Upper control limit (UCL)
CONTROL CHART Average
or Lower control limit (LCL)
SHEWART CHART
Basic types of Data
Discrete Data or Attribute Data Continuous Data
• Sort items into distinct, separate non overlapping • Measured on a infinitely divisible continuous
categories
scale
• Possible values are clearly separated from one
• Unit of Measure can be subdivided and it has
another
a meaning.
• Come is discrete units that cannot be subdivided
• Mostly Measured using an instrument
• Based on Counts
• Ex – Length, Weight, Time, Diameter
• Ex – No. of Defects; Good, Average or Bad; Ratings
Continuous <-> Discrete
• Some things that can be measured on a continuous scale can be converted into discrete measures.
• So, if you start with continuous data, you can always convert it to discrete categories.
• In contrast, if you start with discrete data, it’s usually impossible to convert it to continuous data.
• Commonly used Data Collection method
• Testing every single product is expensive, impossible and
sometimes highly destructive.
Sampling
Method
One grain is enough to know if the whole pot is cooked.
Types of Sampling
Sampling according to
Every Item of the some defined norm
population has a
equal chance Ex - Every 10th unit or
Fist / Middle / Last
The population is first split into
Diving population groups. The overall sample
into subgroups consists of every member from
based on some some of the groups.
common
characteristics An airline company wants to
survey its customers one day,
Ex - Stratification by so, they randomly select 5 flights
Location, RM, Shift, that day and survey every
Machine etc. passenger on those flights.
Control Charts
• Purpose
• To help people manage systems to make the right
decisions.
• Benefits :
• Proactive monitoring of process
• Indicates the stability of the process
• Shows how the system reacts to the change.
(Sampling mean is
shifting upward,
but range is
Xbar – R
consistent)
Chart X-chart
(X-chart detects
shift in central
Example tendency)
(R-chart does not
R-chart detect change in
mean)
(Sampling mean
is constant, but
dispersion is
Xbar – R increasing)
Chart (X-chart
Example X-chart indicates
no change in
central tendency
R-chart (R-chart detects
increase in
dispersion)
Objective of using Control Chart
Stable Process Unstable Process
Only “common” “Special” causes of
causes of variation variation are present
are present
Data Consideration
1) The data should be continuous
2) The data should be in time order
• Enter the data in the order it was collected
3) The data should be collected at appropriate time intervals (Frequency)
• Collect data at equally spaced time intervals, such as every hour, every shift, every day.
• Select a time interval that is short enough that you can identify changes soon after the changes occur.
4) The subgroup size should be 8 or fewer observations
• Size between three and eight is statistically efficient. Most commonly-used size is five
5) The data should be in rational subgroups
• Sample produced in a short period of time under same conditions (operator, equipment, supplier)
• Representative of the output from a process.
Control Chart selecting chart
Type of Data
Count or
Measurement
Classification
(Variable Data)
(Attribute Data)
Subgroup Size of Subgroup Size < 8 Subgroup
Count Classification
1 - 10 Size > 8 - 10
Defects Defectives I-MR X-bar & R X-bar & S
Variable Sample Variable Sample
Fixed Sample Size Fixed Sample Size
Size Size
C Chart U Chart NP Chart P Chart
X - R Chart
For the averages chart: For the range chart:
CL = X CL = R
How do I UCL = X + AR 2 UCL = DR 4
calculate LCL = X − A R 2
LCL = D R 3
the control n
2
3
D4
3.27
2.57
D3
0
0
A2
1.88
1.02
limits? 4
5
6
2.28
2.11
2.00
0
0
0
0.73
0.58
0.48
7 1.92 0.08 0.42
8 1.86 0.14 0.37
9 1.82 0.18 0.34
X = average of the subgroup averages
UCL = upper control limit
R = average of the subgroup range values LCL = lower control limit
A = a constant function of subgroup size (n)
2
For varied sample size: For constant sample size:
How do I P(1 − P)
UCLnp = n P + 3 n P(1 − P)
calculate UCLp = P + 3
n
the control LCLp = P − 3
P(1 − P)
LCLnp = n P − 3 n P(1 − P)
limits? n
p and np Note: P charts have an individually calculated control limit for each point plotted
Charts P = number of rejects in the subgroup/number inspected in subgroup
P = total number of rejects/total number inspected
n = number inspected in subgroup
How do I For varied opportunity (u):
For constant opportunity (c):
calculate UCLu = U + 3
U
UCLC = C + 3 C
the control n
U
limits? LCLu = U − 3
n LCLC = C − 3 C
c and u Note: U charts have an individually calculated control limit for each point plotted
Charts C = total number of nonconformities/total number of subgroups
U = total number of nonconformities/total units evaluated
n = number evaluated in subgroup
Chart interpretation for abnormalities – Rule 1
Point outside the limits
Chart interpretation for abnormalities – Rule 2
7 Points in a row above or below the mean line.
Chart interpretation for abnormalities – Rule 3
7 Points in a row descending or ascending
Chart interpretation for abnormalities – Rule 4
Too Close to the average
Chart interpretation for abnormalities – Rule 5
Too far away from the average
Chart interpretation for abnormalities – Rule 6
Cyclic pattern
Out of Control Action Plan
Benefits • Proven technique for improving productivity.
of • Effective in defect prevention.
Control • Prevents unnecessary process adjustment.
• Provide diagnostic information.
Charts
• Provide information about process capability
