Proceedings of the 13th International Conference on Environmental Science and Technology

Athens, Greece, 5-7 September 2013

MULTIPLE CRITERIA DECISION MAKING USING VIKOR METHOD.

APPLICATION IN IRRIGATION NETWORKS IN THE THESSALONIKI PLAIN

C. ΤZIMOPOULOS, D. ZORMPA* AND C. EVANGELIDES

Laboratory of Hydraulics and Environmental Management

Aristotle University of Thessaloniki, 54124, GREECE
E-mails: *[email protected]

EXTENDED ABSTRACT

A multicriteria model is developed for analyzing the planning strategies for reducing future
social, economic and environmental costs. The developed multicriteria decision-making
procedure consists of generating alternatives, establishing and ranking criteria, assessing
criteria weights, and application of the compromise ranking method.

The paper emphasizes compromise based on cooperativeness as a solution of a decision

problem in water resources planning. A decision making process committed to consensus
utilizing compromise solution as a consensus proposal is presented.

The presented methodology is utilized to obtain the optimal irrigation network among a
finite number of networks in the area of Thessaloniki plain.

The VIKOR method is applied here to determine compromise solution with

noncommensurable and conflicting criteria, including economic, environmental and social
criteria. It focuses on selecting and ranking from a set of alternatives, and determines
compromise solutions for a problem with conflicting criteria, which can help the decision
makers to reach a final decision.

The VIKOR method is based on L1, j (Manhattan distance) and L , j (Chebyshev

distance) metrics, where j  1, 2,..., J is the number of alternatives. The solution norm Qj
is introduced as a linear combination of the L1, j and L , j norms and the final solution is
min Qi, where min L1, j represents the maximum ‘‘group utility’’ (majority rule) and
min L , j represents the minimum individual regret of the ‘‘opponent’’.

As an application, a finite number of networks in the area of Thessaloniki plain is selected

for obtaining the optimal irrigation network. The networks should be graded for certain
evaluation criteria in order to be incorporated in the Economic Plan of the GOEB
Thessaloniki (General Organisation of Hydraulics Field Thessaloniki Lagada). The criteria
have been selected for each alternative and as optimal solution was considered the
network satisfying mostly the selected criteria. Besides, the importance of each criterion
is considered and the method A.H.P. (Analytic Hierarchy Process) proposed by Saaty
was applied for the assessment of the weights.

Keywords: Multi-criteria analysis, compromise programming, VIKOR method,

Thessaloniki plain, irrigation networks, A.H.P.

CEST2013_0143
1. INTRODUCTION

Land and waters in the world are under increasing pressure from the continuous growth
in demand for many different purposes, and the allocation of water in the river basin is a
complex management problem, with conditions that may foster conflicts. Conflict over the
management of a shared water resource arises mostly because of differing objectives
among different interest groups.

The agricultural sector in Greece is the largest water consumer and essentially it has the
lion's share in the management of water. At the beginning of the second half of the 20th
century started all modern land reclamation projects. The organized collective irrigation
networks in Greece were mainly built during the decade of 1960 (Konstantinides, 1989;
Balioti, 2009). They are characterized by age but also by an old technology and the
design does not take into account the factor of protection of both water and the
environment. It is difficult today to characterize the functioning of an irrigation network,
taking into account only one criterion. For the management of irrigation water and repair
of irrigation systems, knowledge of a set of criteria is required that will help the
sustainable disposal of water.

The design of an irrigation management system is characterized by: a) a large degree of

uncertainty, b) sophisticated subject design with often multidimensional objectives, c)
difficulty in determining the individuals or groups that contribute to the decision and d)
sophisticated structure alternatives which combine in sequence several elementary
actions and planning horizons (Netto et al., 1996).

To solve a problem for MCA, many methods were developed, but in this article the Vikor
method is used, as analyzed by Opricovic and Tzeng (2002; 2004; 2006) and Opricovic
(2009). In the area of water resources management, management problems have been
addressed by Duckstein and Opricovic (1980); Opricovic (1993); Maraveas (1998);
Karasavvidis (2003), Baka (2006); Zarghami (2006); Schiau and Wu (2006 );
Karasavvidis (2009) etc.

2. MCDM

Multicriteria optimization (MCO) is considered as the process of determining the best

feasible solution according to established criteria which represent different effects.
However, these criteria usually conflict with each other and there may be no solution
satisfying all criteria simultaneously. Thus, the concept of Pareto optimality was
introduced for a vector optimization problem (Pareto, 1896; Kuhn and Tucker, 1951;
Zadeh, 1963). Pareto optimal solutions have the characteristic that, if one criterion is to
be improved, at least one other criterion has to be made worse. In such cases, a system
analyst can aid the decision making process by making a comprehensive analysis and by
listing the important properties of the Pareto optimal (noninferior) solutions. However, in
engineering and management practice there is a need to select a final solution to be
implemented. An approach to determine a final solution as a compromise was introduced
by Yu (1973), and other distance-based techniques have also been developed (Chen and
Hwang, 1992).

Many papers have proposed analytical models as aids in conflict management situations.
Among the numerous approaches available for conflict management, one of the most
prevalent is multicriteria decision making. Multicriteria decision making (MCDM) may be
considered as a complex and dynamic process including one managerial level and one
engineering level (Duckstein and Opricovic, 1980). The managerial level defines the
goals, and chooses the final ‘‘optimal’’ alternative.

CEST2013_0143
The multicriteria nature of decisions is emphasized at this managerial level, at which
public officials called ‘‘decision makers’’ have the power to accept or reject the solution
proposed by the engineering level. These decision makers, who provide the preference
structure, are ‘‘off line’’ from the optimization procedure done at the engineering level.

Very often, the preference structure is based on political rather than only technical
criteria. In such cases, a system analyst can aid the decision making process by making
a comprehensive analysis and by listing the important properties of noninferior and/or
compromise solutions (Yu, 1973). The engineering level of the MCDM process defines
alternatives and points out the consequences of choosing any one of them from the
standpoint of various criteria. This level also performs the multicriteria ranking of
alternatives.

3. VIKOR method

The VIKOR method was developed for multicriteria optimization of complex systems. It
determines the compromise ranking-list, the compromise solution, and the weight stability
intervals for preference stability of the compromise solution obtained with the initial
(given) weights. This method focuses on ranking and selecting from a set of alternatives
in the presence of conflicting criteria. It introduces the multicriteria ranking index based on
the particular measure of ‘‘closeness’’ to the ‘‘ideal’’ solution (Opricovic, 1998).

Assuming that each alternative is evaluated according to each criterion function, the
compromise ranking could be performed by comparing the measure of closeness to the
ideal alternative.

The multicriteria measure for compromise ranking is developed from the Lp-metric used
as an aggregating function in a compromise programming method (Yu, 1973; Zeleny,
1982).

The various J alternatives are denoted as a1; a2; . . . ; aj . For alternative aj, the rating of
the i aspect is denoted by fij, i.e., fij is the value of i criterion function for the alternative aj;
n is the number of criteria.

Development of the VIKOR method started with the following form of Lp-metric:

1
n p
   wi  fi *  fij  /  fi *  fi    
p
Lp , j , 1  p   ; j=1.2….j.
 i 1 

Within the VIKOR method L1,j =Sj and L , j = Rj are used to formulate ranking measure.
The solution obtained by minj Sj is with a maximum group utility (‘‘majority’’ rule), and the
solution obtained by minj Rj is with a minimum individual regret of the ‘‘opponent’’.

The compromise solution Fc is a feasible solution that is the ‘‘closest’’ to the ideal F*, and
compromise means an agreement established by mutual concessions, as is illustrated in
Figure 1 by f1  f1*  f1c and f 2  f 2*  f 2c .

CEST2013_0143
 Figure 1. Ideal and compromise solution.

The compromise ranking algorithm VIKOR has the following steps:

(1) Determine the best f i * and the worst f i  values of all criterion functions, i = 1,2,. . . , n;

fi *  max fij , fi   min fij if the i-th function represents a benefit;

j j

fi  max fij , fi  min fij if the i-th function represents a cost.
*
j j

(2) Compute the values Sj and Rj, j = 1,2,. . . ,J, by the relations

S j   wi  fi *  fij  /  fi *  fi  
n

i 1

R j  max  wi  fi*  fij  /  fi*  fi  

i

where wi are the weights of criteria, expressing the DM’s preference as the relative
importance of the criteria.

(3) Compute the values Qj, j = 1,2,. . . ,J, by the relation

Q j  v  S j  S *   S   S *   1  v   R j  R*  R 
 R* 

where S *  m i nS j , S   max S j , R*  min R j , R   max R j and v is introduced as a

j j j j

weight for the strategy of “the majority of criteria” (or “the maximum group utility”),
whereas 1 − v is the weight of the individual regret.

(4) Rank the alternatives, sorting by the values S, R and Q in decreasing order. The
results are three ranking lists.

(5) Propose as a compromise solution the alternative [F(1)] which is the best ranked by the
measure Q (minimum) if the following two conditions are satisfied:

C1. “Acceptable Advantage”:

Q( F (2)  F (1) )  DQ

where: F(2) is the alternative with second position in the ranking list by Q and

CEST2013_0143
DQ  1/ (m  1) .
C2. “Acceptable stability in decision making”:

The alternative F(1) must also be the best ranked by S or/and R. This compromise
solution is stable within a decision making process, which could be the strategy of
maximum group utility (when v > 0.5 is needed), or “by consensus” v ≈ 0.5, or “with veto”
(v < 0.5). Here, v is the weight of decision making strategy of maximum group utility.

If one of the conditions is not satisfied, then a set of compromise solutions is proposed,
which consists of:

 Alternatives F(1) and F(2) if only the condition C2 is not satisfied.

 Alternatives F(1), F(2),. . . , F(M) if the condition C1 is not satisfied. The F(M) is
   
determined by the relation Q F ( M )  Q F (1)  DQ for maximum M (the positions of
these alternatives are “in closeness”).

4. CHOICE OF OPTIMAL IRRIGATION NETWORK

4.1. Description of problem

The study area consists of 5 irrigation networks of the plain of Thessaloniki, specifically:
a) F1 = the irrigation network of Agios Athanasios, b) F2 = the irrigation network of
Brachias, c) F3 = the irrigation network of Malgara d) F4 = the irrigation network of
Chalastras and e) F5 = the irrigation network of Chalcedonas.

A common feature of five irrigation networks (options / scenarios) is that they are supplied
with water from the river Axios to serve the water needs of crops. Additionally, the
construction took place during the first phase of land reclamation projects in the plain of
Thessaloniki during the decade of 1960 (Konstantinidis, 1989). Therefore, the problem is
reduced to the calibration and classification of these networks in terms of financial
planning.

To find the ideal solution to the problem, it is necessary to establish criteria that will cover
both the economic-environmental and productive part of the evaluation of each scenario.
Thus, we use the following basic criteria: a) X1 = productivity per acre based on the crop,
b) X2 = cost of production, c) X3 = cost of maintenance, d) X4 = crop needs in water, e) X5
= products’ availability.

The calibration criteria was partially made using the A.H.P method (Analytic Hierarchy
Process) which was introduced in 1980 by Saaty, for both scenarios and for weights as
shown in Table 1. The worst score is 1 and the maximum 9. The main reason this
calibration was used was to achieve as much objectification of compensatory weights and
criteria as possible.

Table 1. Pair-wise comparison scale for AHP preferences

Numerical Verbal judgments of Numerical Verbal judgments of
rating preferences rating preferences
9 Extremely preferred 5 Strongly preferred
8 Very strongly to extremely 4 Moderately to strongly
7 Very strongly preferred 3 Moderately preferred
6 Strongly to very strongly 2 Equally to moderately
1 Equally preferred

CEST2013_0143
4.2. Evaluation of the problem

The F (J, I) is the matrix whose elements indicate the value of different scenarios for
different criteria. The index J is for different scenarios (alternatives), while the index I for
the different criteria. Shows that the F (J, I) the 'calibration' or evaluation of each scenario
based on various criteria. Table 2 shows the numerical calibration for the study area,
while Table 3 gives the calibration logic problem.

Table 2. Decision matrix

X1 X2 X3 X4 X5
F1 8 3 5 2 8
F2 6 3 3 2 4
F3 1 3 3 2 5
F4 7 1 1 3 9
F5 3 4 2 1 1

Table 3. Logical evaluation of problem

X1 X2 X3 X4 X5
Very strongly to Moderately Strongly Equally to Very strongly to
F1
extremely preferred preferred moderately extremely
Strongly to very Moderately Moderately Equally to Moderately to
F2
strongly preferred preferred moderately strongly
Equally Moderately Moderately Equally to Strongly
F3
preferred preferred preferred moderately preferred
Very strongly Equally Equally Moderately Extremely
F4
preferred preferred preferred preferred preferred
Moderately Moderately to Equally to Equally Equally
F5
preferred strongly moderately preferred preferred

For finding the weights by Saaty (1980) the normalization of problem initially took place
and for this reason the program Matlab was used, which resolved the 5x5 matrix (Table
4), and calculated: a) the maximum principal value λmax = 5.9019 and b) the eigenvector
or anchor vector, which was considered the same for each alternative activity, where W 1
= 0.919, W 2 = 0.3452, W3 = 0.1684, W 4 = 0.0765 and W 5 = 0.0635. This vector is
normalized and weights were:

5
wi ,1 =0.584, wi ,2 =0.219, wi ,3 = 0.107, wi ,4 =0.049, wi ,5 =0.041 και W
j 1
i,j =1.

Table 4. Dimensionless assessment of the problem

X1 X2 X3 X4 X5
F1 1.000 0.667 1.000 0.500 0.875
F2 0.714 0.667 0.500 0.500 0.375
F3 0.000 0.667 0.500 0.500 0.500
F4 0.857 0.000 0.000 1.000 1.000
F5 0.286 1.000 0.250 0.000 0.000

For the numerical computation of distances, for different weights and p, drafted the
program "Compromi" programming language Visual Fortran (Tzimopoulos, 2013). Table 5
shows the distances of alternatives in relation to the optimum point for p = 1 and p = 19.

CEST2013_0143
The value p = 19 is quite large value, beyond which the norm is constant and therefore
applies: ( L19  L ).

Table 5. Classification by VIKOR method

S sorting S R sorting R Q sorting Q
F1 0.103 0.103 0.073 0.073 0.000 0.000
F2 0.343 0.343 0.167 0.167 0.276 0.276
F3 0.755 0.410 0.584 0.220 1.000 0.379
F4 0.410 0.586 0.220 0.417 0.379 0.707
F5 0.586 0.755 0.417 0.584 0.707 1.000

According to Table VIKOR ranking in the ranking is as follows:

F1  F4  F2  F3  F5

• Under the condition C1 we have:

Q( F (2) )  Q( F (1) )  DQ  (0.276-0) > 1/(5-1)=0.25.

• According to the second condition rankings of S, R coincides with the classification of Q

and thus fulfilled the condition C2.

The irrigation network of Agios Athanasios outperforms the two norms in relation to the
other four.

5. CONCLUSIONS

The VIKOR method introduces the ranking index based on the particular measure of
‘‘closeness’’ to the ideal solution. The VIKOR method used linear normalization,
additional to the normalized value in the VIKOR method does not depend on the
evaluation unit of a criterion function.

Summarizing all the above, the method leads the user to find the ideal solution and the
prioritization of irrigation to the needs of the maintenance program, thus enabling the
recipient of the decision in this case in GOEB Thessaloniki make the rational planning of
maintenance of irrigation networks, ranking the different networks depending on their
importance.

REFERENCES

1. Baka M., (2006) Water Resources Management Methods of Multicriteria Analysis

(Compromise Programming) and Application in Watershed River Nestos. Master
thesis, School of Rural and Surveying Engineering, AUTH, 115-117.
2. Balioti B., (2009) Management of irrigation systems by means of linear programming.
Application to Irrigation Network Gefyras. Master thesis, School of Rural and
Surveying Engineering, AUTH, 36.
3. Chen J. and Hwang L., (1992) Fuzzy Multiple Attribute Decision Making: Methods
and Applications. Springer-Verlag, Berlin.
4. Duckstein, L. and Opricovic, S., (1980) Multiobjective optimization in river basin
development. Water Res. Res., 16(1), 14-20.

CEST2013_0143
5. Karasavvidis P., (2003) Application of Multicriteria Analysis (Compromise
Programming) Management of Water Resources in Livadiou Larisas. Selecting
optimal position dam. Master thesis, School of Rural and Surveying Engineering,
AUTH, 89-91.
6. Karasavvidis P., Tzimopoulos C. and Evangelides C., (2009) Application of
Multicriteria Analysis (Compromise Programming) Management of Water Resources
in Livadiou Larisas. Selecting optimal position dam. Proceedings of the 6th National
Conference of Agricultural Engineering, E.G.M.E., Thessaloniki, 123-130.
7. Konstantinidis K., (1989) The reclamation works in Thessaloniki plain. Publications
GEOT.E.E, Thessaloniki, 133-134.
8. Kuhn W., and Tucker W., (1951) Nonlinear Programming. In: Neyman, J. (Ed.).
Proceedings of the Second Berkeley Symposium on Mathematical Statistics and
Probability. University of California Press, 481–492.
9. Maravea E., (1998) MCA method in the management of water resources. Thesis,
NTUA, 97.
10. Netto O., Parent E. And Duckstein L., (1996) Multicriterion Design of Long-Term
Water Supply in Southern France. Water Resources Planning and Management,
122, 403-413.
11. Opricovic S., (1993) Dynamic programming with application to water reservoir
management. Agricultural Systems, 41, 335-347.
12. Opricovic S., (1998) Multicriteria Optimization of Civil Engineering Systems, Faculty
of Civil Engineering, Belgrade.
13. Opricovic S. and Tzeng GH (2004) The compromise solution by MCDM methods: a
comparative analysis of VIKOR and TOPSIS. Eur J Oper Res 156(2), 445–455.
14. Opricovic S, Tzeng GH (2006) Extended VIKOR method in comparison with
outranking methods. Eur J Oper Res 178(2), 514–529.
15. Opricovic S., (2009) A Compromise Solution in Water Resources Planning. Water
Resour Manage, 23, 1549–1561
16. Pareto V., (1896) Cours d'Economie Politique. Droz, Geneva.
17. Tzimopoulos C., (2013) Compromi. Interior Report, School of Rural and Surveying
Engineering, AUTH.
18. Saaty T.L., (1980) The Analytic Hierarchy Process. Mc Graw-Hill, New York.
19. Schiau T. and Wu C., (2006) Compromise Programming Methodology for
determining instream flow under multiobjective water allocation criteria. JAWRA
Journal of the A.W.R.A., 42(5), 1179-1191.
20. Yu L., (1985) Multiple-Criteria Decision Making, Concepts, Techniques, and
Extension. Plenum Press, Lawrence, Kansas, 388.
21. Zadeh A., (1963) Optimality and non-scalar-valued performance criteria. IEEE
Transactions on Automatic Control 8(1), 59–60.
22. Zarghami M., (2006). Integrated water resources management in polrud irrigation
system. Water Resources Management, 20, 215-225.
23. Zeleny M., (1982) Multiple Criteria Decision Making. McGraw-Hill, New York.

CEST2013_0143