You are on page 1of 17

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/274350027

Hub port potential of Marmara region in Turkey by network-based modelling

Article  in  Transport · January 2014


DOI: 10.1680/tran.13.00043

CITATIONS READS

3 86

2 authors:

Nil Guler Firat Bolat


Istanbul Bilgi University Istanbul Technical University
7 PUBLICATIONS   49 CITATIONS    25 PUBLICATIONS   12 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Directorate General of Coastal Safety of Turkish Republic - Istanbul Technical University (ITU) NOVA Educational Management Project View project

Maritime Education for Energy Efficiency (MARED) View project

All content following this page was uploaded by Nil Guler on 25 June 2020.

The user has requested enhancement of the downloaded file.


Transport Proceedings of the Institution of Civil Engineers

Hub port potential of Marmara region in http://dx.doi.org/10.1680/tran.13.00043


Turkey by network-based modelling Paper 1300043
Bolat and Guler Received 13/06/2013 Accepted 04/01/2014
Keywords: mathematical modelling/ports, docks & harbours/transport
planning

ICE Publishing: All rights reserved

Hub port potential of Marmara


region in Turkey by network-
based modelling
j
1 Firat Bolat j
2 Nil Guler
Flag, Port and Coastal State Surveyor, Ministry of Transportation, Dean, Istanbul Technical University Maritime Faculty, Istanbul, Turkey
Maritime Affairs and Communications of Turkish Republic, Tuzla
Headquarter, Istanbul, Turkey

j
1 j
2

The increase in the efficiency of main hub ports is a result of home base theory. Put another way, a network
comprising linked nodes that are in collaboration is the issue of concern. In a sense, the bases and facilities of the
main hub port serve as the trans-shipment and assembling centres for the departure destination flows. Based on this
theory, this study puts forth the main hub port potentials for the Marmara region ports in Turkey by analysing the
connection-rich nodes in the region through a transition from a random graph network structure to a non-scaled
network structure using the port connectivity index and the port collaboration index.

Notation Istanbul, Izmit and Tekirdag), is a promising region for analysis


Cij competitiveness index between port i and port j from the perspective of hub–spoke potential. The Turkish Straits
ni number of major ports that are accessible from port i region has a vessel transit 164 nautical miles ( 300 km) long,
n ij common nodes that can be accessed via port i and port j comprising the Istanbul Strait (17 nautical miles ( 31 km)), the
nj number of major ports that are accessible from port j Bosporus and the Canakkale Strait (37 nautical miles
Si connectivity index of port i ( 300 km)) in the Dardanelles. This unavoidable maritime route
is of great importance to the economies of the countries border-
1. Introduction ing the Black Sea. Considering factors such as
A hub port differs from other types of ports in terms of location-
logistics, trans-shipment volume and hinterland properties. With j the very high density of maritime traffic
the evolution of linear shipping network structures, the hub port j the increase in maritime accidents
concept has become prominent. The main factors that triggered j increases in size and tonnage of ships
hub ports to come to the fore also shed light on why countries j increases in the transit of vessels carrying hazardous cargo
actually need hub ports. In this respect, it is crucial to realise the j complex and difficult trafficking
relationship between globalisation and containerisation, which j adverse weather
impacts on today’s commercial development: the urge to globalise j sea, current and climate conditions
has created a highly suitable environment for containerisation and, j regional dangers
in turn, containerisation provided access to globalisation. With the j national and international developments
transition from open-sea shipping to total logistics shipping, the j other maritime activities in the Bosporus region
accretion of trade routes towards ports that can handle containers,
the globalisation of manufacturing plants and the emergence of it appears obvious that there is a great need for a hub port that
the supply chain as a discipline, the relationship between can distribute container cargo to other ports in the Sea of
globalisation and containerisation has increased even further. As a Marmara and to ports in the Black Sea without transiting the
result of this relationship, the efficiency of the main hub ports has Bosporus (i.e. leaving the cargo in the Sea of Marmara).
increased, shifting towards a ‘hub-and-spoke’ structure. Predicating on this need, this study aims to define and analyse
features of container-handling ports in the Marmara region in
In these circumstances, the Turkish Straits region, which accom- terms of their potential to be a hub port, and determine their
modates the Marmara Sea ports (principally Ambarli, Gemlik, current status and predict their future potential in a scientific way.

1
Transport Hub port potential of Marmara region in
Turkey by network-based modelling
Bolat and Guler

Offprint provided courtesy of www.icevirtuallibrary.com


Author copy for personal use, not for distribution

Accordingly, this study assesses the hub port potential of the information on the factors that affected the selection process
Marmara region ports by analysing the connection-rich nodes in (Low et al., 2009). Although these studies defined and listed the
the region through a transition from a random graph network factors important for linear transportation companies, they could
structure to a non-scaled network structure. The port connectivity not reveal how these factors affected the selection of ports. As an
index and the port collaboration index, which can be defined as extension of these studies, the analytic hierarchy process model
an accessibility index and a sustainability index respectively was used in analysing the survey data. Song and Yeo (2004)
(Low et al., 2009; Taylor et al., 2006), are used as driving combined the data obtained from surveys with various factors in
variables to obtain the network structure. These indices are port selection. In recent studies, multi-nominal logit models have
explained in the paper in detail while focusing on the network- been used to show the effects of important factors in port
based hub port assessment (NHPA) model, one of the scientific selection. A logit model is a statistical regression model that
analysis and prediction methods for port selection and analysis. analyses tabled or raw data sets to enable classification fitting in
probability rules by calculating the probability of the estimated
The analysis and prediction using the aforementioned method value of the dependent variable (Inal et al., 2006; Ozdamar,
was based on container cargo movements at container-handling 1999). The multi-nominal logit model is used when the dependent
ports in the Marmara region between 2004 and 2008. The variable is nominal and comprises more than two categories. For
connections between the ports were then calculated and the instance, Tiwari et al. (2003) investigated port selection beha-
network structure between these ports defined. The data obtained viours using this model and fitted data obtained from a survey,
were analysed using the NHPA model and the features of the conducted on the shipper party, onto a multi-nominal logit model.
ports in terms of their potential to be a hub port were defined and They found that distance and port intensity were the primary
forecasts made of their future situation. factors that affect port selection. In addition to this, Malchow and
Kanafani (2004), again using the multi-nominal logit method,
The results of the study enable construction of a non-scaled argued that the inland distance and the intensity of transportation
network structure. Therefore, the primary analysis focused on was inversely proportional to the probability of shipper and port
network structure of the ports rather than their capacity, and the selection.
notions of home base theory (the starting point of the main hub
port concept) and non-scaled networks were applied accordingly When the studies are examined, it is obvious that even the multi-
nominal logit cannot be used as a discrete selection model. The
2. Literature review fact that the multi-nominal logit models ignore the service
Generally, a hub port is part of a regional port system comprising network structure, which is the main element of the maritime
a supply port and some secondary ports and their hinterlands. industry, especially the container port industry, is their major
Hub ports have been a basic research interest in transportation disadvantage and has caused the model to remain limited (Low et
studies. Studies on this topic can be categorised into two classes. al., 2009). Therefore, more recent studies on port selection have
begun to include network-based models for port selection in the
The first class investigated how a port system develops and how international transportation industry. For instance, Taylor et al.
it interacts with its hinterland and land transportation systems. (2006) introduced the network-based heuristic selection model in
The preliminaries of this class of studies are exemplified in which a connectivity index was developed and used, which
Rimmer (1977) and Taaffe et al. (1963). The main ports in these supports the acceptability of a transportation network concept.
studies were located in developing countries with a history of This model was supported by revealing the important dimensions
colonisation and these ports had become hub ports for their in port management by factor analysis (Lattin et al., 2003). Low
regions. Therefore these studies focused, naturally, on the func- et al. (2009), in the network-based holistic selection model, used
tions of the ports as a transit point. the port collaboration index in addition to the port connectivity
index and put forth the NHPA model.
The second class of studies, including the current study, is
interested mostly in the evolution of the port system – namely
3. Methodological approach
how ports compete to become the hub port and how new
technologies, such as container transport, change the area. Indeed, 3.1 NHPA model
the development inclinations of container port systems may The NHPA model can be seen as an extension to studies that
change according to regional conditions. These kinds of variances revealed the reasons supporting regional or nationwide port
can be accepted as the results of sea transport cycles that do or competition or collaboration and port competitiveness – or
do not fit into the context of regional economic development and keeping it in the background. In previous studies, various ports
growth included in a global economy (Rodrigue et al., 1997). The were compared in terms of competition or collaboration in the
selection criteria of the main hub/loading centrals are variable selection of main hubs. These comparisons varied regionally or
due to the variance of the criteria of being a hub port. When the from one nation to another. For instance, in a study on the USA
studies on port selection processes are examined, it can be seen and Western Europe, Fleming and Baird (1999) explained why
that the early studies were survey studies that attempted to obtain some ports were inevitably superior in terms of competitiveness.

2
Transport Hub port potential of Marmara region in
Turkey by network-based modelling
Bolat and Guler

Offprint provided courtesy of www.icevirtuallibrary.com


Author copy for personal use, not for distribution

Cullinane et al. (2002) investigated the ports of Shanghai and pairs on which the ports serve as points of departure and
Ningbo for China and scrutinised the competitiveness of these destination, and by evaluating the network structure that then
two ports in terms of port costs and service quality. Song (2003) arises. Direct investigation of the connectivity indices of ports
surveyed Hong Kong and Shenzen ports in terms of collaboration enables us to comprehend the competition between these ports
of neighbouring ports. While these studies aimed to reveal the and measures the probability of a port in terms of the aspects of
important aspects for main hub selection, the importance of being a global or regional main hub for container transport (Low
network structure was discarded. This shows that the NHPA et al., 2009). The connectivity index is also an indicator of the
model fills in the missing parts of these studies via the port degree of potential of a hub, as a container port’s main duty is to
connectivity and collaboration indices it contains. Thus, the carry the cargo of the carrier from the point of supply to the
NHPA model as a holistic model has some advantages over other point of demand in maritime lines (Wang and Cullinane, 2006).
models (Low et al., 2009). These advantages are as follows. Furthermore, the relation between connectivity index and the
observed crucial aspects of port competitiveness is presented in
(a) The NHPA model bases its results not on some subjective terms of its effects on the connectivity index by conducting a
survey data but on the activities of users of the ports. factor analysis as an extension to the multi-nominal model by
(b) It is used in order to determine how much various port Malchow and Kanafani (2004) and Tiwari et al. (2003), which
features contribute to the port being a focus of selection and discarded the network structure.
adopting a network approach ensures that the relationship of
one port to another is taken into consideration. In addition to the port connectivity index for assessing main hub
(c) The connectivity index in the NHPA model can be interpreted performance, the port collaboration index should also be ob-
as a measure of port competitiveness as the logistic main hub. tained. The degree of the collaboration index is used as a
This is because the linear transport companies are the main criterion for the sustainability of the current network connection.
users of the ports and the connectivity index is calculated as
a result of the itinerary data used by the linear transport Low et al. (2009) assessed global and regional main hubs using
companies for their maritime cargo. these indices. For this study, we degraded global hubs to regional
ones and regional hubs to national ones. This kind of reduction is
The process of the NHPA model, presented schematically in valid since any observation made on the minimal element of a
Figure 1, comprises the assessment of the two network-based system would be valid for the maximal element in that system
indices, namely the connectivity index and the collaboration according to chaos and complexity theory. On the other hand, an
index. These indices are evaluated in order to dynamically assess observation on the maximal element enables one to better under-
a port’s hub performance and potential in the model. stand the minimal element. This set of relations between bigger
systems and the smaller systems they contain is the mappings
The port connectivity index is evaluated by counting the line that constitute the framework of uniformity (McMaster, 1996).

Port network
Port accessibility Service
schedules

Port connectivity index and port collaboration


index

NHPA model (hub port assessment model)

Key operating dimensions of port

Monitored port statistics


Factor analysis
Port attributes

Figure 1. NHPA modelling process

3
Transport Hub port potential of Marmara region in
Turkey by network-based modelling
Bolat and Guler

Offprint provided courtesy of www.icevirtuallibrary.com


Author copy for personal use, not for distribution

Based on these mappings, there is no problem in performing the connectivity index of port i as Si , in other words as a dividend of
mentioned reductions on the assessments of Low et al. (2009). the Dep–Des pairs served by port i, as

As seen in Table 1, ports with a high connectivity index are taken P


as regional hubs, but the sustainability of these ports depends on j 2(ni þ nij )(nj þ nij )
Si ¼ P P
the collaboration index. Ports with a low connectivity index but a 1: j i 2(ni þ nij )(nj þ nij )
high collaboration index are considered national hubs.

The calculation methods for port connectivity index were pre-


The port collaboration index, which refers to the competitiveness
sented by Low et al. (2009). In accordance with these calcula-
index between port i and port j can be calculated by the equations
tions, as shown in Figure 2, when a hub–secondary port network
given in the model proposed by Low et al. (2009). The intensity
comprising two independent ports i and j is taken into considera-
of the competitiveness between port i and port j can be expressed
tion, two different point of departure and point of destination
as the ratio of the Dep–Des pairs they obtain, independently from
pairs (Dep–Des) are designated.
each other in the combined network (Low et al., 2009). There-
fore, the competitiveness index between port i and port j can be
The accessibility of port i is calculated by the expression
P calculated as
j 2(ni þ nij )(nj þ nij ), which is a variation of the integrated
accessibility index defined by Taylor et al. (2006); n i and n j are
given as the number of major ports that are accessible from port i nij (ni þ nj þ nij )
C ij ¼
and port j (including the ports themselves). There is a certain 2: (ni þ nij )(nj þ nij )
number of common nodes that can be accessed via port i and port
j, represented as n ij . Therefore, we express the accessibility or
In this study, these two indices were calculated for ports in the
Marmara region based on the NHPA model. The assessment of
Connectivity index the connectivity index and collaboration index was conducted
through a comparison of port features for evaluating the potential
Low High of the Marmara region ports to be a hub.

Concentration High Sustainable Sustainable 3.2 Marmara region ports and networks
index national port, regional hub In this study, the term ‘port’ is treated as a port area in order to
potential for port status obtain a commercial framework. There are five different ports in
regional hub status the Marmara region – Ambarli, Gemlik, Istanbul, Izmit and
Low Unstable Unstable Tekirdag – as illustrated in Figure 3. These ports include a total
national hub regional hub of nine container-handling terminals that are operated by public
port port and private entities. Information about the ports is given in the
appendix.
Table 1. Assessment of port using collaboration and connectivity
indices
3.3 Data collection
The empirical analysis of the Marmara region ports was con-
ducted using vessel movement data obtained from the Under-
secretariat of Maritime Affairs of Turkish Republic (UMATR).
The data sets in this study were from the income–procedure
inventories of the UMATR. The data cover the departure and
destination ports used by carriers each year between 2004 and
2008. Information on whether there were service lines between
these ports and container ports in the Marmara region (Ambarli,
Port i Port j Gemlik, Istanbul, Izmit and Tekirdag) for the years 2004–2008
were inserted into the data sets and the networks obtained are
illustrated in Figures 4–8. The number of arrows indicates the
number of destination points from the departure port, enabling
determination of n i , n j and n ij for Equations 1 and 2.

4. Results
Figure 2. Hub–secondary port configuration of port i and port j Based on the aforementioned data, all ports were compared with
each other with regard to connectivity and collaboration indices

4
Transport Hub port potential of Marmara region in
Turkey by network-based modelling
Bolat and Guler

Offprint provided courtesy of www.icevirtuallibrary.com


Author copy for personal use, not for distribution

Black Sea

Istanbul Strait
Tekirdag port
Ambarli port
Istanbul port

Marmara Sea
Aegean Sea
Izmit port

Canakkale Strait Gemlik port

86 km

Figure 3. Marmara region ports (# Googlemaps)

2004

Finland

Norway Poland Ukraine


United Kingdom Netherlands Russian Federation
Slovenia Bulgaria Romania Georgia
Belgium
Albania
Germany
Trabzon
Greece AMBARLI
France TEKIRDAG ISTANBUL
Canada Italy
China
Spain IZMIT
Karabiga
Israel
USA
Izmir GEMLIK Malaysia
Syria
Malta Mersin
Brasil Lebanon Hong Kong
Liberia
Tunisia Algeria Saudi Arabia
Libya Egypt
Bahamas
Singapore

Figure 4. Network structure of Marmara region ports in 2004

given by Equations 1 and 2. For example, to compare Ambarli After obtaining these parameters for each port (Table 2), the port
with the other ports, three variables were required connectivity index and collaboration index were calculated for
each port between the years 2004–2008 by using Equations 1 and
j the number of destinations from only Ambarli port (n i ¼ N(I)) 2. The calculated results are presented in this paper via radar
j the number of destinations from only the compared port plots and graphics.
(Izmit, Gemlik, Istanbul or Tekirdag) (n j ¼ N(J))
j the number of common destinations of Ambarli and the The radar plot of port connectivity weighted indices of the ports,
compared port (n ij ¼ N(I, J)). shown in Figure 9 for each year, is a key indicator of the

5
Transport Hub port potential of Marmara region in
Turkey by network-based modelling
Bolat and Guler

Offprint provided courtesy of www.icevirtuallibrary.com


Author copy for personal use, not for distribution

2005

Ukraine
United Kingdom Russian Federation
Slovenia
Yugoslavia Bulgaria Romania Georgia Mongolia
Belgium
Germany Albania
Trabzon Sri Lanka
Greece AMBARLI
France TEKIRDAG
Canada ISTANBUL China
Italy
Spain IZMIT
Karabiga South Korea
Israel
USA
Izmir GEMLIK
Portugal Syria
Malta Mersin
Lebanon
Hong Kong
Tunisia Algeria Saudi Arabia
Libya
Egypt
Equador Singapore

Figure 5. Network structure of Marmara region ports in 2005

2006

Netherlands Ukraine
United Kingdom Russian Federation
Slovenia Romania
Bulgaria Georgia
Belgium Croatia
Germany Albania Turkmenistan China
Trabzon
Greece AMBARLI
France TEKIRDAG ISTANBUL
Canada Italy South Korea
Spain IZMIT
Karabiga
Israel
Sri Lanka
Izmir GEMLIK
USA Portugal Syria Malaysia
Malta
Mersin
Lebanon Hong Kong
Argentina Tunisia Saudi Arabia
Algeria
Libya
Egypt
Singapore

Figure 6. Network structure of Marmara region ports in 2006

accessibility. The port collaboration indices for the five ports are container handling was low until 2008, but then container
presented in Figure 10. amounts received from other ports increased in 2008. Therefore,
while the collaboration index was high before 2008 (i.e. the need
Investigating the Marmara region ports areas using Figures 9 and for lines with other ports was high and it seemed that it could
10 and Figures 11–15 shows that Ambarli has the highest overcome more lines in the future), in 2008 there had been a
connectivity index for most years, but its low collaboration direct fall in the collaboration index. On the other hand, Tekirdag
indices indicate that collaboration of this port with other ports is gained the potential to become a main hub due to the increase in
not productive in terms of sustainability. Istanbul, Izmit and its connectivity index.
Gemlik are in a rather stable state in terms of both collaboration
and connectivity indices. Tekirdag port area was a port where Investigation of the 2004 data indicates that the Ambarli port

6
Transport Hub port potential of Marmara region in
Turkey by network-based modelling
Bolat and Guler

Offprint provided courtesy of www.icevirtuallibrary.com


Author copy for personal use, not for distribution

2007
Sweden

Netherlands Ukraine
United Kingdom Slovenia Russian Federation
Bulgaria Romania
Belgium Slovak Republic Georgia China
Germany Kazakhstan
Trabzon South Korea
France
Greece TEKIRDAG AMBARLI
Canada Italy ISTANBUL
IZMIT Sri Lanka
Spain
Karabiga
Israel
Portugal
USA
Izmir GEMLIK Syria Malaysia
Malta Antalya Mersin Japan
Lebanon
Hong Kong
Tunisia Algeria Saudi Arabia
Libya Senegal Egypt
Equador Singapore

Figure 7. Network structure of Marmara region ports in 2007

2008
Finland

Netherlands Norway Poland


Ukraine
United Kingdom Russian Federation
Slovenia
Croatia Bulgaria Romania
Belgium Georgia
Turkmenistan
Germany Albania China
Trabzon
Greece AMBARLI
France TEKIRDAG South Korea
Canada Italy ISTANBUL
Sri Lanka
Spain IZMIT
Karabiga
Israel Philippines
Portugal Antalya
USA
Izmir GEMLIK Syria Malaysia
Colombia Malta Mersin
Lebanon
Venezuela Liberia Hong Kong
Morocco
Tunisia Algeria Saudi Arabia
Libya Equador
Bahamas Egypt Singapore
South Africa Sudan UAE

Figure 8. Network structure of Marmara region ports for 2008

region shows regional hub potential rather than national hub Conversely, the connectivity and collaboration indices of Gemlik
potential. The connectivity index is higher than that of the other were stable over the years. The mid-level connectivity and
ports but its collaboration index is lower. A low collaboration collaboration indices show that this port has a higher potential
index signals that it is not sustainable and the property of being a to become a regional hub than a national one. The stability of
regional hub will be lost in the future in terms of capacity. this port has been proven in terms of sustainability and, by
Although the potential to be a regional main hub increased in 2005 increasing the collaboration index and pegging the connectivity
and 2006, the sustainability decreased as the collaboration index index, the port increased its potential to become a sustainable
remained the same or decreased. The 2007 and 2008 data indicate regional hub.
that the potential of Ambarli to be a regional main hub decreased
gradually, but its sustainability potential remained the same. When the connectivity and collaboration indices of Istanbul port

7
Transport Hub port potential of Marmara region in
Turkey by network-based modelling
Bolat and Guler

Offprint provided courtesy of www.icevirtuallibrary.com


Author copy for personal use, not for distribution

Ports Years
N(I)
N(J) 2004 2005 2006 2007 2008
N(I, J)

Ambarli–Gemlik
N(Ambarli) 9 12 14 11 12
N(Gemlik) 5 2 4 2 5
N(Ambarli, Gemlik) 27 23 24 28 30
Ambarli–Istanbul
N(Ambarli) 14 14 13 17 16
N(Istanbul) 4 3 1 1 1
N(Ambarli, Istanbul) 22 22 25 22 26
Ambarli–Izmit
N(Ambarli) 19 19 24 20 11
N(Izmit) 3 3 1 1 9
N(Ambarli, Izmit) 17 16 14 19 31
Ambarli–Tekirdag
N(Ambarli) 29 30 36 37 26
N(Tekirdag) 1 1 1 1 1
N(Ambarli, Tekirdag) 7 5 3 3 16
Gemlik–Istanbul
N(Gemlik) 12 9 9 13 12
N(Istanbul) 7 7 7 6 4
N(Gemlik, Istanbul) 19 17 19 17 23
Gemlik–Izmit
N(Gemlik) 16 12 17 13 11
N(Izmit) 6 5 3 3 15
N(Gemlik, Izmit) 15 14 11 17 24
Gemlik–Tekirdag
N(Gemlik) 26 22 27 28 20
N(Tekirdag) 3 2 2 1 2
N(Gemlik, Tekirdag) 5 4 2 3 15
Istanbul–Izmit
N(Istanbul) 11 13 15 10 5
N(Izmit) 6 7 4 7 17
N(Istanbul, Izmit) 15 12 12 14 22
Istanbul–Tekirdag
N(Istanbul) 21 22 25 21 14
N(Tekirdag) 3 3 2 1 4
N(Istanbul, Tekirdag) 6 3 2 3 13
Izmit–Tekirdag
N(Izmit) 16 14 14 18 23
N(Tekirdag) 3 1 2 1 1
N(Izmit, Tekirdag) 6 5 2 3 17

Table 2. Three variables of the ports for calculation of port


connectivity and collaboration indices

between 2004 and 2008 are analysed, it can be seen that this port connectivity index of Izmit increased after 2006, the potential of
has the potential for being both a regional and national hub. That Istanbul being a regional hub gives way to potential of being a
is, when the 2004 and 2005 data are examined, both indices national hub. The connectivity and collaboration indices in 2008
increased, and Istanbul therefore had more potential for being a indicate that Istanbul had the potential of being an unstable
sustainable regional hub. However, as the sustainability and national main hub.

8
Transport Hub port potential of Marmara region in
Turkey by network-based modelling
Bolat and Guler

Offprint provided courtesy of www.icevirtuallibrary.com


Author copy for personal use, not for distribution

2004 2005
Ambarli Ambarli
0·30 0·30
0·25 0·25
0·20 0·20
0·15 0·15
0·15 0·15
Tekirdag 0·10 Gemlik Tekirdag 0·10 Gemlik
0·05 0·05
0 0

Izmit Istanbul Izmit Istanbul

2006 2007
Ambarli
Ambarli
0·30
0·4 0·25
0·3 0·20
0·15
0·2 0·15
Tekirdag Tekirdag 0·10 Gemlik
0·1 Gemlik 0·05
0 0

Izmit Istanbul Izmit Istanbul

2008
Ambarli
0·25
0·20
0·15
0·10
Tekirdag Gemlik
0·05
0

Izmit Istanbul

Figure 9. Network characterisation with port connectivity index

Izmit had the potential of being a national main hub with areas. Therefore, it can be said that this port area was under
unstable sustainability between 2004 and 2007. However, the capacity between 2004 and 2007: it has the capacity to serve
situation changed in 2008 since the connectivity index of Ambarli more container transport and its potential for being a national
decreased and a direct rise was seen in the connectivity indices of hub can increase. The data from 2008 seem to support this
Tekirdag and Izmit. Izmit gained the potential of being a argument.
sustainable regional main hub by 2008; the authors do not have
full data for the following years to understand whether this 5. Conclusion and future scope of work
situation is permanent or not and therefore cannot state that Izmit The main aim of this study was to assess whether Ambarli,
could be a sustainable regional hub. Tekirdag port appears to have Gemlik, Istanbul, Izmit and Tekirdag port regions have the
potential for being a national hub with high sustainability be- potential to be hub ports using the NHPA model. Container ship
tween 2004 and 2007. The high sustainability can be inferred activities in the ports that handle containers in the Marmara
from the high collaboration index compared with the other port region were investigated using scientific methods. Their current

9
Transport Hub port potential of Marmara region in
Turkey by network-based modelling
Bolat and Guler

Offprint provided courtesy of www.icevirtuallibrary.com


Author copy for personal use, not for distribution

2004 2005
Ambarli Ambarli
1·0 1·0
0·8 0·8
0·6 0·6
0·4 Tekirdag 0·4
Tekirdag Gemlik Gemlik
0·2 0·2
0 0

Izmit Istanbul Izmit Istanbul

2006 2007
Ambarli Ambarli
2·0 1·0
1·5 0·8
1·0 0·6
Tekirdag Gemlik Tekirdag 0·4 Gemlik
0·5 0·2
0 0

Izmit Istanbul Izmit Istanbul

2008
Ambarli
0·30
0·25
0·20
0·15
Tekirdag 0·10 Gemlik
0·05
0

Izmit Istanbul

Figure 10. Network characterisation with port collaboration


index

conditions were determined, their activities were assessed using economy, activity and convenience, Ambarli comes out first in
the network-based model and port regions with the potential of the Marmara region. The high connectivity index for this port as
being main hubs were identified. Evaluation of the Marmara calculated by the network structure shows that the criteria consid-
region ports in terms of connectivity and collaboration indices is ered in this study have a direct effect on the connectivity index.
shown in Table 3. In addition, the constant low level of the port’s sustainability,
namely its low collaboration index, agrees with the arguments put
By examining the Marmara region ports in order to select hub forward by Barke (1986).
area, the port area that would be the most productive hub can be
determined by evaluating port connectivity indices in terms of Barke (1986) argues that port areas that expand and develop
both network structure and port features. In terms of scale rapidly begin to suffer from overload; some port services will

10
Transport Hub port potential of Marmara region in
Turkey by network-based modelling
Bolat and Guler

Offprint provided courtesy of www.icevirtuallibrary.com


Author copy for personal use, not for distribution

Ambarli Izmit Ambarli Izmit


Gemlik Tekirdag Gemlik Tekirdag
Istanbul Istanbul
1·2 1·0
0·9

Collaboration index
Collaboration index

1·0 0·8
0·7
0·8 0·6
0·6 0·5
0·4
0·4 0·3
0·2
0·2 2004 0·1 2007
0 0
0 0·05 0·10 0·15 0·20 0·25 0·30 0 0·05 0·10 0·15 0·20 0·25 0·30 0·35
Connectivity index Connectivity index

Figure 11. Connectivity and collaboration indices of Marmara Figure 14. Connectivity and collaboration indices of Marmara
region ports in 2004 region ports in 2007

Ambarli Izmit
Gemlik Tekirdag
Ambarli Izmit
Istanbul
1·2 Gemlik Tekirdag
Collaboration index

1·0 0·35 Istanbul


Collaboration index

0·8 0·30
0·6 0·25
0·4 0·20
0·15
0·2 2005 0·10
0 0·05
0 0·05 0·10 0·15 0·20 0·25 0·30 2008
Connectivity index 0
0 0·05 0·10 0·15 0·20 0·25 0·30
Connectivity index
Figure 12. Connectivity and collaboration indices of Marmara
region ports in 2005 Figure 15. Connectivity and collaboration indices of Marmara
region ports in 2008

Ambarli Izmit
Gemlik Tekirdag
Istanbul
1·8
1·6 Connectivity index
Collaboration index

1·4
1·2 Low High
1·0
0·8
0·6 Concentration High Tekirdag port Gemlik port
0·4 index Low Izmit and Istanbul Ambarli port
0·2 2006
0
ports
0 0·05 0·10 0·15 0·20 0·25 0·30 0·35
Connectivity index Table 3. Evaluation of Marmara region ports in terms of port
collaboration and connectivity indices
Figure 13. Connectivity and collaboration indices of Marmara
region ports in 2006
increased, but its connectivity index decreased, while the
connectivity indices of Izmit and Tekirdag increased. There-
then leave the city centres and move to suburban port systems fore, Ambarli does not have the features of being a productive
or other port systems in the area. Examining Ambarli, its national or regional main hub. Istanbul has been gradually
connectivity index was too high and its sustainability too low decreasing in terms of connectivity. The port shares its
between 2004 and 2007. In 2008, the sustainability of Ambarli operational volume with other ports and the decrease in the

11
Transport Hub port potential of Marmara region in
Turkey by network-based modelling
Bolat and Guler

Offprint provided courtesy of www.icevirtuallibrary.com


Author copy for personal use, not for distribution

annual number of Port calls directly affected the connectivity potential to be a national main hub, but its improvement depends
index. on the possibility of it increasing its connectivity index and
sustainability.
It is also possible to see the effects of the Belde port and
Nato Marmara Ereglisi port projects on Ambarli and Istanbul. For Gemlik, the stability of its scale economy, activities and its
In Belde, Liman Co. has completed 450 m of port area in convenience status were reflected in the port’s connectivity and
Dilovasi, and the annual capacity is 350 000 containers per collaboration indices. Gemlik had the properties of being a
year. The planned length is 960 m and the planned capacity sustainable regional hub between 2004 and 2008. The increase in
of the port is 750 000 containers per year; it is estimated to its collaboration index in 2008 while its connectivity index
be the second biggest port in the eastern Mediterranean remained the same shows that the port expanded its capacity. The
outside Italy and to replace Istanbul. The port in Marmara scale economy, activity and convenience of the port have increased
Ereglisi, where construction started in the beginning of 2000 and the port has gained a high potential to increase collaboration.
with a request from Nato which was later retracted, still Gemlik, however, lacks combined links with its hinterland. If
remains idle but could be reactivated via a build–operate combined links to the hinterland were to be added to the current
transfer model. This port could handle half the load of condition, Gemlik’s connectivity index would increase and its
Istanbul and would thus have a great influence on logistics potential to be a sustainable main hub would further increase.
and container transport.
Acknowledgements
There is the possibility that the rapid increase in the connectivity This work was completed at the Institute of Science and
and collaboration indices of Izmit port in 2008 was related to an Technology at Istanbul Technical University and is part of the
increase in scale economy. However, this could be a one-year first author’s doctoral dissertation work. The authors appreciate
situation and the unstable condition of Izmit port between 2004 the guidance provided by Professor Dr Ilhan Avci of Istanbul
and 2008 means it cannot be defined as the most convenient port Technical University.
as a main hub. Regarding Tekirdag, its prospects are better if
evaluated by container activity in 2008. Increases in activity and Appendix: Ports of the Marmara region
scale economy were reflected in the connectivity index as a result Ambarli port is the biggest port area in Istanbul, with three main
of the increase in container handling. The connectivity index container-handling terminals – Kumport, Mardas and Marport. It
increased and the collaborative index decreased as a result of has an effective hinterland, shown in Figure 16, home to
instant and active use of this port. Tekirdag port will develop its 20 million people. The location of Ambarli is an advantage for a

21·7 km

Figure 16. Hinterland of Ambarli port (# Googlemaps)

12
Transport Hub port potential of Marmara region in
Turkey by network-based modelling
Bolat and Guler

Offprint provided courtesy of www.icevirtuallibrary.com


Author copy for personal use, not for distribution

Kumport Mardas Marport

Port operator (public/private) Private Private Private


Container berth length: m 1930 910 1950
Port area: m2 400 000 194 224 409 000
Coordinates 408589N, 288419E 408579N, 288409E 408599N, 288329E
Maximum port draught: m 15.5 15.0 14.5
Container dock equipment capacity: teu/yeara 1 000 000 550 000 1 700 000
Realised handling in 2008: teu 660 989 387 447 1 203 101
Capacity utilisation rate: % 66.1 70.4 70.1
2008 port handling: % 29.0 17.2 53.4

teu ¼ twenty-foot equivalent unit (20 ft ¼ 6.1 m).

Table 4. Techno-economic data for Ambarli port

port area as it is a key point in the intersection of trade routes economic data for the port are given in Table 5 (UMATR,
and highways. Technical and economic data for Ambarli port are 2009).
given in Table 4 (UMATR, 2009).
Haydarpasa is the only container-handling terminal of Istanbul
Gemlik port hosts the two container-handling terminals of port that is publicly owned. Railway transportation is combined
Gemport and Borusan. Gemlik is a natural gate for the north- with road and sea transportation modes in this port. Due to space
eastern part of Turkey where population and asset levels are constraints, delays in container service occur. However, the num-
high. Accordingly, it is an effective port area for Marmara, ber of containers in Istanbul has increased in recent years because
Anatolia and the North Aegean. Additionally, its location is ships have been directed to Istanbul instead of Ambarli, as
favourable as it is linked to the national highways strategically Ambarli port frequently reaches peak levels. The hinterland of
reinforcing north-east Anatolia and the North Aegean. The Istanbul port is shown in Figure 18 and Table 6 gives techno-
hinterland of Gemlik port is illustrated in Figure 17 and techno- economic data (UMATR, 2009).

17·2 km

Figure 17. Hinterland of Gemlik port (# Googlemaps)

13
Transport Hub port potential of Marmara region in
Turkey by network-based modelling
Bolat and Guler

Offprint provided courtesy of www.icevirtuallibrary.com


Author copy for personal use, not for distribution

Borusan Gemport

Port operator (public/private) Private Private


Container berth length: m 780 845
Port area: m2 54 547 700 000
Coordinates 408259N, 298059E 408259N, 298079E
Maximum port draught: m 11 36
Container dock equipment capacity: teu/year 300 000 600 000
Realised handling in 2008: teu 132 383 310 510
Capacity utilisation rate: % 44.1 51.8
2008 port handling: % .
28 6 67.0

Table 5. Techno-economic data for Gemlik port

18·3 km

Figure 18. Hinterland of Istanbul port (# Googlemaps)

Izmit port has two container terminals, namely Evyap and


Haydarpasa Yilport. These terminals are located 80 km east of Istanbul Strait
and they have links to highways. Izmit port is in the centre of the
Port operator (public/private) Public
biggest industrial region of Turkey. It is also the closest port to
Container berth length: m 1142
Ankara, Turkey’s capital. The hinterland of Izmit port is shown in
Port area: m2 398 420
Figure 19 and techno-economic data are presented in Table 7
Coordinates 408599N, 288579E
(UMATR, 2009).
Maximum port draught: m 12
Container dock equipment capacity: 300 000
Akport is the only container-handling terminal of Tekirdag port.
teu/year
As Tekirdag is the closest port to Bulgaria and Greece, it will
Realised handling in 2008: teu 342 519
gain importance in the event of an increase in its capacity. The
Capacity utilisation rate: % 100
main service area of Tekirdag port lies in industrial areas in the
2008 port handling: % 99.5
Thrace region of Turkey and Istanbul. The hinterland of Tekirdag
Table 6. Techno-economic data for Istanbul port port and its techno-economic indicators are shown in Figure 20
and Table 8 respectively (UMATR, 2009).

14
Transport Hub port potential of Marmara region in
Turkey by network-based modelling
Bolat and Guler

Offprint provided courtesy of www.icevirtuallibrary.com


Author copy for personal use, not for distribution

13.9 km

Figure 19. Hinterland of Izmit port (# Googlemaps)

Evyap Yilport

Port operator (public/private) Private Private


Container berth length: m 830 325
Port area: m2 214 000 202 500
Coordinates 408469N, 298429E 408469N, 298329E
Maximum port draught: m 16 30
Container dock equipment capacity: teu/year 615 000 300 000
Realised handling in 2008: teu 108 204 93 656

Table 7. Techno-economic data for Izmit port

19·6 km

Figure 20. Hinterland of Tekirdag port (# Googlemaps)

15
Transport Hub port potential of Marmara region in
Turkey by network-based modelling
Bolat and Guler

Offprint provided courtesy of www.icevirtuallibrary.com


Author copy for personal use, not for distribution

Akport Malchow M and Kanafani A (2004) A disaggregate analysis of


port selection. Transportation Research Part E 40(4): 317–
Port operator (public/private) Private 337.
Container berth length: m 2000 McMaster MD (1996) The Intelligence Advantage: Organizing for
Port area: m2 130 000 Complexity. Butterworth-Heinemann, Newton, MA, USA.
Coordinates 408579N, 278309E Ozdamar K (1999) Paket Programlar ile İstatistiksel Veri Analizi.
Maximum port draught: m 14 Kaan Kitabevi, Eskişehir, Turkey (in Turkish).
Container dock equipment capacity: 250 000 Rimmer PJ (1977) A conceptual framework for examining urban
teu/year and regional transport needs in South-East Asia. Pacific
Realised handling in 2008: teu 30 186 Viewpoint 18: 133–147.
Capacity utilisation rate: % 12 Rodrigue JP, Comtois C and Slack B (1997) Transportation and
2008 port handling: % 98.7 spatial cycles: evidence from maritime systems. Journal of
Transport Geography 5(2): 87–98.
Table 8. Techno-economic data for Tekirdag port Song DW (2003) Port competition in concept and practice.
Maritime Policy and Management 30(1): 29–44.
Song DW and Yeo KT (2004) A competitive analysis of Chinese
REFERENCES container ports using the analytic hierarchy process. Maritime
Barke M (1986) Transport and Trade. Oliver & Boyd Press, Economics and Logistics 6(1): 34–52.
Edinburgh, UK. Taaffe EJ, Morrill RLY and Gould PR (1963) Transport expansion
Cullinane K, Song DW and Gray R (2002) A stochastic frontier in underdeveloped countries: a comparative analysis.
model of the efficiency of major container terminals in Asia: Geographical Review 53(4): 503–529.
assessing the influence of administrative and ownership Taylor MAP, Sekhar SVC and D’Este GM (2006) Application of
structures. Transportation Research Part A 36(8): 743–762. accessibility based methods for vulnerability analysis of
Fleming D and Baird AJ (1999) Some reflections on port strategic road networks. Networks and Spatial Economics
competition in the United States and western Europe. 6(3–4): 267–291.
Maritime Policy and Management 26(4): 383–394. Tiwari P, Itoh H and Doi M (2003) Shippers’ port and carrier
Inal EM, Topuz D and Ucan O (2006) Doğrusal olasilik ve logit selection behaviour in China: a discrete choice analysis.
modelleri ile parametre tahmini. Sosyoekonomi 2: 47–72 (in Maritime Economics and Logistics 5(1): 23–39.
Turkish). UMATR (Undersecretariat for Maritime Affairs of Turkish
Lattin J, Carroll DJ and Green PE (2003) Analyzing Multivariate Republic) (2009) Kabotaj Taşimaciliği Saha Etüdü. UMATR,
Data. Duxbury Press, Pacific Grove, CA, USA. Ankara, Turkey (in Turkish).
Low JMW, Lam SW and Tang LC (2009) Assessment of hub status Wang TF and Cullinane KPB (2006) The efficiency of European
among Asian ports from a network perspective. container terminals and implications for supply chain
Transportation Research Part A 43(6): 593–606. management. Maritime Economics and Logistics 8(1): 82–99.

WHAT DO YOU THINK?


To discuss this paper, please email up to 500 words to the
editor at journals@ice.org.uk. Your contribution will be
forwarded to the author(s) for a reply and, if considered
appropriate by the editorial panel, will be published as a
discussion in a future issue of the journal.
Proceedings journals rely entirely on contributions sent in
by civil engineering professionals, academics and students.
Papers should be 2000–5000 words long (briefing papers
should be 1000–2000 words long), with adequate illustra-
tions and references. You can submit your paper online via
www.icevirtuallibrary.com/content/journals, where you
will also find detailed author guidelines.

16
View publication stats

You might also like