Deploying Simulation to Compare among Different Risk Reduction
Strategies for Supply Chains
Cameron MacKenzie, Defense Resources
Management Institute, Naval Postgraduate School
Japanese earthquake and tsunami
2
Previous research
• Global macroeconomic impacts MacKenzie, C.A., Santos, J.R., & Barker, K. (2012). Measuring changes in international production from a disruption: Case study of the Japanese earthquake and tsunami. International Journal of Production Economics, 138(2), 293-302.
• Disruption management strategies in the automobile sector MacKenzie, C.A., Barker, K., & Santos, J.R. (2013). Modeling a severe supply chain disruption and post-disaster decision making with application to the Japanese earthquake and tsunami. IIE Transactions, in press.
3
Risk management strategies
4
But this chart just shows benefits! What about costs of strategies?
Challenges of determining strategy
• No single objective function
– Maximize profit
– Maximize demand satisfied
– Maximize percentage of demand versus competitors
• Different costs for each strategy
– Spend $1000 on one strategy to save $10,000 in profit
– Spend $500 on another strategy to save $7500 in profit
• Preparedness versus response strategies
– Spend $1000 on inventory to save $10,000 in profit if a disruption occurs
– Spend $1000 to purchase from alternate suppliers once disruption occurs to save $5000 in profit
5
Challenges of determining strategy
• Uncertainty over length of disruption, customer actions, and competitors’ strategies
• No functional relationship between strategies and objective function and/or relationships are highly nonlinear
6
Supply chain disruption in auto sector
7
Illustrative example
Supplier 1
Supplier 2
Supplier 3
Supplier 4
Firm 2
Firm 3
Firm 1
Final consumers
55
21
8
12
8
26
13
8
Severe supply chain disruption: multiple suppliers cannot produce and multiple firms are impacted
Simulate severe supply chain
disruption
Select response strategy
Choose response strategy
that performs the best
according to an
objective
Repeat but assume that response strategy is being
implemented
Simulation
9
Discretize each strategy so that each
strategy costs the same “small” amount
Preparedness strategies
Select one preparedness
strategy
Response strategies
Choose preparedness strategy that performs the
best
Repeat but assume that preparedness strategy is being implemented
Firm 2’s preparedness strategies
1. Hold raw materials inventory
– Supply 1
– Supply 2
– Supply 3
2. Hold finished goods inventory
3. Arrange to purchase from alternate suppliers in case of disruption
– Alternate supplier 1
– Alternate supplier 2
– Alternate supplier 3
– All alternate suppliers
10
Firm 2’s response strategies
1. Purchase from alternate suppliers at higher cost
– Alternate supplier 1
– Alternate supplier 2
– Alternate supplier 3
– All alternate suppliers
2. Help supplier 3 recover more quickly
11
Objective: maximize profit
• If no preparedness strategy
• Then best response strategy
1. Help supplier 3 recover more quickly
2. Purchase from alternate supplier 3
3. Purchase from alternate supplier 3
4. Purchase from alternate supplier 3
5. Purchase from alternate supplier 3
12
Objective: maximize profit
13
Preparedness strategies
1. Arrange purchase from all alternate suppliers
2. Arrange purchase from alternate supplier 3
3. Arrange purchase from alternate supplier 1
4. Arrange purchase from alternate supplier 3
5. Arrange purchase from alternate supplier 3
6. Arrange purchase from all alternate suppliers
7. Arrange purchase from all alternate suppliers
8. Arrange purchase from alternate supplier 3
Response strategies
1. Purchase from alternate supplier 1
2. Purchase from alternate supplier 2
3. Help supplier 3 recover more quickly
4. Purchase from alternate supplier 1
5. Help supplier 3 recover more quickly
1. Purchase from alternate supplier 3
2. Help supplier 3 recover more quickly
3. Help supplier 3 recover more quickly
4. Purchase from alternate supplier 2
5. Help supplier 3 recover more quickly
Objective: maximize demand
14
Preparedness strategies
1. Arrange purchase from all alternate suppliers
2. Raw material inventory for supply 2
3. Raw material inventory for supply 2
4. Raw material inventory for supply 3
5. Raw material inventory for supply 3
6. Raw material inventory for supply 2
7. Raw material inventory for supply 2
8. Raw material inventory for supply 3
Response strategies
1. Help supplier 3 recover more quickly
2. Help supplier 3 recover more quickly
3. Help supplier 3 recover more quickly
4. Help supplier 3 recover more quickly
5. Help supplier 3 recover more quickly
Greedy algorithm for preparedness
15
Arrange purchase
from alternate supplier 3
Arrange purchase from
alternate supplier 1
Arrange purchase from
all alternate suppliers
Raw material inventory
Finished goods inventory
Arrange purchase from
alternate supplier 1
Arrange purchase from
all alternate suppliers
Raw material inventory
Finished goods inventory
But algorithm could be too greedy
16
Arrange purchase
from alternate supplier 3
Arrange purchase from
alternate supplier 1
Arrange purchase from
all alternate suppliers
Raw material inventory
Finished goods inventory
Arrange purchase from alternate
supplier 1
Arrange purchase from all alternate
suppliers
Raw material inventory
Arrange purchase from alternate
supplier 1
Arrange purchase from all alternate
suppliers
Raw material inventory
Future research
• Refine simulation further
– Expand current path versus explore new path
– Investigate number of simulation runs to obtain certain degrees of confidence
• Use simulation to generalize insights about optimal risk management strategies
– When is holding inventory optimal?
– When is buying from alternate suppliers optimal?
Email: [email protected]
17
Simulation
1. Simulate response strategies given no preparedness strategy
2. Discretize each strategy so that each strategy costs the same “small” amount
1. Keep 10 units of raw material inventory for $100
2. Buy 5 units of supply from alternate supplier for $100
3. Spend $100 to help primary supplier recover more quickly
3. Simulate severe disruption with firm selecting one strategy
4. Choose strategy that performs the best according to an objective (e.g., profit, customer demand) or weighted combination of objectives
5. Repeat step 2 but assume that strategy chosen in step 3 is being pursued
18
Results from maximizing profit
19
Preparedness strategy Total expected profit lost
due to disruption
Arrange purchase from all alternate suppliers 392
Arrange purchase from alternate supplier 3 334
Arrange purchase from alternate supplier 1 397
Arrange purchase from alternate supplier 3 335
Arrange purchase from alternate supplier 3 328
Arrange purchase from all alternate suppliers 429
Arrange purchase from all alternate suppliers 331
Arrange purchase from alternate supplier 3 309
Results from maximizing profit
20
Response strategy given best set of preparedness strategies
Total expected profit lost due to disruption
Purchase from alternate supplier 1 350
Purchase from alternate supplier 2 312
Help supplier 3 recover more quickly 309
Purchase from alternate supplier 1 312
Help supplier 3 recover more quickly 326
Results from maximizing profit
21
Response strategy given no preparedness strategies
Total expected profit lost due to disruption
Help supplier 3 recover more quickly 471
Purchase from alternate supplier 3 466
Purchase from alternate supplier 3 462
Purchase from alternate supplier 3 458
Purchase from alternate supplier 3 454
Purchase from alternate supplier 3 450
Purchase from alternate supplier 3 446
Purchase from alternate supplier 3 442
Purchase from alternate supplier 3 440
Purchase from alternate supplier 3 437
Results from maximizing demand
22
Strategy Total expected demand lost due to disruption
Purchase from all alternate suppliers 674
Raw material inventory for supply 2 437
Raw material inventory for supply 2 562
Raw material inventory for supply 3 489
Raw material inventory for supply 3 659
Raw material inventory for supply 2 693
Raw material inventory for supply 2 575
Raw material inventory for supply 3 597
Results from maximizing profit
23
Response strategy given best set of preparedness strategies
Total expected demand lost due to disruption
Help supplier 3 recover more quickly 795
Help supplier 3 recover more quickly 749
Help supplier 3 recover more quickly 640
Help supplier 3 recover more quickly 606
Help supplier 3 recover more quickly 597
Outline
24
1. Motivation
2. Research contribution
3. Model and simulation
4. Application
Supply chain risk management
• Qualitative [1, 2]
• Quantitative
– Production and inventory models [3]
– Game theory [4]
[1] Y. Sheffi, 2005. The resilient enterprise: Overcoming vulnerability for competitive advantage. Cambridge: The MIT Press. [2] C. S. Tang, 2006. Robust strategies for mitigating supply chain disruptions. International Journal of Logistics Research and Applications 9 (1):33-45. [3] B. Tomlin, 2006. On the value of mitigation and contingency strategies for managing supply chain disruption risks. Management Science 52 (5): 639-657. [4] V. Babich, 2006. Vulnerable options in supply chains: Effects of supplier competition. Naval Research Logistics 53 (7):656-676.
Disruption management [1]
• Disruptions cause operation plans to deviate
• Disruption management studies optimal way to react in the midst of disruptions
– What should be done once a disruption occurs?
– How to minimize the impacts and return to normal production?
[1] G. Yu and X. Qi, 2004. Disruption management: Framework, models and applications. River Edge, NJ: World Scientific Publishing.
What is new with this research?
Mitigation Preparedness
Recovery Response
Supply chain risk
management
Decision and actions by suppliers and firms during
and after disruption
Research questions
• How can we model the supply chain where
– Some facilities are inoperable?
– Other firms experience a supply shortage?
• What can firms do to mitigate the impacts of inoperable facilities and supply shortages?
Outline
29
1. Motivation
2. Research contribution
3. Model and simulation
4. Application
Simulation
Supply shortage for
firms
Move production to
alternate facility?
Firms receive required supplies
Buy from alternate supplier?
No
Yes Supplier’s facility is
closed
Finished goods
inventory?
Demand not satisfied or
customers buy from other firms
Supplier’s facility
reopens?
No
No
Yes
Yes
No
Supply inventory?
No
30
Produce at alternate facility?
Per-unit cost of producing at alternate facility
Fixed cost of moving production to alternate facility
Expected lost revenue of not producing
Per-unit cost of producing at primary facility
Probability primary facility opens next period
Probability supplier’s customers buy from other suppliers
Cost at alternate facility
Expected cost at primary facility
31
Produce at alternate facility?
Per-unit cost of producing at alternate facility
Fixed cost of moving production to alternate facility
Expected lost revenue of not producing
Per-unit cost of producing at primary facility
Probability primary facility opens next period
Probability supplier’s customers buy from other suppliers
Cost at alternate facility
Expected cost at primary facility
32
Threshold parameters for supplier
𝑝 =𝑟 − 𝑐+ 𝜃
𝑐+ − 𝑐 1 − 𝜃
If probability that primary facility will open next period is greater than 𝑝 , supplier will not produce at alternate facility
Per-unit cost of producing at
alternate facility
Per-unit cost of producing at primary
facility
Probability supplier’s customers buy from
other suppliers
Per-unit revenue
33
Threshold parameters for supplier
𝐶 =𝑝𝑍 + 𝑧 𝑟 − 𝑐+ 𝜃 − 𝑝 𝑐+ − 𝑐 1 − 𝜃
𝑝 1 − 1 − 𝑝 1 − 𝜃
If fixed cost of moving production is greater than 𝐶 , supplier will not produce at alternate facility
𝑝 =𝑟 − 𝑐+ 𝜃
𝑐+ − 𝑐 1 − 𝜃
Cost of alternate
facility
Cost of primary facility
Probability of buying from
other suppliers Revenue
Probability primary facility opens next period
Per-period demand Backorders
34
Threshold parameters for supplier
Probability of primary facility opening
𝑝 =𝑟 − 𝑐+ 𝜃
𝑐+ − 𝑐 1 − 𝜃
𝑝
Never produce at alternate
facility
𝐶
Fixe
d c
ost
of
mo
vin
g to
alt
ern
ate
faci
lity
𝐶 =𝑝𝑍 + 𝑧 𝑟 − 𝑐+ 𝜃 − 𝑝 𝑐+ − 𝑐 1 − 𝜃
𝑝 1 − 1 − 𝑝 1 − 𝜃
Produce at alternate facility but may wait some length of time
35
Never produce at alternate
facility
Firm’s influence diagram
How much to produce?
Maximize profit in
current period
Satisfy demand
Value
Time when suppliers’ facilities reopen Customer
loyalty
Inventory on hand
Selling price
Cost of alternate suppliers
36
Modeling insights
• Incorporating business decisions in midst of supply chain disruptions
• Solving for optimal production decisions as function of model parameters
• Measuring impact of preparedness decisions on firm’s ability to respond during disruption
37
Outline
38
1. Motivation
2. Research contribution
3. Model and simulation
4. Application
Supply chain disruption in auto sector
39
Application inspired by auto sector
• Supplies required for production
• Several model parameters gleaned from news reports
• More precise information needed for cost and revenue parameters
Supplier 1
Supplier 2
Supplier 3
Supplier 4
Firm 2
Firm 3
Firm 1
Final consumers
55
21
8
12
8
26
13
40
Simulation results
Average production when suppliers do not move to alternate facility
41
Simulation results
42
Average production when suppliers do not move to alternate facility
Simulation results
Average production when suppliers do not move to alternate facility
Simulation results
Average production when suppliers move to alternate facility
44
Sensitivity on parameters for Firm 2
Parameter Low Base High
Tradeoff between objectives
Maximizes profit Equally prefer
both objectives Satisfies demand
Final goods inventory 0 periods 6 periods 12 periods
Cost of alternate supplier Primary supplier
+ 6 Primary supplier
+ 3 Equal to primary
supplier
Selling price Equal to cost Cost + 1 Cost + 2
Primary supplier’s recovery (expected time)
36 periods 26 periods 3 periods
Supply inventory 0 period 2 periods 4 periods
Customer loyalty (probability firm’s customer does not buy from competitor)
0.01 0.61 0.99
45
Application insights
• Illustrative example reflects actual situation
– Toyota and Honda’s share of production in North America fell from 10% to 7% each
– Nissan’s share of production in North America remained constant
– Detroit 3 automakers increased their share of production in North America by 4%
• Application provides insights into best strategies for response and recovery
– Buying from an alternate supplier may be a better long-term strategy than inventory
– Costs of different strategies should be incorporated
46
This work was supported by
• The National Science Foundation, Division of Civil, Mechanical, and Manufacturing Innovation, under award 0927299
• The Center for International Business Education and Research (CIBER) at The George Washington University
Email: [email protected]
47