Docstoc

csx_or_overview

Document Sample
csx_or_overview Powered By Docstoc
					 The OR Group and Applications of OR Models at CSX

Dharma Acharya, Kamalesh Somani and Suneil Kuthiala
               Operations Research
                 December 5, 2007
    Presentation Outline

       CSX Overview
       OR Vision
       Current OR Projects
       An Example of OR Modeling in Railroad
       Internship Program
       Questions and Answers




2
    CSX Transportation Today


                                                                                 Montreal
                                                                                                              Operates in 23 states,
                                                                                                               the District of
                                                                                    Orlando Jacksonville

                                                             Buffalo
                                                                       Syracuse
                                                                                                               Columbia and two
                                                Detroit
                          Chicago                                        Newark New York City                  Canadian provinces
                                                      Willard     Harrisburg
                                                                             Philadelphia
                                                         Pittsburgh
                                 Indianapolis Columbus Cumberland Baltimore
                                                                                                              22,000-mile rail network
                                              Cincinnati                Washington D.C.
      East St. Louis                                       Charleston
                                                    Huntington
                                                                              Norfolk
                                                                                                              36 major classification
                                       Louisville
                                                                         Rocky Mount
                                                                                                               yards and terminals
                                    Nashville Knoxville
                                                          Charlotte     Hamlet
                                                                                                              32,000 employees
                       Memphis
                                           Atlanta
                                   Birmingham
                                                                      Charleston                              3,700 locomotives
                                  Montgomery
                                                 Waycross
                                                              Savannah
                                                                                                              103,000 freight cars
                                                              Jacksonville
                                  Pensacola
                          New Orleans
                                                               Orlando
                                                     Tampa           West Palm Beach

                                                                         Miami



3
    Surface Transportation Operations

       Operations delivered more than 7.5 million carloads
       Generated annual revenue of more than $8.6 billion
       Four primary business segments:
         Merchandise delivers nearly 3 million annual carloads of
           aggregates, metals, phosphates, fertilizers, food, consumer,
           agricultural, forest and chemical products
         Coal delivers more than 1.7 million carloads of coal, coke and
           iron ore to electric utilities and manufacturers
         Automotive delivers one-third of North America’s light vehicle
           production, more than 500,000 carloads
         Intermodal accounts for over 40% market share of intermodal
           traffic carried in the eastern U.S.




4
    Rail Renaissance

     Traffic Growth
     Rising fuel costs
     High driver turnover
     Rising insurance costs
     Hours of Service changes
     Hazardous material certification
     Aging infrastructure


    Rail is the solution…



5
Operations Organization Structure
                                 EVP &
                          Chief Operating Officer


         Engineering                                 Transportation



         Mechanical                                  Service Design



       Customer Service                             Process Excellence



            Safety



6
Service Design Organization Structure

                             Service Design



     Operations Research &
                                              Integrated Planning
           Planning



        Service Planning                       Car Management




      Advanced Engineering



7
Operations Research Group at CSX
 Vision: To be a catalyst in leveraging state-of-the-art
    analytical and modeling techniques to aid CSX
    Transportation in becoming the safest and most
    progressive North America railroad.




8
Operations Research

    Departments we help
        Locomotive Management

        Crew Management

        Service Planning

        Car Management

        Engineering

        Customer Operations

        Passenger Operations

        And many others…


9
Railroad Problems

      Resource Planning and Utilization
           Locomotives
           Crews
           Cars
           Track
           Maintenance Teams
           Shops
           Yards


      Train Scheduling

      Shipment Routing

10
     Operations Research Projects & Tools

 Service Design
         Forecasting Tool
         Train Analysis Tools
         Handling Reporter
         Joint Facilities
         Gateway Optimization


 Crew Management
       Crew Planning System
       Taxi & Lodging Reporting Tool




11
Operations Research Projects & Tools (contd…)

 Locomotive Management
      Locomotive Fuel and Plan System (LFPS)

      Locomotive Scheduling and Planning Model

      Locomotive Simulation Model (LocoSim)
      Locomotive Shop Routing Model




12
     Operations Research Projects & Tools (contd…)

 Safety
       Hazmat Routing and Analysis
       Defect Detection Equipment
          Wheel Impact Load Detector (WILD)

          Optimal Locations for Supersites

       Locomotive Incident Reporting Tool (LIRT)

       Car Incident Reporting Tool (CIRT)




13
 Operations Research Projects & Tools (contd…)

  Other Support Initiatives
      Crew Qualification Tool

      Track Maintenance Teams Scheduling Model

      Flowmaps




14
Using Optimization for Railroad Problems
 Linear Programming
        Linear Programming (LP) problems are optimization problems in which the
         objective function and the constraints are all linear

        Standard form consists of the following three parts:
                A linear function to be maximized/minimized
                  Maximize         c1 x1  c2 x2     c                     x 
                                                    c 1     c          x   1
                                                              2              x2 


                Problem constraints of the following form
                        a11 x1  a12 x2  b1             a11 a12               b1 
                        a21 x1  a22 x2  b2        A  a21 a22            b  b2 
                                                                                
                         a31 x1  a32 x2  b3           a31 a32                b3 
                                                                                

                Variable bounds
                           l1  x1  u1                  l              u 
                                                     l   1         u   1
                          l2  x2  u2                    l2             u2 

        The problem is usually expressed in matrix form, and then becomes:
                    Maximize              cT x
                    Subject to           Ax  b
                                          l  xu

16
     Network flow problems
  Networks are everywhere
          Telephone network
          National highway systems
          Distribution networks
          Airline networks
          Rail networks

  We wish to move some entity from one point to another
   as efficiently as possible
        This objective is achieved by modeling the problem as
         mathematical objects known as network flow problems

  Network: Nodes, Arcs


17
     Network flow models

      Static models
         No time dimension
            Find cost effective sources of ballast for maintenance jobs
            Matching problem (source and sinks)

      For applications like scheduling of people, jobs or
       projects time is an essential ingredient
         We need to account for evolution of the underlying system over
          time
         Use time expanded network flow models
            Create a node each time you have a event starting
            Create links between nodes (links can have some time length or
             zero time length)



18
     Solving approach


       Raw Data                                               Minimize     cT x
                                    Network
       (Database,     Transform               Transform       Subject to    Ax  b
                                    design
       Spreadsheet,                                                         l  xu
       GUI)


                                                                              (c,A,b,l,u)




                                    Network                        ILOG
       Raw Data         Transform

       (Database,                   design                         CPLEX
                                                   (x, cTx)
       Spreadsheet,
       GUI)



19
Crew Scheduling
Crew Change points



     1     2         3   4   5   6   7




     1     2         3   4   5   6   7




21
Operational Definitions

          A crew consists of a conductor and an engineer
          A territory consists of, 2 crew change points

          Call time
          Off duty time
          Crew on duty



                                   train travel time
                 train departure                       train arrival

     Call time                                                     Off duty time


                                    Crew on duty


22
Operational Definitions (contd…)




      Crew Pool: Engineer or conductor pool




23
Understanding Crew Planning


               Home terminal               Away terminal
                    1

                               Train 1         2


                                                     Rest
                               Deadhead
                    7

                                               3



                                 Train 2
        Rest        4




                    5


                                Train 3        6

     Time



24
Crew Planning


                       Constraints:
                       • RR and FRA regulations
     Train Schedules   • Minimum stay at home and away      Train Assignments
                       • Maximum held away time             • Assignment of crews to
                                                              trains
                                                            • Deadheading and stay in hotel
                                                              decisions

      Crew Change
          Points
                                  Crew
       (Territories)
                                 Planning


                                                             Crew Roster
                                                             • Roster of a crew over the
                                                               planning horizon
       Crew Pools
                       Objective Function:
                       • Minimize crew costs –Hotel Cost,
                        deadhead and straight-time




25
Formulation

   Minimize:
     Hotel cost
        Lodging
        Held away
     Taxi cost
        Taxi fixed
        Crew deadhead
     Crew fixed cost




26
Formulation (contd.)

   Constraints and bounds:
     All trains have crews assigned
     RR and FRA rules (e.g.: Minimum crew rest hours)
     Union agreements
     Combination or non-combination deadhead
     Safety stock of crews




27
     Minimum Cost Flow Problem
 Given a territory (Jacksonville – Orlando) and trains running between these
cities, find the best way to utilize the crews

     Arcs
       Train arcs
       Rest or Idling arcs                          Jacksonville      Orlando
       Taxi arcs
                                          0001 Mon    1
                                                                                0745 Mon
     Jacksonville to Orlando                                             2

         Q113, dep time:
          0001 Mon, arr time:   Time
          0745 Mon
         Q115, dep time:
          1945 Fri, arr time:
          0345 Sat



                                                      3
                                         1945 Fri                         4
                                                                                0345 Sat


28
     Train Arcs
        Orlando to Jacksonville
                 Q114, dep time: 2215 Mon, arr time: 0730 Tue
                 Q116, dep time: 0700 Wed, arr time: 1620 Thu


                              Jacksonville       Orlando

                   0001 Mon    1
                                                        0745 Mon
                                                    2
     Time
                                                    5
                                                         2215 Mon
                              6
                  0730 Tue




                                                    7
                              8                            0700 Wed
                 1620 Thu



                               3
                  1945 Fri                          4
                                                        0345 Sat




29
     Taxi Arcs (contd…)
        Rule 1: combination deadhead at away location -> crew goes directly on taxi
         after train arrives

                             Jacksonville     Orlando

                  0001 Mon    1
                                                     0745 Mon
                                                 2
     Time        1045 Mon    9
                                                 5
                                                      2215 Mon
                             6
                 0730 Tue




                                                 7
                             8                          0700 Wed
                1620 Thu



                              3
                 1945 Fri                        4
                             10                      0345 Sat
                 0645 Sat




30
     Taxi arcs (contd…)
        Rule 2: deadhead crew from home (taxi travel time + crew rest time = 3 hrs
         + 10 hrs = 13 hrs) prior to the away location train departs

                             Jacksonville     Orlando

                  0001 Mon    1
                                                      0745 Mon
                             11                  2
     Time       0915 Mon
                             9
                 1045 Mon                        12
                                                       1215 Mon

                                                 5
                                                       2215 Mon
                             6
                 0730 Tue
                1800 Wed     13
                                                 14     2100 Wed


                                                 7
                             8                          0700 Wed
                1620 Thu


                              3
                 1945 Fri                        4
                             10                       0345 Sat
                 0645 Sat



31
     Idling Arcs
        Rule: minimum idling time = 10 hrs, maximum idling time = 72 hrs


                             Jacksonville     Orlando

                  0001 Mon    1
                                                      0745 Mon
                             11                  2
     Time       0915 Mon
                             9
                 1045 Mon                        12
                                                       1215 Mon

                                                 5
                                                       2215 Mon
                             6
                 0730 Tue
                1800 Wed     13
                                                 14     2100 Wed


                                                 7
                             8                          0700 Wed
                1620 Thu


                              3
                 1945 Fri                        4
                             10                       0345 Sat
                 0645 Sat



32
     Objective Function
     Minimize:
      Hotel cost
          Lodging                           c x   l l
                                          lIdlingArcs
            40 x25  3 * 40 x27  .....                         Jacksonville   Orlando
          held away                                             1
           50 * number of hrs * x27  .....          0001 Mon
                                                                                      0745 Mon
                                                                11               2
                                                   0915 Mon
                                                                9
     
                                                    1045 Mon                     12
         Taxi cost                                                                     1215 Mon

          taxi fixed                                           6
                                                                                 5
                                                                                       2215 Mon

               (2 * 200 ) x29  .....               0730 Tue

          crew deadhead                           1800 Wed     13
                                                                                 14      2100 Wed

             400 x29  .....                                                     7
                                                                8                        0700 Wed
                                                   1620 Thu


                   c x l l
               lTaxiArcs
                                                                 3
                                                    1945 Fri
                                          Time                                   4
                                                                                      0345 Sat
                                                                10
                                                    0645 Sat




33
Formulation (contd.)

        Constraints:
          Node balance constraint
                   ( x )  ( x )
                  lI [i ]
                             l
                                     lO[i ]
                                               l   i  AllNodes

          All trains have crews assigned
                   xl  1  l  TrainArcs

          Minimum crew rest hours and maximum idling time
             Covered in the network design
          Combination or non-combination deadhead
             Covered in the network design
          Safety stock of crews
                    x l
                                 l   k             l  IdlingArcsAtASnapshot

          Crews on taxi and idling arcs
                       xl  {0,1}                    l  Taxi & IdlingArcs



34
     Why CSX


     Competitive salaries
     Vast opportunity for future advancement
     High attrition rate due to retirees
     Fortune 500 company
     Affinity Groups
     Company stability




35
Internship/Co-op Program



     Internship- A full-time student enrolled at a university (or
     community college) who wants the practical experience of working
     full time during the summer break from classes.




     Cooperative Education (co-op)- A student that is
     enrolled at a university and works alternate semesters at CSX
     to fulfill a graduation requirement and receives academic
     credit.




36
Benefits to Students



      Relevant work experience
      Potential full-time opportunity
      Exposure to careers in the railroad industry
      College credit towards graduation
      Paid opportunity




37
What do we look for in potential candidates?

  Web development experience
      VB.NET / ASP.NET


  Strong database skills
      SQL Server 2000

  Eager to learn and willing work hard



  Two semester requirement



38
 How to apply for the Internship/Co-Op Program
       1. Applications and resumes accepted by e-mail
          Suneil_Kuthiala@csx.com
       2. Interviews will be conducted at CSX with Operations
          Research representatives




39
     Questions




40

				
DOCUMENT INFO
Shared By:
Categories:
Stats:
views:16
posted:3/21/2011
language:English
pages:40