A 10× Recharge Rate Overruns the Last 4% of the 15 MW Service Cap

Battery recharge appears in the heat-rejection basis as a 233 kW [MODELED] transient at ten hours, or 581.8 kW [MODELED] at four, computed on the 2,379 kWh [MODELED] three-system floor for a recovery to 90 % [MODELED] state of charge at an assumed 0.92 [MODELED] charger efficiency. Shorten that recovery to one hour — ten times the ten-hour baseline — and the same conversion gives 2,327 kW [SENSITIVITY], which the margin cannot absorb at any point in the envelope. A separate industry custom restores a 5-minute [MODELED] module in 50 minutes [MODELED], 219 kW [MODELED] per 2 MW [MODELED] module, 1,754 kW [MODELED] across eight. Against the 14,391 kW [MODELED] facility figure the 15,000 kW [TARGET] service has 609 kW [MODELED] left — 4.1 % [MODELED] of the cap — and that custom overruns it by 1,145 kW [MODELED]. The 50-minute custom is not the title's 10×. Those two clocks must not be added as if they were one rate.

The 50-Minute Custom Is 11% Against a 4% Remainder

The first electrical basis is explicit about that 50-minute custom, ten times a five-minute discharge rather than ten times the ten-hour recovery. Discharge energy on a 2,000 kW module for five minutes is 166.7 kWh [MODELED] on the alternating-current side and 173.6 kWh [MODELED] on the direct-current side. Restore in 50 minutes to about 95 % [MODELED] at an assumed charger efficiency of 0.95 [MODELED] and the module asks for 219 kW, which is 11 % [MODELED] of its rating. Eight modules at that rate are 1,754 kW. The 609 kW remainder is 4.1 % of 15,000 kW. The custom is not a battery parameter that happens to be aggressive. It is larger than the entire remaining service.

The suggested freeze on that first table is 3.5 % [MODELED] of module rating, 70 kW [MODELED] per module, 560 kW [MODELED] across eight. That fits 609 kW with 49 kW [MODELED] left. Fourteen modules at the same 70 kW would be 980 kW [MODELED] and would have to charge in batches of eight or fewer. Twelve modules would be 840 kW [MODELED] and would also have to batch. The 609 kW remainder is not spare. It is almost entirely spoken for once the 3.5 % cap is in.

The 10× Case Is One Hour, Not the 50-Minute Custom

The heat-rejection basis does not reuse that 50-minute custom. It sets a single conversion — input power equals battery energy times the state-of-charge recovery fraction, divided by recovery time and by charger efficiency — and applies it to the 2,379 kWh floor: 2,379 × 0.9 ÷ 10 ÷ 0.92 gives approximately 233 kW, and 2,379 × 0.9 ÷ 4 ÷ 0.92 gives 581.8 kW.

That second figure is itself a correction. The revision 0 text carried 535 kW [MODELED], which omitted the efficiency divisor and so stated battery-side output rather than charger input — a +46.8 kW [MODELED] error, logged as an errata item and reflected in the shared engineering register. On the strict 2N battery basis of 3,172 kWh [MODELED] the same formula gives roughly 310 kW [MODELED] at ten hours and 775.6 kW [MODELED] at four.

Applying the identical formula at a one-hour recovery — ten times the ten-hour baseline — gives 2,379 × 0.9 ÷ 1 ÷ 0.92 = 2,327 kW [SENSITIVITY]. Nothing in the source names a one-hour case. The arithmetic is performed here and carries a sensitivity label for that reason. That one-hour case is the title's 10×, of the heat-rejection ten-hour clock, not of the 50-minute industry custom. The 2,327 kW figure is included because the ten-hour value is a schedule choice rather than a plant property, and a sensitivity that spans an order of magnitude on the input deserves to be seen next to the margin it is being compared against.

The 642 kW Remainder Already Includes Recharge

Two margins exist against the 15,000 kW cap and they are not interchangeable. Steady facility power of 14,125 kW [MODELED] leaves 875 kW [MODELED]. The charging-inclusive peak of 14,358 kW [MODELED] leaves 642 kW [MODELED], which is 4.3 % [MODELED] of the cap. The difference between the two is exactly the 233 kW of recharge, which means 642 kW is the remainder that already has recharge subtracted from it. The basis says so directly, and prohibits subtracting the 233 kW a second time. 14,125 kW and 14,358 kW are the powers. 875 kW and 642 kW are the remainders.

Against that 642 kW remainder, the four-hour case at 581.8 kW leaves 60 kW [SENSITIVITY]. That four-hour rate consumes 91 % [SENSITIVITY] of the charging-inclusive remainder. The ten-hour rate consumes 36 % [SENSITIVITY] of it. Roughly 0.4 % [SENSITIVITY] of the service cap separates a legitimate accelerated-recovery decision from the physical input limit, and the decision is made by whoever writes the recovery time into the control specification.

The basis already treats recharge as a capacity event on the generation side. Recharge is designated a mandatory lockout item during generator operation. Without that lockout, generator-side demand rises from 14,125 to 14,358 kW and the N+1 margin falls from 1,625 kW [MODELED] to 1,392 kW [MODELED]. In the least favourable modelling column the lockout is what stops the unit count turning over a second time.

Limits and open items

Confirmed: the two remainders are 875 kW steady and 642 kW charging-inclusive, they differ by exactly the recharge component, and they may not be subtracted from one another. [FACT]

Modelled: the 233 kW and 581.8 kW recharge figures, the 14,125 and 14,358 kW facility figures, the 1,625 and 1,392 kW generator margins, the 609 kW remainder, the 1,754 kW fifty-minute custom, and the 3.5 % / 70 kW suggested cap. The 0.92 [MODELED] charger efficiency is explicitly an assumption with no supplier documentation behind it, and the basis states that every one of these figures must be recomputed when a nameplate arrives. The 2,379 kWh battery energy is itself an unfrozen five-minute floor. The 0.95 efficiency inside the fifty-minute table is a different assumption from the 0.92 in the ten-hour table. [MODELED]

Derived here and not in any source: the 2,327 kW one-hour case, the 91 % and 36 % remainder-consumption fractions. [SENSITIVITY]

Held: the recovery time itself, the supplier efficiency curve, and the site-correct recomputation of the column these figures come from — the shared engineering register records that its weather chain was developed against the wrong location, leaving weather-dependent outputs as formula templates without site effect until that recomputation is complete. [HOLD]

Nothing here is an operating measurement, a supplier confirmation or a released capacity position.

Who Owns the Recovery-Time Setting

Charger efficiency running below the assumed value moves all of this the wrong way and moves it quietly. At 0.85 [SENSITIVITY] rather than 0.92, the ten-hour case rises from 233 to 252 kW [SENSITIVITY], the charging-inclusive peak moves from 14,358 to 14,377 kW [SENSITIVITY] and the remainder falls from 642 to 623 kW [SENSITIVITY] — small, but in the same direction as every other open item on this list.

The specification question is therefore about ownership rather than arithmetic. Recovery time will be written by whoever drafts the control specification. The number it consumes is owned by whoever answers to the service cap. Are those the same document?

© 2026 K&K Data Service Inc. All rights reserved. Reproduction or republication is permitted only with clear attribution to K&K Data Service Inc. and a working hyperlink to the canonical URL of this article. Excerpts must preserve the technical context, maturity labels, assumptions, and limitations. No excerpt may imply project approval, field validation, certification, or endorsement that the original article does not state.

Source: K&K Data Service Inc., “A 10× Recharge Rate Overruns the Last 4% of the 15 MW Service Cap,” https://www.kkdatasvc.com/lab/ups-battery-and-generation/recharge-at-10-x-blows-the-last-4-percent-of-the-service-cap/.