Page 90 - Six Sigma Advanced Tools for Black Belts and Master Black Belts

OTE/SPH
OTE/SPH
2:55
Char Count= 0
August 31, 2006
JWBK119-06
Introduction 75
6.1.2 Measurement variation
Another main source of variation is measurement variation. In continuous improve-
ment programs, design and appropriateness qualities gain their significance only after
measurement activities.
The quality of measurement data depends on the statistical properties of multiple
measurements. There are many factors leading to the variability of the measured
quality characteristic against the real value. If the instrument fails to repeat the same
value of measurement, this is defined as within-instrument variation and is called
2
“repeatability”inmeasurementsystemsanalysis(MSA). Ontheotherhand,variation
amongthemeasurementinstrumentsoccurswhendifferentinstrumentsgivedifferent
readings for the same quality characteristic of a specimen. The measuring device
has to be calibrated from time to time. Different calibration methods will result in
different degrees of precision of the measuring device and thus lead to the variation
between calibrations. The variability due to different operators using the instrument
is between-operators variation and is called “reproducibility” in MSA. Also, there will
be unavoidable variability across time.

6.1.3 The observed variation
Generally, in any activity involving measurements, some of the observed variability
will be inherent in the units or items that are being measured, and some of it will result
2
from the measurement that is used. The variance in the recorded observations, σ ,
tot
is comprised of two components: the variance of the product (process characteristic),
σ 2 and the variability of the measurement system σ 2 :
proc meas
2
2
2
σ tot = σ proc + σ meas .
A possible impediment to the effect implementation of the DMAIC approach in
problem-solving is measurement error. Consequently, it is important that measure-
ment systems be in a state of statistical control. Figure 6.2 shows the relationship of
the two components with the variance of the recorded observations.
With statistical procedures in general and control charts in particular, a primary
objective is to analyze components of variability so that variability due to assignable
causes can be detected, and to maintain the process with only natural variability, often
referred to as a ‘stable system with chance causes’. The chance causes of variation

σ 2 proc σ 2 tot

σ 2 meas

Figure 6.2 Relationship between total σ process σ, and measurement system σ.