← Survey overview · Published May 12, 2026

Key Terms & FAQ

Definitions of core concepts and answers to common questions about data center exterior CFD.

This page collects the definitions and questions that come up most often in conversations about exterior data center CFD. It supplements the methodology page with terminology and context.

Key terms

Plume recirculation

When discharged hot air from heat-rejection equipment is drawn back into the same or adjacent equipment intake, raising local intake temperature and reducing cooling capacity. Recirculation can be self-induced (a unit's own exhaust entrained into its own intake under low wind) or cross-unit (one unit's exhaust entering a neighbor's intake).

Generator exhaust dispersion

The dispersion of diesel generator exhaust gas from stack and radiator outlets in the surrounding airspace, governed by stack temperature, exit velocity, ambient wind, and atmospheric stability. Stack-tip plume rise dominates under high wind; buoyancy dominates under calm conditions.

OSAU intake

Outside Air Handling Unit intake — the air entry point of a dedicated outside-air conditioning unit. In data centers, OSAU intake exposure to nearby exhaust plumes is a primary exterior CFD concern, since contaminated or hot makeup air can compromise interior environmental control.

ASHRAE 0.4% design condition

The ambient temperature (dry-bulb or wet-bulb) exceeded 0.4% of hours during a typical meteorological year. Used as the baseline ambient condition for HVAC sizing and as the canonical input for CFD design scenarios. Equivalent to roughly 35 hours per year.

Atmospheric boundary layer (ABL)

The lowest portion of the atmosphere where ground roughness, terrain, and surface heating shape wind profiles. Modeled in steady-state CFD with a logarithmic velocity profile parameterized by the von Kármán constant, an aerodynamic roughness length, and a reference height.

Realizable k-ε turbulence model

A Reynolds-Averaged Navier-Stokes (RANS) turbulence closure widely used for atmospheric and built-environment flows. Standard baseline for exterior data center CFD because of its balance of computational cost and predictive fidelity for separated and recirculating flows.

k-ω SST turbulence model

A RANS turbulence model favored for adverse-pressure-gradient flows and buoyancy-dominated cases. Often used as a sensitivity comparator to realizable k-ε in exterior data center CFD when convergence is poor or near-wall behavior matters.

Pumped-refrigerant cooling

A waterless data center cooling architecture in which refrigerant is mechanically circulated between indoor evaporators and outdoor condensers, eliminating cooling towers and chilled-water loops. The exterior discharge is a dry, sensible-only plume rather than a wet evaporative plume.

Radiator discharge

The cooling-air outflow from a diesel generator radiator, distinct from the engine exhaust stack. Typically horizontal, lower temperature, and higher volume than stack exhaust. Radiator discharge direction relative to neighboring intakes is a key CFD input.

Stack tip downwash

A flow phenomenon where exhaust plumes are drawn down into the low-pressure wake region behind a stack under low-momentum or low-temperature conditions, reducing effective plume height and increasing local ground-level concentration.

Wind rose

A diagram showing the frequency distribution of wind direction and speed at a location. Used to identify prevailing and worst-case wind cases for CFD; typically drawn from NOAA or similar weather-station records.

Building wake

The low-velocity, high-turbulence region immediately downwind of a building, where plumes can become trapped and recirculated rather than dispersing into the freestream. Wake structure depends on building geometry, wind direction, and atmospheric stability.

Frequently asked questions

How is data center exterior plume behavior modeled?

Steady-state RANS CFD on a domain that includes the building, surrounding terrain, neighboring structures, and all major exterior heat sources — rooftop heat-rejection equipment and ground-level generator stacks and radiators. The model resolves bulk plume trajectories, recirculation zones, and intake exposure under representative wind and operating cases.

What inputs does the survey use?

Public sources only — published floor plans and data sheets, satellite imagery, publicly disclosed equipment counts, ASHRAE design conditions, USGS elevation data, and manufacturer-typical exhaust specifications. No proprietary drawings or operational data are used.

What can and can't exterior CFD tell you?

It can identify the relative tendency for plume recirculation, plume entrainment into nearby intakes, and wake-driven flow patterns around the building. It cannot tell you the actual as-built operational performance of a facility, predict failure conditions, or substitute for engagement-grade engineering analysis with proprietary drawings and operational data.

How do generator exhaust plumes affect rooftop cooling equipment?

Under low-to-moderate wind, vertical exhaust columns from a ground-level generator yard can loft to roof level and be entrained into adjacent rooftop heat-rejection equipment intakes. The risk is highest when prevailing wind directions push yard plumes toward the building face below rooftop equipment, and when generator yard spacing is tight enough that radiator plumes interact with neighboring unit intakes before dispersing.

What is waterless pumped-refrigerant cooling?

A data center cooling architecture that circulates refrigerant between indoor evaporators and outdoor condensers instead of using chilled water and cooling towers. The exterior discharge is a dry, sensible-only plume rather than a wet evaporative plume, which simplifies plume buoyancy modeling but increases discharge temperature for a given heat load.

What wind cases are typically modeled?

At minimum: the prevailing summer direction at typical wind speeds (drawn from the regional wind rose), a winter prevailing direction, a low-wind or calm case (~2 mph) for buoyancy-dominated plume behavior, and a directional sweep of four to eight additional cases per facility to identify worst-case directions.

How does data center exterior CFD differ from interior CAC modeling?

Interior modeling resolves rack-row airflow and cold-aisle/hot-aisle thermal management at the data hall scale. Exterior modeling resolves what happens outside the building envelope — how heat-rejection plumes interact with the surrounding airspace, neighboring buildings, and the building's own air intakes. The two are complementary and increasingly relevant together as rack power density and rooftop equipment density both rise.

Why does rooftop equipment density matter for exterior CFD?

Higher density reduces spacing between heat-rejection units, increasing the chance that one unit's exhaust is drawn into the intake of an adjacent unit under low-wind conditions. Modern hyperscale data center rooftops concentrate substantially more heat-rejection load per square foot than designs of even a decade ago, raising the salience of exterior CFD analysis at the design stage.