Rack power went from 10–15 kW to 100 kW+.
We supply the loop itself: hose, couplings, seals, filtration, sensing rope and the instruments that read them, not the CDU and not the cold plate.
One business day, with the file attached
Three things moved together with the rack power
10–15 kW per rack in the hall as designed
100 kW+ per rack once the trays are accelerated
1 µm absolute the fluid still has to meet
A hall designed around 10–15 kW per rack, loaded to 100 kW+. The medium changed; the sealing problem did not.
A joint and its reading
Each rack adds its own pair. Sealing faces grow with rack count, not with floor area.
Pressure differential is the first reading that moves when a circuit stops delivering what the drawing says.
What the density moved
What we will put in writing for a rack at this density
| No | Condition | Value | Unit | Basis |
|---|---|---|---|---|
| 01 | Leak location resolutionLocating sensing rope | 0.1 | m | On 500 m of rope per channel, alarm inside 30 s |
| 02 | Cartridge absolute ratingFilter grade in circuit | 1 | µm | Grades from 0.05 to 100 µm on one housing |
| 03 | Cleanliness at chargeLoop fluid | 17/15/12 | ISO 4406 | Sampled and gated before the loop is charged |
| 04 | Filter change differentialAcross the housing | 0.5–0.7 | bar | With a transmitter on two impulse tubes |
| 05 | Shell design pressure304 and 316L | 1.0 | MPa | Hydrostatic proof per batch |
| 06 | Hose working pressureDASH −4 to −10 | 1.38 | MPa | About 200 psi, burst above 5.5 MPa |
| 07 | Hose temperature bandWater-based coolant | −40 to +99 | °C | The band claimed for this application |
| 08 | Coupling countDry-break pairs | One per rack | linear | Grows with rack count, not with floor area |
Where a tolerance is not published the cell stays empty rather than carrying an estimate.
Where those figures are produced
Hose is held to 1.38 MPa working and burst above 5.5 MPa before a batch leaves.
The cleanliness gate is a sample, not a claim: ISO 4406 17/15/12 before any charge.
The loop as one rack row is actually built
Consumables on this section
- 1Row manifold, dry-break coupling pair
- 2Rack drop hose, EPDM, 1.38 MPa
- 3Cold plate branch, seal face at the joint
- 4Filter on the return leg, 1 µm absolute
- 5Differential pressure transmitter, 0.5–0.7 bar
- 6Locating sensing rope, 0.1 m resolution
The positions on that section that wear
Every rack takes its own pair, so the fitting count tracks the rack count.
The rope runs where the coolant would arrive, not where the inspection route walks.
The failure this industry reports is at the face, not at the heat exchanger.
What the section says about this industry
The standards this configuration is written against
How those standards are demonstrated
Burst and hydrostatic work is done per batch, and the record travels with the shipment.
Five quantities on the loop: flow, pressure drop, conductivity, pH and level.
Configuration files, not case studies
Three configurations recur at this density. Each is held as a file: the loop drawing, the consumable list, the cleanliness gate and the sample that cleared it.
File A. Couplings at the cabinet, hose inside the rack line, isolation at the rack.
File B. One manifold per row, isolation at the head of the row, drops along the tray.
File C. The last branch to the accelerated trays: shortest run, most joints.
What every one of those files consumes
Parts shown are the supply-side samples held against the specification sheet: EPDM seals, a gap filler pad at 1.2–7.5 W/m·K, locating rope at 0.1 m resolution, and one of the five loop instruments.
The file behind this sheet
The file is a document set, not a brochure: drawings, certificates and the test record.
Every batch carries a record of what was measured, and what it measured at.
Four documents are held for this sheet. Tick the ones you need.
Tick a document, name the row or the configuration you are planning against, and the file comes back attached.
One business day, with the file attached.
Nothing else on this sheet is a commitment. Price, stock and lead time are not promised here.