Impacts of Port Productivity and Service Level on Liner Shipping Operating. Cost and Schedule Reliability

Transcription

1 Impacts of Port Productivity and Service Level on Liner Shipping Operating Cost and Schedule Reliability Dun Yang, Abraham Zhang, Jasmine Siu Lee Lam (corresponding author) Division of Infrastructure Systems and Maritime Studies, School of Civil and Environmental Engineering, Nanyang Technological University, Singapore Corresponding ABSTRACT Operating cost and schedule reliability are the two major concerns of container liner shipping operations. Port congestion and port productivity below expectations are among the main causes of schedule unreliability in the liner shipping industry, and they also result in higher operating cost of shipping lines. So far, little research has been conducted to quantify the impacts of port productivity and service level on liner shipping operating cost and schedule reliability. The problem is difficult because liner shipping operations involve many uncertainty time factors and relationships between these factors often lack closed-form expressions. This research builds a discrete event simulation model for quantitative analyses. Port productivity is measured by the number of moves per hour during a vessel s berth time window. Port service level is measured by the percentage of berth-on-arrival for the service of arriving vessels. The simulation model incorporates bunker cost functions and three major time components, namely sailing time, port time and buffer time. Extensive simulation experiments with an Asia-Europe route show the significant effects of port productivity and service level on liner shipping operating cost and schedule reliability. Keywords: Liner shipping; schedule reliability; port productivity; berth-on-arrival; simulation 1

2 1. Introduction A fundamental characteristic of liner shipping is to provide regular and quality services. Among the service requirements by the customers, there has been an increasing emphasis on schedule reliability since schedule unreliability can cause substantial losses for the shippers and consignees especially for those who practise just-in-time production and delivery (Lam & Van de Voorde, 2011; Wang & Meng, 2012 b). On the other hand, higher schedule reliability in most cases also means higher operating cost for the liner companies of which the rise in fuel cost is often the most substantial. Depending on the market economics (demand and supply of liner transportation) and total operating cost, the optimal schedule reliability adopted by liner companies may also change. Despite a steady improvement over the last few years, the overall schedule reliability of the liner shipping industry is still low. According to the survey conducted by the Drewry Shipping Consultant (2012) based on more than 5000 ship arrivals, the average schedule reliability is only 72.3%. Although low schedule reliability can be caused by a number of reasons, in the survey conducted by Notteboom (2006), it is found that over 90% of delayed arrivals are related to port access and terminal operations. This suggests that given higher port productivity and service level, the schedule reliability and cost performance of liner companies can be significantly improved. However, it remains problematic as to how much benefit liner companies can receive from port service improvement, or to what extent ports should improve their services. This research is motivated by the aforementioned challenges faced by major hub ports currently. Ports are expected from shipping lines and shippers to achieve and improve container loading/unloading productivity (Lam and Dai, 2012). However, we do not find any research in quantifying the direct benefits of port productivity that can be brought to port operators in terms of economic performance. Particularly, it is useful for a port operator to measure how much it can charge shipping lines for port service improvements, which can be based on the quantifiable benefits to shipping lines. This research attempts to address this practical need by examining how the increase in port productivity and service level would affect the liner shipping operating cost and schedule reliability. Port productivity is measured by the number of moves per hour during a vessel s berth time window. Port service level is measured by the percentage of berth-on-arrival for the arriving vessels. To the best of our knowledge, this paper is the first quantitative analysis on liner shipping operating cost and schedule reliability in view of port productivity and service level. Changes in operating cost and schedule reliability are simulated and analyzed under different port conditions. Secondly, this paper addresses the practical problems of port service level and productivity planning by estimating the resulted benefits for liner companies at various port service levels. The simulation experiments are conducted for an Asia-Europe route. However, the simulation model can be easily adapted for any other shipping routes and thus serves as a useful planning and reference tool for both port operators and liner companies. This paper is organized in five sections. Section 2 presents a review of previous work in liner shipping and simulation as contrasted to optimization. Section 3 explains the details of the simulation model. In section 4, the simulation results are analyzed and discussed. Lastly, conclusions are drawn in section 5 to summarize the study. 2

3 2. Literature Review Maritime transportation consists of two major stages: ship sailing at sea and ship staying in ports. To increase operating efficiency and reduce transportation cost, numerous studies have been conducted to simulate or optimize shipping and port operations (Alexis, 1981). However, limited research effort has been dedicated to liner shipping operations (Christiansen, Fagerholt, & Ronen, 2004; Wang & Meng, 2012a). Among optimization studies, an early work was done by Boffey et al. (1979) to solve the liner scheduling problems on the route from North West Europe to North East Coast and Canada. By considering speed of the ships and different conditions of ports to be called, an optimization model was built to provide information about optimizing the profitability, transit time, and total buffer time of the route. The model was relatively easy to understand and has been used to by managers to calculate port charges and other values. Perakis and Jaramillo (1991) developed a linear programming model to minimize the annual operating costs for a fleet. The cost components include port dues, canal fees, bunker cost, crew and other costs. Consequently, different methods for adjusting service frequency and vessel speed were also provided. Another example is presented by Ting and Tzeng (2003). In their study, a dynamic programming model was formulated to determine optimal scheduling strategy including cruising speed and quay crane dispatching decisions. Recently, Qi and Song (2012) studied how to design an optimal vessel schedule. The objective function of their optimization model was to minimize the total fuel cost and emission while taking into account the uncertainty of port times and service frequency requirements. An optimization approach has the merit of finding the optimal or near optimal solutions. However, the effectiveness of an optimization model for analyzing liner operations was limited by several constraints. Firstly, due to the high degree of uncertainty and complex geographical and economic constraints on liner operations (Gkanatsas, 2004), the construction and application of an optimization model may be extremely difficult or sometimes even unfeasible at all. Secondly, the complexity of planning and implementation of liner operations lies often not in the optimization of a single objective of cost or profit, but in achieving a fine balance between several inter-related or even contradicting objectives (e.g. speed and cost). As a result, simulation is often preferred and extensively used to evaluate liner shipping and container port operations (Carranza, 2008; Gkanatsas, 2004; Mclean & Biles, 2008). Next, we review the relevant literature on cost performance and schedule reliability of liner operations with simulation approaches. Despite the extensive literature on maritime transportation using simulation methods, there is a dearth of operational research on liner shipping operations (Mclean & Biles, 2008; Wang & Meng, 2012a). Among the relevant scarce literature, an early study on liner operation using simulation modeling was conducted by Datz et al.(1964). In this research, a simulation model was developed to generate and evaluate the profitability of a liner service schedule by taking into account cargo availability. In the research conducted by Bendall and Stent (1999), a simulation model approach was adopted to determine the optimal container fleet type and deployment for a short haul hub-and-spoke transport network. In the quest to achieve an efficient design of schedules and routes, Gkanatsas (2004) developed a discrete event simulation model to help decision making for more robust shipping schedules and to cope with disruptions. A cost penalty function was incorporated to explore and evaluate all possible alternatives (e.g. omit a port, increase nominal speed) to increase the robustness of shipping schedules. In the process, it also considered the tradeoff between the cost of increasing schedule robustness and customers inventory cost which signifies their willingness to pay. A more recent simulation study in liner shipping was conducted by Kleywegt et al. (2005). In this research, a discrete event computer simulation of ocean container movements including liner shipping was 3

4 developed to evaluate all shipping events including vessel events and container events. Voyage costs are evaluated by incorporating separate cost functions and charges on containers such as lashing, lifting and wharfage fees are also considered. The main performance criteria in this model are the total voyage cost and total freight revenue. Recently, Mclean and Biles (2008) developed a simulation model to evaluate the operational costs and performance of liner shipping by considering the movement of intermodal containers, operation in ports and the contribution of each service route to the operation of the network. The results showed that their model can considerably reduce the operating cost of the network without significant effects on the fleet performance. In summary, existing simulation studies focused on the design of schedules and routes in liner shipping. However, it is unclear how port productivity and service level would impact liner shipping operating cost and schedule reliability. Given the great influence of port operations on liner shipping operations and the growing emphasis on schedule reliability, there is an urgent need to quantify the impact of port productivity and service level on liner operating cost and schedule reliability. 3. Model Description The Simulink Software under Matlab is used to develop the discrete event simulation model. The model simulates the stochastic times that a vessel spent at ports and sea legs. The schedule reliability and operating cost of the entire route service are evaluated under different sailing schemes and port conditions. In this study, the AE9 (Asia-Europe) round-trip currently operated by Maersk Line is chosen. Table 1 shows the sequence of westbound port calls evaluated in the application. The container ship specifications are summarized in Table 2. Table 1 Calling port sequence Calling ports Distance to the next port call (nautical miles) Ningbo ---- Shanghai 101 Yantian 760 Tanjung Pelepas 1460 Suez Canal 5313 Rotterdam 3342 Bremerhaven 212 Felixstowe 300 Zeebrugge 83 4

5 Capacity (TEU) Type Table 2 Container ship parameters for simulation LOA (m) Designed service speed (Knots per hour) Fuel consumption rate at designed speed (Metric tons per hour) 8112 Post-Panamax General assumptions to the liner shipping model Due to the focus of port productivity and service level, the simulation model and its application are based on the following assumptions: 1. The routes and distances between container ports do not change during the simulation. 2. The bunker price does not change during simulation and the fuel cost is only affected by the ship speed and sailing time. For the rest of the group (e.g. docking fees, canal transit, routine maintenance and other operation costs), the cost remains constant. 3. The service speed of the container ship is not affected by the weather, sea, or vessel conditions. 4. The vessel will call at each port according to the predetermined sequence and schedule, and will not change or skip any calling ports. 3.2 Definition of variables In this model, two sets of time variables are identified and defined (as shown in figure 1). The first set of variables is termed real time variables. These variables represent the actual time that a ship has spent going through each of the ship operations, defined as follows: Port to Port Time (PPT) is defined as the duration in hours starting from the time when the ship has just berthed at one terminal to the point where the ship was berthed in the terminal of the next calling port. The PPT was further divided into seven components: Port Time (PT): the total time for unloading, movement and loading of containers. Pilot Out (PO): the time for the ship to sail out of the port. Its length is modeled as following a uniform distribution. Sailing Time (ST): the time spent for the movement of container ship from a departing port to the next port in the sequence. Anchorage Time (AT): the time that the ship needs to wait for berthing. It is modeled as a probabilistic event of which the chance of berth-on-arrival depends on the total capacity and congestion level of the calling port as well as the punctuality of the arriving ship. Pilot In (PI): the time for the ship to sail from anchorage to berth in the port. Its length is modeled as following a uniform distribution. 5

6 Window Time (WT): the time that a ship needs to wait to transit through the Suez Canal when it fails to arrive at the canal within its allocated time slot. In this experiment, it is assumed that the time slot allocated to a container ship to transit through the Suez Canal is the same on each day. This means that if a ship misses the allocated time slot, say 1:00pm to 3:00 pm today, it will have to wait until 1:00pm on the next day to pass through the canal. Cargo Ready Time (CRT): the time that a ship needs to wait until the cargo is ready. This happens when the ship arrives at a port so early that the cargo has not been ready to be loaded onto the ship. In this experiment, it is assumed that for every calling port, the cargo will be ready two days before the scheduled arriving time. On the other hand, before the ship actually starts moving containers between ports, the ship operator must plan on the amount of time needed for the aforementioned processes. These pre-planned time components are defined as planned time variables, including Estimated Port to Port Time (EPPT), Estimated Port Time (EPT), Estimated Pilot-out (EPO), Estimated Pilot-in (EPI), Estimated Sail Time (EST), Estimated Anchorage Time (EAT) and Buffer Time (BT). The Buffer Time is defined to account for the sum of Window Time, Cargo Ready Time and normal delays due to their probabilistic or contingent nature. PT i PO i ST i WT i CRT i AT i PI i Porti Port i+1 EPT i EPO i EST i BT i EAT i EPI i Figure 1 Actual & planned voyage time sequence from port i to port i+1 Since the actual amount of time is uncertain, the planned time variables can only be estimated based on historical data with certain confidence levels. Confidence level is defined as the probability that the actual amount of time spent will be less than the planned amount of time. It follows that the more the amount of time assigned to a particular shipping process, generally the higher the confidence level will be, which also leads to higher overall schedule reliability. 3.3 Model Logic For the i th voyage segment in which the container ship travels from port i to port i+1, the two sets of time variables defined in section 3 satisfy PPT i = PT i + PO i + ST i + AT i + PI i + (CRT i + WT i ) (1) EPPT i = EPT i + EPO i + EST i + EPI i + BT i (2) Note that for CRT and WT in formula (1), only WT i will be considered when the (i+1) th destination is Suez canal. In other cases, only CRT i is considered. Here we define the difference (DIF i ) as DIF i = PPTi - EPPT i (3) 6

7 Consequently, the DIF of a particular voyage segment will have a cascading effect on all the following voyage segments. A negative DIF for the previous voyage segment means the ship now has more time for the next voyage segment. Thus the ship can choose either to sail at a lower speed to save fuel cost or sail at the planned speed to ensure higher schedule reliability. In the second case, the ship will either have to increase speed to catch up with the planned schedule (higher fuel cost) or does not speed up and face a greater risk of delay during subsequent voyage segments. 3.4 Service attribute and sailing modes To analyze how the experiment results may be affected by the service attributes, the ship would be required to sail through the voyage under two different total voyage time, namely 10 weeks and 11weeks. Under each case, the ship was designed to sail through the voyage with either constant speed mode or variable speed mode. Thus, it will be much more convenient for us to quantify and compare the impact of port operations on liner schedule reliability and voyage fuel cost Constant speed mode Under constant speed mode, the ship would sail with a predetermined speed between two container ports regardless of any contingent events (the fuel cost being unchanged). Ergo the impact of port productivity (measured by container handling rate) and service level (measure by percentage of berth-on-arrival) on the schedule reliability can be quantified and analyzed. The schedule reliability is computed as schedule reliability = number of ports upon which the ship arrives on time the total number of calling ports (4) Where the ship is considered on time if it arrives within 24 hours of the scheduled time of arrival Variable speed mode Under the variable speed mode, the ship will automatically adjust its speed based on the estimated sailing time available so as to ensure an optimal level of schedule reliability. Since the schedule reliability is maintained on a relatively constant level, the impact of port operations on liner shipping cost (e.g. fuel cost) can be assessed under different port conditions. The ship will adjust its speed based on the following calculations. Firstly, for the i th segment, the available sailing time (AST) is estimated by AST i = PPPT i PT i EPI i BT i (5) Since the distance (D) between port i and i+1 is known, the resulting sailing speed (Vi) can be determined: v i = D i,i+1 AST i (6) However, the speed of the container ship can only be adjusted within a certain operation range (e.g. 14knots/h ~ 24knots/h). Hence, if the above speed is within this operational range, the ship will sail at the estimated speed. Otherwise the ship will sail at either its lower or upper speed limit. 7

8 3.5 Container ship consumption model The fuel consumption of a ship is estimated according to the cubic rule (Stopford, 2009): F = F S S 3 (7) Where F = actual fuel consumption rate at metrics tons per hour; S = actual speed; F* = designed fuel consumption; S* = designed speed. 4. Results and discussion The simulation results are organized into four sections. The influence of port productivity and service level on schedule reliability was examined in scenarios 1 and 2. In scenarios 3 and 4, the impact of port productivity and service level on total voyage fuel cost was investigated. Due to the stochastic nature of each time component, the simulation model was run 100 times for each of the four scenarios and the average values of schedule reliability and voyage total fuel cost were computed. Scenario 1: the ship adopts the constant speed mode under a 10 week service schedule In this case, the total voyage time is 10 weeks and the ship sails at a constant speed of 17 knots per hour. The total fuel cost remains constant at 7.7 million US dollars since the total distance of the voyage does not change. Figure 2 Average schedule reliability vs first port productivity with BOA = 0.3 for all other ports 8

9 A B Figure 3 Average schedule reliability vs first port productivity with BOA = 0.5 for all other ports As shown in figure 2, when the berth-on-arrive (BOA) rate of the first port is at a low or medium level (0.3 and 0.5 respective), the average schedule reliability increases more rapidly at the two ends as the productivity of the first port increases (1 represents a cargo handling rate of 120 containers per hour and 2 represents a cargo handling rate of 240 containers per hour). However, when the BOA of the first port reaches a very high level (BOA = 0.8), the impact of its productivity on average schedule reliability becomes less significant and uncertain (as the curve fluctuates more against changes in port productivity). When the average BOA of all other ports was increased to 0.5, the average schedule reliability all improved significantly regardless of first port BOA levels (see figure 3). Notice that in some cases, an increase in port productivity actually resulted in a decrease in average schedule reliability. This may be explained as follows. Firstly, due to the uncertainty of port time and other voyage time components, the ship arrives within schedule for some calling ports and generally arrives behind schedule for the others. Due to the increase in port productivity, the ship will tend to arrive earlier at all the subsequent ports. As a result, for those ports where the ship previously arrives late, the schedule reliability tends to be increased because the ship now has a greater chance of arriving earlier (the positive side). On the other hand, for those ports where the ship previously arrives within schedule, there is a risk of decline in schedule reliability since the ship may tend to arrive too far ahead of schedule (the negative side). Nevertheless, it was observed that the positive effect generally outweighs the negative effect for a given voyage. It is also possible that by adjusting port productivity and liner voyage schedule, the negative effects will be reduced to minimum. In addition, it is found that the pattern of change in average schedule reliability is also dependent on the service level of other calling ports (measured by BOA). This means for different voyages the optimal combination of productivity and service level for a particular port may also change. Scenario 2: the ship adopts the constant speed mode under a 9 week service schedule Due to the decrease in the total voyage time, the ship now must sail at a higher speed to compensate for the time reduction. As a result, a speed of 19.5 knots per hour was adopted and the total voyage fuel cost is 10.4 million US dollars. 9

10 When the BOA of the first port equals 0.3 and 0.5, it was observed that the schedule reliability generally increases with port productivity (see figures 4 and 5). However when the BOA of the first port is increased to a high level of 0.8, the schedule reliability decreases as the port productivity increases. The results in both scenario 1 and scenario 2 show that a mere improvement on the port side may not lead to higher schedule reliability without operational adjustment in liner operations. Figure 4 Average schedule reliability vs first port productivity with BOA = 0.3 for all other ports Figure 5 Average schedule reliability vs first port productivity with BOA = 0.5 for all other ports Scenario 3: the ship adopts the variable speed mode under a 10 week service schedule In the following two scenarios, the ship will constantly adjust its actual speed to the planned schedule. The resulting schedule reliability at each port service level was between and As clearly shown by the solid curves in figure 6, at each BOA level for the first port, the average total voyage fuel cost would decrease as the first port productivity increases. In addition, by comparing each curve with a straight line connecting their respective endpoints (the dotted lines), it was observed that the rate of decrease in voyage fuel cost gradually diminishes as the productivity of the first port increases to higher levels. 10

11 Figure 6 Average voyage fuel cost vs first port productivity with BOA = 0.3 for all other ports In addition, when the schedule reliability and fuel cost in scenario 3 were compared to those in scenario 1 at each level of port productivity and service level, an average increase of 83% in schedule reliability was observed. On the other hand, the average increase in total fuel cost was only about 10% to 12%. Scenario 4: the ship adopts the variable speed mode under a 9 week service schedule Given a shorter voyage time, the resulting schedule reliability not only decreased at each port service level (between and ) when compared to scenario 3, but also became much more sensitive to both changes in port productivity and service levels. In addition, it was found that the voyage fuel cost at each port service level and port productivity also increased significantly when compared to the results in previous scenario (see figures 6 &7). Figure 7 Average voyage fuel cost vs first port productivity with BOA = 0.3 for all other ports 6. Conclusion A quantitative analysis on liner shipping operating cost and schedule reliability in view of port productivity and service level is conducted with the simulation approach. The discrete event simulation model simulates the stochastic times that a vessel spent at ports and sea legs. Port productivity and service level are measured by container handling rate and the percentage of berth-on-arrival (BOA) for arriving vessels, respectively. 11

12 To the best of our knowledge, this is the first simulation study that quantifies the impact of port productivity and service level on liner shipping operating cost and schedule reliability. The study is beneficial to both port/terminal operators and shipping lines in an era when there is growing competition in cost and schedule reliability in the liner shipping industry. The key findings are summarized as follows. Firstly, schedule reliability can be enhanced by improving either port productivity or service level, and their effects are interrelated. This implies that the improvement of port productivity and service level must be synergized to arrive at the most cost-effective solution. In addition, the impact of a particular port s productivity on liner shipping schedule reliability is affected by other ports service levels. This implies that port/terminal operators should consider other ports conditions for their port development planning. On the other hand, as far as liner schedule performance is concerned, the potential benefits of an improvement on the port side will not be fully exploited without a corresponding adjustment in liner operations. In other words, close cooperation must be present for both parties to achieve better operational performance. Secondly, when the ship sails at a variable speed mode, an increase in either port productivity or port service level would lead to a decrease in fuel cost. However, the rate of decrease in total voyage cost gradually diminishes with the increase in port productivity (at a constant port service level). Lastly, at each level of port productivity and service level, a shorter total voyage time will result in higher fuel cost and lower schedule reliability. The schedule reliability will also become more sensitive to port productivity and service level with a shorter total voyage time. This study may be extended by future studies in two dimensions. Firstly, the analysis may be expanded to an entire liner shipping network. Secondly, the cost benefit analysis may be extended to include indirect benefits received by liner companies in view of the entire logistics chain including customers and service partners affected by the chain. Acknowledgements We wish to acknowledge the funding support for this project from Nanyang Technological University under the Undergraduate Research Experience on CAmpus (URECA) programme. We also thank the anonymous reviewers for their valuable comments. References Alexis, G. A. (1981). A survey of routing and scheduling models in ocean transportation. Master Unpublished Mater's Thesis, M.I.T. Bendall, H. B., & Stent, A. F. (1999). Longhaul Feeder Services in an Era of Changing Technology: An Asia-Pacific Perspective. Maritime Policy & Management, 26(2), Boffey, T. B., Edmond, E. D., Hinxman, A. I., & Pursglove, C. J. (1979). Two Approaches to Scheduling Container Ships with An Application to the North Atlantic Route. Journal of the Operational Research Society, 30(5), Carranza, A. A. M. (2008). Discrete Event Simulation Approach for the Analysis of Liner Shipping Services of Containerized Cargo. [Doctoral Dissertation]. Christiansen, M., Fagerholt, K., & Ronen, D. (2004). Ship Routing and Scheduling: Status and Perspectives. 12

13 Transportation Science, 38(1), Datz, I., Fixman, C. M., Freidberg, A., & Lewinson, V. (1964). A DESCRIPTION OF THE MARITIME ADMINISTRATION MATHEMATICAL SIMULATION OF SHIP OPERATIONS. Paper presented at annual meeting of the Society of Naval Architects and Marine Engineers. 72, Gkanatsas, E. (2004). Designing Robust Shipping Schedules. Msc Maritime Economics & Logistics Master Thesis, Erasmus University Rotterdam Lam, J.S.L. and Dai, J. (2012). A decision support system for port selection, Transportation Planning and Technology, 35(4), Lam, J.S.L. & Van de Voorde, E. (2011). Scenario analysis for supply chain integration in container shipping, Maritime Policy & Management, 38(7), Mclean, A. A., & Biles, W. E. (2008). A Simulation Approach to the Evaluation of Operational Costs and Performance in Liner Shipping Operations. Paper presented at the Simulation Conference, WSC Winter. Perakis, A. N., & Jaramillo, D. I. (1991). Fleet deployment optimization for liner shipping Part 1. Background, problem formulation and solution approaches. Maritime Policy & Management, 18(3), Qi, X., & Song, D.-P. (2012). Minimizing fuel emissions by optimizing vessel schedules in liner shipping with uncertain port times. Transportation Research Part E: Logistics and Transportation Review, 48(4), Ting, S.C., & Tzeng, G.H. (2003). Ship Scheduling and Cost Analysis for Route Planning in Liner Shipping. Maritime Economics & Logistics (formerly International Journal of Maritime Economics), 5(4), Stopford, M. (2009). Maritime Economics, 3rd ed. London ; New York : Routledge. Notteboom, T. (2006). The Time Factor in Liner Shipping Services. Maritime Economics & Logistics, 8(1), Van Rensburg, J. J., He, J., & Kleywegt, A. J. (2005). A computer Simulation Model of Container Movement by Sea. Paper presented at the Proceedings of the 2005 Winter Simulation Conference. Wang, S., & Meng, Q. (2012a). Liner ship route schedule design with sea contingency time and port time uncertainty. Transportation Research Part B: Methodological, 46(5), Wang, S., & Meng, Q. (2012b). Robust schedule design for liner shipping services. Transportation Research Part E: Logistics and Transportation Review, 48(6),