MSA: Excessive variation can generate out-of-spec results

By Jackie Graham, Ph.D.

Recently we discussed variation in measurement systems, raising the question of when a measurement study should be completed.

The rule is very simple:

Whenever a measurement is being used to assess the quality or quantity of a product, a measurement system study is required.

This does mean that all measurement systems should be assessed statistically.

Of course, some kind of priority has to be set, as it is obviously impossible to complete an assessment of every system in an organization immediately. So, we suggest the following priorities:

New measurement systems/equipment.

Assessment of new equipment is a good way to ensure that it meets the organization’s needs. One positive outcome of statistically assessing equipment is that it shows whether different people can work effectively with the equipment. It also gives a performance base line for the equipment, so if there is deterioration in the equipment, the study will be able to quantify the problem.

Measurement systems/equipment being used for SPC.

If the variation from the measurement system is high, then control charts will show changes in the measurement equipment, not changes in the process! So, it is essential to assess measurement systems statistically prior to implementing SPC.

Since trends and changes apparent in SPC charts can come from the measurement system, it is important to understand the effect of measurement variation when trying to track down issues.

Measurement systems/equipment used at critical decision points.

If a measurement is taken to assess whether to pass or fail a batch, then it is essential that the measurement system is able to complete the task consistently and reliably.

A common comment we hear from customers is: “We’ve been using this equipment for years; it has never been a problem. We regularly calibrate it and it has certificates. Why should I assess it?” Remember, a statistical assessment is an accurate picture of the everyday variation in measurements. Equipment can pass calibration easily and yet fail the statistical assessment. Often, measurement systems are viewed as being correct, beyond question. Don’t be blinded to this critical area of variation.

You can also use this kind of assessment in other circumstances, these include:

  • Ensuring that you and your customer use similar methods of measurement
  • Ensuring that you and your suppliers use similar methods of measurement
  • Ensuring that different locations within the organisation measure in a similar way
  • Assessing a measurement system before and after repair
  • Preventing measurement deterioration
  • Ensuring that a new tester is fully trained
  • Assessing two different methods of testing
  • Assessing the impact of changing environmental conditions

In the next article, we will look at how to set up a measurement study. GAGEpack is an excellent software package for measurement systems analysis.

Originally published in the April 2007 edition of Quality eLine, our free monthly newletter.

Subscribe now!