Technical 7th September 2026

Why Is Liquid Cooling Turning Manual Ball Valves Into Actuated Ones?

Q: Why does breakaway torque matter more than seated torque?

Seated torque is published for a clean valve at ambient temperature that has recently been cycled. A cooling loop valve often sits in one position for weeks at loop temperature in glycol, and the seat takes a compression set against the ball. The torque to start it moving is then higher than the published figure.

Q: Is PTFE acceptable as a seat material on an actuated valve?

Virgin PTFE seals well but deforms under sustained compressive load, which raises breakaway torque after standing closed and accelerates wear on a repeated cycle path. Glass-filled RPTFE at 15 or 25 percent is the usual first step. PEEK goes further again, at the cost of higher operating torque.

Electric quarter-turn actuator lifted away from a three-piece stainless steel ball valve, exposing the machined ISO 5211 mounting pad, its bolt circle and the stem drive

Key Takeaway

Liquid cooling does not simply add valves. It moves them out of the plant room and into the rack row, where four conditions make a lever impractical: no operator is present in the aisle, a coolant leak beside energised electronics has to be isolated in seconds, redundant paths have to be switched on command, and fill and flush sequences run to a schedule. Each condition converts a valve that would have carried a handle into a valve that carries an actuator and a signal wire. That conversion changes the body, not just the top. An actuated valve needs an integral ISO 5211 pad rather than a bracket, a stem sized for actuator torque and side load, a seat that survives thousands of cycles instead of dozens, and live-loaded packing. A body specified for manual duty will pass its shell test and still fail in service, because the failure shows up as stem leakage and rising breakaway torque after a few thousand cycles, which no factory acceptance test looks for.

Where the New Valves Actually Come From

In an air-cooled hall the water stops at the CRAH coil. Isolation valves live in the chilled water plant and in the riser. A technician reaches them on foot, and their population is set by the number of air handlers.

Direct-to-chip cooling breaks that arrangement in two. A coolant distribution unit sits between the facility water system and a separate technology cooling system, and the secondary side carries treated fluid out to manifolds, down each rack, and into cold plates sitting on the processors. Every one of those branch points is a place where a technician must eventually be able to shut off flow without draining the row.

The count follows from topology rather than from any published ratio. A CDU needs isolation on both sides of both loops. A manifold pair needs isolation at the top and the bottom. A rack branch needs isolation if a single rack is ever to be serviced while its neighbours keep running. Published per-rack valve counts vary widely between designs, and a rear-door heat exchanger retrofit produces a different figure again, so the honest statement is that valve population scales with the number of independently serviceable branches, and liquid cooling multiplies those branches.

Valve population grows to the right of the CDU. The facility side keeps a small number of large manual isolation points. The technology cooling side adds an isolation point at every branch that has to be serviced independently, and those are the points that cannot wait for a person to walk to them.

Why Those Particular Valves Cannot Stay Manual

Four constraints apply on the secondary side, and each one on its own is enough to rule out a lever.

Condition on the secondary side	Why a lever fails	What the valve needs instead
No operator in the aisle	The cost of operating the valve is the trip to it, not the valve	Remote command over the building network
Leak beside energised hardware	An alarm that ends in a phone call responds in minutes	Actuator wired to leak detection, with a defined fail position
N+1 CDU changeover	Redundancy becomes only as fast as the staffing	Actuator with position feedback to confirm the path
Fill, flush and commissioning	A valve outside the controls has to be worked around	Actuator on the control network, in the sequence

No operator is present in the aisle

A valve inside hot aisle containment is behind a door, in a restricted area, at a height that may need a platform. The cost of operating it is not the valve, it is the trip.

Leak response is measured in seconds

Coolant in a direct-to-chip loop runs within the rack envelope, above and beside energised hardware. Leak detection cable or a tray sensor can raise an alarm quickly, but an alarm that ends in a phone call has a response time measured in minutes. An actuator wired to the same alarm closes on the signal.

Redundant paths have to be switched, not walked to

Where CDUs are deployed N+1, one unit can be taken out of service without dropping the load it was carrying. The changeover is a valve operation. If it is manual, the redundancy is only as fast as the staffing.

Fill, flush and commissioning run to a sequence

A secondary loop is filled, vented, flushed and proven before it carries load, and the same sequence repeats after a component is replaced. Sequences of that kind are written into the controls, and a valve that is not in the controls has to be worked around.

What Changes in the Valve Body

This is the part that gets specified last and causes the most trouble. Bolting an actuator onto a valve designed for hand operation changes the duty in four measurable ways.

Property	Manual isolation valve	Actuated valve on a cooling loop
Operations over service life	Tens to a few hundred	Thousands, set by the control sequence rather than by people
Actuator interface	Lever and stop plate	ISO 5211 pad, sized F03 to F07 in the small line sizes
Stem loading	Hand torque, applied in line	Actuator output torque plus the side load of a cantilevered mass
Seat duty	Static sealing, occasional movement	Repeated wear track, creep under sustained load
Stem seal	Single packing set, adjusted by hand if it weeps	Live-loaded, so spring washers hold gland compression as the packing creeps
Failure that ends its life	Seat leakage	Stem leakage and rising breakaway torque

The mounting pad decides whether the valve is genuinely actuator-ready

ISO 5211 defines the attachment between a part-turn actuator and a valve: a bolt circle, a pilot diameter and a drive that is square or double-D. The flange sizes used on small line sizes are F03 at a 36 mm bolt circle, F04 at 42 mm, F05 at 50 mm and F07 at 70 mm, with F10 at 102 mm and F12 at 125 mm above them.

ISO 5211 flange	Bolt circle diameter
F03	36 mm
F04	42 mm
F05	50 mm
F07	70 mm
F10	102 mm
F12	125 mm

The full table, F03 to F100 with the spigot recess, the bolt thread and the flange heights, together with the hole positions and the designation letters, is in the ISO 5211 reference.

The bolt circle figures are the ISO 5211 values. Confirm the flange size and the drive dimension against the valve drawing before ordering an actuator.

A valve can meet that dimensionally in two ways. The pad can be cast and machined as part of the body or the bonnet, or it can be a bracket bolted on with a coupling to the stem. The bracket version is cheaper and it is what a catalogue often means by actuator-ready.

The difference shows up in lost motion. A bracket adds a joint and a coupling between the actuator output and the ball, and any clearance in that chain becomes angular play. The actuator reports ninety degrees of travel while the ball travels slightly less, and the stop position shifts with the direction of the previous stroke, so the seat contact band moves instead of repeating. It also transfers the actuator mass through the bracket into the stem as a bending load rather than into the body.

Seat material stops being a temperature decision

On a manual valve the seat is selected against media and temperature. Under cycling it is also selected against creep and wear.

Virgin PTFE deforms under sustained compressive load, so a valve that has stood closed takes a compression set against the ball and its breakaway torque rises above the value measured on a recently cycled valve. Filled grades reduce that. RPTFE with 15 or 25 percent glass raises compressive strength and wear resistance at the cost of some conformability, and PEEK raises both a great deal further, asking for more torque in return. Glycol content in a secondary loop pushes the decision as well, because the mixture raises viscosity and changes the wetting behaviour at the seat face.

Seat material	Creep under sustained load	Operating torque	Where it fits on a cooling loop
Virgin PTFE	Highest of the four	Lowest	Low operation counts, clean water, manual duty
RPTFE, 15% glass	Reduced	Moderate	General actuated duty, the usual first step up
RPTFE, 25% glass	Lower again	Higher	Higher seat load or higher cycle counts
PEEK	Lowest of the four	Highest	High cycle counts where the actuator has torque headroom

Cycle count is a specification, and there are standards that measure it

Valve standards mostly test pressure containment, not endurance. ASME B16.34 sets pressure-temperature ratings, API 598 and ISO 5208 define shell and seat testing, and none of them tells you what the valve does after five thousand operations.

The standards that do measure cycling come from fugitive emission testing. ISO 15848-1 defines endurance classes for isolating valves at 205, 1 500 and 2 500 mechanical cycles, tested with a measured leakage limit at the stem. API 641 type tests quarter-turn valves over 610 mechanical cycles with stem emissions held below 100 ppmv. Neither standard was written for cooling water, and neither is a cooling qualification. They are useful because they are the only widely recognised way to state a cycle count with a stem leakage limit attached, which is exactly the pair of numbers an actuated cooling valve is judged on.

Standard	What it measures	What it does not tell you
ASME B16.34	Pressure-temperature ratings by material and class	Anything about cycle life
API 598	Shell, backseat and seat leakage at manufacture	Anything about cycle life
ISO 5208	Seat leakage rate classes A to D	Anything about cycle life
ISO 5211	Actuator attachment dimensions and drive form	Torque, endurance or whether the pad is integral
ISO 15848-1	Stem leakage over endurance classes at 205, 1 500 and 2 500 cycles	Behaviour in water or glycol; the test fluid is helium or methane
API 641	Stem emissions below 100 ppmv over 610 mechanical cycles	Behaviour in water or glycol; the test fluid is methane

The Sizing Error Worth Correcting

Actuator sizing is commonly done by taking the seated torque from the valve catalogue and applying a safety factor. The figure in the catalogue is normally a clean, ambient, water-service number taken on a valve that has just been cycled.

The condition that decides whether the actuator is big enough is different. It is breakaway torque on a valve that has stood in one position at loop temperature, in a glycol mixture, for weeks. The seat has taken a compression set against the ball, and the torque needed to move it the first few degrees is higher than the catalogue value. An actuator chosen on the clean number and a modest factor can stall on a valve that has done nothing wrong.

Ask for the breakaway figure after a dwell period at temperature, in the actual fluid, and size against that. If a supplier can only give the ambient number, that is worth knowing before the order rather than after commissioning.

What Belongs on the Enquiry

An enquiry that carries these items can be quoted against. One that does not will be quoted as a manual valve with a pad on top.

Line size, pressure class and the fluid, including glycol type and concentration
Loop temperature range, and the temperature the valve sits at when it is not moving
Expected operations per year, and whether the valve holds one position for long periods
ISO 5211 flange size required, and whether the pad must be integral to the body
Stem drive form, square or double-D, and the dimension across flats
Breakaway torque after dwell at temperature, not only seated torque at ambient
Stem leakage limit, and whether an ISO 15848-1 or API 641 result is required
Whether the valve must be serviceable in line, which decides two-piece against three-piece
End connections, and whether the actuator envelope clears the adjacent rack

What This Does Not Settle

Electric and pneumatic actuation are not decided by the arguments above. Both meet all four constraints. The choice between them turns on whether compressed air is available at the rack and on what the valve must do when power or air is lost. It is covered separately in the actuation comparison linked below.

Valve counts per rack are also not settled here. They are a function of how many branches a given design makes independently serviceable, and that varies between direct-to-chip layouts, rear-door retrofits and immersion. A number taken from one reference design does not transfer to another.

Frequently Asked Questions

Does a valve need a different body to take an actuator, or only a bracket?

A bracket will mount an actuator and will pass a functional test. The difference appears over cycles. A bracket and coupling add clearance between the actuator output and the ball, which becomes angular play, and they carry the actuator mass into the stem as a bending load. A pad machined into the body or bonnet removes both. On a valve that will be operated a few times a year the bracket is not a problem. On one that runs to a control sequence it is worth specifying the integral pad.

How many cycles should an actuated cooling valve be specified for?

There is no cooling-specific figure in the pressure standards. The practical approach is to calculate operations per year from the control sequence, including commissioning and any leak-test routine, then specify a cycle count with a stem leakage limit using ISO 15848-1 endurance classes, which run at 205, 1 500 and 2 500 cycles, or an API 641 result at 610 cycles. Those tests use methane rather than water, so they qualify the stem seal design rather than the whole valve in service.

Why does breakaway torque matter more than seated torque?

Seated torque is normally published for a clean valve at ambient temperature that has recently been cycled. A valve on a cooling loop often sits in one position for weeks at loop temperature in a glycol mixture, and the seat takes a compression set against the ball during that time. The torque to start it moving is then higher than the published figure. Sizing an actuator on the published number with a modest safety factor can leave it unable to open a valve that is otherwise sound.

Is PTFE acceptable as a seat material on an actuated valve?

Virgin PTFE seals well and has low friction, but it deforms under sustained compressive load, which raises breakaway torque after a valve has stood closed and accelerates wear on a repeated cycle path. Glass-filled RPTFE at 15 or 25 percent raises compressive strength and wear resistance and is the usual first step. PEEK goes further again, at the cost of higher operating torque and less conformability, which matters where the actuator is already close to its limit.

Does this apply to rear-door heat exchangers as well as direct-to-chip?

The access and redundancy arguments apply to both. The leak response argument is weaker on a rear-door unit, because the coolant is at the back of the rack in a door rather than routed to cold plates on the processors, so the consequence of a leak is different. Rear-door retrofits also tend to have fewer independently serviceable branches, so they add fewer valves per rack than a direct-to-chip layout.

Published by LINS Valve Industrial Co., Ltd., Taichung, Taiwan. Last Updated: 2026-09-07

Referenced standards: ISO 5211 (part-turn actuator attachments), ISO 15848-1 (fugitive emission type testing and endurance classes), API 641 (type testing of quarter-turn valves for fugitive emissions), ASME B16.34 (pressure-temperature ratings), API 598 and ISO 5208 (valve inspection, testing and leakage rates), API 608 and ISO 17292 (metal ball valves, blowout-proof stem requirement).

Liquid cooling loop terminology follows ASHRAE TC 9.9 datacom guidance on facility water and technology cooling systems, and Open Compute Project cooling environments material. Bolt circle dimensions are the nominal ISO 5211 values and are not a substitute for a specific actuator or valve drawing. Seat material behaviour is stated from published polymer property data and is not a service life prediction.