Measurement System Analysis (MSA)

We communicate that we are data-driven managers. We are modern, digitised and proud of that. We decide based on these data.

Are the data trustworthy? Not necessarily. AF447 flight on 1 June 2009 tragically crashed into the ocean, killing 228 people. The investigation concluded that the airspeed was incorrect in the cockpit due to pitot tube icing, and the pilots did not handle the situation correctly.

Such a catastrophe demonstrates how important acquiring reliable data is. Several of my articles (not only on Operational Excellence but also on Artificial Intelligence) emphasised this point. We may not induce such accidents in the corporate world, but the financial effect can be enormous.

Why does it happen?

When we perform an uncomplicated measurement, we may make mistakes. Everyone has a custom way of handling the gage. Although several methods have been invented and introduced to eliminate these differences, occasionally, these controls fail due to a lack of training, ability, or complex processes.

What shall we do?

Statistics can help us. The answer is in the automotive industry's widely-used Measurement System Analysis (Gage R&R). The method does not require extreme effort: we need to select 10 parts, number them and measure by three persons, everybody twice. Sound statistical software (like Minitab) guides us through (a random order) the measurements and analyses the results. It identifies how much variation came from the parts, operators and method. The software converts results into a number of distinct categories (NDC). If it is over 10, we are fine.

I show what that means in an example. I drew a line and selected several points on this line as if they were measurement data. Then, I added noise according to 8 and 2 NDCs. You can see in the Figures that the expected R2 is reduced to 96,8 and 73,4%, respectively. We should question the relationship under 90% of R2, but we know it was a dependency.

?

What does it cause? This additional variation pushes good parts into defective parts. On such an occasion, we started from 3 NDCs, improved and stabilised the measurement system and saved significant money for the company (in the range of 200k EUR).

?

#projectmanagement #interim #sixsigma #medicaldevices #operationalexcellence

?