Sample carryover, also called memory effect, is the contamination of a sample by residual product left behind in the sampling system from the sample before it. Probes, sample lines, and grab chambers all retain a small volume of the last product they held, and that residue bleeds into the next sample and shifts its composition. The effect is a memory of the previous condition intruding on the current one. This page explains how carryover happens, when it does the most damage, and how flushing and dead volume control keep it in check.
Sample Carryover in one line: Sample carryover, or memory effect, is when residual product held in a sampler's probe, lines, and chambers contaminates the following sample, so the new grab carries a trace of the previous batch or condition. It matters most at batch changes and interface cuts, and it is controlled by flushing the sample path before each grab and by minimizing the dead volume that retains residue.
A sampling system is a set of wetted volumes: the probe bore, the sample line, valves, filters, and the grab chamber that meters each dose. After a sample is taken, none of these empty completely. A film clings to the walls, and pockets and unswept branches hold small standing volumes. That retained product is a physical memory of what the system last handled, and unless it is displaced, it becomes part of whatever is sampled next. When the next grab is drawn, it picks up that residue along with fresh product, and the analysis reports a blend rather than the true current stream.
The size of the effect depends on how much residue there is relative to the sample and how different the residue is from the current product. A tight, well swept sample path with little dead volume retains little, so the carryover is small and easily flushed away. A path riddled with dead legs, oversized fittings, and stagnant pockets retains a lot and holds it stubbornly, so the memory persists across several samples. And when the product has genuinely changed, a light grade following a heavy one, or clean product following one with high water, the residue is chemically unlike the new stream, so even a small retained volume can visibly shift composition, water content, or gravity.
Carryover is insidious because it does not look like a failure. The sampler runs its schedule, the grab is taken, the receiver fills, and nothing alarms. The only evidence is in the analysis, where a grade reads slightly wrong, a water content is pulled toward the previous batch, or a composite carries a signature that does not match the metered product. Because the contamination is small each time but systematic, it can bias results consistently in one direction, which is exactly the kind of quiet error that survives until someone reconciles the sample against an independent check.
Carryover is nearly harmless on a long, steady, single product run, because the residue is the same product as the current stream, so mixing it in changes nothing meaningful. The problem sharpens precisely when the product in the line changes. At a batch change, the sample path is full of the old batch when the new one arrives, and without a thorough flush the first grabs of the new batch carry old product into the receiver, biasing the new batch's composite toward the batch that came before it.
Interface cuts are the most demanding case. On a pipeline carrying successive products, the interface is the transition zone between two grades, and sampling around it is used to decide where one product ends and the next begins. Memory effect blurs exactly this decision: residue from the trailing grade bleeds into grabs taken in the leading grade and vice versa, smearing a boundary that operations needs to see sharply. A sampler with significant carryover can make an interface look wider and less defined than it really is, which affects how the interface is cut and how much product is downgraded.
Water and sediment content is another place carryover bites, because water tends to hang up in low points and dead legs and then bleed into later grabs long after the line water has passed. A slug of water retained in a pocket can pull the reported water content of subsequent samples upward, misrepresenting the stream. In all these cases the common thread is a change in the line that the sampling system is too slow to forget, so the discipline is to force the system to forget, by flushing, before every grab that could be affected by what came before.
Two design and operating choices control carryover. The first is minimizing dead volume: designing the sample path short and tight, eliminating dead legs, oversized fittings, and stagnant pockets, and putting the probe and grab chamber where flow keeps them swept. Less retained volume means less to carry over and less effort to clear. The second is flushing: running current line fluid through the path before each grab so the residue is displaced to waste rather than into the receiver. Together they attack carryover from both sides, reducing how much residue exists and then removing what remains before it can contaminate the next sample.
When results suggest carryover, the symptom pattern points to the cause. Grabs that consistently read like the previous batch after a change point to an inadequate pre grab flush or a dead pocket the flush does not reach. Water contents that stay elevated after the line water has passed point to a low point holding a water slug. A composite that carries a faint signature of the prior product points to residue accumulating grab over grab because the flush is too short. Diagnosis walks the path: confirm the flush actually sweeps the whole wetted volume including the dead legs, lengthen it after product changes, and hunt for and remove the pockets that retain residue.
Flushing is the main control, and flushing is observable, which ties carryover management to continuous monitoring. A cloud SCADA platform such as Merobix reads the sampler's valve states, flush flow, and the count of grabs and flushes from the field controller, so operators can confirm that a proper flush preceded each grab, especially the critical grabs right after a batch change or around an interface. A flush that quietly stopped working, a stuck valve or a lost flush supply, is the precondition for carryover, and it shows up in those states before it shows up as a contaminated result. Watching flush integrity across a fleet of samplers turns memory effect from a defect found at reconciliation into a condition prevented in real time.
Memory effect is another name for sample carryover: residual product retained in the probe, sample lines, and grab chamber from the previous sample intrudes into the next one, so the new grab carries a trace of the earlier product. The system effectively remembers what it last held and passes that memory into the following sample, shifting its composition, water content, or gravity.
It matters most when the product in the line changes, at batch changes and at interface cuts between two grades, and when water or sediment content swings. On a long, steady, single-product run the residue is the same as the current product, so it does little harm. At a change, the residue is different from the new stream, so even a small retained volume can visibly bias the sample toward the previous condition.
Control it two ways: minimize dead volume by designing the sample path short and tight with no dead legs or stagnant pockets, and flush the path with current line fluid before each grab so residue is displaced to waste rather than into the receiver. After a product change the flush should be lengthened, because the whole path is full of the previous batch and needs several turnovers to clear before a clean grab can be taken.
Merobix reads your field devices into a cloud SCADA - the real thing behind these terms, live in days from any browser.