Computational Logistics: Examples from Container Terminal Operations. Stefan Voß, IWI

Transcription

1 Computational Logistics: Examples from Container Terminal Operations Stefan Voß, IWI 09 November 2011

2 HongKong, May, 29th 2012 Container Terminal Operations and Scalability: Efficiency means Quality Introduction and Motivation History and Background Terminal structure and handling equipment Terminal logistics and optimization methods Scalability and Research Issues Stefan Voß Institut für Wirtschaftsinformatik (Information Systems), University of Hamburg

3 1. Introduction and Motivation History and Background 3

4 Computational Logistics Computational logistics concentrates on methods and information systems that support or perform the planning and control of logistics systems and processes. 4

5 Computational Logistics Computational Logistics involves the use of information systems and modern information and communication technology (IT) for the design, planning and control of logistics networks as well as the complex tasks within them. 5

6 Containerization Trend: High Growth of Container Turnover Container port turnover is growing much faster than the world trade, world GDP or seaborne trade. 11 Data sources: UNCTAD - Review of Maritime Transport; Drewry Shipping Consultants; Clarkson Research Services

7 Competitiveness of a container seaport Main success factors: time in port for ships (transshipment time) combined with low rates for loading and discharging reliability 17 Crucial competitive advantage: rapid turnover of the containers reduction of the transshipment time reduction of the costs of the transshipment process overall objective with respect to terminal operations: efficient usage of all resources minimization of the time a ship is at the berth!

8 2. Terminal structure and handling equipment 18

9 Operation areas of a seaport container terminal and flow of transports Truck and Train Operation Area Hinterland Operation Landside Transshipment Process (LTP) 21 Yard Import/Export Stock Yard Moves Quayside Operation Ship Operation Area Empty Stock Sheds Waterside Transshipment Process (WTP)

10 Example of a partially automated container terminal (Brisbane, Australia) 22

11 23 Example of an automated container terminal (Container Terminal Altenwerder (CTA), Hamburg, Germany)

12 Waterside Transshipment Process Incoming ships are assigned to a berth. Containers are cleared by cranes (bridges) at the ship. Special vehicles (e.g. straddle carriers) perform all transports within the yard and to the ship. 25 Logistic problems berth planning stowage planning yard planning sequencing of all containers (loading and unloading sequence) assignment of (transshipment) containers to straddle carriers

13 Storage Area Handling Equipment Cranes Double/Triple Rail Mounted Gantry Cranes (DRMG, TRMG) 34 Cranes operating in storage area Rail mounted gantry cranes (RMG) stable new development: Double RMG consists of two RMGs of different height and width able to pass each other avoiding a handshake area slightly higher productivity of the system man-driven or automatic span up 8-12 rows allow for stacking 4-10 containers high technical performance: ~ 20 moves/h

14 Horizontal transport means Reachstackers and forklifts/sidelifter 38

15 Automated guided vehicle 39

16 Straddle carriers Handling Equipment 42

17 Container terminal systems 45

18 Container terminal systems schematic side view 46 Is this a Supply Chain?

19 3. Terminal logistics and optimization methods 47

20 New Real-world Problem Research No systems, no impact This profession [Nievergelt, [OR] 1994] has presently more algorithms than applications...? [Wolsey, 1979]?? 48 Solution

21 Basic structure of handling procedures for container flows 49

22 Methods /Algorithms Assignment problems (exact solution) Vehicle routing problems (incl. TSP, TDTSP; simple heuristics) MIP models (reference solutions, subproblems; standard solvers) 50 Online optimization (simple priority rules) Characteristics of container terminal operation demand online (realtime) optimization and decision. Reason: Most of the processes occurring at container terminals cannot be foreseen for a longer time span in general the planning horizon for optimization is very short. Methods: Heuristics, Metaheuristics, (Discrete Event) Simulation etc.

23 CTA at a Glance Overview: Remotely Controlled Transfer 61

24 4. Scalability and Research Issues 62

25 Scalability on a small scale (vs. Scalability on a large scale) Double Rail Mounted Gantry Cranes (DRMG) Allow for stacking 4-10 containers high Technical performance: ~ 20 moves/h Net productivity?? 63 The more we utilize the space the more we need to account for making the right decisions where to locate containers. That is, for bad locational decisions we might have to relocate.

26 The Blocks Relocation Problem (BRP) 64 Given A bay with n blocks which have to be retrieved Assumptions (Kim and Hong, 2006) Retrieval order is given (1, 2,, n) LIFO Only the upper blocks may be moved No pre-marshalling Objective Retrieval order such that the number of relocations is minimized

27 The Blocks Relocation Problem (BRP) Assumptions discussion Assumptions (Kim and Hong, 2006) Only the upper blocks may be moved No pre-marshalling 65

28 The Blocks Relocation Problem (BRP) Dynamic Programming Formulation State variable: s = (k, i, t,c) 66 k {1,..., n} is the block to be retrieved i {1,...,m} is the stack in which the target block is found t is the list of blocks above the target block C is the configuration of the remaining blocks (Example: k = 1, i = 2, t = {5, 4}, and C = {{3, 2}, {7, 6}})

29 The Blocks Relocation Problem (BRP) Dynamic Programming Formulation State transition function T(s, x): Let s = (k, i, t,c ) be the state obtained by applying decision x D(s) to the current state s. s = T(s, x) 68 (i) t = : k = k + 1, i is the stack in which block k+1 is currently located, t is the list of blocks above k + 1 and C = C is the configuration of the remaining blocks (ii) k = k, i = i, t = t \ {τ}, and C depends on the application of move x to block τ (e.g., with respect to the figure let us suppose that x = 1 s = T(s, 1) = (1, 2, {4},C ), where C = {{5, 3, 2}, {7, 6}})

30 Dynamic Programming 69

31 Move-based vs. Method-based Neighborhoods (Paradigms?) Move-based neighborhoods: Neighborhood search based on topological concepts relating a given solution to similar solutions via local changes or moves. (Example: small homogeneous neighborhoods, very large scale neighborhoods) 1. Define a neighborhood 2. Find efficient methods for its exploration 71 Method-based neighborhoods: The basic structure of a neighborhood is determined by the needs and requirements of the method used to search it. 1. Assume a given method. 2. Define a neighborhood that suits the method.

32 The Corridor Method (CM) Sniedovich/Voß (2006) The CM can be described as a local search method where neighborhoods are relatively large sets whose structure and size are compatible with the optimization method operating on it. The optimization problem under consideration is large. There is an optimization method for efficiently solving smaller instances of the problem. It is easy to generate (an initial) feasible solution. There is an efficient method for generating suitably large neighborhoods around feasible solutions to the problem on which the optimization method can be used. 72

33 The Corridor 73

34 Numerical Results 74

35 A New Talk? 75

36 Extending towards Pre-Marshalling Numerical Results Only very few benchmark instances available. Example: Y. Lee and S.-L. Chao. A Neighborhood Search Heuristic for Premarshalling Export Containers. European Journal of Operational Research. doi: /j.ejor (Lee/Chao) 19 (CM) 15 (optimal)

37 79

38 Scalability Scalability on a large scale (vs. Scalability on a small scale) Technology combinations 80 Horizontal transport - quay Container yard operations Horizontal transport - rail TTU RTG TTU AGV Auto. RMG TTU Mini SC Auto. RMG TTU Typical system characteristics RTG + TTU: RMG + AGV: pure SC system (incl. truck handling) high yard capacity / high stuff require. / flexible in peaks high civil work invest. / high degree of auto. / low staff require. RMG + Mini SC: high civil work invest. / auto. yard operations / reduced staff requ. Pure SC system: poor yard capacity / high stuff require. / flexible in peaks

39 Scalability is frequently quantified by Terminal Cost (cost / box) and Handling Capacity (max. throughput / time) TERMINAL-related LEVEL Technical and functional attributes Application Conditions Terminal Organization REGION-related LEVEL e.g., stacking height, transport speed, energy consumption, manning requirements, e.g., level of wages and energy prices, inflation rate, terminal space, length of quay wall e.g. organizational concepts, methods for equipment control, human experiences, 81 Competitiveness of Handling Technology

40 Scalability General aspects of investigation: Equipment variants: TTU+RTG // AGV+RMG // SC+RMG // pure SC Replacement of equipment and civil work components by technical lifetime Terminal processes: Quay crane operations Horizontal transport at quay- and landside (excl. MTs) Container handling in the yard Truck handling 82 Investment in equipment and civil work Operations cost: Labor cost Energy cost M&R cost Capital cost Terminal capacity

41 Scalability Basic assumptions (collaboration with Jürgen Böse): common organized terminal resources + modern equipment pieces (state of the art) further assumptions: equipment productivities, space requirements for yard stacks, energy consumption, staff requirements, etc. Planning horizon: 25 years quay wall length: 800m; capacity: 1.47 mill TEU (9 STS) 83 Yard: capacity: area 800m (parallel to quay), 500m depth RTG (2.78 mill TEU), RMG (2.48 mill. TEU), pure SC (1.94 mill TEU) => Terminal throughput is limited by quay wall

42 Scalability Conditions of Central Europe relating to wages, prices for energy / equipment / M&R services, inflation rate (3%), capital cost) Terminal capacity is equal for all equipment variants Region-related competition level is measured by cost/box 120% 100% 110% Total Cost per Box (25 years) 83% 96% 100% 84 80% 60% 40% RMG yard systems deliver best results! 20% 0% RTG + TTU RMG + AGV RMG + Mini SC pure SC

43 Scalability Condition changes: 5% of European wage level (e.g. less developed countries) Total Cost per Box (25 years) 120% 100% 110% 88% 83% 106% 96% 109% 100% 100% 85 80% 60% 40% competitiveness changes! RTG system takes pole position 20% 0% RTG + TTU RMG + AGV RMG + Mini SC pure SC

44 Scalability Yard area is halved => terminal throughput is limited by yard Region-related competition level is measured by yard capacity 150% 135% 143% -15%-points Yard Capacity (limited by yard) 128% 128% 120% % 90% -28%-points 100% RTG + TTU RMG + AGV RMG + Mini SC pure SC RTG system offers best yard efficiency

45 Summary / Hypotheses Container turnover is expected to continue to grow. The size of container vessels will increase. Shortages in capacity at terminals and in surface distribution networks are viewed as the main constraints to current and future growth in containerized trade. There are also capacity issues with inland transportation by rail and truck. Government and port community stakeholders may present additional obstacles to capacity enhancement (road and rail infrastructure, security, environmental protection and safety). It is a challenging area for joint work in practice and research let s face it. 87

46 Questions? Answers? Post them towards 88

47 89

48 90

49 Green Maritime Transport and Environmental Paths to 'Green Logistics' Air Pollution (Emissions) Oil Spill Slow Steaming 92

50 Slow Steaming Slow Steaming: 21 ± 1 Knots Extra Slow Steaming (ESS): 18 ± 1 Knots Super Slow Steaming (SSS): 15 ± 1 Knots On 45 loops 445 container vessels (> TEU) were operated under ESS in cf. Maersk Line (2009)

51 Slow Steaming 120, , , , , , TEU TEU TEU TEU 94 cf. Notteboom (2009)

52 Environmental Paths to 'Green Logistics' International Maritime Organisation (IMO) Base: Intergovernmental Maritime Consultative Organisation (IMCO) The IMO aims at controlling all issues related with merchant shipping (except pure commercial aspects), reducing the pollution of the sea as well as improving safety and conditions of work. The IMO collaborates with other UN organisations as well as with NGOs. For example, the IMO enacted the International Convention for the Prevention of Pollution of the Sea by Oil (OILPOL, 1967) resulting in protected zones. No ship is allowed to drain oil-containing waste water/alloy inside of those zones. However, the IMO has no executive power. 95 MARPOL as Guideline The major international convention with respect to green maritime transport is the International Convention for the Prevention of Pollution from Ships (MARPOL Convention). It is a combination of two treaties adopted in 1973 and Currently, the convention includes six technical annexes: I - Regulations for the Prevention of Pollution by Oil, II - Regulations for the Control of Pollution by Noxious Liquid Substances in Bulk, III - Prevention of Pollution by Harmful Substances Carried by Sea in Packaged Form, IV - Prevention of Pollution by Sewage from Ships, V - Prevention of Pollution by Garbage from Ships, and VI - Prevention of Air Pollution from Ships. Annex VI was subsequentliy introduced in 1997 since the public and politics were not reasonably aware of air pollution caused by emissions not till then.

53 Technical Aspects for Green Maritime Transport 96 Green logistics in maritime transport aims at reducing negative impacts of maritime traffic on environment and nature. Maritime traffic and transport offers a broad range of possible improvements by applying modern technical instruments as well as modern logistical concepts. Some technical means for reducing pollution in building and operating vessels can be directly applied, some require a modication ('retrofitting') of the ship, and some others can only be applied for newbuilding of ships.

54 Some innovations will make it Think of possible innovations! Skysails? Foldable Containers? Others? 97

55 Some innovations will make it Skysails? 98

56 Some innovations will make it Sailing rotors use wind energy to propel the ship and works on the the principle of the Magnus effect named after a German physicist. Enercon? 99

57 Some innovations will make it Solar panels? For instance, on Auriga Leader

58 Some innovations will make it Rulfs (2009) Slow Steaming Kit Wärtsilä Switzerland Ltd. (2009) 101

59 Empty Container Management: Foldable Container? Leanbox Idea: automated (dis)assembling 102 Source: Leanbox GmbH

60 Economical Approaches for Green Maritime Logistics 103

61 IMO: Stepwise reduction of maximum permissible sulphur concentration in marine fuel Since October 2008, IMO enactments show limit values such that emissions of NOx and SOx have to be reduced gradually until Rules for limiting the emitted amount of, e. g., SOx directly require complex and costly measuring instruments aboard. Alternatively, the amount could be calculated based upon ship's characteristics but this approach requires complex documentation and would be inaccurate. Therefore, the IMO enactments aim at a regulation of the sulphur content of the fuel. (Sulphur) Emission Control Area ((S)ECA) 104

62 IMO: Stepwise reduction of maximum permissible sulphur concentration in marine fuel 105

63 Panama Canal Expansion 106

64 Panama Canal Expansion 107

65 Management Science Issues 108 Policy instruments aiming at a reduction of emissions such as CO2 caused by maritime transport can be broadly divided into two categories. Some instruments affect only investment decisions while others, e.g., fuel taxes or cap-and-trade systems, affect the running of vessels and investments in ships. In other aspects OR may help.

66 Trade, Transport and Ports Container Flows The explosion of trade in containerized maritime transportation is increasing pressure on the leading sea ports. Major challenges of the coming years: Trade imbalances, congestion, etc. Estimated cargo flows on major global containerized cargo routes (Million TEU 2008) ~40.0 (2007, est.) North America < > Europe < > Asia < > North America 109 Asia, particularly China, is still leading the growth in cargo flows Volume imbalances large export flows, but only small import flows (but new developments in 2008/09 due to restructuring of economies; but no reversal) <East-West TOTAL West-East> Data source: UNCTAD - Review of Maritime Transport 2009

67 Empty Container Repositioning Critical Factors 110 Root cause: Trade imbalance (geographic) Dynamic trade effects (demand changes over time; seasonality, etc.) Uncertainty (unpredictable effects, e.g., customer demands, container processing, weather, traffic ) Different container types Blind spots in transport chain (lack of data, in particular w.r.t. rail, truck, inland terminals ) Carrier s operational and strategic practices (alliances, sharing/exchanging containers, etc.) Imbalances inside containers, in transport costs 20% of total container flows at sea are empty (worldwide) The proportion of empty containers is expected to be nearly 23% in 2015 Costs of repositioning: US$ 400 per empty container (2002, Drewry Shipping Consultants) Total costs per empty container (transport, handling): US$ 675 (Behenna 2001) Containers are empty more than 50% of their total cycle time (maintenance/repair, storage, transit)

68 Imbalances not only in volume but also inside the containers and in transport costs Transport costs (June, 2007) : China Rotterdam: US$ 1,600 Rotterdam China: US$ B. Kuipers, TNO - China, Containers, congestion, CO2 and chains July, 2007

69 Empty Container Time Series Forecasting Model Development ABC-XYZ Analysis Objective: estimation of the need for empty containers (of a particular type) in particular areas Idea: Forecasting demand based upon past data Preliminary analysis of assortment of all time series (importance, predictability) based upon their characteristics Analysis of the complete time series set for prioritization of forecasting methods Structuring forecasting approach to align data & suitable methods 112

70 Empty Container Time Series Forecasting Model Development Time Series Analysis 113 Congestion effects Calendar effects for Asia (example major shipping company) Chinese New Year Christmas & Gregorian New Year Chinese National Holiday (different lead & lag per year) Japanese Day of the Children (different lead and lag per year) Encoding of many calendar effects for different regions Statistical testing of significance of calendar effects outlier outlier

71 Trade, Transport and Ports Issues Issues of leading sea ports (not exhaustive) Strong growth in containerized cargo Congestions (US, Europe) (radical) Innovations? 114 Highly fragmented industry with a large number of agents Precise information for planning as well as operations is needed 9/11 Increased pressure to comply with international security requirements Current global economic crisis: even more pressure for improvements in order to increase efficiency and reduce costs

72 Perspective of seaports >2050 (?) and summary less congestion more automated; employing less port workers (+ shift of skills that are needed in order to control automation) smoother trends (less driven by steep and upward growth patterns) driven by new feedstock (e.g., gas, biomass) less emissions (NO x etc.) more dispersed (virtual port facilities, dry ports, hinterland activities) more in line with common supply chain practises on one hand: resemble ports of today quite a lot on the other hand: might be based on totally new infrastructure 115

73 116 ZPMC, Yangshan Ports, etc.