Journal of Environmental Management 119 (2013) 36e46
Contents lists available at SciVerse ScienceDirect
Journal of Environmental Management
journal homepage: www.elsevier.com/locate/jenvman
Selection of remedial alternatives for mine sites: A multicriteria decision analysis
approach
Getnet D. Betrie a, *, Rehan Sadiq a, Kevin A. Morin b, Solomon Tesfamariam a
a
b
School of Engineering, University of British Columbia, Kelowna, BC, Canada
Minesite Drainage Assessment Group, Vancouver, BC, Canada
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 9 August 2012
Received in revised form
21 January 2013
Accepted 27 January 2013
Available online
The selection of remedial alternatives for mine sites is a complex task because it involves multiple criteria
and often with conflicting objectives. However, an existing framework used to select remedial alternatives lacks multicriteria decision analysis (MCDA) aids and does not consider uncertainty in the selection
of alternatives. The objective of this paper is to improve the existing framework by introducing deterministic and probabilistic MCDA methods. The Preference Ranking Organization Method for Enrichment
Evaluation (PROMETHEE) methods have been implemented in this study. The MCDA analysis involves
processing inputs to the PROMETHEE methods that are identifying the alternatives, defining the criteria,
defining the criteria weights using analytical hierarchical process (AHP), defining the probability distribution of criteria weights, and conducting Monte Carlo Simulation (MCS); running the PROMETHEE
methods using these inputs; and conducting a sensitivity analysis. A case study was presented to
demonstrate the improved framework at a mine site. The results showed that the improved framework
provides a reliable way of selecting remedial alternatives as well as quantifying the impact of different
criteria on selecting alternatives.
2013 Elsevier Ltd. All rights reserved.
Keywords:
Multicriteria decision analysis
PROMETHEE
Probabilistic PROMETHEE
Remedial alternatives
Mine sites
1. Introduction
Acid Rock Drainage (ARD) from mine wastes that contains elevated metals and metalloids poses human and environmental risks
(Gray, 1996; Azapagic, 2004). ARD is produced when sulfidebearing material is exposed to oxygen and water during mining
activities (Morin and Hutt, 1997; Price, 2009). Examples of human
health risks include increased chronic diseases and various types of
cancer. On the other hand, examples of ecological risks range from
the elimination of species to a significant reduction of ecological
stability, and to the bioaccumulation of metals in the flora and
fauna (Gray, 1996). In the eastern and western U.S., between 7000
and 16,000 km of streams are affected by acid generating waste
rocks and between 8000 and 16,000 by ARD (USEPA, 1994). In
Canada, an estimated 351 million tonnes of waste rock, 510 million
tonnes of sulfide tailings, and more than 55 million tonnes of other
mining sources have the potential to cause ARD (Minewatch, 2006).
Liability costs of potentially acid-generating wastes at mining sites
are estimated to be US$ 530 million in Australia, between US$ 1.2
and 20.6 billion in USA, and between US$ 1.3 and 3.3 billion in
* Corresponding author. Tel.: þ1 250 807 9013; fax: þ1 250 807 9850.
E-mail addresses: [email protected], [email protected] (G.D. Betrie).
0301-4797/$ e see front matter 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.jenvman.2013.01.024
Canada (USEPA, 2006). Therefore, it is important to use remedial
technologies to control the impacts of ARD caused by the considerable amounts mine waste rocks.
ARD remedial technologies are categorized into source control
and water treatment technologies (USEPA, 2006). The source control technologies chemically stabilize reactive rocks, or physically
isolate waste rocks from water or oxygen, while water treatment
technologies reduce contaminants in mine waters. However,
selecting the optimal technology for a mine site is quite complex
(USEPA, 2006). This is because most environmental decisionmaking (i.e., the selection process) involves multiple and conflicting objectives (e.g., minimizing risk and cost, maximizing benefit,
and maximizing stakeholder preferences) (Kiker et al., 2005; Sadiq
and Tesfamariam, 2009). Moreover, input information for each
objective is often obtained in different forms (i.e., quantitative and
qualitative), which are non-commensurable and thus exacerbate
the decision making process (Tesfamariam and Sadiq, 2008). It is
worth noting that there is a move away from decision-making
based on single objective (e.g., cost or technical feasibility) toward selecting a sustainable remediation technology by considering multiple objectives such as environmental, economical and
social objectives.
A framework for selecting remedial alternatives at mine sites,
which is recommended by USEPA (1988), is shown in Fig. 1. This
G.D. Betrie et al. / Journal of Environmental Management 119 (2013) 36e46
framework consists of series of steps: scoping, site characterization,
treatability investigations, development and screening of alternatives, detailed analysis of alternatives, and selection of optimum
alternative. The scoping step consists of collecting existing data,
identifying the boundary of a study area, identifying the remedial
action objectives, assembling a “technical advisory committee,” and
preparing the project plan. The site characterization step involves
conducting field investigation, defining the nature and extent of the
contamination, identifying regulatory standards and requirements,
and conducting a baseline risk assessment. If there is no adequate
data to evaluate technologies, the treatability investigation step is
followed, where the evaluation of bench or pilot-scale technologies
is performed. In the next step, Development and Screening of Alternatives, potential technologies are identified and screened.
These, technologies are then assembled into potential alternatives
and screened. Next, in the Detailed Analysis of Alternatives, the
developed alternatives are analyzed against established criteria,
and then, using the results of this analysis, compared with each
other. In the final step, Select Optimum Alternative, the optimum
alternative is selected based on information obtained from the
37
detailed analysis of alternatives. In this framework, experts conduct
remedial selection analysis without multiple criteria decision
analysis (MCDA) aids. However, the literature shows that humans
are not capable of solving multiple objectives unaided. When they
attempt to do so, opposing views are often discarded (McDaniels
et al., 1999).
MCDA methods deal with a problem whose alternatives are
predefined and decision-makers rank available alternatives based
on the evaluation of multiple criteria (Tesfamariam and Sadiq,
2006; Sadiq and Tesfamariam, 2009). There are many MCDA
methods, most of which consist of four steps e (i) structuring the
decision problem, (ii) articulating and modeling preferences, (iii)
aggregating the alternative evaluations (preferences); and (iv)
making recommendations (Guitouni and Martel, 1998). The main
differences between these methods lie in the type of algorithm
used to aggregate alternative evaluations, the types of input data
required, and their final results (i.e., a single alternative vs. rank of
alternatives). The literature classifies the MCDA methods into elementary, utility theory, and outranking (Belton and Stewart, 2002;
Figueria et al., 2005). The elementary methods identify a non-
Fig. 1. The USEPA (1988) framework for selecting remedial alternatives at mine sites.
38
G.D. Betrie et al. / Journal of Environmental Management 119 (2013) 36e46
dominated alternative based on the performance of alternatives in
at least one criterion. These methods are suitable for reducing the
number of alternatives into potential alternatives; however, they
are not suitable for environmental problems that involve many
alternatives and criteria (Kiker et al., 2005). Examples of elementary methods include maxmin, conjunctive and other methods.
Utility theory methods use a utility/value function for each criterion to evaluate alternatives and to aggregate the utility/value of
each criterion in order to identify the best alternative (Keeney and
Raiffa, 1993). Examples of these methods include MAU/VT (multiattribute utility/value theory), AHP (analytical hierarchy process),
and other methods. The outranking methods build an outranking
relation (i.e., a binary relation that is used to evaluate the performance of the outranking character of one alternative over another
alternative), and then exploit this relation to identify the best
alternative, sort alternatives into groups, or rank them (Belton and
Stewart, 2002). Examples of outranking methods include ELECTRE
(elimination et choix traduisant la realite), PROMETHEE (preference
ranking organization method for enrichment of evaluations), and
other methods. Guitouni and Martel (1998) presented a comprehensive review of MCDA methods and their description. A comprehensive literature review of MCDA applications in the
environmental decision making was conducted by Kiker et al.
(2005) and Huang et al. (2011). Recently, Yang et al. (2012) has
developed a simulation-based fuzzy multicriteria decision analysis
to select remedial alternatives for a benzene-contaminated site.
Uncertainty is an unavoidable and inevitable component of any
environmental decision making process (Sadiq and Tesfamariam,
2009). MCDA methods require input data such as weights of criteria and preference of alternatives with respect to each criterion
provided by decision makers. However, these data have uncertainty
because of decision makers’ judgment and subjectivity. Before
informed decisions can be made, this uncertainty must be quantified, through the application of available techniques.
The objective of this paper is to improve the existing framework
used for selecting remedial alternatives by introducing deterministic and probabilistic MCDA methods. Although there are many
MCDA methods, as previously discussed, the PROMETHEE methods
are implemented in this study. The PROMETHEE methods are
implemented for two reasons (i) they are easy to apply compared to
other outranking methods such as ELECTRE since PROMETHEE
methods require fewer parameters from decision makers; and (ii)
they rank alternatives as well as identify the best alternative, where
as the utility theory methods only identify the best alternatives.
The remainder of this paper is organized as follows. In the next
section, the proposed framework and methods that are used to
implement the MCDA framework are discussed. Then a case study
is presented to demonstrate the proposed framework at a mine site.
Finally, the results and conclusion of this study are discussed.
2. Proposed framework
The proposed framework for selecting remedial alternatives is
shown in Fig. 2. This framework improves the existing one by
introducing deterministic and probabilistic multicriteria decision
analysis methods instead of “Detailed Analysis of Alternatives”
block shown in Fig. 1. Moreover, the proposed framework introduces “Result Analysis” block. The deterministic multicriteria
block consists of identifying criteria to evaluate alternatives, identifying the weight of criteria using AHP, and applying the PROMETHEE I and II methods. The probabilistic multicriteria decision
analysis block consists of identifying criteria to evaluate alternatives, defining the weight of criteria by using AHP, defining probability distribution for the weight of criteria, conducting Monte
Carlo Simulation, applying PROMETHEE II, and conducting
a sensitivity analysis. In the subsection below, the elements of the
deterministic MCDA and the probabilistic MCDA blocks are presented. Since the probabilistic MCDA block contains the first two
elements of the deterministic block, the elements of the probabilistic MCDA block beginning from “defining the probability distribution criteria weight” are discussed. Finally, the analysis block is
discussed.
2.1. Deterministic multicriteria decision analysis
2.1.1. Preference Ranking Organization Method for Enrichment
Evaluation (PROMETHEE)
The PROMETHEE I and II methods are multicriteria decision
making techniques developed by J. P. Brans (Brans et al., 1986).
These methods build a valued outranking relation and exploit this
relation to obtain a partial ranking (PROMETHEE I) or complete
ranking (PROMETHEE II). When A of n alternatives (i.e., a1, a2, ., an)
have to be ranked, and G of k criteria (i.e., g1, g2, ., gk) have to be
maximized or minimized (shown in Fig. 3), the resulting multicriteria problem has the following form as shown in Eq. (1) (Brans
et al., 1986):
maxfg1 ðaÞ; g2 ðaÞ; .; gk ðaÞja˛Ag
(1)
The basic steps of PROMETHEE methods to solve such multicriteria problem are (Brans et al., 1986):
(i) Defining a preference function for each criterion,
(ii) Calculating the multicriteria preference index as a weighted
average of the preference functions,
(iii) Calculating a leaving flow, an entering flow, and a net flow for
each alternative, and
(iv) Assigning a partial or complete ranking.
A preference function ðPj ða; bÞða; bÞ˛AÞ associated with criterion
gi gives the degree of preference, expressed by decision-makers, for
alternative a over b. In other words, the preference function takes
the deviation of the preference a over b as an input and provides
a normalized output as shown by Eqs. (2) and (3), respectively.
Pj ða; bÞ ¼ Pj ½gi ðaÞ gi ðbÞ
(2)
0 Pj ða; bÞ 1
(3)
If Pj ða; bÞ ¼ 0, there is no preference a over b and if Pj ða; bÞ ¼ 1,
there is a strict preference of a over b. Brans and Mareschal (2005)
have provided six types of preference functions as displayed in
Table 1.
If a weight wj is given, which shows the relative importance of
each criterion, and can be used to calculate the multicriteria preference index as defined by Eq. (4):
pða; bÞ ¼
k
X
wj Pi ða; bÞ
(4)
i¼1
The leaving (outgoing) flow that shows how an alternative ai is
outranking other alternatives is defined by the equation below:
fþ ðaÞ ¼
X
pða; xÞ
(5)
x˛K
The incoming (leaving) flow that shows how an alternative ai is
outranked by other alternatives is defined by the following
equation:
G.D. Betrie et al. / Journal of Environmental Management 119 (2013) 36e46
39
Fig. 2. Proposed framework for selecting remedial alternatives at mine sites.
f ðaÞ ¼
X
pðx; aÞ
(6)
x˛K
PROMETHEE I is used to derive a partial ranking which is
obtained by comparing the leaving and incoming flows, as defined
by Eq. (7):
8
8
>
>
< fþ ðaÞ > fþ ðbÞ and f ðaÞ < f ðbÞ;or
>
>
>
I b iff
>
ap
fþ ðaÞ ¼ fþ ðbÞ and f ðaÞ < f ðbÞ;or
>
>
>
>
: fþ ðaÞ > fþ ðbÞ and f ðaÞ ¼ f ðbÞ;
<
þ
>
aI I b iff f(
ðaÞ ¼ fþ ðbÞ and f ðaÞ ¼ f ðbÞ
>
>
>
>
>
fþ ðaÞ > fþ ðbÞ and f ðaÞ > f ðbÞ;or
>
I
>
: aR b iff
fþ ðaÞ < fþ ðbÞ and fþ ðaÞ < fþ ðbÞ;
(7)
where PI, II and RI represent preference, indifference, and
incomparability, respectively.
PROMETHEE II is used to derive a complete ranking which is
calculated as a net flow between outgoing and incoming flows, as
shown in Eq. (8). If the net flow of a is greater than b, a outranks b.
Otherwise, a is indifferent to b.
fðaÞ ¼ fþ ðaÞ f ðaÞ
(8)
2.1.2. Analytical hierarchy process (AHP)
AHP is one of the MCDA methods that assists decision makers in
solving complex problems by organizing thought, experiences,
knowledge, and judgments into a hierarchical framework, and by
guiding them through a sequence of pairwise comparison judgments (Saaty, 1982). In general, it derives the ratio scale from
a pairwise comparison. In this study, AHP is used only to generate
weight. The process of generating weight using AHP consists of the
following steps (Saaty, 1982):
(i) Identify the criteria to generate weights,
40
G.D. Betrie et al. / Journal of Environmental Management 119 (2013) 36e46
values should be modified until the CR value is within an
acceptable range.
2.2. Probabilistic multicriteria decision analysis
2.2.1. Define probability distribution for criteria weight
The criteria weights are generated using the AHP technique
from preference information provided by actors involved in the
decision process. The definition of probability distribution for criteria weight solely depends on the number of actors who are participating in the decision process. If an adequate number of actors
participated in the decision process, the obtained information
on preference is fitted to a probability distribution function using
available techniques or software. In a situation where there are no
adequate actors participating in the decision process, the type of
probability distribution that is recommended is uniform distribution (Rietveld and Ouwersloot, 1992).
Fig. 3. Evaluation matrix.
(ii) Make a pairwise comparison for each criterion on a scale of 1e
9. Note that the relative preference of one criterion over
another is obtained from decision makers,
(iii) Compute Eigen values and vectors to obtain a normalized
weight for each criterion, and
(iv) Compute the consistency ratio (CR). If the CR is more than 10%,
then the results are inconsistent. Thus, the assigned weight
Table 1
Types of preference function (Brans and Mareschal, 2005).
Preference function
Type I:
Usual
Criterion
Definition
P
1
PðdÞ ¼
P
1
PðdÞ ¼
0
Type III:
V-shape
Criterion
d0
d>0
0;
1;
dq
d>q
d
q
8
0;
>
>
>
>
<d
;
PðdÞ ¼
>p
>
>
>
:
1;
P
1
d
p
0
Type IV:
Level
Criterion
0;
1;
8
0;
>
>
>
<
1
PðdÞ ¼
;
>
2
>
>
:
1;
P
1
0.5
0
q
p
Type V:
P
V-Shape
1
with indifference
Criterion
0
q
p
d
d
Type VI: P
Gaussian 1
Criterion
(
d
d0
0dp
P
6 ni¼ 1 d2i
n n2 1
r ¼ 1 (9)
d>p
where di is the difference between the rank of values of alternatives
and weights of criteria, and n is the total number of MCS
realizations.
dq
q<dp
d>p
8
0
>
>
>
>
<d q
PðdÞ ¼
;
>
pq
>
>
>
:
1
PðdÞ ¼
0
2.2.3. Sensitivity analysis
Sensitivity analysis is used to identify the contribution of each
criterion in the overall ranking of alternatives. A sensitivity analysis
is often carried out using MCS and a rank correlation method is
applied to the results of MCS in order to identify the contribution of
parameters (i.e., criteria) to the output (Tesfamariam et al., 2006).
The latter method involves the determination of correlation coefficients that measure the strength of the linear relationship of
two variables. In this study, the Spearman rank correlation coefficient (see Eq. (9)), which measures the linear relationship of two
ranked variables, was used.
d
0
Type II:
U-shape
Criterion
2.2.2. Conduct Monte Carlo Simulation
Monte Carlo Simulation (MCS) is used to propagate uncertainties in random variables. In this technique, the weight of the
criterion is represented by a probability density function that defines both the range of values which the weight can take and the
likelihood that the weight has a value in any subinterval of that
range (Aral, 2010). A random sampling is used to select the weight
of each criterion set at different probability levels, and repeated
several times. For each combination of sampled weights, the
PROMETHEE II method is run. The MCS runs until sufficient sample
values have been collected to approximate the input parameter
distribution.
dq
q<dp
d>p
d0
0;
1e
2
d2
2s
;
2.3. Results analysis
In the result analysis block, the deterministic alternative rankings
and probabilistic alternative rankings are analyzed. The deterministic alternative rankings are obtained from partial and complete
rankings of PROMETHEE methods. The probabilistic alternative
rankings are obtained from a complete ranking of the PROMETHEE II
method. These results provide probability distributions for the
ranking of alternatives. In addition, the results provide the probability of given alternatives belonging to various ranks.
3. Case study
d>0
In this section, the proposed framework was implemented for
Mine Site B, located in British Columbia, Canada.
G.D. Betrie et al. / Journal of Environmental Management 119 (2013) 36e46
3.1. Site characterization
Site characterization information was obtained from the closure
plan report of the mine site. The remediation objective at this mine
site is to reduce environmental risk. The conceptual model for the
risk assessment is shown in Fig. 4. It shows the sources of contaminants, release mechanisms, media, exposure routes, and receptors. According to the mine’s closure plan report, the chemical of
concern (COC) is elevated copper concentrations from reactive
waste rocks and tailings. The water quality prediction obtained
using data from kinetic tests, heap leach tests, and historical water
quality analysis indicates a maximum of 8 mg/l during neutral pH
periods when neutralization potential was being consumed. The
possible release mechanisms are leaching and erosion. The media
that could be contaminated are surface water and soil. The two
potential exposure routes that are considered at this site are
ingestion and contact. The major ecological entity that should be
protected is fish. Thus, the maximum allowable discharge criterion
of Cu concentration was 0.003 mg/l CCME (1999).
The fate and transport of metals in the environment are determined using an aquivalence-based model. The aquivalence/fugacity
approach was developed by Mackay (1991) for the estimation of
chemical distributions in multimedia based on the complexity of
transport and transformation processes. The aquivalence/fugacitybased model is widely applied to predict the fate and transport of
toxic substances in the environment. This model describes the fate
of chemicals in any well-mixed water body for which the hydraulic
and particle flows are defined. Mackay and his co-workers (Mackay
et al., 1983 and Mackay and Diamond, 1989) applied a fugacity/
aquivalence-based model to rivers and lakes (e.g., Lake Ontario
and Hamilton Harbor) to study the distribution of detergent
chemicals, Polychloro-biphenyls and heavy metals.
A steady state aquivalence-based model developed for this mine
site by Betrie et al. (2012) was used in this case study to predict the
fate and transport of copper in soil and water media. The mine
system consists of tailings and waste rocks, a soil layer, and an
aquifer as shown in Fig. 5. The tailings and waste rocks are sources
of copper metal emissions. Intermedia transport, water-to-soil, and
soil-to-water mass flow through diffusion and advection were
considered. A conceptual model showing different mechanisms of
transport and transformation in two compartments is shown in
41
Fig. 5. The intermedia D values are contaminant transfer from
water-to-soil (DWS) and soil-to-water (DSW). The advection in soil
(DAS) and groundwater (DAW) values in soil and water compartments were calculated based on flow rates.
In this case study, a baseline risk assessment was done following
ecological risk assessment guidance for superfund sites (USEPA,
1997). The ecological risk was estimated using the Hazard Quotient (HQ) approach. The HQ is a ratio that can be used to estimate if
harmful effects, due to the contaminant in question, are likely
(USEPA, 1997). The HQ is calculated using the equation below:
HQ ¼
EEC
Screening Benchmark
(10)
where EEC is the estimated (maximum) environmental contaminant concentration at the site (i.e., contaminant concentration
in the soil, sediment, or water). The screening benchmark is generally a No-Adverse Effects Level concentration. Note that for
a calculated value of HQ > 1, harmful effects cannot be ruled out; if
HQ ¼ 1, contaminant alone is not likely to cause ecological risk;
and if HQ < 1, harmful effects are not likely (USEPA, 1997).
3.2. Development and screening of alternatives
Table 2 shows the alternatives that were considered in this
study and their effectiveness to reduce contaminants. The literature
review (e.g., Skousen et al., 1998; Skousen et al., 2000; USEPA,
2000; Johnson and Hallberg, 2005; USEPA, 2006) shows various
remedial alternatives. These alternatives reduce exposure of reactive wastes to oxygen and water, or reduce the amount of contaminants so that any adverse effects on humans and the
environment are eliminated. The potential effectiveness (in percent) of the alternatives to reduce acid generation is shown in
Table 2. These values are obtained from the literature as presented
in this table. Note that in this step, Development and Screening of
Alternatives, potential technologies should have been identified,
screened, and then assembled into alternatives as shown in Fig. 2.
In this case study, however, this was not done; instead, a single
technology was considered as an alternative in order to make the
illustration of proposed framework simple.
Fig. 4. Conceptual model for risk assessment.
42
G.D. Betrie et al. / Journal of Environmental Management 119 (2013) 36e46
Table 3
Criteria and scoring scale.
Fig. 5. Schematic representation of the mine site.
3.3. Deterministic MCDA
Criteria to evaluate the alternatives were identified by an expert
from the mine site and the authors in order to ensure that the
proposed remedial objectives are achieved. These criteria are presented below:
Minimize risk posed to fish,
Minimize the cost of an alternative,
Maximize the short-term performance of an alternative,
Maximize the long-term performance of an alternative,
Maximize the implementability of an alternative,
Maximize the reduction of toxicity, mobility, or volume of
wastes, and
Maximize the future use and aesthetic of the mine site.
Table 3 shows these criteria and their respective scoring scale.
Each criterion except risk was rated by experts on a scale of 1e9, as
shown in Table 3. An alternative that has low cost rated as 9 and the
high cost rated as 1. For short-term performance, an alternative that
has immediate effect after implementation is rated as 9 and an
alternative that would take a longer time to be effective is rated as
1. For long-term performance, an alternative that is effective for
longer durations (e.g., over 20 years) is rated as 9; otherwise, it is
rated close to 1. If an alternative is easy to implement, it is rated as
9; otherwise, it is rated at a lesser value. An alternative that has an
excellent effect on the reduction, mobility, and transport of contaminants is rated as 9; otherwise, it is rated at a lesser value. An
alternative that would increase future use and aesthetic of the site
is rated as 9.
Table 4 presents the score of the alternatives with respect to
each criterion. This table shows that the cost associated with Donothing (DN) is the lowest (1), Soil cover (SC) is high (6), and
Table 2
Alternatives considered and their effectiveness to reduce contaminants.
Alternatives
Effectiveness (%)
Sources
Do-nothing (DN)
Soil cover (SC)
Membrane system (MC)
Water cover (WC)
0
33
70
99
Excavation and on-site
disposal (EON)
Excavation and off-site
disposal (EOF)
Water treatment (WT)
Constructed wetland (CW)
99
e
MEND, 1999a
Meek, 1994
Vigneault et al., 2007;
Yanful and Simms, 1997
USEPA, 1995
99
USEPA, 1995
90
30
USEPA, 2000
MEND, 1999b; Skousen
et al., 1998
USEPA, 2006
Sulfate reducing
bacteria (SB)
80
Criteria
Score
Risk
Cost
Short-term performance (SP)
Long-term performance (LP)
Implementability (IM)
Reduction of toxicity (RT)
Future use and aesthetic (FU)
HQ
1 (Low) to 9 (Extremely high)
1 (Poor) to 9 (Excellent)
1 (Poor) to 9 (Excellent)
1 (impossible) to 9 (absolutely possible)
1 (Poor) to 9 (Excellent)
1 (Poor) to 9 (Excellent)
Excavation and off-site disposal (EOF) is the highest (9). The shortterm performance of DN is poor (1) but the Water cover (WC)
alternative has an excellent (9) short-term performance. The longterm performance of DN is poor (1) but the Excavation and on-site
disposal (EON) and EOF alternatives have an excellent (9) long-term
performance. The Membrane system (MC) alternative is difficult (2)
to implement, while the DN and Sulfate reducing bacteria (SB) alternatives are the easiest (9) to implement. The DN alternative is
the least effective (1) but EOF is most effective (9) in reducing
toxicity. The DN and SB alternatives have the least impact (1) on
improving the future use of this site, while EOF and EON have the
most impact on improving the future of this site.
The Hazard Quotient (HQ) value of the DN alternative was calculated using the EEC value obtained from the aquivalence model
output and the screening benchmark value. The aquivalence model
results showed that the EEC of water and soil was equal to 0.75 mg/l
and 0 mg/l of copper, respectively. The screening benchmark value
for water, which is equal to 0.003 mg/l, was obtained from CCME
(1999). Note that the HQ value was calculated only for water medium since concentrations of copper in the soil were equal to zero.
For the alternatives other than DN, the EEC value was adjusted by
the effectiveness factor (i.e., EECadjusted ¼ EEC EEC*Effectivenes
½%) when calculating the HQ value of alternatives. In addition, the
scoring shown in Table 3 is highly site specific and may not necessarily apply to other mine sites.
Table 5 presents a comparison matrix of the criteria. The comparison matrix shows the relative importance of one criterion
against another and reflects the preference of actors who participated in the decision making process. A pairwise comparison was
done on a scale of 1e9. For instance, a decision-maker strongly
prefers (9) the Risk criterion compared to the Future use and aesthetic (FU) criterion. The preference of FU over Risk (i.e., 0.11) was
obtained by taking the reciprocal preference value of Risk over FU.
The normalized weights of the criteria were obtained by computing
eigenvectors of the matrix. The remainder values of the matrix can
be explained in a similar fashion.
The input information for multicriteria methods (e.g., PROMETHEE) should be provided as an evaluation table as shown in
Table 6. This evaluation table consists of criteria; whether the
Table 4
Alternatives and their respective score at each criterion.
Criteriaa
DN
SC
MC
WC
EON
EOF
WT
CW
SB
Risk
Cost
SP
LP
IM
RT
FU
260
1
1
1
9
1
1
174
6
2
5
5
4
4
78
9
2
7
2
4
6
13
3
9
8
8
6
5
2.6
8
6
9
7
8
8
0
9
8
9
7
9
9
26
8
9
6
5
8
2
182
3
4
7
8
3
7
52
3
7
4
9
5
1
a
Do-nothing (DN), Soil cover (SC), Membrane system (MC), Water cover (WC),
Excavation and on-site disposal (EON), Excavation and off-site disposal (EOF), Water
treatment (WT), Constructed wetland (CW), Sulfate reducing bacteria (SB), Shortterm performance (SP), Long-term performance (LP), Implementability (IM),
Reduction of toxicity (RT), Future use and aesthetic (FU).
G.D. Betrie et al. / Journal of Environmental Management 119 (2013) 36e46
computed by evaluating the preference of one alternative over
another, using the preference function of each criterion. Next, the
obtained value was multiplied by the weight of each criterion and
summed. A preference index indicates the preference of an alternative over another. For instance, Do-nothing (DN) is preferred over
Soil cover (SC) by 0.13, while SC is preferred over DN by 0.248. A
higher preference value indicates a stronger preference for that
alternative.
The computed outgoing ðfþ Þ, incoming ðf Þ, and net ðfNET Þ
flows for the deterministic PROMETHEE analysis are shown in the
bottom of Table 7. The outgoing flow of an alternative was obtained
by summing the preference indexes in that row. For example, the
incoming flow of DN (0.967) was obtained by summing the preference indexes in the first row. On the other hand, the incoming
flow of an alternative was obtained by summing the preference
indexes in that column. For example, the incoming flow of DN
(4.095) was obtained by summing the preference indexes of in the
first column. The incoming and outgoing flows provide an indication of whether an alternative outranks, or is outranked, by other
alternatives, respectively. If an alternative is characterized by
a greater incoming flow and a smaller outgoing flow, the better this
alternative dominates other alternatives. These values were used
for partially ranking the alternatives. The net flow was obtained as
a net difference between the incoming and the outgoing flows and
it indicates whether an alternative globally outranks other alternatives. The higher the net flow of an alternative, the more this
alternative dominates other alternatives. The net flow is used to
completely rank the alternatives.
The deterministic partial ranking (PROMETHEE I) of the alternatives is shown below:
Table 5
Comparison matrix of criteria.
a
Risk
Cost
LP
SP
IM
RT
FU
Risk
Cost
LP
SP
IM
RT
FU
1
0.14
0.25
0.17
0.13
0.33
0.11
7
1
0.2
0.14
0.12
0.17
0.33
4
5
1
0.11
0.11
0.33
0.2
6
7
9
1
3
6
5
8
8
9
0.33
1
0.33
0.33
3
6
3
0.17
3
1
0.11
9
3
5
0.2
3
9
1
43
a
Long-term performance (LP), Short-term performance (SP), Implementability
(IM), Reduction of toxicity (RT), Future use and aesthetic (FU).
criteria have to be minimized (Min) or maximized (Max); criteria
weight that shows relative importance one criterion over other
criteria; the alternatives’ evaluation outcome with respect to each
criterion; preference function that gives the degree of preference of
an alternative against another alternative; and preference parameter value that has to be fixed. The criteria weights were obtained
from the AHP computation. The obtained results of AHP show that
Risk is the most prioritized (i.e., 0.41) criterion followed by Cost,
Long-term performance (LP), Reduction of toxicity (RT), Implementability (IM), Future use and aesthetic (FU), and Short-term
performance (SP). Type III (i.e., V-shape) preference function was
selected for all criteria. This preference type was selected because it
best represents the data compared to other preference functions.
The parameter value of type III function that was assumed for the
Risk criterion was equal to 230 and for the other criteria was equal
to 8.8.
3.4. Probabilistic MCDA
WCPI DN; WCPI SC; WCPI MC; WCPI EON; WCPI EOF; WCPI
WT; WCPI CW; WCPI SB;
A uniform probability distribution function was defined for the
weights of criteria. Inputs to this distribution were defined by
multiplying the original weight of each criterion by 10%. Each
probability distribution was randomly sampled 5000 times using
the MCS technique. For each combination of randomly sampled
weight, the MCDA method (PROMETHEE II) was run.
The results of criteria weights and PROMETHEE II outputs
obtained from 5000 MCS run were ranked and then correlation
coefficients were estimated using the Spearman rank correlation
method. The estimated correlation coefficients of each alternative
and criterion were normalized and multiplied by 100 to obtain the
contribution (in percent) of the criteria on overall ranking of the
alternatives.
SBPI DN; SBPI SC; SBPI MC; SBPI EON; SBPI WT; SBPI CW;
EOFPI DN; EOFPI SC; EOFPI MC; EOFPI WT; EOFPI CW;
EONPI DN; EONPI SC; EONPI MC; EONPI CW;
WTPI DN; WTPI SC; WTPI MC; WTPI CW;
CWPI DN; CWPI SC; CWPI MC; and
MCPI DN; MCPI SC;
The Water cover (WC) alternative dominates all the alternatives.
The Sulfate reducing bacteria (SB) alternative dominates all the
alternatives except WC and Excavation and off-site disposal (EOF);
the SB alternative is dominated by WC and is incomparable with
EOF. The EOF alternative dominates the Do-nothing (DN), Soil cover
(SC), Membrane cover (MC), Water treatment (WT) and Constructed wetland (CW) alternatives, is dominated by WC alternative, and is incomparable with the SB and Excavation and on-site
4. Results and discussion
Table 7 presents the computed preference indexes of the
deterministic PROMETHEE analysis. These preference indexes were
Table 6
Evaluation table for deterministic PROMETHEE analysis.
Criteriaa
Min or max
Criteria weight
DN
SC
MC
WC
EON
EOF
WT
CW
SB
Preference
function
Preference
parameter
Risk
Cost
SP
LP
IM
RT
FU
Min
Min
Max
Max
Max
Max
Max
0.41
0.23
0.02
0.16
0.06
0.08
0.04
260
1
1
1
9
1
1
174
6
2
5
5
4
4
78
9
2
7
2
4
6
13
3
9
8
8
6
5
2.6
8
6
9
7
8
8
0
9
8
9
7
9
9
26
8
9
6
5
8
2
182
3
4
7
8
3
7
52
3
7
4
9
5
1
III
III
III
III
III
III
III
230
8.8
8.8
8.8
8.8
8.8
8.8
a
Do-nothing (DN), Soil cover (SC), Membrane system (MC), Water cover (WC), Excavation and On-site disposal (EON), Excavation and Off-site disposal (EOF), Water
Treatment (WT), Constructed Wetland (CW), Sulfate Reducing Bacteria (SB), Short-term performance (SP), Long-term performance (LP), Implementability (IM), Reduction of
toxicity (RT), Future use and aesthetic (FU).
44
G.D. Betrie et al. / Journal of Environmental Management 119 (2013) 36e46
Table 7
Preference indexes, incoming, outgoing, and net flow for deterministic PROMETHEE analysis.
a
DN
SC
MC
WC
EON
EOF
WT
CW
SB
DN
SC
MC
WC
EON
EOF
WT
CW
SB
e
0.248
0.454
0.642
0.628
0.678
0.662
0.259
0.523
0.967
4.095
3.128
0.13
e
0.206
0.523
0.442
0.523
0.43
0.185
0.449
0.653
2.888
2.235
0.209
0.078
e
0.42
0.283
0.311
0.282
0.25
0.355
1.127
2.191
1.064
0.052
0.02
0.045
e
0.059
0.078
0.039
0.061
0.052
2.688
0.408
2.28
0.183
0.066
0.034
0.185
e
0.055
0.068
0.183
0.201
2.105
0.975
1.130
0.209
0.092
0.034
0.175
0.026
e
0.058
0.209
0.209
2.42
1.013
1.408
0.182
0.061
0.041
0.172
0.076
0.094
e
0.208
0.183
2.121
1.018
1.103
0
0.043
0.235
0.422
0.418
0.472
0.435
e
0.304
1.397
2.330
0.933
0
0.043
0.077
0.149
0.172
0.208
0.146
0.041
e
2.277
0.838
1.439
fþ
f
fNET
a
Do-nothing (DN), Soil cover (SC), Membrane system (MC), Water cover (WC), Excavation and on-site disposal (EON), Excavation and off-site disposal (EOF), Water
treatment (WT), Constructed wetland (CW), Sulfate reducing bacteria (SB).
disposal (EON) alternatives. The EON alternative dominates the DN,
SC, MC and CW alternatives, is dominated by the WC and SB alternatives, and is incomparable with EOF and WT. The WT alternative dominates the DN, SC, and CW alternatives, is dominated by
the WC, SB, and EOF alternatives, and is incomparable with the EON
and MC alternatives. The CW alternative dominates the DN and SC
alternatives, is incomparable with MC, and is dominated by the WC,
SB, EOF, EON and WT alternatives. The DN and SC alternatives are
incomparable. The alternatives are incomparable whenever they
have conflicting relative rankings. Incomparability between the
alternatives occurs if an alternative is superior according to some
criteria and inferior according to other criteria compared to other
alternatives. For instance, the incomparability of SB and EOF occurs
because the EOF alternative performs better on the Risk, Long-term
performance (LP), Reduction of toxicity (RT) and Future use and
aesthetic (FU) criteria than does the SB alternative, whereas the SB
alternative performs better on Cost and Implementability (IM)
criteria.
The deterministic complete ranking (PROMETHEE II) shows that
WC ranked first followed by the SB, EOF, EON, WT, CW, MC, SC, and
DN alternatives. It is interesting to note that WC and DN have the
same rank in both partial and complete ranking methods. However,
the incomparability information is lost in the complete ranking
method which may affect decision-makers ability to make
informed decision.
The results of the probabilistic complete ranking of the alternatives are displayed in Table 8. This table shows the probability of
an alternative obtaining a certain rank. The WC alternative obtains
the first rank with a probability of 100%. The chances of the SB
alternative ranking second and third are 70% and 30%, respectively,
whereas the chances of the EOF alternative ranking second and
third are 30% and 70%, respectively. The probability of the EON
alternative obtaining fourth and fifth ranks is 67% and 33%,
respectively, whereas the WT alternative has a probability of 33%
and 67% of obtaining the fourth and fifth ranks, respectively. The
chance of the CW alternative ranking sixth and seventh is 97% and
3%, respectively, whereas the MC alternative ranks similar to the
CW alternative with the values of probability reversed. The SC and
DN alternatives rank eighth and ninth with a probability of 100%. It
is interesting to note that this method clearly showed the degree of
conflict in relative ranking that led to incomparability in the
deterministic partial ranking which is presented above.
Fig. 6 shows the cumulative distribution functions of the net
flow for the probabilistic PROMETHEE analysis. It shows a possible
range of net flow for the alternatives. For example, the net flow of
SB has a possible range of between 1.305 and 1.573, and the EOF
alternative has a possible range of between 1.24 and 1.569. It is
interesting to compare the probabilistic net flow values with the
deterministic net flow results. The information obtained from the
probabilistic analysis explains the reason for the conflicting relative
ranking among alternatives (e.g., SB vs. EOF and EON vs. WT). This
probabilistic output enables decision-makers to obtain net flow
values at the required confidence levels and to rank the alternatives
with a certain level of confidence.
The contribution (in percent) of the criteria on ranking the alternatives is shown in Fig. 7. For instance, the three most sensitive
criteria that impact the WC alternative are Risk, SP and Cost.
Although SP has the least weight among the criteria, its effect is
greater than the Cost criterion which is the second most highly
weighted criterion. It is worth noting that the ranking of each
alternative could be sensitive to different criteria as is seen in this
figure. For instance, the SB alternative is sensitive to Cost followed
by Risk, SP, FU, IM, and RT criteria, and the CW alternative is sensitive to Risk followed by Cost, RT, LP, SP, FU and IM criteria.
Table 8
Probabilistic complete ranking of alternatives in percentage.
Rank
a
1
2
3
4
5
6
7
8
9
0
0
0
0
0
0
0
0
100
DN
SC
MC
WC
EON
EOF
WT
CW
SB
0
0
0
0
0
0
0
100
0
0
0
0
0
0
3
97
0
0
100
0
0
0
0
0
0
0
0
0
0
0
67
33
0
0
0
0
0
30
70
0
0
0
0
0
0
0
0
0
33
67
0
0
0
0
0
0
0
0
0
97
3
0
0
0
70
30
0
0
0
0
0
0
a
Do-nothing (DN), Soil cover (SC), Membrane system (MC), Water cover (WC),
Excavation and on-site disposal (EON), Excavation and off-site disposal (EOF), Water
treatment (WT), Constructed wetland (CW), Sulfate reducing bacteria (SB).
Fig. 6. Cumulative distribution function of net flow for probabilistic PROMETHEE
analysis.
G.D. Betrie et al. / Journal of Environmental Management 119 (2013) 36e46
45
makers will be able to leave insensitive criteria for further analysis
in order to reduce the complexity of decision making. Meanwhile,
they can collect more information on the most sensitive criteria in
order to reduce the uncertainty associated with the criteria. The
study also showed that the assigned weight of criteria has little effect on ranking the alternatives. However, a limitation of the
improved framework is that it addresses the uncertainty associated
with the weight of the criteria only. Future work should be done to
address the uncertainty associated with input information. In its
present form, this improved framework could be used by decisionmakers to select remediation alternatives for mine sites and to
select alternatives for mine closures. Moreover, the improved
framework could be used in other systems (e.g., contaminated sites)
either in its present form or by making a little modification. In the
latter case, the improved framework could be used in other systems
that use multicriteria decision analysis by replacing the PROMETHEE method with the MCDA method in use to fit their systems.
Fig. 7. Contribution (%) of criteria on ranking of alternatives.
5. Summary and conclusions
In this paper, the existing framework used for selecting remediation alternatives at mine sites was investigated and improved.
The major improvement of the framework includes an introduction
of the deterministic and probabilistic multicriteria decision analysis
(MCDA) methods. The multicriteria decision analysis consists of
identifying criteria to evaluate alternatives, calculating criteria
weight using analytical hierarchical process (AHP), defining the
probability distribution of criteria weights, conducting Monte Carlo
Simulation (MCS), applying Preference Ranking Organization
Method for Enrichment Evaluation (PROMETHEE) I and II methods,
conducting a sensitivity analysis, determining the deterministic
and probabilistic rankings of alternatives, and identifying the sensitivity of the criteria. Finally, a case study was used to demonstrate
the improved framework.
The results of the deterministic PROMETHEE I analysis showed
and ranked the comparable alternatives, and showed the incomparable alternatives. The deterministic PROMETHEE II results only
showed rankings of the alternatives without incomparability information. For instance, the application of the deterministic
PROMETHEE II method at Mine B showed that the WC (Water
cover) alternative ranks first followed by Sulfate reducing bacteria
(SB), Excavation and off-site disposal (EOF), Excavation and on-site
disposal (EON), Water treatment (WT), Constructed wetland (CW),
Membrane system (MC), Soil cover (SC) and Do-nothing (DN) alternatives. In both methods, the uncertainty in the ranking of the
alternatives due to input information variation was not shown to
decision-makers. On the other hand, the probabilistic PROMETHEE
II results showed the probability level at which an alternative could
yield certain rankings. For instance, the application of the probabilistic complete ranking at Mine B showed that WC ranked 1st
with a probability level of 100%, SB and EOF ranked 2nd and
3rd with a probability level of 70%, EON and WT ranked 4th and 5th
with a probability level of 67%, CW and MC ranked 6th and 7th with
a probability level of 97%, and SC and DN ranked 8th and 9th with
a probability level of 100%. Also, this method was used to rank the
alternatives in the form of a cumulative distribution function that
will help decision-makers rank alternatives with the required level
of confidence.
The sensitivity analysis showed the impact of the criteria on
ranking the alternatives. For instance, the Risk and Future use and
aesthetic (FU) criteria impacted the WC alternative by 44% and 2%,
respectively. Based on the contribution of the criteria, decision
Acknowledgments
The financial support from Discovery Grant program of NSERC is
highly appreciated. We are grateful to Dr. Carolyn Labun, the director of Center for Scholarly Communication at the University of
British Columbia for editing this manuscript.
References
Aral, M.M., 2010. Environmental Modeling and Health Risk Analysis (ACTS/RISK).
Springer, New York.
Azapagic, A., 2004. Developing a framework for sustainable development indicators
for the mining and minerals industry. J. Clean. Prod. 12, 639e662.
Belton, V., Stewart, T., 2002. Multicriteria Decision Analysis: an Integrated
Approach. Kluwer, Boston.
Betrie, G.D., Sadiq, R., Kevin, A.M., Tesfamariam, S., 2012. Uncertainty analysis of an
aquivalence-based fate and transport model using a hybrid fuzzy-probabilistic
approach. In: Proceedings of 12th International Environmental Specialty Conference, Edmonton, AL, Canada.
Brans, J.P., Vincke, P.H., Mareschal, B., 1986. How to select and how to rank projects:
the PROMETHEE method. Eur. J. Oper. Res. 24, 228e238.
Brans, J.P., Mareschal, B., 2005. PROMETHEE methods. In: Figueria, J., Greco, S.,
Ehrgott, M. (Eds.), Multiple Criteria Decision Analysis: State of the Art Survey.
Springer Science þ Business Media, Inc., Boston, pp. 21e36.
CCME (Canadian Council of Ministers of the Environment), 1999. Canadian Environmental Quality Guidelines. Available at: http://www.ccme.ca/publications/
ceqg_rcqe.html.
Figueria, J., Greco, S., Ehrgott, M., 2005. Introduction. In: Figueria, J., Greco, S.,
Ehrgott, M. (Eds.), Multiple Criteria Decision Analysis: State of the Art Survey.
Springer Science þ Business Media, Inc., Boston, pp. 21e36.
Gray, N.F., 1996. Field assessment of acid mine drainage contamination in surface
and ground water. Environ. Geol. 27, 358e361.
Guitouni, A., Martel, J., 1998. Tentative guidelines to help choosing an appropriate
MCDA method. Eur. J. Oper. Res. 109, 501e521.
Huang, I.B., Keisler, J., Linkov, I., 2011. Multi-criteria decision analysis in environmental sciences: ten years of application and trends. Sci. Total Environ. 409,
3578e3594.
Johnson, D.B., Hallberg, K.B., 2005. Acid mine drainage remediation options: a review. Sci. Total Environ. 338, 3e14.
Kiker, G.A., Bridges, T.S., Varghese, A., Seager, T.P., 2005. Application of multicriteria
decision analysis in environmental decision making. Integr. Environ. Assess.
Manag. 1 (2), 95e108.
Keeney, R.L., Raiffa, H., 1993. Decisions with Multiple Objectives: Preferences and
Value Trade-offs. Cambridge University Press, Cambridge.
Mackay, D., Joy, M., Paterson, S., 1983. A quantitative water, air, sediment interaction
(QWASI) fugacity model for describing the fate of chemicals in lakes. Chemosphere 12, 277e291.
Mackay, D., Diamond, M., 1989. Application of the QWASI (Quantitative water air
sediment interaction) fugacity model to the dynamics of organic and inorganic
chemicals in lakes. Chemosphere 18, 1343e1365.
Mackay, D., 1991. Multimedia Environmental Models: the Fugacity Approach. Lewis
Publishers, Chelsea.
McDaniels, T.L., Gregory, R.S., Fields, D., 1999. Democratizing risk management:
successful public involvement in local water management decisions. Risk Anal.
19, 497e510.
Meek, F.A., 1994. Evaluation of acid prevention techniques used in surface mining.
In: Proceedings of International Land Reclamation and Mine Drainage Conference, Pittsburgh, Pennsylvania.
46
G.D. Betrie et al. / Journal of Environmental Management 119 (2013) 36e46
MEND (Canadian Mine Environment Neutral Drainage), 1999a. Construction and
Instrumentation of a Multi-layer Cover. Report 2.22.4a.
MEND (Canadian Mine Environment Neutral Drainage), 1999b. Review of Passive
Systems for Treatment of Acid Mine Drainage. MEND Report 3.14.1.
Minewatch, 2006. Acid Mine Drainage: Mining and Water Pollution Issues. Available
at: http://www.miningwatch.ca/sites/www.miningwatch.ca/files/ARD_0.pdf.
Morin, K.A., Hutt, N.M., 1997. Environmental Geochemistry of Minesite Drainage:
Practical Theory and Case Studies. MDAG Publishing, Vancouver.
Price, W.A., 2009. Prediction Manual of Drainage Chemistry from Sulphidic Geologic
Materials. Canadian Mine Environment Neutral Drainage (MEND). Report 1.20.1.
Rietveld, P., Ouwersloot, H., 1992. Ordinal data in multicriteria decision making:
a stochastic dominance approach to siting nuclear power plants. Eur. J. Oper.
Res. 56, 249e262.
Saaty, T., 1982. Decision-making for Leaders: the Analytic Hierarchy Process for
Decision in a Complex World. Lifetime Learning, Belmont, Calif.
Sadiq, R., Tesfamariam, S., 2009. Environmental decision-making under uncertainty
using intuitionistic fuzzy analytic hierarchy process (IF-AHP). Stoch. Environ.
Res. Risk Assess. 23, 75e91.
Skousen, J., Rose, A., Geidel, G., Foreman, J., Evans, R., Hellier, W., 1998. Handbook of
Technologies for Avoidance and Remediation of Acid Mine Drainage. The National Mine Land Reclamation Center Located at East Virginia University in
Morgantown, West Virginia.
Skousen, J.G., Sexstone, A., Ziemkiewicz, P.F., 2000. Acid mine drainage control and
treatment. In: Branhisel, R.I., Darmody, R.G., Daniels, W.L. (Eds.), Reclamation of
Drastically Disturbed Lands. American Society of Agronomy, Madison, pp. 1e42.
Tesfamariam, S., Sadiq, R., 2006. Risk-based environmental decision-making using
fuzzy analytic hierarchy process (F-AHP). Stoch. Environ. Res. Risk Assess. 21,
35e50.
Tesfamariam, S., Rajani, B., Sadiq, R., 2006. Possibilistic approach for consideration
of uncertainties to estimate structural capacity of ageing cast iron water mains.
Can. J. Civ. Eng. 33, 1050e1064.
Tesfamariam, S., Sadiq, R., 2008. Probabilistic risk analysis using ordered weighted
averaging (OWA) operators. Stoch. Environ. Res. Risk Assess. 22 (1), 1e15.
USEPA (US Environmental Protection Agency), 2006. US EPA Engineering Issue
Management and Treatment of Water from Hard Rock Mines. Available at:
http://nepis.epa.gov/Exe/ZyPURL.cgi?Dockey¼600006RV.txt.
USEPA (US Environmental Protection Agency), 1994. Technical Document of Acid
Mine Drainage Prediction. EPA530-R-94-036.
USEPA (US Environmental Protection Agency), 1995. Contaminants and Remedial
Options at Selected Metal-contaminate Sites. EPA/540/R-95/512.
USEPA (US Environmental Protection Agency), 1988. Guidance for Conducting
Remedial Investigations and Feasibility Studies Under CERCLA. EPA/540/G89/004.
USEPA (US Environmental Protection Agency), 1997. Ecological Risk Assessment
Guidance for Superfund: Process for Designing and Conducting Risk Assessments. EPA 540-R-97-006.
USEPA (US Environmental Protection Agency), 2000. Abandoned Mine Site Characterization and Cleanup Handbook. EPA 910-B-00-001.
Vigneault, B., Kwong, Y.T.J., Warren, L., 2007. Assessing the Long Term Performance
of a Shallow Water Cover Limit Oxidation of Reactive Tailings at Louvicourt
Mine. MEND Report 2.12.2.
Yang, A.L., Huang, G.H., Qin, X.S., Fan, Y.R., 2012. Evaluation of remedial options for
a benzene-contaminated site through a simulation-based fuzzy-MCDA
approach. J. Hazard. Mater. 213e214, 421e433.
Yanful, E.K., Simms, P., 1997. Review of Water Cover Sites and Research Projects.
MEND Project 2.18.1.
List of acronyms
MCDA: multicriteria decision analysis
PROMETHEE: Preference Ranking Organization Method for Enrichment Evaluation
AHP: analytical hierarchical process
MCS: Monte Carlo Simulation
ARD: Acid Rock Drainage
MAU/VT: multi-attribute utility/value theory
ELECTRE: elimination et choix traduisant la realite
ai: alternative i
gi: criterion i
COC: chemical of concern
DWS: Diffusion from water-to-soil
DSW: Diffusion from soil-to-water
DAS: Advection in soil
DAW: Advection groundwater
HQ: Hazard Quotient
EEC: Estimated (maximum) environmental contaminant concentration
DN: Do-nothing
SC: Soil cover
MC: Membrane system
WC: Water cover
EON: Excavation and on-site disposal
EOF: Excavation and off-site disposal
WT: Water treatment
CW: Constructed wetland
SB: Sulfate reducing bacteria
SP: Short-term performance
LP: Long-term performance
IM: Implementability
RT: Reduction of toxicity
FU: Future use and aesthetic
Min: Minimized
Max: Maximized