Measurement System Analysis

Measurement System Analysis (MSA)

Verifies measurement systems before trusting data

4 MSA questions

does tool give accurate readings? are results consistent? do different operators get similar results? does system remain stable overtime?

operator element

how ___ skill, training and consistency affect measurements

equipment element

Account for ___ limitations and calibration

method element

Standardize measurements ___ to reduce variation

environment elements

Consider environmental factors like temperature and humidity

accuracy

closeness to true or reference value

precision

repeatability of measurements under the same conditions

Accuracy without precision

scattered measurements around true value

Precision without accuracy

tight cluster far from true value

Inaccurate and scattered measurements

neither precision nor accuracy

measurement resolution

smallest detectable measurement change; higher = more detailed data

apply the 10:1 ruler

resolution should be 10 times finer than tolerance

linearity

consistency of bias across range; output changes proportionally with the input

Slope of 1

error does not scale with measurement

Intercept of 0

no constant bias exist across measurements

Slope = 1, intercept = 0

perfect linearity

Slope ≠ 1

proportional error changes constantly

Intercept ≠ 0

constant bias across range

Stability

consistent results over time

stability

consistent results over time

how to conduct a stability study

measure unchanged references over time, use control charts to visualize, monitor samples, display results, and spot patterns easily

bias

difference between average of measured values and true values

 Gauge Repeatability and Reproducibility (GR&R) Studies

Evaluate variation from the measurement process itself

Repeatability

same operator uses same equipment to measure same part

Reproducibility

different operators measure same part using same equipment

<10% variation

acceptable system threshold

10-30% variation

address both equipment and operator issues and focus on dominant variation source

>30% variation

needs improvement; consider alternative measurement methods