Beruflich Dokumente
Kultur Dokumente
i 1
,
`i 1 is a convexity constraint which specifies the VRS framework
(Mostafa, 2009,Heidari, et al., 2012). Without this convexity constraint, the BCC model
will be a CCR model (Heidari, et al., 2012) like in Eq. (4).
3) Scale Efficiency
Scale efficiency measures can be obtained for each terminal by conducting both
CCR and BCC DEA models, then decomposing the TE score obtained from the CCR
DEA into two components. One component is due to scale inefficiency, and the other
is due to Pure Technical inefficiency. If there is a diffrence in the CCR and BCC
scores for a particular terminal, this indicates that the firm has scale inefficiency
(Timothy, et al., 2005), Eq. (3)
Eq. (3) can be also defined by:
(6)
Journal of Korea Trade Vol. 17, No. 4, November. 2013
58
4) Super Efficiency
The CCR and BCC models dichotomize container ports into inefficient and efficient
units. However, it is not possible to differentiate between the efficient terminals, since
all efficient units receive the same efficiency score of one (1). For example in our
case, we can differentiate between the most inefficient terminals and the least inefficient
terminals according to their location from the efficiency frontier, but we cannot
differentiate between the most efficient terminals and the least efficient terminals, since
all of them are located on the efficient frontier
5
.
To overcome this limitation, Andersen and Petersen proposed the SuperEfficiency
ranking method for only efficient DMUs. The SuperEfficiency measures how much
inputs can be increased (or the outputs decreased) while not becoming inefficient (So,
et al., 2007).
The superefficiency model is identical to the DEA model previously described, but
a DMU under evaluation (k) is excluded from the reference set. The formulation for
the superefficient model follows Equation (4), but is evaluated without unit k (for
i=1,...,n, i k). For an efficient unit, its exclusion from the reference set will alter
the frontier and allow the unit to be located above the efficient frontier that is to be
superefficient.
. Results and Discussion
1. Results with CRS
As a first step, we used DEA with Constant Returns to Scale (CCR). The results
are as shown in <Figure1>.
<Figure 1> Efficiency Scores under CRS
Efficiency Graph
Hyundai 83.9 % 84%
Hanjin Shipping 75.6 % 76%
PNIT 100 % 100%
PNC 86 % 86%
BNCT 76.6 % 77%
SBTC 52.2 % 52%
KBCT 73.8 % 74%
DPCT 100 % 100%
UTC 68.4 % 68%
Hutchison 100 % 100%
: Efficient * : Weak Efficient
5
The efficient frontier is formed by all (efficient) units that have score 1.
Measuring Energy Use Efficiencyof Container Terminals
59
The results show that the EE scores varied from 0.52 for the most inefficient to
1 for the most efficient
6
, with an overall average EE of 81.65%. Indicating that, on
the average, the terminals could minimize their energy consumption by 18.35% to be
100% efficient.
Among the ten terminals, only three are relatively efficient (PNIT, DPCT,
Hutchinson), while the seven other terminals are inefficient, with two terminals
representing scores above the average, and five terminals having scores below the
average. As a first deduction, we can conclude that in Busans ports, 70% of the
terminals represent waste in energy use. <Table3> shows the possible reduction in
energy consumption for each inefficient terminal.
In order to improve the practices in terms of energy use, for each inefficient terminal,
an efficient terminal or a set of efficient terminals (Benchmarks) with a corresponding
intensity () are selected
7
, and used as efficiency improvement reference(s), as shown
in <Table4>.
<Table 3> Possible Reduction in Energy Consumption for Inefficient Units
Total Energy Use "GJ"
Hyundai 236951.88 to 198706.218
Hanjin Shipping 302380.77 to 228731.528
PNIT 134690.7 to 134690.7
PNC 368021.02 to 316567.149
BNCT 99351.45 to 76097.847
SBTC 167565.75 to 87490.607
KBCT 274827.93 to 202896.137
DPCT 105875.48 to 105875.48
UTC 76970.33 to 52668.798
Hutchison 128543.66 to 128543.66
<Table 4> Inefficient Terminals & Corresponding Benchmarks
PNIT DPCT Hutchison
Hyundai 0.108 0 1.433
Hanjin Shipping 0 0.201 1.614
PNIT 1 0 0
PNC 0 0.106 2.376
BNCT 0 0 0.592
SBTC 0 0.558 0.221
KBCT 0 1.151 0.63
DPCT 0 1 0
UTC 0 0.162 0.276
Hutchison 0 0 1
6
Score 1 means 100% efficient, and score 0.52 means efficient at 52% or a lack of 48%
in efficiency.
7
enchmarks are selected on the basis of their comparable inputs and outputs to the inefficient
DMUs.
Journal of Korea Trade Vol. 17, No. 4, November. 2013
60
To better illustrate our results, lets expose for instance the case of the UTC terminal
from North Port that represents an inefficient score of 68.4% <Figure1>. To reach
a score of 100%
8
, this terminal should reduce its energy consumption by 31.6% which
corresponds to 24,301.532 GJ <Table 3>, while keeping operating the same amount
of equipment, and the same handling throughput. To find an appropriate improvement
pattern to that, UTC terminal has a reference composite DMUs formed by the two
efficient terminals, DPCT (0.162), and Hutchison (0.276)
9
. While the values between
brackets indicate the respective weights () of each reference terminal, that show their
respective contribution amounts in evaluating the UTC terminal as inefficient <Table
4>, i.e., UTC Terminals decision makers may establish their energy consumption policy
based on the practices performed by the two Terminals, DPCT and Hutchison, that
are evaluated to be the most suitable references, in terms of energy use efficiency.
It allows reducing about 31.6% of its total energy use as compared to the consumption
of the year 2012, while using the same amount of inputs to produce the same amount
of outputs.
2. Results with VRS
As a second step, in order to identify whether the origin of the waste in energy
is due to the PTE (managerial deficit), SE (operating at an incorrect RTS), or both,
we used a BCC model to evaluate the PTE and SE of each terminal. The results are
as shown in <Table 5>.
<Table 5> Scores under the BCC Model
CCR BCC Scale
Hyundai 83.9 % 92.1 % 91 %
Hanjin Shipping 75.6 % 85 % 89 %
PNIT 100 % 100 % 100 %
PNC 86 % 100 % 86 %
BNCT 76.6 % 97.5 % 78.6 %
SBTC 52.2 % 58 % 90.1 %
KBCT 73.8 % 84.7 % 87.2 %
DPCT 100 % 100 % 100 %
UTC 68.4 % 100 % 68.4 %
Hutchison 100 % 100 % 100 %
The decomposition of the overall technical efficiency into its components, PTE and
SE, shows that the overall inefficient terminals also have Pure Technical Efficiency
8
100% is score 1 expressed in a percentage.
9
The UTC terminal is closer to the efficient frontier segment formed by these two terminals
on the efficient frontier.
Measuring Energy Use Efficiencyof Container Terminals
61
representing a constant return to scale, and among the terminals estimated inefficient
under CCR model, two terminals have Pure Technical Efficiency (UTC, and PNC),
indicating that the origin of their energy waste is due to unfavorable returns to scale
conditions, and the other terminals represent both scale and Pure Technical inefficiencies.
But two of them (Hyundai, and BNCT) have their PTE score bigger than the SE score.
This suggests that their overall inefficiency is mostly due to operating at incorrect
scale, while the other three terminals (Hanjin Shipping, KBCT, and SBTC) have their
SE score bigger than their PTE score which indicates that the most dominant reason
for the overall inefficiency is probably due to managerial failure. For example, in their
initiatives to improve energy efficiency, decision makers of inefficient units that
represent a higher SE score than PTE score, should concentrate their efforts on
developing a better use of inputs to reduce their energy consumption. On the other
hand, for those representing a PTE score that is higher than their SE score, decision
makers should concentrate their actions on upgrading their inputs themselves to reach
an optimal return to scale and reduce their energy consumption.
To better illustrate our results under VRS, lets expose, for instance, the case of
the BNCT terminal from the New Port, which represents both Pure Technical and
Scale inefficiencies with the respective scores of 97.5 %, and 78.6 %. Even though
the most dominant reason of its overall inefficiency is probably SE
10
, we cannot omit
the contribution of technical factors in the waste of energy especially when the difference
between the two scores is small.
As the first step, we checked the container yard side, where we could see that all
the terminals in the New Port use a similar mix of inputs in terms of technology, capacity,
and type of vehicles. Thus, we can assume that on the yard side, if there is any waste
in energy by BNCT, it is rather due to an inefficient management. Effectively if we
compare the consumption of diesel used for 2012 at the New Port, we can see that
the consumption of diesel by Liter/TEU at BNCT is evaluated at 163% more than the
average consumption made by its pair of terminals in the New Port <Table 6>.
<Table 6> NP Diesel Consumption in 2012
DMUs
Variables
2012 Diesel use
"Liter"
2012 Throughput
"TEU"
Diesel Use
"Liter/TEU"
Average
Liter/TEU
Ratio
Hyundai 2,319,757 2,078,010 1.12
1.20 1
Hanjin Shipping 3,161,931 2,432,255 1.30
PNIT 1,353,687 1,220,000 1.11
PNC 4,260,000 3,353,330 1.27
BNCT 994,093 506,526 1.96 1.96 1.63
10
Because SE score is smaller than PTE score.
Journal of Korea Trade Vol. 17, No. 4, November. 2013
62
After adjusting its yard operation management and energy use policy, at the beginning
of 2013, BNCT could reduce its diesel fuel consumption by about 37% to an average
of 1.24Liter/TEU instead of an average consumption of 1.96 Liter/TEU in the year
of 2012 <Table 7>. It may confirm the assumption that, at the yard level, the failure
in energy use is mostly due to technical failure.
<Table 7> BNCT Average Diesel Consumption in 2012 & 2013
DMUs
Variables
2012 Diesel
Use
"Liter"
2012
Throughput
"TEU"
2013
(Jan,Feb)
Diesel Use
"Liter"
2013
(Jan,Feb)
Throughput
"TEU"
2012 Diesel
Use
"Liter/TEU"
2013
(Jan,Feb)
Diesel Use
"Liter/TEU"
BNCT 994,093 506,526 222,441 179,920 1.96 1.24
At the quay side, managerial performance cannot have a big influence on the technical
performance of the STS cranes since all kinds of cranes are operated by the same
way within a limited amplitude of movement. Thus, we can suppose that energy waste
is caused by operating at an incorrect return to scale conditions, probably due to a
technological disadvantage.
In general for container terminals, most yards equipment represents big similarities,
but STS container cranes may vary in type (high/low profile), size (Panamax, Post
panamax, and SuperPost panamax), Safe Working Load (SWL) capacity, volume of
TEUs handled per hour, type of trolleys and spreaders, etc. This variety in equipment
can affect the energy consumption for operating the STS crane. According to the
collected data, we can see that the average electricity amount consumed per TEU by
BNCT to operate its STS cranes is about 178% higher than its pair of terminals in
the New Port <Table 8>.
<Table 8> NP Electricity Consumption in 2012
DMUs
Variables
2012
Electricity
use "Kwh"
2012
Throughput
(TEU)
Kwh/TEU
Average
Kwh/TEU
Ratio
Hyundai 40,882,580 2,078,010 19.67
18.75 1
HanjinShipping 50,003,899 2,432,255 20.56
PNIT 22,987,582 1,220,000 18.84
PNC 56,433,062 3,353,330 16.83
BNCT 16,911,124 506,526 33.39 33.39 1.78
Measuring Energy Use Efficiencyof Container Terminals
63
After checking the type of STS cranes used in the New Port
11
, we found that all
terminals STS cranes represent big similarities in term of size (all of them are Super
Post Panamax) and type (all of them have high profile). However, a main difference
resides in the type of spreaders, where BNCT is the only terminal in New Port that
is equipped with TwinLift spreaders for the integral of its cranes while all other
terminals use Tandemlift spreaders. In fact, operating a TwinLift spreader consumes
less electricity per move than a TandemLift spreader (as mentioned previously).
Nevertheless, it always needs to make two times more moves to handle the same volume
of TEUs which may considerably affect the energy consumption, especially in time
of busy container traffic. Thus we can estimate that BNCT decision makers may be
able to reduce its energy consumption by upgrading the terminals quayside equipment
(e.g., equipping its STS cranes with TandemLift spreaders rather than TwinLift
spreaders).
3. Ranking Analysis by SuperEfficiency Model
In order to obtain a full ranking of the terminals in our sample, we applied the
DEA SuperEfficiency model that allows the efficient units under evaluation to be
excluded from the reference set, and thus be able to get a score that is bigger than
(1). This method will allow us to rank the terminals in the sample under study from
the relatively most efficient to the relatively most inefficient, which allows a full
comparison of the performance of the terminals and identifies the terminal representing
the best practices among the selected sample. <Figure2> represents the result obtained
with the SuperEfficiency model.
<Figure 2> Efficiency Scores by SuperEfficiency
Efficiency Graph
Hyundai 83.9 % 84%
Hanjin Shipping 75.6 % 76%
PNIT 122.7 % 123%
PNC 86 % 86%
BNCT 76.6 % 77%
SBTC 52.2 % 52%
KBCT 73.8 % 74%
DPCT 106.7 % 107%
UTC 68.4 % 68%
Hutchison 142.7 % 143%
: SuperEfficiency *: Weak Efficiency
11
The analysis was limited only to the New Port terminals because the terminals at the
North Port use a different type of T/C.
Journal of Korea Trade Vol. 17, No. 4, November. 2013
64
The result shows that the SuperEfficiency scores of the terminals. Hutchison, DPCT
and PNIT, of which all efficiency indices are equal to 1 under CRS, are respectively
about 1.43, 1.07, and 1.23.
12
Therefore, the Hutchison terminal is evaluated to be the most efficient unit and
both the practice performed and the results obtained by the Hutchinson Terminal in
terms of energy use may be respectively used as a sound basis and realistic target
in establishing a comprehensive energy policy and issuing more rational regulations
by decision makers (e.g., Busan Port Authority) in the ports of Busan.
. Conclusion
Ports have always been considered as important logistics platforms for international
commerce, as they are the unique interface between sea and land transportation modes.
Merchandise trade and seaborne shipments continue to move in tandem (UNCTAD,
Review of Maritime Transport 2012) and any increase in the international commercial
exchanges will have a direct impact on the port industry through the intensification
of port commercial activities. As a consequence, a ports energy consumption will
increase exponentially, because the port industry always reacted to the increase in the
international trade volume by increasing the amount of the ports equipment or using
more advanced technologies that consume more energy, With respect to these
circumstances, in this study the degree of performance in terms of energy use efficiency
of ten terminals located in Busans ports in Korea had been investigated. Our approach
described an indepth application of inputoriented Data Envelopment Analysis that not
only allows the determination of terminals representing bad practices and the origin
of their inefficiency, but also the way to improve their performance using the terminals
representing the best practices as references. It may be useful for the policy makers
to design their policies differently depending on the type of inefficiency that
characterizes each terminal. Also the superefficiency results allow the Port Authority
to establish a more realistic energy policy and effective regulation that permits terminal
operators to reach a maximal energy saving and emissions reduction for both ports,
at general level based on the practices and results of the best performing unit in the
sample.
The results also indicate that, in general for container terminals, there is no dominant
cause of inefficiency and the lack of energy performance can be driven either by
mismanagement of inputs (terminals layout, horizontal transport that connect the
terminal yard to the quay, stacking criteria, etc.) or by unfavorable scale conditions
mostly related to technological issues. It means that size of terminals doesnt have
12
The SuperEfficiency scores of the inefficient container terminals are the same as the
efficiency scores indicated in CCR model.
Measuring Energy Use Efficiencyof Container Terminals
65
a big influence in term of energy efficiency.
Finally within the framework of our sample on an average, a policy designed in
order to induce Busan Port terminals to move towards less intensiveenergy terminals
will save up to 18% of energy, resulting not only in energy cost reduction, but also
in generating less emissions from the port sector, and consolidating sustainability by
using more environmentally friendly practices.
. Study limitation and Future research
The DEA method allows researchers the ability to distinguish the efficient units
from the inefficient ones. It also can designate a benchmark(s) for each inefficient
unit in order to get the best use of its inputs. Furthermore, it dichotomizes the technical
inefficiency into puretechnical and scale inefficiencies, allowing us to determine
whether the lack in efficiency is due to managerial underperformance or related to
technological issues. It is the ultimate limit of the DEA method, i.e., DEA doesnt
precisely indicate among the various factors affecting the PTE which one needs to
be focused on as waste generators, or which part of the inputs is representing a
technological disadvantage that might lead to scale inefficiency and need to be upgraded.
For the above mentioned reason, based on DEA results, future research using
regression analysis needs to be conducted to precisely determine the respective causes
of energy waste and to reach a more accurate analysis and provide substantial
recommendations for strategy planning in terms of energy use, for each inefficient
terminal.
References
Button, K., (1993), Transport, the Environment and Economic Policy, Aldershot, Hants,
England and Brookfield, Vermont, USA: Edward Elgar Publishing Ltd.
Cannon J. S., (2009), Container Ports and Air Pollution, Presented at the Green Port
2009 Conference.
Cannon J. S., (2008), U.S. Container Ports and Air Pollution: A Perfect Storm, An
Energy Futures, Inc. Study.
Charnes, A., Cooper, W.W., Rhodes, E., (1978), Measuring the efficiency of decision
making units. European Journal of Operational Research, Vol. 2, pp. 429444.
Chen Fu Chien, Feng Yu Lo, and J. T. Lin, (2003), Using DEA to Measure the Relative
Efficiency of the Service Center and Improve Operation Efficiency through
Reorganization, IEEE Transactions on Power Systems, Vol. 18, No. 1, pp. 366373.
Chiu, Y. and M.F. Wu, (2010), Environmental Efficiency Evaluation in China:
Application of Undesirable Data Envelopment Analysis, Polish J. of Environ.
Journal of Korea Trade Vol. 17, No. 4, November. 2013
66
Study, Vol. 19, No. 6, pp. 11591169.
Cooper, W.W., Seiford, L.M., and Tone, K., (2006), Introduction to DEA and its uses
with DEASolver software and references, New York: Springer.
Hart, S. (1995), A NaturalResourcebased view of the firm, Academy 01 Management
Review, Vol. 20, No. 4, pp. 9861014.
Hecht J., (1997), The Environmental Effects of Freight, Paris: Organization for Economic
Cooperation and development.
Kang W.S. and Cheong J.P., (2011), Estimating TransportationRelated Greenhouse
Gas Emissions in the Port of Busan, S. Korea, Asian Journal of Atmospheric
Environment, Vol. 51, pp. 4146.
Lee, W.S. (2008), Benchmarking the energy efficiency of government buildings with
data envelopment analysis, Energy and Buildings, Vol. 40, pp. 891895.
Lianga Q.M, Fana Y, and Wei Y.M, (2007), Multiregional inputoutput model for
regional energy requirements and CO
2
emissions in China, Energy Policy, Vol.
35, pp. 16851700.
Linares P., and X. Labandeira, (2010), Energy Efficiency: Economics and Policy,
Journal of Economic Surveys, Vol. 24, No. 3, pp. 573592.
Magali D. and M.W. Toffel, (2004), Stakeholders and Environmental Management
Practices: An Institutional, Business Strategy and the Environment, Bus. Strat.
Env. 13, pp. 209222.
Merk O., C. Ducruet, P. Dubarle, E. Haezendonck and M. Dooms, (2011), The
Competitiveness of Global PortCities: The Case of the Seine Axis (Le Havre,
Rouen, Paris, Caen) France, OECD Regional Development Working Papers.
Mohammad D.H., M. Omid and A. Mohammadi, (2012), Measuring productive
efficiency of horticultural greenhouses in Iran: A data envelopment analysis
approach, Expert Systems with Applications, Vol. 39, pp. 10401045.
Monjezi N., M. Javad Sheikhdavoodi and M. Taki, (2011), Energy Use Pattern and
Optimization of Energy Consumption for Greenhouse Cucumber Production in Iran
Using Data Envelopment Analysis (DEA), Modern Applied Science, Vol. 5, No.
6.
Nassiri, S.M. and Singh, S. (2009). Study on energy use performance for paddy crop
using data envelopment analysis (DEA) technique, Applied Energy, Vol. 86, No.
78, pp. 13201325.
Onut, S. and S. Soner, (2006), Energy efficiency assessment for the Antalya Region
hotels in Turkey, Energy and Buildings, Vol. 38, pp. 964971.
Partnerships in Environmental Management for the Seas of East Asia (PEMSEA),
(2007), Port Safety and Health and Environmental Management Code (PSHEM
Code).
Patterson, M., (1996), What is energy efficiency?, Concepts, issues and
methodological indicators, Energy Policy, Vol. 24, No. 5, pp. 377390.
Porter, M.E., Van Der Linde C., (1995), Green and Competitive: Ending the stalemate,
Measuring Energy Use Efficiencyof Container Terminals
67
Harvard Business Review.
So, S. H. and J.J. Kim, Geon Cho and Do Kwan Kim, (2007), Efficiency Analysis
and Ranking of Major Container Ports in Northeast Asia: An Application of Data
Envelopment Analysis, International Review of Business Research Papers, Vol.
3 No. 2, pp. 486 503.
The port of Long Beach and The port of Los Angeles, (2006),San Pedro Bay Ports
Clean Air Action Plan Overview.
United Nations Conference on Trade and Development (UNCTAD), (2012), Review
of Maritime Transport.
Zhou P., B.W. Ang, and K.L. Poh, (2008), A survey of data envelopment analysis
in energy and environmental studies, European Journal of Operational Research,
Vol. 189, pp. 118.
Zhou, P. and Ang, B.W., (2008), Decomposition of aggregate CO
2
emissions: a
production theoretical approach, Economics, Vol. 30, pp. 10541067.
<Appendix> Initial Terminals DATA for the year 2012
DMUs
Variables
Throughput
"TEU"
Diesel
"liter"
Electricity
"kwh"
N STS/C
Units
N T/C
Units
N Yard
Vehicles
Units
STS Total
Box/Lift
"TEU"
Total Yard
Equipment
"Units"
Hyundai 2,078,010 2319757 40882580 11 38 97 44 135
Hanjin
Shipping
2,432,255 3161931 50003899 12 42 109 48 151
PNIT 1,220,000 1342000 22987582 9 28 57 36 85
PNC 3353330 4260000 56433062 17 58 146 68 204
BNCT 506526 994093 16911124 8 38 36 16 74
SBTC 966341 3279082 11295910 7 21 40 14 61
KBCT 2230306 4410068 28932862 15 42 107 30 149
DPCT 1193690 1517623 13095408 7 19 42 14 61
UTC 568753 1479095 5480376 5 13 29 10 42
Hutchison 1358431 1810844 16240000 14 33 92 28 125