I'm seriously looking at making my first foray in to watercooling and have read numerous threads here regarding the same (BTW, thanks Nexxo!) but am a touch curious regarding a couple of things not mentioned. Nexxo has stated (and rather accurately, from the point of view of an amateur aeordynamics bod) that having the fans too close to the radiator will result in as much as an 11% reduction in surface area available to airflow owing to the size of the fan hub. Obviously, the answer is to move the fans further away but there are still losses, that would be where shrouds make a difference. I've dicussed this with an aeronautical engineer and fluid dynamics expert (my father) and he suggested creating a plenum chamber upstream of the radiator and a diffusion chamber downstream to maximise the benefit of the airflow. Admittedly, you'd need a case a couple of feet wide to achieve this or, alternatively, a lot of space on top of the case to allow for shrouding etc. I'm looking at only using a diffusion chamber (i.e. pull configuratiion with serious shrouding) with the entire radiator as an intake (no need for a plenum chanber as the airflow is already even) and adding a cone to each fan hub (RC plane prop spinners Cheesecake) to direct airflow and reduce losses. Has anyone tried this yet and should I be barking up the wrong tree can they please direct me to the right one? Pretty please?
I'd also consider myself an aerospace engineer and fluid dynamics expert (though probably not as experienced as your dad!). I'm personally working on a watercooled case at the moment. I opted not to use a shroud, but I'm size limited. Instead I'm using a deeper rad. In an ideal world I'd have used a shroud. Based on the 120mm fan I measured up, the intake area (ie ignoring hub and corners) of the fan is 9550mm^2. The area of the whole fan is 14400 (not inc mounting brackets on the rad), so using a shroud will give you 50% more flow over through the radiator that direct mounting, using very simple logic. In reality it's more complex. There's a small gap between the fans and the fins which allows some recovery of airflow to the fins not exposed to the intake. Using a shroud would match the fan to the radiator much better, but the exact length required to achieve that is a grey area. The longer the better, though improvements will tend towards a limit. Also, the longer you make it, the more boundary layer build up there will be on the shroud walls, reducing effectiveness. Any aspect of losses in the flow build up resistance for the fans to overcome. The nose cones on the fan will be effective, avoiding a dead zone over the fan centre and associated flow losses, but I doubt it will make much difference in measurable terms. Give it a try if you can. Are they pointy cones (ala Trent 900 (Cheesecake )) or rounded cones? (ie GP7000 (boooo!)) Won't make much difference, the rounded ones would be better for a shorter shroud, as very blunt cones would have loses as the flow turns into the intake. Longer, pointy cones would be fine. So, to summarise - a shroud and nose cones will help, if you have the room in your case and flexibility in your design, go for it. How much improvement will it give? The max improvement over fans directly mounted to the rad would be 50%, considering the surface area increase, though you'll never acheive that. How deep a shroud would I recommend.... 5-6cm will get you about 45% more potential cooling power, you won't get much improvement above that. 3cm will probably still be about 35% effective, which seems to be what you'd get from the comercially available shrouds. I don't really have a source for that, it's just from instinct. I'd love to see some test results on this. Hope that helps.
pistol_pete, you'd have to be seriously older to beat my father on age (75). Handley Page, Hawker Siddley and the MoD have a lot to answer for! He still works for Marshall's in Cambridge, despite the fact that he should have retired 10 to 15 years ago! He's still godlike in his own field but that doesn't necessarily involve ours. They cross over occasionally but it's not always relevant. I'm looking at the radiator mounted on one side of the case and the fans on the other, about 7 and half to 8 inches for the housing, probably only around 6" or so for airflow to be realistic (allowing for fan depth and shroud). Standard case width is definitely a contributing factor as I'd prefer not to have to get the dremel out more than necessaey. It could be messy.
6" should be more than enough for the airflow to settle after exiting the radiator. If you really want to go for the to go for the ultimately effective radiator, you should go for a push-pull config, ie fan-shroud-rad-shroud-fan. The extra pressure from doubling the fan count will help to provide more air to the increased radiator area. You should also still have enough of a gap for the shrouds to be effective.
I realise that push-pull config is going to give the best performance but I may have given the wrong impresion. Excessive fan noise is what I'm rather hoping to eliminate. It all looks nice on paper but what I was atually asking was "has anyone tried it?" If so, I'd rather like to hear the oucome. If not, I'll be the guinea pig and see what happens.
