General Motors Company · NYSE:GM
General Motors — Technical Presentation (December 2005)
26 slides from General Motors's technical presentation, Dec 9, 2005
Slide 1: E-Business and Reverse Logistics
E-Business and Reverse Logistics Mark S. Daskin, Kristin Sahyouni and R. Canan Savaskan Northwestern University, Evanston, IL 60208 USA GM ISL WORKSHOP ON FRONTIERS OF E-BUSINESS Bangalore 8-9 December 2005 Outline of the talk Overview of e-commerce and reverse logistics Integrated forward/reverse l
Slide 2: What is Reverse Logistics?
What is Reverse Logistics? Suppliers Manufacturers Distribution End-Users Forward Flow Reverse Flow The process of planning, implementing, and controlling all logistics activities to collect, disassemble, and recover “used” products, product parts, and/or materials in an efficient, economically sust
Slide 3: Another look at reverse logistics
Another look at reverse logistics electronics, paper recycling computer construction waste equipment clothing, auto parts books © Majumder, 2005 GM ISL Workshop on Frontiers of E-Business India Science Lab Logistics network design decisions Logistics Network Design Decisions © Majumder, 2005 GM ISL
Slide 4: Designing a Reverse Logistics Strategy depends on …
Designing a Reverse Logistics Strategy depends on … Nature of product returns : Push vs. Pull Commercial Returns Legislation Driven Warranties Recycling / remanufacturing Repairs Profit potential Driven Contractual Agreements Inbound Quality Defects PUSHED to RL PULLED via RL GM ISL Workshop on Fron
Slide 5: Size of the Commercial Returns Problem
Size of the Commercial Returns Problem Value of returned goods is $60 billion, annually in the U.S. An additional $40 billion dollars spent on reverse logistics Web returns alone cost between $1.8 and $2.5 billion in 2002. Source: Enright (2003) GM ISL Workshop on Frontiers of E-Business India Scien
Slide 6: WEEE Directive covers a large group of products
WEEE Directive covers a large group of products… • Large household appliances • Small household appliances • IT and telecommunications equipment • Consumer equipment • Lighting equipment • Electrical and electronic tools • Toys, Leisure and sports equipment • Medical Devices • Monitoring and Control
Slide 7: Why e-commerce and reverse logistics?
Why e-commerce and reverse logistics? Reverse logistics (closed loop supply chain) is critical in any modern logistics system Particularly relevant with e-commerce Increased rate of customer returns Entire business models based on reverse logistics (Netflix – DVD rentals) GM ISL Workshop on Frontier
Slide 8: Life Cycle Considerations
Is there any value in Integrated Forward/Reverse Logistics Design ? “Can we show that in many cases one needs to design the forward SC together with the reverse one over the lifecycle of the product?” Presentation by V. Daniel R. Guide, Jr. and Luk N. Van Wassenhove entitled, “The Evolution of Close
Slide 9: Remainder of the talk: Review: Fixed Charge Location Model
Remainder of the talk Overview of e-commerce and reverse logistics Integrated forward/reverse location models (formulation, solution algorithm) Application of models to India When does integration matter? Future research GM ISL Workshop on Frontiers of E-Business India Science Lab Review: Fixed Char
Slide 10: Integrated model with cost incentive for co-location
Review: Fixed charge location model Min ∑ f j X j + ∑ ∑ hi cij Y ij Fixed plus transport j i j costs S.t. ∑ Y ij = 1 ∀i j Each demand assigned Y ij − X j ≤ 0 ∀i, j X j ∈ {0,1} ∀j Only to open sites Y ij ≥ 0 ∀i, j Integrality Non- Non-negativity GM ISL Workshop on Frontiers of E-Business India Scienc
Slide 11: Lagrangian relaxation: Relax assignment constraints
Lagrangian relaxation: Relax assignment constraints Opt ∑ f j X Fj + ∑ ∑ (hi cij − λ ij )Y ijF + ∑ β f j X Rj + ∑ ∑ (α hi γ cij − µ ij )Y ijR j i j j i j + ∑ ∑ λ ij + ∑ ∑ µ ij − ∑ s j X cj i j i j j S.t. Y ijF − X Fj ≤ 0 ∀i, j Y ijR − X Rj ≤ 0 ∀i, j X Fj ∈ {0,1} ∀j X Rj ∈ {0,1} ∀j Y ijF ≥ 0 ∀i, j Y
Slide 12: Algorithm summary
Must have at least one of each 1. ∑ X Fj > 0 but ∑ X Rj = 0 (Forward, but no reverse) j j a. Find best reverse to add to existing forward b. Find best stand - alone reverse c. Find best new joint facility PICK BEST OF THOSE THREE 2. ∑ X Fj = 0 but ∑ X Rj > 0 (Reverse, but no forward) j j 3. ∑ X Fj =
Slide 13: Data for India
Data for India Population of 446 large cities (~155,365,000 people or about 14% of India’s 1.1 billion people) Missing Pimpri and Kalyan in Maharashtra with combined population of 2.547 million All other cities of 1 M or more Largest – Bombay: 12.692 million Midsized – Bangalore: 4.931 million Small
Slide 14: Example from India
Example from India 9 Forward sites; 22 Reverse sites; 7 Joint sites National Oceanographic and Atmospheric Administration http://oas.ngdc.noaa.gov/mgg/plsql/extract.search_screen?p_lat_min=3.5&p_lat_max=36.0&p_lon_min=64.5&p_lon_max=97.5 GM ISL Workshop on Frontiers of E-Business India Science Lab S
Slide 15: Reduced return transport cost and increased return rate scenarios
50% reduced return transport cost 9 Forward sites; 14 Reverse sites; 8 Joint sites Fewer reverse sites GM ISL Workshop on Frontiers of E-Business India Science Lab 60% increase in return rate 9 Forward sites; 29 Reverse sites; 6 Joint sites More reverse sites GM ISL Workshop on Frontiers of E-Busine
Slide 16: Forward dominant model
50% increase in reverse facility fixed cost 9 Forward sites; 16 Reverse sites; 9 joint sites Fewer Reverse sites; More joint sites GM ISL Workshop on Frontiers of E-Business India Science Lab Forward dominant model Min ∑ f j X j + ∑ ∑ hi cij Y ij + ∑ β f j X Rj + ∑ ∑ α hi γ cij Y ijR F F j i j j i j
Slide 17: Reverse dominant model
Reverse dominant model Min ∑ f j X j + ∑ ∑ hi cij Y ij + ∑ β f j X Rj + ∑ ∑ α hi γ cij Y ijR F F j i j j i j S.t. ∑ Y ijF = 1 ∀i ∑ Y ijR = 1 ∀i j j Y ijF − X Fj ≤ 0 ∀i, j Y ijR − X Rj ≤ 0 ∀i , j X Fj ∈ {0,1} ∀j X Rj ∈ {0,1} ∀j Y ijF ≥ 0 ∀i, j Y ijR ≥ 0 ∀i, j X Fj − X Rj ≤ 0 Forward is Forward Model
Slide 18: Integrated vs. Sequential or Independent Decisions
Integrated vs. Sequential or Independent Decisions Integrated – Based on Sequential or independent models outlined above 1. Forward dominant: locate forward first then reverse given forward 2. Reverse dominant: locate reverse first then forward given reverse 3. Co-location: locate forward and revers
Slide 19: Measuring similarity between sites
Measuring similarity between sites in Xm and Xn Cost of “moving” moving” ∑ ∑ cij Z ij facilities Min i∈ X m j∈ X n S.t. ∑ Z ij ≥ 1 ∀j ∈ X n Each facility in i∈ X m Xn assigned ∑ Z ij ≥ 1 ∀i ∈ X m j∈ X n Each facility in Xm assigned Z ij ≥ 0 ∀i ∈ X m , j ∈ X n Non- Non-negativity GM ISL Workshop on F
Slide 20: Similarity Measure and Computational Study
Similarity Measure Solution 1 Solution 3 0 of 4 locations in common Total difference between solutions (in miles) = 190 New York 81 Allentown Chicago 86 Milwaukee Los Angeles 7 Burbank Houston 16 Pasadena GM ISL Workshop on Frontiers of E-Business India Science Lab Computational Study Network Design
Slide 21: Computational Study
Computational Study Data Sets: 263 Nodes 150 Nodes GM ISL Workshop on Frontiers of E-Business India Science Lab Test cases Each parameter level computed for LL, LH, HL, HH levels of other parameters = 4•(100 + 100 + 100) = 1,200 parameter combinations 3 models / 2 data sets / 2 formulations = 3•2•2•
Slide 22: Summary Results
Summary Results Solution Time (sec) Optimality Gap Model Data Set Average Maximum Average Maximum Forward 150 nodes 0.84 16.68 0.01% 0.10% Dominant 263 nodes 5.24 104.16 0.00% 0.08% Reverse 150 nodes 0.65 2.55 0.00% 0.07% Dominant 263 nodes 6.49 12.05 0.00% 0.05% Co-Location 150 nodes 0.93 6.85 0.01
Slide 23: Forward Dominant Results
Forward Dominant Results Impact of Gamma on Value of Integration 35% 30% 25% Percent Difference Low alpha; low beta 20% Low alpha; high beta 15% High alpha; low beta High alpha; high beta 10% 5% 0% 0 1 2 3 4 5 6 Reverse/Forward Transportation Ratio As reverse transport becomes more expensive Integra
Slide 24: Forward Dominant Summary and Reverse Dominant Summary
Forward Dominant Summary 1. For large reverse fixed costs and high reverse transport costs, integrated approach is advantageous 2. When reverse fixed costs and transport cost ratios are small (<0.5), sequential approach is satisfactory GM ISL Workshop on Frontiers of E-Business India Science Lab Rev
Slide 25: Co-location Model Results and Future Research
Co-location Model Results 1. Value of integration increases slightly with reverse/forward fixed cost ratio (up to 3.5%) 2. Highly similar forward networks 3. Slightly dissimilar reverse networks GM ISL Workshop on Frontiers of E-Business India Science Lab Future Research 1. Integrated model over tim
Slide 26: Final thoughts
Final thoughts …. U.S. Online Retail Sales to Reach $130 Billion by 2006, $104 Billion in 2005 http://retailindustry.about.com/library/bl/q2/bl_jup052201.htm $104 Billion is That’ That’s a LOT 0.15% of first 10 of money! months of 2005 Wish I had that! U.S. retail sales Get the fundamentals right; t
General Motors Company — Technical Presentation (December 2005) on 6ix