Battery limits, controls, FAT and commissioning for packaged data-centre water plants

Mark Cullens

7 days ago

4 min read

Aerial view of a large data-centre campus under construction with cooling infrastructure and retention ponds

Direct answer
1. Establish physical battery limits
2. Define utilities and design conditions
3. Allocate instruments and control ownership
4. Write the sequence of operation
5. Design alarms, trips and safe states
6. Prepare a meaningful FAT
7. Plan installation and pre-commissioning
8. Prove site integration
9. Conduct performance and reliability tests
10. Complete handover and lifecycle records
11. How Crescent Engineering can support the project
12. References and further reading

Direct answer

A packaged water plant is complete only when its physical boundaries, utilities, controls, tests and handover evidence are unambiguous. Battery limits define where responsibility changes. The controls schedule defines how the package interacts with the data centre. FAT and site commissioning then prove the equipment and the complete installed system in different stages.

For AI data centres, these details protect programme and cooling availability. A skid can pass a factory demonstration yet fail on site because the feed pressure, drain capacity, communications, product-demand signal or residual route was never assigned.

1. Establish physical battery limits

Mark every process connection on the P&ID and layout: feed, product, recycle, overflow, backwash, concentrate, sludge, chemicals, vents and drains. State flange standard, size, rating, material, elevation and whether the mating connection is supplied.

Include access, lifting, maintenance withdrawal, bunding and hazardous-area boundaries. A connection list without physical coordinates or responsibility for final valves still leaves programme risk.

The productised modules guide provides the wider modular-delivery context.

2. Define utilities and design conditions

Schedule normal, minimum and maximum power, water, air, drainage, ventilation, communications and chemical requirements. Identify start-up peaks and simultaneous demands during cleaning or regeneration.

State environmental conditions, feed pressure and temperature, discharge backpressure, floor loading, noise and heat rejection. Define who supplies transformers, cabling, isolators, earthing and backup power.

3. Allocate instruments and control ownership

For each measurement and command, identify the instrument, owner, signal type, range, accuracy, location, calibration and destination. Distinguish package control, supervisory monitoring and site protection.

Resolve ownership of tank level, product demand, source selection, cooling status and residual capacity. Avoid two controllers acting on the same valve or neither controller owning a failover.

4. Write the sequence of operation

Describe start, normal operation, turndown, duty rotation, standby, cleaning, source change, off-spec diversion, controlled shutdown and restart after power loss. Include timing, permissives and operator actions.

Use cause-and-effect tables for critical sequences, but retain narrative context so intent is clear. Simulate future module additions and degraded communications.

5. Design alarms, trips and safe states

Classify alarms by consequence and required response. A quality warning may initiate sampling; a confirmed off-spec condition may divert product; loss of cooling make-up availability may require escalation to site operations.

Define fail position for valves and outputs on loss of power, air or communications. The safe state should protect people, cooling assets and the environment without creating an unnecessary data-centre trip.

6. Prepare a meaningful FAT

Factory acceptance should verify fabrication, materials, instruments, panel construction, software, interlocks, alarm handling, train sequencing and communications. Use simulated inputs to test normal and abnormal cases.

Agree the FAT protocol before the test. List prerequisites, records, acceptance criteria, punch classification and retest rules. FAT cannot prove site hydraulics, final water quality or integration with live utilities; those belong to site testing.

7. Plan installation and pre-commissioning

Inspect delivery condition, preservation, foundations, alignment, pipe cleanliness, cable termination, earthing, drains, chemical storage and access. Close documentation and safety prerequisites before wet commissioning.

Flush and pressure-test with compatible fluids. Record temporary strainers, passivation, disinfection and waste disposal. Protect membranes, resins and biological media from excessive storage or unsuitable chemicals.

8. Prove site integration

Site acceptance should test actual feed and product hydraulics, tank controls, source switching, residual routing, communications, remote commands, alarms and standby changeover. Challenge loss of instrument signals and utility interruptions.

Coordinate with the cooling-infrastructure integration guide so water-plant sequences are tested against the real cooling response.

9. Conduct performance and reliability tests

Define feed cases, stabilisation time, sampling, laboratory methods, meter accuracy and correction factors. Test net compliant product, quality, recovery, energy, chemicals, residuals and availability.

A reliability run should be long enough to observe routine backwash, duty rotation, cleaning or other ordinary events. Record excluded periods and operator intervention transparently.

The proof and assurance guide and recovery-guarantee guide provide complementary evidence and measurement frameworks.

10. Complete handover and lifecycle records

Handover should include approved drawings, calculations, software versions, settings, alarm lists, calibration, test records, certificates, spares, procedures and training. Close high-priority defects before operational responsibility transfers.

Link asset tags to the maintenance system and historian. Retain raw test data so future performance can be compared with the accepted baseline. Schedule post-handover review after loads and water quality become representative.

11. How Crescent Engineering can support the project

Crescent Engineering can define battery limits, utilities, control narratives, cause-and-effect schedules, FAT and site protocols, then support installation, commissioning, reliability testing and final performance evidence.

Crescent Engineering — Water Treatment