Factory Physics?
Perfection of means and confusion of goals seem to
characterize our age.
– Albert Einstein
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000
http://factory-physics.com
1
What is Factory Physics?
Quantitative Tools:
• probability
• queueing models
• optimization
Operations Management:
•
•
•
•
inventory management
shop floor control (MRP, JIT)
scheduling, aggregate planning
capacity management
Manufacturing Principles:
• characterize fundamental logistical behavior
• facilitate better management by working with, instead of against, natural
tendencies
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000
http://factory-physics.com
2
Why Study Factory Physics?
Ideal: sophisticated
technology
Reality: blizzard of
buzzwords
gurus
automation
information
technology
Lack of
System
control
methods
benchmarking
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000
http://factory-physics.com
3
Performance
Can’t Rely on Benchmarking
Benchmarking
can result in an
increasing gap
in performance
when standard
is accelerating.
Leader
Follower
Time
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000
http://factory-physics.com
4
Need for a Science of Manufacturing
Goals
Perspective
• rationalize buzzwords
• recognize commonalties
across environments
• accelerate learning
curve
• basics
• intuition
• synthesis
Practices change, but principles persist!
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000
http://factory-physics.com
5
Scope of Factory Physics
Process
Line
System
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000
http://factory-physics.com
6
Factory Physics
Definition: A manufacturing system is a network of processes through
which parts flow and whose purpose is to generate profit now and in
the future.
Structure: Plant is made up of routings (lines), which in turn are made
up of processes.
Focus: Factory Physics is concerned with the network and flows at the
routing (line) level.
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000
http://factory-physics.com
7
Conclusions
Factory Physics is:
• a set of manufacturing principles
• tools for identifying leverage in existing systems
• a framework for designing more effective new systems
• still being developed…
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000
http://factory-physics.com
8
Manufacturing Matters!
Watch the costs and the profits will take care of themselves.
– Andrew Carnegie
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000
http://factory-physics.com
9
Conventional Wisdom
Popular View: We are merely shifting to a service economy, the same way we
shifted from an agrarian economy to a manufacturing economy.
Statistic:
• 1929 — agriculture employed 29% of workforce
• 1985 — it employs 3%
Interpretation: Shift was good because it substituted high productivity/high
paying (manufacturing) jobs for low productivity/low paid (agriculture) jobs.
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000
http://factory-physics.com
10
Problems with Conventional Wisdom
Offshoring: Agriculture never shifted offshore in a manner analogous to
manufacturing jobs shifting overseas.
Automation: Actually, we automated agriculture resulting in an enormous
improvement in productivity. But the production stayed here.
Measurement:
• 3% figure (roughly 3 million jobs) is by SIC
• But, this does not include crop duster pilots, vets, etc.
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000
http://factory-physics.com
11
Tight Linkages
Economist View: linkages should not be considered when evaluating an
industry, since all of the economy is interconnected.
Problem: this ignores tight linkages:
• Many of the 1.7 million food processing jobs (SIC 2011-99) would be lost
if agriculture went away.
• Other jobs (vets, crop dusters, tractor repairmen, mortgage appraisers,
fertilizer salesmen, blight insurers, agronomists, chemists, truckers,
shuckers, …) would also be lost.
• Would we have developed the world’s largest agricultural machinery
industry in the absence of the world’s largest agricultural sector?
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000
http://factory-physics.com
12
Tight Linkages (cont.)
Statistics:
• Conservative assumptions – e.g., tractor production does not require
domestic market, truckers only considered to first distribution center, no
second round multiplier effects (e.g., retail sales to farmers) considered at
all.
• 3-6 million jobs are tightly linked to agriculture.
• Since agriculture employs 3 million. This means that offshoring agriculture
would cost something like 6-8 million jobs.
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000
http://factory-physics.com
13
Linkages Between Manufacturing and Services
Direct: Manufacturing directly employs 21 million jobs
• about 20% of all jobs.
• down from about 33% in 1953 and declining.
Tightly Linked: If same “tight linkage” multiplier as agriculture holds,
manufacturing really supports 40-60 million jobs, including many service
jobs.
Impact: Offshoring manufacturing would lose many of these tightly linked
service jobs; automating to improve productivity might not.
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000
http://factory-physics.com
14
Linkages Between Manufacturing and Services
(cont.)
Services tightly linked to manufacturing:
•
•
•
•
•
•
•
•
•
design and engineering services for product and process
payroll
inventory and accounting services
financing and insuring
repair and maintenance of plant and machinery
training and recruiting
testing services and labs
industrial waste disposal
support services for engineering firms that design and service production
equipment
• trucking firms that move semi-finished goods from plant to plant
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000
http://factory-physics.com
15
Magnitudes
Production Side: Manufacturing represents roughly 50% of GNP in terms of
production.
• Manufacturing represents 24% of GNP (directly)
• Report of the President on the Trade Agreements Program estimates 25%
of GNP originates in services used as inputs by goods producing
industries.
Demand Side: Manufactured goods represent 47% of GNP (services are 33%)
in terms of final demand.
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000
http://factory-physics.com
16
Magnitudes (cont.)
$64,000 Question: Would half of the economy go away if manufacturing
were offshored?
• some jobs (advertising) could continue with foreign goods
• lost income due to loss of manufacturing jobs would have a serious
indirect multiplier effect
• lost jobs would put downward pressure on overall wages
• effect of loss of manufacturing sector on high-tech defense system?
Conclusion: A service economy may be a comforting thought in the abstract,
but in reality may be an oxymoron.
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000
http://factory-physics.com
17
The Importance of Operations
In 1979
Ford
Toyota
Labor ($/vehicle)
Capital ($/vehicle)
WIP ($/vehicle)
$2,464
$3,048
$ 536
$ 491
$1,639
$ 80
• Toyota was far more profitable than Ford in 1979.
• Costs are a function of operating decisions---planning, design, and
execution.
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000
http://factory-physics.com
18
Takeaways
•
A big chunk of the US economy is rooted in manufacturing.
•
Global competition has raised standard for
competitiveness.
•
Operations can be of major strategic importance in
remaining competitive.
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000
http://factory-physics.com
19
Modeling Matters!
I often say that when you can measure what you are speaking about,
and express it in numbers, you know something about it; but when
you cannot express it in numbers, your knowledge is of a meager
and unsatisfactory kind; it may be the beginning of knowledge, but
have scarcely, in your thoughts, advanced to the stage of Science,
whatever the matter may be.
- Lord Kelvin
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000
http://factory-physics.com
20
Why Models?
State of world:
• Data (not information!) overload
• Reliance on computers
• Allocation of responsibility (must justify decisions)
Decisions and numbers:
• Decisions are numbers
– How many distribution centers do we need?
– Capacity of new plant?
– No. workers assigned to line?
• Decisions depend on numbers
– Whether to introduce new product?
– Make or buy?
– Replace MRP with Kanban?
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000
http://factory-physics.com
21
Why Models? (cont.)
Data + Model = Information: Managers who don't understand models
either:
• Abhor analysis, lose valuable information, or
• Put too much trust in analysis, are swayed by stacks of computer output
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000
http://factory-physics.com
22
Goldratt Product Mix Problem
P
Q
15 D
10
15
$5
Product
Price
Max Weekly Sales
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000
C
A
$20
5
15
5 D
5
C
B
15
$20
P
Q
$90 $100
100
50
C
B
$20
15
10
B
A
$20
•Machines A,B,C,D
•Machines run 2400 min/week
•fixed expenses of $5000/week
http://factory-physics.com
23
Modeling Goldratt Problem
Formulation: Xp = weekly production of P, Xq = weekly production of Q
max
s.t.
(90  45) X p  (100  40) X q  5000 Weekly Profit
Time on Machine A
15 X p  10 X q  2400
15 X p  30 X q  2400
Time on Machine B
15 X p  5 X q  2400
Time on Machine C
15 X p  5 X q  2400
Time on Machine D
X p  100
Max Sales of P
X q  50
Max Sales of Q
Solution Approach:
1. Choose (feasible) production quantity of P (Xp) or Q (Xq).
2. Use remaining capacity to make other product.
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000
http://factory-physics.com
24
Unit Profit Approach
Make as much Q as possible because it is highest priced:
X q  50
(*)
10 X q  2400

X q  240
30 X q  2400

X q  80
5 X q  2400

X q  480
A
B
C,D
15 X p  10(50)  2400

X p  126.67
A
15 X p  30(50)  2400

X p  60
15 X p  5(50)  2400

X p  143.33
X p  100
(*) B
C,D
X p  60
X q  50
Z  45(60)  60(50)  5000  700
25
Bottleneck Ratio Approach
Consider bottleneck: If we set Xp =100, Xq =50, we violate capacity
constraint
15(100)  10(50)  2000  2400
15(100)  30(50)  3000  2400
15(100)  5(50)  1750  2400
15(100)  5(50)  1750  2400
(*)
A
B
C
D
Profit/Unit of Bottleneck Resource ($/minute):
Xp : 45/15 = 3
Xq : 60/30 = 2
so make as much P as possible (i.e., set Xp =100, since this does not
violate any of the capacity constraints):
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000
http://factory-physics.com
26
Bottleneck Ratio Approach (cont.)
X q  100
15 (100 )  10 X q  2400

15 (100 )  30 X q  2400

15 (100 )  5 X q  2400

X q  90
A
X q  30 (*) B
C,D
X q  180
X p  100
X q  30
Z  45 (100 )  60 (30 )  5000  1300
Outcome: This turns out to be the best we can do. But will this
approach always work?
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000
http://factory-physics.com
27
Modified Goldratt Problem
P
25
10
C
15
$5
Q
D
A
$20
Product
Price
Max Weekly Sales
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000
5
15
C
B
$20
P
Q
$90 $100
100
50
14
5
20
C
B
$20
Note: only minor
changes to times.
D
15
10
B
A
$20
•Machines A,B,C,D
•Machines run 2400 min/week
•fixed expenses of $5000/week
http://factory-physics.com
28
Modeling Modified Goldratt Problem
Formulation:
max
45 X p  60 X q  5000
Weekly Profit
s.t.
15 X p  10 X q  2400
Time on Machine A
15 X p  35 X q  2400
Time on Machine B
15 X p  5 X q  2400
Time on Machine C
25 X p  15 X q  2400
Time on Machine D
X p  100
Max Sales of P
X q  50
Max Sales of Q
Solution Approach: bottleneck method.
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000
http://factory-physics.com
29
Bottleneck Solution
Find Bottleneck:
15(100)  10(50)  2000  2400
15(100)  35(50)  3250  2400
15(100)  5(50)  1750  2400
25(100)  15(50)  3250  2400
A
(*) B
C
(*) D
Note: Both B and D are bottlenecks! (Does this seem unrealistic in a
world where line balancing is a way of life?)
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000
http://factory-physics.com
30
Possible Solutions
Make as much P as possible:
25 X p  2400

X p  96
15(96)  10 X q  2400

X q  96
15(96)  35 X q  2400

15(96)  5 X q  2400

25(96)  15 X q  2400

A
X q  27.43 B
C
X q  192
D
Xq  0
X p  96
Xq  0
Z  45(96)  60(0)  5000  680
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000
http://factory-physics.com
31
Possible Solutions (cont.)
Make as much Q as possible:
35 X q  2400

X q  87.5
so make Xq = 50 (can’t sell more than this)
15 X p  10(50)  2400

X p  126.67 A
15 X p  35(50)  2400

X p  43.33
15 X p  5(50)  2400

25 X p  15(50)  2400

B
X p  143.33 C
D
X p  66
X p  43.33
X q  50
Z  45(43.33)  60(50)  5000  50
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000
http://factory-physics.com
32
Another Solution
Make Xp =73, Xq =37: (Where in the heck did these come from? A
model!)
A 15(73)  10(37)  1465  2400
B 15(73)  35(37)  2390  2400
15(73)  5(37)  1280  2400
C
D 25(73)  15(37)  2380  2400
all constraint s satisfied
Z  45(73)  60(37)  5000  505
Conclusions:
• Modeling matters!
• Beware of simplistic solutions to complex problems!
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000
http://factory-physics.com
33