A Water-Rated 1.5 MW CDU Is 1.20–1.28 MW on PG25

The hydraulic basis calls this its most important item in the fluid chapter, and states it as an absence rather than a finding: no document on this project proves what fluid the 1.5 MW [MODELED] cooling-distribution-unit nameplate was calibrated on. If it was calibrated on pure or treated water and the secondary loop is charged with the assumed 25 % propylene glycol mixture, the arithmetic gives 1.5 × 0.80–0.85 = 1.20–1.28 MW of real capability [MODELED]. The band is not a tolerance. It is wide enough to move the unit count. A later parametric engine then treats the bottom of that band as a 1.2 MW [MODELED] working derate and still stamps every passing row as not engineering-accepted. The item is registered for closure before the request for quotation goes out.

The 1.5 MW Rating Is Held Without Its Four Conditions

The preceding section states why a single figure cannot carry the rating. Exchanger duty is Q = U·A·LMTD, so capacity is a function of primary inlet temperature, secondary supply temperature, secondary temperature difference and fluid — four variables. A one-kelvin rise in primary inlet temperature removes about one kelvin of available log-mean difference [MODELED].

That framing is what makes the fluid question answerable at all. It is not a claim that glycol is worse in general. It is a claim that a number quoted without its quadruple does not identify a capability, and the project holds the number without the quadruple.

A 31% Film-HTC Drop Becomes a 15–20% Capacity Drop

The derate is built in two steps rather than asserted. Against treated water, the glycol mixture costs about 32 % more pump power, about 26 % more pipe pressure drop, and about 31 % less in-tube heat transfer coefficient on the secondary side [SENSITIVITY]. Only the third feeds the exchanger.

A 31 % fall in one side's film coefficient does not translate one-for-one into overall conductance, because the other side is water and unaffected. The resulting fall in overall heat transfer coefficient is 15 % to 20 %, and therefore the fall in duty at the same log-mean difference is the same 15 % to 20 % [MODELED]. Subtracting that from unity and multiplying gives the 1.20 to 1.28 MW band directly. The step from film coefficient to overall coefficient is the one a reviewer should re-derive first, because it is where a correlation assumption does the work.

Derating Turns the 3+1 Zone's 7.3% Margin Into a Deficit

The consequence is stated against a specific group rather than in general. The larger zone is arranged three working plus one spare and holds 4,500 kW after a single unit failure. Derated, that becomes 3,600 kW to 3,825 kW against a zone load of 4,194 kW at a liquid fraction of 0.90 — a margin of 7.3 % turning into a deficit of 9 % to 14 %, at which point a single unit failure is no longer satisfied [MODELED].

That flip produces a procurement conclusion rather than a cooling one: this item alone invalidates the twelve-unit branch and forces the comparison to the fifteen-unit branch. A separate note in the blocker register adds that this item and the owner's parallel-maintainability requirement act on the same conclusion in the same direction. A project that resolves both unfavourably does not face a marginal gap. The three branches remain held in parallel. None is selected here.

The Parametric Engine Already Applies a 1.2 MW Derate

The parametric engine that expands thirty-six scenarios across branches, liquid fractions and temperature differences does not use 1.5 MW for capacity checking at all. It carries both figures per object and applies a 1.2 MW design derate to the redundancy criterion, with the stated rule that nameplate must not be passed off as available capacity [MODELED].

That is the bottom of the band, adopted as a working convention before the band's cause is closed. It is a defensible convention and it is also a trap for a later reader, because a derate applied in a machine-checked table looks like a resolved question. The engine labels its own positive-margin scenarios as not engineering-accepted, which is the safeguard. It holds the unit's capacity curve, flow and pressure drop under the same unclosed item.

Closure Needs a Capacity Curve, Not a Single Rating Point

A review of a supplier package in the read set shows what an insufficient answer looks like. The package's constant overall conductance figure is arithmetically the rated duty divided by the log-mean of the stated end differences — an algebraic restatement of the rating, not an independent result across operating points. Back-calculating the listed terminal temperatures and flows reproduces the rating to within 0.6 % on both sides, which demonstrates internal consistency and nothing about behaviour on another fluid [MODELED].

So the closure evidence has to be a curve, not a point. The register asks for the calibration fluid and calibration condition plus capacity as a function of the four variables above, owed by the unit supplier before the quotation package is frozen [HOLD].

One condition would change the conclusion entirely. If the nameplate turns out to have been calibrated on the glycol mixture in the first place, there is no derate, the larger zone keeps its 7.3 %, and the whole item closes at a stroke — which is precisely why the question is worth one letter rather than three more calculations. Nothing above reflects Engineer of Record review, equipment-supplier confirmation, commissioning acceptance or measured capacity. No unit count here is released for procurement.

© 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 Water-Rated 1.5 MW CDU Is 1.20–1.28 MW on PG25,” https://www.kkdatasvc.com/lab/direct-liquid-cooling-and-cdu/a-water-rated-1-5-mw-cdu-is-1-20-1-28-mw-on-pg25/.