How can SP help you?
- Review the measurement chain and assess it against the product requirements
- Review the methods of measurement
- Review the methods of calibration
- Review the measuring equipment itself
- Appraising the entire chain
- Production of an uncertainty of measurement budget
- Advice on modifying methods and procedures, if necessary
- Optimisation of uncertainties of measurement
- Support for decision-making based on measured values
What is traceability?
Traceability is an important concept that provides comparisons of the accuracy of measured results, regardless of where they have been made. Accuracy applies both to equipment and to the methods used, and is determined by an unbroken chain of comparisons back to national or international standards, with each link of the chain having its own known and specified uncertainty of measurement.
What is uncertainty of measurement?
If the result of a measurement is to be used as a reliable basis for decision, it need to be expressed as a measured value, qualified by information on the uncertainty of measurement of the value. Calculation of uncertainty of measurement is based on knowledge of the various effects that can affect the measurement. As the uncertainty of measurement of each comparison must be added to produce the final uncertainty of measurement, its value increases the further down the chain (i.e. away from the national/international standard) that the measurement is made. The requirement for more exactly known uncertainty of measurement increases the further up the chain that the measurement is made.
Why is uncertainty of measurement important?
When making decisions that require comparison with permitted limit values, problems can arise if the measured values are close to the tolerance limit and where the uncertainty of measurement overlaps the limit. It is therefore necessary to know how to respond to such cases.
Does the uncertainty of measurement overlap the tolerance limit? Reduced uncertainty of measurement would be one way of reducing the risk of wrong decisions. But uncertainty of measurement can never be zero, so who should take the risk? The "manufacturer" or the "user" of the result?
The most usual response is to share the risk, i.e. directly to compare the measured value with the tolerance limit. However, the parties should be in agreement on a maximum uncertainty of measurement, so that the risk presented by the decision can be estimated.
Other alternatives include adding the uncertainty of measurement to the measured value before comparing the result with the tolerance limit, thus reducing the risk of an incorrect approval, or subtracting it before making the comparison, thus reducing the risk of an incorrect failure.
Ultimately, of course, the 'optimum' uncertainty of measurement can be decided by finding a balance between the costs of measurements and the costs of the consequences of incorrect decisions, when deciding on whether a result is within or outside the permitted value.
How can measured data be processed?
The values of some parts of the uncertainty of measurement chain can be estimated by using statistical methods (Type A) by investigating the standard deviation of the measurements. Estimates of other components can be based only on experience or other information, such as specifications from instrument manufacturers (Type B). Errors in measured/displayed values of which the magnitude and sign are known should be corrected rather than be incorporated in the uncertainty of measurement. Note that the specification of an instrument is often only a small part of the entire uncertainty of measurement.
How can SP assist your application of measurement methods?
It is not entirely easy to decide on the correct calibration interval for measuring equipment. In some sectors, the choice can be decided by legal requirements or by requirements specified by the particular industry sector concerned. If your company is not subject to such requirements, you need to ask yourself a number of questions before deciding on a suitable calibration interval.
Correct calibration intervals
Calibration intervals are normally in the range of six months to three years. However, the need for calibration can change during the life of an instrument or process, and should be determined by the following factors:
- Are there legal or industry-sector requirements?
- How important is the quality of the results? What are the (economic) risks that would result from a measurement error remaining undetected until the next calibration?
- How carefully is the measuring equipment handled and used? Is it in constant use, or only occasional use?
- Does the company make its own checks between the formal calibrations?
In many cases, information on uncertainty of measurement is available or provided for only one particular part of the chain of measurement. It is important to obtain an overall picture of the entire process in order to arrive at a budget for the whole, and to show how the traceability of the chain of measurement can be established in as optimum a manner as possible.
Reduce the amount of measured data
Based on our experience we find that, in many cases, although companies collect a large quantity of measured data from many different measurement points, there is no proper connection between them. The measurement chain can often be shortened, and the quantity of data reduced, by assembling the whole into a proper system.
SP can help you to find the most effective processes
The purpose of this work is to create a process having a suitable number of measurement points, relevant methods and suitable calibration intervals (or own inspections) as needed to achieve the desired quality. Review the number of calibrations and applicable calibration intervals, and relate the accuracy of calibration to an appropriate level as needed by the production process. This is where SP can provide unconditional assistance in finding the most efficient processes.