Measurement Variation

Jay Arthur
888-468-1537
303-756-9144
KnowWare International, Inc.
DBA LifeStar
2253 S. Oneida
Ste 3D
Denver, CO 80224

We work with companies
that want to fire up their profits using
Lean Six Sigma

Copyright © 2011

One of the "five M's" in root cause analysis is measurement. Measurement error can reject good products and accept bad products. Measurement error can be split into two main causes: equipment (i.e., gage) causes and process (i.e., appraiser) causes. The QI Macros Gage R&R template can help you evaluate your measurement system.

Gage R&R

The Gage R&R template offers three main ways to evaluate your measurement system:

• Average and Range Method
• Specification Tolerance Method
• Anova Method

You'll find instructions for completing a GageR&R study on the first sheet tab in the template:

The first step is to pick a gage to evaluate, two or three appraisers and five to ten parts using these guidelines.

Important: the five-to-ten parts must span the range or tolerance you are trying to measure. If the parts are identical (i.e., no variation), all of the error will fall on equipment and appraisers.

Then, at random, have your appraisers measure each of the parts. Do all ten for the first trial and then, in a different order, do the second and or third trials. Enter your data into the Gage R&R worksheet.

NOTE: Even though you measure the parts in a different order for each trial, you still input the measurements for each part in the same column. For example, part 1 might be measured first in trial 1, eighth in trial 2 and fourth in trial 3 but you still input the measurements for that part in the same column C. In the worksheet below I entered the Ford test data from c:\qimacros\testdata\AIAG SPC.xls. There are 10 parts, three appraisers and three trials.

Number of Distinct Categories
Scrolling down, you'll find the NDC measure. For Gage R&R to accurately evaluate R&R, NDC must be 5 or greater. Ideally, NDC should be 10 or greater. Otherwise, there won't be enough part variation which will cause equipment and appraiser variation to shoot up.

Average and Range Method
As you can see from the results above, %R&R (Using TV-Total Variation) at 10.3% is greater than 10% (but not by much). So the gage system is probably acceptable. Equipment variation (EV) is greater than appraiser variation (AV), so that might be the place to start.

Most Gage R&R users rely on the Average and Range method. Unfortunately, most people don't like the insights they receive. Below are the results sent in by one customer. First, there weren't enough distinct categories (NDC = 4). This throws suspicion on the other results, but %R&R is greater than 30% which means the gage system needs improvement. %EV is 33.8% and %AV is only 7.3%. This suggests that the problem is with the gage, not the measurement process. Recommended improvements are listed on the Gage R&R template in cells T13:T16.

Specification Tolerance Method
As you can see from the results below, %R&R (using the specification tolerance of 2) at 8.3% is less than 10%. So the gage system is acceptable using the Spec Tolerance method. The Specification tolerance is entered into cell B43.

We had one customer who had Six Sigma quality (Cp=2), but their Average and Range method said they're gage system needed improvement. (Your product may be great, but your measurement process can still be suspect.) When they entered the specification tolerance (USL-LSL), however, their measurements were acceptable using the tolerance formulas.

ANOVA Method
If you scroll down on the Gage R&R form using the Ford data, you can see the ANOVA calculations with and without interactions between equipment (repeatability) and appraiser (reproducibility):

In either case, %R&R is less than 10% so the measurement system is acceptable.

Ford's expected answers to this data are:

Measurement Systems Analysis is Easy

• If you choose enough parts that vary across the tolerance
• If you conduct the study randomly
• If you choose the right method for your measurement system
  

One-Day QI Macros Workshop
If you need help using the QI Macros, consider scheduling a One-Day QI Macros Workshop at your company. Download a course outline and pricing information here or give us a call to schedule your session.To sign up for this series tell your friends to send an email to qimacros@aweber.com

Download the FREE 30-day Evaluation copy of the QI Macros Excel SPC Software for Six Sigma