MIT OpenCourseWare

____________
http://ocw.mit.edu

2.830J / 6.780J / ESD.63J Control of Manufacturing Processes (SMA 6303)

Spring 2008

For information about citing these materials or our Terms of Use, visit: ________________
http://ocw.mit.edu/terms.
 Control of
Manufacturing Processes
Subject 2.830/6.780/ESD.63
Spring 2008
Lecture #9

Advanced and Multivariate SPC

March 6, 2008

Manufacturing 1
 Agenda
• Conventional Control Charts
– Xbar and S

• Alternative Control Charts

– Moving average
– EWMA
– CUSUM

• Multivariate SPC

Manufacturing 2
 Xbar Chart
Process Model: x ~ N(5,1), n = 9

6.0 UCL=6.000

5.5
Mean of x

5.0 µ0=5.000

4.5

4.0 LCL=4.000

1 2 3 4 5 6 7 8 9 10 11 12 13
Sample

• Is process in control?

Manufacturing 3
 Run Data (n=9 sample size)

11

9
8
Run Chart of x

5 Avg=4.91
4
3

1
0
0 9 18 27 36 45 54 63 72 81 90 99
Run

• Is process in control?

Manufacturing 4
 S Chart

Standard Deviation of x 3.0

2.5

2.0
UCL=1.707
1.5

1.0 µ0=0.969

0.5
LCL=0.232
0.0
1 2 3 4 5 6 7 8 9 10 11 12 13
Sample

• Is process in control?

Manufacturing 5
 Alternative Charts: Running Averages
• More averages/Data
• Can use run data alone and
average for S only
• Can use to improve resolution
of mean shift
j+n
1
xRj = ∑ xi
n i=j Running Average
n measurements
at sample j 1 j +n
SR j 2 = ∑ (x
n − 1 i= j i
− x ) 2
Rj Running Variance

Manufacturing 6
 Simplest Case: Moving Average
• Pick window size (e.g., w = 9)

9
8
7
Moving Avg of x

6 UCL
5 µ0=5.00
4 LCL
3
2
1
8 16 24 32 40 48 56 64 72 80 88 96
Sample
Manufacturing 7
 General Case: Weighted Averages
y j = a1 x j −1 + a2 x j −2 + a3 x j −3 + ...
• How should we weight measurements?
– All equally? (as with Moving Average)
– Based on how recent?
• e.g. Most recent are more relevant than less
recent?

Manufacturing 8
Consider an Exponential Weighted Average

0.25

Define a weighting function

Wt − i = r (1 − r )
0.2 i

0.15

Exponential Weights
0.1

0.05

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Manufacturing 9
Exponentially Weighted Moving Average:
(EWMA)

Ai = rxi + (1 − r)Ai −1 Recursive EWMA

time
⎛ σ x2 ⎞ ⎛ r ⎞
σA = ⎜ ⎟⎝
⎝ n ⎠ 2 − r⎠
1 [
− (1 − r )2t
]
σx ⎛ r ⎞
2
σA =
n ⎝ 2 − r⎠
UCL, LCL = x ± 3σ A
for large t

Manufacturing 10
 Effect of r on σ multiplier
0.9

0.8
plot of (r/(2-r)) vs. r
0.7

0.6

0.5

0.4
wider control limits
0.3

0.2

0.1

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
r

Manufacturing 11
 SO WHAT?
• The variance will be less than with xbar,
σx ⎛ r ⎞ ⎛ r ⎞
σA = = σx
n ⎝ 2 − r⎠ ⎝ 2 − r⎠

• n=1 case is valid

• If r=1 we have “unfiltered” data
– Run data stays run data
– Sequential averages remain
• If r<<1 we get long weighting and long delays
– “Stronger” filter; longer response time

Manufacturing 12
 EWMA vs. Xbar
1

0.9 r=0.3
0.8 Δμ = 0.5 σ
0.7

xbar
0.6
EWMA
UCL EWMA
0.5 LCL EWMA
grand mean
UCL
0.4
LCL

0.3

0.2

0.1

0
0 50 100 150 200 250 300

Manufacturing 13
 Mean Shift Sensitivity
EWMA and Xbar comparison
1.2
xbar

1 EWMA

UCL
0.8 EWMA

LCL
EWMA

0.6 3/6/03
Grand
Mean

UCL
0.4
LCL

Mean shift = .5 σ
0.2
n=5
r=0.1
0
1 5 9 13 17 21 25 29 33 37 41 45 49 14
Manufacturing
 Effect of r
1

0.9 xbar

0.8 EWMA

0.7 UCL
EWMA

0.6 LCL
EWMA

0.5 Grand
Mean
0.4 UCL

0.3 LCL

0.2

0.1

0
r=0.3
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49

Manufacturing 15
 Small Mean Shifts

• What if Δμx is small with respect to σx ?

• But it is “persistent”

• How could we detect?

– ARL for xbar would be too large

Manufacturing 16
 Another Approach: Cumulative Sums

• Add up deviations from mean

– A Discrete Time Integrator
j

C j = ∑ (x i − x)
i=1

• Since E{x-μ}=0 this sum should stay near

zero when in control
• Any bias (mean shift) in x will show as a trend

Manufacturing 17
 Mean Shift Sensitivity: CUSUM
8

7
t
6 Ci = ∑ (x i − x )
i =1

4
Mean shift = 1σ
3
Slope cause by
2 mean shift Δμ
1

0
1

11

13

15

17

19

21

23

25

27

29

31

33

35

37

39

41

43

45

47

49
-1

Manufacturing 18
 Control Limits for CUSUM
• Significance of Slope Changes?
– Detecting Mean Shifts
• Use of v-mask
– Slope Test with Deadband ⎛1 − β⎞
2
d = ln⎜ ⎟
δ ⎝ α ⎠
Upper decision line
Δx
δ=
θ σx
⎛ Δx ⎞
θ = tan ⎜ ⎟
−1
⎝ 2k ⎠
d where
Lower decision line k = horizontal scale
factor for plot
Manufacturing 19
 8 Use of Mask
7

θ=tan-1(Δμ/2k)
6
k=4:1; Δμ=0.25 (1σ)
5
tan(θ) = 0.5 as plotted

0
1

11

13

15

17

19

21

23

25

27

29

31

33

35

37

39

41

43

45

47

49
-1

Manufacturing 20
 An Alternative
• Define the Normalized Statistic
Xi − μ x Which has an
Zi =
σx expected mean of
0 and variance of 1
• And the CUSUM statistic
t

∑Z i
Which has an
expected mean of
Si = i =1
0 and variance of 1
t

Chart with Centerline =0 and Limits = ±3

Manufacturing 21
 Example for Mean Shift = 1σ
5
Normalized CUSUM

0
1

11

13

15

17

19

21

23

25

27

29

31

33

35

37

39

41

43

45

47
Mean Shift = 1 σ

-1

Manufacturing 22
 Tabular CUSUM

• Create Threshold Variables:

Ci = max[0, xi − ( μ 0 + K ) + Ci −1 ] Accumulates
+ +

deviations
Ci = max[0,( μ 0 − K ) − xi +
− −
Ci −1 ] from the
mean
K= threshold or slack value for
accumulation
Δμ
K= Δμ = mean shift to detect
typical 2
H : alarm level (typically 5σ)

Manufacturing 23
 Threshold Plot
6
μ 0.495

5
σ 0.170
k=δμ/2 0.049
4 h=5σ 0.848

C+
C+ C-

3 C-
H threshold

0
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49

Manufacturing 24
 Alternative Charts Summary
• Noisy data need some filtering
• Sampling strategy can guarantee
independence
• Linear discrete filters been proposed
– EWMA
– Running Integrator
• Choice depends on nature of process
• Noisy data need some filtering, BUT
– Should generally monitor variance too!

Manufacturing 25
Motivation: Multivariate Process Control
• More than one output of concern
– many univariate control charts
– many false alarms if not designed properly
– common mistake #1
• Outputs may be coupled
– exhibit covariance
– independent probability models may not be
appropriate
– common mistake #2

Manufacturing 26
 Mistake #1 – Multiple Charts

• Multiple (independent) parameters being

monitored at a process step
– set control limits based on acceptable
α = Pr(false alarm)
• E.g., α = 0.0027 (typical 3σ control limits), so 1/370 runs
will be a false alarm
– Consider p separate control charts
• What is aggregate false alarm probability?

Manufacturing 27
 Mistake #1 – Multiple
Tests for Significant Effects
• Multiple control charts are just a running
hypothesis test – is process “in control” or
has something statistically significant
occurred (i.e., “unlikely to have occurred
by chance”)?
• Same common mistake (testing for
multiple significant effects and
misinterpreting significance) applies to
many uses of statistics – such as medical
research!

Manufacturing 28
 The Economist (Feb. 22, 2007)

Text removed due to copyright restrictions. Please

see The Economist, Science and Technology.
“Signs of the times.” February 22, 2007.

Manufacturing 29
 Approximate Corrections for
Multiple (Independent) Charts

• Approach: fixed α’
– Decide aggregate acceptable false alarm rate, α’
– Set individual chart α to compensate

– Expand individual control chart limits to match

Manufacturing 30
 Mistake #2: Assuming Independent
Parameters
• Performance related to many variables
• Outputs are often interrelated
– e.g., two dimensions that make up a fit
– thickness and strength
– depth and width of a feature (e.g.. micro embossing)
– multiple dimensions of body in white (BIW)
– multiple characteristics on a wafer
• Why are independent charts deceiving?

Manufacturing 31
 Examples

• Body in White (BIW) assembly

– Multiple individual dimensions measured
– All could be OK and yet BIW be out of spec

• Injection molding part with multiple key

dimensions

• Numerous critical dimensions on a

semiconductor wafer or microfluidic chip

Manufacturing 32
 LFM Application

Rob York ‘95

“Distributed Gaging
Methodologies for Variation
Reduction in and
Automotive Body Shop”

Body in White “Indicator”

Manufacturing 33
Manufacturing 34
 Independent Random Variables

4 4

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53

2 -2

-4

1
-6 x1
x2 -4 -3 -2 -1
0
0 1 2 3 4
-8

-1

4
-2

-3
1

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53

-4 -1

-2

x1 -3

-4
x2
-5

Proper Limits?
Manufacturing 35
 Correlated Random Variables

3
4

1
3

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53

-1
2

1
-2

-3
x1
-4

x2 -4 -3 -2 -1
0
0 1 2 3 4

4
-1

2
-2

0
-3 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53

-1

-4
-2

-3
x2
x1 -4

Proper Limits?
Manufacturing 36
 Outliers?
4

4
2

3 0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53

-1

2 -2

-3

-4
1

x2 -4 -3 -2 -1
0
0 1 2 3 4

-1
3

-2 1

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53

-3 -1

-2

-4 -3

-4

x1

Manufacturing 37
 Multivariate Charts

• Create a single chart based on joint

probability distribution
– Using sample statistics: Hotelling T2

• Set limits to detect mean shift based on α

• Find a way to back out the underlying
causes
• EWMA and CUSUM extensions
– MEWMA and MCUSUM

Manufacturing 38
 Background
• Joint Probability Distributions
• Development of a single scale control chart
– Hotelling T2
• Causality Detection
– Which characteristic likely caused a problem
• Reduction of Large Dimension Problems
– Principal Component Analysis (PCA)

Manufacturing 39
 Multivariate Elements
• Given a vector of measurements

• We can define vector of means:

where p = # parameters
• and covariance matrix:

variance

covariance

Manufacturing 40
 Joint Probability Distributions

• Single Variable Normal Distribution

squared standardized
distance from
mean

• Multivariable Normal Distribution

squared standardized
distance from mean

Manufacturing 41
 Sample Statistics
• For a set of samples of the vector x

• Sample Mean

• Sample Covariance

Manufacturing 42
 Chi-Squared Example - True Distributions
Known, Two Variables
• If we know μ and Σ a priori:

will be distributed as
– sum of squares of two unit normals
• More generally, for p variables:

distributed as (and n = # samples)

Manufacturing 43
 Control Chart for χ2?
• Assume an acceptable probability of Type I
errors (upper α)
• UCL = χ2α, p
– where p = order of the system

• If process means are µ1 and µ2 then χ20 < UCL

Manufacturing 44
 Univariate vs. χ2 Chart
Joint control region for x1 and x2
UCLx1

x1

LCLx1

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
x2

LCLx2
UCLx2
2
UCL = xa.2 2
3
4
5
6
7
x02
8
9
10
11
12
13
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 14
Sample number 15
16

Figure by MIT OpenCourseWare.

Manufacturing 45
Montgomery, ed. 5e
 Multivariate Chart with
No Prior Statistics: T2
• If we must use data to get x and S
• Define a new statistic, Hotelling T2

• Where x is the vector of the averages for

each variable over all measurements
• S is the matrix of sample covariance over all
data

Manufacturing 46
 Similarity of T2 and t2

vs.

Manufacturing 47
 Distribution for T2

– Given by a scaled F distribution

α is type I error probability

p and (mn – m – p + 1) are d.o.f. for the F distribution
n is the size of a given sample
m is the number of samples taken
p is the number of outputs
Manufacturing 48
 F - Distribution

Fν1 ,ν2 ν1 = 8,ν 2 = 16

Significance level α

Tabulated Values Fα , p,( mn − m − p +1)

Manufacturing 49
 Phase I and II?
• Phase I - Establishing Limits

• Phase II - Monitoring the Process

NB if m used in phase 1 is large then they are nearly the same

Manufacturing 50
 Example
• Fiber production
• Outputs are strength and weight
• 20 samples of subgroups size 4
– m = 20, n = 4
• Compare T2 result to individual control charts

Manufacturing 51
 Data Set
Output x1 Output x2
Sample Subgroup n=4 R1 Subgroup n=4
1 80 82 78 85 7 19 22 20 20
2 75 78 84 81 9 24 21 18 21
3 83 86 84 87 4 19 24 21 22
4 79 84 80 83 5 18 20 17 16
5 82 81 78 86 8 23 21 18 22
6 86 84 85 87 3 21 20 23 21
7 84 88 82 85 6 19 23 19 22
8 76 84 78 82 8 22 17 19 18
9 85 88 85 87 3 18 16 20 16
10 80 78 81 83 5 18 19 20 18
11 86 84 85 86 2 23 20 24 22
12 81 81 83 82 2 22 21 23 21
13 81 86 82 79 7 16 18 20 19
14 75 78 82 80 7 22 21 23 22
15 77 84 78 85 8 22 19 21 18
16 86 82 84 84 4 19 23 18 22
17 84 85 78 79 7 17 22 18 19
18 82 86 79 83 7 20 19 23 21
19 79 88 85 83 9 21 23 20 18
20 80 84 82 85 5 18 22 19 20

Mean Vector Covariance Matrix

82.46 7.51 -0.35

20.18 -0.35 3.29

Manufacturing 52
 Cross Plot of Data
X2 bar
23.00
23.00

22.00 22.00

21.00
21.00
20.00 X2 bar

20.00 19.00

18.00
19.00
17.00
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

18.00
X1 bar

17.00 88.00
78.00 80.00 82.00 84.00 86.00 88.00 86.00

84.00

82.00
X1 bar

80.00

78.00

76.00

74.00
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Manufacturing 53
 T2 Chart

18

16

14

12

10
T2
UCL
8

6 α = 0.0054
4
(Similar to ±3σ limits on
2 two xbar charts combined)
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Manufacturing 54
 Individual Xbar Charts
88.00

86.00

84.00

82.00 X1 bar
UCL X1

80.00 LCL X1

78.00

76.00

74.00
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

24.00

23.00

22.00

21.00 X2 bar
UCL X2

20.00 LCL X2

19.00

18.00

17.00
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Manufacturing 55
 Finding Cause of Alarms
• With only one variable to plot, which
variable(s) caused an alarm?
• Montgomery
– Compute T2
– Compute T2(i) where the i th variable is not included
– Define the relative contribution of each variable as
• di = T2 - T2(i)

Manufacturing 56
 Principal Component Analysis

• Some systems may have many measured

variables p
– Often, strong correlation among these variables:
actual degrees of freedom are fewer
• Approach: reduce order of system to track only
q << p variables
– where each z1 … zq is a linear combination of the
measured x1 … xp variables

Manufacturing 57
 Principal Component Analysis

• x1 and x2 are highly correlated in the z1

direction
• Can define new axes zi in order of
decreasing variance in the data
• The zi are independent
• May choose to neglect dimensions with
only small contributions to total variance
⇒ dimension reduction

Truncate at q < p

Manufacturing 58
 Principal Component Analysis
• Finding the cij that define the principal
components:
– Find Σ covariance matrix for data x
– Let eigenvalues of Σ be
– Then constants cij are the elements of the ith
eigenvector associated with eigenvalue λis
• Let C be the matrix whose columns are the eigenvectors
• Then
where Λ is a p x p diagonal matrix whose diagonals are the
eigenvalues
• Can find C efficiently by singular value decomposition (SVD)
– The fraction of variability explained by the ith principal
components is

Manufacturing 59
 Extension to EWMA and CUSUM

• Define a vector EWMA

Z i = rxi + (1 + r)Z i −1
• And for the control chart plot

−1
where Ti = Z i Σ Z i
2 T
Zi

r
Σ Zi i = [1 − (1 − r) ]Σ
2

2−r

Manufacturing 60
 Conclusions

• Multivariate processes need multivariate

statistical methods
• Complexity of approach mitigated by
computer codes
• Requires understanding of underlying process
to see if necessary
– i.e. if there is correlation among the variables of
interest

Manufacturing 61