POI capability envelope

What it is (precise)

Read the one-line and the objects between the two boundaries are countable: the PCS AC terminals at low voltage, the unit transformer stepping up to the collection voltage, the MV collection cables and switchgear, the main power transformer where the project has one, and the revenue meter at the point of interconnection.

The terminal envelope is a region in the P-Q plane, and the P-Q capability entry covers its shape and what clips it. The POI envelope is that same region moved and shrunk — displaced inward on the P axis by real-power losses, displaced on the Q axis by reactive absorption, and conditional on ambient temperature, DC bus voltage and AC voltage in a way the requirement it must satisfy is not.

The real-power reduction is small and predictable. POI power sits roughly 2 to 4 percent below the summed inverter nameplate: the MV transformer and the collection take about 0.5 to 1 percent each, and station auxiliaries another 1 to 3 percent depending on climate and duty, with the stages rarely all at their maxima in the same hour.

One shipping 4,140 kVA / 690 V medium-voltage skid transformer publishes 4.6 kW of no-load loss and 43 kW of load loss at rated power — 1.15 percent of its own rating with both counted, and about 1.0 percent for the eco-design variant of the same frame. Load loss falls with the square of loading while no-load loss is drawn whenever the unit is energized, so the percentage haircut is heaviest at full output and never reaches zero at standstill.

Reactive absorption is the term most often left out, because nothing on a transformer nameplate is labelled Mvar. What the plate does carry is impedance voltage, and its reactive component does the work: a transformer absorbs vars in proportion to loading squared times its own reactance, so a unit whose impedance voltage is 5.5 to 8 percent at rated current takes on the order of 220 to 320 kvar out of a 4 MVA-class block at full load.

One nameplate in hand sits above that band: its reactive component alone runs 8.48 to 9.14 percent across the rating range, which is 340 to 370 kvar from the same 4 MVA block.

Read the plate rather than the typical band, and read the reactive component rather than the headline impedance voltage — the reactive component is never the larger of the two, 9.14 against 9.17 percent on that plate. Two properties make this consequential at the meter. The absorption is quadratic, so it peaks exactly when the plant is at full output and the reactive obligation is hardest to meet. And its sign does not reverse when real power reverses: a transformer absorbs vars while the plant charges just as it does while the plant discharges.

Why the requirement sits at the meter and the curve sits at the terminals

The grid operator's interest in the plant ends at the meter, and so does its ability to verify anything. The requirement is therefore written at a point the operator can measure and the owner cannot relocate — internal losses, converter count and transformer choices are the developer's problem, not a defence. In the United States, FERC Order 827 removed the reactive-power exemption for non-synchronous generation and set the obligation on newly interconnecting non-synchronous resources as a plus or minus 0.95 power-factor range at the high side of the generator substation.

IEEE 1547-2018 for distribution-connected resources and IEEE 2800-2022 for transmission-connected inverter-based resources both formalize the idea with a defined reference point of applicability, the bus at which each requirement is evaluated, rather than leaving it to be argued during witness testing. Grid codes in other markets name their own point; neither its location nor its name travels between jurisdictions.

The vendor draws at the terminals for the mirror-image reason: the terminals are the only boundary it controls, and everything past them is somebody else's design. Datasheets say so on their face.

One 3,950-class central PCS publishes a reactive capability of plus or minus 2,305 kvar at zero power factor and annotates the figure with 40 degrees C design temperature, 1,159 Vdc minimum battery voltage, 0.9 pu AC voltage, an installation altitude of 0 to 1,000 m, and the explicit note that all values are at inverter terminal level. That is a complete and honest statement of capability. It is also not the quantity the interconnection agreement asks about.

The gap between the two documents is conditionality, not just distance. The same converter family delivers 3,960 kVA at 25 degrees C, 3,610 kVA at 40 degrees C and 3,365 kVA at 50 degrees C on a 1,200 Vdc bus — a 15 percent spread with no change to the hardware — and the reactive figure above is valid only down to a stated minimum battery voltage, which means low state of charge trims it too.

The interconnection agreement carries no ambient row and no SOC clause. Compliance is owed at the site's worst hour and at whatever state of charge the plant happens to be at, so the binding case is the hot, low-SOC, off-nominal-voltage corner rather than the reference row that headlines the sheet.

Translating a terminal envelope to the meter

The translation is mechanical once the conditions are fixed. Pick the design ambient, the minimum DC bus voltage the EMS will permit, and the AC voltage band the POI is expected to swing across, then read the terminal envelope for one converter at that combination rather than at reference conditions. Multiply by the number of blocks that will actually be in service, not the number installed. Subtract auxiliary real power and the transformer and cable losses at the loading each corner implies, remembering that load losses scale quadratically while no-load loss is flat.

Subtract the var absorption of every series reactance in the path on the same quadratic basis, and add back the charging vars the MV collection cables contribute, which help most at light load and are modest against transformer absorption at full output. What comes out is a region, and the compliance test is whether that region contains the required one at its corners — full discharge, full charge, and zero real power — not whether two curves look similar in the middle.

Where station service is tapped decides whether the auxiliary load appears in the metered number at all. Fed from the collection bus inside the fence, it is netted out of export and shows up directly in the POI envelope; fed from a separate utility service, it never crosses the revenue meter and the plant's export is not reduced by it, though somebody still pays for the energy.

The general rule behind this belongs to the measurement-boundary entry; what matters here is that the same plant produces two different POI envelopes depending on a single-line decision made by the electrical contractor, and the number in the interconnection study has to match the wiring that was actually built.

How it shows up in specs, studies and contracts

In the interconnection process the envelope is a study deliverable. The plant-level P-Q study takes the fleet of terminal curves, models transformer impedances and collector reactive losses, and reports the resulting capability at the reference point in power-flow cases across the required voltage range.

Utilities test the corners of that claim, not its interior, and the plant controller has to be able to dispatch there in practice: a POI-referenced Q or voltage setpoint means the controller is regulating a quantity it measures downstream of every loss it must compensate, with the deadband and response time the code specifies.

Contractually, the phrase to look for is the point at which each obligation is measured. A reactive-capability clause naming the inverter terminals and an interconnection agreement naming the substation high side are not the same commitment, and the difference is worth several percent of MVA plus the whole var absorption of the step-up.

The commissioning test settles it: the witnessed reactive demonstration happens at a defined meter, at the day's ambient rather than the design ambient, and whether the result is temperature-corrected — and by whose method — should be written down before anybody schedules it. Where scope is split between a PCS supply agreement and an EPC contract, establish who owns the gap if the delivered POI envelope is smaller than the modelled one, because the equipment can be exactly on datasheet while the plant misses its obligation.

Common pitfalls

The charging quadrant is where POI-referenced power-factor bands are missed most often, and the reason is the non-reversing sign of var absorption. A design checked only at full discharge sees the transformer taking vars while the converters are already working hard to supply them; swing to full charge and the absorption is unchanged while the converter's own limits and the plant's loss chain have moved.

Check both signs of P at every voltage in the required band. The zero-P corner deserves the same treatment: standby reactive support is a real obligation in several markets, and at P near zero the plant's losses are small but the transformer still absorbs whatever the reactive flow through it demands.

Two further traps are structural rather than arithmetic. The first is treating the envelope as fixed: a converter block out for maintenance, a container isolated after a fault, or a derated skid on a hot afternoon each shrink the POI envelope, so reactive compliance is partly an availability question and belongs in the same conversation as the availability guarantee.

The second is checking the envelope once, at the reference conditions on the datasheet, and carrying that result through detailed design unchanged — the vendor curve moves with ambient, DC voltage and AC voltage, and the requirement it is being compared against does not move at all.