General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright
owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.
Users may download and print one copy of any publication from the public portal for the purpose of private study or research.
You may not further distribute the material or use it for any profit-making activity or commercial gain
You may freely distribute the URL identifying the publication in the public portal
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately
and investigate your claim.
Downloaded from orbit.dtu.dk on: Aug 09, 2022
The liquefied natural gas infrastructure and tanker fleet sizing problem
Koza, David Franz; Røpke, Stefan; Molas, Anna Boleda
Published in:
Transportation Research. Part E: Logistics and Transportation Review
Link to article, DOI:
10.1016/j.tre.2017.01.003
Publication date:
2017
Document Version
Peer reviewed version
Link back to DTU Orbit
Citation (APA):
Koza, D. F., Røpke, S., & Molas, A. B. (2017). The liquefied natural gas infrastructure and tanker fleet sizing
problem. Transportation Research. Part E: Logistics and Transportation Review, 99, 96-114.
https://doi.org/10.1016/j.tre.2017.01.003
An exact solution approach for the liquefied natural gas infras-
tructure sizing and tanker routing problem
David Franz Koza
a∗
, Stefan Ropke
a
, Anna Boleda Molas
b
a
DTU Management Engineering, Technical University of Denmark, Produktionstorvet 424, 2800 Kgs. Lyngby,
Denmark
b
L’Oréal Danmark A/S, Havneholmen 25, 1561 Copenhagen V, Denmark
Abstract In this work we present a combined infrastructure sizing and tanker routing problem in the
liquefied natural gas (LNG) business that is based on a business case study with a major liner shipping
company. The decision problem is of strategic nature and consists of selecting the LNG storage capacity
at each port of demand as well as defining the size and number of tankers and their shipping routes used
to transport the LNG from its source port to the ports of demand. The goal is to minimize long term
investment and operational costs.
The introduction of global limits on sulphur and nitro oxide emissions has increased the interest in
LNG as an alternative fuel for vessels, including container ships. As the global LNG infrastructure is still
underdeveloped, it requires both strategic investment as well as tactical routing decisions to make LNG
available at the points of demand. We propose mathematical models for determining the capacities of the
necessary LNG infrastructure as well as the size and routes of LNG tankers needed for transportation.
Two models are presented. The first, non-linear model represents an intuitive formulation of the
optimization problem, but is hard to solve. The second formulation is based on the set-partitioning
model and is very attractive from a computational point of view. First, a set of partial solutions is
generated through enumeration. In the second step a set-partitioning problem is solved to determine the
best combination of the previously generated partial solutions. Results for the case study are presented
and an extensive sensitivity analysis is conducted to account for the limited predictability of key parameter
values, to analyse the robustness of the obtained solutions and to derive basic decision rules.
Keywords: Liquefied Natural Gas (LNG) as fuel, liner shipping, infrastructure planning, tanker routing,
mixed integer programming
1 Introduction
In 2008 the International Maritime Organization (IMO), a specialized agency of the United
Nations, has introduced new regulations for the prevention of pollution from ships that aim at
reducing the emission of sulphur oxides, amongst others (International Maritime Organization,
2008). The regulations apply in so-called Emission Control Areas (ECA) since 2015 already and
will become binding globally in 2020 or, if compliance appears to be impossible in 2018, in 2025.
Currently the majority of ships (80-85%, Chryssakis et al., 2014), including container vessels,
are run on heavy fuel oil (HFO). As a consequence of the new limits on sulphur emissions, ships
will no longer be able to operate as of today, because emissions due to the use of HFO exceed
the limits.
In various industrial strategic papers and research studies (see e.g. Andersen et al., 2013;
Rozmarynowska and Oldakowski, 2012; Chryssakis et al., 2014) three viable solutions to meet
the new requirements have been identified. The first one are exhaust gas aftertreatment systems
as e.g. scrubbers. The installation of scrubbers, however, is costly and requires additional space
on the ship. Further, they can increase the fuel consumption of a vessel by 2-3% (Chryssakis
et al., 2014). The second and most straightforward solution is the use of cleaner marine diesel
oil (MDO) or marine gas oil (MGO), as these can usually be used without the need of any
modification to the vessels. MDO and MGO are, however, about 1.5-2.0 times more expensive
than HFO, with prices expected to increase even further once the sulphur emission limits apply
globally.
The third solution is the one that motivates our study and considers liquefied natural gas
(LNG) as an alternative fuel. LNG is natural gas that is converted to liquid form by cooling
it down to approximately −162
◦
C (−260
◦
F ). It is the cleanest form of fossil fuels and if used
1
Arrives
MON
TUE
THU
SAT
WED
SAT
THU
MON
FRI
SUN
WED
MON
TUE
THU
SAT
THU
THU
FRI
WED
FRI
MON
Departs
TUE
WED
FRI
SUN
WED
SUN
THU
TUE
SAT
MON
SAT
MON
WED
FRI
SUN
SAT
FRI
SAT
FRI
FRI
TUE
Transit
--
--
2
4
8
11
23
27
31
33
36
41
42
44
46
51
58
73
78
80
83
Port
Busan, Korea
Kwangyang, Korea
Ningbo, China
Shanghai, China
Yantian, China
Tanjung Pelepas, Malaysia
Suez Canal, Egypt
Algeciras, Spain
Rotterdam, Netherlands
Bremerhaven, Germany
Gdansk, Poland
Aarhus, Denmark
Gothenburg, Sweden
Bremerhaven, Germany
Rotterdam, Netherlands
Port Tangier, Morocco
Suez Canal, Egypt
Tanjung Pelepas, Malaysia
Yantian, China
Hong Kong, Hong Kong
Busan, Korea
Asia - Europe (AE10) -
Roundtrip
Rotterdam
Bremerhaven
Gothenburg
Gdansk
Yantian
Hong Kong
Aarhus
Suez Canal
Kwangyang
Shanghai
Ningbo
Port Tangiers
Tanjung Pelepas
Busan
Algeciras
Figure 1: Example of a liner shipping service between Asia and Europe. The whole round trip takes 12
weeks and hence 12 vessels are operating on the service to ensure a weekly frequency.
to fuel ships, no further measures are needed to satisfy the new regulations concerning the
emission of pollutants. LNG is considered a realistic option for deep sea trades in the long term,
particularly for liner trades (Lloyd’s Register, 2012).
Liner shipping networks consist of cyclic shipping routes, called services, that are operated
periodically. Figure 1 shows an example liner shipping service that connects Asia and Europe.
The individual services are connected through ports, where cargo can be transshipped between
different services, and thus provide an extensive, wide-ranging transportation network. Typically
each port on a service will be visited once per week and the container liner company publishes
the weekly berthing time for each port. As a single round trip can take several weeks, each
service is operated by a corresponding number of container vessels. The structure and way of
operating container shipping networks is very similar to that of buses in public transport, with
the containers being the equivalent of passengers.
The current lack of LNG infrastructure for marine bunkering and the uncertainty about
future availability still is a major drawback of using LNG as a fuel for liner shipping companies.
This work is motivated by and based on a case study with a major liner shipping company that
considers filling that gap by building up and operating the needed infrastructure by themselves.
The study forms the basis of a future scenario in which the company uses LNG fuelled container
vessels on some of their services.
The liner shipping company is responsible for the transport of LNG to predetermined ports
where container vessels will refuel. The transport is done via sea using special purpose LNG
tankers. The tanker fleet needs to be ordered or chartered by the company. Furthermore,
the infrastructure at the majority of the ports of demand will be built and run by the liner
shipping company. The problem is of strategic nature with a time horizon of 2-15 years and
combines strategic infrastructure and tanker investment decisions with tactical tanker routing
and inventory management decisions. The lack of existing infrastructure allows to simultaneously
optimize strategic investment decisions and interdependent tactical decisions.
Strategic infrastructure planning and tactical planning of operations have traditionally been
looked at separately. For tactical and operational problems, the infrastructure and the fleet of
vehicles is usually fixed to a large extent. Furthermore, strategic decisions often do not solely
depend on quantifiable parameters but are subject to many qualitative arguments (legal issues,
local regulations, political decisions, etc). This work aims at providing decision rules of thumb
and identifying important relationships between operational/tactical and strategic decisions for
2
the problem studied. The presented models also allow to evaluate manually developed solutions
and their sensitivity to changes in input parameters. Hence, an important requirement of our
industrial collaborator towards the solution method are fast running times that allow to evaluate
large numbers of different scenarios within reasonable time.
The class of maritime inventory routing problems (MIRPs) is the one closest to the problem
addressed in this work. In fact, if the strategic infrastructure and tanker investment decisions
were fixed, the remaining problem would classify as a MIRP. The first studies on the MIRP
are by Christiansen and Nygreen (1998) and Christiansen (1999) and deal with the production,
shipping and inventory management of ammonia. Since then, most of the contributions are mo-
tivated by some specific application. Furman et al. (2011) present a mixed-integer programming
formulation for vacuum gas oil routing and inventory management. Agra et al. (2013) and, more
recently, Agra et al. (2015) consider a MIRP in the fuel oil distribution business, with the latter
contribution assuming sailing and port times to be stochastic. An overview of maritime inven-
tory routing problems together with examples of applications can be found in Christiansen and
Fagerholt (2009). Several contributions to the MIRP literature with applications in the LNG
business have been made during the recent years. Grønhaug and Christiansen (2009), Grønhaug
et al. (2010) and Andersson et al. (2016) present different exact solution approaches for an LNG
inventory routing problem. Rakke et al. (2011), Stålhane et al. (2012), Halvorsen-Weare and
Fagerholt (2013), Halvorsen-Weare et al. (2013), Goel et al. (2012) and Rakke et al. (2015) and
Andersson et al. (2015) have developed exact and heuristic solution approaches for LNG annual
delivery program planning problems. The paper by Andersson et al. (2010) provides a general
description of the LNG supply chain and presents two related problems. What these research
studies have in common is that they address tactical or operational problems with planning
horizons of at most several months. Different to our work, the onshore infrastructure and the
available fleet of LNG tankers is generally assumed to be given and fixed. By contrast, the
strategic infrastructure and fleet investment decisions play a key role in the problem. We be-
lieve that including these decisions change the nature of the problem significantly compared to
existing MIRPs.
A study closer to our work is the one by Jokinen et al. (2015). They present a mixed integer
linear programming model that aims at minimzing the cost of a small-scale LNG supply chain
in southern Finland, including both annual terminal investment as well as transportation costs.
LNG needs to be distributed from a large regasification terminal to several inland consumers,
using both sea and land based transport through smaller satellite LNG terminals along the coast
and LNG trucks that connect the ports with the points of demand. Even though the focus and
scale of their study differ significantly from our work, their results underline the importance
of simultaneously considering strategic investment and tactical routing decisions in the still
underdeveloped LNG business.
The significant number of recently published industrial studies and surveys that address
the feasibility and prospects of LNG as a fuel in deep sea container shipping demonstrate the
practical relevance of the problem studied (e.g. DNV-GL, 2014; Andersen et al., 2013; Lloyd’s
Register, 2012, 2014).
The contribution of this paper is twofold: firstly, we introduce the infrastructure sizing
and tanker routing problem that combines strategic and tactical decisions. We present two
mathematical models for solving the problem, with the latter formulation allowing very short
solution times. Secondly, we report computational results from a real-life case study. The results
show that it pays off to optimize for strategic and tactical decisions simultaneously, because in the
long perspective transportation costs can be reduced significantly through optimal investment
decisions.
In Section 2 the problem is described in detail and corresponding assumptions are introduced.
In Section 3 we present an arc-based model for conceptual reasons and derive a path based
formulation that is used for the computational tests. The computational results are reported
3
Extraction &
Liquefaction
Storage Shipping Storage
Small scale supply
Regasification
End usage
Figure 2: The LNG supply chain and the part considered in this work (following Andersson et al., 2010)
Add Presentation Title
in Footer via ”Insert”;
”Header & Footer”
Shanghai
Jebel Ali
Salalah
Port Said
Malta
Algeciras
Rotterdam
Qatar
Singapore
Figure 3: Planned refuelling stops.
in Section 4, together with a detailed description of the underlying data and cost. Concluding
remarks are given in Section 5.
2 Problem description
The typical LNG supply chain in a maritime context can be described as follows (see e.g. Ander-
sson et al., 2010): natural gas is extracted, purified and then turned into liquefied natural gas at
a designated liquefaction plant. After liquefaction the LNG is stored in full-containment tanks
that keep the gas in its liquefied state. It is then transported by special-purpose LNG tankers to
its destination terminal, where it is unloaded into onshore storage tanks again. Nowadays most
of the LNG is regasified at its destination and then distributed to end customers via pipelines
or trucks. In our scenario, however, a large portion of the liquefied natural gas would finally be
used to fuel container vessels (see Figure 2). The container ships would either be refuelled at
ports with LNG bunkering facilities or by LNG bunker vessels.
All the LNG needed at the ports of demand in this study is considered to be extracted as
natural gas (NG) and transformed into LNG at liquefaction plants by a third party in Qatar, the
world’s largest producer of LNG. In Qatar the LNG is loaded onto LNG tankers that are either
owned or chartered by the liner shipping company. The destinations of the LNG are ports located
along the route Asia-Europe. The number of ports that shall provide LNG refuelling services
and their locations have been set by the company, based on their estimates of how large the LNG
tanks on the container vessels will be and how often the ships will have to refuel. Generally,
it is expected that LNG fuelled container vessels refuel more often because of the lower energy
content per unit of volume. Depending on the LNG tank type used, the compartment housing
of the tanks can be up to four times that of HFO tanks of equivalent energy content (Hannula
et al., 2006). The eight ports at which onshore LNG storage tanks are planned to be built are
shown in Figure 3. The rather large number of ports is explained by the fact that the ports
will provide refuelling facilities to several liner shipping services along the Asia-Europe trade
but also serve external small scale demand. The scenario of LNG produced onshore in Qatar
4
