good idea?
 

if you read this article, http://www.overclockers.com/articles481/
he tested and found radiators to do better at low flowrates, adn blocks to do beter at high flow rates. this would mean we have to find a point were the two ballance out, or i had another idea.

it would use seperate pumps for the rad, and wb. the rad pump would be low flow, the wb pump would be high flow. the water woiuld be mixed in a liquid exchanger, could just be a resovoir.
i know that 2 pumps adds more heat, but the lowgflow pump wouldnt add much heat, seign as its low flow.

alas...
 

But you would also introduce heat from another pump.....

yeah i know,

That makes sense, in a way. Try it, and make sure water in the "fluid exchanger" gets properly mixed - ie dont have corresponding inlet / outlets face each other. Cross the flux... (*no* , *never* cross the flux hmmmm sorry it was a reflex).

The radiator should more than overcome the heat from the second pump. Good idea.

Actually, the article says that SOME rads perform better at low flow rates.

If you look at those graphs, on page 5, it's clear that some rads perform better at a lower flow rate, but they haven't been tested for even lower flow rates, so it's hard to tell.

Also on the same page, it's clear that the graphs, if extended, would show that some rads would do better, and some worse. The difference is very small though, and I think that it can be attributed to an error factor.

All in all, it seems clear that any rad is designed to be run for a specific flow rate (+/- 5%), and I'd very much like to hear from JimS about this.

JimS?

Athlonnerd, that's an amazing idea. I can't believe no one thought of it before. That fluid exchanger can be incorporated as an air trap, killing 2 birds with one stone, sort of speak. Clearly we needed to isolate the flow to the rad, from the flow to the block(s). Excellent suggestion... I'll get to work on an airtrap soon, knowing all this!

I will defer to my statement in a previous post that says minimum flow rate is what is important with radiators. Originally I thought minimum flowrate throughout the system was important, but after doing more research and reading many posts in this forum, I now am a believer in higher flow rate through the block being better.

That is what makes this a great idea, you get the best of both worlds.

Ok, thanks JimS, but would you say that there is such a thing as an optimal flow rate for radiators? In other words, are radiators designed for a specific flow rate of coolant?

I think radiators are designed to be effective at a minimal flow rate. Almost every analysis and graph I have seen would indicate that even radiators have to have a minimal flow to be effective, with a diminishing return as the flow gets to a certain point.

It would be interesting if someone could get some concrete info on this. A radiator manufacturer would probably be the place to start. Maybe myv65 knows more about this.

Here's an extension to the original idea:

Since rads may have a sweet spot, add a valve from the rad side's pump input, to this new res: it would create a bypass valve that'd allow us to fine tune the flow to the rad.

Good idea. I think we need a grant or something to develop watercooling theories and methods. Anyone? Anyone?

It would not work.

1: Coolant from the central res would be free to flow through the high speed line feeding the block. There is nothing preventing the hot exhaust fluid from immediately reentering the high speed loop.

2: In order to prevent this occurrance, you would need a radiator inline after the water block to guarantee some cooling to the fluid in that loop. Otherwise, you risk a high probability of system failure.

3: Because of the prior cooling from the radiator inline with the water block, the rad in the slow loop would be much less efficient since radiators are more efficient at higher temperature differentials. Using a standard radiator in this loop would probably just absorb heat from the pump.

SOLUTION:

Use a large (think in the order of 2x3 FEET like on the front of a F-Series pickup ... expect to pay $3-400 for a new high quality one) radiator inline with the second pump as a COOLING RESIVOIR. It would serve two purposes: it would increase the coolant volume and further cool (albeit slowly) the coolant. No fan would be required to use this unit due to its size (and absolutely massive surface area) and it could be bolted to the side of your case. The hard part would be adapting the inlet/outlet to the size of hose you're using, though that could be done through PVC fittings, I believe.

Here is a diagram:

Note, the yellow is the central res (note the baffles to prevent direct recirculation), the light blue are the rads (note size diff ;) ), the red is the block, and the green is the pumps.

Oh, and you'll need valves on the outlets of the two circuits to prevent feedback. Should have mentioned that earlier.

Ok...

I think that a 2x3 rad would be overkill, but hey, maybe that's just me...

Either way, the point was to take the rad out of the wb loop.

Seeing that the rad has an ideal (relatively low) flow rate, and that the wblock needs the highest rate possible (without causing cavitation in the pump), then the above idea is excellent, and with a bypass valve, the rad can be fine tuned for maximum cooling.

The alternative is to put everything in one loop and use multiple heatercores in parallel, and that's just not practical.

I'm going to have to disagree with this theory... I read that article, but it just doesn't make any sense to me as to why a rad would dissipate more heat with lower flow. Lets consider what the temp of a rad with a low-flow would look like (assuming you straightened a rad out to make it easy to think) - water would have more time to cool down as it passes through the rad. Asuming flow from left to right (there should be a picture in your head now of a pipe with fins - a rad - going left to right with hot water going into the left side) - the left side of the rad would be hot, and the right side of the rad would be fairly close to ambient. As flow rate gets lower and lower, the temp of the rad on the right side would approach ambient, and so coolant leaving the rad would be closer to ambient temp. Coolant at ambient is good, but a rad in these circumstances is a BAD thing (I will get to why)

Consider a high-flow through this straight rad. The water flowing through the rad would not decrease in temp much (as we approach infinite flow, the temp wouldn't change at all), and so the entire rad surface would be hotter, and coolant leaving the other side would be hotter. Hot coolant going back into our system is not what we want - but really, this IS how we should be running our systems.

In the first case (low-flow), only the first part of the rad is above ambient temp, and is therefore actually doing work in removing heat from the system. The farther down the rad you move, the lower the delta temp to ambient, and the less heat that part of the rad is dissipating. In the high-flow case, the entire rad has a high delta-T to ambient, and will be working as efficiently as it can.

As for your water temp leaving the rad, it's irrelevent. We should not be concerned with the temp of a specific volume of water at some specific posistion within our cooling system, but with the overall amount of heat that our system is absorbing and dissipating. The higher the flow-rates, the bigger the delta-T's we will get in both the rad and the block, and the cooler our CPU's will run.

Getting back more into a real-life rad, when we start bending it around to fit into an acceptable area we also start introducing turbulence, which helps the water dissipate it's heat to the rad, and higher flows will help with this as well.

If you all have problems understanding what I'm picturing in my head, I can whip out MS Paint for a bit and see what I can make - just ask.

MS-Paint? I don't think that it'll be necessary...

Well, there's a lot of different takes on this whole rad issue. Some people believe that rads perform better at a faster flow rate, and some (like me) believe that regardless, rads probably have a sweet spot, as seen in the OC radiator roundup.

Now, I can't explain this sweet spot, and I haven't seen anyone come up with a reasonable explanation for it either.

Regardless, since inducing a high flowrate will cause additional heat to be induced, then any alternative is better.

So the question is, is there a sweet spot for rads, or do we have to shoot for the most flow for the cooling?

How do you figure that inducing the higher flow creates more heat? Now I do agree that of course it will make some heat - but if you are trying to convince me that the additional friction (both between water and rad/tubing, and between water/water from turbulence) is creating that heat then I don't think that it will be a measurable amount of extra heat, and we should ignore it. If you are trying to convince me that this extra heat is because we need bigger pumps to create the flow, then it is definatly a possibility - but perhaps then we need to look at optimizing the system in other ways to get more flow out of the smaller pumps we already have.

A slow flow rate through the rad is beneficial the same way a high flow rate through the block is beneficial. In the block, you want the newly heated water OUT of the block as soon as possible so that new COLD water can absorb the fresh heat. In a rad, the opposite is desired. You want the warm water in the rad as LONG as possible to disperse as much heat from the fluid to the rad as possible. I don't know how to REALLY make that simple to understand, but that touches on the idea. Trust me, it is correct.

As far as the reason that I redesigned the flow like I did, I wanted to ensure that no water that is freshly heated from the block gets immediately recirculated back to the block. That will make a keychain very quickly. The only way to prevent that is to have a rad inline with the block before the res. Now, this rad needn't be a huge heater core or nothing, but SOME cooling is absolutely required to prevent this scenario.

The reason I chose a super-sized radiator for the secondary cooling loop is twofold: 1) I wanted to maximise heat transfer out of the water by allowing it the maximum amount of time in the radiator, and 2) prevent Athlonnerd from having to hook up fans to the thing. With a large core like that, heat transfer to the surrounding air can be achieved if it is just exposed to the outside air ... no forced airflow is required. It'll just give you a quieter box with maximum results. Feasably, with that setup, al you would need is a couple low CFM fans in your case circulating air, and t would be nearly silent (especially with a setup like Antec has with the fan/temp controls). You can get away with running a much smaller radiator, if you don't mind running fans through it. In the modified design idea I proposed, the second loop will serve as nothing more than a secondary water cooler, further dropping coolant temperature.

But as far as completely isolating your rad from your water block loop, it is a fool's errand. You will run a severe risk of damaging your processor. In the modified design with the large core, you could potentially run multiple blocks in parallel through the high-flow pump and still see outstanding results.

My .02

But you see, I don't want the water in the rad as long as possible. To put my previous post into a very short sentence... I want the still somewhat warm water in the rad out as soon as possible to get even warmer water back into it.

The warmer the rad is, the more heat it can disperse into the air. The more heat it can disperse into the air, the lower the overall temp of the system will be. I'm not aiming at the lowest water-temp coming out of the rad, I'm aiming at the lowest overall system temp where the heat input from the cpu (and pump if ya wanna be picky) stabalizes with the heat output from the rad.

Oh, and Ben, you can pretty much count on the fact that the slower the flow through the rad, the more heat is wicked away.

There are a few rules that apply to this that go into calculating the sweet spot.

1) The longer the fluid is in the rad, the more heat is xferred (until temp is normalized between coolant/air). This is just plain common sense.

2) The higher the dT bewwen the coolant/air, the faster the heat will xfer. Again, this is simple thermodynamics.

3) The wildcard is the rate that heat is xferred into the coolant. This is your hot processor cooking your block.

To calculate the sweet spot, you have to balance the rate that heat enters the system with the rad's capability to exhaust that heat. Since different systems enter different amounts of heat to the coolant, there is no way to really give a firm sweet spot to the rad, especially since flow speed affects both the rad and block in inverse proportions. Also, the temp of both your coolant and the surrounding air play a large part in efficiency. You can take a crappy rad and stick it in a block of ice and it'll work great, if you catch my drift.

That hopefully made some sense, and I'm sure one of the EngMonkeys will provide a formula, but rad performance is specific to each system.

Ah, and Cova, that is the hell of it. The reason people with big rads do well is that not only can they have faster flow, but they keep their coolant in there longer. I see your point, though. Your best bet in those systems is a slow flow system that, say, has the equivalent flow diameter of 1-1.5" hosing on your .5" system. While your pressure will remain high, you'll have a maximum amount of water in it at a time. That is similar to what happens in your car. The downside of a rad like that is that they are large (this is why I suggested the full size radiator in my previous example as a good idea ... maximum time for cooling).

airspirit - could you like, re-word that last post of yours. Maybe it's just me not thinking straight right now (air-conditioner is broken in this building right now and it's like 35C in my office, and I'm really tired cause work is boring), but I have no idea what you were just trying to tell me.

He he... I used to think that way too, until I came across this thread .

(You can also see the process by which I made sense of it all!)

I also read most of that thread as it was being made, and though I disagree with a few small points in it, I didn't feel like posting about them at the time. Anyways - which part of my idea did you "used to think that way" about, that you now think differently? The part about bigger pumps adding more heat for increased flow, or the part about friction of the water not making a difference worth worrying about?

I understand and agree with what you said.

The best explanation that I've been able to figure out is that if the flow rate is high in the rad, then it is turbulent, therefore can dissipate more heat, but that's countered by the heat that the rad's restrictions induce to the water (not very significant, but enough to skew a few numbers).

Did you see the radiator roundup at OC?