DEVELOPMENT OF A LNG DISTRIBUTION NETWORK FOR
INLAND SHIPPING
Master thesis, MSc Supply Chain Management
University of Groningen, Faculty of Economics and Business
April 28, 2015 THOMAS GOLLACKNER Student number: 2557320 t.gollackner@student.rug.nl Supervisor Dr. Evrim Ursavas Co-assessor Prof. Dr. I.F.A. Iris Vis
ABSTRACT
Liquefied natural gas (LNG) is considered as one of the best options to curb emission levels from inland shipping, which is expected to experience increased transport volumes in the future. In order for LNG to be adopted as an alternative propulsion fuel however, corresponding refueling infrastructure and a distribution network has to be developed. Location choices for these bunkering facilities and mode selection for replenishment and transportation are crucial decisions in the process of developing this distribution network. The goal of this paper is to review existing facility location models to identify what type of model is best suited for the inland shipping sector. By performing an analysis for the Arnhem Nijmegen region in the Netherlands this study provides required data and information as input for the previously identified model.
TABLE OF CONTENTS
1. INTRODUCTION ... 1
1.1.Aim of this study ... 3
1.2.Research Question ... 3
2. THEORETICAL BACKGROUND ... 4
2.1.Designing a Distribution Network ... 4
2.2.LNG bunkering methods ... 7
2.2.1. Static and dynamic bunkering methods ... 7
2.2.2. Portable bunkering methods ... 8
2.3.Literature review ... 8 3. METHODOLOGY ... 13 3.1.Case selection ... 14 3.2.Data collection ... 14 3.3.Data analysis ... 18 4. RESULTS ... 18 4.1.Traffic analysis ... 19
4.1.1. Origin and Destination points ... 19
4.1.2. Ships suitable for conversion to LNG ... 20
4.2.Findings expert interviews ... 23
4.2.1. Perception of LNG ... 23
4.2.2. Route and bunkering planning ... 25
4.2.3. Arnhem Nijmegen as a LNG bunkering location ... 26
5. DISCUSSION ... 27
6. CONCLUSION ... 30
1. INTRODUCTION
Emission levels of the transportation sector have been rising steadily over the last decades. From 1990 to 2007 transport emissions rose by 29%. In order to prevent a further rise, new regulations on emission levels and attempts to promote the use of alternative energy sources have been undertaken by policy makers (Arteconi & Polonara, 2013). For maritime transport, the International Maritime Organization (IMO) who is responsible for regulations on ship emission applies new regulations from 1 January 2015 regarding the emission levels for ships sailing in the North Sea, the English Channel, and the Baltic Sea. Under these regulations the sulphur content of marine fuels has to be reduced from 1,0% to 0,1% (Danish Maritime Authority, 2012).
Transport volume to and from the European main ports is expected to increase significantly over the next decades. As a direct consequence of this increase of volume, the importance of hinterland transport through trucks, rail, and inland shipping is also expected to increase significantly. European inland waterways currently have a lot of unused capacity even though it is considered as a more sustainable option than roadside transport due to lower emission levels. In order to cope with increased freight movements along the inland waterways, substantial infrastructure investments are required. These infrastructure investments should also serve as a way to make inland shipping more sustainable through the use of alternative fuels with lower emission levels. Since the destination of goods however rarely is located in a port, it is necessary that inland shipping becomes a more integrated part of the supply chain. Achieving this integration requires considerable improvements in different forms, which could be improved infrastructure for multiple modes of transportation or improved planning of operations through the use of communication and information systems.
Alternative fuels, such as biofuels, hydrogen, and liquefied natural gas (LNG) are considered as suitable options to reduce emissions levels and increase sustainability of shipping traffic. Especially LNG is considered among the most promising options because it not only significantly lowers the emission level of a ship but also successfully competes with traditional fuels on additional performance measures. Compared to the traditionally used Heavy Fuel Oil (HFO), LNG reduces the emission of nitrogen oxides (NOx) by approximately 80-85%, of carbon dioxide
due to the nonexistence of sulphur in LNG (Burel, Taccani, & Zuliani, 2013). In addition to the significantly reduced emission level, LNG as a propulsion fuel also has the potential for improved energy efficiency as well as substantial cost savings due the lower price of LNG compared to traditional fuels. These cost savings however only materialize after payback periods of multiple years, depending on the development of the LNG price, since the installation costs for the propulsion system are higher than for traditional diesel solutions (Burel et al., 2013).
Although LNG seems like a favorable solution to reduce emission levels and improve the sustainability of shipping, the required infrastructure to provide ships with LNG does not exist yet. In order to attract and fulfill demand for LNG as a propulsion however, it is necessary to design a sustainable network of bunkering facilities. The main issues to be answered in the design process are concerned with the location, size, and the usability by one or multiple transportation modes of the facilities.
Research on planning and design of a supply chain was previously conducted for different alternative fuels, such as hydrogen and biofuels. Many challenges in designing a supply chain for LNG can be considered quite similar to those of hydrogen. Significant uncertainties in demand and determining the optimal structure of a network capable to fulfill growing demand among others are considered as main complexities for the development of any new fuel technology (Dagdougui, 2012).
demand this factor has to be considered as investments are necessary to recover the gas in order to minimize lost value (Arteconi & Polonara, 2013). Due to these differences and the current lack of models for the design of a LNG distribution network for inland shipping it is necessary to understand the relationship between these factors before an optimal network design solution can be found.
1.1. Aim of this study
The goal of this study is to identify the type of distribution model that is most appropriate for the development of a LNG distribution network for inland shipping. This goal will be accomplished through the use of a literature review to identify the shortcomings of existing models and to investigate the sort of data and information necessary to fill the gaps of the existing models. Through a case study for the Arnhem Nijmegen region in the Netherlands, the previously identified data and information required for the development of this model will be gathered.
In order to provide the required inputs for a network design model a combination of empirical and qualitative data collection methods are carried out. To investigate the potential demand for LNG, the inland waterway traffic is analyzed in order to identify the characteristics of suitable vessels for conversion to LNG in the region. In addition to this empirical analysis, interviews with stakeholders are carried out to gather information about behavior and preferences of ship owners, and to identify regional specific constraints for the distribution network model.
1.2. Research Question
The main research question of this study can be stated as:
What type of distribution model should be used for the development of a LNG distribution network for inland shipping?
3. What LNG specific characteristics need to be considered in a supply chain network design model?
4. What types of bunkering methods are available for LNG?
5. Which specific constraints are required for the model of the LNG supply chain in the Arnhem Nijmegen region?
The remainder of this paper is organized as follows. The next section contains background on LNG specific characteristics for the network design problem an overview of currently existing network design models to identify the type of model that is required for a network of LNG bunker facilities. The following section provides an outline of the methodology of this exploratory case study research.
2. THEORETICAL BACKGROUND
This section provides an overview over necessary steps and decisions for the design of a distribution network. Furthermore different types of bunkering methods for LNG are introduced and existing types of network design and facility location models are presented and related to relevant LNG characteristics for the design of a distribution model. This serves to identify the type of design model that is necessary for the design of a LNG distribution network for inland shipping.
2.1. Designing a Distribution Network
identifies 10 steps for the design of a distribution network. For the design of a LNG network however, not all of these steps are applicable and thus only the relevant ones are described in this study. Before the first step of designing and populating the network optimization database can be carried out however, it is necessary to understand the total market demand for LNG and the potential growth rate (Magrath & Hardy, 1991). This is consequently considered to be the first step in the design of a distribution network.
Understanding the total market demand and the growth rate
Since the demand for LNG as a propulsion fuel is subject to significant uncertainty, it is necessary to first analyze the potential demand before the actual design of a distribution network. An optimal network structure is not only capable of fulfilling the current demand, but has also be serviceable for potentially growing future demand (Dagdougui, 2012). In the case of the LNG supply chain, especially the uncertainty of demand plays an important role. Current demand is very low and the future demand has to be estimated for a suitable design of a distribution network e.g. an analysis of potential scenarios for the German inland shipping demand for 2030 developed two scenarios for the percentage of newly built ships with LNG as a propulsion fuel. In a moderate scenario it is assumed that 50% of new vessels with a tonnage above 2500 tons operate with LNG, while an accelerated scenario assumes that percentage at 75% (Federal Ministry of Transport and Digital Infrastructure, 2014). In addition to newly built vessels, demand can also arise from ships that transition from traditional fuels to the use of LNG as a propulsion fuel. Through analyzing the demand side of the LNG supply chain it is also possible to estimate the potential growth rate of demand. In addition to the total demand of LNG in the network it is important to consider the location of the demand and its development for the design of the supply chain.
Designing and populating the network optimization database
transportation modes and times for replenishment, locations for bunkering facilities and gas stations. In order to collect the required data however, it is necessary to understand the characteristics and uncertainties that exist for a LNG supply chain (Dagdougui, 2012).
Creating network design alternatives
This step in the design process requires the highest level of creativity, as the answers obtained from modeling tools are only as good as the scenarios provided to consider. The majority of support tools for making decisions are only useful to evaluate different scenarios and not to generate them. Hence if only infeasible and inefficient scenarios are considered the obtained result will also be an infeasible and inefficient scenario. Consequently it is necessary to generate as many feasible and efficient scenarios as possible to get the best result possible (Frazelle, 2002). Design alternatives for an LNG distribution network have to consider some unique features of the LNG supply chain as well as they have to take certain regional and local constraints into account. More precisely does the design have to account for currently low but growing demand and therefore the design should have a long-term planning horizon.
Developing a network optimization model
After creating design alternatives it is necessary to formalize the network scenarios into a mathematical optimization model. This requires to set an objective function, which in most cases is to minimize the total costs, and to set constraints for the model (Frazelle, 2002). The next section will provide an overview of various existing optimization models and their specifications.
Strategic Decisions
Type of required facilities Amount of each type Size of facilities
Location of individual facilities
Tactical Decisions
Sourcing
Location of Inventory Quantity of Inventory
Operational Decisions Transportation modes
Table 1: Decision Levels for the LNG distribution network
2.2. LNG bunkering methods
The most critical decisions for the design of a distribution network are concerned with the different types of facilities and their location, which are considered to have a long lasting impact on the whole network (Vidal & Goetschalckx, 1997). The different types of required facilities for a LNG distribution network include LNG depots, gas stations, and LNG bunkering facilities. The different bunkering methods for LNG can generally be divided into static, dynamic and mobile portable bunkering methods.
2.2.1. Static and dynamic bunkering methods
terminal when demand arises. This is particularly useful in the early stages of a LNG distribution network where demand is expected to be relatively low. This circumstance also contributes to one of the key issues in determining the optimal structure of a network of bunkering facilities, to find a configuration that is capable of fulfilling the current and the growing future demand for LNG. An optimal structure usually aims to minimize the costs of development and installation of the infrastructure. This cost minimization approach however could potentially lead to solutions that are inappropriate on other aspects, such as replenishment times (Dagdougui, 2012). Consequently the design of an LNG distribution network should take multiple criteria into account instead of focusing on cost-minimization.
2.2.2. Portable bunkering methods
In addition to the static bunkering methods, refueling of LNG can also be carried out through portable tanks. Those tanks are specialized insulated ISO intermodal portable tank containers that are being filled with LNG to be transported to a port and consequently used as on-board fuel tanks (Det Norske Veritas, 2014). The main advantage of this method is that it allows vessels to convert to LNG without retrofitting the ship for new fuel storage containers onboard. The containers are manufactured in both standard container sizes. The 20-foot container can store slightly more than 20,000 litres of LNG, while the 40-foot container is able to hold around 43,500 litres. One of the main advantages of this bunkering method is that it does not require a LNG bunkering facility nearby. The portable tanks can be driven or lifted on and off a vessel with the regular equipment of a port equipped for intermodal transport. This also means that LNG can be transported along the supply chain without the need to pump and transfer it along the way. The portable containers are able to store LNG for a duration of up to 60 days without experiencing boil-off (Det Norske Veritas, 2014). This method however is currently still in an experimental stage and requires additional ongoing studies. Although the focus on this study is on the dynamic and static bunkering methods, the gathered data and obtained results can also facilitate assessing the feasibility of this solution.
2.3. Literature review
optimal configurations of a distribution network according to previously set criteria (Dagdougui, 2012). As input for these models serves a set of options for facilities and transportation to obtain the type, numbers, location, and capacity of facilities and transportation (Dagdougui, 2012). In addition to these location and transportation options, further input for the decision support is required. Especially the demand for a given geographical area is important for the development of a distribution network. Facility location models can generally be classified through a number of characteristics. Melo et al. (2009) classify facility location models based on their number of layers, number of commodities, single or multiple periods, and deterministic or stochastic parameters. Furthermore it is important to distinguish between capacitated and uncapacitated models. Uncapacitated models assume that a facility can serve an infinite amount of demand, whereas in capacitated models capacity constraints have to be obeyed (Klose & Drexl, 2005). Many practical facility location problems have to deal with the crucial aspect of different types of facilities. Each set of facilities of the same type which also have the same role is typically denoted by a layer or an echelon which defines the level of hierarchy among facilities (Melo et al., 2009). This circumstance is however not the case for the design of a network of bunkering facilities for LNG, which can therefore be seen as a single echelon facility location problem. Similar to the majority of facility location models where the starting point is a given set of potential facility sites, the design of a network of bunker facilities also starts out with a set of potential locations (Klose & Drexl, 2005). These location options however are limited as the potential locations necessarily have to be located along inland waterways.
exact composition of LNG (Querol, Gonzalez-Regueral, García-Torrent, & García-Martínez, 2010). Under consideration of these specific characteristics for an LNG distribution network it can be derived that the facility location problem is an uncapacitated, single echelon, single commodity, single period problem with stochastic parameters.
In existing literature several approaches have already been used to find the optimal locations for refueling stations. Hakimi, (1964) introduced the most common use model in the field of facility location, the p-median model. This location-allocation model locates a given set of p of facilities and allocates demand from nodes i to facilities j. The target of this model is to minimize the total distance travelled by consumers to the facilities (Hakimi, 1964). This model is particularly useful in locating fueling stations close to demand points, which in the case of refueling stations are the homes of customers. For the location of a network of bunkering facilities for LNG however, it is important to consider demand in certain corridors of inland waterways on which freight movement is carried out. The demand for LNG cannot be depicted as a node at a certain location, but should rather be seen as a traffic flow that passes by the bunkering facility. This approach of facility location is called the flow-capturing location model which was first introduced by Hodgson, (1981). The units of demand in this model are not depicted as points in space, but rather flows on paths across the network representing the routes travelled (Upchurch & Kuby, 2010).
origin-destination path of vehicles. In their model the deviations are a must due to the low range of hydrogen vehicles for which the refueling network is designed. Although inland vessels have a significantly higher range than hydrogen vehicles, this extension could be interesting for this study as ships that cannot bunker at their preferred location could be forced to make deviations from their path. In order to generate input for a facility location model it is therefore necessary to analyze the routes from origin to destination of inland vessels in a certain corridor to identify the locations which show the highest flow volumes. The inland waterway network however differs on the amount of potential locations from the previously studied refueling problems in literature. These studies investigated the optimal location for refueling stations for road vehicles, where all nodes in a network are potential locations facilities and the focus is on reducing the inconvenience of the short range of these vehicles. The characteristics for the inland waterway network however are different because the network only exists of waterways along which facilities can potentially be located. In addition are existing ports considered the best options for locating new bunkering facilities, which reduces the potential locations even further. This enhances the solvability of the model, as linear programming models are more effective in finding optimal solutions for smaller networks opposed to larger ones with hundreds of nodes and facility combinations to evaluate (Lim & Kuby, 2010). Another important difference is that unlike short-ranged alternative fuel vehicles, inland vessels possess a significantly larger tank which makes it unlikely that a vessel has to refuel for every single trip it takes. Due to this high range of the barges, a potential refueling network has to be less extensive than the previously studied networks of refueling networks for cars.
integrates the above mentioned decisions is however relatively scarce and consequently many of the strategic aspects of supply chain design are not sufficiently combined (Melo et al., 2009).
In conclusion it can be inferred that for finding the optimal locations for a LNG distribution network for inland shipping, a flow capturing facility location model seems to be the most suitable approach. Such a model however has to be adapted for the characteristics of the inland waterway network. Moreover is an investigation into preferences and behaviors of ships required in order to generate valid assumptions about location preferences, willingness to deviate from set paths, and criteria that constitute the attractiveness of facilities. Furthermore should the model be able to integrate the problem of mode selection for replenishment while taking the formation of boil off gas and the available mode options for transportation of LNG into account.
3. METHODOLOGY
While a significant amount of research on supply chain network design and facility location problems exists, literature tackling the problem of facility location for inland waterways is scarce. Especially in case of the introduction of a new bunkering fuel for inland shipping which has specific characteristics for the design of a distribution network, a literature study should find out if the existing models are appropriate to capture the situation on hand. The literature study starts with general facility location problems and focuses especially on articles that target the issue of flowing demand that is not fixed at one point in the network. In addition to articles about so called flow capturing facility location problems, the literature review looks at articles that tackled the problem of designing a supply chain for different alternative fuels, such as biofuel and hydrogen supply chains. This is done with the aim to identify similarities and most importantly differences between these fuels and LNG for the design of a new distribution network.
to be most suitable research methodology, as the region that is selected for the study can be considered as a representative case (Yin, 2009).
3.1. Case selection
The Arnhem Nijmegen region in The Netherlands serves as the single case in this study in order to get insight about the potential demand and specific characteristics for LNG distribution network. The research took place within the Arnhem Nijmegen city region, which is situated in the east of the Netherlands. This region was especially suited for performing the case study as it is situated in close proximity the German border with access to the important inland waterways of the Nether Rhine and the Waal. Due to the short distance to Germany it is also important to consider the surrounding area of the Rhine for the design of a potential LNG distribution network. Additionally does the region play an important role for inland shipping on the European level, as it is situated between the European hinterland and a few of the most important ports of Europe in Amsterdam, Antwerp, and Rotterdam. This location can be considered as strategically important, as the Nether Rhine and Waal are two of the most important connections between the sea ports and the European waterway network. Due to this the region registers a significant amount of shipping traffic coming from and going to the German inland ports along the Rhine as well as from Basel in Switzerland. The city of Nijmegen was more than willing to cooperate and open to share the necessary information and data. As a result this setting is appropriate to be examined because it provides the author not only with first-hand information and direct access to stakeholders involved in a potential LNG distribution network but also with issues that can be considered representative for multiple regions along the European inland waterways.
3.2. Data collection
author. This data consists of two datasets that contain the position information and additional information for 303 ships with an annual bunkering volume above 500m³. The tables below show a summary of the information that was provided to the author by the EICB for the ship position dataset.
MMSI LAT LON Speed Course Status Timestamp
Table 2: Information dataset ship positions
The term MMSI in the table above refers to the Maritime Mobile Service Identity, which is a nine digit number that serves to uniquely identify a ship. LAT and LON stand for the latitude and longitude coordinates, while speed, course, and status refer to the movement speed, direction, in degrees, and status at the time the information was recorded. In addition to this information the dataset contains timestamps with the date and the exact time of recording for each entry of the database. The duration between timestamps is approximately one hour, but can be extended if it was not possible to retrieve the position data of the ships for an extended period of time.
The above described dataset will be used to analyze the shipping traffic through the Arnhem Nijmegen area and this information will then be combined with the second dataset provided by the EICB. This dataset contains additional details about the individual ships from the first dataset, as can be seen from the table below.
ENI-Nr. MMSI-Nr. Ship name Ship type Owner Bunker volume
Table 3: Ship information dataset
are therefore excluded from this analysis. In addition to the provided information, additional information about the individual ships has been gathered via online resources and company communication. This information concerns the year the ships were built in and the dimensions, especially the length, of the ships as this additional information is required for the analysis of the data, which will be described in the next section.
Figure 1: Overview Data Sources
135 meters and a remaining lifetime of more than 20 years. Additionally are the companies operating different types of ships (tanker barges, cargo barges). As this study targets a region in the Netherlands, the focus in selecting the companies was also on companies located in the Netherlands. The table below provides an overview of the characteristics of the four companies and its ships.
Company A B C D
Traffic in region Yes Yes Yes Yes
Ship type Tanker Cargo Cargo Tanker
Ships suitable for LNG conversion Yes Yes Yes Yes
Table 4: Characteristics of selected companies
The logic behind the selection of these cases is to allow for literal as well as theoretical replication logic (Yin, 2009). While the selection of two companies that focus on a certain ship type serves to verify whether similar results occur for companies of the same context, the selection of tanker as well as cargo shipping companies aims to verify whether the results between the two types are in contrast to each other.
3.3. Data analysis
The first step in analyzing the data was the analysis of the shipping traffic through the Arnhem Nijmegen area. This was carried out through the use of the software Microsoft Excel and Google Maps to identify the locations of the recorded ship position coordinates. This step does not only aim to identify the ships that are actively operating in the area but also to identify the origin and destination points of those vessels as those are required for the design of a flow capturing facility location model. In addition to the origin and destination pairs, the ships were also categorized based on their size, age, and type of ship to identify which existing ships in the area are suitable for the conversion to LNG. This analysis will be discussed in the following chapter.
The analysis of the qualitative data collection started with coding of the information in order to keep an oversight of all gathered information and to help drawing conclusions in a reliable manner. The coding table which is used for the four embedded cases is displayed in Appendix II. The gathered data and information was already analyzed while other interviews still were conducted, as this did not only accelerate the result finding but additionally gave insights that could be used for the following interviews (Miles & Huberman, 1994). The coding is followed by a within-case analysis that allowed the author to become familiar with each of the cases independently. Following the within-case analysis a cross-case analysis was conducted, where the results of the individual cases were compared and patterns between them were identified in order to conclude and generalize the findings (Karlsson, 2009; Yin, 2009). The main points of the cross-case analysis will be discussed in chapter 4.2.
4. RESULTS
4.1. Traffic analysis
4.1.1. Origin and Destination points
One of the most important inputs for a flow capturing facility location model are the origin and destination (O-D) pairs of the traffic that flows through a network (Kim & Kuby, 2012). Based on the ship position dataset provided by the EICB those O-D pairs were identified each time a ship sailed through the Arnhem Nijmegen region. In order to see when ships sailed through the area a corridor of coordinates had to be set that includes the whole region. The corridor used in this analysis includes latitude coordinates between 51.84 and 52.02, and longitude coordinates of 5.8 to 6.15. The corridor was chosen rather large to include ships sailing on the Waal past Nijmegen as well as ships sailing on Nether Rhine past Arnhem. Out of the 303 ships of the original dataset, 221 were recorded to be inside this corridor at least once during the one month tracking period between 20.10.2014 and 20.11.2014. Each time a ship passed through the corridor, the corresponding origin and destination points were identified based on the position data. Since the recordings of the position data however are not complete for all ships it was not always possible to identify one of the two locations. After deleting the O-D pairs with missing locations, data from 207 ships was remaining. These ships account for a total amount of 955 trips, with identified origin and destination points, through the area. This total amount of trips translates to 171 unique O-D pairs. The table below presents the 10 most common O-D pairs, whereas a full table of the O-D pairs, including the number of ships that sailed this path can be found in Appendix III.
Origin Destination Nr. Ships
Duisburg Rotterdam 27 Rotterdam Duisburg 22 Rotterdam Karlsruhe 22 Rotterdam Cologne 17 Cologne Rotterdam 14 Antwerp Mannheim 13 Karlsruhe Antwerp 12 Karlsruhe Rotterdam 12 Mannheim Antwerp 12 Rotterdam Düsseldorf 11
From this table it becomes evident that the most traffic in the Arnhem Nijmegen area is created by ships that sail from the large ports of Antwerp, and Rotterdam to the Ruhrgebiet in Germany and further south to Karlsruhe and Mannheim.
4.1.2. Ships suitable for conversion to LNG
The use of LNG as a propulsion fuel for inland shipping at the time this study was conducted seems only feasible for a part of the inland navigation fleet. The main reasons for this are the high investment costs and the additional space requirements. The investment costs for the tank and the engine for LNG are roughly twice as high as the costs for diesel (BMVI, 2014). The operational costs however, are lower for LNG which leads to the consequence that ships with high mileage and substantial fuel consumption are more likely to benefit. Furthermore does the storage of LNG require additional space which was not accounted for at the time the ship was built. Consequently the conversion of LNG can lead to a loss of freight capacity, which is especially concerning for smaller ships (BMVI, 2014). Consequently the use of LNG seems most suitable for large vessels and particularly tanker and container vessels as storage of LNG may be most convenient for those types of ships.
According to a study by Panteia (2013), tanker vessels generally show the highest amount of operational hours and therefore have a substantial fuel consumption. Additionally are safety regulations and the placement of fuel tanks less complex as the vessel and the crew are already equipped and experienced in dealing with dangerous goods. Dry cargo ships show on average less operational hours than tanker vessels. Moreover does the placement of the fuel tank lead to a loss of freight capacity for those ships. Although large ships container vessels operating on the Rhine also show a high number of operating hours, the potential loss of payload is a serious concern for those ships. While push boats also show substantial operational hours as well as high fuel consumption, those vessels do not have the required space to place an LNG tank. The main problem for push boats is that for stability reasons the LNG tank would have to be placed at the gravitational center of the ship, where however the engine room is located. Consequently the conversion of existing push boats is considered not possible for technical reasons (Panteia, 2013).
2015 the price for a liter of diesel was down to € 1,289 while the price for a kilogram of LNG remained relatively constant throughout the past years and could be purchased for € 1,220 in February 2015. This significant change of the price difference between LNG and diesel also has an impact on the calculated payback times of the study conducted by Panteia, (2013). According to this study, ships with a length of 135 meters and bigger with substantial fuel consumption are expected to have payback periods below 20 years for retrofitting existing ships with new engines. Furthermore can also tanker ships of the 110 meters vessel class with continuous operation and therefore higher fuel consumption reach payback periods below 20 years. Due to the currently significantly lower price difference however, this paper only considers vessels of the largest vessel class of 135 meters and longer as potentially suitable for conversion. As the future development of the price difference between LNG and diesel can have a substantial impact on the amount of suitable ships for conversion, it should be closely monitored combined with a reassessment of the established specifications of suitable ships.
In order to identify the size of the fleet in the Arnhem Nijmegen area that is considered as suitable for conversion to LNG the existing dataset containing detailed information about the individual ships has to be extended with additional information. Specifically the size and the age of the ships are not included in the dataset. The average life-span of an inland ship is considered to be 40 years (Finger & Holvad, 2013). Consequently the remaining lifetime of ships to be considered suitable for conversion has to be a minimum of 20 years.
Figure 2: Criteria for Conversion to LNG
Based on the established criteria the database was filtered to only show ships that are suitable for conversion. From the 221 ships that were active in the area, a total of 95 ships can be considered as suitable for conversion. These 95 ships however are not equally distributed among the three types of ships, which can be seen from the table below.
Ship type Number of ships
Cargo 83
Container 1
Tanker 11
Table 6: Suitable ships per ship type class
however accumulates to 22,676.85 m³ of diesel. This number drastically increases up to 120,674.26 m³ if the conversion to LNG for cargo vessels is assumed feasible.
4.2. Findings expert interviews
This section presents and analyses the results obtained from the cross-case analysis of the interviews with shipping companies. For increased comprehensibility this section is divided into three parts. The first part is concerned with how shipping companies view the prospect of LNG as an alternative bunkering fuel. The second part deals with route planning and selection of bunkering locations, while the final part is devoted to information regarding the Arnhem Nijmegen as a potential bunkering location.
4.2.1. Perception of LNG
The main advantage shipping companies see in connection with the use of LNG is the environmental benefit of decreased emission levels. There are however also some concerns that can be interpreted as disadvantages of LNG. Two of the companies explicitly mentioned the availability of LNG as a disadvantage, as traditional bunkering fuel is available at multiple locations along the route of their ships. This is however not the case for LNG and could pose a threat to the adaption of LNG as shipping companies are concerned over the availability in regions that are further away from the big sea ports in the Netherlands, as the following statement from one interviewee shows: “…the disadvantages or problems which i see is that the logistics to renew all the bunker stations into LNG stations will be not so easy. Especially in the hinterland in Germany or southern Germany it is more and more difficult to get the LNG over there. So therefore it is not so easy to implement.”
Based on the interview answers the majority of shipping companies agrees with the report by Panteia (2013) that tanker barges are especially suitable for the use of LNG. The only respondent who did not mention tanker barges as the most suitable ships was from company C because the focus of the company is on dry bulk cargo transportation and he therefore could not make a qualified assessment of tanker barges. Similar to the consensus about tanker barges, all companies except of the respondent of company D were in agreement that LNG does not seem like a suitable option for dry bulk vessels. This is, as the respondent of company B explained, due to the fact that the LNG tank has to be placed in a way that it uses a significant amount of space which leads to a loss of too much capacity for the barges.
While converting existing ships for the use of LNG can have a payback period below 20 years, the respondents agreed that only a very small number of companies will actually choose to do so. The main reason for this is the high investment costs that are required for the conversion of an existing ship and the relatively long time before the investment pays off. This however could be different if the conditions change and the payback time of the investment shorten to roughly 10 years, as the respondent from company A indicated.
statement from company B essentially captures the opinion of all four interviewed shipping companies.
4.2.2. Route and bunkering planning
The selection of a bunkering location is, as indicated by all companies, not included in the process of planning routes for ships, which has two main reasons. First do inland barges have a large bunker capacity which allows them to sail multiple trips without bunkering. Moreover did all four respondents state that for each trip there are multiple bunker facilities on the way which can be used, so it is not required to include bunkering into the planning of routes. As the interview showed do shipping companies have agreements with bunker companies that they regularly frequent. While two of the interviewed companies have an agreement with a single bunker company in place, company A and B have such agreements with multiple bunker companies from which they choose on a daily basis. The respondents agreed however that they do not have preferences for specific locations, as long as the facility is along the route of the ship. The main criterion on which shipping companies choose the bunkering locations is the price of the fuel. Another important issue for shipping companies is that bunkering facilities have to open 7 days a week and 24 hours a day. Most of the bunker facilities are operating around the clock according to the respondents, but there still exist some examples in the Netherlands where incoming ships have to wait until the next morning to be bunkered, which according to the interviewee of company B is a time loss that is too expensive for shipping companies. Besides the requirements of a competitive price and continuous operating hours, shipping companies do not consider any additional services offered by bunker facilities in their decision making. While all respondents acknowledge that the majority of bunker locations offer some extra services, mostly in the form of an attached shop that sells ship supplies, they do have no influence on the shipping companies.
4.2.3. Arnhem Nijmegen as a LNG bunkering location
The area encompassing Nijmegen, which includes the area of Millingen towards the German border, is unanimously considered as a very good potential location for a LNG bunker facility by the shipping companies. This is due to the fact that a facility in this area would capture all traffic going from the big seaports of Amsterdam, Antwerp, and Rotterdam to Germany, as the remark from company B illustrates: “That is the best area you can do it, because every barge that is going to Germany will pass by Nijmegen, Millingen or that area.” While it is considered as a very good location for a bunkering facility, the respondent from company A also mentioned that the area of Dordrecht would potentially be better because a facility there would also capture traffic from ships that only sail between the three main ports and therefore increase the potential market. The strategic location of the area of Nijmegen between the big seaports and the European hinterland clearly influences the future potential for a LNG bunkering facility. This can be inferred from the fact that all four interviewed companies considered stopping at a bunker facility in this area as very likely if their company is operating LNG fueled ships in the future. In addition do all four companies expect the transport volume and traffic through the area of Nijmegen to increase, which can have a positive influence on the future demand for a LNG bunkering facility.
5. DISCUSSION
The results of this study gave interesting insights into what kind of distribution and especially what kind of facility location model is required for the development of a LNG distribution network for inland shipping. Particularly facility location problems are a widely researched topic with applications in multiple fields and areas. There exists a variety of approaches on how to solve these problems, from the most commonly used -median model (Hakimi, 1964) to flow capturing location models that see demand as flows passing through a network (Hodgson, 1981; Kuby & Lim, 2005; Upchurch & Kuby, 2010). Various existing flow capturing facility models deal with the problem of locating refueling facilities for land based traffic (Kim & Kuby, 2012; Kuby & Lim, 2005; Upchurch & Kuby, 2010). Consequently the author claims that an adapted flow capturing facility location model is the most appropriate for the location of LNG bunker facilities for inland shipping.
bunker location is however the price of the fuel, which is subject to daily change and therefore the preference can also shift from one bunkering company to another on a daily basis. The results show one characteristic that defines the attractiveness of bunkering locations for ship owners in addition to the price, the operating hours of the facility. The operating hours of the facility are a crucial aspect for shipping companies, and locations with continuous operating hours for 24 hours a day, 7 days a week are therefore more attractive than facilities with closing times. Due to the value based decision making in the shipping industry do services that are offered in addition to fuelling not increase the attractiveness of bunkering locations, as those services do not add significant value in the eye of shipping companies. Another important aspect of bunker facilities is their size, as currently not all bunker locations are large enough to service the bigger 135 meter barges. Therefore the size of the facility can also contribute to its attractiveness for ships.
Another factor for the design of a distribution network for LNG in addition to the location of bunkering facilities will be the development of demand for LNG from barges. This demand can stem from either new build ships or from existing ships that are converted for the use of LNG as a propulsion fuel. This study conducted an analysis based on the factors of age, fuel consumption, size, and type of the ships to investigate the amount of ships that are potentially suitable for conversion to LNG in the Arnhem Nijmegen area. The criteria for this analysis are based on a study conducted by Panteia, (2013). The results show a large amount of cargo ships as potential ships and only a smaller amount of tanker barges in the area. The suitability of cargo ships however is still questionable as converting those ships would lead to a decrease of their capacity. Based on additional insights from expert interviews it can be assumed that cargo ships are not a good fit for retrofitting LNG propulsion with the current technical solutions. The majority of the interviewed experts see the loss of capacity as a significant obstacle and cargo ships therefore as not suitable for conversion to LNG. In this case the amount of ships that are suitable for conversion however is drastically reduced as there are essentially only tanker barges left.
recent development of the fuel prices. The study by Panteia, (2013) assumed a price difference between LNG and diesel of 20%. The recent drop of diesel prices however led to a drastically smaller price difference of below 10%. This fact was also mentioned during the expert interviews and is one of the reasons why shipping companies do not expect a widespread use of LNG as a propulsion fuel for the immediate future but rather to be a realistic option in 10 to 20 years. Since these opinions however are partly based on the price difference between diesel and LNG, a development in favor of LNG could also lead to a faster growing interest from shipping companies. The expert interviews furthermore revealed that the current economic situation and its effect on the inland shipping industry are not favorable for large investments in alternative propulsion systems. This shows on the one hand through an existing overcapacity in the market that consequently leads to a reduction of the amount of new ships that will enter the market. On the other hand are margins in the industry very tight which sees many smaller ship owners struggling and makes shipping companies vary of making big financial investments for alternative propulsion systems.
The combined information from the data analysis and the expert interviews indicates that demand for LNG in the studied area will be very low for the foreseeable future, which directly affects the design of a distribution network for the Arnhem Nijmegen area. This low demand level especially influences the appropriateness of the different bunkering methods via truck, ship, or terminal. As the truck to ship bunkering method requires the lowest investment costs, the author claims that it will be the most suitable method for the early years of a LNG distribution network. As ship owners however have the opinion that in a timeframe of 10 to 20 years many of the ships that reach the end of their lifetime will be replaced with ships operating on LNG, bunkering via ship or terminal could become preferable with the resulting higher demand. Another important issue that has to be addressed in the design of the distribution network is the formation of boil off gas in the storage tanks. Especially for low demand situations, the reduction of LNG over time is a factor that has to be considered. The amount of boil off gas depends not only on the tank design but also on the LNG level the tank contains, as the boil off rate will increase as the LNG level in the tank decreases. The daily reduction is between 0,1-0,5% of the total volume.
LNG containers transported via train. Similar to the preferred bunkering method, the modal choice for replenishment is also dependent on the demand of LNG. The development of a model for the LNG supply chain for inland shipping should integrate the problem of mode selection into the facility location problem, which however requires additional data and information on transportation costs and times for the different modes to identify the preferred mode for respective transport and demand volumes.
Due to the exploratory nature of this paper, the data analysis and interview results from the previous sections are collected from a selective area and group of experts from the field. In order to generalize the findings beyond the Arnhem Nijmegen area, a larger scope and area of focus should be taken to increase the reliability of the results. Nevertheless, can the method and criteria of identifying suitable ships for conversion to LNG be applied to different areas and regions.
6. CONCLUSION
The purpose of this paper was to identify the distribution model type that should be used for the development of a LNG distribution network for inland shipping. Due to the steady increase in transport volume and consequently emissions from the transport sector, alternative fuels are a suitable method cope with increasing transport volumes while simultaneously reducing emission levels. Especially LNG is considered as one of the most promising alternatives to traditional fuels. The necessary infrastructure for the widespread use of LNG as a propulsion fuel in inland shipping however is currently not established. This paper therefore investigated the existing distribution network design approaches and identified the most appropriate model and investigated what sort of data and information is required for the development of this model.
insights into preferences and behavior of ship owners. Based on this information, assumptions and constraints that are necessary for a flow capturing facility location model can be formulated.
While this study gave insights into how a LNG distribution network for inland shipping should be formed, these results are bound to certain limitations that need to be considered. One of those limitations concerns the empirical scope and time of this research. The study itself was limited to three month of data collection and analysis, and position information of ship traffic was limited to 1 month of data. This circumstance had an impact on the amount of information the author was able to acquire. Information from companies operating in the LNG bunkering industry could not be obtained, which left the author unable to answer all previously formulated sub questions of this study. Specifically, detailed information about the transportation and replenishment of LNG in addition to previously existing knowledge from literature is missing from this study. Consequently was the scope of the information not sufficient to propose specific constraints for a distribution network related to the studied Arnhem Nijmegen area, as sub question 5 intended. A further limitation of this study is the focus on inland shipping as a sole demand source for LNG. Examining the perspective of companies active in truck transportation, might lead to other interesting results as LNG can be used as a fuel for multiple modes of transportations and facilities can be designed to serve multiple transportation modes.
REFERENCES
Almansoori, a., & Shah, N. (2009). Design and operation of a future hydrogen supply chain: Multi-period model. International Journal of Hydrogen Energy, 34(19), 7883–7897. Arteconi, a., & Polonara, F. (2013). LNG as vehicle fuel and the problem of supply: The Italian
case study. Energy Policy, 62, 503–512.
Awudu, I., & Zhang, J. (2012). Uncertainties and sustainability concepts in biofuel supply chain management: A review. Renewable and Sustainable Energy Reviews, 16(2), 1359–1368. Bersani, C., Minciardi, R., Sacile, R., & Trasforini, E. (2009). Network planning of fuelling
service stations in a near-term competitive scenario of the hydrogen economy. Socio-Economic Planning Sciences, 43(1), 55–71.
BMVI. (2014). LNG as an alternative fuel for the operation of ships and heavy-duty vehicles. Burel, F., Taccani, R., & Zuliani, N. (2013). Improving sustainability of maritime transport
through utilization of Liquefied Natural Gas (LNG) for propulsion. Energy, 57(December 2005), 412–420.
Dagdougui, H. (2012). Models, methods and approaches for the planning and design of the future hydrogen supply chain. International Journal of Hydrogen Energy, 37(6), 5318–5327. Danish Maritime Authority. (2012). North European LNG Infrastructure Project.
Det Norske Veritas. (2014). Liquefied Natural Gas ( LNG ) Bunkering Study.
Eisenhardt, M. (1989). Building Theories from Case Study Research. The Academy of Management Review, 14(4), 532–550.
Finger, M., & Holvad, T. (2013). Regulating Transport in Europe. Edward Elgar Publishing.
Ghiani, G., Laporte, G., & Musmanno, R. (2004). Introduction to Logistics Systems Planning and Control (1st ed.). Sussex: John Wiley & Sons Ltd.
Hakimi, S. L. (1964). Optimum locations of switching centers and the absolute centers and medians of a graph. Operations Research, 12(3), 450–459.
Hodgson, M. J. (1981). The location of public facilities intermediate to the journey to work. European Journal of Operational Research.
Hugo, a, Rutter, P., Pistikopoulos, S., Amorelli, a, & Zoia, G. (2005). Hydrogen infrastructure strategic planning using multi-objective optimization. International Journal of Hydrogen Energy, 30(15), 1523–1534.
Karlsson, C. (2009). Researching Operations Management (1st ed.). New York: Taylor & Francis.
Kim, J.-G., & Kuby, M. (2012). The deviation-flow refueling location model for optimizing a network of refueling stations. International Journal of Hydrogen Energy, 37(6), 5406–5420. Klose, A., & Drexl, A. (2005). Facility location models for distribution system design. European
Journal of Operational Research, 162(1), 4–29.
Kuby, M., & Lim, S. (2005). The flow-refueling location problem for alternative-fuel vehicles. Socio-Economic Planning Sciences, 39(2), 125–145.
Lim, S., & Kuby, M. (2010). Heuristic algorithms for siting alternative-fuel stations using the Flow-Refueling Location Model. European Journal of Operational Research, 204(1), 51–61. Magrath, A. J., & Hardy, K. G. (1991). Six Steps to Distribution Network Design, (February). Melo, M. T., Nickel, S., & Saldanha-da-Gama, F. (2009). Facility location and supply chain
management – A review. European Journal of Operational Research, 196(2), 401–412.
Panteia. (2013). Contribution to impact assessment of measures for reducing emissions of inland navigation.
Querol, E., Gonzalez-Regueral, B., García-Torrent, J., & García-Martínez, M. J. (2010). Boil off gas (BOG) management in Spanish liquid natural gas (LNG) terminals. Applied Energy, 87(11), 3384–3392.
Santoso, T., Ahmed, S., Goetschalckx, M., & Shapiro, A. (2005). A stochastic programming approach for supply chain network design under uncertainty. European Journal of Operational Research, 167(1), 96–115.
Upchurch, C., & Kuby, M. (2010). Comparing the p-median and flow-refueling models for locating alternative-fuel stations. Journal of Transport Geography, 18(6), 750–758.
Vidal, C., & Goetschalckx, M. (1997). Strategic production-distribution models: A critical review with emphasis on global supply chain models. European Journal of Operational Research, 2217(96).
Voss, C., Tsikriktsis, N., & Frohlich, M. (2002). Case research in operations management. International Journal of Operations & Production Management, 22(2), 195–219.
Yin, R. K. (2009). Case study research: design and methods (4th ed.). Thousand Oaks, CA: Sage Publications Inc.
APPENDIX I Interview Questions 1. Introduction
What is your function within the organization?
What are your tasks and responsibilities within the organization?
2. General questions regarding LNG
To what extent do you see LNG as a viable option as an alternative bunkering fuel for inland shipping?
Which advantages and disadvantages do you see regarding the use of LNG as a fuel?
Which ships do you see as particularly suitable to use LNG?
How viable do you see the option of retrofitting existing ships for the use of LNG?
3. Questions regarding Organization
How are the current plans of your organization for the use of LNG as a fuel in the near future?
How do you plan routes for your ships, and does bunkering play a role in planning those routes?
How do you select bunkering locations for your ships? Are there preferences for specific locations?
Which, if any, additional services would you like to receive from bunkering locations that would influence your decision?
4. LNG in Nijmegen area
How interesting would and LNG bunkering facility in the area of Nijmegen be as a bunkering option for your fleet?
What characteristics would a bunkering facility in this area need to have to be a realistic bunkering option for your ships?
How realistic do you consider stopping at facilities in this area with the sole purpose of bunkering LNG?
APPENDIX II Coding Table
Category Code Representative Data
LN G In la n d Sh ip p in g Investment costs
"The investment costs are quite huge. I understand for the gas barges they
are talking about 1,6 million Euros extra per barge, only for the use of LNG."
Company A Price difference oil and LNG
"The price difference is simply too big. I mean, the low price of gasoline in general is such that it kills any initiative." Company D
Availability
"The logistics to renew all the bunker stations into LNG stations will not be easy. Especially in the hinterland in Germany or Southern Germany it is more and more difficult to get the LNG over there." Company B
Environmental benefits "The advantage is clear, it is better for the environment" Company C
Suitable ships
"Mostly tanker barges and in the future maybe also some dry bulk barges. Maybe because right now dry bulk and container barges are the same, but i think that container barges will look a bit different to the dry bulk barges right now. So maybe in the future it is more and more for container barges. But for dry bulk barges i do not see it that positive." Company B
LNG plans
"We are always looking what the possibilities are. But LNG is at this moment not a real option for us." Company A
Retrofitting
"I wouldn't expect the big amount from converting as it requires substantial redesigning and adaptions to your ship. I would expect acquisitions on new built ships." Company D
Fleet development
"For the coming few years, due to the economic problems, i do not expect many ships will be built new." Company B
B u n ke ri n g Selection
"I must say we are looking for the best price and for service. And thereby we select them. And as service you can understand that they have to be open 7 days a week, 24 hours a day." Company B
Location preference
"The location itself is less important. I mean, you would make it work. You select based on the financial perspective." Company D
Route deviation
"We only bunker when the location is on the route of the ship. No deviations." Company C
Additional services
"I can not imagine a service at the moment that adds value but maybe the bunker owners see some other possibilities." Company A
Ar n h e m N ijme ge n Location
"Geographically it would make sense if you are looking for bunkering to do it
in the Millingen Nijmegen area because you always pass there." Company A
Characteristics
"Around the clock availability to provide bunkers. And if they would make a new one, I would simply advise them to take on the bigger ships as well."
Company D
Bunkering "I wouldn't see any constraints in bunkernig over there" Company D
Traffic volume
"In my opinion there will be more traffic. So the expectation for the future is that the transport flow will grow a little bit". Company B
Vlissingen Duisburg 1 Vlissingen Karlsruhe 1 Vlissingen Krefeld 1 Vlissingen Leverkusen 1 Wanssum Rotterdam 1 Werkendam Düsseldorf 1 Worms Rotterdam 1 Worms Sluiskil 1 Zwijndrecht Duisburg 1
