Managing Inventory in Global Supply Chains Facing Port-of-Entry - - PowerPoint PPT Presentation

▶
managing inventory in global supply chains facing port of
SMART_READER_LITE
LIVE PREVIEW

Managing Inventory in Global Supply Chains Facing Port-of-Entry - - PowerPoint PPT Presentation

May 06, 2023 160 likes •486 views

Managing Inventory in Global Supply Chains Facing Port-of-Entry Disruption Risks Co-authors: Brian M. Lewis, Alan L. Erera Chelsea C. White III Schneider National Chair of Transportation & Logistics Georgia Institute of Technology 13

slide-1
SLIDE 1

1 DIMACS/DyDAn/LPS Workshop, 17 November 2008

Managing Inventory in Global Supply Chains Facing Port-of-Entry Disruption Risks

Co-authors: Brian M. Lewis, Alan L. Erera Chelsea C. White III Schneider National Chair of Transportation & Logistics Georgia Institute of Technology 13 October 2008

slide-2
SLIDE 2

2 DIMACS/DyDAn/LPS Workshop, 17 November 2008

Initial comments

  • Prevention, identification, response, recovery from major disruptions
  • Security
  • Ancillary benefits

– More generally, major disruptions – Productivity (economic strength, private sector perspective) – Pilferage

  • Use of information technology – real-time supply chain control,

based on real-time data for the next level of productivity, resilience (downside risk mitigation), and stability

slide-3
SLIDE 3

3 DIMACS/DyDAn/LPS Workshop, 17 November 2008

Importance of trade for economic strength

Growth in Trade as a Percentage of US GDP

2000, 26% 2020, 35% 1990, 13% 0% 5% 10% 15% 20% 25% 30% 35% 40% Percent of GDP

slide-4
SLIDE 4

4 DIMACS/DyDAn/LPS Workshop, 17 November 2008

Supply chain resiliency Supply chain resiliency

slide-5
SLIDE 5

5 DIMACS/DyDAn/LPS Workshop, 17 November 2008

Uncertainty & major disruption

Uncertainty – dealing explicitly with stochastic effects, e.g.,

variability in demand, supply, congestion, driver availability

Major disruption – a loss of nodes &/or links in the global freight

transportation network

Resiliency in supply chains – preventing, gracefully reacting to,

and quickly recovering from major disruptions

Comment: lean supply chains are notoriously fragile Policy implication – the balance in investment between

prevention & quick recovery

R&D challenge – for models of sequential decision making

(e.g., route finding, MDP), a weighted sum of a multiplicative criterion and an additive criterion produces violations of the Principle of Optimality (dynamic programming); games

slide-6
SLIDE 6

6 DIMACS/DyDAn/LPS Workshop, 17 November 2008

Toyota Brake Plant Fire

1997

UPS Labor Strike

1998

Terrorist Attacks & U.S.-Canada Border Closures GM Labor Strike Taiwan Earthquake

1999 2000 2001 2002 2003

Nokia - Ericsson Supplier Fire Longshoreman Strike & West Coast Ports Lockout Iraq War SARS Outbreak

Supply Chain Disruptions

Sarbanes-Oxley Act Business Failures: Enron, Arthur Andersen, Worldcom, Global Crossing, K-Mart, etc. Ford-Firestone Tire Recall NASA Columbia Disaster

slide-7
SLIDE 7

7 DIMACS/DyDAn/LPS Workshop, 17 November 2008

Loss of Key Personnel Restriction of Access / Egress Logistics Provider Failures Dealer Distribution Network Failures Computer Virus / Denial of Service Attacks IT System Failures (Hardware, Software, LAN, WAN) Service Provider Failures Harassment & Discrimination Loss of Key Equipment Tier 1, 2, 3, …n Supplier Problems: Financial Trouble, Quality “Spills”, Failure to Deliver Materials, etc. Warranty / Product Recall Campaigns Logistics Route

  • r Mode

Disruptions Kidnapping Extortion Vandalism Arson

  • Info. Mgmt. Problems

Supplier Bus. Interruption HR Risks – Key Skill Shortage, Personnel Turnovers Loss of Key Supplier

  • Op. Risks

Accounting or Internal Controls Failures Embezzlement Gov’t Inquiries Theft Operator Errors / Accidental Damage Workplace Violence Health & Safety Violations Utilities Failures Communications, Electricity, Water, Power, etc.

Financial Risks

Revenue Management Equip., Facilities, Business Acquisitions & Divestitures Asset Valuation Liquidity / Cash Debt & Credit Rating Fuel Prices Interest Rate Fluctuations Currency & Foreign Exchange Rate Fluctuations Accounting / Tax Law Changes Economic Recession Currency Inconvertibility Credit Default Uncompetitive Cost Structure Financial Markets Instability Inadequate / Inaccurate Financial Controls & Reporting Health Care & Pension Costs Shareholder Activism Adverse Changes in Industry Regulations Adverse Changes in Environmental Regulations Boiler or Machinery Explosion

Hazard Risks

Property Damage

  • Bldg. or Equip. Fire

Building Collapse Asbestos Exposure Mold Exposure Cargo Losses Land, Water, Atmospheric Pollution Geopolitical Risks Severe Hot / Cold Weather Disease / Epidemic Animal / Insect Infestation Blizzard / Ice Storms Hail Damage Lightning Strikes Earthquake Flooding Wildfire Hurricane / Typhoon Heavy Rain / Thunderstorms Tsunami Volcano Eruption Wind Damage Building Subsidence & Sinkholes 3rd Party Liability General Liability Product Liability Directors & Officers Liability Workers Compensation Deductible Limits Terrorism / Sabotage Tornados Loss of Key Facility

Strategic Risks

Customer Relations Corporate Culture Budget Overruns or Unplanned Expenses Product-Market Alignment “Gotta Have Products” Attacks on Brand Loyalty Public Boycott & Condemnation New or Foreign Competitors Market Share Battles Joint Venture / Alliance Relations Pricing & Incentive Wars Ineffective Planning Union Relations, Labor Disagreements & Contract Frustrations Customer Demand Seasonality & Variability Mergers & Industry Consolidation Perceived Quality Inadequate Mgmt. Oversight Negative Media Coverage Product Design & Engineering Program Launch Dealer Relations Timing of Business Decisions & Moves Technology Decisions Product Development Process Supplier Relations Foreign Market Protectionism Ethics Violations Offensive Advertising Loss of Intel. Property

Industry Portfolio of Risks

Enterprise Risks

slide-8
SLIDE 8

8 DIMACS/DyDAn/LPS Workshop, 17 November 2008

Inventory Control with Risk

  • f Major Supply Chain Disruptions

Brian M. Lewis, Alan Erera, Chelsea C. White III

slide-9
SLIDE 9

9 DIMACS/DyDAn/LPS Workshop, 17 November 2008

Outline

  • Motivation and Introduction
  • Part 1: An Inventory Control Model with Border Closures
  • Part 2: An Inventory Control Model with Border Closures

and Congestion

slide-10
SLIDE 10

10 DIMACS/DyDAn/LPS Workshop, 17 November 2008

Motivation and introduction

  • Supply chain security has evolved: from cargo theft to WMD and

border closures

  • Increased focus on supply chain security post-9/11: C-TPAT, CSI,

24-hour rule

  • Research motivated by possibility of port of entry closures

–

September 11 terrorist attacks

  • US-Canadian border delays: minutes to 12 hours
  • US air traffic grounded

–

2003 BAH Port Security Wargame

  • Simulated terrorist attack with “dirty bomb” in containers
  • All US ports closed for 8 days, Backlog takes 92 days to clear

–

2002 10-day labor lockout at 29 Western US seaports

  • Congestion and delays lasted for months
slide-11
SLIDE 11

11 DIMACS/DyDAn/LPS Workshop, 17 November 2008

Motivation and introduction

  • Questions:

–

How can we model major supply chain disruptions (e.g. border closures and congestion) within an inventory control framework?

–

What does an optimal inventory policy look like?

–

How are an optimal policy and the long-run average cost affected by the system parameters?

–

What managerial and policy insights does the model provide?

slide-12
SLIDE 12

12 DIMACS/DyDAn/LPS Workshop, 17 November 2008

Part 1: An Inventory Control Model with Border Closures

Placed Orders Filled Orders (L>0 days) Orders Waiting at Border Closed Border Foreign Supplier Domestic Manufacturer International Border Open Border (0 days) Demand Border Opens (0 days) Observe State: Border Status, Inventory Position

slide-13
SLIDE 13

13 DIMACS/DyDAn/LPS Workshop, 17 November 2008

Problem statement

  • Border system

–

Modeled by a DTMC

–

State space, S ={“O”= Open, “C”= Closed}

–

Exogenous system

O C

pOC>0 pCO>0 pCC>0 pOO>0

slide-14
SLIDE 14

14 DIMACS/DyDAn/LPS Workshop, 17 November 2008

Problem statement

  • Outstanding order vector, z={zkt}

– ke{0,1,2,…, L-1}: orders that have been outstanding for exactly

k days

– L: orders that have been outstanding for at least L days – g: orders that have arrived

  • Order movement function

– Order crossover is prevented

slide-15
SLIDE 15

15 DIMACS/DyDAn/LPS Workshop, 17 November 2008

Problem statement

  • Long-run average cost criterion - no discounting future costs
  • Costs – purchase, holding, penalty
  • Demand - bounded, non-negative, integer-valued, iid
  • Specialize Song and Zipkin (1996) model

–

Stationary state-dependent, basestock policies optimal (denoted, y)

  • Reduced sufficient state information: (it,xt)

–

Ordering decision rule at time t is

slide-16
SLIDE 16

16 DIMACS/DyDAn/LPS Workshop, 17 November 2008

Theoretical results

  • For the border closure model without congestion,
  • The optimal state-invariant order-up-to level ( ) is non-decreasing

in the cost ratio

  • The optimal state-invariant order-up-to level ( ) is non-decreasing

in the penalty cost (p) and non-increasing in holding cost (h).

  • The optimal state-invariant order-up-to level ( ) is non-decreasing

in the minimum leadtime (L).

slide-17
SLIDE 17

17 DIMACS/DyDAn/LPS Workshop, 17 November 2008

Numerical Study

Parameter Values Purchase Cost, c $150,000 Holding Cost, h $100, $500 Penalty Cost, p $1,000, $2,000 Minimum Leadtime, L 1, 7, 15 Transition Probability, pOC 0.001, 0.003, 0.01, 0.02, 0.05, 0.1, 0.2,...,0.8, 0.9, 0.95 Transition Probability, pCO 0.05, 0.1, 0.2,...,0.8, 0.9, 0.95 Demand Distribution Poisson(Mean=0.5), Poisson(Mean=1)

  • Daily review
slide-18
SLIDE 18

18 DIMACS/DyDAn/LPS Workshop, 17 November 2008

Impact of the transition probabilities:

L=15, h=$100, p=$1,000, D~Poisson(0.5)

0.05 0.2 0.4 0.6 0.8 0.95 0.95 0.8 0.6 0.4 0.2 0.05 $75,500 $76,000 $76,500 $77,000 $77,500 $78,000 Long-run Average Cost per Day, g* pOC pCO

0.05 0.2 0.4 0.6 0.8 0.95 0.95 0.8 0.6 0.4 0.2 0.05 12 16 20 24 28 32 36 Order-up-to Level, y* pOC pCO

slide-19
SLIDE 19

19 DIMACS/DyDAn/LPS Workshop, 17 November 2008

Impact of the transition probabilities

  • Observations:

–

Order-up-to level and long-run average cost are non-decreasing in pOC and non-increasing in pCO.

–

The expected duration of a closure (1/pCO) more negatively affects a firm's productivity than the probability of a closure (pOC).

–

Implications for the cooperation between business and government in disruption management and contingency planning.

slide-20
SLIDE 20

20 DIMACS/DyDAn/LPS Workshop, 17 November 2008

Border Workload Queue Foreign Supplier Domestic Manufacturer International Border Open Border Open Border Processed Customers (0 days) Closed Border Closed Border

Part 2: An Inventory Control Model with Border Closures and Congestion

Placed Orders Filled Orders (L>0 days) Demand Closed Border Open Border Observe State: Border Status, Queue Length, Inventory Position

slide-21
SLIDE 21

21 DIMACS/DyDAn/LPS Workshop, 17 November 2008

Results

  • For the border closure model with congestion, the optimal order-up-to

levels (y*(i,n)) are dependent on border state (i) and border workload queue length (n).

  • Order-up-to level and long-run average cost are non-decreasing in pOC

and non-increasing in pCO.

  • The expected duration of a border closure (1/pCO) more negatively

affects a firm's productivity than the probability of a border closure (pOC).

  • Order-up-to level and long-run average cost are more sensitive to the

transition probabilities than in the model without congestion.

slide-22
SLIDE 22

22 DIMACS/DyDAn/LPS Workshop, 17 November 2008

Perishable Product Transportation with Costly Observation

Taesu Taesu Cheong & Chelsea C. White III Cheong & Chelsea C. White III

School of Industrial and Systems Engineering School of Industrial and Systems Engineering Georgia Institute of Technology Georgia Institute of Technology

slide-23
SLIDE 23

23 DIMACS/DyDAn/LPS Workshop, 17 November 2008

Problem

How to most effectively transport perishable freight from origin to

destination

Common practice: try to control temperature in transit. If goods

perish, then discard at the destination.

Question: how valuable would it be to check freight at intermediate

locations between origin and destination and abort transport once it is determined freight is spoiled?

Example: Transport temperature sensitive freight from Japan to

LA/LB to Atlanta.

slide-24
SLIDE 24

24 DIMACS/DyDAn/LPS Workshop, 17 November 2008

Temperature control in reality

Temperatures in an air freight shipment with the instruction to maintain temperatures between 2°C and 8°C (Heap, 2006)

slide-25
SLIDE 25

25 DIMACS/DyDAn/LPS Workshop, 17 November 2008

Economic impact of food spoilage

– 19% of food consumed in U.S. is grown in other countries – Up to 20% of food is discarded due to spoilage (FDA) – U.S. food industry annually discards $35 billion worth of spoiled

goods (Forbes Magazine, April 24, 2006)

– 25% of all vaccine products reach their destination in a degraded

state (Black, 2003, quoting WHO)

Black, A., E‐Logistics in Cold Chain Management, http://www.samedanltd.com/members/archives/EPC/Summer2003/AlastairBlack.htm

slide-26
SLIDE 26

26 DIMACS/DyDAn/LPS Workshop, 17 November 2008

Problem Statement

1 1 n n n+1 n+1 N N … …

W: wholesale purchase cost c(n,n+1): transportation cost from n to (n+1) Rs: reward for state s

n n+1 Decisions Problem Setting NI: no inspection at (n+1) I: perform inspection at (n+1) R: return to the origin

M: inspection cost Ds: disposal cost for state s Origin Destination Intermediate Inspection Points

( ) ( ) ( )

1 , ) 2 , 1 ( 1 ,

1c

n n c n n c n C

n−

+ + − − + − = β β L

States

  • (S+1) states: 0 (fresh), 1, …, S (spoiled)
  • P: State transition probability matrix from location n to (n+1)
slide-27
SLIDE 27

27 DIMACS/DyDAn/LPS Workshop, 17 November 2008

Conclusion

Value of information - Investigated the value of having the choice

to inspect freight quality at intermediate locations in transit

Business implications:

– Better inform decision to invest in IT infrastructure – Better understanding of how to set price; what profit to expect – Operationally, when to optimally inspect

Basic knowledge creation:

– Structure of optimal reward functions & optimal policies – Bound on value of information – Real time algorithmic development

Future research: use of inspection information for:

– Expedite decisions in inventory systems – Security

slide-28
SLIDE 28

28 DIMACS/DyDAn/LPS Workshop, 17 November 2008

Thank you Thank you

slide-29
SLIDE 29

29 DIMACS/DyDAn/LPS Workshop, 17 November 2008

Extra slides Extra slides

slide-30
SLIDE 30

30 DIMACS/DyDAn/LPS Workshop, 17 November 2008

Real-time supply chain control, based

  • n real-time data

Real-time supply chain control, based

  • n real-time data
slide-31
SLIDE 31

31 DIMACS/DyDAn/LPS Workshop, 17 November 2008

Where do the data come from?

Inventory levels Production rates Vehicle, vessel, or trailer

– Position – Speed – Direction – Temperature – Oil or air pressure

Driver alertness Traffic congestion Weather Freight status & visibility

slide-32
SLIDE 32

32 DIMACS/DyDAn/LPS Workshop, 17 November 2008

Real time control, based on real time data

The next level of supply chain efficiency, resilience, stability What’s the value of real-time data? Is it worth the IT

infrastructure investment?

Operationally, how to extract the value (optimally, sub-optimally)

  • f real-time data?

Dealing with data corruption: sensors, transmission, processing What is impact of data processing delay on information value? Are we sure that improved system observation will improve

system performance?