Some notes on fan placement

The forum for non-component-related silent pc discussions.

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Post Reply
LM741C
Posts: 42
Joined: Thu Feb 26, 2009 7:37 pm
Location: C-c-c-california

Some notes on fan placement

Post by LM741C » Tue Mar 10, 2009 6:03 pm

Hi y'all. Some notes taken from our techlib on fan placement. My reference for this is the following:

Steinberg, Dave S.. "Chapter 6 - Forced-Air Cooling for Electronics". Cooling
Techniques for Electronic Equipment, Second Edition. Professional Engineering
Publications. © 1991.

First snippet is on laminar versus turbulent flow:
Laminar flow conditions as well as turbulent flow conditions can exist with forced
convection in liquids and air. Turbulent flow conditions are much more desirable,
because they permit more heat to be removed. However, turbulent flow will usually result
in a higher pressure drop through the system, which requires that larger pumps and fans
and more power be used to overcome the added resistance.
Second is on fan placement:
The position of the fan blades within an axial flow fan housing can be a critical factor in
determining how well a fan will perform. This is especially important in high-speed fans
that have speeds greater than about 8000 rpm.
(Given the 8K rpm speed, I'm not sure how much of the following apply to SPCR readers). Continuing:
An examination of most high-speed axial flow fans will show that the fan blades are not
located at the center of the tubular housing but near one end. When the fan is located
adjacent to a restricted area, such as a 90° bend, the fan blades should be positioned so
that they are at the downstream end of the housing, for the best performance. Air has
weight and kinetic energy, so that the air velocity must be allowed to develop to
effectively overcome the flow resistance. When the fan blades are located at the
downstream end of the fan housing, the air has a slightly longer flow path. This improves
the velocity profile, as shown in Figures 6.3 and 6.4.

Image
Fig 6.3
Image
Fig 6.4

It makes no difference if a blowing or an exhaust fan system is used; the velocity profile
must have a chance to develop to provide an efficient air delivery system.
The reduced flow efficiency for the fan shown in Figure 6.4 will not be obvious to a
casual observer. When the fan is in operation, you can place your hand over the exhaust
and feel a large volume of air flowing through the fan. However, if a thin strip of paper is
slowly passed across the fan exhaust, it will show that some of the air is being shortcircuited.
The airflow at the outer perimeter of the fan will be moving away from the fan,
but the airflow at the center will be moving toward the fan, resulting in a short circuit, as
shown in Figure 6.5.

Image
Fig 6.5

Some good fan installations and some poor ones are shown in Figure 6.6. Test data on
these types of installations have shown that the cooling airflow rate can be more than
doubled just by properly orienting the position of the fan blade within the fan housing
when the fan is located adjacent to an area that restricts the free flow of cooling air.

Image
Fig 6.6
That's all for now, hope somebody found this info useful. I saw some posts in the gallery section where they had double-decked the fans (one emptied out), looks like they're headed in the correct direction. I will conduct some experiments when I get a chance.

Emyr
Posts: 91
Joined: Wed Jun 06, 2007 10:48 am
Location: Cardiff, UK

Post by Emyr » Wed Mar 11, 2009 2:15 am

So in the bottom chamber of a P18X, a PSU whose fan is on the back may perform better than one with a 12cm underneath?

LM741C
Posts: 42
Joined: Thu Feb 26, 2009 7:37 pm
Location: C-c-c-california

Post by LM741C » Wed Mar 11, 2009 6:33 pm

Emyr wrote:So in the bottom chamber of a P18X, a PSU whose fan is on the back may perform better than one with a 12cm underneath?
Unfortunately, I really don't have a clue since fans and cooling are not my area of expertise. <speculate>I would think they're both "bad" since the one underneath (which is what I have on my P182 build) runs into resistance shoving the air into the PSU components and the one in the back is trying to pull air from a restricted area and neither will have a chance to develop peak velocity.</speculate>

Olle P
Posts: 711
Joined: Tue Nov 04, 2008 6:03 am
Location: Sweden

Post by Olle P » Wed Mar 11, 2009 11:51 pm

Emyr wrote:So in the bottom chamber of a P18X, a PSU whose fan is on the back may perform better than one with a 12cm underneath?
It's a matter of definition.

To me it's clear that a PSU design where the general direction of air flow is straight through is, at least in theory, better for cooling the components with any given amount of air flow. There are no inherently dead spots with no air flow inside the PSU.
The downsides are:
- The smaller fan will have to rotate a bit faster (louder) to create that air flow.
- The fan is mounted at the rear, where it's most audible.
- The configuration of air flow makes it a poor choice for aiding in venting hot air away from the CPU in a regular tower case.

With a large fan underneath the PSU you (theoretically) get the same air flow at a lower noise level and better venting of hot air out from the case in normal configuration.
The downsides are:
- Angled air flow provides restriction inside the case. Demands more powerful (read: noisy) fan.
- Inherently dead spots at the corners away from the exhaust. Can be overcome by adding venting holes there too, but those will blow warm air back into the case, increasing the temperature of the air used for cooling the PSU, demanding higher air flow.
- When mounted in P180/182 the limited space above/below the PSU creates a restriction to the air intake as well. (Considerably less prominent in P183, where there's more room above the PSU.)

There are a few high power PSUs that use two fans in push-pull configuration. With a well designed PSU interior and properly balanced fan control that solution could prove to be the very best for mounting in these cases, wrt noise/cooling ratio.

Cheers
Olle

Post Reply