Quality Management: Focus

on SPC
Statistical Process Control
(SPC)
 SPC is a statistics-based methodology for
determining when a process is moving “out of
control.”
 All processes have variation in output.

 Some of the variation is inherent in the process

(common).
 Some of the variation is due to assignable
(special) causes.
 SPC is aimed at discovering variation due to
assignable causes and correcting those causes.
Example: Pizza Delivery
Service design specification requires a
delivered pizza to be between 115
degrees Fahrenheit and 125 degrees
Fahrenheit when it arrives at the
customer’s door. Meeting that
specification would take a careful
planning of production and delivery of
pizza

USL=125 degree and LSL=115 degree

Statistical Quality
Assurance
 Purpose: Assure that processes are performing in an

acceptable manner

 Methodology: Monitor process output using

statistical techniques
 If the results are acceptable, no further action is required

 Unacceptable results call for corrective action

 Statistical Process Control (SPC):

Quality control efforts that occur during

production/service
Variability = Actual -
Expected
 Customer Needs - Service Design

 Service Design - Process Capability

 Process Capability - Process Performance

 Process Performance - Service Performance

 Service Performance - Customer Perception

Each of these deviations is a piece of data

indicating a quality problem
Exercise: The Dice Game
Experiment 1: Roll one die at a time -100
times. Plot the role 1 to 6 as a histogram

Experiments 2 : Roll 3, 4, 5 dice at a time -30

times. Plot the average of each role as a
histogram

What are your observations from the

histograms?
 What is the mean?

 What is the shape?

 What is the impact on the number of dice on the shape?

 Lessons Learnt: Central Limit
Theorem
 If the population form which samples are taken is not

normal, the distribution of sample averages tend

towards normality provided the sample size is at least
4.

 This tendency gets better as n gets larger and

distribution is closer to normal.

 Sample Distribution: Same mean as population,

standard deviation σ/ , where n is the sample size, σ

is the standard deviation of the population.
The Feedback Control
Principle
Disturbances: Normal and Abnormal

Corrective
Action Process Performance

Decision Data &

Information
Variations and Control
 Random Variation: Common variability, Chance variation
 Natural variations in the output of process are created by countless

minor factors, too many to investigate economically

 Assignable variation: In process output, a variation whose

cause can be identified

 Assignable variations are large and their causes must be

investigated, detected and removed

 Assignable cause examples: Tool wear, equipment that

needs adjustment, defective materials, human factors

(carelessness, fatigue, noise and other distractions, failure to
follow correct procedures), failure of pumps, heaters, etc.
Population and Sample
Distribution

σ/

σ
How does the Chart Work?

Out of control
points caused
by assignable
causes
Upper
control
limit

Lower
control
limit

Time 
The Shewhart Control
Chart
 A time-ordered plot of sample statistics
 Helps distinguish between random and nonrandom variability

 It signals when the process is “out of control” and therefore

requires troubleshooting only when necessary

 Four commonly used control charts:

 Xbar and R charts are used for variables--data that are measured, for
example, amount of time needed to complete a task, length or width of a
part.
 p and c charts are used for attributes—data that are counted, for example,
the number of defective parts in a sample, the number of complaint call per
day, number of defects found in a new car, etc.

Type I error with charts:

Concluding a process is not in control when it actually is
Type II error with charts:
Concluding a process is in control when it is not
 86 UCL
84
Average

82
80
78 LCL
76
Sample # 
13
15
17
19
1
3
5
7
9
11
Population and Sample
Distribution

σ/

σ
Three-Sigma Control Limit
Factors
Sales Size, n Factor for Xbar limits, A2 Factor for R LCL, D3 Factor for R UCL, D4

2 1.88 0 3.27
3 1.02 0 2.57
4 0.73 0 2.28
5 0.58 0 2.11
6 0.48 0 2.00
7 0.42 0.08 1.92
8 0.37 0.14 1.86
9 0.43 0.18 1.82
10 0.31 0.22 1.78
11 0.29 0.26 1.74
12 0.27 0.28 1.72
13 0.25 0.31 1.69
14 0.24 0.33 1.67
15 0.22 0.35 1.65
16 0.21 0.36 1.64
17 0.20 0.38 1.62
18 0.19 0.36 1.61
19 0.19 0.40 1.60
20 0.18 0.41 1.59

Eugene Grant and Richard Leavenworth, Statistical Quality Control, McGraw-Hill, 5th Edition, New York, 1980
A quality control manager took five samples (S1, S2, S3, S4, S5),
each with four observations, of the diameter of shafts manufactured
on a lathe machine. The manager computed the mean of each
sample and then computed the grand mean. All values are in cm.
Use this information to obtain three-sigma (i.e., z=3 ) control limits
for means of future times. It is known from previous experience
that the standard deviation  of the process is 0.02 cm.
Example 1 Calculations
using 
12.10  12.12  12.11  12.10  12.12
x 12.11
5
and  0.02 (given). Note that sample size n 4.
 0.02
Hence  x   0.01
n 4
Upper control limit : UCL x  z x 12.11  3 0.01 12.14

Lower control limit : LCL x  z x 12.11  3 0.01 12.08


where  x 
n
Standard deviation of distribution of sample means x
 Process standard deviation and n Sample size.
Observation S1 S2 S3 S4 S5
1 12.11 12.15 12.09 12.12 12.09
2 12.10 12.12 12.09 12.10 12.14
3 12.11 12.10 12.11 12.08 12.13
4 12.08 12.11 12.15 12.10 12.12
Range R 0.03 0.05 0.06 0.04 0.05

0.03  0.05  0.06  0.04  0.05

R Average of sample ranges  0.046
5
Sample size n 4, therefore A2 0.73 from table
Hence, Upper/Lower Control Limits are
UCL  x  A2 R 12.11  0.73 0.046 12.14
LCL  x  A2 R 12.11  0.73 0.046 12.08
Range Chart Control Limits
The control chart used to monitor process dispersion or
precision is the R - chart.
Upper control limit, UCL R D 4 R
Lower control limit, LCL R D 3 R
where D 3 and D 4 are obtained from the three - sigma Control Limit
Factors table (slide 11).
Example 3: R chart Limits
Observation S1 S2 S3 S4 S5
1 12.11 12.15 12.09 12.12 12.09
2 12.10 12.12 12.09 12.10 12.14
3 12.11 12.10 12.11 12.08 12.13
4 12.08 12.11 12.15 12.10 12.12
Xbar 12.10 12.12 12.11 12.10 12.12
Range R 0.03 0.05 0.06 0.04 0.05

0.03  0.05  0.06  0.04  0.05

R Average of sample ranges  0.046
5
n 4. Therefore D3 0.00 and D4 2.28 from table.
Hence, Upper / Lower Control Limits are
UCL R D4 R 2.280 .046 0.105
LCL R D3 R 0.000 .046 0.00
Control Chart
Control Chart: Process is
in control
Control Chart: Mean is out
of control
Control Chart: Variability is out of
control
Control Chart
Both Mean and Variability are out of control
Interpreting Patterns in Control Charts

A process is said to be “in control” when it has the

following characteristics:
1. No points are outside the control limits – the traditional and
most popular SPC chart guideline,
2. The number of points above and below the center line is
about the same,
3. The points seem to fall randomly above and below the
center line,
4. Most points, but not all, are near the center line, and only a
few are close to the control limits.

Operations Management, D.A. Collier, J.R. Evans(200

Samples in a Controlled Process from a Normal
Distribution

Operations Management, D.A. Collier, J.R. Evans(200

Interpreting Patterns in Control Charts

 Evaluating the “patterns in the sample data” using

guidelines such as:

• Eight points in a row above or below the center line

• 10 of 11 consecutive points above or below the center line

• 12 of 14 consecutive points above or below the center line

• Two of three consecutive points in the outer one-third

region between the center line and one of the control limits
• Four of five consecutive points in the outer two-thirds region

between the center line and one of the control limits.

Operations Management, D.A. Collier, J.R. Evans(200

Illustration of Some Rules for Identifying Out-of-
Control Conditions

Operations Management, D.A. Collier, J.R. Evans(200

SPC Out-of-Control
Situations
One One
Sample More
Sample More Than +/ -Than +/-
3 Standard 3 from
Errors Standard
Mean 14 or More Samples
14 or More Oscillating
Samples Oscillating

Errors from Mean

UCL=3 UCL=3

_ _
X=0 X=0

LCL=-3 LCL=-3

Observation Observation

8 or More Samples
8 or More Above
Samples Above (or Below)(or
MeanBelow) 6 or More
6 or More Samples Increasing
Samples Increasing (or Decreasing) (or
Mean Decreasing)
UCL=3 UCL=3

_ _
X=0 X=0

LCL=-3 LCL=-3

Observation Observation

8-32
Healthcare Operations Management, Daniel B. McLaughlin, Julie M. Hays, AUPHA Press.
Fraction Defective or p
Chart
Control chart for attributes , used to monitor th e proportion of defectives in a process.
Upper control limit, UCLp  p  z σ p
Lower control limit, LCL p  p - z σ p
where from Binomial distributi on,
p (1  p)
p 
n
and p is the nominal fraction of defectives in the process.
If p is unknown, it can be estimated as p from history. The estimate, p, replaces p.
Sometimes LCL is negative due to approximat e formula. Use LCL 0.
Example 4: p
chart
A quality control manager counted Sample # Number of
Defectives
the number of defective nuts 1 10
produced by an automatic machine 2 9
3 8
in each of 12 samples. Using the
4 11
shown information, construct a 5 12
control chart that will describe 99.74 6 8
7 13
% of the chance variation in the 8 11
process when the process in control. 9 9
10 10
Each sample contained 200 nuts.
11 8
12 11
Total 120
Example 4 Solution

120
p 0.05
12 200
p (1  p ) 0.05(1  0.05)
p   0.015
n 200
z 3
Upper control limit,
UCLp  p  z σ p 0.05  3 0.015 0.095
Lower control limit,
LCLp  p  z σ p 0.05  3 0.015 0.005
Number of Defects/Unit or
cDiscrete

Chart Quality Measurement:
D = Number of “defects” (errors) per unit of work

 Examples of Defects:
 Number of typos/page, errors/thousand transactions,
 equipment breakdowns/shift, bags lost/thousand flown, power
 outages/year, customer complaints/month, defects/car...

 If n = No. of opportunities for defects to occur, and

p = Probability of a defect/error occurrence in each
then D ~ Binomial (n, p) with mean np, variance np(1-p)
Poisson (np) with mean = variance = np ,
if n is large (≥ 20) and p is small (≤ 0.05)

 With c = np = average number of defects per unit,

Control limits = c + z √c
Example 5: c chart
Day Number of
A bank manager receives a complaints

certain number of complaints 1 3

2 6
each day about the bank’s 3 4
4 5
service. Complaints for 14 days
5 4
are given in the table shown. 6 0
7 2
Construct a control chart using 8 5
three-sigma limits. 9 6
10 0
11 3
12 1
13 0
14 3

Total 42
Example 5 Solution

42
c 3
14
c 1.73
Upper control limit, UCL c c  z c 3  3 1.73 8.2
Lower control limit, LCL c c  z c 3  3 1.73  2.19    0
where c is the mean and number of defects per unit.
c is the standard deviation.
For practical reasons, normal distribution approximation to
Poisson is used.
Use of p Chart and c Chart
Use of a p chart

a. Good or Bad
b. Pass or Fail
c. Operate or don’t operate
d. Dead or Alive

Use of c chart

a. Number of Complaints or failure per unit of time (e.g., hour,

day, month, year)
b. Bacteria per unit of volume (e.g., gallon, cubic foot, cubic
yard)
Process Capability
 The variability of a process can significantly impact customer
satisfaction.

 Three common terms refer to the variability of process output:

 Tolerances: Specifications for range of acceptable values

established by engineering design or customer requirements

 Control limits: Statistical limits that reflect the extent to which

sample statistics such as mean and ranges can vary due to

randomness alone
 Process variability: Natural or inherent variability in a process

 Process capability: The inherent variability of process output

relative to the variation allowed by the design specification
Capability Analysis
 Capability analysis determines whether
the inherent variability of the process
output falls within the acceptable range
of the variability allowed by the design
specifications for the process output.
LSL  USL
 The range of possible solutions:

1. Redesign the process so that it can achieve the desired output

2. Use an alternate process that can achieve the desired output
3. Retain the current process but attempt to eliminate unacceptable
output using 100 percent inspection
4. Examine the specification to see whether they are necessary or
could be relaxed without adversely affecting customer satisfaction.
Process Capability Ratio Cp

Process capability ratio C p

Specification width
Cp 
Process width
USL  LSL
Cp 
6
Motorola Corporation uses Six Sigma management.
For Motorola, C p 2
Process Capability, Cp and
Cpk
USL  LSL
Cp  and is estimated by
6
ˆ USL  LSL
Cp 
6s
 x  LSL USL  x 
C pk min  or  and is estimated by
 3 3 
ˆ  x  LSL USL  x 
C pk min  or
 3s 3s 

Healthcare Operations Management, Daniel B. McLaughlin, Julie M. Hays, AUPHA Press.

Example 6
 A Clinic management has decided that no patient should
wait no more than40 minutes, a waiting time upper
specification limit (USL) of 40 minutes. The Six Sigma team
wants to determine if the process is capable. The team
estimated that the mean and standard deviation of the
process are 30 minutes and 4.1 minutes, respectively. Note
that there is no lower specification limit (waiting time less
than some lower limit would not be a ``defect”).

 a) What is the Cpk index for the process?

 b) Is this process capable? Explain

Specification width or design width or Tolerances

Process width = 6 σ
Specification width or design width or Tolerances

Process width = 6 σ
Example 7
XYZ Lab supplies Treated Elixir Plasma Solution (550-cc bottles) to hospitals.
The quality attributes of this plasma solution are purity, absence of ADIS or
hepatitis virus, and bacterial count and the protein concentration. The
protein concentration is measured by a sophisticated electronic process
known as electrophoresis. American Medical Association (AMA) standards
specify that a 500-cc plasma should contain between 30 and 36 grams of
protein. Both concentrations under and over this range may be hazardous to
a patient’s health. The Lab recently purchased a electrophoresis processing
machine and performed extensive testing to determine the variability in the
process. The standard deviation in the process was estimated to be 1.5
gram. Is this process capable? What is the process capability ratio? What
percentage of defective products would you expect in producing one million
Plasma (550-cc) bottles?
Honda SPC
Honda SPC Quality Control Video:
https://www.youtube.com/watch?v=Sdj-8ZBY
Ymo
References
 Healthcare Operations Management, Daniel B.
McLaughlin, Julie M. Hays, AUPHA Press.
 Managing Business Process flows, Ravi Anupindi, Sunil
Chopra, Sudhakar Deshmukh, Jan Mieghem, Eitan
Zemel; 2nd edition.
 Production/Operations Management by William J.
Stevenson, Sixth Edition, Irwin/McGraw-Hill, 1999.
 Operations Management, David A. Collier, James R.
Evans, 2nd edition, 2007
 Eugene Grant and Richard Leavenworth, Statistical
Quality Control, McGraw-Hill, 5th Edition, New York,
1980