|
A method of asking 'why' five times to
get to the root cause of a problem. First
state the problem:
The machine broke down:
why did it break down?
It seized up for lack of oil
why did it run out of oil?
the maintenance person did not oil it
why didn't the maintenance person oil it?
the oil filler is impossible to access
why is the oil filler impossible to access?
The machine guard conceals it
In this case it has taken four 'whys' to
get to the root cause of the problem. The
solution might be to cut a hole in the guard,
or extend the oil filler tube.
A way of representing activities and dependencies
involved in a complex project. There are
several versions of Activity Network Diagram,
the main types being PERT and CPM.
| Box
Plots (Box and Whisker Plots) |
|
Box Plots are used to represent
relatively small data sets. The standard
format of a box is shown below:
The outliers are points that are more than
1.5 times the interquartile range above
the third quartile or below the first quartile.
The whiskers extend to the largest, and
smallest, data values that are not outliers.
A tally sheet that is used to collect data
. It is a simple method that can be used
with a minimum of instruction:
A Pareto Chart would often be a useful
next step for analyzing the results.
| Consensus
Criteria Method |
|
See Prioritization Matrices
Critical Path Method. Essentially the same
as PERT, although there are some technical
differences.
A method of converting the Voice of the
Customer into specific features (an alternative
to QFD). It involves creating a tree map
with the root being the Voice of the Customer
and the branches leading to specific features:
This is an extension to the Big 'Y' little
'y' concepts, or cascading a process 'x'
into a 'y' on the next level down:
Time based variations with
a repeating pattern:
The chart show Average Length
of Stay for four quarters in each of four
years. The pattern repeats each year, a
pronounced drop in the April quarter increasing
in the July and October quarters.
A graphical tool for helping to chose between
several courses of action. The tree starts
with the decision that has to be made. The
branches are then formed from the future
decisions that the initial decision will
leads to, or to the possible outcomes from
the decisions:
A method of identifying the
root causes of potential faults using a
tree type structure:
It is usual to use the Boolean
'and' and 'or' symbols. A possible cause
of motor failure is overheating (1). This
can happen if the ventilation slots are
blocked (4) OR the motor
load is overloaded (5)
A diagram, or table, that shows the number
of observations falling into each of several
ranges of values. A histogram is a common
method for representing a frequency distribution.
A chart showing the work breakdown against
time. The vertical axis shows the activities
and the horizontal axis the time (in days,
weeks, months etc.)
A graphical method that represents the
distribution of values in a data set. The
data values are grouped into ranges and
shown as bars on a histogram:
See QFD
A model developed by Noritako
Kano. There is an 'expected' quality that
customers take for granted. This is the
minimum functionality, and will not satisfy
the customers:
The 'expected quality' represents a reasonable
level of quality and service, for example
timely and responsive customer service.
The organization should aim for 'Exciting
Quality', giving customers more than they
expected. If you book into a hotel and unexpectedly
find a bottle of wine and a box of chocolates
waiting for you that is 'Exciting Quality',
at least until all hotels adopt the idea
and then it becomes 'Expected Quality'
A diagram used to analyze the associations
between two groups of ideas. It is a simplified
version of Quality Function Deployment (QFD):
A method for removing the turbulence from
data to better reveal underlying trends.
Each point is the average of 'n' consecutive
periods.
Charts that show patterns of variation
from several possible causes. The example
shows a comparison of LOS (Average Length
of Stay) for two procedures at four hospitals
over a four month period:
For more information on interpreting
patterns see Positional , Temporal and Cyclical
variation.
A type of ordered histogram often used
to evaluate the frequency of occurrence
of defects. It is called a Pareto Chart
because the distribution of defects typically
obeys the Pareto Principle, 20% of the defect
categories account for 80% of the defects.
It is usual to include the grouped results
in the bars, and a line graph of the cumulative
total.
The Pareto principle was originated by
an Italian economist, Vilfredo Pareto, who
found that 20% of the population owned 80%
of the resources. This principle applies
to many phenomena, in process improvement
it is often found that 80% of problems will
result from 20% of the causes. Sometimes
expressed as ‘The vital few and the trivial
many’.
The shape of the graph is recursive; once
the major source of defects has been removed
the principle still holds.
Program Evaluation and Review Technique.
A graphical method (often computerized)
for managing large projects:
The lines represent activities, the numbers
on the lines are durations and the nodes
(circles) represent milestones,
The pattern varies with the position, the
width is greatest on the RH Lower quadrant
of the part and least on the RH Top quadrant.
These are used to help prioritize decisions.
There are several forms, the one illustrated
is the Consensus Criteria Method:
Quality Function Deployment is a method
used to relate the characteristics that
customers value to the specific features
and specifications of the product. QFD is
similar to the Matrix Diagram, but contains
more features:
The QFD diagram is often called
the 'house of quality' from its appearance.
| Quality
Function Deployment |
|
See QFD
The idea that problems have a 'root cause'
that is not necessarily obvious at first
sight. For example if a clerical worker
makes mistakes then the apparent cause may
be carelessness. Further investigation might
reveal that the root cause is incorrect
procedures, poor lighting, the need to read
indecipherable documents etc. Tools for
Root Cause Analysis include the 5 Whys and
Fault Tree Analysis.
A chart that plots values in time order:
The center line is the median
value. A run is defined as one or more consecutive
points on the same side of the median. The
runs are analyzed, patterns that indicate
a 'special cause' are:
- the presence of too few, or too many,
runs for the number of points
- too many data points in a run
- an unusually long sequence of ascending
or descending data points
- too many points forming a zigzag pattern
Many tools are used in six sigma to help
with a process improvement activity. This
topic gives a selection of, mainly graphical,
process improvement tools that are useful
at various stages of the DMAIC cycle.
Other process improvement tools have been
included elsewhere in this glossary. You
might often see references to the 'seven
quality tools' or the 'seven old tools';
these are the Cause and Effect diagram,
Check Sheets, Control Charts, Histogram,
Pareto Chart, Scatter Plot and Run Charts.
The list varies from reference to reference,
some lists replace the Run Chart with Flow
Charts (I consider both run charts and flow
charts are essential tools).
A stem and leaf plot is similar to a histogram,
but uses the decimal places to build up
the bars. The advantage of this is that
the plot contains all of the original data.
The dataset:
1.0, 1.1, 1.5. 1.7, 1.9,
2.4, 2.5, 2.5, 2.5, 2.6, 2.8, 2.9, 3.0,
3.3, 3.5, 3.8
can be represented by:
| Stem |
Leaves |
| 1 |
0 |
1 |
5 |
7 |
9 |
|
|
| 2 |
4 |
5 |
5 |
5 |
6 |
8 |
9 |
| 3 |
0 |
3 |
5 |
8 |
|
|
|
The scheme can be modified to suit the
spread of the data values. For example,
the stem could be increased in increments
of 0.5, rather than integers.
There is variation with time (this may,
or may not, also be cyclical):
A variation of the moving average where
the 'n' periods are given different weights,
the purpose being to put more weight on
the most recent period.
|
