A custody gas chromatograph never measures a mole percent directly. It measures a peak area for each component and then multiplies that area by a number, learned during calibration, that converts area into concentration. That number is the response factor, and there is one for every component. Because the response factor is the hinge between raw detector output and the billed composition, a factor that quietly drifts between calibrations biases heating value and volume without ever throwing an obvious fault. This page focuses on that drift and on trending the factors to catch it early.
GC Response Factor in one line: A GC response factor is the per-component ratio that converts a detector peak area into a mole percent, set by running a certified calibration gas of known composition. Each component has its own factor because the detector responds differently to each one. If a factor drifts between calibrations, the resulting mole percent is biased, which silently skews heating value and the energy that appears on the billing statement.
Which page do you need? This page focuses on how response factors are calibrated and validated. For what a GC response factor is, see GC Response Factor in Custody Gas.
When a gas sample passes through the chromatograph, the column separates it and each component reaches the detector in turn, producing a peak whose area is proportional to how much of that component was present. The proportionality constant is not the same for every component, because the detector reacts more strongly to some molecules than to others. The response factor for a component is that constant: the multiplier that turns its measured peak area into a mole percent, established by running a gas of certified, known composition through the analyzer.
During calibration, the analyzer knows the true concentration of each component in the calibration blend, so it can divide the known concentration by the measured area to derive that component's factor. From then until the next calibration, the GC applies those stored factors to every unknown sample. Most custody analyzers work with relative response factors, expressing each component against a reference so the math is less sensitive to changes in overall detector gain, but the idea is identical: a saved multiplier per component, learned from a bottle of known gas.
The factor is only as good as the calibration gas it came from. If the certified blend is mislabeled, has drifted in the bottle, or is run when the analyzer is not at proper operating condition, the derived factors bake that error in and every subsequent sample inherits it. This is why the certificate and handling of the calibration gas matter as much as the analyzer hardware, and why a factor that looks reasonable can still be quietly wrong from the moment it was set.
The danger of a wrong response factor is that it produces a composition that looks entirely plausible. The mole percents still sum near one hundred, the chromatogram still shows clean peaks, and no alarm trips, yet the number for one component is off because its factor is off. Since heating value is computed from the composition, a biased component shifts the calculated BTU per unit volume, and every volume of gas that flows while that bias persists is converted to energy at the wrong rate.
The financial leverage of this error is large because it applies continuously. A custody meter run may move a great deal of gas between calibrations, and a small percentage bias in the heating value multiplies across all of that volume into a meaningful discrepancy in the energy billed. Unlike a hard analyzer failure that stops the measurement and forces a fallback, a drifted factor keeps reporting, so the error accumulates unnoticed on the statement until a proving, a validation, or a dispute exposes it.
Drift is a normal fact of analyzer life, not a rare fault. Detectors age, carrier flow shifts, column performance changes, and ambient conditions swing, and any of these can move the effective response of the detector away from the value captured at the last calibration. The response factors that were correct on calibration day slowly become slightly wrong, and the composition drifts with them. The question is never whether the factors drift but how quickly the drift is noticed relative to how much gas flows in the meantime.
A single response factor read once a month at calibration tells you almost nothing about what happened in between. The value comes from treating each factor as a time series and watching it move. When a validation run or a recalibration produces a fresh set of factors, comparing them against the previous set reveals which components are drifting and how fast, turning an invisible bias into a visible slope. A factor that changes only a hair between calibrations is healthy; one that jumps or trends steadily downward is a component whose measurement, and therefore its share of the heating value, is decaying.
A cloud SCADA host is the natural place to keep that history. By logging the response factors and the calibration and validation results as they occur, the host builds a record that spans many calibration cycles and many analyzers across a fleet. Trending one component's factor over time shows the operator whether a shift is a one-off from a bad calibration gas or a sustained drift from an aging detector, and comparing the same component across similar analyzers flags the one unit that is behaving differently from its peers.
The payoff is catching the drift before it reaches the billing statement rather than after. When the host trends response factors and alarms on a factor that has moved beyond an expected band, the operator can schedule a calibration or investigate the analyzer while the bias is still small, instead of discovering it during a contentious month-end reconciliation. The response factor stops being a number buried in the analyzer and becomes a monitored health signal for the entire custody measurement.
Because the detector does not respond equally to every component. The same amount of one component can produce a larger or smaller peak than the same amount of another, so a single conversion factor would misquantify most components. Giving each component its own factor, derived from a certified calibration gas, lets the analyzer convert every peak area into an accurate mole percent for that specific component.
It biases the calculated composition, which biases the heating value derived from that composition. Because energy billed is heating value times volume, a factor that is slightly off applies that error to every unit of gas that flows until the next calibration corrects it. The result is a continuous, unalarmed discrepancy in the energy on the statement that can grow large over a billing period.
By trending the factors over time rather than reading them once at calibration. A SCADA host that logs each set of factors from calibrations and validation runs can compare successive sets, reveal which components are drifting and how fast, and alarm when a factor moves beyond its expected band. That lets an operator act while the bias is still small instead of finding it during a month-end dispute.
Merobix reads your field devices into a cloud SCADA - the real thing behind these terms, live in days from any browser.