Thoughts on water pressure & flow rate...

[MMZ_TimeLord]

Okay, I'm curious, is there a common theory that a pump that provides higher water pressure and flow rate will be more efficient at cooling a water block? :shrug:

I've been using my system based on the theory that low pressure and flow rate and large copper block with high internal surface area will allow the water to pick up more heat and carry it away. That, coupled with the larger radiator (trans cooler) that I use, seems to work very well.

Any opinions and or theories? Experiences and real numbers are also a plus.

[8-Ball]

Replied over at Bit-tech

8-ball

[Graystar]

You're opening a can of worms with this one!

First off, what do you mean by "efficient"? Higher pressure increases flow, and that affects the system performance. But people have different concepts of what is "efficient", usually confusing efficiency with performance.

If you are considering the power consumed for a given amount of work, you are talking about efficiency. If you are considering the max amount of work that can be done, you are talking about performance. Generally, we talk about performance and not really worry too much about efficiency (although I personally give efficiency a lot of thought.)

So, back to your question. There is a basic physics-based answer, and then there is the more complex application of the physics to our WC systems. Watch the bouncing ball...

Faster flow always transfers more heat. That will always be true. So if you increase the flow rate through your waterblock, you will remove more heat from the CPU. Also, if you increase the flow through your radiator, you will transfer more heat from the water to the radiator. That's the basic physics part.

Now, lets step into reality. Reality tosses in three factors that may kill any benefit from increased flow. First is the difference in performance gain between the waterblock and the radiator.

When you increase flow, the waterblock performs better. That performance increase is attributed solely to increased flow, because we assume that the temperature of the water going into the waterblock remains the same. That's fine. However, the temperature of the water coming *out* of the waterblock has dropped (just happens to be a function of increased flow.) So our radiator will have two factors affecting its performance; increased flow and lower inlet temperature. A lower inlet temperature counters the gain from increased flow.

So when you increase flow the radiator works better, but it doesn't get the same boost as the waterblock. What this means is that the radiator will have to work harder to get rid of all the heat that the waterblock is sending its way.

The second factor is air. In the end, a watercooling system turns into an air cooling system. The amount of heat that you can ultimately remove is based on the relationship between your fans and radiator. So while you may make your waterblock work better with increased flow, you may simply be stuck with that extra heat if you can't get rid of it fast enough.

The last factor is the pump. Increasing pressure is the most costly operation for a pump in terms of watts consumed. The centrifugal pumps that we use are so inefficient for this task that they end up adding heat to the water. Usually, we increase pressure by getting a bigger pump. However, that means more heat added. So the benefits realized from increased flow may be wiped out by the additional heat.

So...not so simple.

[MMZ_TimeLord]

I replied here as well as 8-ball.

Thanks Graystar... I was mainly talking about performance vs. flow rate/pressure.

I was pretty sure that some of those pumps added heat.

My radiator has three thermister controlled fans that move air through it (nearly silent) and they won't go full RPM unless the air temp reaches 110° F. This has moved a LOT of air across it. And it's surface area is probably about 8 times that of my water blocks put together.

And... as to the water taking more heat away as the flow increases... I'd agree. But, as you say, the incerased flow rate also lowers the tempreture of the water coming out compared to a lower flow rate. This DOES indeed increase the requirements of the radiator to remove the heat that is in the water, but it also does not allow the same radiator time to get the same amount of heat from the water. Ergo: your mentioning the need for a higher performing radiator.

Thanks for the replies... this makes me more confident that I do indeed have a fairly balanced setup and will continue to run it as is, other than maybe getting a little more "efficient" and organized case layout.

[8-Ball]

Originally posted by MMZ_TimeLord
And... as to the water taking more heat away as the flow increases... I'd agree. But, as you say, the incerased flow rate also lowers the tempreture of the water coming out compared to a lower flow rate. This DOES indeed increase the requirements of the radiator to remove the heat that is in the water, but it also does not allow the same radiator time to get the same amount of heat from the water. Ergo: your mentioning the need for a higher performing radiator.

Not true.

This comes up a lot. There are a few threads discussing this. Faster flow equals greater efficiency.

It is necessary to think backwards.

You have a fixed air temp.

If the efficiency is improved, then the average water temp will be reduced. Equally, the efficiency of the waterblock will be increased, leading to a reduced delta T between the cpu and the water, which is already cooler.

However, increasing the flow will likely introduce more heat to the water through the pump. When the additional heat from increasing the flow rate exceeds the reduction in temp expected from the improved efficiency, you will reach the point of diminishing returns. Go no further.

It is a common misconception that water needs to spend time in the radiator for optimum performance. You need to try and detach yourself entirely from the flowing coolant and just imagine that there is heat coming in and heat going out. Changing the flow rate effects the efficiency of the heat exchangers carrying out these steps.

Hope this makes sense. (I think I get better at explaining it each time I give it a go, though there may still be mistakes.)

8-ball

[bigben2k]

Aw heck, I struggle with the explanation myself!

The statement isn't entirely innacurate, but the perspective is misleading.

The simplest way to put it, is that a higher flow rate will reduce the thermal resistance, i.e. the ability of the water/coolant to pick up or drop-off heat. So wether it's a waterblock or a radiator, it doesn't matter: more flow is better.

Seen in that light, it also should become more obvious that flow is far, far more critical in the waterblock than in the rad, because the heat source is a lot more concentrated in a block, where in a rad/heatercore, the surface area is relatively huge!

There is a balance point between the rad releasing the heat, and the block putting it in the coolant, but as stated above, the radiator is not the worry point, the waterblock is, and by far!

[MMZ_TimeLord]

Maybe I should get a little bigger version of the same pump I'm using now... and try different flow rates ... then post the results immediatly after a taxing benchmark like 3-DMark or some such.

Sound good?

[MMZ_TimeLord]

The pump I have is the M60AUL and the one I will get is the M200AUL... both are magnetic drive and from the same company.

Here is the company's pump info...

I'll get the pump tomorrow, check my fittings to make sure they will not be "blown off" by the increase in pressure... and post my results as soon as I can.

[bigben2k]

To quote MMZ:

Inside the reservoir is my $20 fountain pump from Home Depot. It doesn't flow much (60gpm) I think with only a 10ft head. But it keeps the water speed slow enough to pick up lots of heat.

(From your first post).

That's a misconception. More flow is better, both for the block, and the rad.

Do you have any specs or model# for that pump? A 10 foot head for $20 sounds pretty good!

[bigben2k]

(Cross replied!)

Ok so for $20, you got 60 gph (not gpm), and a 2.5' head.

Your upgrade doesn't sound too bad, but I think you can do better.

[MMZ_TimeLord]

I remembered incorrectly... 2' head... check the link above...

[bigben2k]

For you only, because you're got a res for the pump, I would highly recommend the MaxiJet 1000, available from BeCooling (link on the ProCooling frontpage), for $20.

It'll give you a bit less than 6' of head, with a max flow rate of 265 gph.

If you really want to crank it up, get two of them, and run them in series (hold on to your clamps!) Total head: approx. 10'.

[MMZ_TimeLord]

Nice pump... problem is I don't think it will fit in my reservior.

My internal measurements for the reservior are 6.5" x 3" x 4.75".

From BeCooling's site... As pictured: 4.5" long, 2.75" wide (3.25" with base), 4" tall (4.5" with base)

So it MAY fit... but MAN that would be tight!

AND... it's $30 instead of $20...

[bigben2k]

Look again: the kit is $30, the retail box is $20. One would fit, two won't, unless you rebuild that res.

[MMZ_TimeLord]

Is that pump adjustable? I want to be able to vary the flow rate. That way I can test our various theroies and understandings or misunderstandings.

[bigben2k]

No, an adjustable pump is extremely rare, within our application. A (recent) exception is the HydroThruster 500, but it's big bucks.

You can simply add a ball valve in your loop although, a needle valve would give you a finer control, but could be a tad more expensive, and hard to locate. You'll also need a flowmeter, if you're going to go that route: I just picked one up on Ebay for $14.

[MMZ_TimeLord]

Okay... the "Mini-Jet" line of pumps on BeCooling are VERY similar to what I have... just a little higher flow rate... I found another site that has several models with 82 – 153 GPH flow... adjustable.

I may give one of those a try first as extremely high flow rates are not something I am prepared to tackle yet.

We'll just have to see how much time I have... :shrug:

[jaydee]

Originally posted by bigben2k
No, an adjustable pump is extremely rare, within our application. A (recent) exception is the HydroThruster 500, but it's big bucks.

Hehehe, now that I have that very pump comming I need to get a flow meter.

[Cathar]

Here's some thoughts to ponder on.

1) Regardless of the flow rate, in a closed loop system the water spends the same amount of time in the radiator and the water-block.

2) It's commonly accepted that faster air-flow on a heatsink cools a CPU better (and more specifically - improves heat transfer), so what makes people suddenly think that the story is different when water is used?

[8-Ball]

Originally posted by Cathar
2) It's commonly accepted that faster air-flow on a heatsink cools a CPU better (and more specifically - improves heat transfer), so what makes people suddenly think that the story is different when water is used?

My point exactly. People seem perfectly happy to accept that faster flow rate in a waterblock is a good thing, but are often stubbonly insistent that the situation is not the same in a radiator.

I believe that it could be a case of those in the know misunderstanding how people agree that increased flow affects efficiency in the waterblock.

I get the impression that a lot of people see it that most of the benefit comes from the water in the block being "replaced" by cooler water more quickly, (please forgive the horrible description/simplification!) rather than the effect of the turbulence arising from increaed flow breaking down the barrier layer at the surface/interface.

Using this same logic, there is a misconception that "because the water spent less time in the water block, it must have picked up less heat", therefore, upon entering the radiator, the temperature difference will be lower and the efficiency will suffer.

That and the insistency on considering a single "packet" of water and how this is "treated" by the heat exchanger, rather that the "water flow" as a whole.

It is beginning t dawn on me that I have been babbling, and before this bottle of wine has eevn more of an effect on my, I shall stop.

8-ball

[Cathar]

Originally posted by 8-Ball
Using this same logic, there is a misconception that "because the water spent less time in the water block, it must have picked up less heat", therefore, upon entering the radiator, the temperature difference will be lower and the efficiency will suffer.

What's extremely false about this misconception is that in a closed loop system the water spends exactly the same amount of time in the block regardless of flow rate (as stated above).

[8-Ball]

Originally posted by Cathar
What's extremely false about this misconception is that in a closed loop system the water spends exactly the same amount of time in the block regardless of flow rate (as stated above).

It's very hard to explain this in conjunction with the other prinicples in such a way that someone might be able to visualise whats going on. It has always been my belief that unless you can truly visualise what's going on, you can't really understand the principles.

8-ball

[Cathar]

Originally posted by 8-Ball
It's very hard to explain this in conjunction with the other prinicples in such a way that someone might be able to visualise whats going on. It has always been my belief that unless you can truly visualise what's going on, you can't really understand the principles.

8-ball

The best way to think of it is like this.

In a fully bled system, name any time in which there is no water in the waterblock?

ie. there is no such time.

Therefore there is water in the block all the time. Regardless of the flow rate, the water spends the same amount of time in the block.

If we want to talk about packetised concepts, think of a car running around a circular race track. For an example, let's assume that the waterblock is a 60m (meter) stretch of road, and the racetrack is 600m long.

How many times per hour is the car inside the 60m section, and for how long?

If the car travels at 10m/s, it's in that 60 meter stretch for 6 second. In an hour the car travels 36000m. The racetrack being 600m long, means that the car go around the track 60 times in an hour. The amount of time spent in the 60m waterblock section is 6 seconds x 60 = 360 seconds.

If the car travels at 20m/s, it's in the 60m stretch for 3 seconds per lap. In an hour it travels 72000m, and will have completed 120laps in that time. 120 x 3 = 360 seconds.

The car has spent exactly the same amount of time in the "waterblock" stretch, regardless of its speed.

[8-Ball]

That's a nice analogy which should be able to get the message across, if drilled in with the point that increased flow rates increase heat exchanger efficiency.

8-ball

[8-Ball]

PS, that's enough for one night. The wine takes its toll and I'm falling asleep.

8-ball

PPS I've got to be in the library, awake and attentive in less than seven hours