The business case for cooling-tower blowdown recovery in AI data centres
Contents
Direct answer
The business case for cooling-tower blowdown recovery is strongest where an AI data centre has a large, stable evaporative load, meaningful water or sewer constraints and a reliable destination for recovered product. The calculation must begin after the tower has been optimised. Otherwise the recovery plant is sized to treat avoidable waste and its projected savings are overstated.
A credible investment case compares the existing and proposed systems at the same cooling duty, weather and water chemistry. It values avoided make-up, discharge and capacity upgrades, then deducts treatment losses, energy, chemicals, membranes, maintenance, labour and final-residual costs. It also tests the value of water security without pretending that recovery removes the evaporation required to reject heat.
The final recommendation should state the conditions under which the project remains attractive, the risks that can change the result and the measurements that will prove savings after commissioning.
Once the case is established, the DC-BR packaged recovery guide converts the process concept into a defined module; projects without a viable concentrate route should also test the DC-ZD minimum and zero-discharge pathway.
1. Define the decision and the baseline
The decision is not simply whether water can be recovered. It is whether a specific recovery arrangement provides better whole-life value than continued discharge or another water strategy while protecting cooling availability.
Establish a measured baseline covering tower make-up, evaporation, drift, blowdown, overflow, source quality, tower conductivity, heat load and weather. Include chemical use, sewer charges, maintenance and current operational problems. A short baseline collected in mild weather may not represent the summer condition that determines connection and treatment capacity.
Reconcile flow-based cycles with chemistry-based cycles. If the make-up-to-blowdown ratio does not match conductivity concentration, investigate unmetered losses, faulty instruments, intermittent bleed or changes in source blend. Recovery economics built on an unreconciled balance are not bankable.
The existing cooling-tower blowdown recovery guide explains the overall technical sequence; this article focuses on turning that sequence into an investment decision.
2. Optimise cycles before sizing recovery
The first saving usually comes from operating the tower correctly. Calibrate conductivity instruments, correct leaking blowdown valves, eliminate unnecessary overflow and confirm the chemical programme. Side-stream filtration or make-up conditioning may allow safe operation at higher cycles without a separate recovery plant.
The benefit of increasing cycles diminishes. With negligible drift, blowdown is approximately evaporation divided by cycles minus one. Moving from two to three cycles halves blowdown relative to evaporation; moving from six to seven produces a much smaller reduction. The chemistry and reliability cost can continue to rise even as the incremental water saving falls.
Document the verified limiting condition: silica, hardness saturation, chloride, corrosion, suspended solids, biological control or discharge. The optimised blowdown flow and composition become the feed basis for recovery. See Increasing cycles of concentration for the detailed method.
3. Quantify recoverable flow and salt load
Model average, peak, minimum and upset conditions. Blowdown changes with heat rejection, source chemistry, tower cycles and operating cells. A recovery plant sized to the annual average may spill during the hours that set sewer capacity; one sized to peak flow may operate inefficiently for much of the year.
Separate hydraulic recovery from net site saving. If 100 cubic metres per hour of blowdown enters a plant operating at 75% recovery, nominal product is 75 cubic metres per hour and concentrate is 25. Pretreatment backwash, cleaning and off-spec product reduce net return. Pumps, tanks and pipe losses may add more.
Complete a constituent mass balance. Recovery does not destroy chloride, sulphate, silica or treatment chemicals; it moves them into product, concentrate, sludge or cleaning waste. Product returned to the same tower creates a recycle loop, so membrane passage and blending must be included in the steady-state calculation.
The final residual route can cap achievable recovery. A sewer may accept 40 cubic metres per hour at moderate salinity but reject 20 cubic metres per hour at twice the chloride concentration. Business cases should model both volume and mass limits.
4. Identify every source of value
Direct benefits may include avoided make-up-water purchase, lower sewer charges, reduced chemical use associated with lower fresh make-up and deferred utility or discharge upgrades. Where potable or desalinated water is displaced, recovery can support corporate water objectives.
Capacity value can exceed tariff savings. A site with a constrained incoming main or sewer may use recovery to accommodate another data hall without major off-site works. That benefit should be tied to documented capacity, programme and alternative cost.
Resilience value requires care. Blowdown recovery is an internal recycle and cannot replace all external water because evaporation continues. It may extend on-site autonomy or reduce the refill rate after an interruption, but only while the cooling towers operate and the recovery plant remains available.
Avoided-risk value should be explicit rather than hidden in optimistic savings. Examples include reduced exposure to future water restrictions, lower dependence on tanker supply during utility work or flexibility to respond to rising discharge tariffs. State the probability and consequence used.
5. Build the full lifecycle cost
Capital cost includes feed segregation, equalisation, pretreatment, membranes or other recovery stages, product and concentrate storage, pumps, pipework, controls, electrical works, buildings, chemicals, safety systems and connection modifications. Retrofit access and outage work can be material.
Operating cost should cover energy, chemicals, consumables, membranes, cleaning, laboratory analysis, sludge or concentrate disposal, maintenance labour, service agreements and critical spares. Include downtime and reduced recovery during cleaning or poor feed conditions.
Use realistic local prices and escalation. Imported membrane or chemical lead times may justify additional inventory. Residual disposal costs can vary sharply if sewer acceptance is lost.
Apply the organisation’s approved discount rate, project life, tax and residual-value assumptions. Present net present value, internal rate of return or payback only after the underlying annual flows and costs are visible. A single payback number conceals sensitivity.
6. Compare treatment concepts on equal terms
At minimum, compare continued optimised blowdown, pretreatment with reuse for a tolerant duty, softening with additional tower cycles, partial membrane recovery and high-recovery membrane treatment. Thermal or minimum-liquid-discharge options should be included only where the residual route justifies their complexity.
Define the same feed cases, product requirement, availability and residual boundaries for every option. Compare net product delivered, not headline process recovery. Include one-train-out capacity and turndown.
Filtration, softening and high-recovery RO should be arranged around the controlling chemistry. Additional stages earn their place by removing a specific constraint, not by making the flow diagram appear comprehensive.
Score operational fit as well as cost. A technically efficient process may be unsuitable if it requires continuous specialist attention, creates hazardous cleaning work or lacks local service support.
7. Price risk, uncertainty and flexibility
Run sensitivity cases for blowdown flow, source-water quality, achievable recovery, membrane life, cleaning frequency, power price, chemical price, discharge tariff, residual disposal and campus load. Report break-even values.
Use separate scenarios for sustained high silica or hardness, a change in source blend, reduced sewer salinity allowance and loss of product reuse. These events can change both capacity and cost.
Pilot testing reduces technical uncertainty but does not eliminate commercial risk. Its value is highest when sustainable flux, cleaning interval or residual quality controls the investment. Define success criteria before the test.
Flexibility can have measurable value. Modular trains, space for a future stage, adjustable blending and a bypass to normal blowdown allow the system to adapt without jeopardising cooling. Price this flexibility against the cost of overbuilding from day one.
8. Test the effect of campus phasing
AI data centres often add halls over several years. Early blowdown may be too small for efficient high-recovery operation, while ultimate flow may exceed the first module. Model each energisation stage and the period between stages.
Modular equipment can align expenditure with demand, but common tanks, electrical infrastructure and headers should be sized deliberately. Oversized equalisation can create biological or sediment problems at low turnover.
The commercial model should distinguish committed load from speculative ultimate capacity. Do not credit savings from future halls without also including the capital and operating cost needed to serve them.
9. Set commercial decision gates
Use gates that can stop or redirect the project. Feasibility requires a reconciled baseline, a defined product use and a plausible residual route. Concept selection requires representative chemistry, process modelling and a lifecycle comparison. Procurement requires an agreed feed envelope, guarantees and acceptance tests.
If the final concentrate lacks written acceptance, the project is not ready for investment approval. If the tower is not optimised, recovery capacity is not fixed. If product use is uncertain, claimed saving should be discounted.
Record assumptions, owner and closure date. This produces a decision trail that can survive changes in personnel and project pressure.
10. Specify guarantees and measurement
Guarantees should cover net product flow and quality, sustainable recovery, availability, energy, chemical consumption, waste volume and key residual characteristics across agreed feed cases. State temperature corrections, sampling methods and test duration.
Savings require a measurement-and-verification plan. Meter blowdown feed, product, concentrate, relevant backwash and returned make-up. Record tower heat load, weather, cycles and tank inventory so post-project performance can be compared with the baseline fairly.
Recovery guarantees and measurable savings should distinguish process performance from site saving. A plant can meet its recovery guarantee while the site saves less because product is spilled, another source increases or tower operation changes.
11. Review a worked decision example
Consider an illustrative site with an optimised average blowdown flow of 60 cubic metres per hour for 8,000 equivalent full-load hours. A 75% net recovery would return 360,000 cubic metres per year before site-level losses. At a combined avoided water and sewer value of AED 6 per cubic metre, gross annual value is AED 2.16 million.
If energy, chemicals, membranes, labour, maintenance, testing and concentrate disposal total AED 1.05 million per year, simple operating benefit is AED 1.11 million before capital recovery. A capital cost of AED 6 million would imply a simple payback of roughly 5.4 years, but this is not yet an investment decision.
If net recovery falls to 60%, residual disposal doubles in cost or the site operates at lower load for three years, the result changes materially. Conversely, deferring a sewer upgrade may add substantial value. The approved model should show these scenarios and use measured data after commissioning.
The numbers are illustrative only. They demonstrate the structure of the calculation and should not be used as project estimates.
12. How Crescent Engineering can support the project
Crescent Engineering can audit tower performance, reconcile water and salt balances, develop recovery concepts, define pilot work, prepare lifecycle costs and translate the selected option into packaged equipment, controls, guarantees and commissioning tests.
13. References and further reading
All financial values, process performance and discharge assumptions must be validated for the specific site.
Related Crescent articles
Continue exploring this topic:
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
Retrofitting blowdown recovery into an operating data centre
Comments