Tactical supply chain planning and robustness analysis in the forest products industry

1
Tactical supply chain planning and robustness
analysis in the forest products industry
Daniel Beaudoin, Ph.D candidate Luc LeBel,
Ph.D Jean-Marc Frayret, Ph.D
May 16 2004Halifax
2
Wood procurement context

• Scierie Montauban
• Abitibi_Consolidated (2)
• Gérard Crête (3)
• Adélard Goyette
• Boiseries Savco
• Coop Jos St-Amant
• Maibec
• Ind. Manufacturières Mégantic
• Plante
• Commonwealth Plywood
• Spécialiste du bardeau de cèdre
• Industries Légaré
• Spruce Falls (2)

41-01
3
Wood procurement context

• Mix stands
• Quality
• Utilization conflicts

4
Wood procurement context

• Cohabitation
• Interdependence

5
Proposed approach
Centralized optimization
Engagements
Automated negotiation

• Negotiation objects
• Negotiation protocol
• Decision making models

Optimal Plan
Integrated coordinated plan
6
Woodlands mission

• To procure the mills with raw material at the
  lowest possible cost (Mercure, 1996).

• How much of what
• When
• FromTo

• When
• Where

Demands

• How much of what
• When
• Where

7
Centralized optimization
Increase profitability
Revenues
Costs
8
Centralized optimization
Profit maximization
Markets
9
Age tracking
Example Block 1, Resource 1, Mill 1
0
t 1
1
1
1
1
t 2
2
2
2
2
0
3
3
3
3
t 3
0
1
1
1
1
4
4
4
4
t 4
0
2
2
2
2
10
Valuation levels

• Revenue net of mills processing cost

Adapted from Maness 1989

• Internal collaboration
• Sale / Marketing Production

11
Case study

• 2 sawmills
• 5 procurement areas
• 50 harvesting blocks
• 14 resources
• 4 resource types
• 28 time periods
• 11 ages
• 3 age classes
• 4 valuation levels per age class

12
CPU time
lt 10 minutes
1
2
3
4

T
1
3
4
T
2

1
3
4
T
2

1
4
2
3

T
13
Inventories
14
Freashness
15
Robustness evaluation
How a plan would unfold in the face of uncertainty

• Capacity
• LRF

• Volumes per resource
• Stumpage fees

• Productivity
• Capacity

• Market conditions

• Capacity

• Productivity
• Capacity

16
Method - 2-stage stochastic decomposition
Determine probability distribution
Formulate LP problem
Solve N problems
Generate N scenarios
N candidate plans
Simulate the N plans in N-1 scenarios
Compute statistics for each candidate plan
17
Average profit S.D.
18
Other criteria
Purchase
Unachievable

• LRF
• Lack of mill inventory

• Capacity
• Harvesting, Transport
• Errors in cruise data

19
Plans feasibility
20
Missing volumes
21
Criteria and ranking
a Max profit
b Min S.D.
c Min-Max
d Max feasibility
e Min missing volumes
22
Harvesting schedule
23
Further development

• In-house testing
• Model anticipating equipment moving costs
• Hierarchical decomposition of the problem
• Different decisions
• Different times
• Different persons
• Negotiation module

24
Thank you!