English (English)

How to Reduce Your Risk by Calculating the Measurement Uncertainty

10 October 2020 /

It is not possible to make a true, perfect measurement. Many factors combine to skew a measurement, and a great deal of effort is necessary to identify all these contributing factors and how they affect the measurement, as well as to normalise them and quantify the magnitude of their effect.

It is understandable, then, why many engineering teams responsible for calibrating heat treatment equipment overlook the importance both of estimating the measurement uncertainty and how it should translate into binary pass/fail compliance statements. In such a high-risk environment, and with so many variables, it is only natural that their inclination steers them towards using just the simple, apparently precise measurement, unburdened by probabilities.

Reversing this line of thinking by acknowledging and estimating the uncertainties at play presents a singular opportunity to improve compliance and reduce risk. Let’s look at how.

The risk of falsely accepting...

On the one hand, a measurement that has been taken without regard for uncertainties has an unquantifiable risk of being falsely accepted as complying with tolerance requirements.

The machine is calibrated, the results apparently within the acceptance limits according to the standards in question, so is approved for operation.The machine is said to comply when there is clearly a probability, perhaps even a large one, that it in fact does not.

Aside from the glaring compliance issues this raises, there are also several commercial, safety and reputational impacts to consider.

    • Any product that has been developed or tested using the non-compliant equipment may itself be out-of-spec, faulty, or potentially harmful to its end-users. The possibility of this will lead to costly product recalls and manufacturing setbacks.
    • The teams operating the non-compliant equipment may be exposed to unknown/unanticipated health and safety risks.
    • In the event of either of the above, the reputational damage to the company could be severe and long-lasting.

...vs the cost of falsely rejecting

Of course, the opposite can also be true. Without the contributing factors taken into consideration, a measurement may drive the conclusion that the equipment is not calibrated within acceptance limits. Without the full picture to inform their decision-making, more risk-averse teams may decide the equipment needs to be repaired / adjusted, then recalibrated, with repercussions for uptime and production.

    • In the case of manufacturing, production may need to be halted while the machines are adjusted or fixed, then recalibrated.
    • Generally speaking, uptime will be impacted as machines are shut down or taken offline to receive the “required” maintenance.
    • Manufactured parts may be unnecessarily recalled or ‘concessioned’ based on the machine having been falsely “failed” as non-compliant.

In either case, the measurement is not reliable without the uncertainties being taken into consideration, with the potential to significantly disrupt day-to-day operations.

Related read: How to Correctly Perform a System Accuracy Test and Stay  Compliant

So how can you comply if no measurement is “perfect”?

If results cannot be measured exactly, then how can they “definitely” comply within a specified tolerance?  Well they can’t, but we can calculate the probability of compliance and match this to customers’ risk appetites.

Perhaps the result is within tolerance, accounting for the uncertainty shows a high probability that the ‘true’ result lies outside of the limits. This is where there is a cost and a benefit of reducing uncertainties. More of the tolerance band is available for the measurement and not taken up by the uncertainty, as it were.

“The smaller the uncertainty, the closer the measurement is likely to be to the ‘true’ measurement. Informed decisions around compliance can then be made.”

The importance of accounting for measurement uncertainty in making compliance statements has increased considerably since the introduction of ISO17025:2017. Many quality managers and the engineer teams supporting them may be unfamiliar with its importance and the processes involved. We hope this article helps explains that importance and its impacts, should that be the case. 

If hands-on support is desired, either with training in-house teams to estimate the uncertainty of measurement, assess the risks in compliance statements or with actually delivering process equipment calibrations with all this included, VFE can also support in these areas.


An impartial, reliable approach to calculating the measurement uncertainty

For upwards of 30 years, VFE has delivered Calibration Services to a range of industries reliant on correctly calibrated process manufacturing equipment for compliance and production.

All our calibration is fully traceable to national standards and with our UKAS accredited laboratory and on-site calibration services (laboratory no. 4338) we can meet almost all industry requirements.

In relation to the uncertainty of measurements, if a customer requires a compliance statement (pass/fail), we will agree (in advance of any measurements being taken) a “decision rule”, which aims to match this risks of false accept/reject with their risk appetite. Importantly, our UKAS accredited laboratory helps our customers to understand the risk involved in accepting/rejecting results, so they can make informed agreements.  

At the same time, it is absolutely essential that calibration services are completely impartial. You can trust that VFE calibration engineers will never produce results that are favourable to a particular aim such as product acceptance or push you to agree a calibration regimen that isn’t appropriate to your needs. 

The decision rule is best discussed by laboratory technical management. 

Building on-site capabilities through bespoke measurement uncertainty training

Some businesses may prefer to develop their on-site capabilities over the use of specialist services. The right training can set up existing engineering teams to better perform calibrations, including estimating the uncertainty of measurement involved. 

However, undertaking these measurements without training can open up a site to unknown, potentially unacceptable, levels of risk.

“It is one thing to invest in a calibration kit and follow the instructions. It is another thing to understand why the process looks like it does, how it might go wrong and what the results mean.”

VFE’s training packages are aimed at in-house engineering teams responsible for maintaining heat treatment equipment such as furnaces, pumps and autoclaves. Courses focus on how to improve machine utilisation and maximise productivity, with bespoke training packages also available. Complementing VFE’s ServiceCare plans, these courses will help customers get the most from their machines between routine service visits.


Uncertainties you can make decisions against

Today, compliance is no longer a simple question of whether a machine has passed or not. Every measurement has uncertainties associated with it. As more and more businesses become reliant on heat treatment equipment work to comply with external standards, they will find themselves subject to questions about those uncertainties of measurement and how they have assessed their compliance risks. 

For an industry underpinned by standards, the notion that uncertainties should be sought after might seem strange, even counterproductive. Instead, it is precisely that measurement uncertainty that enables informed decision-making around risk and compliance, offering quantitative insights into how a machine is performing.

Can you confidently say you know the full range of factors skewing your measurements and how they are impacting your operations?

New call-to-action