Lecture 22
Lecture 22
P2 X4
Process
Product X5
Input P3
X6
Process variables Product
characteristics X7
Dependencies
Which variables affect a product characteristic?
What are operating conditions to get target characteristics? Explained part of variation
Robust Design using Design of Experiments
Controllable variables
• Robust design of a product is (𝑋1 , 𝑋2 ,……)
insensitive to variations in the ….
process environment or in customer
environment Input Process Output (𝑌1 , 𝑌2 ,……)
CTQ
• Design of experiments (DOE) or
….
Determine 𝑋ധ 𝑎𝑛𝑑 𝑅ത
Process
control?
Specs
met?
Identify and remove cause
Continue sampling Review specs Continue sampling
Detect trends any Change process design Update control limits
Determine m/c resetting frequency 100% inspection
Determining control limits
Start base period with atleast 25 samples
Outliers?
Remove outlier data points
R
Outliers?
Set control limits
𝑋ത
Remaining
>=15?
Determining control limits (Contd.)
ഥ chart:
Control limits of 𝒙
If 𝜎𝑥 − 𝑘𝑛𝑜𝑤𝑛 ℎ𝑖𝑠𝑡𝑜𝑟𝑖𝑐𝑎𝑙𝑙𝑦 ,
𝜎𝑥 𝜎𝑥
𝐸 𝑥ҧ ± 3𝜎𝑥ҧ ⇒ 𝜇 ± 3𝜎𝑥ҧ ⇒ 𝐸 𝑥ҧ ± 3 ⇒ 𝑥Ӗ ± 3 ന ± 𝑨𝝈𝒙
⇒𝒙
𝑛 𝑛
where n- sample size
𝑠ҧ 𝑅ത
If 𝜎𝑥 − 𝑢𝑛𝑘𝑛𝑜𝑤𝑛, 𝑡ℎ𝑒𝑛 𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒, 𝜎 ෞ𝑥 = or
𝑐4 𝑑2
𝜎𝑥 𝑅ത
• 𝐸 𝑥ҧ ± 3𝜎𝑥ҧ ⇒ 𝜇 ± 3𝜎𝑥ҧ ⇒ 𝐸 𝑥ҧ ± 3 ⇒ 𝑥Ӗ ± 3 ഥ
ന ± 𝑨𝟐 𝑹
⇒𝒙
𝑛 𝑑2 𝑛
𝜎𝑥 𝑠ҧ
• 𝐸 𝑥ҧ ± 3𝜎𝑥ҧ ⇒ 𝜇 ± 3𝜎𝑥ҧ ⇒ 𝐸 𝑥ҧ ± 3 ⇒ 𝑥Ӗ ± 3 ന ± 𝑨𝟑 𝑺ത
⇒𝒙
𝑛 𝑐4 𝑛
Determining control limits (Contd.)
Control limits of R chart: 𝐸 𝑅 ± 3𝜎𝑅
𝑅ത
It is known that 𝜎
ෞ𝑥 = ; Therefore,𝐸 𝑅 = 𝜎𝑥 𝑑2
𝑑2
It is known that 𝜎𝑅 =𝜎𝑥 𝑑3
If 𝜎𝑥 − 𝑘𝑛𝑜𝑤𝑛 ℎ𝑖𝑠𝑡𝑜𝑟𝑖𝑐𝑎𝑙𝑙𝑦 ,
𝐸 𝑅 ± 3𝜎𝑅 ⇒ 𝜎𝑥 𝑑2 ± 3𝜎𝑥 𝑑3 ⇒ 𝜎𝑥 𝑑2 ± 3𝑑3
Let, 𝐷1 = 𝑑2 − 3𝑑3 and 𝐷2 = 𝑑2 + 3𝑑3 . Then, UCLR=𝐷2 𝜎𝑥 ; LCLR=𝐷1 𝜎𝑥
𝑅ത
If 𝜎𝑥 − 𝑢𝑛𝑘𝑛𝑜𝑤𝑛, then, estimate, 𝜎
ෞ𝑥 =
𝑑2
𝑑3 𝑅 ത
ത
𝐸 𝑅 ± 3 𝜎𝑥 𝑑3 ⇒ 𝑅 ± 3
𝑑2
𝑑3 𝑑3
Let, 𝐷3 = 1 − 3 and 𝐷4 = 1 + 3 , ത LCLR=𝐷3 𝑅ത
Then, UCLR=𝐷4 𝑅;
𝑑2 𝑑2
Determining control limits (Contd.)
Control limits of S chart: 𝐸 𝑠 ± 3𝜎𝑠
𝑠ҧ
It is known that 𝜎
ෞ𝑥 = ; Therefore,𝐸 𝑠 = 𝜎𝑥 𝑐4
𝑐4
It is known that 𝜎𝑠 =𝜎𝑥 1 − 𝑐42
If 𝜎𝑥 − 𝑘𝑛𝑜𝑤𝑛 ℎ𝑖𝑠𝑡𝑜𝑟𝑖𝑐𝑎𝑙𝑙𝑦 ,
𝐸 𝑠 ± 3𝜎𝑠 ⇒ 𝜎𝑥 𝑐4 ± 3𝜎𝑥 1 − 𝑐42 ⇒ 𝜎𝑥 𝑐4 ± 3 1 − 𝑐42
Let, 𝐵5 = 𝑐4 + 3 1 − 𝑐42 and 𝐵6 = 𝑐4 − 3 1 − 𝑐42 , 𝑡hen, UCLs=𝐵6 𝜎𝑥 ; LCLs=𝐵5 𝜎𝑥
𝑠ҧ
If 𝜎𝑥 − 𝑢𝑛𝑘𝑛𝑜𝑤𝑛, then, 𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒, 𝜎
ෞ𝑥 =
𝑐4
𝑠ҧ 1−𝑐42 1−𝑐42
𝐸 𝑠 ± 3 𝜎𝑥 𝑑3 ⇒ 𝑠ҧ ± 3 ⇒ 𝑠ҧ 1 ± 3
𝑐4 𝑐4
1−𝑐42 1−𝑐42
Let, 𝐵4 = 1 + 3 and 𝐵3 = 1 − 3 , then, UCLs=𝐵4 𝑠;ҧ LCLs=𝐵3 𝑠ҧ
𝑐4 𝑐4
ഥ − 𝑪𝒉𝒂𝒓𝒕 and R chart
Why both 𝒙
• 𝑥ҧ − 𝐶ℎ𝑎𝑟𝑡 → Shift in process mean
• R & S - Charts → Variability in the process
Reading a control chart
Process SD 0.139814
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