6
Energy use efciency in greenhouse tomato production in Iran Reza Pahlavan * , Mahmoud Omid, Asadollah Akram Department of Agricultural Machinery Engineering, Faculty of Agricultural Engineering and Technology, School of Agriculture & Natural Resources, Universityof Tehran, Karaj, Iran article info Article history: Received 23 July 2011 Received in revised form 21 October 2011 Accepted 23 October 2011 Available online 10 November 2011 Keywords: Data envelopment analysis Optimization Energy productivity Technical efciency Yield abstract Efcient use of energy in agriculture is one of the conditions for sustainable production. In the present study energy use pattern for tomato production in Iran was investigated and a non-parametric data envelopment analysis (DEA) technique was applied to analyze the technical and scale efciencies of farmers with respect to energy use for crop production. The energy use pattern indicated that diesel, electricity and chemical fertilizers are the major energy consuming inputs for tomato production in the region. Moreover, the results of DEA application revealed that of the average pure technical, technical and scale efciencies of farmers were 0.94, 0.82 and 0.86, respectively. Also the results revealed that by adopting the recommendations based on the present study, on an average, about 25.15% of the total input energy could be saved without reducing the tomato yield. Ó 2011 Elsevier Ltd. All rights reserved. 1. Introduction Tomato is one of the major greenhouse vegetables products worldwide. In Iran, tomato production was 4.83 million tonnes in 2008. From 2002 to 2008, greenhouse areas of Iran increased from 3380 ha to 7000 ha [1]. The share of greenhouse production was as follows: vegetables 59.3%, owers 39.81%, fruits 0.54% and mush- room 0.35% [2]. Agriculture itself is an energy user and energy supplier in the form of bio-energy [3]. Energy is used in every form of inputs such as human, fertilizers, pesticides, machinery and electricity, to perform various operations for crop production. Energy use in agriculture has developed in response to increasing populations, limited supply of arable land and desire for an increasing standard of living. In all societies, these factors have encouraged an increase in energy inputs to maximize yields, minimize labor-intensive practices, or both [4]. Effective energy use in agriculture is one of the conditions for sustainable agricultural production, since it provides nancial savings, fossil resources preservation and air pollution reduction [5]. There are several studies on the energy use pattern and benchmarking of crops production. Energy use for greenhouse vegetables (tomato, cucumber, eggplant and pepper) production were investigated [6e8]. Hatirli et al. and Mohammadi and Omid investigated energy inputs and crop yield relationship to develop and estimate an econometric model for greenhouse tomato and cucumber productions, respectively [9,10]. Also Omid et al. inves- tigated energy use pattern and benchmarking of greenhouse cucumber producers in Tehran city of Iran using data envelopment analysis (DEA) [8]. This paper presents an application of DEA to discriminate ef- cient tomato producers from inefcient ones, recognize wasteful uses of energy inputs by inefcient farmers and suggest necessary quantities of different inputs to be used by each inefcient farmer from every energy source. 2. Material and methods 2.1. Data collection and processing Initial data used in the DEA analysis comprised information on greenhouse tomato producers in the region. They were collected from 31 producers by using a face to face questionnaire method. Before carrying detailed analysis, we attempted to secure homo- geneity by selecting only greenhouses in a specic area. The main drawbacks of deterministic frontier modelsdboth non-parametric and parametric modelsdis that they are very sensitive to outliers and extreme values, and that noisy data are not allowed. Simar [11] pointed out the need for identifying and eliminating outliers when using deterministic models. Outliers are observations that "do not t in with the pattern of the remaining data points and are not at all typical of the rest of the data". By applying sample means, standard deviations, maximum and * Corresponding author. Tel.: þ98 261 2801038, þ98 936 6466720(mobile); fax: þ98 261 2808138. E-mail address: [email protected] (R. Pahlavan). Contents lists available at SciVerse ScienceDirect Energy journal homepage: www.elsevier.com/locate/energy 0360-5442/$ e see front matter Ó 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.energy.2011.10.038 Energy 36 (2011) 6714e6719

Energy use efficiency in greenhouse tomato production in Iran

Embed Size (px)

Citation preview

Page 1: Energy use efficiency in greenhouse tomato production in Iran

at SciVerse ScienceDirect

Energy 36 (2011) 6714e6719

Contents lists available

Energy

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

Energy use efficiency in greenhouse tomato production in Iran

Reza Pahlavan*, Mahmoud Omid, Asadollah AkramDepartment of Agricultural Machinery Engineering, Faculty of Agricultural Engineering and Technology, School of Agriculture & Natural Resources, University of Tehran, Karaj, Iran

a r t i c l e i n f o

Article history:Received 23 July 2011Received in revised form21 October 2011Accepted 23 October 2011Available online 10 November 2011

Keywords:Data envelopment analysisOptimizationEnergy productivityTechnical efficiencyYield

* Corresponding author. Tel.: þ98 261 2801038,fax: þ98 261 2808138.

E-mail address: [email protected] (R. Pahlavan

0360-5442/$ e see front matter � 2011 Elsevier Ltd.doi:10.1016/j.energy.2011.10.038

a b s t r a c t

Efficient use of energy in agriculture is one of the conditions for sustainable production. In the presentstudy energy use pattern for tomato production in Iran was investigated and a non-parametric dataenvelopment analysis (DEA) technique was applied to analyze the technical and scale efficiencies offarmers with respect to energy use for crop production. The energy use pattern indicated that diesel,electricity and chemical fertilizers are the major energy consuming inputs for tomato production in theregion. Moreover, the results of DEA application revealed that of the average pure technical, technical andscale efficiencies of farmers were 0.94, 0.82 and 0.86, respectively. Also the results revealed that byadopting the recommendations based on the present study, on an average, about 25.15% of the total inputenergy could be saved without reducing the tomato yield.

� 2011 Elsevier Ltd. All rights reserved.

1. Introduction

Tomato is one of the major greenhouse vegetables productsworldwide. In Iran, tomato production was 4.83 million tonnes in2008. From 2002 to 2008, greenhouse areas of Iran increased from3380 ha to 7000 ha [1]. The share of greenhouse production was asfollows: vegetables 59.3%, flowers 39.81%, fruits 0.54% and mush-room 0.35% [2].

Agriculture itself is an energy user and energy supplier in theform of bio-energy [3]. Energy is used in every form of inputs suchas human, fertilizers, pesticides, machinery and electricity, toperform various operations for crop production. Energy use inagriculture has developed in response to increasing populations,limited supply of arable land and desire for an increasing standardof living. In all societies, these factors have encouraged an increasein energy inputs to maximize yields, minimize labor-intensivepractices, or both [4]. Effective energy use in agriculture is one ofthe conditions for sustainable agricultural production, since itprovides financial savings, fossil resources preservation and airpollution reduction [5].

There are several studies on the energy use pattern andbenchmarking of crops production. Energy use for greenhousevegetables (tomato, cucumber, eggplant and pepper) productionwere investigated [6e8]. Hatirli et al. and Mohammadi and Omidinvestigated energy inputs and crop yield relationship to develop

þ98 936 6466720(mobile);

).

All rights reserved.

and estimate an econometric model for greenhouse tomato andcucumber productions, respectively [9,10]. Also Omid et al. inves-tigated energy use pattern and benchmarking of greenhousecucumber producers in Tehran city of Iran using data envelopmentanalysis (DEA) [8].

This paper presents an application of DEA to discriminate effi-cient tomato producers from inefficient ones, recognize wastefuluses of energy inputs by inefficient farmers and suggest necessaryquantities of different inputs to be used by each inefficient farmerfrom every energy source.

2. Material and methods

2.1. Data collection and processing

Initial data used in the DEA analysis comprised information ongreenhouse tomato producers in the region. They were collectedfrom 31 producers by using a face to face questionnaire method.Before carrying detailed analysis, we attempted to secure homo-geneity by selecting only greenhouses in a specific area.

The main drawbacks of deterministic frontier modelsdbothnon-parametric and parametric modelsdis that they are verysensitive to outliers and extreme values, and that noisy data are notallowed. Simar [11] pointed out the need for identifying andeliminating outliers when using deterministic models. Outliers areobservations that "do not fit in with the pattern of the remainingdata points and are not at all typical of the rest of the data". Byapplying sample means, standard deviations, maximum and

Page 2: Energy use efficiency in greenhouse tomato production in Iran

Nomenclature

DEA data envelopment analysisGJ Giga jouleDMU decision making unitFAO food and agriculture organizationTE technical efficiencyPTE pure technical efficiencyCCR Banker, Charnes & CooperBCC Charnes, Cooper & RhodesCRS constant returns to scaleVRS variable returns to scaleSE scale efficiencyIRS increasing returns to scaleDRS decreasing returns to scaleRTS return to scaleSD standard deviation

Table 1Energy coefficients of different inputs and outputs used.

Input & output Units Energy coefficient, (GJ unit�1) Reference

A. Input1. Human labour hr 1.96 * 10�3 [4]2. Diesel fuel L 5.63 * 10�2 [13]3. Electricity MW 11.93 [4]4. Fertilizers ton(a) Nitrogen 66.14 [14](b) P2O5 12.44 [14](c) K2O 11.15 [14](d) Micro 120 [15]

5. Chemicals ton 120 [13]B. Output1. Tomato ton 0.8 [9]

R. Pahlavan et al. / Energy 36 (2011) 6714e6719 6715

minimum values and plots of all the variables four outliers wereidentified. The remaining 27 DMUs were used in developing input-oriented DEA models.

The Esfahan province is located within 30e42� and 34e30�

north latitude and 49e36� and 55e32� east longitude. The averagesize of the studied greenhouses has been found to be 0.2 ha. Thecommercial greenhouses surveyed here were made from galva-nized steel. Steel’s greatest value in greenhouse construction is itsstrength. Also they are long-lasting, low cost, and require lessframework (thus less shadowing) than any other framing materialthanks to steel’s natural strength. The top of the greenhouses wascovered with UV stabilized plastic sheet 200 m thickness. Data werecollected from the farmers in the production period of 2009-2010.The size of each sample was determined using the Neyman tech-nique [12]. The inputs used in tomato production were in the formof human labour, chemical fertilizer, chemicals, electricity andmachinery; while the tomato yieldwas the single output. The energyequivalents of these inputs and output were calculated using theenergy equivalent coefficients as presented in Table 1. The previousstudy was used to determine the energy equivalents’ coefficients[4,9,13,14,15]. The total input equivalent can be calculated by addingup the energy equivalences of all inputs in Giga Joule (GJ).

Based on the energy equivalents of the inputs and output, theindicators of energy use including energy ratio (energy use effi-ciency), energy productivity and net energy were calculated asfollow [12, 16]:

Energy Use Efficiency ¼Energy Output�GJ ha�1

�.

Energy Input�GJ ha�1

� (1)

Energy Productivity ¼ tomato output�ton ha�1

�.

Energy Input�GJ ha�1

� (2)

Net Energy ¼ Energy Output�GJ ha�1

� Energy Input�GJ ha�1

�(3)

2.2. Data envelopment analysis

Farrell (1957) proposed a new approach to efficiency measure-ment and the production frontier at the micro level [17]. He divided

economic efficiency into resource use (technical) and allocative(price) components. He proposed a piecewise linear envelopmentof data as the conservative estimate of the production frontierwhich envelopes observation points as closely as possible whichwas estimated by solving a system of linear equations. According toFarrell (1957), technical efficiency (TE) represents the ability ofa decision-making unit (DMU) to produce maximum output givena set of inputs and technology (output-oriented) or, alternatively, toachieve maximum feasible reductions in input quantities giveninput prices and output (input-oriented) [17]. The choice betweeninput and output-oriented measures is a matter of concern, andselectionmay vary according to the unique characteristics of the setof DMUs under study. In this study, input-oriented DEA seemsmoreappropriate, given that it is more reasonable to argue that in theagricultural sector a farmer has more control over inputs ratherthan output levels. DEA allows for the measurement of relative effi-ciency for a group of DMUs that use various inputs to produce outputs.

There are two kinds of DEA models included: CCR and BCCmodels. The CCR model [18] is built on the assumption of constantreturns to scale (CRS) of activities, but the BCCmodel [19] is built onthe assumption of variable returns to scale (VRS) of activities. TheDEAmodels have been described in details by several authors [18,19],thus a detailed description is not provided here. The dual (envel-opment) form of the DEA linear programming problem is simpler tosolve than the ratio and multiplier forms due to fewer constraints.

Efficiency by DEA is defined in three different forms: overalltechnical efficiency (TECCR), pure technical efficiency (TEBCC) andscale efficiency (SE).

2.3. Technical efficiency

Technical efficiency (TE) can be calculated by the ratio of sum ofweighted outputs to sum of weighted inputs [20]:

q ¼PP

p¼1upyp;jPQq¼1nqxq;j

(4)

where ‘x’ and ‘y’ are inputs and outputs, ‘v’ and ‘u’ are input andoutput weights, respectively, ‘q’ is the number of inputs (q¼ 1,2,. ,Q); ‘p’ is the number of outputs (p ¼ 1,2,., P), and ‘j’ represents jthDMU.

The first development of DEA was by Charnes, Cooper andRhodes (CCR) to measure the efficiency of individual DMUs.Mathematically, the CCR DEA model for measuring the input-oriented technical efficiency of a DMU is written as follows [21]:

max qs:t: :Yl � YoqXo � Xl � 0;qfree; l � 0:

(5)

Page 3: Energy use efficiency in greenhouse tomato production in Iran

Table 2Summary of inputs (source wise energy use, GJ/ha) and output (yield, ton/ha).

Diesel Chemicals Fertilizers Human Electricity Yield

Max 24,330.4 35.1 174.8 29.2 12,235.3 500.0Min 9653.1 6.3 28.3 15.0 1817.3 140.0Average 17,190.1 16.4 84.7 24.5 4517.1 291.5STD 3736.6 6.1 39.6 3.8 2597.6 76.8

R. Pahlavan et al. / Energy 36 (2011) 6714e67196716

where q is the technical efficiency of DMU to be evaluated DMUoand l represents the intensity of the efficient DMUs in projectinginefficient DMUs onto the efficient frontier, also called theconvexity constant. The optimal efficiency of a DMU, q*, will be lessthan or equal to 1. DMUs with q* <1 are inefficient while DMUswith q* ¼ 1 form a set of boundary (frontier) points. The envelop-ment problem (Eq. (5)) evaluates a DMUo by comparing it withother DMUs in the group. Themodel allocates aminimumvalue q toDMUo provided that a combination of other DMUs does notconsume more inputs and yield output(s) at least equal to DMUo.The linear programming problem must be repeated for each DMUj,such that (Xo, Yo) ¼ (Xj, Yj) for j ¼ 1,2,., n, where Xo and Yo areinputs and outputs of the DMU to be evaluated, and n is the totalnumber DMUs considered in the DEA analysis.

Pure technical efficiency (TEBCC) is the TE of BCC model. The BCCmodel was initially proposed by Banker, Charnes and Cooper (1984)[19]. The function of input-oriented BCC model for evaluatingefficiency of DMUj (TEBCC) is like CCR model, but in this model theequation

PJj¼1lj ¼ 1 is a convexity constraint, which specifies the

VRS framework [22]. Without this convexity constraint, the BCCmodel will be a CCR model (Eq. (5)) describing a CRS situation.

If there is no restriction on l (l � 0), the solution to Eq. (5)represents constant returns to scale [23]. Constant returns toscale (CRS) imply that a given increase in inputs would result ina proportionate increase in outputs and the feasible region of theenvelopment problem becomes a conical hull. A restriction on l

(l ¼ 1) leads to no condition on the allowable returns to scale, alsocalled variable returns to scale (VRS). Under this condition, theperformance frontier line or hyperplanes are not then restricted topass through the origin [21]. An increase in inputs may not result ina proportionate increase in outputs in this case. Due to convexity,the efficient DMUs form a convex hull on which all inefficientpoints are projected.

Banker et al. (1984) extended the CCRmodel to the estimation ofthe most productive scale size (MPSS) [19]. The MPSS was definedas the scale where CRS prevails and the slope of outputs to inputs is1. Increasing returns to scale (IRS) exists if the slope exceeds 1 anddecreasing returns to scale (DRS) occurs when slope of the line isless than 1. IRS indicates that an increase in the input resourcesproduces more than proportionate increase in outputs. Similarly,DRS suggests a less than proportionate increase in the outputs inresponse to an increase in inputs.

Based on the CCR and BCC scores, scale efficiency defined by[20]:

SE ¼ TECCRTEBCC

(6)

By solving of CCR and BCC models, the weights of inputs(human, diesel, fertilizer, electricity and chemicals) and output(tomato yield) would be calculated so the maximum value of q iscalculated.

In this study, we used DEA-solver software to calculate CRS andVRSwith radial distances to the efficient frontier and determine theamount of energy loss and energy savings of inefficient farmers.

3. Results and discussion

3.1. Energy use pattern analysis

The summarized information on energy use pattern and yieldvalue of tomato production is presented in Table 2. The energy usepattern indicated that diesel, electricity and chemical fertilizers arethe major energy consuming inputs for tomato production in theregion. The average of diesel, chemicals, chemical fertilizer, human

and electricity energy were 17,190.1, 16.4, 84.7, 24.5 and4517.1 GJ ha�1, respectively. Also, the summarized statistics forenergy inputs and output are shown in Table 2. The wide variationbetween energy inputs and output are noticeable. It was due to themismanagement of resource usage between the farmers, indicatingthat there is a potential for improving energy use pattern of tomatoproduction in the region. The energy consumption of diesel andelectricity due to use of heaters and pumps with low efficiency andalso low price of diesel fuel and electricity in Iran, were very high inthe studied area. Also, the high contribution of chemical fertilizerenergy showed that, all of farmers were not fully aware of propertime and quantity of fertilizers usage.

3.2. Identifying efficient and inefficient farmers

The results of BCC model indicated that from the total of 27farmers considered for the analysis, 15 farmers had the pure tech-nical efficiency score of one; while, the remainder of 12 farmerssecured their efficiency scores less than one and were relativelyinefficient in energy use from the different sources. However, thetechnical efficiency estimation indicates that only 8 farmers wereefficient.

The values of the pure technical efficiency, scale efficiency andtechnical efficiency are summarized in Table 3 and Fig. 1.

The average values (for all 27 farmers considered) of puretechnical efficiency (PTE), technical efficiency (TE) and scale effi-ciency (SE) were found to be 0.94, 0.82 and 0.86, respectively. Themean value of SEs for the inefficient farmers (0.66) indicates thatthere is ample scope for improving their operating practices toenhance their energy use efficiency. The results of the RTS indicatedthat all efficient farmers (based on CCR model) were operating atCRS, whereas inefficient farmers were found to be operating at IRSor CRS. The average of SE was as low as 0.86, which indicates that ifinefficient farmers utilize their inputs efficiently, considerablesavings in energy from the different sources is possible without anychange in technological practices.

3.3. Benchmarking

The performance of inputs depends on tomato yields achievedin relation to resources consumed in the process. In general, theperformance assessment may be carried out by comparinga particular system with key competitors having best performancewithin the same group or another group performing similar func-tions [24]. This process is called benchmarking. Table 3 shows theresults of technical efficiency analysis for the 27 tomato productionunit (DMUs). The CRS analysis (CCR model) shows that 8 out of 27DMUs are efficient. These efficient DMUs can be selected by inef-ficient DMUs as best practice DMUs, making them a compositeDMU instead of using a single DMU as a benchmark.

A composite DMU is formed by multiplying the intensity vectorl by the inputs and outputs of the respective efficient DMUs. Forexample, for DMU4, the composite DMU that represents the bestpractice or reference composite benchmark DMU is formed by thecombination of DMUs 1, 8 and 12. This means DMU4 is close to theefficient frontier segment formed by these efficient DMUs,

Page 4: Energy use efficiency in greenhouse tomato production in Iran

Table 3Efficiency estimation results of tomato production in Esfahan, Iran.

DMU Technical efficiency SE Frequencyin referentset

Benchmarks RTS

CRS VRS

1 1.00 1.00 1.00 5 Constant2 0.31 0.71 0.44 5(0.06) 8(0.12)

12(0.09) 27(0.09)Increasing

3 1.00 1.00 1.00 2 Constant4 0.69 0.79 0.87 1(0.13) 8(0.42)

12(0.12)Increasing

5 1.00 1.00 1.00 10 Constant6 1.00 1.00 1.00 0 Constant7 0.73 1.00 0.73 1(0.43) 5(0.05)

8(0.08)Increasing

8 1.00 1.00 1.00 16 Constant9 0.62 1.00 0.62 12(0.41) Increasing10 0.95 1.00 0.95 1(0.33) 5(0.25)

8(0.31)Increasing

11 0.83 1.00 0.83 5(0.16) 8(0.35)12(0.10)

Increasing

12 1.00 1.00 1.00 14 Constant13 0.77 0.88 0.88 5(0.23) 8(0.25)

12(0.28)Increasing

14 0.57 0.82 0.69 8(0.18) 12(0.26)27(0.17)

Increasing

15 0.81 0.92 0.87 3(0.16) 8(0.44)12(0.09)

Increasing

16 1.00 1.00 1.00 5(0.36) 8(0.17)12(0.38)

Constant

17 0.91 1.00 0.91 1(0.27) 5(0.03)8(0.50)

Increasing

18 0.72 0.95 0.76 12(0.16) 27(0.58) Increasing19 0.93 1.00 0.93 5(0.02) 8(0.24)

12(0.52)Increasing

20 1.00 1.00 1.00 1 Constant21 0.57 0.78 0.72 5(0.05) 8(0.29)

27(0.25)Increasing

22 0.77 1.00 0.77 1(0.22) 8(0.34)12(0.02)

Increasing

23 0.88 0.92 0.95 8(0.67) 20(0.20) Constant24 0.83 1.00 0.83 12(0.22) 27(0.44) Increasing25 0.82 0.86 0.95 3(0.16) 8(0.38)

12(0.24)Constant

26 0.53 0.83 0.64 5(0.06) 8(0.27)12(0.14)

Increasing

27 1.00 1.00 1.00 5 ConstantAverage 0.82 0.94 0.86SD 0.18 0.09 0.15

Fig. 1. The overall, pure and scale e

R. Pahlavan et al. / Energy 36 (2011) 6714e6719 6717

represented in the composite DMU. The selection of these efficientDMUs is made on the basis of their comparable level of inputs andoutput yield to DMU4. In Table 3, the benchmark DMU for unit 4 isexpressed as 1(0.13) 8(0.42) 12(0.12), where 1, 8 and 12 are theDMU numbers while the values between brackets are the intensityvector l for the respective DMUs. The higher value of the intensityvector l for unit 8 (¼0.42) indicates that its level of inputs andoutput is closer to DMU4 compared to other DMUs. The summationof all intensity vectors in a benchmark DMU must equal 1. On theother hand, the unit 8 appears 16 times in the reference set ofinefficient DMUs. This places unit 8 closest to the input and outputlevels of most of the inefficient DMUs but uses fewer inputs.

While the DEA results highlight the lower yield of inefficientunits, a more detailed analysis by including the effects of uncon-trollable exogenous variables, such climatic conditions and soilfertility as well as agricultural practices, ownership, producer’sexperience and education should be incorporated in future studiesin order to investigate the causes of inefficiency. Furthermore, theseunits are not perfectly competitive and therefore cannot be treatedon equal grounds. However, by identifying those units with loweryield, this analysis provides a quantification of the yield in theseunits in relation to those performing at the frontier of high yield,thus enabling producers and scientists to focus their attention onthose units with lower performance to determine the actualunderlying causes of that under performance.

3.4. Returns to scale (RTS)

The above comparison, however, does not allow us to discrim-inate between DMUs with IRS or DRS. Therefore, an additional testwas applied to determine whether a DMU has IRS or DRS byapplying the NIRS model. In the NIRS model, the BCC model ismodified by replacing the constraint (l¼ 1) by (l� 1) in Eq. (5). TheBCC model represents both IRS as well as DRS, while the NIRSmodel represents only DRS. If approximately similar values of TEare obtained from the application of the NIRS and BCC models theDMU has DRS. Conversely, if the NIRS and BCC models yielddifferent values of TE the DMU has IRS.

The results of returns to scale indicated that from the totalfarmers considered for the analysis, 11 farmers were operating atCRS, showing the optimum scale of their practices and 16 farmerswere operating at IRS. This indicates that the majority of tomatoproducers in the region were operating below their optimal scale;

fficiencies of inefficient farms.

Page 5: Energy use efficiency in greenhouse tomato production in Iran

Table 4Improvement of energy indices for tomato production.

Items Unit Presentquantity

Targetquantity

Difference(%)

Tomato yield ton ha�1 291.5 291.5 0.00Energy use

efficiencye 0.01 0.01 �33.61

Energy productivity ton GJ�1 0.01 0.02 �33.61Net energy GJ ha�1 �21,599.5 �16,107.6 25.43Total input energy GJ ha�1 21,832.7 16,340.8 25.15Total output energy GJ ha�1 233.2 233.2 0.00

Table 6Energy saving from different sources if recommendations of study are followed.

Inputs Quantity inpresent useGJ/ha

Quantity intarget useGJ/ha

Energysaving,GJ/ha

%

Diesel 17,190.1 13,465.5 3724.5 21.67Chemicals 16.4 11.6 4.7 28.97Fertilizers 84.7 51.8 32.9 38.79Human 24.5 19.8 4.6 18.97Electricity 4517.1 2791.9 1725.2 38.19Total input energy 21,832.7 16,340.8 5491.9 25.15

R. Pahlavan et al. / Energy 36 (2011) 6714e67196718

therefore a proportionate increase in all inputs leads to more thanthe proportionate increase in outputs. Fraser and Cordina appliedDEA to investigate the technical efficiency of dairy farms in Aus-tralia. They reported that a significant number of farmers wereoperating at IRS [25]. Also, Jaforullah and Whiteman examined thescale efficiency of New Zealand dairy farms; they reported that 19%of farms were operating at optimal scale size, 28% at above optimalscale and 53% at bellow optimal scale [26].

Table 4 presents the energy indices for tomato production inpresent and target conditions. The average yield in tomatoproductionwas determined at 291.5 ton ha�1. Energy use efficiency(Eq. (1)) is calculated as 0.01 in present use of energy. Other resultssuch as 0.74 for cotton [27], 0.76 for cucumber, 0.61 for eggplant,0.99 for pepper [28], 0.32 for tomato, 0.31 for cucumber, 0.23 foreggplant, 0.19 for pepper [7] have been reported for different crops,showing the inefficient use of energy in the tomato production inregion. It is concluded that the energy ratio can be increased byraising tomato yield and/or by decreasing energy inputsconsumption. The average energy productivity of tomato (Eq. (2))was 0.01 ton GJ�1 in present use of energy. This means that 0.01units output was obtained per unit energy. Calculation of energyproductivity rate is well documented in the literature such as;soybean (0.18) [29], tomato (0.4), cucumber (0.39), eggplant (0.29),and pepper (0.23) [7]. In addition high consumption of diesel andelectricity, due to the lack of soil analysis, chemical fertilizersenergy was high in the studied area and therefore, energy useefficiency and energy productivity in this study were low. The netenergy (Eq. (3)) of tomato production was �21,599.5 GJ ha�1 inpresent use of energy. Net energy is negative, therefore, it can beconcluded that in tomato production, energy is being lost.

After application of DEA, energy productivity was calculated as0.02 ton GJ�1, in target use of energy, showing an improvement of

Table 5The actual and suggested values of energy use for inefficient farmers.

DMU SE Actual values ofenergy (GJ ha�1)

Suggested values ofenergy (GJ ha�1)

Yield(ton ha�1)

Saving%

2 0.44 24,810.6 7706.2 140.0 68.944 0.87 25,022.6 16,823.9 300.0 32.777 0.73 30,374.9 12,840.2 200.0 57.739 0.62 17,996.2 7717.4 142.9 57.1210 0.95 31,110.7 20,132.8 340.0 35.2911 0.83 17,170.9 14,200.8 260.0 17.3013 0.88 20,886.1 16,056.1 293.0 23.1314 0.69 23,682.2 12,932.6 236.7 45.3915 0.87 20,121.0 16,171.2 300.0 19.6316 1.00 18,024.8 17,914.8 325.0 0.6117 0.91 26,803.3 20,161.2 350.0 24.7818 0.76 24,437.3 13,544.9 241.4 44.5719 0.93 17,819.6 16,492.9 305.0 7.4521 0.72 27,914.7 13,225.4 240.0 52.6222 0.77 21,264.9 14,527.6 250.0 31.6823 0.95 25,584.6 21,761.6 400.0 14.9424 0.83 16,173.8 12,348.5 221.4 23.6525 0.95 21,583.7 17,521.0 325.0 18.8226 0.64 20,441.0 10,861.4 200.0 46.86

33.61%. Also net energy in target conditions was found tobe �16107.5 GJ ha�1.

3.5. Setting realistic input levels for inefficient farmers

The pure technical efficiency score of a farmer that is less thanone indicates that, at present, he is using more energy thanrequired from the different sources. Therefore, it is desired tosuggest realistic levels of energy to be used from each source forevery inefficient farmer in order to avert wastage of energy withoutreducing the yield level. Table 5 gives, for each inefficient farmer,the scale efficiency (SE), the actual energy use (GJ ha�1), the rec-ommended target energy use (GJ ha�1), and the percent saving intotal energy use. From the last column of Table 5 it is evident thatthe percent saving ranges from 0.61% for farmer Number16e68.94% for farmer Number 2. In a similar study, Omid et al.(2011) found the producer wise percent saving was between 5.8%and 69.8% for greenhouse cucumber production [8].

A summary of the information available in Table 5 is presentedin Table 6. Using the information in Tables 5 and 6, it is possible toadvise a farmer regarding the better operating practices followedby his peers in order to reduce the input energy level to the targetvalues indicated in the analysis while achieving the output levelpresently achieved by him. It gives the average energy spent andtargeted (GJ ha�1), possible energy savings and percent of energysaving in each energy source. We note from Table 6 that thepossible overall energy saving is 25.15%. The amount of diesel,electricity, fertilizers, chemicals and human energy saving were3724.5, 1725.2, 32.9, 4.7, and 4.6 GJ ha�1, respectively.

Results shows that reduce in diesel fuel, electricity and fertil-izers consumptions are important for energy saving and decreasingthe environmental risk problem in the area. A saving in diesel fueland electricity by improving heaters and pumps performance andin fertilizer by soil analysis may be possible.

4. Conclusions

In this research, an energy analysis in tomato production inEsfahan province of Iran was conducted and a data envelopmentanalysis was applied to analyze the efficiencies of farmers. Theresults indicated that total input energy, total output energy andenergy use efficiency were 21,833.7 GJ ha�1, 233.2 GJ ha�1 and 0.01,respectively. Diesel, electricity and fertilizers are the main energyconsuming inputs; it is due to the low efficiency of energyconversion by heater and electric motor. Also, the high contributionof fertilizer energy showed that, all of farmers are not fully aware ofproper time and quantity of fertilizers usage.

The results of DEA application showed that the average values ofpure technical efficiency, technical efficiency and scale efficiencywere 0.94, 0.82 and 0.86, respectively. Also, on an average, about25.15% of total input energy could be saved without reducing thetomato yield from its present level by adopting the

Page 6: Energy use efficiency in greenhouse tomato production in Iran

R. Pahlavan et al. / Energy 36 (2011) 6714e6719 6719

recommendations based on the present study. Results shows thatreduce in diesel fuel, electricity and fertilizers consumptions areimportant for energy saving and decreasing the environmental riskproblem in the area. A saving in diesel fuel and electricity byimproving heaters and pumps performance and in fertilizer by soilanalysis may be possible.

Totally, DEA is very suitable to analyze these data and extractmany distinctive features of their practices. It helped in finding thewasteful uses of energy by inefficient farmers and suggest neces-sary quantities of different inputs to be used by each inefficientfarmer from every energy source.

Acknowledgement

The financial support provided by the University of Tehran, Iran,is duly acknowledged.

References

[1] Food and Agriculture Organization (FAO), www.fao.org; 2008.[2] Anonymous. Iran annual agricultural statistics. Ministry of Jihad-e-Agriculture

of Iran, www.maj.ir; 2008.[3] Alam MS, Alam MR, Islam KK. Energy flow in agriculture: Bangladesh. Am J

Environ Sci 2005;1(3):213e20.[4] Esengun K, Gunduz O, Erdal G. Inputeoutput energy analysis in dry apricot

production of Turkey. Energy Convers Manage 2007;48:592e8.[5] Uhlin H. Why energy productivity is increasing: an inputeoutput analysis of

Swedish agriculture. Agric Syst 1998;56(4):443e65.[6] Ozkan B, Fert C, Karadeniz CF. Energy and cost analysis for greenhouse and

open-field grape production. Energy 2007;32:1500e4.[7] Canakci M, Akinci I. Energy use pattern analyses of greenhouse vegetable

production. Energy 2006;31:1243e56.[8] Omid M, Ghojabeige F, Delshad M, Ahmadi H. Energy use pattern and

benchmarking of selected greenhouses in Iran using data envelopment anal-ysis. Energy Convers Manage 2011;52:153e62.

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

[10] Mohammadi A, Omid M. Economical analysis and relation between energyinputs and yield of greenhouse cucumber production in Iran. Appl Energy2010;87:191e6.

[11] Simar L. Detecting outliers in frontiers models: a simple approach. J ProductivAnal 2003;20:391e424.

[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] Heidari MD, Omid M. Energy use patterns and econometric models of majorgreenhouse vegetable productions in Iran. Energy 2011;36:220e5.

[14] Shrestha DS. Energy use efficiency indicator for agriculture. Available from:http://www.usaskca/agriculture/caedac/PDF/mcrae.PDF; 1998.

[15] Mandal KG, Saha KP, Ghosh PK, Hati KM, Bandyopadhyay KK. Bioenergy andeconomic analysis of soybean-based crop production systems in central India.Biomass Bioenergy 2002;23(5):337e45.

[16] Moore SR. Energy efficiency in small-scale biointensive organic onionproduction in Pennsylvania, USA. Renew Agric Food Syst 2010;25(3):181e8.

[17] Farrell MJ. The measurement of productive efficiency. J Roy Stat Soc A Sta1957;120:253e81.

[18] Charnes A, Cooper WW, Rhodes E. Measuring the efficiency of decisionmaking units. Eur J Oper Res 1978;2:429e44.

[19] Banker RD, Charnes A, Cooper WW. Some models for estimating technical andscale inefficiencies in data envelopment analysis. Manage Sci 1984;30:1078e92.

[20] Cooper WW, Seiford LM, Tone K. Introduction to DEA and its uses with DEA-Solver software and references. New York: Springer; 2006.

[21] Charnes A, Cooper WW, Lewin AY, Seiford LM. Data envelopment analysis:Theory, Methodology, and Applications. Boston/Dordrecht/London: KluwerAcademic Publishers; 1994. pp. 513.

[22] Mostafa MM. Modeling the efficiency of top Arab banks: a DEAeneuralnetwork approach. Expert Syst Appl 2009;36:309e20.

[23] Seiford LM, Thrall RM. Recent developments in DEA: the mathematicalprogramming approach to frontier analysis. J Econometrics 1990;46:7e38.

[24] Malana NM, Malano HM. Benchmarking productive efficiency of selectedwheat areas in Pakistan and India e data envelopment analysis. Irrig Drain2006;55:383e94.

[25] Fraser I, Cordina D. An application of data envelopment analysis toirrigated dairy farms in Northern Victoria, Australia. Agric Syst 1999;59:267e82.

[26] Jaforullah M, Whiteman J. Scale efficiency in the New Zealand dairy industry:a non-parametric approach. Aust J Agr Resour Ec 1999;43:523e41.

[27] Yilmaz I, Akcaoz H, Ozkan B. An analysis of energy use and input costs forcotton production in Turkey. Renew Energy 2005;30:145e55.

[28] Ozkan B, Kurklu A, Akcaoz H. An inputeoutput energy analysis in greenhousevegetable production: a case study for Antalya region of Turkey. BiomassBioenergy 2004;26:189e95.

[29] De D, Singh RS, Chandra H. Technological impact on energy consumption inrainfed soybean cultivation in Madhya Pradesh. Appl Energy 2001;70:193e213.