Scheduling in the Pharmaceutical Industry

1
Scheduling in the Pharmaceutical Industry
Kristinn Magnusson Sigrun Gunnhildardottir

• IEOR 4405 Production Scheduling

2
Pharmaceutical Industry

• Most important driver time-to-market
• Highly Competitive
• Very regulated industry
• High amount of cleaning and set up time needed
  between jobs
• Life and death no room for mistakes

3
Real Life Case

• High but uncertain demand
• Suppliers have long lead times
• 40 different product families
• 1000 different product variations (SKUs)

4
Production Process
5
Goals and Objectives

• Determine a campaign plan and schedule customer
  orders within the campaigns
• Provide realistic and accurate models that are
  solvable within acceptable computational time
• General objective of the plans and schedules
• meet the quantity and delivery date of customer
  orders
• minimize the unproductive production time
• ? maximize economic performance of the company

6
Three Level Hierarchical Framework
7
Level 1 Campaign Planning

• Optimize campaign plan
• Fulfill predicted demand
• Minimize production time
• Helpful for purchasing raw material
• The model is updated every 3 months

8
Level 1 Model

• Objective Minimize
• Subject to
• Allocation
• Sequencing
• Delivery
• Capacity
• Campaign
• Mutually Exclusivity

9
Level 2 Campaign Planning and Order Allocation

• Actual orders are known
• Revise campaign plan
• Allocate orders to campaigns
• Specify in which campaign each order are produced
  on every production stage
• It gives the latest allowed completion time for
  the order

10
Level 3 Detailed Schedule

• Actual timing of activities
• Objective to minimize late deliveries
• The model gives
• Machine/Campaign for each order for every
  production stage
• Production sequence of orders
• Start and processing time of tasks
• Setup time required between orders

11
Heuristic Decomposition of Production Stages
12
Improving Lower Bounds...

• ... by adding valid inequalities
• A constraint for the minimum number of campaignes
  needed for a feasible solution
• A constraint for the minimum number of delayed
  jobs

13
Solution Times

• These models have been tested with real data and
  have been shown to be solvable within acceptable
  computational time
• 1. level 14 hours
• 2. level 6 hours
• 3. level 6 minutes

14
References

• P. Jensson, N. Shah and H. Stefansson,
  Multiscale Planning and Scheduling in the
  Secondary Pharmaceutical Industry, Published
  online October 26, 2006 in Wiley InterScience
  (www.interscience.wiley.com)
• N. Shah, Pharmaceutical supply chains key
  issues and strategies for optimisation,
  Computers and Chemical Engineering 28 (2004)
  929941

15
Any Questions ?