An online custody chromatograph runs continuously and is calibrated and validated against reference gas, but there is one more, independent way to prove it is telling the truth: send a physical sample of the same gas to a separate laboratory and compare the answers. Pulling a spot or composite sample and having a certified lab analyse its composition and heating value gives a result derived from entirely different equipment and different people, and comparing it against what the online GC reported for the same gas quantifies any bias in the analyzer. This routine cross-check is a mainstay of good measurement practice because it catches errors that self-validation cannot. This guide explains how the spot-sample-versus-online comparison is done, what constitutes acceptable agreement, why the sampling technique itself is critical, and how a monitoring system logs the comparison as ongoing proof of the analyzer.
Spot sample vs analyzer in one line: Reconciling a spot sample against the online GC means pulling a physical sample of the custody gas, having a certified laboratory analyse its composition and heating value, and comparing that independent result against what the online chromatograph reported for the same gas. The comparison quantifies any bias in the online analyzer using entirely separate equipment, catching errors that calibrating the GC against its own reference gas cannot reveal. The two results should agree within an accepted tolerance, and measurement teams log each comparison over time as ongoing, documented proof that the online analyzer remains accurate.
An online GC is proven day to day against reference cylinders, but every part of that proof lives inside the same measurement chain: the same analyzer, the same sample system, the same calibration approach. A physical sample sent to an independent laboratory breaks out of that chain entirely. The lab uses its own chromatograph, its own reference standards, its own procedures and staff, and it analyses the actual gas from the line rather than a cylinder. When the lab's composition and heating value are compared against what the online GC reported for the same gas at the same time, any disagreement is evidence of a bias somewhere, and because the two determinations are independent, the comparison can reveal errors that the online analyzer would never flag about itself.
This is valuable precisely because some errors are invisible from inside the analyzer's own world. A systematic bias in the online GC's sample handling, a subtle problem in how it resolves a particular component, or even a flaw in its calibration cylinder can leave the analyzer passing all of its internal checks while still reading the process gas wrong. The internal validation confirms the analyzer agrees with its own references; the lab comparison confirms the analyzer agrees with reality as an independent party measures it. The two together are far stronger than either alone, which is why custody practice uses both rather than treating internal validation as sufficient.
The comparison is usually made on both composition and the derived heating value, because the heating value is what settlement ultimately rests on. Comparing component by component shows where any discrepancy lives, for instance in the heavy end or in the inerts, which helps diagnose the cause, while comparing the heating value shows the net effect on the quantity that matters commercially. A comparison that agrees on both gives strong confidence in the online analyzer, while one that disagrees points the investigation toward the component or the part of the system responsible.
The comparison is only as good as the sample, and this is where it most often goes wrong, because pulling a truly representative sample of pipeline gas is harder than it looks. A spot sample captures the gas at one instant, while a composite sample is collected incrementally over a period to represent the average gas, and each has its place; a composite is more representative of a stream that varies, while a spot is a snapshot. Either way the sample must be taken by a sound method that avoids contaminating or fractionating the gas, because a sampling error will make the lab result wrong and create a false disagreement with the online GC. If the sample cylinder is not properly purged, if the gas is allowed to drop heavy ends into liquid, or if the sample is taken from an unrepresentative point, the lab analyses gas that never matched what flowed past the online analyzer.
This is why the industry treats sampling technique with the same seriousness as the analysis, following recognised practices for how gas samples are collected and handled so the sample reaching the lab genuinely represents the line gas. The heavy hydrocarbons are the most vulnerable, because they are the first to condense if the sample cools or the pressure is mishandled, and losing a little of the heavy end in the sampling skews both the composition and the heating value the lab reports. A reconciliation that shows the lab reading lower in the heavies than the online GC should prompt a hard look at whether the sample, not the analyzer, dropped those components. Ruling out sampling error is the first step in interpreting any disagreement.
When the sample is sound, the two results should agree within a tolerance the measurement team sets, typically expressed as a permitted difference in heating value and in the key components. The agreement will never be perfect, because two independent analyses always differ a little, so the tolerance defines how close is close enough to consider the online analyzer proven. A comparison inside the tolerance is a pass that adds to the analyzer's proof record; one outside it triggers an investigation that examines the sampling, the online GC, and the lab in turn to find where the bias lies before assuming the online analyzer is at fault. Setting a sensible tolerance up front is what keeps the process from either ignoring real bias or chasing noise.
A single lab comparison confirms the analyzer on one day, but the practice earns its value when the comparisons are logged and trended as a continuing record. Each reconciliation produces a difference between the lab and the online GC in heating value and in composition, and that difference is a measure of the online analyzer's bias at that time. Kept over months and years, the sequence of comparisons becomes a documented history showing that the online analyzer has repeatedly been checked against independent lab results and stayed within tolerance, which is exactly the kind of evidence a custody audit or a counterparty asks for. It turns an occasional check into an ongoing proof.
A cloud SCADA platform such as Merobix supports this by holding each comparison as a data point alongside the analyzer's continuous operation, so the lab result, the online GC result for the same gas, and their difference are stored together and can be trended. Rather than lab reconciliations living in scattered spreadsheets or paper reports, the platform keeps them next to the composition and heating value the analyzer was producing, so a slow drift in the lab-versus-GC difference is visible as a trend and can be caught before it grows into a dispute. Trending the difference also distinguishes a stable, well-proven analyzer from one whose agreement with the lab is quietly wandering, which is information the raw pass or fail of a single comparison does not convey.
Logging the comparisons this way also strengthens the whole measurement story for a custody point. When the analyzer's day-to-day internal validation is combined with a documented trend of independent lab comparisons, the operator can demonstrate both that the analyzer agrees with its own references and that it agrees with reality as an outside lab measures it, which is a far more complete assurance than either alone. Setting an alert on a lab comparison difference that exceeds the tolerance, and keeping the running history accessible, means the reconciliation is not a formality filed away and forgotten but an active, monitored guarantee that the online analyzer, and therefore the custody energy it feeds, can be trusted.
Calibrating and validating the online GC only proves it agrees with its own reference cylinders inside the same measurement chain. A physical sample sent to an independent lab is analysed on entirely separate equipment by separate staff, so comparing that result against the online GC checks the analyzer against reality rather than against itself. This catches biases the internal checks cannot see, such as a subtle sample-handling error or even a flawed calibration cylinder.
A spot sample captures the gas at a single instant, giving a snapshot of the composition at that moment, while a composite sample is collected incrementally over a period so it represents the average gas over that time. A composite is more representative of a stream whose composition varies, whereas a spot is simpler and suits a steady stream. Both must be taken by a sound method to genuinely represent the line gas.
First rule out the sample, because a poorly taken sample that loses heavy ends or is unrepresentative will make the lab wrong and create a false disagreement, so the sampling technique is examined before blaming the analyzer. If the sample is sound and the difference still exceeds the agreed tolerance, the investigation looks at the online GC and the lab in turn to locate the bias, often using the component-by-component comparison to see where the discrepancy lives. A comparison inside tolerance simply adds to the analyzer's proof record.
Merobix reads your field devices into a cloud SCADA - the real thing behind these terms, live in days from any browser.