Mitigation Solutions Catalog
A reference catalog of technologies that reduce data center water and/or power consumption. Each entry shows the measure's effect range (sourced), when to use it, when not to, implementation considerations, and a one-click "Add to Stack" link that opens the Assessment Tool with this measure pre-toggled. WaterMark does not sell or endorse specific vendors — entries describe technology classes, not products. The same 17 measures appear as toggles in the Assessment Tool's Mitigation Stack and feed the Commitment Sheet exhibit. Methodology §5.5 →
Categories
- Cooling Architecture — 5 baseline cooling system choices set in the calculator dropdown
- Water Source — 5 measures that change where cooling makeup water comes from
- Treatment — 4 measures that reduce blowdown and tower losses
- Power (Scope 2) — 5 measures that reduce or shift upstream power-plant water
- Operations — 3 measures that change how the facility runs
How this catalog works. Effect ranges are sourced from peer-reviewed papers, ASHRAE / EPA / NREL guidance, and trade-association manuals. Ranges reflect typical real-world performance; specific facilities may fall outside these bounds depending on climate, equipment selection, and operational discipline. Citations link to primary sources where possible.
Three special types are flagged in the catalog:
- Source substitution — does not reduce gross volume; shifts demand off potable supply (reclaimed, brackish + desal). Useful for community impact even when not for total water.
- Plant-specific — water effect depends on the specific generation or cooling design and is not modeled in the calculator's headline numbers (BTM gas + CHP, BTM SMR). Listed for forward-looking commitments.
- Compliance lever — no direct volume effect but is the verification mechanism that makes commitments enforceable (real-time telemetry).
Cooling Architecture
Cooling architecture is the baseline choice — set in the calculator's Cooling System dropdown rather than as a stackable mitigation, since the options replace each other rather than compose. Listed here for completeness.
Evaporative cooling tower (open-loop)
Proven · BaselineHeat is rejected by evaporating a portion of recirculating cooling water. Industry default for hyperscale: simple, capex-light, highly effective in any climate. Water consumption is the cost.
| Water rate (LBNL 2021 fleet avg) | 1.8 L/kWh IT load |
| Power penalty | None (this is the baseline) |
| Capex | Lowest of any cooling architecture |
| Best fit | Hyperscale, humid or temperate climate, water available |
| Avoid when | Watershed-stressed basin, drought-prone region, high regulatory scrutiny |
| Citation | LBNL 2021 — US Data Center Energy Usage Report |
Adiabatic / hybrid cooling
ProvenHybrid system that operates as dry-air cooling most of the year and switches to evaporative pre-cooling only during peak heat. Significantly reduces water vs. open evaporative while keeping power penalty modest. Sweet spot in arid climates with hot summers.
| Water rate | ~0.8 L/kWh (~55% reduction vs. evaporative) |
| Power penalty | 0–5% |
| Capex | ~20–35% premium over open evaporative |
| Best fit | Arid climates with hot summers (Phoenix, Tucson, Vegas, Reno) |
| Avoid when | Climate is consistently mild (PNW) — pure dry cooling may be cheaper; or capex-constrained |
| Citation | Energy, 2023 — Adiabatic cooling systems |
Air-cooled / dry cooling
ProvenNo water use for cooling — heat is rejected directly to ambient air. Trades water for energy: chillers and fans work harder, especially in hot climates. Becoming more common as water becomes the binding constraint and air-side economizers are paired with smarter compressor-bypass control.
| Water rate | ~0.01 L/kWh (humidification only) |
| Power penalty | 10–25% over open evaporative; worst-case in 110°F+ ambient |
| Capex | Significant premium for chillers and condenser surface area |
| Best fit | Cool / dry climates (PNW, MN, mountain west); water-constrained sites where any consumption is unacceptable |
| Avoid when | Peak summer wet-bulb >70°F sustains for weeks; PUE-sensitive deployment |
| Citation | LBNL 2021 |
Direct-to-chip liquid cooling
Proven (hyperscale)Liquid coolant circulates through cold plates mounted directly on each chip. Higher heat-transfer efficiency enables denser racks (40–100 kW vs. 10–20 kW for air) and meaningful water reduction. Increasingly the default for AI training clusters where chip TDPs exceed practical air-cooling limits.
| Water rate | ~1.2 L/kWh effective (20–40% reduction vs. evaporative) |
| Power penalty | −5% to −15% (lower PUE) |
| Capex | Per-rack premium offset by fewer racks per workload |
| Best fit | High-density AI / HPC clusters; greenfield builds with cooling distribution units (CDUs) |
| Avoid when | Low-density mixed workloads where retrofit complexity outweighs benefit |
| Citation | LBNL 2021; ASHRAE TC 9.9 liquid-cooling guidance |
Two-phase liquid immersion
EmergingServers fully submerged in dielectric fluid that boils at ~50°C, transferring heat by phase change. Near-zero water use on-site, very low PUE, very high rack density. Still emerging at hyperscale: facility-wide deployments are limited and fluid handling adds operational complexity.
| Water rate | ~0.02 L/kWh (essentially none) |
| Power penalty | −10% to −20% (very low PUE) |
| Capex | Substantial premium; specialized fluid + sealed tanks |
| Best fit | Edge or single-purpose AI clusters; pilot deployments at hyperscale |
| Avoid when | General-purpose colocation where servicing patterns require frequent rack access |
| Citation | GRC 2024; ASHRAE TC 9.9 |
Water Source
Five measures that change where cooling makeup water comes from. Source-substitution measures (reclaimed, brackish + desal) do not reduce gross volume — they shift demand off potable supply.
Reclaimed / recycled municipal water Source substitution
ProvenTreated municipal wastewater (Title 22 in California, equivalent elsewhere) used as cooling-tower makeup in lieu of potable supply. Reduces strain on the potable system but does not reduce gross watershed consumption. Common in arid municipalities where reuse infrastructure exists (Chandler, Phoenix, San Antonio).
Implementation: requires a reclaimed-water pipeline reaching the site (or trucked delivery for small facilities), and tower-side biocide/scale management tuned for nitrogen/phosphorus loadings typical of reclaimed water.
| Volume effect | 0% (substitution only) |
| Power penalty | 0–2% (treatment overhead) |
| Best fit | Arid communities with established reuse infrastructure |
| Avoid when | No reclaimed pipeline within reasonable distance — trucked delivery is rarely cost-effective |
| Citations | EPA WaterSense; AWWA M58 — Reclaimed Water |
Onsite rainwater harvesting
ProvenRoof and site rainfall captured, treated, and used for cooling makeup. Climate-dependent — works much better in the Southeast and PNW than in arid SW. Storage cisterns require footprint; first-flush diversion and treatment add complexity.
| Scope 1 effect | −5% to −15% (volume reduction) |
| Power penalty | +1% to +3% (pumping, treatment) |
| Best fit | Wet climates with substantial roof / impervious surface area; greenfield where storage can be designed in |
| Avoid when | <15 inches annual rainfall, or no available storage footprint |
| Citations | EPA WaterSense |
Greywater reuse onsite
ProvenBuilding greywater (lavatories, condensate from HVAC, equipment-room drips) treated and recirculated to cooling towers. Effective at moderate scale; treatment train needs to handle pathogens and dissolved solids before tower introduction.
| Scope 1 effect | −10% to −25% |
| Power penalty | +2% to +5% |
| Best fit | Large greenfield campuses with significant occupied office / amenity space |
| Avoid when | Small or unoccupied facilities — greywater volume too small to justify treatment |
| Citations | EPA; ASHRAE — water reuse |
Atmospheric water generation (AWG)
ExperimentalMechanical extraction of water from ambient air via condensation or sorbent cycles. Marketed as a magic-bullet solution for arid regions; in practice the energy cost is severe (tens of kWh per gallon produced) and net-water-positive only in very humid conditions. Useful for niche remote applications, not material for hyperscale cooling load.
| Scope 1 effect | −5% to −20% (small absolute volumes) |
| Power penalty | +20% to +50% (very high) |
| Best fit | Pilot / forward-looking commitments; humid coastal sites with carbon-free electricity |
| Avoid when | Arid sites — power cost outweighs water benefit; the basin is what's stressed, not the air |
| Citations | Limited peer-reviewed deployment data; flag for plant-specific assessment |
Brackish / saline source + onsite desalination Source substitution
EmergingCoastal sites can substitute potable freshwater with desalinated brackish or seawater. Significant power cost for the desalination process; useful only where the alternative is severely constrained or contested potable supply.
| Volume effect | 0% (substitution only) |
| Power penalty | +30% to +80% |
| Best fit | Coastal greenfield in water-stressed regions; access to non-thermo power offsets the desal energy |
| Avoid when | Inland sites; thermoelectric grid (Scope 2 water swamps the savings) |
| Citations | NREL water-energy nexus |
Treatment
Four measures that reduce cooling-tower blowdown and total tower water losses. All are real volume reductions and compose with cooling architecture choice.
RO pretreatment (high cycles of concentration)
ProvenReverse osmosis pretreatment of makeup water enables cooling towers to run at higher cycles of concentration before scale and corrosion force blowdown. Doubling cycles of concentration roughly halves blowdown losses. The most cost-effective treatment-side mitigation for most facilities.
| Scope 1 effect | −15% to −30% |
| Power penalty | +1% to +3% |
| Best fit | Open evaporative or adiabatic systems; municipal makeup with elevated hardness or silica |
| Avoid when | Already low-mineral makeup (PNW soft water) — diminishing returns |
| Citations | ASHRAE 90.4; AWWA Cooling Tower Water Treatment |
Sidestream filtration
ProvenContinuous filtration of a portion of recirculating cooling water removes suspended solids, allowing higher cycles of concentration and longer runs between blowdowns. Modest savings; very low operational complexity.
| Scope 1 effect | −5% to −15% |
| Power penalty | 0% to +2% |
| Best fit | Most evaporative systems; pairs well with RO pretreatment |
| Avoid when | Already running clean closed-loop systems |
| Citations | ASHRAE |
Zero-liquid-discharge (ZLD)
ProvenConcentrates and crystallizes blowdown to recover almost all makeup water; produces a solid waste stream rather than a liquid effluent. Highest recovery rates but high capex and meaningful power penalty. Increasingly required in jurisdictions with stringent discharge permits.
| Scope 1 effect | −20% to −40% (net of water-energy nexus) |
| Power penalty | +5% to +15% |
| Best fit | Sites with strict discharge permits; arid regions where every gallon counts |
| Avoid when | Locations with low discharge fees and abundant water — capex doesn't pay back |
| Citations | EPA; IDA — ZLD methods |
Advanced biocide / scale management
ProvenModern oxidizing and non-oxidizing biocide programs combined with scale inhibitors enable longer runs at higher cycles of concentration. Lowest-capex treatment-side improvement; mostly an operational discipline question rather than capital investment.
| Scope 1 effect | −5% to −10% |
| Power penalty | 0% |
| Best fit | All evaporative systems; baseline best practice |
| Avoid when | Never — this should be standard. The "do you do this" question is more about operational discipline than design. |
| Citations | ASHRAE — Cooling tower water management |
Power (Scope 2)
Five measures that reduce or shift the upstream power-plant water consumed to generate the facility's electricity. Scope 2 water is often comparable in magnitude to Scope 1 on thermoelectric-heavy grids — these measures matter as much as cooling choice.
Onsite solar PV
ProvenBehind-the-meter solar generation directly offsets a portion of grid electricity demand, reducing Scope 2 water at thermoelectric plants. Capacity factor limits the share that can be self-supplied; pairs naturally with battery storage. Even small onsite arrays demonstrate commitment in regulatory contexts.
| Scope 2 effect | −5% to −20% (depends on array sizing and capacity factor) |
| Power penalty | 0% (offsets grid) |
| Best fit | Sites with land or roof for arrays; high solar capacity factor regions |
| Avoid when | Greenfield with no land; PNW winter — limited capacity factor |
| Citations | NREL — DC + solar integration |
Onsite battery + load shifting
ProvenBattery storage with peak-shaving control reduces draw from peaking thermoelectric plants (typically the most water-intensive units in the merit order). Useful even without solar pairing; reduces grid stress and Scope 2 water proportionally to peak shaving.
| Scope 2 effect | −2% to −8% |
| Power penalty | ~1–2% round-trip efficiency loss |
| Best fit | Sites on grids with high coincident-peak demand charges; AI workloads with predictable profiles |
| Avoid when | Grid is already heavily renewable (PNW hydro) — marginal benefit |
| Citations | NREL; DOE EERE |
PPA with non-thermoelectric generation (wind / solar / hydro)
ProvenPower purchase agreement physically tied to non-thermoelectric generation — wind, solar, or hydro. The Scope 2 water reduction depends on whether you take a location-based or market-based accounting view, and whether the PPA is additional. WaterMark's calculator uses location-based accounting (grid-average) by default; PPA mitigation in the Stack is intended as a credit toward Scope 2 if the PPA is verifiably additional and dispatch-aligned.
| Scope 2 effect | −20% to −80% (range reflects accounting and additionality debate) |
| Power penalty | 0% |
| Best fit | Anywhere a verifiably additional renewable PPA is achievable |
| Avoid when | If the PPA is a paper credit unlikely to displace marginal generation, the volume claim is hollow |
| Citations | GHG Protocol Scope 2 Guidance; EPA Green Power Partnership |
Behind-the-meter natural gas + CHP Plant-specific
ProvenOn-site combined-heat-and-power gas generation. Water profile depends entirely on the specific cooling design (air-cooled CT vs. wet-cooling CHP); often net-water-positive due to combustion cooling needs. Frequently presented as a sustainability move but requires plant-specific analysis to verify the water claim.
| Scope 2 effect | Varies — flag for plant-specific assessment, not modeled in the headline calculator number |
| Power penalty | Varies |
| Best fit | Sites with significant heat-recovery use (district heating, adjacent industrial process); grid emissions concerns dominate over water |
| Avoid when | Pure water-stress framing — gas combustion is rarely a water improvement |
| Citations | EPA AVERT — generation displacement |
Behind-the-meter SMR (small modular reactor) Plant-specific
ExperimentalSmall modular reactors are not commercially deployed at data center scale as of 2026. NRC licensing is in progress for several designs (NuScale, X-Energy, BWXT); first commercial deployments expected 2028+. Listed for forward-looking commitments — useful in Commitment Sheets where a developer signals long-term intent, but not a credible volume reduction in current operations.
| Scope 2 effect | Highly dependent on cooling design; nuclear typically has high water consumption per kWh |
| Power penalty | 0% (replaces grid) |
| Best fit | Forward-looking 2030+ commitments; sites with land for siting and licensing pathway |
| Avoid when | Near-term operations — not yet commercially available at scale |
| Citations | NRC; NREL — SMR water profiles |
Operations
Three measures that change how the facility runs. Not capital investments — operational discipline. Often the highest-ROI improvements; if the developer does not commit to these, they are leaving water and power on the table for free.
Raised supply temperature setpoint (ASHRAE A2 / A3)
ProvenASHRAE TC 9.9 thermal guidelines define expanded operating envelopes (A2 up to 80°F supply; A3 up to 90°F). Operating at higher cold-aisle supply temperatures reduces both cooling load and water/power simultaneously. Modern servers tolerate these envelopes; the limit is operator confidence and SLA tolerance more than equipment.
| Scope 1 effect | −5% to −15% |
| Scope 2 effect | −2% to −5% |
| Power penalty | −2% to −5% (improvement, not penalty) |
| Best fit | All facilities; especially impactful at hyperscale where 1°F = MW of chiller load |
| Avoid when | Legacy equipment with strict A1 tolerances; mixed-age rooms where lowest common denominator dominates |
| Citations | ASHRAE TC 9.9 — Thermal Guidelines |
Free cooling / economizer hours
ProvenUse ambient air or evaporator-side water for cooling when conditions allow, minimizing chiller and tower runtime. Climate-dependent; PNW and high-desert sites can achieve 80%+ free-cooling hours. The single highest-ROI operational lever in cool climates.
| Scope 1 effect | −10% to −25% |
| Scope 2 effect | −3% to −8% |
| Power penalty | −3% to −8% (improvement) |
| Best fit | Cool / dry climates; high latitude or elevation; any site with significant economizer hours |
| Avoid when | Tropical / consistently hot-humid sites — fewer economizer hours available |
| Citations | ASHRAE; LBNL Data Center program |
Real-time water + power telemetry, public reporting Compliance lever
ProvenNo direct volume effect — but the verification mechanism that makes every other commitment enforceable. Real-time facility-level water and power metering, published on a public dashboard at quarterly or finer cadence, lets municipalities and communities verify whether the developer is actually delivering on their committed mitigations. Without it, the Commitment Sheet is a statement of intent; with it, it's an enforceable exhibit.
| Direct volume effect | 0% |
| Indirect effect | Increases probability that committed mitigations are actually implemented and maintained |
| Power penalty | 0% |
| Best fit | Mandatory inclusion in any Commitment Sheet |
| Avoid when | Never. If a developer resists telemetry, that resistance is itself a signal. |
| Citations | ASHRAE 90.4 — DC efficiency; NIST — DC measurement |
Catalog covers 17 stackable mitigations (cooling architecture in the calculator dropdown plus 14 stackable measures across water source, treatment, power, and operations) plus the 5 cooling architecture choices. To evaluate a specific stack on a specific facility, open the Assessment Tool and toggle measures in the Mitigation Stack section. To export a printable Commitment Sheet exhibit for a siting agreement, click "Generate Commitment Sheet" after selecting your stack.