Dinamica Fabril

Wallace 1. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.
com
1
Basic Factory Dynamics
Physics should be explained as simply as possible,
but no simpler.
Albert Einstein
Wallace 1. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com
HAL Case
Large Panel Line: 5roduces un5o5ulated 5rinted circuit boards
Line runs 24 hr/day (but 19.5 hrs oI 5roductive time)
Recent Performance:
through5ut 1,400 5anels 5er day (71.8 5anels/hr)
WIP 47,600 5anels
CT 34 days (663 hr at 19.5 hr/day)
customer service 75 on-time delivery
What data do we need to decide?
Is HAL lean?
3
HAL - Large Panel Line Processes
Lamination (Cores): 5ress co55er and 5re5reg into core blanks
Machining: trim cores to size
Internal Circuitize: etch circuitry into co55er oI cores
Optical Test and Repair (Internal): scan 5anels o5tically Ior deIects
Lamination (Composites): 5ress cores into multi5le layer boards
External Circuitize: etch circuitry into co55er on outside oI com5osites
Optical Test and Repair (External): scan com5osites o5tically Ior deIects
Drilling: holes to 5rovide connections between layers
Copper Plate: de5osits co55er in holes to establish connections
Procoat: a55ly 5lastic coating to 5rotect boards
Sizing: cut 5anels into boards
End of Line Test: Iinal electrical test
4
HAL Case - Science?
External Benchmarking
but other 5lants may not be com5arable
Internal Benchmarking
ca5acity data: what is utilization?
but this ignores WIP eIIects
eed relationships between WIP, 1H, C1, service!
5
Definitions
Workstations: a collection oI one or more identical machines.
Parts: a com5onent, sub-assembly, or an assembly that moves through
the workstations.
End Items: 5arts sold directly to customers; relationshi5 to constituent
5arts deIined in bill of material.
Consumables: bits, chemicals, gasses, etc., used in 5rocess but do not
become 5art oI the 5roduct that is sold.
Routing: sequence oI workstations needed to make a 5art.
Order: request Irom customer.
1ob: transIer quantity on the line.
6
Definitions (cont.)
Throughput (TH): Ior a line, through5ut is the average quantity oI
ood (non-deIective) 5arts 5roduced 5er unit time.
Work in Process (WIP): inventory between the start and end5oints oI
a 5roduct routing.
Raw Material Inventory (RMI): material stocked at beginning oI
routing.
Crib and Finished Goods Inventory (FGI): crib inventory is
material held in a stock5oint at the end oI a routing; FGI is material
held in inventory 5rior to shi55ing to the customer.
Cycle Time (CT): time between release oI the job at the beginning oI
the routing until it reaches an inventory 5oint at the end oI the
routing.
7
Factory Physics
Definition: A manuIacturing system is a oal-oriented network

oI processes through which parts flow.
Structure: Plant is made u5 oI routings (lines), which in turn are
made u5 oI 5rocesses.
Focus: actory Physics
is concerned with the network and Ilows

at the routing (line) level.
8
Parameters
Descriptors of a Line:
1) Bottleneck Rate (r
b
): #ate (5arts/unit time or jobs/unit time)
oI the 5rocess center having the highest lonterm utilization.
2) Raw Process Time (1
0
): $um oI the lonterm averae
5rocess times oI each station in the line.
) Congestion Coefficient (-): A unitless measure oI
congestion.
Zero variability case, - 0.
'Practical worst case, - 1.
'Worst 5ossible case, -
0
.
ote: e ont use - 6uantitatively,
but point it out to reconi:e that lines
ith same r
b
and T
0
can behave very
differently.
9
Parameters (cont.)
Relationship:
Critical WIP (W
0
): WIP level in which a line having no
congestion would achieve maximum through5ut (i.e., r
b
)
with minimum cycle time (i.e., T
0
).
0
r
b
T
0
10
The Penny Fab
Characteristics:
Four identical tools in series.
Each takes hours 5er 5iece (5enny).
No variability.
CONWIP job releases.
Parameters:
r
b

T
0

0

-
0.5 pennies/hour
8 hours
0.5 L 8 4 pennies
0 (no variability, best case conditions)
11
The Penny Fab The Penny Fab
1
The Penny Fab (WIP1) The Penny Fab (WIP1)
Time 0 hours
13
Time hours
14
Time 4 hours
15
Time 6 hours
16
Time 8 hours
17
Time 10 hours
18
Time 1 hours
19
Time 14 hours
0
Time 16 hours
1
Penny Fab Performance
WIP TH CT
TH
L
CT
1 0.15 8 1

3
4
5
6


Time 0 hours
3
Time hours
4
Time 4 hours
5
Time 6 hours
6
Time 8 hours
7
Time 10 hours
8
Time 1 hours
9
Time 14 hours
30
Time 16 hours
31
Time 18 hours
3
WIP TH CT
TH
L
CT
1 0.15 8 1
0.50 8
3
4
5
6

33
Time 0 hours
34
Time hours
35
Time 4 hours
36
Time 6 hours
37
Time 8 hours
38
Time 10 hours
39
Time 1 hours
40
Time 14 hours
41
WIP TH CT
TH
L
CT
1 0.15 8 1
0.50 8
3 0.375 8 3
4 0.500 8 4
5
6

4
Time 0 hours
43
Time hours
44
Time 4 hours
45
Time 6 hours
46
Time 8 hours
47
Time 10 hours
48
Time 1 hours
49
WIP TH CT
TH
L
CT
1 0.15 8 1
0.50 8
3 0.375 8 3
4 0.500 8 4
5 0.500 10 5
6 0.500 1 6

50
TH vs. WIP: Best Case
0
0.1
0.2
0.
0.4
0.5
0.6
0 1 2 4 5 6 7 8 9 10 11 12
WIP
T
H
r r
bb
WW
00
1/T 1/T
00
51
CT vs. WIP: Best Case
0
2
4
6
8
10
12
14
16
18
20
22
24
26
0 1 2 4 5 6 7 8 9 10 11 12
WIP
C
T
TT
00
WW
00
1/r 1/r
bb
5
Best Case Performance
Best Case Law: 1he minimum cycle time (C1
best
) for a iven
WIP level, w, is iven by
1he maximum throuhput (1H
best
) for a iven WIP level, w is
iven by,
otherwise.
iI
, /
,
CT
0 0
best

r
T
b
otherwise.
iI
,
, /
TH
0 0
best

r
T
b
53
Best Case Performance (cont.)
Example: For Penny Fab, r
b
0.5 and T
0
8, so
0
0.5 L 8 4,
which are exactly the curves we 5lotted.
otherwise.
4 iI
,
, 8
CT
best
otherwise.
4 iI
, 5 . 0
, 8 /
TH
best

54
A Manufacturing Law
Little's Law: 1he fundamental relation between WIP, C1, and
1H over the lon-term is:
Insights:
Fundamental relationshi5
$im5le units transIormation
DeIinition oI cycle time (CT WIP/TH)
CT TH WIP L
hr
hr
parts
parts L
55
Penny Fab Two
10 hr
hr
5 hr 3 hr
56
Penny Fab Two
$tation
Number
Number oI
Machines
Process
Time
$tation
#ate
1 1 hr j/hr
5 hr j/hr
3 6 10 hr j/hr
4 3 hr j/hr
r
b
T
0

0

0.5
0.4
0.6
0.67
0.4 p/hr 20 hr 8 pennies
57
Penny Fab Two Simulation (Time0)
10 hr
hr
5 hr 3 hr
2
58
10 hr
hr
5 hr 3 hr
4
7
59
10 hr
hr
5 hr 3 hr
6
7
9
60
10 hr
hr
5 hr 3 hr
8
7
9
61
10 hr
hr
5 hr 3 hr
8
12
9
17
6
10 hr
hr
5 hr 3 hr
10
12
9
17
63
10 hr
hr
5 hr 3 hr
10
12
14
17
19
64
10 hr
hr
5 hr 3 hr
12
12
14
17
19
65
10 hr
hr
5 hr 3 hr
14
17
14
17
19
22
66
10 hr
hr
5 hr 3 hr
16
17
19
17
19
22
24
67
10 hr
hr
5 hr 3 hr
17
19
17
19
22
24
68
10 hr
hr
5 hr 3 hr
22
19
27
19
22
24
20
69
10 hr
hr
5 hr 3 hr
22
24
27
29
22
24
20
22
70
10 hr
hr
5 hr 3 hr
22
24
27
29
22
24
22
22
ote: ob ill arrive at
bottleneck ust in time
to prevent starvation.
71
10 hr
hr
5 hr 3 hr
27
24
27
29
2
24
25
24
ote: ob ill arrive at
bottleneck ust in time
to prevent starvation.
7
10 hr
hr
5 hr 3 hr
27
29
27
29
2
4
25
27
And so on..
Bottleneck will just
stay busy; all others
will starve periodically
73
Worst Case
Observation: The Best Case yields the minimum cycle time and
maximum through5ut Ior each WIP level.
Question: What conditions would cause the maximum cycle time
and minimum through5ut?
Experiment:
set average 5rocess times same as Best Case (so r
b
and T
0
unchanged)
Iollow a marked job through system
imagine marked job ex5eriences maximum 6ueuein
74
Worst Case Penny Fab Worst Case Penny Fab
Time 0 hours
75
Time 8 hours
76
Time 16 hours
77
Time 4 hours
78
Time 3 hours
ote:
C1 32 hours
4L 8 w1
1H 4/32 1/8 1/1

79
TH vs. WIP: Worst Case
0
0.1
0.2
0.
0.4
0.5
0.6
0 1 2 4 5 6 7 8 9 10 11 12
WIP
T
H
r r
bb
WW
00
1/T 1/T
00
Best Case Best Case
Worst Case Worst Case
80
CT vs. WIP: Worst Case
0
4
8
12
16
20
24
28
2
0 1 2 4 5 6 7 8 9 10 11 12
WIP
C
T
TT
00
WW
00
Best Case Best Case
81
Worst Case Performance
Worst Case Law: 1he worst case cycle time for a iven WIP
level, w, is iven by,
CT
worst
w 1
0
1he worst case throuhput for a iven WIP level, w, is iven
by,
TH
worst
1 / 1
0
Randomness? one - perfectly predictable, but bad!
8
Practical Worst Case
Observation: There is a BIG GAPbetween the Best Case and
Worst Case 5erIormance.
Question: Can we Iind an intermediate case that:
divides 'good and 'bad lines, and
is com5utable?
Experiment: consider a line with a given r
b
and T
0
and:
single machine stations
balanced lines
variability such that all WIP conIigurations (states) are equally
likely
83
PWC Example - jobs, 4 stations jobs, 4 stations
State Vector State Vector
1 11 (1,0,2,0)
2 12 (0,1,2,0)
1 (0,0,2,1)
4 14 (1,0,0,2)
5 (2,1,0,0) 15 (0,1,0,2)
6 (2,0,1,0) 16 (0,0,1,2)
7 (2,0,0,1) 17
8 (1,2,0,0) 18
9 (0,2,1,0) 19
10 (0,2,0,1) 20
clumped
up states
spread
out states
ote: averae WIP at any station is 15/2 .75,
so jobs are spread evenly between stations.
84
Practical Worst Case
Let w jobs in system, no. stations in line, and t
process time at all stations:
C1(sinle) (1 + (w-1)/) t
C1(line) [1 + (w-1)/( t
t + (w-1)t
1
+ (w-1)/r
b
1H WIP/C1
[w/(w+W
-1)(r
b
rom Little's Law rom Little's Law
85
Practical Worst Case Performance
Practical Worst Case Definition: 1he practical worst case
(PWC) cycle time for a iven WIP level, w, is iven by,
1he PWC throuhput for a iven WIP level, w, is iven by,
where W
is the critical WIP.

b
r
w
1
1
CT
0 PWC

,
1
TH
0
PWC b
r
w W
w


86
THvs. WIP: Practical Worst Case
0
0.1
0.2
0.
0.4
0.5
0.6
0 1 2 4 5 6 7 8 9 10 11 12
WIP
T
H
r r
bb
WW
00
1/T 1/T
00
Best Case Best Case
PWC PWC
Good (lean) Good (lean)
Bad (fat) Bad (fat)
87
CTvs. WIP: Practical Worst Case
0
4
8
12
16
20
24
28
2
0 1 2 4 5 6 7 8 9 10 11 12
WIP
C
T
TT
00
WW
00
Best Case Best Case
PWC PWC
Bad (fat) Bad (fat)
Good Good
(lean) (lean)
88
0
0.1
0.2
0.
0.4
0.5
0 2 4 6 8 10 12 14 16 18 20 22 24 26
WIP
TH
Penny Fab Two Performance
Worst Case
Best Case
1/T
0
r
b
W
0
ote: process
times in P2
have var e6ual
to PWC.
But. unlike
PWC, it has
unbalanced
line and multi
machine
stations.
89
Penny Fab Two Performance (cont.)
0
10
20
0
40
50
60
70
80
0 2 4 6 8 10 12 14 16 18 20 22 24 26
WIP
CT
Worst Case
Best Case
T
0
1/r
b
W
0
90
Back to the HAL Case - Capacity Data
Process Rate (p/hr) Time (hr)
Lamination 191.5 1.
Machining 186. 5.9
Internal Circuitize 150.5 6.9
O5tical Test/#e5air - Int 157.8 5.6
Lamination Com5osites 191.5 1.
External Circuitize 150.5 6.9
O5tical Test/#e5air - Ext 157.8 5.6
Drilling 185.9 10.0
Co55er Plate 136.4 1.5
Procoat 146. .
$izing 16.5 .4
EOL Test 169.5 1.8
r
b
, T
0
16.5 33.1

91
HAL Case - Situation
Critical WIP: W
0
r
b
T
0
16.5 L 33.9 4,187
Actual Values:
CT 34 days 816 hours (at 4 hr/day)
WIP 37,000 5anels
TH 45.8 5anels/hour
Conclusions:
Through5ut is 36 oI ca5acity
WIP is 15 times critical WIP
CT is 4.6 times raw 5rocess time
9
HAL Case - Analysis
WIP Required for PWC to Achieve TH 0.6r
b
?
354 , ) 1 187 , 4 (
64 . 0
36 . 0
) 1 (
64 . 0
36 . 0
36 . 0
1
0
0

r r

TH
b b
uch lower than
actual WIP!
Conclusion: actual system is much worse than PWC!
4 . 105 5 . 16
1 187 , 4 400 , 37
400 , 37
1
0
b
r

TH
uch
hiher
than actual
1H!
TH Resulting from PWC with WIP 47,600?
93
HAL Internal Benchmarking Outcome
Lean" Reion
at" Reion
0.0
0.0
40.0
60.0
80.0
100.0
10.0
0 10,000 0,000 30,000 40,000 50,000
WIP
T
h
r
o
u
g
h
p
u
t

(
p
a
n
e
l
s
/
h
o
u
r
)
Best
Worst
PWC
Current
TH 45.8
WIP 37,000
94
Labor Constrained Systems
Motivation: 5erIormance oI some systems are limited by labor or
a combination oI labor and equi5ment.
Full Flexibility with Workers Tied to 1obs:
WIP limited by number oI workers (n)
ca5acity oI line is n/T
0
Best case achieves ca5acity and has workers in 'zones
ample capacity case also achieves Iull ca5acity with '5ick and run
5olicy
95
Labor Constrained Systems (cont.)
Full Flexibility with Workers Not Tied to 1obs:
TH de5ends on WIP levels
TH
CW
(n) A TH() A TH
CW
()
need 5olicy to direct workers to jobs (Iocus on downstream is
eIIective)
Agile Workforce Systems
bucket brigades
kanban with shared tasks
worksharing with overla55ing zones
many others
96
Factory Dynamics Takeaways
Performance Measures:
through5ut
WIP
cycle time
service
Range of Cases:
best case
5ractical worst case
worst case
Diagnostics:
sim5le assessment based on r
b
, T
0
, actual WIP,actual TH
evaluate relative to 5ractical worst case

Dinamica Fabril

Uploaded by

Copyright:

Available Formats

Dinamica Fabril

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Dinamica Fabril

Uploaded by

Copyright:

Available Formats

Wallace 1. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.

Definition: A manuIacturing system is a oal-oriented network

is concerned with the network and Ilows

The Penny Fab (WIP2) The Penny Fab (WIP2)

1H 4/32 1/8 1/1

Wallace 1. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com

is the critical WIP.

Wallace 1. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com

You might also like