Steady Hydraulic Models Miss DLC Interlock, Inertia, and Maintainability

The hydraulic basis spends six chapters on steady heat balance and steady hydraulics, then numbers three items it says that method structurally cannot expose. A Darcy table does not return H-10, the leak interlock logic and who holds the authority to stop; H-11, the thermal inertia of the loop after flow or rejection is lost; or H-12, the service-level definition of how many machines may be offline at once. They belong to the control, transient and contract domains respectively, and the document names that as the reason all three went un-numbered until chapter seven.

Leak Interlock Does Not Appear in a Friction Table

Each of the three was already present in the document as prose before it had a number. The leak response branch sits in section 4.4; the maintenance-overlap question sits in section 5.5. What was missing was not the observation but a register entry with an owner, a closure condition and a gate, because a steady model produces no residual that points at any of them.

H-10 is the one that looks like a valve schedule and is not. Auto-isolating a span is the ring-segment case: the isolation table and conclusion 24 give zero cabinets, because flow detours. About 31 cabinets [MODELED] is the whole-zone-ring case in conclusion 21, a different event. A false drip that auto-closes a single cabinet branch can take a 130 kW [MODELED] rack off cooling in tens of seconds. None of those events appears as a residual in a friction table. The owner has not chosen between auto-close and alarm-only. The facility must not default.

H-06 and H-12 Both Raise Unit Count in the Same Direction

Conclusion 46 shows what that costs when two of them interact. H-06, the effective capacity of a unit if its nameplate was set on pure water, and H-12 both act on the same conclusion — how many units are needed — and they act in the same direction. If a maintenance window must still hold single-fault redundancy, the required spare count rises from one to two; if the nameplate is optimistic, effective capacity per unit falls to 1.20–1.28 MW against a 1.5 MW nameplate [MODELED]. Together the shortfall stops being marginal. Unit count is a defining parameter of the enquiry package, so it may not be handled by pricing one count and adjusting later, because added price and delivery are not bound by the original quotation.

H-12 is the only non-technical item that can independently overturn the unit-count conclusion. The eleven-unit and twelve-unit study branches are both HELD. If planned maintenance must still survive a further fail, both fall, and the step is N+2 at the order of three additional units.

Unclosed, the Three Items Make RFQ Bids Incomparable

The blocker register sorts its items by what they can change. Four sit in the investment-scope layer, meaning each can alter equipment count or system composition, and three of the four are the chapter-seven additions — thermal inertia, the service level definition, and the liquid fraction's companion item on whether mechanical cooling is needed at all. The stated consequence of letting them into a request for quotation unclosed is that the returned bids are not comparable in scope, which makes the comparison itself invalid rather than merely imprecise.

H-11 is the sharpest of the three because it can create a line item that does not exist yet. If loop ride-through is in seconds rather than minutes, stored cooling or an uninterruptibly-supplied secondary pump has to be added, and the heat-rejection basis prices that path at 202 kW of additional protected load for about 7 kW of facility power — an electrical equipment quantity, not an efficiency figure [MODELED].

Sixteen Digits of Agreement Cannot See a Valve Close

A vendor distribution-unit packet was independently recomputed on 864 ring segments. Pressure-drop differences sat at the order of 10⁻¹⁶ kPa [MODELED]. The reviewer states what that proves: arithmetic self-consistency under a fixed topology, an imposed equal-flow assumption and assumed local-loss coefficients. Sixteen digits of agreement about a steady snapshot say nothing about a valve closing. The packet is not this plant's zone ring. Dynamic performance in that review remains unclosed. Still owed are maximum flow interruption, temperature and pressure excursion, valve and pump response, and the representative operating range, with type, unit and site tests not yet separated.

The same review says the same thing in its own words: two steady operating points cannot prove switching without flow interruption or water-hammer. It accepts a wording correction that had over-claimed from a steady-state chart, and then records that dynamic performance itself remains open.

Transients Can Price H-11 and Still Cannot Close It

The heat-rejection basis supplies the counterpart. Its conclusion B-42 finds that whether the primary pump is protected changes facility power by only about 7 kW but protected load by 202 kW, and states the separation that follows: the unclosed thermal inertia item does not block the efficiency conclusion but does block freezing the uninterruptible supply and battery, so the two can advance independently. The same document records the controlled judgement it inherits — battery minutes cannot substitute for liquid-cooling thermal inertia, and thermal transients and the joint integrated test are release conditions.

That is the reciprocal of conclusion 47. Steady hydraulics cannot see the three items; the transient chapter can quantify one of them and still cannot close it, because the closing number belongs to the equipment supplier.

Limits and open items

Confirmed on the current record: H-10, H-11 and H-12 are not hydraulic problems; the first six chapters' method cannot expose them; they were un-numbered until the register; H-06 and H-12 stack on unit count in the same direction. [FACT]

Modelled and not frozen: 1.20–1.28 MW against a 1.5 MW nameplate, spares rising from one to two, 202 kW / 7 kW, 31 cabinets, 130 kW, and the 864-row residual. All three items are open. [MODELED]

This article reports a method limitation stated by a document about its own work rather than an error found in it. Nothing above reflects Engineer of Record review, authority approval, supplier confirmation, commissioning acceptance or observed operation.

One condition would narrow the finding sharply. If the service level item closes in the permissive direction — planned maintenance need not preserve single-fault redundancy — then H-12 stops being able to move equipment count, and the set above reduces from three items to two, of which only one carries an unbudgeted increment. The count in the title is a property of the current register, not of direct liquid cooling, and a reviewer should read it that way.

© 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., “Steady Hydraulic Models Miss DLC Interlock, Inertia, and Maintainability,” https://www.kkdatasvc.com/lab/hydraulics-and-transients/steady-hydraulics-cannot-see-the-three-things-that-break-dlc/.