iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
Short answer: In a common cooling-tower design, heat from servers moves into chilled-water and condenser-water loops. At the cooling tower, some circulating water evaporates and carries heat into the atmosphere. Fresh makeup water replaces evaporation and other losses, while a separate stream called blowdown is drained to keep dissolved minerals from concentrating. Blowdown may be treated and reused, but there is no universal destination: the final handling depends on the facility’s equipment, water quality, permits and local infrastructure.
That distinction matters. “Water used by a data center” can mean water consumed on the site, water discharged from a cooling system, or water associated with the electricity that powers the servers. Those are different quantities.
Where water fits in the heat-removal chain
Servers turn electrical energy into computation and heat. Cooling equipment moves that heat away in several stages rather than sending tap water directly over the electronics.
- IT equipment produces heat. Air or a liquid coolant absorbs heat at the racks. The exact arrangement depends on the server design and density.
- A facility cooling loop transports heat. In a chilled-water example, heat moves from the room or a liquid-cooling heat exchanger into chilled water.
- A condenser-water loop carries heat to the tower. A chiller transfers heat from the chilled-water side to condenser water, which circulates outdoors to the cooling tower.
- The tower rejects heat. Warm condenser water is distributed over fill material while fans or natural draft move air through it. A fraction of the water evaporates; that phase change removes heat from the remaining water.
This simplified path follows the U.S. Department of Energy Federal Energy Management Program (FEMP) description of a chilled-water system, condenser-water loop and tower. Real facilities can combine or omit stages, use dry coolers, or circulate liquid directly at the racks, so the diagram is a representative arrangement rather than a rule for every data center. FEMP’s cooling-water guidance describes the equipment relationships.
What happens to water in a cooling tower?
Evaporation: water that leaves as vapor
Evaporation is the tower’s primary heat-rejection mechanism in an evaporative system. Water changes from liquid to vapor and exits with the tower’s air discharge. That water is consumed at the site in the sense that it is no longer available in the circulating loop; it is not a liquid effluent sent to a drain.
Makeup water: the replacement supply
Automatic controls add makeup water to replace evaporation, intentional blowdown and smaller operational losses such as drift, leaks and maintenance draining. A meter on the makeup line records the incoming volume. Makeup demand rises with heat load, outdoor conditions, tower efficiency, cycles of concentration and the quality of the incoming water.
Blowdown: a separate liquid stream
As water repeatedly circulates, minerals and other dissolved solids become more concentrated because evaporation removes water but leaves most dissolved material behind. The tower periodically drains (“blows down”) part of the circulating water and replaces it with lower-mineral makeup water. Blowdown is therefore a liquid discharge from the loop, not water that evaporated into the air.
Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Fix the driver behind crashes, sound loss and screen glitches3Clear out junk files and repair common Windows errorsFEMP explains that increasing the tower’s cycles of concentration allows more dissolved material to remain in the loop before a drain event is required. The feasible limit depends on scaling, corrosion, biological control, equipment warranties and the chemistry of the supply water. A higher cycle count can reduce both makeup and blowdown, but it is not automatically safe for every tower. See FEMP’s explanation of cycles of concentration and blowdown.
Where does blowdown go?
There is no single answer that applies to all facilities. A site may send blowdown to a permitted wastewater system, pretreat it, combine it with other facility streams, or process it for on-site reuse. The reviewed guidance establishes the blowdown mechanism and a possible treatment route, but not one universal sewer, treatment plant or receiving-water destination. To identify a particular data center’s endpoint, you need that site’s plumbing drawings, discharge permit, water balance and treatment records.
Is cooling water consumed, discharged or reused?
| Water pathway | What happens | How it is normally counted |
|---|---|---|
| Evaporation | Liquid becomes vapor in the tower’s heat-rejection process. | On-site consumptive use; it leaves the facility through the air. |
| Blowdown | Mineral-rich circulating water is intentionally drained. | Liquid discharge unless treated and reused; destination is site-specific. |
| Makeup | New water enters to replace evaporation, blowdown and other losses. | Incoming site water; part becomes evaporation and part may later leave as blowdown. |
| Treated reuse | Blowdown is processed, and a usable fraction returns as tower makeup. | Reduces new makeup demand but does not necessarily eliminate a reject stream or energy use. |
Reverse osmosis (RO) is one possible treatment step. FEMP describes producing a lower-mineral permeate suitable for reuse as makeup while sending a concentrated reject stream to discharge. RO adds pumps, pretreatment, membrane replacement and monitoring, and therefore increases energy and operating requirements. It is a site-specific water-management option, not a promise of a completely closed loop. FEMP discusses RO treatment of cooling-tower blowdown.
How to measure a data center’s water use
Site WUE
Water usage effectiveness (WUE) expresses annual site water use relative to IT energy use:
WUE (site) = annual site water use in liters ÷ annual IT equipment energy use in kilowatt-hours.
Rank #3
The result is reported in liters per kilowatt-hour. The numerator should be defined clearly: an operator may include all facility water or only water consumed by cooling, and those choices affect comparisons. FEMP provides the site-WUE definition and calculation framework. FEMP’s WUE guidance is the reference for the formula.
Source WUE and electricity-related water
Site WUE does not include water consumed upstream to generate electricity. Lawrence Berkeley National Laboratory (LBNL) distinguishes direct, on-site consumption as WUE (site) from WUE (source), which accounts for water consumed in electricity generation. A facility can therefore report a low site-WUE value while drawing power from a water-intensive generation mix, or report higher site water use while relying on electricity with a lower source-water factor. The 2024 United States Data Center Energy Usage Report explains these definitions and the associated methodology on pages 39–45. Read the LBNL 2024 report (PDF).
For a meaningful comparison, examine at least four axes together:
The Tool Desk
Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →- Direct site water consumed, including how evaporation is estimated or metered.
- Water associated with electricity generation.
- Total electricity use and cooling efficiency.
- Climate, water quality, operating hours and other local conditions.
Why a waterless cooler is not automatically the lowest-water choice
Cooling design shifts burdens between water and electricity. LBNL reports that water-cooled chillers and other evaporative systems are generally more energy-efficient than air-cooled chillers and other waterless alternatives, although actual performance depends on design and operating conditions. An air-cooled system can reduce direct site water while requiring more electricity; that extra electricity can carry additional source-water consumption at the power plant.
| Cooling approach | Direct site-water tendency | Energy and operational consideration |
|---|---|---|
| Evaporative cooling tower | Higher, because evaporation and blowdown require makeup water. | Generally energy-efficient for heat rejection; water chemistry and discharge controls are required. |
| Air-cooled chiller or dry cooler | Lower during normal operation because no tower evaporation is required. | Often uses more electricity, especially during hot weather; source-water effects must be counted. |
| Airside economizer | Can reduce mechanical cooling and associated water when outdoor air conditions are suitable. | Availability depends on climate, humidity, air quality and filtration; it may not meet loads year-round. |
| Liquid cooling at or near the rack | Varies by loop design; a sealed primary loop may need little replacement water. | Can alter fan and chiller demand, but requires compatible servers, heat exchangers and controls. |
These categories are not mutually exclusive. A facility may use economizers in cool periods, evaporative towers for peak conditions and liquid cooling for high-density racks. LBNL’s report covers airside economizers, liquid cooling and the water-energy tradeoff. Consult the report’s cooling-system discussion.
Why workload water use varies so widely
A single “liters per workload” number is misleading because water intensity depends on the entire computing and infrastructure stack. A 2025 review by Nuoa Lei, Jun Lu, Arman Shehabi and Eric R. Masanet found more than a 10,000-fold variation in workload-level water use. The authors link that spread to large differences in water consumed per kilowatt-hour of server electricity and to roughly 10-fold differences in server workload efficiency.
Important determinants include:
- Server efficiency: newer or better-configured hardware can perform more work per kilowatt-hour.
- Utilization: lightly loaded servers still draw baseline power, so consolidating or scheduling work changes water per task.
- Grid water factors: the same data-center load can have different source-water impacts depending on the regional generation mix.
- Cooling type: tower evaporation, dry cooling, economizers and liquid loops have different site-water and energy profiles.
- Climate: outdoor temperature and humidity change the hours when evaporative or mechanical cooling is needed.
- Infrastructure and overhead: chillers, pumps, fans, redundancy and power-conversion losses affect the energy denominator.
- Water quality and operating policy: cycles of concentration, treatment targets and allowable discharge chemistry alter makeup demand.
The review’s central conclusion is that “there is no single recipe for minimizing water use”; the best outcome comes from a tailored combination of server, facility, grid and operating choices. Read the 2025 review by Lei, Lu, Shehabi and Masanet.
Quick wins for a faster PC:
Scan for outdated or missing drivers - takes under a minuteDriver Scan →Repair Windows errors before they cause bigger problemsFix Now →Practical measures that can reduce tower water demand
1. Increase cycles of concentration carefully
FEMP says cooling towers commonly operate at two to four cycles, while six or more may be possible when water chemistry and system constraints allow. On the page, FEMP cites a change from three to six cycles associated with a 20% reduction in cooling-tower makeup requirements and a 50% reduction in blowdown. Those are figures for the cited tower-management change, not guaranteed savings for every data center.
- Test makeup and recirculating water for hardness, alkalinity, conductivity, silica and biological indicators.
- Set a chemistry-based conductivity or concentration limit with the water-treatment specialist.
- Adjust blowdown controls gradually rather than disabling them.
- Verify scale, corrosion and microbial-control results after the change.
- Reconcile makeup meters, blowdown estimates and evaporation calculations to confirm the water balance.
2. Avoid unnecessary temperature and humidity control
FEMP identifies less aggressive temperature and humidity control, where appropriate, as an operational opportunity. Any change must remain inside the server manufacturer’s allowable environmental envelope and the facility’s reliability plan. Raising a set point is not a universal recommendation: local climate, rack density, economizer operation and redundancy determine whether it lowers energy or water use without creating equipment risk.
3. Treat blowdown for reuse where the balance works
RO or another treatment train can recover a reusable fraction of blowdown as makeup water. Before installing it, compare the value of avoided makeup and discharge with pretreatment, electricity, membrane replacement, concentrate disposal and operator workload. The reject stream still needs a lawful destination, so “reuse” does not mean that every gallon remains on site.
4. Meter each major stream
Separate meters for makeup, blowdown, potable supply, reclaimed supply and treatment reject make it possible to calculate a defensible water balance. Pair those readings with IT energy, cooling energy, weather and conductivity data. Without stream-level measurement, a low or high WUE figure can conceal a meter boundary, an unmeasured discharge or a change in workload.
Free tools Windows power users keep installed
One-click scans. No signup required.
How climate, water quality and permits change the answer
Two otherwise similar data centers can have different water pathways because they operate under different conditions. A hot, humid climate may require more mechanical heat rejection than a cool, dry location. Hard or silica-rich makeup water can force lower cycles of concentration, increasing blowdown. Local drought rules may limit potable withdrawals or require reclaimed water. Discharge permits can set limits for temperature, conductivity, metals, biocides and other constituents, which in turn influence treatment and operating settings.
For a specific facility, ask for four records: the cooling-system schematic; makeup, blowdown and treatment-meter data; the water-quality and chemical-control plan; and the discharge or reuse authorization. Those documents establish where liquid water actually goes. Public sustainability reports may provide a site-WUE figure, but they do not by themselves identify every stream or the water consumed to generate electricity.
What a complete water account should report
- The boundary: data-center campus, building, cooling plant or only the tower.
- The period: annual total, seasonal interval and whether the value is measured or modeled.
- Site water consumption in liters and the resulting WUE (site).
- Electricity use and the method used to estimate WUE (source).
- Cooling technology, tower cycles, makeup source and treatment process.
- Evaporation, blowdown, drift, leaks and treatment reject as separate pathways where measured.
- Climate zone, workload mix and utilization assumptions when reporting workload intensity.
- Any reclaimed-water supply, discharge permit limits or off-site treatment arrangement.
Reporting these details prevents the most common misunderstanding: treating a direct site-water number as the facility’s complete water footprint, or assuming that all incoming cooling water disappears through evaporation.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
