analysing the optimal location of a hub port in southeast asia

458 Int. J. Logistics Systems and Management, Vol. 6, No. 4, 2010

Copyright © 2010 Inderscience Enterprises Ltd.

Analysing the optimal location of a hub port in Southeast Asia

Peerapol Boontaveeyuwat Transportation Engineering Program, School of Engineering and Technology, Asian Institute of Technology, P.O. Box 4, Klong Luang, Pathumthani 12120, Thailand E-mail: [email protected]

Shinya Hanaoka* Department of International Development Engineering, Graduate School of Science and Engineering, Tokyo Institute of Technology, 2-12-1-14-12, O-okayama, Meguro-ku,Tokyo, 152-8550, Japan Fax: +81 -3-5734-3468 E-mail: [email protected] *Corresponding author

Abstract: This paper applies an integer programming model on the selection of the Pakbara port as an expecting hub competition among the major ports in Southeast Asia. By solving the integer programming model, several origin/destination container flows in the world are serviced via candidate hub ports such that the total system costs, taking into account port costs (entrance and terminal handling charges) and shipping costs (feeder and mainline) are minimised. The result shows the Port Klang or the Pakbara port as the optimal single-hub solution in several scenarios of structural shifts in traffic flows at the Pakbara port.

Keywords: hub port; hub location; container shipping; integer programming.

Reference to this paper should be made as follows: Boontaveeyuwat, P. and Hanaoka, S. (2010) ‘Analysing the optimal location of a hub port in Southeast Asia’, Int. J. Logistics Systems and Management, Vol. 6, No. 4, pp.458–475.

Biographical notes: Peerapol Boontaveeyuwat is a Doctoral Candidate in Transportation Engineering, School of Engineering and Technology, Asian Institute of Technology (AIT), Thailand. He received his Master of Engineering (Transport System Engineering) from University of South Australia in 2004 and Bachelor of Engineering from Suranaree University of Technology, Thailand, in 2001. He served as a transport and traffic engineer in a private company in 2004. His research interests include maritime transport, transport logistics and public transport.

Analysing the optimal location of a hub port in Southeast Asia 459

Shinya Hanaoka is an Associate Professor in the Department of International Development Engineering, Graduate School of Science and Engineering, Tokyo Institute of Technology, Japan. He received a PhD, Master and Bachelor Degree from Tohoku University, Sendai, Japan. He is a Visiting Associate Professor of School of Engineering and Technology, Asian Institute of Technology (AIT), Thailand. He worked as a Researcher in Institute for Transport Policy Studies at Tokyo (1999–2003), a Visiting Researcher in Institute for Transport Studies, The University of Leeds, UK (2002). His research interests include maritime transport, transport logistics, air transport and transport project management.

1 Introduction

Successful industrial countries have deep-sea ports in suitable locations to enhance the competitiveness of carrier’s port choices. Countries, which do not have the deep-sea ports to accommodate such large ships, must tranship cargo offshore through deep-sea ports in their neighbour countries. This incident boosts the significant overall transport costs because of the mainline and feeder costs.

At present, the continuous growth of international trade instigates the increase in world fleet. Sea transport remains the major artery of international trade owing to its ability to move bulky cargo at relatively low cost. Since the international seaborne trade has increased strongly in every year, shipping companies attempt to limit their port of calls at each end to a minimum visit of hub ports, not only to minimise costs through economies of scale but also to avoid the size and capital intensity of modern containerships. Therefore, international transport environment has led to a form of organisation of container transportation known as a ‘hub and spoke’ system, in which cargo is shipped to a major hub port by ultra-large container vessels and trans-shipped to smaller ports by feeder ships (Keceli and Choi, 2008). In Southeast Asia, most major liners select Singapore, Port Klang (Port Kelang) and Tanjung Pelepas as the hub ports where a portion of containers are further forwarded to the local ports by the smaller ships. However, with the fast growth in container volume markets, Singapore and Malaysian ports could possibly face the problem of increasingly congested fleet for seaborne trade in the near future. Consequently, the neighbour countries currently tend to pay more attention for the investment in sea port projects, both the construction of the new port and the development of existing one to attract the container traffics (imports and exports).

The background of this paper is a real problem that the sea transport service of Thailand has currently developed into the open system for corresponding policy to enhance international trade. From statistic, trade values including both exports and imports had increased from 1684 billion to 5339 billion baht since year 1991 until 2003 or annually 10.1% growth rate (SEATEC Consulting Engineering, 2005). However, owing to current lack of deep-sea port on the west coast of southern Thailand, Thailand industries have subsequently lost the chances in developing the international trade at the most efficient level. At present, there are four ports on the west coast of southern Thailand as follows: Kangtang, Ranong, Krabi and Phuket but they are not deep-sea ports. The feeder ships are used to carry the cargoes from these origin ports via Singapore or Malaysian ports as the transhipment ports to the final destination in the world regions. If there is an available deep-sea port on the west coast of southern Thailand, it can be

460 P. Boontaveeyuwat and S. Hanaoka

used as its own origin for direct imports and exports and an efficient option of international transhipment ports for carriers to the countries in South Asia, Middle East, European, Africa and South America regions.

Satun is a province in the south of Thailand that borders Malaysia and situated on the west coast in the vicinity of Malacca Straits in the Indian Ocean. Therefore, its location has a full potential to develop as the distribution centre or transportation hub for international trade. Therefore, a prospective deep-sea port named as Pakbara port is situated in Satun as depicted in Figure 1. Pakbara port is a more convenient hub than using the feeder system for the major shipping lines to transport large volumes of cargo to the southern seaboard area of Thailand, which, consequently, benefits the development of various industries in terms of investment through the rising standard of living in the southern Thailand.

Figure 1 The location of Pakbara port (see online version for colours)

When such additional new port as Pakbara port become available for the shipping companies’ selections as a hub competition against the main ports in Southeast Asia, the changes of hub port selection can emerge. The focus of this paper is to apply a designed integer programming model, which emphasises the difference of shipping and port costs on trunk and feeder routes, on the selection of Pakbara port among the five main ports in Southeast Asia as follows: Singapore, Port Klang, Laem Chabang, Tanjung Pelepas, and Jakarta ports, together with various origin and destination container flows in the world.


2 The model in the literature

Hubs are facilities that serve as the transhipment and switching points, functioning as the load centre to link with various origins and destinations. The hub location problems can significantly be applied in the transportation and telecommunication system. A non-negative flow is associated with several origin and destination port pairs and involved with the attributes for analysis such as distance, time and costs associated with movements (Campbell, 1994).

Campbell (1994) presented integer programming formulations for four main types of discrete hub location problems:

1 P-Hub Median problem (P-HM)

• P-Hub Median problem with spoke flow thresholds (P-HM-TS)

2 Uncapacitated Hub Location Problems (UHLP)

• Uncapacitated Hub Location Problem with flow thresholds (UHLP-T)

3 Capacitated Hub Location Problem (CHLP)

4 P-Hub Centre problems (P-HC1, P-HC2)

• P-Hub Centre problem with spoke flows (P-HC1-T)

5 Hub Covering Problems (HCV)

• Hub Covering Problem with penalty (HCV-P)

• Hub Maximal Covering problem (HMCV)

• Hub set covering problem with flow thresholds (HMCV-T).

It is worth to mention that, in the five problems above, every origin to destination movements are via at least one port and no origin to destination movements are via more than two hubs as long as the cost of movement is an increasing function. The P-HM has the objective of total cost minimisation, which matches to the real problem in the transportation and telecommunication. The UHLP does not consider the number of hubs but instead, the fixed cost of establishing the hub is considered. The CHLP is same as the UHLP only adding the constraint of hub capacity.

The P-HC is suitable for the application like, for instance locating emergency service facilities and vehicles, and the transportation of perishable or time-sensitive items. The basic objective of P-HC is to minimise the time between origin and destination. The HCV have an inverse relationship to the centre problems. Every origin and destination demands are covered by the hubs to serve enough demands. The cost involving the establishment of the hub port is minimised under the available budget in case that the cost to cover all origin-destination demands is excessive. In summary, the first two problems are important for the analysis of location problem for the transportation and transhipment costs whilst the latter two problems focus on the minimisation of service time, which are appropriate for the location problem of emergency service facilities or vehicle. Aversa et al. (2005) summarised the main consideration of the characteristic of location models in Table 1.


Table 1 Model characteristics

Model Number of hubs Hub fixed cost Fixed cost of feeder line

Minimum flow of feeder line

P-HM x P-HM-TS x x x UHLP x UHLP-T x x CHLP x

P-HC1 x P-HC2 x P-HC1-T x x HCV x HCV-P x HMCV x x HMCV-T x x

Source: Aversa et al. (2005)

The upper part of Table 1 shows the mode class of model application, which focuses on the transportation and transhipment costs for the transhipment terminal. The bottom part focuses on service time for emergency service facilities or vehicles.

The P-HM problem had initially been researched by Hakimi (1964, 1965). The next study of P-HM model had been conducted by O’Kelly (1986, 1987), Klincewicz (1991), Aykin (1990) and Campbell (1994). Aversa et al. (2005) applied a mixed integer programming based on P-HM model to analyse the selection of a hub port in the east coast of South America. The model consists of 3883 decision variables and 4225 constraints and provides the result of optimal single-hub and multiple-hub solutions.

In this paper, both the origin/destinations and the hub candidate locations are specified as discrete sets of points. Planar hub location problems, where hubs could be located anywhere on a plane, have been addressed by Aykin (1988), Aykin and Brown (1992).

3 The location model

Since this paper focuses on the cost minimisation of containerised cargo movements on the hub and spoke system, as well as to analyse the fraction of container flows from origin to destination ports, via one or more hub ports in Southeast Asia, the model concept is based on the P-HM model by Campbell (1994). However, the model developed in this paper aims to segregate the significant difference of port and shipping costs appeared on main and feeder lines by the formation of different mathematic models to focus on the cost analysis of ‘hub and spoke’ system.

The following detail demonstrates the summary of formulation and model used in this paper from a carrier’s perspective as follows: parameters and objective function, which consists of decision variables and constraints.


( ) / 24VIKJ IK KJSt DIS DIS Sp= + × (1)

[ ( 24)] [ /( 24)]V vIKj IK KjSt DIS Sp DIS Sp= × + × (2)

[ ( 24)] [ /( 24)]v viKJ iK KjSt DIS Sp DIS Sp= × + × (3)

[ ( 24)] [ /( 24)].v ViKJ iK KJSt DIS Sp DIS Sp= × + × (4)

StIKJ: Streaming time from major port I to major port J via hub port K (days) StIKj: Streaming time from major port I to feeder port j via hub port K (days) StiKj: Streaming time between feeder port i and feeder port j via hub K (days) StiKJ: Streaming time between feeder port i and major port J via hub K (days) DISIK: The distance between major port I and hub port K (Nautical Mile.) DISKJ: The distance between hub port K and major port J (Nautical Mile.) DISKj: The distance between hub port K and feeder port j (Nautical Mile.) DISiK: The distance between feeder port i and hub port K (Nautical Mile) SpV: The speed of large ship V (knots) Spv: The speed of feeder ship v (knots).

viK iK iKSC CV St= × (5)

vKj Kj KjSC CV St= × (6)

VIK IK IKSC CV St= × (7)

.VKJ KJ KJSC CV St= × (8)

SCiK: Sea cost from feeder port i to hub port K (US$/TEU) SCKj: Sea cost from hub port K to feeder port j (US$/TEU) SCIK: Sea cost from major port I to hub port K (US$/TEU) SCKJ: Sea cost from hub port K to major port J (US$/TEU)

:viKCV Vessel daily operating cost along the feeder route from feeder port i to hub port

K by feeder ship v ($US/TEU/day) :v

KjCV Vessel daily operating cost along feeder route from hub port K to feeder port j by feeder ship v ($US/TEU/day)

:VIKCV Vessel daily operating cost along trunk route from the major port I to hub port K

by large ship V ($US/TEU/day) :V

KJCV Vessel daily operating cost along trunk route from hub port K to major port J by large ship V ($US/TEU/day) StiK: Streaming time from feeder port i to hub port K (days) StKj: Streaming time from hub port K to feeder port j (days) StIK: Streaming time from major port I to hub port K (days) StKJ: Streaming time from hub port K to major port J (days).


v viK i K i KPC EC EC HC HC= + + + (9)

v vKj K j K jPC EC EC HC HC= + + + (10)

V VIK I K I KPC EC EC HC HC= + + +

(11)

V VKJ K J K JsPC EC EC HC HC= + + + (12)

PCiK: Port costs from feeder port i to hub port K (US$/TEU) PCKj: Port costs between hub port K and feeder port j (US$/TEU) PCIK: Port costs between major port I and hub port K (US$/TEU) PCKJ: Port costs between hub port K and major port J (US$/TEU)

:viEC Entrance charge at feeder port i entered by vessel v ($US/vessel)

, :V vKEC Entrance charge at hub port K entered by large vessel V or small vessel v

($US/vessel) :V

IEC Entrance charge at major port I entered by vessel V ($US/vessel) :v

jEC Entrance charge at feeder port j entered by vessel v ($US/vessel) :V

JEC Entrance charge at major port J entered by vessel V ($US/vessel) :iHC Cargo-handling charge at feeder port i ($US/TEU) :IHC Cargo-handling charge at major port I ($US/TEU) :KHC Cargo-handling charge at hub port K ($US/TEU) :jHC Cargo-handling charge at feeder port j ($US/TEU) :JHC Cargo-handling charge at major port J ($US/TEU)

IK IK IKMC PC SC= + (13)

KJ KJ KJMC PC SC= + (14)

Kj Kj KjFSC PC SC= + (15)

iK iK iKFSC PC SC= + (16)

iK iK iKFRC CR DIS= × (17)

Kj Kj KjFRC CR DIS= × (18)

Min( , )IKj Kj Ij Kj IjFC FRC Q FSC Q= × × (19)

Min( , )iKJ iK Ij iK IjFC FRC Q FSC Q= × × (20)

iKj iKj IKjTC FC FC= + (21)

( ) +IKj IK IK IKjTC MC Q FC= × (22)

( )iKJ iKJ KJ iJTC FC MC Q= + × (23)

( )IKJ IK KJ IJTC MC MC Q= + × (24)


MCIK: Mainline costs from port I to hub port K ($US) MCKJ: Mainline costs from hub port K to major port J ($US) FSCKj: Feeder costs of sea transport from hub port K to feeder port j ($US) FSCiK: Feeder costs of sea transport from feeder port i to hub port K ($US) FRCiK: Feeder cost of road transport cost from feeder port i to hub port K ($US) FRCKj: Feeder cost of road transport from hub port K to feeder port j ($US) FCIKj: Minimum feeder cost between road and sea transport from hub port K to feeder port j servicing container flows from port I to j ($US) FCiKJ: Minimum feeder cost between road and sea transport from feeder port i to hub port K servicing container flows from port i to J ($US) Qij: The container flows from feeder port i to feeder port j (TEU) QiJ: The container flows from feeder port i to major port J (TEU) QIj: The container flows from major port I to feeder port j (TEU) QIJ: The container flows from major port I to major port J (TEU) TCiKj: Total costs from feeder port i via hub port K to feeder port j ($US) TCIKj: Total costs from major port I via hub port K to feeder port j ($US) TCiKJ: Total costs from feeder port i via hub port K to major port J ($US) TCIKJ: Total costs from major port I via hub port K to major port J ($US).

The objective function is

Minimisationi K j I K j

i K J I K J

iKj iKj IKj IKji N K N j N I N K N j N

iKJ iKJ IKJ IKJi N K N J N I N K N J N

TC X TC X

TC X TC X∈ ∈ ∈ ∈ ∈ ∈

∈ ∈ ∈ ∈ ∈ ∈

+

+ +

∑ ∑ ∑ ∑ ∑ ∑

∑ ∑ ∑ ∑ ∑ ∑ (25)

Subject to K

kK N

Y N∈

=∑ (26)

1 ,K

iKj i jK N

X i N j N∈

= ∀ ∈ ∀ ∈∑ (27)

1 ,K

IKj I jK N

X I N j N∈

= ∀ ∈ ∀ ∈∑ (28)

1 ,K

iKJ i JK N

X i N J N∈

= ∀ ∈ ∀ ∈∑ (29)

1 ,K

IKJ I JK N

X I N J N∈

= ∀ ∈ ∀ ∈∑ (30)

, ,iKj K i K jX Y i N K N j N≤ ∀ ∈ ∀ ∈ ∀ ∈ (31)

, ,IKj K I K jX Y I N K N j N≤ ∀ ∈ ∀ ∈ ∀ ∈ (32)


, ,iKJ K i K JX Y i N K N J N≤ ∀ ∈ ∀ ∈ ∀ ∈ (33)

, ,IKJ K I K JX Y I N K N j N≤ ∀ ∈ ∀ ∈ ∀ ∈ (34)

{0, 1}KY ∈ (35)

whereas, YK: Binary variable that is equal to one if the location of hub port K is selected and zero otherwise XiKj: Flow fraction from feeder port i to j via hub port K XIKj: Flow fraction from major port I to feeder port j via hub port K XiKJ: Flow fraction from feeder port i to major port J via hub port K XIKJ: Flow fraction from major port I to major port J via hub port K N: The exact number of hub ports that must be open Ni: The set of feeder ports i in the total ports Nj: The set of feeder ports j in the total ports NI: The set of major ports I in the total ports NJ: The set of major ports J in the total ports NK: The set of candidate hub ports K in total ports.

The objective function (25) minimises the total system costs, which combines the sea and port costs. Constraint (26) establishes the number of hub ports. Constraints (27)–(30) assure that every origin/destination container flow is serviced via at least one hub port. Constraints (31)–(34) ensure that the container flows are routed via the hub ports. Lastly, the constraint (35) imposes the binary restrictions on the decision variables.

4 Data

This section describes the entry data used in this paper. The Pakbara port is assumed to be available for carrier’s selection as a hub port competition among the major ports in Southeast Asia as previously mentioned: the ports of Singapore, Port Klang, Laem Chabang, Tanjung Pelepas and Jakarta, which have recently been ranked at top 25 ports in term of annual throughput in 2006 (Fryer, 2007).

Owing to the focus on the hub port competition in Southeast Asia, the container flows from 16 significant feeder ports to the destination ports, which could be hub port candidates or major ports in different regions of the world, are considered in this paper. The major ports in world regions are selected as the representative ports in each region for the purpose of distance calculation. As listed, Busan port is selected as the representative port for Northeast Asian region, which comprises 44 regional ports. Shanghai port is selected for Eastern China region, which consists of six regional ports, Hong Kong port is selected for Southern China region, which consists of 15 ports, Colombo port is selected for Southern Asia region, which consists of four ports. Los Angelis is selected for North America, which comprises five ports. Santos port is selected for South America, which consists of three ports. Dubai is selected for Middle East region. Rotterdam port is selected for Europe, which comprises two ports.


Durban port is selected for Africa and Melbourne port is selected for Oceania, which consists of two ports. The realistic port distance data are conducted by using Netpas software. The origin-destination container flows have been estimated annually by a Japanese Research Institute.

It is worth mentioning that the existing container origin-destination data were collected from various sources as follows: Mitsui O.S.K. Lines, Ltd. for origin-destination flows between world’s continent, Ocean Commerce Ltd. for container origin-destination flows between Asian countries, container flow data between Japan and counterpart countries is obtainable from the report published by Japan’s Ministry of Land, Infrastructure, and Transport. The various statistical documents and reports were collected from the Containerisation International Yearbook. Other existing data were provided by the United Nation Trade Statistics, which show monetary import/export trade amounts for each United Nations member country on the basis of continents, and major counterpart countries. Since Taiwan is not a United Nations member country, the information from Taiwan’s International Trade Bureau is used instead. The data were estimated by different methodologies depending on the data availability. Basically, the gravity and fratar models were used incorporating with some specific methodologies as shown in Shibasaki et al. (2005). Table 2 presents the estimated container flows at the ports considered in this analysis.

Table 2 Estimated container flows at the ports in Southeast Asia (SEA) and other ports in the world regions

Container flows Ports and regions Hub port candidate Major ports in world regions Feeder Ports in SEA Singapore 124,116.56 1,413,014.46 145,115.81 Port Klang 161,541.98 1,143,953.52 125,168.51 Laem Chabang 70,685.93 1,723,066.36 113,296.43 Tanjung Pelepas 7,549.46 65,828.62 2,774.94 Jakarta 103,883.56 972,144.97 90,144.91 North East Asia 552,557.65 8,308,077.53 552,084.64 East China Asia 495,384.79 7,751,926.82 449,109.97 South China 1,907,831.45 27,171,353.55 1,567,187.85 South Asia 161,429.9823 2,188,997.68 90,448.75 North America 458,436.3595 5,846,959.60 312,998.71 South America 105,610.1863 1,038,523.87 96,278.50 Middle East 79,899.50 870,326.58 63,511.64 Europe 576,396.84 2,829,007.63 283,754.12 Africa 18,372.73 103,409.77 11,908.28 Oceania 248,369.41 716,766.48 145,319.74 Feeders in SEA 303,057.25 3,061,675.78 –


Table 3 enumerates all port members in this study. The parameters of vessel daily value rate and the economies of scale on trunk and feeder routes are essential for careful selection in this section. Basically, the ships larger than 1500 TEU can be used in international long-distance shipping (Aversa et al., 2005). However, since the introduction of the leading carrier American President Line (APL) in 1988, the carrier’s share on the Post-Panamax containerships (ship capacity not less than 3999 TEU) has been rapidly increasing. The number of international routes (North America – Asia, Europe – Asia and World routes) operated by different levels of ship sizes (4000–4999, 5000–5999 and over 6000 TEU vessels) was investigated by Hanaoka et al. (2006). The significant findings were the highest average number of international routes operated by 4000–4999 ship sizes appeared on North America (NA) – Asia and World (NA-Asia – Europe) routes and slightly less than both 5000–5999 and over 6000 TEU ship sizes only on Europe–Asia route. Therefore, the range of 4000–4999 TEU can be used as the vessels operating on the trunk route. Nonetheless, the parameters of representative vessel daily value in this paper on the trunk route are used at 4000 TEU (minimum size of Post-Panamax containership) regarding the most accurate value of economies of scale achieved in liner shipping reported in Aversa et al. (2005) as shown in Figure 2. The vessel sizes of feeder ships in Southeast Asia were mostly ranged in 800–1500 TEU (Ocean Commerce Ltd., 2006). The representative feeder ship is used at 1000 TEU in this paper regarding the standard of feedermax ship size (Office of Statistical and Economic Analysis, 2007).

Table 3 The port members in case study

Hub ports Major ports in world regions Feeder ports in SEA Pakpara Busan Manila Singapore Shanghai Haiphong Port Klang Hong Kong Danang Laem Chabang Colombo Ho Chi Minh Tanjung Pelepas Los Angelis Sihanouvkville Jakarta Santos Phnom Penh Dubai Bangkok Rotterdam Johor Durban Penang Melbourne Kota Kinabalu Kuching Thilawan Surabaya Belawan Ujing Pandang Muara


Figure 2 Economies of scale in liner shipping on trunk and feeder routes (see online version for colours)

Source: Based on Aversa et al. (2005), Decker and Hamburg (2001)

Table 4 shows the parameters of daily operating costs of vessels considered in this study. It is relevant to mention that the prevailing daily time or charter rate typically is used to be more visible and more standard to provide much greater clarity and readily as the representative measure of ship daily value. The intuitive additional ship cost element in addition to charter costs has also been described as the fully built-up costs. In essence, the fully built-up running costs are estimated to be equivalent to approximately double a vessel’s charter rates (Baird, 2006). The daily operating costs shown in Figure 2 can be multiplied by two for the approximate fully built-up costs provided by Drewry Shipping Consultants Ltd. (2001).

Table 4 Vessel daily value on trunk and feeder routes

Shipping cost (including fuel costs) Vessel capacity (US$/TEU/day) 4000 TEU 6.96 1000 TEU 8.63

The Post-Panamax containership is deployed in international long-distance shipping on trunk route where the shipping costs between hub ports and major ports on other regions both imports and exports are taken into account. The vessels in the feeder shipping where they are taken along the feeder route between the hub ports and feeder ports in Southeast Asia. The load factor is assumed at 0.8 for both large and feeder ships. An average speed of 18 knots is assumed for the feeder ship as (Spv) and 20 knots for the large ship (SpV). Road transport is used in the feeder shipping only for the possible transportation between the hub port and its regional port. An average of US$ 1 per km has been assumed for this analysis.

The container flows between the feeder and feeder ports in Southeast Asia are not considered in this study regarding the reality. In the real-world case, the flows between feeder and feeder ports in the same region generally are shipped by direct transportation with the small vessel. Furthermore, some cases are neglected for container flows shipped between origin and destination ports, on which the vessel unlikely transit at a hub in Southeast Asia and carry the cargoes to the final destination. For port example, container flows from Busan to Los Angelis ports, these can be carried directly via


North Pacific Ocean and not necessary to transit at a hub port in Southeast Asian region. Another case is the shipment between Busan and Shanghai ports, on which a vessel impossibly sail to tranship the containers at some hub ports in Southeast Asian region before turning back to Shanghai ports rather than taking the economically rational decision of direct shipment between these two ports.

5 Computational results

Since the Pakbara port is expected to be a new hub port in Southeast Asia, it is assumedly located to compete with other hub candidates in Southeast Asia region. Nonetheless, owing to lack of forecasting container flow data for Pakbara port, no direct container flow for both imports and exports at Pakbara port is assumed on the problem basis. The results are presented in the next section, whilst the scenarios and sensitivity analysis are shown in Section 5.2.

5.1 Results of problem basis

The model considers the mainline transfer costs between hub ports and major ports in the world regions served by the Post-Panamax containership (4000 TEU) and the feeder costs between hub ports and feeder ports in Southeast Asia served by the feeder ship (1000 TEU). Finally, the model contains 4045 decision variables and 4286 constraints. The port of Klang is selected as the optimal solution with a total cost of $US 13,062 million.

The cost classification of optimal solution can be seen in Table 5, which demonstrates the detail of cost breakdown for optimal result. The number of interesting points can be observed that the mainline costs dominate the total costs up to 86%, which contains shipping costs between hub ports and world’s ports, amount to 26% of total system costs, and world port costs representing the most significant total system costs at 43%. The remaining cost 17% refers to the hub port costs for the mainline part. The hub port costs for feeder part contains only 5% therefore the total hub port costs add up to 22%. The total feeder costs contain 14% of total costs, i.e., shipping and port costs in the port system of Southeast Asia. This is an evidence to indicate the total port costs, i.e., hub and feeder through world port costs represent a significant portion of overall transport costs at 72% whilst the total shipping costs represent only 28%.

Table 5 Cost classification of optimal solution: Port Klang

Costs 106 US$ Percentage (%)

World port costs 5581.97 43 Main line Shipping costs between hub ports and world’s port 3410.58 26 Large vessel costs for main line 2215.14 17 Hub port costs Small vessel costs for feeder line 710.07 5 Feeder port cost 905.58 7 Feeder line Shipping costs between feeder and hub ports 238.98 2

Total costs 13,062.32 100


5.2 Scenarios and sensitivity analysis

The sensitivity analysis presented in this section is to investigate the allocation of hub port selection when the number of hub ports increases, up to six in Southeast Asia, which is the total number of hub ports considered in this study. The constraint of hub port number is relaxed up to the total number of hub ports. The objective is to investigate the impact of such allocations on total system costs and the possible chance of Pakbara port to become one of the sharing hub ports selection.

Table 6 demonstrates the increasing number of hub ports, up to six, and as a result, the total system costs decrease. When all hub ports are selected, more container flows are shipped to final destination directly via their own origin port. Moreover, other container flows have more optional hub ports, which provide the minimum total system costs for carriers. The single-hub configuration is thereby shown to be 7.31% more costly than the total number of hub ports selected (six ports). An interesting observation is when the number of hub ports gradually increases to six, the mainline costs are gradually increased and feeder costs are declined.

Table 6 Model results regarding Pakbara port located in South-East Asia with no direct container flows

Hub ports No. of hubs

Total costs(106 US$)

Main line costs

(106 US$)

Feeder costs

(106 US$)

% of main line costs

(%)

% of feeder costs (%) PBR SGP PKL LCB TJP JKT

1 13,062.00 11,186.48 1875.88 85.64 14.36 ●

2 12,725.15 11,085.28 1639.92 87.11 12.89 ● ●

3 12,437.73 10,967.26 1470.48 88.18 11.82 ● ● ●

4 12,206.54 10,857.36 1349.18 88.95 11.05 ● ● ● ●

5 12,187.18 10,846.41 1340.77 89.00 11.00 ● ● ● ● ●

6 12,172.32 10,834.40 1336.98 89.01 10.98 ● ● ● ● ● ●

PBR: Pakbara port; SGP: Singapore port; PKL: Port Klang; LCB: Laem Chabang port; TJP: Tanjung Pelepas port; JKT: Jakarta port.

The Pakbara port is selected as the last choice because of no container flows via this port following the assumption. This is very reasonable when no direct container flows at a port, it will be a very rare opportunity that a carrier is going to select that port. At hub port, which has direct container flows, export cargoes can be shipped from its own origin port to final destination. On the other hand, some containers from other origin ports are directly shipped to its own destination port in case of import cargoes. Comparing with a hub port with no direct container flows, every container flow needs to be shipped to that transhipped hub port before shipping to the destination port, which again result in additional time in the shipments and ports through increased port costs.

Other details in Table 6 show the order of increasing the number of hubs: N = 1 (Port Klang), N = 2 (Port Klang and Laem Chabang port), N = 3 (Port Klang, Laem Chabang and Singapore ports), and so on. One can observe that Singapore port, as a top hub port of Asia and the world, is ranked as the third position despite its comparatively high container flows. The parameters (shipping and port costs) play a key role as equal to that of container flows to influence a port as a hub. In this paper, the feasible port costs are used for calculation in the model. It is necessary to mention that Singapore’s port


costs are the most expensive costs among the hub port candidates especially that the cargo-handling charge rate is more costly than the Port Klang and Laem Chabang port, which are selected as a two-hub configuration, amounting to 75% per standard container (TEU).

Although Pakpara port is selected as the last choice, we can observe in the deep information of model results that the container flows shipped from Eastern Asia regions via Pakbara port to major world regions (Southern Asia, Europe, South America and Africa) comparatively achieve the minimum costs as shown in Table 7.

Table 7 Model results regarding the total costs between Asian regions and World’s regions

To destination region; [Total Cost (106 US$)] Origin region

Transhipment port Southern Asia South America Middle East Europe Africa Pakbara 52.63 28.79 90.72 677.72 52.26 Singapore 62.25 32.94 107.44 761.66 61.26 Port Klang 52.83 28.82 90.85 678.68 52.34 Laem Chabang 54.45 29.53 93.93 694.06 53.99 Tanjung Pelepas 59.09 31.56 101.89 733.90 58.27

North East Asia (Busan)

Jakarta 57.15 29.94 98.65 717.21 54.97

Pakbara 62.45 32.27 64.71 604.71 39.36 Singapore 75.18 37.34 78.06 685.32 46.88 Port Klang 62.71 32.31 64.82 605.64 39.42 Laem Chabang 64.86 33.18 67.27 620.38 40.80 Tanjung Pelepas 70.99 35.66 73.63 658.67 44.38

East China (Shanghai)

Jakarta 68.46 33.68 71.07 642.79 41.64

Pakbara 304.38 71.89 370.79 2179.05 154.65 Singapore 358.70 82.13 438.11 2446.17 180.90 Port Klang 305.47 71.98 371.32 2182.11 154.87 Laem Chabang 314.58 73.72 383.58 2230.59 159.65 Tanjung Pelepas 340.83 78.74 415.76 2357.83 172.19

South China (Hong Kong)

Jakarta 330.54 74.85 403.53 2307.95 162.88

It can easily be perceived that the results remain the same in the reverse way when container flows shipped from the major ports in the world’s regions to Asian regions since the distance considered in this study is symmetric.

In addition, the container flows shipped from most of the feeder ports in Southeast Asia to world’s regions as mentioned earlier can achieve the minimum costs by using Pakbara port as transhipment port. The feeder ports mentioned are listed here as follows: the ports of Manila, Haiphong, Danang, Ho Chi Minh, Sihanouvkville, Phnom Penh, Johor, Penang, Kota Kinabalu, Kuching, Thilawan, Surabaya, Belawan and Ujing Pandang.

As explained previously, the main objective of this paper is to determine the minimum requirement for the Pakbara port to be the hub port in Southeast Asia after completely constructed. Since there is no direct container flows via Pakbara port,


the increase in container flows is conducted in different scenarios to investigate the impact of Pakbara port location. The hub port candidates’ traffic flows are assumedly added to Pakbara port, both exports and imports. Five scenarios are conducted and the results are shown in Table 8. The same amount of decision variables and the increasing number of constraints in the model depended on each case appeared.

Table 8 Scenario results

Scenario Description Number of hub ports

Total Cost (US$ million)

Decision Hub ports

1 13,945.04 Pakbara 2 13,596.87 Pakbara Laem Chabang

I Add the Singapore port’s container flows to Pakbara port

6 12,824.94 All 1 13,772.13 Port Klang 2 13,430.63 Port Klang Laem Chabang

II Add the Port Klang’s container flows to Pakbara port

6 12,662.80 All 1 13,901.55 Pakbara 2 13,553.39 Pakbara Laem Chabang

III Add the Laem Chabang port’s container flows to Pakbara port


IV Add the Tanjung Pelepas port’s container flows to Pakbara port


V Add the Jakarta port’s container flows to Pakbara port

6 12,657.88 All

The results of Scenarios (I) and (III) show the achievement of Pakbara port as a single-hub configuration when the traffic flows are equal to Singapore and Laem Chabang ports’ traffic flows. The Port Klang engages the single-hub status for other three scenarios and Laem Chabang port is ranked the second position in every case.

6 Conclusions

In reality, the significant volume of containers, convenient intermodal transport system, adequate feedering network, efficient port facility and competitive port costs have generally been considered as the main factors for the achievement of hub status to be able to handle the very huge surge of container flows and the large ship. This study focuses on the key parameters such as the shipping and port costs through the amount of container flows, suitable to be a basis of decision tool for the analysis of a hub port selection.


This paper has implemented the hub port competition in Southeast Asia region after the Pakbara port is expected to emerge in the near future with the consideration of several origin and destination container flows in the world. In the problem basis, no traffic flows at Pakbara port is assumed and it turns up that Port Klang is the optimal single-hub port and Laem Chabang port as the runner hub port when the two hub ports configuration are considered. The classification of total system costs, which provides the port costs, represents the main portion of overall transport costs at 72% and the shipping costs represent only 28%.

The different scenarios are conducted by adding the amount of traffic flows to Pakbara port, equal to its competitive ports as follows: the ports of Singapore, Port Klang, Laem Chabang, Tanjung Pelepas and Jakarta. Obviously observed, the impact of traffic flows turns up Pakbara port as the optimal single-hub port when Singapore and Laem Chabang ports’ structural traffic flows are used. These results indicate that the location of Pakbara port is more advantageous than Laem Chabang port, which is currently the leading Thai deep-sea port, whilst the cheaper port costs are the key factor to aspire Pakbara port to beat Singapore port as the transhipment hub port for the attractiveness of container flows. The interesting observation of the deep information in the model result is the flows in the main regions of Asia, which needs to be shipped via Indian and South Atlantic Ocean to Southern Asia, Europe, Middle East, South America and Africa as the final destination, it indicates that Pakbara port can offer the minimum total system costs as the transhipment port when compared with other hub port candidates.

Acknowledgements

The authors wish to thank the National Institute for Land and Infrastructure Management, the Ministry of Land, Infrastructure and Transport, Japan, for providing the data of origin-destination container flows to be utilised in this paper.

References Aversa, R., Botter, R.C., Haralambides, H.E. and Yoshizaki, H.T.V. (2005) ‘A mixed integer

programming model on the location of a Hub port in the East Coast of South America’, Maritime Economics and Logistics, Vol. 7, pp.1–18.

Aykin, T. (1988) ‘On the location of hub facilities’, Transportation Science, Vol. 22, pp.155–157. Aykin, T. (1990) ‘On a quadratic integer program for the location of interacting hub facilities’,

European Journal of Operational Research, Vol. 46, pp.409–411. Aykin, T. and Brown, G.F. (1992) ‘Interacting new facilities and location-allocation problems’,

Transportation Science, Vol. 26, pp.212–222.. Baird, A.J. (2006) ‘Optimising the container transshipment hub location in northern Europe’,

Journal of Transport Geography, Vol. 14, No. 3, pp.195–214. Campbell, J.F. (1994) ‘Integer programming formulations of discrete hub location problems’,

European Journal of Operational Research, Vol. 72, pp.387–405. Decker, M. and Hamburg, C. (2001) ‘Charter report’, Containerisation International, Vol. 34,

pp.26–27. Drewry Shipping Consultants Ltd. (2001) Salalah Port Container Market Study, Report for Salalah

Port Service Co., Drewry Shipping Consultants Ltd, Oman, London.


Fryer, S. (2007) ‘Top 120 container ports’, Container Management Magazine, Available from URL: http://www.container-mag.com/top120.php?SID (Access 2 March 2008).

Hakimi, S. (1964) ‘Optimum location of switching centers and the absolute centers and medians of a graph’, Operation Research, Vol. 12, pp.450–459.

Hakimi, S. (1965) ‘Optimum distribution of switching centers in a communication network and some related graph theoretic problems’, Operation Research, Vol. 13, pp.462–475.

Hanaoka, S., Boontaveeyuwat, P. and Ishiguro, K. (2006) ‘Analysis of recent transition in port choice of post-panamax containerships in Asia’, Proceedings of International Association of Maritime Economists (IAME) 2006 Conference, Melbourne, CD-ROM.

Keceli, Y. and Choi, H.R. (2008) ‘Level of information systems in Turkish public ports and direction of improvement’, International Journal of Logistics Systems and Management, Vol. 4, No. 6, pp.673–691.

Klincewicz, J.G. (1991) ‘Heuristics for the p-hub location problem’, European Journal of Operational Research, Vol. 53, pp.25–37.

Ocean Commerce Ltd. (2006) International Transportation Handbook, Ocean Commerce. (In Japanese).

Office of Statistical and Economic Analysis (2007) Containership Market Indicators, August 2005, Government Funded Technical Reports, US Maritime Administration, Washington DC.

O’Kelly, M.E. (1986) ‘Activity levels at hub facilities in interacting networks’, Geographical Analysis, Vol. 18, No. 4, pp.343–356.

O’Kelly, M.E. (1987) ‘A quadratic integer program for the location of interacting hub facilities’, European Journal of Operational Research, Vol. 32, pp.393–404.

SEATEC Consulting Engineering (2005) The Feasibility Project Study of Deep Sea Port at Pakbara, Satun, Marine Department (In Thai), Bangkok.

Shibasaki, R., Ma, L., Kannami, Y., Ieda, H. and Takashi, K. (2005) ‘Regional based international maritime container cargo flow estimation between East Asian Nations’, Proceeding of 1st International Conference on Transportation and Logistics 2005, Singapore, CD-ROM.

analysing the optimal location of a hub port in southeast asia

Documents