Carlos
Francisco Simões Gomes
Department
of Production Engineering, Fluminense Federal University, Brazil
Email: cfsg1@bol.com.br
Luiz
Flavio Autran Monteiro Gomes
Ibmec
School of Business and Economics, Ibmec University Center, Brazil
Email: luiz.gomes@professores.ibmec.edu.br
Luís
Alberto Duncan Rangel
Department
of Production Engineering, Fluminense Federal University, Brazil
Email: luisduncan@id.uff.br
Fabricio
Maione Tenório
Federal
Center of Technological Education Celso Suckow da Fonseca, Brazil.
Email:
fabricio.tenorio@cefetrj.br
Marcos
dos Santos
Military
Engineering Institute, Brazil
Email:
marcosdossantos@ime.eb.br
Submission: 11/6/2020
Revision: 12/16/2020
Accept: 12/1/2020
ABSTRACT
This paper approaches
the problem of ballast water treatment in ships. This has been identified as
one of the four greatest threats to the world’s oceans. Solutions that have
been considered for solving the problem are alternative water treatment
technologies. In the case study reported in this paper three major water
treatment technologies have been evaluated with the help of twentysix criteria, quantitative as well as
qualitative by using two discrete multicriteria
methods, TODIM and THOR 2. The THOR 2 consists of the axiomatic evolution of
the THOR method and both THOR 2 and THOR are made available through the THOR
Web platform. Five groups of evaluation criteria are then considered:
practicality; biological effectiveness; cost/benefit ratio; time frame for the
implementation of standards; and environmental impact of the process'
subproducts. In this paper a case study on choosing a ballast water treatment
technology is presented. Three alternative ballast water management technologies
are proposed by experts in the field and are evaluated with the help of
twentysix criteria, quantitative as well as qualitative. Each ballast water
management method is described by a list of twentysix attributes or criteria. After
setting the problem in a clear way and consulting different experts, the two
separate applications of both TODIM and THOR 2 are performed. What is denoted
as Management Method #1 is indeed chosen as the best alternative according to
both methods. The conclusion is that those two methods, although conceptually
and analytically quite different, lead essentially to the same main results. Two other applications of both TODIM and THOR
have indeed confirmed the convergence of results in spite of the conceptual and
technical differences between the two methods. This suggests that formulating a
decision problem in a correct, clearcut way can be at least as important as
the technical characteristics of the method per se.
Keywords: Maritime transportation; Water pollution, TODIM, THOR 2, MultiCriteria Decision Analysis
1.
INTRODUCTION
A number of research efforts on
ballast water have been carried out due to its global importance (IMO, 2020a).
The introduction of invasive marine species into new environments by ships'
ballast water attached to ships' hulls and via other vectors has been
identified as one of the four greatest threats to the world's oceans. Shipping
moves over 80% of the world's commodities and transfers approximately 3 to 5
billion tons of ballast water internationally each year. A similar volume may also
be transferred domestically within countries and regions each year. Ballast
water is essential to the safe and efficient operation of modern merchant
ships, providing balance and stability to unloaded ships. However, it may also
pose a serious ecological, economic and health threat.
Here it is worth observing that
ballast is any material used to weight and/or balance an object. One example
are the sandbags carried on conventional hotair balloons, which can be
discarded to lighten the balloon's load, allowing it to ascend. Ballast water
is therefore water carried by ships to ensure stability, trim and structural
integrity. Ships have carried solid ballast, in the form of rocks, sand or
metal, for thousands of years. In modern times, ships use water as ballast.
It is estimated that at least 7,000
different species are being carried in ships' ballast tanks around the world.
Research on various types of marine life has been conducted by different authors
(Tenenbaum et al., 2004). The vast majority of marine
species carried in ballast water do not survive the journey, as the ballasting
and deballasting cycle and the environment inside
ballast tanks can be quite hostile to organism survival. Even for those that do
survive a voyage and are discharged, the chances of surviving in the new
environmental conditions, including predation by and/or competition from native
species are further reduced. However, when all factors are favorable, an
introduced species can survive to establish a reproductive population in the
host environment, becoming invasive, outcompeting native species and
multiplying into pest proportions (IMO, 2020b).
The socalled
ballast water cycle comprises the following stages: (1) the ship is unloaded
and therefore with no ballast, the sea water then functioning as ballast water;
(2) the ship moves with ballast water taken from the starting port; (3) the
ship discharges ballast water and places exogenous marine life in the
destination port; (4) the ship is loadless and does
not require ballast water anymore.
As the situation becomes more and
more serious, the International Maritime Organization (IMO) has sponsored
international meetings to find out courses of action to meet this challenge,
where the subject is discussed by the IMO Member States.
As a result, whole ecosystems are
being changed. In the USA, the European zebra mussel dreissena
polymorph has infested over 40% of internal waterways and may have required
between US$ 750 million and US$ 1 billion in expenditure on control
measures between 1989 and 2000. In southern Australia, the Asian kelp undaria pinnatifida is invading
new areas rapidly, displacing the native seabed communities. In the Black Sea,
the filterfeeding North American jellyfish mnemiopsis
leidyi has on occasion reached densities of 1kg of
biomass per m2. It has depleted native plankton stocks to such an extent that
it has contributed to the collapse of entire Black Sea commercial fisheries.
In several countries, introduced,
microscopic, 'redtide' algae (toxic dinoflagellates) have been absorbed by
filterfeeding shellfish, such as oysters. When eaten by humans, these
contaminated shellfish can cause paralysis and even death. The list goes on,
with hundreds of examples of major ecological, economic and human health
impacts across the globe. It is even feared that diseases such as cholera might
be able to be transported in ballast water.
Invasive marine species are one of
the four greatest threats to the world's oceans. The other three are landbased
sources of marine pollution, overexploitation of living marine resources and
physical alteration/destruction of marine habitat. Unlike other forms of marine
pollution, such as oil spills, where ameliorative action can be taken and from
which the environment will eventually recover, the impacts of invasive marine
species are most often irreversible (IMO, 2020a).
In this
paper a case study on choosing a ballast water treatment technology is
presented. Three alternative ballast water management technologies are proposed
by experts in the field and are evaluated with the help of twentysix criteria,
quantitative as well as qualitative. The problem is tackled by a multicriteria approach and two methods are utilized.
Results from applying those two methods are the compared.
2.
CASE STUDY
In
2002 a group of Brazilian scientists and engineers comprising fifteen civilian
and four military experts and including members of the Brazilian delegation to
IMO’s International Conference on Ballast Water Management for Ships was asked
to demonstrate the feasibility of evaluating alternative ballast water
treatment technologies and to produce a ranking of those by using Multicriteria Analysis. The technical leadership was with
one of the authors of this paper; their first challenge was to design the
evaluation process. The multicriteria methods to be
used should allow incorporating their value judgments based on their
experiences as well as the experiences of IMO Member States. The evaluation
process should be taken as a learning process. Finally, it should also provide
a recommendation for selecting the best ballast water exchange and treatment
methods.
The steps below were
followed:
· Step
1: identify in all proposals submitted by IMO Member States the relevant
criteria;
· Step
2: submit this set of criteria to IMO Member States;
· Step
3: obtain the consensus about the criteria set;
· Step
4: identify the alternatives that solve the problem;
· Step
5: submit the alternatives to IMO Member States;
· Step
6: identify the importance to criteria by their relative weights;
· Step
7: order the alternatives by both TODIM and THOR 2.
Two multicriteria methods, TODIM and THOR 2, were used in the
case study as separate analytical tools in this case study. After applying both
methods, their results are compared. In order to apply this methodology to the
case under consideration, relevant groups of criteria have been identified.
They are: Practicality; Biological effectiveness (including pathogens);
Cost/benefits; Time frame within which the standards could be practically
implemented; and Environmental impact of the process' subproducts.
3.
METHODOLOGY
At
the end of the 1970s, in the midst of the need for answers that covered several
dimensions for decisionmaking processes, a new field of Management Sciences
emerged, MultiCriteria Decision Analysis (MCDA). Decision problems are often
complex in nature, as they imply the consideration of multiple assessment
criteria, uncertainties, conflicts of values and interests, asymmetries of
power, and a large volume of data and information that, in turn, are often
incomplete. This has been precisely the domain of action of MCDA (Ehrgott et al., 2010).
The
various analytical methods of MCDA apply to all those complex decision problems
that fit into at least one of the following four problematiques:
choice, ranking, sorting, or description (ROY, 1996). Among these various multicriteria methods, there is one that, based on a robust
psychological foundation, both in its original formulation and in extensions,
has allowed solving problems inserted in those problematiques,
the TODIM method.
The
TODIM method was the very first methodological contribution of Brazilian
researchers to the field of MCDA. The embryo of TODIM was a multicriteria
ranking technique introduced at the end of the 80’s (Gomes, 1989a; Gomes,
1989b; Gomes, 1990a; Gomes, 1990b). In 1991 the creator of such technique
decided to provide a psychological basis for ranking alternatives under
multiple criteria and this innovation led to the TODIM method as it is known
today. Thus TODIM (an acronym in Portuguese for Interactive and Multicriteria Decision Making) (Gomes & Lima, 1991; Gomes
& Lima, 1992; Sudha et al., 2020; Zindani et al., 2020) is a discrete multicriteria
method founded on Prospect Theory (Kahneman & Tversky, 1979).
While
all other discrete multicriteria methods assume that
the decision maker always looks for the solution corresponding to the maximum
of some global measure of value – for example, the highest possible value of a multiattribute utility function, in the case of MAUT
(Keeney& Raiffa, 1993) – TODIM makes use of a
global measurement of value calculable by the application of Prospect Theory.
In this way, TODIM is based on a description, demonstrated by empirical
evidence, of how people effectively make decisions in the face of risk.
Although
not all multicriteria problems deal with risk, the
shape of the value function of TODIM is the same as the value function of
Prospect Theory. The multiattribute value function of
TODIM is built in parts, with their mathematical descriptions reproducing that
gain/loss function. The global multiattribute value
function of TODIM therefore aggregates all measures of gains and losses over
all criteria (Gomes & Rangel, 2009; Gomes, Rangel &Maranhão,
2009; Gomes et al., 2013).
The
concept of introducing expressions of losses and gains in the same multiattribute function, present in the formulation of
TODIM, gives this method some similarity to the PROMÉTHÉE methods, which make
use of the notion of net outranking flow (Brans & Mareschal,
1990). TODIM indeed maintains a similarity with outranking methods, because the
global value of each alternative is relative to its dominance over other
alternatives in the set. In its calculations the TODIM method is suitable to
test specific forms of the losses and gains functions. Each one of the forms
depends on the value of one single parameter. The forms, once validated
empirically, lead to the additive difference function of the method. This
notion of an additive utility function is borrowed from Tversky
(Tversky, 1969).
From
the construction of that additive difference function, which performs as a multiattribute value function and, as such, must also have
its use validated by verifying the condition of mutual preferential
independence (Clemen & Reilly, 2001), the method
leads to a global ordering of the alternatives. It can be observed that the
construction of the multiattribute value function, or
additive difference function, of the TODIM method is based on a projection of
the differences between the values of any two alternatives (perceived in
relation to each criterion) to a reference criterion.
The
TODIM method makes use of paired comparisons between the decision criteria,
using technically simple resources to eliminate occasional inconsistencies
arising from these comparisons. It also allows value judgments to be carried
out in a verbal scale, using a criteria hierarchy, fuzzy value judgments and
making use of interdependence relationships among the alternatives. It is a
noncompensatory method in the sense that tradeoffs are not dealt with in the
modeling process (Bouyssou, 1986).
Consider
a set of n alternatives to be ordered in the presence of m quantitative
or qualitative criteria and assume that one of those criteria can be considered
as the reference criterion. After the definition of these elements, experts are
asked to estimate, for each one of the qualitative criteria c, the
contribution of each alternative i to the
objective associated with the criterion.
This
method requires the values of the evaluation, of the alternatives in relation
to the criteria, to be numerical and to be normalized; consequently, the
qualitative criteria evaluated in a verbal scale are transformed into a
cardinal scale. The evaluations of the quantitative criteria are obtained from
the performance of the alternatives in relation to the criteria, such as, for
example, the level of noise measured in decibels, the power of an engine
measured in horsepower or a student’s mark in a subject etc.
Being
an MCDA tool, TODIM was conceived be used with qualitative as well as
quantitative evaluation criteria simultaneously considered. Verbal scales of
qualitative criteria are converted to cardinal ones and both types of scales
are normalized. The relative measure of dominance of one alternative over
another is found for each pair of alternatives. This measure is computed as the
sum over all criteria of both relative gain/loss values for these alternatives.
The parts in this sum will be either gains, losses, or zeros, depending on the
performance of each alternative with respect to every criterion.
The
evaluation of the alternatives in relation to all the criteria produces the
matrix of evaluation, where the values are all numerical. Their normalization
is then performed, using, for each criterion, the division of the value of one
alternative by the sum of all the alternatives. This normalization is carried
out for each criterion, thus obtaining a matrix, where all the values are
between zero and one. It is called the matrix of normalized alternatives’
scores against criteria.
The
application of TODIM then proceeds towards the computation of the overall value
for each alternative. This is accomplished by making use of expressions for the
gain/loss branches of the prospect theoretical value function. An important parameter
of TODIM is q, the attenuation factor of the
losses; different choices of q lead to different shapes of the
prospect theoretical value function in the negative quadrant.
The
global measures obtained of alternatives’ values computed by TODIM permit the
complete rank ordering of all alternatives. A sensitivity analysis should then
be applied to verify the stability of the results based on the decision makers’
preferences. The sensitivity analysis should therefore be carried out on q as well as on the criteria weights, the choice
of the reference criterion, and performance evaluations.
Mathematical
expressions (1), (2) and (3) constitute the essential modeling underlying the
use of TODIM:
(1)
(2)
(3)
where:
d(A_{i}, A_{j}), measurement of
dominance of alternative A_{i} over alternative A_{j};
m, the number of criteria;
c, a generic criterion;
w_{rc,} tradeoff rate between any criterion taken as a
reference criterion r and any other, generic
criterion c;
P_{ic} and P_{jc}, evaluations of alternatives i and j with respect to criterion c;
q, attenuation factor of the losses; different
choices of q lead to different shapes of the
prospect theoretical value function in the negative quadrant.
F_{c}(A_{i}, A_{j}), contribution of
criterion c to function d(A_{i}, A_{j}), when
comparing alternatives A_{i} and A_{j}.
ξ_{i, }normalized global performance of alternative A_{i},
when compared against all other alternatives.
The
function Fc reproduces the Prospect Theory
value function and replicates the most relevant shape characteristics. First,
it fulfills the concavity for positive outcomes (convexity for negative
outcomes) and second, it enlarges the perception of negative values for losses
than positive values for gains, both value functions are steeper for negative
outcomes than for positive ones. It is worth observing that, together with the
value function, these two authors introduced the weighting function that
measures the subjective perception of probabilities. As TODIM is a
deterministic method in its original formulation, only the value function is
extended.
Each
shape characteristic of the value function models psychological processes: The
concavity for gains describes a risk aversion attitude, the convexity describes
a risk seeking attitude; secondly, the assumption that losses carry more weight
than gains is represented by a steeper negative function side.
Different
kinds of decision makers can be understood in terms of their risk and loss
attitude. Although the TODIM method does not deal with risk directly, the way
the decision maker evaluates the outcomes of any decision can be expressed by
their risk attitude: for instance, a cautious decision maker will undervaluate
a superior result more than a braver one. Apart from parameter θ, the
attenuation factor of the losses, function Φ does not offer other
parameters to delineate the behavior of diverse decision makers, therefore a
generic formulation is proposed.
It is worth noticing that as soon as
TODIM appeared in the MCDA literature the fact that this method has elements
from both the socalled French and North American School was pointed out by Roy
and Bouyssou (1993).
The THOR 2 consists of the axiomatic evolution of the THOR
method (Tenório, 2020). THOR was the second MCDA tool created by a
Brazilian researcher and was conceived as the essential product of a doctoral
thesis (Gomes, 1999). THOR is an analytical approach that relies on concepts of
Rough Set Theory, Fuzzy Set Theory and Preference Theory (Gomes et al., 2008). The mathematical model embedded in THOR comes by a
combination those three theories and allows for the quantification of the
attractiveness of each alternative by creating
a nontransitive aggregation function (Costa et al., 2020).
THOR is
therefore a tool for ranking discrete alternatives under multiple criteria, by
eliminating redundant criteria and by reducing imprecision along the decision
process. The concept of quantifying the imprecision associated to the weights
and to the classification of the alternatives put into operation in THOR arises
from the fact that the judgment values, because of their inherent subjectivity,
cannot always be expressed in crisp ways. When using THOR, the simultaneous
input of data into the process from multiple decision makers is also allowed,
enabling those to express their judgment values in scales of ratios, intervals
or ordinals, in addition to the execution of the decision making process
without necessarily assigning weights to criteria (Gomes, 2005).
The
analytical modeling embedded in THOR is based on the ELECTRE methods of the
French School of MCDA. The following additional elements are then necessary for
the application of THOR: a weight for each criterion, representing the relative
importance among them; a preference threshold (p) and another for
indifference (q) for each criterion; discordance; and pertinence of the
values of the weights attributed to the criterion, as well as the pertinence of
the classification of the alternative in the criterion.
Since
its creation in 1999, THOR has been used for solving a diversity of problems
within a multicriteria framework. Table 1 lists references to journal articles making use of THOR.
Table 1: Articles were THOR is used
and types of applications
Authors 
Methods 
Types of application 
Gomes (2005) 
THOR, ELECTRE, PROMETHÉE II, ELECTRE III, AHP 
Analysis of a ballast
water treatment system 
Gomes (2006) 
THOR 
Construction of an energy
plant and selection of a site for procurement 
Gomes et al.
(2008) 
THOR 
Disposition of plastic
and construction waste 
Cardoso et
al. (2009) 
THOR 
Destination of plastic
and packaging waste 
Gomes et al.
(2010) 
TODIM e THOR 
Destination of natural gas 
Gomes and Maia
(2013) 
THOR 
Choice of biomass for
producing renewable energy 
Gomes and Costa
(2015) 
THOR, ELECTRE I, ELECTRE II, PROMETHÉE II 
Choice of credit card
technology 
Tenório et
al. (2020a) 
THOR 
Strategy for buying a
ship for the Brazilian Navy 
Tenório et
al. (2020b) 
THOR 
Selection of a ship for
the Brazilian Navy 
Costa et al.
(2020) 
THOR 2 
Choice of a ship for
medical assistance for COVID19 
Given two alternatives a and b,
three situations can be considered when THOR is used: S_{1}, S_{2}
and S_{3}. Situation S_{1} only considers the alternatives a
for which aPb , with b being any other
alternative. In this way, comparing a with b, we can identify the
criteria in which aPb, taking into consideration the
thresholds of preference (P designates strict preference and Q
designates weak preference), indifference and discordance, checking if the
condition imposed is satisfied. If satisfied, we know that a dominates b.
The binary relations P, Q, and I are defined as below.
Equations (4), (5) and (6) designate the binary relations P, Q
and I, respectively, where the notation g(.) designates a criterion:
aPb ↔ g(a)  g(b) > p (4)
aQb ↔ q < g(a)g(b)
≤ p (5)
aIb ↔ q ≤ g(a)g(b)
≤ q (6)
We
can therefore write Equation (7) for Situation 1 (or S_{1}):
(7)
The
context S_{1} is characterized by the following fact: the sum of the
weights of the criteria j such what a is strongly preferable b
is bigger than the sum of the weights of the criteria j such what a
is weakly preferable b more the sum of the weights of the criteria j
such what a is indifferent b more the sum of the weights of the
criteria j such what a is not comparable with b any more
the sum of the weights of the criteria j such what b is weakly
preferable a to any more the sum of the weights of the criteria j
such what b is strongly preferable to a.
Situation 2 (or S_{2}) leads
to Equation (8):
(8)
The
context S_{2} is characterized by the next fact: the sum of the weights
of the criteria j such what a is strongly preferable b and
is weakly preferable b is bigger than the sum of the weights of the
criteria j such what a is weakly preferable b more the sum
of the weights of the criteria j such what a is indifferent b
more the sum of the weights of the criteria j such what a is not
comparable with b any more the sum of the weights of the criteria j
such what b is weakly preferable a to any more the sum of the
weights of the criteria j such what b is strongly preferable to a.
Situation 3 (or S_{3})
produces Equation (9):
(9)
The context S_{3} is characterized by the next fact: the sum of
the weights of the criteria j such
what a is strongly preferable b and is weakly preferable b and is indifferent b is bigger than the sum of the weights
of the criteria j such what a is weakly preferable b more the sum of the weights of the
criteria j such what more the sum of
the weights of the criteria j such
what her is not comparable with b any
more the sum of the weights of the criteria j
such what b is weakly preferable her
to any more the sum of the weights of the criteria j such what b is strongly
preferable to a. R stands for noncomparability. w_{j}, w are weight
and j are criteria (j = 1, 2, …, n).
It should be noted that the last two situations (S_{2} and S_{3})
are less rigorous than the first (S_{1}). This would lead to a smaller
difference allowing one alternative to be ranked higher than another (Roy &
Bouyssou, 1993).
Situation S_{1} only consider the alternatives a for
which aPb, with b being any
other alternative. In this way, comparing a
with b, we can identify the criteria
in which aPb. This would
consider the thresholds of preference, indifference and discordance. A checking
would verify if the condition imposed is satisfied. If satisfied, we know that a dominates b. Afterwards, the criteria weights in which this condition was
met are added. For another alternative c,
the same procedure described previously is repeated. The final scoring of
alternative a will be the sum of the values obtained.
For the situation S_{2} the alternatives for which aPb and aQb are considered. In situation S_{3},
the alternatives for which aPb, aQb and aIb are considered.
THOR 2 has an important difference with respect to the original THOR
concerning the assignment of weights in the cases of indifference as well as
weak preference in situations S_{1}, S_{2} e S_{3}. When indifference
prevails, half of the respective criterion weight applies. Similarly, when weak
preference occurs, a proportion between half of the criteria weight and the
total weight value is assigned. (Tenório, 2020).
Additionally, THOR 2 takes into consideration that the value of the
criterion weight is multiplied by a fuzzyrough factor, thus contributing to
downgrading the comparison in direct proportion to the importance and security
of data. In other words, differently from the original THOR formulation, in
THOR 2 situations where either strict preference, weak preference or
indifference prevails, the criterion weight is multiplied by the fuzzyrough
index, thus taking into account the full uncertainty of the model, while in
THOR that weight value is downgraded in the situation of weak preference only (Tenório,
2020). The calculations can be performed using a multicriteria platform named THOR
Web, available
through the website www.thorweb.com and developed at the Brazilian
Military Engineering Institute (IME) located in Rio de Janeiro, Brazil.
Before we proceed showing how the two methods were separately
applied to the same data one must clarify that both TODIM and THOR 2 are
noncompensatory methods in the sense that tradeoffs do not occur (Bouyssou,
1986). Weights should in principle reflect to some
extent the degrees of relative importance or strength as estimated by decision
agents along a numerical scale, such as from 0 to 10. This scale can be either
a linear, cardinal scale or a ratio scale. A comparatively high criterion
weight increases the chance that an alternative well classified according to
that criterion is positioned in a high global rank.
However, in some cases a relatively high weight for
any given criterion does not necessarily mean that this is one of the most
important criteria. For instance, given two conflicting criteria for completing
a project, cost and time for completion, a decision maker initially considers
cost as the most important among the two criteria. He therefore assigns a
weight to cost that is much higher than the weight of time for completion. This
is so because he expects to save some money to be assigned to other projects.
However, although some alternatives are close to
reaching below 80% of the available budget, all alternatives are well above 90%
of the time limit for completion. This is a typical situation in which an
intracriteria analysis points out to the following fact: the criterion that had
originally the smallest weight ends up being the most important between the two
criteria.
Problem
structuring has been a quite important step when solving Management Science
problems, and particularly multicriteria problems (Rosenhead & Mingers,
2001; Belton & Stewart, 2002). In this case study, however, the problem was
set in a straightforward way by posing two questions: (1) given a
wellestablished set of evaluation criteria, a set of technological
alternatives (i.e., ballast water management methods) to reduce pollution
caused by ballast water, and a set of restrictions that would apply in a
variety of practical situations, which management methods would be considered
the best by applying TODIM as well as by applying THOR 2? (2) what can we learn
from comparing the two results?
The
relevance of an effective assessment for water ballast management has been
established in the specialized literature (Globallast, 2010, 2011). The
detailed criteria, referring to the relevant factors identified, for
quantitative measuring in association with a nominal scale or description, are
presented below. Each criterion presented shall be analyzed and represented
using quantitative measuring (Figure 1). This can be done by assigning either a
value in a nominal scale, a yes or no answer, or by making use of an interval
or ratio scale.
For
this study, the following criteria were adopted:
(a)
Restriction (or veto criterion) – the system to be incorporated or selected
shall not present any restrictions unacceptable.
(b)
All criteria have the same weight – although the participating experts had
difficulty in achieving a consensus on the weights of criteria they felt it
would be highly valuable to demonstrate that a wellstructured and transparent
multicriteria analysis could lead to a choice of technology.
Figure 1: THOR Web Decision Matrix
One
week of extensive discussions led the experts to agree on the following list of
evaluation criteria. An analysis then allowed to consider those criteria as
being exhaustive, nonredundant and operational.
a) Practicality
a.1) Quantitative
criteria:
C_{1}
 what ballast flow rate range is the system applicable? (m^{3}/hour)
(specify the minimum and maximum flow rate)
C_{2}
 what is the ship tonnage that the system can be applied to? (specify the
minimum and maximum tonnage)
C_{3}
 what is the additional workload on board? (man/hours)
C_{4}
 what is the highest sea state (in the Beaufort wind scale) on which the
system can operate?
C_{5}
 what is the increase in tank's sediment caused by the system? (specify
percentage)
a.2) Questions that need to be answered by a
nominal scale, subject to association to a numerical scale of intervals or by a
yes/no answer:
C_{6}
 does the system present any risks to the ship's crew safety or to the crew?
(3, high risk; 2, medium risk; 1, low risk; 0, no risk)
C_{7}
 does the system affect the tanks' corrosion rate? (2, increases the rate;
1, does not increase the rate; 0, reduces the rate)
C_{8}
 does the system dispense with the need to keep chemical products on board?
(Yes or No)
C_{9}
 can the system be used in short voyages (up to 12 h)? (Yes or No)
C_{10}
 can the system be operated without complete recirculation of the ballast
water? (Yes or No)
C_{11}
 is the system unaffected by incrustation that could lead to a drop in
pressure and/or to a reduction in the flow rate? (Yes or No)
C_{12}
 is the system being applicable to existing ships? (Yes or No)
C_{13}
 are the ship's other functions independent from the system's operation? (Yes
or No)
a.3) Questions that
require detailed answers:
C_{14}
 does the system present any occupational hazard to the operator? Describe and
quantify. (3, high; 2, medium; 1, low; 0, no hazard)
b) Biological
effectiveness (including pathogens)
b.1) Quantitative
Criteria:
C15
 how effective is the system in relation to the removal, elimination and
inactivation/neutralization of aquatic organisms, apart from pathogens
(according to the various taxonomic groups)? (quantify in terms of percentage,
size and/or concentration of organisms)
C16
 same as 15 for pathogens.
b.2) Questions for which
the answers should be either Yes or No
C17
 does the system eliminate cysts?
C18
 does the system allow the elimination of organisms when the water enters the
tank?
C19
 is the system adequate for the elimination of all species or life stages that
may present a hazard to the environment?
c) Costbenefit:
C_{20}
 what is the purchase cost? (US$)
C_{21}
 what is the cost of installation? (US$)
C_{22}
 what is the operational cost? (US$/ton)
C_{23}
 what is the cost variation per ship size? (US$/ton)
C_{24}
 what is the increase of fuel or oil consumption that is introduced by the use
of this system on board? (percentage)
d) Time frame within
which the standards could be practically implemented
C_{25}
 within which time frame could the standards be practically implemented? (no.
of months)
e) Environmental impact
of the process' subproducts
C_{26}
 is the system free from generating subproducts that can have an impact on
the environment?
Undesirable
outcomes are taken with negative values as well as those that have a negative
impact with higher absolute values. This leads to the following: i) In the criteria 3, 5, 2024 and 25,
negative values are assign for the lowest desirable feature; ii) In the criteria 8 to 13, 17, 18, 19
and 26, where the answers should be either "Yes" or "No", a
value of 1 was assigned to a "Yes" answer (desirable) and a value of
0 to a "No" answer (undesirable); and iii) In the criteria 6, 7 and 14, verbal (or nominal) scales
associated to a numerical scale have been created for test purposes.
Ballast
water management technologies can be of two types: no ballast or zero discharge
methods, and continuous flow methods. Research of economic valuations are
intended to improve decisionmaking processes ranging from community or
industry engagement and ecosystem management to the development of national
strategies and action plans to manage the risk associated with invasive alien
species (Globallast, 2010, 2011, 2020a, 2020b).
After
a carefully, first screening, three competitive, alternative ballast water
management technologies were identified. The nineteen professionals
participated fulltime in that first screening, that took two weeks. The
studies in Brazil took more than five months. The alternatives to be analyzed
are named in this article Management Method #1, Management Method #2, and
Management Method #3. Management methods #1 and #2 are flow methods, while Management
Method #3 is a zero discharge method. They are described in Table 1. These
three alternatives were considered as feasible by experts.
The
evaluation matrix is presented in Table 2. This Table presents
an example utilization of this method using the three Management Methods #1, #2
and #3. It is difficult, just by looking at Table 1, to identify the best
management method. This problem becomes even more complicated if we consider
that there are several ballast water treatment methods currently being
discussed at IMO and not just the three ones used as example.
Table 2: Evaluation matrix.
Criteria 
Alternatives 

Management Method #1 
Management Method #2 
Management Method #3 

C_{1} 
Maximum: 15,000
m^{3}/h Minimum: 100 m^{3}/h 
Maximum: 14,000
m^{3}/h Minimum: 200 m^{3}/h 
Maximum: 13,000
m^{3}/h Minimum: 300 m^{3}/h 
C_{2} 
Maximum: 450,000 DWT Minimum: 450 DWT 
Maximum: 350,000 DWT Minimum: 350 DWT 
Maximum: 250,000 DWT Minimum: 450 DWT 
C_{3} 
90 man/hour 
80 man/hour 
90 man/hour 
C_{4} 
7 
8 
10 
C_{5} 
10 % 
12 % 
5 % 
C_{6} 
1 
2 
3 
C_{7} 
2 
1 
3 
C_{8} 
1 
1 
0 
C_{9} 
1 
1 
0 
C_{10} 
1 
1 
0 
C_{11} 
0 
1 
1 
C_{12} 
0 
1 
1 
C_{13} 
0 
0 
1 
C_{14} 
0 
1 
2 
C_{15} 
93 % 
92 % 
90 % 
C_{16} 
90 % 
88 % 
91 % 
C_{17} 
1 
0 
1 
C_{18} 
1 
0 
0 
C_{19} 
0 
1 
1 
C_{20} 
US$ 200,000.00 
US$ 210,000.00 
US$ 220,000.00 
C_{21} 
US$ 10,000.00 
US$ 21,000.00 
US$ 1,000.00 
C_{22} 
0.02 US$/ton 
0.03 US$/ton 
0.04 US$/ton 
C_{23} 
US$ 9 
US$ 8 
US$ 6 
C_{24} 
3 % 
8 % 
1 % 
C_{25} 
6 months 
8 months 
9 months 
C_{26} 
0 
1 
0 
All
values in the above matrix were normalized and transformed into maximization
criteria for the use of the TODIM method. The THOR 2 method used data as shown
in Table 2. From now in this article on Management Methods #1, #2, and #3 will
be designated as A1, A2 and A3.
The
application of TODIM took into consideration three possible situations: (i)
attenuation factor θ equal to 1.0 (less risk proneness); (ii) θ equal
to 10.0 (greater risk proneness); and (iii) θ equal to 5.0 (an
intermediate value between the two previous extreme situations). Results from
the computations are presented in Table 3.
Table 3: Performance of the three alternatives
according to the TODIM method.
Alternatives 
θ = 1.0 
θ = 5.0 
θ = 10.0 
A_{1} 
1.000 
1.000 
1.000 
A_{2} 
0.808 
0.000 
0.000 
A_{3} 
0.000 
0.210 
0.425 
The application of THOR 2 considered
the situations S_{1}. Table 4 shows the outputs from using THOR 2.
Table 4: THOR 2 results
S_{1} 
S_{2} 
S_{3} 

A1 
1.114 
A1 
1.114 
A1 
1.259 
A2 
0.583 
A2 
0.583 
A2 
0.667 
A3 
0.000 
A3 
0.000 
A3 
0.000 
The three alternatives were
evaluated according to criteria 1, 2, 3, 20, 21, 22, 23 and 25 on a ratio
scale. Alternatives were evaluated according to criteria 5, 15 and 16 by an
interval scale. Alternatives were evaluated with respect to all other criteria
by using a nominal scale associated to an interval scale.
5.
DISCUSSION AND CONCLUSIONS
It
is worth noting that although the two methods rely on different foundations
they produced in essence the same results. Two other
applications of both TODIM and THOR have indeed confirmed the convergence of
results in spite of the conceptual and technical differences between the two
methods. The present result from applying TODIM and THOR 2 is therefore consistent
with the results obtained by Gomes et al. (2009) and Gomes et al. (2010).
The
TODIM method is founded on the paradigm of Prospect Theory and data are
aggregated by means of building an additive value function. On the other hand,
THOR 2 relies on the notion of outranking and does not directly take into
account the attitude of a decision maker facing risk. The
fact that both TODIM and THOR 2 produce similar results suggests that formulating
a decision problem in a comprehensive way and applying a method correctly may
be at least as important as the technical characteristics of the method per
se.
A1
was chosen as the best alternative when both TODIM and THOR 2 are used in the
situation when θ is made equal to 1.0. When θ was equal to 5.0 and
10.0, however, there was a discordance between TODIM and THOR 2 concerning the
last two alternatives: A1 is preferred to A3
and A3 is preferred to A2 from TODIM and A1 is preferred to
A2 and A2 is preferred to A3 from THOR 2.
Nevertheless,
since this was a problem in the choice of a technology, for practical purposes
it was concluded that the two methods produced similar results. The final
results show that A1 should be the best choice. Given that participants in the
evaluation study understood the use of the methods, this convergence of results
led to widely accepted recommendation to decision makers.
6.
ACKNOWLEDGEMENTS
The study presented in
this paper was partially supported by CNPq through Project No.
306562/20170.
REFERENCES
Belton, V., & Stewart, T. J. (2002). Multiple
Criteria Decision Analysis. Kluwer, Boston.
Bouyssou, D. (1986). Some remarks on the
notion of compensation in MCDM. European Journal of Operational Research,
26 (1), p 150160.
Brans, J. P., & Mareschal, B. (1990).
The PROMÉTHÉE methods for MCDM, the PROMCALC GAIA and BANKADVISER software. In Readings
in Multiple Criteria Decision Aid (edited by C.A. Bana e Costa), chapter 2,
Springer, Berlin.
Cardoso, R. S., Xavier, L. H.,
Gomes, C. F. S., & Adissi, P. J. (2009). Uso de SAD no apoio à decisão na
destinação de resíduos plásticos e gestão de materiais. Pesquisa Operacional,
29( 1), 67–95.
Costa, I. P. A., Maêda, S. M. N.,
Teixeira, L. F. H. S. B., Gomes, C. F. S., & Santos, M. (2020). Escolha de
navio de assistência hospitalar no combate à pandemia da covid19. Revista
de Saúde Pública, 54, 79. EPUB, August 10, 2020. https://doi.org/10.11606/s15188787.2020054002792
Clemen, R. T., & Reilly, T. (2001). Making
Hard Decisions with Decision Tools^{®}, DuxburyThomson Learning,
Pacific Grove.
Costa, I. P. A., Gomes, C. F. S.,
Teixeira, L. F. H. S. B., Santos, M., & Maeda, S. (2020). Choosing a
hospital assistance ship to fight the covid19
pandemic. Revista de Saúde Pública, 54, 18.
Ehrgott, M., Figueira, J. R., &
Greco, S. (eds.) (2010). Trends in Multiple Criteria Decision
Analysis, Springer, New
York.
Globallast (2010). Economic Assessments for
Ballast Water Management: A Guideline. GloBallast Partnerships Project
Coordination Unit and International Maritime Organization, London,
GloBallast Monograph Series, N. 19.
Globallast (2011). Establising Equivalency
in the Performance Testind and Compliance Monitoring of Emerging Alternative
Ballast Water Management Systems. GloBallast Partnerships Project
Coordination Unit and International Maritime Organization, London,
GloBallast Monograph Series, N. 20.
Gomes, C. F. S. (1999). THOR
 Um Algoritmo Híbrido de Apoio Multicritério à Decisão para Processos
Decisórios com Alternativas Discretas. Tese de doutorado em Engenharia de
Produção, Universidade Federal do Rio de Janeiro, Brasil.
Gomes, C. F. S. (2005). Using MCDA methods THOR in an
application for outranking the ballast water management options. Pesquisa Operacional, 25, Jan/April, p 1128.
Gomes, C. F. S. (2006).
Modelagem analítica aplicada à negociação e decisão em grupo. Pesquisa
Operacional, 26(3), 537–566.
Gomes, C. F. S., & Costa, H. G. (2015). Aplicação de métodos
multicritério ao problema de escolha de modelos de pagamento eletrônico por
cartão de crédito. Production,
25(1), 54–68.
Gomes, C. F. S., Gomes, L. F. A. M., & Maranhão, F. J.
C. (2010). Decision
analysis for the exploration of gas reserves: merging TODIM and THOR. Pesquisa Operacional, 30(3), 601617.
Gomes, C. F. S., & Maia,
A. C. C. (2013). Ordenação de alternativas de biomassa utilizando o apoio
multicritério à decisão. Production, 23(3), 488–499.
Gomes, L. F. A. M.
(1989). Multicriteria ranking of urban transportation system alternatives. Journal
of Advanced Transportation, 23(1), 4352.
Gomes, L. F. A. M.
(1989). Comparing two methods for multicriteria ranking of urban transportation
system alternatives. Journal of Advanced Transportation, 23(2&3),
217219.
Gomes, L. F. A. M. (1990). Modelling interdependencies
among urban transportation system alternatives. Journal of Advanced
Transportation,24( 1, 7785.
Gomes, L. F. A. M.
(1990). Eliminating rank reversal in multicriteria analysis of urban
transportation system alternatives. Journal of Advanced Transportation,24(2),
181184.
Gomes, L.
F. A. M., Gomes,
C. F. S., Rangel, L. A. D. (2009). A comparative decision analysis with THOR and TODIM:
rental evaluation in Volta Redonda. Revista
Tecnologia, 30(1), 711.
Gomes, L. F. A. M., & Lima, M. M. P. P. (1991). TODIM:
basics and application to multicriteria ranking of projects with environmental
impacts. Foundations of Computing and Decision
Sciences,16(4), 113127.
Gomes, L. F. A. M., & Lima, M. M. P. P. (1992). From
modeling individual preferences to multicriteria ranking of discrete
alternatives: a look at Prospect Theory and the additive difference model. Foundations of Computing and Decision
Sciences,17(3), p 171184.
Gomes, L. F. A. M., Machado, M. A. S., González, X. I., &
Rangel, L. A. D. (2013). Behavioral multicriteria decision analysis: the TODIM
method with criteria interactions. Annals of Operations Research,
211(1), p 531548.
Gomes, C. F. S., Nunes, K. R. A., Xavier,
L. H., Cardoso, R., & Valle, R. (2008). Multicriteria decision making
applied to waste recycling in Brazil. Omega,
36, p 395404.
Gomes, L. F. A. M., & Rangel, L. A. D. (2009). An application of the TODIM
method to the multicriteria rental evaluation of residential properties. European
Journal of Operational Research, 193, 204–211.
Gomes, L. F. A. M., Rangel, L. A. D., & Maranhão, F. J. C.
(2009). Multicriteria analysis of natural gas destination in Brazil: an
application of the TODIM method. Mathematical and Computer Modelling, 50, 92100.
IMO (2020a). Ballast Water Management. http://www.imo.org/en/OurWork/Environment/BallastWaterManagement/Pages/Default.aspx.
Access: 10/09/2020.
IMO (2020b). Welcome to GloBallast. http://globallast.imo.org/index.asp?page=problem.htm&menu=true.
Access: 10/09/2020.
Kahneman, D., & Tversky, A. (1979).
Prospect theory: An analysis of decision under risk. Econometrica, 47,
263292.
Keeney, R. L., & Raiffa, H. (1993). Decisions
with Multiple Objectives: Preferences and Value Tradeoffs, Cambridge
University Press, Cambridge.
Rosenhead, J., Mingers, J. (eds) (2001). Rational
Analysis for a Problematic World Revisited, Wiley, Chichester.
Roy, B. (1996). Multicriteria
Methodology for Decision Aiding, Kluwer, Dordrecht.
Roy, B., & Bouyssou, D.
(1993). Aide Multicritère à la Décision: Méthodes et Cas, Economica,
Paris.
Sudha, A. S., Gomes, L. F. A. M., & Vijayalakshmi,
K. R. (2020). Assessment of MCDM problems by TODIM using aggregated weights. Neutrosophic Sets and Systems, 35, 7898.
Tenenbaum, D. R., Villac, M. C., Viana, S. C.,
Matos, M., Hatherly, M., Lima, I. V., & Menezes, M. (2004). Phytoplankton
Atlas of Sepetiba Bay, Rio de Janeiro, Brazil. GloBallast Partnerships
Project Coordination Unit and International Maritime Organization, London,
GloBallast Monograph Series, N. 16.
Tenório, F. M. (2020). Modelagem multicritério: uma
evolução do método THOR. Dissertação (Mestrado em Sistemas e Computação) –
Instituto Militar de Engenharia, Rio de Janeiro, Available: http://www.comp.ime.eb.br/pos/modules/files/dissertacoes/2020/2020Fabricio.pdf.
Access: 31/12/2020.
Tenorio, F. M., Santos, M.,
Gomes, C. F. S., & Araujo, J. C. (2020a). Estratégia para compra de
oportunidade de uma fragata para a Marinha do Brasil a partir do método
multicritério THOR. Revista
Valore, 5, 4357.
Tenorio, F. M., Santos, M., Gomes, C. F. S.,
Araujo, J. C. (2020b). Navy Warship Selection and Multicriteria Analysis: The
THOR Method Supporting Decision Making. In: Thomé, A. M. T., Barbastefano, R. G.,
Scavarda, L. F., dos Reis, J. C. G., & Amorim, M. P. C. (eds.) Industrial
Engineering and Operations Management. IJCIEOM 2020. Springer Proceedings
in Mathematics & Statistics, 337. Springer, Cham.
https://doi.org/10.1007/9783030569204_3
Tversky,
A. (1969). Intransitivity of Preferences. Psychological Review,
76(1), 3148.
Zindani, D., Maity, S. R., & Bhowmik, S. (2020). Complex interval‑valued intuitionistic fuzzy TODIM approach and its application to group decision making. Journal of Ambient Intelligence and Humanized Computing. Published online: 16 July 2020. https://doi.org/10.1007/s12652020023080.