Running radiators in parallel vs. series...

[airspirit]

The effect would be practically identical whether they are in series or in parallel. The only variables in the system if you have two similar radiators is 1) dT and 2) flow rate/pressure. Everything else is relatively equal and cancels out. All the "buzz" science such as boundary layers and wall turbulence, cooling karma, and psychosomatically induced processor cryonics induced by LEDs and acrylic cases that gets thrown around in these forums will not affect one appreciably more than the other. We might as well start debating whether the price of tea in china affects the thermal transfer properties of pure copper. The fact of the matter is that in a closed system, whether the rads are in parallel or in series, the mass of water will spend an identical amount of time inside of a radiator. Since in parallel configuration there is a higher dT between the fluids, there will be more efficient (BETTER) thermal transfer. This is the only possible result, just like the only possible result of throwing a fat man off a tall cliff is a big mess at the base of the cliff.

[myv65]

airspirit,

What you are missing is that convection in a radiator is nothing more than the integral of h(x) * delta-T(x) * dA. Nowhere does time enter into the calculation of heat transfer.

Some time ago there was a post about water being in the radiator for ten seconds versus a total loop cycle time of 60 seconds. Gee, so the water is in the radiator 1/6 of the time. So what happens if the loop gets longer (ignoring added flow resistance for the moment)? Now the water is in the radiator 1/12 of the time. Does this change the radiator performance? No.

The simple fact is that a radiator has to pass heat from fluid to solid, through the solid, and from solid to air. It is entirely possible that the fluid to solid portion of this is the dominant effect. If so, increasing the water velocity (and consequently DECREASING the time the water spends in the radiator) will result in greater heat transfer.

By your argument we could slow our flow to a crawl in the radiator and all would be wonderful. Yeah, that's right, let the water sit in there for a couple of days. It'll be good and cool by the time it gets out. Yes, slower flow will result in higher fluid to air delta-T but at what cost? If the radiator is capable of getting down to within a couple degrees of ambient, then the only way to increase delta-T is to increase the inlet temperature to the radiator. Is this what you want?

I am not arguing that a radiator is more "efficient" with a higher delta-T. I am merely telling you that more efficient radiator performance does not necessarily equal lower chip temperatures. In fact more efficient radiator operation requires higher chip temperatures. Efficiency within the radiator gets progressively worse the nearer the fluid temperature gets to the air temperature.

If none of this makes sense to you, there is no point in my continuing. I simply don't have the time to rehash a semester of basic heat transfer everytime someone doesn't get it. What I would ask is that people don't rely merely on instinct alone in these matters. Instinct can be a wonderful thing, but there's no way of knowing when your instinct is wrong. This is why they offer classes on stuff, including heat transfer.

And don't forget what I've said all along. Parallel should do better for the vast majority of water coolers given the typical radiators that we use. But to say it's better 100% of the time would be inaccurate.

[bigben2k]

I see we're in for it again!

Parallel is better, I agree. (unless one hits a sweet spot)

This is a complicated issue, because it involves more than 2 variables.

I understand the theory, and I agree that it may be possible, but I think that we're in a range of flow rates that's already pretty high for heatercores, but not as high for waterblocks.

[myv65]

Originally posted by bigben2k
I see we're in for it again!

Nah. I've said all I have to say on the matter. Regardless of where this thread heads from here, I will not reply to it again.

[KnightElite]

It's basically been beaten into the ground already, but parallel is better because of the higher delta T.

And for all of you people who want pics of heatercores in parallel:

[airspirit]

You can throw all sorts of fun and nifty equations out explaining how heat transfer works. In this application it all cancels out between the two setups except for those two things.

I'll break this down simply for you, since your education did you a disservice. The problem with the education system is that they teach you how to solve a problem through equations and calculations but they don't tell you why it works. That's why most calculus students can work out an integral but can't tell you what the meaning of what they are doing is. It really is a sad thing. Believe me, I've been on the education side of this as well as the practical side. The equations are great for calculating numbers and values, but it won't tell you why it works better. If people were taught how to be more than talking calculators, we'd have a much more effective educational system. This is where a true understanding of the basic thermodynamics, not equations, helps.

Picture a simple heat transfer. Calculus is a way of studying a changing/variable system, correct? Further, you can break down a variable system into a "slide show" of individual events. Just like you can graph the results via calculus, you can do so similarly with simple 6th grade algebra if you are patient/stubborn enough. This is the "T-Table" that you did in pre-calc before your instructors stopped teaching you "why" and started filling you up with "how".

If you chop this up into single units of time, you will have a heat transfer with fluids being drawn crosswise across it. The two fluids are your coolant and the air being pumped. Since the air flow is identical in both setups it cancels out, just like the size/efficiency of the radiators is identical in both setups so they cancel out.

Now, you hit the nail on the head with one statement: you'll have nice cool water if it sits there for a few days. You WANT your slug of water to remain in the cooling loop as long as possible. The caveat is that you want the cold water to be pushed through the second heat transfer (your water block) as fast as possible. You want high speed in half of your system, and low speed in the other half. Do you follow me so far?

What we have now, frozen in time, is two heat transfers connected by tubing. For ideal cooling we want the flow rate through the block to be as high as possible with the coolant T (hereafter cT) to be as low as possible. Similarly, in order to get the cT as low as possible we want the flow rate through the radiator to be as low as possible, and for the dT between cT and air T (aT) to be as HIGH as possible. Do you have all this visualized? This is where it gets tricky.

In order to speed up the flow through the block we could shrink the flow channel, forcing the water through at a higher velocity. The problem here is that it would also constrict flow through the entire system. It is much more efficient to slow the flow rate through the radiator by increasing channel width at that point. Does that make sense to you? This fits into the framework described above, does it not? You have the best of both worlds, right? Right.

So you can have one big-mouthed radiator do the job. We're talking about two standard radiators, though. By running parallel, you are in effect doubling the TOTAL channel width of the system in the cooling portion. If everything else is the same (the radiators) between a system with ONE path for the water to travel and a system with TWO paths for it to travel, you will have a bigger total channel width in the one with TWO. This is simple geometry. Before you start gnashing your teeth and tearing at your clothes, stop and think about it. It makes perfect sense.

Got it pictured? Okay, so now you'll see that as far as channel width goes, the parallel system wins. What this allows is for the water velocity to be halved in the parallel system. This is easy to visualize, right? Since the pressure throughout the system is identical, the flow rate is dropped in the wider area. This is beneficial, as was described above. What this allows is for the parallel system to keep the water slug in the heat transfer (remember, we're in a freeze frame of the equation ... ).

Here is where you really need to apply yourself, though, to understand how this works. You understand that the higher the dT in the heat transfer, the higher the efficiency, correct? This is where you get to make your choice, and I'll leave this part for you alone. You have a choice between having all of your water hit the heat exchanger array with a high dT and remain in it for two units of time (it doesn't matter that this is split into two parallel channels), or having your slug of water with the high dT hit the first heat exchanger and remain in it for one unit of time and then hit the second heat exchanger with a low dT for one unit of time. When you bring the system back to speed, you'll see which arrangement is more efficient EVERY SINGLE TIME.

Here is an improvisation of what you said before. Consider yourself educated:

If none of this makes sense to you, there is no point in my continuing. I simply don't have the time to rehash years of advanced thermodynamics everytime someone doesn't get it. What I would ask is that people don't rely merely on equations without having any real understanding of these matters. Equations can be a wonderful thing, but there's no way of knowing when your equations don't tell the whole story if you were daydreaming about beer and Britney Spears when your instructor was teaching you what they explain. This is why they offer classes on stuff, to confuse you with calculus, so when you enter the real world and actually see how things really work, you'll be able to do the basic math that'll probably never help you anyway if you don't know how to apply it.

My friend, you wasted your money on that thermodynamics class. Either you didn't pay enough attention to what was being taught (just paying attention to the equations that you'll need for the next test), or your teacher didn't do a good job teaching the class. You would have been better off taking a course in elementary education. At least then you would have been able to hang out with a bunch of young, wide-eyed and naive young women who would jump in the sack with a chimp if it could buy them beer.

At any rate, experience reigns over a textbook education any day of the week. If you get cocky over a worthless college education, you'll never get anywhere. If you learn how to apply that education, it suddenly isn't worthless anymore.

[airspirit]

Damn, that was a long-@ss post.

[bigben2k]

Thanks Airspirit, you've earned yourself new respect!

I re-read BillA's "Radiator roundup" article, to help me relate to some of what you were saying.

Radiator Roundup, by BillA

All the graphs that Billa posted point to all radiators performing less efficiently at higher flow rates (within a range of 0.5 to 2 gpm) EXCEPT for the elusive Serck, which performs best at 1 gpm coolant flow, and 0.25 inch H2O air pressure, and higher . Billa described it as "an aluminum stacked plate design with brazed fins and very low air backpressure". It was also the top performer.

Most units tested however, performed best at around 0.5 gpm, and less.

(Dang! That means that if I want to run 300 gph (5gpm) (don't ask me why!), I'd get the best performance out of 10 heatercores in parallel!)

*** CORRECTION ***
BillA recently indicated that the graphs have been corrected.

Here's BillA's post

and

Here are the corrected graphs

This makes 2 gpm a pretty good target coolant flow. If I shoot for 300 gph (5 gpm), I'd be better off running 2 rads in parallel, not for the heat transfer ability, which would turn out to be pretty much the same wether I use 1 rad or 2, but for the lower flow restriction.

[mkosem]

ok, this has been done. Bruce at cooltechnica did a lot of testing with multiple rads. He is the guy to contact for solid data on different pump/rad configurations for dual radiator setups.

--Matt

[bigben2k]

See my corrected post above.

So like I said (even further up), we're in a range of flow rates where it would make little difference to a heatercore, but not so with a waterblock.

More flow=> more performance from the WB=>a little bit more from the heatercore (s).

Sorry Kev, we had to do this.

[gmat]

Airspirit: Beware dude, your long post sounded much like a flame.
And as far as i know myv65 is a professional, he does not draw his experience only from the classroom. I wouldnt attack him on thermodynamics without heavy artillery back-up.
Besides, your post makes no point. Sorry, but you *can* and you *must* work with equations even in real life, ask any thermal or fluid dynamics engineer.
As an engineer i work everyday with theory, and i know what it's worth.
Moreover, myv65 pointed out to you that *transient* effects have no place in a closed-loop watercooling system.
Just remember, this is *closed loop*.
So in that closed loop you can describe how the transfer works, like this:
http://www.electronics-cooling.com/h...alccorner.html
Again, these are equations, but they describe *closely* what''s happening in practice. *closely* means they describe what you'll be able to measure, within an error margin, with appropriate instruments.
In short, consider the water as a continuous medium. It will transfer heat from point A (WB) to point B (rad) with a % of losses. Theses losses are due to flow restrictions, frictions, turbulence etc.
The solid->liquid and liquid->solid (solid=copper for example) interfaces are critical. It's where the transfer will lose most of its efficiency. High flow means thinner laminar layer means better transfer. Thats all.

Let's make another point. How do you cool your radiator ? With air. Air goes through a "circuit" made of small passages between fins. Do you want air to "sit longer" between your rad fins ? I don't think so. The higher the air flow, the better. Well, it's the same with water.
"Transient" effects are only valid in open-loop (aka tap water directly from the tap, then thrown away) in which case you must account for the temporal variables.

(edit) ah i found the link i was looking for:
http://www.stewartcomponents.com/htm...tem_basics.asp
interesting part:
"The radiator becomes less efficient as the coolant outlet temperature approaches ambient. Therefore, a low flow rate keeps the coolant in the radiator longer. The longer the coolant stays in the radiator the lower the efficiency of the radiator."

[airspirit]

I know it sounded kind of firey, and that was my point, unfortunately. I wanted to lay out the facts in an indisputable manner. I find it patently offensive that people get a semester of mechanical ENGINEERING under their belts and think they're Einstein reincarnated. It is really sad. It is even worse, though, when they come to a board and start lording it over people when they really don't have a good handle on what they're saying.

Gmat, I understand that equations have their place. While I was still doing that type of work I used them quite a bit, but before you throw out an equation, you need to know the background that it applies to and take that into account. It is like trying to hammer to disassemble an engine ... it doens't work very well.

The point I was trying to make is if things like air flow are equal in both systems the only things that matter are the temperature differential across the heat exchangers (radiators) and the flow rate through them. It would be easy to cripple a parallel loop system by not running air across the rads and pumping 40000 CFM of sub-zero air across the system in series (hurricane in a bottle [TM]), and then crow how the series system works better. I'm talking about a setup where everything else is equal. I think this is where his breakdown was. He wanted to start talking about specific circumstances of one without taking into account that it should affect the other. It was junk science. I find that kind of garbage patently offensive, especially since there are alot of inexperienced types on these boards that depend on the findings of people "in the know" to make their decisions by. For someone to come on and spew junk science and say that a semester of college education backs him up and that everyone else must be chewing on retard sandwiches is just ... pathetic.

edit: ENGINEERING ... jeez, my fingers go faster than my noggin.

[airspirit]

This reminds me of an assignment I had in an early CS class. The instructor was extremely full of himself, even though he was lacking in understanding of the language he was teaching (a Fortran guy teaching an introductory C course). He gave an assignment early on that required that we took as input a 3-15 digit number and output it completely reversed. He recommended that we use the % operator and various math techniques to do it. At this point we hadn't covered strings or arrays, so that SHOULD have been the only way to do it. Seeing the loophole, I wrote the entire program (excluding an #include call and the main() function) in one extremely ugly and convoluted line of code that included all of the calls and definitions for the variables. Even though the solution was perfectly correct (and quite a bit more efficient), he got his panties in a bunch because I didn't use his method. After he demanded I do it "right" I rewrote it using strictly math and when printed, the code (including comments) was over 30 pages long (imagine 10-20 lines of comments per line of code). The sad part was that when it all boiled down to it, he just didn't understand what I had done and went nuts. There is a trap that over-dependance on equations and methodology will snare you in. We can talk about laminar layers and the mineral content of the coolant and the gravitational pull of the moon on the system until we're blue in the face, but if you use the same radiators and blocks with the same pump and assorted other gear, most of those things don't have any bearing on this particular situation. The issue is that they cancel out completely and you're left with a very few things that will have an effect. I'm sure, through physics, the fact that I haven't been laid in the last couple of hours will have some effect on the efficiency of my water cooling equipment (the minor raise in air temperature from excessive calorie consumption on my part, for instance) but as long as I'm equally "frisky" before each experiment, that particular effect will cancel out in both systems. Sometimes we put in too much effort in trying to get to the bottom of a situation because a textbook says to, even when it is completely unnecessary. I guess that was my whole point. It is easier and most of the time more effective to write the one line of targetted code rather than the 30 pages of redundancy.

[gmat]

Mhh for one i dont think he bases his assumptions on "a semester of mechanical under [his] belts", like i said he's a pro. Do not underestimate him.
For two he maybe has badly formulated his thought, please bear with him he's a technical guy. Ya know, all technical guys have difficulties being clear to other people. (and to suits, but thats another problem)

In this very case he tried to be short and to-the-point, since this very topic has been beaten dead and buried 50ft under a long time ago.
What everyone agreed on, was the basic thermal equation:
q=UAdT
where A is the surface area of transfer. The higher the flow, the higher A is (and the lower U is due to laminar vs turbulent etc..)
That's true for WBs, thats true for rads, thats true for air on a heatsink, thats true for air on a rad, thats true for your car radiator.

Putting rads in parallel will increase the dT, while not killing A too much since the mean water path will be shorter (ie backpressure will be lower) - so resultant flow wont be divided by two (compared to flow in series setup).
The result, a parallel is winning.

Ah and please stop flaming ppl , read the first (sticky) thread on this forum....

[Skulemate]

Airspirit... I realise that I am just a student, and am therefore lacking in the experience that accounts for so much, but if you're going to make a point, you should use a little more precision. Using flow rate and velocity interchangably is simply not correct. I'm not an expert in heat transfer (I'm in civil engineering so that's not my thing, though I have learned a thing or two about fluids) and as such I found your lack of clarity in making your point very confusing. Also, if you could please explain what you meant by this quote I'd be very much obliged:

Since the pressure throughout the system is identical, the flow rate is dropped in the wider area.

[airspirit]

Yeah, I butchered that. What I meant is that in every point in a closed loop system you're going to have a certain amount of water pass a point per second. If one section of the loop is wider than the rest, the water doesn't have to flow as fast to move the water at the same rate through it. It is kind of like how at a narrow point of a river it'll run fast and how at a very wide point it'll run very slow. When running two channels of equal area as opposed to one channel of the same area (cross section area is what I'm getting at) as one of the two channels by itself, your water will, in effect be running at half of the speed as if it was running in that single channel. The FLOW RATE will be the same, as in the same amount of water passing through that area in the same amount of time, but overall, the water will be slowed down because the breadth of its path was widened. I don't think that made any more sense, did it? It's easy to visualize, but I'm having a bastid of a time describing it.

[bigben2k]

You mean:
the flow rate is the same, at any/every point in a rig. Given different cross-section sizes however, the flow SPEED will change.

(PS: this doesn't apply to gas flow, since they are compressible.)

[Skulemate]

Hehehe... yeah, I understand that. I was simply using that as an example of your lack of clarity . Not saying you're wrong, but wading through your post was difficult. Still not sure about that pressure thing though.

And a good point Ben, though to be picky all fluids are... Not that it applies here though.

[bigben2k]

Originally posted by Skulemate

And a good point Ben, though to be picky all fluids are... Not that it applies here though.

MOSTLY incompressible, I know!

[airspirit]

When I switch <rant> on, my reasoning centers go to jolly green goo and my fingers start flying at 140WPM. It is a b12n0+c4 to get everything poifect. Plus, if I was perfectly clear, then you guys couldn't regard me as that "@sshole nut-job from Idaho". Gotta keep up my evil image, so when I hold the world hostage for one million dollars, you all will take me seriously.

<evil>Bwa-hahahahahaha!</evil>

[utabintarbo]

Originally posted by airspirit
When I switch <rant> on, my reasoning centers go to jolly green goo and my fingers start flying at 140WPM. It is a b12n0+c4 to get everything poifect. Plus, if I was perfectly clear, then you guys couldn't regard me as that "@sshole nut-job from Idaho".

Come on, we've got plenty of reasons besides that

[Heavy_Equipment]

Well Kev, if you have the means and the patience, you could always try both versions. All the formulas and equations in the free world aren't going to measure up to good old fasioned results.

...I seem to recall that the only significant difference was noticed by the pelt crowd, and that would sure favour the delta-t angle heavily.