Lecture 28:
Supply Chain Scheduling 2
© J. Christopher Beck 2008
1
Outline

Discrete Manufacturing vs. Continuous
Manufacturing


What Difference Does It Make?
A Typical Framework for Supply Chain
Optimization



Medium Term Planning
Short Term Scheduling
Information System Issues
© J. Christopher Beck 2008
2
Readings

P Ch 8.2, 8.3
© J. Christopher Beck 2008
3
Supply Chain Scheduling
© J. Christopher Beck 2008
4
Discrete vs. Continuous
Manufacturing

Continuous (process) production

Main inventory/products are finely divisible


Discrete production

Main inventory/products are individually
countable


Steel, shampoo, paper
Cars, computers, consumer electronics
Scheduling problems are different
© J. Christopher Beck 2008
5
Continuous:
1. Main Processing




Raw materials are
transformed to intermediate products
Machines have high start-up/shutdown
costs and high changeover costs
Often fixed batch sizes
Usually run 24/7
© J. Christopher Beck 2008
6
Continuous:
2. Finishing

Products of main
processes are “specialized”


Cut, bent, extruded, painted, printed, …
Often these are commodities


Many clients
Mix of make-to-stock, make-to-order
Due dates, sequence dependent
changeovers, and inventory
© J. Christopher
Beck 2008
management
are important

8
Discrete:
1. Primary Conversion

Like finishing in continuous



Process is generally pretty
simple
Output is often a part


Stamping, bending, cutting
Car body part, computer case, …
Schedule is often integrated with
downstream processes
© J. Christopher Beck 2008
9
Discrete:
2. Main Production

Many different
operations of many tools





100 step process for semiconductors!
Machines are very expensive
Often organized like a job shop
Each order has its own route, quantity,
due date
Sequence dependent changeovers
© J. Christopher Beck 2008
10
Discrete:
3. Assembly



Put together parts
Machines are cheap but material
handling is important
Assembly lines


cars or consumer electronics
Due dates, changeovers, sequencing, …
© J. Christopher Beck 2008
11
Table 8.1
Segment
Process
Horizon
Clock Speed Differentiation
Continuous:
Main
Planning
Long-medium
Low
Very low
Continuous:
Finishing
Planning/
scheduling
Medium-short
Medium/
High
Medium/low
Discrete:
Conversion
Planning/
scheduling
Medium-short
Medium
Very low
Discrete:
Main
Planning/
scheduling
Medium-short
Medium
Medium/low
Discrete:
Assembly
Scheduling
Short
High
High
© J. Christopher Beck 2008
12
Table 8.2
Segment
Optimization
Problem
Solution Technique
Continuous: Main
Lot-sizing, cyclic
scheduling
MIP
Continuous:
Finishing
Single machine,
parallel machine
Batch scheduling,
inventory control,
dispatch rules
Discrete: Conversion
Single machine,
parallel machine
Batch scheduling,
dispatch rules, CP
Discrete: Main
Flow shop, job shop
IP, CP, shifting
bottleneck, LS
Discrete: Assembly
Assembly line
Grouping, spacing,
sequencing techniques,
CP, LS
13
© J. Christopher Beck 2008
Supply Chain Decomposition
Mediumterm
planning
Shortterm
scheduling
Stage 1
© J. Christopher Beck 2008
Stage 2
Stage 3
Stage 4
14
Medium-term Aggregation

Time abstraction


1 unit = 1 day or 1 week
Product abstraction

Work at product “family” level


litres of Tuborg beer, not 6-pack, 12, 24, …
Cost/job/capacity abstraction
Average processing times
 Sequence dependencies ignored
© J. Christopher
2008
 Beck
Factory
treated as a single resource

15
Medium-term Planning Results

Daily or weekly




Demand for product families at each facility
Inventory levels
Transportation requirements
No detailed scheduling has been done!
© J. Christopher Beck 2008
16
Medium-term Constrains
Short-term
Mediumterm
planning
Shortterm
scheduling
Stage 1
© J. Christopher Beck 2008
Stage 2
Stage 3
Stage 4
17
Medium-term Decouples
Short-term
Mediumterm
planning
Shortterm
scheduling
Stage 1
© J. Christopher Beck 2008
Stage 2
Stage 3
Stage 4
18
Short-term Scheduling Uses
More Precise Data





Time in minutes or seconds
Horizon ≈ 1 or 2 weeks
Jobs and resources are detailed
Set-up time/cost are taken into account
Products not just product families

Demand for each product is represented
© J. Christopher Beck 2008
19
Problem

Short term schedule solution may not
exist!


Why?
May require feedback of information to
the medium-term and a resolve

Carlsberg takes 10-12 hours for a mediumterm solve …
© J. Christopher Beck 2008
20
Feedback Mechanism Needed
Mediumterm
planning
Shortterm
scheduling
Stage 1
© J. Christopher Beck 2008
Stage 2
Stage 3
Stage 4
21
Information Infrastructure
Requirements
Mediumterm
planning
Shortterm
scheduling
Stage 1
© J. Christopher Beck 2008
Stage 2
Stage 3
Stage 4
22