- Statistical Process Control
What is Statistical Process Control?
Statistical Process Control is a quality tool that has been
around since the 1920's. It was developed by Dr. William
Shewhart, as a means of controlling processes, using statistics.
It was also identified as a tool that could easily be used by
those who did the work.
Using simple data collection techniques, straightforward
calculations, and graphing tools, those who make the product,
run the process, or provide the service can get immediate
feedback as to how the product, process or service is
Other tools associated with Statistical Process
Process Control primarily utilizes variable and attribute
control charts to monitor products, processes or services. Other
components of SPC include Pareto Charts, Cause & Effect
diagrams, Histograms and Capability Analyses.
Who is involved in Statistical Process Control?
Anyone within the organization who has a responsibility for the
quality outcome of a product, process or service, should have
SPC knowledge. There should also be some central coordinator,
and a steering committee that ensures the results are shared
with the rest of the organization.
- Design of Experiments
What is a DOE?
A Design of Experiment is a tool that was first developed as a
result of a famine in Great Britain. It has since become a vital
quality tool to test the cause & effect relationship between
the inputs and the outputs. A well-designed DOE eliminates the
effect of all outputs except the ones that can be controlled. If
the output, or response changes significantly, there is a
great possibility that the output is related to an input
variable that has changed, rather than some other independent
variable that has not changed.
A DOE can be used to manipulate process inputs to get a better
understanding of the effects on the process outputs. It is a
test or a sequence of tests, known as runs, where potential
important variables are changed or tweaked in a systematic
matter to again see the effect the adjustments have on the
Typically, experiments are designed to test the interaction of
multiple variables, rather than altering one at a time.
Isolation of one factor at a time could lead to erroneous or
missed information and results.
Advantages of a Design of Experiment
Contrary to belief, results from a DOE can be obtained in a
rather timely fashion, and for a relatively low cost. There is
an excellent chance that the optimal variable levels will be
detected. The results will yield a high level of confidence.
Another advantage realized by a DOE is that there is an
increased ability to identify independent main effects and any
interaction effects, depending on the design chosen.
Who is involved in a DOE?
Design of experiments should be the responsibility of a
cross-functional team that includes members from the area or
areas where the understanding of the product, process or service
is to be understood or improved. The team should also include
anyone else with insight into the product, process or service.
What is Parameter Design?
Parameter Design is a statistical technique of establishing the
optimum parameters of a given product, process, system or
service. The goal of parameter design is to determine the
parameter values of a product, process, system or service such
that it is functional and exhibits a high level of performance
with a minimum sensitivity to external influences (noise
A parameter design experiment will typically involve two types
of factors, the design (or controllable) factors, and the
noise factors. A design factor is a factor whose level
can and will be set and maintained.
noise factor is a factor whose level either cannot or will not
be set or maintained, yet could affect the outcome (or levels of
performance) of the functional characteristics. Generally, these
factors are of the format of being too difficult, too expensive
or impossible to control.
What is Tolerance Design?
Tolerance design is a statistical technique, the tolerances of
the input parameters are adjusted to obtain a desired output
variation; which should be at a minimum. The objective in a
tolerance design experiment is to determine the allowable ranges
of variation for the product, process, system or service
The experiment will identify those parameters whose variation
affects the output variation. This effect could be linear, or
quadratic, or of some other format. As a result, it is possible
to determine which parameters' tolerances can be tightened and
to what limit.
The technique used to determine the amount of total variation
due to each of the factors is a statistical tool known as ANOVA,
ANalysis Of VAriance. The factors that
contribute large amounts of variability are the ones that will
be considered in an effort to tighten tolerances.