Transcript Variability Basics God does not play dice with the universe.
Variability Basics
God does not play dice with the universe.
– Albert Einstein
Stop telling God what to do.
– Niels Bohr
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com
1
Variability Makes a Difference!
Little’s Law:
TH = WIP/CT, so same throughput can be obtained with large WIP, long CT or small WIP, short CT. The difference?
Variability!
Penny Fab One
:
achieves full TH (0.5 j/hr) at WIP=W 0 =4 jobs if it behaves like Best Case, but requires WIP=27 jobs to achieve 95% of capacity if it behaves like the Practical Worst Case. Why?
Variability!
2
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com
Tortise and Hare Example
Two machines:
• subject to same workload: 69 jobs/day (2.875 jobs/hr) • subject to unpredictable outages (availability = 75%)
Hare X19:
• long, but infrequent outages
Tortoise 2000:
• short, but more frequent outages
Performance:
Hare X19 is substantially worse on all measures than Tortoise 2000. Why?
Variability!
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com
3
Variability Views
Variability:
• Any departure from uniformity • Random versus controllable variation
Randomness:
• Essential reality?
• Artifact of incomplete knowledge?
• Management implications: robustness is key
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com
4
Probabilistic Intuition
Uses of Intuition:
• driving a car • throwing a ball • mastering the stock market
First Moment Effects:
• Throughput increases with machine speed • Throughput increases with availability • Inventory increases with lot size • Our intuition is good for first moments
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com
g
5
Probabilistic Intuition (cont.)
Second Moment Effects:
• Which is more variable – processing times of parts or batches?
• Which are more disruptive – long, infrequent failures or short frequent ones?
• Our intuition is less secure for second moments • Misinterpretation – e.g., regression to the mean
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com
6
Variability
Definition:
Variability is anything that causes the system to depart from regular, predictable behavior.
Sources of Variability:
• setups • machine failures • materials shortages • yield loss • rework • operator unavailability • workpace variation • differential skill levels • engineering change orders • customer orders • product differentiation • material handling 7
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com
Measuring Process Variability
t e
mean process time of a job
σ e
standard deviation of process time
c e
e t e
coefficien t of variation , CV
Note: we often use the “squared coefficient of variation” (SCV), c
e 2 © Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com
8
Variability Classes in Factory Physics
®
0 Low variability (LV) Moderate variability (MV) 0.75
1.33
High variability (HV)
Effective Process Times:
• •
actual
process times are generally LV
effective
process times include setups, failure outages, etc.
• HV, LV, and MV are all possible in effective process times
Relation to Performance Cases:
For balanced systems • MV – Practical Worst Case • LV – between Best Case and Practical Worst Case • HV – between Practical Worst Case and Worst Case
c e
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com
9
Measuring Process Variability – Example
Trial 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
t e s e c e
Class
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000
Machine 1 22 25 23 26 24 28 21 30 24 28 27 25 24 23 22 25.1
2.5
0.1
LV Machine 2 5 6 5
35
7
45
6 6 5 4 7
50
6 6 5 13.2
15.9
1.2
MV
http://www.factory-physics.com
Machine 3 5 6 5
35
7
45
6 6 5 4 7
500
6 6 5 43.2
127.0
2.9
HV
Question: can we measure c e this way?
Answer: No! Won’t consider “rare” events properly.
10
Natural Variability
Definition:
variability without explicitly analyzed cause
Sources:
• operator pace • material fluctuations • product type (if not explicitly considered) • product quality
Observation:
natural process variability is usually in the LV category.
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com
11
Down Time – Mean Effects
Definitions:
c t
0 base process time base process time coefficien t of variabili ty 0
m r
0 1 base capacity (rate, e.g., parts/hr)
t
0 mean time to failure
f m r c r
mean time to repair coefficent of variabili ty of repair tim es (
r
/
m r
)
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com
12
Down Time – Mean Effects (cont.)
Availability:
Fraction of time machine is up
A
m f m f
m r
Effective Processing Time and Rate:
r e
Ar
0
t e
t
0 /
A
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com
13
Totoise and Hare - Availability
Hare X19:
t
0 0 = 15 min = 3.35 min
c
0 = 0 /
t
0 = 3.35/15 = 0.05
m
f = 12.4 hrs (744 min)
m
r = 4.133 hrs (248 min)
c
r = 1.0
Tortoise:
t
0 0 = 15 min = 3.35 min
c
0 = 0 /
t
0 = 3.35/15 = 0.05
m
f = 1.9 hrs (114 min)
m
r = 0.633 hrs (38 min)
c
r = 1.0
Availability:
A
=
m f m f
m r
744 744
248
0 .
75
A
=
m f m f
m r
114 114
38
0 .
75
No difference between machines in terms of availability.
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com
14
Down Time
–
Variability Effects
Effective Variability:
t e
t
0 /
A σ e
2
A
0 2 (
m r
2
r
2 )( 1
Am r A
)
t
0
Variability depends on repair times
Conclusions:
c e
2
t e
2
e
2
c
0 2 ( 1
c r
2 )
A
( 1
A
)
m r t
0
in addition to availability
• Failures inflate mean, variance, and CV of effective process time • Mean (
t
e ) increases proportionally with 1/
A
• SCV (
c e
2 ) increases proportionally with
m r
• SCV (
c e
2 ) increases proportionally in
c
r 2 • For constant availability (
A
), long infrequent outages increase SCV more than short frequent ones 15
http://www.factory-physics.com
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000
Tortoise and Hare - Variability
Hare X19: Tortoise 2000
t
e =
t
0
A
15 0 .
75
20 min
t
e =
t
0
A
15 0 .
75
20 min
c
e 2 =
c
0 2
( 1
c r
2 )
A
( 1
A
)
m r t
0
0 .
05
( 1
1 ) 0 .
75 ( 1
0 .
75 ) 248 15
6 .
25 high variabilit y
c
e 2 =
c
0 2
( 1
c r
2 )
A
( 1
A
)
m r t
0
0 .
05
( 1
1 ) 0 .
75 ( 1
38 0 .
75 ) 15
1 .
0 moderate variabilit y
Hare X19 is much more variable than Tortoise 2000!
16
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com
Setups
–
Mean and Variability Effects
Analysis:
N s
t s s c s
average no.
jobs between setups average setup duration std.
dev.
of setup time
s t s t e σ e
2
c e
2
t
0 0 2
t s N s
N s s
2
t e
2
e
2
N s N s
1 2
t s
2
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com
17
Setups
–
Mean and Variability Effects (cont.)
Observations:
• Setups increase mean
and
variance of processing times.
• Variability reduction is one benefit of flexible machines.
• However, the interaction is complex.
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com
18
Setup
–
Example
Data:
• Fast, inflexible machine – 2 hr setup every 10 jobs
t
0 1 hr
N s t r e e
10 jobs/setup
t s
2 hrs
t
0 1 /
t
t s
e
/
N s
1 /( 1 1 2 / 2 / 10 ) 10 1 .
2 hrs 0 .
8333 jobs/hr • Slower, flexible machine – no setups
t
0 1.2
hrs
r e
1 /
t
0 1 / 1 .
2 0 .
833 jobs/hr
Traditional Analysis?
No difference!
http://www.factory-physics.com
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000
19
Setup
–
Example (cont.)
Factory Physics ® Approach:
Compare mean
and
variance • Fast, inflexible machine
–
2 hr setup every 10 jobs
t
0 1 hr
c
0 2
N s
0 .
0625 10 jobs/setup
t s
2 hrs
c s
2 0 .
0625
t e r e
t
0 1 /
t
t s e
/
N s
1 /( 1 1 2 / 2 / 10 ) 10 1 .
2 hrs 0 .
8333 jobs/hr
σ e
2 0 2
t s
2
c
2
s N s
N s N s
2 1 0 .
4475
c e
2 0 .
31
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com
20
Setup
–
Example (cont.)
• Slower, flexible machine
–
t
0 1 .
2 hrs no setups
c
0 2 0 .
25
r c e
2
e
1 /
t
0
c
0 2 1 / 0 .
25 1 .
2 0 .
833 jobs/hr
Conclusion:
Flexibility can reduce variability.
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com
21
Setup
–
Example (cont.)
New Machine:
Consider a third machine same as previous machine with setups, but with shorter, more frequent setups
N s t s
5 jobs/setup 1 hr
Analysis:
r e σ e
2 1 /
t e
0 2 1 /( 1
t s
2
c s
2
N s
1 / 5 )
N s N
0 .
833 jobs/hr
s
1 2 0 .
2350
c e
2 0 .
16
Conclusion:
Shorter, more frequent setups induce less variability.
22
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com
Other Process Variability Inflators
Sources:
• operator unavailability • recycle • batching • material unavailability • et cetera, et cetera, et cetera
Effects:
• inflate
t e
• inflate
c e
Consequences:
Effective process variability can be LV, MV,or HV.
23
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com
Illustrating Flow Variability
Low variability arrivals
smooth!
High variability arrivals
bursty!
t t © Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com
24
Measuring Flow Variability
t a
mean time between arrivals
r a
1
t a
arrival rate
a
standard deviation of time between arrivals
c a
a t a
coefficien t of variation of interarriv al times
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com
25
c a
2 (
i
)
Propagation of Variability
c e
2 (
i
)
c d
2 (
i
) =
c a
2 (
i+
1)
i
Single Machine Station:
c d
2
u
2
c e
2 ( 1
u
2 )
c a
2 where
u
is the station utilization given by
u
=
r a t e
Multi-Machine Station:
c
2
d
1 ( 1
u
2 )(
c a
2 1 )
u
2
m
(
c e
2 1 ) where
m
is the number of (identical) machines and
u
r a t e m i+1
departure var depends on arrival var and process var
26
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com
LV
Propagation of Variability – High Utilization Station
HV HV HV HV HV LV LV LV HV LV
Conclusion: flow variability out of a high utilization station is determined primarily by process variability at that station.
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com
LV
27
LV
Propagation of Variability – Low Utilization Station
HV LV HV HV HV LV LV LV HV LV
Conclusion: flow variability out of a low utilization station is determined primarily by flow variability into that station.
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com
HV
28
Variability Interactions
Importance of Queueing:
• manufacturing plants are
queueing networks
• queueing and waiting time comprise majority of cycle time
System Characteristics:
• Arrival process • Service process • Number of servers • Maximum queue size (blocking) • Service discipline (FCFS, LCFS, EDD, SPT, etc.) • Balking • Routing • Many more
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com
29
Kendall's Classification
A/B/C
A: arrival process B: service process C: number of machines
A
M: exponential (Markovian) distribution G: completely general distribution D: constant (deterministic) distribution.
Queue B Server C
30
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com
Queueing Parameters
r a
= the rate of arrivals in customers (jobs) per unit time (
t a
= 1/
r a
arrivals). = the average time between
c a
= the CV of inter-arrival times.
m
= the number of machines.
r e
= the rate of the station in jobs per unit time =
m/t e
.
c e
= the CV of
effective
process times.
Note: a station can be described with 5 parameters.
u
= utilization of station =
r a /r e
.
31
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com
Queueing Measures
Measures:
CT
q
= the expected waiting time spent in queue. CT = the expected time spent at the process center, i.e., queue time plus process time. WIP = the average WIP level (in jobs) at the station. WIP
q
= the expected WIP (in jobs) in queue.
Relationships:
CT = CT
q
WIP =
r a
+
t e
CT WIP
q
=
r a
CT
q
Result:
If we know CT
q
, we can compute WIP, WIP
q
, CT.
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com
32
The G/G/1 Queue
Formula:
CT
q
V
U
t
c a
2
c e
2 2
u
1
u
t e
Observations:
• Useful model of single machine workstations • • Separate terms for variability, utilization, process time.
• CT
q
(and other measures) increase with
c a
2 and
c e
2 • Flow variability, process variability, or both can combine to inflate queue time.
Variability causes congestion!
33
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com
The G/G/m Queue
Formula:
CT
q
V
U
t
c a
2
c e
2 2
u
2 (
m
1 ) 1
m
( 1
u
)
t e
Observations:
• Useful model of multi-machine workstations •
Extremely
general.
• Fast and accurate.
• Easily implemented in a spreadsheet (or packages like MPX).
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com
34
MEASURE:
Arrival Rate (parts/hr) Arrival CV Natural Process Time (hr) Natural Process SCV Number of Machines MTTF (hr) MTTR (hr) Availability Effective Process Time (failures only) Eff Process SCV (failures only) Batch Size Setup Time (hr) Setup Time SCV Arrival Rate of Batches Eff Batch Process Time (failures+setups) Eff Batch Process Time Var (failures+setups) Eff Process SCV (failures+setups) Utilization Departure SCV Yield Final Departure Rate Final Departure SCV Utilization Throughput Queue Time (hr) Cycle Time (hr) Cumulative Cycle Time (hr) WIP in Queue (jobs) WIP (jobs) Cumulative WIP (jobs)
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 VUT Spreadsheet STATION: r a c a 2 A t e ' c e 2 ' k t s c s 2 t 0 c 0 2 m m f m r r a /k t e = kt 0 /A+t s k*
0 2 /A 2 + 2m r (1-A)kt 0 /A+
s 2 c e 2 u c d 2 y r a *y yc d 2 +(1-y) u TH CT q CT q +t e
i (CT q (i)+t e (i)) r a CT q r a CT
i (r a (i)CT(i)) 0.936
100 0.000
1.000
0.100
9.090
0.773
0.009
0.909
0.181
0.980
9.800
0.198
0.909
9.800
45.825
54.915
54.915
458.249
549.149
549.149
1 10.000
1.000
0.090
0.500
1 200 2 0.990
0.091
http://www.factory-physics.com
0.936
100 0.500
1.000
0.098
9.590
1.023
0.011
0.940
0.031
0.950
9.310
0.079
0.940
9.310
14.421
24.011
78.925
141.321
235.303
784.452
2 9.800
0.181
0.090
0.500
1 200 2 0.990
0.091
6.729
100 0.500
1.000
0.093
10.380
6.818
0.063
0.966
0.061
0.950
8.845
0.108
0.966
8.845
14.065
24.445
103.371
130.948
227.586
1012.038
3 9.310
0.031
0.095
0.500
1 200 8 0.962
0.099
2.209
100 0.000
1.000
0.088
9.180
1.861
0.022
0.812
0.035
0.900
7.960
0.132
0.812
7.960
1.649
10.829
114.200
14.587
95.780
1107.818
4 8.845
0.061
0.090
0.500
1 200 4 0.980
0.092
2.209
100 0.000
1.000
0.080
9.180
1.861
0.022
0.731
0.028
0.950
7.562
0.077
0.731
7.562
0.716
9.896
124.096
5.700
78.773
1186.591
5 7.960
0.035
0.090
0.500
1 200 4 0.980
0.092
35
Effects of Blocking
VUT Equation:
• characterizes stations with infinite space for queueing • useful for seeing what will happen to WIP, CT without restrictions
But real world systems often constrain WIP:
• physical constraints (e.g., space or spoilage) • logical constraints (e.g., kanbans)
Blocking Models:
• estimate WIP and TH for given set of rates, buffer sizes • much more complex than non-blocking (open) models, often require simulation to evaluate realistic systems 36
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com
The M/M/1/b Queue
Infinite raw materials 1
B buffer spaces
2 Note: there is room
for b=B+2 jobs in system, B in the buffer and one at each station .
Model of Station 2
WIP
(
M
/
M
/ 1 /
b
)
u
1
u
(
b
1 )
u b
1 1
u b
1
Goes to u/(1-u) as b
Always less than WIP(M/M/1)
TH
(
M
/
M
/ 1 /
b
) 1
u b
1
u b
1
r a CT
(
M
/
M
/ 1 /
b
)
WIP
(
M TH
(
M
/
M
/
M
/ 1 /
b
/ 1 /
b
) ) where
u
t e
( 2 ) /
t e
( 1 )
Goes to r a as b
Always less than TH(M/M/1) Little’s law
Note: u>1 is possible; formulas valid for u
1
37
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com
Blocking Example
t e (1)=21 t e (2)=20 B=2
u
t e
( 2 ) /
t e
( 1 ) 20 / 21 0 .
9524
WIP
(
M TH
(
M
/ /
M
/ 1 )
u
1
u M
/ 1 )
r a
20 jobs 1 /
t e
( 1 ) 1 / 21 0 .
0476 job/min
M/M/1/b system has less WIP and less TH than M/M/1 system
TH(M/M/
1
/b)
1
-u b
1
-u b
1
r a
1 0 .
9524 4 1 0 .
9524 5 1 21 0 .
039 job/min
WIP
(
M
/
M
/ 1 /
b
)
u
1
u
(
b
1 )
u b
1 1
u b
1 20 5 ( 1 0 .
9524 0 .
9524 5 ) 5 1 .
8954 jobs
18% less TH 90% less WIP
38
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com
Seeking Out Variability
General Strategies:
• look for long queues (Little's law) • look for blocking • focus on high utilization resources • consider both flow and process variability • ask “why” five times
Specific Targets:
• equipment failures • setups • rework • operator pacing • anything that prevents regular arrivals and process times
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com
39
Variability Pooling
Basic Idea:
the CV of a sum of independent random variables decreases with the number of random variables.
Example (Time to process a batch of parts):
t
0
c
0 0 time to process single part standard deviation of time to process single part 0 CV of time to process single part
t
0
nt
0
t
0 (
batch
) 0 2 (
batch
)
n
0 2
c
0 2 (
batch
) 0 2 (
batch
)
t
0 2 (
batch
)
n
2 0
n
2
t
0 2 0 2
nt
0 2
c
0 2
n
c
0 (
batch
)
c
0
n
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com
40
Safety Stock Pooling Example
• • • • • •
PC’s consist of 6 components (CPU, HD, CD ROM, RAM, removable storage device, keyboard) 3 choices of each component: 3 6 = 729 different PC’s Each component costs $150 ($900 material cost per PC) Demand for all models is normally distributed with mean 100 per year, standard deviation 10 per year Replenishment lead time is 3 months, so average demand during LT is
= 25 for computers and
= 25(729/3) = 6075 for components Use base stock policy with fill rate of 99%
41
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com
Pooling Example - Stock PC’s
Base Stock Level for Each PC:
R =
+ z s
= 25 + 2.33(
25) = 37
cycle stock safety stock
On-Hand Inventory for Each PC:
I(R) = R -
+ B(R)
R -
= z s
= 37 - 25 = 12 units
Total (Approximate) On-Hand Inventory :
12
729
$900 = $7,873,200
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com
42
Pooling Example - Stock Components
Necessary Service for Each Component:
S = (0.99) 1/6 = 0.9983
z s = 2.93
Base Stock Level for Each Component:
R =
+ z s
= 6075 + 2.93(
cycle stock safety stock
6075) = 6303
On-Hand Inventory Level for Each Component:
I(R) = R -
+ B(R)
R -
= z s
= 6303-6075 = 228 units
Total Safety Stock:
228
18
$150 = $615,600
92% reduction!
43
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com
Basic Variability Takeaways
Variability Measures:
• CV of effective process times • CV of interarrival times
Components of Process Variability
• failures • setups • many others - deflate capacity
and
inflate variability • long infrequent disruptions worse than short frequent ones
Consequences of Variability:
• variability causes congestion (i.e., WIP/CT inflation) • variability propagates • variability and utilization interact • pooled variability less destructive than individual variability
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com
44