A turbine or positive-displacement meter does not behave identically at every flow rate; its calibration drifts a little as the rate changes, and a single fixed factor cannot correct for that across the whole range. A flow computer solves this by holding a curve of correction factors and choosing the right one for the rate it is seeing, moment to moment. Alongside that, it keeps track of how long flow has actually been present, a quantity called flow time, which matters for accumulating totals correctly. This guide explains what flow time is, how flow-dependent linearization works with a multi-point curve and breakpoint interpolation, and how that runtime interpolation differs from applying one fixed K-factor.
Flow Time and Linearization in one line: Flow time is the accumulated time during which a flow computer has detected actual flow, as distinct from elapsed clock time, and it is tracked so that totals and averages reflect only the periods when the meter was really flowing. Flow-dependent linearization is the technique of correcting a meter's output using a multi-point curve rather than a single factor, so that the meter factor or K-factor applied varies with the flow rate. The flow computer reads the current rate, finds where it falls among the curve's breakpoints, interpolates between them to get the correct factor at that rate, and applies it in real time.
Flow time is exactly what it sounds like: the total time during which the flow computer has recognised that fluid is actually moving through the meter. It is deliberately different from elapsed clock time, because a meter can sit for long periods with no flow, and those idle periods should not be treated the same as periods of genuine flow. The flow computer decides that flow is present when the rate is above a low-flow cutoff threshold, and it accumulates flow time only while that condition holds, pausing the counter when flow stops.
The reason a flow computer bothers to track this separately is that several important quantities should be based on flowing conditions, not on wall-clock time. Flow-weighted averages, for instance, should reflect the properties of the fluid while it was actually flowing, so the averaging is tied to flow time rather than to every second on the clock. Totals are accumulated only while flow is present, and the flow-time counter provides an auditable record of how much of a period the meter was genuinely in service. For custody and allocation measurement this record matters, because it distinguishes a meter that flowed steadily from one that flowed intermittently even if both reported the same total.
The low-flow cutoff that gates flow time also protects the measurement from spurious accumulation. Near zero flow a meter's signal can be dominated by noise, vibration, or a slowly creeping indication that does not represent real throughput, and counting that as flow would inflate totals and distort averages. By requiring the rate to exceed a defined cutoff before flow time and totals accumulate, the flow computer ignores that near-zero region. Flow time therefore is not just a clock; it is a record of the periods the computer judged to be valid, flowing measurement.
A turbine or PD meter produces a pulse frequency proportional to flow, and the factor that converts pulses to volume - the K-factor or, after proving, the meter factor - is not perfectly constant across the meter's range. At low flows the meter tends to behave differently than at high flows, so a single fixed factor will be right at some rates and slightly wrong at others. Flow-dependent linearization addresses this by characterising the meter at several flow rates and storing a curve of factors, each valid near a particular rate, rather than a single number.
The curve is stored as a set of breakpoints, each pairing a flow rate, or the corresponding pulse frequency, with the correction factor that applies there. These breakpoints come from calibrating or proving the meter at multiple flows across its operating range. Between the breakpoints the flow computer interpolates: for a rate that falls between two stored points, it computes a factor by linearly interpolating between the factors at those two points, so the applied correction moves smoothly with rate rather than jumping at each breakpoint. This breakpoint interpolation is the runtime heart of flow-dependent linearization, and it is done continuously as the rate changes.
Because the interpolation happens live, the factor the flow computer applies tracks the flow rate in real time. As the rate rises into a higher part of the curve, the computer selects the surrounding breakpoints, interpolates a factor appropriate to that rate, and uses it to convert pulses to volume; as the rate falls, it shifts to lower breakpoints and a correspondingly different factor. The more breakpoints the curve has, the more finely the meter's real nonlinearity is followed, at the cost of needing more calibration points to define them. The result is a correction that reflects how the meter actually behaves at whatever rate it is running, moment by moment.
The contrast that makes flow-dependent linearization worth understanding is with the simpler alternative of a single fixed K-factor. With a fixed factor, the flow computer multiplies the meter's pulses by one constant regardless of rate. That is adequate when the meter is genuinely linear across its working range, or when it always runs at close to one rate, because then a single factor represents the meter's behaviour well enough. Its appeal is simplicity: one number to configure and prove, and no curve to maintain.
The limitation shows up when the meter is used across a wide range of flows and its calibration is not flat across that range. A fixed factor chosen to be right at mid-range will over- or under-read at the extremes, introducing an error that grows as the rate moves away from the point the factor was set for. Flow-dependent linearization removes that error by applying, at each rate, the factor that was actually determined for that rate, so the correction is right across the whole range rather than at one point. This is the practical reason wide-turndown turbine and PD installations use a curve instead of a constant.
It is worth being precise that this is a runtime, in-computer function, which is what distinguishes it from simply owning a meter factor curve on paper. The curve of factors is the data; the flow-dependent linearization is what the flow computer does with that data every scan, reading the rate, locating it among the breakpoints, interpolating, and applying the result. In a SCADA-monitored installation, a cloud platform such as Merobix can trend the flow rate alongside the applied factor and the accumulated flow time, letting an engineer confirm the meter is spending its time in the part of the curve it was proved over and that flow time and totals are accumulating only during valid flow. That visibility helps verify the linearization is doing its job and flags when a meter starts operating outside the range its curve was built for.
Flow time is the accumulated time during which the flow computer has detected actual flow, meaning the rate is above the low-flow cutoff, as opposed to elapsed clock time. It is tracked separately so that totals accumulate only while flow is present and flow-weighted averages reflect flowing conditions rather than idle periods. It also provides an auditable record of how much of a period the meter was genuinely in service.
Flow-dependent linearization corrects a meter's output using a curve of factors that vary with flow rate, rather than a single fixed factor. The curve is stored as breakpoints pairing flow rates with correction factors, determined by calibrating or proving the meter at several rates. At runtime the flow computer finds where the current rate falls among the breakpoints and interpolates between them to apply the right factor for that rate.
A single K-factor multiplies the meter's pulses by one constant regardless of rate, which is fine only if the meter is linear across its range or always runs near one rate. A multi-point curve stores a different factor for each of several rates and interpolates between them at runtime, so the correction is right across the whole range. Wide-turndown turbine and positive-displacement meters use a curve because a fixed factor would over- or under-read at the extremes.
Merobix reads your field devices into a cloud SCADA - the real thing behind these terms, live in days from any browser.