pEA

ssi

mra

Un

Article history:Received 27 February 2012Received in revised form10 May 2012Accepted 13 May 2012

converted by corn plants into biomass [5]. One of the main factorsthat inuence the sustainable crop cultivation is the amount ofenergy input in any form to produce a unit of a desirable crop [6]. Infact, sufcient food production that can ensure food security highlyrelies on the existence of enough energy. Human beings rely on

renewable energies including solar, fossil, human and other sour-ces of energies is needed to a successful agricultural system [10].Energy use in agriculture can be divided into direct and indirect,renewable and non-renewable [11]. Direct energy refers to the anysources of energies, i.e. diesel fuel, gasoline, electricity, natural gas,animal and human power which use directly in the farm duringcrop production. Indirect energy is used off the farm inmanufacturing, packaging and transporting farm machines, fertil-izers, chemicals, irrigation systems, seed and so on [12]. Most of

* Corresponding author. Tel./fax: 98 728 331 3218.

Contents lists available at

Ener

els

Energy 44 (2012) 672e681E-mail address: houshyar.e@gmail.com (E. Houshyar).1. Introduction

1.1. Energy and sustainability

Agriculture could be meant as the conversion of human, fossiland solar energies into food and ber products [1,2]. The founda-tion of all agricultural production is the unique capability of plantsto convert solar energy into stored chemical energy [3]. However,Plants are not particularly efcient at collecting solar energy. Forinstance, 1 ha of green plants collect on average less than 0.1percent of the solar energy reaching them [4], or during the 120-day growth season, roughly only w 0.7% of the solar energy is

various sources of energy range from human, animal, wind, tidal,and water energy to wood, coal, gas, oil, solar, and nuclear sourcesof fuel and power [7]. Agricultural sector has become more energy-intensive in order to supply more food to increasing population [8].Currently, considerable fossil energy is used in agriculturalproductions. Breeding of higher-yielding plant varieties, increasesin the number of seeds planted per hectare, more intensive use offertilizers and pesticides, irrigation systems and farm machinesdepend on the use of signicant amounts of fossil energy [9].Agricultural productions can be enhanced by nding ways toaugment solar energy using human, animal, and fossil energypower [3]. Indeed, a combination of the both renewable and non-Available online 15 June 2012

Keywords:Fuzzy logicDEAEnergy inputCorn0360-5442/$ e see front matter 2012 Elsevier Ltd.doi:10.1016/j.energy.2012.05.025The goal of this study was to evaluate the sustainability and efciency of corn production with regard toenergy consumption in the Fars province, Southwest Iran. To reach the goal, fuzzy modeling and DataEnvelopment Analysis (DEA) were employed. The fuzzy model included three sub-models whichassessed the sustainability use of energy inputs, i.e. machinery, fuel, fertilizer, chemical, human power,seed, and electricity. Some DEA models like CCR (Charnes, Cooper and Rhodes), BCC (Banker, Charnes andCooper) and SBM (Slack Based Measure) were applied in assessment of efciency scores. Using non-hierarchical cluster analysis, 89 farmers were chosen in two clusters including 47 and 42 farmers asthe rst and second cluster, respectively. The farmers in cluster 2 used 44,238.235 MJ/ha energy inputwhich was 2500 MJ/ha more than the rst cluster. The fuzzy modeling revealed that the sustainabilityindices were higher in the rst cluster. However, the indices were in the ranges of very low to mediumfor all the farmers, representing that the current corn production system is not sustainable. Also, the DEAmodels revealed that the rst cluster also held more efcient farmers. Since DEA denes a reference setof efcient for inefcient farmers to follow, lead inefcient farmers to improve their sustainability up toefcient ones. Therefore, efcient and inefcient farmers must change the trends of energy utilization forapproaching even more sustainable production.

2012 Elsevier Ltd. All rights reserved.a r t i c l e i n f o a b s t r a c tSustainable and efcient energy consumIran: Combination of multi-fuzzy and D

Ehsan Houshyar a,*, Hossein Azadi b, Morteza AlmaFrank Witlox b

aDepartment of Farm Machinery and Mechanization, College of Agriculture, Shahid ChabDepartment of Geography, Ghent University, BelgiumcDepartment of Agricultural Mechanization, Science and Research Branch, Islamic Azad

journal homepage: www.All rights reserved.tion of corn production in Southwestmodeling

c, Mohammad Javad Sheikh Davoodi a,

n University, Ahvaz, Iran

iversity, Simon Bolivar Blvd, Ashra Esfehani Highway, Tehran, Iran

SciVerse ScienceDirect

gy

evier .com/locate/energy

ergystudies on energy consumption in agriculture consider seed,manure, animal and human power as renewable sources of ener-gies [13e17]. However, it seems that these sources of energies arenot totally renewable since they use both the sources of renewableand non-renewable energies. Demand for energy is not onlyincreasing in the agricultural sector, but in all other sectors thatinvolve human activities. Cropping pattern, farm activities and levelof technology are some of the factors that dene the rate of energyrequirements in agriculture. Over time, the risk of energy scarcity isgetting more severe. In such a situation, a sustainable agriculturalsystem can be dened as a system with the lower energy inputwhich is more sustainable from the environmental and theeconomic points of view.

In the coming decades the world faces the challenge to makea transition to sustainable energy use patterns in order to save fossilfuels for future generations and to reduce the negative impacts ofburning fossil fuels on the environment [18]. The scarcity of energyresources and the signicant dependency of agriculture on suchresources have led a lot of researchers to evaluate the energy use ofdifferent crops in different regions. Energy and environmentalsecurity are major problems facing our global economy [19]. Energybalance is essential in corn production since it is a good source offood and biofuel specially bioethanol. Accordingly, optimum andsustainable energy utilization is a challenging issue in corn culti-vation in a large number of regions worldwide. For instance, theenergy input for corn production in Canada [20], Wisconsin andGermany [21] is 1.54,1.43 and 2.16MJ/kg, respectively. It is higher insome countries like India and Thailand by 1.2e3 times [11,22].

Corn is a strategic crop in Iran as it is a main feeding source inthe poultry production and therefore a critical input in the poultryindustry. The total annual cultivated area is around 300,000 hawiththe average yield by 5500e7500 kg/ha. The total countrys need isestimated at four million tons per year of which 30e45% isimported [23]. Energy demand is increasing for most of cropsincluding corn as a summer crop in Iran, and this demand is notsustainable in relation to opinions of many experts. For instance,fertilizer demand increased by 30 times during 1960e2003 (from100 to 3000 thousand tons), and it will be reached to 6200 thou-sand tons by 2017e2018 [24]. Also, chemical use is 20e22 thousandtons per year which is 13 timesmore than the average consumptionin the world [25].

Accordingly, the main objective of this study is to determinewhether the amount of energy input for corn production issustainable and how the energy use and corn cultivation sustain-ability can be improved. To reach the objective, we will rst assessthe range of energy input in the region, then, different groups offarmers will be determined according to their energy use byapplication of non-hierarchical cluster analysis. Afterwards, cornfarming sustainability will be evaluated by using multi-fuzzymodeling. Finally, DEA will be used to highlight efcient and inef-cient farmers and improve corn farmers efciency.

1.2. Why using fuzzy modeling and DEA technique?

Fuzzy set theory was developed by Zade in 1975 [26], and hasbeen successfully applied in many elds of study afterwards[27e31]. The theorymay be regarded as an extension of classical settheory, which is based on bi-valued logic; i.e., in or out. A fuzzy set,on the contrary, is a multi-valued logic dened as membershipfunctions that represent the degree which the specied valuebelongs to the set [32]. Fuzzy logic is a very exible technique thatallows integrating different types of information (qualitative andquantitative) to formalize conclusions [33]. The use of fuzzy logic isbased on the idea that the line between sustainability and unsus-

E. Houshyar et al. / Entainability (or acceptance and rejection) of an agricultural practice2. Methodology

This study was carried out to evaluate energy consumption ofcorn production in the farming year 2010 in the central region ofthe Fars province, Southwest Iran. The region covers 40,000 ha andconsists of two towns (Marvdasht and Pasargad), and ve counties;namely Seyedan, Houmeh, Rahmat, Markazi and Ramjerd. Theprovince was chosen since it is the top corn producer in the countryand presents more than 30% of the total corn production [23]. Theprovince is locatedwithin 27 030 and 31 400 north latitude and 50

360 and 55 350 east longitude. The central region of the provincehas a moderate temperature in summer [57] ranging from 15 to37 C. Due to this suitable weather and the existence of themediumto high irrigation water and fertile soil, the region has the highestaverage yield in the country and produces 35e50% of total corn ofthe province. Therefore, sustainable and efcient corn productionin this region could have a signicant effect on the countrys cornproduction independency.

Eighty nine farmers were chosen through a simple randomsampling without replacement. The desired sample size wascalculated by Eq. (1) [58]:

n

t$sr$yN

21 1=N

t$sr$yN

2 (1)

where

- n is the required sample size;- N is the number of holdings in target population;is not clear but rather blurred or fuzzy [34]. Also, different types ofsubjects (positive or negative) can be combined by fuzzy modeling.Fuzzy models are designed for complicated and ill-dened prob-lems such as sustainability assessments. Since the term sustain-ability is a concept with some sort of vagueness and uncertainty inassessment, fuzzy modeling can be used to overcome the uncer-tainty [32]. Accordingly, fuzzymodeling as a powerful tool has beenused to evaluate different aspects of sustainable agriculture; i.e.farmmachinery and tillage systems [35,36], fertilizer and herbicideapplication [37], energy utilization and sustainability [38,39], andpostharvest technologies [40]. Nonetheless, a large number ofstudies on energy inputs for crop production have used a compu-tational framework without any modeling [11,22,41e47].

The fuzzy modeling applied in this study regards two mainaspects: engaging experts knowledge and experience, and deter-mining a reasonable index that displays the actual status of cornproduction sustainability from the energy input point of view.Furthermore, in order to evaluate the energy use status of cornproduction from an efciency point of view, Data EnvelopmentAnalysis (DEA) technique is used. The DEA technique is a non-parametric method, supplies wealth information in the form ofestimates of inefciencies in both inputs and outputs for everyDMU (Decision Making Unit; i.e. farmers in this study) [48]. Boththe DEA and fuzzy DEA have been used mainly in accounting,management and economic and fewer in other sciences like agri-culture. Most researches in agriculture only evaluated technical andsome other types of efciencies [49e56]. As far as we know, thisstudy is the rst attempt to combine fuzzy and DEA modeling insuch a way that uses outputs from fuzzy models as the main inputsin DEA models. Consequently, approaching both sustainability andefciency is expected.

44 (2012) 672e681 673- S is the standard deviation of the sample;

- yN is the mean of the sample;- t is the reliability coefcient (2.576 which represents the 99%reliability);

- r is the permissible difference between actual and calculatedmean (0.03 in this study).

S and yN were calculated by a primary sample consists of 50farmers as 43,441.57 and 4980.43 MJ/ha, respectively. Finally, usingEq. (1), 88.61 (rounded to 89) farmers were chosen from the total1035 (N) corn farmers.

Two types of questionnaires were designed, one to collect theinformation regarding to energy input in the farms, and the other to

E. Houshyar et al. / Energy674set-up the fuzzy models as discussed in the following sections.

2.1. Energy use analysis

A six-page questionnaire was designed to gather data on theenergy use by corn farmers. Since most of the farmers were illit-erate or had only elementary literacy, ve agricultural engineerswere asked to collect the information via face-to-face interviewswith the farmers. The questionnaire consisted of the questionswhich were asking about seed, fertilizer and chemical consump-tions, soil type, the number of applied farm operations and types offarm machines and equipments, etc.

The amounts of the applied inputs, i.e. fertilizers, chemicals,human power and seed were calculated per hectare, and multipliedby their energy equivalents (Table 1) to convert them to energy unitMJ/ha. Fuel as a direct and machinery as an indirect source ofenergy are twomain energy consumers in farm operations. In orderto measure the energy input in the form of fuel and machinery, thefollowing factors were considered: machine type, speed ofoperation, farm slope and soil texture. Fuel consumption byfarm machines as a direct source of energy was calculated byEq. (2) [59]:

Qavg 0:223 Ppto (2)where

Qavg average diesel consumption; gal/h as multiplied by 3.78to convert it to L/h.Ppto maximum PTO power (hp)

The amount of energy used by machinery production as anindirect energy input was estimated by Eq. (3) [60]:

EID TW CED h RU=UL (3)where

EID indirect energy used for machinery production, MJ/ha;TW total weight of the specic machine, kg;

Table 1Energy equivalents of inputs and outputs in agricultural production.

Reference Energy equivalent (MJ/unit) Input (unit)

[61] 102 Liquid chemical (L)[61] 120 Granular chemical (kg)[6] 1.96 Human power (h)[13] 62.7 Machinery (kg)[62] 66.14 Nitrogen (kg)[62] 12.44 Phosphorus (kg)[62] 11.15 Potassium (kg)[13] 0.3 Manure (kg)[13] 20.9 Zinc sulphate (kg)[13] 56.3 Diesel (L)

[6] 14.7 Corn seed (kg)CED cumulative energy demand for machinery, MJ/kg;UL wear-out life of machinery, h;h specic working hours per run, h/ha;RU runs, number of applications in the considered eldoperation.

Another two essential energy consumers are electricity asa direct and irrigation as an indirect energy inputs to deliverwater to the farms. Electricity shows the amount of energy inputfor pumping water as a source of direct energy. Irrigation pres-ents indirect energy which is used in manufacturing of waterpumps, barriers and water canals. Since the irrigation energywas directly calculated as 20% of electricity, it was not consid-ered in the fuzzy and DEA models. The energy outputeinputratio, energy productivity and specic energy are the mainindices to show energy efciency. The indices are formulated asbelow [11]:

Energy ratio : energy output=energy input (4)

Specificenergy : energy input=grainyieldoutputMJ=kg (5)

Energy productivity : grain yield output=energy inputkg=MJ(6)

Net energy gain : energy output energy inputMJ=ha (7)These indices show the status of a system according to two

aspects: the total energy input and the yield (or energy output).However, it cannot be found out whether the system is sustainableor how it would be improved. For instance, a system may bereached to a better energy ratio by 2 times more yield productionand 1.5 times more energy input. In spite of more energy ratio,more energy input may be resulted in lower sustainability. There-fore, the researchers needed to use other scientic methods likefuzzy and DEA to evaluate the sustainability and efciency of thesystems, more carefully. Additionally, the types of energies, i.e.direct, indirect, renewable and non-renewable are not embedded inthese indices. A useful index would be dened as the ratio ofrenewable to non-renewable energy inputs. This index shows thatto what extent a system rely on renewable or non-renewablesources of energies. Although the category of the direct and indi-rect energy shows the source of energy from these two aspects,seemingly there is no signicant relationship between these twocategories and sustainability.

2.2. Non-hierarchical cluster analysis

Cluster analysis is a multivariate method which aims to classifya sample of subjects (or objects) on the basis of a set of measuredvariables into a number of different groups such that similarsubjects are placed in the same group [63]. Accordingly, in order toassess different attitudes in energy utilization, non-hierarchicalcluster analysis (often known as k-means clustering method) wasused. As the name implies, this method exhibits a group of farmerswith similar energy applications. For instance, a group of farmerswith low energy inputs in the form of fertilizer and chemical andhigh in the form of fuel. Non-hierarchical cluster analysis tends tobe used when large data sets are involved [64].

Since determining the number of appropriate clusters (K) isa challenging issue in this method, a combination of K-means andMANOVA was applied to distinct proper K. By MANOVA, all energyparameters were considered as variables and different Ks (2, 3, etc.)

44 (2012) 672e681as treatments. The K with minimum Wilks coefcient and

maximum F was chosen as the best solution, since it representedmaximum difference between clusters and minimum differencewithin a cluster.

2.3. Multi-fuzzy modeling

The basics of fuzzy sets has already discussed in many books andpapers [26,65e68]. Since there were numerous energy parametersthat should be taken into account, we developed three multi-fuzzymodels as primary models. With such an approach, dening fuzzyrules would be much easier and therefore the outputs of modelswould be improved. As shown in Fig.1, the three primarymodels are:

1. Model F (Fertilizer) with three input variables: all types offertilizers and yield are considered in this model. Fertilizerswere divided to two types and dened as two input variables;namely NPPM (nitrogen, phosphorous, potassium and micro-nutrients) and Ma (manure).

2. Model M (Machinery) with four input variables: machinery,fuel, seed and yield are applied in this model.

3. Model C (Chemical) with four input variables: chemical,electricity, human power and yield are the inputs of this model.

One sustainability index came from each model. These three

As shown in Fig. 2, the experts have dened a triangular andshoulder shape for the fuzzy sets. Almost all the experts (97%)preferred to set-up four linguistic values for input linguisticvariables; namely low, medium, high, and very high. Thevalues for output variables (sustainability indices) and the FMCmodel were determined in ve levels according to InternationalUnion for Conservation of Nature and Natural Resources (IUCN)[69] as very low, low, medium, high and very high. Each expertdetermined the ranges of linguistic values based upon his/herspecialty and background. The mean of all the ranges introducedby the experts was calculated to construct the models usingMatlab Fuzzy Toolbox (version 7.6). Fig. 2, also shows thestructure of (for instance) Model C which includes electricity,chemical, human power and yield as the inputs, and thesustainability index C as its output. Also, the gure displays theranges of functions for chemical energy that dened by theexperts as [0 0 210 310], [190 410 710], [462 757 967] and [810910 1300 1300] for the levels low, medium, high and very high,respectively.

Ifethen rules in a fuzzy model are dened based on inputvariables and their linguistic values. For instance, all possibilitiesfor Model F is 64 since it has three input variables as NPPM, Maand yield that each variable has four linguistic values as low,medium, high and very high. Holding the same procedure, the

E. Houshyar et al. / Energy 44 (2012) 672e681 675primary sustainability indices were combined into a secondarymodel FMC (combination of models F, M and C) as the main inputvariables to reach a total sustainability index as the main output. Inorder to dene the types and the ranges of membership functionsbased on the experts knowledge, we chose almost all of availableagricultural experts in the region who were familiar with thefarming condition in the region from both technical and culturalpoints of view. Accordingly, 26 experts were chosen. The expertsheld bachelor, master of science or doctoral in agriculture, andwerefamiliar with the energy and sustainability issues and some withthe fuzzy modeling. The experts were asked to:

1 dene the labels (linguistic values)2 determine the ranges of each value labels3 express fuzzy ifethen rulesFig. 1. Schematic diagram of primtotal number of all possible ifethen rules for the models areestimated as follows:

Model F: 4 4 4 64Model M: 4 4 4 4 256Model C: 4 4 4 4 256Model FMC: 5 5 5 125

All the above possible combinations of the if-then rules wereprepared by means of a simple programming in Matlab with anempty column as output. To nalize the denition of if-then rules,all experts were asked to ll out the output column. After gatheringthe data from the experts, the designed models were presented tothe experts for their nal revision. At last, all the experts conrmedthe models set-ups.ary and secondary models.

re o

E. Houshyar et al. / Energy6762.4. DEA technique

For assessing energy consumption efciency of each farmer(DMUs),1 DEA2 technique was used. CCR, BCC and SBM3 modelswere analyzed using DEA Excel Solver. DEA was used by sevenenergy inputs, i.e. machinery, fuel, fertilizer, human power, seed,chemicals and electricity. DEA was employed three times withdifferent outputs. First, yield as the output, then sustainabilityindices from the fuzzymodels F, M, and C as the outputs, and at last,both the yield and sustainability indices as the outputs. The bestresults came from DEA when yield and sustainability indices werechosen as the outputs as discussed in Section 3.2.

2.4.1. Technical efciencyTechnical efciency is the efciency in converting inputs to

outputs. It exists when it is possible to produce more outputs withthe inputs used or to produce the present level of outputs withfewer inputs. According to Cooper et al. [70], it can be stated as theratio of the sum of weighted outputs to the sum of weighted inputsas described in Formula (8):

TEj u1y1j u2y2j / unynjv1x1j v2x2j / vmxmj

Pn

r1 uryrjPms1 vsxsj

(8)

where x and y are input and output, and v and u are input andoutput weights, respectively, s is number of inputs (s 1,2,.,m), ris number of outputs (r 1,2,.,n) and j represents jth DMUs(j 1,2,.,k). For solving Eq. (8), the following linear program (LP)was developed by Charnes et al. [71], which called CCR model(Formulas (9)e(13)):

Fig. 2. The structuMax : q u1y1i u2y2i / uryri (9)

Subject to : v1x1i v2x2i / vsxsi 1 (10)

u1y1j u2y2j / uryrj v1x1j v2x2j / vsxsj (11)

u1;u2;.;ur 0 (12)

v1; v2;.; vs 0; and i and j 1;2;.; k (13)where q is the technical efciency and i represents ith DMU.

1 DMU: Decision Making Unit.2 DEA: Data Envelopment Analysis.3 CCR: Charnes, Cooper and Rhodes; BCC: Banker, Charnes and Cooper; SBM:

Slack Based Measure.2.4.2. Pure technical efciencyIn 1984, Banker, Charnes and Cooper introduced amodel in DEA,

which was called BCC model to draw out the technical efciency ofDMUs [72]. The calculation of efciency in BCC model is called puretechnical efciency and can be expressed by Dual Linear Program(DLP) as:

Max : Z uyi ui (14)

Subject to : vxi 1 (15)

vX uY u0e 0 (16)

v 0; u 0 (17)Pure technical efciency shows the DMU management [51].

Generally, the CCR-efciency does not exceed BCC-efciency [48].The result of the BCC model shows that how many percent ofenergy used have contributed to the output. Since we tried todecrease energy input, input-oriented was used for the CCR andBCC models in this study. In such a way, the models reduce theinputs as much as possible and conserve the level of outputs.

2.4.3. Scale efciencyTechnical inefciency is caused by the operation of the DMU

itself (pure technical efciency) or by the disadvantageous condi-tions under which the DMU is operating (scale efciency).According to Cooper et al. [70], scale efciency shows the effect ofconditions on the DMU inefciency and is dened as:

f fuzzy model C.

44 (2012) 672e681Scale efficiency Technical efficiency=Pure technical efficiency(18)

The conditions for a rm are considered as the rm size, thenumber of clerks and so on [73]. In agricultural systems, theconditions can be regarded as the weather, the compatibility offarm machines and equipments with the farm sizes and soilconditions, appropriate irrigation systems and wateravailability [51].

2.4.4. SBMAs the CCR and BCC models calculate efciency scores consid-

ering inputs or outputs separately, the SBM model has beendeveloped to consider both inputs and outputs simultaneously [74].Therefore, in our study, this model tries to determine efcientfarmers with regard to minimum inputs (energy parameters) andmaximum outputs (yield and sustainability indices). The SBMmodel is formulated as below [70]:

MAX :Xmi1

Si Xsr1

Sr (19)

Subject to :Xnj1

ljxij Si xio i 1;2;.;m; (20)

Xnj1

ljyrj Sr yro r 1;2;.; s; (21)

lj ; Si ; ; S

r 0 (22)

where Si and Sr are excess and shortfall in inputs and outputs,

respectively.

3. Results and discussion

3.1. Energy use analysis

The results of the energy use analysis for the corn production aregiven in Table 2. Column Total in this table shows the result of allfarmers (89 farmers) before clustering. The result of non-hierarchical clustering by combination of K-means and MANOVAclears that the best number of clusters (k) is two; i.e. 47 and 42farmers in the rst and the second cluster, respectively.

The total input and output energy for the corn production are42,953.27 and 108,150.54 MJ/ha, respectively. More than 85% ofenergy input is used by the three following main energy

Italy our farmers use themost energy to produce corn (Table 3). Theremarkable point is that the studied farmers consume energy 3e4times more than Thailand and India. Although the average yield ismore than these countries, all the energy indices (i.e. energyproductivity, specic energy, net energy gain and outputeinputenergy ratio) are lower as a consequence of high energy input.Our farmers use much more energy for fertilizers, diesel fuel andelectricity by 3.5e13 times in compare to these two countries. Thefarmers energy use of diesel fuel and fertilizers are almost similarto U.S. and Italy. In spite of these similarities, the mean yield inIrans farms is lower than U.S. and Italy by 1642.82 and 4142.82 kg/ha, respectively. However, the farmers use much more electricitythat cause to difference in energy input of these three countries.

Clustering displays different attitudes on energy consumptionamongst the farmers who, in cluster 2, use around 2500 MJ/hamore energy than those in cluster 1. As discussed most studies onenergy consumption in agriculture consider seed, manure, animaland human power as renewable sources of energies. However,these sources of energies are not totally renewable. With thisregard, our farmers use no renewable energies in corn productionwhich would result in low sustainability of agricultural system [76]in the region. Statistical analysis reveals that there are signicantdifferences (P < 0.05 or P < 0.01) between the two clusters on theamount of energy in some inputs such as seedbed preparation andplanting (machinery and fuel), human power, seed and chemicals.However, there are great differences (by 150e1600 MJ/ha) in threeinputs; i.e. seedbed preparation (machinery and fuel), fertilizer(nitrogen) and chemicals. The typical farm operations in cluster 2are: plowingwithmoldboard plow, three times disking, three timesland leveling and nally planting. In addition to high number of

2-4-Harvesting 1284.687[16.66%] n.s*

E. Houshyar et al. / Energy 44 (2012) 672e681 6773-Fertilizers 16,736.110(38.96%)3-1-Nitrogen 15,498.317[92.6%] *

3-2-Phosphorous 484.854[2.90%] n.s

3-3-Potassium 191.011[1.14%] n.s

3-4-Manure 537.079[3.21%] n.s

3-5-Other (zinc sulphate, iron, etc.) 24.849[0.15%] n.s

4-Human Power 413.915(0.96%) **

5-Seed 354.121(0.82%) **

6-Chemicals 599.377(1.4%) **

7-Electricity 13,113.170(30.53%) n.s

8-Irrigation 2622.634(6.11%) n.s

-Total input energy 42,953.27(100%) *

-Direct energy 21,237.794(49.44%)-Indirect energy 21,715.479(50.56%)-Total output energy 108,150.54-Output-input energy ratio 2.537-Energy productivity (kg/MJ) 0.173-Specic energy (MJ/kg) 5.860-Net energy gain(MJ/ha) 65,197.270-Yield(kg/ha) 7357.18n.sconsumers: fertilizers, electricity and diesel fuel. Specic energyand outputeinput energy ratio are 5.860 MJ/kg and 2.537, respec-tively. In compared to four countries, i.e. India, Thailand, U.S. and

Table 2Energy used status for corn production.

Average (MJ/ha)

Total

1-Machinery 1403.236(3.27%)*1-1-Seedbed preparation 465.629[33.18%] **1-2-Planting 275.910[19.66%] **1-3-Fertilizer application spraying 149.517[10.66%] n.s1-4-Harvesting 512.180[36.50%] *2-Diesel fuel 7710.709(17.95%) n.s

2-1-Seedbed preparation 3602.944[46.73%] **2-2-Planting 2561.784[33.22%] **2-3-Fertilizer application spraying 261.294[3.39%] n.s*, **: Show signicant difference between clusters 1 and 2 in 0.05 and 0.01 probability lfarm operations, most of farmers in this cluster use obsolete heavyequipments with low or high power tractors that result in notice-able difference in fuel energy use for seedbed preparation by

Cluster 1 Cluster 2

1358.745(3.25%) 1453.024(3.28%)389.255[28.65%] 551.095[37.93%]310.660[22.86%] 237.024[16.31%]163.064[12.00%] 134.357[9.25%]495.766[36.49%] 530.548[36.51%]7599.455(18.18%) 7835.207(17.71%)3137.468[41.29%] 4123.833[52.63%]2921.898[38.45%] 2158.800[27.55%]271.477[3.57%] 249.900[3.19%]1268.613[16.69%] 1302.674[16.63%]15,974.031(38.21%) 17,588.914(39.76%)14,731.003[92.22%] 16,356.978[93.00%]519.217[3.25%] 446.400[2.54%]160.106[1.00%] 225.595[1.28%]538.298[3.37%] 535.714[3.05%]25.406[0.16%] 24.226[0.14%]356.596(0.85%) 478.059(1.08%)364.685(0.87%) 342.300(0.77%)711.595(1.70%) 481.689(1.09%)12,866.585(30.78%) 13,389.110(30.27%)2573.317(6.16%) 2677.822(6.05%)41,805.009(100%) 44,238.235(100%)20,822.636(49.81%) 21,702.376(49.06%)20,982.373(50.19%) 22,535.859(50.94%)10,8373.091 107,901.5002.592 2.4390.176 0.1665.682 6.02466,568.083 63,663.2657372.319 7340.238evel, respectively.

1]

e8950e4

9e3

2e2e15

1e1e721e3.6e72.74.25.93.99e1513

ergyaround 1000 MJ/ha between the two clusters. Nonetheless, thefarmers who are in cluster 1, manage their farm time schedules andreduce the number of disking and land leveling by in time farmoperations that result in less energy use for two inputs, i.e.machinery and fuel. Moreover, they use more matched equipmentswith their tractors that result in higher scale efciency scores forthis cluster as discussed in the next section.

Cluster 1, compared to cluster 2, consumes much less fertilizersandmuchmore chemicals. However, around 40% of energy input in

Table 3Energy used status for corn production in some other countries.

Average (MJ/ha)

Iran India [1

1-Machinery 1403.236 114.642-Diesel fuel 7710.709 1033.53-Fertilizers 16,736.110 1734.13-1-Nitrogen 15,498.317 e3-2-Phosphorous 484.854 e3-3-Potassium 191.011 e3-4-Manure 537.079 1141.03-5-Other (zinc sulphate, iron, etc.) 24.849 e4-Human Power 413.915 1592.45-Seed 354.121 569.176-Chemicals 599.377 e7-Electricity 13,113.170 e8-Irrigation 2622.634 51.74-Total input energy 42,953.27 4412.6-Direct energy 21,237.794 2785.4-Indirect energy 21,715.479 1627.2-Total output energy 108,150.54 56,247-Output-input energy ratio 2.537 7.41e1-Energy Productivity (kg/MJ) 0.173 0.14e0-Specic energy(MJ/kg) 5.860 3.98e6-Net energy gain(MJ/ha) 65,197.270 51,834-Yield(kg/ha) 7357.18 1108e

E. Houshyar et al. / En678both the clusters belongs to fertilizers. Although nitrogen is moredangerous for the environment due to its mobility compare to otherfertilizers, it accounts for 93% of the total fertilizers. In spite of highincrease in the prices of energy since 2007 in Iran, as there is nosuitable law to restrict the use of agricultural inputs, the farmerstend to consume energy more and more, especially in the forms offertilizers and chemicals. Banaeian and Zangenehs study approvedthat the average of total energy input of corn production in Iranincreased by 22.66 GJ/ha during 2001e2007 [14]. Around 1e3% offarmers use farm fertilizer and chemical applicators. Traditionalapplication of these inputs by labors on the one hand results inhigher consumptions and wastes, and lead to lower efciency andsustainability on the other hand. The share of electricity is one thirdof the total energy input since 90e99% of farmers in both theclusters use furrow irrigation systems in addition to improper farmslope and soil smoothness. However, the soil condition is better inthe farms of cluster 1 that results in lower electricity energy input.As Iran has recently encountered chronic droughts, farmersnancial problems are the main inhibitor toward using modernirrigation systems.

Correlation analysis shows that linear relationship between thetotal fertilizers, nitrogen, chemicals, total energy input and yield are0.765, 0.83, 0.74 and 0.74 in cluster 1, respectively. Furthermore,there are weaker correlation coefcients between the total fertil-izers, nitrogen, total energy input and yield in cluster 2 (0.499,0.547 and 0.33, respectively). The weaker coefcients in cluster 2demonstrate that those farmers who are in this group do notconduct soil sampling to estimate the correct amounts of fertilizersor other inputs. As a consequence, they use and waste the inputscarelessly that results in their low sustainability of corn production.Also, there is a great gap (by 5000e10,000 MJ/ha) on energy inputamong the farmers with the same farming circumstances, i.e. thesame region, the weather, the water and the soil type. Since 50% ofthis energy difference belongs to nitrogen, shows that there isa chaos on using this input. Also, in spite of similar farming situa-tions, this gap is as a consequence of using unmatched farmequipments and tractors, improper irrigation systems and wastingchemicals and specially fertilizers.

Thailand[22]

Italy[47]

U.S [75].

3.95 e e 1394.210,050.4 6600 5895.01

019.64 [5713.4] 17,700 [12,903.70]4730 14,600 10,383.26854.2 1000 1373.27129.2 2200 1147.18

633.61 e e ee e e

572.99 46.9 e 1934.364.46 290.3 e 2177.14

667.6 700 3768.12e e 142.35e 4100 1339.78

0486.51 12,638.9 29,700 31,275.459.5 e e e227.01 e e e7663.10 66,610.6 218,500 132,353.12

5.3 7.36 4.23e e e

e [2.5] [3.48]67176.59 e e e

4531.33 11,500 9000

44 (2012) 672e6813.2. Sustainability and efciency analysis using multi-fuzzy modelsand DEA

Sustainability analysis by application of multi-fuzzy modelsreveals that the farmers use energy in the low to medium levels ofsustainability (Table 4). The sub-models F and C show that thefarmers act more unsustainably in utilization of fertilizers andchemicals. Model M displays that the status of using energy formachinery, diesel fuel and seed are better than the other terms ofenergy inputs. However, the maximum indices come from themainFMC model and the sub-model C by 0.580 and 0.521, respectivelythat belong to some farmers in cluster 1. The FMC model indicatesthat the farmers in cluster 1 produce corn in a more sustainableenergy manner holding the average sustainability index equal to0.256. Nevertheless, all the outputs means from FMC are in theranges of very low to medium which should be considered asalarming ranges for corn producers, agricultural experts andgovernmental decision makers if sustainability is a goal.

As discussed earlier, DEA is applied in three ways with thefollowing outputs: rst, with yield, second, with sustainabilityindices, and nally with both the yield and sustainability indices.The results of DEA reveal that with the yield or sustainabilityindices as outputs, those with high yield and low sustainabilityindices might be chosen as efcient farmers and vice versa. But byusing the yield and sustainability indices as outputs, those withmedium or high yields and sustainability indices could be chosen asefcient and frontiers farmers. Therefore, both of these outputs areentered to the DEA models. The results of technical, pure technical,

scale and slack based measure efciencies show that around50e60% and 40e50% of the efcient farmers are in clusters 1 and 2,respectively (Table 5). The yield and sustainability indices of thisgroup are 6700e7500 kg/ha and 0.35e0.58, respectively. Further-more, the results also show that around 15% of CCR-inefcientfarmers can improve and move toward the BCC-efcient frontiers.

The percent of the high scored technically efcient farmers, by90% or more, with a great difference are 72.34% and 42.85% inclusters 1 and 2, respectively. Since the technical inefciency wouldbe caused by the management (pure technical efciency) or theconditions underwhich the DMU is operating (scale efciency), thisdifference can be analyzed in more details. As described earlier, theconditions in farming systems can be observed as weather, farmmachinery, soil and water etc. Because the study site covers thesame weather, and almost the same soil and water condition, wefocus on the availability of appropriate farm machines and irriga-tion systems to nd the reasons of scale inefciencies. Pure tech-nical and scale efciencies display that this difference is due to thelower scale efciency in cluster 2. Near to 70e80% of the farmers inboth the clusters are purely technical efcient, reveals that the

Table 4Sustainability indices from multi-fuzzy models for corn production.

Farmers Fuzzy models

Model F Model M

Mean Min Max Mean Min Max

Total 0.25 0.08 0.45 0.316 0.11 0. 35Cluster 1 0.25 0.12 0.45 0.338 0.11 0. 35Cluster 2 0.215 0.08 0.38 0.289 0.11 0.26

E. Houshyar et al. / Energyfarmers farm management to simply prepare inputs and so on isnot bad. Approximately, all the experts conrm the result of the

Table 5Frequency distribution of corn growing farmers according to DEA models.

CCR model(TE) Total Cluster 1 Cluster 2

Efcient 37(41.57%) 25(53.19%) 16(38.09%)Inefcient > 90% 17(19.1%) 9(19.15%) 2(4.76%)

81e90% 9(10.11%) 7(14.89%) e71e80% 2(2.25%) e 4(9.52%)61e70% 13(14.61%) e 7(16.67%)< 60% 11(12.36%) 6(12.76%) 13(30.95%)

No. of farmers 89(100%) 47(100%) 42(100%)

BCC model (PTE)Efcient 49(55.1%) 29(61.7%) 22(52.38%)Inefcient > 90% 24(26.97%) 11(23.4%) 8(19.05%)

81e90% e e 7(16.67%)71e80% 9(10.11%) 7(14.89%) e61e70% e e 5(11.90%)< 60% 7(7.86%) e e

No. of farmers 89(100%) 47(100%) 42(100%)

SEEfcient 37(41.57%) 25(53.19%) 16(38.09%)Inefcient > 90% 15(16.85%) 10(21.28%) 4(9.52%)81e90% 12(13.48%) 1(2.13%) 7(16.67%)71e80% 6(6.74%) 2(4.26%) 5(11.90%)61e70% 11(12.36%) 4(8.51%) 2(4.76%)< 60% 8(8.99%) 5(10.64%) 8(19.05%)

No. of farmers 89(100%) 47(100%) 42(100%)

SBMEfcient 46(51.68%) 28(59.57%) 23(54.76%)Inefcient > 90% 1(1.12%) 1(2.13%) e

81e90% 3(3.37%) e 2(4.76%)71e80% 3(3.37%) 6(12.76%) e61e70% 2(2.25%) 1(2.13%) 5(11.90%)< 60% 34(38.2%) 11(23.40%) 12(28.57%)

No. of farmers 89(41.57%) 47(100%) 42(100%)BCC model since they believe that the corn farmers have enoughexperience from the management point of view. Nonetheless, theydescribe that the farmers need to promote their agricultural basicknowledge especially in the application of inputs at the right time,quantity and quality. Although the average farms size is small thatresult in low scale efciency scores, most experts believe that theexistence of the old and obsolete farm machines are the mainfactors of this problem. Also, they emphasize that the more scaleefcient farmers in cluster 1 is due to availability of new and suit-able farm machines and especially skilled farm machinery opera-tors. To conrm this idea, the distribution of the farmers who are incluster 1 is inspected. Since the most scale efcient farmers havea good access to new and suitable farm machines and essentiallyskilled operators, it seems that the role of farm machines is morethan the irrigation systems to promote the scale efciency of thecorn production in this region. However, the suitable irrigationsystem is another important factor to improve the scale efciency.Therefore, the purely technical inefcient farmers should promotetheir agricultural knowledge and experience and apply a bettermanagement. Also, scale inefcient farmers should use irrigationsystems and proper farm machines which are more matched withtheir farm size that result in the energy input reduction by5000e10,000 MJ/ha as discussed in Section 3.1.

SBM is more practical in this study since it condenses inputs andincreases outputs to determine the efcient farmers. SBM alsoreveals that most of the inefcient farmers should reduce their useof energy by 45e65% and enhance the yield and sustainabilityindices by 0e15% and 0e400%, respectively. SBM solution indicatesthat the most reduction should be applied in fertilizers. It conrmsthat the farmers use this input very extravagantly.

4. Conclusion

This study used a combination of fuzzy logic and DEA models toevaluate the energy use sustainability and efciency of the cornproduction and its energy use improvement. The results revealedthat specic energy and outputeinput energy ratio are 5.860 MJ/kgand 2.537, respectively. However, these indices show the farmingsituation from only input and output energy points of view notsustainability or the types of energy input like renewable or non-renewable. Using fuzzy models, we found that the corn produc-tion is in the ranges of low to medium levels of sustainability. DEA

Model C Model FMC (main)

Mean Min Max Mean Min Max

3 0.307 0.10 0.521 0.170 0.077 0.5803 0.250 0.14 0.521 0.256 0.11 0.580

0.268 0.10 0.281 0.10 0.077 0.25

44 (2012) 672e681 679revealed that sustainability indices promotion is achievable byreduction in energy input by 45e65%. Sustainability of the corncultivation can be promoted by making appropriate energy policiesand laws for agricultural sector with regard to different regions andcropping systems. Approving prohibiting law on fertilizers andchemicals applications especially nitrogen is essential as it isaccounting for 93% of all the fertilizers energy input, and the mostreduction should be applied in this part. Also, it seems that theintroduction of precision fertilizer and chemical applicators arevery benecial for some crops such as corn as a fertilizer-intensivecrop. Moreover, changing current tillage and irrigation methodstoward conservation systems like reduce or no tillage and sprinkler

[3] Pimentel D, Pimentel M, Machan MK. Energy use in agriculture: an overview,dspace.library.cornell.edu/retrieve/229/Energy;; 1998. [accessed April 2012].

[4] Pimentel D. Impacts of organic farming on the efciency of energy use in

ergyagriculture, www.organic-center.org/reportles/ENERGY_SSR.pdf;; 2006.[accessed April 2012].

[5] Patzek TW. Thermodynamics of the corn-ethanol biofuel cycle [accessed April2012], alpha.chem.umb.edu/chemistry/ch471/evans%20les/Patzek.pdf; 2006.

[6] Richard CF. Energy in farm production. Netherlands: Elsevier; 1992.[7] Pimentel D. Energy inputs in food crop production in developing and devel-

oped nations. Energies 2009;2:1e24.[8] Hatirli SA, Ozkan B, Fert C. An econometric analysis of energy input-output in

Turkish agriculture. Renewable and Sustainable Energy Reviews 2005;9:608e23.

[9] Pimentel D, Pimentel M. Global environmental resources versus world pop-ulation growth. Ecological Economics 2006;59:195e8.

[10] Pimentel D. Industrial agriculture, energy ows in. Encyclopedia of Energy2004;3:365e71.

[11] Mani I, Kumar P, Panwar JS, Kant K. Variation in energy consumption inproduction of wheat-maize with varying altitudes in hilly regions of HimachalPradesh, India. Energy 2007;32:2336e9.

[12] Zangeneh M, Omid M, Akram A. A comparative study on energy use and costanalysis of potato production under different farming technologies in Ham-adan province of Iran. Energy 2010;35:2927e33.

[13] Verma SR. Energy in production agriculture and food processing. In:Proceedings of the national conference held at the Punjab AgriculturalUniversity, Ludhiana; October 30e31, 1987.

[14] Banaeian N, Zangeneh M. Study on energy efciency in corn production ofIran. Energy 2011;36:5394e402.

[15] Unakitan G, Hurma H, Yilmaz F. An analysis of energy use efciency of canolaproduction in Turkey. Energy 2010;35:3623e7.

[16] Ozkan B, Akcaoz H, Karadeniz F. Energy requirement and economic analysis ofcitrus production in Turkey. Energy Conversion and Management 2004;45(11e12):1821e30.

[17] Singh H, Mishra D, Nahar NM, Ranjan M. Energy use pattern in productionagriculture of a typical village in arid zone India: part II. Energy Conversionand Management 2003;44:1053e67.

[18] Van Ierland EC, Lansink AO. Economics of sustainable energy in agriculture.Kluwer Academic Publisher; 2003.or drip irrigation systems is necessary in order to decrease theenergy input in addition to water and air quality, food safety andsoil fertility. Due to lack of an accurate weather prediction system,the farmers cannot manage their farms suitably. Some essentialinfrastructures such as powerful weather organization and on timeannouncement are needed to accurate farm management andtimely farm operations.

Since DEA introduces an efcient to an inefcient farmer asa reference, we observed that the majority of the efcient framers,who are selected as the reference sets, belong to cluster 1. Theefcient farmers, who are determined by DEA, cultivate corn in themedium to high levels of sustainability, therefore, if inefcientfarmers follow them, sustainability indices will be promoted tothese levels. Indeed, by using a combination of fuzzy modeling andDEA technique, agricultural sustainability would better be esti-mated and could improve up to level of sustainability which isbeing done by the efcient farmers. In other words, it seems that formore (very high) sustainable corn production, the whole society offarmers including both the efcient and inefcient, must amendtheir trends and reduce input energy as much as possible. Yet, theevaluation of optimum level of the energy utilization remains asa main issue for future studies. Also, the denition of new indiceslike ratio of renewable to non-renewable energy inputs is recom-mended. With such indices the sustainability of agriculturalsystems would be analyzed even more carefully from renewabilityof energy inputs point of view.

References

[1] Fluck RC, Baird CD. Agricultural energetics. West Port, Connecticut: AVIPublishing Company, Inc; 1980.

[2] Pimentel D. Sustainable agriculture annotated bibliography [accessed April2012], www.umich.edu/wnppcpub/resources/compendia/AGRIpdfs/AGRIannobib.pdf; 1997.

E. Houshyar et al. / En680[19] Barbir F, Ulgiati S. Sustainable energy production and consumption: benets,strategies and environmental costing. Springer; 2007.[20] McLaughlin NB, Hiba A, WALL GJ, KING DJ. Comparison of energy inputs forinorganic fertilizer and manure based corn production. Canadian AgriculturalEngineering 1999;42(1):2.1e2.14.

[21] Kraatz S. Energy inputs for corn production in Wisconsin and Germany.ASABE; 2008. Paper No: 084569, 16 pp.

[22] Chamsing A, Salokhe VM, Singh G. Energy consumption analysis for selectedcrops in different regions of Thailand. Agricultural Engineering International:the CIGR Ejournal 2006;VIII:1e18.

[23] Ministry of Agricultural e Jihad. Annual agricultural statistics. Department ofAgronomy. Ministry of Agricultural e Jihad of Iran [accessed March 2012],www.maj.ir; 2010.

[24] MOC. Fertilizer use in Iran. Ministry of commerce of Iran [accessed February2012], www.moc.gov.ir; 2008.

[25] Morrovati M. Chemical use in Iran. http://aent.persianblog.ir/post/429/[accessed September 2011].

[26] Zadeh LA, Fu KS, Tanaka K, Shimura M. Fuzzy sets and their applications tocognitive and decision processes. Academic Press, Inc; 1975.

[27] Cornelissen AMG, van den Berg J, Koops WJ, Grossman M, Udo HMJ. Assess-ment of the contribution of sustainability indicators to sustainable develop-ment: a novel approach using fuzzy set theory. Agriculture, Ecosystems andEnvironment 2001;86:173e85.

[28] Azadi H, van den Berg J, Shahvali M, Hosseininia G. Sustainable rangelandmanagement using fuzzy logic: a case study in Southwest Iran. Agriculture,Ecosystems & Environment 2009;131(3&4):193e200.

[29] Azadi H, Shahvali M, Van den Berg J, Faghih N. Sustainable rangelandmanagement using a multi-fuzzy model: how to deal with heterogeneousexperts knowledge. Journal of Environmental Management 2007;83:236e49.

[30] Ferraro DO, Ghersa CM, Sznaider GA. Evaluation of environmental impactindicators using fuzzy logic to assess the mixed cropping systems of theInland Pampa, Argentina. Agriculture, Ecosystems and Environment 2003;96:1e18.

[31] Witlox F, Derudder B. Spatial decision-making using fuzzy decision tables:theory, application and limitations. In: Petry FE, editor. Fuzzy modeling withspatial information for geographic problems. Berlin: Springer; 2005. p.253e74.

[32] Mendoza GA, Prabhu R. Fuzzy methods for assessing criteria and indicators ofsustainable forest management. Ecological Indicators 2003;3:227e36.

[33] Prato T. Assessing ecosystem sustainability and management using fuzzylogic. Ecological Economics 2007;61:171e7.

[34] Phillis YA, Andriantiatsaholiniaina LA. Sustainability: an ill-dened conceptand its assessment using fuzzy logic. Ecological economics 2001;37:435e56.

[35] Carman K. Prediction of soil compaction under pneumatic tires a using fuzzylogic approach. Journal of Terramechanics 2008;45:103e8.

[36] Marakoglu T, Carman K. Fuzzy knowledge-based model for prediction of soilloosening and draft efciency in tillage. Journal of Terramechanics 2009;3:173e8.

[37] Yang CC, Prasher SO, Landry JA, Perret J, Ramaswamy HS. Recognition ofweeds with image processing and their use with fuzzy logic for precisionfarming. Canadian Agricultural Engineering 2000;42(4):195e200.

[38] Rajaram T, Das A. Modeling of interactions among sustainability componentsof an agro-ecosystem using local knowledge through cognitive mapping andfuzzy inference system. Expert Systems with Applications 2010;37:1734e44.

[39] Jana C, Chattopadhay RN. Direct energy optimization for sustainable agricul-tural operation-a fuzzy linear programming approach. Energy for SustainableDevelopment 2005;9(3):5e12.

[40] Klaus G, La0szlo0 N, Istva0n F. Improved climate control for potato stores byfuzzy controllers. Computers and Electronics in Agriculture 2003;40:127e40.

[41] Barut ZB, Ertekin C, Karaaga HA. Tillage effects on energy use for corn silage inMediterranean Coastal of Turkey. Energy 2011;36:5466e75.

[42] Erdal G, Esengn K, Erdal H, Gndz O. Energy use and economical analysis ofsugar beet production in Tokat province of Turkey. Energy 2007;32(1):35e41.

[43] Ozkan B, Akcaoz H, Cemal F. Energy input-output analysis in Turkish agri-culture. Renewable Energy 2004;29:39e51.

[44] Dyer JA, Kulshreshtha SN, McConkey BG, Desjardins RL. An assessment offossil fuel energy use and CO2 emissions from farm eld operations usinga regional level crop and land use database for Canada. Energy 2010;35(5):2261e9.

[45] Hatirli SA, Ozkan B, Fert C. Energy inputs and crop yield relationship ingreenhouse tomato production. Renewable Energy 2006;31:427e38.

[46] Tabatabaeefar A, Emamzadeh H, Varnamkhasti MG, Rahimizadeh R, Karimi M.Comparison of energy of tillage systems in wheat production. Energy 2009;34:41e5.

[47] Alluvione F, Moretti B, Sacco D, Grignani C. EUE (energy use efciency) ofcropping systems for a sustainable agriculture. Energy 2011;36(7):4468e81.

[48] William WC, Lawrence MS, Kaoru T. Data envelopment analysis. USA:Springer; 2007.

[49] Zhuo C, Wallace EH, Scott R. Farm technology and technical efciency:evidence from four regions in China. China Economic Review 2009;20(2):153e61.

[50] De Koeijer TJ, Wossink GAA, Struik PC, Renkema JA. Measuring agriculturalsustainability in terms of efciency: the case of Dutch sugar beet growers.Journal of Environmental Management 2002;66:9e17.

[51] Nassiri SM, Singh S. Study on energy use efciency for paddy crop using

44 (2012) 672e681data envelopment analysis (DEA) technique. Applied Energy 2009;86:1320e5.

[52] Gul M, Koc B, Dagistan E, Akpinar MG, Parlakay O. Determination of technicalefciency in cotton growing farms in Turkey: a case study of Cukurova region.African Journal of Agricultural Research 2009;4(10):944e9.

[53] Zhou P, Ang BW, Poh KL. Slacks-based efciency measures for modelingenvironmental performance. Ecological Economics 2006;60:111e8.

[54] Thiam A, Boris EBU, Teodoro ER. Technical efciency in developing countryagriculture: a meta-analysis. Agricultural Economics 2001;25:235e43.

[55] Toshiyuki S. DEA non-parametric ranking test and index measurement: slack-adjusted DEA and an application to Japanese agriculture cooperatives. Omega,International Journal of Management Science. 1999;27:315e26.

[56] Mousavi-Avval SH, Raee S, Jafari A, Mohammadi A. Optimization of energyconsumption for soybean production using Data Envelopment Analysis (DEA)approach. Applied Energy 2011;88(11):3765e72.

[57] FMB. Annual meteorological report. FARS Meteorological Bureau. www.fersmet.ir [accessed January 2012].

[58] Amidi A. Sampling methods, vol. 1. Payam Noor University Publication; 2005(in Persian).

[59] Grisso RD, Kocher MF, Vaughan DH. Predicting tractor fuel consumption.Applied Engineering in Agriculture 2004;20(5):553e61.

[60] Bockari-Gevao SM, Wan Ishak WI, Azmi Y, Chan CW. Analysis of energyconsumption in lowland rice-based cropping system of Malaysia. Songkla-nakarin Journal of Science and Technology 2005;27(4):819e26.

[61] Chaudhary VP, Gangwar B, Pandey DK. Auditing of energy use and output ofdifferent cropping systems in India. Agricultural Engineering International:The CIGR Ejournal 2006;VIII:1e13.

[62] Shrestha DS. Energy use efciency indicator for agriculture, http://www.usaskca/agriculture/caedac/PDF/mcrae.PDF;; 1998. [accessed March 2011].

[63] Kalyani S, Swarup KS. Particle swarm optimization based K-means clusteringapproach for security assessment in power systems. Expert Systems withApplications 2011;38:10839e46.

[64] Everitt BS, Landau S, Leese M. Cluster analysis. 4th ed. Arnold; 2001.[65] Dubois D, Prade H. An introduction to fuzzy set systems. Clinica Chimica Acta

1998;270:3e29.[66] Prato T. A fuzzy logic approach for evaluating ecosystem sustainability.

Ecological Modeling 2005;187(2e3):361e8.[67] Chen G, Pham TT. Introduction to fuzzy sets, fuzzy logic, and fuzzy control

systems. CRC Press; 2001.[68] Tran LT, Knight CG, ONeill RV, Smith ER, Riitters KH, Wickham J. Fuzzy

decision analysis for integrated environmental vulnerability assessment of themid-Atlantic region. Environmental Management 2002;29:845e59.

[69] IUCN. IUCN resource kit for sustainability assessment. Gland, Switzerland:IUCN; 2001. 95 pp.

[70] Cooper WW, Seiford LM, Tone K. Data envelopment analysis: a comprehen-sive text with models, applications, references and DEA-solver software. USA:Kluwer Academic Publishers; 2004.

[71] Charnes A, Cooper WW, Rhodes E. Measuring the efciency of decision-making units. European Journal of Operational Research; 1978:2.

[72] Banker RD, Charnes A, Cooper WW. Some models for estimating technicaland scale in efciencies in data envelopment analysis. Management Science;1984:30.

[73] Lee C. Analysis of overall technical efciency, pure technical efciency andscale efciency in the medium-sized audit rms. Expert Systems withApplications 2009;36(8):11156e71.

[74] Morita H, Hirokawa K, Zhu J. Aslack-based measure of efciency in context-dependent data envelopment analysis. Omega 2005;33:357e62.

[75] Pimentel D, Williamson S, Alexander CE, Pagan OG, Kontak C, Mulkey SE.Reducing energy inputs in the US food system. Human Ecology 2008;36:459e71.

[76] Houshyar E. Principles of bioenergies: biodiesel, biogas, bioethanol, bio-hydrogen, gasication. Iran: Nosooh Publication; 2012 [in Persian].

E. Houshyar et al. / Energy 44 (2012) 672e681 681

Sustainable and efficient energy consumption of corn production in Southwest Iran: Combination of multi-fuzzy and DEA modeling1. Introduction1.1. Energy and sustainability1.2. Why using fuzzy modeling and DEA technique?

2. Methodology2.1. Energy use analysis2.2. Non-hierarchical cluster analysis2.3. Multi-fuzzy modeling2.4. DEA technique2.4.1. Technical efficiency2.4.2. Pure technical efficiency2.4.3. Scale efficiency2.4.4. SBM

3. Results and discussion3.1. Energy use analysis3.2. Sustainability and efficiency analysis using multi-fuzzy models and DEA

4. ConclusionReferences