A custody gas chromatograph reports a new heating value every few minutes, and a billing period contains thousands of those analyses. To settle the energy that changed hands, all those readings have to collapse into one representative heating value for the period, and how you do that averaging matters. A flow-weighted average ties each analysis to the volume that flowed while it was in effect, so the readings taken during heavy flow count for more than those taken while the pipe was nearly idle. Getting this right is the difference between billing the gas that actually moved and billing an accident of the clock.
Flow-Weighted Heating Value in one line: A flow-weighted heating value average is a period average of BTU in which each analysis is weighted by the volume of gas that flowed during the time that analysis was current, rather than counting every analysis equally. It differs from a simple time average, which treats every reading the same regardless of flow. The two diverge whenever heating value and flow rate change together, and the flow-weighted result is the one that correctly reflects the energy the pipe actually delivered.
Which page do you need? This page focuses on the calculation, worked through step by step. For the definition and why flow-weighting matters, see Flow-Weighted Average Heating Value.
The simplest way to average a period of heating values is to add up every analysis and divide by how many there were, which gives a time average. Each reading counts equally, so an analysis taken at three in the morning while barely any gas moved carries the same weight as one taken at peak flow. This is easy to compute and perfectly fine when flow is steady, but it answers the wrong question the moment flow varies, because it describes the average quality of the analyses rather than the average quality of the gas that was delivered.
A flow-weighted average asks the right question. Instead of treating every analysis equally, it weights each one by the volume of gas that flowed while that analysis was the operative composition. An analysis in effect during an hour of heavy flow contributes far more to the average than one in effect during an hour of trickle, because far more gas was actually sold at that heating value. The weighted result is the heating value that, multiplied by the total volume, reproduces the true total energy delivered over the period.
The distinction only bites when flow and heating value move together, and unfortunately they often do. If the stream's composition shifts at the same times its flow rate shifts, perhaps because a richer or leaner source comes online with its own flow pattern, then the periods of high flow carry a systematically different heating value than the periods of low flow. A time average blind to flow mixes these on equal terms and lands on a number that matches neither the busy hours nor the quiet ones, biasing the energy attributed to the period.
Picture a day where the gas is richer, with a higher heating value, during the hours the pipe runs hard, and leaner, with a lower heating value, during the hours it barely flows. A simple time average sees an equal number of rich-hour and lean-hour analyses and settles somewhere in the middle. But almost all of the actual gas left the pipe during the rich, high-flow hours, so the true energy content of the delivered gas is close to the rich value, well above the time average. The time average understates the energy, and every unit billed at that lower figure shortchanges the seller.
Reverse the correlation and the error reverses with it. If the gas happens to be leaner during the high-flow hours and richer during the quiet ones, the time average is pulled up by the many rich but low-volume analyses, while the gas that actually moved was mostly lean. Now the time average overstates the delivered energy and the buyer is billed for BTUs that never flowed. In both cases the sign and size of the error depend on how flow and quality happen to line up, which is exactly why a method blind to flow is unreliable.
Flow weighting removes the guesswork by letting the volume decide the weight. Because each analysis contributes in proportion to the gas that flowed under it, the rich high-flow hours dominate the average when they should and recede when they should not. The flow-weighted heating value is not a refinement for its own sake; it is the only average that, combined with the measured volume, yields the energy the contract is meant to settle. That is why custody agreements generally specify it for the billing period.
The natural place to form a flow-weighted average is at the flow computer, because that is the device that sees both inputs at once. The flow computer receives the live heating value from the analyzer and measures the volume flowing at each instant, so it can accumulate energy directly by multiplying the current heating value by the volume as the gas passes, integrating BTUs over the period. Dividing that accumulated energy by the accumulated volume yields the flow-weighted heating value for the period without ever having to reconstruct which analysis applied when.
This local accumulation matters because it is robust to the messiness of real operation. Analyses arrive at their own cadence, flow varies continuously, and communications can hiccup, but a flow computer integrating energy in real time captures the true product of quality and volume moment by moment rather than trying to stitch it together afterward from separate logs. The energy total it maintains is the contract-correct quantity, and the flow-weighted heating value falls out of it as a byproduct.
The SCADA host role is to gather, reconcile, and preserve those results across many meter runs and long periods. Pulling the accumulated energy, the volume, and the flow-weighted heating value from each flow computer into a central host lets an operator produce the billing figure for the period, cross-check it against a time average to see how far flow weighting moved the number, and hold an auditable record of how the period energy was formed. When the two averages diverge sharply, that gap is itself a signal worth trending, because it flags periods where flow and quality were strongly correlated and the choice of averaging method had real money riding on it.
Whenever heating value and flow rate change together over the period. If quality is the same during high-flow and low-flow hours, the two averages agree closely. But when the gas is systematically richer or leaner during the hours it flows hardest, the flow-weighted average follows the gas that actually moved while the time average is pulled toward analyses that carried little volume, and the two diverge.
Custody agreements generally specify the flow-weighted heating value for the billing period, because it is the only average that, multiplied by the measured volume, reproduces the true energy delivered. A simple time average describes the average quality of the analyses rather than of the gas that flowed, so it can bias the energy in either direction depending on how flow and quality line up.
Usually at the flow computer, because it sees both the live heating value from the analyzer and the instantaneous volume, and can integrate energy directly by multiplying them as the gas flows. Dividing accumulated energy by accumulated volume gives the flow-weighted heating value. A SCADA host then gathers those results across meter runs, produces the billing figure, and preserves an auditable record of the period energy.
Merobix reads your field devices into a cloud SCADA - the real thing behind these terms, live in days from any browser.