David Maynard said:
Kony is correct in that you are mixing micro and macro effects
and then assuming the tradeoffs are the same, but they're not.
Turbulence is turbulence. Nothing "micro" and "macro"
about it, except that they are neat terms.
...the primary concern in macro case flow is getting the most
volume of air in and out of the case, and to the parts to be cooled,
with the least wasted energy.
You fashion your "facts" to your conclusion. The primary concern
*for YOU* is getting the most volume of air into and out of the case.
And you assume that by doing that, you are getting the maximum
amount of cooling for the heated parts. But merely running air past
a part does not bring it into close contact with the part. There is still
the boundary layer of air around the part with which to contend.
Simply running a greater volume of air past a part is a brute force
approach to cooling. Adding turbulence to the fluid (in this case air)
that passes the heated part aids the fluid to scrub away that boundary
layer and makes the passage of fluid more efficient in carrying away
calories - if "efficiency" is measured in calories/volume of fluid-per-second.
Your assumption that the addition of turbulence is a "waste of energy"
is presumptuous.
Or, put another way, the air has to get there (and
then removed) before it can be used for cooling purposes.
To put it a better way, the air has to get there and be put into
close contact with the heated part in order to cool the part
efficiently. If the air has merely laminar flow, it will not effectively
or efficiently contact the part, but merely pass by it, with the
part's boundary layer still insulating the part.
Take an extreme example to illustrate the point, like an air-conditioned
computer center where the air-conditioning is floor ducted around the
room and then up into the individual racks for cooling. Clearly,
'turbulence' in the floor ducting does nothing to 'improve cooling' but
does lower the overall airflow to the racks, or causes one to employ
larger, more powerful, nosier, more costly, fans to over come the
resistance caused by the useless turbulence.
That is a good example of turbulence wasted to cool the wrong thing.
Clearly, the purpose of running cold air under the floor is not to cool
the underfloor cabling. Instead, the underfloor should be designed to
maximize laminar flow, and the turbulence should be maximized as
the air approaches or enters the racks to be cooled.
Similarly, creating extraneous turbulence inside the case simply
impedes airflow.
"Extraneous" means "extra, unneeded". So your description
is designed for your conclusion. But well-designed turbulence is
turbulence that impinges primarily the parts to be cooled. You
could accomplish that by "turbulators" that induce the turbulence
at carefully selected upstream points. This is a technique used
in aircraft design to control the point of stall onset in wings and
control surfaces. The slight reduction in lift or the slight increase
in drag is considered to be worth the added control predictability
for low speed handling. Similarly with PC case ventilation - the slight
increase in drag caused by turbulence generation can be offset by
the increase in cooling capacity. But if that makes you worry, you
can actively add turbulence with an internal fan - such as is currently
done to cool CPU heatsinks and GPU heatsinks, and there will be
NO increased drag due to energy used to generate that turbulence.
But if you're willing to design for passive turbulence, you could position
parts to minimize the increase in drag which would occur from the
passive generation of that turbulence. Since turbulence can be produced
by sharp-edged holes in flat plates, the opportunity arises to use the
inevitable turbulence produced at the average case's air intake holes -
which are almost always just holes punched in metal sheets. There is
lots of turbulence there and downstream of it, so by putting hot parts in
that turbulence, one could make maximum use of it before it gets
dissipated by drag from contact with other parts in the case.
Such an opportunity occurs for cooling the main hard drive, as in my
Dell computer. The intake holes are the standard holes punched in the
face of the metal case. Since the hard drive is mounted vertically
immediately behind these holes with its circuit board facing the holes,
the hard drive gets the maximum benefit of the inevitable intake turbulence.
Despite the considerable freedom to position the hard drive elsewhere
in the bottom of the case, the designer put it where it is and oriented it as
he did. Elsewhere low in the case would have gotten the same amount
of fresh air flow, but turbulence would have been cut down by the air's
drag against other parts and with the case wall. If you're willing (and have
the room) to install ducting or tunnels within the case to carry intake air
directly to regions of the case, you could postpone the turbulence prod-
uction until just upstream of a part to be cooled. The air could then be put
through a sieve consisting of holes in a flat plate, or it could be given a
swirl by carefully designed vanes or "dragon's teeth" as is done in aircraft
design. By virtue of the ducting, the air would be forced against the sieve
or turbulating vanes, and it would not just go around those "turbulators".
Except for very hot and critical parts, are ducting and "turbulators"
necessary? Quite probably not. By careful positioning and use of
heatsink fans, just "getting the air in and out" has mostly sufficed.
But should one do anything to maximize laminar flow? Unless that
laminar flow is going only past parts that don't need cooling, NO -
leave the air turbulent. Should one maximize turbulent flow if most
of that turbulence will impinge heated parts? YES.
You want useful turbulence at the dissipating surface but a low
resistance path getting the air to it and back out.
Yes, I agree with that. If one were relying on passive generation
of turbulence, one would have laminar flow until just upstream of
the part to be cooled, and a resumption of laminar flow just
downstream of it. The part itself should be bathed with turbulence.
But how does one revert turbulent flow to laminar flow without just
passively relying on time and viscosity to do it? Practically, the
best one can do is to just keep other parts out of the way of the
turbulence as it comes off the heated parts. The placement of the
CPU's heatsink near the exhaust vent of the case makes that easy.
The placement of the typical GPU heatsink makes it a little harder.
It's not a 'recirculating' fan, per see. It's purpose is to increase the
speed of airflow at that point where it's needed because the
alternative is a gigantic heatsink, which creates it's own practical
problems.
Again I agree. I mention "recirculation" to describe what happens,
NOT to indicate the PURPOSE of internal fans. The purpose of
internal fans, such as those which blow air directly against the fins
of a heatsink, is to increase volume/second of the local air flow.
Both the increased speed and the increased turbulence work to
increase the amount of heat transferred from the hot part to the air
flowing past it. The increased speed increases the velocity gradient
of the boundary layer of air - "thinning it" in effect to increase the
transport of heat across that boundary layer - and the increased
turbulence helps to disrupt and cut through that boundary layer. But
the fan does not increase the bulk air flow through the case. It merely
increases the local air speed - which is another way of saying that
it increases the turbulence because the air is kinetically energized
without any increase in overall translational speed. In other words,
the fan is a "turbolator". It's an active turbulator because it receives
electrical power, but it is a turbulator nonetheless.
It's actually unfortunate, which is why exterior ducting is sometimes
used, and is why it's *purpose* is not to be a 'recirculating fan', per see.
You're 'stuck' with that aspect and hope the case designer gets the
air in and out fast enough to ameliorate the problem.
By pointing out that the air is "pre-heated" or "used", I am NOT
condoning such pre-heating or use, it is to point out that the
re-circulation which occurs is INCONSEQUENTIAL to cooling.
*Konehead* is the one who calls air "pre-heated" or "stagnant"
if it has contacted a heated part multiple times. Re-read the thread,
and you will see that. What I point out is that DESPITE such
"pre-heating" or pre-use or re-circulation, internal fans that blow
against heated parts (such as a heatsink fan) still aid in cooling
those parts.
Semantics. It's a means of directing airflow
If by "it" you mean a fan which blows against a heatsink,
"it's" purpose is to increase the local air flow through the
heatsink. You said that in a statement just above.
and while one can 'analogize' it to 'turbulence' by making a
stretch of comparing what 'laminar' flow through the case
would otherwise look like is to ignore the purpose of it.
Sorry, there are too many "its" in your sentence. Perhaps
you could restate it without the "its". As for the effects of
an internal fan, the effects are identical to turbulence - to
increase the probability of contact between each air
molecule and the heated surface.
And there you are incorrect. It makes all the difference...
Please explain why turbulence generated 1/8" upstream
is different in effect from turbulence generated 1/4" upstream
which is different in effect from turbulence generated 6"
upstream. Why should the turbulence care where it was
generated?
and, no, it's not 'just turbulence'.
'Turbulence' is a 'local thing'. Or, put on the other end of the extreme,
everything looks like 'turbulence' at the scale of the 'universe'
regardless of how laminar you think it is where you are. So saying "it's
just turbulence" means nothing unless you consider the scale and what
you're trying to accomplish.
"Turbulence", as I have been using the term, is localized increase
in fluid flow without an increase in bulk fluid flow. The granularity
that I have assumed for the turbulence ranges from microscopic
to vortices measuring up to an inch in diameter. By "bulk", I mean
volumes on the order of one or two cubic feet. Most of the energy
of the turbulence generated at the case's air intake holes are
probably in vortices measuring 1/32" to 1/4" across. Judging by
the persistence of vortices in such things as smoke rings or
blown candle smoke, these vortices probably last for a few to many
seconds - long enough to traverse the interior of a PC case. Thus,
turbulence generated at the intake vent holes would persist during
the air's transit of the case.
Which is why it's essentially useless in a case, as you put it,
"six inches away" from what you want to cool.
You apparently assume that turbulence decays during the one
to five seconds that it takes air to travel across the case from
the vent holes. Why do you believe air to be so viscous?
Mother nature is not concerned with the 'efficiency' of wind
creation relative to how cold you feel...
How "cold you feel" is a function of how much the cooling is
aided by the air speed and ambient humidity. Disregarding
humidity, the chilling effect of air speed on people is exactly
the chilling effect on inanimate objects.
Bad assumption.
WHY?
All that proves is it does a good job of cooling the hard drive, not that
they put it there to 'create case turbulence'.
Read again. The hard drive's placement wasn't to "create case
turbulence" - it was to take advantage of it.
And that should give you a clue that how well one can get the air in and
out is of importance.
"Of importance" does NOT mean "primary importance" or
"most important". Getting air in and out is "of importance".
So is maintaining turbulence "of importance". And striving
for laminar flow throughout the case in to be avoided. That
is the point - turbulence is not bad for cooling. In fact, it's
good for cooling.
The problem is the broad sweeping generalization...
Turbulence is not bad for cooling - that's the subject of the thread,
and that's my point. Turbulence may reduce bulk air flow if the
turbulence is generated passively, but it increases cooling
enough to make engineers design it into their systems (see
the numerous links that I've provided). You don't see them
designing it OUT of their cooling or heating systems except
in long ductways where the purpose is to transport the air,
not to apply the air for heat transfer. Where the heat transfer
is to take place, they always want turbulence.
But turbulence just anywhere is not necessarily a 'good thing',
as has been previously illustrated.
Turbulence is good where heat transfer is to take place.
Where it's generated is inconsequential to its task of heat
transfer. That's because turbulence doesn't care where it
is - it works the same "here" as "there".
Nope. Just as pouring water on a wood fire at the lumber yard helps
extinguish it but pouring water over the whole state for the same fire
isn't an efficient use of resources even though, to analogize your logic,
'water is water'.
If all the water flows to the same fire, it doesn't matter whether
the water enters the hose from the fire hydrant next to the fire or a
block away - the water still gets to the fire. Similarly for turbulence -
if the distance isn't great enough so that the time of transit isn't
so long that viscosity dissipates the turbulence, it doesn't matter
where the source is. Implicit in all your objections is that turbulence
dissipates in milliseconds and that it cannot traverse a significant
portion of the case before dissipating. Everyday experience
contradicts that.
Component level turbulence pretty much takes care of itself, either by
nature or the component designer's intent, so the case designer is
primarily concerned with getting the air in, through (to all needed areas),
and out efficiently.
Tell that to those PC users who have repeatedly failing hard drives.
If the simple placement and orientation of a hard drive can take
advantage of intake turbulence for cooling, why not use it? Why
strive for "smooth air flow" if it's detrimental? Again, the subject of
the thread is "turbulent flow is NOT BAD for cooling".
*TimDaniels*
