WO2006099387A2

WO2006099387A2 - A system and method for railyard planning

Info

Publication number: WO2006099387A2
Application number: PCT/US2006/009062
Authority: WO
Inventors: Mitchell Scott Wills; Srinivas Bollapragada; Joanne Margaret Maceo; Prescott H. Logan
Original assignee: General Electric Company
Priority date: 2005-03-14
Filing date: 2006-03-14
Publication date: 2006-09-21
Also published as: MX2007010675A; RU2007138025A; US7813846B2; WO2006099387A3; BRPI0608008A2; ZA200708324B; CN101535942A; EP1861302A4; RU2430845C2; EP1861302A2; US20070005200A1; CA2599780A1

Abstract

A system and method for generating a computer-aided rail yard plan for the movement of plural railcars through a rail yard, the rail yard having tracks that are used as receiving tracks, classification tracks and departure tracks and the plural railcars arriving on inbound trains and departing on outbound trains, the plural railcars required to undergo a sequence of tasks to move through the rail yard, the yard plan providing (a) a schedule for receiving trains, (b) a schedule for inspecting cars on receiving tracks, (c) a schedule for humping cars from receiving tracks into classification tracks, (d) a schedule for pulling trains onto departure tracks, (e) a schedule for testing trains, and (f) a schedule for departing trains.

Description

[0001] This application claims the benefit of U.S. Provisional Application No.

60/661 ,008, filed March 14, 2005, the disclosure of which is hereby incorporated by

reference.

[0002] The present disclosure is directed to the planning of operations for a rail

yard in order to optimize the performance of the rail yard and to interface the automated

planning of the rail yard with the automated planning of the line of road.

[0003] In North America, the main competitor against the rail industry is the

trucking industry. The most significant hurdles for the rail industry in capturing more of

the North Atlantic market are reducing transit time and reducing transit time variability.

Rail yard operations are central to any effort to reduce transit time and transit time

variability. Rail yards account for upwards of fifty percent of total car transit time and

transit time variation. Typically, thirty five to fifty percent of all carloads endure one or

more yard-based switch events per trip. For the remaining carloads, mainline fluidity is

contingent upon yards receiving and departing trains as scheduled. As a result, on-time

train departure performance is approximately forty to eighty percent and car connection

performance is approximately thirty to seventy percent. These levels of performance

typically result from a lack of coordination among yard activities. Poor planning is

endemic in the yard because of the inherent complexity of the equation that the planner is

attempting to solve in order to perfectly synchronize the operation. Because of his

limitations, the planner typically reaches a sub optimal solution, which results in poor

utilization of yard resources and ultimately underperformance (relative to some

theoretical capability). The nature of yard operations, i.e. a highly variable inflow and the occurrence of catastrophic events, makes planning more difficult some days than

others. Also, there is significant variability in each yardmaster'"s ability to solve the

planning equation.

[0004] A rail yard consists of a number of sub yards with each sub yard designed

to perform specific tasks. Before a train enters a rail yard, the train is typically under the

control of a network movement plan generated by a line-of-road planner and executed by

a dispatcher. As the train enters the rail yard, the responsibility for the movement of the

train is passed from the dispatcher to rail yard personal. The rail yard personal will

control the movement of the train pursuant to a rail yard movement plan. The rail yard

movement plan is different than the line of road movement plan in that the line of road

movement plan considers a train as a single entity and plans the use of resources to move

the train without conflict through the rail network. In the rail yard, the train consist will

be divided into individual cars and thus the rail yard movement plan must account for the

individual movement of each of the cars and locomotive until a reconstituted train having

different cars is released from the rail yard to the line of road movement planner.

Typically, the movement plan for the rail yard had been generated manually to take into

account the various services and resources that are required to process the incoming cars.

[0005] One typical configuration of a rail yard includes a receiving yard for

receiving a train from a network of tracks. The receiving yard includes one or more sets

of track to receive a train from the line of road tracks and permit rail yard personal to

inspect the train. The locomotives are detached from the railcars and further inspection

and maintenance is accomplished. Rail cars are then moved form the receiving yard to classification tracks. The railcars are classified in blocks of common destination. The

classification yard can be either a flat-switched classification yard (requiring a motive

force) or a hump yard. The hump yard typically includes a hill which feeds into a set of

classification tracks to allow individual rail cars to be gravity fed to the appropriate

classification track as a function of the destination of the railcar. Cars having a common

destination are fed to a common track. A series of switches down stream of the hump

control the track to which the car is routed. Once the railcars are classified in blocks,

they are moved as blocks to the departure yard. The departure yard master directs each

block to a departure track based on its subsequent destinations. At the departure yard the

cars are inspected and the train consist is brake tested and powered up and prepared for

release to the network of line of road track under control of the dispatcher. Although

larger yards may have dedicated tracks used for receiving, classifying and departing

railcars and trains, some yards use common tracks to perform the required tasks and do

not have tracks dedicated to a specific purpose, e.g., common tracks are used for

receiving and classifying.

[0006] Typically, the scheduling of train movement in the yard is largely a manual

effort including (a) estimating train arrival time by conferencing with line-of-road

operations management officials, (b) negotiating between line-of-road and yard officials

about the time at which each train will be accepted by the yard, (c) allocating a set of

receiving tracks to an inbound train based on intuition and static business rules

communicated by word of mouth, (d) assigning workers to inbound car inspection tasks,

reporting completion of inspection tasks, and requesting new assignments by physically reporting to the responsible yard manager, in-person, or by radio, (e) selecting a track or

tracks to combine and hump, (f) communicating humping tasks to the hump engine crew

in-person, or via radio, (g) coupling and pulling selected cars to the hump approach lead,

(h) shoving selected cars over the hump at a prescribed rate, (i) planning trim and pull¬

down operations to move the classified car blocks from their classification tracks to the

departure tracks in preparation for departure, (j) manually communicating trim and pull¬

down assignments to switch engine crews, in-person or via radio, (k) reporting

completion of trim and pull-down assignments, in-person or via radio, (1) scheduling

power and crew assignments to each outbound train, (m) assigning workers to outbound

car inspection and departure preparation tasks, reporting completion of inspection tasks,

and requesting new assignments by physically reporting to the responsible yard manager,

in-person, or by radio, and (n) adjusting departure time estimates based on reported,

estimated and/or actual resource availability times (e.g. crew and engine), and task

completion times. Because many of these tasks are performed by yard personnel who

report to the yard master only upon completion of their assigned task, a common problem

is the excessive dwell time of the rail cars while waiting for the required tasks of

inspecting and servicing to be completed by yard personnel.

[0007] The present application is directed to an automated yard planner which

automated many of the above tasks resulting in a yard plan that maximizes the yard's

business objectives such as minimizing dwell time, maximizing throughput, minimizing

costs, etc in order to optimize the yard's performance that was not previously available. Brief Description of the Drawings

[0008] Figure l is a simplified pictorial representation of a railway control and

management hierarchy.

[0009] Figure 2 is a simplified flow chart of one embodiment of a yard planner

according to the present disclosure.

Detailed Description

[0010] With reference to Figure 1, a typical hierarchical rail system planning

architecture 100 is illustrated. At the top level of the hierarchy on a typical railroad,

resource planning 110 may occur annually, addressing for example what trackage should

be added or retired, what yards to operate, and how many new locomotives will be

needed over the course of the year. Strategic planning occurs every few years,

determining the network blocking plan and train schedules that will operate, to

accommodate expected demand, with weekly and seasonal variations based on the

service design 120 and the maintenance planning 130. Daily, car-trips are planned to

satisfy individual customer orders 140, locomotives are assigned to individual trains in

accordance with a locomotive movement plan balancing the flow of locomotives into and

out of the locations where they will be required 150 and crews are assigned in accordance

with pool rotation and labor agreements 160. Minute by minute, line-of-road 170 and

yard operations 180 plan and execute the train and car movements and myriad support

functions to realize the network operating goals.

[0011] Today, at all levels in the hierarchy, the degree of automation varies by

location. This is particularly true of higher-level car trip planning functions, which are often performed manually with the aid of various offline simulation tools, and in the yard

where little or no computer aided operations management is available.

[0012] The present application is directed to providing computer-aided operations

management to the yard planning process. The yard planner may be a distributed agent,

responsible for a single yard, or may be responsible for multiple yards. The yard

planner's design may facilitate manual input as well as automated message input of car

connection goals, and car value functions as well as yard production status and resource

constraints.

[0013] Planning car movements in the yard through the present disclosure requires

an awareness of the current and planned future state of the status of the yard resources

including (a) yard inventory for the receiving, classification and departing yards; (b) the

current state and immediate operating plans for each yard function including receiving

inspection, car classification, switching movements, coupling/spacing cars, building

outbound trains, departure inspection, air testing and hostling; (c) the current and

available yard crews allocable to each function; (d) the current and available locomotives

and crews allocable to each function; (e) scheduled outbound train departures; (f)

available road power and rested road crews; (g) scheduled inbound trains; (h) local

industry service request including pickups, setouts and spotting cars; and (i) expedited car

blocks and must make connections. In addition, in one embodiment of the present

disclosure, the yard planning system is aware of the line of road situation and

automatically receives information relating to congestion, blockages, delayed departure

requests and early/late arrival plans. The yard planner may also provide feedback to the line of road planner regarding yard fullness vs. capacity, desired inbound spacing, desired

inbound train arrival track, predicted ready for departure time updates and planned

outbound train departure lead. The yard planner may also take into account customer

service operating metrics such as on-time arrival performance of loads and empties at

customer dock, on-time pickup of loads and empties from customer dock. In one

embodiment, the yard planner provides customer service progress visibility including

real-time web based status updates other customer.

[0014] In one embodiment of the present invention, the yard planner generates a

yard-level car movement plan that provides a detailed schedule, including time and

resources, for the movement and processing of each car through the yard during the

planning horizon. The yard plan is optimized according to a set of business objectives

that satisfy the network-operating plan while maximizing efficiency of yard-level

operations.

[0015] Business objectives can be initially taken as car connection performance

plus on-time delivery performance extrapolated to the customer's dock, minus operating

cost. Extrapolation involves estimating the duration of the remaining journey to each

car's destination, adding this to the planned departure time from the yard, and evaluating

the net impact to on-time delivery of the car (loaded or empty) to the customer's location.

In the absence of extrapolation data, delivery performance can be approximated by on-

time departure performance.

[0016] Operating cost may be the sum of the relative cost of each planned car

processing task. For example, the cost of pulling the lead car(s) from a track may be assigned a relative cost of one (1), while the cost of pulling a buried (cut of) car(s) may

have a relative cost of three (3), representing the cost to remove the cars obstructing the

desired car(s), then pull the desired car(s), and finally replace the formerly obstructing

cars.

[0017] Alternate objectives for consideration include maximum yard throughput,

yard resource utilization, terminal car dwell and robustness (tolerance and recovery for

yard anomalies). Inputs for each car include estimated arrival time, position within

inbound train, scheduled departure time, scheduled outbound train, connection value

function, on-time performance value function, final customer, positional constraints in

the outbound train (desired standing order based on destination and/or train-building

rules), and bad order status.

[0018] Detailed knowledge of the processing constraints associated with the yard

plant can be represented in a yard planning database, including current car inventory

(number, location, car block distribution), production status, historical processing rate

and capacity by process step, congestion-dependent and, time-dependent variation of yard

process performance, available yard crews by type, available switch locomotives.

[0019] Maintenance schedules and fluidity of outbound routes are also a

consideration. Maintenance schedules can be input as diminished resource availability

over the affected time. Outbound route congestion can be represented via a peer-to-peer

message interface to the line-of-road planning system. Congestion metrics are to be

determined, possibly including nominal delay by train group and as an aggregate, train

density (trains per mile), and planned track blockages. [0020] Car trip plans are a primary input to the yard planner. Car trip plans

include the origin and destination, as well as a list of yards the car will visit and the

specific train (SCAC, Symbol, Section, and Train Origin Date) the car is planned to ride

on each leg of its journey. As the car trip plans are constructed, target arrival times and

car value can be specified in terms of connection performance and on-time arrival

performance. These connection goals and car value functions guide the yard planner to

optimize execution of the network transportation plan, while maximizing efficiency of

yard-level operations.

[0021] Connection performance is a function of the rolled-throughput yield of the

yard plant model. For a given car at any point in time, the probability of making a

connection with an outbound train is described by a connection-success probability

distribution as a function of the time remaining until train departure. The variance of this

process is proportional to the sum of the variances of each process step in the car's

journey through the yard. The rolled- throughput yield may be referred to as will be

referred to as P _MAK_E, or the probability of making a connection as a function of the time

remaining until train departure. This process characterization will be used to identify

candidate cars for each outbound train.

[0022] Detailed car movements and car processing activities within the yard are

planned so as to maximize connection goals, subject to available resources (yard crews,

yard track, and yard engines). As congestion increases or as anomalies occur, not all

connections will be achievable. Car value functions express the "dynamic priority" of each car block, allowing the yard planner to evaluate the relative cost of delaying or

advancing each car.

Total Plan Value = a x (Sum of Connection Value Function)

+ b x (Sum of On-Time Arrival Value Function)

- c x (Sum of Relative Operating Cost)

[0023] Conceptually, configuring coefficients "a", "b" and "c" has the effect of

weighting the optimization more toward connection performance or toward on-time

delivery, or toward operating cost, or as a balance between the three. Connection

performance scores a car-specific constant value if its connection is made and a separate

(possibly zero) value if the connection is missed. On-Time performance is a piecewise

linear function of the extrapolated estimate of arrival time at the customer's dock,

subtracted from the promised delivery time. For simplicity, on-time delivery may be

approximated as on-time departure, until such time that down line transit time and

congestion metrics are available. Processing cost is a relative measure of the cost of

operations prescribed for a car. For example, pulling a buried car from within a cut of

cars standing on a track might be three times the cost of pulling the lead car(s) from the

same track (assuming the obstructing cars must be removed, the desired cars pulled, and

the formerly obstructing cars replaced). If the plan can arrange to place cars on the

classification tracks in the order that they will be needed, overall cost is reduced.

[0024] The sequence of activities performed on each car as it passes through the

yard are represented as a set of resource reservation and dependency rules, allowing configuration of the general flow of yard plant operations, as well as specific track (and

other resource) reservations required in each step. Site-specific business rules and

unusual dependencies among required resources can be configured as a part of the rule

base.

[0026] Rules for building individual train groups, by car block, class-of-service

and destination can be configured to characterize the strategies unique to a specific

terminal or yard. Car blocks are generally organized in a standing order that places

nearest the engine those cars to be set out first. Other train make-up constraints will be

configurable by individual yard (e.g. required positioning within the train consist for

hazardous material and key train cars, long and short cars, loaded and empty cars, speed-

restricted cars, excess dimension cars, expedited car blocks)

[0027] The resulting yard plan provides a detailed schedule, including time and

departure tracks, for the movement of each car through the yard during a predetermined

planning horizon. Subordinate resource planners translate the plan in to measurable tasks

assigned to individuals, while monitoring their progress in a periodic closed-loop

planning cycle.

[0028] The process of moving cars through the yard can be modeled as a sequence

of activities requiring exclusive use of particular, limited resources. The plan consists of

the sequence of car movement operations and the resource reservations necessary to

accomplish them.

[0029] The yard planner generates a yard plan covering a period known as the

planning horizon. At regular intervals, a new yard plan is generated to account for schedule deviations, yard processing rate changes, and extra trains. The new yard plan

once again extends into the future according to the planning horizon; a concept known as

a rolling horizon. A new yard plan is based on the state of the yard at the time that a new

plan is initiated. This state includes projected arrival and departure schedules, yard

resource levels, the current car inventory (including number, location, and block

designations), and track geometry. The yard planner will adjust the resulting yard plan to

assure that it is compatible with the current state before it is presented as the

recommended yard plan. A yardmaster, yard manager or other authorized user may can

review, revise or reject a recommended yard plan before it becomes the operational plan.

In one embodiment, the general flow of the planning process begins with an enumeration

of the outbound trains scheduled to depart the yard in the planning horizon. For each

outbound train, candidate cars are identified as those cars in blocks assigned to that train.

Each car has a business objective value associated with it. The business objective value

may take into account (a) satisfying outbound train make-up rules, such as length, weight,

standing order, and (b) satisfying yard operating rules, such as static/dynamic track

assignments, (c) satisfying available resources, (d) optimizing business objectives, and

(e) minimizing yard operating cost such as by minimizing the number and cost of moves

by each yard resource. The respective business objective value for each car can be

weighted as a function of the probability of the car making the target connection of an

outbound train. Alternately, candidate cars for an outbound train can be selected as a

function of the probability of the making the target outbound train exceeding some

predetermined value, i.e., 50% or greater. For each candidate car, the resource requirements are identified and a sequence of tasks necessary to place car in the outbound

consist. Next the sequence of resources and tasks for a subset of the candidate cars are

scheduled. Finally specific resources are assigned to each task.

[0030] The probability that a car will make its target outbound connection may be

determined using a number of methods including (a) by evaluating the historical

performance of the railcars to make a connection, (b) by using a time based modulation

function, i.e., one that considers one of time of day and time of season, and (c) by using a

load based modulation function.

[0031] In another embodiment, metric evaluation and analysis is used at each

sequential process as a car traverses the yard. Figure 2 illustrates a simplified, sequential

process of creating a yard plan during one periodic cycle. The yard is assumed quiescent,

with some cars in each sub-yard at various phases of yard processing. Simple rules

mimicking human decision-making processes are used to determine what cars should be

moved at each step. True optimization is not achieved, but a favorable result can be

achieved for normal operations. When a feasible plan cannot be found (usually due to an

incomplete rule base) human intervention is solicited, highlighting the unresolved

conflicts, to draw his attention to where it is needed.

[0032] Based on the current yard state the following steps are performed.

Incoming trains are received on the tracks where they should be received 200. The trains

are inspected on the receiving tracks 210. The locomotives are decoupled and the rail

cars, individually or as blocks, are humped to the classification tracks 220. As the cars

are assembled on the classification tracks, blocks of cars are moved to the departure tracks 230. At the departure tracks the train blocks are inspected 240 and locomotives are

added. If the train is too long to fit in a single departure track, the train must double-out,

a final brake test is performed and the trains depart 250.

[0033] The step of receiving the incoming trains 200 includes computing the

available capacity of the receiving yard. In one embodiment each train that is scheduled

to be received within a predetermined period of time is evaluated to identify the train

with the earliest arrival tie that fits in the receiving yard. Once a train is identified, the

receiving yard is searched to determine the best receiving tracks to receive the train. In

one embodiment, the goal may be to receive the train on as few as tracks as possible. If

the train requires more than one receiving track, the tracks should be chosen to be as

close to one another as possible.

[0034] The step of inspecting the cars on the receiving track 210 includes

evaluating the cars on each receiving track to determine a measure of importance for each

car. A measure of importance can be any metric which is used to determine the relative

importance of the rail cars and can include consideration of qualitative or quantitative

factors. For example, the measure of importance may be based on priority of the car, or

the probability of the car to make a connection. In one embodiment, the measure of

importance is determined as a function of the consist of the received train. For example,

a consist containing a car that has high priority and incurs large penalties if delayed

would greatly influence the importance of the cars on that track. In another embodiment,

the measure of importance is determined as a function of the scheduled departure times of

the outbound trains containing the cars on the receiving track. In yet another embodiment, the measure of importance is determined as a function of the minimum

amount of time needed for the cars to make an outbound connection. Once the measure

of importance for each car is determined, the receiving track having the highest

cumulative measure of importance for all the cars on its track is inspected first. Planned

inbound train consist is viewed as expected future car inventory. Profiled inspection and

inbound processing times are budgeted to anticipate when the cars will be available to be

classified.

[0035] The step of humping a receiving track 220 includes determining a measure

of importance for each car on all receiving tracks that have already been inspected. The

measure of importance can be determined by evaluating the same factors as discussed

above, i.e., consist, departure time or time needed to make outbound connection, or it

may consider other factors such as car dwell time. Once the measure of importance of

each car that has been inspected is determined, the tracks containing the cars are sorted in

decreasing order of cumulative importance. In one embodiment, the classification tracks

are evaluated to identify which classification track is available to receive the cars from

the highest sorted receiving track while maintaining pure blocks of cars.

[0036] The classification tracks of the cars of the highest sorted track are then

identified and cars are humped to their respective classification track. A goal is to ensure

that blocks for an outbound train are in the same class track group. In one embodiment,

for each car on the receiving track being considered, if there is space available on a

classification track where the last cars belongs to the same block as the car under

consideration, the car is humped to that classification track. If there exists a classification track with cars that on the same outbound train as the car under consideration, and no

ordering constraint exists for the blocks on that track or the order constraint is satisfied

and there are no other cars that are the same block to be humped in the near future as the

last car on the classification track, then the classification track is selected. If none of the

above criteria is satisfied, an empty classification track is selected for the car under

consideration. If no classification tracks are available, then next highest train in the

sorted order is selected.

[0037] A classification track 220 is assigned to each destination serviced by the

selected train for a period of time prior to scheduled departure, to allow classification of

car blocks that will be assigned to the train. Once a classification track is assigned to a

destination, the assignment will remain either indefinitely, if a fixed class allocation

strategy is in effect, or until no more cars for that destination remain in the yard

inventory. Advanced strategies may also be used that anticipate additional inbound cars

and retain the allocation longer, dynamically choose to release that allocation while

associated cars remain in the receiving yard, allocate multiple destinations to a single

classification track, e.g. geometric switching, where cuts of cars are repeatedly classified

and reclassified to build multiple pure blocks on each track, or building a train in a

classification track - satisfying the required standing order of an outbound train as the

cars enter the bowl.

[0038] Slough tracks may be dynamically or statically allocated in more simplistic

switching strategies, where the slough track receives all unassigned car destinations. A

RIP (repair-in-place) track is assigned to all bad-ordered cars. [0039] The step of pulling a car or group of cars from the classification track to the

departure tracks 230 includes determining a measure of importance for each car that is

ready to be pulled from the classification tracks. In one embodiment, the measure of

importance is determined as a function of the time until departure of the outbound train.

In another embodiment, the measure of importance is determined as a function of the

order constraint of the outbound train. In yet another embodiment, the measure of

importance is determined as a function of the minimum amount of time needed for the

car to make an outbound connection. Once the measure of importance for each car is

determined, the cumulative measure of classification track having the highest measure of

importance is pulled to the departure yard first.

[0040] For each block of cars on the classification track, a departure track is

selected. In one embodiment, if there is space available on a departure track where the

first car belongs to the same block as the block under consideration, that departure track

is selected. If not, then any empty departure track can be selected. If no empty departure

tracks are available, and if there exists a departure track with blocks o cars that are on the

same outbound train as the current block and there are no other cars that are of the same

block to be pulled in the near future as the first block on the departure track. Otherwise,

if there is no space in the departure yard, the next highest sorted classification track is

selected.

[0041] Each consist assembled on the departure track must be inspected prior to

departure 240. The departure trains may be sorted in increasing order of scheduled time.

Beginning with the highest sorted train the amount of time need to test the train can be computed. The slack time between the expected completion of testing and the scheduled

departure time can be computed. If the slack time is less than some predetermined value,

a departure track that has the least amount of track space left to be tested is selected for

the train.

[0042] The trains then may be departed at their scheduled time to the line of road

when the testing is completed 250. If a departure lead is available, and there are trains

for which scheduled departure times have passed, then the train that is most late can be

selected and if all its departure tracks are tested and the length of the train on the

departure tracks is greater than a predetermined minimum length, the train can be

departed.

[0043] The steps identified above results in the computer-aided generation of a

yard plan that provides (a) a schedule for receiving trains, (b) a schedule for inspecting

cars on receiving tracks, (c) a schedule for humping cars from receiving tracks into

classification tracks, (d) a schedule for pulling trains onto departure tracks, (e) a schedule

for testing trains, and (f) a schedule for departing trains. The steps may be implemented

in computer readable program code modules embodied in a computer usable medium for

use by a general purpose or special purpose computer.

[0044] For each train in the planning horizon, the earliest scheduled outbound train

is assigned a departure track for a sufficient period of time prior to scheduled departure,

to allow building, inspecting and brake testing the train before departure. The assigned

departure track and planned departure time is communicated to the line of road via a yard

update message. When operating exceptions result in the modification of either the assigned departure track or planned departure time, additional yard update messages are

sent.

[0045] Subsequent outbound trains are processed in order of scheduled departure

from the yard. When a conflict occurs between two outbound trains requiring a common

resource to meet their scheduled departure window, the resource allocation is adjusted to

optimize the yard objective function. An alert is raised to the authorized user indicating

the type of conflict and the result.

[0046] In another embodiment receiving yard allocation is managed by

allocating track or tracks to each inbound train based on a set of decision rules. These

decision rules may include receiving trains in the order of expected arrival, receiving

trains in the order their cars are required for outbound trains, receiving trains based on

their length and available room in the receiving yard, receiving trains based on outbound

locomotive requirements, receiving trains based on freight priority, receiving trains based

on their crews' hours of service.

[0047] A yard update message is sent to line of road dispatch system indicating the

desired inbound lead for each train planned to be received from the mainline for some

predetermined horizon. When receiving yard congestion is detected, an alert is raised to

the authorized user. If an automatic overflow strategy is employed, a yard update

message will indicate to the line of road planner the desired holding facility to which the

affected train(s) must be routed. Peer-to-peer messages can be exchanged between yard

planners at opposite ends of a line-of-road indicating congestion and the projected time at

which the yard can accept more trains. [0048] In order to avoid the need for backtracking in the search for optimal car

movement choices, the planner must avoid several resource reservation pitfalls, such as

over-subscribing classification tracks. If the number of class tracks is limited and a class

track is reserved to a car destination that will not be needed for some time, the plan can

suffer from class track starvation. Another example of a pitfall is pulling inappropriate

receiving tracks caused by selecting a receiving track with a mix of car destinations,

many of which will not be needed for some time resulting in a suboptimal allocation of

classification tracks. Another pitfall is poor departure track selection, which can lead to

allocating longer tracks to trains not requiring them.

[0049] In one embodiment, rules are utilized by the yard planner to recognize and

avoid patterns representing these pitfalls, to insure efficient planning. Rules may also be

added to recognize and recommend recovery plans, when human operators intervene and

induce such patterns.

[0050] On some occasions the yard planner may not be able to generate a yard plan

that meets all constraints. In such instances an exception is raised. Yard plan exceptions

can be of three types - resource exceptions, productivity exceptions, and schedule

exceptions. A resource exception occurs when there are insufficient resources to

accomplish the objectives. A switch crew shortage and a derailment on a switch lead are

examples of resource exceptions. A productivity exception occurs when the actual

processing rate of a given resource is less than its processing capacity. A schedule

exception occurs when the boundary constraints cannot be met. A car that arrives into the yard with less than some reasonable period of time before its scheduled departure is

an example of a schedule exception.

[0051] In the process of developing a yard plan, if the yard planner detects a

schedule exception, it evaluates the net impact of accelerating a car through the process

to make its scheduled departure versus not doing so, and instead scheduling it to the next

available train. The yard planner generates a yard plan that satisfies as many boundary

constraints as possible and notifies the Yardmaster or other authorized users of the

unresolved schedule exceptions. When a resource or productivity exception occurs, the

yard planner notifies the Yardmaster and Trainmaster of the exception. A yard plan will

not be generated until the Yardmaster or Trainmaster either resolves the exception or

directs the yard planner to revise its state of the yard to reflect the impact of the

exception.

[0052] The user interface of the yard planner may reside on a personal computer

with some informational displays available via a thin client (e.g. web-browser, hand-held

or wearable display unit). Ultimately, status and exception reporting are expected to

migrate to thin-client input devices, where fully automated process monitoring is not

feasible.

[0053] Interactive graphical displays may be useful to present the plan, to provide

decision support, and to accept direct data input, process status reporting, and plan

deviation input. Cost-effective server hardware, for example Windows XP server can

host automatic plan production, external system interface management, thin-client user

interface management and database management functions. [0054] Server hardware can be deployed in a yard office environment for small

yard installations, or in a climate controlled IT center for larger yards. The yard office

environment is subject to much greater temperature and humidity variations that the IT

center. The yard office may also subject the servers to a moderate amount of dust and

dirt, certainly more so than the IT center deployment. Protective dust and dirt

membranes may be required for keyboard and CPU.

[0055] Thus the present disclosure is directed to an automated yard planner that

interfaces with a line of road movement planner. The movement planner sends yard

update messages to the yard planner including inbound train arrival times and consist

information. The mainline update (which consists of the inbound and outbound train

lineups) is finalized and input into the system by the MTO or his designate. When the

lineup is finalized it can be read directly by the yard planner. This information is used in

conjunction with the current yard status to develop a plan for car movements for the yard.

Information can be transmitted to the control terminal of the responsible individuals.

[0056] While preferred embodiments of the present invention have been described,

it is understood that the embodiments described are illustrative only and the scope of the

invention is to be defined solely by the appended claims when accorded a full range of

equivalence, many variations and modifications naturally occurring to those of skill in the

art from a perusal hereof

Claims

What is claimed:

1. A method of planning the movement of plural railcars through a rail yard,

the rail yard having tracks that are used as receiving tracks, classification tracks and

departure tracks and the plural railcars arriving on inbound trains and departing on

outbound trains, the plural railcars required to undergo a sequence of tasks to move

through the rail yard, comprising:

(a) receiving a plurality of trains in a rail yard, each train having a plurality of

railcars received on a receiving track;

(b) for each group of railcars on a receiving track, assigning an importance value

for each railcar;

(c) for each receiving track, aggregating the importance values for the railcars; and

(d) scheduling the railcars to be moved to the classification tracks as a function of

the aggregated importance values.

2. The method of Claim 1 wherein the importance value is determined as a

function of the scheduled departure time of the railcar.

3. The method of Claim 1 wherein the importance value is determined as a

function of the minimum amount of time needed for the railcar to make an outbound

connection.

4. The method of Claim 1 wherein the importance value is determined as a

function of the probability of the railcar to make an outbound connection.

5. The method of Claim 1 wherein the step of scheduling includes sorting the

receiving tracks in decreasing order of priority as a function of the aggregated importance

values of the railcars on the respective receiving tracks.

6. The method of Claim 5 wherein the step of scheduling further includes

selecting the highest sorted receiving track for which there is capacity available on the a

classification track.

7. The method of Claim 1 further comprising selecting a classification track

for a car on the receiving track as a function of the space available on a first classification

track and the last car on the first classification track,.

8. The method of Claim 7 including selecting the first classification track if

there is space available and the last car on the first classification track belongs to the

same block as the car on the receiving track under consideration.

9. A method of planning the movement of plural railcars through a rail yard,

through the rail yard, comprising::

(a) identifying the outbound trains scheduled to depart the yard during a

predetermined planning horizon;

(b) identifying candidate railcars for each identified outbound train;

(c) determining the business objective value of each identified candidate railcar; (d) ranking the identified candidate cars in decreasing order based on business

objective values; and

(e) selecting an identified candidate railcar for inclusion in an identified outbound

train in order of the ranking of the candidate cars.

10. The method of Claim 9 where in the step of ranking includes:

(i) determining the current availability of each identified candidate railcar;

(ii) determining the probability of the candidate railcar to be ready for a departure

of the identified outbound train as a function of the determined availability; and

(iii) weighting the business objective values of the railcar as a function of the

determined probability of the railcar.

11. The method of Claim 9 further comprising :

(f) identifying a sequence of tasks to prepare the selected candidate railcar for

inclusion in an identified train; and

(g) assigning resources to perform the identified sequence of tasks as a function of

the business objective value of the selected railcar.

12. The method of Claim 11 wherein the business objective value is determined

as a function of a physical characteristic of the identified outbound train.

13. The method of Claim 12 wherein the physical characteristics is at least one of

length, weight, or order of the railcars in the train.

14. The method of Claim 11 wherein the business objective value is a function of

yard operating rules.

15. The method of Claim 11 wherein the business objective value is a function of

the yard operating costs.

16. The method of Claim 10 wherein the step of determining the probability is a

function of historical performance of railcars.

17. The method of Claim 10 wherein the probability is determined using a time

based modulation function.

18. The method of Claim 17 wherein the time based modulation function

considers one of time of day and time of season.

19. The method of Claim 10 wherein the probability is determined using a load

based modulation function.

20. The method of Claim 10 wherein the probability is determined as a function

of a moving average of the historical performance of the railcar.