SlideShare a Scribd company logo
1 of 32
Download to read offline
Robust	
  Empty	
  Reposi.oning	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  
in	
  Large-­‐Scale	
  Freight	
  Consolida.on	
  
                  Networks	
  	
  
               Alan	
  Erera1,	
  Antonio	
  Carbajal1,	
  
                       Mar.n	
  Savelsbergh2	
  
       1	
  School	
  of	
  Industrial	
  and	
  Systems	
  Engineering,	
  Georgia	
  Tech	
  	
  

       2	
  University	
  of	
  Newcastle,	
  Australia	
  




                                  Odysseus	
  2012	
  
What	
  to	
  remember	
  
1.  Robust	
  models	
  for	
  empty	
  mobile	
  resource	
  
    management	
  pragma.c	
  and	
  effec.ve	
  
2.  Empty	
  resource	
  hubs	
  useful	
  for	
  very	
  large-­‐
    scale	
  reposi.oning	
  networks	
  
3.  Rolling-­‐horizon	
  deployments	
  of	
  two-­‐stage	
  
    robust	
  op.miza.on	
  models	
  should	
  u.lize:	
  
       –  short	
  robust	
  horizons	
  
       –  rolling	
  robust	
  constraints	
  


	
  
Consolida.on	
  networks	
  
                                 Customer
                                 Satellite terminal
                                 Hub terminal
origin




                                           destination
Consolida.on	
  networks	
  
Net weekly surplus                           -3                   Satellite terminal
of empty trailers                                                 Hub terminal
                           +5
                                                  -5
                                        -1
                                                                       +2


                 +4                                          +1
                            +1               -1        +1
     +1
                                +3                                      +4
            +3
                                                  -4
-3
                                    0
                      +3                                +2
 +1                                                                     -3
                                0            -6                   +1
                 -2                                         -1
Dynamic	
  trailer	
  reposi.oning	
  
•  Large-­‐scale	
  terminal	
  network	
  
    –  250+	
  satellites	
  and	
  hubs	
  
•  Dynamics	
  
    –  Several	
  decision	
  epochs	
  daily	
  
    –  Today’s	
  projected	
  demand	
  for	
  trailers	
  accurate	
  
    –  Tomorrow’s	
  (and	
  beyond)	
  significantly	
  uncertain	
  
•  Goal	
  
    –  Best	
  empty	
  reposi.oning	
  plan	
  each	
  epoch	
  
Modeling	
  approaches	
  
•  Determinis.c	
  rolling-­‐horizon	
  network	
  flow	
  LP	
  
   –  Assume	
  that	
  trailer	
  demands	
  tomorrow	
  (and	
  
      beyond)	
  behave	
  as	
  expected	
  	
  
•  Stochas.c	
  models	
  
   –  Minimize	
  expected	
  costs	
  given	
  probabilis.c	
  model	
  
      of	
  demand	
  
   –  Powell	
  (87),	
  Frantzeskakis	
  and	
  Powell	
  (90),	
  
      Cheung	
  and	
  Powell	
  (96),	
  Godfrey	
  and	
  Powell	
  (02a,	
  
      02b)	
  	
  	
  	
  
   –  Crainic	
  (93),	
  Di	
  Francesco,	
  et	
  al.	
  (09)	
  
Modeling	
  approaches	
  
•  Robust	
  op.miza.on	
  models	
  
   –  Bertsimas	
  and	
  Sim	
  (03),	
  Atamturk	
  and	
  Zhang	
  (07)	
  
   –  Morales	
  (06),	
  Erera	
  et.	
  al.	
  (09)	
  
       •  Two-­‐stage	
  model	
  
       •  Explicit	
  focus	
  on	
  future	
  feasibility	
  
       •  Minimize	
  cost	
  of	
  planned	
  movements	
  such	
  that	
  a	
  
          feasible	
  set	
  of	
  recovery	
  movements	
  exists	
  for	
  each	
  non-­‐
          extreme	
  scenario	
  
Two-­‐stage	
  robust	
  reposi.oning	
  
      First stage decisions

      t=0        t=1          t=2   t=3   t=4   t=5
  A


  B


  C


  D


  E
Two-­‐stage	
  robust	
  reposi.oning	
  
      First stage net supply         bi
      t=0           t=1            t=2       t=3        t=4       t=5
  A

        Initial trailers
  B
                      +2
                                Known and expected future loaded moves
  C
        +6                 -2                      -1

  D


  E
                                     +2                   +1
Two-­‐stage	
  robust	
  reposi.oning	
  
      Second stage “recovery” decisions

      t=0       t=1       t=2      t=3    t=4   t=5
  A


  B


  C


  D


  E
Two-­‐stage	
  robust	
  reposi.oning	
  
      Second stage uncertain demand

      t=0       t=1      t=2        t=3        t=4         t=5
  A


  B

                      Intervals on future loaded moves   [a , a ]
  C

                                          [0, 2]
  D


  E
Two-­‐stage	
  robust	
  reposi.oning	
  
•  First	
  stage	
  network	
  flow	
  
           
   min            c a xa
           a
                              
                     xa −               x a = bi   ∀i∈N
         a∈δ + (i)          a∈δ − (i)

                     xa ≥ 0       and integer ∀ a ∈ A
•  Second	
  stage	
  “recovery	
  flow”	
  for	
  each	
  scenario	
  
                            
               wa (ω) −                wa (ω) = bi (ω) − bi   ∀i∈N
   a∈δ + (i)               a∈δ − (i)

                     xa + wa (ω) ≥ 0                          ∀a∈A
Two-­‐stage	
  robust	
  reposi.oning	
  
•  Key	
  result:	
  Existence	
  of	
  Recovery	
  Flow	
  
      
                  xa ≥ ν(U )   ∀ U is inbound-closed
   a∈δ + (U )∩I


            t=0       t=1      t=2       t=3        t=4        t=5
        A


        B


        C
Two-­‐stage	
  robust	
  reposi.oning	
  
•  Inbound	
  closed	
  set	
  
   –  node	
  set	
  with	
  no	
  incoming	
  recovery	
  transporta.on	
  
      arcs	
  

           t=0       t=1          t=2      t=3       t=4        t=5
       A


       B


       C
Two-­‐stage	
  robust	
  reposi.oning	
  
•  Inbound	
  closed	
  set	
  
   –  this	
  example	
  not	
  inbound-­‐closed	
  


           t=0       t=1          t=2      t=3         t=4   t=5
       A


       B


       C
Two-­‐stage	
  robust	
  reposi.oning	
  
•  Worst-­‐case	
  vulnerability	
  of	
  inbound-­‐closed	
  set	
  
    
                    xa ≥ ν(U )  ∀ U is inbound-closed
   a∈δ + (U )∩I                                         
                         ν(U ) =         a −        a +   bi
   	
                            a∈δ + (U )         a∈δ − (U )    i∈U

              t=0       t=1      t=2          t=3        t=4     t=5
          A
                                                    [0, 2]

          B
                                  [1, 5]
          C
Challenges	
  
(1) Smart	
  recovery	
  network	
  
       –  Low-­‐cost	
  moves	
  (since	
  costs	
  not	
  modeled)	
  
       –  Opera.onally	
  simple	
  
(2) Appropriate	
  use	
  of	
  two-­‐stage	
  model	
  
       –  Controlling	
  conserva.sm	
  pragma.cally	
  
       –  Special	
  considera.ons	
  for	
  rolling	
  horizon	
  
          implementa.on	
  
       –  Solvable	
  (but	
  very	
  large	
  scale)	
  MIPs	
  
	
  
Smart	
  recovery	
  network	
  
Empty hubs

                        Region 1




                        Region 2




                        Region 3
Controlling	
  conserva.sm	
  

•  Exclude	
  extreme	
  scenarios	
  
   –  Narrow	
  the	
  width	
  of	
  intervals	
   [a , a ]
   –  Limit	
  to	
  k	
  the	
  number	
  of	
  uncertain	
  quan..es	
  that	
  
      may	
  simultaneously	
  take	
  on	
  an	
  extreme	
  quan.ty	
  
•  Challenges	
  for	
  large	
  .me-­‐expanded	
  networks	
  
   –  Very	
  large	
  numbers	
  of	
  inbound-­‐closed	
  sets	
  and	
  
      associated	
  robust	
  constraints:	
       O(τ nS )
   –  Difficult	
  to	
  judge	
  in	
  advance	
  which	
  robust	
  
      constraints	
  will	
  be	
  ac.ve	
  
Two-­‐stage	
  robust	
  reposi.oning	
  
•  Bounded	
  vulnerability	
  of	
  inbound-­‐closed	
  set	
  
                                                                                         
                                                                                      
max                           (a − a )za +                (a − a )za |       za = k
  z                                                                                      
                 a∈δ + (U )                    a∈δ − (U )
      	
  
                   t=0            t=1      t=2              t=3        t=4       t=5
             A
                                                                  [0, 2]

             B
                                               [1, 5]
             C
Appropriate	
  use	
  of	
  two-­‐stage	
  model	
  
        Terminal limit
           known                future
    A


    B


    C


    D


 –  inbound-­‐closed	
  sets	
  with	
  L+1	
  terminals	
  or	
  fewer	
  
Appropriate	
  use	
  of	
  two-­‐stage	
  model	
  
        Robust horizon
           known                future
    A


    B


    C


    D
                                         robust horizon

 –  inbound-­‐closed	
  sets	
  include	
  no	
  nodes	
  aher	
  RH	
  
Appropriate	
  use	
  of	
  two-­‐stage	
  model	
  
         Rolling-horizon robust constraints
              known                       future
     A


     B


     C


     D
                                                     robust horizon
 –  add	
  constraints	
  now	
  for	
  future	
  horizon	
  rolls	
  
      •  assume	
  that	
  demand	
  intervals	
  do	
  not	
  change	
  
Appropriate	
  use	
  of	
  two-­‐stage	
  model	
  
         Rolling-horizon robust constraints
              known                                      future
     A


     B


     C


     D
                                                                            robust horizon
 –  add	
  constraints	
  now	
  for	
  future	
  horizon	
  rolls	
  
      •  assume	
  that	
  demand	
  intervals	
  do	
  not	
  change	
  
Tes.ng	
  the	
  ideas	
  
•  Givens	
  
   –  Historical	
  data	
  from	
  a	
  na.onal	
  consolida.on	
  
      trucking	
  carrier	
  	
  
   –  Loaded	
  moves	
  involve	
  264	
  terminals	
  
   –  Reposi.oning	
  moves	
  (truck	
  and	
  rail)	
  
   –  10	
  empty	
  hubs	
  
   –  At	
  most	
  4	
  daily	
  dispatch	
  .mes	
  per	
  terminal	
  
   –  Wide	
  forecast	
  intervals	
  on	
  loaded	
  demands	
  (+/-­‐	
  
      50%	
  of	
  actual)	
  
Tes.ng	
  the	
  ideas	
  
•  Horizons	
  
    –  14	
  weeks	
  of	
  data	
  
    –  Planning	
  horizon	
  of	
  7	
  days	
  for	
  each	
  model	
  
•  Network	
  size	
  for	
  7-­‐day	
  planning	
  horizon	
  
    –  5,000	
  .me-­‐space	
  nodes	
  
    –  300,000	
  arcs	
  
        •  Primarily	
  reposi.oning	
  arcs	
  
        •  Limited	
  connec.ons	
  
Tes.ng	
  the	
  ideas	
  
•  Simulate	
  
   –  Assume	
  today’s	
  loaded	
  demands	
  known	
  
   –  Solve	
  model,	
  implement	
  today’s	
  decisions	
  
       •  Assume	
  trailer	
  deficits	
  covered	
  by	
  an	
  outsourced	
  trailer	
  
   –  Draw	
  realiza.on	
  of	
  tomorrow’s	
  demands	
  
   –  Repeat	
  
Figure 18: Unmet demands on a given day - Scenario 2

                 Results	
  




 Figure 19: Cumulative unmet demands - Scenario 2
Figure 20: Execution costs on a given day - Scenario 2

                  Results	
  




 Figure 21: Cumulative execution costs - Scenario 2
Short	
  planning	
  horizons	
  




Figure 22: Cumulative unmet demands with different planning horizons - Scenario 1
Next	
  steps	
  
•  Refinements	
  
    –  More	
  reasonable	
  model	
  of	
  true	
  uncertainty	
  in	
  
       demand	
  
    –  Understand	
  sources	
  of	
  cost	
  escala.on,	
  including	
  if	
  
       and	
  where	
  excessive	
  conserva.sm	
  introduced	
  
•  Empty	
  hub	
  selec.on	
  
•  Fleet	
  size	
  versus	
  reposi.oning	
  cost	
  for	
  robust	
  
   plans	
  
What	
  to	
  remember	
  
1.  Robust	
  models	
  for	
  empty	
  mobile	
  resource	
  
    management	
  pragma.c	
  and	
  effec.ve	
  
2.  Empty	
  resource	
  hubs	
  useful	
  for	
  very	
  large-­‐
    scale	
  reposi.oning	
  networks	
  
3.  Rolling-­‐horizon	
  deployments	
  of	
  two-­‐stage	
  
    robust	
  op.miza.on	
  models	
  should	
  u.lize:	
  
       –  short	
  robust	
  horizons	
  
       –  rolling	
  robust	
  constraints	
  


	
  

More Related Content

What's hot

Cmu Lecture on Hadoop Performance
Cmu Lecture on Hadoop PerformanceCmu Lecture on Hadoop Performance
Cmu Lecture on Hadoop PerformanceTed Dunning
 
Oxford 05-oct-2012
Oxford 05-oct-2012Oxford 05-oct-2012
Oxford 05-oct-2012Ted Dunning
 
Improvement of Scheduling Granularity for Deadline Scheduler
Improvement of Scheduling Granularity for Deadline Scheduler Improvement of Scheduling Granularity for Deadline Scheduler
Improvement of Scheduling Granularity for Deadline Scheduler Yoshitake Kobayashi
 
A TRAINING METHOD USING
 DNN-GUIDED LAYERWISE PRETRAINING
 FOR DEEP GAUSSIAN ...
A TRAINING METHOD USING
 DNN-GUIDED LAYERWISE PRETRAINING
 FOR DEEP GAUSSIAN ...A TRAINING METHOD USING
 DNN-GUIDED LAYERWISE PRETRAINING
 FOR DEEP GAUSSIAN ...
A TRAINING METHOD USING
 DNN-GUIDED LAYERWISE PRETRAINING
 FOR DEEP GAUSSIAN ...Tomoki Koriyama
 
The International Journal of Engineering and Science (The IJES)
The International Journal of Engineering and Science (The IJES)The International Journal of Engineering and Science (The IJES)
The International Journal of Engineering and Science (The IJES)theijes
 
Process Algebras and Petri Nets are Discrete Dynamical Systems
Process Algebras and Petri Nets are Discrete Dynamical SystemsProcess Algebras and Petri Nets are Discrete Dynamical Systems
Process Algebras and Petri Nets are Discrete Dynamical SystemsFacultad de Informática UCM
 
Mobicents Media Server theory, practice, cloud considerations, design discuss...
Mobicents Media Server theory, practice, cloud considerations, design discuss...Mobicents Media Server theory, practice, cloud considerations, design discuss...
Mobicents Media Server theory, practice, cloud considerations, design discuss...telestax
 
First Place Memocode'14 Design Contest Entry
First Place Memocode'14 Design Contest EntryFirst Place Memocode'14 Design Contest Entry
First Place Memocode'14 Design Contest EntryKevin Townsend
 

What's hot (9)

Cmu Lecture on Hadoop Performance
Cmu Lecture on Hadoop PerformanceCmu Lecture on Hadoop Performance
Cmu Lecture on Hadoop Performance
 
Oxford 05-oct-2012
Oxford 05-oct-2012Oxford 05-oct-2012
Oxford 05-oct-2012
 
Improvement of Scheduling Granularity for Deadline Scheduler
Improvement of Scheduling Granularity for Deadline Scheduler Improvement of Scheduling Granularity for Deadline Scheduler
Improvement of Scheduling Granularity for Deadline Scheduler
 
A TRAINING METHOD USING
 DNN-GUIDED LAYERWISE PRETRAINING
 FOR DEEP GAUSSIAN ...
A TRAINING METHOD USING
 DNN-GUIDED LAYERWISE PRETRAINING
 FOR DEEP GAUSSIAN ...A TRAINING METHOD USING
 DNN-GUIDED LAYERWISE PRETRAINING
 FOR DEEP GAUSSIAN ...
A TRAINING METHOD USING
 DNN-GUIDED LAYERWISE PRETRAINING
 FOR DEEP GAUSSIAN ...
 
The International Journal of Engineering and Science (The IJES)
The International Journal of Engineering and Science (The IJES)The International Journal of Engineering and Science (The IJES)
The International Journal of Engineering and Science (The IJES)
 
Process Algebras and Petri Nets are Discrete Dynamical Systems
Process Algebras and Petri Nets are Discrete Dynamical SystemsProcess Algebras and Petri Nets are Discrete Dynamical Systems
Process Algebras and Petri Nets are Discrete Dynamical Systems
 
Mobicents Media Server theory, practice, cloud considerations, design discuss...
Mobicents Media Server theory, practice, cloud considerations, design discuss...Mobicents Media Server theory, practice, cloud considerations, design discuss...
Mobicents Media Server theory, practice, cloud considerations, design discuss...
 
Jeroen Zomerdijk
Jeroen ZomerdijkJeroen Zomerdijk
Jeroen Zomerdijk
 
First Place Memocode'14 Design Contest Entry
First Place Memocode'14 Design Contest EntryFirst Place Memocode'14 Design Contest Entry
First Place Memocode'14 Design Contest Entry
 

Viewers also liked

Regnes
RegnesRegnes
Regnesjulut
 
Yooooooooooooooooooooooooooooooooooo
YoooooooooooooooooooooooooooooooooooYooooooooooooooooooooooooooooooooooo
Yoooooooooooooooooooooooooooooooooookevingaleanoyo
 
What's New for Blackbaud Canada
What's New for Blackbaud CanadaWhat's New for Blackbaud Canada
What's New for Blackbaud CanadaAshley Donald
 
I linguaggi delle nostre professioni: l'ingegneria
I linguaggi delle nostre professioni: l'ingegneriaI linguaggi delle nostre professioni: l'ingegneria
I linguaggi delle nostre professioni: l'ingegneriapedagogiaglobale
 
Untitled 1
Untitled 1Untitled 1
Untitled 1wowo1
 
OBOC Press Release
OBOC Press ReleaseOBOC Press Release
OBOC Press ReleaseEBRlhilton
 
Dalelands initiative
Dalelands initiativeDalelands initiative
Dalelands initiativeDM_Bucephalus
 
Predicting baseline d13C signatures of a lake food
Predicting baseline d13C signatures of a lake foodPredicting baseline d13C signatures of a lake food
Predicting baseline d13C signatures of a lake foodLancaster University
 
Sps Algemeen En 16x9
Sps Algemeen En 16x9Sps Algemeen En 16x9
Sps Algemeen En 16x9lindaverweij
 
Beasiswa s2 bagi guru smp 2014
Beasiswa s2 bagi guru smp 2014Beasiswa s2 bagi guru smp 2014
Beasiswa s2 bagi guru smp 2014Rustam Wabula
 
Aprendiendo sobre el cuerpo humano
Aprendiendo sobre el cuerpo humanoAprendiendo sobre el cuerpo humano
Aprendiendo sobre el cuerpo humanoTânia Souza
 
מצגת פודקסטים בהוראה
מצגת   פודקסטים בהוראהמצגת   פודקסטים בהוראה
מצגת פודקסטים בהוראהrivka_o
 

Viewers also liked (20)

Presentation
PresentationPresentation
Presentation
 
Regnes
RegnesRegnes
Regnes
 
Memòria
Memòria Memòria
Memòria
 
Yooooooooooooooooooooooooooooooooooo
YoooooooooooooooooooooooooooooooooooYooooooooooooooooooooooooooooooooooo
Yooooooooooooooooooooooooooooooooooo
 
Mobile devices
Mobile devicesMobile devices
Mobile devices
 
What's New for Blackbaud Canada
What's New for Blackbaud CanadaWhat's New for Blackbaud Canada
What's New for Blackbaud Canada
 
04 2013 alumnesp3p4p5_sils
04 2013 alumnesp3p4p5_sils04 2013 alumnesp3p4p5_sils
04 2013 alumnesp3p4p5_sils
 
I linguaggi delle nostre professioni: l'ingegneria
I linguaggi delle nostre professioni: l'ingegneriaI linguaggi delle nostre professioni: l'ingegneria
I linguaggi delle nostre professioni: l'ingegneria
 
Untitled 1
Untitled 1Untitled 1
Untitled 1
 
OBOC Press Release
OBOC Press ReleaseOBOC Press Release
OBOC Press Release
 
An2292 01
An2292 01An2292 01
An2292 01
 
Dalelands initiative
Dalelands initiativeDalelands initiative
Dalelands initiative
 
Predicting baseline d13C signatures of a lake food
Predicting baseline d13C signatures of a lake foodPredicting baseline d13C signatures of a lake food
Predicting baseline d13C signatures of a lake food
 
Sps Algemeen En 16x9
Sps Algemeen En 16x9Sps Algemeen En 16x9
Sps Algemeen En 16x9
 
Beasiswa s2 bagi guru smp 2014
Beasiswa s2 bagi guru smp 2014Beasiswa s2 bagi guru smp 2014
Beasiswa s2 bagi guru smp 2014
 
Universe Empresarial
Universe EmpresarialUniverse Empresarial
Universe Empresarial
 
Social Issues
Social IssuesSocial Issues
Social Issues
 
Freello | Mobile Marketing 4 MEDIA
Freello | Mobile Marketing 4 MEDIAFreello | Mobile Marketing 4 MEDIA
Freello | Mobile Marketing 4 MEDIA
 
Aprendiendo sobre el cuerpo humano
Aprendiendo sobre el cuerpo humanoAprendiendo sobre el cuerpo humano
Aprendiendo sobre el cuerpo humano
 
מצגת פודקסטים בהוראה
מצגת   פודקסטים בהוראהמצגת   פודקסטים בהוראה
מצגת פודקסטים בהוראה
 

Similar to Robust Repositioning in Large-scale Networks

Lecture8 multi class_svm
Lecture8 multi class_svmLecture8 multi class_svm
Lecture8 multi class_svmStéphane Canu
 
Coherent feedback formulation of a continuous quantum error correction protocol
Coherent feedback formulation of a continuous quantum error correction protocolCoherent feedback formulation of a continuous quantum error correction protocol
Coherent feedback formulation of a continuous quantum error correction protocolhendrai
 
Dsoop (co 221) 1
Dsoop (co 221) 1Dsoop (co 221) 1
Dsoop (co 221) 1Puja Koch
 
04030038 proposal presentation
04030038 proposal presentation04030038 proposal presentation
04030038 proposal presentationSanjay Kikani
 
Possible applications of low-rank tensors in statistics and UQ (my talk in Bo...
Possible applications of low-rank tensors in statistics and UQ (my talk in Bo...Possible applications of low-rank tensors in statistics and UQ (my talk in Bo...
Possible applications of low-rank tensors in statistics and UQ (my talk in Bo...Alexander Litvinenko
 
ACM ICPC 2015 NEERC (Northeastern European Regional Contest) Problems Review
ACM ICPC 2015 NEERC (Northeastern European Regional Contest) Problems ReviewACM ICPC 2015 NEERC (Northeastern European Regional Contest) Problems Review
ACM ICPC 2015 NEERC (Northeastern European Regional Contest) Problems ReviewRoman Elizarov
 
Aaex4 group2(中英夾雜)
Aaex4 group2(中英夾雜)Aaex4 group2(中英夾雜)
Aaex4 group2(中英夾雜)Shiang-Yun Yang
 
VHDL and Cordic Algorithim
VHDL and Cordic AlgorithimVHDL and Cordic Algorithim
VHDL and Cordic AlgorithimSubeer Rangra
 
Tpr star tree
Tpr star treeTpr star tree
Tpr star treeWin Yu
 
Kernel based models for geo- and environmental sciences- Alexei Pozdnoukhov –...
Kernel based models for geo- and environmental sciences- Alexei Pozdnoukhov –...Kernel based models for geo- and environmental sciences- Alexei Pozdnoukhov –...
Kernel based models for geo- and environmental sciences- Alexei Pozdnoukhov –...Beniamino Murgante
 
Biosight: Quantitative Methods for Policy Analysis: Stochastic Dynamic Progra...
Biosight: Quantitative Methods for Policy Analysis: Stochastic Dynamic Progra...Biosight: Quantitative Methods for Policy Analysis: Stochastic Dynamic Progra...
Biosight: Quantitative Methods for Policy Analysis: Stochastic Dynamic Progra...IFPRI-EPTD
 
Financial Networks III. Centrality and Systemic Importance
Financial Networks III. Centrality and Systemic ImportanceFinancial Networks III. Centrality and Systemic Importance
Financial Networks III. Centrality and Systemic ImportanceKimmo Soramaki
 
power system operation and control unit commitment .pdf
power system operation and control unit commitment .pdfpower system operation and control unit commitment .pdf
power system operation and control unit commitment .pdfArnabChakraborty499766
 
Hardware Acceleration for Machine Learning
Hardware Acceleration for Machine LearningHardware Acceleration for Machine Learning
Hardware Acceleration for Machine LearningCastLabKAIST
 

Similar to Robust Repositioning in Large-scale Networks (20)

Teknik Simulasi
Teknik SimulasiTeknik Simulasi
Teknik Simulasi
 
Lecture8 multi class_svm
Lecture8 multi class_svmLecture8 multi class_svm
Lecture8 multi class_svm
 
Coherent feedback formulation of a continuous quantum error correction protocol
Coherent feedback formulation of a continuous quantum error correction protocolCoherent feedback formulation of a continuous quantum error correction protocol
Coherent feedback formulation of a continuous quantum error correction protocol
 
Dsoop (co 221) 1
Dsoop (co 221) 1Dsoop (co 221) 1
Dsoop (co 221) 1
 
04030038 proposal presentation
04030038 proposal presentation04030038 proposal presentation
04030038 proposal presentation
 
Possible applications of low-rank tensors in statistics and UQ (my talk in Bo...
Possible applications of low-rank tensors in statistics and UQ (my talk in Bo...Possible applications of low-rank tensors in statistics and UQ (my talk in Bo...
Possible applications of low-rank tensors in statistics and UQ (my talk in Bo...
 
ACM ICPC 2015 NEERC (Northeastern European Regional Contest) Problems Review
ACM ICPC 2015 NEERC (Northeastern European Regional Contest) Problems ReviewACM ICPC 2015 NEERC (Northeastern European Regional Contest) Problems Review
ACM ICPC 2015 NEERC (Northeastern European Regional Contest) Problems Review
 
Alex1 group2
Alex1 group2Alex1 group2
Alex1 group2
 
Time series Forecasting using svm
Time series Forecasting using  svmTime series Forecasting using  svm
Time series Forecasting using svm
 
Aaex4 group2(中英夾雜)
Aaex4 group2(中英夾雜)Aaex4 group2(中英夾雜)
Aaex4 group2(中英夾雜)
 
Chapter 5
Chapter 5Chapter 5
Chapter 5
 
Ydstie
YdstieYdstie
Ydstie
 
Aaex3 group2
Aaex3 group2Aaex3 group2
Aaex3 group2
 
VHDL and Cordic Algorithim
VHDL and Cordic AlgorithimVHDL and Cordic Algorithim
VHDL and Cordic Algorithim
 
Tpr star tree
Tpr star treeTpr star tree
Tpr star tree
 
Kernel based models for geo- and environmental sciences- Alexei Pozdnoukhov –...
Kernel based models for geo- and environmental sciences- Alexei Pozdnoukhov –...Kernel based models for geo- and environmental sciences- Alexei Pozdnoukhov –...
Kernel based models for geo- and environmental sciences- Alexei Pozdnoukhov –...
 
Biosight: Quantitative Methods for Policy Analysis: Stochastic Dynamic Progra...
Biosight: Quantitative Methods for Policy Analysis: Stochastic Dynamic Progra...Biosight: Quantitative Methods for Policy Analysis: Stochastic Dynamic Progra...
Biosight: Quantitative Methods for Policy Analysis: Stochastic Dynamic Progra...
 
Financial Networks III. Centrality and Systemic Importance
Financial Networks III. Centrality and Systemic ImportanceFinancial Networks III. Centrality and Systemic Importance
Financial Networks III. Centrality and Systemic Importance
 
power system operation and control unit commitment .pdf
power system operation and control unit commitment .pdfpower system operation and control unit commitment .pdf
power system operation and control unit commitment .pdf
 
Hardware Acceleration for Machine Learning
Hardware Acceleration for Machine LearningHardware Acceleration for Machine Learning
Hardware Acceleration for Machine Learning
 

Recently uploaded

Cash Payment 9602870969 Escort Service in Udaipur Call Girls
Cash Payment 9602870969 Escort Service in Udaipur Call GirlsCash Payment 9602870969 Escort Service in Udaipur Call Girls
Cash Payment 9602870969 Escort Service in Udaipur Call GirlsApsara Of India
 
Keppel Ltd. 1Q 2024 Business Update Presentation Slides
Keppel Ltd. 1Q 2024 Business Update  Presentation SlidesKeppel Ltd. 1Q 2024 Business Update  Presentation Slides
Keppel Ltd. 1Q 2024 Business Update Presentation SlidesKeppelCorporation
 
Lucknow 💋 Escorts in Lucknow - 450+ Call Girl Cash Payment 8923113531 Neha Th...
Lucknow 💋 Escorts in Lucknow - 450+ Call Girl Cash Payment 8923113531 Neha Th...Lucknow 💋 Escorts in Lucknow - 450+ Call Girl Cash Payment 8923113531 Neha Th...
Lucknow 💋 Escorts in Lucknow - 450+ Call Girl Cash Payment 8923113531 Neha Th...anilsa9823
 
Catalogue ONG NUOC PPR DE NHAT .pdf
Catalogue ONG NUOC PPR DE NHAT      .pdfCatalogue ONG NUOC PPR DE NHAT      .pdf
Catalogue ONG NUOC PPR DE NHAT .pdfOrient Homes
 
Insurers' journeys to build a mastery in the IoT usage
Insurers' journeys to build a mastery in the IoT usageInsurers' journeys to build a mastery in the IoT usage
Insurers' journeys to build a mastery in the IoT usageMatteo Carbone
 
Call Girls Pune Just Call 9907093804 Top Class Call Girl Service Available
Call Girls Pune Just Call 9907093804 Top Class Call Girl Service AvailableCall Girls Pune Just Call 9907093804 Top Class Call Girl Service Available
Call Girls Pune Just Call 9907093804 Top Class Call Girl Service AvailableDipal Arora
 
Call Girls In Radisson Blu Hotel New Delhi Paschim Vihar ❤️8860477959 Escorts...
Call Girls In Radisson Blu Hotel New Delhi Paschim Vihar ❤️8860477959 Escorts...Call Girls In Radisson Blu Hotel New Delhi Paschim Vihar ❤️8860477959 Escorts...
Call Girls In Radisson Blu Hotel New Delhi Paschim Vihar ❤️8860477959 Escorts...lizamodels9
 
Monte Carlo simulation : Simulation using MCSM
Monte Carlo simulation : Simulation using MCSMMonte Carlo simulation : Simulation using MCSM
Monte Carlo simulation : Simulation using MCSMRavindra Nath Shukla
 
0183760ssssssssssssssssssssssssssss00101011 (27).pdf
0183760ssssssssssssssssssssssssssss00101011 (27).pdf0183760ssssssssssssssssssssssssssss00101011 (27).pdf
0183760ssssssssssssssssssssssssssss00101011 (27).pdfRenandantas16
 
Call Girls in Gomti Nagar - 7388211116 - With room Service
Call Girls in Gomti Nagar - 7388211116  - With room ServiceCall Girls in Gomti Nagar - 7388211116  - With room Service
Call Girls in Gomti Nagar - 7388211116 - With room Servicediscovermytutordmt
 
Call Girls In Connaught Place Delhi ❤️88604**77959_Russian 100% Genuine Escor...
Call Girls In Connaught Place Delhi ❤️88604**77959_Russian 100% Genuine Escor...Call Girls In Connaught Place Delhi ❤️88604**77959_Russian 100% Genuine Escor...
Call Girls In Connaught Place Delhi ❤️88604**77959_Russian 100% Genuine Escor...lizamodels9
 
Lowrate Call Girls In Sector 18 Noida ❤️8860477959 Escorts 100% Genuine Servi...
Lowrate Call Girls In Sector 18 Noida ❤️8860477959 Escorts 100% Genuine Servi...Lowrate Call Girls In Sector 18 Noida ❤️8860477959 Escorts 100% Genuine Servi...
Lowrate Call Girls In Sector 18 Noida ❤️8860477959 Escorts 100% Genuine Servi...lizamodels9
 
Sales & Marketing Alignment: How to Synergize for Success
Sales & Marketing Alignment: How to Synergize for SuccessSales & Marketing Alignment: How to Synergize for Success
Sales & Marketing Alignment: How to Synergize for SuccessAggregage
 
GD Birla and his contribution in management
GD Birla and his contribution in managementGD Birla and his contribution in management
GD Birla and his contribution in managementchhavia330
 
M.C Lodges -- Guest House in Jhang.
M.C Lodges --  Guest House in Jhang.M.C Lodges --  Guest House in Jhang.
M.C Lodges -- Guest House in Jhang.Aaiza Hassan
 
RE Capital's Visionary Leadership under Newman Leech
RE Capital's Visionary Leadership under Newman LeechRE Capital's Visionary Leadership under Newman Leech
RE Capital's Visionary Leadership under Newman LeechNewman George Leech
 
Mondelez State of Snacking and Future Trends 2023
Mondelez State of Snacking and Future Trends 2023Mondelez State of Snacking and Future Trends 2023
Mondelez State of Snacking and Future Trends 2023Neil Kimberley
 
Russian Faridabad Call Girls(Badarpur) : ☎ 8168257667, @4999
Russian Faridabad Call Girls(Badarpur) : ☎ 8168257667, @4999Russian Faridabad Call Girls(Badarpur) : ☎ 8168257667, @4999
Russian Faridabad Call Girls(Badarpur) : ☎ 8168257667, @4999Tina Ji
 

Recently uploaded (20)

Cash Payment 9602870969 Escort Service in Udaipur Call Girls
Cash Payment 9602870969 Escort Service in Udaipur Call GirlsCash Payment 9602870969 Escort Service in Udaipur Call Girls
Cash Payment 9602870969 Escort Service in Udaipur Call Girls
 
Forklift Operations: Safety through Cartoons
Forklift Operations: Safety through CartoonsForklift Operations: Safety through Cartoons
Forklift Operations: Safety through Cartoons
 
Keppel Ltd. 1Q 2024 Business Update Presentation Slides
Keppel Ltd. 1Q 2024 Business Update  Presentation SlidesKeppel Ltd. 1Q 2024 Business Update  Presentation Slides
Keppel Ltd. 1Q 2024 Business Update Presentation Slides
 
Lucknow 💋 Escorts in Lucknow - 450+ Call Girl Cash Payment 8923113531 Neha Th...
Lucknow 💋 Escorts in Lucknow - 450+ Call Girl Cash Payment 8923113531 Neha Th...Lucknow 💋 Escorts in Lucknow - 450+ Call Girl Cash Payment 8923113531 Neha Th...
Lucknow 💋 Escorts in Lucknow - 450+ Call Girl Cash Payment 8923113531 Neha Th...
 
Catalogue ONG NUOC PPR DE NHAT .pdf
Catalogue ONG NUOC PPR DE NHAT      .pdfCatalogue ONG NUOC PPR DE NHAT      .pdf
Catalogue ONG NUOC PPR DE NHAT .pdf
 
Insurers' journeys to build a mastery in the IoT usage
Insurers' journeys to build a mastery in the IoT usageInsurers' journeys to build a mastery in the IoT usage
Insurers' journeys to build a mastery in the IoT usage
 
Call Girls Pune Just Call 9907093804 Top Class Call Girl Service Available
Call Girls Pune Just Call 9907093804 Top Class Call Girl Service AvailableCall Girls Pune Just Call 9907093804 Top Class Call Girl Service Available
Call Girls Pune Just Call 9907093804 Top Class Call Girl Service Available
 
Call Girls In Radisson Blu Hotel New Delhi Paschim Vihar ❤️8860477959 Escorts...
Call Girls In Radisson Blu Hotel New Delhi Paschim Vihar ❤️8860477959 Escorts...Call Girls In Radisson Blu Hotel New Delhi Paschim Vihar ❤️8860477959 Escorts...
Call Girls In Radisson Blu Hotel New Delhi Paschim Vihar ❤️8860477959 Escorts...
 
Monte Carlo simulation : Simulation using MCSM
Monte Carlo simulation : Simulation using MCSMMonte Carlo simulation : Simulation using MCSM
Monte Carlo simulation : Simulation using MCSM
 
0183760ssssssssssssssssssssssssssss00101011 (27).pdf
0183760ssssssssssssssssssssssssssss00101011 (27).pdf0183760ssssssssssssssssssssssssssss00101011 (27).pdf
0183760ssssssssssssssssssssssssssss00101011 (27).pdf
 
Call Girls in Gomti Nagar - 7388211116 - With room Service
Call Girls in Gomti Nagar - 7388211116  - With room ServiceCall Girls in Gomti Nagar - 7388211116  - With room Service
Call Girls in Gomti Nagar - 7388211116 - With room Service
 
Call Girls In Connaught Place Delhi ❤️88604**77959_Russian 100% Genuine Escor...
Call Girls In Connaught Place Delhi ❤️88604**77959_Russian 100% Genuine Escor...Call Girls In Connaught Place Delhi ❤️88604**77959_Russian 100% Genuine Escor...
Call Girls In Connaught Place Delhi ❤️88604**77959_Russian 100% Genuine Escor...
 
Lowrate Call Girls In Sector 18 Noida ❤️8860477959 Escorts 100% Genuine Servi...
Lowrate Call Girls In Sector 18 Noida ❤️8860477959 Escorts 100% Genuine Servi...Lowrate Call Girls In Sector 18 Noida ❤️8860477959 Escorts 100% Genuine Servi...
Lowrate Call Girls In Sector 18 Noida ❤️8860477959 Escorts 100% Genuine Servi...
 
Sales & Marketing Alignment: How to Synergize for Success
Sales & Marketing Alignment: How to Synergize for SuccessSales & Marketing Alignment: How to Synergize for Success
Sales & Marketing Alignment: How to Synergize for Success
 
GD Birla and his contribution in management
GD Birla and his contribution in managementGD Birla and his contribution in management
GD Birla and his contribution in management
 
M.C Lodges -- Guest House in Jhang.
M.C Lodges --  Guest House in Jhang.M.C Lodges --  Guest House in Jhang.
M.C Lodges -- Guest House in Jhang.
 
RE Capital's Visionary Leadership under Newman Leech
RE Capital's Visionary Leadership under Newman LeechRE Capital's Visionary Leadership under Newman Leech
RE Capital's Visionary Leadership under Newman Leech
 
Best Practices for Implementing an External Recruiting Partnership
Best Practices for Implementing an External Recruiting PartnershipBest Practices for Implementing an External Recruiting Partnership
Best Practices for Implementing an External Recruiting Partnership
 
Mondelez State of Snacking and Future Trends 2023
Mondelez State of Snacking and Future Trends 2023Mondelez State of Snacking and Future Trends 2023
Mondelez State of Snacking and Future Trends 2023
 
Russian Faridabad Call Girls(Badarpur) : ☎ 8168257667, @4999
Russian Faridabad Call Girls(Badarpur) : ☎ 8168257667, @4999Russian Faridabad Call Girls(Badarpur) : ☎ 8168257667, @4999
Russian Faridabad Call Girls(Badarpur) : ☎ 8168257667, @4999
 

Robust Repositioning in Large-scale Networks

  • 1. Robust  Empty  Reposi.oning                       in  Large-­‐Scale  Freight  Consolida.on   Networks     Alan  Erera1,  Antonio  Carbajal1,   Mar.n  Savelsbergh2   1  School  of  Industrial  and  Systems  Engineering,  Georgia  Tech     2  University  of  Newcastle,  Australia   Odysseus  2012  
  • 2. What  to  remember   1.  Robust  models  for  empty  mobile  resource   management  pragma.c  and  effec.ve   2.  Empty  resource  hubs  useful  for  very  large-­‐ scale  reposi.oning  networks   3.  Rolling-­‐horizon  deployments  of  two-­‐stage   robust  op.miza.on  models  should  u.lize:   –  short  robust  horizons   –  rolling  robust  constraints    
  • 3. Consolida.on  networks   Customer Satellite terminal Hub terminal origin destination
  • 4. Consolida.on  networks   Net weekly surplus -3 Satellite terminal of empty trailers Hub terminal +5 -5 -1 +2 +4 +1 +1 -1 +1 +1 +3 +4 +3 -4 -3 0 +3 +2 +1 -3 0 -6 +1 -2 -1
  • 5. Dynamic  trailer  reposi.oning   •  Large-­‐scale  terminal  network   –  250+  satellites  and  hubs   •  Dynamics   –  Several  decision  epochs  daily   –  Today’s  projected  demand  for  trailers  accurate   –  Tomorrow’s  (and  beyond)  significantly  uncertain   •  Goal   –  Best  empty  reposi.oning  plan  each  epoch  
  • 6. Modeling  approaches   •  Determinis.c  rolling-­‐horizon  network  flow  LP   –  Assume  that  trailer  demands  tomorrow  (and   beyond)  behave  as  expected     •  Stochas.c  models   –  Minimize  expected  costs  given  probabilis.c  model   of  demand   –  Powell  (87),  Frantzeskakis  and  Powell  (90),   Cheung  and  Powell  (96),  Godfrey  and  Powell  (02a,   02b)         –  Crainic  (93),  Di  Francesco,  et  al.  (09)  
  • 7. Modeling  approaches   •  Robust  op.miza.on  models   –  Bertsimas  and  Sim  (03),  Atamturk  and  Zhang  (07)   –  Morales  (06),  Erera  et.  al.  (09)   •  Two-­‐stage  model   •  Explicit  focus  on  future  feasibility   •  Minimize  cost  of  planned  movements  such  that  a   feasible  set  of  recovery  movements  exists  for  each  non-­‐ extreme  scenario  
  • 8. Two-­‐stage  robust  reposi.oning   First stage decisions t=0 t=1 t=2 t=3 t=4 t=5 A B C D E
  • 9. Two-­‐stage  robust  reposi.oning   First stage net supply bi t=0 t=1 t=2 t=3 t=4 t=5 A Initial trailers B +2 Known and expected future loaded moves C +6 -2 -1 D E +2 +1
  • 10. Two-­‐stage  robust  reposi.oning   Second stage “recovery” decisions t=0 t=1 t=2 t=3 t=4 t=5 A B C D E
  • 11. Two-­‐stage  robust  reposi.oning   Second stage uncertain demand t=0 t=1 t=2 t=3 t=4 t=5 A B Intervals on future loaded moves [a , a ] C [0, 2] D E
  • 12. Two-­‐stage  robust  reposi.oning   •  First  stage  network  flow   min c a xa a xa − x a = bi ∀i∈N a∈δ + (i) a∈δ − (i) xa ≥ 0 and integer ∀ a ∈ A •  Second  stage  “recovery  flow”  for  each  scenario   wa (ω) − wa (ω) = bi (ω) − bi ∀i∈N a∈δ + (i) a∈δ − (i) xa + wa (ω) ≥ 0 ∀a∈A
  • 13. Two-­‐stage  robust  reposi.oning   •  Key  result:  Existence  of  Recovery  Flow   xa ≥ ν(U ) ∀ U is inbound-closed a∈δ + (U )∩I t=0 t=1 t=2 t=3 t=4 t=5 A B C
  • 14. Two-­‐stage  robust  reposi.oning   •  Inbound  closed  set   –  node  set  with  no  incoming  recovery  transporta.on   arcs   t=0 t=1 t=2 t=3 t=4 t=5 A B C
  • 15. Two-­‐stage  robust  reposi.oning   •  Inbound  closed  set   –  this  example  not  inbound-­‐closed   t=0 t=1 t=2 t=3 t=4 t=5 A B C
  • 16. Two-­‐stage  robust  reposi.oning   •  Worst-­‐case  vulnerability  of  inbound-­‐closed  set   xa ≥ ν(U ) ∀ U is inbound-closed a∈δ + (U )∩I ν(U ) = a − a + bi   a∈δ + (U ) a∈δ − (U ) i∈U t=0 t=1 t=2 t=3 t=4 t=5 A [0, 2] B [1, 5] C
  • 17. Challenges   (1) Smart  recovery  network   –  Low-­‐cost  moves  (since  costs  not  modeled)   –  Opera.onally  simple   (2) Appropriate  use  of  two-­‐stage  model   –  Controlling  conserva.sm  pragma.cally   –  Special  considera.ons  for  rolling  horizon   implementa.on   –  Solvable  (but  very  large  scale)  MIPs    
  • 18. Smart  recovery  network   Empty hubs Region 1 Region 2 Region 3
  • 19. Controlling  conserva.sm   •  Exclude  extreme  scenarios   –  Narrow  the  width  of  intervals   [a , a ] –  Limit  to  k  the  number  of  uncertain  quan..es  that   may  simultaneously  take  on  an  extreme  quan.ty   •  Challenges  for  large  .me-­‐expanded  networks   –  Very  large  numbers  of  inbound-­‐closed  sets  and   associated  robust  constraints:   O(τ nS ) –  Difficult  to  judge  in  advance  which  robust   constraints  will  be  ac.ve  
  • 20. Two-­‐stage  robust  reposi.oning   •  Bounded  vulnerability  of  inbound-­‐closed  set       max (a − a )za + (a − a )za | za = k z   a∈δ + (U ) a∈δ − (U )   t=0 t=1 t=2 t=3 t=4 t=5 A [0, 2] B [1, 5] C
  • 21. Appropriate  use  of  two-­‐stage  model   Terminal limit known future A B C D –  inbound-­‐closed  sets  with  L+1  terminals  or  fewer  
  • 22. Appropriate  use  of  two-­‐stage  model   Robust horizon known future A B C D robust horizon –  inbound-­‐closed  sets  include  no  nodes  aher  RH  
  • 23. Appropriate  use  of  two-­‐stage  model   Rolling-horizon robust constraints known future A B C D robust horizon –  add  constraints  now  for  future  horizon  rolls   •  assume  that  demand  intervals  do  not  change  
  • 24. Appropriate  use  of  two-­‐stage  model   Rolling-horizon robust constraints known future A B C D robust horizon –  add  constraints  now  for  future  horizon  rolls   •  assume  that  demand  intervals  do  not  change  
  • 25. Tes.ng  the  ideas   •  Givens   –  Historical  data  from  a  na.onal  consolida.on   trucking  carrier     –  Loaded  moves  involve  264  terminals   –  Reposi.oning  moves  (truck  and  rail)   –  10  empty  hubs   –  At  most  4  daily  dispatch  .mes  per  terminal   –  Wide  forecast  intervals  on  loaded  demands  (+/-­‐   50%  of  actual)  
  • 26. Tes.ng  the  ideas   •  Horizons   –  14  weeks  of  data   –  Planning  horizon  of  7  days  for  each  model   •  Network  size  for  7-­‐day  planning  horizon   –  5,000  .me-­‐space  nodes   –  300,000  arcs   •  Primarily  reposi.oning  arcs   •  Limited  connec.ons  
  • 27. Tes.ng  the  ideas   •  Simulate   –  Assume  today’s  loaded  demands  known   –  Solve  model,  implement  today’s  decisions   •  Assume  trailer  deficits  covered  by  an  outsourced  trailer   –  Draw  realiza.on  of  tomorrow’s  demands   –  Repeat  
  • 28. Figure 18: Unmet demands on a given day - Scenario 2 Results   Figure 19: Cumulative unmet demands - Scenario 2
  • 29. Figure 20: Execution costs on a given day - Scenario 2 Results   Figure 21: Cumulative execution costs - Scenario 2
  • 30. Short  planning  horizons   Figure 22: Cumulative unmet demands with different planning horizons - Scenario 1
  • 31. Next  steps   •  Refinements   –  More  reasonable  model  of  true  uncertainty  in   demand   –  Understand  sources  of  cost  escala.on,  including  if   and  where  excessive  conserva.sm  introduced   •  Empty  hub  selec.on   •  Fleet  size  versus  reposi.oning  cost  for  robust   plans  
  • 32. What  to  remember   1.  Robust  models  for  empty  mobile  resource   management  pragma.c  and  effec.ve   2.  Empty  resource  hubs  useful  for  very  large-­‐ scale  reposi.oning  networks   3.  Rolling-­‐horizon  deployments  of  two-­‐stage   robust  op.miza.on  models  should  u.lize:   –  short  robust  horizons   –  rolling  robust  constraints