IEM 4103 Quality Control & Reliability Analysis

IEM 5103 Breakthrough Quality & Reliability

Quality Control Methods

Instructor: Dr. Chenang Liu

Email: Chenang.Liu@okstate.edu
 Outline

 Chance & Assignable Causes of Quality Variation

 Statistical Basis of Control Charts
 Textbook: Chapter 5, 6

2/3/2021

2
 4.1 Chance Cause & Assignable Cause

 Chance causes/common causes/system faults/chronic problems

 System problems/inherent problems (natural variation/background noise)
 “In statistical control”
 Generally small, but unavoidable

 Assignable causes/special causes/local causes/sporadic problems

 Problems arise in somewhat unpredictable fashion (operator error, material
defects, machine failure)
 “Out of statistical control”
 Generally large, should be eliminated

2/3/2021

3
 Example

2/3/2021

4
 Statistical Process Control (SPC)

 Objectives
 To quickly detect the occurrence of assignable causes or process shifts so that
investigations of the process and corrective actions may be undertaken before
many nonconforming units are manufactured.
 Process Variation Reduction
 Procedures (Control chart is one of the primary techniques)
 Monitoring the process and detecting process changes
 Diagnosing the assignable causes
 Providing corrective actions plans
 Dealing with resistance to changes/actions
 Instituting controls to hold the gains
2/3/2021

5
 4.2 Statistical Basis of Control Chart

 Concept of Control Chart

 Choice of Control Limits
 Rational Subgroups
 General Model of Control Chart
 Control Chart Performance
 Process Out of Control: The Rules
 Phase I and Phase II of Control Chart Application

2/3/2021

6
 Why use control charts?

 Improve process and reduce process variation

 Most processes do NOT operate in a state of statistical control.

 Consequently, the routine and attentive use of

control charts will assist in identifying assignable
causes. If these causes can be eliminated from
the process, variability will be reduced and the
process will be improved.

 The control chart will only detect assignable causes. Management, operator, and
engineering action will usually be necessary to eliminate the assignable causes.
2/3/2021

7
 Why control charts are popular?

 Control charts are a proven technique for improving productivity: it will reduce
scrap and rework, which are the primary productivity killers in any operation.
 Control charts are effective in defect prevention: it helps keep the process in control.
 Control charts prevent unnecessary process adjustment: it can distinguish between
background noise and abnormal variation; no other device including a human operator
is as effective in making this distinction.
 Control charts provide diagnostic information: the pattern of points on the control
chart will contain information of diagnostic value to an experienced operator or engineer.
 Control charts provide information about process capability: it provides information
about the value of important process parameters and their stability over time.

2/3/2021

8
 4.2.1 Concept of Control Chart

 Control chart:
 A graphical display of a quality characteristic that has been measured or
computed from a sample versus the sample number or time.

 Center line:
 It represents the average value of the quality characteristic corresponding to the
in-control state (only chance causes are present)

 Upper control limit (UCL) & Lower control limit (LCL):

 These control limits are chosen so that if the process is in control, nearly all of the
sample points will fall between them.

2/3/2021

9
 Example

A typical control chart with

• Center line
• Upper control limit
• Lower control limit

2/3/2021

10
 Control limit VS. Specification limit

 Control Limits (CL) are used to determine if the process is in a state of

statistical control (i.e., is producing consistent output). They are statistics
calculated based on sample measurements.

 Specification Limits (SL) are used to determine if the product will function in
the intended fashion. Specified by designers.

2/3/2021

11
 4.2.2 Choice of Control Limits

 Moving the control limits farther from the center line:

 Decrease the risk of a type I error, but increase the risk of a type II error
 Moving the control limits closer to the center line:
 The risk of type I error is increased, but the risk of type II error is decreased
 Two Limits on Control Charts
 Action limits: when a point plots outside of this limit, a search for an assignable
cause is made and corrective action is taken if necessary.
 Warning limits: increase the sensitivity of the control chart; but may be confusing
to operating personnel.

2/3/2021

12
 Example (real world application)

 In semiconductor manufacturing, an important fabrication step is photolithography, in

which a light-sensitive photoresist material is applied to the silicon wafer, the circuit
pattern is exposed on the resist typically through the use of high-intensity UV light,
and the unwanted resist material is removed through a developing process. After the
resist pattern is defined, the underlying material is removed by either wet chemical or
plasma etching. It is fairly typical to follow development with a hard-bake process to
increase resist adherence and etch resistance. An important quality characteristic in
hard bake is the flow width of the resist, a measure of how much it expands due to
the baking process. Suppose that flow width can be controlled at a mean of 1.5
microns, and it is known that the standard deviation of flow width is 0.15 microns.

2/3/2021

13
 Example (cont’d)

 A control chart for the average flow width is shown as below. Every hour, a sample of
five wafers is taken, the average flow width 𝑥𝑥̅ computed, and plotted on the chart.
Because this control chart utilizes the sample average 𝑥𝑥̅ to monitor the process mean,
it is usually called an 𝑥𝑥̅ control chart. Note that all of the plotted points fall inside the
control limits, so the chart indicates that the process is considered to be in statistical
control.

2/3/2021

14
 Review of hypothesis test errors

 Type I error:
 𝛼𝛼 = 𝑃𝑃 type I error = 𝑃𝑃 reject 𝐻𝐻0 𝐻𝐻0 is true
= 𝑃𝑃 conclude out of control although the process is truely in control

 Type II error:
 𝛽𝛽 = 𝑃𝑃 type II error = 𝑃𝑃 fail to reject 𝐻𝐻0 𝐻𝐻0 is false
= 𝑃𝑃 conclude in control although the process is truely out of control

 Power of the test:

 Power = 1 − 𝛽𝛽 = P{reject 𝐻𝐻0 |𝐻𝐻0 is false}

2/3/2021

15
 Control charts vs. Hypothesis testing

 Hypothesis Testing
 Hypothesis testing has a rejection region
 𝐻𝐻0 is rejected if the data follow in the rejection region
 Control Charts
 Control chart has UCL and LCL
 The process is out of control if the data beyond the control limits
 Think: What is the relationship between rejection region and control limit?

2/3/2021

16
 Example (wafer cont’d)

 We know: 𝜎𝜎 = 0.15, 𝑛𝑛 = 5, 𝜎𝜎𝑥𝑥̅ = 0.0671

 Hypothesis Testing
𝐻𝐻0 : 𝜇𝜇 = 1.5
𝐻𝐻1 : 𝜇𝜇 ≠ 1.5
 Control Charts
𝑈𝑈𝑈𝑈𝑈𝑈 = 1.5 + 3(0.0671) = 1.7013
𝐿𝐿𝐿𝐿𝐿𝐿 = 1.5 − 3(0.0671) = 1.2987

 Think: what is the relationship between confidence interval and control limit?
Based on this predefined control limit, what is the level of significance?

2/3/2021

17
 4.2.3 Rational Subgroups

 Rational subgroup: if assignable causes are present,

 The chance for differences between subgroups will be maximized.
 The chance for differences due to these assignable causes within a
subgroup will be minimized.
 Time order is frequently a good basis for forming subgroups because it allows
us to detect assignable causes that occur over time.
 Two general approaches to constructing rational subgroups are used.

2/3/2021

18
 Rational subgroups: approach 1

 Each sample consists of units that were produced at the same time (or as
closely together as possible). Ideally, we would like to take consecutive units of
production.
 This approach is used when the primary purpose of the control chart is to
detect process shifts.

2/3/2021

19
 Rational subgroups: approach 2

 Each sample consists of units of product that are representative of all units that
have been produced since the last sample was taken. Essentially, each
subgroup is a random sample of all process output over the sampling interval.
 This method of rational subgrouping is often used when the control chart is
employed to make decisions about the acceptance of all units of product that
have been produced since the last sample.

2/3/2021

20
 4.2.4 General Model for Control Chart

 Let 𝑤𝑤 be a sample statistic that measures some quality characteristic of

interest, and suppose that the mean of 𝑤𝑤 is 𝜇𝜇𝑤𝑤 and the standard deviation of
𝑤𝑤 is 𝜎𝜎𝑤𝑤 . Then the centerline, the upper control limit, and the lower control
limit become
UCL = 𝜇𝜇𝑤𝑤 + 𝑘𝑘𝜎𝜎𝑤𝑤
Center line = 𝜇𝜇𝑤𝑤
LCL = 𝜇𝜇𝑤𝑤 − 𝑘𝑘𝜎𝜎𝑤𝑤
 where 𝑘𝑘 is the "distance" of the control limits from the center line, expressed
in standard deviation units
 3-sigma control limits:
 Action limits: 𝑘𝑘 = 3 (𝛼𝛼 = 0.0027)
2/3/2021
 Warning limits: 𝑘𝑘 = 2 (𝛼𝛼 = 0.0455)
21
 Shewhart control charts

 This general model was first proposed by Walter A. Shewhart, and control charts
developed according to this model are often called Shewhart Control Charts.
 The control chart is a device for describing in a precise manner exactly what is
meant by statistical control.

Example: How the control chart works

2/3/2021

22
 Example (wafer cont’d)

 The upper and lower warning limits are located at

𝑈𝑈𝑈𝑈𝑈𝑈 = 1.5 + 2(0.0671) = 1.6342
𝐿𝐿𝐿𝐿𝐿𝐿 = 1.5 − 2(0.0671) = 1.3658
 When probability limits are used, the action limits are generally 0.001 limits and
the warning limits are 0.025 limits.
 What is the action limits?

2/3/2021

23
 4.2.5 Control Chart Performance

 How to measure the performance of a control chart? One could the Type I error
(𝛼𝛼) and Type II error (𝛽𝛽), the same as for evaluating hypothesis test.
 However, a control chart is actually a series of sample-by-sample hypothesis
tests, could we develop something more intuitive to measure the performance
of a chart more directly?
 For that, people like to use the average run length (ARL) as a performance
measure of control charts.

2/3/2021

24
 Average run length (ARL)

 Run length (RL), “L” is the sample number on which a control chart first signals,
that is, when a data point is outside the control limits.
 RL is a random variable, taking only integer values, {1,2,3, ⋯ }
 Using RL itself cannot serve the performance benchmarking purpose. We need
to use its expected value as performance measure, namely average RL (ARL).
 ARL is defined as the expected value of RL, or expected number of samples
until a control chart first signals.

2/3/2021

25
 Two types of ARL

 In-control ARL (ARL0): is the expected number of samples until a control

chart signals, under the condition that the actual process is truly in control
1
ARL0 = Think: How to get this equation?
𝛼𝛼
 Out-of-control ARL (ARL1): is the expected number of samples until a
control chart signals, under the condition that the actual process is in fact
out-of-control
1
ARL1 = Think: How to get this equation?
1 − 𝛽𝛽
 We'd like ARL0 to be as large as possible and ARL1 to be as small as
possible. Why?
2/3/2021

26
 Example

 Suppose that a control chart with 2-sigma limits is used to control a process.
If the process remains in control, find the average run length until a false out-
of-control signal is observed. Compare this with the in-control ARL for 3-
sigma limits.

2/3/2021

27
 Review: OC Curve (an example)

2/3/2021

28
 4.2.6 Process Out of Control – The Rules

 Out of control process

 When one or more points fall beyond the control limit
 Plotted points exhibit some nonrandom pattern of behavior

 Description of nonrandom pattern

 Run: a sequence of observations with same type (e.g., above CL)
 Run Up: a sequence of increasing observations
 Run Down: a sequence of decreasing observations
 Run of length: the number of samples in a run
 Cyclic patterns

2/3/2021

29
 Example for nonrandom pattern

 A run of length 8 or more points has a very low probability of occurrence in a

random sample of points. Consequently, any type of run of length 8 or more
is often taken as a signal of an out-of-control condition. For example, eight
consecutive points on one side of the center line may indicate that the
process is out of control.

Cyclic patterns

2/3/2021

30
 Some sensitizing rules

2/3/2021

31
 Discussion of sensitizing rules

 The overall type I error: the process is concluded out of control if any one of
the rules is applied if all 𝑘𝑘 rules are independent
𝑘𝑘

𝛼𝛼 = 1 − � 1 − 𝛼𝛼𝑖𝑖
𝑖𝑖=1

 where 𝛼𝛼𝑖𝑖 is the type I error of using one rule 𝑖𝑖 alone

2/3/2021

32
 Example

 An x-bar control chart with 3-sigma control limits is used to monitor process
mean. Inspection decision is made based on two successive samples using the
following rules:
 Rule 1: If one or two sample means exceed either the upper or lower control limit.
 Rule 2: If two sample means fall on the same side of the center line.

 (1) What is the Type I error using rule I alone?

 (2) What is the Type I error using rule II alone?
 (3) What is the overall Type I error based on the two rules if the two rules are
independent?
 (4) If the process has a mean shift of one process standard deviation, what is
2/3/2021 the Type II error using rule I?
33
 Phase I and Phase II of Control Chart Application

 Standard control chart usage involves phase I and phase II applications, with
two different and distinct objectives.
 Phase I: a set of process data is gathered and analyzed, constructing trial
control limits to determine if the process has been in control over the period
of time during which the data were collected, and to see if reliable control
limits can be established to monitor future production.
 Phase II: begin after we have a “clean” set of process data gathered under
stable conditions and representative of in-control process performance. In
this phase, we use the control chart to monitor the process by comparing the
sample statistic for each successive sample as it is drawn from the process
to the control limits.
 More details of application will be discussed in next lecture.
2/3/2021

34
 Thank you!

Any Questions?

2/3/2021

35