Before a custody gas chromatograph scales its results to sum to exactly one hundred, it has a raw sum, and that raw sum carries a signal the final composition throws away. A healthy analyzer measures each component well enough that the unadjusted total lands very close to one hundred percent on its own. When that raw sum starts drifting away from one hundred, it is often the earliest sign that something physical has gone wrong, and it is the one number normalization is about to erase. This page is about using that number as an alarm.
Unnormalized Total in one line: The unnormalized total is the sum of a GC's component mole percents as measured, before the analyzer scales them to add up to exactly one hundred percent. In a healthy custody GC this raw sum sits very close to one hundred, so a drift away from it is one of the earliest warnings of a fault such as carrier gas loss, a valve leak, or column degradation. Alarming on the pre-normalization sum catches problems that normalization would otherwise hide.
A chromatograph measures each component independently, applying that component's response factor to its peak area to get a mole percent, and then it adds those mole percents up. Because every measurement carries a little error, the raw sum almost never comes out to exactly one hundred, but on a healthy analyzer it comes out extremely close, because the individual errors are tiny and partly offsetting. That closeness is a genuine health signal: it means every component was measured about right and only small residual errors remain.
Normalization is the step that follows. The analyzer divides each component by the raw sum and multiplies by one hundred, so the reported composition sums to exactly one hundred by construction. This is a legitimate and standard operation, because a real gas does sum to one hundred percent and the small residual errors have to go somewhere. On a healthy GC the normalization adjustment is minute, nudging each component by a whisker, and the reported composition is essentially the same as the raw one.
The problem is that normalization is as much a cosmetic operation as a mathematical one. If the raw sum came out low, say because the whole analysis was reading light, normalization scales everything up to reach one hundred and hands back a composition that sums perfectly and looks flawless, even though every component is now biased. The raw sum was the honest record of how far off the measurement really was, and normalization is about to erase it. That is exactly why the unnormalized total deserves to be captured and watched before the scaling is applied.
A drop in the unnormalized total is a symptom shared by several of the faults that most threaten a custody analyzer. If carrier gas pressure falls or the carrier supply is running out, the whole analysis reads differently and the raw sum moves off one hundred. If a sample valve leaks or the sample volume is not being delivered correctly, less of the sample reaches the detector and the raw sum sags. If the column degrades and peaks broaden or shift, quantitation suffers across the board and again the raw sum drifts. One number responds to a whole family of physical problems.
What makes the unnormalized total uniquely valuable is that it moves before the reported composition looks wrong. Because normalization repairs the appearance of the composition, an operator watching only the final mole percents may see nothing amiss even as the analysis quietly degrades, since the reported numbers still sum to one hundred and stay in familiar ranges. The raw sum has no such disguise. It reflects the real state of the measurement directly, so it is often the first indicator to move when a fault begins and the last one anyone should ignore.
It is not a diagnosis by itself, and it does not need to be. The unnormalized total tells you that the measurement is no longer trustworthy and roughly how badly, which is enough to trigger an investigation before biased gas has been billed. Combined with other diagnostics such as retention-time shifts and response-factor trends, a drifting raw sum usually points quickly toward the specific cause. Its role is to be the alarm that fires first, buying the time to find and fix the underlying fault.
To act as an early warning, the unnormalized total has to be captured before normalization and carried out of the analyzer to somewhere it can be watched. Many analyzers compute and expose the raw sum as a diagnostic value, and the job of the SCADA layer is to bring that value into the monitoring host alongside the reported composition rather than letting it vanish inside the analyzer once the normalized result is produced. Once the raw sum is a live tag, it can be trended and alarmed like any other measurement.
A practical alarm sets a band around one hundred and fires when the raw sum leaves it. Because a healthy analyzer holds the raw sum tightly near one hundred, even a modest excursion is meaningful and worth flagging, and a widening band or a steady drift toward one edge is worth flagging sooner than a single reading. Trending the raw sum over time is as important as the instantaneous check, because a slow slide from one hundred toward, say, ninety-eight over many days is a decaying carrier supply or aging column revealing itself before it becomes a hard failure.
Watching the pre-normalization sum from a cloud host also scales across a fleet in a way a local panel cannot. An operator overseeing many analyzers can compare each unit's raw sum against its own history and against its peers, so the one analyzer whose raw sum is drifting stands out immediately. This turns the single best early-warning signal on a custody GC into a fleet-wide health screen, catching the carrier loss, valve leak, or column decay at one site before it biases a billing statement anywhere.
Because the composition sums to one hundred only because normalization forces it to. That scaling repairs the appearance of the numbers even when the underlying measurement is off, so the final composition can look perfect while every component is biased. The raw sum before normalization is the honest record of how accurate the measurement actually was, which is why it is worth capturing and alarming on.
A whole family of physical problems shows up in the raw sum. Carrier gas pressure loss or a depleting carrier supply, a sample valve leak or an under-delivered sample volume, and column degradation that broadens or shifts peaks all move the raw sum off one hundred. That is why the unnormalized total is such a useful single early-warning number, though it points to a cause only when combined with other diagnostics.
In the SCADA host, on the value captured before normalization. Many analyzers expose the raw sum as a diagnostic, and bringing that into a cloud host lets an operator set a band around one hundred, alarm on excursions, and trend slow drift over days. From a host the raw sum can also be compared across a fleet, so the one drifting analyzer stands out before it biases any billing.
Merobix reads your field devices into a cloud SCADA - the real thing behind these terms, live in days from any browser.