A custody chromatograph is trusted to give a clean, independent analysis every cycle, with each run reflecting only the sample it just took. Carryover breaks that assumption. When heavy components from one analysis are not fully cleared before the next cycle begins, they linger in the column and appear in the following chromatogram as peaks that do not belong to the fresh sample. These extra peaks are called ghost peaks, and they inflate the reported composition with material that was actually left over from before. Because composition drives heating value and therefore custody energy, carryover is not just a nuisance; it biases the measurement. This guide explains how carryover produces ghost peaks, the column, backflush, and cycle-time causes behind it, how it subtly skews heating value between streams, and the chromatogram diagnostics a monitoring system can surface to catch it before it corrupts the record.
GC carryover / ghosting in one line: GC carryover is the persistence of components from one analysis into the next cycle, usually heavy hydrocarbons that are not fully cleared from the column before a new sample is injected. Those residual components show up in the following chromatogram as ghost peaks, extra peaks that do not belong to the fresh sample and that inflate the reported composition. Because composition determines heating value and custody energy, carryover biases the measurement, and it typically stems from incomplete backflush, column condition, or too short a cycle time, all of which leave diagnostic signatures in the chromatogram that SCADA can help surface.
A chromatograph separates a gas sample by carrying it through a column where different components move at different speeds and emerge, or elute, at characteristic times, each producing a peak the analyzer integrates into a concentration. The heavier the component, the more slowly it moves and the longer it takes to clear the column. Carryover happens when some of those slow, heavy components have not finished eluting or been swept out before the analyzer starts the next cycle and injects a fresh sample. The leftover material from the previous run is still in the system, and when the new run is processed, that residue emerges and is counted, producing a peak that belongs to the old sample rather than the new one.
These leftover peaks are called ghost peaks precisely because they appear without a corresponding component in the current sample; they are phantoms from the previous analysis. A ghost peak can show up at the retention time of a heavy component that should not be present at that level in the fresh sample, or it can add to a real peak and make it larger than it should be. Either way the analyzer, which cannot tell a ghost from a genuine peak, integrates it and reports it as part of the current composition. The result is a composition that contains material the current sample did not actually have, or has too much of a component the sample had only a little of.
The reason this matters more for heavy components is both that they are the slowest to clear, making them the most likely to be carried over, and that they carry disproportionate energy. Even a small carried-over amount of a heavy hydrocarbon adds meaningfully to the heating value, because the heavies have high heat of combustion per mole. So carryover tends to inflate exactly the components that most influence the energy calculation, which is why a small memory effect in the heavy end of the chromatogram can move the reported energy by more than its size in the composition would suggest.
The most direct cause of carryover in a custody GC is incomplete backflush. Many gas chromatographs use a backflush arrangement in which, once the lighter components of interest have passed through, the flow through part of the column is reversed to sweep the heavy components back out rather than waiting for them to elute forward, which would take far too long and risk them lingering into the next run. If the backflush is mistimed, if it starts too late or runs too briefly, or if a valve is not switching cleanly, the heavies are not fully flushed and some remain to appear as ghosts in the next cycle. A backflush problem is one of the first things to suspect when ghost peaks appear in the heavy end.
Column condition is another contributor. As a column ages or is contaminated, its ability to cleanly separate and release components degrades, and heavy components can be retained more strongly or released more slowly than the method assumes, increasing the chance that some are carried over. Related to this is column bleed, where the column's stationary phase itself slowly breaks down and releases material that raises the baseline and can produce spurious features; while column bleed is distinct from sample carryover, both corrupt the chromatogram and both point to a column that is no longer performing as it should. A column that is dirty, degraded, or past its service life makes carryover more likely and harder to control.
Cycle time ties these together. The analysis cycle has to be long enough for everything from the current sample, including the backflush of the heavies, to complete before the next injection. If the cycle is set too short in an effort to get analyses more frequently, it can begin the next run before the previous one has fully cleared, guaranteeing carryover regardless of how good the column and backflush are. There is a genuine tension here between wanting frequent updates and needing each cycle to finish cleanly, and setting the cycle time too aggressively is a self-inflicted cause of ghost peaks. A well-configured GC has a cycle time chosen so that every component, especially the slow heavies, is fully accounted for before the next sample enters.
Carryover is insidious in custody service because it biases heating value quietly and, on a multi-stream analyzer, it can bias one stream by material from another. When a single GC analyses several streams in rotation, heavy components left over from a rich stream can carry into the analysis of a leaner stream that follows it, inflating the lean stream's apparent heavy content and its heating value. The lean stream is then credited with energy that actually belonged to the rich stream's residue, a systematic error that follows the sequence in which the streams are analysed. Because it is small per cycle and consistent, this kind of cross-stream carryover bias can persist for a long time without an obvious alarm, quietly distorting the energy split between streams.
The signatures of carryover live in the chromatogram itself, which is why keeping and examining the chromatogram matters rather than trusting only the final composition numbers. Telltale signs include peaks appearing where the fresh sample should have little or none of that component, a baseline that does not return cleanly to where it started, heavy-end peaks that seem larger than the stream should produce, and shifts in retention time or peak shape that indicate the separation is not behaving as the method expects. An analyst who compares the current chromatogram against a known-good reference chromatogram for that stream can often spot a ghost peak or a raised baseline directly, where the summarised composition alone would just look slightly off.
A cloud SCADA platform such as Merobix helps by surfacing these diagnostics rather than hiding them behind a single composition report. Trending each component and the heating value per stream over time makes a slow inflation of the heavy end or a creeping rise in energy visible, and on a multi-stream analyzer, a pattern where a lean stream's heating value tracks the richness of the stream analysed just before it is a strong fingerprint of carryover that a trend view exposes. Where the platform can carry chromatogram data or the diagnostic values a GC produces, such as baseline and retention-time information, it lets a measurement team catch a developing carryover problem while it is still small, prompting a check of the backflush timing, the cycle time, or the column condition before the bias grows large enough to force a prior-period correction. Catching carryover early through trending is far cheaper than discovering it as an unexplained energy imbalance months later.
A ghost peak is a peak that appears in a chromatogram without a corresponding component in the current sample, usually because residual material from a previous analysis was not cleared before the next cycle began. The analyzer cannot tell a ghost from a genuine peak, so it integrates it and reports it as part of the current composition, inflating the result. Ghost peaks most often come from heavy components, which are the slowest to clear the column.
The most common cause is incomplete backflush, where the reversal that sweeps heavy components out of the column is mistimed or too brief, leaving heavies to appear in the next cycle. Column condition also contributes, since an aged or contaminated column releases heavy components slowly or unevenly, and too short a cycle time can start the next analysis before the previous one has fully cleared. Any of these leaves residual material that shows up as ghost peaks.
Carryover most affects heavy hydrocarbons, which are both the slowest to clear the column and the highest in heat of combustion, so even a small carried-over amount adds disproportionately to the calculated heating value. On a multi-stream analyzer, heavies left over from a rich stream can inflate the heating value of a leaner stream analysed next, quietly shifting energy from one stream to another. Because the bias is small and consistent, it can persist unnoticed and distort custody energy over time.
Merobix reads your field devices into a cloud SCADA - the real thing behind these terms, live in days from any browser.