Re: Ambient temperature control

On 7/2/2024 7:33 AM, legg wrote:

[Gotta wonder why data centers in northern latitudes don't
exploit outside air more agressively during the winter
months!]

>
What are the HVAC costs in data processing and server facilities?
 That's just to maintain ambient <40C.

But datacenters have big heat *generating* loads that they are
trying to offset.  So, you would expect to have sufficient
cooling (and the costs thereof) to remove that "added" load.
Imagine setting a device out in the desert.
Or, *in* the passenger compartment of a parked car.
Or, in an uninsulated attic.
The device's dissipation isn't a significant factor in the
ambient temperature that it experiences -- the "ambient volume"
around it is sufficiently large that it doesn't add to the
problem.  Adding a *fan* (in the device) won't do squat to
improve the situation.
If you're just trying to control the temperature of a component,
you have different options than if you are trying to control
the ambient temperature that a "device" experiences.
If you're trying to ensure the ambient for the device is such
that it "encourages" reliability, then you have a different
problem.  Think of the environment your KWHr meter experiences;
the builder chose its location -- likely without any concern
over sun exposure, etc.
In the 60's, one could state that you needed a particular
environment to operate a particular device (e.g., "a computer
room" for the computer).  Nowadays, your device has to tolerate
the environment (e.g., factory floor -- even in factories that
have few "organic" occupants that could bias the ambient
towards a more comfortable level).  *Or*, modify the "local"
environment to a degree that lets it achieve its performance/longevity
goals.