Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Stanley B. Gershwin
Massachusetts Institute of Technology
gershwin@mit.edu
web.mit.edu/manuf-sys
• stochastic processes,
• approximation methods,
• nonlinear analysis and optimization,
• statistics,
• and other mathematical fields.
Machine Buffer
M1 B M2
• The machines are unreliable — they fail at random times and
are repaired at random times.
• As a consequence, machines cause each other to be forced
idle at random times for random durations. This reduces
production.
• Goal of analysis: calculate production rate, in-process
inventory.
Method of analysis:
n=0 n=1 n=2 n=3 n=4 n=5 n=6 n=7 n=8 n=9 n=10 n=11 n=12 n=13
=N−1 =N
(α1 ,α2)
(0,0)
(0,1)
(1,0)
(1,1)
key transitions
0.8
up to a limit.
• However, increasing N
Zero buffer
increases inventory and other
costs.
0.6
20 40 60 80 100 120 140 160 180 200
N
Note: The upper limit is what would be estimated by a simple
capacity formula.
⋆ Imagine the material flow process that the observer sees entering and
the material flow process that the observer sees leaving the buffer.
• We construct a two-machine line L(i)
⋆ ie, we find machines Mu(i) and Md(i) Mu (i) B(i) Md (i)
such that an observer in its buffer will see almost the same
processes.
• The parameters are chosen as functions of the parameters of
the long line and of the parameters of the other two-machine
lines.
• A Hewlett Packard ink jet printer factory was built using MIT
techniques. The estimated economic impact was hundreds of
millions of US dollars.
• These tools have also been used in frozen food and medical
device production.
Yes.
Solution
60
Line Space
Buffer Size
40
30 Case 1 430
20
Case 2 485
Case 3 523
10
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
Buffer
B1 M2 B2 M3 B3
M1 M4
B6 M6 B5 M5 B4
Raw Material
Raw Material FG
Raw Material
Machine
Buffer
Material flow
Token (information) flow
Demand
Type 1
Type 2
Type 1 Type 2
UP/Good UP/Bad
Versions:
G B
Note: The operator does not know when the machine is in the
bad state until it has been detected.
Opinions:
• Quantity-oriented people tend to assume that
increasing a buffer increases the production rate.
Beneficial Buffer
0.8
Effective Production Rate
0.75
0.7
0.65
0 5 10 15 20 25 30 35 40 45 50
Buffer Size
Harmful Buffer
Effective Production Rate
0.5
0
0 5 10 15 20 25 30 35 40 45 50
Buffer Size
Mixed-Benefit Buffer
0.395
Effective Production Rate
0.39
0.385
0.38
0.375
0.37
0 5 10 15 20 25 30 35 40 45 50
Buffer Size
0.375
0.365
Best locations
0.355
Good Production Rate
0.345
0.335
0.325
Worst locations
0.315
0.305
0.295
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Number of Inspection Stations
Ubiquitous inspection:
I I I I I I I I
DOWN G DOWN B
• Another extension is
Distribution
of measured
parameter
Gk G k−1 G6 G5 G4 G3 G2 G1 B
Quality
Repair
Dk D k−1 D6 D5 D4 D3 D2 D1 D