010001 LIBRARY
Michigan State

University

 

 

 

This is to certify that the
thesis entitled

NUMERICAL EVALUATION OF PREFERENTIAL FLOW
THROUGH EVAPOTRANSPIRATIVE LANDFILL COVERS

presented by

Carolyn M. Hardt

has been accepted towards fulfillment
of the requirements for the

MS. degree in Civil Engineering

 

 

’9' /
Major Profefior’s Signature

lzjgjoz

l

 

Date

MSU is an Affirmative Action/Equal Opportunity Employer

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 K.lProj/Acc8Pres/ClRC/DaleDue,Indd

NUMERICAL EVALUATION OF PREFERENTIAL FLOW THROUGH
EVAPOTRANSPIRATIVE LANDFILL COVERS

By

Carolyn M. Hardt

A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
MASTER OF SCIENCE
Civil Engineering

2008

ABSTRACT

NUMERICAL EVALUATION OF PREFERENTIAL FLOW THROUGH
EVAPOTRANSPIRATIVE LANDFILL COVERS

By
Carolyn M. Hardt

Landfill final covers (caps) are regulated by federal and state regulations with the
purpose of minimizing the amount of infiltration into the waste (percolation).
Conventional cover designs typically consist of a compacted clay layer overlain by a
geomembrane. While these covers restrict percolation to relatively small quantities,
desiccation cracking of the clay barrier layer is a common long-term shortcoming. Hence,
as an alternative, evapotranspirative (ET) covers have been routinely explored. While ET
covers can decrease desiccation cracking potential, preferential flow is still a threat.
Macropores formed by desiccation, roots, insects, and earthworms are present in almost
all soil covers including ET covers.

In this study, preferential flow through an ET cover is evaluated using HYDRUS 1D
and 2D (beta version) software which included single-permeability (capillary flow) and
dual-permeability (capillary and preferential flow) modeling options. Simulation
duration, presence of preferential flow, preferential flow depth, fracture hydraulic
conductivity, the amount of fractures, the fraction of surface flow entering the fractures,
hydraulic conductivity of the storage layer and waste layer, and waste layer thickness
were varied to evaluate their effect on percolation. The key conclusion of this study is
that preferential flow can significantly increase percolation through the cap and into the
waste layer of a landfill. Ignoring preferential flow, whether it is currently present or will

be in the future, can be detrimental when designing and evaluating landfill final covers.

ACKNOWLEDGMENTS

I would like to thank all individuals who helped me in making this project possible. I
wish to express my sincere gratitude to my advisor, Dr. Milind V. Khire, for his guidance
and patience. I am also thankful to my committee members, Dr. Mantha and Dr. Wolff,
for their valuable time. I also would like to thank Professor Jirka Simfinek for providing
me the beta version of HYDRUS-2D with the dual-permeability modeling option. I
would also like to thank Dr. Khire’s PhD students Ramil G. Mijares and Moumita
Mukherjee for their assistance during the entire process. I acknowledge and am grateful
for the financial support extended to ‘me by the Michigan State University for three
semesters in the form of a Graduate Teaching Assistantship.

I would like to thank my parents Richard and Margaret and my siblings Tim, Brian
and Danielle for their continuing love and encouragement. Apart from them, I would like
to thank all my friends and family, especially Robert Aylsworth, Kerry Lewis and Brenda
Dill, for their support and encouraging words. Finally, I would like to thank my
professors from the Engineering and Geology Departments at the Michigan State
University and the Geology Department at the University of Pittsburgh. To everyone I

would like to say thank you.

iii

TABLE OF CONTENTS

 

 

 

 

 

 

 

 

List of Tables vi
List of Figures vii
List of Abbreviations ix
List of Symbols x
CHAPTER 1: Introduction 1
1.1 Background ............................................................................................................... 1
1.2 Final Covers .............................................................................................................. 3
1.2.1 Compacted Clay Barriers ................................................................................... 3
1.2.2 Evapotranspirative Covers ................................................................................. 3
1.2.3 Capillary Barrier ET Covers .............................................................................. 4

1.3 Preferential Flow ....................................................................................................... 8
CHAPTER 2: Estimating Percolation 9
2.1 Lysimeters ................................................................................................................. 9
2.2 Numerical Modeling ............................................................................................... 11
2.2.1 Uniform Flow ................................................................................................... 12
2.2.2 Non-equilibrium Flow Models ..................................................................... ' 13
2.2.2.1 Dual-Porosity ............................................................................................ 13
_____2.2.2.2 Dual-Permeability ..................................................................................... 13
CHAPTER 3: HYDRUS 18
3.1 HYDRUS-2D Overview ......................................................................................... 18
3.1.1 Water Transport ............................................................................................... 19
3.1.2 Dual-Permeability ............................................................................................ 20

3.2 HYDRUS-1D Overview ......................................................................................... 21
3.3 Simulations ............................................................................................................. 21
3.3.1 Design .............................................................................................................. 22
3.3.2 Soil Hydraulic Properties ................................................................................. 22
3.3.3 Climate ............................................................................................................. 26
3.3.4 Boundary Conditions ....................................................................................... 30
3.3.5 Time Step, Error Tolerance, Mesh Design, and Initial Conditions ................. 30
3.3.6 Fracture Properties ........................................................................................... 31
3.3.7 HYDRUS-1D ................................................................................................... 31
CHARTER 4: Results and Discussion 33
4.1 Presence or Absence of Fractures ........................................................................... 35
4.2 Fracture Properties .................................................................................................. 39
4.2.1 Fracture Depth ................................................................................................. 39

iv

4.2.2 Hydraulic Conductivity of Fractures ............................................................... 43

 

4.2.3 Volume of Fractures ........................................................................................ 46
4.2.4 Effect of Fraction of Surface Flow Entering Fractures .................................... 48

4.3 Hydraulic Conductivity of Storage Layer ............................................................... 50
4.4 Waste Layer Parameters ......................................................................................... 53
4.4.1 Hydraulic Conductivity of Waste Layer .......................................................... 55
4.4.2 Thickness of Waste Layer ................................................................................ 59
CHAPTER 5: Summary and Conclusions 62
REFERENCES 65

 

LIST OF TABLES

Table 3-1: Saturated and Unsaturated Soil Properties ...................................................... 24

Table 4-1: Summary of Simulations Carried out Using HYDRUS. ................................. 34

vi

LIST OF FIGURES

Figure 1-1: Cross-Section of a Typical MSW Landfill and Conventional Final Cover

(Patel 2005) ................................................................................................................. 2
Figure 1-2: Hydraulic Conductivities of Fine and Coarse Grained Capillary Barrier Soils
(Morris and Stormont 1997). ...................................................................................... 6
Figure 1-3: Water-Entry Matric Suction of Two Typical Capillary Barrier Soils (Khire et
a1. 2000) ...................................................................................................................... 7
Figure 2-1: Cross-Section of a Typical Lysimeter (Albright et al. 2004). .......................... 9
Figure 3-1: Schematic of the Conceptual Models used for Numerical Modeling. ........... 23
Figure 3-2: Soil Water Characteristic Curves. .................................................................. 25
Figure 3-3: Daily and Cumulative Precipitation Data. ..................................................... 27
Figure 3-4: Daily and Cumulative PET. ........................................................................... 28
Figure 3-5: Daily Average Temperature and Daily Net Solar Radiation. ........................ 29
Figure 4-1: Simulated Percolation through Cap versus Lysimeter with and without
Preferential Flow on a Log Scale (a); and Linear Scale (b) ...................................... 36
Figure 4-2: Simulated Cumulative Evaporation and Runoff (a); and Simulated
Cumulative Percolation through the Bottom Most Boundary (b) ............................. 38
Figure 4-3: Depth of Preferential Flow through Lysimeter and Cap. ............................... 41
Figure 4-4: Effect of Fracture Depth on Simulated Percolation for Lysimeter (a); and
Cap (b). ..................................................................................................................... 42
Figure 4-5: Effect of Waste Layer Hydraulic Conductivity and Fracture Depth on
Simulated Percolation. .............................................................................................. 44
Figure 4-6: Effect of Hydraulic Conductivity of Fractures on Simulated Percolation. 45
Figure 4-7: Simulated Effect of the Volume of Fractures on Percolation for Lysimeter (a);
and Cap (b). ............................................................................................................... 47
Figure 4-8: Simulated Effect of qTop for a Lysimeter (a); and Cap (b). ........................... 49

vii

Figure 4-9: Simulated Effect of w and qmp on Percolation through a Lysimeter (a); and

Cap (b). ..................................................................................................................... 51
Figure 4-10: Effect of Hydraulic Conductivity of the Storage Layer on Simulated
Percolation of Lysimeter (a); and Cap (b). ............................................................... 52
Figure 4-11: Simulated Effect of Hydraulic Conductivity of Waste Layer on Cap
Percolation without Preferential Flow (a); and with Preferential Flow (b). ............. 54
Figure 4-12: Ratio of Cumulative Percolation with Varying Fracture Depths and Waste
Layer Hydraulic Conductivity. ................................................................................. 58
Figure 4-13: Effect of Waste Layer Thickness on Simulated Percolation ........................ 61

viii

LIST OF ABBREVIATIONS

ACAP = Alternative Cover Assessment Project

ALCD = Alternative Landfill Cover Demonstration

CL = Inorganic clays of low to medium plasticity; gravelly, sandy, silty or lean clays
DOE = US. Department of Energy

ED = Equivalency Demonstration

ET = Evapotranspirative/Evapotranspiration

HELP = Hydrologic Evaluation of Landfill Performance

ML = Inorganic silts and vary fine sands, rock flour, silty or clayey fine sands or clayey
silts with slight plasticity

MSW = Municipal Solid Waste

NOAA = National Oceanic and Atmospheric Administration
PET = Potential Evapotranspiration

RCRA = Resource Conservation and Recovery Act

SL = Storage Layer

SM = Silty sands, sand-silt mixture

SWCC = Soil Water Characteristic Curve

USCS = Unified Soil Classification System

USEPA = United States Environmental Protection Agency
VL = Vegetation Layer

WL = Waste Layer

ix

LIST OF SYMBOLS

or = van Genuchten Function Fitting Parameter
‘1’ = Matric Suction

‘P3 = Water Entry Matric Suction

0 = Volumetric Water Content

0, = Residual Water Content

0S = Saturated Water Content

yw = Dimensionless Scaling Factor

AS = Change in Water Storage

a = Aggregate characteristic length

[3 = Shape Facture

ET = Evapotranspiration

G = Deep Drainage or Percolation into Waste
h = Pressure Head

K = Unsaturated Hydraulic Conductivity

Ka = Interface Hydraulic Conductivity

K134 = Components of a Dimensionless Anisotropy Tensor KA

KFR = Fracture Hydraulic Conductivity
K ,. = Relative Hydraulic Conductivity
Kw = Saturated Hydraulic Conductivity

KSL = Storage Layer Hydraulic Conductivity

KVL = Vegetation Layer Hydraulic Conductivity

KWL = Waste Layer Hydraulic Conductivity

n = van Genuchten Function Fitting Parameter

P = Precipitation

PrCAp = Cumulative Percolation through Cap

PrLys = Cumulative Percolation through Lysimeter

qmp = Amount of Applied Precipitation Entering Directly into the Fractures

Ri = Surface Run-on

R0 = Surface Runoff

S = Sink Term

t = Time

TSL = Storage Layer Thickness
TVL = Vegetation Layer Thickness

Tw1,= Waste Layer Thickness

FW = Transfer Rate
w = Ration of the Volume of the Fracture to the Total System

xi = Spatial Coordinates

xi

CHAPTER 1
INTRODUCTION

1.1 BACKGROUND

Landfilling is one of the oldest and most common methods of waste disposal used
today. In order to assure this system of waste disposal is safe and reliable in the long
term, certain measures and regulations must be met. The key components of modern
landfills include a bottom liner and leachate collection system (LCS), leachate
recirculation system (LRS), gas collection system, surface water management system,
and final cover (cap).

The final cover system of a municipal solid waste (MSW) landfill is constructed for
the purpose of minimizing exposure of the waste, controlling gas emissions, and
preventing the infiltration of water into the waste which could potentially result in the
creation of contaminated leachate. Figure 1-1 illustrates a typical MSW landfill and final
cover system. Closure criteria for MSW landfills are regulated by Subtitle D, Part 258.60
of the federal Resource Conservation and Recovery Act (RCRA). The final cover must:

(1) have a hydraulic conductivity less than or equal to that of any bottom liner system or
natural sub-soils present, or a hydraulic conductivity no greater that 10'5 cm/s, whichever

is less; (2) minimize infiltration through the cover by the use of an infiltration layer that
contains a minimum 45 cm of earthen material; and (3) minimize erosion of the final
cover by the use of an erosion layer that contains a minimum 15 cm thick soil layer that is
capable of sustaining native plant growth. State regulation guidelines may require

additional components and restrictions.

 

Vegetation/T op Soil

Geomembrane —:m '

  
     
  

 

 

Compacted Clay Barrier —> T; r 3». ‘- .
Geocomposite Drainage N - '|
Waste
Surface Water Gas Collection System
Management System .
\ Final Cover System
- _
Waste

Bottom Liner and Leachate Collection System

 

 

 

Figure 1-1: Cross-Section of a Typical MSW Landfill and Conventional Final Cover
(Patel 2005).

1.2 FINAL COVERS
1.2.1 Compacted Clay Barriers

The resistive, low hydraulic conductivity barrier required by RCRA is intended to
minimize water migration into the underlying Waste layer. This barrier layer is commonly
constructed of compacted clay which often cracks in arid and semi-arid climates due to
low precipitation and relatively high evapotranspiration. Clay barriers have been known
to crack in sub-humid and humid environments as well (Scanlon et a1. 2005). As an
alternative to these composite covers consisting of a geomembrane and compacted clay
barrier, also known as conventional covers, alternative covers, such as evapotranspirative

(ET) covers, have been permitted on a case by case basis.

1.2.2 Evapotranspirative Covers

The RCRA regulations allow for alternative final covers to be implemented as long
as their ability to retard infiltration and prevent erosion is equivalent to conventional
covers. Instead of relying on “impermeable” barriers, ET covers employ soil water
storage capacity and natural evaporation and transpiration to minimize infiltration into
the waste layer. These covers are designed to be permeable to precipitation and local run-
on which is stored within the soil(s) and later removed by plant transpiration and soil
evaporation. A critical component of ET covers is the water storage capacity of the soil.
The soil layer(s) must have enough storage capacity to contain all infiltrated water (Khire
et a1. 1997; 1999). If the storage capacity is exceeded before the water can be removed,
water will drain vertically into the underlying waste. In order for ET covers to perform as

desired, they must have enough storage capacity to prevent water advancement into the

waste and sufficient evapotranspiration to remove excess water from storage. The

governing water balance equation behind ET covers is given in Eq. 1-1:

P+Ri=ET+R0+G+AS (1-1)

where P = precipitation; Ri = surface run-on from adjacent areas; ET =

evapotranspiration = direct evaporation + transpiration from plants; R0 = surface runoff;

G = deep drainage or percolation into waste; and AS = change in the amount of water
stored in the cover soils (Jacobson et a1. 2005).

Multiple designs of ET covers have been developed over the years including
monolithic ET covers, capillary barrier ET covers, and anisotropic barrier ET covers
GBenson and Khire 1995). The simplest, monolithic ET covers, consist of only one soil
type while more complex designs like the capillary barrier ET cover are comprised of

multiply soil types divided into specific layers (Scanlon et a1. 2005).

1.2.3 Capillary Barrier ET Covers

Capillary barrier covers consist of a fine-grained soil layer overlying a coarse-grained
soil layer which, under unsaturated conditions, acts as a barrier restricting the downward
flow of water (Khire et al. 2000). The infiltrating water is held in the fine soil layer until
it can be removed by evapotranspirative gradients.

This natural barrier occurs due to the contrast in hydraulic conductivities of the two

soils. At high matric potentials, the fine soil will have a small yet finite hydraulic

conductivity while the coarse soil will have a hydraulic conductivity even smaller to the
point where it becomes orders of magnitude smaller as seen in Figure 1-2.

While both layers are relatively dry or have a high matric suction, water will not flow
downward into the coarse soil but instead will stay in the fine soil, gradually increasing
its moisture content which in turn decreases the layer’s matric suction and increases its
hydraulic conductivity. The decreasing matric suction will eventually reach the effective

water-entry suction of the coarse grained soil along the soil interface. The water-entry

suction of the coarse soil, 1P3, corresponds to the matric suction located at the bend in the

soil water characteristic curve near the residual water content, as shown in Figure 1-3.

Once the fine soil reaches ‘PB, water begins to enter the coarse soil and the capillary

barrier undergoes breakthrough. Just as with the fine soil, the water content and hydraulic
conductivity of the coarse soil begins to increase while the matric potential of the layer
decreases. Water can now move freely through both cover layers and into the underlying
waste layer. The moisture must be removed from the fine grained soil layer before the
water-entry matric suction of the coarse soil is reached and the cover is compromised.
Removal of moisture can be achieved through evapotranspiration at the surface or lateral
drainage along the interface if the surface is sloped or ofien a combination of the two is

used (Morris and Stormont 1997; Khire et a1. 2000).

 

10-2 ! I l .r

 

 

 

 

 

 

A """""""" """"""""" —CoarseSoil ""
1’ 10“ — ....... FineSoil _,__
E i E E .
>5
4: -6
"5
8 .3
C 10
o
O
2 10'10
:3
co
5
> 10'12
I
10-14

 

Matric Potential (m)

Figure 1-2: Hydraulic Conductivities of Fine and Coarse Grained Capillary Barrier Soils
(Morris and Stormont 1997).

 

 

Matric Suction (cm)

 

 

 

 

 

 

 

0 0.1 0.2 0.3 0.4
Volumetric Water Content (9)

Figure 1-3: Water-Entry Matric Suction of Two Typical Capillary Barrier Soils (Khire et
a1. 2000).

1.3 PREFERENTIAL FLOW

One of the main failures of conventional final covers is desiccation cracking in the
compacted clay layer allowing water to infiltrate into the underlying waste creating
potentially harmful leachate. For this reason, much attention has been placed on natural
covers such as ET covers. Although ET covers are designed to decrease the chances of
desiccation cracking by employing native soils that contain no plastic fines, often it is not
possible to eliminate plastic fines. Also, large non-capillary pores (macropores) exist.
These macropores are mainly created by desiccation and biological activities due to root
channels, insects, and earthworms (Carneira et al. 2000).

Macropores have much higher hydraulic conductivities than the surrounding
micropores (referred to as matrix onwards) because they have relatively low water—entry
suctions and are often connected to the surface. This allows water to enter easily
especially during long or high intensity precipitation events. Depending on their depth,
they can transport a significant amount of water at rates orders of magnitude higher than
in the matrix. Ignoring infiltration through macropores can lead to an underestimation of
percolation, an overestimation of water ponding at the soil surface and high runoff
predictions and, as a result, an inadequate prediction of the covers performance (Khire et
al. 2000; Novak et al. 2000; Khire and Mijares 2008). In this thesis, all nonuniform
infiltration, whether it be cracks, fractures or macropores, is referred to as preferential

flow or fracture flow.

CHAPTER 2
ESTIMATING PERCOLATION

Before any alternative cover can be used in place of a RCRA cover its ability to
minimize infiltration of water into the underlying waste, known as percolation, must be
proven to be up to RCRA standards using an approach called the equivalency
demonstration (ED). An ED consists of the use of lysirneters and/or instrumented test
sections without lysimeters. Sometimes simple analytical or numerical models are also

used for the quantitative assessment.

2.1 LYSIMETERS

The use of instrumented lysirneters is the most common method of evaluating the
water balance of alternative landfill covers including assessing the amount of percolation.
A lysimeter is an impermeable barrier, usually in the shape of a pan, placed under a
section of a landfill cover built onsite. A lysimeter usually consists of a drainage layer
placed above a geomembrane to collect and measure percolation through the cover (Khire

and Mijares 2007). Figure 2-1 shows a schematic of a typical lysimeter.

    

 

    
    
 

 
 
 

 

 

 

Instrument
, , Cover
Drversron - nest
sorl
'3‘ '7?- 3' '.. I -
Surfac +. * ‘-:Aé—'+‘-'.-.-.'-.é"-.-.::.'**fs".f':r‘:"':'??v‘g-r :_ .
v2: - ---- . .-
Runoff ' A ,. ... .- .- :-.Ii1="~ 4—
. - ' " ‘ \ \ Roo
Percolation / . . -
Geomembrane Geocomposrte Interim barrier

liner drain cover soil

Figure 2-1: Cross-Section of a Typical Lysimeter (Albright et al. 2004).

Between July 1995 and July 2000, the US. Department of Energy (DOE) conducted
the Alternative Landfill Cover Demonstration (ALCD) in Albuquerque, New Mexico. Six
types of covers were built in the field, each 13 m wide and 100 m long, for side by side
comparison of performance. Among these, six covers were RCRA Subtitle D and one
was a capillary barrier cover. After three years of continuous monitoring, the RCRA
covers had an average percolation of 4.82 mm/yr, a 99.98% efficiency rate, and a total

percolation of 6,724 L. The efficiency rate is defined as:

l -(percolation/precipitation) x 100 (2-1 )

The capillary barrier cover had an average percolation of 0.87 mm/yr, an efficiency
rate of 99.99%, and a total percolation of 804 L. This ongoing demonstration has shown
that not only does the capillary barrier cover prevent percolation as well as a RCRA
cover but also can have lower percolation compared to RCRA covers (DOE 2000).

In 1998, the United States Environmental Protection Agency (USEPA) began a
similar project, the Alternative Cover Assessment Project (ACAP), where ten
conventional cover lysimeters and 14 alternative cover lysimeters were built at 11
different sites in seven states. The primary goal was to measure percolation. Climatic
conditions ranged from arid to humid and native soil and native vegetation were used at
all sites. Percolation was monitored fi'om July 1 to June 30 of the following year
beginning in 2000 and ending in 2003. At the sub-humid site in Polson, MT, a capillary
barrier cover and a conventional cover were constructed. The conventional cover had a

total percolation of 15 mm during the designated monitored time and an average of 0.4

10

mm/yr. The capillary barrier performed better having a total percolation of 0.2 mm
during the three year period (Albright et al. 2004). As with the DOE study, the capillary
barrier cover proved to be more than equivalent to a RCRA Subtitle D cover during the
three year monitoring period.

With the use of lysimeters, it has been demonstrated that capillary barrier
evapotranspirative covers can be considered equivalent to the RCRA Subtitle D cover.
Not only do they prevent a significant amount of percolation from passing through the

cover but also reduce percolation, exceeding the RCRA standards.

2.2 NUMERICAL MODELING

Relying solely on the use of lysimeters for an estimation of percolation can be
misleading. Khire and Mijares (2008) demonstrated using the water balance model
Vadose/W (by Geo-Slope 2007) that the presence of an underlying waste layer in an
actual cover influences percolation. Because lysimeters have a geomembrane underneath
the proposed cover system instead of a waste layer, as soon as percolation passes through
the cover it is removed from the system. In an actual cover, a waste layer is present
which can hold water and allow it to migrate upwards back into the cover due to
evapotranspirative gradients. All simulations in the study (Khire and Mijares 2008)
resulted in the over-estimation of percolation by the lysimeter compared to percolation
amounts of a cover and waste layer system for a simulated site located in Detroit,
Michigan. In this study only capillary flow was considered. Flow through macropores
was not simulated because Vadose/W (Geo-Slope 2007) does not have an option to
simulate preferential flow. Khire and Mijares (2008) have proposed a numerical approach

to estimate percolation from a cover and waste layer system.

11

2.2.1 Uniform Flow

Uniform, equilibrium, or single-porosity/single-permeability models are based on the
concept that water and/or solutes enter and penetrate through a soil profile in a uniform
manner. This type of flow can naturally occur in granular soils but is often incorrectly
applied when modeling finer soils and soils containing plant roots and insect or animal
burrows. Most uniform flow models apply Richards equation to the entire porous media
as a whole. These models are easier to use because they only require one set of soil
hydraulic properties but can produce misleading results when preferential flow is
significant (Gardenéis et al. 2006). HELP (Hydrologic Evaluation of Landfill
Performance), UNSAT-H, LEACHM, Vadose/W and HYDRUS v2.007 are some of the
most popular uniform flow models today (Simfinek et al. 2003; Kodesova et al. 2005;
Ogor'zalek et al. 2008).

Numerical modeling with uniform models can be a helpful tool when assessing a
landfill’s performance but should be used with caution. HELP is often used by
geotechnicians to predict percolation through landfill covers. However, Khire et al.
(1997) found that HELP significantly over-predicts percolation. In the same study,
UNSAT-H was also analyzed and was found to slightly under-predict percolation
compared to measured field data. Benson et al. (2005) re-confirmed these findings with
data from the ACAP study. Yet in another study by Khire et al. (1999), UNSAT-H over-
predicted percolation. In other studies, Vadose/W was shown to under-predict percolation
(Benson et al. 2005) and HYDRUS v2.007, LEACHM v4.0 and UNSAT-H v.3.0 all
over-predicted percolation (Ogorzalek et al. 2008).

Uniform flow numerical models can be a useful tool when it comes to predicting

landfill cover percolation but the accuracy will depend on how significant the macropore

l2

flow component is for the problem. A margin of error always exists and should be taken
into consideration. Some inaccuracy may be due to the varying levels of complexity and
sophistication between the models. Other models such as HELP have known limitations
due to the simplistic water routing approach used (Morris et al. 1997 and Khire et al.
1997). One of the flaws uniform models share is not accounting for preferential flow
which can lead to an underestimation of percolation. This common blemish in numerical

modeling is significant and hence has been the focus of this thesis.

2.2.2 Non-equilibrium Flow Models

Non-equilibrium flow was defined by Jarvis (1998) as any system where “for various
reasons, infiltrating water does not have sufficient time to equilibrate with slowly moving
resident water in the bulk of the soil matrix.” Non-equilibrium or preferential flow is
simply the phenomenon of infiltrating water entering a fracture or macropore of some
sort in the soil causing the water in the fracture to penetrate the soil faster then the water
in the surrounding soil matrix. In order to achieve more accurate predictions when
modeling water balance in soils and landfill covers, it is critical to account for
preferential flow. The difficulty in modeling cracks, fractures and macropores lies in the
accuracy of estimating parameters that can accurately characterize the preferential flow
paths and numerical algorithms that are able to carry out the necessary computations in a
reasonable time frame with an acceptable level of accuracy. Not only does preferential
flow affect the estimation of percolation through landfill covers but can also affect solute
transport modeling and irrigation efficiency modeling for agricultural and other

applications (Novak et al. 2000; Sirni’rnek et al. 2003).

13

2.2.2.1 Dual-Porosity

Dual-porosity models separate the soil matrix and fractures into two different water
regimes. Water flow in the soil matrix is stagnant allowing water to only flow in the
fracture network (Simi’rnek et al. 2003). The soil matrix can retain, store and exchange
water but does not allow for continuous flow. Many dual-porosity models apply Richards
equation to the water flow in the fractures and a mass balance equation to the soil matrix.
Alternatively, a kinematic wave equation describing gravitational water flow can also be
applied to the fiactures. This approach eliminates the need for soil hydraulic properties of
the fracture system, which can be difficult to determine, but can not be applied to two-

dimensional flow (Simi’rnek et al. 2003).

2.2.2.2 Dual-Permeability

Dual-permeability models are similar to dual-porosity models except water can flow
in the soil matrix as well as in the fractures. Although this addition can lead to more
realistic results, more input parameters such as those relating to the matrix, fracture and
matrix-fracture interface are required that often cannot be directly measured, increasing
the possibility of error (Simi’mek et al. 2003). Dual-permeability models are becoming
more common for modeling preferential flow in soils. The benefits of using dual-
permeability versus single-permeability are amplified not only in modeling preferential
flow in landfill covers but also in irrigation and soil contamination. Much of the
evaluation of dual-permeability models has been in the agricultural industry (Simi’rnek et
al. 2003; Kodesova et al. 2005).

Soil contaminated by the use of pesticides is a major environmental concern in

agriculture. Although contaminant concentrations in the soil can be monitored, it is an

14

expensive and time consuming process which has led to the use of numerical models to
simulate the distribution of pesticide in the soil profile as an alternative. However, the
distribution of contaminants in the soil often varies between what is observed and what is
simulated due to preferential flow. Kodesova el al. (2005) used HYDRUS-1D to simulate
the distribution of chlorotoluron in the soil profile using both single-porosity and dual-
permeability methods and compared the results to measured field data. The single-
porosity model predicted that chlorotoluron reaches to a depth of 8 cm while the dual-
perrneability predicted that it reaches to a depth of 60 cm which was much closer to the
observed behavior of the chlorotoluron. It was concluded that the dual-permeability
model preformed significantly better and more closely matched observed data than the
single-porosity model.

Gerke and Kohne (2004) conducted a bromide tracer experiment to analyze if DUAL,
a relatively simple dual-permeability model, could accurately model the preferential flow
of bromide. Simulations were also carried out using HYDRUS with the single-
permeability model for comparison. 16.75 kg of potassium bromide was applied to a tile-
drained glacial till field near the Northern Germany city of Kiel. After 138 days, the
measured cumulative outflow amounted to 27.7 cm and DUAL predicted an outflow of
28.6 cm but over predicted outflow in the beginning and under predicted it towards the
end. HYDRUS had similar results. However, when it came to predicting bromide
concentrations DUAL was able to closely simulate observed temporal patterns while
HYDRUS failed to estimate any bromide concentrations.

Gardenas et al. (2006) compared the ability of four preferential flow modeling

approaches to simulate drainage and pesticide leaching into a tile-drained agricultural

15

field. The four models, carried out in HYDRUS-2D, include: (1) an equilibrium approach
model: (2) a non-equilibrium mobile-immobile transport model combined with standard
equilibrium flow; (3) a non-equilibrium dual-porosity approach model; and (4) a non-
equilibrium dual-permeability model. The numerical models were compared to field data
from a site in Nasbygard, Sweden. The site has a warm-temperate climate and an average
annual precipitation rate of 662 mm. The non-equilibrium mobile-immobile model
produced the same water balance as the equilibrium model. The equilibrium model
captured the correct cumulative drainage for a peak period in June but only estimated
one-third of the flow of the largest peak flow measured. The equilibritun flow model also
simulated the peak flows days later then when they actually occurred. The dual-porosity
and dual-permeability models both captured the timing of the first peak correctly but
overestimated the amount of drainage. The dual-porosity model overestimated the
drainage by a factor of 2.5 and the dual-permeability model overestimated the drainage
by a factor of 5.6. The models produced similar results when it came to simulating
pesticide concentrations. In the end, the dual-permeability model simulated the dynamics
of the pesticide most accurately but overestimated the total drainage.

Now that dual-permeability models are emerging as the more favorable way to model
preferential flow, their application to modeling percolation through landfill covers must
be analyzed because preferential flow has been reported in many studies. It is critical to
understand how preferential flow can affect percolation into the waste layer especially
when proposing the use of an alternative landfill cover. The purpose of this thesis is to

numerically evaluate preferential flow though ET covers using the dual-permeability

16

model of HYDRUS-2D and HYDRUS-1D and how percolation is affected by specific

aspects of cover design.

17

CHAPTER 3
HYDRUS

HYDRUS-2D (beta v1.04) and HYDRUS-1D (v4.05) were used to evaluate
preferential flow in ET landfill covers in this study. Simulations modeled these two ET
cover systems: (1) a vegetative soil layer underlain by a storage layer with a lysimeter
boundary (without waste layer), and (2) a vegetative soil layer underlain by a storage
layer which is underlain by a waste layer (simulating a cap). HYDRUS was selected for
multiple reasons: (1) HYDRUS is a water balance model which has been used to simulate
the water balance of earthen caps (Sadek et al. 2007); (2) HYDRUS also has a dual-
permeability option which was confirmed by Simi’rnek et al. (2003) and Gardenas et al.
(2006); and (3) it is the only commercially available water balance model for which
currently a dual-permeability option is available to simulate preferential flow. The key
purpose of these simulations was to quantify the effect of preferential flow on percolation
for lysimeters versus for actual ET caps. The following chapter describes the numerical
model HYDRUS and how it was used in this study. The simulations results are discussed

in Chapter 4.

3.1 HYDRUS-2D OVERVIEW
HYDRUS-2D (beta v1.04) was used for the majority of simulations. This finite

element program is designed to model water, heat, and solute transport in a two-
dimensional variably saturated media. For the purpose of this study, only water transport

was simulated.

18

3.1.1 Water Transport

Water flow in HYDRUS-2D is governed by a modified form of Richards equation:

69—1 K[K£‘.§1’_+Kg] —s (3-1)
1

__. y.
616:: ax]

where 6: volumetric water content [L3L'3]; h = the pressure head [L]; S = sink term

[T'l]; JC-(i=1,2) = spatial coordinates [L]; = time [T]; K A = components of a
1 1]

dimensionless anisotropy tensor KA; and K = unsaturated hydraulic conductivity [LT'I]

given by:
K(h,x,z) = Ksat (x,z)K,. (h, x,z) (3-2)

where K, = relative hydraulic conductivity; and K S at = saturated hydraulic

conductivity [LT-l]. These equations were designed for a two-dimensional, isothermal

Darcian flow of water in a variably saturated porous medium. It assumes the air phase is
insignificant when considering water flow.

The user can select between multiple analytical models for describing the unsaturated
hydraulic properties of the soils such as van Genuchten-Mualem, Modified van
Genuchten, and Brooks-Corey models. All simulations in this study used a dual-
perrneability model not available to the public in the 2D version. The option of hysteresis

is also available but was not used (Simfinek et al. 1999).

19

3.1.2 Dual-Permeability

To simulate dual-permeability, the model assigns two soil properties to the porous
media, one pertaining to the fracture network (macropores) and one to the soil matrix
(micropores). The soil matrix and fiacture network are modeled as a continuum. Richards
equation is applied to both the matrix region and the fracture (preferential flow) region.
This modified Richards equation solves for the changes in volumetric water content in
the soil matrix with time by combining the flow of water due to advective and dispersion
(for solute transport) forces with the water transferred from the fracture domain. A sink
term is also added to account for any water loss fi'om the matrix. The water flow

equation for the matrix (subscript m) is:

60m 6 6h F
—=— K -—"’+K —S +—“’— 3-3
at 62[ m 62 m) m l—w ( )
and the equation for flow within the fractures (subscript f) is:

60 612

f a f 1“w
—=— K —+K —S +— 3-4
a: 62 f 62 f f w ( )

where w= the ratio of the volume of the fracture domain to the total volume of the

system; and Fw= the transfer rate of water from the fracture to the matrix and is

calculated as follows:

20

I“w =0%K07w(hf’hm) (3‘5)

where ,6 = the aggregate shape factor; a = the aggregate characteristic length; Ka =

interface hydraulic conductivity; and 7w = the dimensionless scaling factor. The dual-

permeability model is computationally much more complicated than the general flow
equation (Eq. 3-1) because now the characterization of hydraulic conductivity fimctions
and water retention are computed for both regions along with the hydraulic conductivity
function for the matrix-fracture interface (Simi’mek et al. 2003; Kodesova et al. 2005; and

Vogel et al. 2000).

3.2 HYDRUS-1D OVERVIEW
HYDRUS-1D was released in February of 2008 with a publicly accessible dual-

permeability option. This model was released after the majority of simulations in this
study were carried out using the HYDRUS—2D beta version. Hence, only a small number
of simulations were carried out using the 1D model, primarily to “confirm” the results of
the 2D beta version. The 1D version of HYDRUS is similar to HYDRUS-2D previously

discussed except that it simulates only one-dimensional water flow.

3.3 SIMULATIONS
Specific properties and input parameters required for HYDRUS-2D and HYDRUS-

ID are discussed below. Unless otherwise stated, these properties remained constant

throughout all simulations.

21

3.3.1 Design

2D representations of ET covers with preferential flow were created using HYDRUS-
2D software to evaluate percolation when specific parameters were varied. The climatic
data collected from a landfill site in Detroit, Michigan was used for all simulations. This
site is located in a sub-humid region. The covers consisted of a 0.3 m thick vegetation
layer (V L) above a 1 m thick storage layer (SL). In simulations that simulated an actual
cap system, a 6.1 m thick waste layer (WL) was added underneath the cover. However,
some simulations purposely excluded this waste layer in order to simulate a lysimeter and
in four simulations the waste layer thickness was decreased to 3.05 m or increased to 12.2
m to evaluate the sensitivity of this parameter. The width and horizontal depth of the
models were fixed at 1 m. Mesh lines were placed across all horizontal boundaries and in
between all layers to output water flux across the boundary. A schematic of the

conceptual models is presented in Figure 3-1.

3.3.2 Soil Hydraulic Properties

Various hydraulic conductivities of the storage layer (1x104, 1x106, and 1x10-6

cm/s) and waste layer (1x103, 1x104, and 1x10.5 cm/s) were input in several
simulations, while the hydraulic conductivity of the vegetation layer was maintained at
1x10"3 cm/s. The saturated and unsaturated soil properties and fitting parameters used for

describing the soil water characteristic curves (SWCC) corresponding to each hydraulic

conductivity are listed in Table 3-1. The SWCCs are plotted in Figure 3-2.

22

 

 

Lysimeter Actual Cap

 

 

 

 

 

 

 

 

Evaporation Precirlitation Evaporation Precipitation
Vegetation Layer 0.3 m Vegetation Layer
F ............................... L. ...............................
Storage Layer Storage Layer
1.0 m i
_ Upward Flux
Percolation I
I Percolation I I
Boundary Percolation
3.05 m; Waste Layer
IV 6.1 m; <
Free Drainage or
Boundary 12.2 m

 

 

 

 

 

 

Figure 3-1: Schematic of the Conceptual Models used for Numerical Modeling.

23

 

 

 

 

 

 

 

 

 

 

mohamvmwww 23M ADJ: «Em; Nnocd :3 o nb3:

um ooommmwvmo 23 42-3% mm: Sod mmd Nod rod: 2 .8 .o .m A?

was? WWW remaom BmE NNN. wmd mmd _ mo mot:

om ooomwmwmo REM 1:2 mm; mmod mmd mod obi:

moamwmflflw 23M 15-42 N303 Hood :3 o 0.35 _ gm

a. ooomwmmvmo 05 42-35 mm: Sod mmd Nod rot:

SEE/WWW 8:9 40.42 Nwfimg good :3 o moi: md q>
850m :oumWMMEU a 9.83 6 AMEQmEoV mo AAEQMEQV .o 388 sax c5 mans—03H .593

 

 

 

 

 

 

 

 

 

 

83.8““on mom 333823 98 @8838 ”Tm 2an

24

Matric Suction, ‘11 (cm)

_ _ -3 - -5 = -s
VL (KSAT - 10 cmls), SL (KSAT - 1O cmls), and WL (KSAT 1O cmls)

 

 

 

SL & WL (KSAT = 104cm/s)
SL (KMT = 1060m/s)
= ‘3 Is)
I + Wu'fsn 10 cm I

 

 

 

 

 

 

 

 

0 0.1 0.2 0.3 0.4 0.5 0.6

Volumetric Water Content, 0

Figure 3-2: Soil Water Characteristic Curves.

25

3.3.3 Climate

Precipitation and potential evaporation data was derived from measured field data
from a landfill site in Detroit, Michigan from January to December 2006. Collection of
the meteorological data was completed by Khire ‘and Mijares (2008). Hourly
meteorologic data was recorded from February 2005 to December 2006 including the
maximum and minimum temperatures, wind speed, and precipitation. Relative humidity
data was obtained from a local National Oceanic and Atmospheric Administration
(N OAA) weather station. GeoSlope’s Vadose/W (2007) was used to estimate the
potential evapotranspiration (PET) and net solar radiation based on a latitude of 42.3 80
from the measured meteorological data. Effects of transpiration were ignored by
assuming a bare ground surface to produce conservative PET rates. For the purpose of
this study, only data from January 1, 2006 to December 31, 2006 was used. In the case of
a two year simulation, the 2006 meteorological data was repeated. The total precipitation
for 2006 was 857.5 mm and the total estimated PET was 1021.9 mm. The daily and
cumulative distribution of precipitation, PET, and daily temperature and net solar

radiation are plotted in Figures 3-3, 34 and 3-5, respectively.

26

Daily Precipitation (mm)

40

35

 

 
   
   

 

l l l l l l ! 100°

...............................................

g lllllllll Cumulative

800

/ ------------ §-— 600

 

 

I ------------- -« 400

Ill o

150 200 250 300 350

 

 

i
N
o
o

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Julian Day

Figure 3-3: Daily and Cumulative Precipitation Data.

27

(mm) uogreridioard annejnurng

Oc3c_m:<m poflmsgm.
m<mno=mzmuam=oa 3:3

 

 

 

 

 

 

all“ _,
b _ L _
2. 6 8. 6. 4. 2. . 0
7 4 3 2 1

AEEV cozmaamcmboagm
.9298 .58

150 200 250 300 350
Julian Day

100

Figure 34: Daily and Cumulative PET.

28

Average Daily Temperature (°C)

40

30

20

10

.'..
o

N
O

 

 

 

 

l i l l I I l
I SolarRadiation
— ------------------------------ 1. l- -------- ------------- — 15
‘3 “I” '4“ Temperature
— .................... l ° ii ..... ‘ . ............ L 10
, j S I g '
r .............................................. . -. —— 5
l ‘ l
l
I. 4 I '1 PL
- ' ................................................... ...... .'-.. 0
I ‘ s i i
l l l l 1 _5
0 50 1 00 1 50 200 250 300 350

Julian Day

Figure 3-5: Daily Average Temperature and Daily Net Solar Radiation.

29

(aw/rw) uonelpea Janos leN Alien

3.3.4 Boundary Conditions

Atmospheric boundary conditions were applied for the top boundary (top of the
vegetation layer). 365 time-variable boundary records were used correlating to the daily
precipitation and PET input for the top boundary. Heat flow and ground freezing was

ignored. A free drainage boundary (unit gradient) was assigned to the bottom boundary

(Figure 3-1).

3.3.5 Time Step, Error Tolerance, Mesh Design, and Initial Conditions

Time was measured in days with an initial time of O and final time of 365 beginning

on January 1, 2006 and ending on December 31, 2006. Initial, minimum and maximum
time steps were kept constant at 1x10'15, 1x10'12 and 0.5 days, respectively.

The maximum number of iterations was 20, water content tolerance was 0.001 and
pressure head tolerance was 0.01 m. HYDRU-2D default values were used for all other
iteration criteria. I

A rectangular mesh domain was used for all simulations but the vertical discretization
of the mesh varied depending on the performance of the simulation. When possible, the
number of mesh nodes in the vertical direction was 750 for covers with a waste layer and
200 for the lysimeters. These values were used because increasing the mesh nodes to
greater than 750 and 200 when possible made no significant difference in the simulation
results. Due to numerical error, some simulations did not run to completion at these mesh
sizes. For those situations, the number of vertical nodes was reduced until the simulation
ran. A reduction in the number of vertical mesh nodes had a significant effect on the
estimated percolation in some cases. This may be due to numerical error. The mesh

discretization of each simulation is discussed in Chapter 4.

30

The initial condition for all soils was set at -10 m of pressure head (suction). Initial
pressure head conditions should match the matric suction corresponding to the measured
water content of the soil being modeled. In this situation, no real site or soil was being
simulated. This initial condition corresponds to a volumetric water content of about 0.25

for all cover soils (Figure 3-2).

3.3.6 Fracture Properties

Parameters regarding the fracture-matrix interface were kept constant at w = 0.5, B =

3, y = 0.4, and a = 1 for all simulations. The interface hydraulic conductivity, Ka, was
1/100th of the soil matrix hydraulic conductivity. When a landfill cover system with
preferential flow was being simulated, the fracture hydraulic properties were 0Ir = 0.0, 95

= 0.5, a = 0.1 cm’], n = 2, and I = 0.5. For all simulations with preferential flow, the

hydraulic conductivity of the fracture was kept at 2 cm/s except when this parameter was
varied to observe its effect on percolation. These parameters were used by Simfinek et al.
(2003) when preferential flow was simulated. For the simulations when only capillary
flow was modeled, the fracture parameters were set equal to the corresponding matrix

parameters.

3.3.7 HYDRUS-1D

HYDRUS-1D with dual-permeability option was released after a majority of
simulations had been completed in HYDRUS-2D. For this reason, HYDRUS-1D was
used to verify a small number of simulations carried out in HYDRUS-2D. Almost all cap
simulations and lysimeter without preferential flow simulations produced closely

matching results in HYDRUS-1D and HYDRUS-2D. However, simulations of a

31

lysimeter with preferential flow almost always had difficulty running to completions in
HYDRUS-2D. In many cases, the mesh had to be decreased to 75 nodes vertically in
order for the simulations to run to completion which probably yielded erroneous results.
For this reason, all simulations of a lysimeter with preferential flow were executed in
HYDRUS-1D at a mesh of 150. In these simulations, iteration criteria had to be altered so
the simulations would run. The water content tolerance was decreased to 0.0001 and the

pressure head tolerance was decreased to 0.001 m. The upper limit of the tension interval
was increased from 100 m as used in HYDRUS-2D to 1x106 m. In HYDRUS-1D an
additional input parameter regarding the fraction of precipitation that goes into the
fracture has been added. This controls how much of the applied precipitation enters
directly into the fractures. In HYDRUS-2D, this parameter is referred to as qTop but is

“hard-coded” and hence cannot be changed easily. The HYDRUS-2D default value of 0.5
was used for all simulations. All other parameters were maintained the same as that used

for HYDRUS-2D.

32

CHARTER 4
RESULTS AND DISCUSSION

The key objective of this study is to numerically evaluate the effect of various soil
and waste hydraulic parameters on the percolation through evapotranspirative landfill
covers and lysimeters when preferential flow is present. All simulations were carried out
using HYDRUS-2D and HYDRUS-1D. The numerical model simulations are
summarized in Table 4-1.

These two key hypotheses were tested in this study: (1) percolation measured using
lysimeters does not equate to the percolation across the cover waste boundary because
lysimeters exclude the waste layer and evapotranspirative gradients at the interface; and
(2) when only capillary flow is considered (and flow through fractures is ignored)
percolation is under-estimated. The results of the simulations listed in Table 4-1 are

discussed in this chapter.

33

 

 

 

 

 

 

 

 

 

 

 

 

acorn—38mm .3 com com: 33 E Sod co E 86 mo 8:829 30: Eamoa 28 586 .5 Sad mo 8:828 «588 53>» .m
.5333 3:5 05 we acorn—Ea =m Em com: 33 E S- 8 3:3 30: 2385 .v
.EouaEEm an.» N com @833.— 83 Ba Eons—3E7. .3 Eu com: 83 cooN E 380:8 Emu BEE—c.8322 .m
«noun—=86 =a Lou E _ u 15,—. 98 E Nd n .5... .N
SSS—£5. u H can manage...”— u M."— ugwq 83>» u A? ”BEA ems—8m N Am comma cow80w0> n A> ._ “802
N.N_
24.. .wE md m6 _ 1:5 N $3 92 M.2 fie m m .5;
3%
0.3
S-.. a... . . . a...
:4 .wE m o m o _ 15, N v.3 02 Nb— _ o m m M.
n-..—
3
2... a... no no . 43 N we as we 3 e a do.
“I
mad
a-.. .m... 2. .
. S S . .85
wé .wE N6 n o _ A? N m- “.2 m- _ e v v
m .e
3.:
- . . 3. .
a a w... n o 2. _ as N m.2 so. we _ a _ m a
m .c
N E
0% .wE We nd _ AB Nd v.3 92 WE #0 m m EM
w N3. m.
We . E 1:5
- . . . . . Seam
a a a... a o m c . Am N no. we re . a a e 3305
A>
a-.. a; . . . sage...
Tv .mE m o m o N 43 N m-o_ #2 We _ e N N Lo 853$
AmvoEwE .55 3 3896 Eme @E8 383 $23 3E3 AEV Lugosi»:— qeu uotm>
mam—0:38.50 22339 238..“— mn—v— .53— Ex a>¥ .5; Ecru—25m SEEN—mm

 

 

 

 

 

 

 

 

 

 

 

 

 

.mBEE was: as 3.50 22.25% .a Seas-a a-.. 2.3

34

4.1 PRESENCE OR ABSENCE OF F RACTURES

Four simulations were executed in HYDRUS to evaluate the effect of capillary flow
only versus capillary and preferential flow. These simulations represent a time period of

two years. The simulated cumulative percolations versus time are shown in Figure 4-1.
All simulations have a hydraulic conductivity of 10'3 cm/s and 10'5 cm/s for the

vegetation layer and storage layer, respectively. Cap simulations included a waste layer

having a hydraulic conductivity of 10'3 cm/s. All simulations have a waste layer

thickness of 6.1 m when present. When preferential flow was simulated, fractures
extended through the waste layer for the cap simulations and through the storage layer for
the lysimeter simulations. Vertical discretization of mesh nodes was 750 for the cap with
no preferential flow and 600 for the cap with preferential flow. 200 mesh nodes were
used for the lysimeter without preferential flow and 150 for the lysimeter with
preferential flow. Simulations in HYDRUS-2D for the lysimeter with preferential flow
contained unrealistic results of upward cumulative percolation which was attributed to
numerical error and therefore were repeated in HYDRUS-1 D. To demonstrate the
accuracy between the two HYDRUS versions, the simulation of the cap with preferential
flow was also repeated in HYDRUS-1D and produced similar results as seen in Figure
4—1 along with the results of the other four simulations. The water balance error for the
simulations was negligible (< 0.1%) for the lysimeters and approximately 27% for the
cap, which is relatively high.

To insure that simulation width does not affect percolation, selected simulations were
repeated with a width of 1 cm. These simulations confirmed that simulation width is a

non-influential parameter.

35

 

 

 

  
 

 

 
 
 

 

Cumulative Percolation (mm)

 

 

   

 

 

1 capmetawimfraetures) _ _ . .
1° ”* Cap(1Dwithfracturas) ‘3 """"""
102 ----------- ;—

Lysimeter(withfractures : : : -
103 1 1 l l l l l

 

o 100 200 300 400 500 600 700
Julian Day

 

-100 . .
(b) w = qTOP- = 0.5

0 ECap (with no fractures\
\TLyslmeter (with no MMM

 

 

 

 

 

E

E,

S

2:. 100

L“

O

CGL, 200 b Cap (Betawithfractures) _

an . g g ‘ ‘

.2

*5 300 * """"" Cap(1Dwithfractures) j ----------------------------- ~

g Lyssmeter(withfractures) . .

a 400 ----- - ------

0 2 6 5 2 2 2 i
500 l l l l l l l

 

Julian Day

Figure 4-1: Simulated Percolation through Cap versus Lysimeter with and without
Preferential Flow on 3 Log Scale (a); and Linear Scale (b).

36

Figure 4-1 shows that percolation follows a cyclical pattern each year when only
capillary flow is considered. In the cases with preferential flow, a cyclical pattern was not
observed. Figure 4-lb plots the same cumulative percolation of Figure 4-la on a linear
scale to confirm that a cyclical pattern was not observed when preferential flow was
present. Figure 4-1 also illustrates that the lysimeter without preferential flow had
percolation that was about two orders of magnitude higher compared to the cap without
preferential flow. However, once preferential flow was introduced, percolation increased
as hypothesized but with no significant difference in percolation between the lysimeter
and cap. For the remaining simulations, simulation length was reduced to one year due to
time constraints.

The cumulative evaporation and runoff and the cumulative percolation through the
bottom most boundary of the four simulations are presented in Figure 4-2. The cap and
lysimeter with no preferential flow have the same amount of cumulative evaporation and
runoff because the infiltrating water is being stored in the storage layer of the
evapotranspirative cover as it is designed to do. They also have the highest amount of
cumulative evaporation and runoff because they lack preferential flow which carries
water away from the surface at a much faster rate than the surrounding soil matrix. The
cap with preferential flow has the second most cumulative evaporation and runoff. This is
because the preferential flow moves the water to deeper depths reducing direct
evaporation and runoff than the cap and lysimeter without preferential flow. It has more
cumulative evaporation and runoff than the lysimeter with preferential flow because

water is allowed to be stored in the waste layer and later be removed by evaporation.

37

 

1200 l l T l l i I
1000— .......... ...... ........... .......... ......... /

_______ ,_ ............ _ CaporLysimeter
800 f 2 2 2 - s 5 s = 'flymhnofiaaums)

5 i i E E a - -089
600 — ---------- ----------- -------- .-----------'-— (withfractures)

, bLysimeter
E E E ' 3 3 § (withfractures)
400 —~--~~--; ------------ ------ -:- ~ —

200 _ .......... . .......................................................... ........... E
I D I I I l

 

 

 

Cumulative Evaporation & Runoff (mm)

o 50 100 150 200 250 300 350
Julian Day

 

 

 

 

10'1 5 i 5 i .

10‘ i i IE! iwamuErumhfiadumsi§_

—.. ...........................................
n n u n
I

.__ ‘

 

Cumulative Percolation Through the
Bottom Most Boundary (mm)

 

0 50 1 00 1 50 200 250 300 350

 

Julian Day

Figure 4-2: Simulated Cumulative Evaporation and Runoff (a); and Simulated
Cumulative Percolation through the Bottom Most Boundary (b).

38

The lysimeter lacks the waste layer and all water passing though the storage layer is
immediately removed from the system.

The cumulative percolation through the bottom most boundary, the storage layer for
the lysimeter and waste layer for the cap, is presented in Figure 4-2. These percolation
amounts follow a similar trend as the percolation in Figure 4-1. The differences in
percolation among the simulations can be attributed to the same reasons mentioned in

regards to the cumulative evaporation and runoff.

4.2 FRACTURE PROPERTIES

Parameters specific to the pathways of preferential flow, such as the depth through
which they extend, the hydraulic conductivity of the fi'actures, the volume of fractures,
and the fraction of surface flow entering the fractures can also have an effect on the
amount of percolation into the waste layer. Hence, the effect of fracture depth, fiacture
hydraulic conductivity, fracture volume, and the amount of surface flow entering the

fractures were evaluated.

4.2.1 Fracture Depth

The depth of fractures or preferential flow paths, although difficult to asses in the
field, can influence percolation into the waste. For the lysimeter, fracture depth was
simulated to extend though the vegetation layer, the vegetation layer and half way
through the storage layer, and through the entire vegetation and storage layer. For the
cap, fracture depth was simulated to extend thOugh the vegetation layer, the entire
vegetation and storage layer, one third of the way into the waste layer, two thirds into the
waste layer, and through the entire model profile. The effect of fracture depth is

illustrated in Figure 4-3. Although HYDRUS simulates a continuous network of fractures

39

rather than individual fractures, Figure 4-3 shows the fractures as one large individual
fracture for illustration purposes only.
Figure 4-4 represents cumulative percolations from ten simulations showing the

affect of preferential flow depth on percolation. All simulations were carried out with a
hydraulic conductivity of 10'3 cm/s and 10'5 cm/s assigned to the vegetation layer and
storage layer, respectively. Cap simulations included a waste layer having a hydraulic
conductivity of 10'3 cm/s. A waste layer thickness of 6.1 m was used in all cap

simulations. Preferential flow networks were assigned a hydraulic conductivity of 2 cm/s.
Vertical discretization of mesh nodes was 600 for all cap simulations and 150 for
lysimeter simulations. The water balance error for the simulations was negligible (<
0.1%) for the lysimeters and approximately 22% for the cap.

A noticeable pattern emerges in Figure 4-4 for both the lysimeter and cap
simulations. As preferential flow depth increases, so does the amount of percolation. This
occurs because preferential flow carries infiltrating water to deeper depths at faster rates
than the surrounding capillary pore matrix. The deeper the preferential flow, the faster the
water is carried towards or across the storage layer-waste layer boundary where
percolation is assessed. These results indicate that the effectiveness of the storage layer to
store the infiltration and reduce percolation is impaired as the depth of preferential flow

networks increase.

40

 

Lysimeter Cap

Preferential flow extending through: Preferential flow extending through:
1/3 2/,
VL SL WL WL WL

 

 

VL 'ASL

 

 

 

 

 

 

 

SI.
1 ._ .....
l a m.

Percolation

Boundary

 

 

 

 

 

 

 

 

 

[:I Preferential Flow Channel
Arrows indicate direction of percolation through the system.

 

Figure 4-3: Depth of Preferential Flow through Lysimeter and Cap.

41

 

 

 

 

-5
10 (a) ‘ All simulations of lysimeter. '
A 10-4 _________ _________ w=qT03 =O'_5 ..'__‘_4
E _ i : 3
E s s s s
V ‘3 _.-.... ............ .-....---.:.E
C 10 E E E E
2 z 2 a 2
E 102 ------- ------------ ~
8 § § NoFractures §
'5 101 — ------ ------- -------- /—
; . : FracturesThroughVL -
D. Z I \
9 10° ----------- —
a 9 :
3 10‘ _-... ....................................................................... :2
E f Fractures Through 1I2 of SL
3 2 : _
o 10 ~ —
5 iFracturesThroughSL i
103 1 1 I l l l l

 

 

0 50 100 150 200 250 300 350
Julian Day

 

All simulations of up.

 

 

 

 

Fractures Through 1B of WL 5r
Fractures Through 2I3 of WL

 

Cumulative Percolation (mm)

 

Fractures Through WL

 

 

 

 

0 50 100 150 200 250 300 350
Julian Day

Figure 44: Effect of Fracture Depth on Simulated Percolation for Lysimeter (a); and
Cap (b).

42

To insure that the results were not just a product of the layer’s soil water

characteristic properties, the waste layer hydraulic conductivity was decreased from 10'3

cm/s to 10‘4 cm/s. It can be seen in Figure 4-5 that despite the change in waste layer

hydraulic conductivity, a similar pattern in percolation was observed. Regardless of the
soil and waste matrix hydraulic properties, percolation increases as the depth of the

fracture network increases.

4.2.2 Hydraulic Conductivity of Fractures

A second parameter specific to the pathways of preferential flow is the hydraulic
conductivity of the fractures. Because HYDRUS simulates preferential flow by
representing the fracture network as a separate material, a hydraulic conductivity is
required to model the rate of water flow through the system. Six simulations were
executed, three for a cap and three for a lysimeter, with three different fi'acture hydraulic
conductivities. The results are displayed in Figure 4-6.

Three fracture hydraulic conductivities, 2 cm/s, 0.2 cm/s, and 0.02 cm/s, were applied

to the cap and lysimeter simulations. All simulations were carried out with hydraulic
conductivity equal to 10'3 cm/s and 10'5 cm/s assigned to the vegetation layer and storage
layer, respectively. Cap simulations had a waste layer with a hydraulic conductivity of
10'3 cm/s and a waste layer thickness of 6.1 m. Fractures extended through the waste

layer in the cap and through the storage layer in the lysimeter. 750 mesh nodes were used
to discretize the cap and 150 for the lysimeter. The water balance error for the
simulations was negligible (< 0.1%) for the lysimeters and approximately 20% for the

cap.

43

 

 

 

 

 

 

 

 

 

 

 

 

 

All simulations of cap with fractures. I I
-6 Z _ _ I :_J
E 10 ~ -------- qump -'0-5- .
é .......................................................... ............ __
: _. -3 5
S ; .ThroughVL(KWL— 10 cmls).--.§-2
E ...................................... Throu hSL =10‘3cmls __.§-_
8 9 m ’ a
Li.) /Through VL(KWL=1O'4chs) ;
a, ....................... ............ L-.------ -._.
.2 4 \s\
*5 ........ Through SL(KWL= 10 cmls)—”g:
'5 - : - '
E - -------------------------------------------------------- z ----------------
10 - 3 —
' 3 Through WL(Kw1= 10‘30rnls or 104cmls)/5'~ 5 5
10 l l I
0 5O 1 00 1 50 200 250 300 350

Julian Day

Figure 4-5: Effect of Waste Layer Hydraulic Conductivity and Fracture Depth on
Simulated Percolation.

 

 

 

 

 

 

 

 

 

. AllIsimulatiIons contain fractures. I I
.6 I I = = I Z 2
’E‘ 10 ““““““ """"" w qrop. 05“
é ____________________________________________________________ __________________________ E
c . 3
.9 ““““““ : """""""""""""" “
5 Cap (KFR - 0.02 cmls) _______________ _
O : : . .
e I I E :
a: Cap(KFRn=0.20rnls)-;.-----...---.g-2
as ‘ _ = : _
'5 Cap(KFR-2cmls)or
g ' Lysimeter (KFR = 0.02, 0.2, 2 cmls)‘
3 k I n a
E ------------- ' ----------- f ----------- T- ---------- 75>
l I i l

 

 

 

 

Julian Day

Figure 4-6: Effect of Hydraulic Conductivity of Fractures on Simulated Percolation.

45

Figure 4-6 shows that the hydraulic conductivity of the fractures affects the
percolation by a significant amount for the cap. For the cap simulations, percolation
decreases as the hydraulic conductivity of the fracture network decreases. For the
lysimeter simulations, the hydraulic conductivity of the fractures had little effect on the
percolation. The length of travel through fractures for infiltrated water in the lysimeter is
relatively small compared to that for the cap. Hence, for a given fracture conductivity the

percolation through the lysimeter is greater than that through the cap.

4.2.3 Volume of Fractures

The ratio of the volume of the fracture domain to the total volume of the system (w)
is an important parameter because it controls the relative proportions of preferential flow
versus capillary flow in the soil profile. To evaluate the effect of volume of fractures,
eight simulations were carried out with varying w.

w was varied (0.5, 0.2, 0.1 and 0.05) in eight simulation and the results are

presented in Figures 4-7. All simulations used a hydraulic conductivity of 10'3 cm/s and
10'5 chs for the vegetation layer and storage layer, respectively. Cap simulations

included a waste layer having a hydraulic conductivity of 10'3 cm/s with a waste layer

thickness of 6.1 m. Fractures extended through the waste layer in the cap and through the
storage layer in the lysimeter. The hydraulic conductivity for the fiactures was assigned 2
cm/s. Vertical discretization of mesh nodes was 600 for the cap except, when w was
0.05, the mesh was 500. Vertical discretization of mesh nodes was 150 for the lysimeter.
The water balance error for the simulation was negligible (<0.1%) for the lysimeters and

approximately 16% for the cap.

46

 

(a) All simulations of lysimeter with fractures extending
through the storage layer. _
= 0.5.

Cumulative Percolation (mm)

 

 

 

 

,0. 1 1 1 1 i 1 1
0 50 100 150 200 250 300 350
Julian Day

 

 

 

 

 

10'5 . . . . . ,
b) All simulations of cap with fractures extending I
A 104 __________ through the waste layer. A
E \ qropzo‘s' i
g 10-3 ............ ............ _
g I; 3 a a § é : s
E 10‘2 — ------------ §w=oo§ —~
8 1 1 2
a 10 _ ............................................................................ _.
0- a
0 :
02) 10 _ ..................................................................... a
E 1 . i
3 10 — ..... ‘ 5 ........................................................................ _
g = —
0 102 w--- ................................................... ' ................. “a
103 l 1 1 l L 1 l

 

 

 

0 50 100 150 200 250 300 350
Julian Day

Figure 4-7: Simulated Effect of the Volume of Fractures on Percolation for Lysimeter (a);
and Cap (b).

47

From these simulations, it was observed that the ratio of the volume of the fractures
to the total volume of the system can influence percolation. For both the lysimeter and
cap, the larger the fracture domain, the less cumulative percolation. The only exception
being when w is 0.05 for the cap. This simulation seams to vary from the pattern seen in
the other seven simulations and is probably due to the coarser mesh size used for that
particular simulation which may have reduced the accuracy of the results. It should be

noted that qmp, the fraction of surface flow entering into the fracture network, was

maintained at 0.5 for every value of w so even as the amount of fractures Changed, the
amount of water entering into the fractures remained the same. The unchanging (“‘01)

value might be the reason for the unexpected results of these simulations.

4.2.4 Effect of Fraction of Surface Flow Entering Fractures
The fraction of surface flow entering directly into the fracture domain (qmp) is a
fourth fracture property that could potentially affect the amount of percolation. qmp was

varied in eight simulations and the results are displayed in Figure 4-8.

qTop was varied between 0.5, 0.2, 0.1 and 0.05. w remained constant at 0.5 in all
simulations. For all simulations, hydraulic conductivities equal to 10'3 cm/s and 10'5 cm/s
were assigned for the vegetation layer and storage layer, respectively. Cap simulations
included a waste layer having a hydraulic conductivity of 10'3 cm/s. All simulations have
a waste layer thickness of 6.1 m, when present. Fractures extended through the waste
layer in the cap and through the storage layer in the lysimeter. The hydraulic conductivity

for the fractures was assigned 2 cm/s. Vertical discretization of mesh nodes was 750 for

the cap except when qmp was 0.05 the mesh was 600. The mesh was 75 for all lysimeter

48

 

(a) I All simulations of lysimeter with fiacturES.

 

Cumulative Percolation (mm)

 

 

 

l i i 1
0 50 100 150 200 250 300 350

 

Julian Day

    

All simulations of cap with fracturesi
. w = 0.5. :

 

...............

 

 

 

 

 

Cumulative Percolation (mm)

 

 

 

0 50 100 150 200 250 300 350
Julian Day

Figure 4-8: Simulated Effect of qmp for a Lysimeter (a); and Cap (b).

49

simulations and were completed in HYDRUS-2D because they would not run in
HYDRUS-1D once qmp was changed. The water balance error for the simulations was
negligible (< 0.1%) for the lysimeter and approximately 20% for the cap.

As seen in Figure 4-8, qmp has a greater effect on percolation in the cap than in the
lysimeter. In the cap simulations, percolation increases as qTop increases. Figure 4-9
illustrates the effect on percolation when w and (“‘01) are varied together. In these

simulations, as w and qTO-p increase so does the percolation as expected.

4.3 HYDRAULIC‘CONDUCTIVITY OF STORAGE LAYER

Design properties of the evapotranspirative cover also influence the amount of
percolation into the waste layer. Here the hydraulic conductivity of the storage layer was
varied in 12 simulations to demonstrate its effect on percolation and the results are
displayed in Figure 4-10.

Three storage layer hydraulic conductivities (10'4 cm/s, 10'5 curls, and 10'6 cm/s)
were applied to the cap and lysimeter, with and without preferential flow. All simulations

have a vegetation layer with a hydraulic conductivity equal to 10'3 cm/s and cap

simulations have a waste layer with a hydraulic conductivity equal to 10'3 cm/s. All cap
simulations have a waste layer with a thickness of 6.1 111. When preferential flow was
simulated, fractures extend through the waste layer in the cap and through the storage
layer in the lysimeter with a hydraulic conductivity of 2 cm/s. Vertical discretization of
mesh nodes was 750 for the cap except when the storage layer hydraulic conductivity was

104 cm/s the mesh was reduced to 250 in order for the simulation to run.

50

 

 

 

 

-5 I I ' I I
10 (a) I All simulations of lysimeterwith fractures.
A 4 i é i 2 : é
E 10 ” """""""""""""""""""" ‘
é a i E -
8 10—3 _ -. .................. ......................... fi
'fi 2 3 costs. 005
_ ---------- ; ------------------------ w: _ = ------ —
2 1 s s s w=0.1&qmp=01
a) —-- ---------- . ------ —
n. 10 g g g w-ozaqmp=02
a, I I I I
> —--- ------------ : ---------- w=osa =05 ------ —
g : 5 5 : . qTOP
a — ----------------------------- —
E I ;
a 5 . :
l l 1 1 l l l

 

o 50 100 150 200 250 300 350
Julian Day

 

 

 

 

Cumulative Percolation (mm)

 

 

 

 

 

 

Julian Day

Figure 4-9: Simulated Effect of w and qTop on Percolation through a Lysimeter (a); and
Cap (b)-

51

 

 

   

 

 

1 '7 - . . . . . .

0 (a) 1 All Simulations of lysrmeter. I I
E 10-6 __________ w=qTGP=o'5 _
E -5 ____________ I I i E
Z 10 KSL=1O'6chs
,9 104 “ ' "§‘(Betaand10withnofmctures)“““’g'*
"" ' I I I '
L3 103 """""" " 'm‘. """"""" """" . :
3 10.2 ___________ :KSL=1O’5chs(wilhnofraqures)m_§__
ct’ \ 2 s
a) 10'1 ~— ' ---------- :KSL=10“chs(withnofmctures)---g-—
.2 ' 1 . . . g
E 100 — -------------------------------------------- ——l
3 s
E 101 — ----------------------------------------- “a;
3 :
O 2 _ r : ' . ___________________ E__

10 , , .

K =104,1O’5,or10'6cmls(withfractures) s s
103 $1 1 1 1

 

 

 

o 50 100 150 200 250 300 350
Julian Day

 

6 (b) T; All simulationsofcap.w=q;op=0.5 3
10 ‘ I """""" """" I‘m: d
10-5 -- ____________ g._-_-__-___§.__/_.K&=1o‘mtIs(withnohactures)_.

 

 

KSL= 10'5cmls (with no fractures). —

----.-._.. .--..-.-..--.........--..r- -.-...-~1...-----..‘...-.......--- ..............

 

 

 

 

 

Cumulative Percolation (mm)

 

 

 

KSL= 10"cmls (with no fractures) _

KSL= 10‘6cmls (with fractures)?“

Ksr= 10'5 or 104cmls

(withfractures) ‘
1 l i l

 

o 50 100 150 200 250 300 350
Julian Day

Figure 4-10: Effect of Hydraulic Conductivity of the Storage Layer on Simulated
Percolation of Lysimeter (a); and Cap (b).

52

Vertical discretization of mesh nodes was 150 for the lysimeter except when the storage

layer hydraulic conductivity was 10'6 cm/s the mesh was reduced to 25 in order for the

simulation to run. The cap simulation with a storage layer hydraulic conductivity of 10-6

cm/s and no preferential flow was also ran in HYDRUS-1D as a check to insure the two
HYDRUS programs produce the same results. The water balance error was negligible
(<0.1%) for simulations ran in HYDRUS-1D and approximately 30% for the remaining
cap simulations.

As seen in Figure 4-10, when preferential flow is absent, percolation decreases as the
hydraulic conductivity of the storage layer decreases. This is expected because the lower
the hydraulic conductivity, the more restricted the water flow through the layer. Once
preferential flow is simulated, percolation is just over 100 mm for all three storage layer
hydraulic conductivities. This happens because the water has a direct route through the
preferential flow pathways to the bottom of the storage layer. Hence, the percolation is
independent of the hydraulic conductivity of the matrix of the storage layer when

preferential flow is significant.

4.4 WASTE LAYER PARAMETERS

Waste layer parameters can influence percolation even though it is not part of the
cover system (Khire and Mijares 2008). The hydraulic conductivity and soil-water
retention properties of the waste layer and its thickness are parameters that could

potentially affect percolation and were explored in the simulations in Figure 4-11.

53

 

; g All simulations of caplwithout fractures.
— . w=qTOP=O.5*

Beta Version
(MSW SWCC)

 

 

:l ' 10 Version
(SoilSWCC)

Beta Version
(Soil SWCC)

 

 

 

Cumulative Percolation (mm)

5%: 10 cmlsand10'5 crnls

 

 

iKM=1o '4cmlsand10’5cmls

1 03 l l l - _ A
0 50 1 00 1 50 200 250 300 350

Julian Day

 

 

 

 

10'7 , ._ I . . . .
(b) . . All srrnulatrons ofcapwrthfractures.
E 10*6 _ -------- “1:qu =o.5-
E 10‘5 1<WL=10°chsqu3wswccm ..... a
g 10“ —---'M=1O’3crnls(3etaMSWSWOC)or ----- ------------ ~
g 10'3 ""igNL=1o“cm/s(8etaSoiisw00)or ----- ‘‘‘‘‘‘‘‘‘‘‘ -
2 102 . _ .5 ..... .ss
an KM- 10 cmls (Beta Soil SWCC)
m
.2
4.0
E
3
E
3
O

 

 

 

 

o 50 100 150 200 250 300 350
Julian Day

Figure 4-11: Simulated Effect of Hydraulic Conductivity of Waste Layer on Cap
Percolation without Preferential Flow (a); and with Preferential Flow (b).

54

4.4.1 Hydraulic Conductivity of Waste Layer

Six simulations were used to illustrate the effect the waste layer hydraulic
conductivity has on percolation. Because the cap simulation results were not as expected,

all three simulations were repeated in HYDRUS-1D and are discussed below.
Three waste layer hydraulic conductivities (10'3 cm/s, 10.4 curls, and 10'5 cm/s) were
input for the waste layer of the cap with and without preferential flow. Hydraulic

conductivity of the vegetation layer and storage layer were input equal to 10'3 cm/s and

10'5 cm/s, respectively. The simulated waste layer thickness was 6.1 m. When

preferential flow was simulated, fractures (hydraulic conductivity = 2 cmls) extended
through the waste layer. Vertical discretization of mesh nodes was 750 for all simulations
in both versions of HYDRUS. The water balance error was negligible (<0.1%) for
simulations ran in HYDRUS-1D, approximately 30% for simulations ran in HDYRUS-
2D without fractures, and approximately 21% for simulations ran in HYDRUS-2D with
fractures

Figure 4-11 shows that when preferential flow is absent, the largest waste layer
hydraulic conductivity, 10'3 cm/s, has the least amount of percolation. The HYDRUS-1D
simulations produced similar results. Even though the amount of percolation produced by
the 10'3 cm/s waste layer varies between the two HYDRUS versions, they both show that
the greater the waste layer hydraulic conductivity the smaller the amount of percolation.
Because these results are not intuitive, simulations were repeated using the van
Genuchten-Mualem single-porosity modeling option as opposed to the dual-permeability

modeling option. The simulations using the van Genuchten-Mualem single-porosity

modeling option produced the same trend but are not shown in Figure 4-1 1. One possible

55

explanation is the contrast in hydraulic conductivities between the storage and waste
layer. The storage layer has a hydraulic conductivity of 10'5 cm/s so the larger the

hydraulic conductivity of the waste layer, the larger the contrast in the soil-water
retention properties between the two. This may be producing an effect that is similar to a
capillary barrier. The water in the storage layer can not move into the waste layer until a
water-entry matric suction is reached (Fig. 1-3). As the waste layer hydraulic
conductivity decreases, it more closely matches that of the storage layer so the water-
entry matric suction is reached more quickly allowing water to infiltrate into the waste
layer. A second explanation may be the effect of the waste layer SWCC. For a waste
layer hydraulic conductivity of 10'3 cm/s, an SWCC developed by Benson and Wang
(1998) representing MSW was applied (Table 3-1). When the waste layer hydraulic
conductivity was 10'4 cm/s and 10'5 cm/s, SWCCs of soils with the same hydraulic
conductivity were applied because there is very little published date on MSW SWCCs as
the hydraulic conductivity of the MSW decreases when the waste settles and undergoes
decomposition. The assumption that SWCC of MSW resembles that of an equivalent soil
when the waste has undergone degradation may not be realistic.

In order to demonstrate the effect the waste layer SWCC may have on the
percolation, the simulation with a waste layer hydraulic conductivity of 10'5 cm/s was
repeated with the MSW SWCC. As shown in Figure 4-11, this had a significant effect on
the percolation demonstrating that waste layer SWCCs can greatly influence simulated
percolation. Unfortunately, predicting waste layer SWCCs as waste undergoes physical

and biological changes is difficult and there is a lack of data in relavent literature. Once

56

preferential flow is introduced, the waste layer parameters have very little influence and

all simulations produced approximately the same amount of percolation (Figure 4-11b).

57

I
—L

Ratio of Cumulative Percolation, (PrCAP/PrLYS)

 

M: 10'3 WSUM=E105WS(W1MSWSVW3CTUM)

 

 

 

 

 

 

"""""""""""""""""""""""" 1;; 2m
w = q“? = 0 5
No Fractures VL SL Entire Profile
Fracture Depth Extending Through

0002/15 \‘ -: 0.4/1834— Pr .
_. ............... a g ............ MLYS.-..2 Lysrmeter
; 0.005to 0.02/7.3 sea/18:3 WW
I g \ Percolation
E fig183\ l

Lysimeter
Lhderestirmtes

Figure 4-12: Ratio of Cumulative Percolation with Varying Fracture Depths and Waste
Layer Hydraulic Conductivity.

58

Figure 4-12 shows the ratio of cumulative percolation through the cap to that for the
lysimeter for varying depths of the fracture network and varying waste layer hydraulic
conductivities and SWCCs. This graph is intended to demonstrate that lysimeters may
over predict or under predict percolation depending upon the extent of cracks and the
unsaturated hydraulic properties of various components. In Figure 4-12, when the ratio is
grater than 1, that means the lysimeter over estimated percolation. Figure 4-12 shows that

the lysimeter over estimates percolation except when the waste layer hydraulic
conductivity is 10'4 cm/s and 10'5 cm/s when the SWCCs of soils were adapted as input
of SWCC for the waste. When the simulation with waste layer hydraulic conductivity of

10'5 cm/s was repeated with an MSW SWCC as opposed to a soil SWCC, the lysimeter

over estimated percolation. Figure 4-12 demonstrates that lysimeters typically over
estimate percolation and also shows the influence the waste SWCC can have on
simulated percolation. Clearly, measurement of representative SWCCs and waste
hydraulic conductivities is critical if we want to predict the percolation fiom an actual cap

based on the percolation measured from a lysimeter.

4.4.2 Thickness of Waste Layer

Six simulations were preformed to evaluate the effect of waste layer thickness on
percolation. Waste layer thicknesses equal to 3.05 m, 6.1 m and 12.2 m were simulated.
All simulations had a hydraulic conductivity of 10'3 cm/s, 10'5 cm/s, and 10'3 cm/s for the
vegetation, storage, and waste layer, respectively. When preferential flow was simulated,
fractures extended through the waste layer with a hydraulic conductivity of 2 cm/s.
Vertical discretization of mesh nodes was 300, 750 and 900 for the 3.05 m, 6.1 m, and

12.2 m waste layer respectively. The water balance error for the simulations was

59

approximately 30%. As seen in Figure 4-13 waste layer thickness does not influence

percolation amounts into the waste layer for the thicknesses simulated in this study.

60

Cumulative Percolation (mm)

 

 

 

 

 

 

 

 

 

 

Julian Day

Figure 4-13: Effect of Waste Layer Thickness on Simulated Percolation.

61

CHAPTER 5
SUMMARY AND CONCLUSIONS

In this study, parameters impacting percolation into the waste layer of an
evapotranspirative landfill cover with preferential flow was evaluated using HYDRUS—
2D (beta v1.04) and HYDRUS-1D (v4.05). Simulations consisted of either a two layer
soil cover without an underlying waste layer (lysimeter) or the same cover with a waste
layer (cap). The purpose of these simulations was to evaluate: (1) percolation measured
using lysimeters does not equate to the percolation across the cap and waste boundary
because lysimeters exclude the presence of a waste layer and evapotranspirative gradients
at the interface; and (2) when only capillary flow is considered and flow through fi'actures
is ignored percolation is under-estimated.

Simulation duration, the presence of preferential flow, preferential flow pathway
depth, hydraulic conductivity of fractures, the volume of fractures, surface flow entering
the fiactures, hydraulic conductivity of the storage layer, hydraulic conductivity and soil-
water retention properties of the waste layer, and waste layer thickness were varied
throughout multiple simulations. The cumulative percolation of water across the storage
layer-waste layer boundary (bottom of cap) was used to assess the performance of the cap
versus a lysimeter. Based on the results of the simulations, the following conclusions
were drawn:

1. When preferential flow is absent, lysimeters over-predict percolation

compared to a cap. When preferential flow is present, lysimeters either over-

62

predict percolation or the percolation is about the same depending upon the
fiacture depth and hydrologic properties of the storage and waste layer.

. When preferential flow is considered, percolation in both the lysimeter and
cap increases.

. Waste layer thickness does not affect percolation when it is greater than the
thickness of the cap.

. Fracture hydraulic conductivity has minimal influence on percolation when it
is > 0.2 cm/s.

. As the depth of preferential flow paths is increased, the magnitude of
percolation increases.

. The relative volume of the fracture domain had little influence on percolation
when the fiaction of precipitation entering the fiacture network is not
proportionately changed.

. The hydraulic conductivity of the storage layer influences percolation only
when preferential flow is not present. As the storage layer hydraulic
conductivity increases, so does the percolation. When preferential flow in the
storage layer is considered, the preferential flow dominates and the hydraulic
conductivity of the matrix does not influence percolation.

. The hydraulic conductivity of the waste layer influences percolation when
only capillary flow is considered. As the waste layer hydraulic conductivity
increases, percolation decreases. When preferential flow is considered, it
dominates water flow and the hydraulic conductivity of the waste does not

influence percolation.

63

9. The SWCC and hydraulic conductivities of the waste greatly influences
percolation and hence the relative percolation between a cap and an

equivalent lysimeter.

The key conclusion of this study is that preferential flow can significantly increase
percolation through the cap into the waste layer of a landfill. Ignoring preferential flow,
whether it is currently present or will be in the future, can be detrimental when designing
and evaluating landfill final covers. HYDRUS was able to simulate percolation trends
through an ET cover but with some numerical challenges. The dual-permeability model
was often unstable. Simulations ofien crashed and the number of vertical mesh nodes had
to be decreased making the mesh coarse which produces less accuracy in the results.

Simulations that would run in HYDRUS-2D would not run in HYDRUS-1D unless the

water and pressure tolerances were changed. In HYDRUS-1D, qmp was a parameter that
could be changed in the input menu but all simulations crashed when qTop was anything

but 0.5. (“‘01) is “hard-coded” in HYDRUS-2D as 0.5 and hence could only be changed

from the “back-end” of the input.

With a program that crashes frequently at what appears to be a random manner, it is
difficult to have confidence in the results. When a simulation does crash, there is no
indication as to why and the user must blindly change the tolerances, iteration number,
and number of mesh nodes in an attempt to make the simulation run. In order for dual-
perrneability models to become a reliable method of predicting percolation through a

landfill cover, numerically stable models need to be developed.

REFERENCES

Albright, W.H., Benson, C.H., Gee, G.W., Roesler, A.C., Abichou, T., Apiwantragoon,
P., Lyles, BF, and Rock, SA. (2004). “Field water balance of landfill final covers.”
J. Environ. Qual., 33, 2317-2332.

Benson, CH, and Wang, X. (1998) “Soil water characteristic curves for solid waste.”
Environmental Geotechnics Report 98-13, Dept. of Civil and Env. Eng., U of
Wisconsin-Madison.

Benson, C.H., Bohnhoff, G.L., Ogorzalek, A.S., Shackelford, C.D., Apiwantragoon, P.,
and Albright, W.H. (2005) “Field data and model predictions for a monolithic
alternative cover.” Waste Containment and Remediation, ASCE Geotechnical Special
Publication No. 47.

Carneira, M.R., Ahuja, L., Fernando, R.M. and Pereira, LS. (2000). “Evaluating field
measured soil hydraulic properties in water transport simulations using the
RZWQM.” J. of Hydrology, 236, 78-90.

Daniel B. Stephens & Associates, Inc. (2006) “Laboratory report for Waste Management,
Inc. (Woodland Meadows samples)” Daniel B. Stephens & Associates, Inc.,
Albuquerque, New Mexico.

Gardenas, A., Sirni’rnek, J., Jarvis, N., and van Genuchten, M.Th. (2006) “Two-
dimensional modeling of preferential water flow and pesticide transport from a tile-
drained field.” J. of Hydrology, 329, 647-660.

Gerke, H.H., Kchne, J .M. (2004) “Dual-permeability modeling of preferential bromide
leaching from a tile-drained glacial till agricultural field.” J. of Hydrology, 289, 239-
257.

Hauser, V.L., P.E., Weand, BL. and Gill, M.D., RE. (2001). “Natural covers for landfills
and buried waste.” J. of Env. Eng, Vol. 127, No. 9, 768-775.

Jacobson, J.J., Heydt, H., Piet, S., Sehlke, G., Soto, R., and Visser, J. (2005). “Dynamic
modeling of an evapotranspiration cover.” Practice Periodical of Hazardous, Toxic,
and Radioactive Waste management, Vol. 9, No. 4, 223-236.

Jarvis, NJ. (1998) “Modeling the impact of preferential flow on nonpoint source
pollution.” In: Selim, H.M., Ma, L. (Eds.), Physical Nonequilibrium in Soils:
Modeling and Application, 195-221.

Khire, M.V., Benson, CH, and Bosscher, P.J. (1997). “Water balance modeling of
earthen final covers.” J. of GeoTech. and GeoEnv. Eng, Vol. 123, No. 8, 744-754.

65

Khire, M.V., Benson, CH, and Bosscher, P.J. (1999). “Field data from a capillary barrier
and model predictions with UNSAT-H.” J. of GeoTech. and GeoEnv. Eng, Vol. 125,
No. 6, 518-527.

Khire, M.V., Benson, CH, and Bosscher, P.J. (2000). “Capillary barriers: design
variables and water balance.” J. of GeoTech. and GeoEnv. Eng, Vol. 126, No. 8, 695-
708.

Khire, M.V. and Mijares, R.G. (2008). “Influence of the waste layer on percolation
estimates for earthen caps located in a sub-humid climate.” Waste Management and
Remediation, ASCE GeoCongress 2008, New Orleans.

Kodesova, R., KozaK, J., Simunek, J., and Vacek, o. (2005) “Single and dual-
perrneability models of chlorotoluron transport in the soil profile.” Plant and Soil
Env., 51, 7, 310-315.

Morris, CE, and Stormont, J.C., Members, ASCE (1997). “Capillary barriers and
subtitle D covers: estimating equivalency.” J. of Env. Eng, Vol. 123, No. 1, 3-10.

Novak, V., Simunek, J ., and van Genuchten, M. Th. (2000). “Infiltration of water into
soil with cracks.” J. or Irrigation and Drainage Eng, Vol. 126, No. 1, 41-47.

Ogor'lalek, A.S., Bohnhoff, G.L., Shackelford, C.D., Benson, CH, and Apiwantragoon,
P. (2008) “Comparison of field data and water-balance predictions for a capillary
barrier cover.” J. of GeoTech. and GeoEnv. Eng, Vol. 134, No. 4, 470-486.

Sadek, S., Gharrimeh, S., and El-Fadel, M. (2007) “Predicted performance of clay-barrier
landfill covers in arid and semi-arid environments.” Waste Management, 27, 572-583.

Scanlon, B.R., Reedy, R.C., Keese, K.E and Dwyer, SF. (2005). “Evaluation of
evapotranspirative covers for waste containment in arid and semiarid regions in the
southwestern USA.” Vadose Zone Journal, 4, 55-71 .

Simunek, J., Sejna, M., and van Genuchten, M.Th. (1999) “The HYDRUS-2D software
package for simulating two-dimensional movement of water, heat, and multiple
solutes in variably saturated media.” Version 2.0, IGWMC-TPS-53, International
Ground Water Modeling Center, Colorado School of Mines, Golden, Colorado.

Simunek, J ., Jarvis, N.J., van Genuchten, M.Th. and Gardenéis, A. (2003) “Review and
comparison of models for describing non-equilibrium and preferential flow and
transport in the vadose zone.” J. of Hydrology, 272, 14-35.

Simunek, J., Sejna, M., Saito, H., Sakai, M., and van Genuchten, M.Th. (2008a) “The

HYDRUS-1D software package for simulating the one—dimensional movement of
water, heat, and multiple solutes in variably-saturated media.” Version 4.0,

66

Department of Environmental Sciences, University of California Riverside,
Riverside, California.

Simunek, J ., van Genuchten, M.Th., and Sejna, M. (2008) “Development and
applications of the HYDRUS and STANMOD software packages and related codes.”
Vadose Zone J., Vol 7, 587-600.

US. DOE (U. S. Department of Energy), Office of Environmental Management and the
Office of Science and Technology (2000). “Alternative Landfill Cover.” Innovative
Technology Summary Report.

USEPA (U .8. Environmental Protection Agency) (1997) Resource Conservation and
Recovery Act Information System (RCRAIS) database, EPA National Oversite
Database.

Vogel, T., Gerke, H.H, Zhang, R., and van Genuchten, M.Th. (2000) “Modeling flow and

transport in a two-dimensional dual-permeability system with spatially variable
hydraulic properties.” J. of Hydrology, 238, 78-89.

67