Radiators in parallel

[Nomad2000]

I have two radiators - BIX and BI Prime. Can I use them in parallel? With Eheim 1250 and 1248? Thank everybody.

[jeff0628]

How many waterblocks are in your system? I would say if you're just running one cpu block then dont waste your money on running a second radiator and a second pump, because the cpu block cannot generate enough heat to warrant the practical use of a second radiator / pump.

Now if you're planning to go all out and have a cpu block, a gpu block a northbridge block, a hd block, etc... then you could benefit from running 2 rad's in parallel with two pumps. BTW, if you do run 2 pumps and if they will share the same resevoir together then run equal gph pumps. If they're independent closed loops then it doesnt matter.

[bigben2k]

Originally posted by jeff0628
How many waterblocks are in your system? I would say if you're just running one cpu block then dont waste your money on running a second radiator and a second pump, because the cpu block cannot generate enough heat to warrant the practical use of a second radiator / pump.

Now if you're planning to go all out and have a cpu block, a gpu block a northbridge block, a hd block, etc... then you could benefit from running 2 rad's in parallel with two pumps. BTW, if you do run 2 pumps and if they will share the same resevoir together then run equal gph pumps. If they're independent closed loops then it doesnt matter.

You're cruisin' for a flamin'!

Multiple rads, even for one CPU wb is fine. It's actually better... because it lowers the flow to each rad, allowing more time for the coolant to 'cool'. The problem is that they will both (ideally) require a source of fresh air, so the whole thing ends up taking so much space, that it's increasingly difficult to fit it in one case.

The benefits are negligeable of course, but with a powerful pump like the 1250, it certainly can't hurt.

As for using two pumps... it's a worthy idea. One of the pumps becomes a booster pump. Not very efficient, but it does work. (pumps in series).

Pumps in parallel would be a bad idea in this case, because these two pumps are very different. (pumps sharing a res)

Pumps in seperate loops: it's easier to control the flow that way. Otherwise, you're looking at the old debate of series versus parallel, with valves and fittings, and so on...

[schoolie]

Sorry, I still think that the theory that states lower flows through the radiator produce more cooling is plain wrong. Higher flows through the rad will *always* produce more cooling because the the average coolant temp will always by higher.

[edit[
|| radiators are better because the average coolant temp is higher in the rads and the flow resistance is lower
[/edit]

[bigben2k]

Originally posted by schoolie
Sorry, I still think that the theory that states lower flows through the radiator produce more cooling is plain wrong. Higher flows through the rad will *always* produce more cooling because the the average coolant temp will always by higher.

[edit[
|| radiators are better because the average coolant temp is higher in the rads and the flow resistance is lower
[/edit]

Let's see if I get you right...

"Higher flow through the rad will always produce more cooling because the average coolant temp will always be higher."

What if the flow is so high, that the temp at the outlet of the rad is not much different than when the coolant went in?

In theory, you are partially right, in that the intake temp needs to be high, because rads can only cool something that's hot. But if the outlet temp is just as high, what's the point?

What if the flow was so low, that the coolant temp would go very high, enter the rad, have plenty of time to cool down, and exit at a temp slightly above ambiant?

[redleader]

What if the flow is so high, that the temp at the outlet of the rad is not much different than when the coolant went in?

Performance would be uneffected most likely. It doesn't matter if the water is at .99999c or 10 MPH, theres always the same amount of coolant in the radiator. As long as you can keep throwing more heat into the rad, it shouldn't matter how fast the coolant moves or how long the coolant stays in.

This is because every H20 molecule is the same. It doesn't matter if one warm mole spends 10 seconds, or two equally warm ones get 5 seconds each, either way the same amount of heat passes out of the system.

But if the outlet temp is just as high, what's the point?

What if the flow was so low, that the coolant temp would go very high, enter the rad, have plenty of time to cool down, and exit at a temp slightly above ambiant?

Bot these situations should transfer just as much heat. Remember all that matters is hoiw much heat leaves the system, not which indivdual water molecules carried it. Either way the Delta T across the radiator will be the same, so both perform the same.

I agree, the idea that high coolant velocity somehow hurts effciency makes no logical sense. How ever parallel rads make sense because they diminish flow resistance somewhat, thus allowing for better flow to the waterblock.

[schoolie]

Thanks for considering my opinion BigBen OK here goes:

Assuming that the dissipation of heat from the water to the radiator is proportional to the difference in temp between the water and the radiator ( and the air flowing over the fins). If one acepts this, then the heat dissipation would increase with increasing water temps in the radiator, and with lower air temps flowing over the radiator fins.

Consider the steady state of a simplified system, where the coolant temp is only a function of position in the cooling loop, and not time dependent. A lower coolant flow should produce a larger difference in inlet and outlet coolant temps in the radiator, and a lower total heat dissipated by the radiator.

I'l finish my thoughts after I get back, but I'd be curious what other people think.

Thanks

[bigben2k]

Originally posted by schoolie
Thanks for considering my opinion BigBen OK here goes:

Assuming that the dissipation of heat from the water to the radiator is proportional to the difference in temp between the water and the radiator ( and the air flowing over the fins). If one acepts this, then the heat dissipation would increase with increasing water temps in the radiator, and with lower air temps flowing over the radiator fins.

Consider the steady state of a simplified system, where the coolant temp is only a function of position in the cooling loop, and not time dependent. A lower coolant flow should produce a larger difference in inlet and outlet coolant temps in the radiator, and a lower total heat dissipated by the radiator.

I'l finish my thoughts after I get back, but I'd be curious what other people think.

Thanks

I agree with the first part, but I don't believe that the proportion is linear.

As for the second, if the outlet temp is much lower than the inlet temp, then where did the heat go? It is a function of time, any way you look at it.

Let me see if I can pump out an overview:

Facts:
1-Ambiant air is at 20C
2-Coolant temp is increased by the power emitted by the CPU.
3-Coolant is cooled by transmiting its heat to the metal that composes the radiator.
4-The rad cannot lower the temp of the coolant below the temp of the ambiant air.
5-The heat from the coolant is transmitted to the metal of the rad at the same rate, regardless of the rate of movement of the coolant.
6-The rad lowers its metal temp to the ambiant air.
7-The fan helps the above purpose.

It therefore follows that the longer the coolant sits in the rad, the more heat can be transmitted to the metal of the rad. The result is a lower coolant temp.

It is however increasingly pointless to reduce the flow rate, if the coolant temp exits the rad at a temp close to ambiant.

Let's see if we all agree on that before I go on.

[jtroutma]

So far I agree

[schoolie]

When I said steady state, I meant a system where the soltuion to the temp of the coolant at any point in the system would not change as a function of time, assuming that the CPU power and the ambient air temp remains constant.

First, no matter what the rate of flow, the time that the water spends in the rad is the same

OK, I think #5 is where I don't agree:

5-The heat from the coolant is transmitted to the metal of the rad at the same rate, regardless of the rate of movement of the coolant.

The heat flux is proportional to the temp gradient. The lower the coolant temp, the lower the heat transfer. If one contends that a low flow produces better heat shedding, then that implies that the temp diff between outlet and inlet temps is higher in a low flow system, right? Doesn't this to a contradiction because this would logically mean that a low flow system is shedding less heat?

[bigben2k]

Originally posted by schoolie
When I said steady state, I meant a system where the soltuion to the temp of the coolant at any point in the system would not change as a function of time, assuming that the CPU power and the ambient air temp remains constant.

First, no matter what the rate of flow, the time that the water spends in the rad is the same

OK, I think #5 is where I don't agree:



The heat flux is proportional to the temp gradient. The lower the coolant temp, the lower the heat transfer. If one contends that a low flow produces better heat shedding, then that implies that the temp diff between outlet and inlet temps is higher in a low flow system, right? Doesn't this to a contradiction because this would logically mean that a low flow system is shedding less heat?

I'm really trying, but I have a hard time understanding what you're trying to say.

Given a fixed period of time, I agree that the amount of time that the coolant spends in the rad is the same, in proportion to the other components. i.e. within say 1 minute, wether the flow rate is 1 gpm or 2 gpm, the water will spend a proportionally equal amount of time in the rad.

I agree that the higher temp difference (between coolant and ambiant air) will make a rad more efficient.

The part I don't get is: if the flow is faster, then the coolant doesn't have as much time to cool down. Given the above, that might be ok, since the coolant would come across twice within the same period of time, but the problem is that the semi-cooled coolant gets heated up again, before coming back for a second pass.

Here's a link that might spark this conversation up a bit:
http://www.overclockers.com/articles481/

[redleader]

The part I don't get is: if the flow is faster, then the coolant doesn't have as much time to cool down. Given the above, that might be ok, since the coolant would come across twice within the same period of time, but the problem is that the semi-cooled coolant gets heated up again, before coming back for a second pass.

Yes but if flow has doubled, then it only gets heated half as much each pass through the waterblock. See where this is leading . . .?

Regardless of flow each drop of water spends the exact amount of time in the waterblock, in the radiator and absorbs the same amount of heat. Thus it will have the exact same temperature as it leaves the radiator regardless of flow (within reason anyway).

[bigben2k]

Originally posted by redleader


Yes but if flow has doubled, then it only gets heated half as much each pass through the waterblock. See where this is leading . . .?

Regardless of flow each drop of water spends the exact amount of time in the waterblock, in the radiator and absorbs the same amount of heat. Thus it will have the exact same temperature as it leaves the radiator regardless of flow (within reason anyway).

I'm beginning to see...
It seems to me that the flow must be in a specific range. Not too slow, but not too fast.

So how does that explain the sweet spot of rads, as per the link above?

Also, if we're going to use higher flow rates, then we'd have to consider the pressure rating of the rads. It's not a concern to anyone, yet...

[bigben2k]

Originally posted by redleader


Yes but if flow has doubled, then it only gets heated half as much each pass through the waterblock. See where this is leading . . .?

Regardless of flow each drop of water spends the exact amount of time in the waterblock, in the radiator and absorbs the same amount of heat. Thus it will have the exact same temperature as it leaves the radiator regardless of flow (within reason anyway).

Semi-heated, semi-cooled.

[Sirpent]

Originally posted by schoolie:
Assuming that the dissipation of heat from the water to the radiator is proportional to the difference in temp between the water and the radiator ( and the air flowing over the fins). If one accepts this, then the heat dissipation would increase with increasing water temps in the radiator, and with lower air temps flowing over the radiator fins.

That's a reasonable assumption (of course, we leave out turbulence here). The radiator itself is not important. We can assume that the heat transfer is proportional to
(Tair -Twater), where Tair = (Tair-in +Tair-out)/2, Twater = (Twater-in +Twater-out)/2.

Consider the steady state of a simplified system, where the coolant temp is only a function of position in the cooling loop, and not time dependent.

That's fine.

A lower coolant flow should produce a larger difference in inlet and outlet coolant temps in the radiator, and a lower total heat dissipated by the radiator.

Not necessarily. With lower flow Twater-in will go up improving heat transfer. It's not that your answer was wrong, just the argument was not convincing.

I'll finish my thoughts after I get back, but I'd be curious what other people think.

My 2 kopecks.

Well, the problem is, we have much fewer equations then variables in our system. In the real life, there are additional relations between those variables but they are much more complex. It seems hard to get reliable answers from this model. It should be easier to analyse the system mathematically in the "limit case", where the flow is very high and (Twater-in -Twater-out) is very low.

Originally posted by bigben2k:
So how does that explain the sweet spot of rads, as per the link above?

Frankly, that sweet spot phenomenon looks weird. If the measurements are correct, the only possible explanation is that the "volume" of turbulence behaves in a non-monotonic way: water turbulence in the radiator has "resonances" at certain values of the flow.

You see, a "sweet spot" in a real life system is conceivable because the waterblock is also involved in some obscure way. But Bill kept (Twater-in -Tair-in) constant, and this is an entirely different story.

[JimS]

These flow rate arguments are really getting interesting. Here is the way I look at it.

Imagine for a minute that you are taking a small quantity of your coolant, say a droplet and using it as a reference. Say the droplet makes the loop throughout your system in 10 seconds, and spends 2 seconds of that time inside the radiator. In one minute you will have cooled your coolant for a period of 12 seconds. 6 x 10 = 60 seconds, 6 x 2 = 12 seconds.

Now you double your flow rate. Your droplet makes its way through your system in 5 seconds, and spends one second in the radiator. In one minute, you will have cooled your coolant for a period of 12 seconds. 12 x 5 = 60 seconds. 12 x 1 = 12 seconds.

Flow rates in watercooled systems are fairly consistent throughout. There is not a drastic difference in flow rate from any one point in the system to another.

This simple example is why the only thing we watercoolers need to worry about is MINIMUM flow rate. As long as your pump can sustain a flow over 25 gph( this includes the resistance of all the restrictions in a typical watercooled system), your temperature will not fluctuate much no matter how much you increase flow.

There was a test done a while back which pretty much proved this. I think it was an article by Bill Adams from overclockers.com.

[gone_fishin]

Originally posted by JimS
These flow rate arguments are really getting interesting. Here is the way I look at it.

Imagine for a minute that you are taking a small quantity of your coolant, say a droplet and using it as a reference. Say the droplet makes the loop throughout your system in 10 seconds, and spends 2 seconds of that time inside the radiator. In one minute you will have cooled your coolant for a period of 12 seconds. 6 x 10 = 60 seconds, 6 x 2 = 12 seconds.

Now you double your flow rate. Your droplet makes its way through your system in 5 seconds, and spends one second in the radiator. In one minute, you will have cooled your coolant for a period of 12 seconds. 12 x 5 = 60 seconds. 12 x 1 = 12 seconds.

Flow rates in watercooled systems are fairly consistent throughout. There is not a drastic difference in flow rate from any one point in the system to another.

This simple example is why the only thing we watercoolers need to worry about is MINIMUM flow rate. As long as your pump can sustain a flow over 25 gph( this includes the resistance of all the restrictions in a typical watercooled system), your temperature will not fluctuate much no matter how much you increase flow.

There was a test done a while back which pretty much proved this. I think it was an article by Bill Adams from overclockers.com.

He has since revised his findings to show that a radiators efficiency does indeed increase with increased flow rate albeit on a gradual curve and he only tested up to between 2 and 3 gpm.
I have an extremely high flow rate system (6 gpm) compared to the norm and can say that it is very responsive to airflow at the rad.
Here is a chart of an observance in slight water temp change caused by an increase in airflow at the rad and a subsequent drop in cpu temps after they had stabalized at 42.5C.
9:50 is the idle temps just before the load started. It stabalized at 42.5C at 10:08 with two equal readings of 29.9 water temps. At this point I doubled the fan intake to 100cfm at the rad. Water temps started to drop and by 10:10 a drop in cpu temp was recorded by 10:12.
I used an acurite digital thermometer to record water temps and room temps. Room temps were also compared to a digidoc5 sensor along with the acurite readout. The system stabilizes quickly and responds quickly. There is only a total of a half gallon in the loop which all goes through once every five seconds at this flowrate.

[gmat]

The very same person proved that *pressure* plays an important role.
In short you'd want a high pressure in the waterblock, and a low pressure in the rad.
Considering this, putting rads in parallel will reduce the pressure in each individual rad. So far so good.

Now, it goes without question that higher flow rate means better heat transfer. Why ?
* Increased turbulence
* Higher contact rate between water molecules and rad walls, less "dead spots"
* Better temp gradient through the *whole* water circuit because it's "averaged"
* My fluid dynamic courses are so many years behind i can remember everything
Since putting rads in parallel will divide flow... No good.

So what's to say ?
We could go math-crazy and dump endless thermodynamic equations. (Ahhh my friends Reynolds and Boltzmann)
Or you could test it out
Simply put the 2 rads in series, and without changing anything else put em in parallel.
Beware of tube length and elbows and so on... The ideal would be an "open air" test where rads would sit freely on a bench (to avoid tubing / elbows problems)
Get the temps in both configurations. And post the results here

NOTE: this common mistake of "low flow rate = better temps" has been thrown out in forums zillions times. Should be a FAQ, IMHO.

[bigben2k]

I'm glad that we're having this debate. I've always felt that the low flow rate theory needed to be challenged. (even if it's been done before)

So far, I think we all agree that there is a minimum flow rate for a rad, for it to operate efficiently. If the flow is completely stopped, then there is no heat dissipated.

I also think(hope?) that most of us now believe that a higher flow rate, beyond the minimum, will have no/little significant impact.

I would certainly hope that more of us see that the highest flow rate would create an increasing amount of turbulence, thereby increasing the rad efficiency. On the other hand, that turbulence just might be that sweet spot, and the higher flow rate would do nothing. I think this is why every rad behaves differently. This turbulence will depend A LOT on the internal design.

As for the graphs seen on the OC article (link above), I was thinking about it last night, and it seems that the curves don't have enough points to come to any kind of conclusion, and that the curve drawn is extrapolated (Excel?) and could be way off.

Testing the two rads parallel versus series:
We need a volunteer!
Anyone?

[MeltMan]

Since putting rads in parallel will divide flow... No good.

But... You want the water to spend the MOST time in the rads to remove the most heat... Dividing flow amongst 2 radiators would half the flow to each, but after going through the rads the flow comes back to gether.

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

[bigben2k]

Originally posted by MeltMan


But... You want the water to spend the MOST time in the rads to remove the most heat... Dividing flow amongst 2 radiators would half the flow to each, but after going through the rads the flow comes back to gether.

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

Yes, but the point here is that the flow MIGHT fall under the minimum flow for a single rad, therefore reducing the efficiency.

If the flow is high enough, it'll make little to no difference, except that the overall flow is higher, which the CPU WB likes.

[MeltMan]

But the flow resistance should be 1/2 of running a single rad which is good. Running in series would double the flow resistance which is bad.

You have to keep in mind that most radiators already run multiple parallel channels within (heatercores). The most flow restrictive radiators are those with a single channel. Adding another radiator in parallel is going to be just like buying a bigger one.

:shrug: maybe im just confusing myself, but water acts similar to electrotrickery in the way that it flows.

[MeltMan]

I just thought of a GREAT analogy.

1. Take a straw. Blow through it.
2. Take 2 straws. Blow through them.
3. Take 2 straws and stack them. Now blow.

Tell me the results.

Whats that you say? 2 straws in parallel are easier to blow through? Wow! But the air flows only half as fast through each straw? Oh. Well thats good!

:shrug:

[bigben2k]

Originally posted by MeltMan
But the flow resistance should be 1/2 of running a single rad which is good. Running in series would double the flow resistance which is bad.

You have to keep in mind that most radiators already run multiple parallel channels within (heatercores). The most flow restrictive radiators are those with a single channel. Adding another radiator in parallel is going to be just like buying a bigger one.

:shrug: maybe im just confusing myself, but water acts similar to electrotrickery in the way that it flows.

Almost there...

Yes, because the rads in parallel would make an overall lesser resistance to flow, it's better.

Individual components of the rads aside, they still need a minimum amount of flow. If you run them in parallel, and the flow through one rad falls below the minimum flow rate for the rad to be efficient, then you're loosing out on some cooling.

On the other hand, if the flow through the rad is still above the minimum, then everything is ok, in fact, it'll be better because the overall flow rate will be higher.

[MeltMan]

the flow through one rad falls below the minimum flow rate for the rad to be efficient, then you're loosing out on some cooling.

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.