Control Charts Template
Control Charts Template
In monitoring, the main focus is on the control of the improved process. For this we use
Statistical Process Control (SPC).
SPC is the application of statistical techniques for measuring and analyzing the variation
in processes.
The primary tools in SPC are control charts. They were devised by W. A. Shewhart and
are based on the fact that variability does exist in all repetitive processes.
What do you think is the main difference between a control chart and a run chart?
a. A control chart has specification limits while a run chart has control limits.
b. A control chart has control limits while a run chart has specification limits.
c. A control chart has control limits while a run chart has neither control nor
specification limits.
d. There is no difference.
(The correct answer is C)
Control limits are upper and lower bounds on a control chart within which variation is
acceptable and beyond which they are unacceptable.
This forms the basis for plotting the control limits. Any point beyond these limits can be
attributed to special causes, which need to be investigated and then controlled.
Control charts are primarily used to distinguish between common and special causes of
variation and thus control processes.
Control Limits v/s Specification Limits
Represent the voice of the process Represent the voice of the customer
Control limits determine the stability of your improved process whereas specification
limits reveal the process capability.
Process A: Here the control limits fall within the specification limits. This means that the
process is so capable inherently that no defects are produced.
For determining stability, the processes would have to be plotted on a control chart over
time.
If defectives are being considered and sample sizes are equal, then use np-
charts
If defectives are being considered and sample sizes are unequal, then use p-
charts
If defects are being considered and sample size are equal, then use c-charts
If defects are being considered and sample sizes are unequal, then use u-charts
Remember!
A defect is a “non-conformance”.
Whereas a defective is a unit of product that contains one or more defects.
Question:
Component Defects
1 5
2 7
3 0
4 2
5 0
A phone center supervisor selected 4 calls at random and counted the rings before pick-
up for 20 days. The following data was collected:
Day s1 s2 s3 s4
1 1 4 6 4
2 3 4 5 1
3 4 5 5 3
4 6 4 4 5
5 1 3 7 5
6 8 3 6 3
7 7 3 6 9
8 5 7 4 3
9 4 5 9 5
10 7 5 6 5
11 4 5 6 5
12 6 3 8 2
13 3 5 7 6
14 6 3 2 6
15 7 6 4 4
16 6 2 6 3
17 5 5 0 5
18 6 7 6 7
19 6 4 4 5
20 6 5 5 4
Reason:
Since the data is of continuous nature (there can be any number of rings starting
from zero to infinity before the call is picked up) and sample size is 4 (which is
less than 10), an X-Bar and Range chart should be used for analysis.
The X-Bar chart plots the average of the values for a day. In the example above,
average for the first day (the first plotted point) is (1+4+6+4)/4 = 15/4 = 3.75
The Range chart plots the range of the value where
Range = (highest value – lowest value). In the example above, range for the first
day (the first plotted point) is (6-1) = 5
Q: Why should we use a combination of two charts like X-Bar and R-Chart or X-Bar and
S-chart for continuous data?
A: One chart monitors the variation between data points over time.
The second chart monitors the variation within data points over time.
For example, suppose we are tracking cycle time to solve production abends. We
randomly select 20-30 production abends each day and plot the average. The X-
bar chart will monitor the variation in the abend on each day.
The S-chart (or the R chart) on the other hand will monitor the variation (standard
deviation or Range) from sample to sample. It is possible that the averages could
remain fairly consistent, but the spread within the samples could vary.
By monitoring both the variation between data points and variation within data points, we
can detect special variation more effectively.
Control Chart Example 2
The online attendance application and the timesheet application were created in TCS,
SEEPZ replacing the manual system.
The tools group monitored the compliance to the above-mentioned tools. 250 employees
are chosen randomly and reviewed each day for a period of 20 days and the following
data for the no of defaulters was compiled. (Any employee not adhering to the usage of
both the tools will be considered as defaulters).
No of No of No of No of
Day Defaulters Day Defaulters Day Defaulters Day Defaulters
1 12 6 14 11 2 16 11
2 22 7 16 12 15 17 17
3 14 8 19 13 8 18 12
4 15 9 11 14 12 19 11
5 12 10 1 15 12 20 21
Q: Which control chart will be used to plot the data and why?
A: As the data is of discrete nature (number of defaulters will vary from zero to 250 only)
and sample size is constant (250), c-chart is used for data analysis.
C Chart for C1
25
3.0SL=23.60
20
Sample Count
15
C=12.85
10
-3.0SL=2.096
1
0 1
0 10 20
Sample Number
A web page design team reviews the developed interface pages for detecting the defect
and eliminating it before delivery to the customer. The data collected for the number of
defects in the total number of Interface pages reviewed per day is as below:
Q: Which control chart will be used to analyze the above data and why?
A: The data is discrete and the sample size is not constant, so u-chart will be the best
tool to analyze this data as shown below:
Q. Lets see if you can select the correct control charts for these cases:
1. The technical support area is monitoring the average response time on help
requests. Five calls are sampled every hour.
2. A team tracks the number of fields left blank on an application. Each day a sample of
100 applications is audited.
3. Another team does the same thing (counting the number of fields left blank on an
application), only they are looking at all the applications that come in daily. Each day
50-100 applications are turned in.
4. The team is tracking the number of abandoned calls to a customer call center in a
shift. Each shift takes from 200-400 calls.
Options are:
1. X Bar and R
2. IMR
3. X Bar and S
4. P Chart
5. NP Chart
6. C Chart
7. U chart
Correct Answers:
1. X Bar and R Chart
2. C Chart
3. U Chart
4. P Chart
Through the use of the scientific method, extended to take account of modern statistical
concepts, it has been found possible to set up limits within which the results of routine
efforts must lie if they are to be economical.
-Walter A. Shewhart, 1931
Control charts have control limits that are placed +/- 3 estimated standard
deviations (SD) from the average line for both types of data.
Control limits are calculated using control chart factors and the Range-Bar (an estimate
of short-term sigma)
Shewart invented control chart factors in the 1920s to avoid long calculations.
There are standard tables for control chart factors according to the sample size for each
subgroup.
Individual control charts are assumed to have a sample of size 2, the number of data
points make up the moving range
When special cause variation is identified and controlled and the process is
operates at a new level or with reduced variation
When a fundamental change has been made to the process and the change has
had a detectable impact on the performance of the process
As long as the process does not change, the control limits should not be recalculated.
X3
X1
X2
Consider the temperature of water delivered by your bathroom geyser as shown in the
figure:
Y = temperature of water at the outlet of the geyser
X1 = temperature of water at the inlet of the geyser (Input X)
X2 = Flow of water at inlet
X3 = Power input to coil (Process X)
The temperature at outlet of geyser (Y) is directly dependant on the temperature of water
at the inlet (X1), the flow of water at inlet (X2) and the power input to coil (X3).
If you want to control your Y to a specific value, it is pertinent that you focus on your Xs
(X1 and X2 in this case, assuming X3 to be fairly constant).
It is essential to plot control charts for the vital Xs in order to ensure firm control over Y.
Dividing the control chart into zones can aid in detecting special cause variation. Each of
the zones represents standard deviations from the mean. Zone C, for example, is + and
- one standard deviation from the average.
Moreover, whenever any data point crosses the +/- 3 standard deviation limits, it is a
special cause that needs to be investigated.