Retrofitting blowdown recovery into an operating data centre
Contents
Direct answer
Blowdown recovery can be retrofitted into an operating AI data centre without compromising cooling when the new plant remains hydraulically and operationally subordinate to the existing tower system. The site must be able to return to conventional blowdown immediately if recovery is unavailable.
Successful retrofits begin with measured surveys, verified tie-in points, a temporary-works plan and staged commissioning. They account for live cooling demand, restricted access, existing controls, water-quality transitions and the limited outage windows of a critical facility.
The concept can use a standard DC-BR packaged module, but its live-site interfaces should be proven using the battery-limits, FAT and commissioning framework.
1. Define the retrofit boundary and success criteria
State whether the project includes blowdown segregation, equalisation, treatment, recovered-product storage, return to make-up, concentrate handling, controls and civil or electrical work. Identify systems that remain under existing service contracts.
Success should include cooling continuity, net recovered flow and quality, residual compliance, safe fallback, acceptable maintenance and measured savings. A skid delivered to site is not a completed retrofit.
Use the general blowdown recovery guide for process fundamentals and the business-case guide for investment boundaries.
2. Survey the live water and cooling systems
Verify pipe sizes, materials, routes, elevations, valve condition, pump curves, tank volumes, drains, electrical capacity, control panels and communications. Compare drawings with the installation.
Use non-intrusive flow measurement and data logging where meters are missing. Check access for isolation, lifting, membrane replacement, chemical delivery and waste removal.
Identify asbestos, confined spaces, hazardous chemicals, buried services, fire routes and security restrictions. Data-centre operating procedures can affect work hours, tools, dust, noise and personnel access.
3. Establish the real operating baseline
Collect blowdown and make-up flow, chemistry, tower conductivity, heat load, weather, cells in service, chemical dosing and discharge data. Cover representative high-load operation.
Reconcile flow and chemistry cycles. Correct avoidable blowdown before freezing the recovery capacity. Characterise intermittent drains and cleaning separately.
Sample at the proposed treatment feed point. Existing basin or utility data may not represent the combined blowdown header after solids settle or chemical pulses mix.
4. Select tie-ins that preserve conventional operation
The existing blowdown path should remain available and independently operable. The recovery system may intercept flow, but a trip or isolation must not prevent the tower from discharging at its validated rate.
Recovered product should enter a controlled make-up or storage point with backflow protection and off-spec diversion. Avoid a tie-in that can pressurise or contaminate the existing system.
Concentrate and cleaning wastes need their own approved route. Confirm hydraulic capacity and prevent return flow from drains or sewers.
Provide isolation and drains for every new connection. Valve failure positions and manual bypass should be clear to facilities operators.
5. Plan layout, access and modular construction
Operating sites often have limited plot, headroom and crane access. Survey delivery routes, floor loading, bunds, drainage, ventilation, noise and heat rejection from the treatment equipment.
Modular skids can reduce site work and shutdown time. Module dimensions should reflect real doors, lifts and turning radii. Locate consumables and membranes where they can be replaced without entering critical cooling areas.
Prefabricate pipe spools and control panels after survey verification. Use laser scans or coordinated models where congestion makes conventional measurement unreliable.
Separate construction from operating water and chemicals. Dust, flushing debris and welding residues must not enter the tower or product system.
6. Integrate controls without creating common failures
Define the signals exchanged with tower, make-up, building-management and electrical systems. The recovery plant needs blowdown availability and tank status; the cooling system needs recovery availability and product-quality status.
Use clear fail states. Loss of communications should return the tower to conventional blowdown, not leave a critical valve in an uncertain position. A recovery PLC should not gain unnecessary authority over cooling operation.
Provide local manual control for safe isolation and maintenance, with monitored positions where they affect risk. Rationalise alarms so facilities staff receive actionable information.
Cybersecurity, network access, time synchronisation and historian ownership should be agreed before commissioning.
7. Manage shutdowns and temporary works
Create a tie-in register with location, isolation, drain volume, outage, temporary cooling consequence, method statement and rollback. Confirm valve isolation through testing where possible.
Schedule work against cooling load and redundancy. A short pipe cut can require a longer drain, refill, flushing and chemistry-restoration period. Include contingency for a seized valve or unexpected pipe condition.
Temporary hoses, tanks, pumps and electrical supplies require engineered capacity, protection and monitoring. Avoid temporary arrangements that bypass backflow or water-quality controls.
Use hold points before cutting, energising or introducing recovered product. The operations authority should control release.
8. Commission in safe stages
Commission utilities, controls and treatment off line where possible. Flush and clean new pipework to an approved destination. Prove instruments, valves, interlocks and off-spec diversion before connecting product to cooling make-up.
Start with conventional blowdown maintained. Introduce a limited blowdown feed, establish stable treatment, verify product quality and then return a controlled quantity. Increase stepwise while monitoring tower chemistry.
Simulate loss of power, analyser failure, high conductivity, low product tank, blocked residual route and communications failure. Confirm immediate fallback.
A reliability run should include normal backwash, at least representative cleaning behaviour and operator shift changes.
9. Protect operation during transition
Recovered product can change make-up hardness, alkalinity, chloride and inhibitor demand. Coordinate with the cooling-chemistry provider and define permitted ramp rates.
Track tower conductivity, individual ions, pH, corrosion evidence, biological control and heat-transfer performance. Maintain a rollback threshold.
Do not reduce backup-water or discharge capacity until sustained operation is proven. Early savings should not be converted into dependency prematurely.
10. Verify savings and maintainability
Apply the recovery guarantee and M&V plan. Reconcile feed, product, concentrate, make-up, blowdown and tank inventory at equivalent load.
Observe maintenance access, cleaning duration, chemical handling and alarm response during the reliability period. Correct layout or procedure issues before final acceptance.
Train operations and maintenance teams using actual scenarios. Deliver setpoints, calibrations, spares, drawings, code backups and baseline performance data.
11. Plan future capacity
Reserve tie-ins, power, controls and plot for later modules only where the campus programme justifies them. Confirm common infrastructure at each phase.
Early trains may operate at low turndown. Rotate standby equipment, manage tank turnover and preserve idle membranes. Update the business case as data halls energise.
Use operating evidence to refine later modules rather than copying the first design unchanged.
12. Apply formal change and hazard management
Treat the retrofit as a modification to a live critical utility. Use the site’s management-of-change process to review process safety, water quality, controls, cybersecurity, maintenance, environmental compliance and operator responsibilities.
Conduct an appropriate hazard and operability review. Scenarios include wrong valve line-up, chemical overdose, blocked concentrate, reverse flow, loss of ventilation, tank overflow, analyser failure and return of off-spec product. Track safeguards into drawings, code and tests.
Update operating procedures, training, permits, emergency response and asset registers before handover. Mark redundant lines and remove temporary bypasses that are no longer required.
Review supplier access and remote support against site security. A treatment service connection should not create an uncontrolled route into critical operational networks.
13. Protect business continuity during construction
Coordinate works with other live projects and maintenance. Competing demands for cranes, electrical isolations, drains or access routes can create risk even when each job is safe alone.
Use off-site fabrication and factory testing to reduce time near operating equipment. Inspect modules before delivery and verify the site interfaces are ready.
Prepare rollback for every critical cutover. Materials, spools, valves, gaskets, instruments and competent personnel should be present before isolation. Define the maximum decision time to restore the original configuration.
Communicate planned changes to facilities, IT operations, security and health-and-safety teams. After each stage, confirm cooling redundancy and alarm status before releasing the next activity.
14. How Crescent Engineering can support the project
Crescent Engineering can survey live sites, develop retrofit and temporary-works concepts, supply modular recovery equipment, coordinate controls and support staged commissioning without compromising the existing cooling duty.
15. References and further reading
Retrofit methods, isolation and commissioning must be approved through the operating site’s change-control and safety procedures.
Related Crescent articles
Continue exploring this topic:
The business case for cooling-tower blowdown recovery in AI data centres
How far should cooling-tower cycles of concentration be increased?
Filtration, softening and high-recovery RO for cooling-tower blowdown
Managing silica, hardness and salinity in cooling-tower blowdown
Brine minimisation and discharge control for blowdown recovery
Recovery guarantees and measurable water savings for data centres
Comments