Flow rates and Temps

There has been a debate that reducing the flow rates would result in better load temps because the water has more time to absorb the heat from the load and It also has more time in the radiators to reduce the temps
I have long argued that the above will give a appearance of lower temps but not result in lower load (CPU) temps than at higher flow rates (excluding any extra heat generated to move the water faster)

At low flow rates the water exiting the radiators will be lower than at high flow rates. but at the same time the water exiting the load will be higher, As the same effect is being applied to both the heating and the cooling parts in the loop. The result is the reduction in temps leaving the rads will be cancelled out by the increase in temps leaving the load.

In a simple world one would think the results of high and low flow rates would result in the same average water temps and therefore load temps would be the same.


In the test below you can see that the average water temps between high and low flow tests are similar but the load temps ( C Hot ) have rocketed up.

So you can see that higher flow rates result in lower load temps. (excluding any extra heat generated to move the water faster)


How i've tested
below is a pic of my quad 226watts Qmax TEC chiller. You will also see two gpu blocks with temp probs measuring the water temps of the hot side of my tec rig . One before the chiller and one after it.

I have then reduced the flow rate a little by turning the tap close to the PSU

in the next pic you can see the resulting cpu temps and the t-balancer navigation

C Cold is the temps of the cold side of the chiller as close as i can get
C Hot is the temps of the hot side of the chiller as close as i can get

H Water is the temps of the water exiting the hot side of the Chiller block
C Water is the temps of the water exiting the cold side of the Chiller block

http://www.overclock.net/attachment....1&d=1234662466
http://www.overclock.net/attachment....1&d=1234662466
http://www.overclock.net/attachment....1&d=1234662466

Quote:

---

Originally Posted by Ultrasonic2

The result is the reduction in temps leaving the rads will be cancelled out by the increase in temps leaving the load.

---

that sums this pretty much.

Thank you for clearing this up, excellent post by the way.

+1, thanks for taking time to do and post that, very helpful!

Nice to have what physics tells us validated. Kudos for taking the time.

Nice chiller too, too bad mine is like 1/4 finished and it has no load. You'd have to see my thread as to why I'm building a chiller. :D nice to see some serious TEC action.

Thx for sharing the testing results! nice job

One of the reasons for this test this is I’ve recently moved all my radiators outside of the nouse to take advantage of the lower ambient temps and since the rads are out side I no long have to worry about fan noise so I can run my fans at 100% all the time. I have also added an extra radiator

I was expecting to get a 10 degrees C drop in temperatures but I hardly saw any gains. The logical cause of this was a reduction in flow rates due to the extra rad and 30meters of extra tubing

I plan to put a budget but powerful A/C pump in line to see how much difference this will make. Im also going to try a semi split loop. I will have my rad pumps pushing through the rads and around the 30m of tubing . I will then install a pump which rotates water around the chillier and water blocks. The two loops will then come together at a point to transfer the heart from each loop.

In the rad loop I should see an improvement In flow rates as I’ll be reducing the tubing and water blocks that these pumps need to pushed though.

The PC loop should see significant improvements in flow rates as there will be little restriction to flow.

Now I’ve considered doing a semi split loop before but the numbers have never made it viable. However in this case I think it might create better temps than if I had all the pumps in series (in line) with everything else. I think a split loop could work this time because of the significant reduction in flow around the extra tuning and radiators.

Well im going to test it and see

ok here's a pic of my very messy split system

http://www.overclock.net/attachment....1&d=1234766886

here's a pic of the temps without the pump going at all ( Pump off, Valve closed)
http://www.overclock.net/attachment....1&d=1234766886

here's a pic of the temps with the pump in series ( Pump on, Valve closed)
http://www.overclock.net/attachment....1&d=1234858480

Here's a pic of the temps with the system in a semi split mode ( Pump on, valve open )
http://www.overclock.net/attachment....1&d=1234766886

So as predicted the semi split system produced the lowest temps.

thanks for shutting those guys up who think low flow is better

Quote:

---

Originally Posted by eligray

thanks for shutting those guys up who think low flow is better

---

Unfortunately that is not the case. The debate still rages on in the Boars Thread :confused:

care to give a link to that thread?

I think the hardware has to be used for the purpose it was intended. ie: If you have a system designed with all high flow rate components, and then simply reduce the flow, the temps will increase. I believe the same is true for the opposite, build a system with low flow high restriction components, then stick a huge pump on it, the temps will increase as well. I've been searching for 1366 capable water blocks for awhile, I seen one test where the Aqua CUplex using their recomended Eheim 1046 outperformed the GTZ block by 5c, then when I seen a comparison here, where they swapped the same block into a high flow system, they stated that it wasn't even in the same league as the GTZ. So, either one of them didn't get the correct contact in one of their tests, or mixing and matching high flow components with low flow doesn't work as hoped ? Until I see a high flow, and low flow system tested on the same cpu/mobo for correct baseline results, then start mixing and matching, to me the question remains unanswered. There's simply too many variables involved.

Quote:

---

Originally Posted by FragMagnet

I think the hardware has to be used for the purpose it was intended. ie: If you have a system designed with all high flow rate components, and then simply reduce the flow, the temps will increase. I believe the same is true for the opposite, build a system with low flow high restriction components, then stick a huge pump on it, the temps will increase as well. I've been searching for 1366 capable water blocks for awhile, I seen one test where the Aqua CUplex using their recomended Eheim 1046 outperformed the GTZ block by 5c, then when I seen a comparison here, where they swapped the same block into a high flow system, they stated that it wasn't even in the same league as the GTZ. So, either one of them didn't get the correct contact in one of their tests, or mixing and matching high flow components with low flow doesn't work as hoped ? Until I see a high flow, and low flow system tested on the same cpu/mobo for correct baseline results, then start mixing and matching, to me the question remains unanswered. There's simply too many variables involved.

---

wut

More water passing through the block = greater delta T between the block and the coolant stream = more efficient thermal transfer = better temps.

It's basic physics. There is no situation in which having lower flow-rates will give you better temps, except if the pump is now working so hard that it's dumping heat into your loop.

theoretical physics and applied physics are two different things. I'm basing my opinion on a few thousand hours of airflow testing on a flow bench. That, and the fact that in some tests indicate that larger pumps, larger rads, larger hoses, in a high flow system, they only yield 1-2 degrees better temps. Either due to the pump heat, or you've gotten to the point of diminishing returns.
All I'm saying is that I haven't seen enough data yet to draw a conclusion. In the case of the OP's, physics said adding an additional rad, faster fans, and moving the rads outside would give lower temps, in his case it didn't.
It's not just the parts used, but how they work together that gives the end result.
Before SteveRo's testing there were a lot of high flow believers swearing up and down that the Boreas wouldn't work for one reason or another. Yet that low flow system produced better results than his high flow system ? Sure it had low wattage air cooled TECs. But the end results are similar, and so is the price point.

Quote:

---

Originally Posted by FragMagnet

...the fact that in some tests indicate that larger pumps, larger rads, larger hoses, in a high flow system, they only yield 1-2 degrees better temps. Either due to the pump heat, or you've gotten to the point of diminishing returns.

---

Well, duh. Obviously you're going to get diminishing returns, again as your system reaches equilibrium and the increase in thermal transfer efficiency reaches the narrow end of the logarithmic curve. But that's not what you posted above:

Quote:

---

Originally Posted by FragMagnet

...build a system with low flow high restriction components, then stick a huge pump on it, the temps will increase as well.

---

You were saying that in some situations lower flow rates will give better temps - which is wrong. Higher flow-rates will not always give you appreciably better temps, but you're never going to get better temps with lower flow than you would with high.

Quote:

---

In the case of the OP's, physics said adding an additional rad, faster fans, and moving the rads outside would give lower temps, in his case it didn't.

---

Indeed - to quote, diminishing returns. I can guarantee that the temps were lower, the difference was just not significant enough to register on the measurement instruments used.

Quote:

---

It's not just the parts used, but how they work together that gives the end result.

---

Well, yes. Clearly: a waterblock on its own is useless, and the Supreme will suck big-time when used with something like a piddly Tt P500 or a low-end Eheim. But that doesn't magically mean that something like a FuZion will give you better temps with the P500 than it would with, for example, an Iwaki. All dogs are not cats, but not all things that are not cats are dogs.

Quote:

---

Before SteveRo's testing there were a lot of high flow believers swearing up and down that the Boreas wouldn't work for one reason or another. Yet that low flow system produced better results than his high flow system ? Sure it had low wattage air cooled TECs. But the end results are similar, and so is the price point.

---

I'm losing track of your argument here - the Boreas uses TECs, which will give you temps better than the most efficient ambient cooling system ever designed (as they're not limited to ambient) provided they're adequately cooled, but this proves low flow is better than high...how? And what does relative cost of product have to do with liquid dynamics and physics?

Edit: I should reiterate that for the sake of argument I'm ignoring pump heat-dump issues and focusing merely on low/high flow.

[QUOTE=SoulsCollective;3691455]
I'm losing track of your argument here \QUOTE]

No kidding ! Probably because I wasn't arguing, just trying to exchange ideas based on my experience in different fields. There is no 1 "absolute" answer for all situations, you keep stating your position as an absolute, then go on to say "except" this, and "except" that. Like I said I don't come to xtreme to argue, there's dozens of forums online for that, xtreme used to be the exception, but clearly that's changed since I was here last !

Perhaps people here should respect that each person will believe different things and there is no apparent reason why we should force that on another person. I do believe in any situation that higher flow WILL give better temps it just makes logical sense.