You are on page 1of 8

Energy xxx (2013) 1e8

Contents lists available at ScienceDirect

Energy
journal homepage: www.elsevier.com/locate/energy

A quantitative discussion on the assessment of power supply


technologies: DEA (data envelopment analysis) and SAW (simple
additive weighting) as complementary methods for the “Grammar”
Hamed Shakouri G. a, b, *, Mahdis Nabaee a, Sajad Aliakbarisani a
a
School of Industrial and Systems Engineering, College of Engineering, University of Tehran, Iran
b
Institute for Resource, Environment and Sustainability, UBC, Canada

a r t i c l e i n f o a b s t r a c t

Article history: The growing concern about the negative effects of fossil fuels on the environment, and the limited re-
Received 20 May 2013 sources of them have forced more intensive use of other energy sources. In absence of sufficient
Received in revised form economically feasible renewable energies, nuclear power may play essential role in this field. Recently,
27 August 2013
the advantages and disadvantages of nuclear power and fossil fuels regarding their efficiencies have been
Accepted 5 October 2013
Available online xxx
attracted researchers’ interest. This paper discusses on the findings from “A Grammar for assessing the
performance of power supply systems: comparing nuclear energy to fossil energy” (Diaz-Maurin F,
Giampietro M. 2013). Although the “Grammar” is a very valuable approach, it can be accomplished by
Keywords:
Power generation technologies
using helpful quantitative methods. In this discussion, we apply quantitative decision-making ap-
Quantitative decision-making proaches to compare the same fossil fuel (coal) power plants with nuclear power plants. Economic
DEA variables are also taken into consideration. The DEA (data envelopment analysis) and SAW (simple ad-
SAW ditive weighting) are the methods applied. Results confirm the results of the reference paper in most
Nuclear power plant cases and show that the fossil fuel power plants with CCS (carbon capture and storage) are slightly more
Fossil-fuel power plant efficient than nuclear power plants. However, selection of input and output variables is disputable.
Assuming job creation as a desired output can change the ranking results.
Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction as France continued with nuclear energy and enjoyed success with
it. Though alternative energy is making progress, nuclear power
The growing concern in the negative effects of fossil fuels on the and fossil fuels will be important for decades to come.
environment and their exhaustibility has forced more intensive use The combustion of the fuels releases large amounts of CO2 and
of other energy resources, such as nuclear power. The generation of other pollutants to the atmosphere. Fossil resources are also
electricity worldwide is heavily dominated by the use of fossil fuels, limited, with finite amounts in existence. However, its technology is
in spite of blooming renewable energies, which yet make uncer- well-known, diversified and easily developed.
tainty problems in practice, and hence are improper as base load On the other hand, nuclear energy has a major problem named
power plants. radioactive emissions and the process of electricity generating
The shares of these three main categories of electricity resources through these power plants is very complicated. Moreover, it
have not faced a large change during the last three decades and confronts scarcity of primary materials. As an advantage, a tiny part
have been around 65%, 10% and 25% for fossil, nuclear and renew- of it can produce tremendous amounts of energy.
able (with hydro dominance), respectively. Though fossil fuels are Now some questions arise: which kind of these technologies is
deeply entrenched in the world’s economy, problems such as global more preferable? Do merely the equalized costs suffice to make a
warming have made many people take a second to look at nuclear judgment?
power. Events in the late 1970s turned public opinion against nu- In order to meet growing energy demands with minimal envi-
clear plants in the U.S., and their use declined. Some countries such ronmental impacts, some changes in the current energy generation
practices are essential. The changes need to include increased en-
ergy efficiency from fossil fuel combustion technologies through
introduction IGCC (Integrated Gasification Combined Cycle) tech-
* Corresponding author. #436, No. 2202, Main Mall, UBC, Canada. Tel.: þ1
6043587071. nologies. Analogous unit operations of the electricity process are
E-mail addresses: h.shakouri@ubc.ca, hshakouri@ut.ac.ir (H. Shakouri G.). essentially the same between nuclear energy and fossil fuels

0360-5442/$ e see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.energy.2013.10.022

Please cite this article in press as: Shakouri G. H, et al., A quantitative discussion on the assessment of power supply technologies: DEA (data
envelopment analysis) and SAW (simple additive weighting) as complementary methods for the “Grammar”, Energy (2013), http://dx.doi.org/
10.1016/j.energy.2013.10.022
2 H. Shakouri G. et al. / Energy xxx (2013) 1e8

Nomenclature ε a non-Archimedean value

n number of DMUs Abbreviations


m number of input parameters (attributes) DEA data envelopment analysis
c number of all criteria (m þ p) IE internal energy
p number of output parameters (attributes) SAW simple additive weighting
aij weight of technology i in criteria j RW radioactive wastes
j index for DMUj CCS carbon capture and storage
zij normalized weight of technology i in criteria j LWRP light water reactor power plant with processing
i index for input parameters (attributes) IGCC integrated gasification combined cycle
k inverse of logarithm of n IGCCS integrated gasification combined cycle with carbon
k index for output parameters (attributes) capture and storage
Pij estimate of the probability distribution for the PES primary energy resource
attributes OE output energy
ej efficiency of DMUj EC electricity
Ej weight of DMUj in SAW method LW light water reactor power plant
xij input i of DMUj MADM multi attribute decision making
dj degree of deviation of the data for index j CAPEX capital expenditure
ykj output k of DMUj DMU decision-making unit
wj weight of each criterion j FOM fixed operation and maintenance
vi weight of input i CC Charnes Cooper and Rhodes (DEA model)
Ai overall attribute of DMUi VOM variable operation and maintenance
uk weight of output k FF fossil fuel
zij complementary normalized attribute I for DMUj TIE total internal energy

energy. For that reason, it is easier to compare a nuclear power in the process of electricity production. The various phases of the
plant and a fossil-fuel power plant. process are almost the same. The only difference is related to the
There are many papers that compared different types of energy characteristics of the process associated with the generation of heat
sources like nuclear and renewable energy technologies [1]; or which are completely different in the two technologies. This report
contained information about nuclear and fossil fuels power plant at presented the evaluation of the biophysical requirements for two
the point of different aspects like costs [2], pollutants, CO2 emis- power plants: nuclear energy and fossil energy. It focuses on the
sions etc [2,3]. following biophysical requirements for both of them: (i) electricity,
Recently Diaz-Maurin and Giampietro (2013) have published a (ii) fossil fuels, (iii) labor, and (iv) materials [5].
paper entitled: “A grammar for assessing the performance of As a matter of fact, it becomes possible to compare the perfor-
power-supply systems: comparing nuclear energy to fossil energy”, mance of nuclear and fossil energies for making electricity because
that presented an approach for the characterization and compari- they have similar set of energy conversions taking place within
son of the performance of power-supply systems [4]. This paper different power-supply systems. Both of them have a set of similar
generated a double assessment referring to: (i) external conversions shown simply by Fig. 1.
constraints-the consumption of PES (primary energy resource) and
the generation of waste and pollution; and (ii) internal constraints-  Conversion #1: PES to heat (process heat or enthalpy)
the requirements of production factors such as human labor, power  Conversion #2: Heat to mechanical energy
capacity, and internal consumption of EC (electricity) for making  Conversion #3: Mechanical to gross electricity
EC. They provided the comparisons of generating EC with each of  Step #4: Gross to net electricity (final output of net supply of
nuclear and fossil energy. According to their findings, for both in- electricity, after subtracting internal electricity consumption)
ternal and external constraints, nuclear energy requires more re-
quirements than fossil energy. When considering internal It is worthy to mention that the internal use of heat and elec-
constraints, nuclear energy requires about twice as much as power tricity is apart from the loss in each conversion process which is
capacity (5.9 KW/GWhe9.5 KW/GWh vs. 2.6e2.9 KW/GWh) and 5e unavoidable even for the most ordered kind of energy conversion,
8 times more labor (570 h/GWhe640 h/GWh vs. 80 h/GWhe115 h/ i.e. mechanical energy.
GWh). Things do not improve for nuclear energy when looking at
external constraints e.g. the relative scarcity of PES, and various
types of radioactive wastes. The results of this study explain the 2. Assessment methodologies
difficulties faced by nuclear energy to gain interest from investors
[4]. There are some alternative methods helping decision-makers to
This paper is a discussion on these results, based on the same prefer a choice compared to other choices. For a proper decision
data, by applying some quantitative approaches. We have made our making, it is necessary to study characteristics of several alterna-
calculations based on the same information gained from Fig. 6 in tives and choose one or some among them based on the criteria,
Ref. [4] of the original paper, considering parameters of each case values and preferences of the decision maker [6]. There are always
and comparing the four technologies introduced there. some decision variables and some criteria in a decision-making
Diaz-Maurin has also illustrated an analysis to the process of problem. In this regard, since the Second World War, managers
producing electricity in modular elements for different power have paid a lot of attention to “optimization problems” with
plants systems, using semantic and formal categories [5]. In this considering an objective function, whereas nowadays, they pay
way, it becomes possible to individuate similarities and differences more attention to optimization problems with consideration of

Please cite this article in press as: Shakouri G. H, et al., A quantitative discussion on the assessment of power supply technologies: DEA (data
envelopment analysis) and SAW (simple additive weighting) as complementary methods for the “Grammar”, Energy (2013), http://dx.doi.org/
10.1016/j.energy.2013.10.022
H. Shakouri G. et al. / Energy xxx (2013) 1e8 3

Fig. 1. A simplified process for power plants; similar conversions from PES to net-electricity.

numerous criteria and/or objectives, some of which are so inhar- input, respectively. Here, ε is the predetermined smallest amount
monious in some cases. given to each weight of criteria; and ej, is the efficiency calculated
One of these methods is MADM (Multi Attribute Decision for each DMU.
Making) with its variety of models. These models are used when The DEA index can be calculated in various ways. In this study,
there are some alternatives for choosing, and some criterion and we start with a CCR (Charnes, Cooper and Rhodes) input-oriented
decision matrix are available. Generally, decision making process model. The classical DEA model for evaluating the efficiency of
can be divided into the following steps: the jth DMU, i.e. the CCR ratio, is [8]:
X
p
 Define the problem Max eo ¼ uk ykj (4)
 Determine requirements k¼1
 Establish goals
 Identify cases X
m
Subject to : vi xij ¼ 1 (5)
 Define criteria
i¼1
Select a decision making tool
There are many models for this purpose among which we have X X
m
uk ykj  vi xij ; cj (6)
selected SAW (simple additive weighting) and DEA (data envelop-
k¼1 i¼1
ment analysis) to making the comparison.

uk ; vj  ε; ci; k (7)
2.1. DEA (data envelopment analysis)

where the subscript o denotes the DMU which is under assessment.


Efficiency measurement has been a subject of tremendous in-
Many kinds of DEA are developed and discussed in the literature;
terest as organizations have struggled to improve productivity. In
however, it is not aimed in this study to use all for ranking the
the electricity-generating industry, due to the high costs associated
under evaluation technologies. We are merely opening a question
with the construction and operating of plants, regulations that
that “can a spread sheet of numerous quantified characteristics
enforce productivity evaluations, and environmental issues, effi-
help a decision-maker to choose a desired technology among from
ciency are key managerial concepts.
a set of competing technologies?”
To estimate the efficiency of companies or branches of a com-
There are many works that have applied DEA in its various
pany, various methods have been developed during the past two
versions for assessment of electricity generation/distribution sys-
decades. These methods are generally classified as parametric and
tems. Sarica K. Or, (2007) analyzed and compared the performance
non-parametric methods. In the parametric methods, the cost of
of electricity generation plants in Turkey, and they showed that
production function is estimated, whereas in the non-parametric
coal-fired plants have lower efficiency values than natural gas-fired
methods, it is not necessary to estimate the cost or production
ones [9]. Yadav et al. (2010) applied the DEA approach to evaluate
function. For example the latter is applied to evaluation of the
the relative performance of twenty nine electricity Distribution
renewable energy technologies [7].
Divisions of an Indian hilly state. Using input-oriented DEA they
DEA (data envelopment analysis) is the major non-parametric
indicated that numerous divisions had scope for improvement in
method among the others. Since the advent of DEA in 1978, there
overall efficiency [10]. Lins et al. (2011) used the DEA method to
has been an impressive growth both in theoretical developments
incorporate social, environmental and technological aspects and
and applications of the ideas to practical situations. The DEA ana-
variables in the Brazilian government’s energy policies to demon-
lyzes each DMU (decision-making unit) separately and identifies
strate the advantage of promoting residue based generation above
those that exhibit the best practice. The use of the DEA allows us
that from the other sources regarding sustainable development
not only to compare individual DMUs to the best-practicing DMU,
[11]. As the final example, we may refer to Fallahi et al. (2011) who
but also to identify the sources of inefficiency.
provided an empirical analysis of the determinants of energy effi-
The original model is a nonlinear programming problem to find
ciency in 32 Iranian electric power generation companies over the
optimal weightings attempting to maximize all DMU’s efficiencies:
period 2005e2009 by DEA. They estimated changes in the relative
Pp technical efficiency and productivity of these companies. They
uk ykj
Max ej ¼ Pkm¼ 1 (1) showed that the low increase of productivity is more related to low
i ¼ 1 vi xij efficiency rather than technology changes [12]. There are many
other good works in the field which are not referred to here, for the
Subject to : ej  1; cj (2) sake of brevity.

uk ; vi  ε; ci; k (3) 2.2. SAW (simple additive weighting)

which considers a set of n DMUs, with each DMUj, (j ¼ 1, ..., n) using SAW is one of the most popular and most widely used methods
m inputs xij (i ¼ 1, ..., m) and generating p outputs ykj (k ¼ 1, ..., p). because of its simplicity. It assumes additive aggregation of deci-
Parameters uk and vi are the weights given to each output and sion outcomes, which is controlled by weights expressing the

Please cite this article in press as: Shakouri G. H, et al., A quantitative discussion on the assessment of power supply technologies: DEA (data
envelopment analysis) and SAW (simple additive weighting) as complementary methods for the “Grammar”, Energy (2013), http://dx.doi.org/
10.1016/j.energy.2013.10.022
4 H. Shakouri G. et al. / Energy xxx (2013) 1e8

importance of criteria. SAW uses all c criteria values of an alterna- contains for decision making, i.e. how much informative it is. This
tive and uses the regular arithmetical operations of multiplications value is obtained simple as:
and summations. The values must be both numerical and
comparable. dj ¼ 1  Ej ; cj (12)
At the first stage, given aij’s; i ¼ 1, ., c(¼p þ m), j ¼ 1, ., n, as the
original attributes for the n alternatives (i.e. number of DMU’s), Then, the weight of each criterion is calculated by:
construct an n  c decision matrix that includes n alternatives and c
criteria and calculate the normalized decision matrix by using a
dj
wj ¼ Pn ; cj (13)
simple linear normalization method: i¼1 dj

aij Finally, the overall attribute, Aj, is calculated for each alternative as
zij ¼ ; ci; j (8)
Max aij follows:
X
To compare with the terminology applied for DEA, aij’s include Aj ¼ zij wj ; cj (14)
all the inputs and outputs. i
As the second step, Shannon’s entropy formula is applied to
obtain the weights of each criterion based on a rough estimate of where zij ’s are the complementary normalized attributes for the
probability distribution for the values of each criterion. The idea alternatives (DMU’s); i.e.:
behind this computation is that the more is dispersion of an index,
the more important than others it is. zij ¼ 1  zij ; ci; j (15)
The original Shannon’s entropy, comes with the concept of in-
formation theory, where it is used to measure uncertainty This is necessary when the attributes are applied with negative
expressed by distribution of pi, and find how much storage is meanings. After all, the solution to the decision-making problem is
needed to capture all uncertain information. However, Carla achieved by finding the DMU pertaining of the maximum overall
Balocco and Daniele Verdesca (2007) have applied this method for attribute:
energy technology assessment before the present study [13]. The  
calculations can be expressed in a series of formulas that are A* ¼ arg max Aj (16)
j
expressed below.
As mentioned above, to obtain the weightings, Shannon’s En-
Hence, the overall attributes are ranked and the winner alter-
tropy is calculated by:
native(s) is the one maximizing the overall attribute function,
X
n    which is found by multiplication of the complementary
Ej ¼ k Pij  ln Pij ; cj (9) normalized attributes’ matrix by the weightings vector, as
i¼1 defined by (14).

where, each Pij is calculated as an estimate of the probability dis-


3. Preparation of data
tribution for the attributes by:

aij For comparing four technologies introduced by Table 1, based on


Pij ¼ Pn ; ci; j (10) the physical characteristics introduced by Table 2, the above
i¼1 aij
mentioned methods are applied. The four cases are considered as
DMUs (decision making units), which should be ranked and
and k is a constant coefficient equal to:
selected as the best technologies for electric power generation. A
1 description for each DMU is also presented in Table 1.
k ¼ (11) The baseline cases and the findings considered here are the
lnðnÞ
same as the reference papers [4,5]. All the characteristics have
Uncertainty or degree of deviation of the data for index j, namely negative meanings, i.e. the fundamental criterion is that the less
dj, states how much information each of the criterion indexed by j quantity for each characteristic is a promotion to choose the

Table 1
DMUs (decision making units) and their descriptions.

DMUs (decision making units) Abbreviation Description of DMUs

1. Nuclear energy, light water LWR For the nuclear-based power-supply system, a typical LWR power plant is
reactor power plant assumed along with a one through nuclear fuel cycle. In OTC
(one-through cycle) scheme, UO2 assemblies are loaded in the thermal
spectrum light water cooled reactors, irradiated for a period of a few years,
discharged and left in “cooling storage” for a few years.
2. Nuclear energy, light water LWRP This power plant includes a reprocessing phase into the nuclear fuel cycle.
reactor power plant In the TTC (twice-through cycle), after a minimum cooling time, the fuel discharged
with reprocessing) from UO2 fueled LWRs is sent to reprocessing plants where both the uranium and plutonium are extracted.
3. Fossil fuel energy, IGCC For the fossil-based power-supply system, an IGCC (Integrated Gasification
IGCC power plant Combined Cycle) power plant using coal has been considered. The IGCC technology
consists in turning the coal to gas in order to remove impurities before it is combusted,
improving the overall efficiency of the power plant compared to conventional coal-fired power plants.
4. Fossil fuel, IGCC power IGCCS This power plant reduces the CO2 emissions of the power plant (DMU3) by 90% adding a
plant with CCS: 90% CCS (carbon capture and storage) technology that requires a certain amount of process
of CO2 capturing heat depending on the rate of CO2 being captured, mainly due to the gas-compression
needed before injecting the carbon into the ground.

Sources: [4,5,14].

Please cite this article in press as: Shakouri G. H, et al., A quantitative discussion on the assessment of power supply technologies: DEA (data
envelopment analysis) and SAW (simple additive weighting) as complementary methods for the “Grammar”, Energy (2013), http://dx.doi.org/
10.1016/j.energy.2013.10.022
H. Shakouri G. et al. / Energy xxx (2013) 1e8 5

Table 2
Description of characteristics used to assess cases.

Characteristics Abbreviation Unit Description I/O

1 Primary energy source PES kg kg of PES needed for starting the electricity generating Input
2 Input of energy carrier EC GJ Requirements of fossil-fuels for fund elements (making and maintenance of facilities) Input
3 Labor L h The labor needed for the control of flows, production and maintenance Input
of fund elements (including both direct and indirect labor)
4 Internal consumption IEC MWh Electricity consumed by power plant (internal consumption) Input
of electricity
5 Material (concrete, M (CR, ST, CP) kg Material requirements for the production and maintenance of fund elements Input
steel, copper)
6 Electric power capacity PC KW Electric power needed for internal usage, including both direct and indirect consumptions Input
7 Fossil fuel FF MJ Indirect usage of fossil fuels Input
8 Total internal energy TIE GJ Estimate of total primary (fossil fuel) energy consumption Iutput
consumption
9 Radioactive wastes RW (RHW, kg Amount of radioactive waste produced (including all high-level, low-level Output
RLW, RVW) and very-low-level wastes)
10 CO2 emissions CO2 kg Including both direct and indirect emissions Output
11 Net produced electricity GWh Net produced electricity Output

Source: [4,5].

Table 3
Values of physical characteristics for each DMU.

PES M La PCb FFc IEC RW CO2

U Coal CR ST CP FFD FFI RVLW RLW RHW

DMUs kg Tonne kg kg kg h KW GJ GJ MWh kg kg kg kg

LWR 21 0 13,000 490 35 640 2 250 110 33 16,000 340 34 24.6


LWRP 17 0 12,000 490 31 570 5.8 480 100 34 14,000 330 31 39.6
IGCC 0 350 1200 360 6 80 2.6 160 2.3 3.3 0 0 0 700,000
IGCCS 0 470 2200 590 12 115 2.8 210 4 120 0 0 0 110,000

Source: [4,5].
a
Indirect labor is assumed to have half influence as an input, so L ¼ L_direct þ L_indirect/2.
b
PC is calculated by averaging all values given by Refs. [4,5] for direct, indirect, minimum and maximum capacities needed.
c
FFD and FFI are the average of minimum and maximum amounts of internal direct/indirect fossil fuel energy usages.

corresponding DMU. This criterion is also used based on the 4. Results


reference paper [4].
The values for inputs and outputs are given by Table 3. For using With applying these two quantitative approaches, DEA, and
DEA approach, the characteristics are divided to two categories: SAW, we get interesting results discussed below.
inputs and outputs. Because of using a CCR input-oriented model,
the objective is to maximize output. Since the outputs are nega- 4.1. Assessment by DEA
tive, we consider the reversed of real data of waste radioactive and
CO2 emissions as output to convert these negative data to have Applying the original data in Table 3, DEA_CCS results indicate
positive meaning for maximizing the outputs. In addition, in the that all technologies are fully efficient. Obviously, this result
reference paper, the amount of CO2 for nuclear power plants and cannot convince any decision-maker. To obtain useful results, first
the amount of radioactive waste for fossil fuel power plants are we have merged the two kinds of PES into one input variable,
zero. Here, for doing the DEA, these amounts are considered small avoiding any zero in the inputs. Moreover, all materials are sum-
real numbers to avoid division by zero when reversing the outputs. med up leading to another merged input represented by M. The
In fact, all nuclear power plants have CO2 emissions due to con- third modification is to apply the total estimate internal energy
sumption of both fossil fuel and electricity. In this study this
amount is also estimated based on the same original data. To do so,
the average fossil fuel energy used by the system (including both Table 4
Inputs/outputs quantities processed (merged) for DEA/SAW calculations.
direct and indirect usages) is calculated first, and the internal
electricity consumption is converted to GJ assuming 45% of effi- DMUs PES M L PC TIEa RWb CO2 OEc
ciency for its generation to be added then. The outcome of these kg kg h KW GJ kg 1
kg 1
GWh
calculations is also given in the next tabulation, Table 4, repre-
LWR 21 13,525 640 5.9 624 1/2280 1/24.6 1
sented by TIE (total internal energy). LWRP 17 12,521 570 9.5 852 1/2040 1/39.6 1
To apply SAW, both data sets (the original: Table 3 and the IGCC 350K 1566 80 2.6 189 Mb 1/700K 1
merged data: Table 4) are used to achieve extra results enables IGCCS 470K 2802 115 2.9 1174 Mb 1/110K 1
determining the best technology. For SAW, there is no need to Source: Table 3.
reverse the data like DEA, because all of them are considered as a
Total internal energy usage is estimated by adding fossil fuel energy usage to
an attribute wanted to compare with each other. The only change fossil fuel needed to generate the electricity consumption by the plant, assuming
is converting zeros to a small number, say d, facilitating calcula- 45% efficiency for electricity generation system.
b
To sum up all kinds of RW pollutants, RVLW is divided by 10 and RHW is
tion of Shannon’s entropy. First, the decision matrix is normalized multiplied by 10. In addition, M is replaced for infinity, which represents a large
by (8) and then the weights of each criterion are calculated via positive real number, say 1010, to avoid not-a-number in the results.
c
(9)e(16). Electricity is the main useful output of all the power supply system.

Please cite this article in press as: Shakouri G. H, et al., A quantitative discussion on the assessment of power supply technologies: DEA (data
envelopment analysis) and SAW (simple additive weighting) as complementary methods for the “Grammar”, Energy (2013), http://dx.doi.org/
10.1016/j.energy.2013.10.022
6 H. Shakouri G. et al. / Energy xxx (2013) 1e8

Table 5 Table 6
DEA_CCS weightings and efficiencies. DEA results by solving (17) to (19).

DMUsa PES M L PC TIE RW CO2 Efficiency DMUs LWR LWRP IGCC IGCCS

kg kg h KW GJ kg1 kg1 % Efficiency [%] 24.98 16.14 100 60.94


Std 0.082 0.056 0.026 0.078
LWR 1.65E-9 2.40E-4 8.18E-3 1.49E-5 3.97E-4 6.38E-3 8.28E-4 100
LWRP 2.01E-8 2.96E-4 5.88E-2 1.00E-9 1.00E-9 1.00E-9 1.00E-9 74.8
IGCC 1.00E-9 1.39E-4 2.81E-6 5.95E-5 9.7E-4 9.00E-2 1.35E-3 100
IGCCS 1.01E-9 1.46E-4 4.16E-8 6.38E-6 1.43E-5 5.98E-1 5.05E-6 92.7
a
The value of ε in constrain (6) is set to 109. u3j ¼ 3600; v5j ¼ 1; cj (20)

where k ¼ 3 and i ¼ 5 indicate OE and TIE, respectively. Since the


use, TIE, instead of FF (fossil fuel) and IE (internal energy) sepa- feasible solution is sensitive to the value ofε, the changing results
rately. Finally, we have applied RW (radioactive wastes), which is obtained by its variation from 1E-9 to 1E-1 are illustrated by Fig. 2.
calculated in a way described below Table 4. This way, by data in The model is solved several times starting random numbers close to
Table 4 (except the last column), the efficiencies obtained for LWRP the lower bound. The mean values of efficiencies are given in
(light water reactor power plant with processing) and IGCCS (in- Table 6 along with their standard deviations. It is seen in the figure
tegrated gasification combined cycle with carbon capture and that weightings for PES and RW are kept plugged to the lower
storage) fall to 0.7481 and 0.9267, respectively. Corresponding bounds. Recalling that PES weight is an advantage for nuclear
weightings are given in Table 5. It is seen that DEA_CSS tries to plants, while radioactive waste is a disadvantage, it is deducible
maximize efficiencies of all DMUs by changing weightings due to that the model has tried to maximize all efficiencies as much as
each input/output. possible, using similar weightings.
As the next attempt to gain more logical results, we add the As a short preliminary conclusion, it seems that IGCC performs
main output of the power systems, which is 1 GWh of electricity, to better than the other technologies, and conversely, based on the
the set of outputs (the last column in Table 4). given characteristics, which all are physical, nuclear plants have
We have also run the original nonlinear model in (1) applying a lower performances.
minor modification to the objective:
0 1
4.2. Assessment by SAW
n B
Pp C
X
n X B k ¼ 1 uk ykj C
Max ej ¼ B Pm C (17) The obtained weights for the quantitative analysis based on
@ i ¼ 1 vi xij A
j¼1 j¼1 Shannon’s entropy are presented by Tables 7 and 8. As can be seen,
the weight of attributes considered on input category have bigger
amount than output category. Also, the weights for labor, power
Subject to : ej  1; cj (18)
capacity and material are bigger than others. These criteria are
important among others to be lowered on production process of
uk ; vi  ε; ci; k (19) each alternative.
The results of applying SAW method for data in Table 3 show
while forcing all weightings of each input/output to be the same for that the fossil fuel power plants (IGCC and IGCCS) totally have
all DMUs. Moreover, to give a common physical meaning to all the maximum overall attribute among others and perform better than
elements in both the numerators and denominators of the effi- nuclear power plants (LWR and LWRP). Based on the obtained re-
ciencies, i.e. uk ykj and vi xij, we assume that all weightings are such sults IGCC have maximum overall attribute, the next one is IGCCS
that the dimension of all these terms becomes “energy”. Therefore, and after these LWR and LWRP are placed; also, these calculations
none of TIE and OE needs unknown coefficient to be estimated, and are separately done for the merged data (Table 4) to obtain better
the corresponding coefficients are fixed: results. Similarly, the results of SAW for merged data show that
IGCC has maximum overall attribute and the other power plants
have the same sequence as former (SAW based on original data) at
the point of overall attribute. As can be seen both of them confirm
the better performance of fossil fuel power plants. The results of
this part are presented on Table 9.
All of this paper’s results obtained from applying DEA and SAW
method show that the fossil fuel power plants have better perfor-
mance and these confirm the results of reference paper: “When
considering internal constraints, nuclear energy requires about
twice as much as power capacity (5.9e9.5 KW/GWh vs. 2.6e
2.9 KW/GWh) and 5e8 times more labor (570e640 h/GWh vs. 80e
115 h/GWh). Things do not improve for nuclear energy when
looking at external constraints e e.g. the relative scarcity of PES.
The results of this study explain the difficulties faced by nuclear
energy to gain interest from investors” [4] (See also [14]).

5. Discussion

By this point, all the criteria are taken into consideration same as
the reference paper’s criteria. However, there are some discussions
Fig. 2. Illustration of weightings variation with increasing tolerance in DEA. on the selection of inputs and outputs the considering of which can

Please cite this article in press as: Shakouri G. H, et al., A quantitative discussion on the assessment of power supply technologies: DEA (data
envelopment analysis) and SAW (simple additive weighting) as complementary methods for the “Grammar”, Energy (2013), http://dx.doi.org/
10.1016/j.energy.2013.10.022
H. Shakouri G. et al. / Energy xxx (2013) 1e8 7

Table 7
The weight of each attribute calculated by (13) based on data in Table 3.

Chars PES M L PC FF IEC RW CO2

U Coal CR ST CP FFD FFI RVLW RLW RHW

kg Tonne kg kg kg h KW GJ GJ MWh kg kg kg Tonne

Weightsa 0.1074 0.1082 0.051 0.002 0.029 0.0452 0.0212 0.014 0.087 0.064 0.107 0.107 0.107 0.152
a
To calculate entropies by (9), Pij ¼ 0 is replaced by d ¼ 1010, wherever needed.

be useful for better comparison. These data can be considered in and 420, respectively. Then adding inverse of these figures to the
another way. This has been discussed in the following tips: SAW algorithm as new attributes to Table 4, the overall perfor-
mances will be: LWR ¼ 0.56, LWRP ¼ 0.58, IGCC ¼ 0.48, and
1.The unit of PES considered on the reference paper, for gener- IGCCS ¼ 0.62. Thus, the conclusion made by the reference paper
ating 1 GWh electricity is kg, but it is disputable if we compare is still hold admissible.
1 kg of uranium with 1 kg of coal. They say [4]: 4.The reference paper fairly states that it has not considered an
important characteristic, which is the cost of generating elec-
“In this way it becomes possible to individuate similarities and
tricity [4]:
differences in the process of production of electricity, and then
measure and compare “apples” with “apples” and “oranges” “. the comparison is based on a “steady-state” narrative and
with “oranges”.” therefore it does not provide information in relation to turnover
times. Indeed, information like the payback time e which is
extremely important for investors e would require expressing
We think this statement is in contrast to comparing 1 kg of
the characteristics of the power-supply systems over a larger
uranium with 1 kg of coal, for many reasons, some of which are
time scale (several decades) so as to capture their overall
mentioned in the reference paper: a) because they have different
behavior .”
heat values, b) the effort needed to extract each differs; c) their
costs are incomparable, and d) their EROEIs1 have huge differences.
Probably, considering each of these parameters for PES makes Economy of any technology is not negligible. In a static analysis, as
better comparison. in the existing case, two key economic characteristics play the main
However, it is clear that replacing the mass of PES by the energy role. The criteria are a) to pay less as annualized capital cost, and b) to
content for each of uranium and coal, the nuclear power plants will have less expense for fuel, operation, and maintenance annually.
have even lower efficiencies. For addressing this problem, the By consideration of CAPEX (Capital Expenditure), FOM (Fixed
dimensionless MCDM (Multi-Criteria Decision-Making) methods Operation & Maintenance), and VOM (Variable Operation and
(weighted product method, WPM) or preference ranking methods Maintenance) as given by DoE of US, along with electricity gener-
including AHP (Analytic Hierarchy Process), PROMETHEE (Prefer- ation capacities, regarded as outputs (as discussed by the previous
ence Ranking Organisation METHod for Enrichment of Evaluations) paragraph), assessment is repeated. The final results obtained by
are proposed as future works [15]. SAW are given in Table 10. It is seen that still IGCCS stands on the
top with a slightly higher performance compared to that of LWRP.
2.The reference paper considers labor as a negative factor. The On the other hand, the nonlinear DEA with common weights led to
amount of labor required for nuclear energy is 5e8 times more superiority of the advanced steam power plant without CCS, as it is
than the number needed for fossil fuel plants (between 570 and presented in the same table.
640 h per GWh vs. between 80 and 115 h per GWh) and finally
concludes that fossil fuel are better than nuclear energy because
6. Conclusion
it requires less amount of labor. However, from the view point of
“entrepreneurship”, higher labor need could be a positive
In this paper, a quantitative discussion on the “Grammar” pro-
characteristic for political economists, as job creation is one of
posed by Ref. [4] is performed. We tried to provide a transparent
their missions, particularly in developing countries with high
decision-making framework, applying two distinct assessment
population and low employment. Therefore, it can be put on the
methods in order to obtain a more systematic evaluation. As it is
outputs, or it can be replaced by its inverse. Doing so, SAW
well-known in the literature of decision-making problems, without
(based on data in Table 4) leads to: LWR ¼ 0.64, LWRP ¼ 0.64,
quantified decision (control) variables it is confusing to make qual-
IGCC ¼ 0.41, and IGCCS ¼ 0.60; with not significance changes in
ified decisions, which can satisfy the politicians and stakeholders. To
the weightings, meaning that it depends how we look at the
convince real decision-makers in the reality, it seems essential to
problem, and what is more important for the policy makers.
provide reasonable calculations to show performance of each DMU.
3.Another tip that can be considered here is the power gener-
To do so, we chose SAW, as one of the latest methods used by
ation capacity of each plant. In the reference paper, PC is
managers to evaluate their options, in parallel with DEA, which is
considered as an input requirement with negative effect, which
one of the earliest methods of assessment. Moreover, since the
makes sense. However, it can be considered as a positive crite-
linear versions of DEA may lead to useless results, a nonlinear DEA
rion and on output category because it in correlation with the
electricity generation capacity. The more is the installed ca-
pacity, the more it needs power for internal use of electricity. Table 8
Regarding the data given for cases 1 to 4 in Tables 1e4 in Refs. The weight of each attribute based on Table 4.
[4], the plant capacity of these four DMUs are 1300, 1300, 480 Characteristics PES M L PC TIE RW CO2

kg kg h KW GJ kg kg

1 Weights 0.2162 0.0886 0.0904 0.0423 0.0451 0.2136 0.3038


Energy returned over energy invested.

Please cite this article in press as: Shakouri G. H, et al., A quantitative discussion on the assessment of power supply technologies: DEA (data
envelopment analysis) and SAW (simple additive weighting) as complementary methods for the “Grammar”, Energy (2013), http://dx.doi.org/
10.1016/j.energy.2013.10.022
8 H. Shakouri G. et al. / Energy xxx (2013) 1e8

Table 9 Acknowledgment
Ranking DMUs based on attributes obtained applying SAW method by (8)e(16) to
data in Tables 3 and 4.
The authors would like to acknowledge the financial support of
DMUs LWR LWRP IGCC IGCCS University of Tehran for this research under grant number
A3 0.3212 0.3723 0.7369 0.7599 8109923/1/08. They have to appreciate the very valuable work
A4 0.5572 0.5713 0.4946 0.6436 published by François Diaz-Maurin and Mario Giampietro intro-
ducing a “Grammar” for assessment of power supply systems
which was the main motivation for this work. We are also grateful
is applied to confirm more logical comparison. In this model, equal
of valuable supports of IRES, the Institute for Resource, Environ-
weightings are assigned for each attribute of all technologies.
ment and Sustainability, UBC, for facilitating fulfillment of this
Meanwhile, weights for all the inputs and outputs in the form of
research in time.
energy (internal energy use including both electricity and fossil fuel
and the output GWh) are constantly equalized to enforce all attri-
butes have the same dimensions of energy. Results obtained by
using the original data almost admitted conclusion made by the References
reference paper. Moreover, the weightings approach to steady
[1] Pappas Ch, Karakosta Ch, Marinakis V, Psarras J. A comparison of electricity
values when the tolerance for the lower bounds tends to very small
production technologies in terms of sustainable development. Energy Convers
values near zero. Manag 2012;64:626e32 [last accessed 18.05.13], http://dx.doi.org/10.1016/j.
On the other hand, it is observed that the scores obtained for the enconman.2012.06.006.
competing technologies by SAW are very close together. Hence, [2] Sophister J. The economics of nuclear power. Student Econ Rev 2005;19:111e
23. Available from: www.tcd.ie/Economics/SER/sql/download.php?key¼10%fd.
although the ranks are similar to that of the reference paper, [3] Hidden cost of energy: unpriced consequences of energy production and use.
sensitivity to the attributes values is not so high. Washington, USA: Committee on health, environmental, and other external
Following up such quantitative methods, we need clarity of the costs and benefits of energy production and consumption; national research
council; 2010, ISBN 978-0-309-14640-1; 2010. Available from: The National
inputs (with negative meanings) and the outputs (with positive Academies Press at: www.nap.edu/catalog.php?record_id¼12794 [last
meanings) first. Thus, a useful outcome of our discussion is to accessed 18.05.13].
wisely select inputs and outputs among from all the characteristics [4] Diaz-Maurin F, Giampietro MA. Grammar for assessing the performance of
power-supply systems: comparing nuclear energy to fossil energy. Energy
of the technologies under evaluation. There are case-dependent 2013;49:162e77. http://dx.doi.org/10.1016/j.energy.2012.11.014 [last
choices. For instance, higher needs for “labor” may have positive accessed 18.05.13].
meaning when unemployment is a serious problem in a region. [5] Diaz-Maurin F. Biophysical requirements of power-supply systems: nuclear
energy and fossil energy. Spain: Institut de Ciencia I Technologia Ambientals,
Similarly, economic variables may have or may have not high Universitat Autonoma de Barcelona; 2012. p. 1e57. Available from: www.
importance for some decision-makers. recercat.net/handle/2072/179690 [last accessed 18.05.13].
However, by consideration of key economic elements, [6] Harris R. Introduction to decision making, virtual salt. www.virtualsalt.com/
crebook5.htm; 1998 [last accessed 18.05.13].
including capital expenditure and fixed and variable operational
[7] San Cristobal JR. A multi criteria data envelopment analysis model to evaluate
expenditure as important inputs, also adding the power plant the efficiency of the renewable energy technologies. Renew Energy 2011;36:
capacities SAW method confirmed that IGCCS is the best choice. 2742e6. http://dx.doi.org/10.1016/j.renene.2011.03.008 [last accessed
Other economic criteria, such as the value added, labor factor [17], 18.05.13].
[8] Cook WD, Seiford LM. Invited review data envelopment analysis (DEA) e
payback period, lifecycle costs, etc. can be also included in the thirty years on. Eur J Op Res 2009;192:1e17. http://dx.doi.org/10.1016/
assessment as well. It should be mentioned that we are j.ejor.2008.01.032 [last accessed 18.05.13].
comparing the contemporary commercial power supply tech- [9] Sarica K, Or I. Efficiency assessment of Turkish power plants using data
envelopment analysis. Energy 2007;32:1484e99. http://dx.doi.org/10.1016/
nologies, through the selected DMUs (LWR, LWRP, IGCC), with the j.energy.2006.10.016 [last accessed 18.05.13].
technology IGCCS which is still in a pilot testing phase and not [10] Yadav VK, Padhy NP, Gupta HO. A micro level study of an Indian electric utility
spread worldwide. for efficiency enhancement. Energy 2010;35(10):4053e63. http://dx.doi.org/
10.1016/j.energy.2010.06.011 [last accessed 18.05.13].
Although there are many other works done by DEA to evaluate [11] Lins ME, Oliveria LB, M.Silva AC, Rosa LP, Pereira Jr AO. Performance assess-
the energy technologies efficiency [18,17], the problem is still open ment of alternative energy resources in Brazilian sector using data envelop-
to be solved by other methods such as AHP, TOPSIS (Technique for ment analysis. Renew Sustain Energy Rev 2012;16:898e903. http://
dx.doi.org/10.1016/j.rser.2011.09.010 [last accessed 18.05.13].
Order Preference by Similarity), PCA (Principal Components Anal- [12] Fallahi A, Ebrahimi R, Ghaderi SF. Measuring efficiency and productivity
ysis), etc. to facilitate robust decision-making by comparison of the change in power electric generation management companies by using DEA: a
results obtained by these various methods. In addition, it would be case study. Energy 2011;36:6398e495. http://dx.doi.org/10.1016/j.en-
ergy.2011.09.034 [last accessed 18.05.13].
much more beneficial if other power generation technologies,
[13] Balocco C, Verdesca D. Shannon entropy for energy technologies ex-ante
particularly the renewable energy technologies, are entered evaluation. Int J Environ Technol Manag 2007;7(1/2):197e217. http://
together into a wider race field. dx.doi.org/10.1504/IJETM.2007.013245 [last accessed 18.05.13].
[14] Kazimi (Co-chair) MS, Moniz (Co-Chair) EJ, . Forsberg CW, Ansolbehere S,
Deutch J, Driscoll M, et al. The future of the nuclear fuel cycle, an interdisci-
Table 10 plinary MIT study. Cambridge: Mass: Report for Massachusetts Institute of
Performance (attributes) obtained by SAW, adding the output capacities and annual Technology (MIT); 2011, ISBN 978-0-9828008-4-3; 2011. Available from:
costs. web.mit.edu/mitei/research/studies/nuclear-fuel-cycle.shtml [last accessed
18.05.13].
DMUs CAPEXa FOMa VOMa OPCb Performances Efficiencies [15] Fülöp János. Introduction to decision making methods. BDEI-3 Workshop.
by SAW by NL-DEAc [%] Washington: Laboratory of research and Decision Systems. Computer
and Automation Institute, Hungarian Academy of Sciences; 2005. p. 1e15.
LWR 88,800 11,300 11,600 1300 0.5617 35.0 Available from:, citeseerx.ist.psu.edu/viewdoc/download?doi¼10.1.1.86.
LWRP 92,500 12,500 12,500 1300 0.5732 33.3 6292&rep¼rep1&type¼pdf [last accessed 18.05.13].
IGCC 75,200 6600 29,200 480 0.4718 96.6 [16] Annual energy outlook 2012. Released. Available from: www.eia.gov/
IGCCS 93,300 9300 36,800 420 0.5957 73.1 forecasts/aeo/electricity_generation.cfm; Jan. 2013 [last accessed 18.05.13].
a
[17] Wu F, Fan LW, Zhou P, Zhou DQ. Industrial energy efficiency with CO2
Source: EIA [17]. The values of economic characteristics are given in US$ per emissions in China: a nonparametric analysis. Energy Policy 2012;49:164e72.
GWh. http://dx.doi.org/10.1016/j.enpol.2012.05.035 [last accessed 18.05.13].
b
Output power capacities in GW. [18] Sözen A, Alp I, Kilinc C. Efficiency assessment of the hydro-power plants in
c
Nonlinear DEA with common weightings is done summing up the economic Turkey by using data envelopment analysis. Renew Energy 2012;46:192e202.
characteristics and setting ε to 1010. http://dx.doi.org/10.1016/j.renene.2012.03.021 [last accessed 18.05.13].

Please cite this article in press as: Shakouri G. H, et al., A quantitative discussion on the assessment of power supply technologies: DEA (data
envelopment analysis) and SAW (simple additive weighting) as complementary methods for the “Grammar”, Energy (2013), http://dx.doi.org/
10.1016/j.energy.2013.10.022

You might also like