Parallel Radiators: no waterblocks

Charlie Hanson · 8 Nov 2017 at 14:20

My new build allows for a few different radiator options. I'm hoping somebody here has some legitimate technical advice, or cold hard experience, to help me decide what to do.

I have the space for any of the following combinations:

(A) 4x 120 radiators: EK's SE, HWLab's GTS, or any other radiator less than 32mm;
(B) 4x 140 radiators: only EK SE will fit (140mm wide) as the GTS is too wide (153mm);
(C) 3x 140 radiators: allows for wider radiators such as the GTS
(D) 2x 240 radiators: again, EK's SE and HWL GTS.
(E) 1x 520 radiator: there are some that fit, but not a great deal of details on their performance exist
(F) 1x 480 radiator: plenty will fit, but again their performance is hard to judge

The question is: do I go for a single radiator, multiple in series, or multiple in parallel?

'Parallel vs Series' has been debated forever, with neither side victorious. However, in my situation there is a catch: I won't have any waterblocks in the radiator's loop. My components are on one loop with their own pump, and the radiators will have a separate pump. The two loops will share a reservoir. The logic behind this layout is:
1) so I can blast water round the 'hot' loop at whatever speed gives the most heat removal, i.e. whatever makes the water the hottest the fastest. I expect that to be on the upper scale of speed;
2) so I can nudge water round the 'cold' loop at whatever speed gives the most heat removal, which I expect to be on the lower speed scale;
3) that rather than one D5 doing all the work I can get away with a DDC for the hot loop and an even smaller pump for the cold loop, as the flow rate in the cold loop doesn't need to be very high;
4) this allows for parallel radiators without worrying how it affects the flow in the hot loop and, in my head, parallel radiators will give me greater cooling power.

I have used a popular 2016 Radiator Roundup to choose the HardwareLabs Black Ice GTS series as the best Watt dissipators given their performance against other slim radiators such as the Magicool G2 and Alphacool's offering. Sadly, there is no comprehensive inclusion of what EK has to offer, namely the SE range, which is why the first three combinations above are there. From what I've gleaned from the aforementioned roundup, (A) and (C) will transfer the same heat given a 2-3% error margin. (B) sounds attractive, but I lack the data on EK's SE range to know for definite.

Any advice? Has anybody done or seen a similar hot-cold loop? Has anyone measured the performance of the same radiator in parallel vs series? All thoughts welcome.

skyripper · 8 Nov 2017 at 14:34

~~What do you mean by no waterblocks? The waterblock is there to take the heat off a component (CPU, Graphics card or motherboard) and put it into the water.~~

Edit: ok, no blocks on the radiator loop... and no radiators on the waterblock loop... I have no clue why you would want to do that though.

And what do you mean by a hot loop and a cold loop? The water is going to mix in the reservoir and equalize the temperature across both 'loops'.

Confudes am I.

Edit: still confudes

Radox-0 · 8 Nov 2017 at 14:46

Okay, so what your saying, is at the bottom of a resivior, one path will have water flow to a pump which will push liquid around the actual hardware block's, then a second path from same res go to a second pump which push's liquid around the radiator loop. If so I would agree with Skyripper, the water would all surely be mixing in the reservoir anyways.

In terms of radiator options, I would ignore A, B, C and D to be honest. A single large radiator would be the way forward IMO in the same loop as water blocks. What components are you looking at? Depending on your loop, a single pump is more then enough for most peoples loop while providing solid cooling whereby the water temp is very good.

pastymuncher · 8 Nov 2017 at 19:01

I think you are making things extremely complicated for no gain at all. Just have a single loop with a decent 360/480mm radiator inline and it will work fine and look a damn sight neater too.

VKleita · 8 Nov 2017 at 19:35

My head hurts.

More pumps add heat to a loop more pumps = more waste heat.

If the old loop is feeding in water at say 1 litrea minute but the hot loop is pulling water at 2 litres a minues, wont that mean half the water being sucked into the hot loop will actually be from the hot loop with zero cooling applied to it?

Jokester · 8 Nov 2017 at 19:40

Radiators are their most efficient with a bigger deltaT across them. Having them in parallel is better.

Holst1981 · 8 Nov 2017 at 20:09

Jokester said:
Radiators are their most efficient with a bigger deltaT across them. Having them in parallel is better.

This is how I run my rads as long as the piping is sensible.

As for dual loops, personally I wouldn't bother.

The efficiency of the rads is barely influenced by flow rate, and any gains will be tiny.

A single loop only needs one pump, or two gives redundancy.

If your doing it because it's fun or will look cool... That's OK

Just don't expect any real performance gains and post some photos.

The largest influence real temps is airflow.
If you want to overthink something, overthink fitting the rads and getting the airflow right.

But this really is case dependent.

Obviously more surface area is better..

Safetytrousers · 8 Nov 2017 at 20:19

Jokester said:
Radiators are their most efficient with a bigger deltaT across them. Having them in parallel is better.

Given that 'In water cooling, DT is simply the difference between the ambient air temperature and the water temperature on the outgoing side of the radiator' I don't know what you mean.
http://www.overclockers.com/guide-deltat-water-cooling/

I've had a CPU and GPU loop with 2x 240 radiators (with a total of 4x 120mm fans), and the same CPU and GPU with one thick 420 radiator (with 3x140mm fans). The temperatures were/are identical with both set ups. There comes a point where adding any more radiator estate will yield no difference in cooling potential - I guess I reached maximum potential each time. But bigger and more fans allow you to run them slower and quieter. These were/are a single loop, maybe parallel loops can get better results, I don't know.

Jokester · 8 Nov 2017 at 20:41

Safetytrousers said:
Given that 'In water cooling, DT is simply the difference between the ambient air temperature and the water temperature on the outgoing side of the radiator' I don't know what you mean.
http://www.overclockers.com/guide-deltat-water-cooling/

Yes, the bigger the delta (the same as heat load in effect) the more heat you can shift.

Putting radiators in series means the delta in subsequent radiators is less, meaning they do less effective cooling.

Having them in parallel means they all have the maximum delta and thus cooling effect.

Charlie Hanson · 8 Nov 2017 at 22:24

Thanks for the input. To address some points:

Radox-0 said:
Okay, so what your saying, is ...

Yes, that's right.

skyripper said:
And what do you mean by a hot loop and a cold loop?

The 'hot' loop is just the loop that contains the waterblocks, i.e. adds heat. The 'cold' loop is the loop with the radiators, i.e. removes heat. Or adds cold, if you want to disregard physics for syntactical symmetry.

skyripper said:
The water is going to mix in the reservoir and equalize the temperature across both 'loops'.

Yes, that's the intention. Hot loop takes cold water and brings it back warm, cold loop takes warm water and makes it cold. That said, I'm aware that the temperature difference between any two points on the loop is going to be negligible due to the volume of liquid. The reservoir is not going to be a heat exchanger, it's just a holder for the water and a pressure leveller for the pumps.

VKleita said:
More pumps add heat to a loop more pumps = more waste heat.

For EK pumps: D5 = 23W; DDC 3.2 = 18W; SPC-60 = 6W. So yes, two DDCs will be adding heat, but two SPCs will be half that of a D5. Even so, there's only 1W of difference between 1x D5 and DDC+SPC. I'll be using an EK Z270 full block and a 1070, so no massive effort. The real benefit I see of two separate pumps is that I can set them both at optimum values for each purpose.

Holst1981 said:
If your doing it because it's fun or will look cool... That's OK

I don't want to jeopordise cooling for looks and vice versa. I want this system to run as quietly as possible because it's for the living room. I don't think that airflow will be a problem.

Maybe I'll figure out a way of testing both.

Charlie Hanson · 8 Nov 2017 at 22:40

Here's a rubbish teaser for the build. **SPOILER ALERT** Yes, it;s going to be one of those.

Dirk Diggler · 9 Nov 2017 at 20:40

This dual loop system is a complete waste of effort, however I feel you may be doing it just because you want to take on the challenge... If so, then fine.

I'm not sure what technical advice you're looking for? As for cold hard experience; what is it that you're trying to achieve? Asthetic pleasure? Cool? Quiet?

Charlie Hanson · 9 Nov 2017 at 21:57

String said:
I'm not sure what technical advice you're looking for? As for cold hard experience; what is it that you're trying to achieve?

Technical advice is "your theory is correct" or "your theory is wrong"; cold hard experience is "i did XYZ and it was OK, but then I tried ZYX and it was better". I want to achieve minimal noise, and also to see if it works how I think it will.

If nobody has tried a hot/cold loop before there's no harm in being the first.

Dirk Diggler · 9 Nov 2017 at 22:17

Charlie Hanson said:
1) so I can blast water round the 'hot' loop at whatever speed gives the most heat removal, i.e. whatever makes the water the hottest the fastest. I expect that to be on the upper scale of speed;

2) so I can nudge water round the 'cold' loop at whatever speed gives the most heat removal, which I expect to be on the lower speed scale

I believe your logic is flawed. In a conventional loop, the fluid gathers heat from the components and it disappates that heat via the radiators. If there is enough radiator capacity to cool the components then that's what will happen. The cooling capacity is increased by the RPM of the fans only; the speed of the pump has negligible effect.

Whilst your "hot" loop will gather all the heat from the components, the only way it can dissipate that heat is by mixing with cooler fluid in the reservoir. Reservoirs have no cooling capacity, they are not designed to do that. A reservoir is not a heat exchanger.

I predict that the fluid in your hot loop will continue to increase in temperature as your components heat it further and further, and the fluid in your cold loop will stay very close to ambient as the radiators will very efficiently keep it at that temperature.

Jokester · 9 Nov 2017 at 22:20

You can avoid that by careful selection of which ports you use to connect both loops to.

Dirk Diggler · 9 Nov 2017 at 22:25

Jokester said:
You can avoid that by careful selection of which ports you use to connect both loops to.

Which ports? Have you actually did this with success?

Jokester · 9 Nov 2017 at 22:31

String said:
Which ports? Have you actually did this with success?

Aye, I’ve done it before it’ll work fine. Just make sure the outlet of one pump doesn’t directly feed the inlet of it’s own loop so that the fluid is mixing in the reservoir.

Dirk Diggler · 9 Nov 2017 at 22:35

Jokester said:
Aye, I’ve done it before it’ll work fine. Just make sure the outlet of one pump doesn’t directly feed the inlet of it’s own loop so that the fluid is mixing in the reservoir.

In that case, I'll follow this thread with interest. What would you say was the benefit when you did it?

Jokester · 9 Nov 2017 at 22:41

String said:
In that case, I'll follow this thread with interest. What would you say was the benefit when you did it?

There isn’t any particular benefit other than slightly better temps due to more flow in both loops, and at least in my case I could turn off one pump and it would still pull fluid through the other loop so there was still some pump redundancy.