Académique Documents
Professionnel Documents
Culture Documents
www.elsevier.com/locate/renene
Abstract
The authors show an application of the multicriteria decision-making methodology used to
assess an action plan for the diffusion of renewable energy technologies at regional scale. This
methodological tool gives the decision-maker considerable help in the selection of the most
suitable innovative technologies in the energy sector, according to preliminary fixed objectives.
In this paper, a case study is carried out for the island of Sardinia. This region presents, on
one hand, a high potential for energy resources exploitation, but on the other hand, it represents
a specific case among other Italian regions, because of its socio-economic status and history.
Three decision scenarios have been supposed, each one representing a coherent set of
actions, on the basis of which strategies of diffusion are developed.
2003 Elsevier Science Ltd. All rights reserved.
Keywords: Multicriteria decision-making; Renewable energy technologies
1. Introduction
Energy planning processes usually include a study of sectorial demand and supply,
forecasts of the trends of inputoutput items, based on economics and technological
models, and a list of actions, collecting several measures voted to fulfill the main
0960-1481/03/$ - see front matter 2003 Elsevier Science Ltd. All rights reserved.
doi:10.1016/S0960-1481(03)00102-2
2064
objectives of the energy plan. This action plan (AP) is addressed to specific strategies
and interventions, which are able to fit, in the best way, demand and supply, according to the many constraints and factors.
The addressing of these needs could be supported by the adoption of multicriteria
approaches in the selection of the most suitable action among all the alternatives.
The selection of the alternative options derives from the goal set identified by the
decision-maker (DM), with regard to the environmental, technical and economical
spheres.
A decision support approach, called Electre III [1], is presented for energy planning application. This method also represents the first methodology with fuzzy concepts incorporated in it, able to help the decision-maker to select the most suitable
innovative technologies in the energy sector [2].
2065
systems at different levels of intervention. In such a way, the DM can have a complete framework of different performances of planning alternatives.
3.2. Criteria of evaluation and evaluation of alternatives, according to each
criterion
The criteria of evaluation have to provide tools of judgment for DM, which must
verify the consistence of choices with the expectations of the DM and with the needs
of the other involved actors. The main target of the analyst is to show the effects
of each alternative, by means of a set of suitable criteria, F = {1,2,...j,...m}, which
allows constitution of a rank order of the alternatives [6].
Given two alternatives Ai and Ak, assume that gj = gj(Ai) and gj = gj(Ak) express
the performance values of Ai and Ak, respectively, according to the jth criterion. Ai
is predominant over Ak, if and only if:
gj(Ai) gj(Ak)
jF
(1)
The inequality (1) must be verified for at least one criterion. F is consistent if it
is accepted by all the actors involved in the decisional process.
Criteria express qualitative or quantitative viewpoints, objectives, aptitudes, and
constraints of an action, and allow assessment of the alternatives, by means of a
rank order.
A coherent set of criteria has to fulfill the following requirements:
Exhaustivity: criteria must not be insufficient or in excess:
If F, gj(Ai) = gj(Ak) Ai and Ak are indifferent
If the DM does not consider the previous statement to be true, then it implies that
some important evaluation criteria have not been taken into account.
Coherence: the set of decision-makers preferences on each criterion has to be consistent with the global preferences:
If gj(Ai) = gj(Ak) j k and gk(Ai) gk(Ak) Ai is preferred to Ak
Not redundancy: criteria must not be in excess and must not be duplicated. Deleting
some criteria can make invalid one of the previous condition for
at least a pair of actions.
3.3. Definition of aggregation procedure
Aggregation of criteria is necessary to give a synthetic judgment, stemming from
the results of the criteria application. In particular, the Electre III approaches are
characterized by a partial aggregation of preferences.
Under the above considerations, it is possible to define the outranking relation of
the alternative Ai on the alternative Ak, as a binary relation on A, if it is possible
clearly to assert that Ai is at least as good as Ak, given the problem essence, the
DMs preferences and the quality of the assessment about each alternative [7].
The hypothesis of outranking is supported by two test conditions: (1) concordance;
and (2) discordance. An index, which is defined in the range [0,1], provides a judg-
2066
ment on the degree of credibility of each outranking relation and represents a test
to verify the performance of each alternative.
While a criterion is defined as a rigid tool1 according to the classic conception,
the Electre methodology introduces the flexible concept of pseudo-criterion. It defines
an indifference condition in a zone where the difference between Ai and Ak is
rather small.
A zone of weak preference is also defined between the zones of indifference and
strict preference. Such zones represent uncertainty between indifference and strict
preference conditions [8].
The above procedure allows net judgments to be avoided, when data are not completely available and are uncertain.
An indifference threshold qj and a strict preference threshold pj are fixed, with
regard to the jth criterion. In particular, qj indicates the minimum boundary of uncertainty, associated with the performed calculations, while pj can be considered as the
maximum boundary of error, connected to the performed calculations. Therefore, a
pseudo-criterion is a function gj, in which the discriminant capacity is characterized
by two Ai and Ak A thresholds, indicated with qj and pj and defined as follows:
Ai and Ak are indifferent if:
|gj(Ai) gj(Ak)| qj
(2)
(3)
(4)
(5)
(6)
For the jth criterion it can be defined the so-called veto threshold vj, as the limit
value of the difference gj(Ak)gj(Ai), over which it is reasonable to reject the hypothesis of outranking of Ai over Ak, with regard to the considered criterion. It implies
that:
qj pj vj
(7)
The above thresholds are not experimental values, of which the exact score is
required, but they are suitable quantities that experts introduce in order to make the
methodology more flexible, taking into account data uncertainty and approximation [9].
2067
Ak is strictly preferred to Ai
Ak is weakly preferred to Ai
(8)
(9)
(10)
where pj and qj are the strict preference and indifference thresholds, respectively. In
other words, cj(Ai, Ak) shows the degree of concordance with the judgmental statement that Ai outranks (is at least as good as) Ak. It decreases linearly from the top
level as soon as gj(Ak) has passed the indifference threshold, and it arrives at the
bottom level as soon as gj(Ak) has reached the preference threshold.
The index of global concordance Cik represents the amount of evidence to support
the concordance among all the criteria, under the hypothesis that Ai outranks Ak [11].
It is defined as the weighted average of all cj(Ai,Ak) F, with regard to the statement
that Ai outranks Ak:
Cik
Wjcj(Ai, Ak)
j1
(11)
Wj
j1
c, g, l
d
1
2
3
4
5
6
7
3
1
b, f, i, j 4
e
1
a, h
2
k
1
Criteria
Rankinga
1, 2, 3
4
(5)
6, 7, 8, 9
10
11, 12
13
86b
Weight W
Table 1
Example of criteria weighting according to Simos approach
(6 + 7+8 + 9) /4 = 7.5
10
(11 + 12) / 2 = 11.5
13
(1 + 2+3) /3 = 2
4
Average weights W = W /
Nr
9
12
13
15
2
5
49 = 36
112 = 12
213 = 26
115 = 15
100
32 = 6
15 = 5
Test
2068
M. Beccali et al. / Renewable Energy 28 (2003) 20632087
2069
(12)
(13)
dj(Ai,Ak) 0gj(Ak)gj(Ai) pj
(14)
dj(Ai, Ak) shows the degree of discordance with the judgmental statement that Ai
outranks Ak. It increases linearly from the bottom level as soon as gj(Ak) has passed
the preference threshold, and it arrives at the top level as soon as gj(Ak) has reached
the veto threshold [12].
3.7. Flexibility in the outranking relation
Flexibility allows verification if the outranking relation between two alternatives
is indisputable, not very convincing, or included between the previous conditions.
It is expressed as an index dik, termed credibility degree of outranking. As regards
to Ai and Ak, it is defined as:
1dj(Ai, Ak)
when dj Cik
1 Cik
jF
dik Cik
(15)
2070
if Ai is preferred to Ak in both two orders, then it is also preferred in the final order;
if Ai and Ak are indifferent in one of the two orders, while Ai is preferred to Ak
in the other one, then Ai is preferred to Ak in the final order, too;
if Ai is preferred to Ak in one of the two orders, while Ak is preferred to Ai in the
other one, then Ai and Ak are incomparable in the final order.
The rank order of the alternatives is presented in a diagram, where scores on the
two reference axis represent the position of alternatives, derived from both distillation
phases. The best alternatives are situated on the upper right side, while the worst
alternatives are positioned on the bottom left of the diagram. The more an alternative
is far from the bisecting line, the more it will be incomparable with the others.
The diagram area, with regard to the best actions, contains the following actions
(Fig. 1):
excellent actions in the two orders
Fig. 1.
2071
Energy source
Technology/Action
Size
1
2
Solar energy
Small
Mediumlarge
Wind energy
Hydraulic energy
6
7
8
Biomass
9
10
Animal manure
Energy saving in
residential and
industry sectors
11
12
13
14
CHP
Mediumlarge
2072
selected actions. The order number in the first column will be used to synthesize
the position of each action in the diagram.
4.2. Definition of evaluation criteria
A process of diffusion of an innovative technology needs the following requirements:
compatibility with political, legislative and administrative situation;
consistence with the local technical and economic condition, which depends on the
local capacity of managing the innovation both at technical and financial levels;
consistence with energy demand predictions, which will have to confirm or reject
the expectations of lasting development for the considered innovation;
compatibility with the existing environmental and ecological constraints.
According to the above considerations, 12 criteria are identified and collected in
Table 3.
Table 3
Groups of criteria
Technological criteria
Sustainability according to
greenhouse pollutant emissions
Sustainability according to other
pollutant emissions
Land requirement
labour impact
Market maturity
Compatibility with political,
legislative and
administrative situation
2073
2074
(18)
It must be highlighted that earnings from the sale of energy are not considered in
this parameter. In this way the effects of tariff policies are avoided.
5.6. Sustainability according to greenhouse pollutant emissions (criterion f)
This criterion is introduced to measure the equivalent emission of CO2, which is
avoided by the examined action. Therefore it is a reference index, expressed in
grCO2/MJ of saved primary energy. Also in this case, reference volumes of emission
of substituted conventional technologies have been considered.
5.7. Sustainability according to other pollutant emissions (criterion g)
Pollutants are divided into the following categories:
air emissions mainly due to combustion process;
liquid wastes, which are associated mainly with secondary products by fumes
treatment or with process water;
solid wastes, which are generated during the life cycle of actions.
Type and quantity of emissions, and costs associated with wastes treatments are
assessed. In order to have a synthetic index, the score is expressed through the following qualitative scale of values:
1.
2.
3.
4.
2075
2076
1 Domestic
solar water
heaters
2 Solar water
heating for
large
demands at
low
temperatures
3 PV roofs:
grid
connected
system
generating
electric
energy
4 Wind
turbines
(grid
connected)
Alternatives
649
1842
2790
Technical
maturity,
reliability
(1-5)
1255
Targets of
primary
energy
saving in
regional
scale
[TJ/year]
Consistence
of
installation
and
maintenance
requirements
with local
technical
know-how
(1-5)
Continuity
and predictability of
performances (1-3)
0.013
0.077
0.022
0.017
Cost of
saved
primary
energy
( /MJ)
Table 4
Matrix of evaluation of alternatives, according to the fixed criteria
48
48
63.6
49
Sustainability
according
to
greenhouse
pollutant
emissions
(gCO2/MJ)
Sustainability
according
to other
pollutant
emissions
(1-4)
10.0
Sustainability
according to
other
environmental
impacts (15)
0.0
0.0
0.0
Land
requirement
[m2/KW]
30
268
160
160
Compatibility
with
political,
legislative
and
administrative
situation (13)
labour
Market
impact [n. maturity
engaged
(1-5)
persons/MJ
year]
5 hydro plants
in derivation
schemes
6 hydro plants
in existing
water
distribution
networks
7 high
efficiency
wood boilers
8 CHP plants
fed by
agricultural
wastes or
energy crops
9 CHP plants
fed by
biogas
Alternatives
574
921
1884
1047
Technical
maturity,
reliability
(1-5)
574
Targets of
primary
energy
saving in
regional
scale
[TJ/year]
Table 4 (continued)
Consistence
of
installation
and
maintenance
requirements
with local
technical
know-how
(1-5)
Continuity
and predictability of
performances (1-3)
0.028
0.015
0.003
0.004
0.004
Cost of
saved
primary
energy
( /MJ)
55.8
56.7
63.6
48
48
Sustainability
according
to
greenhouse
pollutant
emissions
(gCO2/MJ)
Sustainability
according
to other
pollutant
emissions
(1-4)
12.5
70.0
0.3
3.5
0.0
Sustainability
according to
other
environmental
impacts (15)
Land
requirement
[m2/KW]
20
45
1000
1200
labour
Market
impact [n. maturity
engaged
(1-5)
persons/MJ
year]
Compatibility
with
political,
legislative
and
administrative
situation (13)
2078
M. Beccali et al. / Renewable Energy 28 (2003) 20632087
10 Building
insulation
11 high
efficiency
lighting
12 high
efficiency
electric
householders
appliances
13 high
efficiency
boilers
14 Plants
coupled with
refrigerating
adsorption
machines
Alternatives
2784
2215
1172
703
Technical
maturity,
reliability
(1-5)
4187
Targets of
primary
energy
saving in
regional
scale
[TJ/year]
Table 4 (continued)
Consistence
of
installation
and
maintenance
requirements
with local
technical
know-how
(1-5)
Continuity
and predictability of
performances (1-3)
0.005
0.014
0.009
0.002
0.021
Cost of
saved
primary
energy
( /MJ)
55.8
63.6
48
48
63.6
Sustainability
according
to
greenhouse
pollutant
emissions
(gCO2/MJ)
Sustainability
according
to other
pollutant
emissions
(1-4)
0.3
Sustainability
according to
other
environmental
impacts (15)
0.0
0.0
0.0
0.0
Land
requirement
[m2/KW]
294
44
labour
Market
impact [n. maturity
engaged
(1-5)
persons/MJ
year]
Compatibility
with
political,
legislative
and
administrative
situation (13)
2080
Table 5
Levels of priority in the three decisional scenarios
Environmental-oriented scenario
Economy-oriented scenario
High priority
Sustainability according to
greenhouse pollutant emissions
High priority
Compatibility with political,
legislative and administrative
situation
Market maturity
High priority
Continuity and predictability
of performances
Middle priority
Sustainability according to
greenhouse pollutant
emissions
Sustainability according to
other pollutant emissions
Land requirement
Low priority
Sustainability according to
greenhouse pollutant emissions
Low priority
Market maturity
Labour impact
Compatibility with political,
legislative and administrative
situation
actions. It can be supposed that it assigns the highest importance to labor impact
among the social and economical criteria, locating them in the middle priority rank.
The other criteria of the same group are associated with the low priority rank.
All the criteria, which assess technical reliability of actions, are assigned to the
middle priority rank, except for the criterion of energy saving targets. In particular,
the reduction of fossil fuel consumption represents not only an economic target, but
also one of the most relevant issue of environmental sustainability.
2081
8. Economy-oriented scenario
One of the main targets of decision-makers is the promotion of RET, as a means
to increase enterprise capacity and generate new profit. In this sense, the possibility
to create new employments, the economical efficiencies of projects and the consistence with the constraints and the legislative facilities represent the most relevant criteria.
Besides, a good judgment on market maturity implies that the selected actions
show a high capability of market penetration.
The technical maturity criterion describes the reliability of a given technology,
which is associated with the safety of investment. The cost of primary energy saved
is a reliable indicator for the energetic and economical efficiency of the examined
technology.
The two criteria land requirement and consistence with local technical know-how
are assigned in the middle priority rank. The first one is an environmental indicator,
which also has an economic implication. In fact, the increase of the occupied land
often involves an increase of initial operating costs. The last one measures the presence of a local technical know-how, suitable to allow the introduction of the given
technology. The remaining criteria are associated with the low priority rank.
2082
does not assume the highest priority in the global evaluation, while the last one takes
value when it reaches the highest priority (Table 6).
11. Results
11.1. Environmental-oriented scenario
The outcomes of the distillation procedure are the final order for each decisional
scenario. In each order a best actions area is defined as the area within which the
best alternatives are placed for both distillations. These alternatives represent the
actions that fulfill the objectives that the decision-maker has fixed.
In Fig. 2 the dotted area at the top right is the best alternative area. Actions 1, 4,
6, 10, 11 and 12 belong to such an area. As regards to the excluded ones, actions
2 and 7 are the nearest to the best actions area.
Action 13 is the only excluded action that deals with building energy saving. The
other alternatives are excluded because their performances are too low in the most
significant criteria for the considered scenario. However, they also have a consistent
weight in the criteria of middle and high priority. For example, action 3 is excluded
from the best actions area, but it reaches high performances on average for the priority criteria in the considered scenario.
11.2. Economy-oriented scenario
Actions 1, 2, 5, 6, 10, 11, 12 and 13 are the best ones (Fig. 3). Action 4 is the
best among the excluded ones, mainly depending on the evaluation in the criteria
of land requirement and labor impact. It must be remembered that earnings from
energy sales are not considered.
The low score of actions 11, 12, and 13 in labor impact is balanced by the good
values that these actions reach in the criterion cost of saved primary energy.
Table 6
Thresholds of veto, indifference and preference for the different criteria
a
3
0
0.5
3
0
0.5
100
5
15
120
20
40
4
0
0.5
71
0
3
5
0
0.5
1000
0
100
5
0
0.5
3
0
0.5
1.9
0
0.5
20
0
10
50
0
5
1.9
0
0.5
0.1
0
0.1
1.9
0
0.5
1000
0
100
1.9
0
0.5
1.9
0
0.5
Fig. 2.
2083
12. Discussion
Table 7 shows a comparison among the results of the aggregation procedure for
the three examined scenarios. It can be noted that the following actions always belong
to the best actions zone:
action
action
action
action
6;
10;
11;
12.
2084
Fig. 3.
action
action
action
action
action
2;
4;
5;
7;
13.
Therefore, the four most recurring actions and the best excluded one are characterized by more relevant robustness than the others. In other words, they are not
much dependent on weights variation or other constraints that characterize the three
scenarios. They are considered consistent with the priority expectations of all the
three decision scenarios, each one representing a hierarchy of different values and
targets.
The last group of actions (actions 2, 4, 5, 7, and 13) contains potential best
Fig. 4.
2085
Outcomes of distillation procedure in the energy saving and rational use scenario.
Table 7
Panel of results
Scenario
1. Environmentaloriented scenario
2. Economyoriented scenario
3. Energy saving and
rationalization scenario
1, 4, 6, 10, 11, 12
1, 2, 5, 6, 10, 11, 12, 13
6, 7, 10, 11, 12
2, 7
4, 14
1, 5, 13, 14
actions, for which an improvement in one of the priority criteria could bring a
significant variation in the overall evaluation.
13. Conclusions
A MDMM is applied in order to assess groups of actions focused on the implementation of RET innovative technologies voted to use energy renewable resources. The
2086
References
[1] Roy B. Me todologie multicrite`re daide a` la de cision. Economica, Collection Gestion, Paris 1985.
[2] Lootsma FA, Schuyt H. The multiplicative AHP, Smart, and ELECTRE in a common context. The
Journal of Multi-Criteria Decision Analysis 1997;6.
[3] Eom Hyun BT. he current state of multiple criteria decision support systems. Human Systems Management 1989;8:1139.
[4] Beinat E. Value functions for environmental management. The Netherlands: Kluwer Academic Publishers, 1997; 239 pp.
[5] Maystre L, Pictet J, Simos J. Me thodes multicrite`res Electre. Lusanne: Presse Polytechniques et
Universitaires Romandes, 1994.
[6] Roy B, Bouyssou D. Comparison of two decision-aid models applied to a nuclear power plant siting
example. Cahier 47, LAMSADE, Universite de Paris-Dauphine.
[7] Roy B, Bouyssou D. Aide multicrite`re a la de cision: me thodes et cas. Economica, Collection Gestion,
Paris 1993.
[8] Ostanello Borreani A. Elementi di analisi multicriteriale e teoria di aiuto alla decisione. Appunti
del corso di ricerca operativa, anno accademico 1977/1978. Torino: Libreria Editrice Universitaria
Leprotto & Bella, 1983; 104.
[9] Cusin R. Methodes mathematiques de laide multicritere a la decision. Chania: GREQE Groupe de
recherche en Economie Quantitative et Econometrie, 1993; 188203.
[10] Simos J. Evaluer limpact sur lenvironnement. Une approche originale par lanalyse multicrite`re de
ne gotiation. Lausanne: Presse Polytechniques et Universitaires Romandes, 1990.
[11] Triantaphyllou E. Multi-criteria decision making methods: a comparative study. Dordrecht, The
Netherlands: Kluwer Academic Publishers, 2000; Vol. 44, 263 pp.
[12] Lootsma Freerk A. In: Fuzzy logic for planning and decision making. P.O. Box 17, AA Dordrecht,
The Netherlands: Kluwer Academic Publishers; 1997. p. 14960.
[13] Regione Autonoma Sardegna. Sistema informativo per lenergia: Bilancio energetico e quadro di
riferimento per la pianificazione energetica regionale. Snam Progetti and Enel/CESEN, 1997.
[14] Ciocca B, Zucca C. Cogenerazione di energia elettrica, termica e frigorifera per la comunita` di S.
Patrignano. In: Proceedings of Aicarr Conference Le centrali frigotermoelettriche negli impianti di
climatizzazione: esperienze e idee, Padova 1994.
[15] Ambiente Italia. Sustainable alternative fluent energy based development of small Mediteranean
islands. Istituto di ricerche, Cec DG XVII, Milano, 1995.
2087
[16] Beccali M. Nuove tecnologie energetiche e sviluppo sostenibile, un approccio multicriteria per la
valutazione delle probabilita` di successo di una pianificazione innovativa. PhD thesis, 1994.
[17] Roy B. Classement et choix en pre sence de point de vue multiple (la me thode Electre). Revue
Informatique et Recherche Operationelle 1968;8.
[18] Beccali G, Cellura M, Mistretta M. A decision support system software based on multi-criteria
analysis for the selection of urban sustainability scenarios. In: Proceedings of the International Conference RIO 02 World Climate and Energy Event. 2002. p. 3018.