You are on page 1of 11

640 Int. J. Shipping and Transport Logistics, Vol. 9, No.

5, 2017

Optimal design of container terminal gate layout

Chu Cong Minh*


Department of Bridge and Highway Engineering,
Ho Chi Minh City University of Technology,
268 Ly Thuong Kiet Street, District 10,
Ho Chi Minh City, Vietnam
Email: ccminh@hcmut.edu.vn
*Corresponding author

Nathan Huynh
Department of Civil and Environmental Engineering,
University of South Carolina,
300 Main Street, Columbia, SC 29208, USA
Email: huynhn@cec.sc.edu

Abstract: Over the last 50 years, international trade of goods has grown
significantly in the USA, and as a result, US ports have become bottlenecks in
the freight supply chain and logistics. A particular issue that most US ports are
contending with is gate congestion (i.e., queuing of trucks outside the container
terminal gate). This research provides planners and engineers of container
terminals a set of methodologies to design an optimal gate layout to reduce gate
congestion. The use of the methodologies to determine the optimal number of
service gates is illustrated considering various truck arrival rates, gate service
rates, and waiting time thresholds.

Keywords: queuing model; pooled queues; gate layout; optimisation; queuing


time.

Reference to this paper should be made as follows: Minh, C.C. and Huynh, N.
(2017) ‘Optimal design of container terminal gate layout’, Int. J. Shipping and
Transport Logistics, Vol. 9, No. 5, pp.640–650.

Biographical notes: Chu Cong Minh is an Associate Professor at the Ho Chi


Minh City University of Technology, Vietnam. He received his PhD in
Transportation Engineering from the Nagaoka University of Technology,
Japan. His research interests include terminal design and operations, and
transportation planning and management.

Nathan Huynh is an Associate Professor at the University of South Carolina.


He received his PhD and MS in Transportation Engineering from the
University of Texas at Austin. His research interests include intermodal freight
terminal design and operations, intermodal freight transport, and freight
transportation planning and logistics.

Copyright © 2017 Inderscience Enterprises Ltd.


Optimal design of container terminal gate layout 641

1 Introduction

Over the last 50 years, international trade of manufactured goods has grown significantly,
and as a result, the global supply chains and the underlying support infrastructure,
particularly ports, have been strained. Due to the fact that ports have to operate at or near
capacity, its ability to withstand ‘shocks’ (in the form of high number of trucks arriving
to drop off export containers or pick up import containers) is greatly diminished. The
resultant congestion at the terminal gate can have deleterious impacts on air quality due
to increased idling from drayage trucks. Also, excessive truck delay leads to a loss of
profitability for truckers and can eventually impact consumer prices.
Consider the typical gate layout at US marine container terminals (Figure 1), once the
trucks arrive at the entry gate they must stop to present transaction information, be
inspected, and then be directed to a specific location in the container yard. When the
truck arrival rate exceeds that of the service gate, long queues of trucks will begin to form
across multiple lanes. Another contributing factor to long queues of trucks is trouble
transactions; trouble transactions are generally a result of documentation, container
location, or equipment issues and constitute approximately 5% of all transactions. In the
typical gate layout (will be referred to as non-pooled queuing strategy), when a truck
being serviced has a trouble transaction, all the vehicles in the lane directly behind the
‘trouble’ vehicle must wait for the trouble transaction to clear; trouble transactions take
much longer to complete than a normal transaction. Thus, a trouble vehicle will hold up
all trucks waiting behind. This phenomenon can lead to situations of ‘socially unjust’
queuing where a truck arriving earlier receive service later because it was stuck behind
the trouble vehicle.

Figure 1 Satellite image of the entry gate layout of APM Terminal, Port Elizabeth, New Jersey
and USA (see online version for colours)

Source: https://www.google.com/maps/@40.6695181,74.1602534,209m/
data=!3m1!1e3
Now consider the scenario where all trucks upon arrival are put into a single queue and
trucks at the front of the queue are served by the next available gate associates (Figure 2).
642 C.C. Minh and N. Huynh

This queuing setup is often employed at places like banks, airports, and theme parks.
Using this gate setup, the trouble vehicles no longer prevent vehicles behind it from
receiving service.

Figure 2 Pooled queuing strategy (see online version for colours)

This study seeks to provide a methodology for terminator operators and port planners to
design a gate layout that will reduce truck queueing time and gate operation costs. The
chief advantage of the developed methodology is that it provides practitioners with a
more accurate approach than the state-of-the-practice which assumes deterministic truck
arrival rate and gate service rate. The developed methodology can be used for design of a
new terminal gate or expansion of an existing terminal gate. Specifically, the developed
methodology can be used to:
1 evaluate the performance of different truck queuing strategies
(pooled vs. non-pooled)
2 determine the number of gates necessary to achieve average truck waiting time for a
given truck arrival rate and gate service rate.
The remainder of this paper is organised as follows. The literature review section
summarises past studies that are related to this work. The methodology section presents
the queuing models and the gate optimisation model. Numerical results and sensitivity
analyses are presented in the model application section. Lastly, the conclusion section
provides concluding remarks and recommendation for further studies.

2 Literature review

A number of studies have addressed the truck queuing problem at the inbound gates of
marine container terminals. However, these studies have focused mainly on operational
strategies to reduce truck queuing time, such as the use of a truck appointment system,
extended gate hours, or truck arrival management (e.g., Giuliano and O’Brien, 2007;
Chen et al., 2011, 2013a; Zhang et al., 2013; Chen et al., 2013b). This study deals with
the planning and design of gate layout to reduce truck queuing time. The three studies
most related to this work are discussed in more detail below.
Guan and Liu (2009a, 2009b) developed a model to measure costs of congestion at
the gates, provided alternatives to improve gate operation and investigated ways to
Optimal design of container terminal gate layout 643

reduce gate congestion at the Port of New York. A multi-server queuing model was used
to analyse congestion at the gates and to estimate truck waiting cost. The authors applied
the M/Ek/n model given that the truck inter-arrival times followed the exponential
distribution and the service times followed the Erlang distribution. The proposed
optimisation model was to minimise the total gate system cost, which considered both the
cost to trucks (waiting time) and cost to the terminal operator (number of service gates to
operate).
Fleming et al. (2013) developed an agent-based simulation model to study the
terminal gate system with two different queuing strategies, a pooled queue and
non-pooled queues. Their simulation model overcomes a key shortcoming of queuing
models – vehicle movements within the queue. Using a car-following model, a realistic
representation of how trucks move within the queue was captured. The developed
simulation model was used to evaluate queuing strategies under various operational
conditions. Results indicated that using a pooled queue yielded significantly lower
average truck queuing times and variability in queuing times.
Minh and Huynh (2014) developed a planning-level tool that can be used by design
engineers, terminal operators, port authorities, and transportation planners to assess the
effectiveness of different gate layout as well as determine the optimal layout for marine
container terminals. Specifically, their analytical tool can be used to determine the
average truck queuing time for a given gate configuration or determine how many service
gates and queuing lanes are needed to achieve a desired level of service for a given truck
arrival rate and truck service rate. This tool accounts for the non-deterministic nature of
truck arrivals and service times. Specifically, it accounts for the fact that trucks arrive as
a Poisson stream, will be served by one of n gates and that the gate service times are
independent and identically distributed random variables.
This study expands the work of Guan and Liu (2009a, 2009b) by providing design
engineers with a methodology to
1 assess the trade-offs between non-pooled and pooled queuing strategies
2 determine the minimum number of gates needed to ensure that the average queuing
time does not exceed a specified threshold.
In contrast to the work by Fleming et al. (2013) which used simulation, the developed
methodology is analytical and thus much simpler to apply. A key difference between this
work and that of Minh and Huynh (2014) is that this work assumes the gate service time
of trucks follows the Erlang distribution (Huynh et al., 2011; Guan and Liu, 2009a,
2009b) instead of a general distribution. Thus, this study uses the M/Ek/n queuing model
instead of the M/G/n queuing model previously proposed by the authors. The advantage
of using the M/Ek/n queuing model is that it provides a simple analytical formula as
opposed to the more complicated M/G/n diffusion approximation formulas.

3 Methodology

This section describes the queuing models that can be used to analyse the truck queuing
situation at container terminal gates. These models assume the following.
1 there is no truck balking after arrival
644 C.C. Minh and N. Huynh

2 there are no obstacles in the queue-to-gate paths


3 arriving trucks will join the shortest queue
4 the utilisation factor (traffic intensity/number of gates) is less than one.

3.1 Non-pooled queues


If truck arrivals follow a Poisson process and the gate service times are Erlang distributed
(Ek), with k as the shape parameter, the M/Ek/1 model can be used to analyse the
non-pooled queuing situation. If is the truck arrival rate (trucks/min) to the terminal and
n is the number of gates, then the truck arrival rate for each lane is /n (trucks/min). The
average truck queuing time can be calculated using the following equation (Gross et al.,
2008):
1+ k /n
QTM / Ek /1 = × (1)
2k ( − / n)

where,
mean of gate service rate of trucks (trucks/min)
mean of truck arrival rate (trucks/min)
k shape parameter of the Erlang distribution
n number of gates.

3.2 Pooled queues


In the pooled queuing situation, the gate will have n gates working in parallel to serve a
single queue of trucks. Assuming that truck arrivals are Poisson distributed and the gate
service times are Erlang distributed, the M/Ek/n model can be used to analyse the pooled
queuing situation. The following approximation formula can be used to calculate the
average truck queuing time with a relative percentage error less than 2% (Cosmetatos,
1976):
⎡1 + v 2 (1 + v 2 ) (1 − ρ)(r − 1) ( 4 + 5r − 2 ) ⎤
QTM / Ek / n ≈ ⎢ ⎥ × QTM / M / n (2)
⎣⎢ 2 ⎦⎥
+
16 ρr

QTM/M/n corresponds to the M/M/n queuing system and can be calculated by using the
following formula:


⎡ n −1 a i ⎤
−1

⎢ ⎥
an an
(n − 1)!(n − a) ⎢⎣ i = 0 i ! (n − 1)!(n − a) ⎦⎥
QTM / M / n ≈ 2
× + (3)

where,
a = /μ

v = var( Ek ) / coefficient of variation of gate service rate, 0 < v < 1.


Optimal design of container terminal gate layout 645

ρ = /n < 1 traffic intensity or server utilisation


The gate service times of trucks are assumed to follow the Erlang distribution with shape
(k) and rate parameters (θ). The mean and the variance of the distribution are k/θ and
k/θ2, respectively. The coefficient of variation of gate service times of trucks, v, is:

var ( Ek ) k θ 1
v= = × = (4)
θ2 k k

Finally, the average truck queuing time is calculated as


⎡ n −1 a i ⎤
−1

⎢ ⎥
an an
(n − 1)!(n − a ) ⎣⎢ i = 0 i ! (n − 1)!(n − a ) ⎦⎥
QTM / Ek / n ≈ 2
× +

⎡ 1 ⎛ 1⎞ ⎤
⎢1 + ⎜ 1 − ⎟ (1 − ρ)(n − 1) ( 4 + 5n − 2 ) ⎥
(5)
×⎢ k + ⎝ ⎠ ⎥
k
⎢ 2 16 ρn ⎥
⎢ ⎥
⎣ ⎦

3.3 Gate optimisation model


The required number of service gates needed under the pooled queuing situation to meet
a specified waiting time threshold (T in minutes) can be determined by solving the
following mathematical program

minimize n (8)

subject to


⎡ n −1 a i ⎤
−1

⎢ ⎥
an an
(n − 1)!(n − a) ⎢⎣ i = 0 i ! (n − 1)!(n − a) ⎥⎦
QTM / Ek / n ≈ 2
× +

⎡ 1 ⎛ 1⎞ ⎤
⎢1 + ⎜1 − ⎟ (1 − ρ)(n − 1) ( 4 + 5n − 2 ) ⎥
(9)
×⎢ k + ⎝
k⎠
⎥ ≤T
⎢ 2 16 ρn ⎥
⎢ ⎥
⎣ ⎦

, , a , ρ, T ≥ 0 (10)

The stated mathematical program can be solved for by simply evaluating different values
of n (from low to high). The optimal value of n is the one that satisfies the expression:

n* = arg min{n : QT (n) ≤ T }.


n
646 C.C. Minh and N. Huynh

4 Model application

The models presented above are applied using data collected from one of the container
terminals on the US East Coast. Huynh et al. (2011) used the provided terminal webcam
to obtain the truck arrival rate and gate service time of trucks. It was found that the truck
arrival rate follows the Poisson distribution with mean = 1.4096 trucks/min, and the
gate service time of trucks follows the Erlang distribution with shape parameter k = 5 and
rate parameter θ = 0.9112. That is, the mean ( ) of gate service rate of trucks at this
terminal is 0.18 trucks/min.

4.1 Non-pooled queues vs. pooled queues


To assess the performance of non-pooled queuing strategy versus pooled queuing
strategy, the terminal layout, truck arrival data, and gate service data collected by Huynh
et al. (2010) are used. This particular terminal has 10 service gates, and as stated, the gate
service rate follows the Erlang distribution with mean = 0.18 trucks/min. For this
analysis, the mean truck arrival rate ( ) is varied from 0.1 to 1.8 trucks/min. As shown in
Figure 3, the average truck queuing time of the pooled queuing strategy is always lower
than that of non-pooled queuing strategy. This finding is consistent with the results
reported in Minh and Huynh (2014).

Figure 3 Comparison of average queuing times between pooled and non-pooled queuing strategy
300

Pooled
250 Non-pooled
Average Waiting time (min)

200

150

100

50

0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8

Arrival Rate (trucks/min)

4.2 Gate optimisation


To examine the effect of truck arrival rate and time threshold on the number of service
gates and average queuing time, we used a mean gate service rate of 0.18 trucks/min, and
the mean truck arrival rate is varied between 0.1 and 2.0 trucks/min. The time threshold,
T, is assumed to range between five and 30 min in five-min increments. Table 1 shows
the relationships between truck arrival rate, number of service gates, and average truck
queuing time.
It is clear that the level of n* is dependent on the levels of T and for the pooled
queuing system. With a specified time threshold, increasing arrival rate may increase the
Optimal design of container terminal gate layout 647

number of gates accordingly, but it is not always true. For example, in case of the time
threshold is 15 min, when the truck arrival rate increases from 0.6 to 0.7 trucks/min, the
number of gates increases from four to five gates. However, when the truck arrival rate
increases from 0.7 to 0.8 trucks/min, the number of gates remains the same (five gates).
This is because when the truck arrival rate increases (from 0.7 to 0.8 trucks/min), the
queuing time also increases (from 1.55 min to 4.55 min) but it is still less than the waiting
time threshold (15 min). When the truck arrival rate reaches 0.9 trucks/min, if the number
of gates remains at five, then the server utilisation would reach one ( /(n ) = 0.9/
(5 * 0.18)) and the queue would grow to infinity. Therefore, the number of gates needs to
increase by one to six, and therefore reduces the average queuing time to 2.4 minutes.
Table 1 Effect of truck arrival rate and waiting time threshold on optimal number of service
gates and average queuing time

Waiting time threshold (min)


Truck
arrival rate 5 10 15 20 25 30
(trucks/min)
No. AQT No. AQT No. AQT No. AQT No. AQT No. AQT
0.1 1 4.17 1 4.17 1 4.17 1 4.17 1 4.17 1 4.17
0.2 2 1.57 2 1.57 2 1.57 2 1.57 2 1.57 2 1.57
0.3 3 0.83 2 7.69 2 7.69 2 7.69 2 7.69 2 7.69
0.4 3 2.48 3 2.48 3 2.48 3 2.48 3 2.48 3 2.48
0.5 4 1.16 4 1.16 3 13.09 3 13.09 3 13.09 3 13.09
0.6 4 3.41 4 3.41 4 3.41 4 3.41 4 3.41 4 3.41
0.7 5 1.55 5 1.55 5 1.55 5 1.55 5 1.55 4 28.32
0.8 5 4.55 5 4.55 5 4.55 5 4.55 5 4.55 5 4.55
0.9 6 2.06 6 2.06 6 2.06 6 2.06 6 2.06 6 2.06
1.0 7 1.18 6 6.16 6 6.16 6 6.16 6 6.16 6 6.16
1.1 7 2.55 7 2.55 7 2.55 7 2.55 7 2.55 7 2.55
1.2 8 1.42 7 8.74 7 8.74 7 8.74 7 8.74 7 8.74
1.3 8 3.15 8 3.15 8 3.15 8 3.15 8 3.15 8 3.15
1.4 9 1.62 9 1.62 8 13.81 8 13.81 8 13.81 8 13.81
1.5 9 3.91 9 3.91 9 3.91 9 3.91 9 3.91 9 3.91
1.6 10 1.92 10 1.92 10 1.92 10 1.92 10 1.92 9 28.87
1.7 10 4.96 10 4.96 10 4.96 10 4.96 10 4.96 10 4.96
1.8 11 2.37 11 2.37 11 2.37 11 2.37 11 2.37 11 2.37
1.9 12 1.42 11 6.50 11 6.50 11 6.50 11 6.50 11 6.50
2.0 12 2.82 12 2.82 12 2.82 12 2.82 12 2.82 12 2.82
Notes: No. = number of service gates (n ) and AQT = average queuing time (min)
*

If a marine terminal provides a vehicle booking system, that is, truck drivers can make
appointments by telephone or via the internet before they arrive at the terminal, the gate
service time at the terminal can be reduced. On the contrary, if trucks arrive at the
terminal without sufficient and appropriate transaction documents, the overall gate
service time will be increased. To understand the impacts of these two scenarios, the
relationships between gate service rates ( ), number of service gates (n*) and time
648 C.C. Minh and N. Huynh

threshold (T) are examined and the results are shown in Table 2. For this analysis, the
gate service rate is again assumed to be Erlang distributed, with varied between 0.135
and 0.230 trucks/min, and T is varied between five and 30 minutes, in five-minute
increments. The truck arrival rate is assumed to be Poisson distributed with mean
= 1.41 trucks/min.
Table 2 Effect of gate service rate and waiting time threshold on optimal number of service
gates and average queuing time

Waiting time threshold (min)


Gate
service rate 5 10 15 20 25 30
(trucks/min)
No. AQT No. AQT No. AQT No. AQT No. AQT No. AQT
0.135 12 1.61 11 6.63 11 6.63 11 6.63 11 6.63 11 6.63
0.140 11 3.36 11 3.36 11 3.36 11 3.36 11 3.36 11 3.36
0.145 11 1.98 11 1.98 10 13.53 10 13.53 10 13.53 10 13.53
0.150 11 1.34 10 5.39 10 5.39 10 5.39 10 5.39 10 5.39
0.155 10 3.10 10 3.10 10 3.10 10 3.10 10 3.10 10 3.10
0.160 10 1.95 10 1.95 10 1.95 9 18.47 9 18.47 9 18.47
0.165 10 1.38 9 6.75 9 6.75 9 6.75 9 6.75 9 6.75
0.170 9 3.84 9 3.84 9 3.84 9 3.84 9 3.84 9 3.84
0.175 9 2.54 9 2.54 9 2.54 9 2.54 9 2.54 9 2.54
0.180 9 1.73 9 1.73 9 1.73 8 18.54 8 18.54 8 18.54
0.185 9 1.30 8 7.38 8 7.38 8 7.38 8 7.38 8 7.38
0.190 8 4.34 8 4.34 8 4.34 8 4.34 8 4.34 8 4.34
0.195 8 2.94 8 2.94 8 2.94 8 2.94 8 2.94 8 2.94
0.200 8 2.14 8 2.14 8 2.14 8 2.14 8 2.14 8 2.14
0.205 8 1.56 8 1.56 8 1.56 8 1.56 7 22.50 7 22.50
0.210 8 1.23 7 8.85 7 8.85 7 8.85 7 8.85 7 8.85
0.215 8 0.99 7 5.25 7 5.25 7 5.25 7 5.25 7 5.25
0.220 7 3.59 7 3.59 7 3.59 7 3.59 7 3.59 7 3.59
0.225 7 2.66 7 2.66 7 2.66 7 2.66 7 2.66 7 2.66
0.230 7 2.06 7 2.06 7 2.06 7 2.06 7 2.06 7 2.06
Notes: No. = number of gates (n ) and AQT = average queuing time (min)
*

It is clear that the level of n* is dependent on the levels of T and for this system. With a
specified waiting time threshold, increasing the gate service rate may decrease the
number of gates accordingly, but again, it is not always true. In the case of 15 minutes
waiting time threshold, if the gate service rate increases from 0.140 to 0.145 trucks/min,
the number of gates reduces from 11 to ten gates. In this instance, the average queuing
time increases from 3.36 min to 13.53 min; note that this waiting is acceptable since it is
still less than the waiting time threshold of 15 minutes. However, if the gate service rate
increases to 0.15 trucks/min, and the number of gates needs to remain at ten gates and not
reduce to nine. This is because if the number of gates reduces to nine, the server
utilisation would exceed one ( /n = 1.41/(9*0.15) = 1.04), and the queue would grow to
Optimal design of container terminal gate layout 649

infinity. The number of gates, therefore, needs to remain the same as before (ten gates),
and thus the increase service rate reduces the average queuing time to 5.39 minutes.
Figure 4 shows the relationship between the waiting time threshold and the number of
gates. When the average truck arrival rate is 1.41 trucks/min and the gate service rate of
trucks is 0.18 trucks/min, the number of gates is a function of the waiting time threshold.
If the waiting time threshold increases, the number of gates may decrease, so the average
queuing time may increase accordingly. However, it is not always true. If the waiting
time threshold increases from five to 15 min, the number of gates remains the same,
i.e., nine gates. This is because if the number of gates is reduced, the average queuing
time will be higher than the specified waiting time threshold. It is noted that if the
permitted waiting time increases from 20 to 30 minutes or more, the number of gates
would remain the same, i.e., eight gates. This is because for the given truck arrival rate
and gate service rate, that is, 1.41 and 0.18 trucks/min, respectively, the minimum
number of gates needs to be n = roundup(a) = eight gates to keep the server utilisation
less than one. In the other words, the minimum number of gates in this case is eight gates,
no matter how high the waiting time threshold is.

Figure 4 The relationship between waiting time threshold and number of gates
10 20
9 18
8 16

Average Waiting Time (min)


7 14
No of gates

6 12
5 10
4 8
3 6
2 4
1 2
0 0
5 10 15 20 25 30

Truck processing time (trucks/min)


Truck waiti g ti e threshold trucks/ i
No of Gates
Average WT

5 Conclusions

This paper developed a methodology for terminal operators and port planners to
1 investigate the possible benefit of using a pooled queuing strategy for inbound trucks
at maritime terminal gates
2 determine the optimal number of service gates for different truck waiting time
thresholds.
Numerical results indicated that the pooled queuing strategy yields lower truck queuing
time than the non-pooled queuing strategy. This strategy can be implemented by terminal
operators and the practice implications are discussed in our previous work (Minh and
650 C.C. Minh and N. Huynh

Huynh, 2014). The developed optimisation model provides a methodology for terminal
operators to reduce gate operation costs. That is, for a specified truck waiting time
threshold, the terminal operators can use the model to determine the minimum number of
gates needed to meet this constraint for different truck arrival rates.
The developed methodology is limited to situations where the truck inter-arrival times
follow the Poisson distribution and the gate service times follow the Erlang distribution.
It does not have the capability to analyse complex gate setup that segregates truck
transactions by type (e.g., export vs. import, full vs. empty, appointment vs.
non-appointment). Given that more and more container terminals in the USA are
adopting the use of truck appointment systems, future work could consider developing
queuing models that accommodate trucks with different priorities.

References
Chen, G., Govindan, K. and Yang, Z. (2013a) ‘Managing truck arrivals with time windows to
alleviate gate congestion at container terminals’, International Journal of Production
Economics, Vol. 141, No. 1, pp.179–188.
Chen, G., Govindan, K., Yang, Z., Choi, T. and Jiang, L. (2013b) ‘Terminal appointment system
design by non-stationary M(t)/Ek/c(t) queueing model and genetic algorithm’, International
Journal of Production Economics, Vol. 146, No. 2, pp.694–703.
Chen, X., Zhou, X. and List, G.F. (2011) ‘Using time-varying tolls to optimize truck arrivals at
ports’, Transportation Research Part E, Vol. 47, No. 6, pp.965–982.
Cosmetatos, G.P. (1976) ‘Some approximate equilibrium results for the queue (M/G/r)’,
Operations Research Quarterly, Vol. 27, No. 3, pp.615–620.
Fleming, M., Huynh, N. and Xie, Y. (2013) ‘An agent-based simulation tool for evaluating pooled
queue performance at marine container terminals’, Transportation Research Record: Journal
of the Transportation Research Board, Vol. 2330, pp.103–112.
Giuliano, G. and O’Brien, T. (2007) ‘Reducing port-related truck emissions: the terminal gate
appointment system at the ports of Los Angeles and Long Beach’, Transportation Research
Part D: Transport and Environment, Vol. 12, No. 7, pp.460–47.
Gross, D., Shortle, J.F., Thompson, J.M. and Harris, C.M. (2008) Fundamentals of Queueing
Theory, 4th ed., John Wiley and Sons, New York.
Guan, C.Q. and Liu, R. (2009a) ‘Container terminal gate appointment system optimization’,
Maritime Economics & Logistics, Vol. 11, No. 4, pp.378–398.
Guan, C.Q. and Liu, R. (2009b) ‘Modeling Gate congestion of marine container terminals, truck
waiting cost, and optimization’, Transportation Research Record: Journal of the
Transportation Research Board, Vol. 2100, pp.58–67.
Huynh, N., Harder, F., Smith, D., Sharif, O. and Pham, Q. (2011) ‘An assessment of truck delays at
seaports using terminal webcams’, Transportation Research Record: Journal of the
Transportation Research Board, Vol. 2222, pp.54–62.
Minh, C. and Huynh, N. (2014) ‘Planning-level tool for assessing and optimizing marine container
terminal gate layout’, Transportation Research Record: Journal of the Transportation
Research Board, Vol. 2409, pp.31–39.
Zhang, X., Zeng, Q. and Chen W. (2013) ‘Optimization model for truck appointment in
container terminals’, 13th COTA International Conference of Transportation Professionals
(CICTP 2013) Procedia - Social and Behavioral Sciences, Vol. 96, pp.1938–1947.

You might also like