Content uploaded by Dimas Dharmowijoyo
Author content
All content in this area was uploaded by Dimas Dharmowijoyo on Jul 01, 2015
Content may be subject to copyright.
Asian Transport Studies, Volume 1, Issue 1 (2010), 18-32
© 2010 ATS All rights reserved
18
Policy Evaluation of Multimodal Transportation Network: The Case of
Inter-island Freight Transportation in Indonesia
Ade SJAFRUDDIN
a
, Harun Al-Rasyid S. LUBIS
b
, Russ Bona FRAZILA
c
,
Dimas B. DHARMOWIJOYO
d
Transportation Research Group, Institut Teknologi Bandung (ITB),
Jl. Ganesha 10 Bandung – 40132, Indonesia
a
ades@trans.si.itb.ac.id
b
halubis@si.itb.ac.id
c
frazila@si.itb.ac.id
d
dimas@trans.si.itb.ac.id
Abstract: This paper outlines the development of an analytical model of multimodal, multi-
commodity freight transportation in Indonesia. The model is used to evaluate strategic
planning of inter-island freight transportation. The demand for inter-island freight
transportation is estimated using econometric demand models which account for the socio-
economic characteristics of the regions and commodity types. The supply side is concerned
with modeling the inter-island transportation network, including links, such as the land access
network, and transfer points which are connected by the links, and a generalized cost function.
System optimization is obtained by identifying the combination of links and transfers that
minimize the total generalized transport cost. The model is applied to the evaluation of the
impacts of several policy scenarios which contribute to the improvement of the inter-island
freight transportation network. The model has shown its capability of estimating system costs of
transportation network that can be utilized to assess the policy implications.
Keywords: freight transportation, multimodal, system optimum, regional network
1. INTRODUCTION
Sea freight transportation network plays a vital role in Indonesian economy since the country
covers over 5 million square kilometers area with over 17,000 islands. Since 1980s the annual
growth of freight transportation ranged 10-20% by railways, 5-10% by air, and 10-15% by sea.
Connectivity and efficiency of the existing inter-island transportation network, however, has
faced serious problems. This has led to the high cost of transportation that in turn affects the
competitiveness of the economy.
With respect to inter-island freight transportation network, the situation is more
problematic as transport of commodities normally involves two or more different modes. At
the transfer points, problems related to loading and unloading of commodities, scheduling,
warehousing, different destinations, and so forth are of significant interest.
This research aimed to develop a methodological approach to evaluate optimum freight
transportation network in terms of a generalized transport cost. A model is formulated in such
a way that the network and the demand are interrelated to each other. This is so that the model
will be capable of evaluating transport policy changes. A multimodal system and the cost
model are developed based on the respective characteristics of each mode. STAN (Strategic
Transportation ANalysis, INRO Consultants, 1997) is the software used in this research, since
it can be used to model multimodal transportation network comprehensively and carry out a
strategic planning model.
SJAFRUDDIN, A., et al. / Asian Transport Studies, Volume 1, Issue 1 (2010), 18-32
19
In more detail, the research objectives are stated as follows:
• to develop an inter-island freight transportation demand model by commodity
types;
• to develop a generalized cost model for inter-modal freight transportation;
• to develop a method to analyze the optimality of the freight transportation network
based on costs;
• to conduct policy analysis of alternatives for strategic development of freight
transportation network in Indonesia.
Previous researches on freight transportation in Indonesia have been conducted.
Sjafruddin et al. (1997, 1999) developed regional freight transportation demand models for
various transportation modes. These models estimated the demand for intercity freight
transportation with region’s social economic factors and transport systems attributes. Factors
identified as significant in generating freight transport demand include population, GRDP,
GRDP per capita, proportion of industrial sector product, production surplus, and deficit. The
calibration of these models, however, was based upon the quantities of total commodities
without commodity classification and this limits its use for a policy evaluation. Prasetyo
(1999) developed an approach to analyze multi-mode multi-commodity freight transportation
network for Java Island with an emphasis on modeling of transfer points. He specified an
appropriate way to represent a multimodal network and to model a transfer point in the
network model. Prasetyo applied the methodology in evaluating some land transportation
network development in the Java Island. His findings, however, were limited as the model
only represents total commodity flow. Hermawan (1999) developed mode choice models of
intercity freight transportation on Java Island. The model choice model was calibrated for two
modes, namely truck and train, based on a binary logit model using stated preference data.
The resulting models were regarded as reasonable enough to be applied for evaluating policy
scenarios, but the methodology did not model the transfer point explicitly. Lubis et al. (2002,
2003) examined the optimization of national freight transportation network based on a
multimode multi-commodity approach and utilized the STAN software. Eastern Indonesia
was studied with respect to its transportation development plans including port system
improvement and a new regional railway network on Kalimantan Island. The demand was
estimated based on the available inter district O-D data and expanded for the future using a
growth factor. This research made conclusions about the preferable development scenario
based on network performances and benefit-cost analysis. Frazila (2005) developed regional
freight transportation network model for Indonesia utilizing a genetic algorithm approach. He
derived detailed cost parameters for the link cost functions and transfer time estimation which
is assumed consisting of loading-unloading time and affected by capacity at transfer point.
Furthermore, a genetic algorithm procedure was developed to model the optimal flows in the
network. Crainic et al. (1990) reported findings related to the application of STAN for
strategic transportation analysis in Brazil. This research developed a methodology to establish
the cost function and the assignment procedure for the regional commodity flow and shown
the capability of the model to evaluate strategic planning of multimodal, multi-commodity
freight transportation network.
2. RESEARCH OUTLINE
The analysis covers following aspects:
• the pattern of the demand for regional freight transportation in Indonesia;
SJAFRUDDIN, A., et al. / Asian Transport Studies, Volume 1, Issue 1 (2010), 18-32
20
• the existing system of freight transportation network consisting of inland links, transfer
points, inter-island links;
• the development of the system to cope with the growth of demand for inter-island freight
transportation.
Figure 1 describes the general outline of the steps and aspects of the research.
Figure 1. Research outline
3. MODELING APPROACH
3.1 Demand Model
An econometric model utilizing variables describing characteristics of the freight and/or
regional conditions is employed. The resulting models are therefore sensitive to social-
economic changes that are represented by the variables.
The total amount of commodity movement between a pair of cities is assumed to be a
function of certain social economic characteristics of these cities. The general form of the
demand model is specified as follows:
Model 1:
γβ
α
).().(
jijiijk
YYXXT =
(1)
Model 2:
β
α
).(
jiijk
XXT =
(2)
Model 3:
γ
α
).(
jiijk
YYT =
(3)
Land Use System National Transport
System
Freight Transport
System
Network System
Evaluation of
Network Performance
Freight Transport
Pattern
Freight Transport
Demand Model
Alternative
Policy Scenario
Simulation
Networ
k Dbase
STAN
Socio-Economic
Condition
SJAFRUDDIN, A., et al. / Asian Transport Studies, Volume 1, Issue 1 (2010), 18-32
21
where:
T
ijk
: volume of commodity k produced in zone i and transported to zone j, in
tons/year,
X
i
, X
j
: number of population of zone i and zone j, in 1,000 people,
Y
i
, Y
j
: GRDP (Gross Regional Domestic Product), total or industrial sector, of zone
i and zone j, in 10
9
Rupiah, and
γ
β
α
,,
: model parameters.
Considering that the industrial sector induces a very significant impact to the flow of
commodities from and to a region, GRDP of industrial sector was also utilized as an
alternative variable in spite of total GRDP. The model, which is an aggregate model and
utilizes aggregated data, is considered useful as a tool to evaluate impacts of policy changes to
the existing system rather than to generate absolute estimates of freight tonnage.
3.2 Network Model
The basic components of the modeling framework and related data requirement are illustrated
in Figure 2. The network represents the infrastructure and services that form the supply side,
mode represents how activities are carried out, node and link represent the spatial structure of
the transport system, and transfer shows operational characteristics of interchange between
modes. The demand side is identified as products or groups of products, production and
consumption rate per unit area of analysis, and the demand for respective commodities
transported from one place to another.
(Source: INRO Consultants, 1997)
Figure 2. Main model components, STAN (Strategic Transportation ANalysis)
Network
Mode
Zone Centroid
Node
Link
Transfer
Connector
Route
Demand
Zone
Product
Matrix
Platoon
Function
Unit cost function
Multi mode, multi commodity
assignment
System optimum
Flow and cost prediction
SJAFRUDDIN, A., et al. / Asian Transport Studies, Volume 1, Issue 1 (2010), 18-32
22
The simulated network consists of a number of elements. Details of the elements are
shown in Figure 3. The main elements consist of mode, node and link, vehicle, transfer, and
functions that follow a particular structure and are defined hierarchically. Every element is
specified by certain attributes that determine how they interact to each other.
A simple illustration of a network is shown in Figure 4. In this figure two modes are
indicated, namely road and rail. City A to B is served by all modes, whilst A to C and B to C
are served by road only. In the simplified representation all cities are connected each other by
direct links and modes are allocated as attributes of links.
The transfer deserves special attention. Transfers are defined as a separate process of
movement between a node and mode. Figure 5 shows a mode interchange and its
development at the node.
(Source: INRO Consultants, 1997)
Figure 3. Network elements in STAN, hierarchy and attributes
3.3 Cost Function and Assignment Procedure
The network optimization model that is used to simulate network flows in STAN is a
nonlinear multiproduct multimode assignment formulation that minimizes the total
generalized system cost of shipping the products considered, from origins to destinations, via
the permitted modes, while satisfying the usual flow conservation and non-negativity
Centroid & Node
Standard Attribute
Additional Attribute
Mode
Standard Attribute
Additional Attribute
Product
Standard Attribute
Additional Attribute
Link
Standard Attribute
Additional Attribute
Results:
Volume
Unit cost
Marginal cost
Vehicle
Standard Attribute
Additional Attribute
Function Set
Link set function
Transfer set function
Transfer
Standard Attribute
Additional Attribute
Results:
Volume
Unit cost
Marginal cost
SJAFRUDDIN, A., et al. / Asian Transport Studies, Volume 1, Issue 1 (2010), 18-32
23
constraints. Capacity constraints and congestion phenomena on various transportation modes
are specifically considered by using volume/delay and penalty functions.
Figure 4. Real and simplified network representation
Figure 5. Node expansion to explicitly describe transfers between modes
The total generalized system cost to be minimized is:
total generalized system cost = total cost on links + total cost on transfers (4)
where the total costs of a product on a link (or transfer ) are:
total link costs = link total unit product cost x product volume (5)
total transfer costs = transfer total unit product cost x product volume (6)
3
4
2
1
1
2
3
4
Transfers (1.3) (1.4)
(2.3) (2.4)
Road
Rail
Real Network
A
B
C
Rail
Road
(a) Combined representation (b) Parallel representation
Simplified Network
Jl,
Rl
Jl,
Rl
Jl
Jl
Road
Rail
A
B
C
A
B
C
SJAFRUDDIN, A., et al. / Asian Transport Studies, Volume 1, Issue 1 (2010), 18-32
24
The link and transfer unit costs for a given product are a user-defined combination of up to
three functions:
s
pa
(v) = a x link “operating” cost (v) (1st cost function)
+ b x link “delay” cost (v) (2nd cost function)
+ c x link “other” cost (v) (3rd cost function) (7)
where other costs may be energy consumption, potential hazards, environmental impacts, or
others, and v denotes the flow of all products on the multimodal network.
The total cost of the flow on arc a, a Є A (a set of arcs), for the product p, p Є P (a set of
all products considered), is the product s
pa
(v) v
pa
, while the total cost of the flow on transfer t,
t Є T (a set of transfers), is s
pt
(v) v
pt
. We then have:
∑
∈
=
Pp
p
a
p
aa
vvsvs )()(
(8)
∑
∈
=
Pp
p
t
p
tt
vvsvs )()(
(9)
The link (transfer) function set indicates the functions associated with each product on
each link (transfer), where :
s
a
(v) : total cost of the flow on the link
s
t
(v) : total cost of the flow on the transfer point
s
pa
(v) : unit cost of product p on the link
s
pt
(v) : unit cost of product p on the transfer point
v
pa
, v
pt
: the flow of product p on the link, or on the transfer point
The total generalized system cost, that is the total cost of the flows of all products over the
multi-modal network, is the function F:
( ) ( )
∑ ∑ ∑
∈ ∈ ∈
+=
Pp Aa Tt
p
t
p
t
p
a
p
a
vvsvvsF
(10)
The multiproduct, multimode assignment model is carried out based on minimizing total
general system cost as defined in equation (10).
4. DATA AND MODEL PARAMETERS
4.1 Data
Data on inter-island commodity flows through ports were obtained from a domestic sea
transportation study (JICA, 2003). The JICA study specified the commodities into 13
categories (Petroleum, CPO, Other Liquid, Coal, Other Mines, Rice, Agri Grains, Fertilizer,
Cement, Other Grains, Fresh Products, Forestry Products, and Others (General Cargo) with
four packaging types, namely container, break bulk, dry bulk, and liquid bulk. The JICA study
estimates the annual origin-destination of commodity flows by ports.
SJAFRUDDIN, A., et al. / Asian Transport Studies, Volume 1, Issue 1 (2010), 18-32
25
Data on existing transportation networks were obtained from Ministry of Communication
and various transport operators. The data includes network condition, services attributes, and
fares. The data was used to derive parameters for the cost functions and variables of the link
and transfer services of the networks under consideration.
Indications of potential policies for transportation network development were obtained
from varying previous studies and plans. These include the development plans for the port
system, highways, toll roads, and railways in the areas under study.
4.2 Resulting Demand Model
The demand models for this research were developed for 5 commodity types, namely
CPO+Other Liquid, Rice+Agri Grains, Other Grains+Fresh Product, Forestry Product, and
Others (General Cargo). These commodities are common and associated with general
consumption by communities in these regions, and can be accommodated by most general
ports in Indonesia. Other commodities such as petroleum, coal, other mines, fertilizer, and
cement were not modeled as these are related to specific areas or industries and are normally
handled by special ports. O-D tables were built based on this data, and disaggregated across
42 zones covering entire area of Indonesia. Each zone represents a port hinterland covering
the area of one or more districts. The resulting O-D table for each commodity is then used to
calibrate the freight transportation demand model.
A number of combinations of explanatory variables are tested in the demand model
calibration. Based on some statistical tests and reasonableness of parameters, Table 1 presents
the best demand models derived by commodity type. These demand models are used to
estimate the inter-island freight transportation flows in the baseline year 2006.
Table 1. Inter-island freight demand model by commodity type
No
Commodity Model R
2
Variable
1 CPO + Other
Liquid
T
ij
= 0.01 x 10
-
3
(X
i
X
j
)
0
.
51
(Y
i
Y
j
)
0
.
69
0.66
Population,
GRDP Industry
2 Rice + Agri
Grains
T
ij
= 0.10 x 10
-
3
(X
i
X
j
)
0
.
45
(Y
i
Y
j
)
0
.
74
0.46
Population,
GRDP Industry
3 Other Grains +
Fresh Product
T
ij
= 2.40 x 10
-
3
(X
i
X
j
)
0
.
15
(Y
i
Y
j
)
0
.
63
0.64
Population,
GRDP Industry
4 Forestry
Product
T
ij
= 0.123
(X
i
X
j
)
0
.
45
(Y
i
Y
j
)
0
.
30
0.58
Population,
GRDP Industry
5 Others (General
Cargo)
T
ij
= 1.106
(X
i
X
j
)
0
.
28
(Y
i
Y
j
)
0
.
38
0.60
Population,
GRDP Industry
Note : X denotes number of population (in 1,000), Y denotes GRDP (in 10
9
Rp)
4.3 Cost Functions
Unit generalized cost is expressed as a function unit link cost, unit transfer cost, and the travel
time value of commodity, and specified as follows:
C
p
=
α
lp
+
α
tp
+
β
p
(t
l
+ t
t
) (11)
where:
C
p
= unit generalized cost (Rp/ton)
α
lp
= link unit cost (Rp/ton)
SJAFRUDDIN, A., et al. / Asian Transport Studies, Volume 1, Issue 1 (2010), 18-32
26
α
lp
= transfer unit cost (Rp/ton)
β
p
= value of time of product p (Rp/hr/ton)
t
l
= travel time at link (hr)
t
t
= transfer time at node (hr)
By assuming that a node or transfer point can be considered as a link (or transfer link), the
cost function at a transfer link can be specified as follows:
c
ap
=
α
p
l
a
+
β
p
t
a
(12)
where:
c
ap
: generalized cost (Rp/ton) at link a
α
p
: fare (Rp/ton/km) at a transfer link
β
p
: value of time (Rp/ton/hr)
t
a
: travel time at link (hr)
l
a
: length of link a (km), l
a
= 1 for transfer link
Based on the data obtained from transport operators, Frazila (2005) derived the following
cost parameters for the link cost function (which were used in this study):
1. Link parameters of roads:
α
p
= 0
,
β
p
= Rp 7 /hr/ton;
2. Link parameters of toll roads:
α
p
= Rp 200/ton/km
,
β
p
= Rp 8 /hr/ton;
3. Link parameters of sea :
α
p
= Rp 43.9 /ton/km
,
β
p
= Rp 5/hr/ton for container cargo,
β
p
= Rp 5/hr/ton for dry bulk cargo,
β
p
= Rp 6/hr/ton for break bulk / general cargo,
β
p
= Rp 5 hr/ton for liquid bulk;
Transfers identified in this study are the following:
• Transfer between land transportation and sea transportation; and
• Transfer between ferry and sea transportation.
Transfer time is assumed to consist of loading-unloading time and affected by the capacity
at the transfer point which in turn depends on the number of berths and delays during the
process. Following equation is used to estimate transfer time at a port (Frazila, 2005):
n
cd
vol
tbm
tt
o
⋅
−
+=
1
(13)
where:
t : time at port
t
o
: delay at port for inspection, administration, berthing, etc which is assumed 6 –
48 hr
tbm : loading-unloading time (hr)
vol : total loading-unloading flow (ton/hr)
SJAFRUDDIN, A., et al. / Asian Transport Studies, Volume 1, Issue 1 (2010), 18-32
27
cd : capacity of loading-unloading (ton/hr/berth)
n : number of berth (unit)
The value of time at transfer points is assumed to be equal to link value of time for all
packaging. Cost parameters for transfer points at ports are obtained based on the available
data as presented in Table 2.
Table 2. Cost at transfer point (port)
Description Fare /
α
p
(Rp./ton)
Unloading at major ports 2,023
Loading at major ports 1,025
Unloading at other ports 2,492
Loading at other ports 862.5
5.1 Simulated Policy Scenarios
Policy scenarios of network development in Java Island consist of:
A1. Existing condition, where the condition of the road network is much better than that
of other modes. The existing railway network connects most main cities and regions
in Java, but most links are single tracks and the level of service is considerably low.
A2. Trans Java Toll Roads Development (Figure 6) that fully connects all provincial
cities in Java and improves road network capacity especially in the northern part of
Java. This includes the existing toll roads plus 6 new links (Jakarta-Semarang,
Semarang-Demak, Demak-Solo, Solo- Surabaya, Surabaya-Malang, Gempol-
Banyuwangi).
A3. Java Railway Development, a capacity expansion by building double tracks railways
in Java including access links to ports, especially in the busy, northern part and
several links in the southern area where the majority of cities and districts are
connected by railways, and improvement of services is done to meet the demand.
Figure 6. Trans Java toll roads development
Policy scenarios of network development in Sumatera Island consist of:
B1. Existing condition, where the road network is used much more by traffic. The
existing railways in Sumatera are situated at three regions where railway networks
are unconnected to each other and the services are very limited.
B2. Improvement of Trans Sumatera Highway Network, especially in the west, east, and
middle routes.
SJAFRUDDIN, A., et al. / Asian Transport Studies, Volume 1, Issue 1 (2010), 18-32
28
B3. Development of a fully-connected Trans Sumatera Railways (Figure 7), consisting of
the improvement and construction of new links, to fully connect all provinces in
Sumatera and the improvement of services so as to meet the demand. The existing
railways in Sumatera are situated in 3 areas and are unconnected to each other; and
this development adds 13 new links of railways.
Policy scenarios for the inter-island network development consist of:
C1. Existing condition, where the land transportation in the existing condition and the
port system does not have a hierarchy of a hub-spoke system.
C2. Implementation of Hub-Spoke Ports with 15 hubs (Figure 8), namely Medan, Batam,
Palembang, Padang, Bengkulu, Jakarta, Surabaya, Cilacap, Pontianak, Banjarmasin,
Balikpapan, Bitung, Makassar, Ambon, and Sorong. This is to represent the policy
to assign a number of ports as hub that services longer distance inter-island
movement and every hub connects to several spokes as feeder.
C3. Implementation of Hub-Spoke Ports with 9 hubs (part of Figure 8), namely Medan,
Batam, Palembang, Jakarta, Surabaya, Banjarmasin, Makassar, Ambon, and Sorong.
C4. Development of Trans Java Toll Roads as in A2.
C5. Development of Fully Connected Trans Sumatera Railways as in B3.
BANDA ACEH
Lhokseumawe
Langsa
Besitang
0 50km
SKALA
BANDAR LAMPUNG
Tanjung Api-api
MEDAN
PEKANBARU
PADANG
BENGKULU
PALEMBANG
Binjai
Belawan
Tebingtinggi
P.Siantar
Kisaran
Tjg.Balai
Rantauprapat
Sibolga
Pdg.Sidempuan
D u m a i
D u r i
Pdg. Panjang
Pariaman
Solok
Muaro
Muaralembu
Telukkuantan
Rengat
Kuala Enok
Muarabungo
Muarabulian
JAMBI
Sengeti
Muarasabak
Bangko
Lbk.Linggau Tebingtinggi
Lahat
Muaraenim
Tanjungenim
Blimbing
Sekayu
Betung
Baturaja
Kotabumi
LEGENDA:
Ibukota Propinsi
Kota-kota lainnya
Jalan kereta api yang ada (eksisting)
Koridor jalan kereta api baru skema awal
SAMUDERA HINDIA
Figure 7. Trans Sumatera railways development
5.2 Simulation Results
SJAFRUDDIN, A., et al. / Asian Transport Studies, Volume 1, Issue 1 (2010), 18-32
29
The simulation was carried out to determine the optimum condition of the modeled policy
scenario based on the total (link plus transfer) multimodal transportation cost.
Figure 9 shows that the development of double tracks in the Java Railway Network
significantly reduces the system cost and this is much better than the impact of the
development of the Trans Java Toll Roads. This is mainly brought about by greater cost
efficiency of the railways. As the double-track railways become available, most bottlenecks of
inland transportation, as access to ports in Java, will be relieved and a substantial portion of
existing road freight traffic will be diverted to railways. The railway network is also covering
most areas of production centers in Java and the railway development will certainly provide
better direct access to inter-island ports. And, the benefit of using the railway will increase as
travel distance becomes longer. The Trans Java Toll Roads development will also improve the
inland transportation network, but its impact will not be as substantial as the railways
development since the toll roads are less cost effective than railways and mainly improve
capacity in the northern area.
Figure 8. Implementation of hub-spoke ports with 15 hubs
As seen in Figure 10 the Trans Sumatera Railway development will substantially reduce
the total system cost of inland freight transportation. This is mainly due to the high cost
efficiency of railways and the policy will divert a substantial portion of inter-island road
freight traffic to railways especially for longer distance traffic. The system improvement is
also obtained from better intermodal transfers as the simulated railway network provides
direct access to inter-island ports from main production centers in Sumatera.
The performance of inter-island network will be improved by the implementation of hub
ports together with designated spokes as feeder to the hubs. The implementation of the 9-hub
system, as seen in Figure 11, results in lower system costs than the 15-hub system which
suggests that more hubs do not necessarily mean more efficient inter-island port system. The
SJAFRUDDIN, A., et al. / Asian Transport Studies, Volume 1, Issue 1 (2010), 18-32
30
Figure 9. Intra-Java commodity transportation,
optimum system cost by policy scenario
Figure 10. Intra-Sumatera commodity transportation,
optimum system cost by policy scenario
Figure 11. Inter-island Indonesia commodity transportation,
optimum system cost by policy scenario
SJAFRUDDIN, A., et al. / Asian Transport Studies, Volume 1, Issue 1 (2010), 18-32
31
implementation of hub-spoke ports represents an improvement to the system hierarchy of
inter-island port system and this increases the efficiency of freight transportation system
through better inter-modal or intra-modal transfer at ports. This policy also reduces the
number of inefficient direct links between inter-island ports and results in better economies of
scale of services in both main and feeder links. The improvement of the inland transportation
network simulated here, both the Trans Java Toll Roads and the Trans Sumatera Railways
Development, will also bring about some benefit to the inter-island transportation as these
provide better access to the ports.
6. CONCLUSIONS
Several remarks could be drawn from this research as follows:
• Demand models by commodity type (CPO+Other Liquid, Rice+Agri Grains, Other
Grains+Fresh Product, Forestry Product, and Others/General Cargo) for inter island freight
transportation in Indonesia have been developed and can be utilized to estimate the volume
of commodity in a certain time.
• The development of the railway network will substantially reduce the inland transportation
cost and also induce some benefit to the inter-island transportation network. The potential
system improvement by railways development is apparently much higher than by road
development.
• The implementation of hub-spoke ports, representing an improvement to the system
hierarchy of inter-island port system, increases the efficiency of freight transfer at ports and
reduces the number of inefficient direct links between inter island ports which therefore
improves the economies of scale of services in both main and feeder links.
• The simulated policies in this research, however, cannot be interpreted as final for the
purpose of actual implementation. Further analysis of economic and financial feasibility of
the plan still needs to be conducted.
ACKNOWLEDGEMENTS
This research was supported by the Competitive Research Grant Program (Hibah Bersaing)
Batch XIV 2007 of the Directorate of Research and Community Services, Directorate General
of Higher Education, Ministry of National Education.
REFERENCES
Crainic, T.G., Florian, M., Guelat, J., Spess, H. (1990) Strategic planning of freight
transportation: STAN, An interactive graphic system. Transportation Research Record,
1283, 97-124.
Frazila, R.B. (2005) Optimization of Freight Transportation Network, Ph.D. Dissertation,
Hiroshima University, Japan.
Guelat, J., Florian, M., Crainic, T.G. (1990) A multimode multiproduct network assignment
model for strategic planning of freight flows. Transportation Science, 24 (1), 25-39.
Hermawan, W. (1999) Model kompetisi moda angkutan barang antara kereta api dan truk
dengan teknik stated preference, Master Thesis, Institut Teknologi Bandung (ITB),
Indonesia (in Indonesian).
SJAFRUDDIN, A., et al. / Asian Transport Studies, Volume 1, Issue 1 (2010), 18-32
32
INRO Consultants Inc. (1997) STAN User’s Manual Software Release 5, Montreal.
JICA (2003) STRAMINDO (Study on the development of domestic sea transportation and
maritime industry in the Republic of Indonesia), Final Report.
Lubis, H.A.S., Sjafruddin, A., Frazila, R.B. (2002) Port Planning with Multimodal Analysis,
Research Report – Institut Teknologi Bandung (ITB), Indonesia.
Lubis, H.A.S, Sjafruddin, A., Putra, A.I., Pramono, D. (2003) Developing Eastern Indonesia
multimodal freight transport network. The 8
th
JSPS Seminar on Marine Transportation
Engineering, Hiroshima, October 2003.
Prasetyo, L.B.B (1999) Modelling Intermodal Transfer of Freight Transportation Network,
Master Thesis, Institut Teknologi Bandung (ITB), Indonesia.
Sjafruddin, A., Frazila, R.B., Astuti, R.D. (1999) Regional freight transport demand modeling
in the Java Island. Journal of the Eastern Asia Society for Transportation Studies, Vol.3,
303-313.
Sjafruddin, A., Tamin, O.Z, Pudjianto, B., Astuti, R.A., B.F., Russ (1997) Pemodelan kebutuhan
angkutan barang regional di Pulau Jawa. Research Report – Institut Teknologi Bandung
(ITB), Indonesia (in Indonesian).
