a numerical study of soil cover perfomance.yanful, mousavi and de souza.2006
TRANSCRIPT
-
7/21/2019 A Numerical Study of Soil Cover Perfomance.yanful, Mousavi and de Souza.2006
1/21
Journal of Environmental Management 81 (2006) 7292
A numerical study of soil cover performance
Ernest K. Yanful, S. Morteza Mousavi, Lin-Pei De Souza
Geotechnical Research Centre, Department of Civil and Environmental Engineering, The University of Western Ontario, London, ON, Canada N6A 5B9
Received 17 June 2003; received in revised form 8 October 2005; accepted 11 October 2005
Available online 23 March 2006
Abstract
In the investigation of soil cover design options for final decommissioning of reactive mine waste, it is often necessary to analyze or
predict the anticipated cover performance as a function of the cost of implementation, which is governed by the type, number andthickness of the layers in the cover system. An example of such investigation is presented in this study where one-dimensional
evaporation from hypothetical moisture-retaining cover systems is simulated to assess the influence of several cover properties and
hydrogeologic parameters on performance. The commercially available transient flow model, SoilCover, was used to compute suction
and water content profiles for different cover design scenarios. The predicted water content profile and porosity of layers were then used
to estimate the oxygen diffusion coefficients of the various layers. The oxygen diffusion coefficients were used to estimate oxygen flux
through the cover systems. The oxygen flux was, in turn, related to the maximum acid flux.
The studied cover and hydrogeologic parameters included soil type, thickness of barriers, and water table elevation. Two types of
infiltration and oxygen barrier and two types of capillary layer with different thicknesses were studied. The water table was either kept
constant at the base of the waste (tailings) or dropped by 0.5, 1, 2, and 3 m over 120 days. The results showed that the relationship
between water table depression and the thickness of capillary layers, on one hand, and desaturation of the infiltration and oxygen barrier,
on the other, is not linear. Relationships between oxygen flux and barrier thickness and between cost increase and performance
improvement of the studied cover systems are presented. Finally, a method that outlines steps for site-specific and economically feasible
design of multi-layer cover systems is introduced.r 2006 Elsevier Ltd. All rights reserved.
Keywords: Soil cover; Sand-bentonite; Optimum design; Infiltration barrier; Capillary layer; Oxygen diffusion; Water level; Environmental management
1. Introduction
1.1. Layered soil cover systems
The contaminated effluent generated when reactive
sulfide-bearing mine waste comes into contact with water
and oxygen is called acid rock drainage (ARD). Ifpreventative and control measures are not taken, ARD can
contaminate surface and ground water in communities
around a mine site. Limiting accessibility of water and
oxygen to the waste can reduce ARD production.
Covering waste (for example, tailings) with a soil layer
that has a low hydraulic conductivity limits accessibility of
water to the tailings. In addition, when the soil cover is
placed close to saturation and maintained at this high
water content, oxygen accessibility is also restricted. Thus,
a fine-grained soil cover with minimum hydraulic con-
ductivity but maximum degree of saturation is the best soil
cover to reduce oxygen ingress in a net-infiltration
environment.
The concept of soil cover is not new. Brown (1970),Nicholson et al. (1989),Aubertin et al. (1995a, b),Khire et
al. (1997),Chapuis (2002), andMbonimpa et al. (2003)are
among many researchers that have studied soil covers.
The challenge to implementing a soil cover above the
water table for a reactive mine waste is maintaining a high
degree of saturation in the cover over a long period. To do
this, both downward drainage of water from the cover
towards the waste and upward evaporation from the cover
towards the atmosphere, especially during dry periods,
must be prevented or at least minimized. Nicholson et al.
ARTICLE IN PRESS
www.elsevier.com/locate/jenvman
0301-4797/$- see front matter r 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jenvman.2005.10.006
Corresponding author.
E-mail address: [email protected] (E.K. Yanful).
http://www.elsevier.com/locate/jenvmanhttp://www.elsevier.com/locate/jenvman -
7/21/2019 A Numerical Study of Soil Cover Perfomance.yanful, Mousavi and de Souza.2006
2/21
(1991)suggested that the selection of cover materials must
be site specific and recommended further study of the
hydraulic processes.
Yanful (1993) observed that a moisture-retaining multi-
layer soil cover over acid-generating tailings was able to
reduce the oxygen flux by 98%. This cover consisted of a
middle layer of fine-grained material that served asmoisture-retaining barrier layer and two protective
coarse-grained layers at the top and bottom of the barrier.
In the present study, the barrier layer is called infiltration
and oxygen barrier, while the two protective coarse-
grained layers at the top and bottom of the infiltration and
oxygen barrier are referred to as capillary layers.
Placing coarse-grained soil layers, or capillary layers,
above and below the infiltration and oxygen barrier
generally helps preserve the degree of saturation of the
latter under a wide range of climatic conditions (Rasmuson
and Eriksson, 1986;Collin, 1987;Yanful and Aube , 1993;
Benson et al., 1994; Shackelford et al., 1994; Choo and
Yanful, 2000). The role of capillary layers will be discussed
further in Section 2.1.
Layered soil covers have also been studied in a number
of field investigations in the last 15 years. Two- and three-
layer soil covers were constructed over acid-generating
waste rock at Rum Jungle Mine in Australia (Harries and
Ritchie, 1987), Heath Steele Mine site in New Brunswick
(Yanful et al., 1993a, b), Equity Silver Mine site in British
Columbia (OKane et al., 1998) and Bersbo mine in
Sweden (Lundgren, 1997). Test plots of a three-layer soil
cover were constructed on acid-generating tailings at the
decommissioned Waite Amulet site near Rouyn-Noranda,
Que bec in 1990 and monitored over a 3-year period(Yanful and St-Arnaud, 1991; Yanful et al., 1994). The
three layers consisted of an uppermost fine sand layer
underlain sequentially by a compacted silty clay and a
coarse sand. A blanket of gravel was placed over the fine
sand to prevent erosion. Monitoring data indicated that the
silty clay (infiltration and oxygen barrier), compacted close
to 2% wet of the optimum water content, maintained its
placement degree of saturation of approximately 95% and
hydraulic conductivity of 1 107 cm/s during the 3
years. As a result, the cover reduced oxygen diffusion and
water percolation significantly and allowed only 4% of
precipitation to percolate into the underlying tailings
(Woyshner and Yanful, 1995).
In all the soil cover systems described above, the nearly
saturated infiltration and oxygen barrier with low perme-
ability performed well because of the presence of protective
coarse-grained layers, which functioned as capillary break
layers. Other mine waste cover systems that do not use the
capillary layer concept have also been investigated. For
example, OKane et al. (2000), Wels et al. (2002), and
OKane and Waters (2003) have demonstrated the perfor-
mance of a store and release cover for reducing acid
generation in arid climates.
In spite of the good performance reported in the above
cases, soil covers have not been widely used in mine waste
decommissioning projects in temperate climates. Two main
reasons account for this trend. First, the potentially high
upfront cost of a soil cover: for example, for a 20-ha
tailings site, the cost of installation of a three-layer soil
cover is estimated to be $100,000$250,000/ha (site
specific), while the cost of implementation of a water cover
for the same site is $120,000/ha. Although these estimatesare for construction costs only and do not consider site-
specific details and maintenance, they show that, generally,
the upfront cost of implementation of a soil cover is more
than that of a water cover. The second disadvantage of soil
cover is uncertainty about long-term performance. Very
little information is available on long-term performance of
soil covers.
Significant long-term maintenance costs and detrimental
environmental impacts may result if a soil cover is not
properly designed and constructed. The primary motiva-
tion for the present work was to examine how an optimized
soil cover design can improve performance and lead to
possible cost reductions.
1.2. The present study
This numerical study attempts to correlate cover
performance with cost to assist the designer in selecting
case- and project-dependent optimum designs. Also, the
study examines covers under critical long-term scenarios
where the covers are allowed to evaporate during a 4-
month period with no rain. This is likely the worst-case
scenario and is similar to the study period used bySwanson
et al. (2003), who examined soil cover performance over a
153-day period that included an extreme dry summer. Theresults can help achieve cost-effective designs and signifi-
cantly reduce the number of scenarios that must be
examined in a pilot or test plot study. In decommissioning
projects where time is a limitation and test plots cannot be
constructed and monitored over a long period of time prior
to implementation, the method described in the present
study can be used along with sound engineering judgment
to select the final cover.
A number of researchers have highlighted the important
role of numerical modeling in the analysis of layered soil
covers. McMullen et al. (1997) observed that because the
hydraulic behavior of a layered cover system is complex,
computerized calculations based on numerical models may
be used to establish optimum configurations. Bussie` re et al.
(1995) analyzed covers with different material types, and
showed how numerical methods can be used to evaluate the
performance of capillary layers. Mbonimpa et al. (2003)
elucidated the benefits of using numerical simulations for
predicting soil cover performance. Also, Swanson et al.
(2003)reported the results of similar simulations to predict
the performance of the soil cover system at Equity Silver
Mine, British Columbia, Canada.Mbonimpa et al. (2003)
presented sample calculations of oxygen flux in cover
systems and suggested that changing key cover properties
and design parameters, such as the degree of saturation of
ARTICLE IN PRESS
E.K. Yanful et al. / Journal of Environmental Management 81 (2006) 7292 73
-
7/21/2019 A Numerical Study of Soil Cover Perfomance.yanful, Mousavi and de Souza.2006
3/21
the barrier and thickness of the various layers, could be an
effective way of predicting cover performance.
In the present work, the effect of material type and
thickness of capillary layers and infiltration and oxygen
barrier were studied during a 4-month period. The study
involved the use of the flux boundary model SoilCover
(Geo-Analysis 2000 Ltd., 2000) and appropriate materialproperties to simulate several scenarios.
The inclusion of evaporation is one important aspect
that distinguishes the present study from previous studies
(e.g.Shackelford et al., 1994;Bussie` re et al., 1995;Aubertin
et al., 1996). The present work is also different from
previous studies in the sense that it considers the key
aspects of cover design and analysis (material selection,
flow modeling, oxygen transport modeling, and cost
analysis). This approach provides an integrated picture
that allows the designer or proponent to eliminate less cost
effective options without recourse to costly long-term
laboratory and test plot studies. The following section
highlights the key considerations in the study.
2. Key aspects of cover design
2.1. Considered mechanism
Using a transient numerical model, Nicholson et al.
(1991) showed that in a layered soil cover static non-
equilibrium conditions would prevail in the coarse-
grained lower capillary layer for prolonged periods of
time, such that an overlying fine-grained infiltration and
oxygen barrier would not drain. In the numerical analyses
presented in this study, transient evaporation was con-sidered as top boundary condition at the soil surface and
static non-equilibrium condition was assumed in the
coarse-grained lower capillary layer, in accordance with the
observations of Nicholson et al. (1991) andBarbour and
Yanful (1994).
2.2. Importance of evaporation
Isothermal models may be used to predict responses of a
soil cover system over the short term; however, for long-
term analyses under extended drying conditions, coupled
heat and water transport soilatmosphere modeling should
be used to analyze soil covers (Swanson et al., 2003).
Water content within a cover system is not only a
function of the soil water characteristic curves of the cover
materials but is also influenced by the water flux that
develops through the cover due to climatic factors. Water
flux across the cover system will develop as a result of
precipitation or evaporation, and the pressure within the
cover will respond to accommodate these fluxes. Therefore,
if this point is neglected, the pressure profiles and the
resulting water contents within the cover system may
deviate significantly from those postulated. Consequently,
the adequacy of the selected materials can be fully assumed
only when their soilwater characteristic curves and their
hydraulic conductivitysuction functions are considered in
the light of surface infiltration and evaporation (Barbour,
1990).
Although Nicholson et al. (1991) did not study
evaporation, they recognized that evaporation from the
surface of a cover system was the primary process of
moisture loss from the cover. They recommended furtherinvestigation of the hydraulic concepts articulated in their
paper to include more realistic scenarios, such as the effect
of evaporation. Mbonimpa et al. (2003) confirmed that
layered soil covers with capillary layers could be employed
to reduce vertical percolation of water into the underlying
waste.
Choo and Yanful (2000) demonstrated that vapor
transport exceeded liquid water flow a few days after
drying.Swanson et al. (2003)performed a soilatmosphere
modeling of an engineered soil cover on acid-generating
mine waste in a humid alpine climate. They confirmed that
vapor flow was the dominant flow mechanism near the
surface of the cover a few days after the start of drying.
Swanson et al. (2003) cautioned the danger of under-
estimating water loss from covers by ignoring evaporation.
These findings emphasize the need to examine coupled heat
and water transport in soil cover analysis and design.
Although evaporation was not incorporated in most of
the early studies on soil covers (includingAkindunni et al.,
1991; Nicholson et al., 1991; Shackelford et al., 1994;
Bussie` re et al., 1995; McMullen et al., 1997; Mbonimpa
et al., 2003) those studies all recognized its importance. In
the present work, the movement of liquid water and water
vapor due to evaporation within a soil cover is included in
the analysis.
2.3. Estimating oxygen-diffusive flux
A knowledge of soil moisture content and hence oxygen
flux is necessary for determining the final (optimal)
configuration of a soil cover (Mbonimpa et al., 2003). In
fact, the production of acid in sulfidic mine waste is
controlled by the availability of oxygen at the waste
surface. The oxygen flux can be related to the theoretical
maximum acid flux on the basis of the stoichiometry of the
overall sulfide oxidation reaction (Yanful, 1993):
FeS2s154
O2g 72
H2Ol ! FeOH3s 2H2SO4aq.
(1)
Eq. (1) indicates that 3.75 mol of gaseous oxygen (O2)
oxidize 1 mol of pyrite (FeS2) in the presence of water
(H2O) to produce 2 mol of sulfuric acid (H2SO4). Thus, the
oxygen flux may be correlated to the acid flux to give an
indication of cover performance (Yanful, 1993).
The primary mode of oxygen transport in tailings is
diffusion through pore spaces. In unsaturated fine-grained
soils, such as the soils used in infiltration and oxygen
barriers (i.e., silts and clays) oxygen transport is generally
controlled by molecular diffusion (Collin, 1987;Collin and
ARTICLE IN PRESS
E.K. Yanful et al. / Journal of Environmental Management 81 (2006) 729274
-
7/21/2019 A Numerical Study of Soil Cover Perfomance.yanful, Mousavi and de Souza.2006
4/21
Rasmuson, 1988; Mbonimpa et al., 2003). In porous
geologic materials the most important factor that controls
the magnitude of the oxygen diffusion coefficient is the
degree of saturation (Nicholson et al., 1991). Modeling the
moisture flow in a cover system is a necessary step to
determining the oxygen flux through the cover (Swanson et
al., 2003). Due to the relatively low solubility of oxygen inwater and the four orders of magnitude decrease in oxygen
diffusion coefficient from air to water, soil covers at or near
water saturation exhibit reduced oxygen fluxes (Nicholson
et al., 1989). This led to the common practice of using soil
materials with low oxygen diffusivities to reduce oxygen
transfer from the atmosphere to reactive sulfide mine
waste. Several studies (for example, Yanful, 1993; Yanful
et al., 1993b;Yanful et al., 1999) have confirmed that the
reduced oxygen transfer leads to a decrease in oxidation
and acid generation. Measurement of the effective diffusion
coefficient of oxygen is required to evaluate the flux for
optimum design of soil cover systems (Mbonimpa et al.,
2003).
In the present study, modeled moisture profiles and
measured or estimated porosity of various soil layers in a
cover system were used to calculate oxygen diffusion
coefficients using the method of Millington and Shearer
(1971). Collin and Rasmuson (1988) indicated that the
Millington and Shearer (1971) equations are reasonably
accurate over a wide range of water contents, from dryness
to saturation. The resulting oxygen diffusion coefficient
and selected thickness of the various layers were then used
to estimate oxygen flux across the cover system. The
relative efficiency of a cover can be evaluated by the total
oxygen flux through the cover, which governs the amountof acid generation.
2.4. Cost analysis
Unit cost assumptions for geologic materials and earth-
works (Rowe, 1993) were used to estimate the cost of the
considered soil cover scenarios. These assumptions were
based on 1993 costs; however, prices and final cost may be
easily adjusted for inflation to bring them to 2005 figures.
Because the analysis was based on relative cost estimate
and the results are presented in percent, this should not
affect the conclusion on cover performance. The unit costs
do not take into account minor components and unknown
variables, which generally cannot be accurately known
until at the stage of detailed design. The main considera-
tions in the cost computations are the type and thickness of
the various soil layers present in the cover. For illustration,
cost increase and performance improvement versus thick-
ness of capillary layers and infiltration and oxygen barriers
for a few cases are presented. Correlations between cost
increase and performance improvement are also high-
lighted.
Optimum cover design has been an issue of importance
to many engineers and researchers including Bussie` re et al.
(1995)andMbonimpa et al. (2003).McMullen et al. (1997)
noted that as the hydraulic behavior of a layered cover
system is fairly complex, computerized numerical simula-
tions might be used to establish the optimum cover
configuration. However, it is obvious that there cannot
be a unique optimum design for every soil cover applica-
tion. Selecting an optimum design is definitely a site-
specific and project-specific issue and case-dependentprocess that hinges on several factors including
(i) site hydrology and hydrogeology, climate, and materi-
al availability and type;
(ii) intended short- and long-term functions of the cover;
(iii) construction constraints and challenges;
(iv) available project budget.
In general, one operator may want to prevent ARD
production, while another operator may find that imple-
menting a cover that is less than 100% effective and
combining it with collection and treatment of the resulting
residual ARD may be more economical than trying to
achieve complete ARD elimination.
Construction challenge is another factor that can affect
the implementation of a cover system. For example,
working on a tailings surface with heavy construction
equipment can be difficult if not impossible. Trafficability
is an issue because high capillarity can result in an elevated
water table in the tailings, which can complicate the
placement of relatively thin cover layers with adequate
compaction and homogeneity. Thus, material placement
should be an important consideration in all design options
(McMullen et al., 1997). In some cases, winter construction
has been adopted to ensure that the tailings surface issufficiently frozen to support equipment.
Also, since homogeneity and free-drainage of the top
capillary layer are important for it to function as a good
capillary break, it is necessary to avoid pumping of the
underlying tailings into the overlying sand or gravel
capillary layer during placement and compaction. To
achieve this, the top capillary layer should be placed
during the coldest winter months when the tailings surface
is frozen. Available budget also has a critical role in the
selection of an optimum cover design. The proportion of
initial investment and maintenance cost to total cost is
dependent on many factors including the expected perfor-
mance of the cover and the length of the project. Therefore,
the concept of an optimum soil cover design is case-
dependent
2.5. Proposed approach
The approach proposed in the present work is the
development of site- and project-specific optimum designs
that take the above factors into consideration. It empha-
sizes the role of material type, layer thickness, water level
location, and involves the quantitative analysis of the
correlation between cost and performance for a given soil
cover scenario. Such an approach should help the designer
ARTICLE IN PRESS
E.K. Yanful et al. / Journal of Environmental Management 81 (2006) 7292 75
-
7/21/2019 A Numerical Study of Soil Cover Perfomance.yanful, Mousavi and de Souza.2006
5/21
select an optimum case-specific design for a cover project.
Fig. 1 presents the conceptual flowchart of the computa-
tional sequence used in the study.
A necessary requirement for maintaining a soil cover in a
high saturated state after prolonged drainage is that the
magnitude of the air-entry value (AEV) of infiltration and
oxygen barrier be greater than the sum of the thickness ofthe infiltration and oxygen barrier and the absolute value
of the suction at which the underlying coarse layer
approaches the residual moisture content (Akindunni
et al., 1991). The AEV is the suction required to overcome
the capillary forces exerted by the largest pore and initiate
drainage in the medium (Nicholson et al., 1989). Similarly,
the AEV of the infiltration and oxygen barrier should be
greater than the absolute value of the suction at which the
overlying coarse layer approaches the residual moisture
content. The thickness of the lower capillary layer should
be greater than the absolute value of its AEV to ensure that
drainage occurs to reach residual saturation (maximum
suction) and hence minimum hydraulic conductivity in the
lower capillary barrier, which would minimize water loss
from the infiltration and oxygen barrier (Bussie` re et al.,
1995).
A reliable and technically feasible cover design must
meet the above criteria. The design process for a particular
cover alternative would normally start with a considerationof site specifications and material type and availability. It
should also consider expected cover function, construction
challenges, and available budget. The design alternative is
then checked against the above criteria. The data needed
for these criteria include: AEV of infiltration and oxygen
barrier, thickness, hr (residual saturation head) of upper
capillary layer, and finally hr and AEV of lower capillary
layer. Although it is possible to obtain these parameters
from laboratory tests, it is useful to obtain first approx-
imation estimates from published empirical solutions.
Lambe and Whitman (1969) presented a range of
capillary heads data for soils draining from the bottom.
ARTICLE IN PRESS
NO YES
* Air entry value ** residual saturation head
Preliminary cost estimation
Transient flow numerical analysis using model and input data
Site-specific data (hydrology, hydrogeology, etc.)
Material type and availability
Expected cover function
Construction constraints and challenges
Available budget
Schematics of preliminary soil cover design
Profiles of: Suction, saturation degree,
volumetric water content
Diffusion modeling using volumetric water
content and porosity as input
Oxygen diffusion coefficient (De) of layers
Calculation of oxygen flux using De and
thickness of layers
Performance-Cost correlation analysis for soil cover system
Site specific Optimum soil cover design
Is thickness of infilt. & O 2barr.its| |AEV ?
Fig. 1. Conceptual flowchart of computational sequence.
E.K. Yanful et al. / Journal of Environmental Management 81 (2006) 729276
-
7/21/2019 A Numerical Study of Soil Cover Perfomance.yanful, Mousavi and de Souza.2006
6/21
Nicholson et al. (1991) plotted the data on log-linear axes
and obtained the following relation:
Log d10 0:01AEV 0:421, (2)
whered10is the particle size (in mm) than which 10% of the
soil is finer, and AEV is the air entry value of the material
in cm of water. Nicholson et al. (1991) suggested thatEq. (2) may be used to estimate AEV of an infiltration and
oxygen barrier. However, it should be pointed out that the
data presented by Lambe and Whitman (1969) was for
cohesionless soils, and therefore the equation may not be
suitable for cohesive soils. Therefore, in the present study it
is suggested that Eq. (2) be used for cohesionless soils and
laboratory test be used for cohesive soils.
Lambe and Whitman (1969)also presented data on the
maximum height of rise of water in cohesionless soils
subjected to downward vertical drainage. This rise is
similar to residual water height (hr) in an upper capillary
barrier in a three-layer cover consisting of a middle lowpermeability layer. If the data are plotted on log-linear
axes, the log (d10) value (d10 in mm) versus the hr (cm
water) exhibits a linear trend over the range of sandy gravel
to silt (Fig. 2). Therefore, the d10 value presents an easily
measured parameter to provide an initial estimate of the hrof cohesionless materials (such as sand) used in upper
capillary layer.
The design steps integrating the above requirements for
the capillary barriers and infiltration and oxygen barrier in
a multi-layer cover system are outlined inFig. 1. The use of
Fig. 2and Eq. (2) at the first stage of the design process can
reduce the amount of subsequent laboratory testing. If the
answer to the three questions in the flowchart ofFig. 1isYES, then it is recommended that laboratory tests (soil
water characteristics tests) be run to obtain more reliable
values of AEV and hr.
The numerical analyses were performed using a satur-
atedunsaturated evaporation flux model, say SoilCover
(Geo-Analysis 2000 Ltd., 2000). Typical input include
climatic, hydrological and geotechnical data. Profiles of
suction, degree of saturation and volumetric water content
resulting from the simulation are used along with porosities
of the layers to calculate oxygen diffusion coefficients (De),
using the method of Millington and Shearer (1971). Next,
the oxygen flux through the cover system is estimated usinga steady-state oxygen flux model, De and thickness of
layers as input data. Finally, an analysis of cover
performancecost relation, similar to what is performed
in the present study for the considered cases, can be
performed to identify the optimum site- and project-
specific design. The flowchart in Fig. 1 summarizes these
steps.
3. Methods and materials
3.1. Numerical model
In a previous study,Yanful et al. (2002)presented results
that showed good agreement between experimental eva-
poration data and numerical simulations obtained using
the soilatmosphere model SoilCover (Geo-Analysis 2000
Ltd., 2000). AlsoSwanson et al. (2003)assessed the ability
of the same model to simulate field conditions and found
reasonable agreement between the model and field results.
Therefore, it was decided in the present study that
SoilCover could be used to predict evaporation and suction
and water content profiles in multi-layer soil covers.
SoilCover is a one-dimensional finite element package
that models transient liquid and water vapor flow, based ona theoretical model for predicting the rate of evaporation
from a soil surface (Wilson et al., 1994). The model
framework is a system of equations describing coupled heat
and mass transfer in a soil. The flow of water vapor and
liquid water are described on the basis of Ficks law and
Darcys law, respectively. Fouriers law is used to describe
conductive heat flow in the soil profile below the soil/
atmosphere boundary. Temperature is evaluated on the
basis of conductive and latent heat transfers. The model
calculates the vapor pressure in the soil using the relation-
ship provided by Edlefsen and Anderson (1943), which
calculates vapor pressure from the total suction in the
liquid phase.
Atmospheric coupling is achieved by calculating the soil
evaporative flux from environmental input data and soil
hydraulic properties. Evaporation is calculated using a
modified Penman formulation (Wilson, 1990), and eva-
porative flux can be calculated from either a saturated or
an unsaturated soil surface. Other user-defined soil
parameters used in SoilCover include specific gravity,
porosity, and saturated hydraulic conductivity. The var-
ious features of the software and the governing equations
are described in detail in the user manual (Geo-Analysis
2000 Ltd., 2000). In the present study the software was
used to evaluate hypothetical multi-layer cover systems
ARTICLE IN PRESS
Residual saturated value (hr) (cm water)
0 50 100 150 200 250 300 350 400
d(10)particlesize
(mm)
0.001
0.01
0.1
1
log(d10)=.00479(AEV)-0.579
r 2=0.97
Fig. 2. Relationship betweend10 grain size and residual saturation head
(hr) for dry, cohesionless soils undergoing bottom drainage (Data from
Lambe and Whitman, 1969).
E.K. Yanful et al. / Journal of Environmental Management 81 (2006) 7292 77
-
7/21/2019 A Numerical Study of Soil Cover Perfomance.yanful, Mousavi and de Souza.2006
7/21
overlying mine waste (tailings). Evaporation was consid-
ered as top boundary condition at soil surface, and suction
as bottom boundary condition at the covertailings inter-
face. Temperature, relative humidity, radiation, and wind
speed were used as climatic input data. Other input soil
properties included saturated hydraulic conductivity, por-
osity, soil water characteristic curve and hydraulic con-ductivitysuction function. Profiles of suction, degree of
saturation and water content obtained as output were used
to calculate oxygen diffusion coefficient and then oxygen
flux and, finally, to evaluate the performance of the cover
systems.
3.2. Oxygen diffusion coefficient and flux calculation
procedure
Millington (1959) proposed the following equation for
diffusion of gases in non-aggregated porous media:
DD0
1 Sw2PT y2x, (3)
whereDis the diffusivity of a gas in the porous medium, D0is the diffusivity of the same gas in air, and D/D0 denotes
the normalized diffusivity. Also, Sw defines the degree of
saturation, y the volumetric water content, andPTdenotes
the total porosity of the medium, while x is obtained from
the following relation (Millington and Quirk, 1961):
P2xT 1 1 PTx (4)
i.e., the minimum or effective pore area, P2xT cm2 per cm2, is
associated with a maximum area occupied by solid,
(1PT)x
cm2
per cm2
.In the present study, Eqs. (3) and (4) were used to
calculate the diffusion coefficient of each soil layer in the
cover systems, following the method of Millington and
Shearer (1971). Diffusion coefficient and thickness of the
layers were then used to compute oxygen diffusion flux as
described in the sections that follow.
The computation of steady-state diffusion through a
multi-layered sequence is based on the steady-state diffu-
sion through a finite layer (Ficks first law):
F DeqC
qx, (5)
whereFis the mass flux [M/L2
T] and other parameters areas previously defined. The concentration profile through
the sequence of layers is calculated by equating the flux
from each consecutive pair of layers at the interface
between them. The flux can be equated at the interface for
any two adjacent layers with the resulting general equation:
Di
Li
Ci1
Di1
Li1
Di
Li
Ci
Di1
Li1
Ci1 0, (6)
where the subscript is notation for any layer. This provides
a system of equations (equal to the number of layers minus
one), which can be solved simultaneously. The boundaryconditions are represented by a constant concentration at
surface and a concentration of zero at the base of the
deepest layer.
The needed input data are the number of layers, the
thickness (m) and diffusion coefficient (m2/s) for each layer
starting at the top of the sequence. A computer program
based on the above equations was developed by Gillham
and Nicholson (1990) for analyzing transient and steady-
state diffusion through a single-layer cover, and steady-
state diffusion through a multi-layer cover system. This
program was adopted for the present study. The analysis
focused on steady-state oxygen flux that is more dealt with
in long-term analyses.
3.3. Modeled soil cover scenarios
The conceptual cover system used in the present study
typically consisted of an infiltration and oxygen barrier
placed between upper and lower capillary layers. The two
soil types used as infiltration barrier were sandbentonite
(Cases A1A5), and silt (Cases C1C10). Tailings and
Waite Amulet fine sand were used to texturally represent
mine waste, and Waite Amulet fine sand and coarse sand
were used as upper and lower capillary layers, respectively
(Choo and Yanful, 2000). Porosities of fine sand, sand
bentonite, coarse sand, silt, and tailings were 0.40, 0.39,
0.39, 0.45, and 0.45, respectively. Saturated hydraulic
conductivities, soilwater characteristic curves, and un-
saturated hydraulic conductivitysuction functions of the
studied soils are presented in Table 1 and Fig. 3. The
unsaturated hydraulic conductivitysuction functions for
all soils were determined using the method of van
Genuchten (1980). The properties of silty tailings given
by Gonzalez and Adams (1980) were used to represent
those of the mine waste and the alternate silt infiltration
and oxygen barrier.
A schematic of each modeled soil cover system and watertable location are shown in Fig. 4 and beyond. Plots of
suction and degree of saturation profiles are also presented
for some cases. The suction profile is presented to facilitate
the discussion of cover performance using the soilwater
ARTICLE IN PRESS
Table 1
Laboratory-measured saturated hydraulic conductivities of study soils
Soil type WA coarse sand WA fine sand sand-bentonite Tailings/silt
Ksat(cm/s) 5.3102 6.1103 5108 5.8106
Taken fromGonzalez and Adams (1980).
E.K. Yanful et al. / Journal of Environmental Management 81 (2006) 729278
-
7/21/2019 A Numerical Study of Soil Cover Perfomance.yanful, Mousavi and de Souza.2006
8/21
characteristic curve and the unsaturated hydraulic con-
ductivitysuction function. The volumetric water content
profiles and the porosities of the layers were used to
estimate oxygen diffusion coefficients using the procedure
proposed byMillington and Shearer (1971). The estimated
diffusion coefficients and the selected thickness of the
layers were then used to estimate oxygen flux through the
cover systems. This flux is a maximum value (worst-case
scenario) and in reality the flux would be less because of the
lower flux during transient state. The results are presented
in terms of cases related to the type of infiltration barrier
(sand-bentonite, or silt) implemented in the cover system. It
should also be noted, however, that oxygen fluxes
calculated using the average Sr are only approximate
and, in some cases, may not necessarily represent the actual
flux through the cover.
3.4. Initial and boundary conditions
A perched water table condition was simulated in the
uppermost fine sand to ensure that it was initially
saturated. As noted by Gardner and Fireman (1958),
Wilson et al. (1997) andYang and Yanful (2001), a high
rate of evaporation cannot exist when the water table is
deep. Hence for the cases where the water table was deep,
the uppermost fine sand layer had to be initially saturated
to promote evaporation. This would represent the situation
where a brief rainfall period follows prolonged drying that
leads to water table depression.
Environmental fluctuations such as infiltration andevaporation influence the performance of a cover. How-
ever, infiltration would generally contribute to maintaining
a high degree of saturation in the infiltration and oxygen
barrier and may not be a critical performance factor for
moisture retaining covers. Thus, evaporation, and not
water infiltration, was applied as a top boundary condition.
This condition would be analogous to the most critical field
situation where, following a period of precipitation,
evaporation occurs on a soil surface for a long time before
the next rain. Air temperature of 1825 1C, relative
humidity of 1651%, and potential (pan) evaporation rate
of 18.3 mm/day were used as climatic input data. This
selected potential evaporation was a conservative value
derived from previous laboratory column studies (Choo
and Yanful, 2000).
Both constant and changing water table conditions were
considered for the bottom boundary conditions during the
simulations. Therefore, the bottom boundary conditions
included either a constant or time-dependent pressure
head. The datum was located at the base of the tailings in
Cases A and at the interface of the tailings and the cover
system in Cases C. In some of the modeled cases, the water
table was placed at the surface of the tailings, that is, at the
interface of soil cover and tailings. In these cases, the water
table was gradually lowered to 1 m below the surface of thetailings during the simulation period. This drop in water
table was found to be typical of some tailings deposits in
Canadian temperate climates, such as the Waite Amulet
site (Yanful et al., 1990;Woyshner and Yanful, 1993). To
evaluate the effects of different water table conditions,
water table drops of 0.5, 1, 2, 3, and 4 m below the surface
of the tailings were also examined. Other cases involving
constant water level during the analysis period were also
investigated.
4. Analysis of soil cover scenarios
4.1. Sand-bentonite as infiltration barrier
Sand-bentonite was used as an infiltration barrier
because of its generally low hydraulic conductivity, ability
to maintain a high degree of saturation (Yanful and
Shikatani, 1993) and apparent resistance to the influence of
freezing and thawing (Wong and Haug, 1991; Chapuis,
2002). Furthermore, the soilwater characteristic curve
(Fig. 3) shows that sand-bentonite can maintain signifi-
cantly high water content even at high suctions. Thus this
type of material may be used as infiltration and oxygen
barrier when the water table is deep.
ARTICLE IN PRESS
Suction (kPa)
0.1 1 10 100 1000
Volumetricwatercontent
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40
0.45
0.50
WA coarse sand
WA fine sand
tailings and silt
sand bentonite
0.1 1 10 100
Hydraulicconductivity(m
/s)
10-2010-1910-1810-1710-1610-1510-1410-1310-1210-1110-1010-910-810-710-610-510-4
WA coarse sand
WA fine sand
tailings
sand bentonite
Suction (kPa)
Fig. 3. Soilwater characteristic curves and suctionhydraulic conductiv-
ity functions for the soils used in soil cover scenarios.
E.K. Yanful et al. / Journal of Environmental Management 81 (2006) 7292 79
-
7/21/2019 A Numerical Study of Soil Cover Perfomance.yanful, Mousavi and de Souza.2006
9/21
In Cases A1A5 a 60-cm thick sand-bentonite layer was
used as infiltration barrier. In Cases A1A3, the water
table, initially located at the surface of the tailings, was
decreased to 1 m below the surface of the tailings over 120
days. However, in Cases A4 and A5, the water table was
kept constant at 1 m below the surface of the tailings at all
times.
4.1.1. Cases A1 and A2: decreasing water level
4.1.1.1. Influence of lower capillary layer. In Case A1 the
soil cover scenario consisted of 30 cm of Waite Amulet fine
sand overlying 60 cm of sand-bentonite, which in turn
overlaid 30 cm of Waite Amulet coarse sand (Fig. 4). The
thickness of the sand layers (capillary layers) was based on
field design data presented by Yanful and St-Arnaud
(1991). The 60 cm thickness of sand-bentonite (infiltration
and oxygen barrier) was selected on the basis of both field
experiment and diffusive flux calculations performed by
Yanful (1993), which showed that this thickness reduced
the oxygen flux substantially. The suction profile in Case
A1 (Fig. 4) shows steadily increasing suction in the lower
capillary layer (coarse sand) over time, in accordance with
the gradual drop in the water level until the suction at the
residual water content is attained. Due to the low AEV of
the coarse sand (Fig. 3), it loses water rapidly by drainage,
resulting in a decrease in hydraulic conductivity and overall
flux of water and, hence, a high degree of saturation in the
infiltration barrier.
In Case A2 (Fig. 4) the Waite Amulet coarse sand layer
was removed from the cover system studied in Case A1, in
order to evaluate the feasibility of reducing costs by
removal of the lower capillary layer. Comparison of
suction profiles for Cases A1 and A2 in Fig. 4shows that
the suction at the base of the sand-bentonite layer increased
by 39%, from 6.9 kPa for Case A1 to 9.6 kPa for Case A2.
When a lower capillary layer is used, its steep hydraulic
conductivitysuction function results in a pressure profile
that starts at zero at the water table, and decreases to
negative pressure until near the residual suction value.
4.1.1.2. Influence of upper capillary layer. To investigate
the influence of the upper capillary layer in the perfor-
mance of a cover system with a 60-cm thick sand-bentonite
infiltration barrier, when the water level drops 1 m from the
ARTICLE IN PRESS
initial water level
tailings
Case A1
WA coarse sand
sand-bentonite
WA fine sand30 cm
60 cm
30 cm
100 cm
final water level
Case A1-Suction (kPa)
-10 -5 0 5 10 15
Elevation(m)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
2.2
day 1
day 15
day 30
day 60
day 90
day 120
Case A2-Suction (kPa)
-10 -5 0 5 10 15
Elevation(m)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
initial water level
60 cm
30 cm
100 cm
final water level
tailings
sand-bentonite
WA fine sand
Case A2
Fig. 4. Schematic diagrams of soil cover Cases A1 and A2 and profiles of suction.
E.K. Yanful et al. / Journal of Environmental Management 81 (2006) 729280
-
7/21/2019 A Numerical Study of Soil Cover Perfomance.yanful, Mousavi and de Souza.2006
10/21
top to the bottom of the tailings during the simulation
period, covers with 0, 15, 30 and 50cm thick upper
capillary layers were analyzed.Fig. 5shows that at the end
of 120 days, the average volumetric water content of the
infiltration barrier increased from 0.335 for a cover system
without upper capillary layer, to 0.369, 0.386, and 0.387 for
cover systems with 15, 30 and 50-cm-thick upper capillarylayers respectively. Accordingly, the average oxygen diffu-
sion coefficient of the infiltration and oxygen barrier also
decreased from 1.2 108 m2/s to 6.71010, 3.4 1011,
and 3.0 1011 m2/s respectively. Fig. 5 illustrates the
influence of the upper capillary layer, and also presents a
non-linear relation between the thickness of the upper
capillary layer and the performance of the soil cover system
in terms of the volumetric water content of the infiltration
and oxygen barrier.
Fig. 6shows that including a 15-cm-thick upper capillary
layer in a cover system increases cost by as much as 25%
but improved performance by only 3.7%. This apparently
marginal improvement in performance is an artifact of the
modeling boundary condition used for the case without a
top capillary barrier. In order to maintain the sand-
bentonite layer at the same boundary condition in the two
scenarios (with and without upper capillary layer), a
fictitious thin (1 mm) upper capillary barrier with a perched
water table was assumed to exist at the top of the sand-
bentonite barrier. The modeling results suggest that this 1-
mm-thick saturated layer apparently protected the sand-
bentonite against desiccation within the modeling period,although it is inconceivable that this would happen in
practice.
Fig. 6 also shows that increasing the thickness of the
upper capillary layer to 30 cm increases cost by 50% and
improved performance by 38%, which appeared to be cost
effective. However, a greater increase in the thickness of
the upper capillary layer from 30 to 50 cm increased cost
from 50% to 83%, but improved performance only
marginally, from 38% to 49%. Thus the relation between
the thickness of the capillary layer and cover performance
was not linear.
4.1.2. Case A3: influence of grain-size contrast between
layers
Impedance to drainage in a multi-layer cover is due to
the contrast in the hydraulic properties of two soil layers in
contact with each other (Nicholson et al., 1991). As shown
inFig. 7, Case A3 was similar to Case A2, except that in
Case A3 the importance of grain-size contrast between the
sand-bentonite layer and the underlying layer was eval-
uated by replacing the silt tailings with sand tailings
possessing the same hydraulic properties as the Waite
Amulet fine sand.
Since the AEV of the fine sand tailings was approxi-
mately only 2 kPa (Fig. 3), the 1-m drop in the water tablethat induced a suction of 10 kPa at the base of the tailings
was enough to initiate drainage in the tailings. The degree
of saturation which remained nearly constant in the silty
tailings in Case A2 (Fig. 7) now decreased rapidly over time
in the fine sandy tailings in Case A3 (Fig. 7) and reached its
residual saturation value. However, even in this case the
infiltration barrier (sand bentonite) did not lose saturation
(Fig. 7) because the suction in the fine sand at residual
saturation was about 10 kPa, which was still much less than
the AEV (100 kPa) of the sand bentonite (Fig. 3).
4.1.3. Case A4: fixed water level
Selective layering can be used to maintain near saturated
conditions in covers regardless of the depth of the water
table (Nicholson et al., 1991). The soil cover configuration
for Case A4 (Fig. 8) is exactly the same as in Case A1
(Fig. 4). However, in Case A4, unlike the previous case, the
water table was kept at the base of the tailings during the
simulation period. As the induced suction at the base of the
coarse sand is higher than the suction at its residual
saturation, the coarse sand would drain to residual
saturation. However, in this case residual saturation was
reached throughout the entire coarse sand profile in Case
A4 (Fig. 8) instead of just in the lower part in Case A1
(Fig. 4). The results show that the maximum suction at the
ARTICLE IN PRESS
0 10 20 30 40 50 60
Volum
etricwatercontent(%)
0.33
0.34
0.35
0.36
0.37
0.38
0.39
Oxygen
diffusioncoefficient(m/s)
Volumetric water content
Oxygen diffusion coefficient
0.0
4.0x10 -9
6.0x10 -9
8.0x10-9
1.0x10-8
1.2x10-8
1.4x10-8
2.0x10 -9
Thickness of top capillary layer (cm)
Fig. 5. Volumetric water content and oxygen diffusion coefficient in
infiltration barrier versus thickness of top capillary layer.
0 10 20 30 40 50 60
Performanceimprovemen
t(%)
0
20
40
60
80
100
Costincrease(%)
0
20
40
60
80
100
Cost increase
Performance improvement
Thickness of top capillary layer (cm)
Fig. 6. Percentage of cost increase and performance improvement versus
thickness of top capillary layer in a soil cover with sand-bentonite
infiltration barrier and without bottom capillary layer.
E.K. Yanful et al. / Journal of Environmental Management 81 (2006) 7292 81
-
7/21/2019 A Numerical Study of Soil Cover Perfomance.yanful, Mousavi and de Souza.2006
11/21
top of the coarse sand which was 6.9 kPa in Case A1 (Fig.
4) increased to about 10 kPa (45% more) in Case A4 (Fig.
8). The degree of saturation profiles for both Cases A1 and
A4 presented inFig. 8show that the infiltration barrier was
nearly 100% saturated in both cases, because the suctions
of 6.910.7 kPa induced at the contact between the coarse
sand and the infiltration and oxygen barrier was still
smaller than its AEV value, which is approximately
100 kPa (Fig. 3).
4.1.4. Case A5: influence of upper capillary layer with fixed
water level
AsFig. 9shows, Case A5 was similar to Case A4 in the
sense that the elevation of the water table was kept
constant at the base of the tailings. However, the thickness
of the upper capillary layer (Waite Amulet fine sand) was
reduced from 30 to 15 cm.Table 2shows the influence of
the thickness of upper capillary layer on the performance
of a cover with a 60-cm-thick infiltration barrier and a 30-
cm-thick lower capillary layer when the water table was
kept at the bottom of the tailings. Increasing the thickness
of upper capillary layer from 0 to 15 cm increased the cost
of the cover system by 16.6% but improved the perfor-
mance by only 3.5%. However, while doubling the
thickness of the upper capillary layer (from 15 to 30 cm)
increased cost by only 33.3%, it resulted in 38% improve-
ment in performance, again confirming a non-linear
relationship between the thickness of capillary layer and
performance, even when the water level was constant.
Although doubling the thickness of the upper capillary
layer from 15 to 30 cm does not double the volumetric
water content of the infiltration barrier, it does improve
cover performance by more than 10 times. This is similar to
what was the case inFigs. 5 and 6. The reason probably is
that in a steady-state analysis of a layered soil cover, the
total oxygen flux is determined not only by the volumetric
water content of the infiltration barrier, but also by the
volumetric water content and oxygen diffusion coefficients
of other layers.
This emphasizes the importance of ensuring that
adequate thickness of the upper capillary layer is used,
taking into account the type of material used for the
infiltration and oxygen barrier (Fig. 10). Another reason
for using enough thickness for the top capillary layer is that
otherwise under dry conditions, cracks may form at the
surface of the infiltration and oxygen barrier, which may
facilitate evaporation, thereby resulting in desaturation of
the infiltration barrier to some extent.Swanson et al. (2003)
ARTICLE IN PRESS
Case A2-Degree of saturation (%)
0 20 40 60 80 100
Elevation(m)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
day 1
day 15
day 30
day 60
day 90
day 120
Case A3-Degree of saturation (%)
0 20 40 60 80 100
Elevation(m)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
initial water level
60 cm
30 cm
100 cm
final water level
tailings
sand-bentonite
WA fine sand
Case A2
Case A3
initial water level
60 cm
30 cm
100 cm
final water level
WA fine sand
sand-bentonite
WA fine sand
Fig. 7. Schematic diagrams of soil cover Cases A2 and A3 and profiles of degree of saturation.
E.K. Yanful et al. / Journal of Environmental Management 81 (2006) 729282
-
7/21/2019 A Numerical Study of Soil Cover Perfomance.yanful, Mousavi and de Souza.2006
12/21
reported differences between measured and predicted
cover responses and attributed them to desiccation
induced cracking at the cover surface. Most saturatedun-
saturated flow models, including that used by Swanson et
al. (2003) and the one used in the present study, consider
soil as a continuous material and ignore the presence of
any cracks.
4.2. Silt as infiltration barrier
Almost invariably, the selection of a soil type for a cover
system is governed by its availability at or near the project
site. At many Canadian mine sites where crystalline rocks
dominate, there is a limited quantity of fine grained, clay
type soil materials while silts may be present in sufficient
ARTICLE IN PRESS
Case A4-Suction (kPa)
-4 -2 0 2 4 6 8 10 12 14
Elevation(m)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
2.2
day 1
day 15
day 30
day 60
day 90
day 120
Case A4
tailings
WA coarse sand
sand-bentonite
WA fine sand30 cm
60 cm
30 cm
100 cm
constant water level
Case A1-Degree of saturation (%)
0 20 40 60 80 100
Elevation(m)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.82.0
2.2
Case A4-Degree of saturation (%)
0 20 40 60 80 100
Elevation(m)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.82.0
2.2
Fig. 8. Schematics and suction profile for Case A4 and profiles of degree of saturation for Cases A1 and A4.
tailings
WA coarse sand
sand-bentonite
WA fine sand15 cm
60 cm
30 cm
100 cm
constant water level
Case A5
0 5 10 15 20 25 30 35
Volumetricwatercontent(%)
31
32
33
34
35
36
37
38
39
Oxygendiffusioncoefficient(m2/s)
0
5e-9
1e-8
2e-8
2e-8
2e-8
3e-8
Volum. water content
Oxygen diffusion coeff.
Thickness of top capillary barrier (cm)
Fig. 9. Schematics of Case A5 and volumetric water content and oxygen diffusion coefficient versus thickness of top capillary layer.
E.K. Yanful et al. / Journal of Environmental Management 81 (2006) 7292 83
-
7/21/2019 A Numerical Study of Soil Cover Perfomance.yanful, Mousavi and de Souza.2006
13/21
quantities to merit consideration in cover projects (Nichol-son et al., 1989). Thus in a number of cases, it may be
necessary to evaluate non-clayey materials, such as silt, as
potential barriers in multi-layer cover systems.Mbonimpa
et al. (2003) used artificial silt as infiltration and oxygen
barrier in a cover system. In Cases C1C10 of the present
study, silt was investigated as a potential alternative to a
clay-type barrier such as the sand-bentonite.
4.2.1. Cases C1, C2, C3, and C4: influence of water depth
A single-layer silt cover (60 cm thick) placed directly on
3 m of tailings (mine waste) was investigated as shown in
Fig. 11. The influence of water table depression on the
drainage behavior of the silt barrier was evaluated. The
elevation of the water table, initially at the silttailings
interface in all cases, was decreased to 0.5, 1, 2 and 3 m
below the interface in Cases C1, C2, C3, and C4,
respectively.
The water table was lowered by 0.5 m in Case C1. The
suction at the bottom of the silt infiltration and oxygen
barrier in Case C1 at Day 120 was 5 kPa (Fig. 11), which
was less than the AEV of the silty tailings (approximately
10kPa) (Fig. 3). As a result, the average degree of
saturation in the silt infiltration and oxygen barrier in
Case C1 at Day 120 was still 90.3% (Fig. 11).
The water table was lowered by 1.0, 2.0, and 3.0 m over120 days in Cases C2, C3, and C4, respectively, as indicated
in Fig. 12. Lowering the water table increased suction at
the base of the cover, promoted drainage and decreased the
degree of saturation. The average degrees of saturation of
the silt infiltration and oxygen barrier in Cases C2, C3 and
C4 (Fig. 12) at Day 120 were approximately 81.1%, 66.9%,
and 57.9% respectively. Fig. 13 shows the desaturation
that occurred in the silt cover in Cases C1C4.
The data are further highlighted in Fig. 14 as a plot of
the average degree of saturation in the single silt cover
versus water level drop below the base of the cover in Cases
C1C4.Fig. 14shows that at the end of the first day, the
degree of saturation was the same for all four cases of the
modeled single silt cover. The reason for this behavior
probably is that during the first day the loss of water by
evaporation for all cases was the same and equal to the
potential evaporation, which is controlled by environmen-
tal conditions and not by soil properties and/or the depth
of water level. At later times, the loss of water from covers
is controlled by the actual evaporation, which is governed
by soil properties and water level location. Significant
reductions in the degree of saturation occurred when the
water level was lowered deep enough to exert suctions that
exceeded the AEV of the silt (10 kPa or 1 m of head), as
shown inFig. 14. This shows that the elevation of the water
ARTICLE IN PRESS
Cost increase (%)
0 2 4 6 8 10 12 14 16 18
Performanceimprov
ement(%)
0
10
20
30
40
0 5 10 15 20 25 30 35
Costincreaseorpe
rformance
improvement(%)
0
10
20
30
40
Cost increase
Performance improvement
Thickness of top capillary layer (cm)
Fig. 10. Percentage of cost increase and performance improvement versus thickness of top capillary layer, and correlation between cost increase and
performance improvement due to adding top capillary layer.
Table 2
The influence of thickness of top capillary layer on performance and cost of a soil cover system with a 60 cm thick sand-bentonite infiltration barrier and a
fixed water level
Soil cover
case
Thickness of top
capillary layer
(m)
Thickness of
bottom capillary
layer (m)
Average vwc in
infiltration
barrier (%)
Oxygen diffusion
coeff. of
infiltration
barrier (m2/s)
Oxygen flux
through soil
cover (kg/m2/s)
Performance
improvement
with thicker, top
capillary layer
(%)
Cost increase
with thicker, top
capillary layer
(%)
A0 0 0.30 32.06 2.4108 9.41012
A4 0.30 0.30 38.58 3.41011 5.81012 38.0 33.3
A5 0.15 0.30 36.89 6.91010 9.01012 3.5 16.6
Note: (i) vwc volumetric water content; (ii) A0, A4, and A5 modeled cover scenarios (Figs. 8 and 9).
E.K. Yanful et al. / Journal of Environmental Management 81 (2006) 729284
-
7/21/2019 A Numerical Study of Soil Cover Perfomance.yanful, Mousavi and de Souza.2006
14/21
table need not be very deep for a single layer soil cover with
a relatively low AEV, such as a silt cover, to desaturate.
4.2.2. Cases C5, C6, and C7: influence of thickness of lower
capillary layer
As mentioned before, Case C4 (Fig. 12) modeled a soil
cover consisting of only a silt infiltration and oxygen
barrier (without any capillary layer). The infiltration and
oxygen barrier desaturated when the water table was
lowered to a depth that caused suctions in the silt to exceed
its AEV. In Cases C5C7, the 60-cm-thick silt infiltration
and oxygen barrier was placed above different thicknesses
of Waite Amulet coarse sand lower capillary layer, as
shown in Fig. 15. Similar to Case C4, the covers were
subjected to a gradual 3-m drop in water level over a 120-
day period. The objective of studying these cases was to
evaluate the influence of different thicknesses of a lower
capillary layer on performance of silt as infiltration and
oxygen barrier. Three cases, C5, C6 and C7, with respective
lower capillary layers 15, 30 and 50cm thick were
considered. For a large site, the use of reduced thickness
of capillary layer can lead to significant cost savings, if
performance can be shown to be acceptable.
Fig. 15presents the profile of degree of saturation in the
infiltration barrier for Case C5 when the cover system
included a 15-cm-thick lower capillary layer. The figure
shows that the lower capillary layer desaturated rapidly
and reduced desaturation of the infiltration barrier.
Schematics of Cases C6 and C7 (30- and 50-cm-thick,
respectively) are shown inFig. 15.
Including a 15-cm-thick lower capillary layer consider-
ably decreased the total oxygen flux through the cover
system; however, further increase in the thickness of this
layer did not result in measurable further improvements.
The reason for this trend is illustrated in Fig. 16showing
that the provision of a thicker lower capillary layer only
marginally increases the degree of saturation of the silt
cover from 70% to 71%. These results imply that a single
silt cover would not be a suitable oxygen barrier in
situations where the water table is deep enough to induce
suctions equal to or greater than the AEV of the silt.
Fig. 17 presents percent cost increase and performance
improvement versus thickness of the lower capillary layer
for Cases C4C7, while Fig. 18 shows a plot of percent
performance improvement against cost increase.
4.2.3. Cases C8, C9, and C10: influence of thickness of
upper capillary layer
These cases were studied to examine the influence
of the thickness of the upper capillary layer on cover
ARTICLE IN PRESS
60 cm
300 cmtailings
Case C1
siltinitial water level
final water level
50 cm
Degree of saturation (%)
0 20 40 60 80 100
Elevationofinfiltrationbarrier(m)
0.0
0.1
0.2
0.3
0.4
0.5
0.6
day 1
day 15
day 30
day 60
day 90
day 120
Suction (kPa)
-5 0 5 10 15 20 25
Elevationofinfiltrationbarrier(m)
0.0
0.1
0.2
0.3
0.4
0.5
0.6
Fig. 11. Schematic of Case C1 and profiles of degree of saturation and suction for Case C1.
E.K. Yanful et al. / Journal of Environmental Management 81 (2006) 7292 85
-
7/21/2019 A Numerical Study of Soil Cover Perfomance.yanful, Mousavi and de Souza.2006
15/21
performance. First, an unprotected 60-cm-thick silt cover
was studied in Case C8 (Fig. 19). Then a 15-cm-thick fine
sand layer was placed on the silt cover in Case C9 and,
finally, the fine sand thickness was increased by 100% to
30cm in Case C10 (Fig. 19). The water level was kept at
1 m below the base of the cover during the simulation
period.
As the profiles of degree of saturation show (Fig. 20), the
unprotected silt cover (silt) in Case C8 lost quite a bit of
water because of evaporation. In fact, the average degree of
saturation of the infiltration barrier decreased to about
75.7% in 120 days. By adding 15 and 30 cm thick protective
layers in Cases C9 and C10 respectively, the average degree
of saturation of the silt cover increased to approximately
94% and 95% at the end of 120 days, as indicated in
Fig. 20. AsFig. 21demonstrates, the inclusion of a 15-cm-
thick upper capillary layer decreased the total oxygen flux
significantly. However, further increase in the thickness of
the upper capillary layer did not appear to decrease the
oxygen flux further.
Fig. 22 presents the computed percent cost increase
versus performance improvement with the addition of an
upper capillary layer to the single-layer silt cover. The
results suggest that, everything else being the same, the
ARTICLE IN PRESS
60 cm
300 cm tailings
Case C3
silt
initial water level
200 cm
final water level
60 cm
300 cm tailings
Case C4
silt
initial water level
final water level
60 cm
300 cm tailings
Case C2
siltinitial water level
100 cm
final water level
Fig. 12. Schematics of soil cover Cases C2C4.
Degree of saturation (%)
20 40 60 80 100
Elevation(m)
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.5 m (Case C1)
1 m (Case C2)
2 m (Case C3)
3 m (Case C4)
Fig. 13. Profiles of degree of saturation versus decreasing water level
(Cases C1C4) in a single layer silt cover overlying mine tailings at day
120.
E.K. Yanful et al. / Journal of Environmental Management 81 (2006) 729286
-
7/21/2019 A Numerical Study of Soil Cover Perfomance.yanful, Mousavi and de Souza.2006
16/21
provision of a 1530 cm thick upper capillary layer
increased the cost of the cover system by 2550%, but
improved performance by nearly 99%. Although the
results demonstrate the important role of the upper
capillary layer in preventing desaturation of the oxygen
barrier, they also show that increasing the thickness of the
upper capillary layer considerably more than an optimum
value would not be cost-effective, because the resulting
increase in the degree of saturation of the silt infiltration
and oxygen barrier would be only marginal.
ARTICLE IN PRESS
Degree of saturation (%)
0 10 20 30 40 50 60 70 80 90 100
Elevationofcoversystem
(m)
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
day 1day 15
day 30
day 60
day 90
day 120
60 cm
300 cm
tailings
Case C7
silt
50 cm WA coarse sand
initial water level
final water level
60 cm
300 cm tailings
Case C5
silt
15 cm WA coarse sandinitial water level
final water level
60 cm
300 cmtailings
Case C6
silt
30 cm WA coarse sand
initial water level
final water level
Fig. 15. Schematic diagram of soil cover Cases C5C7 and profile of degree of saturation for Case C5.
Water level drop
Degreeofsaturation(%)
0
10
20
30
4050
60
70
80
90
100
day 1
day 15
day 30day 60
day 90
day 120
0.5 m(Case C1)
1 m(Case C2)
2 m(Case C3)
3 m(Case C4)
Fig. 14. Influence of water level drop on the degree of saturation of the infiltration barrier.
E.K. Yanful et al. / Journal of Environmental Management 81 (2006) 7292 87
-
7/21/2019 A Numerical Study of Soil Cover Perfomance.yanful, Mousavi and de Souza.2006
17/21
5. Advantages and limitations of the proposed approach
Evaporation is an important factor that controls the
performance of covers designed to retain moisture and hence
reduce oxygen transport. The present work is different from
many previous studies of soil covers because it includes
evaporation as flux-boundary condition, and also considersall key aspects of cover design and analysis (material
selection, flow modeling, oxygen transport modeling, and
cost analysis). This integrated approach allows the designer
or proponent to eliminate less cost effective options without
the need for costly long-term laboratory and test plot studies.
The steps outlined in the conceptual flowchart of
computational sequence (Fig. 1) and the approach described
in this paper can be a helpful guide in the design of an
economically feasible case- and project-dependent soil cover.
It is based on known and proven theories and published
methods. Also, it is fairly easy to use and the needed input
data are generally routinely available from most field
measurements or standard laboratory testing. The results
of this study, when combined with in situ monitoring and
physical study of the effects of freeze-thaw, desiccation, and
erosion, can constitute a rigorous evaluation of the long-term
performance of a soil cover system. The use of the method
along with well-instrumented field test covers can eliminate
performance uncertainties in full-scale applications.
Despite these advantages, there are a number of
limitations with the study. The use of the one-dimensional
model provides only an approximate analysis of water flow
as it is recognized that a two-dimensional analysis, or even
a three-dimensional analysis, would be more appropriate.
Also, the numerical model used in the study considered soilas a continuum and did not consider processes such as
desiccation, cracking, and freeze-thaw. Hysteretic effects
caused by alternate cycles of wetting and drying conditions
were also not considered.
The oxygen flux through single-layer and multi-layer
cover cases was calculated using a steady-state analysis.
The steady-state flux provided a long-term assessment of
possible reductions in oxygen ingress and was considered
adequate. In a previous study, Yanful (1993) computed
both laboratory and field transient oxygen fluxes and
confirmed reductions in oxygen ingress by a multi-layer
cover. Finally, the present study dealt with the hydraulic
performance of soil cover systems and their impact on
oxygen transport, but did not include chemical transport
analysis. It was assumed that a reduction in oxygen flux
due to the presence of a cover would necessarily lead to
decreased mine waste oxidation and hence contaminant
release. This is a reasonable assumption and has been
confirmed by several field and laboratory studies (for
example, Yanful, 1993; Payant et al., 1995; Yanful et al.,
1999;Yanful and Orlandea, 2000). It should be noted that
although numerical modeling was used here to predict the
performance of different designs, it is recognized that field
data from constructed test covers may be used to confirm
or modify the results of the numerical analysis.
ARTICLE IN PRESS
0 5 10 15 20 25 30 35 40 45 50 5556
58
60
62
64
66
68
70
72
Averagedegre
eofsaturation
atinfiltrationbarrier(%)
Thickness of lower capillary layer (cm)
Fig. 16. Influence of thickness of lower capillary layer on degree of
saturation at the infiltration barrier.
Thickness of lower capillary barrier (cm)
0 10 20 30 40 50 60
Costincrease(%)
0
20
40
60
80
100
Performanceimprovement(%)
0
20
40
60
80
100
Cost increase
Performance improvement
Fig. 17. Cost increase and performance improvement versus thickness of
lower capillary layer for a silty infiltration barrier (Comparison of Cases
C4C7).
Cost increase (%)
0 25 50 83
Performanceimprovem
ent(%)
0
5
10
15
20
25
Fig. 18. Performance improvement versus cost increase for a silty cover
with lower capillary barrier.
E.K. Yanful et al. / Journal of Environmental Management 81 (2006) 729288
-
7/21/2019 A Numerical Study of Soil Cover Perfomance.yanful, Mousavi and de Souza.2006
18/21
6. Summary and application.
(1) Single cover: A single-layer soil cover consisting of only
an infiltration or oxygen barrier could be used if
sufficient quantity of the material is available at or
within economic hauling distance of the mine site.
However, the cover must be protected with an over-
lying soil layer against erosion, freeze-thaw, desiccation
and cracking. A single silt cover cannot be used as an
infiltration and oxygen barrier in situations where the
ARTICLE IN PRESS
Case C8-Degree of saturation (%)
0 10 20 30 40 50 60 70 80 90 100
Co
verelevation(m)
0.0
0.1
0.2
0.3
0.4
0.5
0.6
day 1
day 15
day 30
day 60
day 90
day 120
60 cm
100 cm
tailings
Case C8
silt
constant water level
60 cm
15 cm
100 cm
constant water level
tailings
silt
WA fine sand
Case C9
60 cm
30 cm
100 cm
constant water level
tailings
silt
WA fine sand
Case C10
Fig. 19. Schematic diagram of soil cover Cases C8C10 and profile of degree of saturation for Case C8.
Thickness of upper capillary barrier (cm)
Degreeofsaturation(%)
0
10
20
30
40
50
60
70
80
90
100
day 1
day 15
day 30
day 60
day 90
day 120
0 15 30
Fig. 20. Influence of the thickness of upper capillary layer on degree of saturation of silt cover.
E.K. Yanful et al. / Journal of Environmental Management 81 (2006) 7292 89
-
7/21/2019 A Numerical Study of Soil Cover Perfomance.yanful, Mousavi and de Souza.2006
19/21
water table is deep enough to induce suctions equal to
or greater than the AEV of the silt.
(2) Layered soil cover: The largest component of the cost of
a soil cover is likely the construction of the infiltration
and oxygen barrier. The effectiveness of a material as
an infiltration and oxygen barrier is dependent on its
compaction and water retention characteristics includ-
ing the magnitude of its AEV. Generally, this barrier
need not be thick (typically, 0.51.0 m) if protective
upper and lower capillary layers are provided to
prevent its desaturation. A layered soil cover is
generally the preferred cover system for acid generating
mine waste. Infiltration and oxygen barrier may consist
of clay and a sand-bentonite mixture.
In this study, a sand-bentonite barrier with a saturated
hydraulic conductivity of 1 107 cm/s was observed to
retain its moisture. The movement of water in the capillary
layers caused by upward evaporation and downward
drainage did not produce significant water flow within
the sand-bentonite. In fact, a large change in suction
resulted in only a small change in the water content or
degree of saturation. The hydraulic conductivity of
1107
cm/s was realistic as it is similar to typical field
measured values for well-compacted, clayey soils (Elsbury
et al., 1990).
At many Canadian mine sites there is a limited quantity
of fine grained, clay type soil materials while silts may be
present in sufficient quantities to merit consideration in
cover projects. In the cases considered in the present study,
where silt was the infiltration and oxygen barrier, a lowercapillary barrier effectively stabilized the degree of satura-
tion of the silt. A contrast in grain-size distribution would
promote rapid drainage in lower capillary layer and
maintain a high degree of saturation in the infiltration
and oxygen barrier.
Silt is more susceptible to frost heaving than clay.
Therefore, when it is used as an infiltration and oxygen
barrier in wet and cold areas, it may be necessary to place it
below the zone of frost penetration. This could translate
into using upper capillary and erosion protection layers of
at least 1.2 m thick in many parts of Canada where frost
can penetrate that deep. In such an application though, the
primary role of the silt would be to function as an oxygen
barrier, although it could also reduce water infiltration to
some extent. Besides, non-plastic silt and/or silt with low
plasticity have a self-healing potential that allows them to
recover their hydro-geotechnical properties after freeze-
thaw cycles (Eigenbrod, 2003).
From the results of the present study, the relation
between the thickness of a capillary layer and the degree of
saturation of the infiltration and oxygen barrier was not
linear implying that increasing the thickness of a capillary
layer considerably more than an optimum amount only
marginally increases the degree of saturation of the
infiltration and oxygen barrier.It is obvious that the capillary layers and infiltration and
oxygen barrier did not perform independently; rather, they
had mutual hydraulic interactions that influence cover
performance. For example, it was found that simply
doubling the thickness of the capillary layer did not
produce a corresponding doubling in performance im-
provement.
(3) Minimum design criteria and proposed approach: The
properties of interest in the selection of candidate soil
cover materials include the pressure head at which the
lower capillary barrier (coarse sand in this study)
approaches its AEV and residual saturation head (hr),
the AEV of the infiltration and oxygen barrier, and hrof the upper capillary barrier.
7. Conclusions
An investigation of the influence of a number of soil
cover properties and hydrogeologic parameters on perfor-
mance has been presented. The investigation simulated
one-dimensional evaporation from hypothetical moisture-
retaining cover systems using the commercially available
transient flow model, SoilCover, to compute suction and
ARTICLE IN PRESS
Thickness of upper capillary barrier (cm)
0 5 10 15 20 25 30 35
Costincrease(%)
0
20
40
60
80
100
Performanceimprovement(%)
0
20
40
60
80
100
Cost increase
Performance improvement
Fig. 22. Percentage of cost increase and performance improvement versus
thickness of upper capillary layer (Comparison of Cases C8C10).
Time (day)
0 20 40 60 80 100 120
TotalOxygenFlux(kg/m2/s)
0 cm Case C8
15 cm Case C9
30 cm Case C10
Thickness of upper capillary layer
0.5x10-9
1.0x10-9
1.5x10-8
2.0x10-8
2.5x10-8
3.0x10-8
3.5x10-8
0.0
Fig. 21. Total oxygen flux through soil cover with top capillary layers of
0, 15, and 30 cm thick.
E.K. Yanful et al. / Journal of Environmental Management 81 (2006) 729290
-
7/21/2019 A Numerical Study of Soil Cover Perfomance.yanful, Mousavi and de Souza.2006
20/21
water content profiles. Water content was used as input to
an oxygen diffusion model to calculate oxygen flux for the
different design scenarios. A cost analysis was then
performed for the various scenarios to obtain a relation-
ship between performance and cost. Based on the results,
the following is concluded:
(1) The key properties in the selection of candidate soil
cover materials are the saturated hydraulic conductiv-
ity, compaction characteristics, oxygen diffusion coeffi-
cient, hydraulic conductivitysuction function, AEV,
and the pressure head at which the various layers (for
example, lower and upper capillary barriers) ap-
proaches their AEVs.
(2) A single-layer silt cover would not perform well as
infiltration and oxygen barrier in situations where the
water table is deep enough to desaturate the cover. Silt
may be a good oxygen barrier if it is protected against
drainage with an underlying coarse layer and against
erosion, freeze-thaw, desiccation and cracking with an
overlying coarse soil.
(3) A multi-layer soil cover consisting of an infiltration and
oxygen barrier, drainage barrier and an upper protec-
tive layer performs better than a single-layer cover.
(4) Although increasing the thickness of the upper
capillary barrier in a multi-layer cover improves cover
performance significantly, it only does so to a certain
point. A greater increase in thickness, say from 30 to
50cm, increases cost significantly (by 33%), but
improves performance marginally (by 11%). Thus the
relationship between the thickness of the upper
capillary barrier and cover performance is not linear.(5) Sand-bentonite can serve as an excellent infiltration and
oxygen barrier, if the economics would allow. A sand-
bentonite layer with a saturated hydraulic conductivity
of 1107 cm/s can retain its moisture, even if it is
subjected to a large change in suction. The change in
suction results in only a small change in the water
content or degree of saturation.
(6) The step-by-step method of analysis proposed in this
paper would be a useful tool for making technical and
economic decisions on site-specific cover designs prior
to, or in combination with, pilot scale testing. Such an
approach can lead to a cost-effective and technicallyfeasible design of multi-layer cover systems.
References
Akindunni, F.F., Gillham, R.W., Nicholson, R.V., 1991. Numerical
simulations to investigate moisture-retention characteristics in the
design of oxygen-limiting covers for reactive mine tailings. Canadian
Geotechnical Journal 28, 446451.
Aubertin, M., Chapuis, R.P., Aachib, M., Bussie` re, B., Ricard, J.-F.,
Tremblay, L., 1995a. E valuation en laboratoire de barriers seches
construites a` partir de re sidus miniers. Ecole Polytechnique, CDT
P1622. Final Report, Mine Environment Neutral Drainage Program
Programme de Neutralisation des Eaux de Drainage dans lEnvir-
onnement Minier (MEND/NEDEM).
Aubertin, M., Ricard, J.-F., Chapuis, R.P., 1995b. A study of capillary
properties of mine tailings: measurements and modeling. In: Proceed-
ings, 48th Canadian Geotechnical Conference, Vancouver, BC, pp.
1724.
Aubertin, M., Bussie` re, B., Aachib, M., Chapuis, R.P., Crespo, J.R., 1996.
Une mode lisation nume rique des e coulements non sature s dans des
couvertures multicouches