What about using normal house radiators?

The alternative to direct air cooling

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Post Reply
Gnep
Posts: 23
Joined: Thu Apr 01, 2004 3:54 am
Location: UK

What about using normal house radiators?

Post by Gnep » Fri Jun 11, 2004 1:05 pm

Has anyone ever tried using a normal household radiator in a watercooling setup? I may be being naive, but surely they could dissipate as much heat as a Reserator at about a tenth of the price? OK they are designed for significantly higher temperature differentials, and you might well need a beefy pump, but what about it?

I mean rads like this:
Image

Input appreciated!

Thanks

Gnep

MoJo-chan
Posts: 167
Joined: Wed Apr 30, 2003 3:49 pm

Post by MoJo-chan » Fri Jun 11, 2004 2:55 pm

It's an interesting theory. Once nice thing about these rads is that they have a bleed valve too.

It all comes down to how well water will flow through one of these things. To be honest, I'd avoid the unknown and just get a car radiator (not a heatercore, the actual radiator that sits in front of the engine). Car rads have been used before sucessfully.

Gnep
Posts: 23
Joined: Thu Apr 01, 2004 3:54 am
Location: UK

Post by Gnep » Sun Jun 13, 2004 3:53 am

True, but then these are absolutely designed to be passive - car radiators will work best with forced air, which is what I am trying to avoid...

Nowhere_man
Posts: 190
Joined: Tue Mar 23, 2004 5:50 am
Location: USA

Post by Nowhere_man » Sun Jun 13, 2004 6:56 am

I've always wanted to try using one of these :shock: to liquid cool a computer.

I think you're radiator idea has merit.

Gnep
Posts: 23
Joined: Thu Apr 01, 2004 3:54 am
Location: UK

Post by Gnep » Mon Jun 14, 2004 4:24 am

OK, I've done a little bit of research. It would seem that the rating (in W) under British standards is for a delta-T of 50 deg C:

http://www.plumbingpages.com/featurepages/BSEN442.cfm

Now for a typical watercooling system in a PC, we would probably be looking at more like a delta-T of 15 deg C between room and water/radiator. Now comes the tricky bit - how to scale down the heat dissipated for the reduced delta-T. Digging around my old textbooks (it's been a long long time), I reckon that heat dissipation under natural convection in air is proportional to about delta-T^1.25 or so, assuming all other properties (height, area) of the radiator are the same. This would also be assuming that the heat transfer between water and steel radiator will never be our limiting factor... It is also consistent with the above page's mention of a move in the standard from delta-T of 60 to 50 deg C reduced heat outputs by just over 80%.

Thus for a delta-T of 15, I reckon a radiator rated at 500W would be good for just over 100W from the computer. Which should be good enough for my current CPU + GPU. And that's one of the smaller radiators too. So it could well be worth trying... and as they only cost about 20 quid nothing to lose really... :) (apart from the cost and hassle of fittings etc.). Might well do so in a couple of months.

Rusty075
SPCR Reviewer
Posts: 4000
Joined: Sun Aug 11, 2002 3:26 pm
Location: Phoenix, AZ
Contact:

Post by Rusty075 » Mon Jun 14, 2004 6:05 am

Yeah, I can't think of a reason why it wouldn't work. The issue you're likely to run into is flow rate. Those rad's have lots of linear feet of tubing in them, much more than a typical PC radiator would. To keep the flow rate at a reasonable level you may want to plan on running dual pumps.

Ideally you'd find one of the house radiators that use copper tubes instead of steel. That would improve performance and reduce the corrosion, but may counteract the cost savings you're looking for.

Post Reply