Estimating the uncertainty of measurement certainly sounds like an oxymoron. You need to measure the uncertainty of measurement. Still with me? ISO 17025:2017 accounts for the uncertainty of measurement under the sub-clause 5.4.6, where it is prescribed for labs to understand, document, and apply procedures to estimate the uncertainty of measurement reasonably.
The essential requirements related to the uncertainty of measurement are stated below:
- Labs must make use of appropriate steps to estimate the uncertainty of measurement.
- Methods used that are statistical are not always to be used to estimate the uncertainty of measurement.
- All the uncertainty components shall be identified and estimated to the closest match for each of the product testing.
- Approximating the estimation of uncertainty in measurement will rely on knowing the methods used to perform the measurements
- Validation data methodology can be used to determine the uncertainty of measurement.
The necessary part of measuring the uncertainty of measurement is to identify all sources of uncertainty of measurement. This consideration includes seeking every component, no matter how minor it is. If it’s contributing to the uncertainty of measurement and its estimation it needs to be considered.
Few Basic Concepts Relevant to Uncertainty of Measurement
Let’s brush up on some beginner’s guide terminology related to the uncertainty of measurement.
Measurements are no more than numbers, usually acquired through apparatus such as scales, thermometers, stopwatches, or rulers.
Measuring an object will give us valuable information about what it is. A measurement is made up of two components – the value (a number) and the units related to a larger, defined measurement standard.
It is the specific quantity subject to measurement.
Uncertainty of measurement
Uncertainty of measurement deals with the quality of measurement. As engineers will tell you, there’s no such thing as an exact value for a measurement, only the most probable value. Uncertainty of measurement is that doubt that happens with each measurement.
Error is the difference between the measured value and the actual value. Error and Uncertainty of measurement are not interchangeable, no matter how closely they may look like one another. Error deals with flaws in the measuring process, but uncertainty is the doubt about result measurement.
Sources of Errors and Uncertainties
- The measuring instruments
- The item being measured
- The Measurement Process
- Calibration method
- Operators’ skill and competencies
- Sampling Issues
- The Environment
General Kinds of Uncertainty in any Measurement
Random uncertainty of measurement occurs when you get a different unexpected result after repeating a measurement.
Systematic Uncertainty of Measurement occurs when a value consistently shows up as more or less than the actual value.
Standard Uncertainty may be defined as the uncertainty of a measurement’s outcome, which may be expressed as a Standard Deviation.
Combined Standard Uncertainty
Combined Standard Uncertainty may be described as the result of a measurement that is estimated standard deviation equivalent to the positive square root of total variance obtained by adding all the uncertainty components.
Calibration is a set of operational standards set up under specified conditions. The relation between quantities’ values indicated by the instrument used to measure or any values represented by reference material and corresponding values obtained by the standard.
The difference between the expected outcome of testing and the real accepted value obtained.
Validation is a process for proving the characteristics of the method’s capacity, which is appropriate for solving any analysis problem.
Reference material is a material or a substance used to calibrate a measuring apparatus or assessment of a measurement method or assign values to the material.
Essential Aspects to Estimate Uncertainty of Measurement
Following are the steps of the procedure for estimating the uncertainty of measurement:
The Measurand needs to be defined
Defining the measurand is essential in getting a proper estimate for uncertainty. The correlation between the measurand and input quantity should be established through the model equation.
Identification of Sources of the uncertainty of measurement
Some of the more common sources for uncertainty of measurement include:
- Unclear defining criteria of the measurement
- Sampling method
- The way the sample is being transported and stored
- Sample preparation method
- Internal Lab conditions and external environmental conditions under which the measurement has been carried out
- Lab operators
- Any change in the testing procedures
- Equipment that is used to get measurement done
- Reference material used
Stepwise Method to Estimate Uncertainty of Measurement According to ISO 17025
Following is the eight-step approach to estimate the uncertainty of measurement:
- Decide what the measurement will allow you to conclude. If possible, estimate the actual output for the measurement and determine what calculations may be necessary to give you the desired outcome.
- Conduct the measurements, nothing the results.
- Draw an estimate of each input quantity’s uncertainty that provides a feed to the final desired result.
- Determine what errors may have affected the measuring process and if those errors are independent of each other. If the errors are dependent, you may be required to perform more calculations for error-correction.
- Calculate the outcome of your measurement exercise.
- Evaluate the combined standard uncertainty from all the individual aspects.
- Denote uncertainty with uncertainty interval and state level of confidence.
- Mention measurement results along with uncertainty and state how the measurement is done and the uncertainty of measurement.
How to Minimize Uncertainty in Measurement?
Here are a few simple stepwise techniques that can be used to minimize uncertainty in measurement:
- Calibrating the Measuring Instrument: It is crucial to perform a calibration on measuring instruments before measuring anything with a reference standard or agent.
- Do Corrections: Make corrections that can affect your measurement practice. For example: if you find zero error in your measuring instrument, you should make your measurement outcome likewise after removing zero error.
- Sync your Measurements to National Standards: Make your measurement outcomes traceable and in sync with the national standards.
- Measuring Instrument: Choose the best of the best measuring instrument and make the maximum use of the lab’s calibration facilities.
- Repeat Measurement Process: Check the accuracy and precision of measurement by repeating them repeatedly. Multiple iterations may be able to rule out errors or spot systematic errors.
- Remove base reference numbers and assumed measurements. Rely on the physical measures and not those obtained from secondary sources.
- Be aware of the fact that during multiple calibration chains, the uncertainty increases after each step.