Controlled Random Stacking at an RTG terminal
Improving the APM Algeciras Container Terminal performance by implementing the Controlled Random Stacking strategy
Delft, October 10, 2009
Master Thesis Operations Research & Management
Stephan Vermeulen stephan.vermeulen@tba.nl stephan.vermeulen@student.uva.nl student nr: 0418498 06-42753595 Supervisors Ir. J.A.M. Hontelez Dr. C. Duin Ir. R. Chan Dr. C. Boer
Abstract
Yard planning is the way the stacking of containers is organized at a container terminal. Re-cently a new yard planning strategy is introduced: Controlled Random Stacking (CRS). By spreading the piles of similar containers over the yard, this strategy provides better use of yard space and a better workload distribution for the involved stacking equipment, which improves the quay crane productivity. This strategy is tested by emulation for the container terminal Algeciras, Spain, a terminal that uses rubber tired gantries for stacking containers. Therefore, CRS is implemented in its terminal operating system in three different forms. CRS gives better results than the common strategy of filling whole bays with similar containers in quiet times, but similar results in busier times. The following processes have to be improved to achieve the benefits of CRS: automation, RTG job selection and stowage planning.
M
A N A G E M E N TS
U M M A R Y1. Introduction
Overseas container transport is the main mode to transport goods. The seaborne trade volume grows faster than the world economy already for decades. Containers are mainly shipped by container vessels between container terminals all over the world. At a container terminal, containers are transhipped between vessels and other transport modes as trains and trucks. This fast growing and the large volumes of containers handled, forces container terminals to improve the efficiency continuously.
TBA is a company that is specialized in optimizing container terminal operations and logistics. TBA developed CONTROLS, an extensive software application by which container terminals processes can be emulated.
The goal of this research is to improve the performances of an RTG terminal by implementing a new yard planning strategy derived from theory: Controlled Random Stacking (CRS). The Algeciras Terminal in Spain is the case: CRS is implemented in the Terminal Operating System (TOS) of this terminal and tested by emulation with CONTROLS, that is connected to this TOS. 2. The Algeciras terminal
The Algeciras terminal is located in the south of Spain, near Gibraltar. This is a very strategic location; many shipping lines pass the Strait of Gibraltar. Because of this and the thinly popu-lated hinterland, the terminal handles mainly transshipment containers. These containers arrive and depart by vessel.
The Algeciras terminal is an RTG terminal. The following equipment is used at the terminal: • Quay Cranes (QC’s) Cranes that load and discharge containers onto vessels. • Terminal Trucks (TT’s) Trucks for internal transport between QC’s and yard. • Rubber Tired Gantries (RTG’s) Mobile yard cranes that stack and pick up containers. 3. Current yard planning strategy
Containers stored at a terminal to be picked up by vessel, train or truck, are stacked in the yard. Yard planning is the process of determining the locations of the containers in the yard. The containers are stacked in blocks according to different rules, in order to be handled efficiently by the equipment used at the terminal.
The current, manually performed yard planning strategy at the Algeciras terminal is gathering containers with similar properties on all piles in the same bay, see Figure 1.
Bay Pile
Figure 1: Container stacking blocks with marked bay and pile slots
Containers, that are stacked together, are containers that have to be loaded simultaneously. Therefore the properties, by which containers are grouped for stacking, are the following:
• Port of discharge Next terminal where container is discharged • Depart vessel Vessel by which container leaves
• Service Standard shipping route along world ports • Stow code Code for placement location in vessel • Weight class Class of weight
• Equipment type Container type (20’ or 40’ foot, cooled or not)
This currently at the Algeciras Terminal used yard planning strategy has two main problems: • Many reserved ground slots
In the current strategy complete bays are reserved for containers of the same group. When the first container is placed in a bay, containers with other properties are not al-lowed to be stacked in this bay due to the reserved slots.
• Inefficient RTG workload distribution
The containers are stacked by RTG’s. Because of the stacking per complete bay, an RTG’s handling this bay has a great workload, while many other RTG’s are idle. The working RTG can be so busy that TT’s have to wait and cannot serve the QC continu-ously. This way the QC productivity is limited.
4. Controlled Random Stacking strategy
Analysing the current problems and yard planning strategies in literature, Controlled Random Stacking (CRS) is chosen to be implemented as yard planning strategy in the Terminal Operat-ing System (TOS). CRS can not be applied manually, so it has to be performed by the
Al-CRS is a yard planning strategy that spreads piles of containers over (a part of) the yard. So containers are not gathered by complete bays. This means that every single free ground slot can be used to stack a container and that containers have to be handled at more locations in the yard simultaneously, so more RTG’s can do the same amount of work. This results in less or even no reserved ground slots and better distribution of the RTG workload.
5. Experiment settings
Four main settings are implemented in SPARCS and tested by CONTROLS, see Table 1.
Experiment SPARCS included # Matching piles per bay Matching containers in bay Matching containers in pile
Current strategy Yes 6 ++ ++
CRS 1 Yes 1 --- ++
CRS 2 Yes ≤2 -- ++
CRS 3 Yes ≤3 - ++
Table 1: Experiment settings with their properties
CS is the current strategy with only similar containers in a bay. CRS 1 prefers only one pile of matching containers in a bay, CRS 2 allows two piles and CRS 3 even three. For CRS 2 and 3, these two or three piles in a bay share their service, not per definition the other properties. 6. Results
Two working shifts are emulated; one quiet shift with two vessels berthed and a busy shift with five vessels berthed. The main performance indicators have the following results, see Tables 2 and 3. The RTG workload distribution is indicated by the standard deviation of the productivity in moves per hour per RTG. The QC productivity in moves per hour is the most important performance indicator; the faster QC’s work, the shorted a vessel is berthed. A rehandle is a shuffle of a container by an RTG in order to pick up a container that is not on top of the pile.
Setting % Unreserved ground slots Standard deviation RTG productivity QC productivity % Rehandles of all RTG moves Current strategy 7.24% 3.7 23.6 33,53% CRS 1 7.14% 3.0 30.1 16,43% CRS 2 12.61% 2.7 27.6 21,22% CRS 3 12.06% 2.6 28.3 29,37%
Table 2: Experiment results for the quiet shift with productivity in moves per hour
Setting % Unreserved ground slots Standard deviation RTG productivity QC productivity % Rehandles of all RTG moves Current strategy 8.83% 4.5 24.6 19,57% CRS 1 8.39% 3.4 24.7 11,19% CRS 2 11.09% 3.9 24.3 7,99% CRS 3 10.48% 4.0 24.6 12,11%
7. Conclusions
For both shifts, the percentage of unreserved ground slots is not improved by CRS 1, but it is by CRS 2 and 3. With CRS 2 and 3, only half a bay of one third of a bay can be reserved, which leaves the rest of the bay unreserved. The unexpected low percentage of unreserved ground slots with CRS 1 is caused by the high and preference for each ground slot: containers are allowed to be stacked on every ground slots very easy, so ground slots are filled quickly. With CRS 2 and 3, there still is some preference for certain bays, which keeps relatively more ground slots unoccupied.
The equipment performances are all improved by CRS for the quiet shift. For the busy shift, the RTG workload distribution (lower standard deviation) is improved and the percentage of rehandles is decreased. The average QC productivity is not improved by the CRS settings, but is comparable to the result of the current strategy in busy times.
The QC productivity is not improved by the better RTG workload distribution, because of the insufficient RTG job selection. With CRS, containers have to be handled at more locations than with the current strategy. This forces RTG’s to move more often and work shorter time on one location. The RTG job selection could be adjusted with more frequent reassignment of RTG’s to jobs and less block changes by RTG’s. Because of the many and often changing locations where containers have to be handled with CRS, RTG’s could better be assigned to yard area’s mainly instead just to jobs.
The percentage of rehandles is highest for the current setting and decreases when the number of matching piles per bay decreases for the CRS settings. A greater number of containers that have to be handled in one bay gives more rehandles. The stowage planning lists all containers that have to be loaded onto a vessel. This planning is not optimal for the Algeciras Terminal: these containers are not picked very efficient. Also, the load sequence has to be refreshed in order to avoid many more rehandles.
8. Recommendations
The Algeciras terminal is not recommended to implement CRS at the moment. The terminal is not automated at the moment, while an well-functioning automation is crucial to apply CRS. The following conditions have to be satisfied and tested in relation with CRS for a well-supported implementation for an RTG terminal.
• A well functioning automated yard planning (with the current strategy). • Adjusted RTG job selection.
TABLE OF CONTENTS MANAGEMENT SUMMARY ... I 1 INTRODUCTION ... 3 1.1 CONTAINER TRANSPORT ... 3 1.2 PROBLEM BACKGROUND ... 4 1.3 RESEARCH QUESTIONS ... 5 1.4 OVERVIEW ... 5
2 CONTAINER TERMINAL IN GENERAL ... 7
2.1 INTRODUCTION ... 7
2.2 PROCESSES ... 7
2.3 LAYOUT ... 8
2.4 CONTAINERS ... 9
2.5 EQUIPMENT ... 11
2.6 TYPES OF CONTAINER TERMINALS... 15
3 OVERVIEW YARD PLANNING ... 18
3.1 INTRODUCTION ... 18
3.2 LITERATURE REVIEW YARD PLANNING STRATEGIES ... 19
3.3 YARD PLANNING STRATEGIES SUPPORTED IN TOS’S ... 27
3.4 SUMMARY &CONCLUSION ... 32
4 YARD PLANNING IN SPARCS ... 35
4.1 YARD LAYOUT AND ALLOCATION ... 35
4.2 SECTION AND STACKING FACTORS ... 37
4.3 EXPERT DECKING ... 39
4.4 SUMMARY ... 41
4.5 ALGECIRAS TERMINAL ADJUSTMENTS ... 41
5 ALGECIRAS CONTAINER TERMINAL ... 43
5.1 GENERAL DESCRIPTION ... 43
5.2 EQUIPMENT ... 44
5.3 LAYOUT ... 44
5.4 CURRENT YARD PLANNING STRATEGY ... 45
5.5 CURRENT PROBLEMS LEADING TO CRS... 47
6 EMULATION ... 51
6.1 EMULATION VS.SIMULATION ... 51
6.2 CURRENT USE OF SPARCS ... 51
6.3 START SITUATION BY SCENEMA ... 52
6.4 EMULATION WITH CONTROLS ... 52
6.5 KEY PERFORMANCE INDICATORS ... 54
6.6 OVERVIEW ... 54
7 EXPERIMENTS ... 55
7.1 HISTORICAL REPLAY ... 55
7.2 CS:GROUPING ON POD AND SERVICE PER BAY ... 56
7.3 CRS1: GROUPING ON POD AND SERVICE PER PILE ... 59
7.4 CRS2: GROUPING ON SERVICE PER TWO LANES AND POD PER PILE... 61
7.5 CRS3: BY GROUPING ON SERVICE PER THREE LANES AND POD PER PILE ... 63
7.6 ADDITIONAL SETTINGS ... 63
8 RESULTS & ANALYSIS ... 66
8.1 VALIDATION OF THE RESULTS ... 66
8.2 YARD DENSITY ... 67
8.3 RTG WORKLOAD DISTRIBUTION ... 70
8.4 QC PRODUCTIVITY ... 74
8.5 NUMBER OF REHANDLES ... 77
8.6 DRIVEN DISTANCE TT’S ... 78
9 CONCLUSIONS, RECOMMENDATIONS & FURTHER RESEARCH ... 82
9.1 CONCLUSIONS ... 82 9.2 RECOMMENDATIONS ... 83 9.3 FURTHER RESEARCH ... 84 LIST OF ABBREVIATIONS ... 86 REFERENCES ... 88 APPENDICES
A CONTAINER PROPERTIES ... A.1 B EXPERT DECKING PENALTIES &BONUSES ... B.1 C ADDITIONAL EXPERT DECKING VARIABLES ... C.1 D SECTION &STACKING FACTOR ... D.1 E CONTROLSVALIDATION ... E.1 F RTGJOB SELECTION BY CONTROLS ... F.1 G EXPERT DECKING VALUES ... G.1 H RESULTS YARD DENSITY ... H.1
1
I
N T R O D U C T I O N1.1 Container transport
Overseas container transport is a very common way of transporting goods around the world. Big advantage is the large volume that can be handled. Since the standardization of the container dimensions in 1968, the volume of seaborne trade has grown rapidly (Figure 1.1). The last years, it still grows faster than the world economy and the industrial production, see Figure 1.2.
Figure 1.1: Containerisation: volumes handled in 10 major container terminals in TEU (Twenty-foot Equivalent Unit: one container)
Source: Internal TBA report (2005)
Figure 1.2: Indices for world economic growth (Gross Domestic Product), industrial production (Organization for Economic Cooperation and Development) and world seaborne trade (volume), 1994=100.
As a result of containerisation, the container ships and the terminals around the world have increased in size and throughput as well. Container transport has proved itself as very efficient. These developments have led to a lot of research regarding container terminals. Many articles are written on container terminals and their operations, and container terminals and its operators continuously try to improve these operations. Because of the growth of container transport and the great volume of containers, it is very important to handle the involved vessels and terminals efficiently. One of the terminals attempting to improve their performance is the APM Algeciras Container Terminal in Spain.
1.2 Problem background
TBA BV is a company that designs and optimizes logistic processes. Container terminals are their main specialization. By simulating the processes at a container terminal, these processes can be optimized. The term 'optimization' is used, real optimization however is never achieved. Despite the fact that 'improvement' would be a more suiTable term, 'optimization' is common used at container terminal en operators.
During the last years TBA designed and developed a product called CONTROLS (CONtainer Terminal Real-time Optimization of the Logistics System) for simulation of container terminals. CONTROLS provides a virtual environment to test and tune a terminal operating system of a terminal. A terminal operating system (TOS) is a software application supporting the planning, scheduling and equipment control activities of a container terminal and by this being responsi-ble for accurate operations within the terminal.
Recently TBA created a complete simulation model of the APM Algeciras Container Terminal in CONTROLS. This terminal is using the application SPARCS as a TOS. In order to optimize the planning, this TOS system provides possibilities for different parameter settings. By adjust-ing the parameters, the user can optimize (improve) plannadjust-ing and scheduladjust-ing activities of the TOS. To set the parameters as good as possible, the CONTROLS model of the terminal in Algeciras will be used. This process of testing is called emulation (Auinger, Vorderwinkles and Buchtela, 1999).
The part of the planning that will be focused on in this research project is the placements of containers in the yard. Yard planning is the term used for determining placement of containers. Earlier studies done by TBA have shown that there is a potential improvement in both produc-tivity and working hours by new automated yard planning strategies instead of the manually placement that is used currently at the RTG (Rubber Tired Gantry) terminal of Algeciras. An RTG is the kind of equipment by which containers can be stacked at a terminal.
1.3 Research questions The objective of this research is:
Improve the performances at an RTG container terminal by adjusting the yard planning strategy.
The objective has led to the following research questions:
1. What different yard planning strategies for different types of container terminals are discussed in literature?
2. How are yard planning strategies supported in terminal operating systems (TOS’s)? 3. How is yard planning exactly performed in SPARCS?
4. What is the actual yard planning strategy used at the APM Algeciras Container Terminal and what improvements are desired?
5. Which improved yard planning strategy could be used and how can this strategy be implemented in SPARCS?
6. What is the optimal tuning of the SPARCS yard planning parameters for this strategy?
The first three theoretical questions do not apply to the Algeciras Terminal specifically and are answered in general. The following three questions concentrate on the case itself: improving the performances the Algeciras terminal. In 1.4, an overview of this thesis on basis of the research questions is given.
1.4 Overview
First, a general overview of the subject has to be obtained. So the container terminal with its processes is described in chapter 2.
This research is focussed on the yard planning. Many articles on this subject are published in literature. Different strategies to organize yard planning are investigated in chapter 3. The capabilities of main terminal operating systems to implement yard planning strategies are also discussed in chapter 3. So research questions 1 and 2 are answered in this chapter.
In chapter 4, research question 3 is discussed. The focus is on the way SPARCS performs the yard planning. All steps, parameters and calculations of this process are described.
The actual yard planning strategy of the Algeciras terminal is described with its problems and desired improvements in chapter 5 in order to answer question 4. This leads to the answer of research question 5, a choice for an improved yard planning strategy: controlled random stack-ing (CRS).
This strategy is implemented in SPARCS and tuned to improve the performance indicators as much as possible. The tunings are tested by emulation with the simulation model of the
Al-geciras terminal in CONTROLS. Chapter 6 describes this emulation process and chapter 7 gives the settings in SPARCS in order to test the (tuning of) the controlled random stacking strategy. In chapter 8, the results of the emulations are presented and analyzed. This leads to an answer to question 6. In chapter 9, the conclusions and recommendations are given.
Figure 1.3 shows a schematic overview of the research by subject, pointing to the different chapters (Ch) and research questions (Q).
2
C
O N TA I N E R TE R M I N A L I NG
E N E R A L2.1 Introduction
A container terminal is a facility where containers are transhipped between overseas container ships (vessels) and inland transport modes like trucks, trains and barges. The content of the containers doesn’t change at a container terminal. Because it is impossible to load all containers directly from the ship on a vehicle or the other way around, the containers have to be stored at the terminal to wait for further transportation. So in a container terminal adequate planning and scheduling of container movements is necessary. Also, the cranes and vehicles performing these container movements have to be scheduled.
Steenken et al. (2004), Vis and de Koster (2003), Stalhbock and Voβ (2004) and Meersmans and Dekker (2001) give a (literature) review of operations and decisions of a container terminal. They all state that the increasing importance of container terminals makes the contribution of operations research more and more valuable. Murty et al. (2000) give an overview of all deci-sions that have to be made at a container terminal at the operational level. These authors, along with Geevers (2008) and internal TBA data and knowledge provide the information for this overview.
In the next three paragraphs different aspects of containers terminals are discussed; first the processes, next the layout followed by the containers and finally the equipment. Different types of container terminals are categorized in 2.6.
2.2 Processes
In this paragraph, the main physical processes that happen at a container terminal are discussed: vessel arrival, loading and discharge, transportation and stacking.
2.2.1 Vessel arrival
When the arrival of a vessel is known, a planning has to be made. This planning has three parts. 1. Berth planning
With a terminal operation system (TOS) the planning for all operations at the container terminal can be made. In the berth planning, a part of the berth (quayside) is allocated to the arriving vessel, due to the properties and current state of the vessel and the berth (loca-tion of the cranes). With a terminal opera(loca-tion system (TOS) the planning for all opera(loca-tions at the container terminal can be made. In the berth planning, a part of the berth (quayside) is allocated to the arriving vessel, due to the properties and current state of the vessel and the berth (location of the cranes).
2. Stowage planning
The stowage planning consist the placement of the containers in the vessel. This can be done when it is known which containers have to be loaded and discharged (unloaded). With respect to the properties of the containers and the constraints to balance the vessel, the op-timal placement of the containers is determined.
3. Crane split.
The crane split is the planning of the loading and discharge of containers by quay cranes (QC’s).
When these plans are made and the vessel has berthed, the planned operations are performed.
2.2.2 Loading and discharge
The loading and discharge of containers is done by quay cranes. This process is planned in the crane split planning. The loading and discharge work is distributed a certain number of quay cranes, depending on the availability, the amount of work and the size of the vessel.
2.2.3 Transportation
During loading and discharging, the containers have to be transported from the vessel to the yard where the containers are stored and vice versa. From the yard is also transportation to the other modes of transport like trucks and trains and vice versa. This transportation planning by vehicles is also done with the TOS.
2.2.4 Stacking
Stacking is the process of storing the containers on the yard. Containers can be stored on a chassis or they can be stacked on the ground. When containers are stored on a chassis they can not be piled up, so this system requires a lot of space. At most container terminals, there is not enough space for this system, so the containers are piled up (stacking). It is tried to do the stacking in such manner that it optimizes different performances, like efficient use of space and equipment, turn around time of a vessel or the number of rehandles. The way the stacking of containers is organized is called yard planning, which is the main subject of the literature study. In a TOS, the placement of containers in the yard can be determined.
2.3 Layout
Different areas can be distinguished at a container terminal. First there is the berth (quayside) where the vessels (ships) arrive and the loading and discharge of containers takes place. Also there is the landside area where the containers are loaded on and unloaded from trains and trucks. Because it is impossible to transfer all containers directly from one mode of transporta-tion to another, they have to be stored in the yard. It is here in the yard where the containers are
stacked. In Figure 2.1, the different operating areas and container flows are given. A container movement from one location to another is called a job.
Containers can have different routes of transportation (categories) through a container terminal. The most important ones are:
• Export Arrival by vessel and departure by truck, train or barge. • Import Arrival by truck, train or barge and departure by vessel. • Transshipment Arrival and departure by vessel.
In a yard, the areas for import and export/transshipment are separated in most container termi-nals. In most container terminals, the areas for export/transshipment containers are close to the berth (vessel) and the areas for import are further away from the berth, close to the truck gate and trains. Handling smaller container ships (barges) for IWT (Inland Waterway Transport) is an inland operation, but is performed at the berth/quayside. The handling of external trucks doesn’t need to be done in a special area; it can also be done in the yard.
Figure 2.1: Different areas and container flows in a container terminal
2.4 Containers
The dimensions of containers are standardized. They can be 20, 40 or 45 feet long and they all have the same width of 8 feet, so they can be stacked easily. The 45 foot containers are pretty rare and mostly are stored in a special area. The containers can be 8’6’’ (8 foot and 6 inches: 2.6 m) and 9’6’’ (2.9 m) high, but this difference is too small to affect the stacking possibilities in most cases. Containers can be distinguished by many properties; these properties are described in Appendix A.
Containers can be categorized or grouped by these properties and so areas in the yard can be assigned to containers with certain properties. There are special areas for reefers (cooled containers that need power supplies), empty containers, etc.
The volume of one 20 foot container is expressed as one TEU (Twenty-foot Equivalent Unit). So the capacity and occupation of a yard, vessel or another transportation mode can be ex-pressed in TEU, just as the throughput of a container terminal per time period. The exact location of a container in a yard can be expressed in the following terms, which are illustrated in Figure 2.2 and Figure 2.3:
• Stack lane A series of stack blocks that is adjacent in one line. • Stack column A series of stack blocks that is adjacent side-to-side.
• Stack block A series of stack blocks that is in one stack lane and one stack column. • TGS One TEU Ground Slot.
Figure 2.2: Yard overview with stack lane, column and block.
The following definitions are used to identify series of slots with in a stack block:
• Pile A series of containers that is located on top of each other, occupying one TGS. • Bay A series of containers that is adjacent side-to-side and on top of each other. • Row A series of containers that is adjacent in line and on top of each other. • Tier The layer in which a container can be placed.
The exact location of a container in a stack block is expressed in a five digit number: the first two numbers denote the bay, the second two the row and the last one the tier.
Bay Row Pile Tier
Figure 2.3: Different container series.
2.5 Equipment
At a container terminal, different kind of equipment is used to handle the containers and per-form all needed operations. This equipment can be divided into terminal equipment, which operates only at the terminal, and equipment for transportation to or from the container terminal, such as vessels, barges, trucks and trains. These transportation modes for importing and export-ing containers are clear and won’t be discussed in detail, except for the vessels. The internal terminal equipment will be discussed extensively.
A TOS sends operation orders for jobs to the internal equipment and when a job is fulfilled the equipment sends a message back. So there is continuing communication and knowledge of the status of all containers and equipment. The assignment of internal equipment to jobs is called job selection.
2.5.1 Vessel
A vessel is a container ship, see Figure 2.4. They can carry up to 15.000 TEU. An important property of a vessel is that the containers are stored on deck and below deck in its hull. Between these two storage areas are hatch covers to cover the below deck containers. When containers below deck are loaded or discharged, the hatch covers first are removed by the quay cranes. They are placed back when the container handling below deck is finished.
Figure 2.4: Vessel
2.5.2 Quay crane
Quay cranes (QC’s, see Figure 2.5) are the cranes located on rails at the quayside to load and discharge the vessels. They are able to move containers from transport trucks to the vessel and vice versa. They move along the quayside via the rails, so they can’t pass each other. Quay cranes can move one 40/45 foot container or two 20 foot containers (twinlift) at a time. They can also remove and place the hatch covers. When a hatch cover has been removed, it is placed on the quay.
Figure 2.5: Quay crane
2.5.3 Transport equipment
There are two main types of transport equipment:
• Terminal trucks (TT) are manually driven and can transport one 40/45 foot container or two 20 foot containers at a time, (Figure 2.6).
TT’s and AGV’s can not lift containers themselves, so the container always has to be placed and removed by other equipment. They mainly transport containers between the yard and the operating area’s (quay side, trains). TT’s and AGV’s are internal trucks (IT), external trucks can also operate in the yard to pick up or deliver a container.
Figure 2.6: Terminal truck (TT) Figure 2.7: Automated Guided Vehicle (AGV)
2.5.4 Stacking equipment
Stacking equipment consist of devices that handle the containers. They take containers of a transport vehicle and stack them and vice versa. They can also reshuffle the containers in the stacks. The different devices:
• Rail mounted gantry (RMG, Figure 2.9) • Rubber tired gantry (RTG, Figure 2.8) • Overhead bridge crane (OBC)
• Automated stacking crane (ASC, Figure 2.9)
RMG’s and OBC’s can’t switch between stack lanes, while RTG’s can by turning their wheels 90º. On the other side, RTG’s move slower because of the lack of rails. RMG’s and OBC’s can theoretically be regarded as equal equipment. OBC’s are used very seldom. Automated RMG’s are called automated stacking cranes (ASC’s). The stacking height of this equipment can be up to four, five or six containers.
Figure 2.8: Rubber tired gantry (RTG) Figure 2.9: Rail mounted gantry (RTG) / auto-mated stacking crane (ASC)
2.5.5 Transport and stacking equipment
Transport and stacking equipment is able to handle the containers independently. They can lift, stack and transport containers by themselves. But the transport is less efficient than trucks and the stacking is less efficient than specific stacking equipment. Different types are the following:
• Straddle carrier (SC, Figure 2.10) • Reach stacker (RS, Figure 2.11) • MT-handler (Figure 2.12)
SC’s are the only transport and stacking equipment that is used as the main equipment at a terminal. A SC can only stack up to two or three containers height. Reach stackers handle odd containers or are used in deviating yard locations where stacking equipment (see 2.5.4) can’t be used. An MT-handler is a kind of forklift truck that only handles empty containers.
2.6 Types of container terminals
All container terminals can be categorized in on-chassis terminals and stacking terminals.
2.6.1 On-chassis terminals
In an on-chassis (wheeled) terminal, containers are placed on a chassis and pulled to the yard location by TT’s. The containers remain on the chassis when stored, so it is not possible to stack the containers. This means that the storing yard has to be relative large, because the space is not used efficiently. Advantage is that no stacking is required, which saves equipment costs and operating time. On-chassis terminal are mainly seen in North America, because of the sufficient availability of low cost land. A couple of on-chassis container terminals are in Long Beach, Vancouver and Boston.
2.6.2 Stacking terminals
In stacking terminals, the containers can be stacked (piled up) in the yard, so stacking equip-ment is required at these terminals. According to the used equipequip-ment, stacking container terminals can be categorized in three main types of terminals.
1. SC terminals
In these terminals, a lot of work is done by straddle carriers, see 2.5.5. They take care of the stacking and transport of the containers. Advantage is that less transshipment is needed between different types of equipment. Disadvantage is that SC’s don’t drive very fast, compared to TT’s and AGV’s. Another disadvantage is the quite inefficient usage of the yard; when SC’s do the stacking, open spaces are required between each stack lane for the legs of the SC’s, see Figure 2.13. This reduces the capacity of the yard, just as the lower stacking height of SC’s. Last years, SC terminals don’t satisfy anymore in some cases, due to the growing volume in container trade. The SC terminals of Antwerpen and North Port, Kuala Lumpur, prepare to switch to automated, RMG and RTG terminals nowadays.
2. RTG terminals
The stacking of the containers at these terminals is done by RTG’s (2.5.4) while the trans-portation is done by TT’s, see 2.5.3. With RTG’s, the stacking height can be higher and the usage of the yard ground can be more efficient compared to SC and on-chassis terminals. In an RTG terminal, stacking blocks mostly are located parallel to the berth and have a driving lane. RTG’s work over a stacking block that can be four to six containers wide and includes a driving lane, see Figure 2.14. In this driving lane, TT’s and external trucks can operate by picking up and delivering containers to the RTG’s. This way a container can be delivered or picked up exactly at the yard location where it is handled by the involved RTG. Transporting a container with an RTG is not very efficient, because it drives slowly, so this is avoided. RTG terminals are widely used, for example at the harbours in Sydney, Hong Kong and Algeciras. Disadvantage of these terminals is that a lot of manpower is re-quired compared to other terminals.
Figure 2.14: Stacking block of an RTG terminal with lanes for TT’s and external trucks
3. Automated/RMG terminals
A complete automated terminal uses ASC’s and AGV’s, see 2.5.3 and 2.5.4. A terminal with RMG’s (2.5.4) or OBC’s and AGV’s or TT’s is very similar, so these terminals can be regarded as the being equal. The processes at these terminals are comparable to the proc-esses at RTG terminals. Difference is that at most automated terminals, AGV’s and exter-nal trucks don’t use a driving lane parallel to the blocks but pick up and deliver the con-tainers at the beginning and end of a stacking block (Figure 2.15). These blocks are located right-angled to the berth, so one end is at the quayside operating area and one end is at the landside operating area. The reason for this is that RMG and ASC’s can drive fast with containers, especially compared to RTG’s.
Figure 2.15: Stacking block of an automated terminal with AGV area
Automated terminals are very efficient in usage of yard space and stacking height; because of the rails, ASC’s and RMG’s can be even larger than RTG’s, and no driving lane is re-quired. Another advantage is the less requirement of manpower, but the disadvantage is the big investment that is needed.
RMG terminals are used all over the world, where complete automated terminals are much less in number, but this number is growing. Because of the high labour costs in Western Europe, automated terminals are mostly seen in harbours in this area, like Rotterdam, Hamburg and London.
3
O
V E R V I E W YA R DP
L A N N I N G3.1 Introduction
Most of the literature handles decisions for organizing and controlling the operations inside a container terminal. The term ‘optimization’ is often used, while real optimization almost never is achieved. The term ‘improvement’ would be better, but because of the fact that the term ‘optimization’ is commonly used in the literature and on the work floor, it will be used in this literature review as well. In literature, authors mainly focus on one optimization decision, like the berth planning or vehicle planning. In some cases, it is tried to consider the container terminal as a whole for optimization, but this appears to be very complex and is often split into more than one research questions.
TBA is a consulting company that designs and optimizes logistic processes. Their main spe-cialization is container terminals. By simulating the processes at a container terminal, these processes can be improved. One of the main processes is container stacking: placing containers in the yard. Yard planning is the way container stacking is organized. This is done by the equipment discussed in 2.5. Different decisions have to be made when placing a container. These decisions can be split into three levels, see Vis and de Koster (2003):
1. Strategic level (design/layout of the container terminal). 2. Tactical level (allocation of areas to groups of containers). 3. Operational level (exact storage of a container)
The results of the literature study on yard planning strategies are presented in 3.2. Most of the literature handles decisions on the last two levels, and the layout of a terminal is given in most cases. The storage of transshipment containers can be done the same way as is done for export containers, see Kim et al. (2000). Besides published articles, internal information of TBA is a main source as well.
In 3.3, three main terminal operating systems are investigated. This is done by using the sys-tems, study the manuals and discus with their users inside TBA.
Different performance indicators are used to measure the performance of the yard planning. Dependent on the type of terminal and its properties, certain key performance indicators (KPI’s) are considered more important than others. These KPI’s lead to the specific focuses of yard planning strategies. Main KPI that is used at container terminals is the quay crane productivity. This is because the control of quay cranes is very expensive and the quay crane process is the bottleneck at most terminals.
Other KPI’s used for yard planning (strategies) are:
• Yard density: percentage of reserved/occupied (ground) slots • Workload distribution stacking equipment
• Number of rehandles by stacking equipment (RTG/RMG) • Travel distances transport equipment (TT/AGV)
In this chapter, first the different yard planning strategies with their properties, advantages, disadvantages and involved types of terminals are discussed. The next paragraph handles the different TOS’s and in 3.4 this chapter is closed by a summary and conclusions.
3.2 Literature review yard planning strategies
One of the main processes at a container terminal is container stacking. In the container terminal yards, thousands of containers can be stored. Due to the costs for handling the containers and operating the vessels, efficient and effective storage of the containers can be very valuable. Two main yard planning strategies can be distinguished: ‘pre-marshalling’ strategies and ‘sort and store’ strategies.
3.2.1 Pre-marshalling strategies
When export and transshipment containers (see 2.4) arrive, they are assigned to a temporary area without consideration of their status (pre-stacking). The discharge of these containers from the vessel can be done quickly because the containers can be stored with few restrictions. A container remains in the temporary pre-stacking area until it is known that it has to be loaded on some depart vessel. Before this vessel arrives a stowage plan (see 2.2.1) is made. According to this loading plan, the involved containers are stacked in the pre-marshalling area at a certain time before loading. When involved containers arrive after this moment, they are stored at the marshalling area immediately. An overview is given in Figure 3.1. Merit of the pre-marshalling strategy is that his stacking at the pre-pre-marshalling area is done in such way that the containers can be loaded rapidly on the vessel.
Main goal of the pre-marshalling strategy is minimizing the vessel-time by minimize the discharge and loading time. In order to minimize the vessel-time, the quay crane productivity has to be optimized, which is very desirable.
The movement of containers from the pre-stacking area to the pre-marshalling area is often called housekeeping (3.2.2). According to Chen (1999) is the great number of these container movements (rehandles) the big disadvantage of the pre-marshalling strategy. Each pre-stacked container is stacked at least twice, so double handling is required. The pre-marshalling strategy is mainly used at SC terminals, because a SC can transport and stack the containers. It is very inefficient, regarding equipment use, to apply this strategy to a terminal with internal transport vehicles and stacking cranes (RMG, RTG, ASC). The advantage is the simplified storage planning, according to Chen. The temporary storage can be done pretty regardless (manually). Taleb-Ibrahimi et al. (1993) first focus on minimizing the space used. This implies that contain-ers are pre-stacked as long as possible; in the pre-stacking area blocks can be filled completely, while the stacking in the pre-marshalling area is done according to a number of rules, which requires more space. In their second strategy they reduce the number of rehandles, which implies minimizing the number of containers that are pre-stacked. As many containers as possible are stacked in the pre-marshalling area. Disadvantage is that information about the loading of these containers must be known early.
An optimization model for the pre-marshalling problem is given by Lee and Hsu (2007). They minimize the number of pre-marshalling movements by finding a shortest path in a directed graph. The pre-marshalling strategy is a strategy on both the tactical (reservations of blocks for just discharged containers) and operational level (exact placement in pre-marshalling).
3.2.2 Sort and store strategies
Unlike pre-marshalling strategies, export containers are placed directly in their final position in sort and store strategies. The sort and store strategy can be static or dynamic. Static means that containers are only moved when needed for transport. When the strategy is dynamic, there can be housekeeping moves. Housekeeping is reshuffling the already stored containers because of changed information or shortage of space. A rehandle is a movement of a container in order to reach another container in the stack. The sort and store strategy mainly has three focuses:
• Maximizing quay crane productivity. • Minimizing the number of rehandles. • Balancing workload yard cranes.
Most authors do not combine these focuses in one research, but they can be combined. • Maximizing quay crane productivity
Preston and Kozan (2001) try to optimize the quay crane productivity by minimizing the average, i.e. total, setup and travel time from the locations of the export containers to the quay.
(QC’s), so the quay crane productivity is optimized. This implies minimizing the time a vessel is berthed at the terminal. The setup time is the time needed to place the container on the travelling vehicle, including needed rehandles. Preston and Kozan determine the best location by solving a genetic algorithm and conclude that the container handling order (FCFS, LCFS, random) doesn’t affect the results. But they do not take into account the different properties (Appendix A) of the containers.
Instead of minimizing the travel distances, Kim and Park (2003) minimize the travel costs of a container from storage location to berth. They do this by a decision-based heuristic regarding the expected duration of stay (DOS) of a container and by sub-gradient optimization. For the DOS method, the information about the departure time of the containers must be known. Both Preston and Kozan (2001) and Kim and Park (2003) claim their strategy is applicable to differ-ent types of terminals. The purpose of maximizing quay crane productivity is indeed a purpose that all types of terminals have. But minimizing travel distances is more important for on-chassis (see 2.6.1) and SC terminals (see 2.6.2) than for RTG and automated terminals, because of the relatively larger yards.
Duinkerken et al. (2001) give a very simplistic strategy; closest position. In this strategy, every container is placed in the closest available slot. This is very inefficient, because containers that are discharged after each other have to be stacked by the same handling equipment, so this equipment is very busy, while other equipment has no work to do. Furthermore, many rehandles can occur because containers are not stacked by any property. The closest position strategy therefore is not a very realistic strategy.
Minimizing travel distances/costs mostly is applied to groups of containers and blocks (tactical level) instead of individual containers and locations (operational level). Disadvantage of this strategy is that many stacking locations need to be reserved in advance, so the number of occupied locations is the actual occupied location plus the reservations. This results in less free stacking locations. Also, this strategy is often applied manually, which results in non-optimal decisions in most cases. Minimizing travel distances for blocks of containers of certain groups is often applied at RTG and automated terminals (see 2.6.2).
Duinkerken et al. (2001) determine the location of the containers by minimizing the reduction in remaining stack capacity (RSC). The RSC is the number of empty slots in a pile times the number of classes that are allowed to be stacked on top. The classes could be different classes of weight (heavier on top) or expected Duration Of Stay (DOS, shortest on top) for containers that are to be loaded on the same vessel. Containers are only allowed to be stacked on top of a higher class. So information about the containers is required to categorize it. Because of the varying container placement locations, this strategy is mainly used for automated terminals, see 2.6.2.
In the example in Figure 3.2, the containers are grouped in 10 ascending classes and the maxi-mum stacking height is four containers. In the first pile, only one slot for a container of class 1 (heaviest or shortest DOS) is available, so the RSC is 1*1=1. In the second pile, two slots for class 2 or lower (class 2 is the lowest class in the pile) are available; RSC=2*2=4. For the third and the last pile, the RSC is respectively 2*10=20 and 4*10=40. A new container will be stacked on the location where the reduction of the RSC is minimal. So the eventual RSC after placement has to be calculated for each pile. A new container of class 2 is not allowed in the first pile. For the other piles:
(2) RSC after placement: 1*2=2 reduction: 2 (3) RSC after placement: 1*2=2 reduction: 18 (4) RSC after placement: 3*2=6 reduction: 34 The new container will be placed in the second pile.
Figure 3.2: Example of calculation of the RSC in a bay Source: Duinkerken et al. (2001)
• Minimizing number of rehandles
Dynamic programming is used by Kim et al. (2000) to derive the optimal decisions for storing an export container. Main rules are that containers can not be stacked on heavier containers, regarding the loading sequence of a vessel (heavier first for stability), and that container is stacked on the lowest allowed location, see Figure 3.3. This is a form of category stacking, where the containers are grouped by depart vessel or by Port Of Discharge (POD). The optimal decisions for a location minimize the number of rehandles. This decision tree is usable in real-time decision making.
Figure 3.3: Optimal storage locations for heavy (H), medium (M) and light (L) containers Source: Kim et al. (2000)
Kim and Kim (1999) present a programming model to minimize the number of rehandles by determining the optimal stacking height for import containers. They use the segregation space allocation strategy, i.e. it is not allowed to stack containers on containers that have arrived earlier. The segregation strategy could be seen as form of category stacking, but in this research it will be treated separately. This strategy could also be applied to export containers.
The blocks to store the containers are mostly manually and sometimes by automatic calculating with the strategy of minimizing travel distances. Minimizing the number of rehandles is a strategy that needs to reserve space for (groups of) containers. These containers are mostly grouped on POD. This is the same disadvantage that the minimizing travel distances strategy has. Advantage is that the decision rules for stacking in a block are not very difficult.
The higher the stacks can be, the more important minimizing the number of rehandles is. So the strategy of minimizing the number of rehandles is more common at RTG and automated terminals, see 2.6.2. Advantage of this strategy is that no ground slots have to be reserved. The slots above containers that are already stacked are reserved for containers with same desired properties. Minimizing the number of rehandles is a strategy on operational level.
• Balancing workload
The balancing workload strategy focuses on dividing the yard work over the blocks and so over the yard equipment. In most strategies, groups of containers are assigned to the same block by certain properties. So the traveling vehicles and stacking cranes have a great workload at this block, but the equipment for the other blocks has far less to do. Balancing the workload results in more efficient use of equipment, because the containers of one group are divided over different blocks. This dividing can be one equally or random.
Murty et al. (2000), Zhang et al. (2003) and Bazzazi et al. (2009) all divide the containers equally over the blocks. They do not take into account the properties of the containers, but Bazzazi does regard the fact that some types of containers (reefers) need to be stored at certain designated locations. This strategy is applied to the RTG container terminal at Hong Kong, which has a small yard with a high throughput, so the yard (stacking equipment) has to be
handled efficiently. The assumption is that every block is served by one RTG, so the workload is distributed very equal. This strategy could be applied to automated terminals as well, because it also has specific stacking equipment that has to be used efficiently (see 2.6.2).
A random strategy is used by Duinkerken et al. (2001) and Dekker et al. (2006). They both use a complete random strategy, which does not work very well, because stacking all containers just allover the yard without rules returns in many rehandles when picking up containers.
Dekker et al. (2006) improve this random strategy to a category random stacking, which is also used by TBA, where it’s called controlled random stacking (CRS). This strategy randomly assigns containers to blocks where they can only be stacked on piles of the same property, mostly POD, or on the ground when there are no such piles yet.
The CRS strategy is a recently developed strategy that requires a lot of automation. Therefore it is only used by some new automated terminals (AGV’s and ASC’s). Applying CRS at a termi-nal with manual yard planning is very hard and takes much effort, because the yard location (block) to store a container differs for each individual container. Also, in container terminals with manually controlled equipment, it is common to assign groups of containers to a block or a bay, so it is easy for the drivers to know the location to deliver the container. There is resistance of TT and SC drivers against changing deliver locations for each new container. So the disad-vantage of CRS for RTG terminals is the difficult implementation, but it appears to be efficient enough to some RTG terminals, like Algeciras, to consider the implementation. Advantage of the category/controlled random strategy is that reservations of ground slots are not needed, because the location of a container is chosen just when it needs to be stored.
The balancing workload strategies are mainly on the operational level, because a location is determined for each individual container. On the tactical level, another strategy can be used. For example, a part of the yard where the containers of a certain vessel are allowed to be stacked, can be allocated the berth location of this vessel by a minimizing distances strategy.
Finally, Duinkerken et al. (2001), give a yard planning strategy that is not very realistic for export containers. The strategy is levelling, which just fills the block by tier. First the ground is filled, then the second tier and so on. This strategy minimizes the maximum stacking height. This strategy does the stacking regardless of any property, so very inefficient container place-ment could occur. It however could be applied to import containers, when they are picked up randomly. This strategy can be applied to all types of container terminals, with a one tier height for on-chassis terminals.
3.2.3 Summary
A reasonable number of authors have paid attention to yard planning strategies. Most of them apply their yard planning strategy on an example terminal and not on a real one, so most strategies are applied to models with lots of assumptions. Optimal use of the strategy is difficult in real terminals. In most container terminals (except automated terminals), a lot of the planning is done manually, especially the assignment of groups of containers to blocks. Besides, pretty often situations appear that cannot be handled by the strategy, so additional manual planning is needed.
Furthermore, it’s difficult to tell a strategy is good or bad. You can, choosing a best strategy for a container terminal, compare different strategies regarding the properties of the terminal. A yard planning strategy usually isn’t the unique strategy of a container terminal. Mostly, a tactical strategy can be combined with an operational strategy. For example, minimizing travel distances can be applied to groups of containers on the tactical level and minimizing the number of rehandles to individual containers of a certain group or block on the operational level. A schematic overview of the yard planning strategies can be found in Figure 3.4.
3.3 Yard planning strategies supported in TOS’s
Three commonly used terminal operating systems will be investigated. These TOS’s are: • SPARCS
• SPACE (CTCS) • CATOS
The focus is on the way the yard planning can be done in the TOS’s and capabilities the TOS’s have for implementation of the yard planning strategies.
3.3.1 SPARCS
SPARCS, a TOS by Navis, is the most common used TOS at container terminals. It has the Expert Decking (ED) function, see the SPARCS and the SPARCS ED manual. ED is one of the main elements that is used to perform the yard planning at a terminal.
• Yard planning in SPARCS
Three main elements are involved in yard planning performed by SPARCS. For a detailed description of the yard planning in SPARCS, see chapter 4.
1. Yard allocation
Containers are categorized in allocations groups by certain properties. Locations, mostly (parts of) blocks, in the yard are assigned to the different groups; these are the allocation ranges. These settings are set manually. When a container arrives, it is assigned to the allo-cation range that is reserved for the group this container belongs to.
2. Section and stacking factor
Applying the section (bay) and stacking (pile) factor is the second step. With these factors, groups of containers that are desired to be stacked together in a bay and a pile are defined. In the section and stacking factor, different container properties (see Appendix D) by which they are matched can be included; factor elements. Mostly, the stacking factor is more spe-cific than the section factor and therefore has more properties included. The factors can be set for different groups of containers, defined by certain properties.
3. Expert Decking (ED)
The third step is the ED strategy. After the second step, matching piles are known for a specific container that has to be placed. ED applies a penalty-based strategy to choose the best pile. This strategy has tens of penalties and a few bonuses (negative penalties) for matching bays and piles, travel distance, high stacking, block fullness, stacking by weight, etc. For an overview of the penalties, see Appendix B and Appendix C. The available slot that has the lowest sum of penalties is the best location. Different yard planning strategies
can be implemented by tuning the parameters (penalties). An overview of the steps is given in Figure 3.5.
Figure 3.5: Steps in SPARCS for defining container placement strategy
• Implementation of yard planning strategies in SPARCS
Because of the many containers properties and penalties that can be set and tuned, quite many yard planning strategies can be implemented in SPARCS. For pre-marshalling strategies the section factor can be set in such manner that all containers from a certain vessel are placed in the same bays. Also, there are bonuses for having bays with the same section in the same row and place the following bay next to the last one. Complete blocks of containers coming from the same vessel can be built this way. The moment of pre-marshalling containers of these blocks have to be set manually (just as housekeeping moves), but with the right penalty tuning, the containers can be placed in the right vessel loading order.
With ED, a number of travel penalties are available, so minimizing travel time of individual containers is possible. These penalties are given to individual containers, so it is not possible to apply minimizing travel distances for yard allocation automatically. Furthermore, maximize RSC appears possible. Penalties can be given for stacking on a certain height and for stacking on a higher class. It is however not possible to multiply these penalties.
Minimizing the number of rehandles can be implemented in SPARCS too. Setting the penalties high or infinite for stacking on containers with different properties, will avoid rehandles because similar containers are stacked in one pile as much as much as possible. The focus can be on the properties of which mixed piles cause many rehandles.
The balancing workload strategies are more difficult. Equally distribution of containers of one group over the yard has to done manually. The random strategy is possible. The yard allocation for groups of containers can be let go or very rough, so containers of one group may be placed (almost) everywhere in the yard. If no properties are selected for the section factor, a container won’t be matched to bays. The stacking factor can be very specific, so only a very ‘similar’ container (same POD, same depart vessel) may be stacked on a certain matching pile. When no matching pile can be found, an empty TGS has to be appointed. This TGS cannot be chosen randomly, since a random function doesn’t exist in SPARCS. Penalties can be set in such
manner that piles with the same stacking factor won’t be placed close to each other. This is possible because penalties can be negative (so actually a bonus).
3.3.2 SPACE
SPACE is part of the Container Terminal Control System (CTCS), a TOS by Cosmos. SPACE is the yard planning tool of CTCS.
• Yard planning in SPACE
Just like in SPARCS, the yard planning strategy in SPACE is implemented in three main steps (SPACE-TRAFIC Tutorial version 1.0).
1. Combinations of container properties
The first step is the definition of combinations of container properties; this is pretty similar to the defining of groups of containers to groups in SPARCS. All containers are catego-rized in groups by properties.
2. Allocation
Secondly, locations (mostly lanes or complete blocks) in the yard are reserved for all com-binations; allocation.
3. Placement algorithm
The last step is defining the placement algorithm that is appointed to a combination. This algorithm consists of parameters and patterns. These parameters are not penalties like in ED. The parameters give permission for or prohibit a location when it has or hasn’t the property belonging to this parameter. The patterns mainly specify the order in which the containers of the same group are placed. For example they can be placed horizontally, ver-tically, on the same POD, on weight, etc.
When a container arrives, SPACE first assigns the container to a combination (group). Then SPACE looks for an area that is reserved for this combination. If there are no free slots anymore in this area, SPACE can find a new block/row/bay to reserve for this combination or the con-tainer can be placed manually. When the area is found, according to the algorithm the exact slot for the container is determined. An overview of the steps for a container placement is given in Figure 3.6.
Figure 3.6: Steps in SPACE for determining container placement
A lot of configurations in SPACE have to be set manually. The combinations of containers have to be set, areas for the combinations have to be reserved and an algorithm for each combination has to be established. There can be a lot of combinations, reserved areas and algorithm, so this is a lot of work, but very specific combinations and algorithms can be set. When the settings are done, SPACE can handle the placement on itself. However, it is wise to review the settings continuously, for example in busier or quieter times. An advantage is that it is possible to assign more than one area to a combination and prefer certain areas over others. The other way around is it also possible to assign more than one combinations to one area and prefer certain combina-tions.
• Implementation of yard planning strategies in SPACE
SPACE is a TOS that can handle a few yard planning strategies itself. The strategies for mini-mizing the number of rehandles can be applied, because it is possible to set rules for stacking on same POD, stacking on lighter or same weight or not stacking on older containers for imports. The reservations for groups of containers are done manually in SPACE; therefore minimizing travel times/distances is difficult. There is the boundary option which assigns berth locations to blocks, so containers won’t be transported all over the yard.
For pre-marshalling strategies, manual intervention is required. It is possible to make a combi-nation of all containers arriving by the same vessel and reserve an area close to this vessel. But afterwards the restacking/pre-marshalling has to be done manually.
Applying random strategies is very difficult in SPACE. The random function doesn’t exist. It is possible to assign individual slots/piles to combinations of containers, but SPACE will still assign containers to these locations by an algorithm. Furthermore, the reservations for contain-ers are still fixed, so when there are changes in the number (of containcontain-ers) of a combination many reservation areas have to be adjusted.
Finally, the yard planning strategies of maximizing RSC and distributing containers equally over the yard are just not applicable in SPACE. It could be done manually, but that’s not desirable, due to the amount of calculations.
3.3.3 CATOS
TSB is the editor of the TOS CATOS. One of the main parts of CATOS is the planning system, which includes yard planning.
• Yard planning in CATOS
The yard planning in CATOS is set in three main steps (Yard Planning Manual for CATOS). 1. Grouping pattern
In the grouping pattern, containers are grouped by properties. Also, boundaries in the yard for different groups are defined. Inside a set of boundaries, more groups can be stacked. CATOS itself can choose the block to stack a certain container. This is done by choosing the best location inside the boundaries that satisfies the next steps.
2. Positioning rules
Stacking orders (directions) and heights are defined here, just as prohibitions and permis-sions for mixed stacking of containers with the same certain property. Numerous position-ing rules can be set.
3. Assignment of positioning rules to groups
The last step is the assignment of positioning rules to a certain group along with other in-formation, like the estimated number of containers of one group and the number of blocks where containers of this group may be stacked inside the boundaries. When a container ar-rives, it is assigned to a group and thus to a positioning rule. By applying this positioning rule inside the boundaries, CATOS determines a slot to place the container. The steps for setting the yard planning can be found in Figure 3.7.
Figure 3.7: Steps in CATOS for defining container placement strategy
• Implementation of yard planning strategies in CATOS
CATOS has a couple of options that are really helpful for implementing some yard planning strategies. The first is the pre-marshalling option. With this option, containers with certain properties, like POD or depart vessel, can be chosen to undergo a pre-marshalling (housekeep-ing). This way, containers can be picked out and restacked easily for loading. So pre-marshalling strategies can be used well in CATOS.
Also, CATOS has the Random Grounding option. When a container arrives without a pre-planned location, CATOS can find a position that agrees with the relevant properties and equipment availability. So is appears that CRS can be applied with CATOS. Equally distribu-tion of containers over (parts of) the yard can also be applied. In CATOS a number of blocks can be given for stacking a group of containers and there is an option of scattering these con-tainers over these blocks. So the balancing workload strategies can be implemented.
The strategies of minimizing the number of rehandles can be used in CATOS. In the positioning rules, mixed stacking can be allowed or prohibited for certain container properties. Also, it is possible to stack containers only on the same or lighter weight.
Maximizing RSC is not an option in CATOS. Minimizing travel time/distance does not seem to be an option of CATOS either. By setting the boundaries, which is done manually, the travel distances need to be taken into account also manually.
3.4 Summary & Conclusion
Different yard planning strategies and terminal operating systems are discussed in this chapter, along with the capabilities of the TOS’s to have the yard planning strategies implemented. A complete overview is given connecting yard planning strategies with their focuses and features to types of terminals and TOS’s. In Table 3.1 an overview of yard planning strategies for each type of container terminal, following from the literature review is given.
