Radiators in parallel

[bigben2k]

Originally posted by MeltMan


Why?

Who "defines" what the minimum flow rate for a radiator to be efficient is? You arent going to "miss out" on any cooling. Your radiator will just cool the water closer to ambient longer. So what if the water stays in the radiator at close to ambient longer? That is a good thing in case you add more heat load. It's a buffer. You arent missing out on cooling. The water wont warm up in the rad. You are just gaining flow rate which in turn is cooling better on the chip.

Oh, so close...

Ok. Try to picture a graph of the performance of a rad, where X is the flow rate, and Y is the amount of heat dissipated.

Would you agree that, at no flow, the heat dissipated is also zero?

At a very, very low flow, (I need JimS here, to complete what I'm trying to say!) the coolant would get very hot, enter the rad, and be cooled down with time to spare BUT the waterblock has allowed the coolant to become very hot, and your CPU temps would suffer.

If you allow the coolant to go a little faster, then the CPU wb still emits the same heat to the coolant, but it'll do so to a larger amount of coolant, so the coolant doesn't get so hot. In that scenario, the rad would still have time to cool down the coolant.

Now... this is where we jump to a faster flow rate. The idea is that for time X, wether the coolant passes through once or twice, the coolant spends the same amount of time in the rad and in the CPU, so it shouldn't make a difference. The coolant is half cooled, but half heated. The only difference is that the coolant is travelling faster, which will create more pressure throughout the rig.

The higher pressure is good, because we're getting into turbulent flow, which would allow the water to collect or dissipate more heat.

This pressure thing is where we're at...

[MeltMan]

warmer...colder.. colder... you're an ice cube.

BUT the waterblock has allowed the coolant to become very hot, and your CPU temps would suffer.

NO

I am not talking about flow through the waterblock at all. What I am saying is that with equal flow through the waterblock, slower flow through multiple radiators is a GOOD THING since it allows the water more time to cool. Read the last 2 sentences in my previous post. What Im trying to convey is that your flow will increase with multiple radiators because there is less flow resistance. This will cool the waterblock better since you are relieving the biggest bottleneck of the system: the rad.

When watercooling you want 2 things:
1. High flow through the waterblock for turbulance.
2. Low flow through the rad for time to cool.

When running multiple rads in parallel you achieve this.

[bigben2k]

If that's all you're trying to say, then I agree, as long as the rad gets its minimum flow rate.

[MeltMan]

yeah, as long as there is flow through the rad(s) then everything is peachy.

[jtroutma]

It took you two how many posts to come up with that?!

I could have recaped that at the very beginning
(or most of use could have for that matter)

BigBen, I think your saying would fit quite well here:

K.I.S.S!

[gmat]

Yup more flow = the better.

There's no such thing as "good minimal flow". Indeed some rads (not all) have a "sweet spot" obviously produced by a flow/turbulence factor. This sweet spot could be placed quite high on the flow line, BTW.

Now bb2k you still make one major mistake. When considering the whole cooling solution, the only place i see a "time" variable is in flow, and it must be the highest possible.
Just as a waterblock, if your "time" increases (ie your flow drops) the efficiency of heat transfer drops.

Now consider the WHOLE rad and the WHOLE lot of water in a CLOSED LOOP (yes put a pump and a heat source). If you lay down heat tranfer equations you'll see that *absolute time* comes NOWHERE in the equations (if you set apart the transient parts, ie startup / shutdown stages). You'll end up with heat resistance variables, and THATS ALL. Look ma, no time.
And that heat resistance is an *inverse* factor of water flow (fluid dynamics here). Ho ho ho, what would that be, higher flow is better ? .... (note you see 'time' variables in the terms of a /dt like in dQ/dt, if you dont know what that means dont even bother to reply)
If you insist in not believing that i may go frenzy and actually layeth down the math equations but that will require an effort (thus, upsetting me).
Repeat 100x times to yourself: "more flow is better". Thanx.

Its like Ohm's law. Two resistors (rads) in parallel equal 1/2 the resistance (flow cut)

Yes yes ! I was only talking about flow in the rads. And i told that could be either good or bad, pressure loss (good in a rad) vs flow drop (bad). On the other hand the 'ohm law' in fluid dynamics is still valid and putting rads in parallel will result in better overall flow.
Some experiences (dont remember the site) proved that was a good idea for that very reason.

[redleader]

Wow this thread went peachy while I was away. Frankly I'm amazed everyone into agreement through rational, pleasent debate. I love this place.

As for sweet spots, I suspect they are more a result of individual setups and not universally applicable. I'd also like to point out that higher performance rads seem to lack them, rather they feature normal diminished returns as would be expected due to increased bottlenecking by other invaribal factors.

[BillA]

perhaps its easier to understand with some numbers

it is (now) generally accepted that a higher (volumetric) flow rate through a wb will result in 'better' cooling
not 'more' cooling, not 'faster' cooling; it is actually decreased thermal resistance
- the higher flow rate, in a given wb, results in higher fluid velocity which increases the heat convection rate
- as the convection rate increases the thermal resistance decreases necessitating a lower thermal gradient for the heat to move
- the decreased gradient results in a lower wb bp temp, which - given that the TIM joint's thermal resistant is a constant - then is seen as a reduction in the CPU temp



at 70W with a heat die (~ 100W per Radiate)

so what then is the effect of the same increase in the flow rate on the performance of the rad ?



this is a graph not in the original rad article but shown because it is the most (apparently) simple
[all of the revised graphs can be seen here and were updated due to the incorrect calibration of the flow meter used; the article itself is several weeks away from completion]

the dissipation was calculated using the equation on pg 1 of the article
while a slight benefit (almost) always derives from higher coolant flow, it becomes appreciable really when accompanied also with higher air flow rates

what is not apparent is the coolant temperature drop across the radiator, which is in contradiction to the dissipation;
as the dissipation increases with an increase in the flow rate, so is the temperature difference between the inlet and outlet decreasing

from a system viewpoint this is not really so significant since the temperature rise across the wb is also decreasing as the flow rate increases
- except that a higher coolant equilibrium temperature is going to result, in turn raising the wb bp temp and the CPU temp

the way that this higher temperature can be avoided is by having substantial overcapacity in the radiator

changes in the capacity can most easily be visualized in terms of the coolant residence time
- the most simple would be to make the rad's tubes longer (2 rads in series), but this will increase the flow resistance
- a better approach is to add additional tubes (2 rads in parallel), providing the additional benefit of decreased flow resistance

must close for now as its WAY too late
be cool

[gmat]

Amen to that. Thank you BillA !

[gone_fishin]

BillA, for it being so late that is one of the best descriptive write ups describing your testing I have seen. Nice job

[WebMasta33]

Originally posted by Nomad2000
I have two radiators - BIX and BI Prime. Can I use them in parallel?.

If you run them in parallel, you're not going to have even water flow to both radiators.

First off, the prime has 3/8" barbs, the BIX has 1/2" barbs. So that's more resistance on the prime, so more water will flow to the BIX, due to the fact that a fluid always flows the path of least resistance.

Secondly, the BIX is a double pass rad, where the Prime is a quad pass rad. So that is also a volume limiter. So even if they both had 1/2" barbs, you couldn't push water in that fast because it's going into smaller channels.

Just thought I'd throw that into the mix....

[bigben2k]

Right on.

The BI prime will resist more because of
a) it's a quad pass (versus 2)
b) it has 3/8 barbs, versus 1/2

But it will work. It seems wasteful though.

I'm missing something here... Does the current setup have the BIX or the BI prime?

If the current setup is with the BIX, then adding a BI prime, in parallel, would have little to no effect, as most of the heat would still come out of the BIX.

If the current setup is with the BI prime, then adding a BIX in parallel would cause most of the flow to go through the BIX, where most of the heat would be dissipated.

In either case, adding another rad in series, would restrict the overall flow, which would cause everything to perform a bit less. I think...

[bigben2k]

however...

in both cases, running the rads in parallel would increase the overall flow, which should increase the performance.

but...

if the current setup has the BIX, then adding a BI prime would have very little effect on the overall flow.

I'm getting a headache...

[gmat]

As far as choices go, here's a reminder of total heat efficiency:
one rad < 2 rads in series < 2 rads in parallel
So if you can put the 2nd rad in // do it.

Moreover, adding a *less* flow restrictive rad in // will improve the overall flow of the whole circuit (remember ohm's law ?)

Note: 2 rads in series is better than one rad because of the greater exchange surface....

[MeltMan]

Note: 2 rads in series is better than one rad because of the greater exchange surface....

But, at the cost of flow.

[BillA]

the BI Prime is not just 'any' radiator
it is absolutly one of the most flow restrictive rads ever offered for WCing



the negative impact of the coolant flow reduction in the wb will far exceed any gain from the increased rad area

one should consider such options from an overall system's viewpoint

be cool

[Cova]

Originally posted by gmat
As far as choices go, here's a reminder of total heat efficiency:
one rad < 2 rads in series < 2 rads in parallel
So if you can put the 2nd rad in // do it.

Moreover, adding a *less* flow restrictive rad in // will improve the overall flow of the whole circuit (remember ohm's law ?)

Note: 2 rads in series is better than one rad because of the greater exchange surface....

I'd have to disagree with this one...

2 rads in series < one rad < 2 rads in parallel

IMHO it's better to have a good high-flow rad and make the WB as efficient as possible, most any semi-decent rad will have no problem dissipating 100W of heat that a CPU puts out. But putting 2 rads in series cuts your overall flow-rate through the system down considerably. In my experience, it's not the block, but the rad that is the most restrictive to water-flow (not counting tiny chipset/GPU blocks, some of those are way too small)

Then again, my rad is a BI Prime - so I might be biased in my judgement of where the flow-restriction is in a system after looking at billa's graph there.

[jtroutma]

How about this.....

I am planing sometime in the near future to add a Black Ice II (modded) into my system. Since my case is hurting for space I HAVE to run the rads in series (a BE Cooling AquaCoil + BI II)

I will post my results as soon as I get around to getting the parts, modding the rad, and putting it into my system.

I will also try and experiment with running the two in parrallel before I finish my setup.

This SHOULD HELP bring some answers.

BTW the modding I will be doing is turning the Black Ice II from a 4 pass w/ 3/8" barbs to a 2 pass with an intake of 1/2" and (2) 3/8" outtakes.

[lead]

Well to add to what Billa said about the BI prime-Just got a D-Tek heater core and had the puter apart so decided to check flow as compared to BIp. With 1048 pump 2 ft of suction hose in a 2 gal bucket, 1 ft of hose to rad, 2 ft. discharge to bucket from rad with 3/8 fittings on pump. Shook everthing to get air out and pull dicharge hose out to 1 gal milk jug--1 min and 15 sec to fill--did this several times . I know thst is not a test procedure like Billa uses but it would show the difference with the core. same test with the core and 55 sec. Had 3/8 fittings on pump as to run vinyl 1/2 od --so put stock fittings on(1/2) and re-tested--no difference on the prime but the core dropped to 50 sec. (on re-test and before used 3/8 silicon so it would fit on core 1/2 fittings.) The 2 gal bucket was below the pump and rad and at start of test water level was equeal to pump in-take but had to go uo about 3 inches to get out of bucket. Did have some times with the Via Aqua 1300 but couldn't get consistent results as via kept getting air and the via would noticeably slow down(flow) as the water level in bucket went down(so won't say now till I re-tape the threads(only place I can think of getting air)but it looks like it might do 2 GPM w/core and 5/8 fittings on block and pump.
And I am not knocking the Prime just sharing what little I have learn --Am follwing the threads about flow and dual Rads as that is what I am putting together and learning as i go along.
thought i might better edit this--these are my results done on my carport and on a rainey wed. afternoon using my Timex and my 1048 and my BI prime and my core- -your results might vary--but I don't think so by much.