An analyzer is only useful if you can trust its readings, and a validation check is how that trust is confirmed on a routine basis. In a validation, the analyzer measures a known reference and its result is compared against the certified value, proving it is still reading true, without changing its calibration. This guide explains what a validation check is, how it differs from a recalibration, why custody and emissions analyzers schedule it regularly, and how pass or fail results and drift trends are logged and alarmed in the control system.
Analyzer Validation Check in one line: An analyzer validation check is a routine test in which an analyzer measures a known reference gas or standard and its result is compared to the certified value to confirm the analyzer is still accurate. Crucially, a validation only verifies accuracy; it does not adjust or recalibrate the analyzer. Custody-transfer and emissions analyzers run validations on a schedule so that any drift is caught and documented, and the pass, fail, and drift results are logged and alarmed in the control system.
A validation check introduces a known reference to the analyzer, most often a certified gas of known composition, and lets the analyzer measure it as if it were sample. Because the true value is known, the analyzer's result can be compared directly against it. If the reading matches the certified value within the allowed tolerance, the analyzer passes and is confirmed to be measuring accurately; if it is outside tolerance, it fails, signalling that something has drifted or gone wrong.
The defining feature of a validation is that it does not change the analyzer. Unlike a calibration, which adjusts the analyzer's response so it reads the reference correctly, a validation leaves the analyzer's settings untouched and simply records how it performed. This is important because it means a validation is an honest, independent check: it reports the analyzer's actual accuracy at that moment rather than forcing it to agree. An analyzer that passes validation has demonstrated it is still trustworthy on its existing calibration.
The result of a validation is therefore evidence, not an adjustment. A passing check is a documented statement that the analyzer was accurate at that time; a failing check is a flag that it was not and that action, such as investigation or recalibration, is needed. Keeping validation and calibration distinct matters because it preserves the meaning of the check: a validation you also calibrated during would no longer tell you whether the analyzer had drifted, only that you had corrected it.
Some analyzers carry consequences heavy enough that their accuracy must be demonstrated continually, not assumed. Custody-transfer analyzers help determine the quantity or quality of product that changes hands and is paid for, so an undetected drift translates directly into money measured wrong. Continuous emissions monitoring analyzers produce the numbers reported to regulators, so their accuracy underpins compliance. For both, being able to show the analyzer was reading true, day after day, is part of the job.
Scheduled validation is how that demonstration is made. Running a known reference at regular intervals and recording the result builds an ongoing record of the analyzer's accuracy over time, so at any point there is recent evidence it was in tolerance. Just as importantly, a validation catches drift early: an analyzer that has slowly wandered will start to show larger deviations from the reference before it is badly out, giving a chance to act before bad data accumulates. This is why these applications treat validation as a standing routine rather than an occasional exercise.
Many of these validations are automated, sometimes called autovalidation, where the analyzer periodically switches itself to the reference, measures it, records the result, and returns to sampling, all without an operator present. Automation makes frequent checks practical, keeps the schedule from slipping, and produces a consistent, timestamped record. It is what allows a remote or unmanned custody or emissions analyzer to keep proving its own accuracy between site visits.
A validation only fulfils its purpose if its outcome is captured and visible, and that is where the control system comes in. Each validation produces results, the measured value against the reference, the deviation, and a pass or fail verdict, and those are reported from the analyzer to the control system so they enter the record rather than living only in the analyzer's local memory. A failed validation is treated as an alarm, so operators are told the moment an analyzer can no longer be trusted.
Beyond the individual pass or fail, the sequence of validations over time tells a story. Plotting how far each validation deviated from the reference reveals drift as a trend: a deviation that grows check after check shows the analyzer slowly wandering, often before any single check has failed. Watching that trend lets a technician plan a recalibration or service proactively, rather than waiting for an outright failure that may have already tainted some data.
A cloud SCADA platform such as Merobix reads and historizes each validation result and the analyzer's status, alarms on a failed check so a drifting or faulty analyzer is flagged immediately, and preserves the drift trend for review. That record is exactly what custody and emissions applications need to prove their analyzers stayed accurate, and it lets the team see a slow drift developing across many checks. On remote and unmanned sites, where nobody witnesses the autovalidation happen, having every result captured, alarmed, and reviewable from any browser is what turns routine validation into dependable, defensible evidence of analyzer accuracy.
A validation measures a known reference and compares the analyzer's result to it to verify accuracy, but leaves the analyzer's settings unchanged. A calibration goes further and adjusts the analyzer's response so it reads the reference correctly. Keeping them separate matters because a validation is an independent check of whether the analyzer has drifted, while a calibration corrects it; validating without adjusting is what makes the check honest.
Because their readings carry heavy consequences: custody analyzers help determine product that is bought and sold, and emissions analyzers produce numbers reported to regulators. Routine validation builds an ongoing record that the analyzer was accurate and catches drift early, before the analyzer is badly out. Many run automated validations, or autovalidation, so the checks happen on schedule and are documented even on unmanned sites.
A failed validation means the analyzer read the known reference outside the allowed tolerance, so it can no longer be trusted, and it is treated as an alarm so operators are told right away. The failure signals that something has drifted or gone wrong and that action is needed, such as investigation or recalibration. Because the validation itself does not adjust anything, the failure is preserved as evidence and the surrounding drift trend can be reviewed to understand it.
Merobix reads your field devices into a cloud SCADA - the real thing behind these terms, live in days from any browser.