For custody transfer applications, specify a pressure transmitter with an accuracy class of ±0.075% of calibrated span or tighter, paired with a long-term drift below 0.05% URL per year, so the metered volume error stays inside the contractual measurement uncertainty budget [S2][S4].
Capacitive sensing technology delivers 0.04–0.1% full-scale accuracy, the narrowest band in industrial pressure measurement, and its very low long-term drift makes it the preferred transducer class for custody transfer and regulatory metering points [S4]. The decision is driven by the overall measurement uncertainty budget of the metering skid, not by the sensor headline number alone [S5].
What "Accuracy Class" Actually Means on a Custody Transfer Skid
Accuracy class is the vendor's reference figure, typically expressed as ±X% of calibrated span or ±X% of URL (upper range limit), and it describes performance under reference conditions of 20–25 °C, stable line pressure, and a fixed mounting orientation [S7]. On a custody transfer skid, the figure that matters is total probable error in the field: reference accuracy, combined with thermal zero shift, static line pressure effect (footprinting), long-term drift, and turndown from URL, all combined via root-sum-of-squares inside the meter uncertainty budget [S6][S5]. The reference number on the data sheet is therefore necessary but never sufficient on its own.
Static line pressure is the single largest hidden error source on differential pressure transmitters used on orifice runs. Even a high-accuracy class unit can lose effective performance if footprinting is not corrected at the operating static pressure [S6]. A common engineering rule is to size the calibrated span between 30% and 80% of URL so that the turndown penalty does not inflate the reference accuracy by a factor of two or more.
Selection Criteria Tied to the Uncertainty Budget
The first selection step is to fix the allowable process error in physical units, for example, "liquid level must stay within ±2 cm on a 5 m tank" or "flow must stay within ±0.5% of indicated rate", before touching any catalog number [S8]. From that target, work back to the required reference accuracy, then add margin for thermal effects, drift over the calibration interval, and the effect of remote seals if the transmitter is mounted away from the primary element [S8].
Three decision criteria dominate custody transfer specification: (1) reference accuracy, (2) long-term drift per year, and (3) static line pressure immunity. Piezoresistive silicon sensors typically deliver 0.1–0.25% FS with 0.02–0.05% per °C zero shift, capacitive sensors reach 0.04–0.1% FS with minimal drift, and bonded foil strain gauge sensors sit at 0.3–1.0% FS but survive harsh high-pressure, high-temperature service where silicon fails [S4]. Capacitive remains the reference-class choice; piezoresistive covers high-volume process lines; strain gauge covers hydraulics, boilers, and compressor housings.
Calibration interval is typically 12 months for critical custody points, and instrument drift is reported to drive a large share of process upsets, which is why low-drift classes justify the price premium [S4].
How the Class Numbers Map to Skid Roles
On an orifice-based liquid custody skid, the practical threshold is a transmitter with ±0.075% of span or better, combined with a turndown limited to roughly 4:1 from URL, which keeps the worst-case flow error in the ±0.5% to ±2% range typical of well-designed orifice plate systems [S2][S5]. Below ±0.075% of span, the transmitter stops being the dominant term in the uncertainty budget and the residual error lives in the primary element, the flow computer, and the proving.
On a gas custody transfer run, where differential pressure is often the calculated variable, calibration against a traceable reference is mandatory and the same ±0.075% class holds; Fluke's custody transfer calibration procedure for DP meters is built around verifying reference, line pressure, and thermal performance as separate line items [S3]. Skid designers should also evaluate absolute pressure transmitters for densitometer and compressibility inputs, where the same accuracy-class logic applies, and level transmitters for prover tanks, where the same drift discipline prevents volume bias between proving cycles.
Sourcing, Standards, and What the Audit Will Check
Custody transfer is governed by weights-and-measures rules that demand traceability, not a particular sensor brand. The American Petroleum Institute and ISO guidelines create the consistent demand for traceable differential pressure measurement on liquid hydrocarbon and gas runs, with regulators focused on accuracy class, calibration interval, and the documented uncertainty budget [S5][S1]. Field audits will look for the calibration certificate, the as-installed zero and span values, the static pressure footprint correction factor, and proof that the operating span is inside the vendor's specified turndown envelope.
Practical specification for new builds: pick a transmitter with ±0.075% of span reference accuracy, capacitive or high-grade piezoresistive silicon, long-term drift below 0.05% URL per year, HART or Foundation Fieldbus output, 4:1 maximum turndown, and a 12-month calibration interval tied to a documented uncertainty budget that includes the temperature transmitter input if a multivariable unit is in use [S2][S4][S5]. For the same selection applied to a level skid, see the capacitance level transmitter reference and the related sizing guide for DP ranges in DP transmitter range sizing for orifice flow meters, since the calibrated-span logic carries over directly.
Trackable signals to watch: API MPMS chapter revisions that retighten the uncertainty budget for hydrocarbon skids, and any expansion of weights-and-measures approval for wirelessHART custody-grade transmitters on gas runs [S1][S5]. On the calibration side, footprinting correction curves and the shift from 12-month to 6-month verification on high-pressure gas runs remain the practical pressure points to monitor over the next two quarters [S6].