CRAC and CRAH identify different air-handling interfaces, not complete data center cooling designs.
In the examples described by the U.S. Department of Energy, CRAC commonly refers to direct-expansion equipment using a refrigerant path, while a CRAH uses a chilled-water coil. This article follows a hypothetical rack row to show what to trace after the air handler: airflow at the racks, the refrigerant or water loop, and the equipment that ultimately rejects heat.
Data center fundamentals series
Short answer: CRAC and CRAH are clues about the air-handling path
CRAC and CRAH are not simply two names for the same equipment. In the DOE guidance examples, packaged direct-expansion, or DX, cooling equipment is commonly described as CRAC equipment. A CRAH is an air-handling unit that uses chilled water.
That distinction is useful, but it is only the first step. A CRAC label suggests that the room air is being cooled through a DX refrigerant path; a chilled-water CRAH suggests that room air passes across a chilled-water coil. The actual refrigerant or water interface should still be confirmed from equipment documentation and the connected piping system.
The name alone does not establish the cooling architecture for the whole facility.
U.S. Department of Energy — Best Practices Guide for Energy-Efficient Data Center Design
A DX CRAC does not necessarily mean the facility rejects heat only to air
DX identifies an indoor heat-transfer path, not necessarily the final heat-rejection method. DOE describes air-cooled CRAC configurations with remote air-cooled condensers. It also describes CRAC units with water-cooled condensers that may be paired with outdoor drycoolers, as well as configurations that use cooling towers.
So the statement “this is a CRAC facility, therefore it is air cooled” skips important parts of the system. Separate the indoor DX unit from the condenser-side arrangement. Until the condenser type, outdoor equipment, and any water loop are identified, the final heat-rejection path remains uncertain.
A chilled-water CRAH does not describe the entire central plant
A chilled-water CRAH provides an important fact about the air side: it uses a chilled-water coil. It does not, by itself, say how chilled water is produced or where the captured heat ultimately goes.
Where a central chilled-water plant is involved, trace the downstream path separately: CRAH coil, chilled-water loop, chiller, condenser-water side, and then a cooling tower or another heat-rejection device. The actual configuration must be checked in facility design records and equipment documentation. The presence of a CRAH alone is not proof that a particular chiller or cooling-tower arrangement exists.
Worked example: trace a hypothetical rack row instead of stopping at the equipment label
This is a hypothetical example, not a measurement or description of an operating facility. Assume that servers in one rack row draw air in at the front and discharge warmer air at the rear. Whether the room contains a CRAC or CRAH, the useful investigation is to record the heat path in order rather than to treat the unit name as the conclusion.
1
- Boundary and question to examine
- Are rack intake air and server exhaust air mixing or recirculating?
- What the CRAC or CRAH label cannot answer by itself
- The label does not show the condition of airflow at the racks.
2
- Boundary and question to examine
- What supply-air condition does each air handler provide to the row?
- What the CRAC or CRAH label cannot answer by itself
- Installed equipment does not by itself prove the rack intake condition.
3
- Boundary and question to examine
- Is the air handler connected to a refrigerant circuit or a chilled-water circuit?
- What the CRAC or CRAH label cannot answer by itself
- Naming conventions alone do not confirm the installed interface.
4
- Boundary and question to examine
- After that circuit receives heat, where does it go: a condenser, chiller, drycooler, cooling tower, or another device?
- What the CRAC or CRAH label cannot answer by itself
- The indoor unit name does not fully describe final heat rejection.
DOE guidance emphasizes air management that minimizes mixing and bypass between cooling air and server exhaust air. It also identifies monitoring at IT-equipment air inlets and monitoring of supply-air temperature and humidity for each CRAC or CRAH. The practical distinction in this example is therefore threefold: rack-side air conditions, the heat-transfer medium at the air handler, and the connected plant boundary.
Why neither name establishes cooling performance or PUE
CRAC and CRAH are equipment descriptions, not performance grades. PUE measures supporting-infrastructure efficiency as the ratio of annual data center facility energy to annual IT-equipment energy. It does not represent the performance of one air-handling unit, and it does not define the efficiency of every activity or service delivered by a data center.
Cooling-system efficiency can also be evaluated at a different boundary: average cooling-system power in kW divided by average data center cooling load in tons. Two facilities can both use equipment described as CRAC or CRAH while having different airflow management, fan and pump energy, chiller or condenser arrangements, operating conditions, and cooling loads. The shared label is therefore not evidence of equal cooling performance.
Direct liquid cooling is a separate system-boundary question
Direct liquid cooling should not automatically be placed on either side of a CRAC-versus-CRAH comparison. DOE describes direct liquid cooling as approaches that transfer waste heat to a fluid at, or very near, where the heat is generated instead of first transferring that heat to room air for conditioning.
Some liquid-cooling arrangements are hybrid systems: liquid removes most, but not all, of the IT heat, while traditional air cooling handles the remaining load. The relevant question is therefore not only which room air handler exists, but whether heat is captured near the IT heat source and what residual air-cooling path remains.
The useful conclusion: follow heat to its final rejection boundary
The CRAC–CRAH distinction is valuable because it distinguishes a DX refrigerant path from a chilled-water coil on the air-handling side. It does not determine rack-air mixing, chilled-water production, condenser-side design, final heat rejection, or facility efficiency.
When reviewing a facility description or design proposal, ask the next question after identifying the air handler: what medium receives the heat, and which connected equipment carries that heat to its final rejection point? That trace makes it harder to mistake an equipment acronym for evidence of a complete cooling design or a performance result.
No comments:
Post a Comment