Revenue meter

What it is (precise)

The device is a static electricity meter built to a general-requirements standard plus an accuracy standard — IEC 62052-11 with IEC 62053-22 for active energy in IEC markets, ANSI C12.1 as the general code with ANSI C12.20 supplying the accuracy classes in North America. It samples voltage and current at instrument-transformer secondaries, nominally 1 A or 5 A and around 100 to 120 V, multiplies them into instantaneous power, integrates, and accumulates the result into registers after applying the CT and VT ratios.

A battery plant needs at least two energy registers rather than one, because it is both a generator and a load: delivered and received are separate quantities that settle at different prices under different tariff terms, and netting them inside the meter is a configuration decision with money attached.

Four-quadrant meters add reactive registers per quadrant, which matter wherever the market prices reactive energy or the connection agreement penalises power factor. Whether the meter is configured in the generator or the load convention is worth confirming rather than inferring — the sign convention entry covers the reconciliation.

How many measuring elements the installation needs is settled by Blondel's theorem: an N-wire circuit requires N−1 elements. A three-wire medium-voltage tie is metered with two elements; a four-wire circuit needs three. Installations that economise below that count are accurate only while the system stays balanced, which is exactly the assumption a fault, an unbalanced auxiliary board or a single-phase-loaded station service breaks.

The instrument transformers feeding the meter are also not the ones feeding the relays. A metering core is specified to saturate early — the instrument security factor in IEC 61869-2 exists precisely to keep fault current out of the meter — which is the opposite requirement to a protection core that has to stay linear at twenty times rated current. Both cores can live in the same physical CT unit; they are different windings doing incompatible jobs, and the protection relay entry owns the protection side of that pair.

What settlement consumes is not an instantaneous value but interval energy, and the interval has to match the market's settlement period: 15 minutes across the EU under the electricity balancing guideline, Regulation (EU) 2017/2195 (the Irish SEM excepted at 30 minutes), and half-hourly settlement periods in Great Britain under the Balancing and Settlement Code. Interval data is collected by the party the market rules designate — a utility, an ISO, or an accredited meter operator — then validated, estimated and edited before it settles.

That pipeline is the reason a revenue meter is not a telemetry stream, whatever the two have in common physically. Telemetry is an instantaneous sample sent for operational awareness and tolerates gaps; the meter's registers are a cumulative, auditable record where a gap has to be estimated under a documented rule and defended later. The telemetry entry owns the real-time path.

Accuracy class, and what the class actually binds

Start with the naming, because conflating the two systems is the most common error in a metering specification. IEC 62053-22 covers static meters for AC active energy, with classes 0.1S, 0.2S and 0.5S in its 2020 edition; the S denotes the classes intended for transformer-operated metering, which hold their error band down to a small fraction of rated current — the condition a grid-scale plant lives in for most hours, since a fleet sized for its peak spends a great deal of time well below it.

ANSI C12.20 states classes 0.1, 0.2 and 0.5 with no S. Writing "class 0.2S per ANSI C12.20" or "class 0.5 per IEC 62053-22" names a class the cited standard does not define, and a specification that cannot be tested against its own citation is not a requirement. Reactive energy is a separate standard with its own class list again, IEC 62053-24, so a project that needs reactive settlement needs both cited.

The class binds the meter, not the number that settles. The current and voltage transformers ahead of it contribute ratio errors that add directly to the meter's own, and their classes come from their own standards: metering classes including 0.2S and 0.5S for current transformers under IEC 61869-2, voltage-transformer classes from 0.1 to 1.0 under IEC 61869-3, and in North American practice a class paired with a burden designation under IEEE C57.13 — written as, for example, 0.3 B-0.5.

The burden half of that designation is load-bearing: the class holds only at or below the stated burden, so a long secondary run from a switchyard to a control building can push a CT out of its class without a single component failing or alarming. Some market rules deal with this by stating an overall accuracy limit for the whole metering installation rather than component classes, which is a different obligation to satisfy; read which one the tariff imposes before ordering anything.

The term that surprises people is phase displacement. Active power is V × I × cos φ, so an angular error δ introduced by the instrument transformers shifts the measured value by approximately −δ × tan φ in relative terms. At unity power factor tan φ is zero and the phase error contributes nothing at all; at 0.5 power factor tan φ is about 1.73, and ten arcminutes of displacement — 0.0029 radians — works out to roughly 0.5% on active energy, wider than the meter's entire class band.

A battery exchanging reactive power at low real power sits exactly where that term is largest, which is an argument for specifying phase-displacement limits on the instrument transformers and not only a tight class on the meter. It also explains why a metering installation that tests clean at full power factor during commissioning can behave differently during reactive-support duty.

Revenue-grade, check meter, and the measurement the controller uses

Three different populations of measurement live at the same interconnection and they are not substitutes. Revenue-grade metering is certified against the market's rules, sealed against tampering, fed from dedicated metering cores, tested at commissioning and re-tested at whatever interval the tariff sets, and read by or under the control of a party other than the plant's own SCADA.

The specific obligations are jurisdictional and do not transfer: ERCOT's settlement-metering requirements in Texas, PJM's metering requirements in its footprint, and in Great Britain the Balancing and Settlement Code's Codes of Practice, which band requirements by circuit size with the most demanding tier on the largest connections. Copying a metering specification from a project in another market is how a plant arrives at commissioning with equipment that is entirely competent and not registrable.

A check meter is an independent second metering installation of the same class, taken from separate transformer cores or windings, so that a failure shows up as disagreement between two records rather than as a settlement query months afterwards.

Where a market requires one, the contract still has to say which record governs, what magnitude of disagreement triggers a test, and how the affected intervals are estimated when a meter is found faulty. That last point deserves attention because estimation rules decide who absorbs the uncertainty, and the party writing the rule is rarely the party who will be paying under it.

The third measurement is the one the plant controller uses. A power plant controller closes its loop on a sub-second measurement of real and reactive power at the interconnection, taken from transducers or from the controller's own inputs, and that measurement exists to regulate rather than to settle. It will not agree exactly with the revenue meter — different transducers, different filtering, different sampling windows — and the sensible response is to characterise the offset at commissioning rather than to discover it in a settlement dispute.

A plant can regulate itself neatly inside its export limit on its own screens and still show an exceedance in the settlement record, and it is the settlement record the compliance conversation will use. The power plant controller entry owns the control loop; what belongs here is the boundary between a value used to act and a value used to invoice.

When the meter is not at the boundary: loss compensation

The point of interconnection is commonly defined at the high-voltage side of the main power transformer, while the practical place to install a metering set is the medium- or low-voltage side, where the instrument transformers are cheaper and reachable. Loss compensation closes that gap in the meter's firmware instead of in copper. No-load losses are treated as roughly constant while the transformer is energised; load losses scale with the square of current; the coefficients come from the transformer's factory test report, entered once and applied to every interval thereafter.

The same technique compensates a length of line or cable between the metering point and the named boundary. Two consequences follow immediately: the settled quantity is a computed number rather than a measured one, and the correctness of that number depends on the test-report values and the direction convention someone entered during commissioning.

The battery-specific consequence is that the compensation is applied against the plant twice per cycle. With metering on the plant side of the transformer, exported energy arrives at the boundary smaller than the meter read it, so the compensation subtracts; imported energy was drawn from the boundary larger than the meter read it, so the compensation adds. A one-way generator crosses that loss once per megawatt-hour sold. A battery crosses it on the way in and again on the way out.

The loss chain used elsewhere on this site puts roughly 0.5 to 1 percent in a transformer each way, so a full cycle metered on the plant side carries about one to two points of compensated loss over the round trip — a figure the revenue model has to include explicitly, because a model built on the assumption of a boundary meter simply does not contain it. Where compensation is configured to apply no-load losses whenever the transformer is energised, some of that debit continues at zero dispatch.

The other topology question is where auxiliary load taps relative to the metering point. Station service fed from a bus inside the metering point is netted out of the reading automatically, so every capacity and efficiency figure taken from that meter is already net of HVAC, thermal management and controls; station service fed from outside it is a separate meter, a separate bill, and a different definition of the same guarantee.

Neither arrangement is wrong and both are common, so the instruction is to trace the auxiliary tap on the one-line diagram against the metering CTs before agreeing to any test protocol that will be scored on that meter. The measurement-boundary entry owns the general rule about which bus a number belongs to; this is the specific drawing check that makes the rule enforceable.

Contracts, commissioning, and the ways it reads wrong

The contractual questions are short and worth settling early: which meter governs, which accuracy class under which standard with the jurisdiction named, who owns and seals the equipment, at what interval it is tested and by whom, what disagreement between main and check triggers a test, how intervals are estimated when a meter fails, and where the capacity test's energy is read.

That last item is where this entry meets the contract-energy entry, which owns the guaranteed quantity: the demonstrating number is whatever the meter's registers say between two timestamps, so the test protocol has to name the register, the interval resolution and the correction treatment, not merely the megawatt-hours.

It is also worth recording that a class is a limit on error rather than an expected value — two installations each sitting inside ±0.2% can differ from one another by twice that — which is why a tolerance for main-versus-check disagreement belongs in the document rather than in an argument.

The failure modes to plan for are the ones that produce plausible readings. On a three-element installation, a blown VT fuse removes one element and registers roughly a third less on a balanced load, which looks like an underperforming plant rather than a broken meter. A reversed CT polarity turns export into import on one element. A swapped pair of phases puts current against the wrong voltage and shifts the apparent power factor without changing any magnitude on any nameplate.

None of these announce themselves, and none are caught by a factory certificate. What catches them is a phasor check at commissioning with real power actually flowing, compared against the check meter and the controller's own reading. A battery has an advantage over most plant here: it can be commanded to produce controlled flow in all four quadrants, so a test that visits export, import, injection and absorption is a scheduling matter rather than an opportunity waited for.

The last item is time, because the meter's clock decides which settlement interval a megawatt-hour lands in, and adjacent intervals can carry very different prices. A clock adrift by minutes moves real energy between real prices while every magnitude in the record stays correct. The time-synchronization entry owns the mechanism and the sources; what belongs here is the reason to care, and the commissioning step of confirming that the meter, the check meter and the plant's own records all timestamp the same event the same way before anyone tries to reconcile them.