Why analyser validation is becoming a bigger process problem

For years, process-stream analysers have been sold on a simple promise: faster data, faster decisions, and less dependence on waiting for laboratory results. In petrochemicals, that promise still holds.

But as feedstocks and fuels become more variable, the harder question is no longer whether online measurement is useful. It is whether the online analyser and the laboratory method are still measuring the same thing closely enough for operational and commercial decisions to be trusted. 

ASTM’s current standards framework shows how seriously the sector now treats that issue. ASTM D6299, now listed by ASTM in an active 2025 version, applies not only to laboratory test methods but also to validated process-stream analysers and to monitoring the differences between two analytical systems that purport to measure the same property. ASTM D6708, now listed in an active 2024 version, provides the statistical framework for assessing and improving expected agreement between two test methods measuring the same petroleum property.

Find your next process monitoring instrument in our international directory.

That matters more now because “agreement” becomes harder when materials become less predictable. In a conventional refinery or fuels environment, a process analyser and a laboratory method may disagree occasionally because of drift, maintenance issues, sampling lag or calibration problems. 

In circular and alternative-fuel settings, disagreement can also arise because the material itself is more compositionally diverse, less stable, or less well matched to the assumptions built into existing methods. ASTM D6708 itself warns users not to apply its calculated between-method reproducibility casually to materials significantly different in composition from those actually studied, and notes that where sample-specific biases are present the range and type of samples may need to be expanded substantially. 

That is an important point for anyone dealing with recycled streams, unconventional feedstocks or newer fuels: a tidy validation exercise carried out on a narrow set of materials may say less than it appears to say.

This is where ASTM D3764 comes into sharper focus. The current D3764 practice for validation of the performance of process-stream analyser systems is explicitly based on the statistical principles of D6708 and is designed to test whether the agreement between the analyser system and an independent test method meets user-specified requirements. 

It treats validation as a performance question rather than a marketing claim. The standard also makes clear that validation only applies to the type and property range of the materials actually used in the study. 

That limitation is easy to overlook, but it is precisely where many modern problems begin. A refinery-grade analyser correlation that performs well on a stable hydrocarbon slate may not remain robust when the plant starts co-processing waste-plastic-derived oils, variable bio-based intermediates, or less familiar alternative-fuel streams.

ASTM D6299 then picks up the story after initial validation. Its role is not simply to say whether an analyser once matched the lab, but whether the whole analytical measurement system remains in statistical control over time. ASTM says the practice can be used to monitor precision and bias continuously, and that users need to consider factors such as frequency of use, parameter criticality, system stability, business economics, and regulatory or contractual requirements. 

That last point is especially relevant for product release, custody transfer and specification compliance. In those settings, disagreement between the line and the lab is not just a technical inconvenience; it can affect shipment decisions, off-spec calls, giveaway, and customer disputes. As newer fuels and feedstocks enter the system, the value of disciplined control-charting and statistical surveillance rises because the cost of trusting a drifting analyser also rises.

The broader implication is that analyser validation is becoming a more strategic task than many plants have treated it in the past. The traditional approach was often linear: install analyser, build correlation, compare with the lab, then move on. The standards landscape suggests a more demanding model. 

D3764 points towards robust performance validation against an independent method; D6708 addresses whether two methods really agree and what bias correction may be justified; and D6299 addresses whether that agreement remains under control in routine operation. Together, they imply that analyser confidence has to be earned repeatedly, not assumed permanently. That is particularly true where the stream itself may be evolving faster than the validation plan.

For petrochemical and alternative-fuels readers, the message is straightforward. As the industry moves towards more variable raw materials and more diverse fuel slates, the online analyser becomes more important, not less. But its value depends increasingly on proving what it agrees with, over what range, under what conditions, and for how long. 

The real competitive edge is not merely having more analysers on the line. It is having a validation and quality-assurance regime strong enough to know when the analyser is telling the truth, when it is drifting, and when the material has changed enough that yesterday’s agreement no longer means very much. ASTM’s current standards do not make that problem disappear, but they do make one thing clear: in a less predictable process world, analyser validation is no longer background statistical housekeeping. It is becoming part of the core discipline of process control.