimations is more clearly stated in stochastic models whereas deterministic models have a fixed output and at most may give a range of possible answers. The makeup of both stochastic models and deterministic models are approaches that relate input to output. Apart from being deterministic or stochastic, these approaches may be (1) conceptual (based on simplified physical, chemical or biological phenomena), (2) empirical (based on observations), (3) black box type (statistical relationships between input and output), (4) gray box type (combination of physical relationships with statistical relationships) (Butler and Davies, 2000).

There are several aspects to model development. The mathematical model involves relating input and output variables as mentioned in the previous paragraph. Once the model is coded, it needs to be calibrated. Calibration of a model requires estimation of parameters used in the model. Field data can be used to determine these parameters. Model development and calibration is typically implemented by the modeler. When the model is being applied to a specific system, verification is necessary. This stage is similar to calibration except that verification is specific for a system and is typically implemented by the user. The user verifies the model by comparing model estimates with field measurements. Once, the model is verified it can be used to predict conditions for other scenarios.

The pavement-water models discussed in this chapter are mostly deterministic models. Section 10.1 covers simplified modeling approaches for pavement water flows or conditions whereas section 10.2 includes comprehensive models that incorporate details related to pavement moisture conditions or strength properties. Many of the models in sections 10.1 and 10.2 have not been verified by various data sets, and thus, prediction capability of models is not clear. Consequently, emphasis of the discussion is not on determining which model predicts field conditions with greatest precision and accuracy. Instead, models’ mathematical expressions are discussed to present a broad background on various approaches. Finally, section 10.3 includes commercially available unsaturated zone water flow and contaminant transport models that may be used on pavements.

10.2 Examples of Simplified Pavement Water Movement Models Simulation of Subsurface Drainage of Pavements

A deterministic approach for simulating subsurface drainage of pavements is presented by McEnroe and Zou (1993) and McEnroe (1994). Drainage of a pavement is modeled based on conservation of mass using the continuity equation and Darcy’s Law. Application of the model to a theoretical pavement demonstrates that the important question about drainage should not be about how long draining will take place, but instead it should address the extent to which the pavement will drain because of capillarity.

70

Model analysis suggests that if a pavement drains at all it will drain fast. To model drainage, the continuity equation is used and values for drainable porosity, hydraulic conductivity, and elevation of the phreatic head are estimated. Drainable porosity is predicted as a function of water content, which in turn is determined from residual saturation, bubbling head and pore size distribution indices. The bubbling head is the suction head below which the material remains fully saturated. Since, the model is based on bubble head, it provides a useful analysis for drainability of a pavement structure.

One of the important variables in this approach is the lowest degree of saturation that can be achieved in the field through drainage, smin. The matric potential in unsaturated pores create a negative pressure due to surface tension between water and pore walls and below a certain water content level, the water in the pores do not drain due to this surface tension. Thus, the minimum degree of saturation, smin, plays an important role in drainage modeling. The mathematical definition of smin is:

nnS d−= 1min (Eq. 10.1)

where: nd = drainable porosity, [L3 L-3], and n = porosity, [L3 L-3]. Drainable porosity can be estimated by the following empirical relationship (Moulton, 1980):

nd=0.0355k0.235 (Eq. 10.2)

where: k = hydraulic conductivity, [L T-1]. Substituting for drainable porosity in equation 10.1;

235.0min

0355.01 kn

S −= (Eq. 10.3)

Continuity equation is presented as follows:

0),(),()(n e =∂

∂−

∂∂

xtxq

ttxhx (Eq. 10.4)

where: t = time [T], x = horizontal distance from the edge drain [L], ne (x) = the depth averaged drainable porosity of the base, h(x,t) = the elevation of the phreatic surface, and q(x,t) = the lateral discharge (per unit width) in the saturated part of the drainage layer.

71

In short, estimation of smin as a function of material properties and its relation to residual saturation forms the basis for the McEnroe model. The continuity equation takes a special form for drainage layers and is solved by an implicit nonlinear finite-difference scheme of SUBDRAIN. The approach allows estimation of drainage as a function of hydraulic conductivity and time. Subbase Moisture Conditions

The approach presented in Waters (1998a) and Waters (1998b) is significantly different than the other models reviewed in this paper. First, an innovative approach for prediction of moisture conditions in flexible pavements is presented, then this approach is used to analyze drainage conditions in a pavement. Strength related properties are utilized and water movement with time is not considered. The entire approach is based on linear relationships of defined terms.

Prediction of moisture conditions is achieved by relation of moisture content to mechanical properties of the pavement. Unlike other models mentioned in Section 10, a mass balance approach is not used. Instead material properties for moisture content are determined and expressed in terms of moisture content (w), liquid limit (LL), plastic limit (PL), and fraction of fines (F). Relationships are presented for optimum moisture content (OMC), modified liquid limit (MLL), moisture sensitivity index (MSI), moisture condition index (MCI), California bearing ratio (CBR) at OMC, and field CBR:

OMC = 0.9 [(PL x F)+2(1-F)] (Eq. 10.5) MLL = (LL x F) +2(1-F) (Eq. 10.6) MSI = (MLL/OMC) x 100 (Eq. 10.7) MCI = (w/MLL) x 100 (Eq. 10.8) CBROMC=6700/MSI (Eq. 10.9) CBRField=CBROMC0.5(0.9755)RMC-100 (Eq. 10.10)

Using these relationships relative moisture content (RMC) can be calculated.

RMC = (w/OMC)x100 (Eq. 10.11) or

RMC = (MCI x MSI)/100 (Eq. 10.12)

A short analysis of factors affecting matric potential is also presented by Waters (1998a). Effect of water table, rainfall and MCI on matric potential is discussed. Waters (1998a) notes that matric potential can be expressed in units of pF. The following equations are used:

MP = 2 +log H (Eq. 10.13) H = 28 (1000/R)5.684 (Eq. 10.14) MCIc= 67.9 x H-0.141 (Eq. 10.15) where: MP = matric potential , pF [L], H = depth of water table [L], R = rainfall [L], and MCIc= MCI at the center of pavement.

72

A combination of the above equations can be used to determine field moisture conditions. The approach is relatively simple compared to detailed analyses of other models based on conservation and equilibrium laws. However, this approach demonstrates the use of empirical relationships and simplified concepts for estimation of pavement drainage conditions. Water Movement in PCC

Savage and Janssen (1997) present a deterministic semi-empirical model based on soil physics principles. The model is used to estimate unsaturated hydraulic conductivities and moisture movement characteristics of PCC. Model parameters are estimated from an experimental study. To predict unsaturated hydraulic conductivity, the van Genuchten equation (see Appendix A) is used and coefficients are determined from the best-fit curve equation. Less than 10 percent difference is observed between model estimations and experimental results for unsaturated hydraulic conductivity. Similarly, data for change in moisture content as a function of time is plotted and fit on a trendline that has two parameters (Eq. 10.16). The curve trendline equation presented next is used for modeling the time change of moisture content.

tB

jw t

+= (Eq. 10.16)

where: w = cumulative proportion of available moisture lost, percent, t = elapsed drying time, [T], J = coefficient [dimensionless], B = coefficient, [T].

10.3 Examples of Comprehensive Pavement Water Movement Models Water Flow Modeling in Unsaturated Base and Subbase Materials

The base and subbase water flow model adopted by Lebeau et al. (1998) can be used to model one and two-dimensional infiltration. The model does not seem to have been validated on an existing pavement system, however simulation of water movement in various drainage systems (e.g. daylighting, subgrade level pipe drain at the pavement edge, subgrade level pipe drain at the shoulder edge, geocomposite edge drain at the pavement edge, and geocomposite edge drain at the shoulder edge) is presented as a theoretical approach. The model is based on a finite element method solution of a complicated partial differential equation.

The one and two dimensional unsaturated flow models by Lebeau et al. (1998) consider physical deformation of base and subbase material under the influence of water. The effect of soil water on shear strength of the material is modeled by the non-linear shear strength equation proposed by Fredlund et al. (1996). Simply, the equation addresses the reduction in apparent cohesion by an increase in water content and subsequent diminished water suction.

The water movement is modeled using a form of Richards Equation (Eq. 10.17). First, the appropriate form of Darcy’s Law is inserted into the continuity equation, resulting in Richards Equation. Then, the constitutive equation for the water phase of an isotropic unsaturated soil is introduced with the assumption that the compressibility of both water and soil matrix are negligible. Stress and shear and other soil mechanics terms and their consequent modifications of Richards

73

Equation are not within the scope of this paper. However, they are mentioned here to demonstrate that water movement models can be considered in light of the effect of water on material properties. The equation of water flow used by Lebeau et al. (1998) is provided in equation 10.17 to familiarize the reader with various forms of Richards Equation that is applicable for highway water movement models;

dtdhmghuk www ⋅⋅⋅=∇⋅⋅∇ ρ))(( (Eq. 10.17)

where:

operatorgradient =∇ k(uw) = hydraulic conductivity as a function of pore water pressure [L T-1], h= total hydraulic head [L], ρw = density of water [M L-3], g = gravitational acceleration [L T-2], mw = dθ/dUw = coefficient of water volume change with respect to a change in pore water pressure [F-1 L2], θ = volumetric water content [L3 L-3].

Unsaturated conditions are difficult to model because water content, matric potential and hydraulic conductivity are functions of each other. However, estimation of these variables is crucial for the modified Richards Equation (Eq. 10.17). The equations adopted for the base and subbase water movement model are those proposed by Fredlund. Lebeau et al. (1998) use equations 10.18 and 10.19 to estimate water content and unsaturated hydraulic conductivity noting that the latter equation is selected because it provides superior predictions in association with the former equation.

wvw

o

rw

rw

w

ueu

uu

+

+

+

−=

α

θθ

ln101ln

1ln1

,

6

, (Eq. 10.18)

∫

∫

=

=

⋅⋅−

⋅⋅−

=)10ln(

)ln(

)10ln(

)ln(6

6

)(')(

)(')()(

)(

w

w

uy

yy

sy

uy

yy

wy

wr

dyee

e

dyee

ue

uk

θθθ

θθθ

(Eq. 10.19)

where: e = natural number, 2.71828, α, v, w = three different soil parameters, θa= saturated volumetric water content [L3/L3], uw,r = pore water pressure corresponding to the residual water content, θr, [F L-2], kr(uw)/ks = relative hydraulic conductivity as a function of pore water pressure, ks= saturated hydraulic conductivity [L T-1], θ(ey) = equation 10.18 evaluated at ey [L3L-3], θ’(ey) = derivative of equation 10.19 evaluated at ey [F-1L2], and

74

y = dummy variable of integration representing the logarithm of pore water pressure.

Lebeau et al. (1998) used their model to simulate different drainage systems by accounting for capillary nature of drainage systems, omitting the capillary nature of drainage systems, and accounting for capillary nature of drainage systems while considering a crack in the surface course. The model needs to be verified to evaluate its applicability. Base and Subgrade Moisture Regimes

The model presented by Alonso (1998) is one of the most comprehensive pavement water movement models in the literature. The model demonstrates coupling of mass and heat balances with mechanical deformation in unsaturated bases and subgrades. The basic flow phenomena considered are air and water flow in a Darcy type porous media, vapor diffusion, and liquid-vapor phase changes (see table 10.1). For heat flow, conductive transport (Fourier law) and advective transport in liquid and gas phases are considered. To calculate the mass and heat balances, change in porosity or volumetric deformation is needed. Thus, mechanical deformation due to suction and temperature changes is estimated.

The constitutive equations in the model are numerous. Major equations will be mentioned by

stating the parameters considered and relationships required to solve the equations. It is also noted that the formulations are solved using the finite element approach. For a more detailed discussion of the equations, the reader is referred to the original paper.

75

Table 10.1 Properties and parameter for flow, temperature and deformation analysis (adapted from Alonso, 1998))

Phenomenon/Parameter Determination Suggested Relative Importance Water Flow

Liquid density Physical Constant -Vapor density Physical Constant -

HighVapor diffusion Tortuosity

Physical Constant Special test Low/intermediate

Vapor dispersion Special test LowConvective flow (Darcy) Intrinsic hydraulic conductivity Relative hydraulic conductivity

Test Test. Approximation from characteristic curve

Intermediate

Characteristic curve Test. Approximation from grain size Very HighAir Flow

Gas density Physical constantAir molecular diffusion Physical constantAir dispersion Special test

Convective flow (Darcy) Relative hydraulic conductivity Air viscosity

Special test. Also from water relative hydraulic conductivity Physical constant

Generally low

Heat Transfer

Thermal conductivities From physical constants; generally HighSpecific heats Physical constants High

Mechanical Behavior Thermal expansion coefficient Special test Intermediate/high State surface for volume changes Tests with controlled suction HighElastic moduli Test with controlled suction High

The mass balance approach is based on an equation that considers many factors such as liquid

and gaseous phases of water as a function of porosity as well as liquid and gaseous phases of convective flow of water and diffusive flow of the vapor. The relationships required to solve the water mass balance equation are (1) water density as a function of temperature and pressure, (2) vapor density as a function of gas and liquid pressure, temperature, (3) Van Genuchten water retention curve, which depends on model parameters as well as water surface tension as a function of temperature, (4) Darcy’s Law for estimating convective water flow as a function of hydraulic conductivity and water retention curve, and (5) Fick’s law for estimating vapor diffusion through air, which depends on dispersion coefficient, tortuosity parameter, mechanical dispersion, vapor molecular coefficient in the air, and temperature. Similarly, another complicated equation is used to represent the air mass balance. The relationships of concern for the air mass balance are (1) air density, (2) air mass dissolved in water (estimated by Henry’s law), (3) Darcy-type convective airflow, and (4) Fick-type air diffusion in gas phase.

The energy balance equation and formulation of the mechanical problem are also detailed. Specific energies, non-advective heat conduction and advective heat conduction are considered for the energy balance equation. Thermal conductivity is modeled by Fourier’s Law, which is modified to account for each dry and saturated phase conditions. Important parameters for mechanical behavior are thermal expansion coefficient, state surface for volume changes and elastic moduli.

76

Mechanical deformation is coupled to mass and energy balance equations because all of them contain terms that depend on porosity. Since changes in suction induce irreversible volumetric deformations simultaneous analysis of mass, energy and stress variables is adequate.

Capabilities of the model are presented by simulation of a pavement behavior under a representative Mediterranean climate. Determination of initial and boundary conditions for density, moisture, and stress are shown. Detailed representations of conservation of mass and energy for liquid and gaseous phases of water and inclusion of the mechanical formulation are helpful in identifying the important variables that affect the water movement in pavements. Conclusions of the simulation reveal the importance of soil water suction as an independent variable. Water Migration Model in Cracked Concrete

The water migration models presented in (Oshita and Tanabe 2000a, b, c) are micro-scale deterministic models that focus on concrete. The models are based on mass conservation and force equilibrium for a porous concrete composite that consists of aggregate, cement, paste, water and cracks. First a model for homogenous concrete without cracks is developed. In the latter papers, incorporation of cracks into the existing model and its calibration are presented.

Similar to the approach adopted by Alonso (1998), mass conservation is modeled for both liquid and gaseous states of water. An expression is used to represent each of the following conditions: (1) total strain changes, (2) volumetric changes due to hydrostatic pressure, (3) changes in water content, (4) changes in liquid and gas volume due to changes in hydrostatic pressure, and (5) changes in liquid and gas volume due to changes in temperature. Changes of liquid and gas volume due to creep strain are also considered. In contrast to the model by Alonso (1998), heat transfer is not considered. Only mass balance and force equilibrium equations are coupled.

Calibration of the model is based on experimental results on concrete specimens with a width, length and height of 40cm, 60cm, and 15cm respectively. Variations in flow rates and hydraulic conductivity as a function of crack width are considered. Model predictions match closely with experimental results for the relationship between total flow rate and crack width. Both flow rate and hydraulic conductivity appears to increase exponentially as a function of crack width (figure 10.1).

1.0E-111.0E-101.0E-091.0E-081.0E-071.0E-061.0E-051.0E-041.0E-031.0E-021.0E-01

1.0E+000 0.1 0.2 0.3 0.4 0.5 0.6

Crack width (mm)

Hydr

aulic

con

duct

ivity

(cm

/s)

Figure 10.1 Non-linear increase in hydraulic conductivity with increase in crack width (adapted from Oshita and Tanabe, 2000c)

77

Integrated Model of the Climatic Effects on Pavements

A comprehensive climatic model of water movement in pavements was developed by Pufahl et al. (1993). This model is an integration of four other existing models: (1) the precipitation model, (2) the infiltration and drainage model, (3) the climatic-materials-structural model (CMS Model), and (4) the U.S. Army Cold Regions Research and Engineering Model (CRREL Model). The integrated model is a successful attempt to combine these four independently developed modules into one model. Comparison of measured data with those computed using the model show close agreement for field sites in Illinois and Texas (Pufahl et al. 1990).

The input data required for the model is divided into three groups: (1) rainfall data, (2) pavement data, and (3) meteorological data (see figure 10.2). Rainfall and average monthly wind speed data are provided in data files of the program and are specific for each of the nine climatic regions of the U.S. A rainfall estimation program uses these past climatological data and employs stochastic processes and random methods to estimate rainfall patterns and its effect on moisture of the pavement (Liang and Lytton, 1989). The pavement specifics data is input by the user by selecting from the list of typical values. The input data are processed by the four modules and the output is provided in two windows. Structural sections of pavements are modeled by different modules depending on temperature conditions (see figure 10.3).

ment infiltration parameters

Input 1 Rainfall Data

- Monthly amount - Number of wet days - Number of thunderstorms

Input 2 Pavement geometry Physical and thermal material properties Initial soil suction profileInitial soil temp. profile Heat transfer coefficient Rainfall intensity coefficient Pave

Precipitation Model

Infiltration – Drainage Model

CMS Model CRREL Model

Output Soil temp. profile with time Soil suction profile with time Frost penetration with time Thaw depth with time Surface heave with time Degree of drainage with time Dry and wet probabilities of base course Adequacy of base course design

Output Asphalt stiffness with time Base and subbase mod. With time Subgrade mod. With time Climatic data

Input 3 Average monthly wind speed Sunshine percentage Max. and min. air temperature Solar radiation

Figure 10.2 Schematic interpretation of interaction of modules (adapted from Pufahl et al., 1990)

78

Above freezing Below Freezing

Moisture Temperature Moisture Temperature

Upper Weather Boundary Condition Precipitation Model CMS CMS

Asphaltic Concrete (or PCC) Infiltration-Drainage Model CMS CMS

Base Course Infiltration-Drainage Model CRREL CRREL Subbase Course Infiltration Drainage Model CRREL CRREL Intermediate Boundary Conditions

Subgrade CRREL CRREL CRREL CRREL

Bottom Boundary Conditions CRREL CRREL CRREL CRREL

Figure 10.3 Pavement segments where component models are used (adapted from Pufahl et al. 1990)

There are three boundary conditions in the integrated model (Pufahl and Lytton, 1991). The

first is at the surface of the pavement, the second is at the top of the subgrade and the third is in the order of 3.0 to 4.6 m into the subgrade. The second boundary condition is included because infiltration of water through the pavement during or after a rainstorm often produces free water in the base and subbase which forms a steep gradient in water pressure at the subbase – subgrade interface. Subbase has a positive water pressure and subgrade has a negative water pressure or suction. This model has been updated since its first implementation in 1989 and its most current version can be downloaded from the world wide web without a cost (http://uiairpave.ce.uiuc.edu/icm/). The current version is a finite-difference, two dimensional Windows™ based model. It comes with a highly developed Help menu. It outputs temperature, resilient modulus, pore water pressure, water content, frost and thaw depth, frost heave, and drainage performance for up to 30 pavement layers. Output data may be presented in graphical or tabular fashion. The program is capable of using either metric or English units.

Relevant Modules of Hazardous Waste Identification Rule (HWIR) Unsaturated Zone Module

The unsaturated zone module is only one of the many modules in the HWIR model. The major role of the unsaturated zone module within HWIR is to provide inputs to the saturated zone module. The unsaturated zone module simulates the migration of water and a contaminant between the top of the unsaturated zone and the water table. The module estimates annual average contaminant mass flux from the source to the water table and feeds this data as input to the saturated zone module. The unsaturated zone model is a one dimensional, steady state model that assumes that the flow from underneath the source travels vertically towards the water table. Contaminant transport is modeled using advection and dispersion. The module can simulate both steady state and transient transport, with single or multiple species chain decay reactions and linear or nonlinear sorption. One limitation of the module is that it does not consider partitioning of contaminant into the air phase. Thus, it is assumed that no mass transfer occurs between the soil vapor and air above the soil and the mass entering the groundwater conservatively is overestimated.

79

Saturated Zone Model The saturated zone model estimates annual average concentrations of contaminants (1) at one or

more water supply wells, and (2) influent into a single, hydraulically-connected, intercepting (gaining) stream. Input data could be from infiltration from the bottom of a waste source (unsaturated zone model) or recharge from outside the source area. The model simulates horizontal flow in an unconfined aquifer with approximately uniform saturated thickness that is bounded by impermeable aquitard at the bottom. Contaminant transport is modeled by advection, hydrodynamic dispersion, and degradation. Contaminants may be modeled as single species or multiple species, chain-decay reactions, and linear sorption. Both steady-state and transient three-dimensional transport in the aquifer can be simulated. Recycled Materials Fate and Transport Model (IMPACT)

At Oregon State University, a comprehensive laboratory and model development research program was conducted on environmental impact of construction and repair materials on surface and ground waters (Huber et al., 2001). As part of the research, a numerical fate and transport model, IMPACT, was developed in Visual Basic for Applications for use in conjunction with the Excel 7.0 spreadsheet (Microsoft has incorporated Visual Basic for Applications into Excel 7.0 as the primary macro language). The model predicts changes in aquatic toxicity and contaminant concentrations as they migrate to soil and possibly to ground and surface water near highways. A finite difference scheme is used to simulate the changes in concentration of the constituent as it migrates through the soil.

IMPACT is for use in the near-highway environment (over a scale of meters) and consists of fate and transport analyses related to removal, reduction, and retardation (RRR) processes, plus generation of initial pollutant loadings. Transport processes of advection and dispersion in soil are coupled to the RRR processes of sorption, biodegradation, photolysis and volatilization. Six different field conditions are considered. Source term contaminant leaching is modeled for (1) runoff from impermeable highway surfaces or bridges, (2) infiltration through permeable highway surfaces, (3) infiltration through permeable highway surfaces and subsurfaces, (4) flow in culverts, (5) embankments, and (6) flow through an exploration drill hole (see figure 10.4). Generation of constituents in the runoff uses leaching functions determined in the laboratory. After generation, the constituents are routed down the appropriate surface and/or subsurface pathway where they may be subject to RRR processes. Model output consists of flows, loads (mass), concentration of surrogate chemical, and toxicity of tested construction and repair materials in their appropriate reference environments.

80

Rain

Pavement

Description: Runoff from impermeable highway surface.Pathways: Surface flowPrimary Processes: Photolysis, VolitilizationSource Term Model Parameter:Maximum leaching capacity

(leachate extraction), Flat Plate Leaching (typical value)

Runoff

Rain

Pavement

Description: Runoff through permeable highway surface.Pathways:Subsurface, Surface flowPrimary Processes:Sorption, Biodregradation, Photolysis, VolitilizatioSource Term Model Parameter:Maximum leaching capacity

(leachate extraction), mass transfer rate (column test)

Rain

Pavement

Description: Runoff through permeable highway surface.Pathways:Subsurface flowPrimary Processes:Sorption, BiodregradationSource Term Model Parameter:Maximum leaching capacity

(leachate extraction), mass transfer rate (flat plate leachin

Recycled Fill

Pavement

Description: CulvertPathways: Surface flowPrimary Processes:Sorption, Photolysis, VolitilizationSource Term Model Parameter:Maximum leaching capacity

(leachate extraction), Mass transfer rate (flat plate leachin

Water Flow

Rain

Description: PilingPathways:Subsurface flowPrimary Processes:Sorption, BiodregradationSource Term Model Parameter:Flat Plate leaching

mass transfer rate (flat plate leachin

Pile

Rain

Description: Bore Hole.Pathways:Subsurface flowPrimary Processes:Sorption, BiodregradationSource Term Model Parameter:Maximum leaching capacity

(leachate extraction), Mass transfer rate (flat plat leaching

Exploration Drill Hole

Ground Water Ground Water

Ground Water

Rain

Ground Water Ground Water

Figure 10.4 Highway reference environments for fate and transport model application

For verification, results of the model were compared to ten column studies and it was found

that the model estimated solute concentrations equal to or exceeding actual concentrations in solution. A more comprehensive verification of the model especially with field data is required. The authors suggest that the model be used as a screening tool since its predictions are possibly a good representation of a worst-case scenario. The authors believe that the model is probably capable of providing a sense of the potential for harmful interaction of leachate from new road construction with the surrounding biota.

81

The model has several disadvantages. It cannot simulate heterogeneous or structured soils

(except for layering), preferential flow, changes in soil moisture content, rate-limited sorption, and chemical reaction and complexation within the soil matrix. The model extrapolates leaching rates for large highway surfaces from small (76cm ) flat plate studies and column studies for fill materials in the laboratory. The empirical factors used in the model also represent a source of uncertainty. The following assumptions apply for the model: (1) the effects of leaching and individual environmental effects, which are tested in dependently, can be superimposed, (2) the sorbed and dissolved solute is in equilibrium, sorption phenomena are not rate limited, and sorption is reversible, (3) the flow is uniform and unidirectional (downward from reference environment), and (4) soil moisture is constant over the course of a model run.

2

0.33 Asphalt Concrete (high leakage)

The IMPACT model refrains from sophistication for modeling vadose zone water flow and

does not use Richards Equation because (1) the primary interest is in the total mass of constituent transported as opposed to its vertical distribution and (2) the model is too simple to account for uncertainties about hydraulic properties of the soil. If surface infiltration is less than the saturated hydraulic conductivity of the soil, then the vertical water velocity (Darcy velocity) is set equal to infiltration. If surface infiltration is greater than the saturated hydraulic conductivity then the vertical water velocity is set equal to the saturated hydraulic conductivity. To calculate seepage or pore velocity, the rate at which a drop of water migrates through the soil pores and the rate at which constituents are advected through the soil, the Darcy velocity is divided by effective porosity. In short, flow through the vadose zone is modeled simplistically by assuming rate- and supply-limited infiltration and does not employ a mass balance approach (’s equation) commonly used in more sophisticated models.

It is worthwhile to present the approaches used in IMPACT for crack infiltration and vadose

zone transport modeling because they relate closely to the rest of this report. In addition, IMPACT is the first model developed specifically for predicting the impact of beneficial use of recycled materials. IMPACT allows the user to pick one of the three methods available to model crack infiltration. The first available approach is Ridgeway’s equation (equation 9.1) described in section 9.2 of this report. The second approach is quite simplistic; Cedergren’s rainfall factors for different pavement types (multiplied by the 1-hour duration 1-year frequency precipitation rate for a particular location) are used to estimate infiltration (see table 10.2). This approach is suggested for use in the total lack of site-specific field measurements. The third approach uses runoff calculations of Huber (1993) and estimates infiltration by subtracting runoff from precipitation. The IMPACT model requires that for any of the previous methods, the user must manually input the desired infiltration rate, with guidance (help screens) provided by the model. This requires the user not to blindly use the approximations and include a professional assessment of the field situation to be modeled.

Table 10.2 Rainfall factors for design of highway subbase drainage recommended by Cedergren (1974)

Pavement Type Factor Asphalt Concrete (low leakage)

0.5 Portland Cement Concrete (low leakage) 0.5 Portland Cement Concrete (high leakage) 0.67

82

PURDRAIN Model PURDRAIN is an unsaturated medium model written in PASCAL at Purdue University in

1992. The theory of transient moisture flow in unsaturated porous media is modeled by the Brooks and Corey (1964) soil water retention approach and the Van Genuchten (1980) conductivity approach. The model was calibrated and validated based on data for hydraulic properties of base/subbase materials and subgrade soils that was created from laboratory and field data as discussed by Ahmed (1993).

PURDRAIN model can handle one and two-dimensional analyses of moisture infiltration and subsequent redistribution in a multi-layer system. Relative degrees of saturation, piezometric heads and moisture contents are evaluated for pavement systems with various geometry, material and hydraulic properties. Outflow from a pavement subdrainage system can also be predicted for precipitation events on a time basis. The governing equation in the model is as follows:

(Eq. 10.20)

where:

C = water retention coefficient,

Equation 10.20 is a formulation of Darcy’s Law modified for unsaturated media (Richards Equation). This equation is a second order differential equation with non-linear coefficients C and K . Because it is a highly non-linear differential equation and coefficients C and K depend on the solution of the problem, an iterative procedure is used to solve for the unknowns. In PURDRAIN, a finite difference method solution to the problem can be achieved by the Gauss-Seidel, Jacobi, or over relaxation procedures (Espinoza and Bourdeau 1992). This model has not been updated and a windows based version of it does not exist at this time.

w

w

+

=

dzdK

dzd

dxdK

dxd

dtdC zxw

φψφψφ )()(

ψθ

ddCw −=

wΦ= Piezometric head = z-Ψ [L], Ψ = matric suction =Ψ(θ) [L], K(Ψ) = hydraulic conductivity as a function of Ψ [L/T], θ= volumetric moisture content.

w w

10.4 Need for Commercially Available Models and Model Selection A wide range of approaches was presented for modeling water movement in highways. The models contained empirical equations, mathematical expressions developed from statistical analyses of field data, and equations that represented our theoretical understanding of the problem. In conceptual models, mass and energy balance equations were coupled and solved in light of their effect on mechanical properties. The most problematic variable in these approaches was determined to be the unsaturated hydraulic conductivity. Without a universal agreement on a single equation, prediction of unsaturated hydraulic conductivity from theoretical and empirical equations remains ambiguous. Once complicated partial differential equations are set up, their numerical (and analytical) solution is yet another challenge.

83

The models presented in sections 10.2 and 10.3 aid in understanding approaches for modeling water movement in pavements. The simple approaches presented in section 10.2 are useful but too simple to be adapted to meet the purposes of the present research. Models based on soil physics (section 10.2) and strength of the pavement provide a nice framework for how strength and moisture content are inter-related. However, strength of pavements is not the focus of the paper and dwelling on mechanical properties would be a significant deviation from the goal of the present research, which is to develop a risk based, source term leaching model.

For purposes of the present study, a robust model that will handle contaminant transport in

addition to water flow is necessary. The first step in the current research is to model water movement; next step will be to couple water movement with transport of metals and possibly organic chemicals. The Integrated Model of the Climatic Effects on Pavements model would have been a strong candidate to be used in the present study if it could model contaminant transport in addition to water and temperature. Once another model is selected, it might be interesting to compare the selected model’s water movement results with those of the Integrated Model of the Climatic Effects. The IMPACT model was developed for a similar purpose as the present study; however, it is too simplistic. It may be worthwhile to evaluate the IMPACT model by comparing its results with a more sophisticated model. The HWIR model also seems to be suitable for the present project, however, its limited spatial capability (one-dimensional modeling) and semi-analytical solution feature are not desirable. None of the models described so far were capable of satisfying the criteria described in the next paragraph.

Based on discussions in this report, a model to be used for the present research would ideally be

capable of: - modeling unsaturated water movement and contaminant transport, - modeling infiltration through cracks and joints in presence and absence of ponding, - handling hysteresis of soil moisture, - modeling primarily metals and preferably organics, - modeling transport terms (advection, dispersion, diffusion), - modeling retarding surface reactions (dissolution/precipitation, sorption), - calculate activity or aqueous phase reaction (complexation, redox, acid/base reaction), and

- modeling heterogeneous media (e.g. different layers of the pavement structure such as wearing surface, base, subbase, subgrade, and different vertical sections of the pavement such as shoulders and the rest of the pavement), and

- varying groundwater table depth as it changes after rain events. To account for crack infiltration, the spatial capability of the model should be at least two

dimensional. Among analytical, finite difference and finite element techniques, the finite element method is preferable because irregular geometries such as cracks and variable surface geometry (crowning of the road, transverse and lateral cracks) can be handled more easily. Analytical models should not be considered because they cannot handle complicated criteria.

Some modeling features that are desired but not as important are (1) modeling of vapor transport of water and organics, (2) heat transport modeling, (3) saturated water movement and contaminant transport modeling, (4) modifiable code preferably written in an object-oriented language (e.g. C++), (5) capability to incorporate geographical information systems (GIS) data, and

84

(6) statistical output instead of a single value corresponding to a single set of input. Statistical output may not be as crucial since it can be simulated manually. In addition, a strong customer support is crucial and pre- and post-processors are necessary.

Eventually, the model will be used to determine the source term of leaching for recycled materials used in the highway environment. Thus, it is also crucial that the model allow placement of contaminants at varying locations (e.g. base, subbase, shoulders, and embankment) and concentrations. Some of the desired outputs of the model will be an analysis of (1) contaminant concentration and water content as a function of depth, distance and time, (2) the effect of various initial contaminant concentrations and water content at various pavement sections (asphalt or PC concrete, base, shoulders, embankments), and drainage structures. Web sites of International Groundwater Modeling Center (www.mines.edu/igwmc), Geotechnical and Geoenvironmental Software Directory (www.ggsd.com), Scientific Software (www.scisoftware.com), and Waterloo Hydrologic (www.flowpath.com) provide a collective comprehensive resource for groundwater or pavement drainage models. Among 50 unsaturated zone models selected from these sites, only 10 contaminant transport and water flow models could account for heterogeneity and preferential flow (see Table 10.3). HYDRUS-2D, a highly sophisticated two-dimensional model recommended by other researchers modeling pavement water flow, and FEMFAT3D, a reasonably priced three-dimensional model were selected based on price information in addition to criteria stated previously.

Chapter 10 Synopsis Sections 10.2 and 10.3 reviewed simplistic and more comprehensive approaches to modeling water movement in the highway environment. Soil mechanics approaches, pure hydrology approaches, and simplistic empirical approaches were presented; however none were a good match with the purposes of the present study. The IMPACT model, HWIR model, and the Integrated Model of the Climatic Effects on Pavements may be used in conjunction with another, more sophisticated model that will be selected. Commercially available unsaturated zone water flow and contaminant transport models achieve high sophistication in the sense that they can simulate heterogeneous media, preferential flow (crack infiltration) and sometimes hysteresis using various modules. Variable surface geometry may also be modeled if a finite element model is selected. Based on these criteria and model prices, among ten models that were compared HYDRUS-2D, and FEMFAT 3D were selected for use in the next phase of the present study.

85

Table 10.3 Comparison of commercially available unsaturated zone groundwater flow and contaminant transport models

HYDRUS 2D with MESHGEN

SEEP/W and CTRAN/W Model FATMIC 3D FEFLOW FEMFAT 3D Hydrogeochem SUTRA R-UNSAT VAM2D VS2DT or VS2DI

Solution technique

Hybrid: finite element & finite difference

2D-finite difference, 1D-analytical Finite element Finite element Finite element Finite Element Finite element Finite element Finite element Finite difference

Advection, dispersion, diffusion, aqueous

complexation, adsorption,

desorption, ion exchange, ppt,

dissolution, redox and acid base reactions

Contaminant transport

Chemical and biological

transformation

Convective and dispersive contaminant

transport with adsorption,

hydrodynamic dispersion, first order chemical reaction

Advective-dispersive transport, first order decay, equilibrium reactions between liquid and solid

phases and liquid and gaseous phases

SEEP/W computes water velocity,

content flux while CTRAN/W computes

contaminant migration. The two together compute

advective and dispersive flux,

molecular diffusion, and adsorption.

Advection, hydrodynamic dispersion, equilibrium sorption, and first-order

degradation

Sorption, advection/convection, dispersion/diffusion

Advection, dispersion, decay, adsorption, ion-

exchange

Solute absorption, production, and decay Models organics only

Saturated or Unsaturated

Saturated and unsaturated

Saturated and unsaturated

Saturated and unsaturated

Saturated and unsaturated

Saturated and unsaturated

Saturated or unsaturated Unsaturated Unsaturated and saturated Unsaturated and

saturated

Dimensions 3D 2D, 3D 3D 2D 2D 2D 2D 2D

Variable boundary

conditions of evaporation,

infiltration, or seepage on the

soil-air interface for

the flow module and

variable boundary

conditions of inflow and outflow for the transport

module automatically

Mass (1st type,

Dirichlet), flux (2nd

type, Neumann),

transfer (3rd type,

Cauchy), well (4th type)

Flow and concentration:

prescribed concentration-head,

gradient flux, total flux, river boundaries

Saturated and unsaturated

3D 2D

Prescribed total analytical

concentrations on Dirichlet boundaries, Prescribed

fluxes on flow-in boundaries,

natural advective fluxes

on flow-out boundaries, all

boundary values are spatially-

and temporally-dependent

Constant or time-varying) prescribed

head and flux boundaries,

boundaries controlled by atmospheric

conditions, as well as free drainage

boundary conditions. Soil surface

boundary conditions may change during the simulation from prescribed flux to

prescribed head type conditions.

Seep/W:Constant hydraulic head

and/or flux, Transient head and/or flux as a

function of time, Flux as a function of

computed head

Fixed pressure heads, infiltration with

ponding, evaporation from the soil surface, plant transpiration, or

seepage faces.

A time-dependent-concentration,

specified-flux, or specified-concentration

boundary condition

Seepage faces, water table, recharge

infiltration, evapotranspiration,

pumping wells

Boundary conditions 1

May vary with time or be constant

Heterogeneous media

Yes (also anisotropic) Yes Yes Yes (also anisotropic) Yes Yes Yes Yes Yes Yes

Pre and postprocessors None Yes Yes Yes Yes Yes with ARGUS

ONE No Yes Yes

Heat transport Heat transport Isothermal Isothermal Heat transport Isothermal Isothermal Isothermal Isothermal

Code language FORTRAN Fortran Fortran Fortran Fortran Fortran Fortran - -

Temperature Nonisothermal

ANSI C and C++

86

11. Summary and Conclusions The purpose of this study was to describe the state of the knowledge about water movement in

the highway environment so that this knowledge may be incorporated into fate/transport models for use in risk assessment. The literature review covered a wide variety of topics including drainage systems, measurement techniques, hydraulic conductivity of pavement layers, and affects of cracking, pumping, temperature, and mechanical properties on water movement.

Measurement techniques for moisture content, pore water pressure, and rainfall were discussed.

For measuring in situ water content, TDR is a growing and promising technique. However, there is not a universal agreement on a TDR calibration equation. Many studies use the common Topp et al. (1980) equation even though it may not be accurate for base and subbase measurements. To ensure quality data, TDRs may be installed before construction and may be calibrated for specific composition of the material tested. Information on common techniques for measuring pore water pressure in pavement structures is sparse. Most of the studies suffice by measuring moisture content only. Loi et al. (1992) discussed applicability of thermal conductivity sensors for measuring pore water pressure in pavement studies. In pavement water movement studies, precipitation also needs to be measured to quantify drainage or specify climate conditions of pavements. Mostly, a tipping bucket is used for measuring rainfall or edge drain outflow.

Hydraulic conductivity of pavement layers and its measurement was discussed in detail

throughout the report. Laboratory techniques for measuring saturated hydraulic conductivity center on constant and falling head permeameter tests that should be conducted at low hydraulic gradients, and in addition, laminar and horizontal flow conditions if field conditions are to be simulated. Modifications of constant and falling head procedures to measure horizontal hydraulic conductivity in base/subbase layers are presented in the literature. For pavement structures, there are no established field techniques for measuring saturated hydraulic conductivity, and discussion of measurement techniques for laboratory and field unsaturated hydraulic conductivity was found to be minimal in the literature. In lack of field or laboratory data, hydraulic conductivity can be estimated using Lindly correlations, Moulton monograph, or Cedergren chart. In almost all cases, attempts focus on measuring saturated hydraulic conductivity. However, unsaturated conditions in pavements are widely observed in the field. Thus, more research is required to collect unsaturated hydraulic conductivity data. Standardization of both saturated and unsaturated, field and laboratory techniques for measuring hydraulic conductivity is also required.

Hydraulic conductivity data for asphalt concrete, PCC, base/subbase and subgrade layers presented, displayed wide variations due to high variability in mixture designs and in measurement techniques. In determining void content, compaction techniques (gyratory versus field compacting) should be taken into account. Many studies suggest that void content is not linearly related to hydraulic conductivity. Factors affecting hydraulic conductivity are shape, size, and interconnectivity of air voids, type of raw materials and in PCC, the curing temperature and the type and extent of chemical reactions during hardening. The void contents (6-15 percent) used in asphalt concrete designs in the U.S. result in poorer performance than designs of both free-draining or impervious asphalt concrete layers. Most of the evidence suggests that addition of recycled materials to PCC decreases hydraulic conductivity of the material after curing. Hydraulic conductivity of base/subbase layers is discussed in great detail in the literature because newer designs, including a permeable base,

87

are replacing the impervious dense-graded base layer that was used commonly until the late 1970s. Many researchers agree that at least a hydraulic conductivity of 0.34 cm/s is required for adequate drainage. Hydraulic conductivity of embankments is not widely discussed in the literature. The few values noted are quite low, suggesting that embankments are not free -raining structures.

Water enters pavements despite efforts to prevent it, but the extent of pavement deterioration can be reduced by proper drainage and maintenance. The major water ingress routes are infiltration through the pavement surfaces (through joints and cracks) and shoulders, melting of ice during the freezing/thawing cycles, capillary action, and seasonal changes in the water table. In the literature the most emphasis is placed on infiltration through crack, joint and shoulders and drainage through edge drains. Use of an impermeable material to seal surfaces can slow (not prevent) infiltration. Sealing may also cause moisture accumulation within the asphalt concrete mixture by blocking evaporation. Drainage systems (permeable base, edge drains, geotextiles) are used to remove excess water that has entered the pavement. Drainage appears to occur within 4-7 hours after precipitation. Ratio of precipitation to outflow volume from edge drains varies significantly (6-70%) depending possibly on pavement and edge drain type, geometry and condition as well as intensity and duration of precipitation.

Moisture in pavements increases or decreases after construction until equilibrium is reached.

Seasonal moisture content variations and subsequent changes in pavement performance that occur after equilibrium are more pronounced in cold regions. Groundwater conditions may also influence pavement and especially subgrade water content if the ground water table is within approximately six meters from the surface.

Water pumping, temperature variations, geotechnical properties of base/subbase and subgrade

layers, and infiltration through cracks and joints significantly affect the water movement in the highway environment. Water pumping occurs due to differential pore water pressures as a consequence of moving wheel loads and causes exchange of fines between base/subbase and subgrade layers. Importance of temperature as a variable affecting water movement is often linked to freeze-thaw phenomena, yet the temperature may also be required for modeling evaporation from the pavement. Various methods for estimating pavement temperature are presented in the literature. Strength of pavement layers is linked to moisture content because matric suction pulls soil particles together increasing shear strength. Many models in the literature are developed from a soil mechanics perspective. Infiltration through cracks and joints was discussed in great detail. Variables affecting crack infiltration are hydraulic conductivity of base; crack length, width, and spacing; amount of infilling material; duration and intensity of precipitation; pavement geometry; hydraulic properties of the pavement layers; and conditions of the edge drains. The crack infiltration equation by Ridgeway (1982) is still used in the literature.

Lastly, simplistic and more comprehensive approaches to modeling water movement in the highway environment were reviewed. Among all modeling attempts presented, none were a good match with the purposes of the present study. The IMPACT model, HWIR model, and the Integrated Model of the Climatic Effects on Pavements may be used as supplemental programs once a more sophisticated model is chosen. Ultimate goal of the present project is to estimate the risk posed by use of recycled materials in the pavement. To achieve this goal, a highly sophisticated model capable of simulating unsaturated water flow and contaminant transport in heterogeneous media in at least

88

two dimensions is required. Among commercially available models, HYDRUS-2D and FEMFAT 3D were selected for use in the next phase of the present study.

12. Research Needs The report identified several measurement-technique-related areas that may benefit from more

research. As the TDR method gains acceptance, research will be needed to standardize the procedure and calibration equations. Not much information exists on unsaturated hydraulic conductivity measurement techniques both in the field and in the lab. Development and standardization of unsaturated hydraulic conductivity techniques to be used for pavement studies is warranted. In measuring saturated hydraulic conductivity, some authors use horizontal flow, but its acceptance is not resolved. More research comparing horizontal and vertical saturated hydraulic conductivities is required. Hydraulic conductivity values of asphalt concrete, PCC, base/subbase and subgrade layers abound in the literature. However, data on hydraulic conductivity of embankments is sparse. Embankment hydraulic conductivity data and comparisons of lab, field, and theoretical saturated hydraulic conductivities would supplement existing information and possibly aid in developing a universal conclusion about the relationships among lab, field and theoretical hydraulic conductivity values.

More research is required on relative significance of ingress and egress routes; mainly crack

infiltration, evaporation, and capillary rise. To model crack infiltration, many studies still utilize equations developed by Ridgeway (1982) even though contradictory evidence hinting on more sophisticated nature of crack infiltration exists. Variables affecting crack infiltration are: hydraulic conductivity of base; crack length, width, and spacing; amount of infilling material; duration and intensity of precipitation; pavement geometry; hydraulic properties of the pavement layers; and conditions of the edge drains. Research is needed to develop quantitative relationships between these factors and infiltration and drainage rates. Relative contribution of capillary rise and evaporation to moisture content seems to be ignored in the literature by assuming they are minor. Research examining these two routes would provide a solid evidence for validity or inappropriateness of this assumption. The use of sophisticated commercially available unsaturated zone models to examine water ingress and egress routes and estimate moisture contents seems possible although not widely practiced. Models presented in section 10.3 should be tested to verify their applicability to modeling water movement in pavement structures.

89

Appendix

A. Recommended Hydraulic Conductivity Values Koch and Sandford (1998) cited a report by Manion et al. (1995), which also has a compilation of recommended values for free-draining subbase layers that include some references not used in the present study. The table in Koch and Sandford (1998) is included here. Recommended hydraulic conductivity (cm/s) Author

Yoder and Witczak (1975) At least 0.02 to 0.04 Barksdale (1986) 0.25 to 2.8 Ridgeway (1982) 0.35 to 1.06 typical Crovetti and Dempsey (1993)

Cedergren (1994) 1.7 to 35 Cedergren (1987) Greater than 0.35 FHWA (1990)

B. ASTM and AASHTO Standards Cited in This Report ASTM standards: D 1241 –68 (Reapproved 1994) Standard specification for materials for soil-aggregate subbase, base, and surface courses

D 2434 Test method for hydraulic conductivity of granular soils (Constant head) D 3017-96 Standard test method for water content of soil and rock in place by nuclear methods (Shallow depth) D 3152-72 (Reapproved 1994) Standard test method for capillary-moisture relationships for fine-textured soils by pressure-membrane apparatus D 3404-91 (Reapproved 1998) Standard guide for measuring matric potential in the vadose zone using tensiometers D 4643-93 Standard test method for determination of water (moisture) content of soil by the microwave oven method D 4959-89 (Reapproved 1994) Standard test method for determination of water (moisture) content of soil by direct heating method D 5084-90 (Reapproved 1997) Test method for measurement of hydraulic conductivity of saturated porous materials using a flexible wall permeameter D 5126-90 (Reapproved 1998) Standard guide for comparison of field methods for determining hydraulic conductivity in the vadose zone

D 5298-04 Standard test method for measurement of soil potential (suction) using filter paper D 5856-95 Standard test method for measurement of hydraulic conductivity of porous material using a rigid-wall, compaction-mold permeameter AASHTO Standards: T 99 The Moisture-Density Relations of Soils Using a 2.5 kg (5.5 lb) Rammer and a 305 mm (12 in) Drop T 239-91 Moisture content of soil and soil-aggregate in place by nuclear methods (Shallow depth)

0.35

3.5 to 35

D 2325-68 (Reapproved 1994) Standard test method for capillary-moisture relationships for coarse- and medium-textured soils by porous-plate apparatus

D 5220-92 Standard test method for water content of soil and rock in-place by the neutron depth probe method

90

T 255-92 Total moisture content of aggregate by drying T 265-91 Laboratory determination of moisture content of soils T 273-86 (1993) Soil suction T 255-92 Total moisture content of aggregate by drying

C. Standard Specifications for Materials for Soil-Aggregate Subbase, Base, and Surface Courses (ASTM D1241-68) Type I mixture: Mixtures consisting of stone, gravel or slag with natural or crushed sand and fine mineral particles passing a No. 200 (75µm) sieve and conforming to the requirements of Table E.1 for gradation A,B,C, or D. Type II mixture: Mixtures consisting of natural or crushed sand with fine mineral particles passing a No. 200 (75µm) sieve, with or without stone, gravel, or slag, and conforming to the requirement of Table D.1 for gradation E or F. Coarse aggregate: Coarse aggregate retained on a No. 10 (2.00-mm) sieve, for use in Type I and Type II mixtures, shall consist of hard, durable particles or fragments of stone, gravel, sand, or slag; materials; materials that break up when alternately frozen and thawed or wetted and dried shall not be used. Fine aggregate: Fine aggregate passing a No. 10 (2.00-mm) sieve, for use in Type I and Type II mixtures, shall consist of natural or crushed sand and fine minerals particles passing the No. 200 (75µm) sieve. The fraction passing the No. 200 sieve shall not be greater than two thirds of the fraction passing the No. 40 (425µm) sieve. The fraction passing the no. 40 sieve shall have a liquid limit not greater than 25 and a plasticity index not greater than 6.

Table C.1 Gradation requirement for soil-aggregate materials Weight percent passing square mesh sieves

Type II (Square openings) Gradation A Gradation B Gradation C Gradation D Gradation E Gradation F

2-in. (50-mm) 100 100 … … … 1-in (25.0 mm) … 75 to 95 100 100 100 100

30 to 65 40 to 75 50 to 85 60 to 100 … … 35 to 65 50 to 85 55 to 100 70 to 100

No. 10 (2.00mm) 15 to 40 40 to 70 40 to 100 55 to 100 No. 40 (425µm) 8 to 20 15 to 30 15 to 30 25 to 40 20 to 50 No. 200 (75µm) 2 to 8 5 to 15 5 to 15 8 to 15 8 to 15

D. Terminology Blast furnace slag: Non-metallic by-product of iron production in blast furnaces, consisting essentially of silicates and alumina-silicates of lime and other bases. Boiler slag: Molten ash from high-temperature pulverized coal burning power plants (wet-bottom furnace) that is water-quenched, forming angular, black, “glassy” particles.

Type I Sieve Size

…

3/8 in (9.5 mm) No. 4(4.75mm) 25 to 55 30 to 60

20 to 45 25 to 50 30 to 70

6 to 15

Bottom ash: The heavier of two ashes given off by coal-fired electric generation plants. Bottom ash falls to the bottom of the furnace and mixes with slag. Cullet: Waste glass suitable for remelting. Coarse-graded mixes: Superpave mix plotting below the maximum density line and restricted zone

91

Dense-graded aggregate base: Mixture primarily sand and gravel, well-graded from coarse to fine. It can be unstabilized or cement or asphalt stabilized. When compacted the mixture has only low air voids and is essentially impermeable to water. Fill: Material placed to level or raise the height of a site. Fine-graded mix: Superpave mix plotting above the maximum density line and restricted zone Flowable fill: A mixture of sand, fly ash, a small amount of cement and water. Foundry sand: Silica sand used in ferrous and non-ferrous foundries in the moulds, that becomes ‘contaminated’ during the casting process (spent sand or waste sand). Fly ash: Siliceous, fine, solid material given off by coal-fired electric generation plants. Fly ash is the lighter ash that escapes with the flue gas and is trapped in and removed by a bag house. Hot-mix asphalt: Designed aggregate and asphalt cement mix produced in a hot-mix plant where the aggregates are dried, heated and then mixed with heated (fluid) asphalt cement (hot mix), then transported, placed and compacted while still at an elevated temperature (about 125o-135oC) to give a durable, deformation resistant, fatigue resistant pavement course. Open-graded friction course: Special-purpose hot asphalt mixes used to improve surface fictional resistance, minimize hydroplaning, reduce splash and spray, improve night visibility, and lower pavement noise levels. These functions are achieved by high percentage of internal air voids that quickly remove water from the pavement surface. Open-graded mixes are generally characterized by a large percentage of one size of coarse aggregate-usually ½ or 3/8 inch maximum particle size and little or no fines in the mix. Pavement: All elements from the wearing surface of a roadway to the subgrade. Includes the surface pavement (asphalt or PCC), the base and/or permeable base, and the subbase. Pseudo-hydraulic conductivity: During a hydraulic conductivity test, if flow is turbulent, and Darcy’s Law is invalid, pseudo-hydraulic conductivity is measured by polynomial or exponential models. Reclaimed asphalt pavement: Asphalt cement concrete removed by milling machines during pavement rehabilitation. Rutting: Transverse movement of mixtures of surface and binder course generated at wheel path of carriageway in the cases of warmer climate areas, heavily trafficked roads, approaches to intersection, climbing lanes. Slag: Molten by-products, from the smelting or sintering of metallic ores, that are cooled by various methods. Tire chips: Scrap tires that have been cut into pieces with a maximum dimension between 75 to 300mm.

92

References AASHTO. 1993. Standard specifications for transportation materials and methods of sampling and

testing, 16th edition. AASHTO. 1998. Supplemant to the AASHTO guide for design of pavement structures. Part II - Rigid pavement design and rigid pavement joint design. U.S. Department of Transportation. Ahmed, Z., T. D. White, and P. L. Bourdeau. 1993. Pavement drainage and pavement-shoulder joint evaluation and rehabilitation. Purdue University and Indiana Dept. of Transportation, West

Lafayette. Ahmed, Z., T. D. White, and T. Kuczek. 1997. Comparative field performance of subdrainage systems. Journal of Irrigation and Drainage Engineering 123:194-201. Aldea, C.-M., Shah, S. P., and Karb, A. 1999. Effect of microcracking on durability of high-strength concrete. Transportation Research Record, 1668, 86-90. Ali, H. A., and A. Lopez. 1996. Statistical analysis of temperature and moisture effects on pavement

structural properties based on seasonal monitoring data. Transportation Research Record 1540:48-55.

Alobaidi, M., and D. J. Hoare. 1996. The development of pore water pressure at the subgrade-subbase interface of a highway pavement and its effect on pumping of fines. Geotextiles and Geomembranes, 14(1):111-135.

Alonso, E. E. 1998. Suction and moisture regimes in roadway bases and subgrades. Pages 57-105 in E. J. Hoppe, editor. International Symposium on Subdrainage in Roadway Pavements and Subgrades. Grafisaff, Granada, Spain.

Anderson, K. W. 1996. Advocates and aggregates: Washington state includes new partners in recycling. TR News 184:8-13.

Andrew, J. W., N. M. Jackson, and E. C. Drumm. 1998. Measurement of seasonal variations in subgrade properties. Pages 13-38 in A. T. Papagiannakis and C. W. Schwartz, editors. Application of geotechnical principles in pavement engineering. ASCE, Boston, MA, USA.

Arabi, M., S. Wild, and G. O. Rowlands. 1989. Frost resistance of lime-stabilized clay soil. Transportation Research Record 1219:93-102.

Arnold, W., and M. Nishigaki. 1998. Modeling of transport of two-phase flow under non-isothermal conditions. Pages 113-121 in E. J. Hoppe, editor. International Symposium on Subdrainage in Roadway Pavements and Subgrades. Grafistaff, Granada , Spain.

ASTM. 2000. Annual book of ASTM standards, Vol 04.09 Bakker, R. F. M. 1983. Hydraulic conductivity of blended cement concretes. Pages 589-605 in V. M.

Malhotra, editor. Fly Ash, Silica Fume, Slag and Other Mineral By-Products in Concrete. American Concrete Institute, Detroit.

Ball, J. E., R. Jenks, and D. Aubourg. 1998. An assessment of the availability of pollutant constituents on road surfaces. The Science of the Total Environment 209:243-254.

Barber, E. S., and C. L. Sawyer. 1952. Highway Subdrainage. Pages 643-666 in HRB. Bardet, J.-P. 1997. . Upper Saddle River, Prentice Hall. Experimental Soil MechanicsBarksdale, R. D., and R. G. Hicks. 1977. Drainage considerations to minimize distress at the

pavement-shoulder joint. Pages 383-398 in International Confernece on Concrete Pavement Design, Purdue University.

Baumgardner, R. H. 1992. Overview of permeable bases. Pages 275-287 in Materials: performance and prevention of deficiencies and failure, New York.

Bear, J. 1979. , McGraw Hill. Hydraulics of Groundwater

93

Bedient, P. B., H. S. Rifai, and C. J. Newell. 1999. Ground water contamination. Prentice Hall, Upper Saddle River.

Benson, C. H., and M. M. Gribb. 1997. Measuring unsaturated hydraulic conductivity in the laboratory and field. Pages 113-168 in S. L. Houston and D. G. Fredlund, editors. Geotechnical Special Publication No. 68 -Unsaturated Soil Engineering Practice. ASCE, Logan, Utah.

Berbee, R., G. Rijs, R. d. Brouwer, and L. v. Velzen. 1999. Characterization and treatment of runoff from highways in the Netherlands paved with impervious and pervious asphalt. Water Environment Research 71:183-190.

Berry, E. E., and V. M. Malhotra. 1978. Fly ash for use in concrete, Part II - A cricitical review of the effects of flyy ash on the properties of concrete. CANMET Report 78-16, Canada Center for Mineral and Energy Technology.

Beven, K., and P. Germann. 1982. Macropores and water flow in soils. Water Resources Research 18:1311-1325.

Bhat, S. T., and C. W. Lovell. 1996. Design of flowable fill: waste foundry sand as a fine aggragate. Transportation Research Record 1546:70-78.

Bhat, S. T., and C. W. Lovell. 1997. Mix design for flowable fill 1589:26-28. Biczysko, S. J. 1985. Permeable subbases in highway pavement construction. Pages 81-82 in 2nd

Symp. UNBAR, Univ. of Nottingham, Dept. of Civil Eng. Birgisson, B., and R. Roberson. 2000. Drainage of pavement base material: Design and construction

issues. Transportation Research Record 1709:11-18.

Bowders, J. J., and M. A. Othman. 1994. Molding water content and hydraulic conductivity of compacted soil subjected to freeze/thaw. Transportation Research Record 1434:55-60.

Brooks, R. H., and A. T. Corey. 1964. Hydraulic properties of porous mediium, Hydrology paper 3. Colorado State University, Fort Collins.

Bumb, A. C. 1987. Unsteady[state flow of methane and water in coalbeds. University of Wyoming, Laramie.

Butler, D., and J. W. Davies. 2000. Urban Drainage. E & FN Spon, New York. Button, J. W. 1996. Hydraulic conductivity of asphalt surface seals and their effect on aging of

underlying asphalt concrete. Transportation Research Record 1535:124-130. Callahan, K. 2000. Association of State and Territorial Solid Waste Management Officials Beneficial

Use Survey. ASTSWMO, Washington, D.C. Cedergren, H. R. 1974. Drainage of highway and airfield pavements. John Wiley & Sons, New York. Cedergren, H. R., and K. A. Godfrey. 1974. Water: Key cause of pavement failure. Civil Engineering.

Cedergren, H. R. 1994. America's pavements: world's longest bathtubs. Civil Engineering 64:56-58. Chandra, D., K. M. Chua, and R. L. Lytton. 1989. Effects of temperature and moisture on the load

response of granular base course material in thin pavements. Transportation Research Record 1252:33-41.

Chesner, W. H., R. J. Collins, and M. H. MacKay. 1998. User guidelines for waste and by-product materials in pavement construction. Chesner Engineering, New York.

Choubane, B., G. C. Page, and j. A. Musselman. 1998. Investigation of water hydraulic conductivity of coarse-graded superpave pavements. Asphalt Paving Technology 67:254-276.

Bodocsi, A., I. A. Minkarah, A. Amicon, and R. S. Arudi. 1994. Development and comparison of hydraulic conductivity measurement techniques for jointed concrete pavement bases. Transportation Research Record 1434:37-46.

Cedergren, H. R. 1988. Why all important pavements should be well drained. Transportation Research Record 1188:56-61.

94

Christopher, B. R. 1998. Performance of drainable pavement systems: a U.S. perspective. Pages 205-215 in E. J. Hoppe, editor. International Symposium on Subdrainage in Roadway Pavements and Subgrades. Grafistaff, Granada, Spain.

Christopher, B. R., and V. C. McGuffey. 1997. National Cooperative Highway Research Program Synthesis of Highway Practice 239: Pavement Drainage Systems. NCHRP.

Chu, T. Y., W. R. Humphries, and S. N. Chen. 1972. A Study of subgrade moisture conditions in connection with the design of flexible pavement structures. Pages 53-66 in Third International Conference on the Structural Design of Asphalt Pavements, University of Michigan, MI.

Chu, T. Y. and W. K. Humphries. 1972. Investigation of subgrade moisture conditions in connection with the design of flexible pavement structures. Columbia, University of South Carolina College of Engineering.

Collins, R. J., and S. K. Ciesielski. 1994. Recycling and use of waste materials and by-products in highway construction. Transportation Research Board, Washington D.C.

Cooley, L. A., E. R. Brown, and D. E. Watson. 2000. Evaluation of open-graded friction course mixtures containing cellulose fibers. Transportation Research Record 1723:19-25.

Cramer, S. M., and A. J. Carpenter. 1999. Influence of total aggregate gradation on freeze-thaw durability and other performance measures of paving concrete. Transportation Research Record 1668:1-8.

Crossley, A., and S. A. M. Hesp. 2000. A new class of reactive polymer modifiers for asphalt: Mitigation of moisture damage. Transportation Research Record 1728:52-59.

Crovetti, J. A., and B. J. Dempsey. 1993. Hydraulic requirements of permeable bases. Transportation Research Record 1425:28-36.

Daleiden, J. F., and L. L. Peirce. 1997. Subsurface drainage systems in roadway construction. Transportation Research Record 1596:58-61.

Darter, M. I., and S. H. Carpenter. 1987. Techniques for pavement rehabilitation - A traning course. FHWA-HI-90-022, Federal Highway Administration, Washington D.C.

Dawson, A. R. 1985. Water movements in road pavements. Pages 7-12 in 2nd Symp. UNBAR, University of Nottingham, Dept. of Civil Eng.

Dawson, A. R., and A. R. Hill. 1998. Prediction and implication of water regimes in granular bases and subbases. Pages 121-129 in E. J. Hoppe, editor. International Symposium on Subdrainage in Roadway Pavements and Subgrades. Grafistaff, Granada, Spain.

Decker, D. S. 1993. Evaluating the use of waste materials in hot mix asphalt, Special report 165. National Asphalt Pavement Association, Lanham.

Decker, D. S. 1994. Evaluating the use of waste materials in hot mix asphalt. Asphalt Paving Technology 63:12-21.

Dempsey, B. J. 1982. Laboratory and field studies of channeling and pumping. Transportation Research Record 849:1-11.

Diefenderfer, B. K., I. L. Al-Qadi, and A. Loulizi. 2000. Laboratory calibration and in situ measurements of moisture by using time-domain-reflectometry probes. Transportation Research Record 1699:142-150.

Cooley, L. A., and E. R. Brown. 2000. Selection and evaluation of field hydraulic conductivity device for asphalt pavements. Transportation Research Record 1723:73-82.

Dempsey, B. J. 1988. Pavement drainage system design, module I. Drainage survey and evaluation. in Wisconsin DOT Highway Pavement Drainage Sumposium, University of Wisconsion, Madison.

95

Domenico, P. A., and F. W. Schwartz. 1990. Physical and Chemical Hydrogeology. John Wiley & Sons, New York.

Dore, G., j. M. Konrad, M. Roy, and N. Rioux. 1995. Use of alternative materials in pavement frost protection: material characteristics and performance modeling. Transportation Research Record 1481:63-74.

Edvardsen, C. 1999. Water Hydraulic conductivity and Autogenous Healing of Cracks in Concrete. ACI Materials Journal 96:448-454.

Eigenbrod, K. D., and J. J. A. Kennepohl. 1996. Moisture accumulation and pore water pressures at base of pavements. Transportation Research Record 1546:151-161.

Eighmy, T.T. and Chesner, W.H. 2001. Framework for Evaluating Use of Recycled Materials in the Highway Environment, Report No. FHWA-RD-00-140, U.S. DOT, Washington, D.C.

Eikelboom, R. T., E. Ruwiel, and J. J. J. M. Goumans. 2000. The building materials decree: an example of a Dutch regulation ased on the potential impac of materials on the environment. Pages 963-974 in G. R. Wooley, J. J. J. M. Goumans, and P. J. Winwright, editors. WASCON 2000, Waste Materials in Construction: Science and Engineering of Recycling for Environmental Protection. Pergamon, Harrogate, England.

El Tani, M. 1991. A permeameter for unsaturated soil. Transport in porous media. 6:101-114 Esch, D. C. 1995. Long-Term evaluations of insulated roads and airfields in Alaska. Transportation

Research Record 1481:56-62. Espinoza, R. D., and P. L. Bourdeau. 1992. Numerical modeling of moisture infiltration in pavement

systems. Pages 46-41 - 46-48 in 45th Canadian Geotechnical Conference: Innovation conservation & rehabilitation. Canadian Geotechnical Society, Toronto.

Evans, R. P., J. C. Holden, and K. J. McManus. 1996. Application of a new vertical moisture barrier construction method for highway pavements. Road & Transport Research 5:4-13.

Evans, R. P., and K. J. McManus. 1999. Construction of vertical moisture barriers to reduce expansive soil subgrade movement. Transportation Research Record 2:108-112.

Feldman, R. F. 1983. Significance of porosity measurements on blended cement performance. Pages 415-433 in V. M. Malhotra, editor. Fly Ash, Silica Fume, Slag and Other Mineral By-Products in Concrete. American Concrete Institute, Detroit.

Feng, A., J. Hua, and T. D. White. 1999. Flexible Pavement Drainage Monitoring Performance and Stability. Purdue University, West Lafayette.

FHWA. 1990. Technical paper on subsurface pavement drainage. Technical Paper 90-01, Office of Engineering, Pavement Division, Federal Highway Administration, Washington.

Fleckenstein, L. J., and D. L. Allen. 1996. Evaluation of pavement edge drains and their effect on pavement performance. Transportation Research Record 1519:28-35.

Fleming, R. R., C. D. F. Rogers, and M. W. Frost. 1998. Subgrade equilibrium water content and resilient modulus for UK clays. Pages 359-367 in E. J. Hoppe, editor. International Symposium on Subdrainage in Roadway Pavements and Subgrades. Grafistaff, Granada, Spain.

Forsyth, R. A., G. K. Wells, and J. H. Woodstrom. 1987. The Economic impact of pavement subsurface drainage. in Annual Meeting of the Transportation Research Board, Washington, D.C.

Frabissio, M. A., and N. J. Buch. 1999. Investigation of design parameters affecting transverse cracking in jointed concrete pavements. Transportation Research Record 1668:24-32.

Elsayed, A. S., and J. K. Lindly. 1996. Estimating hydraulic conductivity of untreated roadway bases. Transportation Research Record 1519:11-18.

96

Fredlund, D. G., J. K. Gan, and H. Rahardjo. 1991. Measuring negative pore water pressures in a freezing environment. Transportation Research Record 1307:291-299.

Fredlund, D. G., and H. Rahardjo. 1987. Soil mechanics principles for highway engineering in arid regions. Transportation Research Record 1137:1-11.

Fredlund, D. G., and H. Rahardjo. 1993. Soil mechanics for unsaturated soils. John Wiley & Sons, New York.

Fredlund, D. G., and A. Xing. 1994. Equations for the soil-water characteristic curve. Can. Geotech. J. 31:521-532.

Fredlund, D. G., A. Xing, and S. Huang. 1994. Predicting the hydraulic conductivity function for unsaturated soils using the soil-water characteristic curve. Can. Geotech. J. 31:533-546.

Fredlund, D. G., J. K. M. Gan, and P. Gallen. 1995. Suction measurements on compacted till specimens and indirect filter paper calibration technique. Transportation Research Record. 1481: 3-9.

Fredlund, F. G., A. Xing, M. D. Fredlund, and S. L. Barbour. 1996. The relationship of unsaturated soil shear strength to the soil-water characteristic curve. Canadian Geotechnical Journal 33:440-448.

Freeman, R. B., and G. L. Anderton. 1994. Hydraulic conductivity versus unsurfaced stability. Pages 685-692 in Proceedings of the Materials Engineering Conference Infrastructure.

Gardner, W. R. 1958. Some steady state solutions of the unsaturated moisture flow equation with application to evaporation from a water table. Soil Science 85:228-232.

Gardner, W. H. 1987. Water Content. Pages 493-512 in Methods of Soil Analysis Part 1 Physical and Mineralogical Methods, Editor Klute, A., American Society of Agronomy, Inc., Madison Wisconsin

Grogan, W. P. 1992. Performance of free draining base course at Fort Campbell. Pages 434-448 in Materials: Performance and Prevention of Deficiencies and Failure, New York.

Grogan, W. P. 1994. Evaluation of drainable base courses for pavements. Pages 693-700 in Materials Engineering Conference Infrastructure.

Gschwendt, I., and R. Stano. 1998. Evaluation and classification of subgrade moisture regimes. Pages 375-383 in E. J. Hoppe, editor. International Symposium on Subdrainage in Roadway Pavements and Subgrades. Grafistaff, Granada, Spain.

Guan, Y., E. C. Drumm, and N. M. Jackson. 1998. Weighting factor for seasonal subgrade resilient modulus. Transportation Research Record 1619:94-101.

Hagen, M. G., and G. R. Cochran. 1995. Comparison of pavement drainage systems. Minnesota Department of Transportation, St. Paul, Minnesota.

Fwa, T. F., S. A. Tan, and Y. K. Guwe. 1999. Laboratory evaluation of clogging potential of porous asphalt mixtures. Transportation Research Record 1681:43-49.

Hagen, M. G., and G. R. Cochran. 1996. Comparison of pavement drainage systems. Transportation Research Record 1519:1-10.

Hamza, A., and R. Bahar. 1998. Analysis of a transient flow through porous media. Pages 145-153 in E. J. Hoppe, editor. International Symposium on Subdrainage in Roadway Pavements and Subgrades. Grafistaff, Granada, Spain.

Hansen, E. C. 1991. Unsteady pressure-driven viscous flows beneath concrete pavement slabs. Journal of Hydraulic Engineering 117:713-724.

Hansen, E. C., R. Joannesen, and J. M. Armaghani. 1991. Field effects of water pumping beneath concrete pavement slabs. Journal of Transportation Engineering 117:679-696.

97

Henry, K. S. 1996. Geotextiles to mitigate frost effects in soils: A critical review. Transportation Research Record 1534:5-11.

Hermansson, A. 2000. Simulation model for calculating pavement temperatures including maximum temperature. Transportation Research Record 1699:134-141.

Hoffman, G. L. 1982. Subbase hydraulic conductivity and pavement performance. Transportation Research Record 849:12-18.

Houston, S. L., W. N. Houston, and T. W. Anderson. 1995. Moisture and strength variability in some Arizona subgrades. Transportation Research Record 1481:35-46.

Huang, B., L. N. Mohammad, A. Raghavendra, and C. Abadie. 1999. Fundamentals of hydraulic conductivity in asphalt mixtures. Asphalt Paving Technology 68:479-500.

Huber, W. C. 1993. Contaminant Transport in Surface Water. Pages Chapter 14 in D. R. Maidment, editor. Handbook of Hydrology. McGraw-Hill, New York.

Huber, W. C., P. O. Nelson, N. N. Eldin, K. J. Williamson, and J. R. Lundy. 2001. Environmental impact of runoff from highway construction and repair materials: project overview. in 2001 TRB Annual Meeting, Washington D.C.

Humphrey, D. N., and L. E. Katz. 2000. Water-quality effects of tire shreds placed above the water table. Transportation Research Record 1714:18-24.

Jin, M. S., K. W. Lee, and W. D. Kovacs. 1994. Seasonal variation of resilient modulus of subgrade soils. Journal of Transportation Engineering 120:603-616.

Janoo, V., R. L. Berg, E. Simonsen, and A. Harrison. 1994. Seasonal changes in moisture content in airport and highway pavements. Pages 357-362 in Symposium and Workshop on Time Domain Reflectometry in Environmental, Infrastructure, and Mining Applications, Evanston, IL.

Janoo, V., C. Korhonen, and M. Hovan. 1999. Measurement of water content in Portland cement concrete. Journal of Transportation Engineering May/June:245-249.

Janssen, D. J. 1987. Moisture in Portland cement concrete. Transportation Research Record 1121:40-44.

Janssen, D. J. 1985. The effect of asphalt concrete overlays on the progression of durabilit cracking in Portland cement concrete. Urbana, University of Illinois.

Johnson, C. A., M. Kersten, F. Ziegler, and H. C. Moor. 1996. Leaching behavior and solubility - controlling solid phases of heavy metals in municipal solid waste incinerator ash. Waste Management 16:129-134.

Jones, J. A., and R. H. Jones. 1989. Granular drainage layers in pavement foundations. Pages 55-69 in UNBAR3, Unbound Aggregates in Roads.

Kazmierowski, T. J., A. Bradbury, and J. Hajek. 1994. Field evaluation of various types of open-graded drainage layers. Transportation Research Record 1434:29-36.

Kennedy, T. W. 1985. Prevention of water damage in asphalt mixtures. Pages 119-133 in B. E. Ruth, editor. Evaluation and prevention of water damage to asphalt pavement materials. ASTM STP, Philadelphia.

Kenney, T. C., D. Lau, and G. I. Ofoegbu. 1984. Hydraulic conductivity of compacted granular materials. Can. Geotech. J. 21:726-729.

Kersten, M. S. 1944. Survey of subgrade moisture conditions. HRB Proc. 24. Kestler, M., and R. L.Berg. 1995. Case study of insulated pavement in Jackman, Maine.

Transportation Research Record 1481:47-55. Kida, A., Y. Noma, and T. Imada. 1996. Chemical speciation and leaching properties of elements in

municipal incinerator ashes. Waste Management 16:527-536.

98

Kim, I., and B. Batchelor. 2001. Empirical partitioning leachmodel for solidified/stabilized wastes. Journal of Environmental Engineering 127:188-195.

Kim, S.-H., W. E. Wolfe, and F. C. Hadipriono. 1992. The development of a knowledge-based exper system for utilization of coal combustion by-product in highway embankment. Civil Engineering Systems 9:41-57.

Kimball, C. E. 1997. Evaluating groundwater pollution susceptibility of dust suppressants and roadbed stabilizers. Transportation Research Record 1589:64-69.

Klute, A., editor. 1986. Methods of soil analysis Part 1,physical and mineralogical methods, Second edition. Soil Science Society of America, Inc., Madison.

Koch, P. B., and T. C. Sandford. 1998. Infiltration rate of water through pavement cracks. Dept. of Civil and Env. Eng., University of Maine, Orono, Maine.

Kozeliski, F. A. 1992. Permeable bases help solve pevement drainage problems. Aberdeen's Concrete Construction 37:660-662.

Ksaibati, K., J. Armaghani, and J. Fisher. 2000. Effect of moisture on modulus values of base and subgrade materials. Transportation Research Record 1718:20-29.

Lafleur, J., and Y. Savard. 1996. Efficiency of geosynthetic lateral drainage in northern climates. Transportation Research Record 1534:12-18.

Lebeau, M., J. Lafleur, and Y. Savard. 1998. Comparison of different internal drainage systems based on FEM. Pages 153-163 in E. J. Hoppe, editor. International Symposium on Subdrainage in Roadway Pavements and Subgrades. Grafistaff, Granada, Spain.

Ledieu, J., P. DeRidder, P. DeClerck, and S. Dautrebande. 1986. A method of measuring soil moisture by time-domain reflectometry. Journal of Hydrology 88:319-328

Leong, E. C., and H. Rahardjo. 1997. Hydraulic conductivity functions for unsaturated soils. Journal of Geotechnical and Geoenvironmental Engineering 123:1118-1126.

Lessard, M., M. Baalbaki, and P.-C. Aitcin. 1996. Effect of pumping on air characteristics of conventional concrete. Transportation Research Record 1532:9-14.

Li, Y., S. G. Buchberger, and J. J. Sansalone. 1999. Variably saturated flow in storm-water partial exfiltration trench. Journal of Environmental Engineering 125:556-565.

Li, Z., and C.-K. Chau. 2000. New water hydraulic conductivity test scheme for concrete. ACI Materials Journal 97:84-90.

Liang, H. S., and R. L. Lytton. 1989. Rainfall estimation for pavement analysis and design. Transportation Research Record 1252:42-49.

Lindly, J. K., and A. S. Elsayed. 1995. Estimating hydraulic conductivity of asphalt-treated bases. Transportation Research Record 1504:103-111.

Loi, J., D. G. Fredlund, J. K. Gan, and R. A. Widger. 1992. Monitoring soil suction in an indoor test track facility. Transportation Research Record 1362:101-110.

Long, F., J. Harvey, C. Scheffy, and C. L. Monismith. 1996. Prediction of pavement fatigue for California department of transportation accelerated pavement testing program drained and undrained test sections. Transportation Research Record 1540:105-114.

Look, B. G., I. N. Reeves, and D. J. Williams. 1994. Application of time domain reflectometry in the design and construction of road embankments. Pages 410-421 in Symposium and Workshop on Time Domain Reflectometry in Environmental, Infrastructure, and Mining Applications, Evanston, IL.

Lynn, T.D., E.R. Brown, and L. A. Cooley. 1999. Evaluation of aggregate size characteristics in stone matrix asphalt and superpave mixtures. Transportation Research Record 1681:19-27.

99

Lytton, R. L., D. E. Pufahl, C. H. Michalak, H. S. Lian, and B. J. Dempsey. 1993. An integrated model of the climatic effects on pavements. Texas Transportation Institute, Texas A&M University, Texas.

Macmaster, J. B., G. A. Wrong, and W. A. Phang. 1982. Pavement drainage in seasonal frost area, Ontario. Transportation Research Record 849:18-23.

Mahboub, K. C., and P. R. Massie. 1996. Use of scrap tire chips in asphaltic membrane. Transportation Research Record 1530:59-63.

Mallela, J., L. Titus-Glover, and M. I. Darter. 2000. Considerations for providing subsurface drainage in jointed concrete pavements. Transportation Research Record 1709:1-10.

Manion, W. P., D. N. Humphrey, and P. E. Garder. 1995. Evaluation of existing aggregate base drainage performance. Maine Department of Transportation.

Masad, E., B. Muhunthan, N. Shashidar, and T. Harman. 1999. Quantifying laboratory compaction effects on the internal structure of asphalt concrete. Transportation Research Record 1681: 179-185

Mathis, D. M. 1990. Permeable bases prolong pavement, studies show. Roads & Bridges 28:33-35. McEnroe, B. M. 1994. Drainability of granular bases for highway pavements. Transportation

Research Record 1434:23-28. McEnroe, B. M., and S. Zou. 1993. Simulation of subsurface drainage of highway pavements. Pages

483-488 in H. W. Shen, S. T. Su, and F. Wen, editors. Hydraulic Engineering 1993. ASCE, San Francisco, California.

McInnes, D. B. 1972. Some aspects of raod shoulder water entry. Pages 211-229 in A. S. Reiher, editor. Sixth Conference - Pavement Studies. Australian Road Research Board, Canberra.

McKee, C. R., and A. C. Bumb. 1984. The importance of unsaturated flow parameters in designing a monitoring system for hazardous wastes and environmental emergencies. Pages 50-58 in Hazardous materials control research institute national conference, Houston, Texas.

McKee, C. R., and A. C. Bumb. 1987. Flow-testing coal bed methane production wells in the presence of water and gas. Pages 599-608 in SPE Formation Evaluation.

Mitchell G., J. D. Shinn. 1998. Innovative use of the cone penetrometer test for highway environmental site characterization and monitoring. Transportation Research Record 1626:120-128

Mizutani, S., T. Yoshida, S.-i. Sakai, and H. Takatsuki. 1996. Release of metals from MSW I fly ash and availability in alkali condition. Waste Management 16:537-544.

Mohammad, L. N., B. Huang, A. Puppala, and A. J. Allen. 1999. Regression model for resilient modulus of subgrade soils. Transportation Research Record 1687:47-54.

Moulton, L. K. 1980. Highway subdrainage design. FHWA-TS-80-224, U.S. Dept. of Transportation. Moynahan, T. J., and Y. M. Sternberg. 1974. Maryland Drainage Study: Vol. VII - An Investigation

of vertical and horizontal hydraulic conductivies in base courses. University of Maryland, College Park.

Muskat, M. 1937. The Flow of homogenous fluid through porous media. McGraw Hill, N.Y. Nakamoto, J., K. Togawa, T. Miyagawa, and M. Fujii. 1998. Water hydraulic conductivity of high

slag content concrete. Pages 779-798 in V. M. Malhotra, editor. Fly ash, silica fume, slag and natural pozzolans in concrete. American Concrete Institute, Bangkok.

Novak, E. J., and L. E. Defrain. 1992. Seasonal changes in longitudinal profile of pavements subject to frost action. Transportation Research Record 1362:95-100.

OECD (Organization for Economic Cooperation and Development). 1973. Water in roads: prediction of moisture content in road subgrades. Road Research. Paris.

100

Oloo, S. Y., D. G. Fredlund, and J. K.-M. Gan. 1997. Bearing capacity of unpaved roads. Can. Geotech. J. 34:398-407.

Or, D., and M. Tuller. 2000. Flow in unsaturated fractured ferrous media: hydraulic conductivity of rough surfaces. Water Resources Research 36:1165-1177.

Oshita, H., and T.-a. Tanabe. 2000a. Modeling of water migration phenomenon in concrete as homogenous material. Journal of Engineering Mechanics 126:551-553.

Oshita, H., and T.-a. Tanabe. 2000b. Water migration phenomenon model in cracked concrete. I. Formulation. Journal of Engineering Mechanics 126:539-544.

Oshita, H., and T.-a. Tanabe. 2000c. Water migration phenomenon model in cracked concrete. II: Calibration. Journal of Engineering Mechanics 126:544-549.

Ovik, J., B. Birgisson, and D. E. Newcomb. 1998. Relating climate factors to pavement subsurface conditions. in 9th International conference on cold regions Engineering. Cold regions impact on civil works, Duluth, MN.

Ovik, J., B. Birgisson, and D. E. Newcomb. 1999. Characterizing seasonal variations in flexible pavement material properties. Transportation Research Record 1684:1-7.

Ozyildirim, C. 1998. Effects of temperature on the development of low hydraulic conductivity in concretes. Virginia Transportation Research Council, Charlottesville.

Pagotto, C., M. Legret, and P. L. Cloirec. 2000. Comparison of the hydraulic behaviour and the quality of highway runoff water according to the type of pavement. water research 34:4446-4454.

Partridge, B. K., P. J. Fox, J. E. Alleman, and D. G. Mast. 1999. Field demonstration of highway embankment construction using waste foundry sand. Transportation Research Record 1670:98-105.

Picornell, M., and R. L. Lytton. 1987. Characterization of the meteorological demand for the design of vertical moisture barriers. Transportation Research Record 1137:42-51.

Picornell, M., R. L. Lytton, and M. Steinberg. 1983. Matrix suction instrumentation of a vertical moisture barrier. Transportation Research Record 945:16-21.

Picornell, M., and M. A. B. A. Rahim. 1991. Simulation of climatic and vegetation effects on pavements on expansive clays. Transportation Research Record 1307:281-290.

Pigeon, M., and M. Regourd. 1983. Freezing and thawing durability of three cements with various granulated blast furnace slag contents. Pages 979-997 in V. M. Malhotra, editor. Fly Ash, Silica Fume, Slag and Other Mineral By-Products in Concrete. American Concrete Institute, Detroit.

Pleau, R., P. M, A. Lamontagne, and M. Lessard. 1995. Influence of pumping on characteristics of air-void system of high-performance concrete. Transportation Research Record 1478:30-36.

Pratt, C. J., J. D. G. Mantle, and P. A. Schofield. 1995. UK Research into the performance of permeable pavement, reservoir structures in controlling storm water discharge quantity and quality. 32(1):63-69.

Pufahl, D. E., and R. L. Lytton. 1991. Temperature and suction profiles beneath highway pavements: computed and measured. Transportation Research Record 1307:268-276.

Pufahl, D. E., R. L. Lytton, and H. S. Liang. 1990. Integrated computer model to estimate moisture and temperature effects beneath pavements. Transportation Research Record 1286:259-269.

Raad, L. 1982. Pumping mechanisms of foundation soils under rigid pavements. Transportation Research Record 849:29-37.

Rainwater, N. R., and R. E. Yoder. 1999. Comprehensive monitoring systems for measuring subgrade moisture conditions. Journal of Transportation Engineering 125:439-448.

101

Randolph, B. W., J. Cai, A. G. Heydinger, and J. D. Gupta. 1996a. Laboratory study of hydraulic conductivity for coarse aggregate bases. Transportation Research Record 1519:19-27.

Randolph, B. W., E.P. Steinheuser, A.G. Heydinger, and J.D. Gupta. 1996b. In situ test for hydraulic conductivity of drainable bases. Transportation Research Record 1519:36-40.

Ridgeway, H. 1976. Infiltration of water through the pavement surface. Transportation Research Record 616:98-101.

Ridgeway, H. H. 1982. National Cooperative Highway Research Program Synthesis of HIghway Practice 96: Pavement Drainage Systems. NCHRP.

Roberson, R., and B. Birgisson. 1998. Evaluation of water flow through pavement systems. Pages 295-302 in E. J. Hoppe, editor. International Symposium on Subdrainage in Roadway Pavements and Subgrades. Grafistaff, Granada, Spain.

Roberson, R., R. Olson, C. Millner. 2000. Edge-Joint Sealing as a Preventive Maintenance Practice. MnDOT Internal Document

Roberson, R. 2001. Moisture Effects, Mn/Road: The first six years, a workshop. in Minnesota Pavement Conference, Minnesota.

Roy, D. M., D. Shi, B. Scheetz, and P. W. Brown. 1992. Concrete microstructure and its relationships to pore structure, hydraulic conductivity, and general durability. Pages 139-152 in J. Holm and M. Geiker, editors. Durability of Concrete - G. M. Idorn International Symposium. American Concrete Institute.

Sansalone, J. J.. 1999. Adsorptive infiltration of metals in urban drainage - media characteristics. The Science of the Total Environment, 235:179-188.

Saraf, C., C.-p. Chou, and B. F. McCullough. 1987. Effect of rainfall on the performance of continuously reinforced concrete pavements in Texas. Transportation Research Record 1121:45-49.

Savage, B. M., and D. J. Janssen. 1997. Soil physics principles validated for use in predicting unsaturated moisture movement in Portland cement concrete. ACI Materials Journal 94:63-70.

Schimmoller, V. E., K. Holtz, T. Eighmy, C. Wiles, M. Smith, G. Malasheskie, and G. J. Rohrbach. 2000. Recycled Materials in European Highway Environments: Uses, Technologies, and Policies. American Trade Initiatives.

Schreurs, J. P. G. M., H. A. v. d. Sloot, and C. Hendriks. 2000. Verification of laboratory-field leaching behavior of coal fly ash and MSWI bottom ash as a road base material. Waste Management 20:193-201.

Schroeder, R. L. 1994. The use of recycled materials in highway construction. Road & Transport Research 3:12-27.

Slate, F. O., and C. Hover. 1984. Microcracking in concrete. Pages 137-159 in A. Carpenteria and A. R. Ingraffea, editors. Fracture Mechanics of Concrete. Nijhoff Publishers.

Smiles, D. E., and O. T. Denmead. 1972. Infiltration and evaporation of water from soil. Pages 197-215 in A. S. Reiher, editor. Sixth conference - Principal Addresses and invited papers. Australian Road Research Board, Canberra.

Smith, L. L. 1996. Wasting no time: Florida's history of successful recycling for transportation. TR News 184:4-7.

Stephens, D. B. 1996. Vadose zone hydrology. Lewis Publishers, New York. Stormont, J. C., and C. E. Anderson. 1999. Capillary barrier effect from underlying coarser soil layer.

Journal of Geotechnical and Geoenvironmental Engineering 125:641-648.

Simonsen, E., V. C. Janoo, and U. Isacsson. 1997. Prediction of temperature and moisture changes in pavement structures. Journal of Cold Regions Engineering 11:291-307.

102

Tandon, V., and M. Picornell. 1997. Proposed evaluation of base materials for drainability. Transportation Research Record 1596:62-69.

Tawfiq, K., J. Armaghani, and J. R. Vysyaraju. 1996. Hydraulic conductivity of concrete subjected to cyclic loading. Transportation Research Record 1532:51-59.

Terrel, R. L., J. R. Lundy, and R. B. Leahy. 1994. Evaluation of mixtures containing waste materials. Asphalt Paving Technology 63:22-37.

Thadkamalla, G. B., and K. P. George. 1995. Characterization of subgrade soils at simulated field moisture. Transportation Research Record 1481:21-27.

Thom, N. H., and S. F. Brown. 1987. Effect of moisture on the structural performance of a crushed-limestone road base. Transportation Research Record 1121:50-56.

Tian, P., M. M. Zaman, and J. G. Laguros. 1998. Gradation and moisture effects on resilient moduli of aggregate bases. Transportation Research Record 1619:75-84.

Tindall, J. A., J. R. Kunkel, and D. E. Anderson. 1999. Unsaturated zone hydrology for scientists and engineers. Prentice Hall, Upper Saddle River.

Tsai, C.-H., and T. M. Petry. 1995. Suction study on compacted clay using three measurement methods. Transportation Research Record 1481:28-34.

Van der AA, J. P. C. M., and G. Boer. 1997. Automatic moisture content measuring and monitoring system based on time domain reflectometry used in road structures. NDT & E International Independent 30:239-242.

Van Herck, P., B. V. d. Bruggen, G. Vogels, and C. Vandercasteele. 2000. Application of computer modeling to predict the leaching behavior of heavy metals from MSWI fly ash and comparison with a sequential extraction method. Waste Management 20:203-210.

Van Sambeek, R. J., 1989. Synthesis on subsurface drainage of water infiltrating a pavement structure. Braun Pavement Technologies, St. Paul.

Vashisth, P., K. W. Lee, and R. M. Wright. 1998. Assessment of water pollutants from asphalt pavement containing recycled rubber in Rhode Island. Transportation Research Record 1626:95-104.

Virtanen, J. 1983. Freeze-thaw resistance of concrete containing blast-furnace slag, fly ash or condensed silica fume. Pages 923-941 in V. M. Molhatra, editor. Fly Ash, Silica Fume, Slag and Other Mineral By-Products in Concrete. American Concrete Institute, Detroit.

Waters, T. J. 1998a. An innovative approach to the prediction of moisture conditions in flexible pavements. Pages 427-435 in E. J. Hoppe, editor. International Symposium on Subdrainage in Roadway Pavements and Subgrades. Grafistaff, Granada, Spain.

Waters, T. J. 1998b. The Strength of pavement materials following drainage. Pages 435-443 in E. J. Hoppe, editor. International Symposium on Subdrainage in Roadway Pavements and Subgrades. Grafistaff, Granada, Spain.

Xu, X., J. Nieber, J. M. Baker, and D. E. Newcomb. 1991. Field testing of a model for water flow and heat transport in variably saturated, variably frozen soil. Transportation Research Record 1307:300-308.

Terrel, R. L., and S. Al-Swailmi. 1993. Role of pessimum voids concept in understanding moisture damage to asphalt concrete mixtures. Transportation Research Record 1386:31-37.

van Genuchten, M. T. 1980. A closed form equation for predicting the hydraulic conductivity of unsaturated soils. Soil Science Society of America 44:892-898.

Williams, J., R. E. Prebble, R. T. Williams, and C. T. Hignett. 1983. The influence of texture, structure and clay mineralogy on the soil moisture characteristic. Australian Journal of Soil Research 21:15-32.

103

Yoder, E. J., and M. W. Witczak. 1975. Principles of pavement design, Second edition. John Wiley & Sons, New York, p 192.

Zhou, H., L. Moore, J. Huddleston, and J. Gower. 1993. Determination of free-draining base materials properties. Transportation Research Record 1425:54-63.

104