Recorded Lecture- Part 3/3 Process Performance

Control Charts

[Link]
Process Control, A. Asef-Vaziri, Systems & Operations Management. 1
Control Charts
 Process control charts – for any process - define acceptable variations -
around the mean (average).
 Upper Control Limit and Lower Control Limit (UCL & LCL) define a range
within which any variation in performance is interpreted as normal and
random
 How do we know if the variations are normal and the process is in control-
Similar to Tracking Signal- Recall-
 Numbers are within the UCL and LCL
 No systematic pattern is observed
 In Tracking Signal we did not have clear UCL & LCL, and I explained that
+3.75MAD is incorrect.
 The UCL & LCL are clearly defined for process performance control chats.
 Applications of Process Control Charts
 Controlling quality, inventory, throughput, flow and time.
 Stock trading - purchase when the price drops below LCL, and sell when
Process Control, A. Asef-Vaziri, Systems & Operations Management. 2
Statistical Process Control
 If the observed values are within the range and do not show a systematic
pattern
 Accept the variations as normal
 Don’t make any adjustments to the process
 If the observed values are
 Outside of UCL or LCL
 Show a pattern- trend or cycles
 The process is out of control
 Need to investigate what’s causing the problems – the assignable cause
 Optimal Degree of Control depends on 2 things:
 How much variability in the performance measure we consider acceptable
 How frequently we monitor the process performance.

Process Control, A. Asef-Vaziri, Systems & Operations Management. 3

Control Charts
 Xbar = Average of each sample of size (n observations)
 Xdoublebar = Average of all observations (N observations)
 Sigma= Standard deviation of all observations
 SigmaXbar= Sigma/SQRT(n) where n is the # of observations in each
sample

 UCL =
 LCL =

Process Control, A. Asef-Vaziri, Systems & Operations Management. 4

Xbar-Chart UCL LCL
 We have collected 10 samples each containing 9 observations. The
average of all observations is 100 and the standard deviation of all
observations is 30. Find the UCL and LCL for Xbar chart if if we have 9
observation in each sample.
Xdoublebar = = mu= = 100
StdDevX= 30
StdDevxbar = StdDevX/SQRT(n) = =
StdDevxbar = 30/SQRT(9) = 10
UCL = 100+3*10 = 130
LCL= 100-3*10 = 70

Process Control, A. Asef-Vaziri, Systems & Operations Management. 5

Xbar-Chart – Sales of 25 Stores in 11 Weeks (n=25)
Average Throughput per day
Day 1 Day 2 Day 3 Day 4 Day 5 Day 6 Day 7 Day 8 Day 9 Day 10 Day 11
63 114 135 87 69 124 69 118 58 153 80
135 116 94 131 107 132 101 158 112 87 88 Chart Title
134 85 57 50 134 115 99 74 87 55 107
130
65 153 137 56 91 107 46 93 102 119 45
112 85 89 147 82 104 136 76 72 87 170 120
81 134 101 47 106 109 134 119 148 128 90
110
64 78 92 68 127 120 102 80 156 123 46
151 88 88 131 88 162 119 131 137 44 42 100
107 97 104 98 115 104 107 115 90 157 135
90
108 129 155 112 143 151 133 148 148 95 103
59 149 1 117 57 123 131 77 143 122 97 80
100 65 127 124 137 121 129 63 113 87 162
70
135 82 130 114 93 67 120 96 73 151 89
97 66 73 134 119 134 63 67 103 31 169 60
0 2 4 6 8 10 12
133 124 84 100 151 74 134 113 73 90 36
111 81 70 122 82 75 84 129 75 107 54
88 93 108 93 123 108 102 50 107 112 79
96 91 71 127 94 65 39 67 87 110 97
97 129 59 50 53 78 32 100 141 80 95 Xdoublebar 99.89 =AVERAGE(A3:K27)
146 114 93 105 76 87 43 117 107 69 141 StdX 31.10 =STDEV.S(A3:K27)
77 155 111 103 89 130 56 51 141 64 145 n 25 =COUNT(A3:A27)
137 46 80 136 91 107 73 98 88 157 88 StdXbar 6.22 =N22/SQRT(N23)
116 112 96 109 108 100 93 98 115 56 126 UCL 118.55 =N21+3*N24
70 72 50 127 120 100 107 79 57 102 112 LCL 81.23 =N21-3*N24
39 60 59 125 94 96 58 117 91 73 89

100.84 100.72 90.56 104.52 101.96 107.72 92.4 97.36 104.96 98.36 99.4 Xbar =AVERAGE(K3:K27)

Process Control, A. Asef-Vaziri, Systems & Operations Management. 6

p-Chart
Pbar ()= total number of defectives in all samples divided by the total number
of observations. , where D is the total number of defects and N is the total
number of observations
Var(p) = Var p = pbar(1-pbar)/n  = (
StdDev p = Sigma p = SQRT(Var p) = = SQRT(() =
Note that n is not the total number of observations but the number of
observations in each sample.
P-chart
UCL =
LCL =

Process Control, A. Asef-Vaziri, Systems & Operations Management. 7

P-Chart UCL LCL
 Find the upper control limit for a 3-sigma p-chart assuming you sampled 60
items per day for 5 days and found a total of 30 defectives in that 5 day
period.
 Total number of observations = N 5*60 = 300
 Total number of defects = D =30
 Pbar =30/300 = 0.1
 UCL = pbar+3SQRT(pbar(1-pbar)/n) = 3
 LCL = pbar-3SQRT(pbar(1-pbar)/n) = -
 UCL = = 0.1+3= 0.1+0.11619 = 0.21619
 UCL = = 0.1-3= 0.1-0.11619 = -0.01619 = 0

Process Control, A. Asef-Vaziri, Systems & Operations Management. 8

p-Chart
Based on 20 samples of 5 doors each, the number of defective doors in each
sample is 1, 0, 0, 2, 1, 1, 0, 1, 2, 1, 2, 1, 1, 2, 1, 0, 3, 0, 1, 0.
We have had 20*5 = 100 observations. We have found 20 defects in total
=20/100 = 0.2
U

𝐿𝐶𝐿=0.2− 3 √ 0.2(0.8)/5=− 0.1366=0

If the observed fraction defective is less than 0.07366, we conclude the
process is in control, as is the case above.

Process Control, A. Asef-Vaziri, Systems & Operations Management. 9

Samples of 25 Students in D&F Range in 17 Quizzes
Q1 Q2 Q3 Q4 Q5 Q6 Q7 Q8 Q9 Q10 Q11 Q12 Q13 Q14 Q15 Q16 Q17
1 30 90 70 85 70 60 80 70 90 85 75 75 100 95 70 100 95 Defects 96
2 89 90 70 50 90 65 65 80 80 55 75 95 45 60 100 85 70 All 425
3 100 100 95 85 95 85 80 65 80 60 95 95 90 85 95 90 100 pbar 0.2258824 =T2/T3
4 95 85 95 95 55 85 85 95 85 95 75 70 65 95 85 40 40 n 25
5 95 75 70 95 55 75 75 75 70 100 90 75 40 90 85 60 75 Varp 0.0069944 =T4*(1-T4)/T5
6 50 90 95 70 40 55 75 45 40 80 70 85 45 80 50 100 90 StdDevp 0.0836324
7 35 95 75 95 70 50 80 95 90 80 95 85 35 60 75 90 50 UCL 0.4767796 =T4+3*T7
8 90 70 75 85 100 80 75 95 65 90 75 95 100 90 70 80 75 LCL 0 =MAX(0,T4-3*T7)
9 85 35 80 85 45 100 75 75 90 70 95 35 100 45 70 60 70
10 80 95 95 95 80 80 90 35 90 95 75 95 75 80 80 70 100 p-Chart
11 95 80 85 85 45 30 95 70 50 90 75 35 35 80 60 95 85 0.6
12 75 100 100 80 100 70 85 85 85 85 100 90 95 70 90 75 35
13 100 100 75 80 80 45 90 65 70 70 70 50 100 100 85 90 60 0.5
14 55 75 100 95 85 95 100 85 100 55 60 40 95 95 75 80 75
15 100 90 95 80 95 85 35 80 80 80 80 95 95 80 95 100 40 0.4

16 80 95 75 30 45 100 75 40 80 65 95 70 90 80 95 90 35
0.3
17 85 45 40 95 100 90 100 70 65 95 80 85 40 35 80 85 90
18 35 100 85 80 95 40 100 95 100 95 80 65 75 90 95 75 70 0.2
19 70 85 70 40 90 100 35 100 100 90 85 45 65 75 90 55 95
20 35 95 90 35 100 75 85 65 90 50 70 70 90 50 30 86 85 0.1
21 80 75 95 85 85 70 95 70 40 30 80 85 95 60 95 35 85
22 90 60 80 95 90 100 80 35 75 80 45 85 100 45 80 80 75 0
0 2 4 6 8 10 12 14 16 18
23 80 65 100 40 75 80 90 35 75 85 70 40 65 95 100 95 95
24 90 85 100 50 85 80 60 100 90 100 95 70 90 90 85 80 85
25 90 80 45 70 80 60 85 70 75 90 85 85 35 65 75 75 80
c<70 6 4 2 6 6 8 4 8 5 6 2 7 10 8 3 5 6
n 25 25 25 25 25 25 25 25 25 25 25 25 25 25 25 25 25
p<70 0.24 0.16 0.08 0.24 0.24 0.32 0.16 0.32 0.2 0.24 0.08 0.28 0.4 0.32 0.12 0.2 0.24

Process Control, A. Asef-Vaziri, Systems & Operations Management. 10

Performance Variation

Stabl Unstabl Trend

e e

Cyclical Shift
UCL

LCL
Out of Control In Control Improved

Process Control, A. Asef-Vaziri, Systems & Operations Management. 11

Practice
 Assuming you sampled final grades of samples of 25 students per course in
8 core courses. Suppose the average and standard deviation of all the data
are 75 and 12 respectively.
 Find the UCL and LCL for average grades of samples of 25 students in core
courses (Xbar-Chart).
Xdoublebar = = 75
Standard Deviation = Sigma = = 12
Sample Size = n=25
Standard Deviation of Average = Sigmaxbar = 12/sqrt(25) = 12/5 = 2.4
UCL= 75+3*2.4 = 75+7.2 = 82.2
LCL= 75-7.2 = 67.8
If LCL is negative, we set it to 0

Process Control, A. Asef-Vaziri, Systems & Operations Management. 12

Practice
 Assuming you sampled final grades of samples of 25 students per course in
8 core courses. Suppose there are 44 students with D/F grades.
 Find the UCL and LCL for proportion of students with D/F samples of 25
students in core courses (p-Chart).

pbar = 44/(25*8) = 44/200= 0.22

UCL = pbar+3SQRT(pbar(1-pbar)/n) =
LCL = pbar-3SQRT(pbar(1-pbar)/n) = -
UCL = 0.22+3= 0.22+3*0.082849 = 0.22+ 0.248547782 = 0.468547782
UCL = 0.22-3= = 0.22- 0.248547782 = -0.02855 = 0

Process Control, A. Asef-Vaziri, Systems & Operations Management. 13

3-Sigma & 6-Sigma Capability

A higher value of the average indicates a that all doors produced are
consistently heavier. An increase in the value of the standard deviation
means a wider spread of the distribution around the mean, implying that
many doors are much heavier or lighter than the overall average weight.

Process Control, A. Asef-Vaziri, Systems & Operations Management. 14