Rads in parallel or series?

[jman1310]

i've searched the forum extensively but never found a direct answer

i have a lower performing pump (mag2 250gph 6'max head)

cooling system:

mag2
3 blocks (custum cpu, polarflow nb, dd maze4 gpu)
res with a 3/4 feed into pump
2 BIpros
all plumed with 1/2 id hose and 3/4 res to pump
i think there is about 8' to 9' of hose used
pump and rads are in an external radbox

curently the rads are in series but i've got a TDX in the mail so the system will be drained soon anyway

my understanding is that:

parallel gives you less head loss but higher temps
serial gives you more head loss but lower temps

also, in parallel i should be able to remove much of the hose used in the radbox

pics available in this thread

[maxSaleen]

The best way to find out is to run both configurations and find out what yields the best temps. The pressure drop from running those rads in parallel would probably kill flow rates. My personal opinion is to run them in series. Now my answer would be different if the radiators were extremely restrictive. If that were the case I would say parallel. I've looked at your pics but cannot ascertain the loop's order. What is it?

[jman1310]

only the rads would be in parallel
how cirtical is flow in the rads?

does running in series effectively make a 2 pass rad into a 4 pass?
while running in parallel just increases the surface area the rads? (ie like one big rad)

[maxSaleen]

I'm aware of this, though you have to remember that pressure drops in certain parts of the loop can absolutely KILL flow rates. Flow in the rads doesn't really matter. The rad will disipate about the same amount of heat at .5gpm as 2gpm. For a deeper explanation of this look at pH's work in the articles section. The two rads will dissipate the same amount of heat regardless of loop configuration. What you should try to achieve, for the sake of your waterblocks' efficiency, is the highest flow rate in your loop.

[bobkoure]

You have this exactly backwards.
Loop restriction is proportional to flow velocity.
Running rads in parallel effectively halves water velocity through that section.
Unlike many waterblocks rads don't strongly depend on water velocity to transfer heat from coolant to metal (there appears to be some radiator-idiosyncratic relationship, but nobody's reported a big effect, and it's unclear whether less velocity is good or bad).
So... halve radiator water resistance, add back in whatever additional resistance you get from whatever fitting(s) you use (possibly nothing if you run the two outputs from a two-output block each to a separate rad, the output from each rad to a two-input res, like the polyprop one from Swiftech.

[redleader]

There is some dependence on flow for radiator performance:

http://www.overclockers.com/articles778/index04.asp

In this case you're definately going to be on the lower end of the flow axis with that weaker pump, 3 blocks and a radiator. Probably though with 2 radiators you'll have enough dissipation either way that coolant temps will be fairly low (unless you've got a TEC or something in that loop). I'd probably try parallel first.

[jman1310]

ok, so i'll switch to parallel when the new block gets here
i should be able to use 3/4 pipe off the pump to the rads to help offset the split

if i understand the graphs right then higher flow rates will only help at higher airflow which i don't have and won't use

I like my quiet system - all fans runnung at 5v

[bobkoure]

Originally Posted by jman1310

if i understand the graphs right then higher flow rates will only help at higher airflow

Airflow doesn't come into it. If your waterblock is performing more efficiently, then whatever it is cooling will be closer to coolant temp. Many waterblocks become more efficient as flow rates go up (and some, like the Swiftech 6000 series, don't change that much as flow rates go up - look at the performance curves - flat means water flow rates don't matter so much).
Also, just as a basic physics thing, heat energy flows more rapidly as the temperature difference becomes greater. So, with lower airflow, your coolant will be warmer, so less difference between this and your CPU, so the waterblock will be less efficient. On the other hand, the difference between your coolant temp and the air becomes greater, so your radiator will be performing more efficiently...

[jman1310]

Originally Posted by bobkoure

Airflow doesn't come into it. If your waterblock is performing more efficiently, then whatever it is cooling will be closer to coolant temp. Many waterblocks become more efficient as flow rates go up (and some, like the Swiftech 6000 series, don't change that much as flow rates go up - look at the performance curves - flat means water flow rates don't matter so much).
Also, just as a basic physics thing, heat energy flows more rapidly as the temperature difference becomes greater. So, with lower airflow, your coolant will be warmer, so less difference between this and your CPU, so the waterblock will be less efficient. On the other hand, the difference between your coolant temp and the air becomes greater, so your radiator will be performing more efficiently...

when the coolant flow is increased from 1lpm to 12lpm
this graph shows a nearly flat curve a 7v fan which indicates at lower airflow increasing coolant flow only increases the heat dissipated slightly
however, with the same fan at 12v there is an improvement of nearly 50 watts
and 2 of the sames fans at 12v gave a 75 watt difference


image borrowed from here without permission sorry and thanks

[bobkoure]

And your point is...?

[nikhsub1]

I think his point is that fannage does play a factor in rad efficiency in regards to flowrate. I would try the rads both ways and see...

[jman1310]

my point is having less flow in the rad will not hurt performance due to the low fan speed (see redleader's post)

[maxSaleen]

Airflow obviously has an impact on the heat dissipated by a rad. No one can give a logical argument otherwise (though the imperical data above is a stickler). Water flow rates will have almost no effect on heat dissipated. Find the article about water cooling myths on this site (it's under "articles" on the front page). There you will find an excellent explanation as to why flow rate through a rad doesn't change heat dissipation. Flow rates can have a large effect on your waterblocks performance. You said that you were going to throw a tdx in the loop so you will want to achieve all the flow your loop can muster. From this standpoint I still say run your rads in series.

[nikhsub1]

Originally Posted by maxSaleen

Flow rates can have a large effect on your waterblocks performance. You said that you were going to throw a tdx in the loop so you will want to achieve all the flow your loop can muster. From this standpoint I still say run your rads in series.

You've got that backwards. If he ran the rads in parallel, theory is system flow rate should he higher. You halve the flow to the rads, which may be ok, but everything else gets a benefit from the parallel rads (the blocks). In series we introduce more head loss, how much more is unknown. Bottom line is, test both ways, it is the only way to get a clear answer, and then, it may not be so clear.