Bigger res = lower temps??

Hi, I'm planning on redoing my loop soon (when I get a blocked 580/680) and I'm thinking of adding a big res (custom jobbie) to help keep temps down. My reasoning is that more water = more to heat up = cooler all round temps. Is this correct??
Cheers,

 

it is correct, yet it is not.

Long story short, without the physics ect:

You're increase of water volume will only 'delay' the enevitable of the whole water loop reaching it's peak temperature.

to sufficiently be able to keep temps lower you must have a very large surface area for the heat to radiate from the 'hot' water to the 'cooler' surroundings eg: air. Or, just have higher CFM fans.

in other words, either get a seriously large res, like a beer keg and let the heat radiate into the room naturally, or get a larger radiator, which will also require more water in the loop.

I could work it out for you if i get some more details...such as

thermal output of the CPU/GPU/pump
pump's output in litres per minute
total water in the loop
radiator's surface area in mm^2


So, just a larger res, say 2litres is only just slightly delaying the heating of the entire loop

 

The best I could do for that is guess. I think I'll just shove another 240 rad in if the 680 brings temps too high, or get better fans (currently using 1x H100i stock fan and 2x enermax T.B.silence). Or go back to air cooling... HA, as if .
Cheers,

 

RS360 does my 680+2500K fine

but then my ambient temps are much lower due to location

 

maybe just throw a couple of ice cubes into the res before use, as mentioned, going to need a large res for this to happen

 

I really don't think temps will be too bad, I'm only going 20c above idle temps at full load anyway (30c-50c) and I don't really mind my CPU/GPU running at temps up to 70c. If they go over 70c, that's when I'll chuck in another rad or down clock my CPU.
Cheers,

 

Reservoir size makes no difference to temperatures.

All it does is slow down the speed at which the system moves from its idle water temp to the full load water temp.

If your loop holds 2 litres of water it takes a 300 Watt system approx 4 minutes to heat the water by 10C, an extra litre of water adds about another 2 mins.
That is assuming that you have no fans running to fight heating but either way the difference is negligible.

Although the system warms more slowly it also cools back down equally slowly.

Only extra radiator space or faster fans/more airflow will reduce temperatures.

 

The only real argument I can see for a larger res is component life. Longer time constants w.r.t. temperature will always help reduce the thermal stress on components.

One other situation where you might consider using a large reservoir is if you had a system that has something that comes completely on and then goes completely off to cool, such as a basic chiller unit. But it's pretty unlikely and for such scenarios your fluid turnover time typically needs to be greater than 5 minutes. Pretty easy to end up needing a 60-100L res the little bottles put into PC's will look like nothing in comparison !

Luckily people who use chillers just want the absolute maximum level of coolness and don't encounter this problem

 

The CPU and GPU's themselves will still pretty much instantly change temp from idle to load, ... eg 30C to 60C for CPU's
The extra few degrees added as the water changes temp are pretty insignificant compared to that. The speed that that happens might make a statistical difference to component life but I doubt it.

 

Perhaps the only exception is a reservoir like this

Even then you'd need good airflow and difference in temps would probably be negligible.

 

Pretty much bang on, time taken to heat the fluid is proportional to the volume. You see, now that I have had a chance to procrastinate,
I give you....
Little bit of awesome to lighten your day

 

So if you don't run your PC 24/7, you can theoretically get a res so big that before the water even reaches peak temp you would already shutdown for the day

Just for shits n giggles, what volume would this theoretical res need to be? Assume the PC is on for no more than 12hrs straight

 

As long as the res had sufficient heat dissipation to disperse all the heat over the 12hrs the PC was off and return the water to ambient temperature then yes it's possible. No idea exactly how much water that would take but it'd be a hell of a lot if you're talking about 12hrs straight at full load with multiple GPUs.

 

You would need more details,
Specific heat and density of the coolant and power drawn by system versus power dissipated in the radiators.
Not impossible, but not my cup of tea either

Aye, you also need to realise that this works both ways. say it took 12 hours to change the temperature of your super-massive-res by 15degrees from ambient, it would then take 12hours to cool down those 15degrees. So unless you had the water still being cooled by the rads with the PC off, you're on a pretty strict use cycle

 

Alternatively, the start of your loop could be a tap, and the end of your loop a sink... Just eliminate the need for a radiator at all. If you live in an apartment, water isn't metered

 

You sir, are a genius! .... no radiators, no pumps...