How to Validate a Gas Chromatograph Calibration
Validating a GC calibration is the routine that proves the analyzer still reports the truth without changing anything, and it is distinct from recalibrating, which adjusts the response factors. This procedure is for the analyzer technician or measurement engineer running a periodic validation on a process gas chromatograph. It walks running a certified check gas, comparing each reported component against its tolerance, using the unnormalized total as a consistency check, and deciding pass, fail, or recalibrate. The whole point of a validation is to catch a GC that has drifted before its composition feeds a wrong custody or quality result, while leaving a good calibration alone.
Validate a GC Calibration in one line: To validate a gas chromatograph calibration, run a certified check gas of known composition through the stream without adjusting anything and compare each reported component against the check gas value and its tolerance. Confirm the unnormalized total sits near one hundred percent as a consistency check, and read the pattern of any deviations. If the components fall within tolerance the calibration is validated as-is; if they fall outside, the GC has drifted and needs recalibration, and the check gas run is your documented evidence either way.
Choose the Check Gas and Define Tolerances
A validation is only as good as the gas you challenge with and the tolerances you judge against, so establish both first. Use a certified check gas of known composition, ideally independent of the gas used to calibrate the GC so the validation is a genuine second opinion rather than a re-run of the calibration point. Confirm the check gas is in date and its composition spans the range the stream operates over. The distinction between a gas used only to validate and one used to adjust is covered in the note on the validation cylinder versus calibration cylinder.
Define the acceptance tolerance for each component before you run, because a validation without a pass criterion is just a number. The tolerances come from the site measurement procedure and the accuracy the downstream use requires, and they are tighter for components that dominate a custody or heating-value calculation. Know which components matter most for this stream, so a small deviation on a major component is weighed differently from the same deviation on a trace one. The validation itself is the same discipline as the general analyzer validation check, applied to a multi-component analysis.
Run the Check Gas Without Adjusting Anything
The defining rule of validation is that you do not touch the calibration. Run the certified check gas through the stream exactly as a normal sample and record the composition the GC reports, unchanged. The point is to see what the GC says about a known gas in its current state, because that as-found comparison is the evidence of whether it has drifted. Adjusting the response factors during a validation destroys the very thing the validation exists to measure and turns it into an undocumented recalibration.
Let the analysis run its full cycle and capture the reported value for each component. Confirm the check gas actually reached the analyzer cleanly and displaced the previous sample, because a validation run contaminated by residual process gas or a poor stream switch tells you about the plumbing, not the calibration. Where the GC carries composition over from run to run, be alert to memory effects that could bias the validation, a concern described in the note on sample carryover and memory effect.
Compare Each Component and Check the Total
Compare the GC's reported value for each component against the certified check gas value and its tolerance. A component within tolerance validates for that component; a component outside tolerance is a drift finding. Read the pattern, not just individual passes: a single component off may be a response-factor drift for that peak, while several components off in a consistent direction points at a systemic issue such as carrier flow, sample pressure, or a calibration that has shifted across the board. This response-factor stability is the subject of the note on GC response factor validation.
Use the unnormalized total as an independent consistency check alongside the per-component comparison. If the unnormalized total on the check gas has moved away from one hundred percent, the GC is over- or under-measuring overall even if normalization makes the reported composition look plausible, as the note on a GC unnormalized total check explains. A total that has drifted, combined with components out of tolerance, is a clear signal that recalibration is due rather than a one-off fluke.
Decide the Disposition and Record the Evidence
Decide the outcome on the evidence. If every relevant component falls within tolerance and the unnormalized total is healthy, the calibration is validated and left alone, which is the correct and often the desired result. If components fall outside tolerance or the total has drifted, the GC needs recalibration, and only then do you adjust the response factors, followed by another validation run to confirm the fix. Never recalibrate a GC that validated cleanly, because trimming a correct analyzer injects error and destroys the record that told you it was trustworthy.
Record the check gas used, the reported versus certified value for each component, the unnormalized total, and the pass or fail decision, because that record is the documented proof of the analyzer's state and the trend that reveals slow drift. When the validation results, the component deviations, and the unnormalized total are trended in a monitoring platform such as Merobix, a GC creeping toward its tolerance limits over successive validations is visible before it fails one, so the recalibration is planned rather than triggered by a failed custody check. The validated composition is the evidence the reported gas quality can be defended.
Frequently Asked Questions
What is the difference between validating and calibrating a GC?
Validating runs a certified check gas and compares the reported composition against tolerance without changing anything, proving whether the GC still reports the truth. Calibrating adjusts the response factors so the GC matches a known standard. A validation answers whether the calibration is still good; a calibration corrects it when it is not. You only recalibrate after a validation fails, and never recalibrate a GC that validated cleanly, because that injects error into a correct analyzer.
Should the GC check gas be different from the calibration gas?
Ideally yes. Using a certified check gas independent of the gas the GC was calibrated on makes the validation a genuine second opinion rather than a re-run of the calibration point, so it can actually catch a calibration that drifted. Both gases must be certified and in date, and the check gas composition should span the operating range. If only one certified gas is available, the validation still has value but is a weaker independent test of the calibration.
How do I read the pattern of deviations in a GC validation?
Look at whether one component or several are off. A single component outside tolerance often means a response-factor drift for that peak. Several components off in a consistent direction points at a systemic cause such as carrier flow, sample pressure, or a calibration that shifted across the board. Cross-check with the unnormalized total: if it has moved away from one hundred percent along with out-of-tolerance components, recalibration is clearly due rather than a one-off fluke.
Automation services
Need help turning this into a working system?
Merobix integrates SCADA, programs Allen-Bradley and Siemens PLCs, and designs and fabricates industrial control panels.
Meeting requests are reviewed before confirmation.