Intermodal Transportation Teodor Gabriel Crainic ESG UQAM & CIRRELT - CRT CIRRELT Plan What are we talking about? à Container-based intermodal transportation à System design (location) à Fleet Management (empties) à Perspectives à 2 © Teodor Gabriel Crainic 2006 What are we talking about? 3 © Teodor Gabriel Crainic 2006 Intermodal Transportation Simple and straightforward definition: à Movement of a person or a load by a sequence of at least two transportation modes, the transfer from one mode to the next being performed at a (intermodal) terminal à E.g., Door-to-door transportation of containers Ó Over long distances Ó Origin → “land” transport → port → container ship → port → “land” transport → destination à 4 © Teodor Gabriel Crainic 2006 A “Strict” Definition Movement of goods à One and the same loading unit or vehicle à A chain of à Several transportation modes (services) Ó Coordination Ó Interactions à Intermodal terminals Ó No handling of the goods themselves à Door-to-door service à European Conference of Ministers of Transport (1993) 5 © Teodor Gabriel Crainic 2006 A More General Definition Movement of goods à A chain (network) à Several transportation modes (services) Ó Coordination (more or less) Ó Interactions à Intermodal terminals à “Door-to-door” service à 6 © Teodor Gabriel Crainic 2006 Many Things to Many People Major instrument for E.U. policy aimed at switching freight from trucks & highways to more environmentfriendly modes (rail, water) à Dedicated rail services (subdivisions) to move large volumes of containers/trailers over long distances: the trans-continental “land bridges” à Container transportation à Consolidation carriers: local & long-haul operations, several long-haul types of services, with/without use of external services à 7 © Teodor Gabriel Crainic 2006 Many Things to Many People (2) Uncontainerized cargo à Courier (express) services à National planning à City Logistics à Terminal (ports, airports, …) planning and operations à 8 © Teodor Gabriel Crainic 2006 Scope of Presentation Container-based intermodal transportation Ó Illustrative planning/operations management issues Ó Operations research models and methods à A very young field à No definite answers à An open invitation to join in !! ☺ à 9 © Teodor Gabriel Crainic 2006 Plan What are we talking about? à Container-based intermodal transportation à System design (location) à Fleet Management (empties) à Perspectives à 10 © Teodor Gabriel Crainic 2006 Intermodal Transportation – Containers Advantages Ó Reduced cargo handling Ó Increased security regarding damage and loss Ó Increased standardization of transportation and transfer equipment Ó Increased automation of terminal operations ⇒ Cost reduction, more efficient door-to-door transportation à Sustained growth à 11 © Teodor Gabriel Crainic 2006 Evolution of Container Traffic (Koh and Kim 2001) Year 1993 1995 1997 1999 2000 2001 2002 2003 Container traffic (M) 113.2 137.2 153.5 203.2 225.3 231.6 240.6 254.6 Growth rate (%) 12.5 9.8 4.2 10.0 10.9 2.8 3.9 5.8 2004 280.0 10.6 2005 304.0 8.6 12 © Teodor Gabriel Crainic 2006 Intermodal Transportation – Containers (2) Lifeline of world-wide trade and economy à Increasingly larger container ships for inter-continental transportation (liners) Ó These cannot berth at all ports Ó It is not economical to stop at many ports à Container mega ports à New coastal navigation feeder services (“regular” ships): mega ports and huge liners ↔ regular ports ⇒ A new link in the multi-modal chain à “New” long-distance rail services (double-stack) à 13 © Teodor Gabriel Crainic 2006 Intermodal Transportation – Containers (3) Asia (Hong Kong, Singapore, …) to America or Europe: à Origin → truck → port → large container ship (liner) → mega port → “small” container vessel → port → truck/rail/river → destination Þ à Container port terminal transformations for increased efficiency in loading/unloading operations and exchanges with land carriers Ó New terminals / Enhancement of existing ones Ó Automation Ó Intelligent Transportation Systems 14 © Teodor Gabriel Crainic 2006 Notes Container intermodal transportation (& express courier / post services) Ó Customer: Customized service Ó Operator(s): Hub-and-spoke network with consolidation à All long-haul transportation must address the issue of empty vehicle repositioning Ó Trade is unbalanced ⇒ Vehicle flows as well !! Ó Tight fleet size? Long distances? à 15 © Teodor Gabriel Crainic 2006 Plan What are we talking about? à Container-based intermodal transportation à System design (location) à Fleet Management (empties) à Perspectives à 16 © Teodor Gabriel Crainic 2006 System and Service Design Strategic decisions – System Design à Locate (intermodal) terminals à Direct/indirect customer (zone) service à Port/terminal dimensioning Ó Number of berths Ó Size of storage space Ó Type & number of various equipment types à Facility & service abandon, … à 17 © Teodor Gabriel Crainic 2006 System and Service Design (2) Tactical decisions – Service Design Ó Routes served (routes, stops, mode, equipment, …) Ó Service frequency & schedule Ó Cargo routing Ó Terminal workloads à Container port terminal equipment assignment Ó To sea-side and land-side operations à Not many contributions for container transportation à 18 © Teodor Gabriel Crainic 2006 System Design Not many contributions Ó Tactical or operational models to evaluate strategic strategies Ó Ports: queuing, simulation à Discrete location models Ó Consolidation / hub terminals à Network design + location Ó Select direct services/links à Aim to capture economies of scale associated to consolidation of freight à 19 © Teodor Gabriel Crainic 2006 System Design (2) à Location of facilities (terminals) Ó Production-distribution Ó Hub location Ó Location with balancing requirements 20 © Teodor Gabriel Crainic 2006 Location with Balancing Requirements Land part of an intermodal container transportation system (may be generalized) à Use in-land container depots for more efficient operations and reduced empty travel à 21 © Teodor Gabriel Crainic 2006 “Traditional” Operations Loaded Empty Importing customer Exporting customer 22 © Teodor Gabriel Crainic 2006 Operations with In-Land Depots Loaded Empty Empty balancing Importing customer Exporting customer 23 © Teodor Gabriel Crainic 2006 Location with Balancing Requirements (2) Loaded movements are “profitable” à Empty movements are not Ó Customer to depot: Return movement à ISupply of empties Ó Depot to customer: Demand satisfaction IDemand for empties Ó Depot to depot: Repositioning of empty containers due to unbalances in supply/demand 24 © Teodor Gabriel Crainic 2006 Location with Balancing Requirements Network Oi supply Oip i customers cijp s jkp j k depots i' customers demand D i' Dip 25 © Teodor Gabriel Crainic 2006 Oi supply i customers xijp yj wjkp j k depots xki ' p i' customers demand D i' (flows of empty containers) 26 © Teodor Gabriel Crainic 2006 Location with Balancing Formulation Minimise Z = ∑ f j y j j ∈D + ∑ {∑ ∑ (cijp xijp + c jip x jip ) + ∑ ∑ s jkp w jkp} p∈P i ∈C j ∈D Subject to j ∈D k ∈D xijp = Oip ∑ j ∈D i ∈C, p ∈ P x jip = Dip ∑ j ∈D i ∈C, p ∈ P [Demand / Flow conservation] 27 © Teodor Gabriel Crainic 2006 Location with Balancing Formulation (2) xijp ≤ Oip y j i ∈ C, j ∈ D, p ∈ P x jip ≤ Dip y j i ∈ C, j ∈ D, p ∈ P [Linking / Feasibility] ∑x i ∈C ijp + ∑ wkjp − ∑ x jip − ∑ w jkp = 0 k ∈D i ∈C j ∈ D, p ∈ P k ∈D xijp , x jip ≥ 0 [Balancing] i ∈ C, j ∈ D, p ∈ P w jkp ≥ 0 j ∈ D, k ∈ D, p ∈ P y j ∈{0,1} j ∈D 28 © Teodor Gabriel Crainic 2006 Plan What are we talking about? à Brief overview of freight transportation à Container-based intermodal transportation à System design (location) à Fleet Management (empties) à Perspectives à 29 © Teodor Gabriel Crainic 2006 Operational Planning Resource management Ó Crews Ó Vehicles Ó Power (engines, etc.), and so on à Allocation – dispatching, schedules Ó Make sure the required resources are where they need to be when they need to be there Ó Be efficient ! Ó (Satisfy demand, achieve economic and service objectives, implement plan, obey laws, policies, …) à 30 © Teodor Gabriel Crainic 2006 Issues Trade is unbalanced à Moving goods results in unbalanced distribution of resources: crews, vehicles, etc. à One needs to reposition resources for use in the following periods Ó Regular operations (if possible) Ó Balancing operations (vehicles, power units, …) Ó Crews travelling as passengers à The demand in following periods is a forecast à 31 © Teodor Gabriel Crainic 2006 Other Operational Issues Real-time dispatch à Pacing à Real-time routing adjustment à… à 32 © Teodor Gabriel Crainic 2006 Consolidation Transportation Transportation plan “guides” operations à It includes guidelines for repositioning (it should …) à “Indicative” schedules: Ad-hoc (real-time) procedures à “Regular” demand planning: Short-term and real-time adjustment of plans à Scheduled operations: Repositioning must follow and “feed” schedules + real-time adjustment à Well-defined crew (personnel) schedules à 33 © Teodor Gabriel Crainic 2006 Customized Transportation No plan !! à Dynamic management and control of resources: routes, schedules, fleets, personnel, etc. à Uncertainty plays important role Ó Demand Ó Travel times Ó Service times at customers and terminals Ó Weather, … à 34 © Teodor Gabriel Crainic 2006 The Empty Vehicle Repositioning Problem Surpluses and deficits of empty vehicles Ó Observed at terminals “at the end” of the day Ó Computed with respect to the next period demand à Need to reposition for the next period Ó How many vehicles (of what type) to move from a surplus location (origin) to a deficit location (destination)? à Much more decision freedom than in loaded transportation à A cost activity with “no” profit à 35 © Teodor Gabriel Crainic 2006 History à Transportation model – static and deterministic Ó Known surpluses and deficits – No uncertainties Ó No (not important) travel time impact – Static Ó Arrival times known (certain prediction) IAll travel, loaded and empty, occurs during the same period Ó Single or fully substituable resources (vehicles) Ó For certain LTL types, tactical planning, … 36 © Teodor Gabriel Crainic 2006 History (2) à Deterministic, multi-period transshipment model Ó Different movements require different travel times Ó Vehicles become empty at different moments (customer release times) Ó Demand varies in time … Ó The dynamic characteristic of the system represented through (dynamic, time-dependent) space-time networks 37 © Teodor Gabriel Crainic 2006 Space-Time Networks Nodes: Facilities – terminals, customers, etc. – at given time periods Ó A physical point is repeated at each period & activity à Arcs: Movements in space and time Ó Independent, e.g., a truck moving by itself Ó Grouped, e.g., containers on flat cars (rail) or in a ship Ó Holding decisions (vehicles or cargo) à 38 © Teodor Gabriel Crainic 2006 Space-Time Network (Simple) Terminals Time Current period Future periods Inventory (holding) arc Repositioning arc 39 © Teodor Gabriel Crainic 2006 Space-Time Network Terminals Time Current period Future periods Loaded movement End of horizon Inventory (holding) arc Repositioning arc 40 © Teodor Gabriel Crainic 2006 Challenges and Limitations à One may include Ó Capacities Ó Several types of resources Ó Inventory costs IStock out (rent, borrow, …) IEnd of horizon value Ó Substitutions (and costs) Ó Complex cost structures à Linear programming formulations with a few “complications” 41 © Teodor Gabriel Crainic 2006 Challenges and Limitations (2) Planning horizon length? End-of-horizon? Ó Rolling horizon à Everything is deterministic Ó Times (travel, terminal operations, customer, …) Ó Future demand, etc. à Utilization Ó Strictly scheduled railways with bookings à 42 © Teodor Gabriel Crainic 2006 The Uncertainty Factor Times may vary à Demand estimation is rarely precise à Unexpected demands and events à Current decisions impact the future state of the system and future decisions à Need to explicitly consider / model Ó Uncertainty – the stochastic nature – of the system and its environment Ó The impact of current decisions on future system state and decisions à 43 © Teodor Gabriel Crainic 2006 The Uncertainty Factor (2) Stochastic formulations and solution methods à A complex field: General approaches and, often, custom-designed methods à Active research field à Formulations Ó General stochastic programming and solution methods : à IFew efficient applications to transport problems 44 © Teodor Gabriel Crainic 2006 The Uncertainty Factor (3) à Formulations Ó Recourse formulations and rolling horizon methods INice application to regular-type systems (e.g., consolidation) Ó Stochastic formulation and solution strategies based on adaptive dynamic (linear) programming and decision/time-based decomposition ITime-Space multicommodity networks IRecent developments with interesting results 45 © Teodor Gabriel Crainic 2006 The Uncertainty Factor (4) Challenges of stochastic formulations Ó Problem formulation (!!) Ó Resolution (!!) à Plus Ó Representation of resources and attributes Ó Forecasts Ó Availability and reliability of data Ó Validation of models and strategies Ó Implementation and follow up à 46 © Teodor Gabriel Crainic 2006 Container (Empty) Fleet Management Major repositioning decisions over large geographical regions (e.g., inter-continental movements) Ó Similar to consolidation transportation à Allocation of empty containers to customers Ó Similar to customized transportation à Two applications in this talk Ó Allocation and management of a heterogeneous fleet of containers over a land zone Ó Single-commodity dynamic container allocation for liner operators (regular ocean navigation lines) à 47 © Teodor Gabriel Crainic 2006 Heterogeneous Fleet Given region (continent) à Loaded containers arrive (e.g., maritime network) to be delivered to customers à Empty containers arrive or leave to balance system-wide operations (demand) à Customers empty containers that must be moved out à Customers require empty containers for future loaded shipments à One must manage the fleet of containers for maximum profit, while satisfying demand à 48 © Teodor Gabriel Crainic 2006 Heterogeneous Fleet (2) Several types of containers (e.g., 20 or 40 feet, normal box, thermal, refrigerated, etc.) à Substitutions allowed: conditions and costs à “Massive” inter-depot balancing movements à Due-dates at some terminals (e.g., ship schedules) à Time windows at customers à Demand (at least part of) fluctuates in time and is forecasted only à Unloading time at customer: Uncertain à Travel times may be uncertain as well à 49 © Teodor Gabriel Crainic 2006 Heterogeneous Fleet (3) Containers may be damaged partially (repairs) or totally à External sources (buy, rent) of empty containers à Centralized empty container fleet management à Loaded movements not “managed” à Associated problem: global management of the empty & loaded container movements together with vehicle routes à A single model not computationally feasible ⇒ hierarchical DSS à 50 © Teodor Gabriel Crainic 2006 Main Movements (No Time/Container Type Specifics) Demand customer Depot j Harbour Supply customer External pool of empty containers Depot k 51 © Teodor Gabriel Crainic 2006 Formulations Crainic, Dejax, Gendreau (93) à Single and multicommodity deterministic formulations à A two-stage, restricted recourse single commodity, stochastic model à Multicommodity stochastic formulations may be defined à 52 © Teodor Gabriel Crainic 2006 Formulations (2) Space-time diagram Ó Generalized network (substitutions) Ó Multiple-period transportation arcs Ó Holding arcs (depots) Ó Inter-depot balancing arcs à Stochastic elements Ó Demand (of known and possible customers) Ó Release time from supply customers ⇒ Inventory levels ⇒ Import and export (border points, external pool) à 53 © Teodor Gabriel Crainic 2006 Formulations (3) Minimize total cost over the time horizon à Flow conservation (over time and space, including access to external pool) Ó Supply at supply customers Ó Demand Ó Container substitution à Depot (and port) inventories (each container type) à Bounds on inter-depot balancing flows à “End-of-horizon” restrictions (e.g., limits on “final” inventories) à 54 © Teodor Gabriel Crainic 2006 Single Commodity Liner Cheung and Chen 1998 à A container liner company offers regular service among a number of ports à Carries loaded and, space permitting, empty containers à Ship schedules known and fixed à 1 ship / period between 2 ports à Demand less than ship capacity à 1 container type à 55 © Teodor Gabriel Crainic 2006 Formulation Two-stage stochastic model Ó Time-space network with random arc capacities Ó Minimize the (expected) total cost Ó Rolling-horizon mode à Sources of randomness Ó Ship residual capacity for taking empty containers (given port and time period) Ó Demand for containers at each port/time Ó Supply of containers at each port /time before unloading from ships à 56 © Teodor Gabriel Crainic 2006 Formulation (2) Minimize total expected (cost – revenue from demand) à Ship container conservation: containers unloaded à Ship container conservation: containers loaded for repositioning à Port container conservation / demand satisfaction à Ship residual capacity for repositioning à Number of leased containers à 57 © Teodor Gabriel Crainic 2006 Plan What are we talking about? à Brief overview of freight transportation à Container-based intermodal transportation à System design (location) à Fleet Management (empties) à Perspectives à 58 © Teodor Gabriel Crainic 2006 Perspectives Intermodal transportation Ó Growing steadily & should continue to grow Ó Containers and other modes à Profound modifications to the economic, regulatory, technological, social and political environment of industry à Globalization, automation, ITS, e-logistics, security, … à Need for innovative and enhanced planning and management procedures à Opportunities for Operations Research and Transportation Science à 59 © Teodor Gabriel Crainic 2006 Perspectives (2) A number of research efforts and important results à Much more work is needed! à Many issues application areas not/little addressed à Industry evolution ⇒ New problems à à Ports & terminals Ó Planning (all levels) Ó Integration of operations & equipment types Ó Automation 60 © Teodor Gabriel Crainic 2006 Perspectives (3) à Carrier strategic & tactical planning Ó More studied than terminals, but Ó Better representation/integration of “local” operations and characteristics Ó Integration of employee scheduling impacts/relations Ó Better representation of time dependencies Ó Better integration of stochastic aspects into long-term planning models Ó Algorithmic developments 61 © Teodor Gabriel Crainic 2006 Perspectives (4) à Short-term planning Ó Time-dependent, stochastic formulations and algorithms Ó Integrated models, e.g., IContainer fleet management over land and sea IVehicles, power, crews, … à Modelling of ITS and e-logistics and integration to planning and management models; e.g., Ó Flow of information Ó Impact on uncertainty 62 © Teodor Gabriel Crainic 2006 Perspectives (5) Modelling the impact of security measures and addressing the new issues à Logistic networks Ó Coordination & synchronization Ó Information flows and uncertainties à Methodology Ó Models Ó Methods exact and (meta) heuristic Ó Parallel computation à 63 © Teodor Gabriel Crainic 2006