The evaluation in the blog is correct in asserting that turbulent flow is much better for heat removal, and the reason goes a bit further as to why laminar flow does not pick up heat as well. In a truely laminar flow situation the fluid molecules next to the surface may not be moving at all, and the next layer are moving very slowly, with a classical velocity gradient up to the fastest moving molecules, which are usually those farthest from the wall. The result is that heat is primarily transfered to the air br conduction through the stagnant layers. This is the mechanism of laminar flow's poorer performance.
Not the most exciting explanation in the world, but some useful background stuff.
William K.'s response is correct. Going a little farther, once the heat is in the plenum air laminar vs turbulent doesn't matter so much. What matters most is the volume of heated air leaving the system. High volume usually means high speed and turbulent.
. I call the flow inside a PC as the path of least air resistance which is generally missing all the hot spots well above the surface.
If a well designed plenum existed to cover the motherboard, air could be forced at a much more efficient lower volume and higher surface speed. The noise is usually on the exit vents and fan blades and not on the paths in between. Which is where my Spoiler design simply force air down under a thin cover sheet over circuit board (open frame PSU in my case)
The 1U high open framed PSU was about half the width of the 19" wide rack.
A folded cover using cut sheets of mylar to make a plenum of smooth air to flow into the dual fans near one end were also covered to make a duct to push air out the vented side. This plenum supports the inertia of air force it in one direction from one end to the other. Although not pure laminar flow, it was fairly uniform in direction and flow.
What made this design unique was restricting the intake area with a sloped vent, to a) increase the linear speed and setup a large eddy current of air to flow fastest at the surface where the hot spots started and continued to the middle of the PSU.
The position of the sloped plastic cover or spolier served to disturb the air mostly on the input and was more uniform albeit not pure laminar towards the intake to the fans.
Thats the best explanation I give without a diagram or more words.
As I said - I doubt you have laminar flow. Fully developed laminar flow in a duct has a parabolic profile. In a smooth duct you don't get a lot of vortex flows because there are no sharp corners to generate it.
Laminar flow has a very specific definition.
It is well known that true laminar flows which occur at the leading edge of a streamlined solar car are good for drag but bad for heat transfer.
Roughly, more turbulence and mixing the better the heat transfer and the larger the drag.
I understand Laminar as a Streamline of uniform flow.
This is how I viewed the bulk of air flow ( revealed by smoke test) flowing over the top of the board mounted components.
The vents on either side for intake and exhaust are turbulent as well as near the fans but owing to the geometry of the PSU being like a rectangular pipe, a fair amount of inertia supported the laminar flow.
Adding the "spoiler" flap on the intake, disturbed the smoke pattern for less than half of the channel and then became more uniform or Streamlined or Laminar.
Although technically it would have to be proven as such with Reynold's numbers. The added drag was not signifcant to me as the exhaust air flow was not reduced significantly, which I measured crudely using suspended paper to measure the deflection, as the temperature rise was improved dramatically.
I experimented with an optimal slope on angle and size of flap for air intake to get the most effective temperature reduction (down stream) while minimizing the change in exhaust pressure drop from drag.
I could only afford 1 day to resolve this issue, given an agressive 7wk schedule from paper napkin concept to customer delivery. Where I was responsible for schematic design, conceptual sheet metal design layout, EMI/safety design to pass UL/CSA and -80dBm Telecom crosstalk for 1.544Mb/s T1 data streams on each 16 channels on front panel.
THe origami ribbon design was fun too.
How route a 40 wire ribbon cable inside a box to interconnect a 4" wide ribbon in a 1.5" space with 6 folds for $5 using custom tooling ( which I sub-contracted to a friend who had the skills & tools in his shoe repair shop! ) He use piano hinge mounted jigs for precision folding, heat guns, Locktite adhesive drops, a precise anvil press for the crimp connector and heat press for the finished product, so both ends after many 90 degree folds were perfectly square to insert in headers. ( One header for PCB and the other for front panel connector adapter. ) He then built 2000 sets for initial build in a week.
Did I mention no one knew I lit a cigarette at night in the office to prove the laminar/turbulent concept, where there was good ventilation? ( The lab was too close to production).. except those few I have told.
I should have made my problem sound more vexing...
I ran the twin fans at 10% over voltage and the only thing that happened was the piano sound board of the rack mount lid just buzzed like crazy and never got cool enough. I could not afford nor fit any larger fans in the design. After racking my brains, I tried the smoke test to verify laminar flow was ineffective.
My dilemna was how to create a turburlent wave down the surface area of an open framed power supply no next to no cost. The Spoiler-like fold on the front end of the polycarb cover. Not quite like a Nascar front bumper but, you get the idea.
I think Design challenges are not allows so dramatic but can be very difficult in finding the best performance, cost and reliability. This should be valuable for those just graduated as well as those with many past memories.
Consider the medium that is being used to remove heat, AIR. The molecules in air are relatively far apart. With laminar flow, very few air molecules will come into contact with the hot surface. Turbulent flow increases the number of molecules that comes into contact with the hot surface and this leads to greater heat transfer. Turbulent flow consumes additional energy, which adds to the ineffiencies to the air flowing system. Some of this additional energy turns into sound, hence noisier. Most of the additional energy will turn into heat. One observation I have noticed on computer equipment cooling is the direction the air is moving thru the cabinet. On a majority of equipment, the fan is position to pull air thru the cabinet. This will lower the air pressure inside the cabinet, with the resulting lower density of air. The net effect is to increase the amount of void spaces between the air molecules. A better system will be to push air into the computer cabinet, increasing the density of the cooling air inside the cabinet. Another benefit of this concept is the ability to pressurize the cabinet with filtered air, keeping the inside of the cabinet cleaner.
We were asked to provde power to a regenitive blower to cool LCD cockpit display demo. I put in a air flow interlock that would cut off LCD power without the air blowing. The LCDs were turning purple and red from heat. The packager asked me about sucking the air out. The regenitive blowers data sheet indicated practically the same either direction. The blowing into the LCD was concentrating the heat. Vs sucking room air into the LCD. I had to flip over the air flow interlock. Determinig the vane size for the micro switch was interesting.
I do apprectiate the up front look at design issues like this. It's one thing to go through the problem solving and deduction of what went wrong that is commonly demonstated in the Sherlock Ohms articles. But I also appreciate the up front approach used in this article. Anything that provides good insight, basic engineering principles used in a way that is ingenuitive (sp) and is enjoyable to read is apprectiated.
Focus on Fundamentals consists of 45-minute on-line classes that cover a host of technologies. You learn without leaving the comfort of your desk. All classes are taught by subject-matter experts and all are archived. So if you can't attend live, attend at your convenience.