Data Centre Water Usage
Effectiveness (WUE):
Cooling Tower vs Adiabatic
vs Dry Coolers

Power Usage Effectiveness (PUE) has been the defining efficiency metric of the data centre industry for fifteen years. It has driven enormous improvements in electrical design — from average PUEs of 2.0 in 2007 to industry best practice of 1.15–1.30 today. But PUE measures only one dimension of resource consumption. It says nothing about water.

A data centre that achieves PUE 1.15 through aggressive evaporative cooling may simultaneously consume 3–5 litres of water per kilowatt-hour of IT load — a figure that, at 10 MW of IT load, represents 720,000 litres of water per day. In Chennai, Hyderabad, or NCR — cities that face genuine water scarcity for significant parts of the year — this is not a theoretical concern. It is a planning authority constraint, a water utility negotiation, and an ESG disclosure obligation that many data centre developers are encountering for the first time as their facilities grow to hyperscale dimensions.

Water Usage Effectiveness (WUE) is the metric that captures this, and the choice between cooling tower, adiabatic, and dry cooler heat rejection strategies is the primary engineering lever for controlling it.

WUE: Definition, Measurement, and Targets

// WUE Definition (The Green Grid, 2011)

WUE = Annual Water Usage (litres) / Annual IT Energy (kWh)

// Units: L/kWh  (litres of water consumed per kilowatt-hour of IT energy)

// What counts as "water usage"?
// — Evaporation from cooling towers and adiabatic coolers  ← primary term
// — Cooling tower blowdown (water discharged to maintain water quality)
// — Drift losses from cooling towers
// — Make-up water for humidification systems
// NOT included: potable water for staff, landscaping, fire suppression testing

// Industry WUE benchmarks (The Green Grid / Uptime Institute):
World-class   : < 0.5 L/kWh   ← dry cooler or immersion + dry cooler
Good          : 0.5 – 1.0 L/kWh ← adiabatic with controlled operation
Average       : 1.0 – 2.0 L/kWh ← cooling tower with free cooling hours
Poor          : > 2.0 L/kWh   ← full evaporative cooling, no free cooling

// Example: 10 MW IT load, WUE = 2.0 L/kWh, 8,760 hr/year
Annual water = 10,000 kW × 8,760 hr × 2.0 L/kWh = 175,200,000 L = 175 ML/year
// Equivalent to ~480,000 litres per day — a significant urban water demand

Cooling Towers: Highest Efficiency, Highest Water Consumption

Evaporative cooling towers are the dominant heat rejection technology in Indian data centres today. They achieve the lowest chiller approach temperature — and therefore the lowest chiller energy consumption and best PUE — of any heat rejection technology. They do so by evaporating water directly into the air, which is why they are also the highest water consumers.

Cycles of Concentration and Blowdown

As water evaporates from a cooling tower, dissolved minerals concentrate in the remaining water. Cycles of concentration (CoC) measures how much more concentrated the tower water is compared to make-up water. Higher CoC means less blowdown required (less water wasted) but higher mineral concentration that risks scale formation on heat exchanger surfaces. The optimum CoC for Indian municipal water quality (typically high TDS) is 3–5 cycles, maintained by controlled blowdown and chemical treatment including scale inhibitor, biocide, and pH adjustment dosing.

// Cooling tower water balance — 10 MW data centre
// Heat rejected: 4 MW  |  Range: 6°C  |  Wet-bulb: 25°C  |  CoC: 4

Evaporation rate  = 4,000 kW × 0.00153 m³/kWh = 6.1 m³/hr  (6,100 L/hr)
Drift loss        = ~0.05% of circulation  ≈  30 L/hr  (modern drift eliminators)
Blowdown          = Evaporation / (CoC − 1) = 6,100 / 3 = 2,033 L/hr
──────────────────────────────────────────────────────────
Total make-up     = 6,100 + 30 + 2,033 = 8,163 L/hr  →  196 m³/day
Annual consumption= 196 × 365 = 71,540 m³/year  =  71.5 ML/year

// WUE from cooling tower alone (IT energy = 10 MW × 8,760 hr = 87,600 MWh)
WUE_tower         = 71,540,000 L / 87,600,000 kWh = 0.82 L/kWh

Adiabatic Coolers: The Middle Path

Adiabatic coolers — also called evaporative pre-coolers or wet/dry coolers — are a hybrid technology that combines a dry cooler (finned coil, no direct water contact with the airstream) with an evaporative pre-cooling pad on the inlet air. Water is sprayed onto the pad when ambient temperature is high; the evaporative cooling of the inlet air reduces the temperature of air entering the dry cooler coil, improving its heat rejection performance without direct contact between water and the process fluid.

Adiabatic vs Cooling Tower: WUE and Performance Comparison

Adiabatic Activation Control Strategy

The key to minimising adiabatic cooler water consumption is an intelligent controls strategy that activates water pre-cooling only when the dry-bulb temperature exceeds the threshold at which the dry cooler alone cannot maintain the design supply temperature. A poorly configured adiabatic cooler that activates water spray at a conservative setpoint (e.g. any time ambient exceeds 25°C) will waste water during conditions where dry cooling was perfectly adequate. The correct activation setpoint is determined by the specific dry cooler coil performance curve at the design water flow rate — not a generic rule of thumb.

// Adiabatic activation logic — example for 35°C CDU supply setpoint

// Dry cooler can maintain 35°C supply when:
// Ambient dry-bulb < (35 − approach) = 35 − 8 = 27°C  →  dry cooling adequate
// Ambient dry-bulb > 27°C  →  activate adiabatic pre-cooling

// Adiabatic pad reduces effective inlet temp toward wet-bulb:
// Dry-bulb = 38°C, Wet-bulb = 27°C  →  saturation efficiency ~80%
T_eff = 38 − 0.80 × (38 − 27) = 38 − 8.8 = 29.2°C
// Cooler now sees 29.2°C inlet instead of 38°C → can achieve <35°C supply ✓

// Hours per year requiring adiabatic activation — Delhi (ASHRAE data):
// Hours with DB > 27°C: ~3,100 hr/year  (35% of annual hours)
// WUE contribution from adiabatic hours only — much lower than cooling tower full-year

Dry Coolers: Zero Water, Higher Chiller Energy

Dry coolers — finned-tube heat exchangers rejecting heat entirely to ambient air with no evaporative component — achieve WUE of essentially zero. They are the correct choice for liquid-cooled AI data centres where the CDU secondary water loop operates at 35–45°C supply temperature, because at this elevated temperature, dry coolers achieve free cooling for a large fraction of the Indian year without any chiller operation at all.

The Temperature Crossover Advantage for Liquid Cooling

// Dry cooler free cooling hours — comparison of conventional vs liquid cooling

// Conventional air-cooled DC: chiller needs 6–12°C condenser water supply
// Free cooling possible when: ambient dry-bulb < 6 − 5°C approach = 1°C
// Delhi annual hours below 1°C dry-bulb: ~150 hr/year → minimal free cooling

// Liquid-cooled DC (DLC/immersion): CDU needs 35–40°C secondary water supply
// Free cooling possible when: ambient dry-bulb < 35 − 8°C approach = 27°C
// Delhi annual hours below 27°C dry-bulb: ~5,200 hr/year → 59% free cooling ✓

// Free cooling hours by Indian city (35°C setpoint, dry cooler):
Delhi       : ~5,200 hr/yr  (59%)   — excellent free cooling potential
Pune        : ~5,800 hr/yr  (66%)   — best free cooling of major DC cities
Hyderabad   : ~4,600 hr/yr  (53%)   — good; moderate summer heat
Bengaluru   : ~6,200 hr/yr  (71%)   — best climate; high altitude benefit
Chennai     : ~2,800 hr/yr  (32%)   — limited; high humidity year-round
Mumbai      : ~2,400 hr/yr  (27%)   — poorest; coastal humidity constrains cooling

Dry Cooler Sizing for Peak Summer Conditions

The dry cooler must be sized to maintain the design supply temperature at the summer peak dry-bulb condition — not the annual average. In Chennai, this means designing for 40°C ambient while maintaining 45°C supply (5°C approach). The resulting coil area is significantly larger than what is needed for 90% of the year — but undersizing the dry cooler means falling back to chiller operation (and increased energy cost) during summer peaks rather than the water spraying that characterises an adiabatic system.

India-Specific Water Context

The water management challenges for Indian data centres are materially different from those in European or North American locations, and the cooling technology selection must be made with Indian water reality — not ASHRAE generic guidance — as the baseline.

Technology Selection Framework by Site Context

WUE Monitoring and Reporting

WUE is not a design target that can be set once and forgotten — it is an operational metric that must be continuously monitored and reported against. The monitoring infrastructure must be installed as part of the cooling plant commissioning, not retrofitted later.

• 01

  Make-Up Water Metering

  Calibrated water flow meter on the make-up water supply to each cooling tower or adiabatic cooler — not just a facility-level water meter that also captures domestic, fire, and irrigation consumption. The metering data must be time-stamped and logged to the BMS at minimum 15-minute intervals. Annual WUE requires the ratio of total metered cooling water make-up to total IT energy from the power metering system — both must be on the same time base.

• 02

  Blowdown Monitoring and Conductivity Control

  Automatic conductivity controller on the cooling tower basin modulates the blowdown valve to maintain design cycles of concentration. Manual blowdown — a common practice in smaller facilities — wastes water (over-blowdown) or causes scale (under-blowdown). Conductivity data logged to BMS enables retrospective analysis of water consumption efficiency and early detection of water quality excursions that drive excessive blowdown.

• 03

  Adiabatic Activation Hours Logging

  For adiabatic systems, log the total annual hours of water activation, the volume activated, and the ambient dry-bulb temperature at activation. This data validates that the control setpoint is correctly optimised — systematic activation at temperatures where dry cooling alone was adequate is identifiable and correctable without any hardware change.

• 04

  Annual WUE Reporting

  Annual WUE calculation and disclosure — alongside PUE — should be included in the data centre’s sustainability report. For hyperscaler tenants operating in India, WUE disclosure is increasingly a contractual requirement in colocation agreements. Establishing the measurement baseline from commissioning date avoids the difficulty of retrospectively justifying earlier years’ consumption when reporting obligations are first introduced.

Conclusion: Water Efficiency Is Now a Design Constraint

WUE has been a voluntary metric since The Green Grid introduced it in 2011. In India in 2025, it is becoming a planning requirement, a tenant specification, and a water utility negotiation item. The data centres that treat it as a voluntary metric today will be retrofitting cooling technology at considerable cost and disruption within five years.

The engineering decision framework is straightforward in its headline: liquid-cooled AI data centres should use dry coolers — their warm water operating temperature makes dry cooling effective for the majority of the Indian year, achieves near-zero WUE, eliminates Legionella risk, and reduces maintenance burden. For air-cooled facilities, adiabatic cooling represents the best current compromise between PUE and WUE for most Indian climates, with cooling towers reserved only for coastal humid locations where adiabatic systems reach their performance limits.

India’s hyperscale data centre campuses are being planned today for a thirty-year operational life. The cities and water sources that will supply them in 2045 will be under greater stress than they are today. Designing for low WUE is not environmental virtue — it is infrastructure prudence.