Summary

  • One Vertiv model designation, 440, prints 440 kW on its data sheet at 75°F return air and 50°F entering water, and 451 kW in its own user guide at 95°F dry bulb and 62°F entering water. In the text retrieved from those two documents, neither carries the sentence that would join them.
  • The small chilled-water manuals publish an entering-water correction table that moves net capacity by 23% to 26% across four degrees of water temperature. The large chilled-water catalogs carry the condition in the table heading and in extra rows instead, and Vertiv's own product manual calls the type number indicative.

The number on the sheet

The figure that reaches a facility schedule is usually the one on the short sheet. For the Liebert CW440 it is 440 kW, printed with its condition note attached in parentheses: 75°F return air temperature, 50°F entering water, 62°F leaving water. A second line on the same sheet offers up to 750 kW at higher return air temperatures. That is a headline with a footnote, and the footnote is doing real work (CW440 data sheet SL-18063).

Open the same product's user guide and system design and the net capacity table prints a different construction of the same machine. The heading reads: based on 62°F entering water, a 12°F water rise, 95°F dry bulb, 70.7°F wet bulb, 30% relative humidity. Under that heading, model 440 lists 1,540 kBTUH, which the table also prints as 451 kW, total and sensible both, with 275 GPM and 18 ft of pressure drop (CW440 User Guide and System Design).

Both documents are the manufacturer's. Both describe the same designation. Both attach conditions. They do not attach the same conditions, and in the text retrieved from those two files neither prints a sentence that reconciles the figures.

The size of the gap is not the finding. Eleven kilowatts separate them, and that is my division of two printed numbers, not the manufacturer's. What matters is where each number sits. The digits in the model name match the data sheet's headline, not the performance table. And the two rest on different water: 50°F entering on the sheet against 62°F entering in the guide, with the same 12°F rise in both, so the sheet's 62°F leaving water corresponds to a guide table whose leaving water would be 74°F. The room moves too, from a 75°F return air to a 95°F dry bulb at 30% relative humidity.

That is what a condition note is for. A machine delivers a different quantity of cooling into different water in a different room, and a document printing one quantity has to say which. The difficulty is administrative rather than thermodynamic. A reader holding only the short sheet holds a number whose basis lives in another document.

How the condition is carried

The design catalogs show the normal method, and it is repetition rather than arithmetic. The Liebert CW System Design Catalog for the 38 to 181 kW range prints its heading flatly: net capacity data based on 45°F entering water, a 10°F water rise, 75°F dry bulb, 61°F wet bulb, 52°F dew point, 44% relative humidity. The same rows then appear again at 80°F dry bulb and 85°F dry bulb with their own wet-bulb and humidity values (Liebert CW System Design Catalog, 38 to 181 kW).

A warmer room therefore gets a different row, not a multiplier. The catalog also states that fan motor heat is factored into the net capacity ratings, that capacity data is factory certified within 5% tolerance, and that the data is rated with the standard filter. The larger book repeats the pattern at 305, 375 and 415 kW, with Table 2.1 based on 45°F entering water and a 10°F rise, rows at 75, 80 and 85°F dry bulb, and separate tables for a 12°F rise and for 36,500 and 42,000 ACFM air volumes (Liebert CW Thermal Management System catalog, 305 to 415 kW).

This is an ordinary way to publish performance, and it is why the heading is not boilerplate. On these documents the heading is the only thing that says which printed number belongs to which room. An engineer who carries the 75°F row into a 95°F hall has not been misled by the catalog. The catalog simply never printed the row for that hall, and the interpolation is the reader's arithmetic.

Where a correction table exists

The Liebert DataMate System Design Catalog, covering the 1.5 to 3 ton class, does what the large chilled-water books do not. It prints net capacity for the DME044C on a stated basis of 45°F entering water and a 10°F rise at three room conditions, then prints Table 4.6, chilled water capacity correction factors based on a 10°F water rise, running across entering-water temperatures from 42°F to 49°F and normalised to 1.00 at 45°F (Liebert DataMate System Design Catalog).

Those factors are not decorative. At 49°F entering water the total and sensible factors are 0.74 and 0.83 at 72°F and 50% relative humidity, and 0.77 and 0.85 at 75°F and 45% relative humidity. At 42°F the printed factors run upwards instead, to 1.27 and 1.14 and to 1.23 and 1.11. Converting those into percentages is my arithmetic: 0.74 is a 26% reduction against the 45°F basis and 0.77 is 23%, while the sensible factors of 0.83 and 0.85 are reductions of 17% and 15%.

Four degrees of entering-water temperature, and a quarter of the printed capacity. That is the order of magnitude at stake when a condition moves, which is why the presence or absence of a correction table is a documentary fact and not a formatting preference.

The legacy Mini-Mate2 books follow the same structure and add a sentence the catalogs omit. The 2 and 3 ton manual states net capacity on 45°F entering water and a 10°F rise, tabulates at 80, 75 and 72°F dry bulb, prints a correction table by entering-water temperature from 42°F to 49°F normalised to 1.00 at 45°F, and then states that 75°F and 45% relative humidity is the standard rating condition for ASHRAE 127-2007 and that all capacities are nominal with actual performance within plus or minus 5% (Liebert Mini-Mate2 capacity technical data manual, 2 and 3 ton). The 8-ton book is the same document with a larger coil, Table 11 giving 0.74 and 0.79 at 72°F and 0.77 and 0.83 at 75°F for 49°F entering water, with the same standard-condition and tolerance notes (Liebert Mini-Mate2 8-ton capacity technical data manual). Both are 2010 files on a legacy server, so they are evidence about how this document class worked rather than proof of what the current book says.

The structural point survives the date. Where the correction table exists, the condition travels inside the document and a reader can move a printed capacity to different water with the manufacturer's own numbers. Where it does not exist in the text retrieved, the condition travels only in the heading and the alternative rows, and the reader moves the number on their own authority.

What the type number means

The most useful sentence about the model name is not in the chilled-water books. It is in the product documentation for the Liebert PDX perimeter system, revision C.01 dated 16 August 2024 and hosted on a distributor portal. That document states that the three-digit identifier tends to identify the net total cooling capacity in kW at standard conditions of 30°C room air, 35% relative humidity, 20 Pa external static pressure and 45°C condensing temperature with compressors and fans at full speed. Then it says what the type numbers elsewhere leave implicit: the correspondence is only indicative, and actual performance requires the selection tools (Liebert PDX product documentation, rev. C.01).

The same document is also the one place in this set where altitude is handled, and its handling is narrower than it looks. Below 1,000 m no derating is required. Between 1,000 and 2,000 m the rule is to decrease output current by 1% per 100 m, or alternatively reduce the maximum ambient temperature by 1°C per 200 m. Above 2,000 m the customer contacts technical support. That is an inverter and compressor derating rule about current and ambient temperature, not a tabulated net-cooling-capacity factor.

My arithmetic on the printed rate gives a 10% current reduction at 2,000 m or a 5°C ambient reduction, and the document converts neither into kilowatts of cooling.

The standard is a comparison instrument

The condition basis points back to a standard that is unusually clear about its own purpose. The ANSI/ASHRAE 127-2007 preview records that the primary rating point, Full Cooling Test A of Table 1, was redefined at 23.9°C and 45% relative humidity against the prior 22°C and 50% (ANSI/ASHRAE Standard 127-2007 preview). A technical account of the 2012 edition records that 2007 used a single return-air rating point of 75°F and 45% relative humidity, that 2012 introduced four application classes running from 75°F for Class 1 to 105°F for Class 4 all at a 52°F dew point, and that the standard separates units using compressors and refrigerant coils from units using chilled-water coils (Testing CRAC and CRAH: Standard 127-2012).

Addendum b to the 2020 edition states the purpose in one sentence: the intent is to enable products to be compared for steady-state capacity and efficiency based on a defined set of standard performance points. Then comes the limit — the document does not define a method of testing equipment for transient response, project-specific applications, control stability or similar parameters, and alternative tests are to be called application tests and not confused with, or inferred to be, tests complying with the standard (ANSI/ASHRAE Addendum b to Standard 127-2020). The scope of the current edition is the same object: uniform test requirements for rating air-conditioning units applied in data-centre and other information technology equipment spaces, by an air enthalpy method across at least one heat exchanger (ANSI/ASHRAE 127-2020).

Put together, the rating is exactly what a comparison needs and exactly not what a room needs. A unit rated at a defined point can be compared with another unit rated at the same point. It cannot be dropped into a hall at another return temperature and another entering-water temperature and expected to deliver the point. The standard says that second calculation is out of scope, and that is the calculation the facility performs.

The other thing called capacity

The same company's paperwork carries a second sense of capacity, and it is a different kind of quantity. Vertiv's fourth-quarter 2025 release, published 11 February 2026, reports net sales of $2,880 million up 23%, organic orders growth of approximately 252% year over year and 117% sequentially, a book-to-bill ratio of approximately 2.9 times, and backlog of $15.0 billion at year end up 109% (Vertiv fourth-quarter 2025 results release). The company furnished that release as Exhibit 99.1 to a Form 8-K filed the same day, which makes it a furnished statement rather than a filed financial statement (Vertiv Holdings Co Form 8-K, 11 February 2026).

On the call, management said the company would stop publishing the quarterly orders series, because the best approach is no longer to report actual orders, orders forecast or backlog with quarterly earnings. The same call described book-to-bill as almost 3 times, backlog as more than double the prior year and up 57% sequentially, and everything in backlog as a binding purchase order, often with advance payment (Vertiv Q4 2025 earnings call transcript).

The filed annual document is more careful, and its care is the interesting part. The Form 10-K for the year ended 31 December 2025 reports estimated combined backlog of $15.0 billion and $7.2 billion at the two year-ends, states that the majority is considered firm and expected to ship within 12 to 18 months, defines backlog as product and services for which a customer purchase order or purchase commitment has been received and not yet delivered, and states that orders may be subject to cancellation or rescheduling. It carries no quarterly book-to-bill and no quarterly orders figure (Vertiv Holdings Co Form 10-K, year ended 31 December 2025).

Two arithmetic notes belong to me and not to Vertiv. Book-to-bill of approximately 2.9 times applied to quarterly net sales of $2,880 million implies orders of roughly $8.35 billion, a figure the release never prints. And $15.0 billion against $7.2 billion is an increase of $7.8 billion, or 108.3%, against a stated 109%, the difference being rounding.

The reason to set these paragraphs beside the data sheets is not to suggest either number is wrong. Capacity in an order book is a contractual quantity with a shipment window and a cancellation clause; capacity on a data sheet is a thermal quantity at a stated condition. Neither measures cooling delivered into a room.

What a reader has to hold

Reproducing the number a facility was designed for takes three things at once: the heading that states the condition basis, the row belonging to the site's room, and, where water or altitude differs, a correction factor from a document that prints one. In this set the first two are present for the chilled-water catalogs, the third is present for the small units and absent from the retrieved text of the large ones, and the paragraph authorising an interpolation is not there.

The manufacturer has a mechanism for the missing step: the selection tools its own manual names. That mechanism is not a public document. Whether the reconciliation exists inside it, and whether a customer can obtain the output with the equipment order, is a question for a project rather than a datasheet review.

The documentary finding stands on its own. One designation prints 440 kW and 451 kW inside the same manufacturer's folder under two condition notes, and neither retrieved text joins them. That is not a scandal; it is the ordinary condition of equipment documentation, and it is why the table heading deserves more attention than the model name. The name is a label. The heading is the claim. The entity record behind this coverage is Vertiv in the directory.

Reading limits, where they bite: publication revision or date is not shown in the retrieved text of the 38 to 181 kW catalog, the CW440 user guide, the DataMate catalog or the large CW catalog. Retrieved PDF text was partial for several files, so the absence of an altitude table describes what was retrieved. The full Table 1 conditions of the 2020 standard were not obtained. The PDX manual is distributor-hosted and the Mini-Mate2 manuals are 2010 legacy copies. The transcript is third-party and its wording is used as reported. Every cooling figure here is a manufacturer-printed rating; nothing measures an installed unit.