Measurement System Accuracy

It is an unbiased true value which is normally reported and is the nearness of measured result and reference value. It has different components as

Bias – It is the systematic difference between the average measured value and a reference value. The reference value is an agreed standard, such as a standard traceable to a national standards body. The distance between the average measurement and reference value on the x-axis is the bias. Bias can be positive or negative, depending on where the average measured value falls. The reference value is the accepted true value determined by measurement with precision for a particular method – for example, knowing the true weight of a bolt to be 0.03 grams and using that to calibrate a scale. When applied to attribute inspection, bias refers to the ability of the attribute inspection system to produce agreement on inspection standards. Bias is controlled by calibration, which is the process of comparing measurements to standards. To calculate bias, you first calculate the average measurement of the part. Bias would equal that average measurement minus the reference value for your gauge.

Linearity – It is the difference in bias through measurements. How does the size of the part affect the accuracy of the measurement method? The linearity errors are, with respect to how much change you’re likely to see over time with measurements. Linearity is a linear change in bias over the operating range of the actual measurement device. This is true in both manufacturing and services settings. Your system should be equally accurate for all of the measuring levels. You need to measure ten different parts or samples at least five times each to get a good set of data to do the calculations. For example, the weigh scale for materials in a warehouse if a foreman finds a 2-kg error when measuring a load of 100 kg. He continues to take measurements and finds they’re off by 6 kgs for a 200 kg load. If the 2-kg error for the first 100 kgs was indicative of performance and measurements are still off 2 kgs with a weight of 200 kgs, that’s one thing. However, a 6-kg difference at 200 kgs would suggest this is not linear. It is found by, calculate bias, and perform linear regression, for a linear regression graph, if the slope is different from one, it’s non-linear. If the intercept is different from zero, the gauge has a bias. A sloped line indicates the presence of linearity in a graph. The graph plots bias on the y-axis and reference value on the x-axis. The formula for calculating linearity (L) is – L = bV(subscript p), where b, would be the slope, multiplied by the amount of process variation, or V(subscript p) that’s measured in the process.

Stability – It is the change of bias over time and usage. How accurately does the measurement method perform over time? You want to see the same results for the same sample over time when using your measurement system. A lack of stability could indicate special cause variation in your data.

Control charts are a powerful way to maintain measurement systems, and evaluate stability on an ongoing basis. Control charts have upper control limits (UCL), lower control limits (LCL), and a target value, or central line (CL). You should use control charts regularly to check the stability of a measurement system. This will help you determine the right recalibration intervals, the measurement devices, and the process for looking for root causes of problems. This could help you separate special cause, such as wobbly bearings in the machine, from assignable causes, for example related to deterioration in your measurement system.