Electrical Ride-Through and Thermal Ride-Through Use Different Clocks

A secondary pump trips. Flow at the plate stops. The uninterruptible bus is still healthy. The plate starts a 5-to-15-second clock [MODELED]. Plot that event on an electrical ride-through axis and it does not appear. Conclusion B-44 of the heat-rejection basis says the two clocks do not overlap, and then says something worse about them. Battery minutes hold up the uninterruptible bus; they do nothing about the fact that primary-side rejection fans, adiabatic auxiliaries and mechanical cooling all sit on the unprotected side and stop the instant utility supply is lost, staying stopped until the generator takes load. During that window the secondary side is still energised and still dumping full duty into a primary loop that is rejecting nothing. Adding battery lengthens the interval over which the secondary side keeps pushing heat, so a larger battery makes the primary side heat up faster.

Electrical and Thermal Clocks Carry Different Numbers

Generator start plus transfer is taken as 10 to 30 seconds, graded an assumption because the on-site generation scheme is not frozen [MODELED]. Against that, conclusion B-58 puts system ride-through at 187 seconds when every pump keeps running, and 59 seconds when the primary circulation pump is not on the uninterruptible supply and the primary loop drops out of the thermal coupling [MODELED]. The third case, flow lost at the cold plate itself, is 5 to 15 seconds [MODELED].

Both numbers reproduce from the stated inputs. Take 35 m³ of secondary fluid at 3,968 kJ per cubic metre per kelvin, allow 5 K, and divide by 11,785 kW: 694,400 ÷ 11,785 = 58.9 s. Take the combined system capacitance of 472,000 kJ/K, the same 5 K and 12,600 kW: 2,360,000 ÷ 12,600 = 187 s [MODELED]. Conclusion B-59 states the ratio plainly — 3.1 minutes against 59 seconds against 10 seconds, roughly twenty-fold and six-fold — and attributes it not to how much fluid is stored but to which pumps are still turning.

Sequence S-3 is the unit-supplier clock. A secondary-pump trip starts the plate's 5-to-15-second band [MODELED]; ten seconds sits inside that band, not beyond it. The allowed changeover time is the interval to put the standby pump in or to re-allocate flow around the ring. If that interval exceeds about 10 seconds [MODELED], isolation is hydraulically true and thermally false: the detour exists on the velocity screen, and the plate clock does not wait for it.

Cooling Needs Fan Re-Energise, Not Generator Start

Conclusion B-64 is the crossing point. What the cooling side needs is not when the generator starts but when the rejection fan motor terminals are re-energised, which is detection plus start plus the load step that actually carries the fans. Step loading lengthens that: to protect the generator from voltage and frequency dip, mechanical loads are conventionally admitted last. Correct electrical design therefore extends the cooling side's loss-of-rejection window, and the basis records that this coupling had not been identified by either discipline.

Conclusion B-61 makes the acceptance criterion longer still. The test is not ride-through greater than generator time, but ride-through greater than generator time plus fan ramp plus compressor anti-short-cycle delay, and where a mechanical cooling stage exists that delay is typically 5 to 10 minutes — potentially an order of magnitude larger than the generator interval, in which case the stored-cooling requirement is set by a compressor protection timer rather than by the generator [MODELED].

A just-feasible rejection route at the study weather point leaves zero allowable rise on the primary. Zero rise is zero time on that level. The temperature axis can be a line of zero length before the electrical axis has started.

A 2.4 K/min Rise Cannot Be Logged at Minute Intervals

Conclusion B-69 sets the instrumentation floor. Primary-loop temperature rises at about 2.4 K per minute once rejection stops — 12,600 kW divided by 80 m³ times 3,968 kJ/m³·K gives 0.0397 K/s [MODELED] — and the cold-plate event is a few seconds long. A management system trending at one to five minute intervals captures zero or one sample across a ten-second event, so the temperature axis cannot be plotted at the resolution the electrical axis is specified to.

Any figure that puts seconds on one axis and kelvins on the other is therefore also a statement about logging rate. Whoever specifies the point list decides whether the crossing can be observed.

Fourteen Joint Scenes Sit in the Index; Zero Are Executable

The commissioning evidence master index carries the cooling-side time quantities as item XI-08 and records all six as not obtained, listing the 187-second and 59-second values explicitly as conditional branch values rather than design data. Its item XI-06 records the whole outage-to-restoration state machine as structurally complete with every time quantity missing, and XI-07 records the generator transient studies and their joint measured items as not started.

The index also states the rule that decides what an eventual test is worth: acceptance criteria flow backwards down the evidence chain, so if the engineering studies arrive later than the equipment, the integrated test runs without a pass criterion and its result can only be recorded, not judged. Fourteen joint scenes currently sit in that index [FACT]. The number presently executable is zero [FACT]. That is a design-input fact, not a commissioning delay. That converts the two-axis figure from a design illustration into a scheduling constraint.

Limits and open items

Confirmed on the current record: the electrical interval and the temperature interval are different clocks; adding battery minutes does not lengthen thermal ride-through; a just-feasible rejection route at the study weather point has t_L4 = 0; fourteen joint scenes currently sit in the commissioning index and the number presently executable is zero. [FACT]

Modelled and not frozen: 187 s and 59 s as conditional branch values, 10 to 30 seconds of start-plus-transfer, 35 m³ / 80 m³ / 5 K compositions, 2.4 K per minute, 5 to 10 minutes of purge on a mechanical cooling stage only, and 202 kW of protected load against about 7 kW of facility power and roughly 50 m³ of stored cooling. The ride-through figures rest on a secondary fill volume estimated at 25 to 45 m³ with ±40% uncertainty, on an allowable supply temperature rise that is the supplier's unobtained number, and on a primary volume that is not fixed. [MODELED]

Recommendation R-B21's case for putting the primary pump on the uninterruptible supply is a comparison, not a decision. One condition changes the shape of the figure rather than its values. If the allowable supply temperature rise returns at the low end, the temperature axis shortens until the 59-second branch falls inside the generator interval itself, and the question stops being how much margin exists and becomes whether any unprotected primary equipment is admissible. Nothing above reflects Engineer of Record review, supplier confirmation, commissioning acceptance or observed operation, and no protected-load figure here is a procurement release.

© 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., “Electrical Ride-Through and Thermal Ride-Through Use Different Clocks,” https://www.kkdatasvc.com/lab/hydraulics-and-transients/two-time-axes-electrical-ride-through-vs-temperature/.