Beruflich Dokumente
Kultur Dokumente
Quality Control
Inspection
before/after
production
Acceptance
sampling
The least
progressive
Corrective
action during
production
Process
control
Quality built
into the
process
Continuous
improvement
The most
progressive
Cost
Total Cost
Cost of
inspection
(appraisal and
Prevention cost)
Optimal
Amount of Inspection
Cost of
passing
defectives
(failure cost)
Inspection
Where/When
Raw materials
Finished products
Inputs
Acceptance
sampling
Transformation
Process
control
Outputs
Acceptance
sampling
Inspection
points
Cashier
Counter area
Eating area
Building
Kitchen
Hotel/motel Parking lot
Accounting
Building
Main desk
Supermarket Cashiers
Deliveries
Characteristics
Accuracy
Appearance, productivity
Cleanliness
Appearance
Health regulations
Safe, well lighted
Accuracy, timeliness
Appearance, safety
Waiting times
Accuracy, courtesy
Quality, quantity
Define
Measure
Compare to a standard
Evaluate
Take corrective action
Evaluate corrective action
6
Mean
x
i 1
N
N
Variance 2
2
(
x
)
i
i 1
Standard deviation
8
x
where : Standard deviation of the population
n
Sampling Distribution
Sampling distribution is the distribution of sample means.
Sampling distribution
Variability of the average scores of
people with last name R and S
Process distribution
Variability of the scores
for the entire class
Mean
Normal Distribution
normdist(x,.,.,1)
normdist(x,.,.,0)
Probab
Mean
95.44%
99.74%
11
normdist(x,mean,st_dev,1)
x
norminv(prob,mean,st_dev)
12
13
Control Limits
Process is in control if sample mean is between control limits.
These limits have nothing to do with product specifications!
Sampling
distribution
Process
distribution
Mean
LCL
Lower
control
limit
UCL
Upper
control
limit
14
Type I Error=Alpha
/2
/2
Mean
Probability
of Type I error
LCL
UCL
Control Chart
Abnormal variation
due to assignable sources
Out of
control
UCL
Mean
Normal variation
due to chance
LCL
Abnormal variation
due to assignable sources
9 10 11 12 13 14 15
Sample number
17
UCL
LCL
1
Sample number
18
Control Charts
Control charts for variables (measurable
quantities), e.g. length, temperature
Mean control charts
To check mean
c-charts
To check the number of defectives (occurrences)
19
20
Multipliers D_4 and D_3 depend on n and are available in Table 10-2.
EX: In the last five years, the range of GMAT scores of incoming PhD class is
88, 64, 102, 70, 74. If each class has 6 students, what are UCL and LCL for
GMAT ranges?
R (88 64 102 70 74) / 5 79.6. For n 6, D 4 2, D 3 0.
UCL D4 R 2 * 79.6 159.2 LCL D3 R 0 * 79.6 0
21
Detects shift
x-Chart
LCL
UCL
R-chart
Does not
detect shift
LCL
22
Sampling
Distribution
UCL
Does not
reveal increase
x-Chart
LC
L
UCL
R-chart
LC
L
Reveals increase
23
Use of p-Charts
p=proportion defective, assumed to be known
When observations can be placed into two categories.
Good or bad
Pass or fail
Operate or dont operate
Go or no-go gauge
UCL p z p
where p
LCL p z p
p (1 p )
, z as before
n
24
Use of c-Charts
c=number of occurrences per unit
Use only when the number of occurrences per unit can be counted.
UCL c z c LCL c z c
if c is not known, use the average c
25
C-chart Example
While the nuclear submarine Kursk was being raised in the
Barents sea (between Svalbard, No and Novaya Zemlya, Ru),
which took 15 hours, engineers took a reading of number of
Geiger counts per hour to detect any increase in radiation
levels. Should they have stopped before 5th or 10th hour given 3sigma control and the readings data: 42, 48, 50, 45, 52, 66, 64,
84, 92, 76.
At the 5th hour, average number of counts=47.4, stdev of counts=6.88,
UCL=47.4+3*6.88=68.05, LCL=47.4-3*6.88=26.75. Do not stop.
At the 10th hour, average number of counts=61.9, stdev of counts=7.87,
UCL=61.9+3*7.87=85.51, LCL=61.9-3*7.87=38.29. Stop, 9 th reading is
out of control.
26
(r=8 runs)
27
EX: What are 3-sigma UCL and LCL for the number of runs in 50 samples?
2K - 1
16 K 29
K 50, E(r)
33 and r
2.92
3
90
UCL E ( r ) z r 33 3 * 2.92
LCL E ( r ) z r 33 - 3 * 2.92
28
Process Capability
Tolerances/Specifications
Requirements of the design or customers
Process variability
Natural variability in a process
Variance of the measurements coming from the process
Process capability
Process variability relative to specification
Capability=Process specifications / Process variability
29
Process Capability:
Specification limits are not control chart limits
Lower
Specification
Upper
Specification
Sampling
Distribution
is used
Upper
Specification
30
31
32
Lower
specification
Process
mean
+/- 3 Sigma
+/- 6 Sigma
33
Chapter 10 Supplement
Acceptance
Sampling
34
Acceptance Sampling
Acceptance sampling: Is a lot of N products good
if a random sample of n (n<N) products contain
only c defects?
For example take a sample of 10(=n) milk bottles out
of every 100(=N). If 1(=c) or more bottles do not fit
specifications, reject the entire lot of 100 bottles.
36