This is short-sighted. How is the oil cooled once it absorbs
heat from the electrical components?
That old system doesn't produce much heat. It was very
nearly low enough to be passively cooled. With such a low
heat generation rate it is sufficient to simply spread the
heat out to a larger surface area (all sides of the oil and
enclosure).
All the setup has now is
convection of the oil and conduction to air at its top surface
and conduction through the glass walls of the aquarium - nothing
at all better than a forced air draft past the electrical components.
True, the whole thing is a novelty rather than a good
design. That's not a "new" system build though, I recall
that linked system from well over a year ago.... we don't
even know if it's still working today.
When he starts using some components that actually do release
appreciable heat, he'll have to start pumping the oil past the
components and then through an oil-air or oil-running water heat
exchanger (i.e. "radiator"). What he has now makes no thermal
sense.
Actually it makes perfect thermal sense. The goal in any
cooling system is not to create some hypothetical "perfect"
solution only from the standpoint of keeping components
cooler than they need to be. Temp is not a "contest", it
merely needs stay cool enough to remain stable and with
aceptible lifespan. Within the context of that particular
(Celeron 5xx) system it would only be worse to pump the
liquid- because it doesn't need a pump and adding one just
increases expense and components subject to failure.
You are correct that a significantly higher heat build would
require more attention to removing heat, but only what is
necessary towards keeping it reliable. Not that a giant
aquarium is in itself a good solution, but adding further to
it with pump and radiator with no specific, realized gain is
just traveling even further down a road not needed relative
to optimized air cooling.
