Originally Posted by
antiacid
Now that we're talking fans and clearly blaming you for bringing up another can of worms, let's jump into it! :)
On a side note, before I start elaborating: Regarding your thick fans, how noisy were they?
Now, on to the real thing...
You know, I have been thinking that there needs to be a way to compare all of those different components in a way that makes sense to the adept. At this time, either there's too much information on the graphs or then the information isn't so clear to pick-up from that graph (which doesn't justify the actual graph). This is a general observation.
My line of thought was to try and come up with some kind of dimensionless number to discuss those radiators in a decent way. I mean, there's already a truck-ton of different dimensionless numbers for general heat transfer theory so there's no reason why we don't use them / recycle them / make up our own.
Concerning the testing of fans, specifically, there seems to be about three or four values of interest: CFM, static pressure, noise and RPM. One could also argue that the values for "free flow test" for those fans don't work on rads because of the added restriction. I'd argue that the "static pressure" doesn't change if you put a wall in front of it. It'll push the wall at exactly the same value as in free flow. I know that the noise doesn't change, the RPM shouldn't and the CFM change in relation to the "impedance" of the radiator tested.
In the classical restrictive pipe example, the pressure diminish in the smallest part of the pipe because the pipe takes momentum away from the fluid and the speed of the fluid increases in that section. In a radiator setup, that tube is very short: From the fan intake up to the radiator exit. In a typical scenario, that's about 5cm or about there. Does Bernouilli's equation works here or is it too short? The extra air would just NOT be pushed through the radiator if it was too restrictive, reducing CFM?
This brings us to an interesting value: This so called "impedance" or "restrictiveness" should be measured at different preset points (I propose low/med/high CFM, so about 35/55/75CFM) to give an idea of how air-restrictive is that radiator. In a way, this is already done by showing graphs of performance vs fan rpm vs fan type but it isn't clear how those results would vary for a fan of slightly different characteristics that didn't get tested. I'd like to get some kind of empirical rule developed for a given radiator and then expand it... One can dream :)
The reason I am bringing this up because I find it totally retarded that predictions cannot accurately be made if you were to test a radiator/fan combo with fan A and I were to use fan B, even while knowing the characteristics of both fans. The difference is not small, as shown in your graph 10c delta vs rpm: at 2000rpm, there's a 100watts+ difference (which accounts for different values of CFM/mmH20 @ given RPM @ particular fan).
BTW, you really should start labeling those graphs for easier commenting on them ;)
The goal of developing a general "impedance to air" figure would be that one dude could shop for a fan, check it's cfm/dba/rpm/static pressure and then shop for a radiator with a given "ITA" (not only am I developing a new unit/concept but I'm also shortening it!) and then plug this in a simple equation (or online calculator, what do I care) and figure out an approx. value of efficiency that he could then relate to the price paid for then fan!
I'm thinking that all fans aren't made equal. xyz CFM @ abc mmH20 is really the more "correct" way of thinking about fans and it shouldn't be reduced to an RPM figure. It is about as useful as the label when comes time to make a decision on what to buy. That's like buying a pack of gum by the color it is and hoping that you'll get the taste you want.
I am suggesting that this ITA value for radiators would work as a coefficient to moderate the values given by the vendors (tested in free flow setups). That way, you'd get straight up an idea of how much static pressure and CFM is actually going towards cooling your computer! I have no idea how a curve of ITA vs RPM would work for a given fan though, so I can't really guess at the form of the equation. I think that once a stable relationship is established, it could be possible to develop an analytical solution that would solve efficiency of forced flow convective heat transfer against flow parameters.
Maybe I'm re-doing what's been already done though. My heat transfer knowledge isn't as good as it used to be (and that already started not that good) so maybe I'm just really coming up with another way to talk about Péclet number or something. I'll do some reading tomorrow morning and see if there's some theory that could be applied in this scenario and that would satisfy my need to quantify everything :P
My end goal would be to develop a program in which you input your fans, rads, cpu, gpu, avg length of tubing, etc, and then automatically get the values around the loop :) Sure, testing needs to be done for every single parts as they come out but once it has been tested and validated, that's all there is to it! I wonder if the watercooling scene is ready for such a unifying vision :P
Anyway, that's my wall of text for tonight and I'm not 100% certain of everything I said in there so I'm really hoping that you guys pick it apart and we can figure out if this "impedance to air" concept is really worth developing :)
