A controlled revision of the hydraulic basis added a three-row table and changed no flow, no pipe size, and no pump power. The table splits the symbol f into three entries it declares non-substitutable. F-CALC is the hydraulic representative value of 0.95 [MODELED], the single figure fed into every flow, pipe size, pressure drop and pump power number in that document. F-SENS is the sensitivity steps 0.70, 0.80, 0.90, 0.95 and 1.00 [SENSITIVITY], whose job is to expose where conclusions flip. F-OEM is the rack-level written split of liquid-side and air-side heat, and it has not been obtained [HOLD]. Neither 0.95 nor 0.90 is a frozen input. Neither is this project's liquid fraction [FACT].
The f Table Is a Disclosure Control, Not a Recalculation
The table exists because of how the previous revision read, not because a number in it was wrong. In that revision, 0.95 appeared in the flow chapter as a design value and 0.90 appeared in the blocker register as the current assumption. Neither location declared how the two differed in maturity. A reader taking either as settled would not have been contradicted by the document.
The separation is therefore a disclosure control, not a calculation change. The revision says so directly: it alters no number and removes only the maturity ambiguity. That is an unusual thing for an engineering revision to admit. It is the reason this entry is a records problem rather than a thermodynamics problem.
Until F-OEM Arrives, 0.95 and 0.90 Are Not Determined
What the third entry costs is visible in the spread the second one produces. Applied to 12,000 kW [TARGET] of billable information-technology load, liquid-side duty across 0.70 to 0.95 runs 8,400 to 11,400 kW, a span of 3,000 kW equal to about 26 % of the liquid system's capacity [SENSITIVITY]. With secondary pump heat added, the secondary-loop duty runs 8,785 kW at 0.70 to 12,385 kW at 1.00, and the representative value of 0.95 gives 11,785 kW [MODELED].
The hydraulic author ranks that single unclosed input as the first blocker of the phase, above fluid, temperature difference and pipe size. It alone decides whether the unit count can fall and whether the air side needs a complete independent terminal system. The instruction attached to it is explicit: until the written split arrives, no downstream document, request for quotation, investment model or external material may cite 0.95 or 0.90 as a determined liquid fraction [FACT].
Baseline 0.95 and Conservative 0.90 Are Different Calibres
The heat-rejection basis then uses both. Its baseline column runs at 0.95 and its conservative column at 0.90 [SENSITIVITY]. The conservative column is constructed as a set of mutually compatible unfavourable premises rather than a worst-case stack: the temperature branch in the middle, the liquid fraction at 0.90, the pump allowance left unreduced, and supplier efficiencies at the low end of guarantee.
That is a legitimate use of the sensitivity steps. It is also exactly the situation the separation table was written to survive. Two columns of one schedule carry two values of one symbol, and the schedule is readable only because the columns are labelled. Strip the column headers — which is what happens when a single figure is quoted onward — and the two values become indistinguishable from a disagreement.
f Blocks Secondary Hydraulics, Not Total Rejection Sizing
The two documents rank the same unclosed input in opposite positions. That looks like a contradiction and is not.
On the secondary side, duty rises with the liquid fraction, so unit count and terminal strategy move with it. On the primary side, total rejection is nearly insensitive to it: heat the liquid loop does not take, the air side takes, and both end up in the atmosphere. The liquid fraction therefore governs how rejection is divided between liquid and air terminals, not how much there is in total. That is why the heat-rejection document can advance ahead of the closure while the hydraulic document cannot.
The same heat-rejection record then ranks the item again, this time against a facility-load miss rather than against total rejection. That miss is 1,686 kW [MODELED] against the 14,400 kW facility-load target [TARGET]. Moving f from 0.95 to 0.90 accounts for 70 kW, or 4.2 % of that miss [MODELED]. The supplier temperature branch accounts for 750 kW, or 44.5 % of the same miss [MODELED]. Those shares are of facility power, not of annual energy.
The same record ranks f as a blocker for annual energy on separate terms, without those figures. The fraction does not change total rejection. It changes the temperature grade of what is rejected, and with it the free-cooling hours. The ranking is by use: not a blocker for rejection-plant sizing, a blocker for annual energy, and a first blocker for secondary hydraulics. It is still not a frozen liquid fraction.
Limits and open items
Confirmed as a records state: three calibres, mutually non-substitutable, with the third unobtained; neither 0.95 nor 0.90 may be cited as determined. [FACT]
Modelled and not frozen: the 8,400 kW to 11,400 kW liquid-side span, the 3,000 kW and 26 % figures, the 8,785 kW to 12,385 kW secondary-duty range, the 11,785 kW representative duty, and the 70 kW / 4.2 % against 750 kW / 44.5 % attribution of the 1,686 kW facility-load miss. All are arithmetic on assumed fractions. They inherit an assumed pump power and an assumed heat-to-fluid conversion. [MODELED]
Open: the rack-level written allocation table, per model and configuration, owed by the information-technology equipment vendor. Until it exists, no calibre of the symbol is a project fact. [HOLD]
Nothing here reflects Engineer of Record review, equipment-supplier confirmation, commissioning acceptance or observed plant behaviour. No fraction is released as a purchase input.
Carry the calibre name
Until that written split exists, the correct citation practice is to carry the calibre name with the value. A reviewer who sees f = 0.95 in a downstream document with no calibre attached should treat it as unattributed. The source document has already declared that the bare symbol does not identify which of the three is meant.
© 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., “The Liquid Fraction f Mixes Design, Sensitivity, and OEM Values,” https://www.kkdatasvc.com/lab/direct-liquid-cooling-and-cdu/liquid-fraction-f-is-three-different-numbers/.
Request the Public Calculation Note. If your project record shows a different result, or the same failure domain under another name, we want to see it. Email inquiry@kkdatasvc.com or use the contact page.