A gas chromatograph measures every component of a gas and computes an accurate heating value from that full composition, but a custody GC is expensive to buy, install, and maintain, and not every point can justify one. Correlative or inferred heating value methods offer a cheaper alternative: they measure a small number of physical properties, such as the speed of sound in the gas together with its carbon dioxide content, and infer the heating value from a correlation rather than from a full component analysis. These methods work well for gas that stays within the range the correlation was built for and can drift when the gas moves outside it. This guide explains how correlative energy measurement works, which property combinations are used, where the correlation holds and where it breaks, and how a host validates an inferred value against periodic GC checks.
Inferred Heating Value in one line: An inferred or correlative heating value is an estimate of a gas stream's energy content derived from a small set of easily measured physical properties instead of a full component gas chromatograph analysis. A common approach measures the speed of sound in the gas along with its carbon dioxide content and maps those to a heating value through a correlation, and other methods combine density with a diluent measurement. These techniques are cheaper than a full GC and work well while the gas composition stays within the range the correlation was calibrated for, but they lose accuracy when the composition drifts outside that range, so they are usually validated periodically against a reference GC.
The heating value of a gas is ultimately a property of its composition, but composition is expensive to measure fully. Correlative methods exploit the fact that several bulk physical properties of a gas are also set by that same composition and are much easier to measure. The speed of sound in a gas, for instance, depends on the mix of molecules present, and so does the gas density. If you can measure a couple of such properties, and you constrain the problem with a measurement of the main inert diluent, you can work backward through a correlation to a heating value without ever resolving the individual hydrocarbons.
A widely used combination pairs the speed of sound with the carbon dioxide content. Speed of sound alone does not uniquely fix the heating value because different compositions can share a sound speed, but the ambiguity is largely resolved once the amount of the main non-combustible diluent is known, and carbon dioxide is measured cheaply and directly by an infrared sensor. With sound speed and carbon dioxide in hand, a correlation returns a heating value that tracks the true value well for pipeline-quality gas. Other implementations combine gas density with a diluent measurement to achieve the same end by a different property route.
The attraction of all these methods is cost and simplicity. A speed-of-sound sensor and an infrared carbon dioxide analyzer are far cheaper and lower maintenance than a full chromatograph with its carrier gas, columns, and calibration standards, and they respond quickly and continuously. For a point where a full GC cannot be economically justified but some energy measurement is needed, inferred heating value gives a usable, continuous number at a fraction of the capital and operating cost, which is exactly why these correlative instruments exist.
A correlation is only valid over the range of compositions it was built from. For gas that stays close to typical pipeline quality, dominated by methane with modest ethane and small amounts of the usual diluents, the relationship between the measured properties and the heating value is stable and the inferred value tracks the truth closely. In that regime the correlative instrument is a genuine substitute for a GC for the purpose of energy measurement, because the composition simply does not vary in ways the correlation was not designed to handle.
The methods lose accuracy when the composition moves outside that expected range. Unusual amounts of heavier hydrocarbons, unexpected diluents the correlation does not account for, or a shift in the balance of components can all push the gas into territory where the mapping from measured property to heating value no longer holds, and the inferred value drifts from the true one. The heavy end is particularly awkward because a small change in heavy hydrocarbon content moves the heating value noticeably but may barely register in the measured bulk properties, so the correlation can miss it. Streams with variable or off-nominal composition are therefore the weak point of any inferred method.
This means the fit of a correlative instrument to a location depends entirely on how well behaved the gas there is. A single-source stream of consistent pipeline gas is an excellent candidate, because the composition rarely leaves the correlation's comfort zone. A point that sees blended or variable gas, or gas with an unusual heavy or inert content, is a poor candidate, because the correlation will be right on average but wrong exactly when the composition swings, which is often when accurate energy measurement matters most. Understanding the composition variability at the point is the first step in deciding whether inference is appropriate at all.
Because a correlative method can drift when the composition leaves its calibrated range, an inferred heating value should not be trusted indefinitely without a reference. The standard safeguard is to compare it periodically against a full GC analysis, either from a portable or shared chromatograph brought to the point or from a spot sample sent to a laboratory. That comparison confirms the correlation is still valid for the current gas and, if the two disagree, reveals that the composition has moved into a range the inference no longer handles well, prompting a recalibration or a rethink of the method's suitability there.
The role of a monitoring host in this is to keep the inferred value and the reference checks side by side so drift is visible rather than hidden. A host that stores every inferred reading and overlays the occasional GC or lab result can show whether the two agree consistently or whether the inferred value is slowly diverging, which is the signature of a composition change the correlation is not tracking. Rather than a single pass-or-fail check, the value comes from watching the difference between inferred and reference over time and reacting when it opens up.
A cloud SCADA platform such as Merobix is well suited to hold this reconciliation. It can trend the continuous inferred heating value for the point, record each periodic GC or lab reference against the same timeline, and alarm when the gap between them exceeds a tolerance, signalling that the correlation may no longer be valid for the current gas. It can also flag when a scheduled reference check is overdue, so an inferred value never runs unchecked for too long. Presenting the cheap continuous estimate and its periodic anchor together lets an operator get the economy of inference at low-value points while keeping enough oversight to know the moment the estimate stops being trustworthy.
Because heating value and several easily measured physical properties are both determined by the same composition, you can measure a small set of those properties and infer the heating value through a correlation. A common method combines the speed of sound in the gas with its carbon dioxide content, and others combine density with a diluent measurement. This avoids the cost of a full chromatograph while still producing a continuous energy number, as long as the gas stays within the range the correlation was built for.
It loses accuracy when the gas composition moves outside the range the correlation was calibrated for, such as unusual amounts of heavy hydrocarbons, unexpected diluents, or a variable blend. The heavy end is especially troublesome because small changes in heavy hydrocarbon content move the heating value noticeably but may barely register in the measured bulk properties. Points with consistent single-source gas are good candidates, while points with blended or variable gas are poor ones because the correlation errs exactly when composition swings.
It is validated periodically against a full GC analysis, either from a portable or shared chromatograph or a spot sample sent to a lab, to confirm the correlation is still valid for the current gas. A monitoring host stores the inferred value and the reference checks together so any slow divergence between them becomes visible, which signals that the composition has drifted outside the correlation's range. Watching that gap and reacting when it widens, rather than trusting the inference indefinitely, is what keeps the method honest.
Merobix reads your field devices into a cloud SCADA - the real thing behind these terms, live in days from any browser.