COMPARING DIFFERENT LNG BUNKERING
SOLUTIONS GIVEN VARIOUS DEMAND SCENARIOS
Master’s Thesis
July, 2014
B.F.B. Kolkman, 1824716
Abstract
After January 1, 2015, vessels have to limit their sulphur emissions to 0.1% in a number of seas due to new IMO regulations. Because most of the conventional maritime fuels do not meet the new emission requirements, new fuels and/or techniques must be considered. Liquefied natural gas (LNG) is an interesting option because of the low emission levels and the, in comparison to conventional fuels, relatively low price. However, a widespread infrastructure to provide LNG to vessels does not exist. In this research, we compare different methods to provide LNG to inbound vessels in the port of Amsterdam. The performance of these LNG bunkering solutions is measured by the delivered cost and the time it takes to get bunkered divided by the demand (i.e. the average time it takes to receive one unit LNG). Since the system is not existing, we perform an experimental design with various factors and factor levels. After performing a trade-off analysis, the best factor levels are chosen for final comparisons.
Table of contents 1. Introduction 4 1.1 Introduction 4 1.2 Research questions 5 1.3 Methodology 6 1.4 Validation 8 2. Problem description 9
2.1 Terminal-to-ship with decentralized storage (PTS) 9
2.2 Ship-to-ship with decentralized storage (STSD) 10
2.3 Truck-to-ship with centralized storage (TTS) 10
2.4 Ship-to-ship with centralized storage (STSC) 11
2.5 General assumptions 12 2.6 Inbound vessels 12 2.6.1 Arrivals 12 2.6.2 Market share 13 2.6.3 Demand 13 2.7 LNG infrastructure 15 2.7.1 LNG terminal 15 2.7.2 Bunker vessel 16 2.7.3 Bunker truck 16 2.8 Performance measures 17 2.8.1 Time 17 2.8.2 Costs 17 3. Simulation model 18 4. Experimental design 23 4.1 Experiments 23
4.2 Warm-up, run-length and replication size 23
4.3 Outcomes of the experimental design 24
5. Results 25
References 29 Appendix A – Interview protocol – Oliehandel Klaas de Boer B.V., Harlingen 33
Appendix B – Interview protocol – Feederlines shipmanagement, Groningen 35
Appendix C – Interview protocol – Oliehandel Klaas de Boer B.V., IJmuiden 37
1. Introduction
1.1 Introduction
On January 1, 2015, new regulations on the sulphur emissions of ships in the English Channel, the North Sea and the Baltic Sea will take effect: in order to prevent pollution, the International Maritime Authority (IMO) limits the sulphur content in maritime fuels to 0.1%. Most of the conventional maritime fuels do not meet those regulations (Corbet & Winebrake, 2008). Therefore, (new) fuels and technologies that satisfy the IMO regulations must be considered (Wang et al., 2007; Danish Maritime Authority1, 2012; Laugen, 2013).
Liquefied natural gas (LNG) is considered an interesting option because of the low emission levels and the lower price compared to oil-based fuels and, thus, lower ship operating costs (Burel et al., 2013; Laugen, 2013). LNG is natural gas cooled down to -162 degrees Celsius to make the gas liquid. Through this liquefaction, the volume decreases to 1/600th of its initial volume; this makes LNG suitable for storage and transport (DMA, 2012).
Although LNG is a feasible solution to satisfy the sulphur emissions regulations, a widespread infrastructure to provide LNG to maritime users does not exist (Laugen, 2013). Potential supplies of natural gas will not invest until sufficient demand exists and ship owners will not use LNG until LNG can be supplied sufficiently: a so-called chicken-and-egg problem (Jensen & Ross, 2010; DMA, 2012). In order to increase the availability of LNG, ‘investments are necessary at all stages of the
LNG supply chain: exploration and development, transmission systems, LNG infrastructures and LNG storage capacity’ (Kumar et al., 2011B).
In order to develop an infrastructure, decisions at different levels must be made. The logistics of such an infrastructure can be divided into three levels of decision making: strategic level, tactical level and operational level (Hendriks, 2009). Applied to the LNG supply chain, strategic decision making concerns the network approach: the number, location and sizes of locations where LNG will be supplied are chosen. On tactical level, the layout of the particular locations will be determined. Decisions on operational level relate to short-term decisions (i.e. day-to-day decisions).
In the maritime sector fuels are supplied to end-users (i.e. inbound vessels) through bunkering. The bunkering operations are performed by bunkering companies. Bunkering is considered to be ‘a core
aspect of the maritime LNG supply chain’ (DMA, 2012). DMA (2012) and Arnet (2013) describe four
LNG-bunkering and is considered to be a temporary solution. Because the different bunkering methods imply different locations from where the LNG is provided and, therefore, different layouts of the considered locations, it is recommended to consider different decision making levels in the development of an LNG infrastructure. Since bunkering companies provide the needed fuels to the ships, a bunkering company has to decide which bunkering solution is most efficient for a port. Costs are assumed to be the issue in the development of a fuel infrastructure (Ogden et al., 1999), therefore, costs are used as a performance measure. Furthermore, time is considered to be an important factor in bunkering operations (Chang & Chen, 2006), so time will also be used as a performance measure. Since the future maritime LNG demand is unknown, multiple demand scenarios will be used in this research.
In this thesis, the most efficient trade-off between the performance measures (time and costs) will be determined for one port and for a certain bunkering company. The decision entails decision-making on strategic, tactical and operational levels, since it affects the location, the size and the layout of the logistic nodes.
The purpose of this thesis is twofold: first, to investigate how different bunkering methods perform in a port under various demand scenarios. Second, to compare the different bunkering solutions in a port given the various demand scenarios. Modelling and simulation will be used in order to reach the purposes. A case of bunkering company Oliehandel Klaas de Boer B.V. (further mentioned as ´KdB´) in the port of Amsterdam will be performed in the simulation model.
1.2 Research questions
The main research question of this study is:
Which LNG bunkering solution has the best trade-off between time and costs from the perspective of a bunkering company, given different demand scenarios, for the port of Amsterdam?
We developed five sub questions to answer the main research question: 1. Which bunkering solutions should be considered to provide LNG to ships? 2. Which factors should be considered in the different LNG bunkering solutions?
3. Which factor levels are realistic for a comparison of different LNG bunkering solutions?
4. How can the performance of a bunkering solution be measured (i.e. development of the performance measures)?
6. How can the model be used to compare different LNG bunkering solutions?
1.3 Methodology
In this section the methods and techniques that will be used in this thesis are described and motivated. First, the overall methodology is described, after which we present a specific method for each sub question.
The aims of this thesis are (1) to investigate the performance of different bunkering solutions in terms of costs and time and (2) to compare the different bunkering solutions. Modelling and simulation will be used to achieve these aims. A widespread LNG infrastructure doesn’t exist at this moment. According to Robinson (2004) simulation is preferred over experimentation with the real system if the real system is not existing. Furthermore, ‘simulation is able to represent variability,
interconnectedness and complexity’ (Robinson, 2005). Last, modelling and simulation is well suited
for comparing layouts and system configurations: Vis & Anholt (2010) show this for terminal operations and Iannoni & Morabito (2006) show this for transportation and logistics systems. Since time can be an important issue in the maritime sector (Chang & Chen, 2006; Notteboom, 2006), time is considered in the simulation model. In this simulation, time points where states change are relevant (for example: the time an inbound vessel arrives at the port and the point of time where a bunker vessel is out of fuel and must be refilled). The time between state-changing time points is irrelevant (for example: if a bunker vessel sails from the import terminal to an inbound vessel that must be bunkered). Therefore, discrete event simulation is used to simulate the LNG bunkering processes. The construction of the components of the different bunkering methods might take several years (Ocean Shipping Consultants, 2013). In order to simulate and compare all bunkering solutions in a realistic way (e.g. it is impossible to bunker LNG by a terminal in a port next year if the port has no terminal development and construction plans at this moment), the year 2020 is chosen as year to simulate. It implies that we simulate a developed infrastructure that, at the time of writing, does not exist.
Since we want to simulate a non-existing system, the parameters and parameter values for the configuration of the bunkering solutions are currently unknown. Therefore, we have to find out what the parameters for the different bunkering solutions are. An experimental design will be developed to compare different values of parameters for every bunkering solution. In the terminology of experimental design, parameters are called factors; the different values of these factors are called
factor levels (Law, 2007). After the experimental design, the bunkering solutions with the best
ships?´ literature research (of the little literature there is available on (LNG-) bunkering) and
interviews with subject-matter experts (SMEs) from bunkering company KdB and shipping company Feederlines shipmanagement will be conducted. KdB is a supplier of bunker fuels and provides the fuels in multiple ports across the Netherlands. Because KdB operates in multiple ports in the Netherlands, it is considered as a representative company to validate the factors gained from literature and, where appropriate, complement these factors. The port of Amsterdam will be used in the simulation, because KdB provides conventional fuels in this port for several years, but, at this moment, does not provide LNG. Furthermore, sufficient information and data is available about his port. Finally, Amsterdam doesn't have an LNG terminal at the moment, in contrast to, for example, the port of Rotterdam (Port of Rotterdam, 2014). Feederlines shipmanagement manages 37 worldwide-operating ships. The company is able to indicate which factors ship owners concern regarding bunkering. The interview protocols are attached as Appendix A and Appendix B.
The second sub question is ´Which factors should be considered in the different LNG bunkering
solutions?´. Literature only cannot provide the answer to this sub question, since little research is done
on the bunkering of LNG to ships. In order to answer this sub question the limited literature on LNG bunkering and literature on LNG infrastructure and on (the developing of) the infrastructure for other (alternative) fuels will be used. Also, KdB will provide data to answer this sub question.
The third question, ´Which factor levels are realistic for a comparison of different LNG bunkering
solutions?´ will be answered by examining the latest information about ports and bunkering and by
information provided by KdB.
For answering the fourth sub question ´How can the performance of a bunkering solution be measured
(i.e. development of the performance measures)?´ the little literature that is available on bunkering,
research on the maritime sector, literature concerning LNG bunkering and the transport of cryogenic materials will be used.
For answering the fifth and sixth sub question (i.e. ´How can the factors and factor levels be captured
in a model?´ and ´How can the model be used to compare LNG bunkering solutions?´) will be
answered by means of modelling and simulation. The model of Mitroff et al. (1974), as described by Karlsson (2009), will be used to structure the modelling and simulation parts of this thesis.
Figure 1 – Visual representation of the methodology 1.4 Validation
The validation will be done by conducting an interview with an SME in the field of bunkering. The
processes used in the models will be presented to this expert and asked whether they have face validity (see Appendix C). Furthermore, we will compare the behaviour of the inbound vessels and the LNG servers with the processes described by existing literature and obtained from interviews.
chapter, chapter 5, consists of the results of the trade-off analysis of the experimental design. The last chapter provides the conclusions and discussion.
2. Problem description
We want to simulate a non-existing system, since a widespread LNG infrastructure doesn´t exist at this moment. In this chapter, the factors and factor levels of the simulation model and the experimental design will be developed. First, the considered bunkering solutions will be discussed. Second, the general assumptions are listed, followed by the descriptions of the inbound vessels and the LNG infrastructure. Last, the performance measures (i.e. time and costs ) are discussed.
By means of interviews and literature, four bunkering solutions are considered in this thesis. The solutions can be classified in two classes: (1) decentralized storage of LNG and (2) centralized storage of LNG. In the case of centralized storage, the LNG is stored in the import terminal and is supplied by LNG servers (bunker vessels or bunker trucks) that serve multiple ports, whereas in the case of decentralized storage, a local LNG terminal is located in the port from which the LNG is provided to bunker vessels or to the inbound vessels directly. The four bunkering solutions are discussed in the next four subsections (2.1-2.4).
2.1 Terminal-to-ship with decentralized storage (PTS)
LNG is transferred directly from a local terminal to an inbound vessel. Disadvantages of this bunkering solution are the inflexibility of the terminal and the high operational costs (Arnet, 2013). The size of a terminal can vary up to 100,000 m3. The local terminal will be replenished by a replenishment vessel (that is provided with LNG by the import terminal) if its LNG level reaches a certain threshold. When an inbound vessel arrives at the port, it navigates to the terminal where it will be bunkered. If the terminal is occupied, the inbound vessel has to wait until the terminal is free. The terminal in this bunkering solution only serves inbound vessels in the port of consideration, hence only one port can be considered in this solution.The PTS bunkering solution is presented visually in Figure 2.
2.2 Ship-to-ship with decentralized storage (STSD)
Bunkering with bunker vessels is the major bunkering method for maritime fuels. This kind of bunkering can take place both onshore and offshore, but is mostly classified as offshore. Discharging
unit is 200 – 10,000 m3 (Arnet, 2013). An advantage of bunker vessels is that ‘bunkering from vessels
is flexible with respect to covering several sizes and locations. This lowers the costs and the time spent on bunkering’ (Arnet, 2013). This bunkering solution consists of a local terminal located in the port of
consideration. The terminal will be replenished by a replenishment vessel that receives its fuel from the import terminal. When an inbound vessel arrives at the port, it sails to a quay where it will moor. If a bunker vessel is available, the bunker vessel navigates to the quay where the inbound vessel is moored and provides the necessary LNG. If no bunker vessel is available, the inbound vessel has to wait until a bunker vessel is free. Since the bunker vessels in this bunkering solution only serve inbound vessels in the port of consideration, we only consider one port: the port of consideration (i.e. the port of Amsterdam). In Figure 3, the STSD solution is presented visually.
Figure 3 – Visual representation of the STSD bunkering solution 2.3 Truck-to-ship with centralized storage (TTS)
LNG is supplied by trucks with an LNG tank size up to approximately 50 – 100 m3. Because trucks
multiple times by the import terminal to fulfill the demand of the inbound vessel. A visual representation of the TTS solution is presented in Figure 4.
Figure 4 – Visual representation of the TTS bunkering solution. 2.4 Ship-to-ship with centralized storage (STSC)
At this moment, bunkering by bunker vessels is the major bunkering method for conventional
maritime fuels. Discharging unit is 200 – 10,000 m3 (Arnet, 2013). Research on optimizing of tanker
routing (for bunkering in general) has been done: Werners & Kondratenko (2009) developed a model to optimize the tanker routing problem. As mentioned in Section 2.2, an advantage of bunkering with bunker vessels is the mobility of the bunker vessels. For this bunkering solution, LNG is provided by the import terminal to one or more bunker vessels. When an inbound vessel arrives at the port, it will navigate to a quay where it will moor. If a bunker vessel is available, it sails to the inbound vessel and provide the vessel with LNG. If no bunker vessel is available, the inbound vessel has to wait until a bunker vessel is free and able to provide the needed LNG. The bunker vessels serve multiple ports. A visual representation of the STSC bunkering solution is showed in Figure 5.
2.5 General assumptions
Since a non-existing system is simulated and since we only want to include relevant aspects of the system in the model, several assumptions are made. The main assumptions of the model are: - The bunker vessels, bunker trucks, replenishment vessel and terminal only supply LNG. - Only LNG fuelled inbound vessels are considered in our model.
- The demand doesn’t change in the considered time interval (see Section 4.2)
- Inbound vessels are served according to the first-come-first-serve policy (see Section 2.7)
- The need for LNG of an inbound vessel is 1.6 times higher than if the same ship uses a conventional fuel. (see Section 2.6.3)
- The arrival of ships is not known in advance by the bunkering company (i.e. ships arrive spontaneously).
- The amount of LNG stored in a local terminal decreases over time through boil-off gasses. (see Section 2.7.1)
- Failures of equipment are not considered in the model (see Section 2.7).
- An inbound vessel will only arrive at the port when a quay is free (see Section 2.6.1)
- Ships sail with a speed of 3.44 m/s (approximately 12 km/h); trucks 22 m/s (about 80km/h) (see Section 2.7.2 and Section 2.7.3)
- The import terminal has an infinite amount of LNG to provide (see Section 2.7.1). - Inbound vessels only leave the system when their demand for LNG is fulfilled. - Units of LNG are expressed in m3; costs in EUR.
2.6 Inbound vessels
2.6.1 Arrivals
Table 1 – Average arrival rate of vessels at the port of Amsterdam for every hour of a day. Extracted from maritimetraffic.com and Port of Amsterdam (2013).
Only inbound vessels which are LNG fuelled and which are clients of the bunkering company are considered in the model. Using the probability that an inbound vessel is LNG fuelled and using the market share of KdB in the port of Amsterdam (see Section 2.6.2), we decide whether the inbound vessel must be bunkered (otherwise, the inbound vessel will be deleted from the system).
In the case of centralized LNG storage bunkering solutions, other ports are considered (see Section 2.6.3). The arrival of ships in other ports are treated quite similar; the only difference is the use of a Poisson process instead of a nonstationary Poisson process due to a lack of data about the other ports.
2.6.2 Market share
In this thesis, different bunkering solutions are compared from the perspective of a bunkering company. Since we do not consider all bunkering operations in a port, but only the bunkering operations executed by the bunkering company of consideration, the market share of the bunkering company in that port must be calculated. In this public version of the thesis, the market share is not
presented, since it is confidential information of KdB.
2.6.3 Demand
The future LNG demand is above all largely dependent on future fuel prices (DMA, 2012). Since infrastructure requirements for refuelling depend on the level of demand (Ogden et al., 1999), the demand of LNG will be discussed in this section.
According to Burel et al. (2013), in 2013, 34 ships used LNG as fuel and the construction of 31 new ships is planned for the period up to 2015. These numbers exclude LNG carriers and inland navigation vessels. Most of these ships are operating in Norway (Burel et al., 2013). The DNV (2012) developed three scenarios for the demand of LNG in 2020 (in terms of the number of LNG-fuelled ships): in the
Interval (time in hours of a day)
Average Interval (time in hours of a day)
Average
Lower bound Upperbound Lower bound Upperbound
low demand scenario 650 ships will be LNG-fuelled, in the base case 1,650 ships will use LNG. In the high demand scenario 5,000 vessels will use LNG as maritime propulsion. Next, the probability that an inbound vessel is a LNG-fuelled can be calculated for each demand scenario, using information of the DMA (2012) and the IMO (2012) about the current size of the total fleet and the forecasted annual growth (which is 2%). The numbers are presented in Table 2.
Table 2 – Forecasted fleet size and the LNG demand scenarios. Extracted from IMO (2012), DMA (2012) and DNV (2012).
Every inbound vessel has an individual demand for LNG. An empirical distribution is used for the demand for LNG of an inbound vessel. The distribution is derived from the historic figures of demand for conventional fuels from inbound vessels that are served in the port of Amsterdam by KdB in the period September 2010 to April 2014. These demand figures for conventional propulsions are transferred into cubic metres (since conventional fuels are given in metric tonnes) and corrected for energy density. This correction for energy density is done because we assume that ships want to be provided by the same amount of energy as conventional fuels, in order to have the same action range compared to conventional fuels. We make this assumption because costs increase as a vessel spends time in a port (Talley & Ng, 2013), therefore vessels want to minimize the time spent in ports (e.g. to bunker). This implies that the need for LNG of an inbound vessel is 1.6 times higher than if the same vessel uses a conventional fuel (GTT, 2013). The empirical distribution is presented in Table 3. The numbers of the intervals are expressed in cubic metres LNG. (This is a public version of the research.
In this version, Table 3 is empty, since the empirical distribution contains confidential information.)
For centralized bunkering methods, five other ports are represented in a simplified way: an empirical distribution to determine the demand for LNG in those ports is derived by accumulating the demand of those ports. The other considered ports are Beverwijk, IJmuiden, Velsen, Scheveningen and Zaandam. The import terminal is located in Rotterdam. The distances between the ports and the import terminal are presented in Table 4.
1 January, 2021 Fraction
Size of total fleet 126,034 100.00%
Number of LNG-fuelled vessels Low demand scenario 650 0.52%
Medium demand scenario 1,650 1.31%
Table 3 – Emperical distibution of the demand for LNG of an inbound vessel in cubic meters. Extracted from historical data from KdB. No figures are presented in this version of the report, since the data is confidential.
Table 4 – Distances (in km) between the different port and locations. Source: Google Maps. 2.7 LNG Infrastructure
In this section, different components of a potential LNG infrastructure will be discussed. Failures are not considered in the model. The inbound vessels will be served according to the first-come-first-served policy (Lorenzoni et al., 2006 & Nishimura et al., 2001). This policy is an oversimplification compared to the policy bunkering companies pursue, but it will hardly affect the outcomes, since the demand for LNG is relative low (max. 3.5% of the total fleet will be LNG-fuelled according to the demand scenarios we consider).
2.7.1 LNG Terminal
The capacity of a local terminal may vary from 100 m3 (Norwegian Maritime Authority, 2012;
Innovation Norway, 2013) up to 100,000 m3 (DMA, 2012). Three factor levels from this interval are
chosen for the experimental design: 1,000 m3 (DNV, 2013; Innovation Norway, 2013), 5,000 m3 (GTT, 2014) and 20,000 m3 (DMA, 2012). The terminal provides LNG to the bunker vessels or inbound vessels. The rate at which the terminal transfers LNG depends on the receiving party: the bunkering rates of fishing vessels and tugs are lower than the transfer rates of container and cargo vessels (DMA, 2012). In this thesis we assume that the LNG transfer rate by terminal and bunker vessels to inbound vessels is 200 m3/h (average taken from, amongst others, DNV, 2013; FKAB, unknown). The bunkering rate to transfer LNG from a bunker truck is 40 m3/h. Before an LNG
Interval (cubic metres LNG) Fraction Interval (cubic metres LNG) Fraction
Lower bound Upper bound Lower bound Upper bound
Beverwijk IJmuiden Scheveningen Velsen Zaandam Rotterdam
transfer starts, a checklist must be completed and the supplier and receiver must be connected by means of a connection hose. After the bunkering operation, some checks will be executed and the hose will be disconnected. For these events thirty minutes are calculated (FKAB, unknown), which is consistent with the results of the interviews. A port can have a maximum of one local terminal.
The reorder policy used in the simulation model is the reorder policy of KdB. The by KdB used reorder policy is not described in this public version of the research, since it is confidential information.
One of the characteristics of LNG is evaporation under normal operation. In this research, evaporation is considered: a boil-off rate of 0.1% per day is applied to the LNG level of the local terminal (Shin et al., 2008). In order to present a realistic terminal fuel level, the vaporization of the terminal content is updated every hour (i.e. the smallest time interval at which the simulation does an action in any case). If the terminal reaches its reorder level due to vaporization, the replenishment vessel will refuel the terminal.
The import terminal (i.e. the LNG supplier of the replenishment vessel in the decentralized bunkering solutions and the supplier of LNG of the bunker vessels and bunker trucks in the centralized solutions) is assumed to have an infinite amount of LNG to provide.
2.7.2 Bunker vessel
Bunker vessels provide LNG to end-users. The capacity of a bunker vessel can vary (DNV, 2013;
DMA, 2012). Three factors levels are chosen to experiment with in the experimental design: 200 m3
(DMA, 2012), 1,000 m3 (Innovation Norway, 2013) and 5,000 m3 (average taken from DMA; 2012).
The replenishment vessel is also a bunker vessel. It has a capacity of 2,500 m3. All bunker vessels and replenishment vessels sail with a speed of about 12 km/h, which is derived from the average speed of a bunker vessel of KdB (maritimetraffic.com). The coupling time of thirty minutes also applies to bunker vessels and the replenishment vessel. The number of bunker vessels may vary. In the decentralized bunkering solutions the maximum number of bunker vessels is two; in the centralized bunkering solutions the maximum number is six (since six ports must be served).
A bunker vessel will be refilled when its fuel level equals zero. In the decentralized bunkering solution the bunker vessel will also be refilled if its fuel level is not equal to its capacity and no LNG fuelled vessel must be bunkered, because the local terminal is located at relatively small distance from the location of the bunker vessel, so it takes relatively little time to refill the bunker vessel.
Driving speed is assumed to be 80 km/h (rijksoverheid.nl). For the experimental design two factors are chosen: 20 m3 (Arteconi & Polonara, 2013)and 40 m3 (DNV, 2013). The number of bunker trucks varies in the experimental design: from one up to ten.
2.8 Performance measures
2.8.1 Time
Chang & Chen (2006) show that bunkering can be the key operation of a port in order to maintain a competitive position. Delays of bunkering operation may cause delays of the departures of the outbound vessel. In their simulation of bunkering operations in the Port of Kaohsiung in China the waiting time of an inbound vessel is used as a performance measure. Furthermore, Notteboom (2006) concludes that time is an important issue in the maritime sector. Since the chosen bunkering solution highly affects the time the bunkering operation takes, we also use time as a performance measure in this study. The time the inbound vessel spends in the port is used as a performance measure. In order to correct this measure for the differences in LNG demand, we divide the time by the LNG demand. This results in a performance measure that measures the average time it takes to receive one cubic metre LNG.
2.8.2 Costs
As mentioned in the Introduction, costs are the issue in developing an infrastructure (Ogden et al., 1999). Since costs are an important factor in the development of an infrastructure, we use it as a performance measure. The costs of (developing) an infrastructure are often presented as delivered costs (i.e. the total costs of the infrastructure divided by the amount of delivered units); examples are Ogden et al. (1999) for a fuel refuelling infrastructure and Stephen et al. (2010) for delivering biomass feedstock to a biomass processing facility. In cost calculations a distinction can be made between investment costs and operational costs.
The investment and operational costs of terminals and bunker vessels are derived from the DMA (2012): we have derived a linear cost function from the financial data the DMA (2012) provides. We are aware that this is an simplification, because the DMA (2012) states that ‘studies show a significant
economy of scale in the LNG terminal business’ and ‘the specific tank cost in EUR/cubic metres decreases when the tank size increases’, but accurate financial data concerning LNG equipment is
hard to find (Mokhatab et al., 2014). However, we believe that the derived cost functions can give good estimations about the delivered cost of a bunkering method and provide the financial information proportional to the bunkering solution.
and therefore, financial data from trucks that transport liquid hydrogen (also a cryogenic material) is used. The data is corrected for currency exchange and inflation. The financial data used in this research is shown in Table 5.
The amount of time it takes to refill the local terminal in the port of consideration is used as the measure for the costs of a replenishment vessel. The costs of the local terminal and the bunker vessels in the decentralized bunkering solutions are only assigned to the port of consideration, since they do not serve inbound vessels in other ports. In the decentralized bunkering solutions the costs of the bunker vessels and bunker trucks are assigned to the port of consideration by using the proportion LNG provided in that port. We chose the delivered units of LNG as a cost driver, because interviews have revealed that in the bunkering industry costs for delivering maritime propulsions are usually calculated per delivered unit of fuel.
Table 5 – Costs and economic lifetime of the different LNG servers. Sources: DMA (2012), Amos (1998) and American Transportation Research Institute (2012).
3. Simulation model
In this chapter, the implementation of the processes described in the previous chapter will be discussed. The terminology of the simulation software package is printed in italic font.
Tecnomatix Plant Simulation 10.1 is used as simulation software. It is discrete event simulation software primarily designed for plant planning. However, because of the flexibility of the objects in the software it is also suitable to simulate port operations. Ha et al. (2007) show this with their simulation model of container terminals. Four simulation models are created (one for every bunkering solution) within one frame: the port of Amsterdam. Inbound vessels are treated as entities and are represented as moving units in the simulation model. Each hour, the number of inbound vessels arriving will be determined by a probability function extracted from maritimetraffic.org and the Port of Amsterdam (2013). This number of inbound vessels will be distributed over one hour according to a negative exponential function. After the creation of the inbound vessel, multiple user-defined attributes are assigned to the inbound vessel: VesselID, LNGDemand, ArrivingTimeInPort,
TimeOfLNGServerAssignment, StartTimeBunkering, EndTimeBunkering. Using data from IMO (2012)
Capacity (in cubic metres LNG)
Investment costs (in EUR * 1,000)
Economic lifetime (in years)
Annual operational costs (in EUR * 1,000)
Section 2.6.3). If it is not, the vessel will be deleted from the system. The bunkering company has a certain market share in the port of consideration. Using that market share, it will be decided whether the LNG vessel is a client of the bunkering company (see Section 2.6.2). If the LNG vessel is not a client of the bunkering company, again, the inbound vessel will be deleted from the system. The demand for LNG of the created vessel is determined by an empirical distribution based on the historic demand of the bunkering company in the port of consideration (see Section 2.6.3). After the creation of the entity at a certain point in time, the inbound vessel moves to the port and to the chosen bunkering solution.
In the PTS case (see Figure 6), the inbound vessel moves to the terminal (which is a resource and is represented as a SingleProc). The terminal has the following user-defined attributes: LNGCapacity,
LNGLevel and TransferRate. The terminal can serve one vessel at a time. If the terminal is occupied,
Figure 6 – Overview of processes regarding the PTS bunkering solution
If the chosen bunkering solution is STSD (see Figure 7), the inbound vessel moves to a quay (represented by a Store). If all available bunker vessels are already assigned to other inbound vessels, the inbound vessel waits in a buffer (i.e. Sorter) and is added to a CardFile (i.e. a queue file that has a first-in-first-out structure). Bunker vessels are treated as resourses in the simulation and represented by
moving units. The following user-defined attributes are assigned to a bunker vessel: LNGCapacity, LNGLevel, SailingSpeed, TimeOfAssignment, AssignedQuay and TransferRate. As soon as a bunker
minimize the time the inbound vessel is in the system. The average distance in the port is used as the value for the distance between the several locations in the port. After each refuelling operation of the terminal, the terminal level is checked and and it is determined whether the terminal needs to be refueled. An inbound vessel only leaves the system if its LNG demand is fulfilled.
Figure 7 – Overview of processes regarding the STSD bunkering solution
If TTS is chosen as the bunkering solution (Figure 8), then the inbound vessel moves to a quay (represented as a Store). Bunker trucks are treated as recourses and are represented as moving units in the simulation program. Multiple user-defined attributes are assigned to bunker trucks: LNGCapacity,
LNGLevel, DrivingSpeed, TransferRate, TimeOfAssignment, AssignedPort and AssignedQuay. If all
system (in the port of consideration nor in the other ports), the bunker trucks are stored in a Store. When an inbound vessel is mooring, the bunker truck(s) will be provided with LNG by the import terminal, a SingleProc, and drive to the port where the inbound vessel is moored. The inbound vessel will be bunkered. If the bunker truck is out of fuel, it will be checked whether another bunker truck is waiting behind the previous bunker truck. If that is the case, this bunker truck will continue the bunkering operation. If the number of bunker trucks is limited, the bunker truck(s) have to drive multiple times to the import terminal and back to the inbound vessel to fulfil the LNG demand of the inbound vessel.
Figure 8 – Overview of the processes regarding TTS bunkering solution
The last bunkering solution, STSC (Figure 9): the inbound vessel moves to a quay (represented by a
Store). Also in this bunkering solution, the bunker vessels are treated as resources and are represented
vessel is operational), then another bunker vessel will be assigned to the inbound vessel. When there are no inbound vessels that have to be bunkered, the bunker vessels are stored in the port where the expected demand is highest.
Figure 9 – Overview of the processes regarding the STSC bunkering solution 4. Experimental design
In Chapter 3, factors and factor levels were developed to emulate different bunkering methods in a realistic way. In this chapter, we will discuss the performed experiments with these developed factor levels and describe the results of the various experiments.
4.1 Experiments
In the experimental design we vary the terminal capacity, the bunker vessel capacity, number of bunker vessels, bunker trucks capacity and the number of bunker trucks within the different demand scenarios, as described in Section 2. In total, 177 possible experiments are found. The experiments in which the bunker vessel capacity exceeds the terminal capacity (in the STSD solutions) are deleted, resulting in a total of 171 experiments. For every bunkering solution, the different possible combinations are compared within the same demand scenario.
4.2 Warm-up, run-length and replication size
the performance of various bunkering solutions under normal operations, therefore a steady-state simulation is built. This implies that demand will not change within the chosen run length. Furthermore, in simulation, it is possible that due to the choices of the initial conditions, the output of the initial transient is not representative of the steady-state mean, resulting in biased estimations (Law, 2007). In our simulation model, the initial situation is representative of the system, since most of the time the system is idle. We illustrate this in Figure 10. The dots in Figure 6 represent when an LNG-fuelled inbound vessel is in the port, in the high demand scenario, in the first year of a randomly chosen simulation replication.
Figure 10 – LNG-fuelled vessels in the port of Amsterdam, in the high demand scenario.
Robinson (2005) states that, as a rule of thumb, the length of a run must be at least 10 times the warm-up period. As noticed, no warm-warm-up period is observed, therefore we assume that a run length of 10 years is sufficient. Because of the run length of 10 years and because a paired t-test (see Section 4.3) requires more than one observation, one simulation run is not sufficient to obtain results with the desired accuracy. As a result, multiple replications must be performed. The Confidence Interval Method is used to determine the number of replications (Robinson, 2005). This method is applied to the outcomes of the time performance measure, since the delivered costs are not calculated during the run of a simulation, but after the experiments are performed. For the experimental design we want to obtain an accuracy of at least 90%. We found that a number of 105 experiments is sufficient for the STSC solutions in the low demand scenario. For the other experiments, 50 replications are sufficient to obtain the desired accuracy.
4.3 Outcomes of the experimental design
Figure 11 – Trade-off points of STSD in the low demand scenario
The ‘X’ marks the best trade-off between costs and time given the demand scenario and the chosen factors levels. In Figure 11, the ‘X’ refers to the STSD solution executed with a local terminal capacity of 1,000 m3 and one bunker truck with a capacity of 200 m3 (which has a distance of 3,566 to the origin). A off is made for every bunkering solution for every demand scenario. The best trade-offs for every bunkering solutions are presented in Table 6.
Table 6 – The best trade-off bunkering solutions divided in the three demand scenarios 5. Results
To do a final comparison between the best bunkering solutions within the demand scenarios, we perform the experiments with 160 replications to obtain an accuracy of at least 95%. Again, the Confidence Interval Method is used to calculate this replication size (Robinson, 2005). The runs are executed with a run length of 10 years.
Bunkering solution Capacity terminal in cubic metres LNG Number of bunker vessels/bunker trucks Capacity of bunker vessels/bunker trucks in cubic metres LNG Delivered cost in EUR/cubic metre LNG Time to bunker a cubic metre LNG (in seconds)
Low demand scenario PTS 1,000 - - € 2,283.64 56.32
STSD 1,000 1 200 € 3,564.83 72.88
TTS - 2 40 € 374.11 316.12
STSC - 1 200 € 666.86 299.61
Medium demand scenario PTS 1,000 - - € 869.44 54.57
STSD 1,000 1 200 € 1,354.67 71.05
TTS - 3 40 € 208.81 292.29
STSC - 1 1,000 € 305.29 118.19
High demand scenario PTS 1,000 - - € 294.00 60.54
STSD 1,000 1 200 € 454.66 75.97
TTS - 4 40 € 93.81 360.51
Table 7 – Results of the 140 runs of the best trade-off points obtained by the previous section
Table 7 shows the results of the final experiments. The average waiting time is the average time before the bunkering operation starts is in all demand scenario the lowest for the PTS bunkering solution: 186 seconds. The waiting time for the TTS bunkering solution is the highest of the waiting times of all bunkering solutions: 9,469 seconds, that is about 2.6 hours. The bunkering operation starts when the LNG server couples with the inbound vessel. The average bunkering time is the time the inbound vessel leaves the system minus the time the bunkering operation starts. Again, the average bunkering time is lowest for the PTS bunkering solution and highest for the TTS solutions. The time an inbound vessels spends in the system is the sum of the waiting time and the bunkering time. The average number of LNG fuelled inbound vessels is the average number of LNG fuelled vessels of all replications. The results show that, even in the high demand scenario, the system is idle most of the time. This implies that the utilization of the LNG servers is low. The simulation shows that even the utilization of the bunker trucks for the TTS solution in the high demand scenario (the solution that also considers other ports and that has the highest average waiting time and bunkering time) is about 11%. The last two columns of Table 7 show the performance measures. Overall, the STSD solution has the highest delivered costs; TTS has the lowest delivered costs.
We performed a paired-t test for the outcomes of the runs with replication size of 160. Within every demand scenario all possible 6 comparisons are made. A Bonferonni correction is made to obtain an overall accuracy of 95%, resulting in an individual confidence interval of 99.17%. All pairwise comparisons are significant; except the comparison between the TTS solution with two bunker trucks
of 40 m3 and the STSC solution with one bunker vessel with a capacity of 200 m3 in the low demand
scenario. However, since the TTS solution has a lower delivered cost, TTS is preferred over STSC in the low demand scenario.
Bunkering solution Average waiting time (in seconds) Average bunkering time (in seconds) Average number of LNG fuelled inbound vessels Total fullfilled LNG demand in cubic metres Delivered cost in EUR/cubic metre LNG
Average time in system (in seconds) divided by demand (in cubic metres)
Figure 12 – Time to bunker one cubic meter plotted against the delivered costs; the crosses mark the best trade-offs.
Again, the best trade-offs are determined, see Figure 12. The red line connects the trade-off points of the different bunkering solutions in the low demand scenario. The blue line connects the medium demand scenario points for the various bunkering solutions. The green line connects the trade-off points of the high demand scenario. An ´X´ marks the best trade-off given that demand scenario (i.e. the smallest distance to the origin). In the low demand scenario, the TTS solution configuration with two trucks with a capacity of 40 m3 each results in the best trade-off between the performance measures. The associated costs are EUR 366.88 per delivered cubic metre LNG and the average time to bunker one cubic metre LNG is 316 seconds. In the medium and high demand scenario the STSC solutions has the best trade-off between costs and time. With a bunker vessel size of 1,000 m3, the delivered costs are EUR 305.48 and EUR 102.18 for the medium and high demand scenarios, respectively. The associated bunkering times (on average, for one cubic metre LNG) are 120 and 142 seconds, for the medium and high demand scenario respectively.
The results are not generalizable to other ports and to other bunkering companies, since the results depend on specific input on KdB and the port of Amsterdam. However, we showed that the methodology, the LNG bunkering processes and the factors described in this research can be used to decide which LNG bunkering solution has the best trade-off between costs and time for a bunkering company in a given port and, therefore, can be applied to other cases.
6. Conclusions and discussion
and factor levels, performing the bunkering operations with two bunker trucks that are refilled by the import terminal in the port of Rotterdam is the most efficient bunkering solution in the low demand scenario. In the medium and high demand scenario, executing the bunkering operations with one bunker vessel (with a capacity of 1,000 m3) that is refilled in Rotterdam is the most efficient bunkering solution, given the chosen factors and factor levels.
It must be emphasized that we simulated a non-existing system and that other (perhaps more suitable) bunkering solutions may emerge in the period between the moment of writing and 2020. Furthermore, a limited number of factors and factor levels are used in the simulation. To determine the optimal configuration of the bunkering solutions, a more extensive factorial analysis should be performed. Also, other refilling and service policies could be carried out to obtain more sophisticated results. Last, obtaining more accurate financial data would increase the accuracy of the results.
Acknowledgements
I would like to thank Prof. dr. Vis for supervising this master’s thesis, for the clear explanations concerning simulation (‘the dice’) and for, when I had questions, finding time in her agenda, even when there was no free time in her agenda. I would also like to thank Simon Thunnissen for the helpful questions, comments and feedback and for all the supportive meetings every week. I would like to thank Jose Lopez for always being available for questions and for the helpful contributions during the development of the simulation model. Also, I want to thank dr. Bokhorst, for answering all my questions concerning Plant Simulation, every time I walked by.
With providing all necessary data and support, Jeroen Heijne, director of Oliehandel Klaas de Boer B.V. was very helpful to me. Without his support and data, it would have been unable to do this research project. I would like to thank Mr. Hartog for doing the interview and validate my models, even after the official appointment. Furthermore, the employees of KdB who learned me to bunker in the port of Harlingen must be thanked. Last, the author wants to thank Mr. Bootsma, coach and advisor at Feederlines shipmanagement, and representatives of Groningen Seaports for the helpful insights in the maritime sector.
References
American Transportation Research Institute, 2012. An Analysis of the Operational Costs of
Trucking: An 2012 Update, http://www.glostone.com/wp-
content/uploads/2012/09/ATRI-Operational-Costs-of-Trucking-2012.pdf
Last retrieved on 23 July, 2014.
Amos, W.A., 1998. Costs of Storing and Transporting Hydrogen,
http://www.nrel.gov/docs/fy99osti/25106.pdf
Last retrieved on 23 July, 2014
Arnet, N.M.L., 2013. Evaluation of technical challenges and need for standardization for
LNG bunkering, Master thesis, Norwegian University of Science and Technology,
Trondheim.
Arteconi, A. & Polonara, F., 2013. LNG as vehicle fuel and the problem of supply: The
Italian case study, Energy Policy, Vol. 62, pp. 503 – 512.
Burel, F., Taccani, R. & Zuliani, N., 2013. Improving sustainability of maritime transport
through utilization of Liquefied Natural Gas (LNG) for propulsion, Energy, Vol. 57,
pp. 412 – 420.
Chang, Y.C. & Chen, C.C., 2006. Knowlegde-based simulation of bunkering services in the port of Kaohsiung, Civil Engineering and Environmental Systems, Vol. 23, Is. 1,
pp. 21 – 34.
Corbett, J.J. & Winebrake, J.J., 2008. Emissions Tradeoffs among Alternative Marine Fuels: Total Fuel Cycle of Analysis of Residual Oil, Marine Gas Oil, and Marine Diesel Oil, Journal of the
Air & Waste Management Association, Vol. 58, pp. 538 – 542.
Danish Maritime Authority, 2012. North European LNG Infrastructure Project – A
feasibility study for an LNG filling station infrastructure and test of recommendations.
http://www.dma.dk/themes/LNGinfrastructureproject/Documents/Final%20Report/
LNG_Full_report_Mgg_2012_04_02_1.pdf
Last retrieved on 23 July, 2014.
DNV, 2012. Shipping 2020, http://www.dnv.nl/binaries/shipping%202020%20-
%20final%20report_tcm141-530559.pdf
Last retrieved on 23 July, 2014.
DNV, 2013. LNG Fuel Bunkering in Australia: Infrastructure and Regulations,
http://www.dnv.com/binaries/public%20version%20-
%20jip%20on%20lng%20bunkering%20in%20australia%20dnv%20andpartners%20january
%202013_tcm4-539390.pdf
Last retrieved on 23 July, 2014
http://www.fkab.com/fileadmin/Files_to_pages/FKAB/Rapporter_Presentationer/FKAB_prese
ntation_-_LNG_fuel_for_shipping.pdf
Last retrieved on 23 July, 2014.
GTT, 2013. LNG bunkering for commercial vessels – solutions for storage tanks, feeder vessels and
vessel fuel tanks,
http://www.lngbunkering.org/sites/default/files/2013%20GTT%20LNG%20BUNKERING%2
0FOR%20COMMERCIAL%20VESSELS%20-
SOLUTIONS%20FOR%20STORAGE%20TANKS,%20FEEDER%20VESSELS%20AND%
20VESSEL%20FUEL%20TANKS.pdf
Last retrieved in 23 July, 2014
GTT, 2014. LNG Bunkering Solutions,
http://www.gtt.fr/wp-content/uploads/2013/02/GTT_Bunkering_Brochure.pdf
Last retrieved on 23 July, 2014
Ha, B.H., Park, E.J. & Lee, C.H., 2007. A simulation model with a low level of detail for
container terminals and its applications, Proceedings of the 2007 Winter Simulation
Conference.
Hendriks, M.P.M., 2009. Multi-step optimalization of logistics networks: strategic, tactical and
operational decisions, dissertatie, Technische Universiteit Eindhoven.
Iannoni, A.P. & Morabito, R., 2006. A discrete simulation analysis of a logistics supply
system, Transportation Research Part E, Vol. 42, 191 – 210.
Innovation Norway, 2013. Experience from Norwegian Small Scale LNG Use and
Distribution, The 10th ASCOPE Conference & Exhibbition.
International Maritime Organization, 2012. International Shipping Facts and Figures –
Information Resources on Trade, Safety, Security, Environment,
http://www.imo.org/KnowledgeCentre/ShipsAndShippingFactsAndFigures/TheRolean
dImportanceofInternationalShipping/Documents/International%20Shipping%20-
%20Facts%20and%20Figures.pdf
Last retrieved on 23 July, 2014.
Jensen, M.W. & Ross, M., 2010. The Ultimate Challenge: Developing an Infrastructure for
Fuel Cell Vehicles, Environment, Vol. 42, No. 7, pp. 10- 22.
Karlsson, C., 2009. Researching Operations Management, Routledge, New York.
Kia, M., Shayan, E. & Ghoth, F., 2002. Investigation of port capacity under a new approach
by computer simulation, Computers & Industrial Engineering, Vol. 42, pp. 533 – 540.
Kumar, S., Kwon, H.T., Choi, K.H., Cho, J.H., Lim, W. & Moon, I., 2011(B). Current status
and future projections of LNG demand and supplies: A global prospective, Energy
Lam, J.S.L., Chen, D., Cheng, F. & Wong, K., 2011. Assesment of the Competitiveness of
Ports as Bunkering Hubs: Emperical Studies on Singapore and Shanghai,
Transportation Journal, Vol. 50, No. 2, pp. 176 – 203.
Laugen, L., 2013. An Environmental Life Cycle Assessment of LNG and HFO as Marine
Fuels, Master thesis, Norwegian University of Science and Technology.
Law, A.M., 2007. Simulation Modelling & Analysis, McGraw-Hill, New York.
Lorenzoni, L.L., Ahonen, H. & Alvarenga, A.G., 2006. A multi-mode resource-constrained
scheduling problem in the context of port operations, Computers & Industrial
Engineering, Vol.50, pp. 55 – 65.
Mitroff, I.I., Betz, F., Pondy, L.R. & Sagasti, F., 1974. On managing science in the systems
age: Two schemas for the study of science as a whole systems phenomenon,
Interfaces, Vol. 4, No. 3, pp. 46 – 58.
Mokhatab, S., Mak, J.Y., Valappil, D.A. & Wood, D.A., 2014. Handbook of Liquefied
Natural Gas, Elsevier Science & Technology, Oxford, 2014.
Nishimura, E., Imai, A. & Papadimitriou, S., 2001. Berth allocation planning in the public
berth system by genetic algorithms, European Journal of Operational Research,
Vol. 131, pp. 282 – 292.
Notteboom, T.E., 2006. The Time Factor in Liner Shipping Services, Maritime Economics
& Logistics, Vol. 8, pp. 19 – 39.
Norwegian Maritime Authority, 2012. Approval of Ships using LNG as fuel.
http://ahk.de/fileadmin/ahk_norwegen/Dokumente/Presentasjoner/lng/KARLSEN_Ho
w_do_authories_approve_LNG_ships.pdf
Last retrieved in 23 July, 2014.
Ocean Shipping Consultants, 2013. LNG as a Bunker Fuel: Future Demand Prospects &
Port Design Options, press release, http://www.maritime-
rh.com/maritime_docs/osc_press_releases/press-release-lng-as-a-bunker-fuel-osc.pdf
Last retrieved on 23 July, 2014.
Ogden, J.M., Steinbugler, M.M. & Kreutz, T.G., 1999. A comparison of hydrogen, methanol
and gasoline as fuels for fuel cell vehicles: implications for vehicle design and
infrastructure development, Journal of Power Sources, 79, pp. 143 – 168.
Port of Amsterdam, 2013. Annual Report 2012,
http://www.portofamsterdam.com/docs/uk/Annual%20Report/Annual%20report%202
012_%20for%20print.pdf
Port of Rotterdam, 2014. Gate terminal to build new facilities for distribution of LNG via vessels and
barges, press release, http://www.portofrotterdam.com/en/News/pressreleases-
news/Pages/gate-terminal-build-facilities-distribution.aspx
Last retrieved on 24 July, 2014.
Robinson, S., 2004. Simulation: The Practice of Model Development and Use, John Wiley & Sons, Ltd., Chichester.
Shin, M.W., Shin, D., Choi, S.H. & Yoon, E.S., 2008. Optimal operation of the boil-off gas
compression process using a boil-off rate model for LNG storage tanks,
Korean Journal Chem. Eng., Vol. 25, pp. 7 - 12.
Stephen, J.D., Sokhansanj, S., Bi, X., Sowlati, T., Kloeck, T., Townley-Simth, L. &
Stumborg, M.A., 2010. The impact of agricultural residue yield range on the delivered
cost to a biorefinery in the Peace River region of Alberta, Canada, Biosystems
Engineering, Vol. 105, pp. 298 – 305.
Talley, W.K. & Ng, M.W., 2013. Maritime transport chain choice by carriers, port and shippers, International Journal of Production Economics, Vol. 142, pp. 311 – 216.
Vis, I.F.A. & Van Anholt, R.G., 2010. Performance analysis of berth configurations at
container terminals, OR Spectrum, Vol. 32, pp. 453 – 476.
Wang, C., Corbett, J.J. & Winebrake, J.J., 2007. Cost-Effictiveness of Reducing Sulfur
Emissions from Ships, Environmental Science & Technology, Vol. 41, No. 24,
pp. 8233 – 8239.
Appendix A – Interview protocol – Oliehandel Klaas de Boer B.V., Harlingen, April 4, 2014
Middels het interview wil ik proberen informatie te verkrijgen over twee soorten informatie. De eerste soort informatie is de ‘procedurele’ informatie over de procedure van het bunkeren een individueel schip (zie Figure 2), uitgezet tegen tijd. Hieronder staat per stage uit Figure 2 vermeld welke informatie ik graag zou willen verkrijgen en welke vragen ik zou kunnen stellen.
Voorafgaand – Hoe geeft een schip aan dat het gebunkerd wil worden?
(Hoe) wordt het bunkeren van een schip gepland? Welke factoren
spelen mee bij het inplannen van te bunkeren schepen?
Welke voorbereidingen moet het bunkerbedrijf doen voordat een schip
aankomt?
Incheck - Wat wordt er gedaan als er een schip aankomt? Volgens literatuur
moeten er procedures worden uitgevoerd (e.g. formulieren invullen).
Welke zijn dat en waarom?
Bunkeren - Hoe wordt er gebunkerd (i.e. technieken en ‘bunkering solutions’)? En
wat bepaald hoe er wordt gebunkerd?
Hoe lang (i.e. tijd van bunkering operation) wordt er gebunkerd? Waar
is dit afhankelijk van?
Uitcheck - Wat moet er nog gebeuren voordat het schip weer kan vertrekken?
Facturering - Hoe wordt er betaald/gefactureerd?
Tweede soort informatie is meer de ‘conceptuele’ informatie en hangt sterk samen met het model van mijn scriptie (Figure 2). Ik ga me focussen op hoe de processen (op geaggregeerd niveau) verlopen bij de huidige conventionele brandstoffen, omdat LNG nog niet gebruikt wordt door Klaas de Boer.
Figure 2 - Conceptueel Conceptueel:
Demand - Wat is de planningshorizon voor het bunkeren van schepen?
- Wat is de verdeling van de verschillende soorten schepen (i.e. size of
ships)?
- Periodiek terugkerende vraag of niet?
- Hoe wordt de hoeveel brandstof dat naar de haven gaat bepaald?
Locatie - Hoe speelt locatie een rol in de keuze voor de bunkering configuratie?
Bunkering solutions - Conventionele brandstoffen zijn niet nieuw, welke solutions worden
gebruikt en waarom? Herinnert u zich nog iets van de introductie van
een van deze brandstoffen? Hoe is dat gegaan?
Bunkering configuraties
Keuze voor bunkering configuratie – Welke factoren spelen hierbij een rol? Waarom is
gekozen voor de bunkering configuraties bij Harlingen haven?
Overig
In mijn scriptie ga ik alleen kijken naar de ‘harde’ kant van het bunkeren (i.e. fysieke proces), maar in hoeverre spelen de niet fysieke aspecten in rol in de beslissingen die worden genomen m.b.t.
Appendix B - Interview protocol – Feederlines shipmanagement, Groningen, April 28, 2014
Doel: inzicht verkrijgen in de verschillen tussen conventionele brandstoffen en LNG, vanuit het perspectief van scheepseigenaren. Dit om de data van conventionele brandstoffen te transformeren naar data die gebruikt kan worden voor het modelleren van een LNG infrastructuur en het ontwikkelen van vraagscenario’s.
Motivatie van keuze voor organisatie: Feederlines is een organisatie betrokken bij het Dinalog-project
Design of LNG Networks en overweegt in de toekomst gebruik gaat maken van LNG als maritieme
brandstof.
Middel: kwalitatief interview met een expert van Feederlines.
Structuur: geprobeerd is om de funnel methode aan te houden. Eerst algemene vragen (over LNG), daarna specifiekere vragen te stellen. De onderstaande vragen zijn richtlijnen: bij bepaalde antwoorden zal er worden doorgevraagd naar motivaties en beweegredenen. Ook als tijdens het interview blijkt dat er bepaalde relevante aspecten niet behandeld worden in dit protocol, dan zullen over deze aspecten ook vragen gesteld worden.
1. Introductie
De vragen in deze sectie betreffen introductievragen, ook om te valideren of de geïnterviewde een expert is.
1.1 Wat is uw functie in de organisatie?
1.2 Wat zijn uw werkzaamheden binnen de organisatie?
2. Algemeen (15 min.)
De vragen in deze sectie betreffen algemene vragen over LNG: de motivatie voor LNG, hoe de organisatie tegenover (een transitie) naar LNG staat, maar ook om te controleren of er een bepaalde bias met betrekking tot LNG bestaat.
2.1 U bent betrokken bij het Dinalog-project Design of LNG Networks, welke redenen heeft uw organisatie hiervoor? Waarom LNG en geen andere mogelijkheden die de zwaveluitstoot verlagen? 2.2 Welke voor- en nadelen ziet u in LNG (dit is al gedeeltelijk tijdens de LNG meeting in Zwolle besproken)?
3. Fysieke verschillen tussen LNG en conventionele brandstoffen voor individuele schepen
Het doel van deze sectie is inzicht te verkrijgen in de veranderingen die gaan optreden als een individueel schip op LNG vaart (in vergelijking met een schip dat op een conventioneel brandstof vaart), dit om de te verkrijgen kwantitatieve data te kunnen omzetten naar data toe te passen op een LNG model.
3.1 Welke verschillen zijn er (of verwacht u) tussen schepen die varen op conventuele brandstoffen en schepen die varen op LNG m.b.t.
- Inhoud brandstoftanks? (Kunt u dit kwantificeren, bijvoorbeeld door een percentage van de
toename/afname te geven? Verwacht u dat dit over tijd gaat veranderen? Hoe gaat dit
veranderen?)
- De tijd die het duurt om te bunkeren? (Kunt u dit kwantificeren, bijvoorbeeld door
een percentage van de toename/afname te geven? Verwacht u dat dit over tijd gaat
veranderen? Hoe gaat dit veranderen?)
- Materiaal van de tanks?
- Motoren (i.e. de ruimte die een LNG-motor inneemt)?
- Overige ruimte die wordt ingenomen in verband met het varen op LNG (bijvoorbeeld
isolatie- of koelingssystemen)? (Kunt u dit kwantificeren?)
4. Verschillen in flexibiliteit van schepen die op LNG en conventionele brandstoffen varen
Doel van deze sectie is inzicht te verkrijgen in de factoren die invloed hebben op de flexibiliteit van een schip dat LNG gebruikt.
4.1 Hoe wordt de route van een schip van Feederlines gepland? Welke rol speelt bunkeren in het plannen van een route?
4.2 Als u zou gaan overstappen naar LNG, hoe ziet u de bevoorrading van LNG voor u (bv. zelf de bevoorrading organiseren of een bunkerbedrijf in de arm nemen; welke manieren van bunkeren zou u overwegen als u de supply zelf gaat organiseren; welke factoren spelen mee in het maken van die keuzes?)
4.3 Welke rol speelt het aanbod van LNG en de manier van LNG bunkeren in een individuele haven in het overstappen naar LNG en het plannen van LNG schepen?
5. Kosten
Deze sectie gaat over de kosten die (een investering van) LNG met zich meebrengen. Doel is om achter aspecten te komen die meespelen met een transitie naar LNG die niet/weinig met de fysieke aspecten te maken hebben, maar ook voor het ontwikkelen van de demand scenario’s.
5.1 Van LNG is bekend dat de investeringskosten hoger zijn dan andere oplossing; LNG is echter goedkoper dan conventionele brandstoffen. Wat verwacht u van de kosten van bunkeren (dus niet de prijs van de brandstof, maar de kosten van de transfer van LNG) – zowel op de lange als korte termijn? Speelt, en zo ja hoe, dit een rol in een eventuele overstap naar LNG?
6. Transitie naar LNG (15 min.)
Deze sectie gaat over een transitie naar LNG als een brandstof. Hoe ziet de organisatie een transitie voor zich? Welke aspecten hebben invloed op hoe de transitie verloopt en de snelheid van een transitie? De antwoorden zullen gebruikt worden voor het ontwikkelen van vraagscenario’s.
6.1 Wat verwacht u dat de markt gaat doen nadat ze nieuwe regelgeving van kracht gaat op 1 jan. 2015? Welke (soort) organisaties zullen op welke manier wanneer gaan overstappen naar LNG? 6.2 Welke policy zou uw organisatie hebben om over te stappen naar LNG (bv. zo snel mogelijk de
Appendix C - Interview protocol – Oliehandel Klaas de Boer B.V., IJmuiden, May 8, 2014
Doel: het valideren van en het leveren van input voor het meest recente conceptuele model.
Motivatie voor de keuze van organisatie: Oliehandel Klaas de Boer B.V. is betrokken bij het Dinalog-project Design of LNG Networks.
Middel: kwalitatief interview met een expert op het gebied van bunkeren
Structuur: geprobeerd is om middels de funnel methode het interview te structureren.
1. Introductie
De vragen in deze sectie betreffen introductievragen; tevens om te valideren of de geïnterviewde een expert is. De sectie zal vooral worden gegaan door een korte inleiding over mijn scriptie, zodat sommige vragen niet vanuit het niets komen.
1.1 Wat is uw functie binnen Oliehandel Klaas de Boer B.V. (afgekort KdB)? 1.2 Wat is uw functieomschrijving?
1.3 Sinds wanneer werkzaam bij KdB? (Het antwoord op deze vraag heeft invloed op de vraag hoe een eventuele infrastructuur in het verleden is ontwikkeld)
2. Bunkeren in het algemeen
Met deze sectie hoop ik inzicht te krijgen hoe een infrastructuur van een maritieme brandstof zich in het verleden heeft ontwikkeld, mits de expert hiervan op de hoogte is.
2.1 Hoe heeft Klaas de Boer in het verleden zijn infrastructuur opgezet? (KdB bestaat al sinds 1914, maar wellicht hebben zij ‘recent’ nieuwe havens aan hun markt toegevoegd. Hoe hebben zij dit gedaan?)
3. Huidige situatie
De antwoord uit deze sectie zullen worden gebruikte om de heuristieken om te bunkeren te ontwikkelen en de parameters te ontwikkelen.
3.1 Welk onderscheid maakt Klaas de Boer in typen schepen dat gebunkerd wordt?
3.2 Hoeveel schepen hebben zich van tevoren aangemeld? Wat is de verhouding tussen incidentele en vaste bunkeroperations?
3.3 Hoe wordt bepaald welk schip wanneer wordt gebunkerd? (I.v.m. priorities – klopt de aanname in het conceptueel model?)
3.4 Hoeveel bunkerschepen zijn er in de verschillende havens waar KdB actief is? Waar is dit afhankelijk van? Medewerkers van deze schepen, zijn dat in alle gevallen ook de operators? Hoeveel personen heeft een bunkerschip nodig om operationeel te zijn?
3.5 Interarrival times van schepen? 3.6 Kostenopbouw
4. Ontwikkeling van een LNG infrastructuur
Een aantal aannames zijn gemaakt in het conceptueel model, zoals de mogelijke bunkerconfiguraties. De vragen in deze sectie valideren te gedane aannames.
4.1 In mijn scriptie zullen vijf bunkerconfiguraties met elkaar worden vergeleken: Gecentraliseerde LNG-opslag met (1) STS of (2) TTS of (3) STS én TTS
