Organizations have processes and every process has a certain amount of risk associated with it due to variations. By managing process variations we can minimize risks to achieve desirable results. Every process has input variables (X) that delivers outputs (Y). Changes in (Y) is dependent on changes in (X) and can be expressed as:

Y=f(X)

MEASURE AND CONTROL VARIABLES
It is obvious from above relationship that to manage output of a process one has to measure and control input variables. To measure input variables a measurement system (measuring equipment, personnel, method etc.) is required. Every measurement system have some degree of risk associated with it due to variations or errors. Quantifying this error will minimize the risk in your measurement process and provide confidence that the variation observed in the process is not related to measurement system but it is actually due to special and common causes acting on the process.

Special causes are those that are not always acting on the process. These are due to uncontrolled variations and can be discovered and removed using control charts (X-bar and R chart etc.) to make the process stable and predictable. Examples of special causes of variations include excessive raw material variations, tool wear, improper set-up etc.

Common causes are those that are always acting on the process. These are due to inherent variation in the process. Common causes of variations are generally not controllable by the process operator. These could only be removed through management commitment to improve the process capability. Examples of common cause variations include normal raw material variations, humidity, ambient temperature etc.

INADEQUATE MEASUREMENT SYSTEM
Don’t be surprised when I say that the measurement system that you have used for years and years may be totally inadequate but no one is aware of it! What’s that? You say that can’t happen in your operation? Don’t bet on it!   I have come across measurement systems in which the variation due to measurement system was 70% of the total tolerance. Countless hours were spent to improve the process by minimizing variations while majority of variation was due to the measurement system.

Therefore, it is imperative that the measurement system error is known prior to any process improvement initiative involving measurements of process variables. There is a technique to quantify and manage error associated with a measurement system. This technique is called GR&R study.

GR&R METHODOLOGY
The GR&R methodology to quantify measurement error is simple. First let me define what is GR&R? GR&R is an acronym for Gage Repeatability and Reproducibility. There are two major elements of measurement system errors; repeatability and reproducibility. Repeatability error is associated with the measuring equipment and could be due to the manufacturing defects, while reproducibility error is associated with operator technique of taking and reading measurements.

There are two techniques to determine GR&R:

1. Long form technique: This involves usually ten (10) parts, three (3) operators and each operator measures each part three (3) times. Typically you start with the long form. If several similar measurement system check out all right, move to the short form for expediency.
2. Short form technique: This usually involves five (5) parts, two (2) operators and each operator measure each part once. The short form technique is a quick way of determining acceptability of gage variation. The drawback in measuring parts only once is that the two measurement system errors (repeatability and reproducibility) cannot be segregated.

A measurement system is acceptable if the calculated GR&R is less than 10%. If it is 10% to 30% it may be acceptable, but only if there is a justification such as technological limitations. Any measurement system having over 30% error is unacceptable.

SHORT FORM
Below is a step-by-step method to determine GR&R using short form technique:

1. Select five (5) parts that have dimensions spanning the range you want to measure
2. Mark them for identification as well as the precise spot that will be measured
3. Select two (2) trained operators to take measurements
4. Ask each operator to measure parts randomly and record results as shown in table 1
5. Calculate range for each part as shown in table 1
6. Calculate sum of ranges as shown in table 1
7. Calculate average range:

Average Range (R bar) = ΣR/5 = 2/5 = 0.4

7. Multiply the average range by 4.33 (a constant). Result represent the combined repeatability and reproducibility error of the measurement system.

GR&R = 0.4×4.33 = 1.73

Note: This method can be used with varying amounts of parts and operators but any change from the above example will change the value of the constant.

8. To convert GRR value to a percentage of tolerance, divide it by the tolerance and multiply by 100. For example if the tolerance is 20 for the study above, then GRR as a percentage of tolerance is:

GRR as percentage of tolerance = 1.73×100 / 20 = 8.66%

In the above example the measurement system is acceptable because error is under 10% of the tolerance.

If the calculated GR&R using a short form study is more than 10%, a long form study shall be carried out.

Table 1: GR&R Study-Short Form

Bio:

Afaq (Fayzee) Ahmed core skill set consists of Operational Excellence (OE) design & implementation, Management Systems implementation, Business Performance Improvements, Assessments, Supply Chain Improvements, Mega Capital Projects QA/QC and Vendor Inspection & Testing. He is holder of M.S degree in Mechanical Engineering from University of Southern California. He is the founding member of Operational Excellence Management System for Saudi Aramco as part of the ATP initiative. Prior to Aramco, he worked for US multi nationals. Afaq is a recognized ASQ trainer. He is a holder of three major certifications from American Society for Quality (ASQ) and two certifications from Exemplar Global. He is senior member of ASQ and Exemplar Global QM Committee member. He is author/reviewer of ASQ Certified Quality Auditor and Quality Engineer certification courses. He is a recipient of several awards and recognitions from multinationals.