Fuel
journal homepage: www.elsevier.com/locate/fuel
h i g h l i g h t s
" Decision making in the electricity generation sector comprises a complex process.
" Evaluation of technologies to support power generation is based on several criteria.
" Delphi method is used to determine importance of criteria based on experts opinion.
" A multi-criteria analysis is then developed to provide ranking of technologies.
" Developed methodology is then applied using as a case study the island of Crete.
a r t i c l e i n f o a b s t r a c t
Article history: Decision making process in selecting the most suitable technological solutions for electricity power gen-
Received 6 November 2011 eration in terms of strategic planning comprises a rather complex procedure with several contradicting
Received in revised form 25 November 2012 factors involved. In this context, development of an appropriate evaluation methodology that will provide
Accepted 26 November 2012
decision makers with a useful tool is the aim of the specic study. To achieve this, the Delphi method is
Available online 19 December 2012
used in order to compare a number of alternative technologies with respect to several characteristics, e.g.
cost, environmental impacts, social impacts and technological status. According to this method, a prop-
Keywords:
erly formed questionnaire is sent to a number of experts, currently related to the power generation sec-
Electricity generation technologies
Decision making
tor. After the evaluation of results, weight factors are determined, so that the different technologies can
Weight factors be ranked according to both their scoring in each criterion category and their global scoring, indepen-
Crete dently of categories, on the basis of a multi-criteria analysis. Emphasis is currently given on the electricity
generation sector of Greece, with application of the developed methodology carried out for Crete, i.e. the
biggest Greek island.
2012 Elsevier Ltd. All rights reserved.
1. Introduction throughout this long period of study has been satised mainly on
the basis of thermal power plants (Fig. 2), while contribution of
National gross electricity consumption of Greece may be deter- large hydropower units should also be taken into account [5] (serv-
mined by a long-term annual increase rate in the order of 3.3% ing however as peak power plants). In the meantime, contribution
since the early 90s (see also Fig. 1), although according to the latest of the rest of renewable energy sources (RESs) is relatively re-
ofcial data [1] a small decrease has been recently noted (2008 stricted [6,7] (given that the RES potential of the entire Greek re-
2009), mainly owed to the impacts of economic recession. At the gion is of mediumhigh quality), with the wind and PV power
same time, reliance of the Greek electricity generation sector shares only recently exceeding the capacity of 1.7 GW [8] and
mainly on the local lignite deposits [2] as well as on imported fossil 1.0 GW respectively [7].
fuels (i.e. oil and natural gas (NG)), implies electrical energy depen- By acknowledging issues such as excess GHG emissions (consid-
dence at the levels of 3540% [3] along with considerable produc- ering Greeces commitment to decrease the corresponding emis-
tion of greenhouse gas (GHG) emissions [4]. Electricity demand sions in 2020 by 4% in comparison to the 2005 levels [9]), high
levels of energy dependence (a long-term, primary energy depen-
dence of around 70% must be considered), gradual retirement of al-
Corresponding author. Tel.: +30 210 5381237; fax: +30 210 5381467. ready existing power stations and nally urgency of meeting RES
E-mail address: jkald@teipir.gr (J.K. Kaldellis). targets adopted at the country level (e.g. 20% and 40% of RES con-
URL: http://www.sealab.gr (J.K. Kaldellis). tribution to the national gross energy and electricity consumption
0016-2361/$ - see front matter 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.fuel.2012.11.073
J.K. Kaldellis et al. / Fuel 106 (2013) 212218 213
2
3
94
95
96
97
00
01
02
03
04
05
06
07
98
99
08
09
9
9
19
19
19
19
19
19
19
19
20
20
20
20
20
20
20
20
19
19
20
20
based on a nancial objective alone, such as minimum costs or
Year
maximum prots, since additional factors like environmental per-
Fig. 1. Time evolution of gross electricity consumption in Greece. formance of each technology and accruing energy dependence
levels also need to be taken into account. As already implied, one
efcient way to deal with such problems, determined by various
possibly conicting objectives, is to dene multiple alternative
Time Evolution of Installed Power Capacity in Greece decision criteria under the umbrella of a multi-criteria analysis.
16.000
Pumped Hydro & PVs More specically, multi-criteria analysis is a term used to de-
14.000 Wind scribe techniques that use more than one criterion to evaluate
Power Capacity (MW)
Hydro
12.000 and judge performance [18]. Multi-criteria decision support tech-
Fossil Fuels
10.000 niques typically involve weighing of the criteria to reect the rela-
8.000 tive importance attributed to each of them, with examples of such
6.000 techniques ranging from simplistic rate and weight to rigorous
4.000
multi-objective optimization. Overall, the scope of such analyses
is to assign each alternative solution with a rate against each of
2.000
the criteria (within a predetermined range of values), after assign-
0
ing a weight factor to each of the criteria, adding up to 100%. In fact,
03
90
91
92
93
94
95
96
97
98
99
00
01
02
04
05
06
07
08
09
Table 1
Results of the Delphi method application.
Criteria Round 3
Median value 1st Quartile 3rd Quartile Weight factor
Environment
High gaseous and particulate (atmospheric) emissions 10 9 10 0.0926
Hot waste water disposal 7 6 8 0.0648
Solid waste generation/disposal 7 5 9 0.0648
Land resource use/exploitation/rehabilitation 7 5 8 0.0648
Ozone layer depletion 8 7 9 0.0741
Eutrophication (sea, lakes, etc.) 7 6 8 0.0648
Water and land contamination 8 8 9 0.0741
Smog generation 8 7 9 0.0741
Disruption to local ecosystems 8 7 8 0.0741
Geological disruption 6 4 8 0.0556
Ecotoxicity 7 7 9 0.0648
Microclimate change 7 6 8 0.0648
Positive impacts (emissions reduction) 10 9 10 0.0926
Solid and liquid waste management (secondary usage) 8 8 9 0.0741
Society
Severe health hazards 10 10 10 0.1351
Possible people relocation 8 6 9 0.1081
Accidents 7 6 8 0.1081
Noise 7 6 7 0.0946
Visual impacts 7 6 8 0.0946
Changes in cultural/ecological behavior 9 8 10 0.1216
High rate of employment in local societies 8 7.5 9 0.1081
Security of supply 9 9 10 0.1216
Regional development 9 8.75 9 0.1216
Finance
High paid cost/tn of CO2 10 9 10 0.0877
Fuel cost dependency (price uctuations) 10 9 10 0.0877
High construction cost/long payback period 8 8 9 0.0702
Material degradation due to pollution (buildings, etc.) 8 7 8 0.0702
High dismantling/transporting/storage cost 6 5 8 0.0526
Money savings (less usage of imported fuels) 9 8 10 0079
Low external costs 8 8 9 0.0702
Collateral nancial benets (nancial exploitation of by-products) 7 6 8 0.0614
No fuel cost 8 7 9 0.0702
Low labor cost 6 5 7 0.0526
Cheap electricity price for the consumers 8 7 9 0.0702
Electricity price stability 9 8 9 0079
Economic growth (for the country or a region) 9 7 9 0079
Low life cycle cost/attractive return on investment 8 8 9 0.0702
Technology
System efciency 10 8 10 0.2941
Capacity factor 8 8 9 0.2353
Fuel availability 9 9 9 0.2647
Existing experience 7 7 8 0.2059
(3) Financial impacts (initial investment cost, operational cost, Considering the above, estimation of weight factors rst at the
etc.). level of category and then at the level of criteria involved (as these
(4) Technological status (efciency, capacity factor, etc.). resulted from the application of the Delphi method, see also Ta-
ble 1) is given in Figs. 59, while accordingly, for the ranking of
Accordingly (see also Fig. 3), for each of the criteria involved technologies, a 04 scale is adopted (0: when a technology is not
in each category, a weight factor is estimated with the applica- related to a criterion at all, or presents a poor performance on it;
tion of the Delphi method [1921] (see also Appendix A), with 4: when a technology has an absolute relation to a criterion, or pre-
the help of experts (total number of 30 experts) currently coming sents an excellent performance on it).
from the academic eld, the Greek Public Power Corporation Synopsizing, inputs of the developed evaluation model include
(PPC) and nally from certain national energy research centres the following:
(Fig. 4). At this point, it must be noted that Delphi comprises a
exible enough method that allows engagement of experts, Weight factors for the importance of the criteria categories
allowing at the same time consensus-building that eliminates (environmental, social, nancial and technological), we, ws,
any bias-driven answers, potentially deriving from the experts wf and wt respectively.
level of commitment to their expertise. During this process, the Weight factors for the importance of each criterion i within
experts are asked to use a 110 scale when providing their an- each category (environmental, social, nancial and technologi-
swers to the questionnaire, with 1 corresponding to the lowest cal), wei, wsi, w and wti respectively.
importance and 10 to the highest (see also Appendix A and Values vij representing performance of each technology j
Table 1). concerning each criterion i.
J.K. Kaldellis et al. / Fuel 106 (2013) 212218 215
7%
13% 6%
5%
4%
3%
2%
70% 1%
PPC 0%
Ozone layer depletion
(emissions reduction)
Geological disruption
generation / disposal
Microclimate change
Land resource use
Disruption to local
High atmospheric
Smog generation
Hot waste water
contamination
Ecotoxicity
management
emissions
Solid waste
disposal
Estimation of Weight Factors for the Four Different Estimation of Social Criteria Weight Factors
15%
Thematic Areas of Evaluation
30%
12%
25%
Weight Factor
9%
Weight Factor
20%
6%
15%
10% 3%
5% 0%
Noise
impacts
Security of
employment
Accidents
development
relocation
Severe health
Changes in
High rate of
Possible
Visual
behavior
people
Regional
supply
hazards
0%
nvironmental Social impacts Financial impacts Technology issues
impacts
Fig. 5. Weight factors assigned to each category. Fig. 7. Weight factors assigned to social criteria.
216 J.K. Kaldellis et al. / Fuel 106 (2013) 212218
Estimation of Financial Criteria Weight Factors Distribution of Crete's Installed Power Capacity (2008-MW)
10%
9% 165 161
8%
Weight Factor
7%
6%
151
5%
4%
3%
2%
1%
0% 479 Steam Generators
Economic growth
Material degradation
Cheap electricity
Low external costs
High paid cost / tn of
Collateral financial
dependency
Money savings
Electricity price
No fuel cost
Fuel cost -
Diesel Units
attractive ROI
stability
Gas Turbines
benefits
price
cost
CO2
Wind Turbines
Fig. 10. Contribution of different power sources in the current electrical power mix
of Crete.
Fig. 8. Weight factors assigned to nancial criteria.
20% 600
15% 500
400
10% 300
5% 200
100
0% 0
experience
availability
efficiency
Capacity
Existing
factor
Fuel
Year
Fig. 11. Time evolution and forecast of annual peak load demand in Crete.
Fig. 9. Weight factors assigned to technological criteria.
The developed methodology is accordingly applied using as a Concerning solar thermal power systems [23,24], energy is col-
case study the island of Crete, so as to investigate different techno- lected by concentrating collectors and is used to operate a heat en-
logical solutions for solving the future power generation problem. gine. The main problem of these systems is that the efciency of
Crete is the fourth largest island in Mediterranean and the biggest the collector decreases as its operating temperature increases,
island of Greece, providing an excellent case study in terms of both while the performance of the heat engine increases as its operating
scale and diverse electricity fuel mix. At this point, it should also be temperature increases. The most promising conguration
J.K. Kaldellis et al. / Fuel 106 (2013) 212218 217
0,5 by the three RES technologies. On the other hand, diesel generators
present the lowest rate, clearly owed to their poor performance in
0,0
the category of environment, which was also assigned with the
-0,5
Environmental Performance highest importance by the experts.
-1,0 Social Performance
Financial Performance
-1,5 Technological Performance
Overall Score
6. Conclusions
-2,0
Solar Thermal Power Cryogenic NG Wind Power PV Systems Diesel Power
Synopsizing, an effort was currently undertaken in order to de-
Fig. 12. Final score (per category and overall) for the ve technologies examined. velop an integrated methodology for the support of decision
218 J.K. Kaldellis et al. / Fuel 106 (2013) 212218
making in the eld of new electricity generation plants. More pre- In this context, resulting weight factors are given in Table 1,
cisely, based on the adoption of multiple criteria for the selection of with negative impacts reected by setting a minus sign (i.e. a
the most appropriate power plants called to satisfy future electri- negative impact should count in an algebraically negative way in
cation needs, a multi-criteria model has been developed. Prior the calculations).
to that, evaluation criteria were assigned with weight factors
decided by the application of the Delphi method. For this purpose, References
a number of eld experts were used in order to assess these weight
factors, which were accordingly applied on the basis of the multi- [1] Eurostat. Energy statistics-supply, transformation, consumption; 2011.
<http://epp.eurostat.ec.europa.eu/portal/page/portal/eurostat/home/>.
criteria analysis to ve different power generation technologies, [2] Kaldellis JK, Zarakis D, Kondili E. Contribution of lignite in the Greek
being candidate for satisfying part of the Cretes island future elec- electricity generation: review and future prospects. Fuel 2009;88:47589.
tricity needs. Results obtained demonstrate the effectiveness of the [3] Chalvatzis KJ, Hooper E. Energy security vs. climate change: theoretical
framework development and experience in selected EU electricity markets.
proposed methodology, which may however be further elaborated Renew Sust Energ Rev 2009;13:27039.
so as to include a greater number of both evaluation criteria and [4] Tsoutsos Th, Papadopoulou E, Katsiri A, Papadopoulos AM. Supporting schemes
experts involved, expanding at the same time to also capture the for renewable energy sources and their impact on reducing the emissions of
greenhouse gases in Greece. Renew Sust Energy Rev 2008;12:176788.
entire electricity power generation system of Greece. [5] Kaldellis JK. Critical evaluation of the hydropower applications in Greece.
Renew Sust Energ Rev 2008;12:21834.
[6] Kaldellis JK. Investigation of Greek wind energy market time-evolution. Energ
Appendix A Policy 2004;7:86579.
[7] Hellenic Association of Photovoltaic Companies. The Greek PV Market. Greek
Delphi is a consensus method. This normally means that at least PV market statistics: H1-2012; 2012. <http://www.helapco.gr/>.
[8] Hellenic Wind Energy Association. Installations-Wind Power Installations in
three rounds of answers to the questionnaire should be expected, Greece-2012; 2012. <http://www.eletaen.gr/>.
although additional rounds may be needed sometimes. During the [9] European Parliament and Council. Decision No. 406/2009/EC on the effort of
rst round, the questionnaire is circulated to the panel of experts. Member States to reduce their greenhouse gas emissions to meet the
Communitys greenhouse gas emission reduction commitments up to 2020;
When answers are collected, the median value and the inter-quar- 2009. <http://eur-lex.europa.eu/>.
tile range are calculated for each answer. Then, results of the rst [10] Kalampalikas NG, Pilavachi PA. A model for the development of a power
round (median and inter-quartile range) are circulated again to production system in Greece, Part II: Where RES meet EU targets. Energ Policy
2010;38:651428.
those experts, whose answers during the rst round fell outside
[11] Point Carbon. European Emission Trading Scheme (EU-ETS) News; 2012.
the inter-quartile range. This time they are asked to give reasons <http://www.pointcarbon.com/news/reutersnews/1.2059305>.
for their answers and to revise their former given values in case that [12] Kaldellis JK, Mantelis N, Zarakis D. Evaluating the ability of Greek power
stations to comply with the obligations posed by the second National
they wish to do so. After receiving the second round forms, the med-
Allocation Plan concerning carbon dioxide emissions. Fuel 2011;90:288495.
ian value and the inter-quartile range are recalculated, while a list is [13] Kaldellis JK. Maximum wind energy contribution in autonomous electrical
also formed with the reasons that the experts gave, explaining their grids based on thermal power stations. Appl Therm Eng 2007;27:156573.
answers. All this data is again circulated to the experts for round [14] Lund H. Large-scale integration of wind power into different energy systems.
Energ 2005;30:240212.
three, asking them to proceed with their nal evaluation. [15] Geman H, Ohana St. Forward curves scarcity and price volatility in oil and
Taking into account the above, in Table 1, the round three med- natural gas markets. Energ Econ 2009;31:57685.
ian value and inter-quartile range for each criterion evaluated are [16] Pohekar SD, Ramachandran M. Application of multi-criteria decision making to
sustainable energy planning a review. Renew Sust Energ Rev 2004;8:36581.
presented, considering that it is the specic round during which [17] Wang JJ, Jing YY, Zhang CF, Zhao JH. Review on multi-criteria decision analysis
consensus was achieved by the experts. The next step corresponds aid in sustainable energy decision-making. Renew Sust Energ Rev
to the estimation of weight factors through the normalization of 2009;13:226378.
[18] Zopounidis C, Pardalos PM. Handbook of multicriteria analysis series: applied
the median values of the last round. More precisely, by using the optimization. 1st ed. Berlin: Springer Verlag; 2010.
symbols me, ms, mf and mt for the median values of the [19] Okoli Ch, Pawlowski SD. The Delphi method as a research tool: an example,
environmental, social, nancial and technology categories respec- design considerations and applications. Inform Manage 2004;42:1529.
[20] Landeta J. Current validity of the Delphi method in social sciences. Technol
tively, the corresponding weight factors may be estimated by:
Forecast Soc 2006;73:46782.
me [21] Rowe G, Wright G. The Delphi technique: past, present, and future prospects-
we A:1 introduction to the special issue. Technol Forecast Soc 2011;78:148790.
me ms mf mt [22] Greek Public Power Corporation. Annual Report on the Electricity Generation
System of Crete; 2008.
ms [23] Chien JCL, Lior N. Concentrating solar thermal power as a viable alternative in
ws A:2 Chinas electricity supply. Energ Policy 2011;39:762236.
me ms mf mt [24] Ummadisingu A, Soni MS. Concentrating solar power-technology, potential
and policy in India. Renew Sust Energ Rev 2011;15:516975.
mf [25] Montes MJ, Rovira A, Muoz M, Martnez-Val JM. Performance analysis of an
wf A:3 integrated solar combined cycle using direct Steam Generation in parabolic
me ms mf mt trough collectors. Appl Energ 2011;88:322838.
[26] Kaldellis JK, Zarakis D, Kondili E. Energy pay-back period analysis of stand-
mt alone photovoltaic systems. Renew Energ 2010;35:144454.
wt A:4 [27] Kaldellis JK, Zarakis D, Kondili E. Optimum autonomous stand-alone
me ms mf mt photovoltaic system design on the basis of energy pay-back analysis. Energy
2009;34:118798.
Similarly, the weight factor for each criterion i within each [28] Weisser D. A guide to life-cycle greenhouse gas (GHG) emissions from electric
criterion category can be estimated by the following expression: supply technologies. Energy 2007;32:154359.
[29] Kaldellis JK, Kavadias KA, Filios AE, Garofallakis S. Income loss due to wind
jmi j energy rejected by the Crete island electrical network-the present situation.
wi P A:5 Appl Energy 2004;79:12744.
i jmi j [30] Kaldellis JK. Maximum wind potential exploitation in autonomous electrical
networks on the basis of stochastic analysis. J Wind Eng Ind Aerod
where mi is the median value of each criterion and + is used for 2008;96:141224.
positive and for negative contribution of each criterion in the [31] Kumar S, Kwon HT, Choi KH, Lim W, Cho JH, Tak K, et al. LNG: an eco-friendly
cryogenic fuel for sustainable development. Appl Energy 2011;88:426473.
evaluation of the solution under investigation. Note that this
[32] Georgakellos DA. Impact of a possible environmental externalities
assumption is used instead of using negative marks in order to facil- internalisation on energy prices: the case of the greenhouse gases from the
itate the entire calculation. Greek electricity sector. Energy Econ 2010;32:2029.